You are here

Factors predictive of response, disease progression, and overall survival after dabrafenib and trametinib combination treatment: a pooled analysis of individual patient data from randomised trials

Lancet Oncol 2016; 17: 1743-54

Summary

Background

Dabrafenib plus trametinib treatment provides significant benefits over BRAF-inhibitor monotherapy in patients with BRAFV600E-mutant or BRAFV600K-mutant advanced melanoma; however, in many patients the disease progresses, leading to death. With many treatment options available, understanding clinical factors that predict long-term response and survival for treatments is important for optimisation of patient management. We aimed to identify clinical factors associated with long-term response and survival using pooled data from randomised trials of dabrafenib plus trametinib in patients with metastatic BRAF-mutant melanoma.

Methods

We did a retrospective individual data analysis based on all published randomised trials that included treatment-naive patients with BRAFV600E-mutant or BRAFV600K-mutant metastatic melanoma who received the approved dose of dabrafenib 150 mg twice daily plus trametinib 2 mg once daily. Data were pooled from patients in the BRF113220 (part C; March 26, 2010, to Jan 15, 2015), COMBI-d (May 4, 2012, to Jan 12, 2015), and COMBI-v (June 4, 2012, to March 13, 2015) randomised trials. Patients with untreated brain metastases were not permitted to enrol in these trials. Baseline factors, identified a priori based on known melanoma clinical or prognostic characteristics, were analysed for association with progression-free survival and overall survival using univariate and multivariate analyses and assessed for hierarchical effect on outcomes using regression tree analyses. We also analysed factors identified after baseline, on treatment, and at progression, for associations with survival after progression. The trials included in this analysis are registered with ClinicalTrials.gov: BRF113220, number NCT01072175; COMBI-d, number NCT01584648; COMBI-v, number NCT01597908.

Findings

617 patients were included in this analysis with a median follow-up of 20·0 months (range 0−48·0, IQR 10·1−24·8); 396 patients had progression events (ie, disease progression or death) and 290 patients had died. Median progression-free survival (11·1 months [95% CI 9·7−12·9]), median overall survival (25·6 months [23·1−34·3]), 1-year progression-free survival (48% [44–52]) and overall survival (74% [71–78]), and 2-year progression-free survival (30% [26–34]) and overall survival (53% [49–57]) were consistent with those in the individual trials. Patients with normal lactate dehydrogenase (LDH) concentration and fewer than three organ sites containing metastases (n=237) had the longest 1-year progression-free survival (68% [95% CI 62–74]) and overall survival (90% [87–94]) and 2-year progression-free survival (46% [40–54]) and overall survival (75% [70–81]), whereas patients with LDH concentration at least two times the upper limit of normal (n=70) had the shortest 1-year progression-free survival (8% [3–19]) and overall survival (40% [29–55]) and 2-year progression-free survival (2% [0–13]) and overall survival (7% [3–19]). Of patients with disease progression (n=379), survival after progression was longest in those with progression in baseline or new non-CNS lesions (n=205; median 10·0 months [95% CI 7·9−12·0]) and shortest in those with new CNS lesions or concurrent progression in baseline and new lesions (n=171; median 4·0 months [3·5−4·9]).

Interpretation

Several patient and clinical characteristics at and after baseline are associated with outcomes with dabrafenib plus trametinib, and durable benefit is possible with targeted treatment in defined patient subsets.

Funding

Novartis.

Introduction

About 40% of patients with advanced melanoma have activating BRAF mutations in their tumours, resulting in constitutive mitogen-activated protein kinase (MAPK) pathway signalling.1, 2, and 3 Dual MAPK pathway inhibition by targeting BRAF-mutant and MEK proteins with dabrafenib combined with trametinib improved response, progression-free survival, and overall survival in patients with BRAFV600E-mutant or BRAFV600K-mutant advanced melanoma compared with BRAF inhibitor monotherapy in three randomised clinical trials.4, 5, 6, 7, 8, and 9 Dabrafenib plus trametinib was assessed in the first-line setting in BRAF-inhibitor-naive patients in one phase 2 trial (BRF113220, part C only)4 and 5 and two phase 3 trials (COMBI-d6 and 7 and COMBI-v8, 9, and 10). For dabrafenib plus trametinib, overall response (complete plus partial response) was reported in 41 of 54 patients (76%, 95% CI not available) in BRF113220, 144 of 210 evaluable patients (69%, 62–75) in COMBI-d, and 231 of 351 (66%, 60–71) in COMBI-v (complete responses in eight [15%], 33 [16%], and 59 [17%], respectively), median progression-free survival was 9·4 months (95% CI 8·6−16·6), 11·0 months (8·0−13·9), and 12·6 months (10·7−15·5), and median overall survival was 25·0 months (95% CI 17·5−36·5), 25·1 months (19·2–not reached), and 25·6 months (22·6–not reached), respectively.4, 5, 6, and 9 Similarly, the combination of vemurafenib, a BRAF inhibitor, and cobimetinib, a MEK inhibitor, has impressive activity in patients with melanoma with activating BRAF mutations.11

