Exposure ResponseRelationshipsoftheEfﬁcacyandSafety...

Cancer Therapy: Clinical

Exposure–ResponseRelationships of the Efficacy and Safetyof Ipilimumab in Patients with Advanced Melanoma

Yan Feng1, Amit Roy1, Eric Masson1, Tai-Tsang Chen2, Rachel Humphrey1, and Jeffrey S. Weber3

AbstractPurpose: This retrospective analysis was conducted to characterize ipilimumab exposure–response

relationships for measures of efficacy and safety in patients with advanced melanoma.

Experimental Design: Data were pooled from 498 patients who received ipilimumab monotherapy at

0.3, 3, or 10 mg/kg in 1 of 4 completed phase II clinical trials. The relationships between steady–state

ipilimumab trough concentration (Cminss), complete or partial tumor response (CR or PR), and safety

[immune-related adverse events (irAEs)] were described by logistic regression models. The relationship

between exposure and overall survival was characterized using a Cox proportional–hazards model.

Results: The steady-state trough concentration of ipilimumabwas found to be a significant predictor of a

CR or PR (P < 0.001). Model-based estimates indicate that the probabilities of a CR or PR at median Cminss

for the 0.3, 3, and 10mg/kg groupswere 0.6%, 4.9%, and11.6%, respectively.Overall survival at themedian

Cminss for ipilimumab at 0.3 mg/kg was estimated to be 0.85- and 0.58-fold lower relative to that at the

median Cminss for 3 and 10mg/kg, respectively. Model-based estimates indicate that the probabilities of a

grade 3 or more irAE at the median Cminss for the 0.3, 3, and 10 mg/kg doses were 3%, 13%, and 24%,

respectively.

Conclusions:Higher doses of ipilimumab produce greater Cminss thatmay be associatedwith increased

tumor responses, longer survival, and higher rates of irAEs. The efficacy and safety of ipilimumab at 3 versus

10 mg/kg in patients with advanced melanoma is being evaluated in an ongoing phase III trial. Clin Cancer

Res; 19(14); 3977–86. �2013 AACR.

IntroductionCytotoxic T-lymphocyte antigen-4 (CTLA-4) is a key

immune checkpoint molecule that downregulates T-cellactivation by binding to its ligands, B7-1 (CD80) and B7-2 (CD86), on antigen-presenting cells (1). Ipilimumab is afully human IgG1monoclonal antibody that blocks CTLA-4from binding to its ligands, thereby augmenting antitumorimmune responses (2). Consistent with its proposedmech-anismof action, ipilimumabhas been shown to increase thepercentage of activated CD4þ and CD8þ T cells in periph-eral circulation, with a concomitant decrease in na€�ve CD4þ

and CD8þ T cells, in patients with advanced (unresectablestage III or IV) melanoma (3). Ipilimumab has shown a

statistically significant improvement in overall survival(OS) in 2 randomized controlled, phase III trials of patientswith advanced melanoma; one with ipilimumab mono-therapy at 3mg/kg in previously treated patients (4) and theother with ipilimumab at 10 mg/kg in combination withdacarbazine in previously untreated patients (5).

Ipilimumab monotherapy has been extensively studiedin phase II clinical trials of patients with advanced mela-noma, most of whom had received prior therapy for met-astatic disease (6–8). The results of one of these trials(CA184-022) suggested that ipilimumab elicits a dose-dependent effect on both efficacy and safety endpoints,where the best overall response (BOR) rates were 11.1%,4.2%, and 0% for ipilimumab at doses of 10, 3, and 0.3mg/kg, respectively (6). Across ipilimumab clinical trials, themost common treatment-related adverse events wereimmune-related (irAEs; refs. 4–8), i.e., adverse events thatare inflammatory innature and consistentwith an immune-basedmechanism of action (2, 9). The incidence of irAEs ofany grade, and irAEs of grade 3 and 4, showed a dose-dependent increase with ipilimumab in study CA184-022(6). These efficacy and safety outcomes correlated with theresults of population pharmacokinetic analyses, whichshowed greater achievement of the target trough concen-tration during the induction dosing period (every 3 weeksfor 4 doses) for ipilimumab at 10 versus 3 mg/kg, and withthe results of pharmacodynamic analyses, which showed a

Authors' Affiliations: 1Bristol-Myers Squibb, Princeton, New Jersey;2Bristol-Myers Squibb, Wallingford, Connecticut; and 3Moffit Cancer Cen-ter, Tampa, Florida

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Current address for R. Humphrey: MethylGene Inc., Montreal, Quebec,Canada.

Corresponding Author: Jeffrey S. Weber, Moffitt Cancer Center, 12902Magnolia Drive SRB-2, Tampa, FL 33612. Phone: 813-745-2007; Fax: 813-449-8260; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-12-3243

�2013 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 3977

on June 9, 2018. © 2013 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 5, 2013; DOI: 10.1158/1078-0432.CCR-12-3243

http://clincancerres.aacrjournals.org/

greater mean rate of increase in absolute lymphocyte count(ALC) at 10 mg/kg compared with lower doses (6).

