Cartilage degradation biomarkers predict ... - Wiley Online Library

ARTHRITIS & RHEUMATISM Vol. 62, No. 10, October 2010, pp 3006–3015 DOI 10.1002/art.27596 © 2010, American College of Rheumatology

Cartilage Degradation Biomarkers Predict Efficacy of a Novel, Highly Selective Matrix Metalloproteinase 13 Inhibitor in a Dog Model of Osteoarthritis Confirmation by Multivariate Analysis That Modulation of Type II Collagen and Aggrecan Degradation Peptides Parallels Pathologic Changes Steven Settle,1 Lillian Vickery,1 Olga Nemirovskiy,1 Tom Vidmar,2 Alison Bendele,3 Dean Messing,1 Peter Ruminski,1 Mark Schnute,1 and Teresa Sunyer1 Objective. To demonstrate that the novel highly selective matrix metalloproteinase 13 (MMP-13) inhibitor PF152 reduces joint lesions in adult dogs with osteoarthritis (OA) and decreases biomarkers of cartilage degradation. Methods. The potency and selectivity of PF152 were evaluated in vitro using 16 MMPs, TACE, and ADAMTS-4 and ADAMTS-5, as well as ex vivo in human cartilage explants. In vivo effects were evaluated at 3 concentrations in mature beagles with partial medial meniscectomy. Gross and histologic changes in the femorotibial joints were evaluated using various measures of cartilage degeneration. Biomarkers of cartilage turnover were examined in serum, urine, or synovial fluid. Results were analyzed individually and in combination using multivariate analysis. Results. The potent and selective MMP-13 inhibitor PF152 decreased human cartilage degradation ex

vivo in a dose-dependent manner. PF152 treatment of dogs with OA reduced cartilage lesions and decreased biomarkers of type II collagen (type II collagen neoepitope) and aggrecan (peptides ending in ARGN or AGEG) degradation. The dose required for significant inhibition varied with the measure used, but multivariate analysis of 6 gross and histologic measures indicated that all doses differed significantly from vehicle but not from each other. Combined analysis of cartilage degradation markers showed similar results. Conclusion. This highly selective MMP-13 inhibitor exhibits chondroprotective effects in mature animals. Biomarkers of cartilage degradation, when evaluated in combination, parallel the joint structural changes induced by the MMP-13 inhibitor. These data support the potential therapeutic value of selective MMP-13 inhibitors and the use of a set of appropriate biomarkers to predict efficacy in OA clinical trials.

1 Steven Settle, DVM, Lillian Vickery, BS, Olga Nemirovskiy, PhD, Dean Messing, BS, Peter Ruminski, MS, Mark Schnute, PhD, Teresa Sunyer, PhD: Pfizer Global Research and Development, St. Louis, Missouri; 2Tom Vidmar, PhD: BioSTAT Consultants, Portage, Michigan; 3Alison Bendele, DVM, PhD: Bolder BioPATH, Boulder, Colorado. Dr. Settle and Ms Vickery contributed equally to this work. Drs. Settle, Nemirovskiy, Bendele, Schnute, and Sunyer, and Ms Vickery, Mr. Messing, and Mr. Ruminski own stock or stock options in Pfizer. Mr. Ruminski and Dr. Schnute are coinventors on a patent for PF152, a selective matrix metalloproteinase 13 inhibitor; the patent is assigned to Pfizer. Address correspondence and reprint requests to Teresa Sunyer, PhD, Pfizer Inc., 700 Chesterfield Parkway BB5B, St. Louis, MO 63017. E-mail: [email protected]. Submitted for publication January 16, 2010; accepted in revised form June 1, 2010.

Osteoarthritis (OA) is a chronic degenerative joint disease affecting primarily aged or injured joints. The disease is characterized by an imbalance between cartilage synthesis and degradation, with increased breakdown of matrix components leading to proteoglycan loss and cartilage fibrillation, eventually resulting in severe cartilage defects. These changes are irreversible, and the only treatment other than palliative symptom control is total joint replacement. Therefore, the discovery of a disease-modifying osteoarthritis drug (DMOAD) would fill a large unmet medical need. Inhibition of collagenase-mediated degradation of type II collagen, the most abundant protein in articular 3006

REDUCTION OF DOG OA LESIONS AND BIOMARKERS BY MMP-13 INHIBITOR

cartilage, would be expected to delay the progression of OA. Type II collagen is actively cleaved in OA cartilage by matrix metalloproteinase 13 (MMP-13; collagenase 3) resulting mainly in a 45–amino acid fragment known as type II collagen neoepitope (TIINE). Collagenase activity, as evaluated by TIINE levels in the urine and synovial fluid, is elevated in OA patients and in animal models of OA (1). Transgenic mice overexpressing MMP-13 in articular cartilage displayed cartilage changes characteristic of human OA (2), whereas MMP13–knockout mice demonstrated significantly reduced tibial cartilage erosions in a surgical model of OA (3). The discovery and development of MMP inhibitors for the treatment of OA has been pursued by a number of groups, but these efforts have been hindered by musculoskeletal syndrome, a dose-limiting side effect associated with broad-spectrum MMP inhibition (4). Recent reports describe a new class of highly selective MMP-13 inhibitors shown to be efficacious in both rat and rabbit models of OA (5–7) without the associated joint toxicity. Demonstrating the therapeutic value of potential OA drugs requires long and expensive clinical trials. The primary clinical outcome of disease progression is the radiographic assessment of joint space narrowing (JSN) in the affected joint. Due to the slow progression of this disease, OA clinical trials are ⬎2 years long and require a large number of patients. Therefore, the identification and validation of biomarkers that predict the efficacy of potential OA drugs faster and/or with fewer patients is important for the development of DMOADs. Many biomarkers reflecting degradation of various tissues in OA joints have been described, but their predictive value as surrogates for JSN is still under investigation (8–15). In this report, we describe a novel potent and highly selective MMP-13 inhibitor (PF152) and its effect on lesion progression in a canine model of OA. Furthermore, we evaluated the utility of novel biomarkers of cartilage degradation to predict joint structural changes. Several models have been developed as tools for understanding the typical OA changes in the cartilage and subchondral bone. Many of these are surgical models of the knee, closely resembling traumatic injury to articular cartilage (16–18). While no model has been fully characterized due to the lack of a validated DMOAD for comparison, the canine models offer advantages over rodent models because the complement of MMPs and degradation products are more consistent with humans (1). The dog partial meniscectomy model generates consistent focal lesions that facilitate the detection of changes with treatment (19,20). We further character-

