Outcomes of patients with metastatic melanoma treated with immunotherapy prior to or after BRAF inhibitors
Presented in part at the 48th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, June 1-5, 2012, and at the 27th Annual Meeting of the Society of Immunotherapy of Cancer, Bethesda, MD, October 26-28, 2012.
See editorial on pages 1617–1619, this issue.
Abstract
BACKGROUND
The immunotherapy (IT) agents ipilimumab and interleukin-2 as well as BRAF inhibitors (BRAFi) vemurafenib and dabrafenib, with or without trametinib (MEK inhibitors), are all FDA-approved treatments for BRAF metastatic melanoma, but there are few studies to guide optimal sequencing. This retrospective analysis describes the outcomes of patients treated with either BRAFi before IT or IT before BRAFi.
METHODS
A cohort of patients treated with BRAFi alone or with MEK inhibitor was retrospectively identified. Response rate (RR), overall survival (OS), and progression-free survival (PFS) were evaluated for the entire cohort, subdivided by BRAFi prior to or after IT.
RESULTS
RR and median PFS and OS calculated from commencement of BRAFi following IT (N = 32) were 57%, 6.7 months (95% confidence interval [CI] = 4.3-9.1 months), and 19.6 months (95% CI = 10.0-undefined months), respectively; whereas for BRAFi initially (N = 242) were 66%, 5.6 months (95% CI = 4.7-6.8 months), and 13.4 months (95% CI = 10.1-17.0 months). Results were similar when controlled for prognostic variables. A total of 193 patients discontinued BRAFi, with OS of 2.9 months (range of 1.8-4.4 months) from day of BRAFi discontinuation. Forty patients subsequently received IT with ipilimumab. Only half could complete 4 doses of ipilimumab; PFS with ipilimumab was 2.7 months (95% CI = 1.8-3.1 months) and OS was 5.0 months (95% CI = 3.0-8.8 months).
CONCLUSIONS
In this retrospective analysis, prior treatment with IT does not appear to negatively influence response to BRAFi. Outcomes for IT with ipilimumab following BRAFi discontinuation are poor. Randomized controlled trials are needed to define if sequencing IT prior to BRAFi therapy is superior to sequencing BRAFi prior to IT. Cancer 2014;120:1695–1701. © 2014 American Cancer Society.
INTRODUCTION
Ipilimumab, a monoclonal antibody against cytotoxic T-lymphocyte antigen 4, and vemurafenib, a potent inhibitor of mutant BRAF, have been shown to improve the overall survival (OS) of patients with metastatic melanoma in phase 3 randomized controlled trials.1-3 Consequently, both drugs have been approved for the treatment of unresectable stage III and stage IV melanoma and have changed the standard of care for these patients. A second BRAF inhibitor (BRAFi), dabrafenib, and the MEK inhibitor (MEKi) trametinib also improve response rates and survival compared with chemotherapy in patients with metastatic BRAF-mutant melanoma and more recently have received US regulatory approval in May 2013.4, 5 Furthermore, the dabrafenib/trametinib combination has improved efficacy over single-agent dabrafenib.6 In the context of this progress, a clinical dilemma has emerged regarding how best to sequence these agents in treating patients with BRAF-mutant melanoma.