 

Research in context

 

Evidence before this study

We searched PubMed up to Sept 6, 2016, for phase 1–3 clinical trials with the terms “combination”, “BRAF”, “MEK”, and “melanoma”, and identified 265 articles, eight of which were primary analyses or updated efficacy and safety analyses of phase 1, 2, or 3 clinical trials of combined BRAF and MEK inhibition in metastatic cutaneous melanoma. BRAF and MEK inhibition improved progression-free survival and overall survival in patients with BRAFV600-mutant melanoma, although the degree of treatment benefit varied. Studies of small groups of patients treated with dabrafenib and trametinib have explored clinical factors associated with response and progression, and identified that sex and lactate dehydrogenase (LDH) concentration were associated with prolonged survival. By contrast, analysis of baseline melanoma tissue has not yielded any tissue biomarker predictive of long-term survival. Pivotal phase 2 (BRF113220) and phase 3 (COMBI-d and COMBI-v) studies of dabrafenib plus trametinib have provided a large homogeneous dataset for assessment of clinical factors associated with response and survival in patients receiving these drugs.

Added value of this study

To our knowledge, this pooled analysis uses the largest dataset available for BRAF and MEK inhibitor combination treatment in melanoma. Efficacy data were consistent across trials, supporting pooling of the data for retrospective analyses. Using statistical analyses that identified risk factors and considered the potential interactions and hierarchy between them, baseline LDH concentration or number of organ sites containing metastases, or both, were identified as the best baseline predictors of progression-free survival and overall survival. Descriptive analyses showed that best response was associated with progression-free survival and overall survival, and although the baseline factors predictive of progression-free survival and overall survival also predicted response, baseline sum of tumour diameters most strongly predicted complete response. Duration of clinical benefit differed between the best and worst prognostic groups identified in this study, and patterns of progression also varied across the pooled population, with corresponding differences in survival after progression.

Implications of all the available evidence

These results provide evidence that patient and clinical factors at and after baseline can predict clinical outcomes in patients treated with dabrafenib plus trametinib and that durable responses are possible in subsets of patients with BRAF-mutant melanoma. These findings could be used to provide a framework for translational studies exploring melanoma drug resistance mechanisms and prolonged patient response and survival, which will play an important part in the design of future treatments and trials to further improve outcomes in patients with melanoma.

Concurrently, immune checkpoint inhibitors targeting PD-1 and CTLA-4 showed statistically and clinically significant improvements in overall survival in molecularly unselected populations of patients with advanced melanoma.12, 13, and 14 Existing data support the notion that these treatments achieve clinical benefit in patients with melanoma who have activating BRAF mutations.13, 14, and 15 However, in the absence of prospective trials directly comparing these treatments with MAPK inhibitors, how to select one treatment over another for patients with BRAF-mutant melanoma is unclear.

The homogeneity of the patient populations and results across the three randomised dabrafenib plus trametinib trials4, 5, 6, 7, 8, and 9 has created an opportunity to improve our understanding of clinical factors associated with long-term benefit with this treatment, which is important for (1) assisting clinicians to decide between and advise patients about active treatments for BRAF-mutant melanoma (eg, BRAF and MEK inhibitor combination treatment, anti-PD-1, and anti-CTLA-4); (2) providing a framework for making appropriate comparisons of outcomes across different advanced melanoma drug treatments; and (3) identifying specific patient subgroups for further translational studies. Thus, in this retrospective individual data analysis based on all randomised trials that included patients treated with the approved dose of dabrafenib plus trametinib, we examined baseline and on-treatment factors in pooled data from the BRF113220, COMBI-d, and COMBI-v trials to identify characteristics associated with long-term response and survival. We also sought to establish which of the independent factors associated with clinical outcomes were most predictive of prolonged response and survival. Finally, we analysed patterns of progression and survival after progression in patients receiving dabrafenib plus trametinib to further understand long-term patient outcomes.

