chronic myeloid leukemia in chronic phase treated with ... - Nature

2 downloads 62 Views 1MB Size Report
Prediction of outcomes in patients with Ph+ chronic myeloid ... Keywords: chronic myeloid leukemia; nilotinib; multivariate analysis; predictive model; imatinib ...
Leukemia (2013) 27, 907–913 & 2013 Macmillan Publishers Limited All rights reserved 0887-6924/13 www.nature.com/leu

ORIGINAL ARTICLE

Prediction of outcomes in patients with Phþ chronic myeloid leukemia in chronic phase treated with nilotinib after imatinib resistance/intolerance E Jabbour1, PD le Coutre2, J Cortes1, F Giles3, KN Bhalla4, J Pinilla-Ibarz5, RA Larson6, N Gattermann7, OG Ottmann8, A Hochhaus9, TP Hughes10, G Saglio11, JP Radich12, D-W Kim13, G Martinelli14, J Reynolds15, RC Woodman16, M Baccarani14 and HM Kantarjian1 The purpose was to assess predictive factors for outcome in patients with chronic myeloid leukemia (CML) in chronic phase (CML-CP) treated with nilotinib after imatinib failure. Imatinib-resistant and -intolerant patients with CML-CP (n ¼ 321) were treated with nilotinib 400 mg twice daily. Of 19 baseline patient and disease characteristics and two response end points analyzed, 10 independent prognostic factors were associated with progression-free survival (PFS). In the multivariate analysis, major cytogenetic response (MCyR) within 12 months, baseline hemoglobin X120 g/l, baseline basophils o4%, and absence of baseline mutations with low sensitivity to nilotinib were associated with PFS. A prognostic score was created to stratify patients into five groups (best group: 0 of 3 unfavorable risk factors and MCyR by 12 months; worst group: 3 of 3 unfavorable risk factors and no MCyR by 12 months). Estimated 24-month PFS rates were 90%, 79%, 67% and 37% for patients with prognostic scores of 0, 1, 2 and 3, respectively, (no patients with score of 4). Even in the presence of poor disease characteristics, nilotinib provided significant clinical benefit in patients with imatinib-resistant or -intolerant CML. This system may yield insight on the prognosis of patients. Leukemia (2013) 27, 907–913; doi:10.1038/leu.2012.305 Keywords: chronic myeloid leukemia; nilotinib; multivariate analysis; predictive model; imatinib intolerance; imatinib resistance

INTRODUCTION The successful introduction of the selective BCR-ABL tyrosine kinase inhibitors (TKIs), which suppress the molecular processes driving Philadelphia chromosome-positive (Ph þ ) chronic myeloid leukemia (CML) in chronic phase (CP) (CML-CP), has revolutionized the management and outcome in CML.1 In patients with newly diagnosed Ph þ CML-CP, imatinib mesylate therapy induced high rates of complete cytogenetic response (CCyR) and major molecular response, and improved survival in Ph þ CML.2–5 Following imatinib treatment, 490% of patients obtained complete hematological response (CHR), and 480% achieved CCyR. With 8 years of follow-up, the results were favorable, resulting in a major change in the natural history of the disease.6 Despite the benefit of imatinib over prior treatments, some patients may develop resistance,7 with a reported annual resistance rate of 2–4% in patients with newly diagnosed CMLCP during the first 3 years, with the incidence decreasing over time thereafter.8 Nilotinib is a potent and highly selective BCRABL kinase inhibitor, approved for the treatment of patients with Ph þ CML-CP or accelerated phase and resistance or

intolerance to imatinib therapy.9,10 After 24 months of follow-up of imatinib-resistant and -intolerant patients treated with nilotinib in the pivotal phase II study, achievement of major cytogenetic response (MCyR) was demonstrated in 59% of patients and CCyR in 44% of patients.11 Despite the success of nilotinib therapy in patients resistant to or intolerant of prior imatinib treatment, some patients did not achieve optimal responses to therapy.12 Early identification of these patients would permit earlier interventions to maximize treatment benefits and improve outcomes for this patient population. The purpose of the analysis was to identify the baseline patient and disease factors and the time-dependent response end points that potentially may predict for differences in long-term outcomes.