3007

ized this model with respect to gross pathology, microscopic lesions, and soluble biomarkers, and we demonstrated the protective effect of PF152 on joint structural damage using a multivariate analysis approach integrating the various gross, histologic, and biomarker measures. MATERIALS AND METHODS Collagenase activity. Apparent inhibition constants (Ki(app)) were determined using the synthetic quenched fluorescent peptide substrate MCA-Arg-Pro-Leu-Gly-Leu-DpaAla-Arg-Glu-Arg-NH2 as described (21). Full-length recombinant human proteins were used for MMPs 1, 2, 9, and 13. The catalytic domain of the protein was used for MMPs 3, 7, 8, 10, 12, 14, 15, 16, 20, 24, 25, and 26. PF152 potency was also evaluated using full-length bovine type II collagen by quantifying the generation of the one-quarter–length and threequarter–length fragments upon digestion with full-length human MMP-13 using sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Selectivity of PF152 against ADAM-17 (TACE), ADAMTS-4 (aggrecanase 1), and ADAMTS-5 (aggrecanase 2) was determined as described (21). Human articular cartilage was obtained from OA patients undergoing knee replacement. Apparently normal cartilage was dissected and cultured as described (22,23) with the following modifications: 40 mg tissue/well was incubated in 96-well plates with 200 ␮l of DMEM media (high glucose, consisting of 25 mM HEPES containing 2 mM L-glutamine and 1 mM sodium pyruvate; Gibco) freshly supplemented with 1⫻ HL-1 (Lonza) and 5 ␮g/ml ascorbic acid (Sigma), with or without 0.1 ng/ml interleukin-1␤ (R&D Systems) plus 50 ␮g/ml oncostatin M (R&D Systems). PF152 was added to cultures in 10 replicates at 4 concentrations. Media were replaced every 3–4 days for 17 days and stored frozen. Biomarkers. Total TIINE. Urine or human articular cartilage media (25 ␮l) were serially diluted and analyzed using a novel chemiluminescence sandwich immunoassay consisting of the neoepitope 9A4 antibody and the capture 5109 antibody as detailed (1,24). Values were calculated from a standard curve prepared from human 45-mer (0.125–10 ng/ml) or dog 30-mer (0.313–20 ng/ml) TIINE peptide. Urine TIINE values were normalized with creatinine levels. TIINE 45-mer assay. Quantitation of 45-mer TIINE peptide with 5-hydroxyproline was performed by 2dimensional liquid chromatography–tandem mass spectrometry (LC-MS/MS) as described elsewhere (1,25), using dog urine or human articular cartilage supernatant diluted 7-fold or 100-fold, respectively, in 50 mM ammonium acetate. Aggrecan neoepitope assays. A novel assay was developed for the quantification of aggrecanase-generated aggrecan neoepitope peptides (26) based on the LC-MS/MS analysis of peptides of specific length generated by chymotrypsin digestion. Dog synovial fluid or human articular cartilage supernatant (25 ␮l) was heated for 10 minutes at 100°C to inactivate endogenous enzymatic activity. Samples were cooled and digested in buffer (100 mM Tris, 10 mM CaCl2, 10% acetonitrile, and 100 ␮g chymotrypsin; Sigma) at 37°C. Digested sample (1 ml) was injected onto 2 sequential immunoaffinity

3008

columns made of the following antibodies: a monoclonal antibody (BC-3) recognizing the N-terminus of aggrecan fragments produced by cleavage at the Glu373–Ala374 bond (a gift from Dr. C. Hughes, Cardiff, Wales) and an in house– developed polyclonal antibody recognizing the N-terminus of cleavage at the Glu1771–Ala1772 bond. The peptides ending in ARGN or AGEG (referred to as ARGN and AGEG) were detected by monitoring high-performance liquid chromatography elution times and ion pairs corresponding to the specific MRM pairs. The peptides measured included ARGNVIL, its internal standard ARGNVIL-d8, AGEGPSGIL, and its internal standard AGEGPSGIL-6C13. Assay for N-propeptide of type II collagen (PIINP). PIINP was measured in dog serum using a novel enzymelinked immunosorbent assay (NPII assay; Pfizer) that quantitatively measures N-terminal type II procollagen peptides as previously described (27). Total aggrecan and C-propeptide of type II collagen (CPII) assays. Assay for total aggrecan (Immunodiagnostic Systems), measuring intact aggrecan and fragments carrying the interglobular domain I and/or II generated during aggrecan turnover, as well as an assay for CPII (Ibex) were performed in dog serum according to the manufacturers’ directions. Canine partial meniscectomy. Sixty skeletally mature female beagles with an average age of 48 months and an average weight of 12 kg were subjected to partial medial meniscectomy of the left stifle join as described previously (20). The dogs were housed 5 per run by treatment group and exercised daily to encourage full use of the affected leg. Beginning 1 day prior to surgery, dogs received test compound orally via capsule every 12 hours. Treatment continued for 28 days after surgery. Animals were observed daily for signs of lameness or adverse reactions. Treatment groups (15 dogs per group) received vehicle (empty capsules), 0.5 mg/kg PF152, 1.5 mg/kg PF152, or 5 mg/kg PF152. Blood and urine were collected from all animals 3 days before surgery and on days 14, 21, and 27 after surgery. Urine was obtained by cystocentesis in conscious dogs before the morning dosing, after placing the dogs in metabolic cages overnight. Immediately after dogs were killed, synovial fluid was collected from both knees ⬃12 hours after the last dose. Left knees were dissected with synovium, evaluated for gross morphology, and placed into 10% neutral buffered formalin for processing. All procedures were approved by the Institutional Animal Care and Use Committee. Gross pathology. The operated knee was disarticulated, and the lesions on the tibial plateaus and femoral condyles were measured, described, and photographed before and after staining with India ink. Scores were assigned to describe the degree of meniscal repair, the gross appearance of osteophytes or peripheral fibrous tissue proliferation, the gross synovial reaction, and cartilage degeneration in the medial tibial plateau and medial femoral condyle. Cartilage lesion areas (in mm2) were measured, and depth was scored as 1 ⫽ superficial, 2 ⫽ medium, and 3 ⫽ deep (lesion score ⫽ length ⫻ width ⫻ depth). General pathology scores (scale of 0–4) were assigned based on India ink staining intensity, lesion area, and depth perception as described (20). Photographs of India ink–stained medial tibial plateaus were used to calculate the percentage area of lesion by image analysis, with lesion areas expressed as