There is little evidence to guide sequencing decisions, other than comparisons across studies and ad hoc analyses of small subsets of patients in larger studies or compassionate access programs.7 The patterns and rates of response are different when comparing immunotherapy (IT) with BRAFi therapy. Only a small subgroup of those treated with IT will benefit, but those who do may experience prolonged remissions. This concept was first described with the use of interleukin-2 (IL-2), a therapy associated with a 16% to 22% response rate and a durable (> 2 years) remission rate that approaches 10%.8 In small series of patients, adoptive T-cell transfer, which is a more intensive IT, has resulted in complete remissions in up to 40% of cases, even when other ITs have failed.9 This approach remains investigational and is available at a limited number of centers. Treatment with ipilimumab is associated with response rates of 15% and 11% in first-line and second-line phase 3 studies and 2-year OS of 28.5% and 23.5%, respectively.1, 2 Recent reports suggest that treatment with inhibitors of the second identified immune checkpoint, programmed death 1 (PD1):PD ligand 1 (PDL1), also are associated with clinical benefit that likely exceeds that of high-dose IL-2 or ipilimumab. These agents remain under investigation; however, the response rates, approaching 30% to 50% or more, and durability of these responses suggest that these agents will likely be approved by the US Food and Drug Administration (FDA) by the end of 2014.10-13
Compared with IT, BRAFi therapy is associated with more frequent and earlier responses (single-agent rate ∼50%, BRAFi and MEKi combination rate ∼75% relative to 10% to 20% response rates for IL-2 or ipilimumab).1-7 Treatment with a BRAFi is associated with symptomatic relief within days of commencing treatment, whereas IT and ipilimumab, in particular, may take several weeks or months before responses and symptom reduction are realized.14, 15 Although BRAF inhibitors produce a disease control rate > 90%, the median progression-free survival (PFS) is only 6 months.3, 4 However, some patients in the initial phase 1 study with vemurafenib have durable responses with a reported 2-year survival of 36%, and a subset of these surviving patients remain on drug without progression.10
There are few published data regarding outcomes of the sequence of BRAFi therapy followed by or prior to IT, with ipilimumab or other agents. In order to determine if the sequence of treatment with a BRAFi and IT affects patient outcomes, we retrospectively examined the response rate and PFS in patients with BRAF-mutant metastatic melanoma treated sequentially with BRAFi, with or without IT. Two cohorts of patients were examined from 5 international melanoma centers; those who received IT prior to a BRAFi, and those who received a BRAFi first. Subsequently, a subset of patients who received BRAFi and then ipilimumab were analyzed.
MATERIALS AND METHODS
Patient Population
A cohort was retrospectively identified of 274 patients with BRAF-mutant metastatic melanoma, who were consecutively enrolled in clinical trials of BRAF inhibitors or treated with standard-of-care vemurafenib at the Dana-Farber Harvard Cancer Center (Beth Israel Deaconess Medical Center, Dana-Farber Cancer Institute, Massachusetts General Hospital Cancer Center) in the United States; Melanoma Institute of Australia; or Westmead Hospital, Sydney, Australia, between 2009 and 2012. Inclusion required documentation of BRAF V600 mutation, measurable disease, and treatment with dabrafenib, vemurafenib, or the combination of dabrafenib and trametinib. Patients who had received prior treatment with other BRAF inhibitors were excluded. Other lines of therapy were allowed before or after the BRAFi treatment, including surgery, radiation, and chemotherapy.
Treatment
All patients received a BRAFi—vemurafenib, dabrafenib, or dabrafenib/trametinib—either as per clinical trial or compassionate use protocol or via commercial use of vemurafenib. Patients may have received one or more of the following treatments that were considered immunotherapies: commercial high-dose IL-2, ipilimumab (clinical trial, compassionate use, or commercial), or clinical trials of PD-1 antibodies (nivolumab, MK-3475, and CT-011), anti-PDL1 antibody (MPDL-3280A), or adoptive T-cell therapy. A subset of patients received IT as first-line therapy before initiating BRAFi, termed “IT initially,” and a different subset of patients received IT following discontinuation of BRAFi, termed “post-BRAFi.” Continuation with BRAFi after progression was allowed on several of the clinical trial protocols, and reinstitution of vemurafenib following removal from trial was feasible after FDA approval.
Statistical Analysis
Outcomes to BRAFi therapy were analyzed by using a retrospective subset analysis. Fisher's exact test calculated P values to compare characteristics of the subsets treated with immunotherapy in the first line prior to BRAFi, “IT initially” versus a larger subset which had not received prior immunotherapy, “BRAFi initially.” Outcome endpoints included objective response rate (RR) as defined by Response Evaluation Criteria in Solid Tumors (RECIST) 1.0, PFS, and OS. BRAFi PFS or OS were defined from the start of BRAFi until progression as defined by RECIST, death, or last documented contact. The Kaplan-Meier method was used to estimate PFS and OS with 95% confidence interval (CI), and survival curves were analyzed with log-rank analysis. A Cox proportional hazards model was used to determine, in a multivariate model, the significance of prognostic variables at time of BRAFi initiation for BRAFi PFS and OS. “Post-BRAFi” ipilimumab PFS or OS were defined from start of ipilimumab until progression using immune-related response criteria, death, or last documented contact, and Kaplan-Meier method was used to estimate PFS and OS with 95% CI. There was no internal control group for this comparison.