Methods

Study design and participants

All treatment-naive patients with BRAFV600E-mutant or BRAFV600K-mutant metastatic melanoma treated with dabrafenib plus trametinib in the intention-to-treat population of randomised trials BRF113220 (part C), COMBI-d, and COMBI-v were included in this analysis. In addition to similarities noted between Kaplan-Meier estimated survival curves, the cohorts from the studies used in this analysis all used the same oral dosage of dabrafenib 150 mg twice daily plus trametinib 2 mg once daily, the same assessment schedule, and were done during the same general time period with regard to treatment guidelines and available subsequent treatments.4, 5, 6, 7, 8, and 9 BRF113220 (part C) was an open-label, randomised phase 2 study4 of dabrafenib plus trametinib versus dabrafenib monotherapy 150 mg once daily in which patients were screened for enrolment between March 26, 2010, and July 7, 2011. COMBI-d was a double-blind, randomised phase 3 study6 and 7 of dabrafenib plus trametinib versus dabrafenib 150 mg twice daily plus placebo in which patients were screened for enrolment between May 4, 2012, and Nov 30, 2012. COMBI-v was an open-label, randomised phase 3 study8 and 9 of dabrafenib plus trametinib versus vemurafenib monotherapy 960 mg twice daily in which patients were screened for enrolment between June 4, 2012, and Oct 7, 2013. All patients received dabrafenib plus trametinib until disease progression, death, unacceptable toxicity, or withdrawal of consent. Enrolled patients in all trials had an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1 and could have had previous systemic treatment in the adjuvant setting; however, previous ipilimumab must have ended at least 8 weeks before randomisation. Patients with untreated brain metastases were not permitted to enrol in these trials. Full study protocols for these trials were included in previous publications.4, 6, 7, and 8 All participants in each study provided written informed consent. The protocol for each study was approved by the institutional review board or human research ethics committee at each site. Follow-up times have been updated from the original publications for each study,4, 7, and 8 and so we have used the more recently reported data cutoff dates in this pooled analysis, which were Jan 15, 2015, for BRF113220, part C,5 Jan 12, 2015, for COMBI-d,6 and March 13, 2015, for COMBI-v.9

Procedures

Factors for analyses identified a priori included baseline clinical characteristics and known prognostic factors for advanced or metastatic melanoma (stage IIIC unresectable or stage IV melanoma; appendix p 1).16 Baseline factors analysed were age, sum of target lesion diameters, lactate dehydrogenase (LDH) concentration, sex, ECOG PS, presence of visceral disease, number of organ sites containing metastases (any organ with at least one metastasis was counted as a single organ site), previous adjuvant immunotherapy, previous adjuvant ipilimumab, stage of disease, site of disease (visceral or non-visceral), and BRAF mutation genotype (BRAFV600E or BRAFV600K). Age was included as a continuous variable because a statistical test used to investigate non-linear trends between clinical outcomes and age did not provide any evidence to suggest that separation into age groups was a better fit than continuous age.

We also analysed the association of the following patient and clinical characteristics during treatment or that were recorded at the time of progression (appendix p 1): LDH concentration, ECOG PS, sum of target lesion diameters, progression pattern (new lesion or not; CNS or non-CNS), enlargement of baseline lesions (target or non-target, or both), and best response.

Tumour response was assessed with Response Evaluation Criteria In Solid Tumors (RECIST) version 1.1 in all studies (best response is defined as a complete or partial response), and progression-free survival (measured from the start of dabrafenib plus trametinib treatment to time of progression or death) and overall survival (measured from the start of dabrafenib plus trametinib treatment to time of death) endpoints were defined as previously reported.4, 6, 7, and 8

Statistical analysis

We assessed the prognostic value for each individual factor (univariate analysis), each factor in conjunction with other factors (multivariate analysis), and interactions between variables (regression tree analysis). Factors assessed after baseline were only considered when investigating characteristics that might affect outcomes after progression. We plotted baseline LDH concentration against LDH concentration at progression for patients with documented progression to assess whether LDH could be a surrogate marker for predicting survival after progression. We explored factors predictive of patient outcomes based on patient response using descriptive subgroup stratification of median progression-free survival or overall survival (medians or at specific timepoints [eg, 1 year or 2 years]) estimated by Kaplan-Meier time-to-event analyses, in which median outcomes were defined as the shortest survival time in months in which the estimated survival probability was 0·5 or lower. We used the Brookmeyer-Crawley method to estimate 95% CIs for medians.