Ipilimumab continues to be evaluated in melanoma andother tumor types, including castration-resistant prostatecancer and non–small cell lung cancer (2, 10, 11). To betterunderstand the risk-benefit profile of ipilimumab mono-therapy across the dose range evaluated in clinical studies,we conducted a retrospective analysis of pooled data fromphase II trials including all patients for whom pharmaco-kinetic samples were available.More than 450 patients withadvanced melanoma were included in the analyses, whichwere done to characterize ipilimumab exposure–responserelationships for important measures of efficacy and safety.Specifically, the exposure–response relationships for tumorresponse and overall survival were analyzed to characterizeefficacy and those for irAEs were analyzed to characterizesafety.

Materials and MethodsPatients

The patients included in the current analyses had partic-ipated in one of the following 4 completed phase II clinicaltrials which evaluated ipilimumab monotherapy in ad-vanced melanoma: CA184-004, bmyCA184-007, CA184-008, and CA184-022 (6–8, 12). Study CA184-004 investi-gated potential biomarkers of clinical activity in patientswho received ipilimumab at 3 or 10 mg/kg (12), and studyCA184-007 evaluated the impact of prophylactic oral bude-sonide on the rate of grade 2 or more diarrhea in patientswho received ipilimumab at 10 mg/kg (7). Both of thesestudies included previously treated and previously untreat-ed patients. In study CA184-022 (6), patients intolerant toor who progressed on prior therapy were randomized toipilimumab at 0.3, 3, or 10mg/kg, whereas in studyCA184-

008, only one dose of ipilimumab (10 mg/kg) was evalu-ated in previously treated patients (8). In all 4 of thesestudies, ipilimumab was given every 3 weeks for up to 4doses (induction phase), followed by a maintenance phase(every 12 weeks beginning at week 24) in eligible patients.The first tumor assessment was carried out at week 12 (endof induction dosing). All patients, or their legal representa-tives, gave written informed consent to participate in theirrespective trials. A summary of patient demographics andlaboratory values for the exposure–response analyses isgiven in Supplementary Table S1.

Data analysisThe exposure–response analysis of efficacy was charac-

terized by 2measures of tumor response and byOS, and theexposure–response analysis of safety included toxicitiescharacterized as irAEs. The twomeasures of tumor responsewere: (i) BOR of partial response (PR) or complete response(CR) by mWHO criteria, and (ii) clinical activity of PR andCR by immune-related response criteria (irRC), an explor-atory endpoint defined to capture the delayed tumorresponses that are sometimes observed with immunother-apy (13). The irRC are derived from mWHO criteria andinclude new lesions in the assessment of total tumor bur-den; in contrast to standard criteria, the irRC do not nec-essarily characterize patients with new lesions as havingprogressive disease (13).

The exposure–response analysis of BOR by mWHO cri-teria was conducted with data from patients in studiesCA184-007, CA184-008, and CA184-022 for whom sum-mary measures of ipilimumab exposure and BOR assess-ment were available (N ¼ 354; �73% of the total treated).Data from study CA184-004 were not included in theexposure–response analyses of BOR as tumor assessmentsin this trial were not adjudicated by an independent reviewcommittee (IRC), and patients who underwent biopsy ofthe index lesions were censored for efficacy. Similarly, theexposure–response analysis of clinical activity by irRC wasconducted with data from patients in studies CA184-007,CA184-008, and CA184-022 for whom summary measuresof ipilimumab exposure and the assessment of clinicalactivity were available [N ¼ 419; 86% of the total numberof patients treated/randomized (487)].

The exposure–response relationships for efficacy (OS)and safety (irAE) were analyzed with data from patients instudies CA184-004, CA184-007, CA184-008, and CA184-022 for whom summary measures of ipilimumab exposurewere available [N ¼ 498; 88% of the total number ofpatients treated/randomized (569)]. The safety endpointreflected the worst reported grade of an irAE in any one ofthe categories: gastrointestinal, hepatic, skin, endocrine,and other.

The measure of ipilimumab exposure employed in theexposure–response analyses was ipilimumab steady-statetrough concentration (Cminss) during the induction phase,as determined by a previously developed population phar-macokinetic model (6), using data from the 4 phase IIstudies. Pharmacokinetic concentrations in samples from

Translational RelevanceIpilimumab monotherapy at 3 mg/kg is currently

approved in several countries for the treatment ofadvanced (unresectable stage III or IV)melanoma, basedon the results of a phase III, randomized controlled trial(MDX010-20) in which ipilimumab at 3 mg/kg showeda statistically significant and clinically meaningfulimprovement in overall survival in previously treatedpatients. We conducted a retrospective analysis of datafrom 4 phase II studies, investigating the relationshipbetween ipilimumab exposure and measures of efficacy(including overall survival) as well as safety. These anal-yses include survival data from a wide dose range (0.3–10mg/kg) that were not available when studyMDX010-20 was initiated, and suggest that higher ipilimumabexposure is associated with better survival, albeit at agreater risk of adverse events. A formal comparison ofipilimumab monotherapy at 3 versus 10 mg/kg is cur-rently being evaluated in a separate phase III trial.

Feng et al.

Clin Cancer Res; 19(14) July 15, 2013 Clinical Cancer Research3978




patients enrolled in the 4 phase II studies were analyzed by avalidated ELISA assay that had a lower limit of quantifica-tion of 0.4 mg/mL (14). Model validation in a previouslypublished linear 2-compartment PPK model with zero-order intravenous infusion and first-order elimination haveshown that it provided a good description of the entireipilimumab serum concentration-time profile for the 3 and10 mg/kg doses (15). The observed and model-predictedPK profile for patients who received ipilimumab at 3 or 10mg/kg is presented in Supplementary Fig. S1. Steady-statepeak and time-averaged concentrations were not selectedfor these analyses, as they are highly correlated with Cminss(correlation coefficient > 0.9). Moreover, the pharmacolog-ic rationale for Cminss was considered to be stronger thanthe other 2 summary measures of exposure, based on theassumption that T-cell activation in the induction phase oftreatment is maximized by blocking the interaction ofCTLA-4 with its ligands over the entire dosing interval.