MATERIALS AND METHODS Patients Three hundred and fifteen patients with Ph þ CML-CP who were resistant to or intolerant of imatinib and not known to have the BCR-ABL T315I

1 Department of Leukemia, MD Anderson Cancer Center, The University of Texas, Houston, TX, USA; 2Medizinische Klinik m.S. Ha¨matologie und Onkologie, Charite´—University of Medicine Berlin, Berlin, Germany; 3HRB Clinical Research Facility, National University of Ireland Galway and Trinity College Dublin, Galway, Ireland; 4Kansas University Medical Center, University of Kansas Cancer Center, Kansas City, KS, USA; 5Malignant Hematology Division, H Lee Moffitt Cancer Center, Tampa, FL, USA; 6University of Chicago, Chicago, IL, USA; 7Heinrich-Heine-Universitat, Du¨sseldorf, Germany; 8Medizinische Klinik II - Johann Wolfgang Goethe-University, Frankfurt, Germany; 9Klinik fu¨r Innere Medizin II, Abteilung Ha¨matologie/Onkologie, Universita¨tsklinikum Jena, Jena, Germany; 10Royal Adelaide Hospital, SA Pathology, Adelaide, South Australia, Australia; 11Division of Internal Medicine & Hematology, University of Turin, Orbassano, Italy; 12Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 13Seoul St Mary’s Hospital, The Catholic University of Korea, Seoul, South Korea; 14Institute of Hematology ‘‘L. e A. Sera`gnoli’’, University of Bologna, Bologna, Italy; 15Novartis Pharma AG, Basel, Switzerland and 16 Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA. Correspondence: Dr E Jabbour, Department of Leukemia, MD Anderson Cancer Center, The University of Texas, 1515 Holcombe Boulevard, Houston, TX 77030, USA. E-mail: [email protected] Presented in abstract form at the 51st annual meeting of the American Society of Hematology, New Orleans, LA, December 7, 2009. Received 17 August 2012; revised 27 September 2012; accepted 28 September 2012; accepted article preview online 6 November 2012; advance online publication, 23 November 2012

Factors predictive of patient outcome on nilotinib E Jabbour et al

908 mutation at study entry were treated with nilotinib 400 mg orally twice daily in an ongoing open-label, single-treatment arm phase II study (N ¼ 321) that has been previously described (study AMN-2101).10,11 CMLCP was defined as o15% blasts in peripheral blood, o20% basophils, o30% blasts and promyelocytes and 4100  109/l platelets.13 Patients were treated on an Institutional Review Board (IRB)-approved protocol. Informed consent was obtained in accordance with the Declaration of Helsinki. Response criteria were as previously described.2 CHR was defined as a white blood cell count o10  109/l, a platelet count o450  109/l, no immature cells (blasts, promyelocytes and myelocytes) in the peripheral blood and disappearance of all signs and symptoms related to leukemia (including palpable splenomegaly). Response was further categorized by the best cytogenetic remission as complete (0% Ph þ ), partial (PCyR; 1–35% Ph þ ), minor (mCyR; 36–65% Ph þ ), and minimal (66–95% Ph þ ). MCyR included CCyR and PCyR (that is, p35% Ph þ ). Response rates were calculated based on the intent-to-treat population. Progression-free survival (PFS) was measured from the start of study drug to the earliest date of the following: progression to accelerated phase or blast phase, discontinuation due to progression (as assessed by investigator), or death from any cause on nilotinib therapy. Event-free survival (EFS) was measured from the start of study drug to the earliest date of the following: progression to accelerated phase or blast phase, discontinuation due to adverse events or laboratory abnormalities, discontinuation due to progression (as assessed by investigator), or death from any cause on nilotinib therapy. Patients for whom none of these events were reported were censored at the cutoff date or at the discontinuation date if they discontinued for other reasons not included in the composite end point. This trial was registered at http:// www.clinicaltrials.gov as NCT00109707.

Statistical considerations Twenty-one factors and covariates were prospectively defined for the modeling exercise, including 19 baseline factors (including the occurrence of grade 3/4 myelosuppression in the first 3 months) and two cytogenetic response factors based on previously published data.14,15 Cox proportional hazards regression models were used to investigate associations of PFS with 19 baseline patient and disease characteristics and two response end points. The two response factors included no mCyR (Ph þ metaphases 465%) by 6 months of nilotinib therapy and no MCyR (Ph þ metaphases 435%) by 12 months. The 19 baseline factors included presence of additional chromosomal abnormalities at baseline; age at study entry (years); duration of prior imatinib treatment (months); baseline CHR status (no baseline CHR (0) vs baseline CHR (1)); time from diagnosis of CML (months); prior highest imatinib dosage (mg); grade 3/4 myelosuppression (anemia, neutropenia and thrombocytopenia) event during the first 3 months from first dose; highest prior imatinib dose (o600 mg (0) vs others (1)); baseline mutation status (no mutation (0) vs any mutation [1] for patients with available mutation data); resistance to (0) vs intolerance of (1) imatinib therapy; hemoglobin at start of nilotinib therapy (g/l); best response to prior CML therapy; percent Ph þ metaphases at start of nilotinib; achievement of prior cytogenetic response; baseline alkaline phosphatase level; baseline percent of blasts in peripheral blood; baseline percent of basophils in peripheral blood; prior interferon-alfa therapy and sex. The effects of these factors on PFS, EFS and overall survival (OS) and the achievement of MCyR by 12 months of therapy were evaluated.

