Protocol

0 downloads 0 Views 4MB Size Report
Placebo patients begin LY450139 (escalated to 140 mg) at or after V14 (Wk 64). ...... V2b V3 V4 V5 V6 V7 V8 V9 V10 V11 V12 V13 V14 V15 V16 V17 V18 V19 V20 EDc ...... tia l. P a ge 92. S tudy. S c he dule. Footnote s. (c onc lude d) jk. A t V.
Missing:

Protocol This trial protocol has been provided by the authors to give readers additional information about their work. Protocol for: Doody RS, Raman R, Farlow M, et al. A phase 3 trial of semagacestat for treatment of Alzheimer’s disease. N Engl J Med 2013;369:341-50. DOI: 10.1056/NEJMoa1210951

Items included in Protocol/Statistical Analysis Plan/Informed Consent PDF 1. Original study protocol 23 August 2007 2. Final study protocol 08 September 2010 (includes previous version of the protocol) 3. Protocol amendment summaries 4. Original statistical analysis plan 5. Final statistical analysis plan 6. Master consent form 7. Final informed consent approved by central IRB

Page 1

Confidential Information The information contained in this protocol is confidential and is intended for the use of clinical investigators. It is the property of Eli Lilly and Company or its subsidiaries and should not be copied by or distributed to persons not involved in the clinical investigation of LY450139 unless such persons are bound by a confidentiality agreement with Eli Lilly and Company or its subsidiaries.

Protocol H6L-MC-LFAN Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo

LY450139 Phase 3, parallel, placebo-controlled, dose/response, delayed-start 88-week study in patients with mild to moderate Alzheimer’s disease.

Eli Lilly and Company Indianapolis, Indiana USA 46285 Protocol Approved by Lilly: 23 August 2007

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 2

Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo Table of Contents Section 1.

Page

Introduction....................................................................................... 11

2. Objectives ......................................................................................... 14 2.1. Primary Objective ........................................................................ 14 2.2. Secondary Objectives and Measures ............................................. 14 3. Investigational Plan ........................................................................... 17 3.1. Summary of Study Design............................................................ 17 3.1.1. Screening Phase (Study Period I).......................................... 18 3.1.2. Initial Treatment Phase (Study Period II).............................. 19 3.1.3. Delayed-Start Phase (Study Period III) ................................. 19 3.1.4. Follow-Up Visit (Visit 801).................................................. 19 3.2. Discussion of Design and Control................................................. 20 4. Study Population ............................................................................... 21 4.1. Inclusion Criteria.......................................................................... 21 4.2. Exclusion Criteria......................................................................... 21 4.2.1. Exclusion Criteria for All Patients ........................................ 21 4.2.2. Rationale for Inclusion or Exclusion of Certain Study Candidates............................................................................ 24 4.3. Discontinuations........................................................................... 24 4.3.1. Discontinuation of Patients................................................... 24 4.3.2. Discontinuation of Study Sites.............................................. 25 4.3.3. Discontinuation of the Study ................................................ 25 5. Treatment.......................................................................................... 26 5.1. Treatments Administered.............................................................. 26 5.1.1. Initial Phase of Study ........................................................... 26 5.1.2. Delayed-Start Portion of the Study ....................................... 27 5.1.3. Additional Dosing Information............................................. 27 5.2. Materials and Supplies.................................................................. 28 5.3. Method of Treatment Assignment................................................. 28 5.4. Rationale for Selection of Doses in the Study ............................... 28 5.5. Selection and Timing of Doses ..................................................... 29

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 3 5.6. Blinding ....................................................................................... 30 5.7. Concomitant Therapy ................................................................... 30 5.7.1. Allowed Medications ........................................................... 30 5.7.2. Excluded Medications .......................................................... 31 5.7.3. Monitoring Medications ....................................................... 31 5.8. Treatment Compliance ................................................................. 32 6.

Efficacy, Health Outcome/Quality of Life Measures, Safety Evaluations, Sample Collection and Testing, and Appropriateness of Measurements..................................................... 34 6.1. Efficacy Measures ........................................................................ 34 6.1.1. Primary Efficacy Measures................................................... 34 6.1.2. Secondary Efficacy Measures: Biomarkers.......................... 34 6.2. Additional Secondary Clinical Outcome Measures ....................... 35 6.3. Safety Evaluations........................................................................ 36 6.3.1. Adverse Events .................................................................... 36 6.3.1.1. Adverse Events in General.............................................. 36 6.3.1.2. Serious Adverse Events .................................................. 37 6.3.2. Other Safety Measures ......................................................... 38 6.3.2.1. Vital Signs ...................................................................... 38 6.3.2.2. Electrocardiograms ......................................................... 38 6.3.2.3. Laboratory Evaluations ................................................... 39 6.3.2.4. Additional Considerations............................................... 39 6.3.3. Safety Monitoring ................................................................ 39 6.3.4. Complaint Handling ............................................................. 40 6.4. Sample Collection and Testing ..................................................... 40 6.4.1. Standard Laboratory Testing (Hematology, Chemistry, Urinalysis).......................................................... 41 6.4.2. Plasma Samples for Assessment of Aβ, LY450139, and Donepezil ...................................................................... 41 6.4.2.1. Timing of Plasma Sampling ............................................ 41 6.4.2.2. Plasma Collection Procedures ......................................... 41 6.4.2.3. Technique Used for Evaluation of Plasma....................... 42 6.4.2.4. Pharmacokinetic Analysis of Plasma LY450139 ............. 42 6.4.3. Stored Samples..................................................................... 42 6.5. Appropriateness of Measurements ................................................ 43

7. Data Quality Assurance..................................................................... 44 7.1. Data Assurance Responsibilities ................................................... 44 7.2. Direct Data Entry and Computerized Systems .............................. 44

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 4 8. Sample Size and Statistical Methods ................................................. 45 8.1. Determination of Sample Size ...................................................... 45 8.2. Statistical and Analytical Plans..................................................... 45 8.2.1. General Considerations ........................................................ 45 8.2.2. Patient Disposition ............................................................... 46 8.2.3. Patient Characteristics .......................................................... 46 8.2.4. Efficacy Analyses................................................................. 46 8.2.4.1. Analysis of Primary Outcome and Methodology for FDA .......................................................................... 47 8.2.4.2. Analysis of Primary Outcome and Methodology for CHMP....................................................................... 48 8.2.4.3. Efficacy Analysis of Secondary Endpoints— Initial Treatment Period .................................................. 49 8.2.4.4. Efficacy Analysis of Secondary Endpoints— Delayed-Start Period....................................................... 50 8.2.4.5. Subgroup Analyses ......................................................... 51 8.2.4.6. Analysis of Plasma Aβ.................................................... 52 8.2.4.7. Analyses of Biomarker Outcomes ................................... 52 8.2.4.8. Statistical Details of the Gatekeeping Strategy ................ 53 8.2.5. Safety Analyses—Initial Treatment Period ........................... 53 8.2.6. Safety Analyses—Delayed-Start Period................................ 54 8.2.7. Safety Analyses—Follow-Up Visit....................................... 56 8.2.8. Interim Analyses and Data Monitoring Committee ............... 56 9.

Informed Consent, Ethical Review, and Regulatory Considerations................................................................................... 57 9.1. Informed Consent ......................................................................... 57 9.2. Ethical Review ............................................................................. 57 9.3. Regulatory Considerations............................................................ 58 9.3.1. Investigator Information ....................................................... 58 9.3.2. Protocol Signatures .............................................................. 58 9.3.3. Final Report Signature.......................................................... 59

10. References......................................................................................... 60

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 5 Table of Contents (Concluded) List of Protocol Attachments Protocol Attachment LFAN.1. Study Schedule Protocol Attachment LFAN.2. Clinical Laboratory Tests Protocol Attachment LFAN.3. Screening Measures Related to Diagnostic Criteria

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 6

Abbreviations and Definitions



Amyloid β peptide

AChEI

Acetylcholinesterase inhibitors

AD

Alzheimer’s disease

ADAS-Cog

Alzheimer’s Disease Assessment Scale—Cognitive subscale

ADCS-ADL

Alzheimer’s Disease Cooperative Study—Activities of Daily Living Inventory

ADL

Activity of Daily Living

ADRDA

Alzheimer’s Disease and Related Disorders Association

AE

Adverse event

AIREN

Association Internationale pour la Recherche et l'Enseignement en Neurosciences

ALT/SGPT

Alanine transaminase

ANCOVA

Analysis of covariance

ANOVA

Analysis of variance

ApoE

Apolipoprotein E

APP

Amyloid precursor protein

Assent

Agreement from a child or other individual who is not legally capable of providing consent but who can understand the circumstances and risks involved in participating in a study (required by some Institutional Review Boards [IRBs])

AST/SGOT

Aspartate transaminase

Audit

A systematic and independent examination of the trial-related activities and documents to determine whether the evaluated trialrelated activities were conducted and the data recorded, analyzed, and accurately reported according to the protocol, applicable standard operating procedures (SOPs), good clinical practice (GCP), and the applicable regulatory requirement(s)

BACE

Beta-site APP amyloid cleaving enzyme

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 7

Blinding/Masking

A procedure in which one or more parties to the trial are kept unaware of the treatment assignment(s). Single-blinding usually refers to the subject(s) being unaware, and double-blinding usually refers to the subject(s), investigator(s), monitor(s), and, in some cases, select sponsor personnel being unaware of the treatment assignment(s).

CDR-SB

Clinical Dementia Rating—Sum of Boxes

CEC

Clinical endpoint committee

CHMP

Committee for Medicinal Products for Human Use

CNS

Central nervous system

Complaint

A complaint is any written, electronic, or oral communication that alleges deficiencies related to the identity, quality, purity, durability, reliability, safety or effectiveness, or performance of a drug or drug delivery system.

Compliance

Adherence to all the trial-related, good clinical practice (GCP), and applicable regulatory requirements

Confirmation

A process used to confirm that laboratory test results meet the quality requirements defined by the laboratory generating the data and that Lilly is confident that results are accurate. Confirmation will either occur immediately after initial testing or require that samples be held to be retested at some defined time point, depending on the steps required to obtain confirmed results.

CRF

Case Report Form (sometimes referred to as Clinical Report Form). A printed or electronic form for recording study subjects’ data during a clinical study, as required by the protocol.

CRO

Contract research organization

CRP

Clinical research physician

CSF

Cerebrospinal fluid

CSR

Clinical study report

CT scan

Computerized tomography scan

CYP450

Cytochrome P450

DMC

Data monitoring committee

ECG

Electrocardiogram

EMEA

European Medicines Agency

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 8

End of Study (Trial)

The date of the last visit or last scheduled procedure shown in the Study Schedule for the last active subject in the study

Enroll

See Study Entry Terms

Enter

See Study Entry Terms

EQ-5D

EuroQol 5-Dimensional Health-related Quality of Life Scale

FDA

Food and Drug Administration

GCP

Good clinical practice

GDS

Geriatric Depression Scale

IB

Investigator’s Brochure

ICD

Informed Consent Document

ICH

International Conference on Harmonisation

Intention to Treat (ITT)

The principle that asserts that the effect of a treatment policy can be best assessed by evaluating on the basis of the intention to treat a subject (that is, the planned treatment regimen) rather than the actual treatment given. It has the consequence that subjects allocated to a treatment group should be followed up, assessed, and analyzed as members of that group, irrespective of their compliance to the planned course of treatment.

Interim Analysis

Any analysis intended to compare treatment groups at any time prior to the formal completion of a trial

Investigator

A person responsible for the conduct of the clinical trial at a trial site. If a trial is conducted by a team of individuals at a trial site, the investigator is the responsible leader of the team and may be called the principal investigator.

IRB/ERB

Institutional review board/ethical review board: a board or committee (institutional, regional, or national) composed of medical professional and nonmedical members whose responsibility is to verify that the safety, welfare, and human rights of the subjects participating in a clinical trial are protected

IVRS

Interactive voice response system

Legal Representative

An individual or judicial or other body authorized under applicable law to consent, on behalf of a prospective subject, to the [patient's/subject's] participation in the clinical trial

MHIS

Modified Hachinski Ischemia Scale

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 9

MMSE

Mini Mental State Examination

NINCDS

National Institute of Neurological and Communicative Disorders and Stroke

NINDS

National Institute of Neurological Disorders and Stroke

NPI

Neuropsychiatric Inventory

Patient

A subject with a defined disease

PD

Pharmacodynamic

PET

Positron emission tomography

PI

Principal investigator

PIB

Pittsburgh Compound-B

PK

Pharmacokinetic/Pharmacokinetics

RUD-Lite

Resource Utilization in Dementia-Lite

SAE

Serious adverse event

Study Entry Terms

Screen

The act of determining if an individual meets minimum requirements to become part of a pool of potential candidates for participation in a clinical trial. In this study, screening involves diagnostic psychological tests, blood draws, electrocardiograms, and similar procedures. For this type of screening, informed consent for these screening procedures and/or tests shall be obtained; this consent may be separate from consent obtained for the study. Enroll/Randomize

The act of assigning a patient to a treatment. Patients who are enrolled in the trial are those who have been assigned to a treatment. Enter

The act of obtaining informed consent for participation in a clinical trial from patients deemed eligible or potentially eligible to participate in the clinical trial. Patients entered into a trial are those who sign the informed consent document directly or through their legally acceptable representatives. Subject

LY450139 H6L-MC-LFAN Confidential

An individual who is or becomes a participant in clinical research, either as a recipient of the test article or as a control. A subject may be either a healthy human or a patient.

Protocol Approved by Lilly: 23 August 2007

Page 10

Treatment-Emergent Adverse Event (TEAE)

Any untoward medical occurrence that either occurs or worsens at any time after treatment baseline and does not necessarily have to have a causal relationship with this treatment (also called treatment-emergent signs and symptoms [TESS])

ULN

Upper limit of normal, meaning upper limit of analyte’s reference range

vMRI

Volumetric magnetic resonance imaging

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 11

Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo 1. Introduction Alzheimer’s disease (AD) is a neurodegenerative disorder causing progressive decline in memory and other aspects of cognition. The average duration from onset of symptoms to nursing home placement is 5 to 7 years and from symptom onset to death is 7 to 9 years (Jost and Grossberg 1995; Smith et al. 2001). Currently available therapies are limited to cholinesterase inhibitors such as donepezil (Aricept ®) and an N-methyl-D-aspartate receptor antagonist, memantine (Namenda®). Although treatment with these drugs can lead to modest symptomatic improvement, their mechanism of action is related to manipulation of neurotransmitter systems, and there is no compelling evidence that they slow the underlying rate of disease progression (Irizarry and Hyman 2001; Schneider 2001). Although the cause of AD is unknown, several lines of evidence suggest that a 38-42 amino acid peptide known as amyloid β (Aβ) plays a central role (Schenk et al. 1995; Selkoe 1999; Hardy and Selkoe 2002; Cummings 2004). A definitive diagnosis of AD is possible only by examining brain tissue after death. Amyloid plaques and neurofibrillary tangles are required neuropathological findings for a definitive diagnosis. Amyloid β peptide, particularly the Aβ1-42 variant, is a major constituent of amyloid plaques. Mutations in the amyloid precursor protein (APP) gene that encode amino acid substitutions near the cleavage sites for Aβ formation and mutations in the gamma (γ)secretase complex have been linked to AD. Families with genetic mutations in the gamma secretase complex (PS-1 and PS-2) have an increased synthesis of Aβ. These findings suggest that Aβ synthesis is related to the pathogenesis of AD. Amyloid β peptide is generated from the APP by the sequential cleavage of APP by two proteases, beta-site APP amyloid-cleaving enzyme (BACE) and γ-secretase. Beta-site APP amyloid-cleaving enzyme first cleaves APP and generates a large, secreted protein product known as sAPPβ and a smaller membrane-bound fragment. This latter fragment is cleaved at several sites by γ-secretase to generate Aβ peptides of various lengths; however, the primary products of this cleavage are known as Aβ1-40 and Aβ1-42. Gamma-secretase is a complex enzyme, appearing to consist of four proteins known as presenilin (PS-1 or PS-2), nicastrin, Aph-1, and Pen-2 (Haass 2004). Although the synthesis of Aβ is becoming better understood, relatively little is known about its clearance. Animal studies suggest that Aβ is actively transported from brain interstitial fluid into blood (Shibata et al. 2000). A common polymorphism in the apolipoprotein E (ApoE) ε4 allele is considered a major risk factor for the development

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 12 of AD, and ApoE may be important in the clearance of Aβ from the brain (Holtzman 2001). Taken together, pathological data showing accumulation of Aβ and amyloid plaques in the brains of AD patients, and genetic data showing the involvement of APP and Aβ in human AD lend support to the amyloid hypothesis, which has attracted substantial attention to the development of new drugs that might slow disease progression by inhibiting Aβ synthesis. LY450139 is a “functional” inhibitor of γ-secretase that inhibits the synthesis of both Aβ1-40 and Aβ1-42 (Gitter et al. 2004). Following oral dosing, LY450139 has demonstrated dose-dependent inhibition of Aβ production in the brain, cerebrospinal fluid (CSF), and plasma in a variety of nonclinical in vivo animal models. A chronic, 5month study using APPV717F transgenic mice given LY450139 that compared LY450139-treated animals to vehicle-treated animals found that LY450139 caused a statistically significant reduction in brain Aβ. Similar time- and dose-dependent reductions of Aβ have been demonstrated in early-phase clinical studies in healthy volunteers and in patients with mild to moderate AD. Potential safety concerns with compounds that inhibit γ-secretase have been identified in part based on inhibition of cleavage of a protein known as Notch (Wong et al. 2004). Despite these theoretical concerns, adverse events (AEs) in studies using doses of LY450139 up to 140 mg once daily have been manageable, as detailed in the Investigator’s Brochure (IB). The present study will include monitoring for these and other potential effects, as described elsewhere in the protocol. More detailed information regarding the above discussions of disease mechanism, nonclinical efficacy and pharmacology, and clinical pharmacology and safety of LY450139 may be found in the IB. Considering the pathological and genetic evidence supporting the amyloid hypothesis, and considering the nonclinical and clinical studies of LY450139 demonstrating a reduction of Aβ in brain, CSF, and plasma, it is hypothesized that LY450139 may slow the rate of progression of AD. These potential benefits must be weighed against possible risks that may be associated with LY450139. The risk-benefit ratio would appear appropriate to begin a pivotal study assessing the effect of 100-mg or 140-mg LY450139 per day on the rate of progression of AD. This Phase 3 clinical trial is designed to assess the effect of LY450139 on AD progression using coprimary endpoints that include both cognitive and functional measures. Several amyloid-related biomarkers will be assessed to document the impact on the underlying cause of the disease. Many study participants are expected to be receiving symptomatic treatments along with the study drug; the protocol requires that treatment with such medications be continued, using stable doses for these patients. The study design has been discussed with the United States Food and Drug Administration (FDA), the European Medicines Agency (EMEA), and Health Canada; there is general LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 13 agreement with these regulatory authorities that the study design is appropriate for assessment of potential disease modification in AD.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 14

2. Objectives 2.1. Primary Objective The primary objective of this study is to test the hypothesis that LY450139 given orally will slow decline of AD as compared to placebo. Because of differences in regulatory requirements across regions, this objective will be assessed using several measures. In one region, one measure will be considered primary and the other secondary, and this relationship will be reversed for the second region. This is discussed more fully in Section 8. Both measures will be evaluated sometime between 64 and 88 weeks after initiation of treatment; the precise timing is specified in an ethical review board (ERB) supplement to this protocol. For the FDA, the primary objective will be assessed using a slope analysis of 2 coprimary outcomes, the Alzheimer’s Disease Assessment Scale—Cognitive subscore (ADASCog11) and the Alzheimer’s Disease Cooperative Study—Activities of Daily Living Inventory (ADCS-ADL), in which the specific hypothesis is that the rate of change over time (slope) for LY450139 will be significantly less than that for placebo, as assessed using a repeated-measures model. For the Committee for Medicinal Products for Human Use (CHMP), the primary objective will be assessed by change from baseline to endpoint using 2 coprimary outcomes, the ADAS-Cog11 and ADCS-ADL. The specific hypothesis is that the change from baseline to endpoint will be significantly less than that for placebo, as assessed by a stratified analysis of baseline-to-endpoint change.

