We describe a randomized three-arm phase I study of ipilimumab administered alone (I group) or in combination with dacarbazine (D group) or carboplatin/paclitaxel (CP group) in patients with previously untreated advanced melanoma. The primary objective was to estimate the effect of ipilimumab on the pharmacokinetics (PK) of dacarbazine and paclitaxel and, conversely, to estimate the effects of dacarbazine and carboplatin/paclitaxel on the PK of ipilimumab. Secondary objectives included evaluation of the safety and anti-tumor activity of ipilimumab when administered alone or with either dacarbazine or carboplatin/paclitaxel, and assessment of pharmacodynamic (PD) effects of ipilimumab on the immune system when administered alone or with either of the two chemotherapies. Ipilimumab was administered at a dose of 10 mg/kg intravenously (IV) every 3 weeks for up to 4 doses. Patients in the D group received dacarbazine 850 mg/m2 IV every 3 weeks. Patients in the CP group received paclitaxel 175 mg/m2 IV and carboplatin [AUC=6] IV every 3 weeks. Starting at week 24, patients without dose-limiting toxicities were eligible to receive maintenance ipilimumab at 10 mg/kg every 12 weeks until disease progressed or toxicity required discontinuation. Of 59 randomized patients, 18 (30.5%) discontinued treatment due to adverse events. Response rates by modified WHO criteria were 29.4% (I group), 27.8% (D group), and 11.1% (CP group). No major PK or PD interactions were observed when ipilimumab was administered with dacarbazine or with the carboplatin/paclitaxel combination. This study demonstrated that ipilimumab can be combined safely with two chemotherapy regimens commonly used in advanced melanoma.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
Ipilimumab, now known as Yervoy™, is an IgG1 human monoclonal antibody that binds to cytotoxic T lymphocyte antigen 4 (CTLA-4), an immune checkpoint protein on T cells (1, 2). It has been tested extensively in stage III and IV melanoma and was shown to have anti-tumor activity at doses of 3 mg/kg and 10 mg/kg (3-5). When administered at 3 mg/kg alone or with a peptide vaccine in a randomized phase III trial, ipilimumab prolonged overall survival in patients with previously treated, unresectable melanoma (6). The results of this randomized phase III trial led to approval of ipilimumab in a number of countries, including the United States, European Union, and Australia, at a dose of 3 mg/kg for patients with unresectable or metastatic melanoma (7-9). Ipilimumab at 10 mg/kg combined with dacarbazine, an alkylating agent, was shown in a separate randomized phase III trial to prolong survival compared to dacarbazine alone in patients with previously untreated advanced melanoma (10). In a randomized phase II trial of 217 patients with advanced melanoma designed to evaluate ipilimumab at doses of 0.3, 3, or 10 mg/kg, the overall response rate was highest in the group that received 10 mg/kg, with nonsignificant differences in overall survival among treatment groups (3). An ongoing randomized phase III trial in advanced melanoma is comparing the safety and efficacy of ipilimumab monotherapy at 3 vs. 10 mg/kg (ClinicalTrials.gov NCT01515189).
Reports of documented drug-drug interactions between biologics and small molecules have been scant in the literature. However, certain chemotherapies (e.g., methotrexate) and other immunosuppressive treatments have been shown to alter the pharmacokinetics (PK) of monoclonal antibodies (11). More recently, it has been recognized that some monoclonal antibodies can alter the PK of small molecules metabolized by cytochrome P450 (CYP450) enzymes by modulating cytokines (12). For example, dacarbazine and paclitaxel are both metabolized by various CYP450 enzymes, which could be downregulated by cytokines (e.g., IL-1, IL-2, IFN-β, and IL-6). The primary goal of this randomized phase I trial was to assess potential PK drug-drug interactions between ipilimumab at a dose of 10 mg/kg and two chemotherapies commonly used for melanoma. In this study, patients were randomized to receive ipilimumab alone (I group), ipilimumab with dacarbazine (D group), or ipilimumab with carboplatin/paclitaxel (CP group). Ipilimumab was given every 3 weeks for 4 doses as an induction regimen over 12 weeks. Chemotherapy was also given every 3 weeks, but was continued for up to 8 cycles over 24 weeks. In addition to assessing PK interactions, safety was evaluated, efficacy measures based on modified World Health Organization (mWHO) and immune-related response criteria (irRC, as discussed in Materials and Methods) were assessed, and a number of biomarkers were explored for their potential utility as pretreatment or pharmacodynamic (PD) indicators of the effects of ipilimumab. They included absolute lymphocyte count (ALC) and T cell phenotypes as assessed by flow cytometry. Additionally, changes in levels of antibodies to ipilimumab and the Cancer/Testis antigen, NY-ESO-1, were evaluated in this trial. Antibodies against ipilimumab were assessed as a safety endpoint. Antibodies against NY-ESO-1 were assessed because previous data suggest that there may be an association between immune response to this antigen and clinical outcome with ipilimumab treatment (13).