Table 1.

Cox proportional hazard regression modeling was used to identify significant factors associated with PFS. Landmark analyses, also using Cox proportional hazard regression models, were used to check that associations between PFS and achievement of cytogenetic response accounted for the possibility of guarantee-time bias.16–18 Dichotomized versions of two of the baseline continuous covariates (hemoglobin and basophils) were also considered as candidate factors. The cut points for dichotomizing these covariates were determined by examining the P-values associated with log-rank tests of PFS conducted over a range of cut points and selecting a rounded value of the cut point in the vicinity of the value that produced the minimum P-value. The regression modeling was used as a guide to develop a simple scoring system based on a small number of factors. Survival probabilities were estimated by the Kaplan–Meier method and compared by the log-rank test.

RESULTS Univariate analyses With a median follow-up of 28 months (range, 1–36 months), the incidence of MCyR was 60% and CCyR was 44%. The estimated 24month PFS rate was 64% and survival rate was 87%. Univariate analyses on the continuous baseline covariates and the dichotomous factors are shown in Tables 1 and 2, respectively. Of the continuous factors considered, low baseline hemoglobin level (hazard ratio (HR) ¼ 0.979; Po0.001), high baseline percent of basophils (HR ¼ 1.078; Po0.001) and high baseline percent of Ph þ metaphases (HR ¼ 1.013; Po0.0009) were significantly associated with the highest risk for progression. Among the dichotomous factors, imatinib resistance, no CHR at baseline, low baseline hemoglobin level, high baseline percent of basophils, high baseline percent of Ph þ metaphases, presence of a baseline mutation, presence of a baseline mutation with low sensitivity to nilotinib (E255K/V, Y253H, F359C/V), presence of a baseline mutation with IC50 X150 nM and highest prior dose of imatinib X600 mg were significantly associated with an unfavorable PFS. High baseline percent of Ph þ metaphases (when considered as a categorical variable with more than two categories) was significantly associated with an unfavorable PFS (HR ¼ 1.342; P ¼ 0.012) (Table 3). Both the 6-month mCyR (HR ¼ 1.342; P ¼ 0.012) and the 12-month MCyR (HR ¼ 0.219; Po0.001) were significantly associated with a better PFS (Table 3).

Multivariate analyses Multivariate analyses were conducted on the nine significant (Po0.05) baseline covariates, incorporating only the dichotomous versions of baseline hemoglobin level and baseline percent of basophils. Backward selection produced a model for PFS with two factors, baseline percent of basophils and baseline mutations with low sensitivity to nilotinib. When baseline hemoglobin level was included in the model, the hazard ratios and P-values were (Table 4):

Univariate analyses of the continuous baseline covariates on progression-free survival

Covariate Age (years) Baseline hemoglobin (g/l) Baseline basophils (%) Baseline blasts (%) Time since diagnosis (months) Highest prior imatinib dose (mg) Baseline Ph þ metaphases (%) Duration of prior treatment (months)

n

Median

Range

HR

P-value

315 313 313 304 315 314 302 265

58 121 2 0 58 600 100 26.87

21–85 77–172 0–29 0–12 5–275 300–1200 0–100 0.03–71

1.004 0.979 1.078 1.074 1.002 1.000 1.013 1.008

0.619 o0.001 o0.001 0.209 0.409 0.506 0.009 0.184

Abbreviation: HR, hazard ratio.

Leukemia (2013) 907 – 913

& 2013 Macmillan Publishers Limited

Factors predictive of patient outcome on nilotinib E Jabbour et al

909 Table 2. Univariate analyses of the effect of dichotomous factors on progression-free survival Level

n

%

HR

Male Female

157 158

50 50

0.928

0.722

Resistant Intolerant

222 93

70 30

0.530

0.015

No Yes

202 113

64 36

0.488

0.003

o120 X120

143 170

46 54

0.506

0.001

X4 o4

97 216

31 69

0.487

o0.001

Other chromosomal abnormalities

No Yes

233 82

74 26

1.296

0.251

BL mutation

No Yes

143 123

54 46

2.594

o0.001

BL mutation with low sensitivity to nilotinib

No Yes

288 27

91 9

10.153

o0.001

BL mutation with IC50 X150 nM

No Yes

286 29

91 9

8.197

o0.001

Prior cytogenetic response

No Yes

130 182

42 58

0.667

0.055

Highest prior dose of imatinib (mg)

o600 X600

86 228

27 73

1.717

0.042

Prior interferon-alfa

No Yes

124 191

39 61

1.356

0.166

No Yes

201 114

64 36

1.183

0.434

Dichotomous factor Sex Imatinib resistant/ intolerant BL CHR BL hemoglobin (g/l) BL basophils (%)

Grade X3 myelosuppression in first 3 months

P-value

Abbreviations: BL, baseline; CHR, complete hematologic response; HR, hazard ratio.