For univariate analyses, we used Cox proportional hazards models17 to model independent predictors of progression or death; hazard ratios (HRs) and 95% CIs for each risk factor were provided for each variable. For multivariate analyses, we used Cox proportional hazards models17 to jointly model all risk factors simultaneously as predictors of progression or death. HRs were based on maximum likelihood estimates. We calculated two-sided χ2 p values based on Wald statistics for univariate and multivariate analyses. Homogeneous prognostic subgroups were identified by Poisson regression tree for categorical outcomes and exponential regression tree for survival outcomes (ie, progression status and death).18 and 19

The regression tree analysis recursively partitioned patients into binary subgroups that best maximised differences between outcomes, and all possible splits for each factor were assessed at each partition and interactions between different variables were assessed exhaustively. The size of the dataset used in this study was not large enough to form a completely separate validation set without losing vital information or avoiding idiosyncrasies of data (ie, noise or artifacts in the data). Thus, to minimise overfitting of idiosyncrasies in data, we cross-validated the regression tree model using a method that mimics prediction of outcome in a future patient when building the model.20 We used surrogate variables in the event of a missing value to allow inclusion of all available data. The regression tree approach is meant to supplement the multivariate approach, because it maps the hierarchy of predictive factors. Similar to the multivariate approach, regression trees take into account all factors simultaneously; however, instead of focusing on variation in the data, regression trees focus on prediction of the outcome of interest by identifying the most influential factor among all factors assessed to divide data into two risk groups. The process is then repeated in each subset of the data recursively. This ordering of factors provides extra insight into the results identified by multivariate analysis by moving beyond relative significance based on a p value, which can often confound the importance of factors identified through multivariate analyses. All statistical analyses were done in SAS (version 9.3) and R (version i386 3.2.1).

The trials included in this analysis are registered with ClinicalTrials.gov: BRF113220, number NCT01072175; COMBI-d, number NCT01584648; and COMBI-v, number NCT01597908.

Role of the funding source

The study was designed by the authors and the funder of the study. Data were collected by the study site staff and monitored by the funder. The funder was also involved in data analysis, data interpretation, and writing of the report. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results

617 patients randomly assigned to dabrafenib plus trametinib in BRF113220 (part C; n=54), COMBI-d (n=211), and COMBI-v (n=352) were included in this analysis (appendix pp 1, 6), with a median follow-up of 20·0 months (range 0−48·0, IQR 10·1−24·8). Median follow-up was 24·4 months (IQR 13·4−43·5) in the BRF113220 trial, 19·7 months (10·0−26·0) in the COMBI-d trial, and 19·8 months (9·5−23·9) in the COMBI-v trial. Kaplan-Meier estimated overall survival curves and baseline characteristics were similar across trials (figure 1A; appendix p 2). In the pooled population of 617 patients there were 396 progression events (ie, disease progression or death; 42 in the BRF113220 trial, 139 in the COMBI-d trial, and 215 in the COMBI-v trial), median progression-free survival was 11·1 months (95% CI 9·7−12·9), 1-year progression-free survival was 48% (44–52) and 2-year progression-free survival was 30% (26–34; figure 1B). In the pooled population, 290 of 617 patients had died (36 in the BRF113220 trial, 99 in the COMBI-d trial, and 155 in the COMBI-v trial). Median overall survival was 25·6 months (95% CI 23·1−34·3), 1-year overall survival was 74% (71–78), and 2-year overall survival was 53% (49–57; figure 1C). 416 (67%) of 617 patients had a RECIST response (100 [16%] with a complete response and 316 [51%] with a partial response), 150 (24%) had stable disease, and 35 (6%) had progressive disease (appendix pp 3, 6).

gr1

Figure 1

Kaplan-Meier estimates of survival

All patients had at least 24 months of follow-up between the time of random allocation and data cutoff. Any survival estimates beyond 2 years should be interpreted with caution. (A) Cross-trial comparison of overall survival in the individual studies included in the pooled analysis. (B) Pooled progression-free survival and (C) pooled overall survival for patients treated with dabrafenib plus trametinib.