2.2. Secondary Objectives and Measures Secondary objectives of this study are listed below. •

To test the hypothesis that LY450139 is a disease-modifying medication independent of acute symptomatic effects, as assessed using a “delayedstart” study design (Leber 1997; Whitehouse et al. 1998; Mohs et al. 2006) in which placebo patients will begin receiving double-blind LY450139 sometime after 64 weeks in the study. The specific hypothesis is that statistically significant differences will be observed in changes from study baseline to endpoint in ADAS-Cog11 and ADCS-ADL scores between patients originally assigned to active treatment and patients originally assigned to placebo who initiate active treatment at least 64 weeks later. This result would be consistent with a slowing of the underlying rate of disease progression due to treatment with LY450139; if LY450139 were a symptomatic treatment, one would expect an improvement in cognitive measures during the delayed start period such that the group originally on placebo would approximate scores from the group originally assigned to LY450139 by endpoint (88 weeks after baseline).

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 15 •

To provide supporting evidence that LY450139 is a disease-modifying compound, multiple biomarkers will be assessed. Some biomarker data will be collected only in subsets of patients and will therefore be fully described in study addenda; all are described here to contribute to greater understanding of the study objectives. Briefly, the following secondary hypotheses will be tested, using the indicated measures, in each case, the specific hypothesis being that LY450139 will be different from placebo. •

LY450139 will acutely reduce Aβ in plasma within 6 hours of administration.



LY450139 will attenuate the accelerated rate of decline in brain glucose metabolism known to occur in patients with AD (De Santi et al. 2001; Herholz 2003; Thal et al. 2006). This will be assessed in a subset of patients as part of optional Study Addendum 1, using fluorine-18 fluorodeoxyglucose ([18F]-FDG) positron emission tomography (FDG-PET).



LY450139 will alter the accelerated rate of decline in brain volumes known to occur in patients with AD (Schott et al. 2005). This will be assessed in a subset of patients as part of optional Study Addendum 2, using volumetric magnetic resonance imaging (vMRI).



LY450139 will reduce brain amyloid burden as compared to placebo. This will be assessed in a subset of patients as part of optional Study Addendum 3, using an amyloid-imaging PET tracer called Pittsburgh Compound-B ([11C]-PIB) (PIB-PET).



LY450139 will reduce the elevated concentrations of CSF tau proteins known to exist in patients with AD (Trojanowski 1996; Green et al. 1999). CSF tau will be collected via lumbar punctures, which are addressed in optional Study Addendum 4.



To compare the safety of LY450139 and placebo, through assessment of treatment-emergent adverse events (TEAEs) and changes in vital signs, laboratory evaluations, and electrocardiograms (ECGs).



To characterize population pharmacokinetics (PK) of LY450139, explore potential factors that may influence variability of PK, and explore the association of PK variables with efficacy, biomarkers, and safety parameters.



To test the hypothesis that LY450139 will slow the rate of decline of AD, compared to placebo, using the extended Alzheimer’s Disease Assessment Scale (ADAS-Cog14), assessed using a slope analysis from a repeatedmeasures model.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 16 •

To assess global clinical benefit of treatment with LY450139 as demonstrated through the Clinical Dementia Rating—Sum of Boxes (CDR-SB) and additional clinical outcomes instruments, including the Neuropsychiatric Inventory (NPI), Resource Utilization in Dementia— Lite, and EuroQol-5D (EQ-5D).

In addition, the following exploratory hypotheses related to biomarkers will be tested. •

An exploratory hypothesis to be tested using PIB-PET (optional Study Addendum 3) is that only patients demonstrating sufficient amyloid burden at baseline will respond to treatment as determined by clinical measures.



An exploratory hypothesis to be tested using CSF (optional Study Addendum 4) is that LY450139 will normalize (increase) CSF Aβ1-42 concentrations, which are known to be reduced in patients with AD (Motter et al. 1995; Galasko et al. 1998; Thal et al. 2006) and thus might be normalized during chronic treatment with a disease-modifying drug. Because treatment with LY450139 would be expected to acutely decrease CSF Aβ concentrations, CSF will be obtained for Aβ measurements approximately 24 hours after dosing (immediately before the next dose). Because the effect of daily dosing may be longer than 24 hours after chronic LY450139 administration, CSF Aβ1−42 concentrations might be decreased rather than increased at this endpoint measurement; thus, this will be considered an exploratory analysis.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 17

3. Investigational Plan 3.1. Summary of Study Design Study H6L-MC-LFAN (LFAN) will be a multicenter, randomized, double-blind, placebo-controlled, Phase 3 study comparing LY450139 140 mg, LY450139 100 mg, and placebo in approximately 1500 patients with mild to moderate AD. Patients who meet entry criteria will be randomized in a 1:1:1 ratio (500 per treatment arm) to 1 of the 3 treatment groups: LY450139 100 mg once daily, LY450139 140 mg once daily, or placebo once daily. Patients will be randomized by site and by mild or moderate MiniMental State Examination (MMSE) scores (for the purposes of this study, moderate will be defined as scores including 16 through 19, and mild as including 20 through 26). The primary hypothesis being tested is that LY450139 will slow the rate of cognitive and functional decline in AD as compared to placebo. In addition, sometime after 64 weeks of treatment, patients receiving placebo will begin receiving LY450139 (escalated to 140 mg/day) for the remainder of the study. This “delayed-start” design feature will evaluate whether patients originally treated with placebo improve during this period such that their cognitive scores subsequently approximate those of the patients initially treated with LY450139. The absence of such an improvement—in other words, maintaining a statistical difference between patients initially assigned to LY450139 and placebo following the delayed start—would support the conclusion that LY450139 slows disease progression and does not affect simply symptoms of disease. Patients and investigators will be blinded to the timing of the delayed start of active treatment for the placebo-treated patients. An ERB supplement will explain the actual timing. Figure LFAN.1 illustrates the study design. The Study Schedule (Protocol Attachment LFAN.1) details procedures and tests occurring at specific times during the study.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 18

Study Period III Delayed Start Period*

Follow-up visit

Study Study Period I Period II Screen Initial Randomized Treatment Period

100 mg LY450139 60 mg (2 wks) No treatment

140 mg LY450139 100 mg (2 wks) 60 mg (2 wks)

placebo

*

Visit: 1

2345678

Week:

0 2 3 4 5 6 8 12 20 28 40 52 64* 68* 72* 76* 80* 84* 88 92

9 10 11 12 13 14* 15* 16* 17* 18* 19* 20 801

Baseline

or ED

*Placebo patients begin LY450139 (escalated to 140 mg) at or after V14 (Wk 64). Abbreviations: ED = early discontinuation; wk = week

Figure LFAN.1.

Study design for Protocol H6L-MC-LFAN.

3.1.1. Screening Phase (Study Period I) At or before Visit 1, the study will be explained to the patient (and his or her legal representative, if applicable) and caregiver. Informed consent must be obtained before any study procedures are conducted or any excluded medications are discontinued. The screening period spans the time between Visit 1 to Visit 2. Ideally, screening will take from 3 to 14 days, but up to 30 days are permitted to allow for completion of Visit 1 screening procedures, assessments, and evaluation of results from laboratory tests and ECGs. Visit 1 is not considered complete until all screening procedures have been completed. Current or planned use of concomitant medications, the effects of vacations or absences on protocol compliance, and general compliance with the protocol will be discussed at Visit 1. Patients must meet all eligibility criteria (Section 4.1 and Section 4.2) to continue to Visit 2. Patients who do not meet all inclusion criteria or who meet any exclusion criteria will be discontinued from the study. At Visit 2, appointments should be made for all remaining visits and should be scheduled as close as possible to the target date, relative to Visit 2.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 19

3.1.2. Initial Treatment Phase (Study Period II) Study Period II is a double-blind treatment phase during which patients who meet entry criteria will be enrolled and randomized (Visit 2) to LY450139 100 mg, LY450139 140 mg, or placebo, administered orally once each day, preferably in the morning. To minimize the possibility of AEs, patients randomized to LY450139 will begin receiving 60 mg each day and will be gradually escalated to their assigned doses (Section 5.1.1). Dose reductions are allowed on a limited basis as described in Section 5.1. No dose reductions will be allowed beginning at Week 64 (Visit 14) of the study. Intent-to-treat (ITT) analysis will be based on initial randomization (140 mg once daily, 100 mg once daily, or placebo). During this study period, certain visits will require patients to take their study medication on-site. In addition, procedures for some visits may take more than 1 day, especially for patients who participate in study addenda related to imaging. See the Study Schedule (Protocol Attachment LFAN.1) for the timing of events and the measures to be assessed.

3.1.3. Delayed-Start Phase (Study Period III) Study Period III will begin sometime after 64 weeks of treatment (Visit 14). In this phase, patients initially randomized to placebo will be given LY450139. These patients will be dose-escalated to 140 mg once daily in a manner identical to that for patients initially randomized to 140 mg LY450139 (Section 5.1.2). This dose will be continued until the end of the study, at Week 88 (Visit 20). Because dose reductions are not allowed at or after Week 64 (Visit 14) of the study to maintain blinding, AEs that may occur in patients initially treated with placebo and entering the delayed-start portion of the study could necessitate patients’ discontinuation from the study. Patients and investigators will be blinded to the timing of the delayed start of active treatment for the placebo-treated patients. An ERB supplement will explain the actual timing.

3.1.4. Follow-Up Visit (Visit 801) A follow-up visit has been included in the study to provide an opportunity for evaluation of patient safety and to allow for concomitant medication adjustment, given the potential for LY450139 to affect concomitant medication exposure (discussed in more detail in Section 5.7.3). Patients who complete this study through Visit 20 may be eligible to participate in an open-label extension. Patients who opt to participate in the open-label extension will not have a follow-up visit because their safety will be further evaluated in the extension. The follow-up visit (Visit 801) will be required, however, for all other patients who are randomized to drug at Visit 2, take at least 1 dose of study medication, and either LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 20 discontinue before Visit 20 or opt not to participate in an open-label extension. Visit 801 should be scheduled to occur approximately 4 weeks after the patient’s last dose of study drug. See the Study Schedule (Protocol Attachment LFAN.1) for the timing of events and the measures to be assessed.

3.2. Discussion of Design and Control Continuing Standard of Care. Patients may take approved AD symptomatic medications (standard-of-care) such as concomitant acetylcholinesterase inhibitors (AChEIs) or memantine if such medications have been given for at least 4 months and stable dosing has been maintained for at least 2 months before Visit 2. Changes in doses of available symptomatic medications for AD will require consultation with the sponsor and should occur only when necessary for the adequate overall care of the patient. The effect of LY450139 treatment plus standard-of-care will be compared to placebo plus standard-ofcare. Dosing. As described more fully in Section 5.5, study drug will be dosed once a day, preferably in the morning. Although the elimination half-life of LY450139 is approximately 2.5 hours, plasma Aβ1-40 reduction from baseline after a single dose of 140 mg lasts about 12 hours. In addition, nonclinical toxicology results suggest that dosing with LY450139 more than once a day may not have a favorable benefit-risk profile. A number of membrane proteins, including Notch, have been identified as possible substrates for γ-secretase. Human Notch1 appears to be involved in multiple blood cell lineages at different stages of maturation, having been found in lymphoid, myeloid, and erythroid precursor populations, as well as in peripheral blood T and B lymphocytes, monocytes, and neutrophils (Milner and Bigas 1999). The clinical relevance of drug-induced Notch inhibition on these tissues, however, remains unknown. Despite a theoretical risk to myeloid differentiation, consistent changes in this process have not been observed in toxicology studies nor in clinical trials. Notch is also involved in cell differentiation in the gastrointestinal mucosa; goblet cell hyperplasia has been described in both the dog and the rat following repeat-dosing with LY450139, possibly because of effects on Notch processing. Clinical studies to date have shown some gastrointestinal effects, including both constipation and diarrhea (Siemers et al. 2005, 2006, 2007; Fleisher et al. 2007), but, as detailed in the IB, these effects appeared generally mild and manageable. Intermittent, rather than continuous, inhibition of Notch may be preferable; thus, once-daily dosing is appropriate for this study.

LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 21

4. Study Population Eligible patients will be men and postmenopausal women with mild or moderate probable AD, as specified in the entry criteria below.

4.1. Inclusion Criteria A patient included in the study must meet all of the following inclusion criteria. [1] Meets National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS/ADRDA) criteria for probable AD [2] Has a Modified Hachinski Ischemia Scale score of ≤4 [3] Has an MMSE score of 16 through 26 [4] Has a Geriatric Depression Scale (GDS) score of ≤6 (on the staffadministered short form) [5] Has a magnetic resonance imaging (MRI) or computerized tomography (CT) scan within the past year on file with the investigator, with no findings inconsistent with a diagnosis of AD [6] Is at least 55 years old. If female, is post-menopausal, as evidenced by a lack of menstruation for at least 12 consecutive months or by having had a bilateral oophorectomy.

4.2. Exclusion Criteria 4.2.1. Exclusion Criteria for All Patients A patient will be excluded from the study if he or she meets any of the following exclusion criteria. [7] Meets National Institute of Neurological Disorders and StrokeAssociation Internationale pour la Recherche et l'Enseignement en Neurosciences (NINDS/AIREN) criteria for vascular dementia [8] Does not have a reliable caregiver who is in frequent contact with the patient (defined as at least 10 hours per week), will accompany the patient to the office and/or be available by telephone at designated times, and will monitor administration of prescribed medications. Note: The caregiver must be able to communicate with site personnel and willing to comply with protocol requirements. Participants living in an assisted-living facility may be included if study medication intake is supervised and if regular contact with a caregiver who accompanies the patient is maintained. [9] Is not capable of swallowing whole oral medications LY450139 H6L-MC-LFAN Confidential

Protocol Approved by Lilly: 23 August 2007

Page 22 [10] Has serious or unstable illnesses including hepatic, renal, gastroenterologic, respiratory, cardiovascular (including ischemic heart disease), endocrinologic, neurologic (other than AD), psychiatric, immunologic, or hematologic disease or other conditions that, in the investigator’s opinion, could interfere with the analyses of safety and efficacy in this study; or has a life expectancy of 474 ms (female) at Visit 1 or predose Visit 2 (baseline value). A patient with prolonged QRS interval (>116 ms if male, >102 ms if female) will be excluded with mean JTc interval >359 ms (male) or >382 ms (female) at Visit 1 or predose Visit 2 (baseline value). (See Section 6.3.2.2 for details on QRS calculation and determination of using QTc or JTc for exclusion.) [16] Has evidence of significant active cardiac disease, uncontrolled hypertension, uncompensated congestive heart failure, or endocarditis [17] Has potassium 474 ms if female (calculate QTc using Bazett’s correction method; see Section 6.3.2.2 for details on use of QTc for exclusion). [16] Has evidence of significant active cardiac disease, uncontrolled hypertension, uncompensated congestive heart failure, or endocarditis [17] Has potassium 474 ms if female at Visit 1 or predose Visit 2 (baseline value). H6L-MC-LFAN (b) Confidential

Protocol Amendment Summary

Page 4 A patient with prolonged QRS interval (>116 ms if male, >102 ms if female) will be excluded with mean JTc interval >359 ms (male) or >382 ms (female) at Visit 1 or predose Visit 2 (baseline value). (See Section 6.3.2.2 for details on use of QRS calculation and determination of using QTc or JTc for exclusion.) [24] Has received AChEIs or memantine for less than 4 months or has less than 2 months of stable therapy on these treatments by Visit 2 (Note: If a patient has recently stopped AChEIs or memantine, he or she must have discontinued treatment at least 2 months before Visit 2.) Refer also to Section 5.7.1 for patients on donepezil. [32] In the investigator’s opinion, lacks adequate premorbid literacy to complete the required psychometric tests.

4.3.1. Discontinuation of Patients In addition, patients will be discontinued from the study drug and from the study in the circumstances noted below. •

For patients with prolonged QRS interval, mean of triplicate ECG resulting in treatment-emergent prolonged JTc of >400 ms with absolute change of >75 ms over baseline (as described in Section 6.3.2.2). For patients with normal QRS interval, m Mean of triplicate ECG resulting in treatment-emergent prolonged QTc >500 ms with absolute change of >75 60 ms over baseline (as described in Section 6.3.2.2).

5.7.1. Allowed Medications Use of approved treatments for AD [including donepezil (Aricept ®), rivastigmine (Exelon®), galantamine (Razadyne™, Razadyne™ ER), tacrine (Cognex®), and memantine (Namenda®)] is permitted during the study, provided that such medications have been given for at least 4 months and the dose has been unchanged for 2 months prior to Visit 2. Any patient taking donepezil should be treated with a dose no less than 10 mg/day, unless this dose is not approved by local regulatory agencies, in which case patients should be treated with no less than the highest dose approved by local regulatory agencies. Doses of these medications should remain constant throughout the study. Starting, stopping, or changing doses of AChEIs or memantine during the study could interfere with outcome measures and may therefore result in discontinuation of the patient from the study. If a patient has recently stopped AChEIs or memantine, he or she must have discontinued treatment at least 2 months before Visit 2. Other vitamins or nutraceuticals given for their possible effects on AD may be continued on stable doses beginning 2 months prior to Visit 2.

H6L-MC-LFAN (b) Confidential

Protocol Amendment Summary

Page 5 Patients who are not receiving any treatment for AD are permitted to enter the study, and must not begin any such treatments for the duration of the study. Before a patient starts, stops, or changes doses of AChEIs or memantine, the sponsor or designee must be contacted to determine whether the patient should continue in the study and whether clinical outcome measures should be performed. 6.3.2.2. Electrocardiograms Twelve-lead ECGs will be obtained in triplicate according to the Study Schedule (Protocol Attachment LFAN.1). QRS interval, QT interval, and heart rate values from these ECGs will be used to calculate either an average QTc interval (for patients with a normal QRS interval) or an average JTc interval (for patients with a prolonged QRS interval—because of bundle branch block or other intraventricular conduction delay or implanted pacemakers, for example). In addition, change from the predose Visit 2 baseline value will be calculated. Mean JTc or QTc interval will be used to determine patient eligibility for the protocol and, These will also be used with change from baseline, to determine if the patient meets discontinuation criteria (Section 4.3) and to guide patient management. The ECGs and mean QTc/JTc will be interpreted by a qualified physician (the investigator or qualified designee) at the site as soon after the time of ECG collection as possible, and ideally while the patient is still present, for immediate patient management and to determine whether the patient meets entry or discontinuation criteria. If a clinically significant increase in the mean QTc or JTc interval from baseline is present, the investigator should assess the patient for symptoms (such as palpitations, near syncope, syncope). The determinations by the site PI or designee will be the final ECG data used for entry criteria or for discontinuation criteria purposes. Arbitration may be necessary in a small group of patients with intermittent pacing or conduction delays for whom automated methods of interval calculations do not perform reliably. The ECGs will subsequently be transmitted electronically via modem to the centralized ECG vendor designated by Lilly. The centralized ECG vendor’s cardiologist will then complete the ECG overread. The central ECG vendor’s overread will be used for data analysis and report-writing purposes; however, mean QTc and JTc will be calculated separately. Once the overread ECG is returned from the centralized ECG vendor, the investigator or qualified designee is responsible for determining if any change to patient management is needed and must document his or her review. If there are differences in ECG interpretation between the investigator or qualified designee and the ECG vendor cardiologist, the investigator or qualified designee's interpretation will prevail for study entry and immediate patient management purposes, and the ECG vendor cardiologist's will prevail for data analysis purposes.

H6L-MC-LFAN (b) Confidential

Protocol Amendment Summary

Page 1

Confidential Information The information contained in this protocol amendment summary is confidential and is intended for the use of clinical investigators. It is the property of Eli Lilly and Company or its subsidiaries and should not be copied by or distributed to persons not involved in the clinical investigation of LY450139, unless such persons are bound by a confidentiality agreement with Eli Lilly and Company or its subsidiaries.

Protocol Amendment Summary H6L-MC-LFAN(c) Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo LY450139

Eli Lilly and Company Amendment (c) Approved by Lilly: 23 January 2009 Note to Investigators: The new protocol is indicated by Amendment (c) and will be used to conduct the study in place of any preceding version of the protocol.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 2

Overview The overall changes and rationale for the changes made to this protocol are as follows: •

Table of Contents was updated to reflect changes to Section 8 and addition of Attachment LFAN.8.