Exposure–response analysis: tumor responseThe exposure–response analysis of tumor response was

characterized by proportional-odds logistic regressionmodels relating Cminss to the probability of achieving BORof (CR or PR), as defined by mWHO criteria as well asclinical activity by irRC, and based on IRC adjudicatedtumor sizemeasurements. The logit (log-odds) are given by:

logit Pri BOR ¼ logPi BOR

1� Pi BOR

� �¼ b0 þ bTXi;

where b0 and b are scalar and vector parameters thatrepresent baseline odds and the effect of Xi on achievingBOR or clinical activity, respectively. Themodel parameterswere estimated by maximum likelihood. This model for-mulation assumes that the predictor variables Xi haveproportional effects on the odds of Pr(BOR) or Pr(clinicalactivity).Model development was conducted in 3 stages. First, a

base model was developed to establish the existence andfunctional form of a relationship between ipilimumabCminss and probability of a tumor response. Second, thecovariate effects that may potentially modulate the expo-sure–response relationship were examined in a full model.Third, the final model was developed by backward elimi-nation to retain only the covariates that were significant at a1% level. The final model was evaluated by assessing theagreement between the observed and predicted proportionof tumor response and the associated 90% model predic-tion intervals (PI) in each dose group (0.3, 3, and 10 mg/kg). Themodel-predicted proportions (and associated 90%PI) in each dose group were obtained by simulation usingthemodel- predicted probability of response of the patientsin each dose group.The full model examined the potential effects of the

following covariates: body weight, age, gender, EasternCooperative Oncology Group performance status (ECOGPS), baseline serum lactate dehydrogenase (LDH) levels,concomitant budesonide, prior systemic anticancer thera-

py, metastatic stage (M-stage), prior immunotherapy, HLA-A�0201 status, and prior interleukin (IL)-2 therapy.

Exposure–response analysis: OSThe exposure–response analysis of OS was characterized

by a Cox proportional–hazards (CPH) model relating ipi-limumab Cminss to the hazard of death. The hazard func-

tion is expressed aslðtÞ ¼ l0ðtÞ expðbTxiÞ, wherel0(t) is thebaseline hazard function and xi is a vector of predictorvariables. The parameter vector b is estimated bymaximumpartial likelihood.

We also assessed the potential effects of the followingcovariates on the exposure–response relationship: age,weight, gender, baseline ALC, HLA-A�0201 status, baselineLDH levels, prior systemic anticancer therapy, prior immu-notherapy, prior IL-2 therapy, ECOG PS, and M-stage atstudy entry.

The CPH model was developed in 3 stages: (i) a basemodel was developed to establish the existence and func-tional form of the exposure–response relationship betweenOS and ipilimumab exposure (Cminss); (ii) a full modelwas developed to assess the effect of all the covariatessimultaneously; (iii) the final model was developed byretaining the covariates that were significant at a 1% level,with appropriate functional forms of their relationshipswith OS. The CPH model was evaluated by comparingmodel-predicted cumulative probability of OS versus timewith that obtained by Kaplan–Meier (KM) analyses.

Exposure–response analysis: irAEsThe exposure–response of irAEs was characterized by a

proportional-odds ordinal logistic regression model relat-ing Cminss to the probability of experiencing grade 2 ormore or grade 3 ormore irAEs. The logit (log-odds) of grade2 or more and grade 3 or more irAEs are given by:

logit Pri Grade�2 ¼ logPi Grade�2

1� Pi Grade�2

� �

¼ bGrade2 þ bTXi; Eq:1

logit Pri Grade�3 ¼ logPi Grade�3

1� Pi Grade�3

� �

¼ bGrade2 þ bGrade3 þ bTXi; Eq:2

where bGrade2, bGrade3 and b are scalar and vector para-meters that represent baseline odds of grade 2 or more andgrade 3 or more irAEs and the effect of Xi [i.e., Cminss, log(Cminss)] on achieving irAEs, respectively. The modelparameters were estimated by maximum likelihood. Thismodel formulation assumes that the predictor variables Xihave proportional effects on the odds of probability ofirAEs.

Model development was conducted in 3 stages, similarto that for the exposure–response analysis of tumorresponse. Covariate-parameter relationships were exam-ined for the baseline covariates of body weight, age, gender,ECOG PS, baseline LDH levels, concomitant budesonide,prior systemic anticancer therapy, metastatic stage, prior

Ipilimumab Exposure–Response Relationships in Advanced Melanoma

www.aacrjournals.org Clin Cancer Res; 19(14) July 15, 2013 3979




immunotherapy, HLA status, and prior IL-2 therapy. Thefinal model was evaluated by assessing the agreementbetween the observed proportion of irAEs by dose and theassociated 90% model PI.

ResultsExposure–response analysis: BOR by mWHO

Cminss was found to be a statistically significant predic-tor of theprobability of aBOR(CRorPR)bymWHOcriteria(P < 0.001), and the probability was found to increase withan increase in log–transformed Cminss (Table 1 and Fig.1A). However, none of the covariates examined wereretained in the final model as none had statistically signif-icant effect on the exposure–response relationship for BORby mWHO criteria.