 

Baseline factors associated with progression-free survival and overall survival in univariate and multivariate analyses are shown in the table. Compared with patients with LDH concentration at least two times the ULN, those with normal LDH concentration and those with at least one to less than two times the ULN had improved progression-free survival and overall survival by univariate analysis (appendix p 3). Multivariate analysis also showed a strong association of LDH concentration with progression-free survival and overall survival (table). Other factors with significant associations with progression-free survival and overall survival by both univariate and multivariate analyses were ECOG PS, number of organ sites containing metastasis, and sex (table). Factors not associated with progression-free survival or overall survival were BRAF mutation status, previous adjuvant immunotherapy, and previous adjuvant ipilimumab by univariate analysis, and baseline disease stage, sum of lesion diameter, previous adjuvant immunotherapy, and previous adjuvant ipilimumab by multivariate analysis (table).

Table

Univariate and multivariate analyses of baseline factors associated with progression-free survival and overall survival in patients treated with dabrafenib and trametinib

 

Patients (n) Progression-free survival Overall survival
Number of events Median (95% CI), months Univariate analysis Multivariate analysis Number of events Median (95% CI), months Univariate analysis Multivariate analysis
HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value
LDH category*
Normal 397 224 17·4 (15·5–21·0) 0·22 (0·17–0·30) <0·0001 0·29 (0·21–0·41) <0·0001 140 45·5 (31·6–NE) 0·16 (0·11–0·22) <0·0001 0·23 (0·16–0·33) <0·0001
≥1 to <2 × ULN 147 108 7·4 (5·8–9·2) 0·49 (0·36–0·67) <0·0001 0·55 (0·39–0·76) 0·0003 94 17·0 (13·2–20·1) 0·42 (0·30–0·59) <0·0001 0·49 (0·35–0·69) <0·0001
≥2 × ULN 69 62 5·5 (4·3–5·7) .. .. .. .. 54 8·8 (7·1–12·8) .. .. .. ..
Baseline ECOG PS
0 438 264 14·1 (12·1–15·9) 0·57 (0·46–0·70) <0·0001 0·69 (0·55–0·86) 0·0009 172 33·0 (29·0–NE) 0·43 (0·34–0·54) <0·0001 0·53 (0·42–0·68) <0·0001
≥1 174 130 7·3 (5·7–9·1) .. .. .. .. 116 13·5 (11·1–17·6) .. .. .. ..
Number of organ sites containing metastasis
<3 313 177 16·6 (14·8–21·1) 0·54 (0·44–0·66) <0·0001 0·78 (0·62–0·99) 0·041 106 45·5 (33·0–NE) 0·40 (0·32–0·51) <0·0001 0·63 (0·48–0·83) 0·0011
≥3 302 219 8·1 (7·4–9·4) .. .. .. .. 184 16·4 (14·6–19·0) .. .. .. ..
Site of disease§
Visceral only 156 98 12·9 (9·5–15·6) 0·80 (0·63–1·02) 0·077 NA NA 62 NE (24·1–NE) 0·64 (0·48–0·86) 0·0034 NA NA
Non-visceral only 119 61 21·3 (15·9–NE) 0·54 (0·40–0·71) <0·0001 NA .. 40 NE (NE–NE) 0·50 (0·35–0·71) <0·0001 NA ..
Visceral and non-visceral 287 195 9·2 (7·4–11·0) .. .. .. .. 152 19·1 (16·3–24·1) .. .. .. ..
Baseline disease stage
III (any), IVM1a, or IVM1b 215 111 19·8 (16·2–27·8) 0·51 (0·41–0·63) <0·0001 0·79 (0·61–1·03) 0·080 66 NE (34·3–NE) 0·42 (0·32–0·55) <0·0001 0·76 (0·55–1·04) 0·088
IVM1c 401 285 9·1 (7·4–9·5) .. .. .. .. 224 19·0 (17·0–23·1) .. .. .. ..
Sum of lesion diameter
<median (58 mm) 302 174 17·4 (15·3–21·1) 0·54 (0·44–0·65) <0·0001 0·94 (0·74–1·20) 0·63 108 36·5 (31·6–NE) 0·45 (0·36–0·58) <0·0001 0·93 (0·70–1·22) 0·59
≥median (58 mm) 304 220 7·4 (7·1–9·2) .. .. .. .. 179 17·5 (15·6–20·7) .. .. .. ..
Sex
Female 264 153 15·5 (12·1–17·1) 0·73 (0·60–0·90) 0·0024 0·67 (0·54–0·82) 0·0001 104 33·0 (25·6–NE) 0·68 (0·53–0·86) 0·0015 0·65 (0·51–0·83) 0·0005
Male 353 243 9·2 (8·2–11·1) .. .. .. .. 186 21·1 (18·3–28·4) .. .. .. ..
Age (continuous variable)** 617 NA 11·1 (9·7–12·9) 0·92 (0·85–0·99) 0·033 0·90 (0·83–0·97) 0·0075 NA 25·6 (23·1–34·3) 0·92 (0·84–1·00) 0·056 0·89 (0·81–0·98) 0·012
BRAF mutation status
V600E only 538 342 11·2 (10·1–14·7) 0·82 (0·62–1·09) 0·18 0·67 (0·50–0·90) 0·0085 249 26·7 (23·5–37·0) 0·88 (0·63–1·22) 0·4298 0·65 (0·46–0·92) 0·016
V600K only, or V600E and V600K 79 54 9·2 (7·7–12·8) .. .. .. .. 41 23·7 (16·7–NE) .. .. .. ..
Previous adjuvant immunotherapy
No 552 352 11·2 (10·1–13·0) 0·83 (0·61–1·14) 0·25 0·91 (0·66–1·27) 0·59 258 25·6 (23·5–36·5) 0·84 (0·58–1·21) 0·34 0·94 (0·64–1·38) 0·74
Yes 65 44 9·1 (7·0–18·7) .. .. .. .. 32 21·1 (14·7–NE) .. .. .. ..
Previous adjuvant ipilimumab
No 608 388 11·1 (9·7–12·9) 0·82 (0·41–1·67) 0·59 0·85 (0·41–1·79) 0·67 283 26·1 (23·5–36·5) 0·54 (0·25–1·14) 0·11 0·55 (0·25–1·21) 0·14
Yes 9 8 8·6 (5·9–NE) .. .. .. .. 7 10·8 (9·3–NE) .. .. .. ..