Baseline hemoglobin in g/l (o120 is 0, X120 is 1); HR ¼ 0.637; P ¼ 0.0458.  Baseline basophils percent (X4 is 0, o4 is 1); HR ¼ 0.580; P ¼ 0.0136.  Baseline mutations with low sensitivity to nilotinib (0 ¼ none or not known, 1 ¼ present); HR ¼ 7.125; Po0.0001. 

Landmark analyses Exploratory multivariate analyses that also incorporated the timedependent response outcomes (mCyR by 6 months and MCyR by 12 months) confirmed their association with PFS. Subset selection procedures identified MCyR by 12 months as more strongly associated with PFS than MCyR by 6 months. Landmark analyses from 12 months confirmed the association of PFS with MCyR by 12 months (Po0.001) and baseline mutations with low sensitivity to nilotinib (Po0.001). These results supported including MCyR by 12 months and nilotinib sensitivity mutations as part of the prognostic scoring system. In these landmark analyses, the dichotomized factors for baseline hemoglobin level and baseline percent of basophils & 2013 Macmillan Publishers Limited

appeared to be less strongly associated with PFS (P ¼ 0.277 and P ¼ 0.055, respectively). Nevertheless, these two baseline factors were included in the scoring system, because their importance as predictors of PFS before the availability of 12-month cytogenetic response assessments was previously established in published studies.13,19 These factors were also shown to be significant in the multivariate analyses presented here. The univariate and multivariate analyses in this study confirmed the effects of three baseline variables that significantly and consistently correlated with PFS among patients treated with nilotinib after imatinib failure. These results supported the inclusion of baseline hemoglobin, percent basophils and nilotinib sensitivity mutations into the prognostic scoring model that was developed and retrospectively evaluated as part of this study. The landmark analyses supported inclusion of MCyR by 12 months. Prognostic model Various factors have been previously described as independent prognostic factors of CML patient outcomes.13–15,19,20 Four of these factors were confirmed in this study as significant dichotomized prognostic markers and were used to create a scoring system that could serve as a model to further evaluate their effects on patient outcomes. These four factors were baseline BCR-ABL mutations associated with low pharmacological sensitivity to nilotinib, baseline hemoglobin o120 g/l, baseline basophils X4% and drug resistance as defined by lack of MCyR by 12 months. Patients were assigned a score of 4 and then 1 was subtracted for the presence of each of the favorable factors. Scores potentially ranged from 0 to 4, with 0 being the best (that is, no unfavorable factors present) and 4 being the worst (that is, four unfavorable factors present). In the rare cases (3 of 315) in which patients had missing values for a factor, their scores were not reduced. In the analysis of all four factors, the majority of patients scored 0 (n ¼ 77), 1 (n ¼ 116) or 2 (n ¼ 82). A restricted score, or ‘baseline score,’ was also calculated in the same way using only the three baseline factors (that is, excluding MCyR status by 12 months). The majority of patients in the analysis of only baseline factors scored 0 (n ¼ 109) or 1 (n ¼ 153). The Kaplan–Meier estimates of PFS and OS in the four-factor score groups are shown in Table 5 and Figures 1a and b. As hypothesized, the 2-year PFS rates strongly correlated with the score groups, in which PFS was best for patients with the lowest scores (score ¼ 0, no unfavorable factors present) and decreased as the scores increased (Figure 1a). The nine patients in the ‘worst’ group (score ¼ 4) did not reach the 12-month landmark: eight patients experienced a PFS event and one was censored before 12 months. The 2-year PFS rates were 89%, 67%, 50%, 19% and 0% for patients with scores of 0, 1, 2, 3 and 4, respectively. The 2-year OS rates were also significantly better among patients with a score of 0 (99%; Po0.001) when compared with the rates of patients with scores of 1, 2, 3 or 4 (88%, 76%, 82% and 87%, respectively). As one of the factors, MCyR status by 12 months, was not known at baseline, an analysis was conducted considering only the baseline factors: hemoglobin level, percent basophils and mutation status (Table 5 and Figures 1c and d). Again, the 2-year PFS rates were best for patients with the lowest baseline scores (score ¼ 0, no unfavorable baseline factors present) and worst for patients with the highest baseline score (score ¼ 3, all three unfavorable baseline factors present) (Figure 1c). The 2-year PFS rates were 79%, 65%, 39% and 0% for patients with scores of 0, 1, 2 and 3, respectively. The 2-year OS rates also appeared to be higher among patients with a score of 0 (93%) vs patients with scores of 1, 2 or 3 (84%, 81% and 90%, respectively), but this difference was not statistically significant (P ¼ 0.06; Figure 1d). This baseline model also predicted for the 12-month probability of achieving MCyR. Similar to the analyses of PFS, the 12-month probability of achieving MCyR significantly correlated with Leukemia (2013) 907 – 913