SETTLE ET AL

the percentage of total medial tibial plateau area. All analyses were done in a blinded manner by 2 independent investigators (SS and AB). Histopathology. The proximal end of the left tibia and the distal end of the left femur were fixed and decalcified. The surface areas at risk for lesion development of the tibia and femur were prepared as described (20). The resulting 8-␮m sections were numbered 1–3 (levels), with 1 being the most anterior piece, and were stained with toluidine blue. Each articular surface was divided visually with an ocular micrometer into fourths, from outside adjacent to synovium (zone 1) to inside adjacent to cruciate ligaments (zone 4). Synovial membrane, subchondral bone, and osteophytes were each described and graded. Cartilage degeneration was scored 1–5 by level and zone based on chondrocyte loss, proteoglycan loss, fibrillation, and relative depth of degeneration, as previously described (20,28). Width measurements of any type of lesions were taken across the tibial and femoral surfaces to give a total lesion width. Compound concentration. Using LC-MS/MS analysis, PF152 was measured in plasma collected on day 14 following protein precipitation. For the quantification of PF152, the peak area of the mass/charge 474.23139.1 transition was measured, and linear regression analysis from the standards was used to determine the concentration. Statistical analysis. Comparison of biomarker levels in serum and urine of animals before and after surgery-induced OA and in synovial fluid from operated and contralateral knees was performed using a paired t-test of values corrected for baseline levels. The nonparametric Kruskal-Wallis test was used to calculate significant differences among dose groups in biomarker, gross, and microscopic lesion measurements. Comparisons between the dose groups were done using the Wilcoxon rank sum test. Multivariate discriminant analysis was performed on variables that were scaled to have the same mean and SD. The statistical analyses were conducted using SAS software (SAS Institute). All analyses were done in a blinded manner with respect to treatment.

RESULTS PF152 is a potent inhibitor of MMP-13 in vitro. The novel MMP-13 inhibitor PF152, N-(4-fluoro-3methoxybenzyl)-6-(2-(((2S,5R)-5-(hydroxymethyl)-1,4dioxan-2-yl)methyl)-2H-tetrazol-5-yl)-2-methylpyrimidine-4-carboxamide, potently inhibited human MMP-13 with a mean ⫾ SEM Ki of 1.5 ⫾ 0.24 nM. When full-length bovine type II collagen and type II gelatin were used as substrates, the 50% inhibition concentration (IC50) was 1.2 nM and 2 nM, respectively. The Ki for MMP-13 was ⬎3,000-fold lower than the Ki for all other enzymes tested (16 MMPs, TACE, ADAMTS-4, and ADAMTS-5). PF152 (10 ␮M) significantly inhibited basal TIINE levels (83%) in media of human articular cartilage cultures. Treatment of human articular cartilage with cytokines rapidly depleted aggrecan content and progressively degraded type II collagen, with in-


3009

Figure 1. Inhibition of human cartilage degradation by PF152. Human articular cartilage explants were incubated with (positive control) or without (negative control) cytokines and the indicated concentrations of PF152. A and B, Total levels of type II collagen neoepitope (TIINE) in the culture media of unstimulated (A) or cytokine-stimulated (B) human articular cartilage following 13 days of culture, expressed as ng/ml supernatant/mg cartilage wet weight. C and D, Levels of the peptides ending in ARGN (C) and AGEG (D) in the culture media of cytokine-stimulated human articular cartilage following the first 4 days of culture, expressed as ng/ml culture supernatant. Values are the mean ⫾ SEM. The percent inhibition from negative control (A) or positive control (B–D) is indicated. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01; ⴱⴱⴱ ⫽ P ⬍ 0.001 versus negative control in A or positive control in B–D, by analysis of variance.

creasing early production of AGEG/ARGN and later generation of TIINE. PF152 completely inhibited cytokine-induced TIINE in a dose-dependent manner in day 13 culture media (estimated IC50 ⱕ300 nM) and to a lesser degree AGEG and ARGN in day 4 culture media (Figure 1). The sequence homology of dog and human MMP-13 is 92%, and within a 9Å radius of the PF152

binding site, both dog and human enzymes have identical residues. As expected, PF152 inhibited the dog MMP-13 catalytic domain and the human MMP-13 catalytic domain in a similar manner (Ki of 0.84 nM and 1.2 nM, respectively). In vivo OA model characterization. Gross pathologic findings. All 60 dogs were fully weight bearing within 24 hours of surgery; there were no differences

Table 1. Biomarker changes in the canine partial medial meniscectomy model of osteoarthritis* Biomarker

Baseline level

Day 28 level

Fold change†

P

Urine total TIINE, ng/mg creatinine Urine 45-mer TIINE, ng/mg creatinine SF ARGN, ng/ml SF AGEG, ng/ml Serum aggrecan, ng/ml Serum CPII, ng/ml Serum PIINP, ng/ml

6.6 ⫾ 2.6 3.2 ⫾ 0.5 0.85 ⫾ 0.14 0.72 ⫾ 0.08 27.5 ⫾ 1.2 1,420 ⫾ 51 9.5 ⫾ 1.3

14.6 ⫾ 8.0 6.3 ⫾ 0.8 1.7 ⫾ 0.2 2.1 ⫾ 0.2 15.4 ⫾ 1.2 1,470 ⫾ 78 11.0 ⫾ 1.17

2.2 ⫾ 0.5 2.0 ⫾ 0.2 2.0 ⫾ 0.2 2.9 ⫾ 0.3 0.60 ⫾ 0.02 1.1 ⫾ 0.1 1.3 ⫾ 0.1

0.008 0.001 0.002 0.035 0.001 0.5 0.054

* Values are the mean ⫾ SEM. TIINE ⫽ type II collagen neoepitope; ARGN ⫽ peptide ending in ARGN; AGEG ⫽ peptide ending in AGEG; CPII ⫽ C-propeptide of type II collagen; PIINP ⫽ N-propeptide of type II collagen. † Calculated for each dog based on pre- versus postsurgical levels (urine and serum markers) or on operated versus contralateral knee levels (synovial fluid [SF] markers).