The odds of a patient achieving a BOR bymWHO criteriaincreased by 2.41-fold for a 2.7-fold increase in Cminss [log(Cminss) increases by one unit for approximately 2.7-foldincrease in Cminss]. The observed responders at 0.3, 3, and10mg/kg were 0%, 6%, and 11.7%, respectively, consistentwith model-estimated probabilities of a BOR at medianCminss for 0.3, 3, and 10mg/kg groups of 0.6%, 4.9%, and11.6%, respectively. Figure 1A shows the results of modelevaluation that compares observed and predicted propor-tion of mWHO responders at each dose level. The resultsfrom a predictive check showed that there is good agree-ment between the model-predicted probability of BOR andthe observed proportion of BOR responders. Estimates ofcovariate effects from the full exposure–response model forBOR are presented in Supplementary Fig. S2, and the resultssuggest that Cminss had a greater effect on achieving atumor response than other covariates.

Exposure–response analysis: clinical activity by irRCCminss was found to be a statistically significant predic-

tor of the probability of clinical activity by irRC (P < 0.001),and the probability increased with higher log–transformedCminss (Table 1 and Fig. 1B). However, as with BOR bymWHO criteria, none of the covariates examined wereretained in the final model as none had statistically signif-icant effect on the exposure–response relationship for clin-ical activity by irRC.

The odds of a patient achieving clinical activity by theirRC increased by 1.76-fold for a 2.7-fold increase inCminss. The observed clinical activities by irRC at 0.3, 3,and 10 mg/kg were 6.3%, 15.0%, and 25.1%, respectively,consistent with the model-estimated probabilities of irRCresponses at median Cminss for the 3 dose groups of 4.5%,15.1%, and 25.6%, respectively. These results suggest thatthe probability of achieving clinical activity by the irRCincreased with dose of ipilimumab. Figure 1B shows theobserved proportion and model-predicted median propor-tion/probability of an irRC response versus Cminss. Theresults from a predictive check showed that there is goodagreement between themodel-predicted probability of irRCresponses and the observed proportion of irRC responses.Estimates of covariate effects from the full exposure–response model for tumor response by irRC are presentedin Supplementary Fig. S3, and similar to BOR by mWHOcriteria, Cminss seemed to have a greater effect on achievinga tumor response than other covariates.

Exposure–response analysis: OSFrom the Kaplan–Meier analysis, OS improved with

increasing ipilimumab exposure (Cminss) and dose asillustrated in Fig. 2, although OS seems to be more closelyassociated with exposure than dose. Almost all the patientsin the 2nd and 3rd tertiles were from the 10 mg/kg dosegroup (approximately 90%and99%, respectively), whereasmost patients in the 1st tertile were from the 0.3 and 3 mg/kgdose groups (approximately 29%and49%, respectively).Patients in the highest tertile of Cminss seemed to havemarkedly better OS than patients in the lower tertiles ofCminss, suggesting the existence of an exposure–responserelationship even within the 10 mg/kg dose group. How-ever, it is important to note that the Kaplan–Meier analysisforOSbyCminss does not account for the potential effect ofconfounding variables (such as LDH levels), and thereforemodel-based analyses were conducted.

Estimates from the full exposure–response model for OSare presented in Supplementary Fig. S4, and the resultssuggest that baseline LDH levels and Cminss had greatereffects on the risk of death relative to other covariates. Theresults of the final model show that Cminss, baseline LDH,and ECOG PS were significant predictors of OS (P < 0.001).

Table 1. Parameter estimates from the logistic regression model for ipilimumab exposure–responserelationships

Exposure–response model Predictor

OR coefficient(95%CI)a P

Median Cminss(5th–95th percentile)[mg/mL]

OR: 5th percentile:median of Cminss(95%CI)

OR: 95th percentile:median of Cminss(95%CI)

BOR Log(Cminss)b 2.41 (1.66–4.26) <0.01 49.3 (1.75–110) 0.0531 (0.00742–0.38) 2.03 (1.26–3.26)irRC Log(Cminss)b 1.76 (1.35–2.54) <0.001 47.7 (1.61–106) 0.146 (0.0551–0.387) 1.58 (1.25–1.99)irAEs Log(Cminss)b 1.91 (1.58–2.38) <0.001 43.9 (1.75–103) 0.124 (0.0653–0.237) 1.74 (1.46–2.06)

a95% CI obtained from bootstrap (N ¼ 500).bCminss was log-transformed. Log(Cminss) increases by one unit for approximately 2.7-fold increase in Cminss.

Feng et al.





Importantly, HLA-A�0201 genotype did not have a signif-icant impact on exposure–response OS in this analysis.OS improved with increasing Cminss, decreased withincreasing LDH, and was worse for patients with ECOG PSmore than 0. Patients in the 5th percentile of Cminss (1.75mg/mL) had an OS hazard ratio of 1.52 relative to patientswith median Cminss of 43.9 mg/mL. Patients in the 95thpercentile of LDH levels (846 IU/L) had an OS hazard ratio

of 3.27 relative to thosewithmedian LDH levels (206 IU/L),which was less than the top limit of normal (225 IU/L).Furthermore, the estimated OS hazard ratio for patientswith an ECOG PS more than 0 was 1.72 relative to patientswith an ECOGPS of 0. The hazard ratios at the 5th and 95thpercentiles of the continuous predictors Cminss and LDH,relative to their median values, indicate that the magnitudeof the effect was greater for LDH than for Cminss (Table 2).