Fifty-nine patients were randomized and received treatment on this trial. Fifteen patients were still on treatment at the time of the data cut. Overall, median age was 56 years, with a 64:36 percent male:female distribution. All patients except for one were Caucasian, all had Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1 and had prior surgery, but none had prior systemic therapy for unresectable melanoma. Table 1 summarizes these data and describes the prior treatments for all patients.
Forty-four patients were off treatment at the time of the data cut, including 14 (70.0%) in the I group, 13 (68.4%) in the D group, and 17 (85.0%) in the CP group. Of the 44 patients, 41 patients did not complete the 24 weeks of induction and an additional 3 patients completed induction and either did not receive subsequent therapy (n = 2) or discontinued therapy during maintenance (n = 1). Thirty-one patients (52.5%) received all 4 induction doses of ipilimumab, including 9 (45.0%) in the I group, 10 (52.6%) in the D group, and 12 (60.0%) in the CP group. The median per patient cumulative ipilimumab dose during induction was 30.1 mg/kg in the I group, 36.7 mg/kg in the D group, and 37.1 mg/kg in the CP group. Eighteen patients (30.5%) received treatment with ipilimumab during maintenance. Fifteen patients (25.4%) continued ipilimumab beyond 48 weeks in the maintenance phase of this study. The overall median cumulative dose in maintenance was 30.0 mg/kg per patient. By week 24, 41/59 patients (69.5%) had stopped treatment; 26 of these patients discontinued due to progression of disease, 11 discontinued due to toxicity, and 4 discontinued for other reasons.
Plasma and serum samples from all 59 randomized patients were available for PK analysis. Of these 59 patients, 51 provided PK data sufficient for statistical analyses. Summary statistics are presented for ipilimumab, dacarbazine, its active metabolite 5-aminoimidazole-4-carboxamide (AIC), and paclitaxel PK parameters in Table 2A-D, respectively. The estimated geometric mean ratios (GMRs) for ipilimumab Cmax and area under the concentration time curve (AUC)(0-21d) changed 0.982- and 0.917-fold, respectively, in the presence of dacarbazine (Table 3A). The estimated GMRs for ipilimumab Cmax and AUC(0-21d) changed 0.934-and 0.868-fold, respectively, in the presence of carboplatin/paclitaxel.
The estimated GMRs for dacarbazine Cmax and AUC[from time zero extrapolated to infinite time(INF)] changed 1.027- and 0.912fold, respectively, in the presence of ipilimumab (Table 3B). Those for its metabolite, AIC, changed 1.058- and 0.970-fold, respectively. The GMRs for paclitaxel Cmax and AUC(INF) changed 0.963- and 1.068-fold, respectively, in the presence of ipilimumab.
There were no treatment-related deaths in this trial. Eighteen patients (30.5%) had an adverse event (AE) leading to discontinuation of treatment: 5 (25.0%) in the I group, 7 (36.8%) in the D group, and 6 (30.0%) in the CP group. The most common AEs of any grade, regardless of causality, were rash (72.9%), fatigue (69.5%), pruritus (66.1%), diarrhea (59.3%), and nausea (57.6%) (Table 4). The most common treatment-related AEs of any grade among the patients were rash (67.8%), pruritus (62.7%), diarrhea (55.9%), and fatigue (57.6%).