RESULTS
Patient Characteristics
A total of 274 patients received BRAFi; 117 received vemurafenib, 99 received dabrafenib, and 58 received dabrafenib/trametinib. The total cohort was subdivided into 32 patients who received immunotherapy in the first-line “IT initially” (15 patients received high-dose IL-2 alone, 10 patients received ipilimumab alone, 1 patient received PD-1 antibody alone, and 6 patients received IL-2 followed by monoclonal antibody) versus 242 patients who received BRAFi as first-line therapy without prior IT “BRAFi initially.” Lactate dehydrogenase (LDH) was statistically different between the 2 groups (Table 1) and therefore is a significant variable in the multivariate analysis. The choice of BRAFi was also statistically different between the 2 groups, because BRAF inhibitors were not commercially available in Australia and were accessed only via clinical trials or compassionate access programs, eg, 62% patients received BRAFi on a clinical trial, 27% on compassionate access, and 11% commercially. In addition, ipilimumab remains unfunded in Australia and was available via a compassionate program up until April 2012, thus restricting treatment options once a patient had progressed on BRAFi to compassionate access in Australia. In the United States, ipilimumab was available through compassionate access until 2011 when it was FDA-approved and became commercially available.
Characteristic | IT Initially (N = 32) | BRAFi Initially (N = 242) | Total (N = 274) | P |
---|---|---|---|---|
BRAFi therapy | ||||
Vemurafenib | 26/32 (81%) | 91/242 (38%) | 117/274 (43%) | 0 |
Dabrafenib | 0/32 (0%) | 99/242 (40%) | 99/274 (36%) | |
Dabrafenib/trametinib | 6/32 (19%) | 52/242 (22%) | 58/274 (21%) | |
ECOG PS | .54 | |||
0 | 14/32 (44%) | 135/242 (56%) | 149/274 (54%) | |
1 | 14/32 (44%) | 83/242 (34%) | 97/274 (35%) | |
2 | 3/32 (9%) | 17/242 (7%) | 20/274 (7%) | |
3/4 | 1/32 (3%) | 7/242 (3%) | 8/274 (3%) | |
Stage | .69 | |||
III | 0/32 (0%) | 6/237 (2%) | 6/274 (2%) | |
IVA | 4/32 (13%) | 19/237 (8%) | 23/274 (9%) | |
IVB | 2/32 (7%) | 21/237 (9%) | 23/274 (9%) | |
IVC | 24/32 (80%) | 191/237 (81%) | 215/274 (80%) | |
Lactate dehydrogenase | 0 | |||
Normal | 29/30 (97%) | 141/228 (62%) | 170/274 (66%) | |
Elevated | 1/30 (3%) | 87/2728(38%) | 88/274 (34%) | |
Prior brain metastases | .19 | |||
Yes | 2/30 (8%) | 51/239 (21%) | 53/274 (20%) | |
No | 23/30 (92%) | 188/239 (79%) | 211/274 (80%) | |
Prior chemotherapy | .34 | |||
Yes | 24/32 (75%) | 199/242 (82%) | 223/274 (81%) | |
No | 8/32 (25%) | 43/242 (18%) | 51/274 (19%) |
- Abbreviations: IT, immunotherapy; ECOG PS, Eastern Cooperative Oncology Group performance status.
Response to BRAFi
In univariate analysis comparing the groups, “IT initially” (n = 32) versus “BRAFi initially” (n = 242), response to BRAFi was not inferior when administered after IT. As noted in Table 2, the relative risk was 57% versus 66% for the groups “IT initially” versus “BRAFi initially,” with a non–statistically significant Fisher's exact P value of .31. PFS was 6.7 months (95% CI = 4.3-9.1 months) versus 5.6 months (95 % CI = 4.7-6.8 months) for the groups “IT initially” versus “BRAFi initially,” with a non–statistically significant log-rank P value of .43 (Fig. 1B). OS was 19.6 months (95% CI = 10.0-undefined months) versus 13.4 months (95% CI = 10.1-17.0 months) for the groups “IT initially” versus “BRAFi initially,” with a non–statistically significant log-rank P value of .40 (Fig. 1B).