3010

between groups in activity levels as observed daily or in weight gain, and there was no evidence of joint infection in any animal. Four weeks after surgery, all of the dogs in the vehicle-treated group had varying degrees of cartilage degeneration on the medial tibial plateau, and 4 of 15 had mild lesions on the medial femoral condyle. The lesions appeared as moderate fibrillation of the cartilage consistently in the area of greatest load. No lesions were noted in the lateral compartments. The lesions on the medial tibial plateau were superficial to moderate in severity (mean ⫾ SEM area 13.1 ⫾ 1.8 mm2 and pathology score 2.5 ⫾ 0.2) and comprised a mean ⫾ SEM 17.3 ⫾ 2% of the total plateau (percentage area of lesion). Lesions on the medial femoral condyle were inconsistent and smaller (mean ⫾ SEM area 0.4 ⫾ 0.2 mm2, mean ⫾ SEM pathology score 0.53 ⫾ 0.27). All dogs showed minimal to moderate fibrotic proliferation of the meniscal fragment (mean ⫾ SEM meniscal repair score 1.20 ⫾ 0.11), no osteophytes, and mild synovial thickening (mean ⫾ SEM synovial reaction score 1.13 ⫾ 0.09). Histopathologic findings. The medial tibial plateaus from all dogs had mild to moderate cartilage degeneration (mean ⫾ SEM degeneration score 1.04 ⫾ 0.06) consisting of chondrocyte and proteoglycan loss and fibrillation and moderate collagen loss, with the most severe lesions generally observed in the central area (level 2, zone 2). Medial femoral condyle lesions were minimal (mean ⫾ SEM degeneration score 0.49 ⫾ 0.06). No osteophytes were observed, there was minimal subacute synovium inflammation, and all but 1 animal had minimal to marked bone sclerosis in the femur (mean ⫾ SEM subchondral bone score 2.47 ⫾ 0.34). Due to the minimal medial femoral condyle lesions, the effect of the inhibitor is presented only for the medial tibial plateau. Biomarker findings. Urinary total TIINE and 45mer TIINE were significantly elevated ⬃2-fold in control animals on day 27, which was already evident, and to the same extent on days 14 and 21 after surgery (data not shown). Synovial fluid biomarkers ARGN and AGEG were significantly elevated (2-fold and 2.9-fold, respectively) in the operated knees at the end of the study. In contrast, the level of total aggrecan in the last serum sample was significantly decreased by 40% when compared with presurgery levels. The biomarkers of cartilage matrix synthesis, CPII and PIINP, showed no significant changes (Table 1). PF152 protects against cartilage degeneration in vivo. Randomization. Evaluation of prestudy total TIINE and 45-mer TIINE (on day –14) showed high levels in younger dogs (age ⬍20 months) and decreased levels

SETTLE ET AL

Figure 2. Biomarker responses. A, Correlation of dog age with basal urinary levels of total and 45-mer TIINE peptides measured 14 days prior to study initiation and expressed as ng/mg of urinary creatinine (creat). Note the difference in scales. B, Effect of PF152 on creatininecorrected levels of urinary TIINE peptides (total and 45-mer). For each animal, TIINE concentration 27 days after surgery was corrected by concentration 3 days before study initiation and expressed as percent of predose (indicated by dashed line). C, Effect of PF152 on synovial fluid (SF) levels of aggrecan biomarkers (ARGN and AGEG). Baseline levels (from contralateral legs of vehicle-treated group) are indicated by dashed lines. Values in B and C are the mean ⫾ SEM. The percent inhibition from the vehicle-treated group is indicated. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.001 versus vehicle-treated group, by analysis of variance. mpk ⫽ mg/kg body weight (see Figure 1 for other definitions).


Table 2.

3011

Effect of PF152 on the medial tibial plateau: gross pathologic evaluation*

2

Lesion area, mm † Percentage area of lesion Lesion score‡ Pathology score

Vehicle

0.5 mg/kg PF152

Inhibition vs. vehicle, %

P

1.5 mg/kg PF152


P

5 mg/kg PF152


P

13.1 ⫾ 1.8 17.3 ⫾ 2.1

7.6 ⫾ 1.5 7.7 ⫾ 1.5

42 55

0.037 0.001

8.9 ⫾ 2.2 8.5 ⫾ 2.0

32 51

0.092 0.002

7.1 ⫾ 1.3 8.0 ⫾ 1.6

45 54

0.021 0.002

16.7 ⫾ 2.1 2.5 ⫾ 0.2

11.6 ⫾ 2.7 1.9 ⫾ 0.3

30 24

0.113 0.172

14.6 ⫾ 4.2 1.9 ⫾ 0.4

12 26

0.133 0.108

8.3 ⫾ 1.7 1.3 ⫾ 0.2

50 47

0.024 0.007

* Except where indicated otherwise, values are the mean ⫾ SEM. † Length ⫻ width. ‡ Length ⫻ width ⫻ depth.

with age, probably reflecting collagenase activity in their growth plates (Figure 2A). Both TIINE assays showed good linear correlation (R2 ⫽ 0.739), although the absolute amount of 45-mer TIINE was lower than total TIINE. Dogs were randomized into 4 groups of 15 based on age, body weight, and urinary TIINE levels. Exposure. PF152 was detected in the plasma of all dogs treated with the MMP-13 inhibitor. The mean ⫾ SEM peak concentration ⬃2 hours after dosing was 816 ⫾ 149 nM, 2,500 ⫾ 752 nM, and 2,850 ⫾ 1,460 nM for the low-, intermediate-, and high-dose groups, respectively. The mean ⫾ SD trough concentration detected 12 hours after dosing was 56 ⫾ 26 nM, 94 ⫾ 36 nM, and 365 ⫾ 360 nM for the low-, intermediate-, and high-dose groups, respectively. PF152 was 52% bound to plasma proteins in the dogs, and the free fraction concentration at trough was calculated to be between 29 nM and 190 nM. Gross pathologic findings. Animals treated with P152 showed decreased cartilage degeneration com-

pared with vehicle-treated animals (Table 2). Three of 15 dogs in both the intermediate- and high-dose groups had no visible lesions. The lesion areas (percentage area of lesion) were significantly reduced to a similar extent with all doses (51–55%). The pathology score, however, was reduced by PF152 in a dose-dependent manner, and only the high dose significantly reduced pathology and lesion scores (inhibition of pathology and lesions of 47% and 50%, respectively). Histopathologic findings. PF152 reduced both the medial tibial plateau lesion width and the cartilage degeneration score in a dose-dependent manner (24% and 26% inhibition of lesions and degeneration, respectively), with the high dose differing significantly from vehicle. By breaking down the scores according to levels (anterior to posterior), the effect of PF152 was most pronounced at the posterior level 3 (86–85% inhibition) and it was insignificant in the weight-bearing middle level 2 (15–11% inhibition). When evaluation was done by zones (inside to outside), the highest inhibition (34%)