* Data missing for four patients.

Data missing for five patients.

Data missing for two patients.

§ Analyses only include data for COMBI-d and COMBI-v (data missing for one patient from these studies). Site of disease data were not available for BRF113220; therefore, this variable was not included in the multivariate analysis.

Data missing for one patient.

Data missing for 11 patients.

** HR reflects increment of 10 years.

ECOG PS=Eastern Cooperative Oncology Group performance status. HR=hazard ratio. LDH=lactate dehydrogenase. NA=not applicable. NE=not estimable. ULN=upper limit of normal.

Regression tree analysis identified four prognostic groups of baseline factors to best predict progression-free survival: normal LDH concentration with fewer than three organ sites with metastasis; normal LDH concentration with at least three organ sites with metastasis; LDH concentration at least one to less than two times the ULN; and LDH concentration at least two times the ULN (figure 2A). Similarly, five prognostic groups of baseline factors were identified to predict overall survival by regression tree analysis: normal LDH concentration with fewer than three organ sites with metastasis; normal LDH concentration with at least three organ sites with metastasis; LDH concentration at least one to less than two times the ULN with an ECOG PS of 0; LDH concentration at least one to less than two times the ULN with and ECOG PS of at least one; and LDH concentration at least two times the ULN (figure 2B). Kaplan-Meier curves of progression-free survival and overall survival stratified by prognostic groups are shown in figures 2C and 2D. Patients with normal LDH concentration and fewer than three organ sites containing metastases (n=237) had the longest 1-year progression-free survival (68% [95% CI 62–74]) and overall survival (90% [87–94]) and 2-year progression-free survival (46% [40–54]) and overall survival (75% [70–81]), whereas patients with LDH concentration at least two times the upper limit of normal (n=70) had the shortest 1-year progression-free survival (8% [3–19]) and overall survival (40% [29–55]) and 2-year progression-free survival (2% [0–13]) and overall survival (7% [3–19]). Although M stage was a significant factor in explaining the variation in the univariate and multivariate models (table), by regression tree analysis, we found that composite M stage was not as predictive or prognostic as its elements (ie, LDH concentration; appendix pp 3, 9) and the number of organ sites containing metastasis (appendix pp 3, 4).

gr2

Figure 2

Regression tree analysis of factors most associated with progression-free survival and overall survival

Regression tree analysis of factors associated with (A) progression-free survival and (B) overall survival, and descriptive Kaplan-Meier estimated (C) progression-free survival and (D) overall survival by subgroups identified. ECOG PS=Eastern Cooperative Oncology Group performance status. LDH=lactate dehydrogenase. NE=not estimable. OS=overall survival. PFS=progression-free survival. ULN=upper limit of normal.