Abbreviations and Definitions were updated.



Based on meetings with FDA and CHMP, the analysis methods for the primary objective were revised, and several other analyses were added or changed, leading to extensive changes throughout objectives and statistical analysis methods sections (Sections 2 and 8). An overview of the relevant changes is as follows: •

Revision of the analytic method to test hypothesis for co-primary outcomes. The previous methods of slope analysis and analysis using stratification method have been replaced with a single analysis using mixed model repeated measures (MMRM).



Addition of sensitivity analyses to validate the assumptions and confirm the results of the primary analysis.



Inclusion of slopes analysis as a secondary analysis.



Inclusion of test of linearity for the slopes analysis.



Revision of other secondary analysis to be done using MMRM so that same analytic method is used consistently for all outcomes.



Revision of analysis for the delayed start period so that the main analysis of the delayed start period consists of analyzing change from baseline (beginning of study) to end of study.



Inclusion of a test for parallelism analysis for the delayed start period.



Revision of the gatekeeping strategy.



Inclusion of an interim analysis for futility (section 8.2.8).



Because of the need for individually numbered country-specific addendum to address regional regulations, the use of numbers to name the study addenda in Section 2.2 has been eliminated.



References to Pittsburgh Compound B (PIB), which is no longer being used in optional Study Addendum 3, were replaced with references to an amyloid imaging tracer.



Exclusion criterion related to alcohol/subtance abuse and dependence was clarified.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 3 •

Exclusion criterion and description of ECG procedures related to QTc were clarified to specify use of Bazett’s correction method.



Description of process for reading and routing of ECGs was clarified.



Exclusion criterion related to previous participation in any LY450139 study was clarified.



Allowable donepezil dosing was clarified.



Table LFAN.2, Concomitant Medications with Recommended Monitoring, was added to increase readability.



Administration of efficacy measures was clarified.



Description of ADAS-cog scoring was revised.



A comment suggesting monitoring of decreases in uric acid was removed because investigators are blinded to uric acid results.



Additional Expected Events section was expanded to include discretionary opthalmalogic examinations and exclusion of potential for teratogenic effects.



Procedures for evaluation of plasma samples were clarified.



References were updated.



Footnote “l” under Study Schedule (Study Attachment LFAN.1) was amended to clarify biomarker storage.



Chloride was added to the clinical chemistry panel. Typically part of a standard clinical chemistry panel, chloride was inadvertently omitted from the original protocol.



As the methylene blue stain is not the only acceptable method to determine fecal WBC/RBC, this specific method is no longer required. Instead, “Clinical Laboratory Tests: Other Tests” simply states that a fecal WBC/RBC should be performed for patients reporting diarrhea.



Protocol Attachment LFAN.4 was added to clarify guidance on discontiuations.



General typographical errors were corrected.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 4

Revised Protocol Sections Note:

All deletions have been identified by strikethroughs. All additions have been identified by the use of underscore. Table of Contents

8.2.4.2.

1.1. MMRM Analysis of Primary Outcome and Methodology for CHMP

8.2.4.3

1.2. Sensitivity Analysis

8.2.4.2.

Efficacy Analysis of Secondary Endpoints—Initial Treatment Period

8.2.4.43.

Efficacy Analysis of Secondary Endpoints—Delayed-Start Period

8.2.4.53.1.

Main Analysis of Delayed-Start Period

8.2.4.3.2.

Test for parallelism during Delayed Start Period

8.2.4.3.3.

Additional Analyses

8.2.4.4.

Subgroup Analyses

8.2.4.65.

Analysis of Plasma Aβ

8.2.4.76.

Analyses of Biomarker Outcomes

8.2.4.87.

Statistical Details of the Gatekeeping Strategy

8.2.4.8

Specification of A Priori Sequence to Test Hypothesis for Biomarker Data

8.2.5.

Safety Analyses—Initial Treatment Period

8.2.5.1.

Adverse Events

8.2.5.2.

Laboratory Analyses

8.2.5.3.

Electrocardiograms

8.2.5.4.

Vital Signs and Weight

List of Protocol Attachments Protocol Attachment LFAN.4. Guidance on Discontinuation

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 5

Abbreviations and Definitions

LOCF

Last Observation Carried Forward

PIB

Pittsburgh Compound-B

2.1. Primary Objective The primary objective of this study is to test the hypothesis that LY450139 given orally will slow decline ofassociated with AD as compared to placebo. Because of differences in regulatory requirements across regions, this objective will be assessed using several measures. In one region, one measure will be considered primary and the other secondary, and this relationship will be reversed for the second region. This is discussed more fully in Section 8. Both measures will be evaluated sometime between 64 and 88 weeks after initiation of treatment; the precise timing is specified in an ethical review boardEthical Review Board (ERB) supplement to this protocol. For the FDA, the The primary objective will be assessed using a slopemixed-model repeated measures (MMRM) analysis of 2 coprimary outcomes, the Alzheimer’s Disease Assessment Scale—Cognitive subscore (ADAS-Cog11) and the Alzheimer’s Disease Cooperative Study—Activities of Daily Living Inventory (ADCS-ADL), in which the specific hypothesis is that the rate of change over time (slope)at the end of the initial treatment phase for LY450139 will be significantly less than that for placebo., as assessed using a repeated-measures model. For the Committee for Medicinal Products for Human Use (CHMP), the primary objective will be assessed by change from baseline to endpoint using 2 coprimary outcomes, the ADAS-Cog11 and ADCS-ADL. The specific hypothesis is that the change from baseline to endpoint will be significantly less than that for placebo, as assessed by a stratified analysis of baseline-to-endpoint change.

2.2. Secondary Objectives and Measures Secondary objectives of this study are listed below: •

To test the hypothesis that LY450139 is a disease-modifying medicationwill slow the rate of decline associated with AD, compared to placebo, for ADAS-Cog11, assessed using a slope analysis from a repeated-measures model.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 6 •

To test the hypothesis that LY450139 will slow the rate of decline associated with AD, compared to placebo, for ADCS-ADL, assessed using a slope analysis from a repeated-measures model.



To test the hypothesis that LY450139 has an effect independent of acute symptomatic effects, as assessed using a “delayed-start” study design (Leber 1997; Whitehouse et al. 1998; Mohs et al. 2006) in which placebo patients will begin receiving double-blind LY450139 sometime after 64 weeks in the study. The specific hypothesis is that statistically significant differences will be observed in changes from study baseline to endpoint in ADAS-Cog11 and ADCS-ADL scores between patients originally assigned to active treatment and patients originally assigned to placebo who initiate active treatment at least 64 weeks later. This result would be consistent with a slowing of the underlying rate of disease progression due to treatment with LY450139; if LY450139 were a symptomatic treatment, one would expect an improvement in cognitive measures during the delayed start period such that the group originally on placebo would approximate scores from the group originally assigned to LY450139 by endpoint (88 weeks after baseline).



To test the noninferiority hypothesis that LY450139 is a diseasemodifying medication independent of acute symptomatic effects, as assessed using a “delayed-start” study design. The specific hypothesis is that patients switched to LY450139 do not improve significantly more than patients treated with LY450139 from beginning of study as assessed using the ADAS-Cog11. This will be assessed by comparing the slope from the beginning of the delayed-start period to the end of the delayedstart period between patients originally assigned to active treatment and patients originally assigned to placebo who are switched to active treatment at least 64 weeks later.



To provide supporting evidence that LY450139 is a disease-modifying compound, multiple biomarkers will be assessed. Some biomarker data will be collected only in subsets of patients and will therefore be fully described in study addenda; all are described here to contribute to greater understanding of the study objectives. Briefly, the following secondary hypotheses will be tested, using the indicated measures, in each case, the specific hypothesis being that LY450139 will be different from placebo. •

LY450139 will acutely reduce Aβ in plasma within 6 hours of administration at Week 6.



LY450139 will attenuate the accelerated rate of decline in brain glucose metabolism known to occur in patients with AD (De Santi et al. 2001; Herholz 2003; Thal et al. 2006). This will be assessed in a subset of patients as part of an optional Study Addendum 1study addendum , using fluorine-18 fluorodeoxyglucose ([18F]-FDG) positron emission tomography (FDG-PET).

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 7 •

LY450139 will alter the accelerated rate of decline in brain volumes known to occur in patients with AD (Schott et al. 2005). This will be assessed in a subset of patients as part of an optional Study Addendum 2study addendum, using volumetric magnetic resonance imaging (vMRI).



LY450139 will reduce brain amyloid burden as compared to placebo. This will be assessed in a subset of patients as part of an optional Study Addendum 3study addendum, using an amyloid-imaging PET tracer called Pittsburgh Compound-B ([11C]-PIB) (PIB-PET).



LY450139 will reduce the elevated concentrations of CSF tau proteins known to exist in patients with AD (Trojanowski 1996; Green et al. 1999). CSF tau will be collected via lumbar punctures, which are addressed in the optional Study Addendum 4lumbar puncture study addendum.



To compare the safety of LY450139 and placebo, through assessment of treatment-emergent adverse events (TEAEs) and changes in vital signs, laboratory evaluations, and electrocardiograms (ECGs).



To characterize population pharmacokinetics (PK) of LY450139, explore potential factors that may influence variability of PK, and explore the association of PK variables with efficacy, biomarkers, and safety parameters.



To test the hypothesis that LY450139 will slow the rate of cognitive decline of associated with AD, compared to placebo, using the as assessed by 2 extended versions of the Alzheimer’s Disease Assessment Scale (a 12-item version, the ADAS-Cog12; and a 14-item version, the ADASCog14) and the Mini-Mental State Examination (MMSE), assessed using a slopeMMRM analysis. from a repeated-measures model.



To assess global clinical benefit of treatment with LY450139 as demonstrated through the Clinical Dementia Rating—Sum of Boxes (CDR-SB) and additional clinical outcomes instruments, including the Neuropsychiatric Inventory (NPI), Resource Utilization in Dementia— Lite, and (RUD-Lite), EuroQol-5D 5-Dimensional Health-related Quality of Life Scale Proxy version (EQ-5D Proxy).



To assess differential rate of functional decline with LY450139 compared with placebo using a subset of items from the ADCS-ADL for instrumental activities of daily living (ADLs) (items 7 through 23).

In addition, the following exploratory hypotheses related to biomarkers will be tested. •

An exploratory hypothesis to be tested using PIB-an amyloid-imaging PET tracer (optional Study Addendum 3imaging study addendum) is that only patients demonstrating sufficient amyloid burden at baseline will respond to treatment as determined by clinical measures.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 8 •

An exploratory hypothesis to be tested using CSF (optional Study Addendum 4lumbar puncture study addendum) is that LY450139 will normalize (increase) CSF Aβ1-42 concentrations, which are known to be reduced in patients with AD (Motter et al. 1995; Galasko et al. 1998; Thal et al. 2006) and thus might be normalized during chronic treatment with a disease-modifying drug. Because treatment with LY450139 would be expected to acutely decrease CSF Aβ concentrations, CSF will be obtained for Aβ measurements approximately 24 hours after dosing (immediately before the next dose). Because the effect of daily dosing may be longer than 24 hours after chronic LY450139 administration, CSF Aβ1−42 concentrations might be decreased rather than increased at this endpoint measurement; thus, this will be considered an exploratory analysis.

3.1. Summary of Study Design Study H6L-MC-LFAN (LFAN) will be a multicenter, randomized, double-blind, placebo-controlled, Phase 3 study comparing LY450139 140 mg, LY450139 100 mg, and placebo in approximately 1500 patients with mild to moderate AD. Patients who meet entry criteria will be randomized in a 1:1:1 ratio (500 per treatment arm) to 1 of the 3 treatment groups: LY450139 100 mg once daily, LY450139 140 mg once daily, or placebo once daily. Patients will be randomized by site and by mild or moderate MiniMental State Examination (MMSE) scores (for the purposes of this study, moderate will be defined as scores including 16 through 19, and mild as including 20 through 26). The primary hypothesis being tested is that LY450139 will slow the rate of cognitive and functional decline in associated with AD as compared to with placebo. In addition, sometime after 64 weeks of treatment, patients receiving placebo will begin receiving LY450139 (escalated to 140 mg/day) for the remainder of the study. This “delayed-start” design feature will evaluate whether patients originally treated with placebo improve during this period such that their cognitive scores subsequently approximate those of the patients initially treated with LY450139. The absence of such an improvement—in other words, maintaining a statistical difference between patients initially assigned to LY450139 and placebo following the delayed start—would support the conclusion that LY450139 slows disease progression and does not affect simplyonly symptoms of disease. Patients and investigators will be blinded to the timing of the delayed start of active treatment for the placebo-treated patients. An ERB supplement will explain the actual timing.

4.2.1. Exclusion Criteria for All Patients [14] Has a history of chronic alcohol or drug abuse or dependence (using DSM-IV TR criteria) within the past 5 years

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 9 [15] Requires the use of concomitant medications that prolong the QT/QTc interval or has a known history of Long QT Syndrome or Brugada Syndrome. Has ECG abnormalities obtained at Visit 1 or at predose Visit 2 (baseline value) that, in the opinion of the investigator, are clinically significant with regard to the patient’s participation in the study; or has a QTc abnormality at Visit 1 or predose Visit 2 (baseline value) as indicated by a mean QTc interval >458 ms if male or >474 ms if female. (See (calculate QTc using Bazett’s correction method; see Section 6.3.2.2 for details on use of QTc for exclusion.) [26] Has previously completed or withdrawn from this study, or a past study of LY450139 if withdrawn due to adverse event.

4.3.1. Discontinuation of Patients •

Severe nausea, vomiting, diarrhea, or abdominal pain (see Protocol Attachment LFAN.4 for guidance on definition of severe)

Patients who discontinue the study early will have early discontinuation procedures performed as shown in the Study Schedule (Protocol Attachment LFAN.1). Patients who discontinue early should be scheduled for a follow-up visit (Visit 801) 4 weeks after last dose of study drug, and a virtual welfare/disposition follow-up 6 months after last dose of study drug.

5.7.1. Allowed Medications Use of approved treatments for AD [including donepezil (Aricept ®), rivastigmine (Exelon®), galantamine (Razadyne™, Razadyne™ ER), tacrine (Cognex®), and memantine (Namenda®)] is permitted during the study, provided that such medications have been given for at least 4 months and the dose has been unchanged for 2 months prior to Visit 2. Any patient taking donepezil should be treated with a dose no less than 10 mg/day, unless this dose is not approved by local regulatory agencies, in which case patients should be treated with no less than the highest dose approved by local regulatory agencies. If the patient has historically not tolerated 10 mg/day of donepezil, the highest tolerated dose approved by local regulatory agencies should be used. Doses of these medications should remain constant throughout the study. Starting, stopping, or changing doses of AChEIs or memantine during the study could interfere with outcome measures and may therefore result in discontinuation of the patient from the study.

5.7.3. Monitoring Medications Because the CYP3A4 induction observed with LY450139 has been shown to be similar in nature to rifampicin, a known potent inducer of CYP3A4 and other isozymes, a review of rifampicin drug interactions (Niemi et al. 2003) has been used as a guide in suggesting examples of medications that are particularly likely to decrease in exposure when administered with LY450139. Based on that review article, it is recommended that drug LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 10 levels be monitored for patients receiving quinidine, nortriptyline, theophylline, phenytoin, valproic acid, divalproex, sodium valproate, valproate semisodium, carbamazepine, digoxin, or digitoxin; PT/INR be assessed for patients receiving warfarin or phenprocoumon; free T4/TSH be assessed for patients receiving thyroxine; and blood glucose (preferably fasting) be assessed for patients receiving glibenclamide, glimepiride, glipizide, or repaglinideany of the medications listed below (Table LFAN.2). Because specific interactions other than CYP3A have not been studied, however, this list is not exhaustive; thus, recommendations for periodic assessments of plasma concentrations or clinical monitoring of biomarkers may be extended to other medications at the investigator's discretion. Table LFAN.2.

Concomitant Medications with Recommended Monitoring

Drugs with Associated Laboratory Monitoring Laboratory Monitoring Phenprocoumon PT/INR Local Lab Warfarin PT/INR Local Lab Thyroxine free T4/TSH a Central Lab Glibenclamide blood glucose (preferably fasting) b Central Lab Glimepiride blood glucose (preferably fasting) b Central Lab b Glipizide blood glucose (preferably fasting) Central Lab Repaglinide blood glucose (preferably fasting) b Central Lab Drugs with Therapeutic Drug Level Monitoring Monitoring Carbamazepine carbamazepine level Local Lab Digitoxin digitoxin level Local Lab Digoxin digoxin level Local Lab Nortriptyline nortriptyline level Local Lab Phenytoin phenytoin level Local Lab Quinidine quinidine level Local Lab Theophylline theophylline level Local Lab Local Lab Valproic acid (including divalproex sodium, valproic acid level sodium valproate, and valproate semisodium) Abbreviations: free T4 = free thyroxine; PT/INR = prothrombin time/international normalized ratio ; TSH = thyroid-stimulating hormone. a Free T4 and TSH are to be collected through the central vendor using laboratory kits supplied. b Blood glucose is already collected as part of the central lab chemistry panel. Please consider fasting for diabetics.

6.1.1. Primary Efficacy Measures Cognitive testing for each patient should be performed at approximately the same time on each day that testing occurs and each individual scale should be administered by the same rater from visit to visit to reduce potential variability. The ADAS-Cog and the ADCSADL must be performed by different raters in order to assure independent assessment of

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 11 cognitive (ADAS-Cog) and functional (ADCS-ADL) capabilities. The following is a brief description of the cognitive testing that will be used in this study. Alzheimer’s Disease Assessment Scale—Cognitive subscale (ADAS-Cog; Rosen et al. 1984). The ADAS was designed as a rating scale for the severity of the dysfunction in the cognitive and noncognitive behaviors characteristic of persons with AD. The cognitive subscale of the ADAS, the ADAS-Cog11, will be used as a primary efficacy measure and consists of 11 items assessing areas of function most typically impaired in AD: orientation, verbal memory, language, and praxis. The scale ranges from 0 to 70, with higher scores indicating greater disease severity. The ADAS-Cog11 score will be derived from an extended ADAS-Cog, the ADAS-Cog14, which will be the actual scale administered to patients. The ADAS-Cog14 is augmented with delayed free recall, digit cancellation, and maze completion measures (Mohs et al. 1997). A score of 0 to 10 for delayed free recall and a conversion code of 0 to 5 for delayed free recall, digit cancellation and maze completion provides total score ranges for this extended ADASCog14 of 0 to 85.90. In addition, a 12-item version of the ADAS-Cog (ADAS-Cog12) will be assessed. This version, also derived from the ADAS-Cog14, is the ADAS-Cog11 augmented with the delayed free recall measure, resulting in a total score ranging from 0 to 80.

6.3.2.2.

Electrocardiograms

Twelve-lead ECGs will be obtained in triplicate according to the Study Schedule (Protocol Attachment LFAN.1). Corrected QT (QTc) will be calculated using Bazett’s correction method (QTcB). Mean QTcB interval will be used to determine patient eligibility for the protocol and, with change from baseline, to determine if the patient meets discontinuation criteria (Section 4.3) and to guide patient management. The ECGs and mean QTcB will be interpreted by a qualified physician (the investigator or qualified designee) at the site as soon after the time of ECG collection as possible, and ideally while the patient is still present, for immediate patient management and to determine whether the patient meets entry or discontinuation criteria. If a clinically significant increase in the mean QTcB interval from baseline is present, the investigator should assess the patient for symptoms (such as palpitations, near syncope, syncope). The determinations by the site PI or designee will be the final ECG data used for entry criteria or for discontinuation criteria purposes. Arbitration may be necessary in a small group of patients with intermittent pacing or conduction delays for whom automated methods of interval calculations do not perform reliably. The ECGs will subsequently be transmitted electronically via modem to the centralized ECG vendor designated by Lilly. The centralized ECG vendor’s cardiologist will then complete the ECG overread. Two of the 3 ECGs obtained will be overread for interval measurements of QT, HR, RR, and QRS duration only; these will not be returned to sites. The third ECG will have a complete safety overread with both interval measurements and a cardiologist’s interpretation; this ECG will be returned to the sites. The central ECG LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 12 vendor’s overread will be used for data analysis and report-writing purposes; however, mean QTcB will be calculated separately. Once the overread ECG is returned from the centralized ECG vendor, the investigator or qualified designee is responsible for determining if any change to patient management is needed and must document his or her review. If there are differences in ECG interpretation between the investigator or qualified designee and the ECG vendor cardiologist, the investigator or qualified designee's interpretation will prevail for study entry and immediate patient management purposes, and the ECG vendor cardiologist's will prevail for data analysis purposes.