Characteristic | IT Initially (N = 32) | BRAFi Initially (N = 242) | Total (N = 274) |
---|---|---|---|
Response | |||
CR | 0/32 (0%) | 9/233 (4%) | 9/274 (3%) |
PR | 17/32 (53%) | 144/233 (62%) | 161/274 (61%) |
SD | 7/32 (22%) | 54/233 (23%) | 61/274 (23%) |
PD | 6/32 (19%) | 26/233 (11%) | 32/274 (12%) |
Progression | |||
No | 11/32 (34%) | 57/242 (24%) | 68/274 (25%) |
Yes | 21/32 (66%) | 185/242 (76%) | 206/274 (75%) |
Discontinuation | |||
No | 13/32 (41%) | 68/242 (28%) | 81/274 (30%) |
Yes | 19/32 (59%) | 174/242 (72%) | 193/274 (70%) |
Death | |||
No | 17/32 (53%) | 113/242 (47%) | 130/274 (47%) |
Yes | 15/32 (47%) | 129/242 (53%) | 144/274 (53%) |
- Abbreviations: CR, complete response; IT, immunotherapy; PD, progressive disease; PR, partial response; SD, stable disease.
Because LDH and choice of BRAFi differed between populations, a Cox proportional hazards model was used to assess the contribution of initial IT (Table 3). The variables, in addition to prior treatment with IT, included LDH within normal limits or elevated, performance status, stage, previously treated stable brain metastatic disease, treatment with vemurafenib versus dabrafenib versus dabrafenib/trametinib. Prior IT was associated with a nonsignificant improvement in PFS and OS (P values of .39 and .86 for PFS and OS, respectively). Choice of BRAFi, prior brain metastasis, and AJCC stage IV M1a versus M1b versus M1c disease were likewise not significantly different. In contrast, elevated LDH and higher Eastern Cooperative Oncology Group performance status were independent predictors of a shorter PFS and OS.
Characteristic | PFS | OS | ||||
---|---|---|---|---|---|---|
HR | CI | P | HR | CI | P | |
BRAFi vs IT initially | 0.94 | (0.5-1.9) | 0.86 | 0.78 | (0.4-1.4) | .39 |
Dabrafenib | 1.34 | (0.8-2.2) | 0.22 | 1.44 | (1-2.2) | .08 |
Combination | 0.71 | (0.4-1.3) | 0.28 | 1.06 | (0.7-1.6) | .79 |
Elevated LDH | 1.7 | (1.1-2.6) | 0.01 | 1.65 | (1.2-2.3) | .01 |
Brain metastases | 0.96 | (0.6-1.5) | 0.85 | 1.11 | (0.8-1.6) | .60 |
ECOG PS | ||||||
1 | 1.55 | (1-2.3) | 0.04 | 1.95 | (1.4-2.8) | .00 |
2 | 2.91 | (1.4-6.2) | 0.01 | 2.2 | (1.1-4.4) | .03 |
3/4 | 119 | (31-452) | 0.00 | 49 | (17-144) | .00 |
Stage | ||||||
IVA | 1.01 | (0.1-9.3) | 0.99 | 1.29 | (0.5-3.7) | .63 |
IVB | 2.32 | (0.3-19) | 0.44 | 0.86 | (0.3-2.6) | .79 |
IVC | 3.47 | (0.5-25) | 0.22 | 1.58 | (0.6-3.9) | .32 |
- Abbreviations: BRAFi, BRAF inhibitor; CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; IT, immunotherapy; LDH, lactate dehydrogenase; OS, overall survival; PFS, progression-free survival.