Figure 1. Model-predictedprobability of BOR by mWHOcriteria (A) or ir-Clinical Activity bythe irRC (B) versus ipilimumabsteady-state trough concentration.The solid line and shaded arearepresent exposure–responserelationship model expected Pr(BOR) and 95% bootstrap CIs(N ¼ 500). The horizontal box plotsrepresent the distributions ofCminss at each dose group asfollows: boxes (25th, 50th, and75th percentiles) and whiskers (5thand 95th percentiles). Open circlesrepresent the observed proportionof responders for each dose group,plotted at the median Cminss ofeach dose. The vertical barsrepresent 90% prediction intervalsof each observed proportionobtained from simulated trials.






At themedian Cminss determined for the 0.3mg/kg doseof ipilimumab, OS was estimated to be 0.85-fold and 0.58-fold lower than at the median Cminss for the 3 mg/kg and10 mg/kg doses, respectively. The results of the final CPHmodel are presented in Fig. 3. In the final model, the effectsof the continuous predictors Cminss and LDH and thecategorical 1 predictor ECOG PS were evaluated by a 4 � 2

stratification of the data according to whether Cminss andLDH were above or below the median values in patientswith ECOG PS of 0 or more than 0 for the dataset. For all ofthe stratified groups, there was generally good agreementbetween the model-predicted cumulative probability of OSand the correspondingKaplan–Meier curves (the latter lyingwithin the 90% model PI; Supplementary Fig. S5). The

Figure 2. Kaplan–Meier estimatesof OS by Cminss tertiles (A) and bydose (B). In both A and B, 1T/2T/3Tindicate tertiles of Cminss andnumbers below the graphsrepresent the patients at risk at 0, 6,12, 18, 24, 30, and 36 months.

Feng et al.





predicted distribution of OS in patients by LDH level,ECOG PS, and ipilimumab exposure (median Cminss at0.3, 3 or 10 mg/kg) indicates that the effect of LDH on therisk of death is greater than that for ECOG PS and Cminss

(Fig. 3). Patients with greater ipilimumab exposure seem tohave lower risk of death at any given value of LDH andECOG PS relative to patients with lower exposure at thesame value for LDH and ECOG PS.

Table 2. Parameter estimates from the CPH model for ipilimumab exposure–response relationships withOS

PredictorPredictormedian (P05–P95)a

HR coefficient(95% CI)b

HR P05: median(95% CI)

HR P95: median(95% CI)

Cminss [mcg/mL] 43.9 (1.75–103) 0.990 (0.986–0.994) 1.52 (1.29–1.8) 0.552 (0.437–0.697)LDH [IU/L]c 206 (131–846) 2.32 (1.91–2.8) 0.684 (0.628–0.745) 3.27 (2.5–4.28)

Comparator: Reference HR Coefficient (95% CI)d

ECOG PS >0: ¼ 0 (N ¼ 176:322) 1.72 (1.37–2.15)

aP05: 5th percentile and P95: 95th percentile.bHR coefficient represents the hazard ratio for one unit of change in the predictor variable.cTheHRcoefficient for LDHcorresponds to log transformedLDH (whichwasemployedas the linearpredictor, as thedistributionof LDHis right skewed).dHR coefficient represents the HR for comparator relative to reference predictor variable.

Figure 3. Model-predicted OS bybaseline LDH, ECOG PS, andipilimumab exposure (medianCminss at 0.3, 3, and 10 mg/kgdoses) in patients with advancedmelanoma. Lines represent thepredicted survival by the CPHmodel.






Exposure–response analysis: irAEsAsdepicted in Fig. 4, the probability of experiencing grade

2ormore and grade 3ormore irAEs seemed to increasewithincreasing ipilimumab exposure; however, patients whodid not experience an irAE during the induction periodwere not likely to experience an irAE during maintenancetherapy with ipilimumab. The Kaplan–Meier analysis ofirAEs show that the majority of events occurred during theinduction dosing period (Supplementary Figs. S6 and S7).

Estimates fromthe full exposure–response irAEmodel aregraphically presented in Supplementary Fig. S8. The resultssuggest that Cminss had the largest effect on the occurrenceof an irAE among the covariates tested. The results of thefinal model (developed by backward elimination) suggestthat, among the covariates tested, only ipilimumab Cminsswas identified as a significant predictor for the occurrence ofan irAE (P<0.001).HLA-A�0201 genotypewas not found tohave a significant impact on exposure–response irAEs in thisanalysis.

The exposure–response analysis indicated that the prob-ability of experiencing a grade 2 or more and grade 3 ormore irAE increased with increasing Cminss (Table 1). Thepoint estimates of the odds ratio for 5th:median Cminss(1.75 mg/mL) and 95th (103 mg/mL):median Cminss (43.9mg/mL) were 0.124 and 1.74, respectively; therefore, theodds of a patient experiencing a grade 3 ormore irAE (vs. noirAE, grade 1 or grade 2) are expected to increase by 1.74-fold for a Cminss increase from 43.9 (median) to 103 mg/mL (95th percentile). Model-predicted probabilities ofexperiencing a grade 2 or more irAE at the median Cminss(5th, 95th percentiles) for the 0.3, 3, and 10 mg/kg treat-

ment groups are approximately 9.8% (5.6, 14), 33% (19,43), and 51% (34, 62), respectively, and the correspondingprobabilities of experiencing a grade 3 or more irAE areapproximately 3.3%(1.8, 4.9), 13%(6.8, 19), and24%(14,33; Fig. 4). The results from the predictive check indicatethat there is good agreement between the model-predictedprobability of grade 2 or more and grade 3 or more irAEsand the observed proportion of grade 2 ormore and grade 3or more irAEs.