 

Because patients typically reach their optimum response soon after starting treatment, and all patients included in the pooled analysis had sufficient follow-up time to achieve a complete response (appendix p 3), time-to-event outcomes were analysed in subgroups stratified by best response. Progression-free survival and overall survival varied by best response to dabrafenib plus trametinib (appendix p 7). 396 of 617 patients had disease progression or died at the time of data cutoff. A progression event was experienced by 33 of 100 patients who achieved a complete response (median progression-free survival 38·0 months [95% CI 31·6–NE]; 2-year progression-free survival 68% [95% CI 59–79]), 217 of 316 with a partial response (12·6 months [11·1−14·7]), 110 of 150 with stable disease (4·6 months [3·8−5·6]), and 33 of 35 with progressive disease (1·8 months [1·8−1·9]). Death was reported in 15 patients with a complete response (median overall survival not reached; 2-year overall survival 88% [82–95]), 149 with a partial response (26·7 months [95% CI 23·7−37·0]), 91 with stable disease (12·5 months [10·6−17·0]), and 27 with progressive disease (6·8 months [6·3−16·8]; appendix p 7).

By regression tree analysis, baseline sum of target lesion diameter was the strongest predictor of complete response (34%; appendix p 8). However, the groups identified to predict progression-free survival and overall survival affected overall response (appendix p 4). Although the proportion of patients with an overall response was highest in the best prognostic group for progression-free survival and overall survival (normal LDH concentration and fewer than three organ sites with metastasis: overall response in 185 [78%, 95% CI 73–83] of 237), responses were also noted in the other, less favourable, groups: overall response in 109 (68%, 60–75) of 161 with normal LDH concentration and at least three organ sites with metastasis; 60 (65%, 55–74) of 93 with LDH concentration at least one to less than two times the ULN and ECOG PS of 0; 27 (48%, 35–61) of 56 with LDH concentration at least one to less than two times the ULN and ECOG PS of 1; and 35 (50%, 40–60) of 70 with LDH concentration at least two times the ULN (appendix p 4).

When assessing the raw data from patients for whom we had LDH values (n=613; missing data for four patients), an overall response was achieved by more patients with normal LDH concentration (294 [74%, 95% CI 70–78] of 397) than by patients with LDH concentration at least one to less than two times the ULN (87 [59%, 51–67] of 147) or by patients with LDH concentration at least two times the ULN (35 [51%, 39–63] of 69). More patients achieved a complete response in the normal LDH concentration group (91 [23%, 95% CI 19–27]) than in group with LDH concentration at least one to less than two times the ULN (nine [6%, 2–10]) or in the group with LDH concentration at least two times the ULN (zero), whereas the proportion of patients who achieved a partial response was similar across groups (203 [51%, 46–56] vs 78 [53%, 45–61] vs 35 [51%, 39–63]; appendix pp 3, 9).

Patients who had disease that progressed for reasons other than death while receiving dabrafenib plus trametinib (n=379) were categorised in four groups by patterns of progression. Patients with progression of baseline lesions (n=182) were divided into those with progression of baseline lesions only (n=114) and those with concurrent new lesions (n=68). Patients with progression in new lesions only (n=194) were divided into patients with new non-CNS lesions only (n=91) and those with new CNS lesions (ie, CNS, spinal cord, brain, or meninges; n=103; appendix p 4). We noted differences in survival after progression between progression site subgroups (figure 3A). Patients with progression in baseline lesions only and those with progression in new non-CNS lesions only (205 [54%] of 379) had longer survival after progression (median 10·0 months [95% CI 7·9−12·0]) than patients who had progression either with new CNS lesions or with both new lesions and progressive baseline lesions (171 [45%], median 4·0 months [3·5−4·9]; figure 3A). No baseline clinical factors were significantly associated with the development of CNS lesions (data not shown). Of the 379 patients who progressed, 33 (9%) subsequently received an anti-PD-1 treatment (nivolumab or pembrolizumab) and none received an anti-PD-L1 treatment.

gr3

Figure 3

Descriptive Kaplan-Meier estimated survival after progression

Survival after progression by (A) site of progression and (B) factors identified as predictive of survival after progression by regression tree analysis. Of 379 patients with documented progression, 11 had missing lactate dehydrogenase concentration at progression. ECOG PS=Eastern Cooperative Oncology Group performance status. NE=not estimable.