6.3.2.4.

Additional Considerations

Laboratory Changes. Various expected laboratory changes (decrease in uric acid and potassium, possible changes in lymphocyte subsets, for example) seen with administration of LY450139 should also continue to be closely monitored. 6.3.2.4. Additional Considerations Additional Expected Events. The potential for AEs that have been observed with LY450139 treatment (including rash, somnolence, hair or skin discoloration, fatigue, and asthenia) will be monitored through the standard collection of AEs. The potential for QTc prolongation will be assessed through collection of ECGs (as discussed in Section 6.3.2.2). The potential for teratogenicity upon fetal exposure to LY450139 is addressed by excluding female patients who are premenopausal or have not had a bilateral oophorectomy. The potential for visual changes will be assessed with fundoscopic exams during physical examination. Patients who develop treatment-emergent visual changes may be referred for ophthalmologic examination at the discretion of the investigator.

6.4.2.3.

Technique Used for Evaluation of Plasma

Plasma concentrations of LY450139, donepezil, and Aβ for PD analysis will be determined from processed blood samples at sponsor-designated laboratories. Plasma concentrations of LY450139 and donepezil will be determined using validated and selective liquid chromatography and tandem mass spectrometry methods. Additional analyses may be conducted for further evaluation of metabolites, but the blood volume will not increase. Plasma concentrations of Aβ will be determined using for LY450139 and donepezil, and validated immunoassay methods for Aβ. The concentration data will be used to estimate the PK and parameters of the study population. LY450139 concentrations will be assayed on those specimens obtained from patients randomized to LY450139. Bioanalytical samples collected to measure study drug concentration and metabolism and/or protein binding will be retained for a maximum of 2 years following last patient visit for the study. Donepezil concentrations will be assayed only for patients who are receiving donepezil. Results will not be provided to investigators.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 13

8. Sample Size and Statistical Methods 8.1. Determination of Sample Size Power and sample size calculations were based on the analysis of the primary objective using the change from baseline to endpoint for the CHMP. This is the most conservative method and thus will also cover the analysis of the primary objective for the FDA. Power and sample size calculations were based on the expected effect size at the end of the initial treatment period. Also, because determination of sample size was calculated based on the duration of the initial treatment phase, all sample size calculations are provided in the ERB supplement to maintain blinding to phase durations.

8.2. Statistical and Analytical Plans 8.2.1. General Considerations All analyses will be performed on an ITT basis unless otherwise specified. An ITT analysis is an analysis of data by the groups to which subjects are assigned by random allocation, even if the subject does not take the assigned treatment, does not receive the correct treatment, or otherwise does not follow the protocol. When change from baseline is assessed, patients will be included in the analysis only if both a baseline and a postbaseline measure are available. For the analysis of the initial treatment period, unless otherwise defined, a baseline measure is the observation prior to the first dose (Visit 2 observation). If the Visit 2 observation is missing, the Visit 1 observation will be considered the baseline measure. The endpoint measure is defined as the patient’s last measure in the appropriate study period. In the event that any investigator has an inadequate number of patients (defined as one or zero randomized patient per treatment group) for the planned analyses, the following strategy will be implemented. Data from all such investigators will be pooled. The pooling will be done first within a country. If the resulting pool is still inadequate, pooling will be done across countries. The resulting pooled data will be included in all analyses as such and treated as though the data were obtained from a single site. If any of the individual items for ADAS-Cog11 are missing or unknown, every effort will be made to obtain the score for the missing item or items. If less than 30% (3 or fewer of 11) of the items are missing, the total score will be imputed as follows: The sum of the nonmissing items will be prorated to the sum from 11 items. (For example, if 8 of 11 items are nonmissing, with a total of 35, and the maximum possible score for those 8 items = 47, the imputed total score for 11 items = (70*35)/47 = 52.) The imputed number will be rounded up to the nearest integer. If the nearest integer is greater than the maximum possible score (70), the imputed score will be equal to the maximum score of LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 14 70. If more than 3 items are missing, the total score for ADAS-Cog11 at that visit will be considered missing. The same imputation technique will be applied to the ADAS-Cog14. If less than 30% (4 or fewer of 14) of the items are missing, the total score will be imputed as the prorated total calculated from the nonmissing items. The imputed total will be rounded up to the nearest integer. If the nearest integer is greater than the maximum possible score (85), the imputed score will be equal to the maximum score of 85. If more than 4 items are missing, the total score for ADAS-Cog14 at that visit will be considered missing. Note: It is possible to have an imputed total score for ADAS-Cog11 but to have a missing total score for ADAS-Cog14. Conversely, if 4 of the items for ADAS-Cog11 are missing and the additional items for ADAS-Cog14 are available, the total score for ADAS-Cog11 will be missing, but an imputed total score for ADAS-Cog14 will be available. The same imputation technique will be applied to the CDR-SB. If only one box (of 6) of the CDR is missing, the sum of the boxes will be imputed by prorating the sum from the other 5 boxes. If the score from more than one box is not available, the CDR-SB at that visit will be considered missing. For all other scales, if any item is missing, any total or sum involving that item will be considered missing. All efficacy and safety hypothesis tests will use a two-sided 0.05 significance level unless otherwise stated. Any change to the data analysis methods described in the protocol will require an amendment ONLY if it changes a principal feature of the protocol. Any other change to the data analysis methods described in the protocol and the justification for making the change will be described in the clinical study report (CSR). Additional exploratory analyses of the data will be conducted as deemed appropriate. All analyses will be performed on an intent-to-treat (ITT) basis unless otherwise specified. An ITT analysis is an analysis of data by the groups to which subjects are assigned by random allocation, even if the subject does not take the assigned treatment, does not receive the correct treatment, or otherwise does not follow the protocol. A gatekeeping strategy using a fixed-sequence approach will be used to test secondary endpoints after the test of the co-primary endpoints (Section 8.2.4.7). When change from baseline is assessed, patients will be included in the analysis only if both a baseline and a postbaseline measure are available. For the analysis of the initial treatment period, unless otherwise defined, a baseline measure is the observation prior to the first dose (Visit 2 observation). If the Visit 2 observation is missing, the Visit 1 observation will be considered the baseline measure. The endpoint measure is defined as the patient’s last measure in the appropriate study period.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 15 For analysis using MMRM, an unstructured covariance matrix will be used to model the within-patient variance-covariance errors. If the unstructured covariance structure matrix results in a lack of convergence, the heterogeneous Toeplitz covariance structure, followed by the heterogeneous autoregressive covariance structure, will be used. The Kenward-Roger approximation will be used to estimate the denominator degrees of freedom. In the event that any investigator has an inadequate number of patients (defined as 1 or 0 randomized patient per treatment group) for the planned analyses, the following strategy will be implemented: Data from all such investigators will be pooled. The pooling will be done first within a country. If the resulting pool is still inadequate, pooling will be done across countries. The resulting pooled data will be included in all analyses as such and treated as though the data were obtained from a single site. If any of the individual items for ADAS-Cog11 are missing or unknown, every effort will be made to obtain the score for the missing item or items. If less than 30% of the items are missing (3 or fewer of 11), the total score will be imputed as follows: The sum of the scores from the non-missing items will first be calculated. This score will then be multiplied by a factor that includes the maximum score for the missing items. For example, if the item, "Word-recall Task", which ranges from a score of 0-10 (maximum = 10) is missing, and another item "Commands", which ranges from a score of 0-5 (maximum = 5) is also missing then the multiplication factor = 70/(70 - (10 + 5)) = 70/55 = 1.27. Thus the total ADAS-Cog11 score for this example will be the sum of the remaining 9 items multiplied by 1.27. The imputed number will be rounded up to the nearest integer. If more than 3 items are missing, the total score for ADAS-Cog11 at that visit will be considered missing. The same imputation technique will be applied to the ADAS-Cog12 and ADAS-Cog14. Note that it is possible to have an imputed total score for ADAS-Cog11 but not have a imputed total score for ADAS-Cog14. Conversely, if 4 of the items for ADAS-Cog11 are missing and the additional items for ADAS-Cog14 are available, the total score for ADAS-Cog11 will be missing, but an imputed total score for ADAS-Cog14 will be available. For CDR-SB and ADCS-ADL, if less than 30% of the items are missing, the total score will be imputed as the prorated total calculated from the nonmissing items. The imputed total will be rounded up to the nearest integer. If the nearest integer is greater than the maximum possible score, the imputed score will be equal to the maximum score for that scale. If more than 30% of the items are missing, the total score for at that visit will be considered missing. For all other scales, if any item is missing, any total or sum involving that item will be considered missing. All efficacy and safety hypothesis tests will use a 2-sided 0.05 significance level unless otherwise stated.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 16 Any change to the data analysis methods described in the protocol will require an amendment ONLY if it changes a principal feature of the protocol. Any other change to the data analysis methods described in the protocol and the justification for making the change will be described in the clinical study report (CSR). Additional exploratory analyses of the data will be conducted as deemed appropriate. 8.2.4.1. Analysis of Primary Outcome and Methodology for FDA The primary objective of this study is to test the hypothesis that LY450139 given orally will slow decline of AD as compared to placebo. For the FDA, this will be assessed using a slope analysis of 2 coprimary outcomes (ADAS-Cog11 and ADCS-ADL), in which the specific hypothesis is that the rate of change over time (slope) for LY450139 will be significantly less than that for placebo, as assessed using a repeated measures model. Specifically, each of the two primary endpoints, ADAS-Cog11 and the ADCS-ADL, will be analyzed separately using a repeated-measures mixed model analysis. The change from baseline score at each visit post baseline during the initial treatment period will be the dependent variable. The linear model for the fixed effects will include terms for 7 independent effects: baseline score, investigator, treatment, MMSE stratification factor at Visit 1 (mild or moderate), time, time-by-treatment interaction, and concomitant AChEI or memantine use (yes/no) at each visit. Time will be assumed to be a continuous variable calculated as number of days between baseline and each visit post baseline during the initial treatment period. The within-patient variance-covariance matrix will be assumed to be unstructured. The null hypotheses are that (1) the contrast of 140 mg versus placebo equals zero, (2) the contrast of 100 mg versus placebo equals zero, and (3) the contrast of 140 mg versus 100 mg equals zero. The estimation method will use restricted maximum likelihood and use the Kenward-Rodgers estimate for the denominator degrees of freedom. Additionally, comparison of least-square means at last visit using results from the abovespecified repeated measures mixed model will be performed. This analysis may be visualized in Figure LFAN.2 as a comparison of the slope between Point A and Point B with the slope between Point A and Point C. 8.2.4.1. Analysis of Primary Outcome The primary objective of this study is to test the hypothesis that LY450139 given orally will slow the decline associated with AD as compared to placebo. 8.2.4.1.1. MMRM Analysis Each of the 2 primary endpoints, ADAS-Cog11 and the ADCS-ADL, will be analyzed separately using an MMRM analysis, in which the specific hypothesis is that the change from baseline at end of initial treatment period for LY450139 will be significantly less

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 17 than that for placebo. The change from baseline score at each visit post-baseline (when the scales were accessed) during the initial treatment period will be the dependent variable. The model for the fixed effects will include terms for 8 independent effects: baseline score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), visit, treatment-by-visit interaction, concomitant AChEI or memantine use at baseline (yes or no), and age at baseline. Visit will be considered a categorical variable with values equal to the visit numbers at which the scales were assessed. The null hypotheses are that (1) the contrast of 140 mg versus placebo equals zero, (2) the contrast of 100 mg versus placebo equals zero, and (3) the contrast of 140 mg versus 100 mg equals zero. A rejection of the null hypothesis in favor of the alternate that shows LY450139 is superior to placebo will demonstrate a treatment effect. An unstructured covariance matrix will be used to model the within-patient variance-covariance errors. If the unstructured covariance structure matrix results in a lack of convergence, the heterogeneous Toeplitz covariance structure, followed by the heterogeneous autoregressive covariance structure, will be used. The Kenward-Roger approximation will be used to estimate the denominator degrees of freedom. This analysis may be visualized in Figure LFAN.2 as a comparison of the decline between Points A and B to the decline between Points A and C. 8.2.4.1.2. Sensitivity Analysis Sensitivity analyses of missing data will be performed to assess the robustness of results to missing at random (MAR) assumption. We define sensitivity analyses as one in which several statistical methods are considered simultaneously and/or where a statistical model is further scrutinized using specialized tools. The simplest procedure is to fit a selected number of (nonrandom) models, which are all deemed plausible or one in which the primary analysis is supplemented with a number of variations. In order to assess the potential impact of missing not at random (MNAR) data on inference of treatment effect, the Diggle and Kenward selection model for informative dropout will be implemented to compare parameter estimates from models with MAR and MNAR missing data mechanisms (Diggle and Kenward 1994). In addition, local and global influence tools will be used to assess the impact of influential patients and/or observations on inferences of treatment effects under the MAR assumption (Shen et al. 2006). We will also perform sensitivity analysis using pattern-mixture modeling (Molenberghs et al. 2004) to complement the selection models. The results from all the nonrandom models will be collectively used to judge the validity of the MAR assumption and its impact on differences in dropout rates between the treatment groups.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 18 8.2.4.2. Analysis of Primary Outcome and Methodology for CHMP The primary objective of this study is to test the hypothesis that LY450139 given orally will slow decline of AD as compared to placebo. For the CHMP, the primary objective will be assessed by change from baseline to endpoint using 2 coprimary outcomes: the ADAS-Cog11 and the ADCS-ADL. The specific hypothesis is that the change from baseline to endpoint will be significantly less than that for placebo, as assessed by a stratified analysis of baseline to endpoint change. Specifically, a summary score approach based on a stratification method (Dawson 1994; Thomas et al. 2000) will be applied to analyze the data. We will stratify the analysis according to patterns of missing data and then combine the strata-specific test statistics into an overall test of group equality. The method has three steps, discussed below. [1] Strata are selected. For this study, three strata will be used (see the ERB supplement for more details). [2] A test statistic is calculated within each stratum comparing the treatment to placebo. For this study, an analysis of covariance (ANCOVA) will be conducted with the dependent variable, the change score from baseline to the last visit. The ANCOVA model for each stratum will include terms for 6 independent effects: baseline score, investigator, treatment, MMSE stratification factor at Visit 1 (mild or moderate), concomitant AChEI or memantine use (yes/no), and age. The test statistic will be the t statistic comparing the least-square means of treatment to placebo. The t statistics for each stratum will be denoted as T1, T2, and T3. [3] Weights are assigned to each stratum, the test statistics from each stratum are combined, and a weighted statistic is calculated. Strata with more patients are given greater weight than strata with fewer patients. The weights for each stratum for this analysis (denoted as W1, W2, and W3) will be chosen using the formula shown below. SQRT ((ns1*ns2)/(ns1+ns2)). Here, (s = 1, 2, 3) represents the three strata; ns1 and ns2 represent the number of patients with observations for the treatment groups that are being compared. The weighted statistic Zs is computed as Z = (W1*T1 + W2*T2 + W3*T3)/(SQRT(W12 + W22 + W32)). Under the null hypothesis of no treatment effect, Zs is asymptotically normal with mean = 0 and variance 1; thus, the calculated values of Zs can be compared with the standard normal

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 19 distribution, and the p-value for the test of hypothesis of no difference between treatments obtained. This method will be applied separately to each of the following comparisons: (1) 140 mg versus placebo, (2) 100 mg versus placebo, and (3) 140 mg versus 100 mg. Each of the two primary endpoints, ADAS-Cog11 and the ADCS-ADL, will be analyzed separately using this stratification approach. The analysis may be visualized in Figure LFAN.2 as a comparison of the change from Point A to Point B with the change from Point A to Point C. 8.2.4.2. Efficacy Analysis of Secondary Endpoints—Initial Treatment Period Secondary efficacy endpoints for the initial treatment period will be evaluated as discussed below. •

Slopes of the 2 coprimary outcomes (ADAS-Cog11 and ADCS-ADL) will be assessed separately using an MMRM analysis. The change from baseline score at each visit post baseline during the initial treatment period will be the dependent variable. The model for the fixed effects will include terms for 8 independent effects: baseline score, investigator, treatment, MMSE stratification factor at Visit 1 (mild or moderate), time, time-by-treatment interaction, and concomitant AChEI or memantine use at baseline (yes or no), and age at baseline. Time will be assumed to be a continuous variable calculated as number of days between baseline and each visit post baseline during the initial treatment period. The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. To test for linearity, a quadratic slopes model will be fitted to the ADAS-Cog11 and ADCS-ADL data separately. The quadratic model will include the linear component of time (TIME) and a quadratic component of time (TIME*TIME), the linear component of time X treatment interaction (TIME*TREATMENT) and quadratic component of time X treatment interaction (TIME*TIME*TREATMENT). If the quadratic component of time (TIME*TIME) and the interaction effect of TIME*TIME*TREATMENT is not significant, then the data can be considered as linear and the linear slopes model will be considered appropriate.



Analysis of patients who were completers of the initial treatment period will be done for the coprimary outcomes of ADAS-Cog11 and ADCS-ADL using an MMRM. The change from baseline at each visit post baseline will be the dependent variable. The model will include terms for the 8 independent effects listed above (Section 8.2.4.1.1). The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2)

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 20 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. •

An MMRM analysis of the coprimary outcomes of ADAS-Cog11 and ADCSADL will be conducted that excludes patients who changed doses of AChEI and/or memantine anytime during the study. The change from baseline at each visit post-baseline will be the dependent variable. The model will include terms for the 8 independent effects listed above (Section 8.2.4.1.1). The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero.



The additional clinical and outcome measurements listed below will be analyzed separately using an MMRM. The change from baseline at each post-baseline visit during the initial treatment period will be the dependent variable. The model for the fixed effects will include terms for the 8 independent effects listed above (Section 8.2.4.1.1). The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. The outcomes that will be analyzed are listed below. o Change-from-baseline difference as obtained from the ADASCog12. o Change-from-baseline difference as obtained from the ADASCog14. o Change-from-baseline difference in each of the 2 components of ADCS-ADL (basic ADL and instrumental ADL). o Change-from-baseline difference in global clinical benefit as obtained from the CDR-SB. o Change-from-baseline difference obtained from the MMSE total score. o Change-from-baseline difference in behavioral disturbance as measured by NPI total score (frequency multiplied by severity). o Change-from-baseline difference in resource utilization as measured by RUD-Lite scores. o Change-from-baseline difference in quality of life as measured by EQ-5D Proxy utility score. o Change-from-baseline difference in quality of life as measured by QoL-AD total score.



For the NPI, separate analysis of patients who were above and below the threshold of 10 points at baseline will be done as follows:

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 21 o For patients without clinically meaningful behavioral or psychiatric disturbances at baseline (NPI < 10), a survival analysis of time to first worsening of symptoms (NPI ≥10) will be done. Time to first worsening of the NPI score is defined as total days between randomization date and date of first visit at which the NPI score worsens (≥10). The null hypothesis is that the median time to first worsening of the NPI score estimated using the Kaplan-Meier product limit method will be identical for all treatment groups. The log-rank test will be used to test the null hypothesis against the alternate hypothesis that the median time to first worsening of the NPI score is not the same between the groups. o For patients who were at or above the threshold of 10 points at baseline, attenuation in the worsening of behavioral symptoms after baseline will be compared between treatment groups using an MMRM analysis. The change from baseline score at each visit post-baseline during the treatment period will be the dependent variable. The model for the fixed effects will include terms for 8 independent effects: baseline NPI score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), visit, treatment-by-visit interaction, concomitant AChEI or memantine use at baseline (yes/no), and age at baseline. Visit will be considered a categorical variable with values equal to the visit numbers at which the scales were assessed. The null hypothesis is that the contrast of (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups at the last visit equals zero. •

The change from baseline to the end of the treatment period using LOCF method for the co-primary outcomes (ADAS-Cog11 and ADCS-ADL) will be assessed separately using an analysis of covariance (ANCOVA) model. The model will include terms for 6 independent effects: baseline score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), concomitant AChEI or memantine use at baseline (yes/no), and age at baseline. The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero.