Outcome Following Disease Progression on or Discontinuation of BRAFi
Of the cohort of 274 patients treated with BRAFi, 193 patients discontinued therapy (Table 2), including 176 patients (91%) who suspended BRAFi due to disease progression and 17 patients (9%) who discontinued BRAFi due to toxicity or personal choice. Of the 81 patients who remain on BRAFi treatment, 30 patients have progressed but remain on therapy following progression, and 51 patients have had sustained responses and will continue to be followed clinically to determine outcomes.
Following discontinuation of BRAFi (n = 193), median time to death was 3.0 months (95% CI = 2.1-4.8 months). Initial poorer performance status and discontinuation due to disease progression were significant hazards. Of the 193 who ceased BRAFi, only 61 patients (35%) went on to receive subsequent lines of therapy despite progression on BRAFi in 58 of 61 cases (95%). For the 61 patients who received additional treatment, second-line therapy included commercial or compassionate use ipilimumab (33 patients), anti–PD-1 antibody trials (4 patients), BRAF- and MAPK-directed therapy trials (14 patients), clinical trial compound E7080 (7 patients), and chemotherapy (2 patients).
Post-BRAFi Ipilimumab Therapy
Outcomes of patients who received ipilimumab following BRAFi are summarized in Table 4. Of the 193 patients who discontinued BRAFi, a subset of 40 patients were subsequently treated with ipilimumab (33 patients with second-line, 7 patients with third-line). Because not all 193 patients were restaged, it is difficult to statistically compare this subset of 40 patients relative to the larger cohort. Of the 27 patients with recorded pretreatment LDH and recorded performance status, 16 (59%) had an elevated LDH and 10 (36%) had a performance status of 2 or greater. No responses to post-BRAFi ipilimumab were seen; 34 (94%) patients had progressive disease, and 2 had stable disease (6%). The full 4-dose treatment of ipilimumab was received by only 50% of patients due to clinical deterioration or toxicity. Twenty-four of the 40 patients are deceased, and the median PFS and OS for post-BRAFi ipilimumab therapy was 2.7 months (95% CI = 1.8-3.1 months) and 5.0 months (95% CI = 3.0-8.8 months), respectively (Fig. 2). Confounding is significant; for example, OS for post-BRAFi ipilimumab was greater than a year for 3 patients: 1 had a response to subsequent reinitiation of vemurafenib with response, 1 had a response to BRAFi/MEKi, and 1 had stable disease with reinitiation of vemurafenib.
Characteristic | Patients (N = 40) |
---|---|
Response | |
CR | 0/36 (0%) |
PR | 0/26 (0%) |
SD | 2/36 (6%) |
PD | 34/36 (94%) |
Progression | |
No | 4/40 (10%) |
Yes | 36/40 (90%) |
No. of doses received | |
1 | 6/40 (15%) |
2 | 8/40 (20%) |
3 | 6/40 (15%) |
4 | 20/40 (50%) |
Death | |
No | 16/40 (40%) |
Yes | 24/40 (60%) |
- Abbreviations: CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.
DISCUSSION
This retrospective analysis is the largest study to examine sequencing of treatment with BRAF inhibitors and IT in patients with BRAF-mutation–positive metastatic melanoma. The results of this multicentered case series have important clinical ramifications. First, using a multivariate model, there was no statistical difference in OS or PFS based on whether patients first received a BRAFi or IT. Second, the response rate of BRAFi therapy was similar in patients who did or did not receive prior IT. Third, the median OS after BRAFi discontinuation was very short. Fourth, ipilimumab therapy following BRAF inhibitors was associated with no tumor responses and poor survival.
Patients treated with ipilimumab after a BRAFi had poorer outcomes than the results reported for ipilimumab in both first- and second-line settings. It is not clear if this result is because these patients had largely discontinued BRAFi due to progression, and in the setting of rapid clinical deterioration and poor performance status have insufficient lifespan to benefit from ipilimumab, which typically requires weeks or months to show response, or whether there is a biological explanation such as an intrinsic resistance to IT after a BRAFi. One potential biological explanation of reduced IT benefit after BRAFi therapy is related to tumor melanocytic antigen expression and T-cell infiltration. Namely, BRAFi therapy has been shown to be associated with an increase in tumor melanocytic antigen expression and tumor CD-8 T-cell infiltration; whereas time-of-progression biopsies have demonstrated that tumor melanocytic expression and T-cell infiltration was no longer evident.16, 17 It is critical that these concepts be tested further in well-controlled randomized clinical trials, which investigate optimal sequencing and timing of therapies and ideally that include correlative analyses that help to identify why patients may or may not respond to therapies.