DiscussionThe current pooled analyses of data fromphase II trials in

advanced melanoma represent the first quantification ofexposure–response relationships for ipilimumab efficacyand safety endpoints in any tumor type. Specifically, thepurpose of this study was to determine if there is anassociation between pharmacokinetics (exposure) and keyclinical outcomes with ipilimumab, and to determine base-line patient characteristics that may impact these relation-ships. While there are 2 completed phase III clinical trials ofipilimumab in patients with advanced melanoma (4, 5),data from the 4 completed phase II studies included hereinprovide the largest dataset across a consistent patient pop-ulation and treatment regimen, and provide a range ofevaluated ipilimumab doses. Exposure–response analysesfor ipilimumab at 10 mg/kg in combination with dacarba-zine in patients with previously untreated advanced mela-noma (5) are the subject of separate publications.

One of the objectives of the exposure–response analysesfor efficacy was to characterize the relationship between

Figure 4. Model-predictedprobability of grade 2 or more andgrade 3 or more irAEs versusipilimumab steady-state troughconcentration. The solid line anddash line represent exposure–response relationship model-predicted probability of grade 2 ormore and grade 3 or more irAEs,respectively. The shaded areasrepresent 95% bootstrapconfidence intervals (N¼ 500). Thehorizontal box plots represent thedistributions of Cminss at eachdosegroupas follows: boxes (25th,50th, and 75th percentiles) andwhiskers (5th and95th percentiles).Open circles and open trianglesrepresent the observed proportionof irAEs grade 2 or more and grade3 or more for each dose group,respectively. The vertical barsrepresent 90% prediction intervalsof each observed proportionobtained from simulated trials(N ¼ 500).

Feng et al.





ipilimumab exposure and objective response. Using eithermWHO criteria or the exploratory irRC, the exposure–response analyses indicated that Cminss during the induc-tion phase was a statistically significant predictor ofresponse. By mWHO criteria, the probability of BORincreased with a higher log-transformed Cminss, eventhough the median Cminss at 10 mg/kg was well abovethe target concentration (20 mg/mL) required for maximalinhibition of CTLA-4 binding to its ligands (6). While asimilar association was observed using irRC, model-basedestimates indicated that the probabilities of a tumorresponse at the median Cminss for 0.3, 3, and 10 mg/kgdoses were greater when evaluated by the irRC (4.3%,15.0%, and 25.6% for irRC vs. 0.6%, 4.9%, and 11.6% forBOR by mWHO criteria, respectively). This observationreinforces the potential value of the irRC as a tool forcharacterizing the efficacy of immunotherapies, wheredurable objective responses are sometimes seen in patientswho are characterized as having disease progression bymWHO criteria. Collectively, regardless of which efficacycriteria are used, these results suggest that the probability ofachieving an objective response is greater with higher ipi-limumab exposure. None of the covariates tested, includingbaseline LDH levels, had a statistically significant effect onthe exposure–response analyses for tumor response.An exposure–response analysis was conducted to char-

acterize the relationship between ipilimumab exposure andOSusing aCPHmodel. The analysis showed that therewas astatistically significant relationship between ipilimumabCminss and the hazard ratio for OS. Patients at the 5thpercentile of Cminss (1.75 mg/mL) had an OS hazard ratioof 1.52 relative to patients with median Cminss (43.9 mg/mL), and theOS of patients at the 95th percentile of Cminss(103 mg/mL) had an OS hazard ratio of 0.552 relative topatients at themedianCminss. These results suggest thatOSimproves with increasing Cminss and ipilimumab dose.TheOS of patients with advancedmelanoma is known to

be associated with several prognostic factors, includingECOG PS and baseline serum LDH (16, 17). LDH has beenshown to be an independent prognostic factor for survivalevenafter accounting for site andnumber ofmetastases, andhas been incorporated into the current American JointCommittee on Cancer (AJCC) staging classification (16).The Kaplan–Meier analysis with respect to dose should beinterpreted as being suggestive of a qualitative doseresponse rather than a definitive quantification of the doseresponse, as the pooled phase II data included in theanalyses are not balanced with respect to prognostic factorsacross doses. Similarly, the differences in OS in Kaplan–Meier analyses by Cminss tertiles should be interpreted asbeing indicative of a qualitative exposure–response,because this univariate analysis does not account for poten-tial imbalances in risk factors between the Cminss groups.However, themultivariateCox-proportional hazardsmodeldoes include the key prognostic factors, and this modelprovides an estimate of the effect of Cminss on reducing therisk of death after adjusting for the presence of these riskfactors.

Among the covariates tested, baseline serum LDH levelsand ECOGPSwere found to be significant predictors of OS.Our results further show that HLA-A�0201 status does notsignificantly impact exposure–response relationships forOS or safety, consistent with the results of a retrospectiveanalysis of ipilimumab clinical trial data showing thatefficacy and safety outcomes were independent of HLA-A�0201 status (18). Thus, while the phase III trial of ipili-mumab monotherapy at 3 mg/kg enrolled only patientsthat were HLA-A�0201-positive (4), the results collectivelysuggest that ipilimumab efficacy and safety are independentof HLA-A�0201 status.