Factors predictive of patient outcome on nilotinib E Jabbour et al

910 Table 3.

Univariate analyses of the effects of ordinal multilevel factors and response outcomes on progression-free survival Level

N

%

HRa

0 40 and p35 435 and p65 465 and p95 495 and p100

9 25 18 75 175

3 8 6 25 58

1.342

0.012

No CHR CHR but no CyR Minimal CyR Minor CyR PCyR CCyR

27 103 19 28 57 77

9 33 6 9 18 26

0.902

0.074

Response outcome Minor cytogenetic response by 6 months

Level No Yes

N 123 192

% 39 61

HRb 0.255

P-value o0.001

Major cytogenetic response by 12 months

No Yes

138 177

44 56

0.219

o0.001

Ordinal factor Baseline Ph þ metaphases (%)

Prior best response to therapy

P-value

Abbreviations: CCyR, complete cytogenetic response; CHR, complete hematologic response; CyR, cytogenetic response; HR, hazard ratio; PCyR, partial cytogenetic response; Ph þ , Philadelphia chromosome-positive. aHR for a unit increase in the category number (according to ordering given in table). bHR for the second level of the factor relative to the first level.

Table 4.

Table 5. Kaplan–Meier estimates of 24-month PFS and OS and 12month MCyR in the overall and baseline score groups

Multivariate analysis for progression-free survival

Variable Baseline hemoglobin, (g/l) Baseline basophils in peripheral blood (%) Baseline BCR-ABL mutations with low sensitivity to nilotinib (0 ¼ none or unknown, 1 ¼ present)

P-value 0.0458 0.0136 o0.0001

Hazard ratio 0.637 0.580 7.125

prognostic score. MCyR by 12 months was achieved by 71%, 55%, 32% and 25% of patients with scores of 0, 1, 2 and 3, respectively, (Po0.001; Table 5). PFS and OS rates from the 12-month landmark analyses in the score groups are shown in Table 6 and Figures 2a and b. The nine patients in the worst prognostic group (score ¼ 4) did not reach the 12-month landmark: eight patients experienced a PFS event and one was censored before 12 months. Thus, this category was not included in the landmark analysis for PFS but was included in the landmark analysis for OS (because all patients were followed for survival). Overall, the 2-year PFS and OS rates were highest for patients with the lowest scores and lowest for patients with the highest score (Figures 2a and b). Patients with a score of 3 (n ¼ 13) had lower PFS rates than the other score groups at all time points, particularly compared with patients with a score of 0 (Figure 2a). Although 2-year OS rates were relatively similar in the 1, 2, 3 and 4 score groups, OS was 100% among patients with a score of 0 (Figure 2b). Achievement of MCyR by 12 months was a critical prognostic factor (Table 6). The 2-year PFS rates among patients with scores of 0, 1 or 2 who had MCyR by 12 months vs those who did not were 91% vs 48%, 79% vs 66% and 78% vs 31%, respectively. Score group 3 was not included in this analysis because of its small sample size (n ¼ 2). We repeated the analysis assessing the validity of this score on EFS. This model also predicted for EFS. The 2-year EFS rates were 81%, 50%, 29%, 8% and 0% for patients with scores of 0, 1, 2, 3 and 4, respectively, (Po0.001). Leukemia (2013) 907 – 913

PFS and OS according to score groupsa % at 24 months (95% CI)a Score 0 1 2 3 4

N (%)

PFS

OS

77 (24) 116 (37) 82 (26) 31 (10) 9 (3)

89 (78–94) 67 (56–76) 50 (35–63) 19 (5–41) 0 (NA)

99 (91–100) 88 (80–93) 76 (65–84) 82 (61–92) 87 (39–98)

PFS, OS and MCyR according to baseline score groupsb % at 24 months (95% CI) Score 0 1 2 3

N (%)

PFS

109 (35) 153 (49) 41 (13) 12 (4)

79 (68–86) 65 (55–73) 39 (21–57) 0 (NA)

OS 93 84 81 90

(86–97) (77–89) (64–91) (47–99)

% at 12 Months MCyR 71 55 32 25

Abbreviations: CI, confidence interval; MCyR, major cytogenetic response; NA, not applicable; OS, overall survival; PFS, progression-free survival. a Adverse features were hemoglobin o120 g/l, percent of basophils X4%, baseline mutations with low sensitivity to nilotinib and no MCyR by 12 months of therapy. bAdverse features were hemoglobin o120 g/l, percent of basophils X4% and baseline mutations with low sensitivity to nilotinib.