Interestingly, the spectrum of side effects appeared to vary among the three groups in this phase I study. Treatment-related elevations of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) of any grade occurred in more patients in the D group (47.4% and 52.6%, respectively) than in the I group (15.0% and 15.0%, respectively) or the CP group (25.0% and 25.0%, respectively). More treatment-related hypersensitivity events of any grade occurred in the CP group (30.0%) compared to the I group (5.0%) or the D group (0%).
Grade 3-4 treatment-related AEs occurred in a total of 39 patients: 15 events in 10 patients in the I group, 28 events in 14 patients in the D group, and 35 events in 15 patients in the CP group. The most common grade 3-4 treatment-related AEs were increased ALT (18.6%), decreased neutrophil count (15.3%), diarrhea (13.6%), and increased AST (10.2%). Grade 3-4 treatment-related skin events, including erythema and rash, were only encountered in the CP group (20.0%), while grade 3-4 treatment-related diarrhea occurred less frequently in the CP group (5.0%) than in the other groups (20.0% and 15.8% in the I and D groups, respectively). Grade 3-4 decreases in neutrophil count occurred more frequently in the CP group (40.0%) than in the other groups (0% and 5.3% in the I and D groups, respectively).
Three patients (15.0%) in the I group and one patient (5.0%) in the CP group had an ipilimumab infusion interruption during induction due to hypersensitivity reactions. One patient (5.0%) in the D group had an ipilimumab infusion interruption due to an infusion reaction during induction. Nine patients (45.0%) in the CP group had temporary interruptions of the paclitaxel infusion during induction: 7 due to hypersensitivity reactions; 1 due to a hypersensitivity reaction, rash, itching, and jittery legs; and 1 due to pruritus. Two patients (10.5%) in the D group had temporary interruptions of the dacarbazine infusion during induction: 1 due to a burning sensation in the arm and 1 due to a hypersensitivity reaction.
Of the 56 patients with post-baseline immunogenicity data, 4 patients (7.1%) developed low-titer anti-ipilimumab antibodies after having tested negative at baseline: 1 patient (5.6%) in the I group and 3 patients (15.8%) in the D group.
Anti-tumor activity was observed in all treatment arms (Table 5). The objective response rate (ORR), based on mWHO criteria, for response-evaluable patients was highest in the I group (29.4%) followed by the D group (27.8%) and the CP group (11.1%). Based on irRC, the ORR was highest in the I and D groups (both 33.3%) followed by the CP group (27.8%). The disease control rate (DCR), based on mWHO criteria, for response-evaluable patients was highest in the D group (55.6%), followed by the I group (52.9%) and the CP group (44.4%). Based on irRC, DCR was highest in the I group (73.3%), followed by the D group (61.1%) and the CP group (55.6%).
Mean ALC increased over time after initiation of ipilimumab treatment similarly in the three groups (Figure 1A). However, ALC increases appeared to be smaller and less frequent in the CP group. Positive associations between immune-related clinical activity (irCA) and increases in ALC, consistent with those seen in other ipilimumab trials, were not statistically significant in this trial (Supplementary Table 1).
The mean relative frequencies and absolute counts of activated (HLA-DR+) CD4+ and CD8+ T cells increased significantly over time after initiation of ipilimumab treatment in a similar manner in the three treatment groups (Figure 1B). No associations between irCA and T cell subset frequencies were observed (Supplementary Table 2).
The titer of serum antibodies to NY-ESO-1 were stable over time in the majority of patients (data not shown). No evidence of any associations between irCA and baseline NY-ESO-1 antibody concentration was observed, although the number of patients in the study was modest (Supplementary Table 3).
The primary objective of this phase I randomized trial was to assess potential PK drug-drug interactions between ipilimumab and two commonly used chemotherapies. No major PK drug-drug interactions between ipilimumab and either paclitaxel, dacarbazine, or AIC, were observed. The largest estimated interaction was a reduction in ipilimumab exposure (AUC[0-21d]) in the presence of carboplatin/paclitaxel (90% CI for GMR = 0.69, 1.09). Remaining confidence intervals (CIs) for the GMRs extended only slightly below the default “no effect” interval of 0.80 to 1.25.