8.2.4.3. Efficacy Analysis of Secondary Endpoints—Initial Treatment Period Analysis of patients who were completers of the initial treatment period will be done for the coprimary outcomes of ADAS-Cog11 and ADCS-ADL using an ANCOVA. The change at endpoint from baseline will be the dependent variable. The ANCOVA model will include terms for the 6 independent effects listed above (Section 8.2.4.2). The null LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 22 hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. We will also analyze additional clinical and outcome measurements (below) that will be important to prescribing physicians and patients to understand the impact of LY450139 on AD treatment. • Change-from-baseline difference as obtained from the ADAS-Cog14 •

Change-from-baseline difference in global clinical benefit as obtained from the CDR-SB



Change-from-baseline difference in behavioral disturbance as measured by NPI total score (frequency multiplied by severity)



Change-from-baseline difference in resource utilization as measured by RUD-Lite scores



Change-from-baseline difference in quality of life as measured by EQ-5D score

Each of the outcomes (ADAS-Cog14, CDR-SB, NPI, RUD-Lite, EQ-5D) will be analyzed separately using a repeated measures mixed model analysis. The change from baseline score at each visit post baseline during the initial treatment period will be the dependent variable. The linear model for the fixed effects will include terms for the 7 independent effects listed above (Section 8.2.4.1). The within-patient variancecovariance matrix will be assumed to be unstructured. The null hypotheses are that (1) the contrast of 140 mg versus placebo equals zero, (2) the contrast of 100 mg versus placebo equals zero, and (3) the contrast of 140 mg versus 100 mg equals zero. The estimation method will use restricted maximum likelihood and use the Kenward-Rodgers estimate for the denominator degrees of freedom. We will also analyze change-from-baseline difference as obtained from the MMSE scores using a repeated measures mixed model analysis. The change from visit 1 score at each visit post baseline during the initial treatment period will be the dependent variable. The linear model for the fixed effects will include terms for the 7 independent effects listed above (Section 8.2.4.1). The within-patient variance-covariance matrix will be assumed to be unstructured. The null hypotheses are that (1) the contrast of 140 mg versus placebo equals zero, (2) the contrast of 100 mg versus placebo equals zero, and (3) the contrast of 140 mg versus 100 mg equals zero. The estimation method will use restricted maximum likelihood and use the Kenward-Rodgers estimate for the denominator degrees of freedom.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 23 8.2.4.3. Efficacy Analysis of Secondary Endpoints—Delayed-Start Period 8.2.4.3.1. Main Analysis of Delayed-Start Period After at least 64 weeks of double-blind, randomized treatment (initial treatment period), patients on placebo will be switched to treatment with 140 mg/day LY450139 (delayedstart period). Patients and investigators will be blinded to the timing of this change. The change from baseline to endpoint in ADAS-Cog11 will be analyzed using an ANCOVA including terms for the 6 independent effects: baseline score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), concomitant AChEI or memantine use at baseline (yes/no), and age at baseline. In this case, however, only patients from the original 140 mg dose group who continued in the study past the point of the delayed start and patients from the original placebo group who continued in the study past the point of the delayed start will be included in the analysis. The hypothesis of interest is that the decline, as measured by change from baseline (Visit 2) at 88 weeks in ADAS-Cog11 and ADCS-ADL, will be worse for the original placebo-treated patients compared to the patients treated with 140 mg from the start of the study. In Figure LFAN.2, this may be understood as a comparison of change from Point A to Point D with change from Point A to Point E. This analysis will be repeated for the co-primary outcome of ADCS-ADL. 8.2.4.3.2. Test for parallelism during Delayed Start Period To test for parallelism of slopes of cognitive decline during the delayed start period, the following analysis will be done using an ANCOVA. The change in ADAS-Cog11 from the end of the initial treatment period to the last observation during the delayed-start period will be the dependent variable. The model will include terms for 8 independent fixed effects: score at the end of the initial treatment period, investigator, treatment, MMSE stratification factor at Visit 1 (mild or moderate), time, time-by-treatment interaction, concomitant AChEI or memantine use at baseline (yes or no), and age at baseline. Time will be assumed to be a continuous variable calculated as number of days between the end of the initial treatment period and last visit during the delayed-start period. The specific hypothesis is that patients switched to 140 mg for the delayed start period do not decline significantly less in cognition than patients treated with 140 mg from beginning of study as assessed using the ADAS-Cog11. In Figure LFAN.2 above this may be understood as a comparison of the slope from Point B to Point D with the slope from Point C to Point E (with intercepts maintained). The noninferiority margin for this hypothesis is specified as 50% of the treatment difference observed at the end of the initial treatment period. The hypothesis will be tested by constructing a 90% 1-sided confidence interval of the difference in slopes from Points B to D and Points C to E. If the lower limit of the confidence interval rules out the difference which would have been

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 24 obtained if 50% of the observed difference had been lost, the slopes from Points B to D and Points C to E will be considered parallel. This analysis will be repeated for the co-primary outcome of ADCS-ADL. 8.2.4.3.3. Additional Analyses Each of the 2 coprimary endpoints (change from baseline to endpoint in ADAS-Cog11 and ADCS-ADL) will also be analyzed separately using an ANCOVA including terms for the 6 independent effects: baseline score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), concomitant AChEI or memantine use at baseline (yes/no), and age at baseline. In this case, however, only patients from the original 100mg dose group who continued in the study past the point of the delayed start and patients from the original placebo group who continued in the study past the point of the delayed start will be included in the analysis. The hypothesis of interest is that the decline, as measured by change from baseline (Visit 2) at 88 weeks in ADAS-Cog11 and ADCSADL, will be worse for the original placebo-treated patients compared to the patients treated with 100 mg from the start of the study. 8.2.4.4. Efficacy Analysis of Secondary Endpoints—Delayed-Start Period Main Analysis of Delayed-Start Period. After at least 64 weeks of double-blind, randomized treatment (initial treatment period), patients on placebo will be switched to treatment with 140 mg/day LY450139 (delayed-start period). Patients and investigators will be blinded to the timing of this change. Each of the two coprimary endpoints (change from baseline to endpoint in ADAS-Cog11 and the ADCS-ADL) will be analyzed separately using an ANCOVA including terms for the 6 independent effects listed in Section 8.2.4.2. In this case, however, only patients from the original 140-mg dose group who completed the initial treatment period and from the original placebo group who continued in the study past the point of the delayed start will be included in the analysis. The hypothesis of interest is that the decline, as measured by change from baseline (Visit 2) at 88 weeks in ADAS-Cog11 and ADCSADL, will be worse for the original placebo-treated patients compared to the patients treated with 140 mg from the start of the study. In Figure LFAN.2, this may be understood as a comparison of change from Point A to Point D with change from Point A to Point E. Absence of Symptomatic Effect of Treatment. To test for an absence of symptomatic effect of treatment, change in each of the two endpoints, ADAS-Cog11 and ADCS-ADL, during only the delayed-start period of the study will be analyzed separately using an ANCOVA model including terms for the 6 independent effects listed in Section 8.2.4.2, as further discussed below. Change from beginning of delayed-start period to endpoint score will be the dependent variable. The hypothesis of interest is that the decline, as measured by change from LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 25 beginning of delayed start to 88 weeks in ADAS-Cog11 and ADCS-ADL, will be no better for the original placebo-treated patients compared to the patients treated with 140 mg from the start of the study. In Figure LFAN.2, this may be understood as a comparison of the change from Point B to Point D with the change from Point C to Point E. Additional Analyses Each of the two coprimary endpoints (change from baseline to endpoint in ADAS-Cog11 and ADCS-ADL) will also be analyzed separately using an ANCOVA including terms for the 6 independent effects listed in Section 8.2.4.2. In this case, however, only patients from the original 100-mg dose group who completed the initial treatment period and from the original placebo group who continued in the study past the point of the delayed start will be included in the analysis. The hypothesis of interest is that the decline, as measured by change from baseline (Visit 2) at 88 weeks in ADAS-Cog11 and ADCSADL, will be worse for the original placebo-treated patients compared to the patients treated with 100 mg from the start of the study. 8.2.4.4. Subgroup Analyses Subgroup analyses for the initial treatment period will be performed as described below. •

To assess the effects of various demographic and baseline characteristics, subgroup analyses of the 2 primary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. Six subgroups will be included for this analysis, provided there are at least 50 patients for all categories of a subgroup in each treatment group: (1) gender, (2) race (dichotomized based on distribution of ethnicity in study), (3) ApoE, (4) family history of AD, (5) depression, and (6) anticholinergic medication use. These analyses will be performed using an MMRM analysis of change from baseline at each visit post-baseline. The MMRM will include the 8 previously described independent effects (Section 8.2.4.1.1) as well as subgroup and treatment-by-subgroup interaction. The treatment-by-subgroup interaction will be tested to determine whether treatment differences in the coprimary outcomes are the same for each subgroup.



To assess the consistency of treatment effects across sites (investigators), a subgroup analysis of the coprimary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. These analyses will be performed using an MMRM analysis of change from baseline at each visit postbaseline. The MMRM will include the 8 previously described independent effects (Section 8.2.4.1.1) as well as a treatment-by-investigator interaction.



To assess the effects of donepezil on treatment, subgroup analysis of the coprimary endpoints (ADAS-Cog11 and ADCS-ADL) will be performed.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 26 Patients who were on donepezil at baseline will be compared to patients who were not on donepezil at baseline for both treatment groups, provided there are at least 50 patients for both categories of donepezil (yes or no) in each treatment group. These analyses will be performed using an MMRM analysis of change from baseline at each visit post-baseline. The MMRM will include terms for 9 independent effects: baseline score, investigator, treatment, MMSE stratification factor at baseline (mild or moderate), visit, treatment-by-visit interaction, age at baseline, concomitant donepezil use at baseline (yes or no), and treatment-bydonepezil interaction. All subgroup analyses will be considered secondary analyses. Additional subgroup analyses may be performed as suggested by the data. 8.2.4.5. Subgroup Analyses To assess the effects of various demographic and baseline characteristics, subgroup analyses of the two primary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. These analyses will be performed using a repeated measures mixed model analysis of change from baseline over time. The repeated measures mixed model will include the 7 previously described independent effects (Section 8.2.4.1) as well as subgroup and treatment-by-subgroup interaction. The treatment-by-subgroup interaction will be tested to determine whether treatment differences in the coprimary outcomes are the same for each subgroup. Six subgroups will be included for this analysis, provided there are at least 50 patients for all categories of a subgroup in each treatment group: (1) gender (2) race (dichotomized based on distribution of ethnicity in study), (3) ApoE, (4) family history of AD, (5) depression, and (6) anticholinergic medication use. To assess the consistency of treatment effects across sites (investigators), a subgroup analysis of the coprimary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. These analyses will be performed using a repeated measures mixed model analysis of change from baseline over time. The repeated measures mixed model will include the 7 previously described independent effects (Section 8.2.4.1) as well as a treatment-byinvestigator interaction. To assess the consistency of treatment effects across MMSE scores at Visit 1 (mild, moderate), a subgroup analysis of the coprimary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. These analyses will be performed using a repeated measures mixed model analysis of change from baseline over time. The repeated measures mixed model will include the 7 previously described independent effects (Section 8.2.4.1), as well as a treatment-by-MMSE stratification at Visit 1 (mild, moderate) interaction.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 27 To assess the effects of donepezil on treatment, subgroup analysis of the coprimary endpoints, ADAS-Cog11 and ADCS-ADL, will be performed. Patients who were on donepezil will be compared to patients who were not on donepezil for all treatment groups. These analyses will be performed using a repeated measures mixed model analysis of change from baseline over time. The repeated measures mixed model will include terms for 8 independent effects: baseline score, investigator, treatment, MMSE stratification factor at Visit 1 (mild or moderate), time, time-by-treatment interaction, treatment-by-donepezil interaction, and concomitant donepezil use (yes/no) at each visit. All subgroup analyses will be considered secondary analyses. Additional subgroup analyses may be performed as suggested by the data. 8.2.4.56. Analysis of Plasma Aβ To evaluate the reduction of plasma Aβ after treatment, an ANCOVA will be used to compare percent change in plasma Aβ from baseline to 6 hours post dose at 6 weeks after randomization. The ANCOVA model will include the following independent effects: baseline plasma Aβ, investigator, treatment, and age. The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. Additionally, percent change in plasma Aβ from baseline to 6 hours post dose at Week 52 will be compared to percent change in plasma Aβ from baseline to 6 hours post dose at Week 6 using a one-sample t-test. The hypothesis of interest is that the difference between percent change in plasma Aβ at Week 52 and Week 6 will be equal to zero for both 140 mg and 100 mg To assess the impact of plasma Aβ on cognition and function with treatment, a correlation analysis using Spearman’s rank correlation will be done on percent change in plasma Aβ from baseline to 6 hours post dose at Week 6 with change from baseline to endpoint for both ADAS-Cog11 and ADCS-ADL; this will be done by combining the 140-mg treatment group and the 100-mg treatment group. Additionally, an ANCOVA will be used to compare percent change in plasma Aβ from baseline to 6 hours post dose at Week 52 between treatment groups. The ANCOVA model will include the following independent effects: baseline plasma Aβ, investigator, treatment, and age. The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and 100 mg groups equals zero. To assess the impact of plasma Aβ on cognition and function with treatment, Spearman’s rank correlation coefficient will be obtained on percent change in plasma Aβ from baseline to 6 hours post dose at Week 6 with change from baseline to end of initial treatment period for both ADAS-Cog11 and ADCS-ADL; this will be done within each treatment group. LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 28 8.2.4.7.6.Analyses of Biomarker Outcomes Additional endpoints related to CSF biomarkers (including CSF Aβ1-42 and CSF tau) and 3 imaging endpoints (vMRI, FDG-PET and PIP-PET amyloid-imaging tracer) will be assessed. The dependent variable for each biomarker is its change from baseline to endpoint. The analysis of the change from baseline will use an ANCOVA model with fixed effects of (1) baseline biomarker result, (2) investigator, (3) treatment, and (4) age. The null hypotheses are that the difference in least-square means between the (1) 140 mg and placebo groups equals zero, (2) 100 mg and placebo groups equals zero, and (3) 140 mg and placebo 100 mg groups equals zero. Also, to assess relationships between the different biomarker outcomes, Spearman rank correlation coefficients between plasma Aβ, FDG-PET, vMRI, the amyloid-imaging tracer, and CSF will be obtained. Additionally, Spearman rank correlation coefficients between biomarker data and clinical outcomes will be obtained at the visits when both the biomarker and clinical outcomes are collected. 8.2.4.8.7.Statistical Details of the Gatekeeping Strategy A gatekeeper strategy will be used to decide the sequence of hypothesis testing to protect against type I error of falsely rejecting the null hypothesis. The gatekeeping strategy will include the following tests done in sequential order. All comparisons will be done for the 140 mg group compared to the placebo group at an alpha equal to 0.05. The significance of the coprimary endpoints of changes in ADAS-Cog11 and ADCS-ADL between the 140-mg and placebo groups will be assessed first. Following the significance of the coprimary endpoints, the change in ADAS-Cog11 from baseline to the end of the delayed-start period will be assessed. Change from baseline to 6 hours post dose at Week 6 in plasma Aβ will then be assessed, followed by assessment of the change from baseline to endpoint of the two imaging biomarkers (vMRI and FDG-PET). The change from baseline to endpoint for the additional clinical outcomes CDR-SB and NPI will be assessed last. As discussed above, ITT comparisons between patients randomized to 140 mg and patients randomized to placebo for analysis of primary and secondary outcomes variables will be included in the gatekeeper strategy; however, if a large proportion of patients in the 140-mg dose group are discontinued from the study for any reason, the 100-mg dose group will be compared with placebo for these analyses. The criteria listing the specific numbers of patients who discontinue needed for this change will be finalized prior to datalock and unblinding. A gatekeeping strategy will be used to decide the sequence for testing hypotheses to protect against type I error of falsely rejecting a null hypothesis. All comparisons will be done for the 140 mg group compared to the placebo group at an alpha equal to 0.05. The gatekeeping strategy will include the following tests done in sequential order (Figure LFAN.3) LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 29 The test of hypothesis of the coprimary endpoints of changes in ADAS-Cog11 and ADCS-ADL between the 140-mg and placebo groups will be assessed first. The test of hypothesis of change in ADAS-Cog11 from baseline to the end of the delayed-start period will then be assessed. The test of hypothesis of changes in NPI from baseline to end of the initial treatment period will be assessed next. The test of hypothesis of changes in CDR-SB from baseline to end of the initial treatment period will be assessed last. MMRM Analysis of coprimary Outcomes During Initial Treatment Period ADAS-Cog11 H0: LS Means (at last visit) for 140 mg LY450139 = LS Means (at last visit) for placebo ADCS-ADL H0: LS Means (at last visit) for 140 mg LY450139 = LS Means (at last visit) for placebo ↓ Delayed-Start Period: Analysis of Change from Baseline (Week 0) to end of DelayedStart (Week 88) ADAS-Cog11 H0: LS Means (at Week 88) for 140 mg LY450139 = LS Means (at Week 88) for placebo ↓ MMRM Analysis of NPI During Initial Treatment Period NPI H0: LS Means (at last visit) for 140 mg LY450139 = LS Means (at last visit) for placebo ↓ MMRM Analysis of CDR-SB During Initial Treatment Period CDR-SB H0: LS Means (at last visit) for 140 mg LY450139 = LS Means (at last visit) for placebo Figure LFAN.3.

Gatekeeping Strategy.

As discussed above, ITT comparisons between patients randomized to 140 mg and patients randomized to placebo for analysis of primary and secondary outcomes variables will be included in the gatekeeping strategy; however, if a large proportion of patients in the 140-mg dose group are discontinued from the study for any reason, the 100-mg dose group will be compared with placebo for these analyses. The specific criteria that would trigger the switch will be pre-specified as follows: –

For the 140 mg group, the drop-out rate is more than 50% and the rate exceeds the rate of the placebo group by at least 20%, and



Drop-out rate of the 100 mg group does not meet the above criteria.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 30 8.2.4.8. Specification of A Priori Sequence to Test Hypothesis for Biomarker Data A prioritized ordering will be used to test the hypothesis for biomarker data. The order will include the test of hypothesis for plasma Aβ followed by the test of hypothesis for FDG-PET, vMRI, and CSF. All comparisons will be done for the 140 mg group compared to the placebo-treated group at the 0.05 significance level.