Table 3. Effect of PF152 on the medial tibial plateau: histologic evaluation*

Lesion width, ␮m Average Level 1 (anterior) Level 2 Level 3 Cartilage degeneration score Average Level 1 (anterior) Level 2 Level 3 Zone 1 (outside) Zone 2 Zone 3 Zone 4

Vehicle

0.5 mg/kg PF152


P

1.5 mg/kg PF152


6,410 ⫾ 314 8,230 ⫾ 345 8,040 ⫾ 459 2,960 ⫾ 782

5,710 ⫾ 321 7,540 ⫾ 572 7,370 ⫾ 509 2,230 ⫾ 664

11 8 8 25

0.347 0.707 0.156 0.922

5,230 ⫾ 372 6,780 ⫾ 332 7,240 ⫾ 589 1,660 ⫾ 581

18 18 10 44

1.04 ⫾ 0.06 1.37 ⫾ 0.09 1.32 ⫾ 0.11 0.43 ⫾ 0.12 0.73 ⫾ 0.06 1.66 ⫾ 0.12 0.96 ⫾ 0.06 0.80 ⫾ 0.07

0.89 ⫾ 0.06 1.13 ⫾ 0.10 1.17 ⫾ 0.10 0.35 ⫾ 0.10 0.53 ⫾ 0.05 1.33 ⫾ 0.17 0.99 ⫾ 0.09 0.69 ⫾ 0.06

15 17 11 19 27 20 ⫺2 14

0.287 0.235 0.549 0.987 0.070 0.184 1.000 0.318

0.84 ⫾ 0.08 1.07 ⫾ 0.08 1.23 ⫾ 0.13 0.23 ⫾ 0.08 0.70 ⫾ 0.16 1.21 ⫾ 0.15 0.89 ⫾ 0.09 0.57 ⫾ 0.07

19 22 6 46 4 27 8 29

0.143 0.049 0.774 0.648 0.522 0.065 0.665 0.016

* Except where indicated otherwise, values are the mean ⫾ SEM.


P

0.057 4,850 ⫾ 224 0.011 7,270 ⫾ 330 0.301 6,850 ⫾ 485 0.585 420 ⫾ 311

24 12 15 86

0.002 0.108 0.035 0.023

0.77 ⫾ 0.06 1.07 ⫾ 0.08 1.18 ⫾ 0.12 0.07 ⫾ 0.05 0.52 ⫾ 0.10 1.09 ⫾ 0.12 0.87 ⫾ 0.10 0.60 ⫾ 0.05

26 22 11 85 29 34 9 25

0.015 0.041 0.538 0.032 0.050 0.013 0.345 0.035

P

5 mg/kg PF152

3012

was detected in zone 2 (Table 3). Synovium pathology, bone scores, and collagen degeneration scores were not different from those in controls. Biomarker findings. All 3 doses of PF152 similarly brought urinary TIINE back to basal levels (Figure 2B). This same effect was observed in a subset of animals 14 days after surgery/treatment and was maintained 21 and 27 days after surgery/treatment. The aggrecan degradation biomarkers in synovial fluid were also inhibited by PF152. The intermediate dose decreased ARGN by 36%, while the high dose inhibited AGEG by 44%, compared with control levels in normal knees (Figure 2C). In contrast, total serum aggrecan was not modified by PF152 treatment. PIINP and CPII were not analyzed since there was no change in controls. Multivariate analysis. Given that each measure (gross, histologic, or biomarker) appeared to identify a different concentration of inhibitor as efficacious, a

SETTLE ET AL

multivariate discriminant analysis was performed to establish whether, when taking all measurements into consideration, there was a dose-dependent effect of PF152. The multivariate discriminant analysis created linear combinations (standardized canonical coefficients) of 10 gross, histologic, and biomarker variables (Figure 3) that explained 95% of the variability found in the data (78.5% explained by standardized canonical coefficient 1 and 16.5% explained by standardized canonical coefficient 2). A clear and significant (P ⬍ 0.001 by Wilks’ ␭) separation was observed between treated and untreated groups, but no significant (P ⬎ 0.10) separation was found among the 3 PF152-treated groups. To understand if biomarkers were representative of structural changes, multivariate discriminant analysis was done separately using only structural measures or biomarkers. Multivariate discriminant analysis performed with the 6 gross and histologic values created a first linear combination that explained 82% of the variability in the data. Multivariate discriminant analysis performed with the 4 biomarkers created a first linear combination that explained 75% of the variability in the data. All PF152-treated groups were significantly different from the vehicle-treated group, but not from each other, by either test. DISCUSSION

Figure 3. Separation between PF152 and vehicle treatment by multivariate discriminant analysis. Plot shows the 2 linear combinations (standardized canonical coefficients 1 and 2) created by multivariate discriminant analysis of 4 gross variables (medial tibial plateau lesion area [length ⫻ width] and percentage area of lesion, lesion score [length ⫻ width ⫻ depth], and general pathology score), 2 histologic variables (lesion width [average] and cartilage degeneration score [average]), and 4 biomarker variables (urinary TIINE [total and 45-mer, expressed as percent of predose] and synovial fluid ARGN and AGEG). Significant (P ⬍ 0.001) separation was observed between the vehicle-treated group (black symbols) and the low-dose group (red symbols), intermediate-dose group (green symbols), and high-dose group (yellow symbols). See Figure 1 for definitions.