In the absence of randomized trials comparing the order of therapies, this retrospective analysis provides some data from which to base decisions regarding sequencing of therapies. The results of this study support the use of IT prior to BRAFi therapy in select patients, because there is no apparent detriment to the effectiveness of either agent when their use is sequenced in this manner, although the results do not refute that front-line BRAF inhibitor therapy is a reasonable alternative. The results also show that survival is very poor for those with BRAF-mutant melanoma who are refractory to BRAF inhibitors. Treatment after BRAFi failure has become the greatest unmet need in this patient population.
This study is limited because it is retrospective and thus susceptible to recall bias and selection bias. Although the inclusion of every patient treated with a BRAFi at the 5 institutions from the prespecified time interval limits this bias, this cohort is unlikely to be representative of all patients with BRAF-mutant metastatic melanoma. Access to vemurafenib, dabrafenib, and ipilimumab via clinical trial, compassionate use, or commercial use was influenced by insurance and federal regulations, which differed between the multiple institutions over the interval 2009 to 2012, possibly contributing to bias in selection of therapy. Specifically, those who did not receive a BRAFi for whatever reason (lack of access prior to regulatory approval, durable response to surgery or immunotherapy, comorbidities, and so forth) are not included in this analysis. In addition, initial selection for first-line therapy was based on several factors including rate of disease progression, performance status, age, other comorbidities, presence of brain metastases, patient preference, clinical trial availability, and “fitness” for IT as judged by the treating physicians. Because many of these potential confounders are difficult to objectively extract from the medical record, they cannot be controlled in the multivariate analysis. These factors likely resulted in patients with a poorer prognosis, eg, rapidly progressive disease or brain metastases, receiving first-line BRAFi therapy.18 Only a well-designed randomized controlled trial to examine optimal sequencing can account for these potential confounders.
Despite these limitations, the data suggest that BRAF inhibitors are as effective after IT as they are prior to IT, and that IT should be considered first-line if a patient with BRAF-mutant metastatic melanoma is an “immunotherapy candidate.” Because recent reports suggest that inhibitors of the second identified immune checkpoint, PD1, are associated with as high or higher response rates, the rationale to treat with upfront IT may have increased support, although it is important to note that these agents may also be associated with more rapid responses and may be associated with a higher rate of clinical benefit at the time of BRAF therapy resistance.10-13 The optimal sequence strategy for patients with BRAF-mutant, metastatic melanoma is the subject of an industry-sponsored study exploring the effect of sequencing vemurafenib prior to ipilimumab (ClinicalTrials.gov identifier NCT01673854), as well as a US Cooperative Group trial (4613) that is comparing the sequence of the combination of dabrafenib and trametinib (DT) followed by the combination of nivolumab and ipilimumab (NI) versus NI followed by DT. In addition, multiple studies are investigating various combinations of BRAF inhibitors with IT with the aim of improving complete response rate and median survival (NCT01400451, NCT01656642, and NCT01767454). However, until these sequencing and combination studies are complete, melanoma treatment providers will continue to need to make treatment decisions regarding the administration sequence of BRAFi and IT.
FUNDING SUPPORT
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
Drs. McDermott and Hodi have been consultants/advisors to Roche-Genentech and Bristol-Myers Squibb; Dr. Flaherty has been a consultant/advisor to Roche-Genentech and GlaxoSmithKline; Dr. Kefford has been a consultant/advisor to Roche-Genentech, GlaxoSmithKline, and Bristol-Myers Squibb; Dr. Atkins has been a consultant/advisor to Bristol-Myers Squibb, GlaxoSmithKline, and Prometheus; and Dr. Long has been a consultant/advisor to Roche-Genentech, GlaxoSmithKline, Bristol-Myers Squibb, Amgen, and Novartis. All other authors made no disclosure.