8.2.5. Safety Analyses—Initial Treatment Period Safety will be assessed by summarizing and analyzing TEAEs, laboratory analytes, vital signs, and ECGs during the initial treatment period. Safety analyses for the initial treatment period will include all pairwise comparisons between the treatment groups (1) 140 mg versus placebo, (2) 100 mg versus placebo, and (3) 140 mg versus 100 mg. All hypotheses will be tested at significance level = 0.05. No adjustments for multiple comparisons will be made. For analyses of TEAEs and change from baseline to endpoint, baseline will be the last visit prior to randomization. Treatment-emergent adverse events during the initial treatment period will be assessed by comparing proportions of patients between treatment groups and placebo using Fisher’s exact test. Treatment-emergent events are defined as those events that were present during the study period but were not observed prior to start of the study period or those events for which severity increased compared to baseline. Subgroup analyses of TEAEs will also be done. The subgroups will include age (dichotomized by median age in study), gender, and race (dichotomized based on distribution of ethnicity in study). For each subgroup analysis, a pairwise Breslow-Day test of homogeneity of odds-ratios between each treatment group will be performed on the incidence of each TEAE. This test will assess the interaction between the subgroup and treatment. Treatment-group mean changes from baseline to endpoint in vital signs (pulse, blood pressure), weight, and laboratory analytes will be assessed using an ANCOVA model with treatment and investigator as independent factors and baseline value as a covariable in the model. Pairwise comparisons between treatment groups will be made by fitting linear contrasts to least-square means from the ANCOVA model. Within-group changes from baseline will also be assessed from this model. Treatment-group mean changes for each ECG parameter will be conducted as discussed here. Change from baseline to each postbaseline visit at which ECG measurements are taken will be assessed using an ANCOVA model with treatment and investigator as independent factors, and baseline value as a covariable in the model. Pairwise comparisons between treatment groups will be made by fitting linear contrasts to least-

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 31 square means from the ANCOVA model. Within-group changes from baseline will also be assessed from this model. Treatment-emergent changes from baseline in vital signs, weight, laboratory analytes, and ECG data during the initial treatment period will be assessed by pairwise comparisons between treatment groups with Fisher’s exact test. A treatment-emergent increase is defined as a change from a value less than or equal to the high limit at all baseline visits to a value greater than the high limit at any time post baseline during the study period. For analyses of treatment-emergent changes, baseline will be considered as all visits prior to randomization. 8.2.5.1. Adverse Events An AE will be defined as any untoward medical occurrence in a patient that was reported after informed consent had been obtained, without regard to the possibility of a causal relationship. Patients will be examined and questioned throughout the study regarding the occurrence of AE. Treatment-emergent adverse events will be defined as events that first occurred or worsened after initiation of therapy. Baseline will be considered as all visits prior to initiation of therapy (Visits 1 and 2). Serious adverse events are defined in Section 6.3.1.2. Fisher’s exact test will be used to evaluate pairwise treatment differences for SAEs (including deaths), discontinuations due to AEs, and incidence of TEAEs. Subgroup analyses of TEAEs will also be done. The subgroups will include age (dichotomized by median age in study), sex, and race (dichotomized based on distribution of ethnicity in study). For each subgroup variable, a Breslow-Day test of homogeneity of odds-ratios between treatment groups will be performed on the incidence of each TEAE. 8.2.5.2. Laboratory Analyses Laboratory measurements will be analyzed using continuous data (change from baseline) and categorical data (proportion of treatment-emergent abnormalities). Treatment-group mean changes from baseline to endpoint in laboratory analytes will be assessed using an ANCOVA model with treatment and investigator as independent factors and baseline value as a covariable in the model. Within-group changes from baseline will also be assessed from this model. Treatment differences in the proportion of patients with treatment-emergent high or treatment-emergent low or treatment-emergent abnormal laboratory values at (1) anytime and (2) endpoint will be assessed using Fisher’s exact test. A treatment-emergent high value is defined as a change from a value less than or equal to the high limit at all baseline visits to a value greater than the high limit either at endpoint or at any time after baseline. A treatment-emergent low value is defined as a change from a value greater

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 32 than or equal to the low limit at all baseline visits to a value less than the low limit either at endpoint or at any time after baseline. A treatment-emergent abnormal value is defined as a change from normal at all baseline visits to abnormal either at endpoint or at any time after baseline. 8.2.5.3. Electrocardiograms ECG measurements will be analyzed using continuous data (change from baseline) and categorical data (proportion of treatment-emergent abnormalities). The analysis will be done for the following ECG measurements: PR, QRS, QT, QTc, JTc, and RR intervals. Analyses of QTc will be done on all patients. Exploratory analyses will be done on JTc patients. JTc patients will be defined as those patients with prolonged QRS at baseline and those who have experienced prolonged QRS during the study. All analyses of JTc and QTc will be carried out using the Fridericia correction method and a correction method calculated from the population studied for QTc and a correction method calculated from the population studied for JTc. Change from baseline to each post-baseline visit at which ECG measurements are taken will be assessed using an ANCOVA model with treatment and investigator as independent factors, and baseline value as a covariable in the model. Within-group changes from baseline will also be assessed from this model. Incidence of treatment-emergent abnormal ECGs during the initial treatment period will be assessed by comparisons between treatment groups with Fisher’s exact test. For analyses of treatment-emergent abnormal ECGs, baseline will be considered as all visits prior to randomization. 8.2.5.4. Vital Signs and Weight Vital sign measurements and weight will be analyzed using continuous data (change from baseline) and categorical data (proportion of treatment-emergent abnormalities). Treatment-group mean changes from baseline to endpoint in vital signs and weight will be assessed using an ANCOVA model with treatment and investigator as independent factors and baseline value as a covariable in the model. Within-group changes from baseline will also be assessed from this model. Treatment differences in the proportion of patients with treatment-emergent high or low vital signs and weight will be assessed between treatment groups using Fisher’s exact test. For analyses of treatment-emergent changes, baseline will be considered as all visits prior to randomization.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 33

8.2.6. Safety Analyses—Delayed-Start Period Safety will be assessed by summarizing and analyzing TEAEs, laboratory analytes, vital signs, and ECGs during the delayed-start period. Safety analyses for the delayed-start period will include all comparisons between the groups of patients initially randomized to 140-mg treatment group and the 100-mg treatment group. All hypotheses will be tested at significance level = 0.05. No adjustments for multiple comparisons will be made. Safety will be assessed by summarizing TEAEs, laboratory analytes, vital signs, and ECGs during the delayed-start period. The summaries will be provided for each of the treatment groups (i) patients randomized to 140 mg, (ii) patients randomized to 100 mg and (iii) patients randomized to placebo and switched to 140 mg during the delayed start period. No comparisons will be made between the treatment groups. For patients treated with placebo during the initial treatment period and switched to 140 mg for the delayed-start period, summary statistics of TEAEs, laboratory analytes, vital signs, and ECGs during the delayed-start period will be provided. No comparisons will be made to patients treated with 140 mg during the initial treatment period or to patients treated with 100 mg during the initial treatment period. For analyses of TEAEs and change from baseline to endpoint in vital signs, laboratory analytes, and ECG parameters, baseline for patients originally treated with placebo and switched to 140 mg will be the last visit in the initial treatment period prior to switching. The baseline for patients originally on 140 mg or 100 mg will remain the last visit prior to start of the initial treatment period. Treatment-emergent adverse events during the delayed-start period will be analyzed by comparing proportions between patients randomized to the 140-mg treatment group and patients randomized to the 100-mg treatment group using Fisher’s exact test. Frequencies of each TEAE during the delayed-start period for patients treated with placebo during the initial treatment period will be provided. Treatment-emergent adverse events during the entire study period (initial treatment period combined with the delayed-start period) will be analyzed by comparing proportions between patients randomized to the 140-mg treatment group and patients randomized to the 100-mg treatment group using Fisher’s exact test. Treatment-group mean changes in vital signs (pulse, blood pressure), weight, and laboratory analytes from baseline to endpoint during the delayed-start period will be compared between patients randomized to the 140-mg treatment group and patients randomized to the 100-mg treatment group. The mean changes will be analyzed using an ANCOVA model with treatment and investigator as independent factors and baseline value as a covariable in the model. Within-group changes from baseline will also be assessed from this model.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 34 Mean changes in vital signs and laboratory analytes during the delayed-start period for patients treated with placebo during the initial treatment period will be analyzed using an ANCOVA. The ANCOVA model will assess within-group change from baseline to endpoint and will include investigator and baseline values in the model. Treatment-group mean changes from baseline for each ECG parameter during the delayed-start period will be compared between patients randomized to the 140-mg treatment group and patients randomized to the 100-mg treatment group. Change from baseline to each visit during the delayed-start period at which ECG measurements are taken will be assessed using an ANCOVA model with treatment and investigator as independent factors and baseline value as a covariable in the model. Within-group changes from baseline will also be assessed from this model. Mean changes in ECG parameters during the delayed-start period for patients treated with placebo during the initial treatment period will be analyzed using an ANCOVA for each visit at which ECG measurements are taken. The ANCOVA model will assess withingroup change from baseline to each visit during the delayed-start period and will include investigator and baseline values in the model. Treatment-emergent changes from baseline in vital signs, weight, laboratory analytes, and ECG data during the delayed-start period will be assessed by comparisons between patients randomized to the 140-mg treatment group and patients randomized to the 100mg treatment group with Fisher’s exact test. For these analyses of treatment-emergent changes, baseline will be considered as all visits prior to randomization. Frequencies of each treatment-emergent change in vital signs, weight, laboratory analytes, and ECG data during the delayed-start period for patients treated with placebo during the initial treatment period will be provided. For this analysis of treatmentemergent changes, baseline will be considered the last two consecutive visits in the initial treatment period prior to switching. Frequencies of each TEAE during the delayed-start period for patients treated with 140 mg during the initial treatment period and patients treated with placebo during the initial treatment period and switched to LY450139 during the delayed-start period will be provided. Within-treatment-group mean changes in vital signs (pulse, blood pressure), weight, and laboratory analytes from baseline to endpoint during the delayed-start period will be analyzed using an ANCOVA model with investigator as independent factor and baseline value as a covariable in the model. Within-treatment-group mean change from baseline for each ECG parameter during the delayed-start period (at each visit when ECG measurements are taken) will be assessed using an ANCOVA model with investigator as an independent factor and baseline value as a covariable in the model.

LY450139 H6L-MC-LFAN(c) Confidential

Protocol Amendment Summary

Page 35 Proportion of patients with treatment-emergent abnormal vital signs, weight, laboratory analytes, and ECG data during the delayed-start period will be summarized for each treatment group. For these analyses of treatment-emergent abnormalities, baseline will be considered as all visits prior to randomization for patients randomized to 140 mg or 100 mg. For patients randomized to placebo and switched to 140 mg during the delayedstart period, baseline will be considered the last 2 consecutive visits in the initial treatment period prior to switching.

8.2.7. Safety Analyses—Follow-Up Visit Data collected at the follow-up visit (Visit 801) will be summarized. Means and standard deviations Absolute mean values of all analytes collected at this visit will be presented.

8.2.8. Interim Analyses and Data Monitoring Committee No interim analyses are planned for this study. If an unplanned interim analysis is deemed necessary, the appropriate Lilly regulatory scientist will be consulted to determine whether it is necessary to amend the protocol. The analysis would be conducted under the auspices of a DMC, and the interim analysis plan would be prespecified before any unblinding occurred. In addition, for safety reasons, the DMC will evaluate unblinded ADAS-Cog results after approximately 50% of the patients have had at least 12 months of treatment. This assessment will allow the DMC to confirm that clinically meaningful cognitive worsening is was not present in subjects treated with LY450139. Complete details The DMC will be available in first look at grouped cognitive measures masked to treatment assignment. The DMC will review unblinded data only if ADAS-Cog scores show differences between masked treatment groups at the p 20 bpm (standing to supine) (i.e. supine minus standing > 20) Increase of > 7% from baseline Decrease of > 7% from baseline > 38.3 °C and > 1.1 °C increase from baseline (> 101 °F and > 2 °F increase from baseline)

> 38.3 °C (or > 101 °F)

> 180 mmHg and increase of > 20 mmHg from baseline < 90 mmHg and decrease of > 20 mmHg from baseline > 105 mmHg and increase of > 15 mmHg from baseline < 50 mmHg and decrease of > 15 mmHg from baseline > 120 bpm and increase of > 15 bpm from baseline < 50 bpm and decrease of > 15 bpm from baseline Decrease of > 20 mmHg in systolic blood pressure (supine to standing)

FINAL

Statistical Analysis Plan

Page 20

Abbreviations: bpm=beats per minute, NA=not applicable.

5.6.6. Analyses of ECG data ECG measurements will be analyzed using continuous data (change from baseline) and categorical data (proportion of post-cessation abnormalities). The analysis will be done separately for the following ECG parameters: PR, QRS, QT, QTc, JTc, and RR intervals. Exploratory analyses will be done on JTc subjects. JTc subjects will be defined as those subjects with prolonged QRS (for males, >116 msec; for females >102 msec) at both baseline and at anytime during the study. All analyses of QTc will be carried out using the Fridericia correction method. All analyses of JTc will be carried out using the following correction JTc =JT/RR0.498. In addition, analyses will also be carried out on study population based corrections of QTc and JTc. Prior to study unblinding, the population based QTc and JTc corrections will be determined through a review of the baseline data for these ECG parameters. The correction methods determined will be documented as an appendix to the final statistical analysis plan. 5.6.6.1. Continuous Data Treatment-group mean changes in ECG from baseline to Endpoint Visit and Visits 802, 803, and 804 will be assessed using MMRM. The MMRM analysis will include treatment, investigator and visit as independent factors, baseline value as a covariable and treatment-by-visit interaction in the model. Within-group changes from baseline will also be assessed from this model. 5.6.6.2. Categorical data Treatment differences in the proportion of subjects with post dose cessation abnormal ECG at any time after cessation of dosing will be assessed using Fisher’s exact test. A post dose cessation abnormal value is defined as a change from normal at all baseline visits to abnormal at any time after cessation of dosing. The analyses of abnormal ECG will be repeated using subgroup categorizations based on a subject’s exposure to semagacestat. The same subgroups of semagacestat exposure specified in Section 5.5.2 will be used. Abnormal ECG criteria are presented in Table SAP.6.A and criteria for abnormal QTcF prolongation are presented in Table.SAP.6.B. Table SAP.6.A Criteria for Abnorm al ECG Param eters ECG Parameter PR Interval QRS Duration QT Interval Semagacestat

Gender Both Male Female Both

Criterion > 200 msec > 116 msec > 102 msec >450 msec FINAL

Statistical Analysis Plan RR Interval

Both

Page 21 > 120 bpm or 450 msec; in adult females, QTcF > 470 msec QTcF > 500 msec Increase of > 0 msec and ≤ 30 msec relative to baseline Increase of > 30 msec and ≤ 60 msec relative to baseline Increase of > 60 msec relative to baseline Increase of > 60 msec relative to baseline and QTcF > 500 msec

Post-cessation emergent abnormal ECG parameters (RR interval, PR interval, QRS duration and QT and QTcF intervals) are those values which are normal at all baseline visits and fall into the abnormal categories in Tables SAP.6.A and 6.B post-cessation. For each ECG parameter at each post-baseline visit, only subjects who had a normal baseline result and had a non-missing result at that post-cessation visit will be included in the denominator when computing the proportion of subjects with post-cessation emergent high, low, or abnormal ECG results. For extreme QTcF interval assessments, when computing proportion of subjects with post-cessation emergent abnormalities, only subjects who were normal at baseline and had a non-missing result at that post-cessation visit will be included in the denominator for criterion numbers 1 and 2 in Table. SAP.6.B respectively. For criterion numbers 3, 4, 5 and 6 in Table SAP.6.B, at each post- cessation visit, only subjects who had a normal baseline result and had a non-missing result at that post- cessation visit will be included in the denominator when computing the proportion of subjects with potentially clinically significant ECG QTcF intervals.

5.7. Data Monitoring Committee for Amended Study The independent, external DMC assigned to the original study will continue periodic study monitoring. The DMC will inform the sponsor if they have any recommendations based their assessments of the data.

Semagacestat

FINAL

Statistical Analysis Plan

Page 22

7. References Aisen PS, Davis KL, Berg JD, Schafer K, Campbell K, Thomas RG, Weiner MF, Farlow MR, Sano M, Grundman M, Thal LJ. 2000. A randomized controlled trial of prednisone in Alzheimer's disease. Alzheimer's Disease Cooperative Study. Neurology 54(3):588–593. Aisen PS, Schafer KA, Grundman M, Pfeiffer E, Sano M, Davis KL, Farlow MR, Jin S, Thomas RG, Thal LJ; Alzheimer's Disease Cooperative Study. 2003. Effects of rofecoxib or naproxen vs placebo on Alzheimer disease progression: a randomized controlled trial. JAMA 289(21):2819–2826. Galasko D, Kershaw PR, Schneider L, Zhu Y, Tariot PN. 2004. Galantamine maintains ability to perform activities of daily living in subjects with Alzheimer's disease. J Am Geriatr Soc 52(7):1070–1076. Reines SA, Block GA, Morris JC, Liu G, Nessly ML, Lines CR, Norman BA, Baranak CC; Rofecoxib Protocol 091 Study Group. 2004. Rofecoxib: no effect on Alzheimer's disease in a 1-year, randomized, blinded, controlled study. Neurology 62(1):66–71. Shen S, Beunckens C, Mallinckrodt C, Molenberghs G. 2006. A local influence sensitivity analysis for incomplete longitudinal depression data. J Biopharmaceut Stat 16:365-384. Thal LJ, Carta A, Clarke WR, Ferris SH, Friedland RP, Petersen RC, Pettegrew JW, Pfeiffer E, Raskind MA, Sano M, Tuszynski MH, Woolson RF. 1996. A 1-year multicenter placebo-controlled study of acetyl-L-carnitine in subjects with Alzheimer's disease. Neurology 47(3):705–711.

Semagacestat

FINAL

Patient Information and Consent Form Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo Introduction You are invited to take part voluntarily in a research study of an investigational study drug known as LY450139. Your participation in this study is expected to last up to 2 years. Up to 1500 patients will be participating in this study. Eli Lilly and Company is sponsoring this study. Before agreeing to participate in this research study, it is important that you read and understand this form. It describes the purpose, procedures, benefits, risks and discomforts, and precautions of the study. It also describes the alternative procedures that are available to you and your right to withdraw from the study at any time. If you participate, you will receive a copy of this form to keep for your records.

Purpose of the Study The primary purpose of your participation in this study is to help answer the following research question(s), and not to provide you treatment for your condition: •

The safety of LY450139 and any side effects that might be associated with it.



Whether LY450139 can help patients with mild to moderate Alzheimer’s disease.

Qualifications to Participate The doctor in charge of this study or a member of the study staff has discussed with you the requirements for participation in this study. You cannot participate in this study if: •

you have a history of HIV;



you have a history within the last 5 years of alcohol or drug abuse;



you have a history within the last 5 years of a serious infection affecting the brain;



you have multiple or severe drug allergies;



you have serious or uncontrolled illnesses including those that affect your heart, kidneys, lungs, blood, liver, brain, nerves, ability to fight off infection, bowels stomach, mental health, abnormal blood/urine tests or other condition that the investigator feels could affect your participation in the trial;



you cannot swallow a whole tablet;



you are taking certain medications or may need to use certain medications during the study that are not allowed in this trial;

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 1 of 16



you have uncontrolled high blood pressure.

If female, in order to participate in this study you must be post-menopausal, as evidenced by a lack of menstruation for at least 12 consecutive months or have had both ovaries removed. You agree to use the study drug only as instructed by your study doctor and staff, and to return any unused study drug and containers at the end of your participation in the study or as otherwise instructed by the study doctor. It is important that you are completely truthful with the study doctor and staff about your past medical history as well as any symptoms experienced during the study. It may be harmful to you or to other people who may take the drug if you are not truthful with the study doctor and staff. You should not participate in this study if you do not meet all qualifications.