From population PK modeling of data from previous studies, geometric mean steady-state (AUC) for ipilimumab administered as 3 mg/kg monotherapy was predicted to be 15.6 mg.hr/ mL (14). In this study, geometric mean AUC(0-21d) for ipilimumab 10 mg/kg in combination with concurrent carboplatin/ paclitaxel was estimated to be 46.9 mg.hr/mL. Since a 3 mg/kg monotherapy regimen of ipilimumab has been shown to extend overall survival with an acceptable benefit/risk profile in patients with pretreated advanced melanoma (6), and mean exposure was substantially greater with 10 mg/kg than with 3 mg/kg, the observed decreases in ipilimumab exposure when administered at 10 mg/kg concurrently with carboplatin/paclitaxel may not be clinically relevant.
Minimal effects of ipilimumab on the PK of paclitaxel or dacarbazine and its major metabolite, AIC, were observed. Paclitaxel and dacarbazine terminal elimination half-life obtained before and after ipilimumab administration (mean half-life ~10 and ~2 hours for paclitaxel and dacarbazine, respectively) were comparable and consistent with previously published studies (15). The observed decreases in paclitaxel, dacarbazine, and AIC exposure resulted in individual exposures that would be within the expected therapeutic ranges of these chemotherapies.
Anti-tumor activity was observed in all three treatment arms. Estimates of ORR were imprecise due to the small size of this study, and differences among treatment arms were not tested statistically. Nonetheless, the lowest response rate by mWHO criteria or irRC was in the CP group, with the I and D groups being similar to each other. A similar trend was also observed for DCR by mWHO and irRC.
Ipilimumab was generally well-tolerated in previously untreated patients with advanced melanoma in this trial when administered alone or when co-administered with carboplatin/paclitaxel or dacarbazine. There were no treatment-related deaths in any of the groups in this study. There were more treatment-related elevations of ALT and AST, both high grade and overall, in the D group than in the I group or the CP group. There were more grade 3-4 treatment-related decreases in neutrophil count in the CP group than in the other groups. In contrast, there were more treatment-related events of diarrhea, both high-grade and overall, in the ipilimumab monotherapy group than in either of the chemotherapy groups. These results, although based on small frequencies, suggest that the spectrum of AEs with ipilimumab may be influenced by the drug with which it is partnered and merit additional evaluation in future combination ipilimumab trials. Consistent with this hypothesis, in a phase III study of ipilimumab in metastatic melanoma patients (10), increases in ALT and AST were more frequent in an ipilimumab plus dacarbazine treatment group than in a placebo plus dacarbazine group. Thus, considering the results of that study and the current study, liver toxicity was more frequent for the combination of dacarbazine with ipilimumab than for one or the other single agent. Future combination trials may help to determine whether the potential toxic effects of these two agents are synergistic or additive.
Several previous studies have demonstrated significant mean increases in ALC after initiation of ipilimumab therapy (3, 16, 17). Similar increases in mean ALC were observed in this study for all three treatment groups, but ALC increases were not significantly associated with any clinical outcome in this small phase I study. Likewise, in contrast to observations made in at least two other small studies noting increased levels of NY-ESO-1-specific T cells and antibody levels (13, 18, 19), concentrations of serum antibodies to NY-ESO-1 were stable over time in virtually all patients in the present study. No evidence of association between baseline or post-baseline NY-ESO-1 antibody concentration and irRC was observed in any group. Increased relative frequencies of activated CD4+ and CD8+ T cells, previously documented with ipilimumab monotherapy (20-23), were also observed in the present study despite the addition of chemotherapy. Alterations in signal transduction pathways including STAT1 and STAT3 in cells of the monocytic lineage have been observed with tremeli-mumab (24), another CTLA-4 abrogating antibody, but such PD changes were not evaluated in the current study.
In summary, no major PK or PD interactions were observed when ipilimumab was administered with dacarbazine or with the combination of carboplatin/paclitaxel. This study also demonstrated that ipilimumab can be combined safely with two chemotherapy regimens that are commonly used in patients with advanced melanoma. Thetoxicityprofiles appearedtodifferamong the groups, suggesting that the agent paired with ipilimumab may have modified the incidences of some of the toxicities. Anti-tumor activity was observed in all three treatment groups, but estimates of efficacy were imprecise due to the small size of this study, and differences among treatment arms were not tested.