PF152 is a non–hydroxamic acid MMP-13 inhibitor that binds the S1⬘ active site pocket of the enzyme and is not a ligand to the catalytic zinc atom, making it highly selective for MMP-13 (5,7). Here we show that PF152 protects against human articular cartilage degradation with or without cytokine stimulation, as demonstrated by the reduction of the cartilage breakdown products TIINE, ARGN, and AGEG. Furthermore, PF152 demonstrated cartilage protection when administered orally to dogs with OA-like lesions. The canine model was chosen because dogs express the same complement of MMPs as humans, while rodents lack MMP-1. Additionally, the growth plates in dogs close with maturity, unlike those in rats. Open growth plates express high levels of MMP-13 and TIINE, reflecting endochondral ossification, which potentially complicates the evaluation of PF152 on cartilage lesions and the utility of TIINE as a biomarker of OA. Therefore, basal urinary TIINE levels were used to determine the appropriate age of dogs for use in the drug trial. The partial meniscectomy surgical model was investigated because it develops consistent lesions localized in the area of the meniscal defect corresponding to


the greatest weight-bearing region, but not in the lateral compartments (19), in contrast to the commonly reported canine anterior cruciate ligament transection model that produces variable lesions in all quadrants of the joint (19,29–31) and requires a longer time to develop OA pathology (32). The progression of the cartilage lesion correlates well with human disease, with initial proteoglycan loss followed by cartilage fibrillation and eventually significant cartilage defects, but it occurs rapidly. Four weeks after surgery, bone changes were negligible in this study, although sclerosis and osteophytes have been reported to develop with time as the joint remodels to redistribute the load on the tibial surface (19). During the 4-week study period, the dogs rarely showed signs of lameness, which adds to the lesion consistency but makes this model less relevant for pain or functional studies. The 2 most abundant components of articular cartilage are type II collagen and aggrecan. The breakdown of collagen in OA is initiated primarily by MMP-13 cleaving the helical protein, resulting in peptides ending in specific neoepitopes (1). Aggrecan fragments are similarly produced at varying sites predominantly by ADAMTS-4/5 but also by MMPs (26,33,34). The breakdown products are released into the synovial fluid and cleared into the serum and urine, where they can be measured as biomarkers for disease progression or response to therapeutic intervention. We implemented 2 novel and complementary assays for TIINE peptides. The 45-mer LC-MS/MS detects the most abundant TIINE peptide generated by MMP-13 and is the assay of choice for clinical samples, while the TIINE immunoassay measures the sum of all peptides containing neoepitopes and is useful for small volumes in preclinical studies. Both assays revealed a 2-fold increase following meniscectomy surgery, showing that sufficient cartilage breakdown occurs in a single joint to give measurable elevation of a biomarker in the urine. Interestingly, similar ratios were observed when urinary TIINE levels were measured in OA patients and compared with those in age-matched healthy subjects (35–37). We also used novel LC-MS/MS–based assays for aggrecan degradation biomarkers ARGN and AGEG. Peptides ending in these sequences were elevated 2–3fold in the synovial fluid of dogs 4 weeks after meniscectomy surgery. Similarly, ARGN and AGEG were elevated in the synovial fluid in rat models of OA and in OA patients (data not shown). Unlike the case with TIINE, we have been unable to detect changes of these aggrecan biomarkers in urine, and therefore, they were evaluated only in synovial fluid. Serum biomarkers eval-

3013

uated in this study included total aggrecan, CPII, and PIINP. Total aggrecan decreased 40% in meniscectomized dogs, a decrease similar to that in patients with inflammatory arthritis (38). We did not observe changes in levels of either PIINP or CPII following development of lesions, although we have described decreased plasma PIINP levels in OA patients (27), and levels of CPII were found to be elevated in OA cartilage and synovial fluid shortly after knee injury (15). PF152 reduced the OA lesions and biomarkers following 4 weeks of prophylactic treatment, but the dose required for significant inhibition varied with the parameters assessed. Clearly, a dose-dependent effect of PF152 was observed when the cartilage lesions were examined histologically or when the depth of the lesion was taken into consideration in the gross pathologic assessments (length ⫻ width ⫻ depth and pathology scores). However, 2-dimensional lesion measurements (evaluated by image analysis) were similarly reduced by all doses of PF152, which correlated with its effect on TIINE. These results suggest that collagenase activity may be important for the superficial progression of OA lesions. Cartilage damage progresses from the area of greatest trauma to the surrounding tissues, most likely driven by collagenases and other biochemical mediators released from damaged chondrocytes. The effect of PF152 on the progression of disease was evident at the periphery of the lesions when the affected areas were subdivided for analysis, with 85% inhibition at the posterior region. However, when all the histologic regions were averaged, the high dose of PF152 reduced the cartilage degeneration score by 26%. The magnitude of this effect was similar to that reported for other MMP-13 selective inhibitors in the rat medial meniscal tear model (6,7) and was not significantly different from that of MMP inhibitors with broader selectivity (ref. 39 and data not shown). Similarly, cathepsin K inhibitor decreased the cartilage degeneration score in this dog model by 21% (20). The limited response is believed to be due to the aggressiveness of these models with no relief of mechanical stress. All doses of PF152 restored TIINE to normal levels, suggesting that inhibition of MMP-13–mediated collagen degradation was maximal even at the lowest dose evaluated, at least in the synovial space and the superficial layers of cartilage. Further studies are required to evaluate concentration of PF152 deeper in the cartilage. Selective MMP-13 inhibition also resulted in the significant but incomplete reduction of the aggrecanase-