Study Procedures If you decide to take part in this study, you will need to visit the hospital or study doctor’s office at least 20 times over the next 22 months. You will first complete a “screening” visit so that your study doctor can determine if you may participate in the study. The initial evaluation will include a detailed medical history, physical exam, three electrocardiograms (ECG), and a routine blood and urine test. You will also be asked to answer two questionnaires. This “screening” visit may take from 3 to 14 days, but could take up to 30 days to have all the tests completed and the results to the doctor. If the study doctor says you are allowed to participate, you will be asked to return to the study site in about one week when you will be randomly assigned to LY450139 or placebo (a pill that has a similar appearance to the study drug but has no medicine). You will asked to return for a visit every week or every other week for 2 months and then you will have visits once a month or less often for the remainder of the study. You will also have a last study visit 4 weeks after you take your last dose of study drug. Your study doctor or nurse will make sure you know when you need to attend. You will be asked to take the study drug in the form of three tablets once each morning at about the same time. You should swallow the study tablets whole; do not crush the tablets. On some days you will not take the drug yourself, but will be given the study drug at the study site during your visit by the doctor or nurse. Neither you nor your study doctor will know what medication you are taking. Whether you receive LY450139 or placebo (a pill that has a similar appearance to the study drug but has no medicine) will be determined by chance. The probability that you will receive LY450139 is 2 in 3. Sometime after Visit 14 (week 64), the placebo group will also be given LY450139. You will have a number of different examinations and tests during the study at different visits. You will be asked questions about what other medications you are taking and how well you are taking your study drug. Safety measures such as blood pressure, pulse, and electrocardiograms will also be taken. On some visits, you will have blood and urine samples taken. You will also be asked to answer questionnaires at some study visits. The study visits may take several hours up to one day, depending on the evaluations made.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 2 of 16

A blood sample will be taken at 13 visits during the study. The doctor or nurse will take your blood (approximately 2 Tbs.) for testing by placing a needle inside a vein in your arm. This will cause a small hole or ‘puncture’. The blood samples will be collected to perform the following tests: chemistry (to make sure the contents in your blood are within normal range), hematology (to make sure the levels of the different kinds of blood cells in your blood are normal), drug monitoring tests (if you are taking certain medications certain blood tests will be performed to look for any changes in how it may be working). A urine sample will be taken at 4 visits during the study. You will be asked to urinate or “pee” into a small cup. A urinalysis (test of the urine) will be done to make sure the contents of your urine are within normal range. APoE genotyping will be done in this study. A blood sample will be taken one time for this test. This test looks for the presence of certain genes (things that determine physical characteristics) that may related to Alzheimer’s disease. You and your doctor will not receive these results. Blood and urine samples collected for specified laboratory tests will be destroyed within 60 days of confirmation of the test results, unless laws, regulations, or international laboratory certification standards require a longer retention period. This confirmation will either occur immediately after initial testing, or may require that samples be held to be retested at a defined later point in time. Certain samples (plasma) are being collected so that the sponsor can measure the amount of drug that is in your body. The sponsor will keep these samples with the patient identifier for a maximum of two years following last patient visit for the study. The remaining plasma from these samples will either be discarded or pooled. The pooled samples will be kept for possible future research related to Alzheimer's disease for a maximum of eight years following last patient visit for the study. At the end of the eight years, any remaining pooled samples will be discarded. Storage of identifiable samples of plasma for possible future research related to Alzheimer’s disease is a mandatory part of this study. The amount of blood not used for Aβ (a protein linked to Alzheimer’s disease) testing will be stored. Stored samples will keep the patient identification number and, therefore, will not be stored indefinitely. Samples will be stored for a maximum of 8 years after last patient visit for the study; any sample remaining at that time will be destroyed. Patients may also choose whether or not to participate in additional studies that will evaluate disease progression (study addenda). This will require the banking of patient samples such as blood, serum, and plasma, the collection of cerebrospinal fluid (fluid surrounding the brain) and also includes imaging studies. Details concerning these additional studies are covered in separate informed consent documents, which you should review and discuss with your study partner/caregiver. Because of the exploratory nature of these analyses and because the results should not change medical management, patients will not receive the test results. Patients choosing not to participate in the addenda studies are not excluded from participation in the main study.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 3 of 16

After the study ends, you may be given the choice to receive LY450139 as part of an additional study. Please read the pages(s) called Study Procedures (Attachment 1). This will give you detailed information about how often you have to come to see the doctor/nurse, how long each visit may take how much blood will be taken and when tests and procedures will be performed on each study visit.

Risks There may be risks to you if you participate in this study. LY450139 is new and has been taken only by a limited number of people. As of 02 August 2007, approximately 237 people have received LY450139. Ten or more people who have taken LY450139 had headache, diarrhea (or loose stools), rash, head cold, dizziness or problems with balance, nausea, tiredness, and muscle pain. Five or more people had sleepiness, constipation (hard stools), vomiting, abdominal (stomach) pain or discomfort, stomach or intestinal gas, cough, sore throat, confusion, and itching. One person was hospitalized for blockage of the intestine. Four people reported signs of blood in the stool or black stool, which may suggest injury to the stomach or intestines. Temporary changes in electrocardiograms (heart-trace readings) have been seen in people given daily doses of LY450139. Changes in heart-trace readings may indicate a change in heart function that could be serious or life-threatening. Three people, taking multiple doses of 200 mg or higher of LY450139, reported visual disturbances. The people described the visual disturbances as seeing brief flashes of light or like seeing foam-like bubbles appearing on the outside edges of their vision in both eyes. These events went away within 2 days to 51 days of stopping the study drug without any permanent damage. An unusual event that was seen in animals and also reported by a few people in studies involving dosing over several weeks was a lightening of hair color. This went away after people and animals stopped taking the drug. Information on bad effects in animals can indicate possible risks to people. Because of limited exposure of people to date to LY450139, information on bad effects in animals may be useful in your decision whether to participate in this study. In animal studies, LY450139 doses similar to those currently being tested in people affected the digestive system. These animals developed soft stools that sometimes contained blood. Other effects of these doses of LY450139 seen in animals include shrinking of some organs (including the spleen, prostate, adrenal glands, and saliva-producing organs), when examined with a microscope, and an irritation of the liver (increase in liver enzymes). These drug effects disappeared when LY450139 was stopped. Reduced immunity (greater possibility of infection) may be possible based on certain effects seen in animals. Based on laboratory studies with similar drugs, it is possible that LY450139 and other compounds like it could cause cancer. Cancer has not been observed with LY450139 in any of the animal or human studies performed to date.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 4 of 16

Based on dog studies, normal function of the ovaries may be damaged and could cause early menopause or infertility.

Study Procedure Related Risks Electrocardiograms (ECGs) (electrical tracings of the heartbeat or heart rhythm) will be done during this study in which you will have pads placed on different parts of your body. There is no pain or discomfort during an ECG; however, removing the pads may cause some irritation to your skin. For most people, needle punctures for blood draws do not cause any serious problems. However, they may cause bleeding, bruising, discomfort, infections, dizziness, and/or pain at the needle site. In addition to the risks already described, LY450139 and the study procedures may have other unknown risks. If you choose to participate in the additional studies there may be additional risk. Details describing risk associated with study addenda are covered in separate informed consent documents which you should review and discuss with your study partner/caregiver. At any time during this study, you may experience or worsening of your Alzheimer’s disease. It may be more likely that you will experience such worsening of your symptoms during the early part of your participation in the study or if you receive placebo (a pill that has similar appearance to the study drug but has no medicine) as your study drug. If you are using other approved treatments for Alzheimer’s disease before entering the trial, you may be able to continue using them during the study. The study doctor will discuss this with you. There may be unknown risks of possible harmful interaction with other medication you may be taking. You should follow the study doctor’s directions carefully for taking this study drug. You should swallow the study tablets whole; do not crush the tablets. You should not give the study drug to other people and should keep it out of the reach of small children.

Reporting Health Experiences If you have any injury, bad effect, or any other unusual health experience during this study, make sure that you immediately tell the nurses or Dr. [study physician’s name], at [phone #]. You can call at any time, day or night, to report such health experiences.

Other Treatments You do not have to take part in this study to be treated for your illness or condition. Other treatments and therapies for your condition are available. Those might include cholinesterase inhibitors such as donepezil (Aricept) and an N-methyl-D-aspartate receptor antagonist, memantine (Namenda). The study doctor can discuss these treatments and therapies with you.

Participation in the Study If any important new information is found during this study that may affect your wanting to continue to be part of this study, you will be told about it right away.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 5 of 16

Your taking part in this study is entirely voluntary. You may refuse to take part in the study or you may stop your participation in the study at any time, without a penalty or loss of benefits to which you are otherwise entitled. Your participation may also be stopped by the study doctor or sponsor without your consent. If this happens, it might be due to a bad reaction you have had to LY450139 or new information about LY450139’s safety or effectiveness. If you stop being part of this study, the study doctor or one of the staff members will talk to you about any medical issues regarding the stopping of your participation.

Treatment and Compensation for Injury If you follow the directions of the study doctor and staff and you are physically injured due to any substance or procedure properly given under the plan for this study, the sponsor will pay the medical expenses for the treatment of that injury that are not covered by your medical insurance, by a government program, or by any other third party.

Possible Benefits Although LY450139 is being tested as a possible treatment for a condition that you may have, you may not receive any medical benefit. Study drug and study procedures will be provided at no cost to you. You and your study partner/caregiver will be paid [insert amount per study visit] to reimburse you for transportation, parking, meal, or others expenses related to your participation in this study. If you withdraw from the study early, you will be paid for these expenses for the portion of the study that you did complete. Information obtained from this study will benefit the sponsor of the study, Eli Lilly and Company, and may benefit subjects/patients in the future. You may receive information about your health from any physical examinations and laboratory tests to be done in this study.

Investigator Payment The sponsor is paying the study doctor and/or [name of Institution or site] for their work in this study.

Questions If you have any questions about this study or your rights, please contact Dr. [study physician’s name] at [address and phone #]. If you have any questions about your rights as a participant in a research study, please contact [ERB contact or other neutral or disinterested party] at [address and phone number].

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 6 of 16

Confidentiality The study doctor and staff will handle your personal health information in a confidential manner. Your health information will be used and disclosed in accordance with the following U.S. Data Privacy Statement.

U.S. Data Privacy Statement A federal government rule has been issued to protect the privacy rights of subjects/patients. This rule was issued under a law called the Health Insurance Portability and Accountability Act of 1996 (HIPAA). This rule is designed to protect the confidentiality of your personal health information. Your personal health information is information about you that could be used to find out who you are. For this research study, this includes information in your existing medical records needed for this study and new information created or collected during the study. This Data Privacy Statement explains how your personal health information will be used and whom it will be given to (“disclosed”) for this research study. It also describes your privacy rights, including your right to see your personal health information. By signing the consent document for this study, you will give permission (“authorization”) for the uses and disclosures of your personal health information that are described in this Data Privacy Statement. If you do not want to allow these uses, you should not participate in this study. If you agree to participate in the research study, your personal health information will be used and disclosed in the following ways: The study doctor and staff will use your medical records and information created or collected during the study to conduct the study. The study doctor and staff will send your study-related health information (“study data”) to the sponsor of the study and its representatives (“sponsor”). Because the sponsor conducts business related to clinical research in many countries around the world, this may involve sending your study data outside of the United States. Other countries may have privacy laws that do not provide the same protections as the laws in this country. However, the sponsor will respect the terms of this Data Privacy Statement in all countries. The study data sent by the study doctor to the sponsor does not include your name, address, social security number, or other information that directly identifies you. Instead, the study doctor assigns a code number to the study data and may use your initials. Some study data sent to the sponsor may contain information that could be used (perhaps in combination with other information) to identify you (e.g., date of birth). If you have questions about the specific health information that will be sent to the sponsor, you should ask the study doctor. The sponsor will use the study data for research purposes to support the scientific objectives of the study described in the consent document, to assess the safety or efficacy of any drug or treatment included in the study, to better understand the disease(s) included in the study, or to improve the design of future studies. LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 7 of 16

Your study data, either alone or combined with data from other studies, may be shared with regulatory authorities in the United States and other countries, doctors at other institutions participating in the study, and the ethical review board overseeing this study. Study data that does not directly identify you may be published in medical journals or shared with others as part of scientific discussions. Your original medical records, which may contain information that directly identifies you, may be reviewed by the sponsor, the ethical review board overseeing this study, and regulatory authorities in the United States and other countries. The purpose of these reviews is to assure the quality of the study conduct and the study data, or for other uses authorized by law. The sponsor works with business partners in drug development. The sponsor may share your study data with these business partners, but only if the business partners need the information as a part of this work with the sponsor, and only if the business partners signs a contract that requires it to protect your study data in the same way as the sponsor. The sponsor will not disclose personal health information to insurance companies unless required to do so by law, or unless you provide separate written consent to do so. Your medical records and study data may be held and processed on computers. Your personal health information may no longer be protected under the HIPAA privacy rule once it is disclosed by your study doctor to these other parties. You have the right to see and copy your personal health information related to the research study for as long as this information is held by the study doctor or research institution. However, to ensure the scientific integrity of the study, you will not be able to review some of the study information until after the study has been completed. You may cancel your authorization at any time by providing written notice to the study doctor. If you cancel your authorization, the study doctor and staff will no longer use or disclose your personal health information in connection with this study, unless the study doctor or staff needs to use or disclose some of your personal health information to preserve the scientific integrity of the study. The sponsor will still use study data that was collected before you cancelled your authorization. If you cancel your authorization, you will no longer be able to participate in the study. However, if you decide to cancel your authorization and withdraw from the study, you will not be penalized or lose any benefits to which you are otherwise entitled. Your authorization for the uses and disclosures described in this Data Privacy Statement does not have an expiration date.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 8 of 16

Signatures To be entered into the study, you or a legal representative must sign and date the signature page (Attachment 2). •

Name and signature of participant or legally authorized representative.



Name and signature of witness to participant signature.



Name and signature of person conducting consent discussion.



Date of signature for each individual.

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 9 of 16

Patient Information and Consent Form Attachment 1 Signature Page To become a part of this study, and to authorize use and disclosure of your personal health information, you or your legal representative must sign and date this page. By signing this page, you are confirming the following: •

You have read all of the information in this Patient Information and Consent Form, and you have had time to think about it.



All of your questions have been answered to your satisfaction.



You voluntarily agree to be part of this research study, to follow the study procedures, and to provide necessary information to the study doctor, nurses, or other staff members, as requested.



You may freely choose to stop being a part of this study at any time.



You allow the study doctor and the sponsor to use and disclose your personal health information as described in this document.



You have received a copy of this signed Patient Information and Consent Form to keep for yourself.

Signatures (page 1 of 3) FOR SUBJECT/PATIENT TO COMPLETE

Signature of Patient

Date

Patient Name (print or type)

Patient Number

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 10 of 16

Signatures (page 2 of 3) FOR LEGAL REPRESENTATIVE TO COMPLETE

Signature of Legal Representative

Date

Legal Representative Name (print or type)

If signed by legal representative, state description of relationship to subject/patient or other basis for legal authority

Name of individual conducting informed consent discussion (print or type)

Signature of individual conducting informed consent discussion

Date

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 11 of 16

Signatures (page 3 of 3)

Signatures for Patient Assent/Impartial Witness FOR ADDITIONAL SIGNATURES

Signature of Impartial Witness

Date

Impartial Witness Name (print or type)

Patient Number

If the patient does not have the capacity to give informed consent to participate in this research study, the legal representative must read and sign the following affidavit:

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 12 of 16

By signing below, I ___________________________ certify that I have read or had this (Print legal representative’s name)

document read to me. I will be given a signed copy. I have been given the chance to ask questions and to discuss the patient’s participation with the investigator. As legal representative for this patient, I hereby give consent for this patient to participate in this study. I conclude that the patient wishes to participate in this research and has indicated his/her assent to do so. (I do/do not have durable power of attorney for health care for the patient. I am/am not the patient’s legal guardian/representative.)

___________________________________________ Name of Legal Representative (Print) ___________________________________________

______________________________

Signature of Legal Representative

Date (Legal Representative must personally date)

___________________________________________

_______________________________

Signature of Subject

Date

___________________________________________

_______________________________

Signature of Investigator

Date

___________________________________________

_______________________________

Signature of Witness

Date

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 13 of 16

Study Partner/Caregiver Information As the patient’s informant, I have important responsibilities that need to be followed in order for the study to be conducted in the safest and best manner possible. These responsibilities are as follows:

1. I must accompany the patient to all clinic visits and must have direct contact with the patient for a minimum of three days per week (for at least 4 hours per day during waking hours). 2. I will receive a schedule for clinic appointments from the study doctor. I must try to help the patient keep all scheduled appointments. 3. I will aid in the storage and administration of the study drug. I will try and ensure that the patient takes each dose of medication. I will make sure the study drug is stored safely and all supplies are returned, including used and unused medication bottles. 4. I am a very valuable source of information about the patient. The study doctor will interview me during the clinic visits. My cooperation in answering these questions is necessary.

Study Partner/Caregiver Consent I, ______________________, have read and understand all of the preceding information that describes both the patient’s participation in the study and my responsibilities as the patient’s informant. I voluntarily consent to participate in the study. (I do/do not have durable power of attorney for health care for the patient. I am/am not the patient’s legal guardian.)

____________________________________

_____________________________

Signature of caregiver/informant

Date

____________________________________ Relationship to patient _____________________________________

______________________________

Signature of Witness

Date

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 14 of 16

Patient Information and Consent Form Attachment 2 Study Procedures H6L-MC-LFAN Screening Visit

This visit will determine if you can participate in the study.

Approximate Visit Length

This visit may take from 3 to 14 days, but could take up to 30 days to have all the tests completed and the resu

Procedures To Be Done

-Answer 2 questionnaires to evaluate memory and mood -Neurological examination -Physical examination -Electrocardiogram (performed 3 times) -Urine sample collected -Blood sample (2 to 5 tablespoons) taken

Study Visit 2

At this study visit you will be entered into the study and randomized to a treatment group. You will be admin tests, then given study drug, then administered the tests again. You will also be given enough study drug to la study visit. You will take 3 capsules of the study drug on your own each morning

Approximate Visit Length

Needs to be estimated

Procedures To Be Done

-Answer questionnaires to evaluated memory and mood -Electrocardiograms (performed 6 times) -Blood pressure and pulse measured -Blood sample (2 to 5 tablespoons) taken -Study partner/caregiver will answer 1 questionnaire

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 15 of 16

Study Visits:

These visits will evaluate you general health and well being. You will be given enough study drug to last unti

3, 4, 5, 6, 7, 8 10, 15, 16, 18, 19 Approximate Visit Length

Needs to be estimated

Procedures To Be Done

-Answer questions about your general health and how well you are taking your study drug -Blood sample (2 to 5 tablespoons) taken (only on visits 3, 7, 15, 16, 18, 19) -Blood pressure and pulse measured -Electrocardiograms (performed 3 times)

Study Visits:

These visits will evaluate you general health, wellbeing, and progression of Alzheimer’s disease. You will be study drug to last until your next visit.

9, 11, 12, 13, 14, 17, 20 Approximate Visit Length

Needs to be estimated

Procedures To Be Done

-Answer questions about your general health and how well you are taking your study drug -Answer questionnaires to evaluate memory and mood -Physical/neurological exam (visits 13, 17, 20 only) -Blood sample (2 to 5 tablespoons) taken -Urine sample taken (visits 13, 17, and 20 only) -Blood pressure and pulse measured -Electrocardiograms (performed 3 times)

LY450139 H6L-MC-LFAN USA 16-November 2007 Eli Lilly and Company © 2007

Confidential

Page 16 of 16

Patient Information and Consent Form Effect of γ-Secretase Inhibition on the Progression of Alzheimer’s Disease: LY450139 versus Placebo Introduction You are invited to take part voluntarily in a research study of an investigational study drug known as LY450139. The study protocol has been amended due to the recent stopping of dosing. Your participation in this amended study is expected to last up to 32 weeks from the time of the dose stop. Up to 1500 patients will be participating in this study. Eli Lilly and Company is sponsoring this study. Before agreeing to participate in this research study, it is important that you read and understand this form. It describes the purpose, procedures, benefits, risks and discomforts, and precautions of the study. It also describes the alternative procedures that are available to you and your right to withdraw from the study at any time. If you participate, you will receive a copy of this form to keep for your records. We will ask you to come to all study visits with a study partner or caregiver. Your study partner or caregiver is someone who knows you well and will validate information we gather at your visits. It is important that you and your study partner/caregiver understand the description of the research study before you agree to participate. In this consent form, “you” always refers to the patient. If you are a legally acceptable representative or study partner, please remember that “you” refers to the study patient.

Purpose of the Study The primary purpose of your participation in this study is to help answer the following research question(s), and not to provide you treatment for your condition: •

The safety of LY450139 and any side effects that might be associated with it.



To determine if the greater cognitive and functional worsening seen in LY450139-treated patients versus placebo-treated patients in the interim and futility analyses persists after LY450139 is discontinued.



To monitor effects on concomitant medications following LY450139 cessation and to allow for concomitant medication adjustment, given the potential for LY450139 cessation to affect concomitant medication exposure.



To monitor other safety measures following cessation of LY450139/placebo treatment through assessment of adverse events (AEs), routine

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 1 of 18

physical/neurologic/skin examinations, and changes in vital signs, laboratory evaluations, and electrocardiograms (ECGs).

Qualifications to Participate The doctor in charge of this study or a member of the study staff has discussed with you the requirements for participation in this study. You cannot participate in this study if: •

you have a history of HIV;



you have a history within the last 5 years of alcohol or drug abuse;



you have a history within the last 5 years of a serious infection affecting the brain;



you have multiple or severe drug allergies;



you have serious or uncontrolled illnesses including those that affect your heart, kidneys, lungs, blood, liver, brain, nerves, ability to fight off infection, bowels stomach, mental health, abnormal blood/urine tests or other condition that the investigator feels could affect your participation in the trial;



you cannot swallow a whole tablet;



you are taking certain medications or may need to use certain medications during the study that are not allowed in this trial;



you have uncontrolled high blood pressure.