The risk of death increased with increasing LDH levels,and was higher for patients with ECOG PS more than 0comparedwith thosewith anECOGPSof 0. LDHseemed tohave a greater effect on the risk of death than Cminss orECOG PS in our exposure–response analyses. Althoughbaseline LDH levels do not have a clinically meaningfulimpact on the systemic clearance of ipilimumab, patientswith lower baseline LDH levels tended to have higheripilimumab concentrations, which may have impacted OS.However, LDH levels and Cminss were accounted for in theexposure–response model and both were found to beindependent covariates for OS. Interestingly, Cminssseemed to be more closely associated with OS than treat-ment group (dose). Based on the exposure–response OSanalysis, a higher dose of ipilimumab may also provide asurvival benefit for patients with poor prognostic features,for example, elevated baseline LDH levels. This result isconsistent with the improved OS in phase III trials ofipilimumab in advanced melanoma, where patients withnormal and elevated LDH levels were enrolled (4, 5).

The exposure–response analysis for safety showed that theprobability of experiencing an irAE of grade 2 or more orgrade 3 ormore, and the probability of first occurrence of anirAE at a given time point, increased with increasing Cminssover the evaluated dose range. The model predicted that theprobabilities of grade 3 or more irAEs were approximately3%, 13%, and 24% at the median Cminss of the 0.3, 3, and10 mg/kg doses. These data are consistent with the frequen-cies of grade 3 and 4 irAEs reported in the phase III trial ofipilimumab monotherapy at 3 mg/kg, which ranged from10.2% to 14.5% (4). IrAEs may reflect ipilimumab’s im-mune-based mechanism of action and the majority occurduring the induction dosing period. They can be severe andlife-threatening, but most were reversible with treatmentguidelines developed in ipilimumab clinical trials (4–8,19). There were no study drug-related deaths in CA184-007 (7), one at 3 mg/kg in CA184-022 (6), and one at 10mg/kg in CA184-008 (8). In the first phase III trial, 12 deathsin511treatedpatients(2.3%)were related to ipilimumabat3mg/kg (4). However, there were no study drug-related deathsin the subsequent phase III trial in 247 patients who receivedipilimumab at 10 mg/kg plus dacarbazine (5). Thus, whilegreater ipilimumabexposure increases the frequencyof grade3 or more irAEs, the availability of treatment guidelines forthe management of irAEs may prevent this increased inci-dence from causing a higher rate of treatment-related deaths.






The results of the current analyses show a positive asso-ciation between ipilimumab exposure, and by extensiondose, and key clinical outcomes. An ipilimumab dose of 10mg/kg produced a greater Cminss with numerically highertumor responses and an increased probability of improvedOS, although these outcomes were associated with a greaterlikelihood of developing an irAE. Given the association ofCminss with both efficacy and safety measures, individu-alized dosing might be an effective approach for the ipili-mumab treatment of patients with advanced melanoma inthe future, provided it is possible to determine a range ofipilimumab exposure that is optimal with respect to bothefficacy and safety with data being collected in an ongoingphase III dose optimization trial (11). Both the 3 and 10mg/kg doses of ipilimumab have been shown to improveOS in phase III clinical trials of patients with advancedmelanoma (4, 5).While the current results provide evidencefor the potential of better efficacy outcomes with greateripilimumab exposure, further clinical studies of the risk-benefit profile are needed to determine the most effectivedose of ipilimumab.

Disclosure of Potential Conflicts of InterestA. Roy is employed (other than primary affiliation; e.g., consulting) as a

group director, clinical pharmacology and pharmacometrics at Bristol-MyersSquibb and has ownership interest (including patents) in Bristol-MyersSquibb. T.-T. Chen is employed (other than primary affiliation; e.g., con-

sulting) as a adjunct assistant professor at Columbia University and hasownership interest (including patents). J. Weber is a consultant/advisoryboard member of Bristol-Myers Squibb. No potential conflicts of interestwere disclosed by the other authors.

Authors' ContributionsConception and design: Y. Feng, A. Roy, E. Masson, T.T. Chen, J. WeberDevelopment of methodology: Y. Feng, A. Roy, E. Masson, R. Humphrey,J. WeberAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.): R. Humphrey, J. WeberAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): Y. Feng, A. Roy, E. Masson, T.T. Chen,R. Humphrey, J. WeberAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): R. HumphreyWriting, review, and/or revision of the manuscript: Y. Feng, A. Roy,E. Masson, T.T. Chen, R. Humphrey, J. WeberStudy supervision: R. Humphrey

AcknowledgmentsThe authors thank StemScientific for providing editorial and writing

assistance.

Grant SupportThis study was supported by Bristol-Myers Squibb Co.The costs of publication of this article were defrayed in part by the

payment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received October 16, 2012; revisedMay 20, 2013; acceptedMay 21, 2013;published OnlineFirst June 5, 2013.

References1. Melero I, Hervas-Stubbs S, Glennie M, Pardoll DM, Chen L. Immu-

nostimulatory monoclonal antibodies for cancer therapy. Nat RevCancer 2007;7:95–106.

2. Hoos A, Ibrahim R, Korman A, Abdallah K, Berman D, Shahabi V, et al.Development of ipilimumab: contribution to a newparadigm for cancerimmunotherapy. Semin Oncol 2010;37:533–46.