3014

SETTLE ET AL

specific biomarkers ARGN and AGEG, both in human articular cartilage culture media and in dog synovial fluid. Breakdown of the collagen network that occurs in OA may facilitate the access of ADAMTS-4/5 to its substrate, and therefore, inhibition of MMP-13 activity by PF152 and preservation of the collagen network may limit ADAMTS-4/5 cleavage of aggrecan and reduce the subsequent production of ARGN/AGEG biomarkers. In addition, MMPs have been shown to cleave aggrecan directly; synovial fluid from OA patients contains aggrecan fragments that are generated by both ADAMTS-4/5 (ARGN) and MMPs (FFGV), consistent with 2 proteolytic pathways for aggrecan degradation (34). Upon MMP cleavage of aggrecan, the newly generated fragment containing the ADAMTS-4/5 sites may diffuse more readily through the cartilage matrix and become more accessible to further ADAMTS-4/5–mediated degradation. Thus, inhibition of aggrecan breakdown at the MMP site by PF152 may in turn decrease generation of ARGN/AGEG biomarkers. These findings indicate that fragments of both collagen and aggrecan are potential and possibly complementary markers of disease progression. In addition to the statistical analysis performed for each of the individual structural and biomarker measures in a univariate manner, multivariate discriminant analysis was performed using all the available data to better understand the overall effect of the inhibitor. When all variables were taken into consideration, the dose response was less evident, as illustrated in Figure 3, in which the vehicle-treated group is clustered apart from the PF152-treated groups, which were less separated from each other. This may be explained by the relatively higher influence in the analysis (loading) of parameters that individually showed no dose response, including lesion score calculated as length ⫻ width ⫻ depth, lesion area calculated as percentage area of lesion, and 45-mer TIINE level, with standardized canonical coefficients of –1.65, 1.21, and 0.77, respectively. Interestingly, similar separation between groups was obtained when only structural parameters or biomarkers were included in the multivariate discriminant analysis. These data support the potential therapeutic value of selective MMP-13 inhibitors and the use of a set of appropriate biomarkers to predict efficacy in OA clinical trials. ACKNOWLEDGMENTS The authors would like to thank Grace Munie, Mike Davies, Vincent Peterkin, Mark Abrams, Linda Branson, and

Poonam Aggarwal (Pfizer) for analyzing enzyme selectivity and potency of PF152, performing pharmacokinetic analysis, or running biomarker assays, and Don Maul (Pre-Clinical Research Services, Fort Collins, CO) for coordinating the animal study. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Sunyer had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. Settle, Vickery, Sunyer. Acquisition of data. Vickery, Nemirovskiy, Bendele, Ruminski. Analysis and interpretation of data. Settle, Vickery, Nemirovskiy, Vidmar, Bendele, Messing, Schnute, Sunyer.

REFERENCES 1. Nemirovskiy OV, Dufield DR, Sunyer T, Aggarwal P, Welsch DJ, Mathews WR. Discovery and development of a type II collagen neoepitope (TIINE) biomarker for matrix metalloproteinase activity: from in vitro to in vivo. Anal Biochem 2007;361:93–101. 2. Neuhold LA, Killar L, Zhao W, Sung ML, Warner L, Kulik J, et al. Postnatal expression in hyaline cartilage of constitutively active human collagenase-3 (MMP-13) induces osteoarthritis in mice. J Clin Invest 2001;107:35–44. 3. Little CB, Barai A, Burkhardt D, Smith SM, Fosang AJ, Werb Z, et al. Matrix metalloproteinase 13–deficient mice are resistant to osteoarthritic cartilage erosion but not chondrocyte hypertrophy or osteophyte development. Arthritis Rheum 2009;60:3723–33. 4. Renkiewicz R, Qiu L, Lesch C, Sun X, Devalaraja R, Cody T, et al. Broad-spectrum matrix metalloproteinase inhibitor marimastat– induced musculoskeletal side effects in rats. Arthritis Rheum 2003;48:1742–9. 5. Johnson AR, Pavlovsky AG, Ortwine DF, Prior F, Man CF, Bornemeier DA, et al. Discovery and characterization of a novel inhibitor of matrix metalloprotease-13 that reduces cartilage damage in vivo without joint fibroplasia side effects. J Biol Chem 2007;282:27781–91. 6. Baragi VM, Becher G, Bendele AM, Biesinger R, Bluhm H, Boer J, et al. A new class of potent matrix metalloproteinase 13 inhibitors for potential treatment of osteoarthritis: evidence of histologic and clinical efficacy without musculoskeletal toxicity in rat models. Arthritis Rheum 2009;60:2008–18. 7. Schnute ME, O’Brien PM, Nahra J, Morris M, Roark WH, Hanau CH, et al. Discovery of (pyridin-4-yl)-2H-tetrazole as a novel scaffold to identify highly selective matrix metalloproteinase-13 inhibitors for the treatment of osteoarthritis. Bioorg Med Chem Lett 2010;20:576–80. 8. Rousseau JC, Delmas PD. Biological markers in osteoarthritis. Nat Clin Pract Rheumatol 2007;3:346–56 9. Hellio Le Graverand MP, Brandt KD, Mazzuca SA, Katz BP, Buck R, Lane KA, et al. Association between concentrations of urinary type II collagen neoepitope (uTIINE) and joint space narrowing in patients with knee osteoarthritis. Osteoarthritis Cartilage 2006;14:1189–95. 10. Otterness IG, Brandt KD, Hellio Le Graverand MP, Mazzuca SA. Urinary TIINE concentrations in a randomized controlled trial of doxycycline in knee osteoarthritis: implications of the lack of association between TIINE levels and joint space narrowing. Arthritis Rheum 2007;56:3644–9. 11. Sowers MF, Karvonen-Gutierrez CA, Yosef M, Jannausch M, Jiang Y, Garnero P, et al. Longitudinal changes of serum COMP


12.

13.

14.

15. 16. 17. 18. 19. 20.

21.

22.

23.

24. 25.