If female, in order to participate in this study you must be post-menopausal, as evidenced by a lack of menstruation for at least 12 consecutive months or have had both ovaries removed. It is important that you are completely truthful with the study doctor and staff about your past medical history as well as any symptoms experienced during the study. It may be harmful to you or to other people who may take the drug if you are not truthful with the study doctor and staff. You should not participate in this study if you do not meet all qualifications.

Study Procedures If you decide to take part in this study, you will need to visit the hospital or study doctor’s office at most 4 times over the 32 weeks from the dose stop. The first visit will be approximately 4 weeks after the dose stop. The second visit will be 8 weeks after the dose stop. The third visit will be 16 weeks after the dose stop. The fourth & final visit will be 32 weeks after the dose stop. Your study doctor or nurse will make sure you know when you need to attend. You will have a number of different examinations and tests during the study at different visits. You will be asked questions about what other medications you are taking and how LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 2 of 18

well you are feeling. Safety measures such as blood pressure, pulse, and electrocardiograms will also be taken. At all visits, you will have blood and urine samples taken. You will have a total-body skin examination at all visits. You will also be asked to answer questionnaires at study visits. The study visits will last up to 6 hours, depending on the evaluations made. A blood sample will be taken at 4 visits during the study. The doctor or nurse will take your blood (approximately 2 tablespoons) for testing by placing a needle inside a vein in your arm. This will cause a small hole or “puncture”. The blood samples will be collected to perform the following tests: chemistry (to make sure the contents in your blood are within normal range), hematology (to make sure the levels of the different kinds of blood cells in your blood are normal) and drug monitoring tests (if you are taking certain medications certain blood tests will be performed to look for any changes in how it may be working). A urine sample will be taken at 4 visits during the study. You will be asked to urinate or “pee” into a small cup. A urinalysis (test of the urine) will be done to make sure the contents of your urine are within normal range. Blood and urine samples collected for specified laboratory tests will be destroyed within 60 days of confirmation of the test results, unless laws, regulations, or international laboratory certification standards require a longer retention period. This confirmation will either occur immediately after initial testing, or may require that samples be held to be retested at a defined later point in time. Certain samples (plasma) are being collected so that the sponsor can measure the amount of Aβ (a protein linked to Alzheimer’s disease) that is in your body and/or protein binding. The amount of blood not used for these tests will be stored. The sponsor will keep these samples with the patient identifier for a maximum of 2 years following last patient visit for the study. At the end of the 2 years, any remaining samples will be discarded. Some samples of plasma will be stored for possible future research related to Alzheimer’s disease. Storage of these identifiable samples is a mandatory part of this study. These stored samples will keep the patient identification number and, therefore, will not be stored indefinitely. These samples will be stored for a maximum of 8 years after last patient visit for the study; any sample remaining at that time will be destroyed. Please read the pages(s) called Study Procedures (Attachment 1). This will give you detailed information about how often you have to come to see the doctor/nurse, how long each visit may take how much blood will be taken and when tests and procedures will be performed on each study visit.

Risks There may be risks to you if you participate in this study. LY450139 is new and has been taken only by a limited number of people. As of 07 September 2010, 367 people (297 healthy subjects and 70 people with Alzheimer’s disease [AD]) have taken semagacestat (LY450139) in completed studies. LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 3 of 18

Risks and Discomforts Associated with Semagacestat An evaluation of ongoing study data showed that while patients taking placebo (sugar pill) showed worsening in memory and ability to perform routine activities at about the rate expected for Alzheimer’s patients, memory and function worsened more in patients taking semagacestat. Therefore, Lilly has decided to permanently stop study drug dosing in all semagacestat studies. For safety reasons, Lilly will continue to do follow-up testing for 7 months after stopping semagacestat. Because the studies are ongoing, the data are still blinded to you, your investigator, and the Lilly study team. The following risks and discomforts have been noted in people taking semagacestat in other studies but most of these risks should diminish once you stop semagacestat. A total of 70 people with AD have taken semagacestat in 2 completed clinical studies, and 297 healthy volunteers have taken semagacestat in 14 completed studies. The most common side effects are listed below. Of the 70 people with AD: 10 to 14 people reported: 5 to 9 people reported:

• • • • • • • • • • • •

Diarrhea (unusually loose or frequent stools) Weakness. Nausea (feeling sick to the stomach) Headache Vomiting Cough Rash Dizziness Sleepiness Confusion Back pain An infection of breathing system such as a cold or bronchitis

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 4 of 18

Of 108 healthy volunteers who took more than one dose of semagacestat: 28 people reported: • Headache 10 to 16 people reported: • Rash • Itchiness • Diarrhea 5 to 9 people reported: • Nausea • Stomach pain • Constipation (unusually hard or infrequent stools) • Stomach or intestinal gas • Sore throat • Visual disturbances (such as blurred vision or seeing brief flashes of light) • Dizziness • Tiredness • Back pain • Indigestion • Muscle pain • Vomiting • Throat irritation

Of 169 healthy volunteers who took just 1 dose of semagacestat: 32 people reported • Headache 11 people reported: • Back pain 5 to 9 people reported: • Pain in extremity • Diarrhea • Nausea • Dizziness • Neck pain • Muscle spasms

One person with Alzheimer’s disease was hospitalized for blockage of the intestine. Five people (3 with Alzheimer’s disease, 2 without) reported signs of blood in the stool or black stool, which may suggest injury to the stomach or intestines. In animals and in a few people, hair color lightened. This unusual event happened after taking the drug for several weeks. This went away after people and animals stopped taking the drug.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 5 of 18

Temporary changes in electrocardiograms (heart-trace readings) were seen in people taking daily doses of semagacestat. Changes in heart-trace readings may indicate a change in heart function that could be serious or life-threatening. Skin cancer has been seen in some people with Alzheimer’s disease taking daily doses of semagacestat or placebo (sugar pill) in 2 ongoing studies. Since these studies are still ongoing, details about the rate of skin cancer with semagacestat are not yet available. Your skin will be checked for changes while you are participating in studies with semagacestat. Semagacestat may change the effectiveness of some medicines by raising or lowering the amount in your blood. If you take one of these medicines, levels of these medicines in your blood and/or their effectiveness will be monitored in studies with semagacestat to be sure you are getting the right amount of medicine. Some people taking medicines that lower a brain protein thought to cause Alzheimer’s disease have experienced an increase in the water content of the brain tissue. Semagacestat works by lowering this brain protein, but no cases of an increase in the water content of brain tissue have been reported in patients taking semagacestat. Symptoms that can happen if someone develops this condition include headache, worsened confusion, change in level of consciousness, seizures, unsteadiness, and vomiting. If you develop a combination of these symptoms at the same time, you will be asked to get a scan of your brain. This would show whether the condition was present. Bad effects in animals can indicate possible risks to humans. A higher rate of cancer of the ovaries and cancer of the uterus was seen in old animals that were given semagacestat every day from the time they were young compared with animals not given semagacestat. The effect on ovaries was likely related to the effect of semagacestat on young, active ovaries. The effect on the uterus might also be related to effects on ovaries and changes in hormones that might occur because of that. While the relevance to humans is not completely understood, human females taking semagacestat must be post-menopausal (that is, have had no menstrual periods for at least 12 months in a row or have had both ovaries removed), so their ovaries will no longer be active when they start taking semagacestat. Including only post-menopausal women is likely to reduce the risk. Ovarian or uterine cancer has not been reported with semagacestat in any of the human studies done so far. Some animals given semagacestat had problems with the digestive system. For example, some had soft stools or soft stools with blood in them. Also, some animals had shrinking of some organs (including the spleen, prostate, adrenal glands, tear glands, uterus, and saliva-producing organs) and irritation of the liver. These drug effects disappeared when semagacestat was stopped. Based on certain effects seen in animals, infection may be more likely.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 6 of 18

Effects on Male Fertility A study of rats given semagacestat every day found that semagacestat may decrease male fertility. No change in the rat testicles was seen. The researchers think that the decreased fertility might be due to changes in semen. The effect on fertility in male rats went away after dosing was stopped. The effect of semagacestat on human male fertility is unknown.

Risks in Premenopausal Women Semagacestat may affect reproduction. Studies in female dogs sometimes showed problems with ovary functioning. This could cause early menopause or infertility. Studies in pregnant rats and rabbits sometimes resulted in loss of the unborn babies, lowweight babies, or babies that did not grow right. Women in the study must be postmenopausal, so these experiences will not affect women in the study. Women who give semagacestat to patients are protected from the drug by a coating over the drug tablet. To keep safe, do not break the tablets into pieces or crush them. Broken tablets should be thrown away or returned to the study site in a sealed bag. If you touch a broken tablet, wash your hands right away to remove any drug material on your skin. Use a wet disposable towel to clean up dry, crushed tablets. If possible, wear disposable gloves when cleaning up spilled material.

Study Procedure Related Risks Electrocardiograms (ECGs) (electrical tracings of the heartbeat or heart rhythm) will be done during this study in which you will have pads placed on different parts of your body. There is no pain or discomfort during an ECG; however, removing the pads may cause some irritation to your skin. For most people, needle punctures for blood draws do not cause any serious problems. However, they may cause bleeding, bruising, discomfort, infections, dizziness, and/or pain at the needle site. In addition to the risks already described, the study procedures may have other unknown risks. At any time during this study, you may experience or worsening of your Alzheimer’s disease. If you are using other approved treatments for Alzheimer’s disease before entering the trial, you may be able to continue using them during the study. The study doctor will discuss this with you. There may be unknown risks of possible harmful interaction with other medication you may be taking.

Reporting Health Experiences If you have any injury, bad effect, or any other unusual health experience during this study, make sure that you immediately tell the nurses or Dr. [study physician’s name], at [phone #]. You can call at any time, day or night, to report such health experiences. LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 7 of 18

Other Treatments You do not have to take part in this study to be treated for your illness or condition. Other treatments and therapies for your condition are available. Those might include cholinesterase inhibitors such as donepezil (Aricept) and an N-methyl-D-aspartate receptor antagonist, memantine (Namenda). The study doctor can discuss these treatments and therapies with you.

Participation in the Study If any important new information is found during this study that may affect your wanting to continue to be part of this study, you will be told about it right away. Your taking part in this study is entirely voluntary. You may refuse to take part in the study or you may stop your participation in the study at any time, without a penalty or loss of benefits to which you are otherwise entitled. Your participation may also be stopped by the study doctor or sponsor without your consent. If you stop being part of this study, the study doctor or one of the staff members will talk to you about any medical issues regarding the stopping of your participation.

Treatment and Compensation for Injury If you follow the directions of the study doctor and staff and you are physically injured due to any substance or procedure properly given under the plan for this study, the sponsor will pay the medical expenses for the treatment of that injury that are not covered by your medical insurance, by a government program, or by any other third party.

Possible Benefits Study procedures will be provided at no cost to you. You and your study partner/caregiver will be paid [insert amount per study visit] to reimburse you for transportation, parking, meal, or others expenses related to your participation in this study. If you withdraw from the study early, you will be paid for these expenses for the portion of the study that you did complete. Information obtained from this study will benefit the sponsor of the study, Eli Lilly and Company, and may benefit subjects/patients in the future. You may receive information about your health from any physical examinations and laboratory tests to be done in this study.

Investigator Payment The sponsor is paying the study doctor and/or [name of Institution or site] for their work in this study.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 8 of 18

Questions If you have any questions about this study or your rights, please contact Dr. [study physician’s name] at [address and phone #]. If you have any questions about your rights as a participant in a research study, please contact [ERB contact or other neutral or disinterested party] at [address and phone number].

Confidentiality The study doctor and staff will handle your personal health information in a confidential manner. Your health information will be used and disclosed in accordance with the following U.S. Data Privacy Statement.

U.S. Data Privacy Statement A federal government rule has been issued to protect the privacy rights of subjects/patients. This rule was issued under a law called the Health Insurance Portability and Accountability Act of 1996 (HIPAA). This rule is designed to protect the confidentiality of your personal health information. Your personal health information is information about you that could be used to find out who you are. For this research study, this includes information in your existing medical records needed for this study and new information created or collected during the study. This Data Privacy Statement explains how your personal health information will be used and whom it will be given to (“disclosed”) for this research study. It also describes your privacy rights, including your right to see your personal health information. By signing the consent document for this study, you will give permission (“authorization”) for the uses and disclosures of your personal health information that are described in this Data Privacy Statement. If you do not want to allow these uses, you should not participate in this study. If you agree to participate in the research study, your personal health information will be used and disclosed in the following ways: The study doctor and staff will use your medical records and information created or collected during the study to conduct the study. The study doctor and staff will send your study-related health information (“study data”) to the sponsor of the study and its representatives (“sponsor”). Because the sponsor conducts business related to clinical research in many countries around the world, this may involve sending your study data outside of the United States. Other countries may have privacy laws that do not provide the same protections as the laws in this country.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 9 of 18

However, the sponsor will respect the terms of this Data Privacy Statement in all countries. The study data sent by the study doctor to the sponsor does not include your name, address, social security number, or other information that directly identifies you. Instead, the study doctor assigns a code number to the study data and may use your initials. Some study data sent to the sponsor may contain information that could be used (perhaps in combination with other information) to identify you (e.g., date of birth). If you have questions about the specific health information that will be sent to the sponsor, you should ask the study doctor. The sponsor will use the study data for research purposes to support the scientific objectives of the study described in the consent document, to assess the safety or efficacy of any drug or treatment included in the study, to better understand the disease(s) included in the study, or to improve the design of future studies. Your study data, either alone or combined with data from other studies, may be shared with regulatory authorities in the United States and other countries, doctors at other institutions participating in the study, and the ethical review board overseeing this study. Study data that does not directly identify you may be published in medical journals or shared with others as part of scientific discussions. Your original medical records, which may contain information that directly identifies you, may be reviewed by the sponsor, the ethical review board overseeing this study, and regulatory authorities in the United States and other countries. The purpose of these reviews is to assure the quality of the study conduct and the study data, or for other uses authorized by law. The sponsor works with business partners in drug development. The sponsor may share your study data with these business partners, but only if the business partners need the information as a part of this work with the sponsor, and only if the business partners signs a contract that requires it to protect your study data in the same way as the sponsor. The sponsor will not disclose personal health information to insurance companies unless required to do so by law, or unless you provide separate written consent to do so. Your medical records and study data may be held and processed on computers. Your personal health information may no longer be protected under the HIPAA privacy rule once it is disclosed by your study doctor to these other parties. You have the right to see and copy your personal health information related to the research study for as long as this information is held by the study doctor or research institution. However, to ensure the scientific integrity of the study, you will not be able to review some of the study information until after the study has been completed.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 10 of 18

You may cancel your authorization at any time by providing written notice to the study doctor. If you cancel your authorization, the study doctor and staff will no longer use or disclose your personal health information in connection with this study, unless the study doctor or staff needs to use or disclose some of your personal health information to preserve the scientific integrity of the study. The sponsor will still use study data that was collected before you cancelled your authorization. If you cancel your authorization, you will no longer be able to participate in the study. However, if you decide to cancel your authorization and withdraw from the study, you will not be penalized or lose any benefits to which you are otherwise entitled. Your authorization for the uses and disclosures described in this Data Privacy Statement does not have an expiration date.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 11 of 18

Patient Information and Consent Form Attachment 1

Signature Pages To be entered into the study, you or a legal representative in addition to your caregiver must sign and date the signature pages which follow. The following are required •

Name and signature of participant OR legally authorized representative.



Name and signature of witness to participant signature. (If applicable / required by IRB/EC. This is not a Lilly requirement.)



Name and signature of caregiver



Name and signature of person conducting consent discussion.



Date of signature for each individual (personally completed)

By signing below, you are confirming the following: •

You have read all of the information in this Patient Information and Consent Form, and you have had time to think about it.



All of your questions have been answered to your satisfaction.



You voluntarily agree to be part of this research study, to follow the study procedures, and to provide necessary information to the study doctor, nurses, or other staff members, as requested.



You may freely choose to stop being a part of this study at any time.



You allow the study doctor and the sponsor to use and disclose your personal health information as described in this document.



You have received a copy of this signed Patient Information and Consent Form to keep for yourself.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 12 of 18

FOR SUBJECT/PATIENT TO COMPLETE

Signature of Patient

Date

Patient Name (print)

Patient Number

FOR LEGAL REPRESENTATIVE TO COMPLETE

Signature of Legal Representative

Date

Legal Representative Name (print)

If signed by legal representative, state description of relationship to subject/patient or other basis for legal authority

FOR INDIVIDUAL CONDUCTING INFORMED CONSENT

Signature of individual conducting informed consent discussion

Date

Name of individual conducting informed consent discussion (print) LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 13 of 18

SIGNATURE OF IMPARTIAL WITNESS (IF APPLICABLE)_

Signature of Impartial Witness

Date

Impartial Witness Name (print)

Patient Number

(If page not applicable: line through, date & initial)

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 14 of 18

If the patient does not have the capacity to give informed consent to participate in this research study, the legal representative must read and sign the following affidavit: By signing below, I ___________________________ certify that I have read or had this (Print legal representative’s name) document read to me. I will be given a signed copy. I have been given the chance to ask questions and to discuss the patient’s participation with the investigator. As legal representative for this patient, I hereby give consent for this patient to participate in this study. I conclude that the patient wishes to participate in this research and has indicated his/her assent to do so. (I do/do not have durable power of attorney for health care for the patient. I am/am not the patient’s legal guardian/representative.)

___________________________________________ Name of Legal Representative (Print)

___________________________________________

______________________________

Signature of Legal Representative

Date

___________________________________________

_______________________________

Signature of Subject

Date

___________________________________________

_______________________________

Signature of Investigator

Date

___________________________________________

_______________________________

Signature of Witness

Date

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 15 of 18

Caregiver Information As the patient’s caregiver, I have important responsibilities that need to be followed in order for the study to be conducted in the safest and best manner possible. These responsibilities are as follows: 1. I must accompany the patient to all clinic visits and must have direct contact with the patient for a minimum of three days per week (for at least 4 hours per day during waking hours). 2. I will receive a schedule for clinic appointments from the study doctor. I must try to help the patient keep all scheduled appointments. 3. I am a very valuable source of information about the patient. The study doctor will interview me during the clinic visits. My cooperation in answering these questions is necessary.

Caregiver Consent I, ______________________, have read and understand all of the preceding information that describes the patient’s participation in the study, my responsibilities as the patient’s informant, and possible risks from exposure to broken tablets. I voluntarily consent to participate in the study. (I do/do not have durable power of attorney for health care for the patient. I am/am not the patient’s legal guardian.) FOR CAREGIVER TO COMPLETE

Signature of Caregiver

Date

Caregiver Name (print) FOR INDIVIDUAL CONDUCTING CAREGIVER INFORMED CONSENT TO COMPLETE

Signature of individual conducting caregiver informed consent discussion

Date

Name of individual conducting caregiver informed consent discussion (print) LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 16 of 18

Patient Information and Consent Form Attachment 2

Study Procedures H6L-MC-LFAN(e)

Endpoint Visit

The Endpoint Visit is the completion of the protocol as originally designed.

Approximate Visit Length

Up to 6 hours

Procedures To Be Done

-Answer 7 questionnaires to evaluate memory and mood -Neurological examination -Physical examination, including body weight and vital signs (blood pressure, pulse, and temperature) -You will have a total-body skin examination. -Electrocardiogram (performed 3 times) -Urine sample collected -Blood sample (2 to 5 tablespoons) taken This visit will occur for patients who choose to discontinue the study at or before the Endpoint Visit.

Study Visit 801

Approximate Visit Length

Up to 1 hour

Procedures To Be Done

-Electrocardiogram -You will have a total-body skin examination. -Blood sample (2 to 5 tablespoons) taken -Answer questions about your general health These visits will evaluate your general health and well being. Up to 6 hours

Study Visits: 802, 803, and 804 Approximate Visit Length Procedures To Be Done

-Answer 2 questionnaires -Neurological examination -Physical examination, including body weight and vital signs (blood pressure, pulse, and temperature) -Answer questions about your general health -Blood sample (2 to 5 tablespoons) taken -Electrocardiogram -Urine sample collected -You will have a total-body skin examination.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 17 of 18

6-Month Post-Study Virtual Follow-up

This visit will be a telephone or certified mail follow-up only if you discontinue before or at the Endpoint Visit.

LY450139 H6L-MC-LFAN USA 23 Sep 2010 Eli Lilly and Company © 2007

Confidential

Page 18 of 18