Materials and methods
This trial was performed at four centers in the United States: Moffitt Cancer Center, Memorial Sloan-Kettering Cancer Center, The Angeles Clinic and Research Institute, and Levine Cancer Institute at Carolinas Medical Center. Of 72 patients who signed informed consent for the trial, 13 patients were screen failures. The remaining 59 patients were randomized: 20 to the I group, 19 to the D group, and 20 to the CP group. Patients were required to have had no prior systemic therapy for advanced melanoma, be age 18 or older, have ECOG PS of 0 or 1, and possess normal hepatic and renal function. Absolute neutrophil count (ANC) was required to be 1500 per cubic millimeter or more, hemoglobin was required to be 9 g/dL or more, and platelets were required to be 100,000 per cubic millimeter or more. Patients could not have a history of autoimmune disease, nor could they require chronic steroid therapy.
For the I group, patients received ipilimumab at a dose of 10 mg/kg intravenously (IV) administered over 90 minutes every 3 weeks for up to 4 doses (week 1, 4, 7, and 10). For the D group, each patient received dacarbazine 850 mg/m2 IV by a 1-hour infusion on day 1 and ipilimumab at a dose of 10 mg/ kg IV administered over 90 minutes on day 3. Further doses of ipilimumab (weeks 4, 7, and 10) and dacarbazine (weeks 4, 7, 10, 13, 16, 19, and 22) were given every 3 weeks, with the ipilimumab infusion first, followed by dacarbazine on the same day. For the CP group, patients received paclitaxel 175 mg/m2 IV by a 3-hour infusion and carboplatin ([AUC]=6) IV by a 30-minute infusion on day 1 and ipilimumab at a dose of 10 mg/kg IV administered on day 3. Further doses of ipilimumab (weeks 4, 7, and 10) and carboplatin/paclitaxel (weeks 4, 7, 10, 13, 16, 19, and 22) were administered every 3 weeks, with patients receiving ipilimumab first, followed by paclitaxel, and then carboplatin on the same day. In the D and CP groups, chemotherapy was limited to a maximum of 8 doses. Patients received histamine-H1 and -H2 receptor blockers immediately after ipilimumab and 30 to 60 minutes before infusion of paclitaxel. Concomitant corticosteroids (e.g., dexamethasone) could be used at the time of paclitaxel administration and as a pretreatment antiemetic with dacarbazine; no corticosteroids were administered in the I group.
Maintenance treatment began at week 24 and consisted of continued dosing with ipilimumab at 12-week intervals until immune-related progressive disease (irPD), drug intolerance, withdrawal of consent, pregnancy, death, loss to follow-up, or study closure. The irRC used in this study define irPD as at least a 25% increase in tumor burden compared with nadir in two consecutive observations at least 4 weeks apart (25). Maintenance treatment was the same for all subjects regardless of previous treatment arm.
In all three groups, tumor assessments (radiographic) were scheduled at baseline; weeks 12, 16, 20, and 24 during induction; and every 6 weeks during maintenance for the first year. For patients who continued dosing beyond the first year, tumor assessments were scheduled every 12 weeks.
Analyses are based on data collected through week 48 of the study. All patients were followed for 70 days after last dose of treatment or a minimum of 48 weeks, whichever was applicable.
Plasma concentration data for ipilimumab, paclitaxel, dacarbazine, and AIC were generated by bioanalytical laboratories and transferred to eToolbox (EP version 2.6.1; Thermo Electron Corporation, Philadelphia, PA). PK parameters were generated using Kinetica version 4.4.1.
Ipilimumab PK parameters were derived using serum concentration-time data collected following the week 7 dose in the I group, D group, and CP group. To assess the effect of paclitaxel on the PK of ipilimumab, a linear model was applied to ipilimumab log(AUC[from time zero to day 21 (0-21d)]) and log(maximum observed serum concentration [Cmax]), using interpatient PK data from the I and CP groups, with treatment as a fixed effect. The ratio of GMR and corresponding 90% CI were provided on the original (anti-log) scale with the I group as reference. A similar analysis was performed to assess the effect of dacarbazine on ipilimumab PK.