and urinary CTX-II predict X-ray defined knee osteoarthritis severity and stiffness in women. Osteoarthritis Cartilage 2009;17: 1609–14. Tanishi N, Yamagiwa H, Hayami T, Mera H, Koga Y, Omori G, et al. Relationship between radiological knee osteoarthritis and biochemical markers of cartilage and bone degradation (urine CTX-II and NTX-I): the Matsudai Knee Osteoarthritis Survey. J Bone Miner Metab 2009;27:605–12. Sharif M, Kirwan J, Charni N, Sandell LJ, Whittles C, Garnero P. A 5-yr longitudinal study of type IIA collagen synthesis and total type II collagen degradation in patients with knee osteoarthritis— association with disease progression. Rheumatology (Oxford) 2007;46:938–43. Meulenbelt I, Kloppenburg M, Kroon HM, Houwing-Duistermaat JJ, Garnero P, Hellio Le Graverand MP, et al. Clusters of biochemical markers are associated with radiographic subtypes of osteoarthritis (OA) in subjects with familial OA at multiple sites: the GARP study. Osteoarthritis Cartilage 2007;15:379–85. Birmingham JD, Vilim V, Kraus VB. Collagen biomarkers for arthritis applications. Biomark Insights 2007;1:61–76. Troyer H. Experimental models of osteoarthritis: a review. Semin Arthritis Rheum 1982;11:362–74. Warskyj M, Hukins DW. Animal models for osteoarthritis— ensuring experimental validity. Br J Rheumatol 1990;29:219–21. Bendele AM. Animal models of osteoarthritis. J Musculoskelet Neuronal Interact 2001;1:363–76. Bendele AM. Animal models. In: Bronner F, Farach-Carson MC, editors. Bone and osteoarthritis: topics in bone biology. London: Springer Verlag; 2007. p. 149–63. Connor JR, LePage C, Swift BA, Yamashita D, Bendele AM, Maul D, et al. Protective effects of a cathepsin K inhibitor, SB-553484, in the canine partial medial meniscectomy model of osteoarthritis. Osteoarthritis Cartilage 2009;17:1236–43. Tortorella MD, Tomasselli AG, Mathis KJ, Schnute ME, Woodard SS, Munie G, et al. Structural and inhibition analysis reveals the mechanism of selectivity of a series of aggrecanase inhibitors. J Biol Chem 2009;184:24185–91. Billinghurst RC, Dahlberg L, Ionescu M, Reiner A, Bourne R, Rorabeck C, et al. Enhanced cleavage of type II collagen by collagenases in osteoarthritic articular cartilage. J Clin Invest 1997;99:1534–45. Dahlberg L, Billinghurst RC, Manner P, Nelson F, Webb G, Ionescu M, et al. Selective enhancement of collagenase-mediated cleavage of resident type II collagen in cultured osteoarthritic cartilage and arrest with a synthetic inhibitor that spares collagenase 1 (matrix metalloproteinase 1). Arthritis Rheum 2000;43: 673–82. Yates MP, Settle SL, Yocum SA, Aggarwal P, Vickery LE, Aguiar DJ, et al. IGFBP-5 metabolism is disrupted in the rat medial meniscal tear model of osteoarthritis. Cartilage 2010;1:55–64. Li WW, Nemirovskiy O, Fountain S, Mathews RW, SzekelyKlepser G. Clinical validation of an immunoaffinity LC-MS/MS assay for the quantification of a collagen type II neoepitope peptide biomarker of osteoarthritis in human urine. Anal Biochem 2007;369:41–53.

3015

26. Tortorella MD, Pratta M, Liu RQ, Austin J, Ross OH, Abbaszade I, et al. Sites of aggrecan cleavage by recombinant human aggrecanase-1 (ADAMTS-4). J Biol Chem 2000;275:18566–73. 27. Nemirovskiy OV, Sunyer T, Aggarwal P, Abrams MA, Hellio Le Graverand MP, Mathews WR. Discovery and development of the N-terminal procollagen type II (NPII) biomarker: a tool for measuring collagen type II synthesis. Osteoarthritis Cartilage 2008;16:1494–500. 28. Bendele AM, Hulman JF. Effects of body weight restriction on the development and progression of spontaneous osteoarthritis in guinea pigs. Arthritis Rheum 1991;34:1180–4. 29. Pond MJ, Nuki G. Experimentally induced osteoarthritis in the dog. Ann Rheum Dis 1973;32:387–8. 30. Pelletier JP, Martel-Pelletier J, Altman RD, Ghandur-Mnaymneh L, Howell DS, Woessner JF Jr. Collagenolytic activity and collagen matrix breakdown of articular cartilage in the Pond-Nuki dog model of osteoarthritis. Arthritis Rheum 1983;26:866–74. 31. Van Valberg AA, van Roermund PM, Marijnissen CA, Wenting JG, Verbout AJ, Lafeber FP, et al. Joint distraction in treatment of osteoarthritis (II): effects on cartilage in a canine model. Osteoarthritis Cartilage 2000;8:1–8. 32. Brandt KD, Myers SL, Burr D, Albrecht M. Osteoarthritic changes in canine articular cartilage, subchondral bone, and synovium fifty-four months after transection of the anterior cruciate ligament. Arthritis Rheum 1991;34:1560–70. 33. Fosang AJ, Last K, Maciewicz RA. Aggrecan is degraded by matrix metalloproteinases in human arthritis: evidence that matrix metalloproteinase and aggrecanase activities can be independent. J Clin Invest 1996;98:2292–9. 34. Struglics A, Larsson S, Pratta MA, Kumar S, Lark MW, Lohmander LS. Human osteoarthritis synovial fluid and joint cartilage contain both aggrecanase- and matrix metalloproteinase-generated aggrecan fragments. Osteoarthritis Cartilage 2006;14:101–13. 35. Pickering E, Szekely-Klepser G, Nemirovskiy O, Li W, Brown R, Sunyer T, et al. Urinary type II collagen neoepitope (uTIINE): a marker of osteoarthritis activity [abstract]. Osteoarthritis Cartilage 2004;12 Suppl B:S93. 36. Sunyer T, Vickery LE, Nemirovskiy OV, Settle SL, Bendele A, Messing DM, et al. Translational pharmacology strategy to evaluate MMP-13 inhibitors for the treatment of OA; use of biomarkers [abstract]. Osteoarthritis Cartilage 2008;16 Suppl 4:S14. 37. Sunyer T, Nemirovskiy O, Abrams M, Vickery L, Dufield D, Fujiwara H, et al. Development and application of novel markers of collagenase activity, potential to support preclinical and clinical studies [abstract]. Osteoarthritis Cartilage 2004;12 Suppl B:S36. 38. Sumer EU, Sondergaard BC, Rousseau JC, Delmas PD, Fosang AJ, Karsdal MA, et al. MMP and non-MMP-mediated release of aggrecan and its fragments from articular cartilage: a comparative study of three different aggrecan and glycosaminoglycan assays. Osteoarthritis Cartilage 2007;15:212–21. 39. Janusz MJ, Bendele AM, Brown KK, Taiwo YO, Hsieh L, Heitmeyer SA. Induction of osteoarthritis in the rat by surgical tear of the meniscus: inhibition of joint damage by a matrix metalloproteinase inhibitor. Osteoarthritis Cartilage 2002;10: 785–91.