DISCUSSION The availability of second-generation TKIs has provided new therapeutic options for patients with CML after imatinib failure. This is the largest study that has evaluated prognostic factors predicting response to nilotinib in patients with CML-CP resistant to or intolerant of imatinib therapy. We described two prognostic scores that can be used to predict PFS in this patient population. Both can be & 2013 Macmillan Publishers Limited

Factors predictive of patient outcome on nilotinib E Jabbour et al

911

Figure 1. Survival of patients by overall scores according to the prognostic model. Kaplan–Meier estimates of PFS by score groups (a); overall survival (OS) by score groups (b); PFS by baseline component of scores (c), which excludes major cytogenetic response by 12 months as a factor because it is not known at baseline, and OS by baseline component of scores (d), which excludes major cytogenetic response by 12 months as a factor because it is not known at baseline. Table 6.

Associations of PFS and OS with MCyR status by the 12month landmark analysis in the baseline score groupsa MCyR at 12 months

N (for PFS)

% at 24 months (95% CI)

N (for OS)

PFS

% at 24 months (95% CI) OS

Score ¼ 0 No Yes

11 69

48 (16–75) 91(81–96)

27 74

84 (62–94) 100

Score ¼ 1 No Yes

27 68

66 (44–82) 79 (67–87)

55 83

87 (74–93) 93 (84–97)

Score ¼ 2 No Yes

11 9

31 (8–59) 78 (37–94)

27 13

83 (60–93) 85 (51–96)

Abbreviations: CI, confidence interval; MCyR, major cytogenetic response; OS, overall survival; PFS, progression-free survival. Baseline adverse features were hemoglobin o120 g/l, percent of basophils X4% and baseline mutations with low sensitivity to nilotinib. aNo analysis is presented for baseline score ¼ 3 because the sample size was too small (n ¼ 2).

used to predict PFS, but at different times (one at baseline and one after 12 months). The first is based on three baseline factors that were identified through univariate and multivariate analyses. The second & 2013 Macmillan Publishers Limited

prognostic score includes achievement of MCyR by 12 months in addition to those three factors. This four part score can be used when 12-month response results are known. The four independent factors identified by our study as predictive of outcomes were the achievement of MCyR by 12 months of therapy (HR ¼ 0.359; Po0.0001), baseline hemoglobin level X120 g/l (HR ¼ 0.985; P ¼ 0.0291), baseline percent of basophils o4% (HR ¼ 1.070, P ¼ 0.0128), and the absence of baseline mutations with low sensitivity to nilotinib (E255K/V, Y253H and F359C/V) (HR ¼ 6.097; Pp0.0001). These findings are consistent with previous experience with imatinib therapy after interferon failure, in which more aggressive forms of disease (for example, high basophil levels, lack of CHR and high percent of Ph þ metaphases) were associated with poor response to imatinib therapy, suggesting the presence of aggressive clones with intrinsic resistance to therapy.13 Our findings are in line with a previous report from the MD Anderson Cancer Center where patients with high IC50 kinase domain mutations treated with nilotinib had a low likelihood of achieving MCyR, with a negative impact on EFS and OS.20 Recently, a multivariate analysis investigating baseline factors predictive of cytogenetic response was reported for patients receiving dasatinib after imatinib failure.21 In this analysis, age, lower percent of Ph þ metaphases, shorter duration of CML, longer duration of prior imatinib therapy, prior Leukemia (2013) 907 – 913

Factors predictive of patient outcome on nilotinib E Jabbour et al

912

Figure 2.

Survival of patients according to the 12-month landmark analysis. Kaplan–Meier estimates of PFS (a) and OS (b) by score groups.

response or intolerance to imatinib, absence of splenomegaly and no prior allogeneic stem cell transplantation predicted for the achievement of MCyR and CCyR. Because of the inclusion of different baseline factors and statistical end points, comparisons between the multivariate analyses of dasatinib and nilotinib cannot be made. In patients with 12-month cytogenetic results available, the four factors were combined to create a single prognostic score that stratified patients into five prognostic subsets. The best groups were patients having none of the three unfavorable disease risk factors and achieving MCyR by 12 months. Patients in the worst group had all three of the unfavorable disease risk factors and did not achieve MCyR by 12 months. This is a simple scoring system that is easily applicable in the clinic, provided that patients were adequately monitored during imatinib therapy, allowing assessment of their response to imatinib. This scoring system could serve to advise patients of their prognosis and treatment options. An important question is whether these factors can predict for significant differences in PFS (for example, subsets with expected 24-month PFS 475% vs o50%). If such is the case, then patients in the favorable group (no adverse factors at baseline) might rely on second-generation TKIs indefinitely, whereas those in the unfavorable group (three adverse factors) would be well advised to pursue an stem cell transplantation at the time of imatinib failure. A trial of nilotinib could be initiated while the patient is preparing for transplant; however, in view of the poor long-term prognosis, stem cell transplantation would be recommended when available. In contrast, in patients with one or two baseline adverse features, close monitoring is needed, and alternative options could be considered, particularly if response is suboptimal after 12 months of therapy with nilotinib. In addition, these factors are not applicable for patients with imatinib intolerance and those in advanced-stage disease. Achieving 12-month MCyR is a known major determinant of outcome in previous generations of therapy including interferonalfa and imatinib,5 and in patients treated with second-generation TKIs.14 The 12-month landmark analysis showed that achieving MCyR by 12 months was the most dominant independent predictive factor for PFS after the start of therapy (Table 6) and may compensate for the presence of unfavorable baseline factors, such as kinase domain mutations with low sensitivity to nilotinib therapy that remain uncommon. This is in line with the results of the pivotal trial in which the 2-year PFS rates were 94% and 79% in patients with and without 12-month MCyR, respectively, and is consistent with a smaller retrospective analysis from the MD Leukemia (2013) 907 – 913