PK parameters for dacarbazine and AIC were derived for the D group from plasma concentration-time data collected following the week 1 (dacarbazine alone) and week 7 (dacarbazine in the presence of ipilimumab) doses. Paclitaxel PK parameters were derived for the CP group from plasma concentration-time data collected following the week 1 (paclitaxel alone) and week 7 (paclitaxel in the presence of ipilimumab) doses. PK of carboplatin was not assessed in this study since carboplatin is dosed individually by AUC and undergoes complete renal elimination (specifically glomerular filtration), which is not expected to be affected by a large molecular-weight monoclonal antibody such as ipilimumab. To assess the effect of ipilimumab on the PK of dacarbazine, a linear mixed model was applied to dacarbazine log(AUC INF) and log(Cmax), using intrapatient PK data in the D group from week 1 and week 7, with week as a fixed effect and an unstructured covariance matrix to model within-patient correlation. The GMR and corresponding 90% CI were provided on the original (anti-log) scale with week 1 as reference. Similar analyses were performed to assess the effect of ipilimumab on AIC and paclitaxel.
AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA) version 13, and graded according to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE) version 3.0. On-study AEs, treatment-related AEs, and treatment-related high-grade AEs (grade 3-4) were tabulated by system organ class, preferred term, and treatment. On-study events were defined as those reported between the first dose date and 70 days after the last day of study treatment.
Efficacy measures were derived from raw tumor measurement data according to both the mWHO and irRC. The mWHO criteria represent a hybrid of the original WHO and Response Evaluation Criteria in Solid Tumors (RECIST) tumor response classification systems, while the irRC represents a further refinement of the mWHO criteria created based on the natural history of clinical responses in patients treated with immunotherapy (25). The best overall response (BOR) was defined as the best response designation from the date of first dose until the patient’s last tumor assessment. A complete response (CR) or partial response (PR) had to be confirmed by a repeat tumor assessment at least 4 weeks after the criteria were first met. The ORR was defined as the percentage of patients with a BOR of confirmed CR or PR. The DCR was defined as the percentage of patients with a BOR of confirmed CR, PR, or stable disease (SD).
Efficacy results are presented for response-evaluable patients, defined as those with i) measurable disease at baseline, ii) histologic diagnosis of melanoma, iii) at least one baseline and one on-study tumor assessment, iv) no resection of index lesions, and v) for irRC, no resection of new lesions. For use in biomarker analyses, irCA was defined as a confirmed immune-related CR or PR, or an immune-related SD ending not earlier than 24 weeks from date of first dose of any study drug.
I. Absolute lymphocyte count (ALC)
The mean change in (ALC) over time for each treatment arm and corresponding two-sided 95% CIs were estimated using an extended (i.e., mixed) linear model that included ALC as response variable, fixed effects of treatment arm, linear splines of time since first dose with knots at the dates of the induction doses, and the treatment-by-time interactions, as well as a spatial-exponential within-patient correlation structure. Conditional F-tests were used to test for mean ALC changes over time in each treatment arm and the difference between treatment arms in the pattern of change in ALC over time.
II. Peripheral blood T cell subsets assessed by flow cytometry
Mean T cell subset frequencies at each of the assessment time points for each treatment arm and corresponding two-sided 95% CIs were estimated using an extended (i.e., mixed) linear model that included T cell frequency as response variable, fixed effects of treatment arm, nominal assessment time, and the treatment-by-time interactions, as categorical variables, and a spatial-exponential within-patient correlation structure. Conditional F-tests were used to test for mean T cell frequency changes over time in each treatment arm and the differences among treatment arms in the pattern of change in T cell frequency over time.
This study was sponsored by Bristol-Myers Squibb. We wish to thank Vafa Shahabi, Ph.D., of Bristol-Myers Squibb for her significant contributions to data generation, analysis, and scientific content of the study report. Additionally, we thank Daniel Brickman, Ph.D.; Tina Chatterjee, M.S.; Mohan Mandava, M.S.; and Guan Xing, Ph.D., of Bristol-Myers Squibb; and Oksana Mokliatchouk, Ph.D., formerly of Bristol-Myers Squibb, for assistance with data analyses. Professional medical writing and editorial assistance was provided by Ami P. Modi, Ph.D., and Rebecca Goldstein, Ph.D., at StemScientific, and this assistance was funded by Bristol-Myers Squibb.
- Copyright © 2013 by Jeffrey Weber