3. Weber JS,HamidO,ChasalowSD,WuDY,ParkerSM,GalbraithS, et al.Ipilimumab increases activated T cells and enhances humoral immunityin patients with advanced melanoma. J Immunother 2012;35:89–97.

4. Hodi FS, O'Day SJ, McDermott DF, Weber RW, Sosman JA, HaanenJB, et al. Improved survival with ipilimumab in patients with metastaticmelanoma. N Engl J Med 2010;363:711–23.

5. Robert C, Thomas L, Bondarenko I, O'Day S, Weber J, Garbe C, et al.Ipilimumab plus dacarbazine for previously untreated metastatic mel-anoma. N Engl J Med 2011;364:2517–26.

6. Wolchok JD, Neyns B, Linette G, Negrier S, Lutzky J, Thomas L, et al.Ipilimumab monotherapy in patients with pretreated advanced mela-noma: a randomised, double-blind, multicentre, phase 2, dose-rang-ing study. Lancet Oncol 2010;11:155–64.

7. Weber J, Thompson JA, Hamid O, Minor D, Amin A, Ron I, et al. Arandomized, double-blind, placebo-controlled, phase II study com-paring the tolerability and efficacy of ipilimumab administered with orwithout prophylactic budesonide in patientswith unresectable stage IIIor IV melanoma. Clin Cancer Res 2009;15:5591–8.

8. O'Day SJ, Maio M, Chiarion-Sileni V, Gajewski TF, Pehamberger H,Bondarenko IN, et al. Efficacy and safety of ipilimumabmonotherapy inpatients with pretreated advanced melanoma: a multicenter single-arm phase II study. Ann Oncol 2010;21:1712–7.

9. Weber J. Ipilimumab: controversies in its development, utility and auto-immuneadverse events. Cancer Immunol Immunother 2009;58:823–30.

10. Shepherd FA, Douillard JY, Blumenschein GR Jr. Immunotherapy fornon-small cell lung cancer: novel approaches to improve patientoutcome. J Thorac Oncol 2011;6:1763–73.

11. ClinicalTrials.gov. Phase III trial in subjects with metastatic melanomacomparing 3mg/kg ipilimumab versus 10mg/kg ipilimumab. Availableat: http://clinicaltrials.gov/ct2/show/NCT01515189.

12. Hamid O, Schmidt H, Nissan A, Ridolfi L, Aamdal S, Hansson J, et al. Aprospective phase II trial exploring the association between tumormicroenvironment biomarkers and clinical activity of ipilimumab inadvanced melanoma. J Transl Med 2011;9:204.

13. Wolchok JD, Hoos A, O'Day S, Weber JS, Hamid O, Lebb�e C, et al.Guidelines for the evaluation of immune therapy activity in solidtumors: immune-related response criteria. Clin Cancer Res 2009;15:7412–20.

14. Weber JS,O'DayS,UrbaW,Powderly J,NicholG,YellinM, et al. PhaseI/II study of ipilimumab for patients with metastatic melanoma. J ClinOncol 2008;26:5950–6.

15. Dai D, Wu C, Parker SM, Jure-Kunkel MN, Pfister M, Berman D,et al. Model-based evaluation of ipilimumab dosage regimen inpatients with advanced melanoma. J Clin Oncol 2008;26: abstr9073.

16. BalchCM,Gershenwald JE, SoongSJ, Thompson JF, AtkinsMB, ByrdDR, et al. Final version of 2009 AJCC melanoma staging and classi-fication. J Clin Oncol 2009;27:6199–206.

17. Korn EL, Liu PY, Lee SJ, Chapman JA, Niedzwiecki D, Suman VJ,et al. Meta-analysis of phase II cooperative group trials in metastaticstage IV melanoma to determine progression-free and overall sur-vival benchmarks for future phase II trials. J Clin Oncol 2008;26:527–34.

18. Wolchok JD, Weber JS, Hamid O, Lebb�e C, Maio M, Schadendorf D,et al. Ipilimumab efficacy and safety in patients with advanced mel-anoma: a retrospective analysis of HLA subtype from four trials.Cancer Immun 2010;10:9.

19. Weber JS, K€ahler KC, Hauschild A. Management of immune-relatedadverse events and kinetics of response with ipilimumab. J Clin Oncol2012;30:2691–7.

Feng et al.





2013;19:3977-3986. Published OnlineFirst June 5, 2013.Clin Cancer Res Yan Feng, Amit Roy, Eric Masson, et al. Ipilimumab in Patients with Advanced Melanoma

Response Relationships of the Efficacy and Safety of−Exposure

Updated version

10.1158/1078-0432.CCR-12-3243doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2013/06/05/1078-0432.CCR-12-3243.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/19/14/3977.full#ref-list-1

This article cites 17 articles, 6 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/19/14/3977.full#related-urls

This article has been cited by 5 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/19/14/3977To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-12-3243

http://clincancerres.aacrjournals.org/content/suppl/2013/06/05/1078-0432.CCR-12-3243.DC1

http://clincancerres.aacrjournals.org/content/19/14/3977.full#ref-list-1

http://clincancerres.aacrjournals.org/content/19/14/3977.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/19/14/3977


Exposure ResponseRelationshipsoftheEfﬁcacyandSafety...

Documents

Transcript of Exposure ResponseRelationshipsoftheEfﬁcacyandSafety...