Anderson Cancer Center, where the achievement of 12-month MCyR constituted the sole independent predictive factor for EFS.19 In summary, the outcome of patients after imatinib failure treated with nilotinib could be predicted. Patients with anemia, high proportion of basophils in peripheral blood or a kinase domain mutation with low sensitivity to nilotinib have a poor PFS and OS when treated with nilotinib and could be offered additional treatment options. The achievement of 12-month MCyR could compensate for the presence of these unfavorable adverse features. Patients achieving MCyR after 12 months of therapy can continue on nilotinib; those not achieving it would consider alternative options. This decision will also depend on other variables, such as the age of the patient, and donor matching. The scoring system described here needs to be prospectively evaluated and validated in future clinical trials.

CONFLICT OF INTEREST EJ received honoraria from Novartis and BMS. PDlC acted as a consultant and received honoraria for Novartis and BMS and received research funding from Novartis. JEC acted as a consultant for Novartis, BMS and Pfizer and received research funding from Novartis, BMS, Pfizer, Ariad and Chemgenex. FJG acted as a consultant, received honoraria and research funding from Novartis. KNB received honoraria and research funding from Novartis. JP-I acted as a consultant for Novartis and BMS and received honoraria from Novartis. RAL acted as a consultant, received honoraria, and received research funding from Novartis. NG received honoraria and research funding from Novartis. OGO acted as a consultant, received honoraria, and research funding from Novartis. AH acted as a consultant for Novartis, BMS, Pfizer and Ariad, and received honoraria and research funding from Novartis, BMS and Pfizer. TPH acted as a consultant and received research funding from Novartis, BMS and Ariad. GS acted as a consultant for Novartis, BMS and Pfizer and received honoraria from BMS and Novartis. JPR acted as a consultant for Novartis, BMS, Ariad and Pfizer and received research funding from Novartis. D-WK received honoraria from Novartis and BMS and received research funding from Novartis, BMS, Pfizer and Ariad. GM acted as a consultant for Novartis, BMS, Pfizer and Genzyme, and received honoraria from Novartis and BMS; and research funding from Novartis. JR and RCW are Novartis employees and stock owners. MB acted as a consultant for Novartis, BMS and Pfizer, and received honoraria from Novartis, BMS and Pfizer, and received research funding from Novartis. HMK acted as a consultant for Novartis and received research funding from Novartis, BMS and Pfizer.

ACKNOWLEDGEMENTS Financial support for medical editorial assistance was provided by Novartis Pharmaceuticals. We thank Michael Mandola, PhD for medical editorial assistance with this manuscript.

& 2013 Macmillan Publishers Limited

Factors predictive of patient outcome on nilotinib E Jabbour et al

913 AUTHORS CONTRIBUTIONS EJ, JEC, FJG, JP-I, OGO, AH, TPH, JPR, D-WK, GM and HMK designed the study; HMK provided administrative support; EJ, PDlC, JC, FJG, JP-I, RAL, NG, OGO, TPH, JPR, D-WK, GM, MB and HMK provided study materials; EJ, PDlC, KNB, NG and AH collected and assembled data; EJ, JEC, AH, GS, D-WK, JR, RCW, MB and HMK analyzed and interpreted data; and all authors drafted/approved the manuscript.

11

12

13

REFERENCES 1 Jabbour E, Cortes JE, Giles FJ, O’Brien S, Kantarjian HM. Current and emerging treatment options in chronic myeloid leukemia. Cancer 2007; 109: 2171–2181. 2 Kantarjian H, Sawyers C, Hochhaus A, Guilhot F, Schiffer C, Gambacorti-Passerini C et al. Hematologic and cytogenetic responses to imatinib mesylate in chronic myelogenous leukemia. N Engl J Med 2002; 346: 645–652. 3 O’Brien SG, Guilhot F, Larson RA, Gathmann I, Baccarani M, Cervantes F et al. Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia. N Engl J Med 2003; 348: 994–1004. 4 Cortes J, Talpaz M, O’Brien S, Jones D, Luthra R, Shan J et al. Molecular responses in patients with chronic myelogenous leukemia in chronic phase treated with imatinib mesylate. Clin Cancer Res 2005; 11: 3425–3432. 5 Kantarjian HM, Talpaz M, O’Brien S, Jones D, Giles F, Garcia-Manero G et al. Survival benefit with imatinib mesylate versus interferon-alpha-based regimens in newly diagnosed chronic-phase chronic myelogenous leukemia. Blood 2006; 108: 1835–1840. 6 Deininger M, O’Brien SG, Guilhot F, Goldman JM, Hochhaus A, Hughes TP et al. International Randomized Study of Interferon vs STI571 (IRIS) 8-year follow up: sustained survival and low risk for progression or events in patients with newly diagnosed chronic myeloid leukemia in chronic phase (CML-CP) treated with imatinib. ASH Ann Meeting Abstracts 2009; 114: 1126. 7 Shah NP. Loss of response to imatinib: mechanisms and management. Hematol Am Soc Hematol Educ Program 2005 183–187. 8 Hochhaus A, Hughes T. Clinical resistance to imatinib: mechanisms and implications. Hematol Oncol Clin North Am 2004; 18: 641–656, ix. 9 Tasigna (nilotinib) prescribing information. Novartis Pharmaceuticals Corporation: East Hanover, NJ, 2012. 10 Kantarjian HM, Giles F, Gattermann N, Bhalla K, Alimena G, Palandri F et al. Nilotinib (formerly AMN107), a highly selective BCR-ABL tyrosine kinase inhibitor, is effective in patients with Philadelphia chromosome-positive chronic

& 2013 Macmillan Publishers Limited

14

15

16 17

18 19

20

21

myelogenous leukemia in chronic phase following imatinib resistance and intolerance. Blood 2007; 110: 3540–3546. Kantarjian HM, Giles FJ, Bhalla KN, Pinilla-Ibarz JA, Larson RA, Gattermann N et al. Nilotinib is effective in patients with chronic myeloid leukemia in chronic phase following imatinib resistance or intolerance: 24-month follow-up results. Blood 2011; 117: 1141–1145. Baccarani M, Cortes J, Pane F, Niederwieser D, Saglio G, Apperley J et al. Chronic myeloid leukemia: an update of concepts and management recommendations of European LeukemiaNet. J Clin Oncol 2009; 27: 6041–6051. Kantarjian HM, Talpaz M, O’Brien S, Smith TL, Giles FJ, Faderl S et al. Imatinib mesylate for Philadelphia chromosome-positive, chronic-phase myeloid leukemia after failure of interferon-alpha: follow-up results. Clin Cancer Res 2002; 8: 2177–2187. Tam CS, Kantarjian H, Garcia-Manero G, Borthakur G, O’Brien S, Ravandi F et al. Failure to achieve a major cytogenetic response by 12 months defines inadequate response in patients receiving nilotinib or dasatinib as second or subsequent line therapy for chronic myeloid leukemia. Blood 2008; 112: 516–518. Milojkovic D, Nicholson E, Apperley JF, Holyoake TL, Shepherd P, Drummond MW et al. Early prediction of success or failure of treatment with second-generation tyrosine kinase inhibitors in patients with chronic myeloid leukemia. Haematologica 2010; 95: 224–231. Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. J Clin Oncol 1983; 1: 710–719. Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response and other comparisons of time-to-event by outcome variables. J Clin Oncol 2008; 26: 3913–3915. Buyse M, Piedbois P. On the relationship between response to treatment and survival time. Stat Med 1996; 15: 2797–2812. Jabbour E, Kantarjian H, O’Brien S, Shan J, Garcia-Manero G, Wierda W et al. Predictive factors for outcome and response in patients treated with secondgeneration tyrosine kinase inhibitors for chronic myeloid leukemia in chronic phase after imatinib failure. Blood 2011; 117: 1822–1827. Jabbour E, Jones D, Kantarjian HM, O’Brien S, Tam C, Koller C et al. Long-term outcome of patients with chronic myeloid leukemia treated with second-generation tyrosine kinase inhibitors after imatinib failure is predicted by the in vitro sensitivity of BCR-ABL kinase domain mutations. Blood 2009; 114: 2037–2043. Jabbour E, Bahceci E, Zhu C, Lambert A, Cortes J. Predictors of long-term cytogenetic response following dasatinib therapy of patients with chronicphase chronic myeloid leukemia (CML-CP). ASH Ann Meeting Abstracts 2009 114: 3296.

Leukemia (2013) 907 – 913