 

When baseline and on-treatment factors were analysed by regression tree analysis (appendix p 1), four prognostic factors related to characteristics at progression best predicted survival after progression, and baseline factors were not predictive, including baseline LDH concentration (appendix p 10). Patients with LDH at least two times the ULN at baseline had an LDH at progression that was less than that at baseline (appendix p 10). Prognostic groups for survival after progression were those patients with progression in baseline or new non-CNS lesions with lesion diameter at progression less than 42 mm (119 [31%] of 379 patients); progression in baseline or new non-CNS lesions with lesion diameter at progression 42 mm or greater (89 [23%]); progression in new CNS or new and baseline lesions with ECOG PS of 0 at progression (112 [30%]); and progression in new CNS or new and baseline lesions with ECOG PS 1 or higher at progression (59 [16%]; appendix p 10). Survival after progression stratified by these prognostic groups is shown in figure 3B). LDH concentration and ECOG PS at progression did not add predictive value for survival after progression (appendix p 5).

Discussion

This pooled analysis of three randomised controlled trials of dabrafenib plus trametinib treatment is, to our knowledge, the largest annotated dataset available of patients with BRAFV600E-mutant or BRAFV600K-mutant melanoma treated with combined BRAF inhibitor and MEK inhibitor treatment. In the pooled dataset, 2-year overall survival was 53% and 2-year progression-free survival was 30%, which is consistent with that previously reported for individual trials.4, 5, 6, 7, 8, 9, and 21

The regression tree analyses identified risk factors and their potential interactions, with easily understandable hierarchical results to facilitate clinical interpretation and use. In the most favourable subgroup identified, patients with normal LDH concentration and fewer than three organ sites with metastasis at baseline had an overall survival of 75% and progression-free survival of 46% at 2 years. Moreover, patients who achieved a complete response had a 2-year overall survival of 88% and 2-year progression-free survival of 68%. 34% of patients with a baseline sum of lesion diameters less than 58 mm and fewer than three organ sites with metastasis had a complete response. By contrast, 2% of patients with a sum of lesion diameter at least 58 mm and elevated baseline LDH concentration (≥1 × ULN) achieved a complete response; long-term outcomes remained poor, as shown by the 2-year overall survival of 7% and 2-year progression-free survival of 2% in patients with LDH concentration at least two times the ULN.

Consistent with findings from previous reports for targeted therapy and immunotherapies in melanoma,4, 5, 6, 7, 8, 9, 11, 12, 22, 23, 24, 25, 26, 27, 28, and 29 an unmet medical need for patients with an elevated LDH concentration was confirmed in this study because patients with elevated LDH concentration had lower progression-free survival and overall survival at 2 years, compared with those patients with normal LDH concentration. However, high overall response in patients with elevated LDH concentration suggests some short-lived clinical benefit. Whether LDH concentration is simply a marker of aggressive disease or has a direct causative role in tumour growth or response to treatment is not known, but is the subject of translational research. We showed that in patients with high baseline LDH concentrations, dabrafenib plus trametinib treatment led to decreased LDH concentrations, including at progression, which might have implications for drug sequencing or combinations with immunotherapy, since immunotherapy has substantially less activity in patients with high LDH concentration compared with normal LDH concentration.25, 26, 27, 28, and 29 Together, these results show that short-term benefit and rapid resistance are not universal with dabrafenib plus trametinib treatment, but instead are associated with key baseline factors associated with aggressive disease. Indeed, durable progression-free survival and overall survival were frequently achieved in patients with more favourable characteristics (ie, normal LDH concentration and fewer organ sites containing metastasis at baseline).

Patients who achieved a complete response with dabrafenib plus trametinib had favourable survival outcomes and over two-thirds of these patients remained progression free at data cutoff; further follow-up will be needed to fully characterise their long-term outcomes. Since the sum of target lesion diameter at baseline was the best predictor of a complete response by regression tree analysis, together these data raise the possibility that earlier diagnosis of stage IV melanoma, potentially with emerging diagnostic instruments (ie, blood-based biomarkers), might offer a benefit in terms of durability of treatment response, assuming that such approaches will identify patients with better prognostic factors. Treatment of earlier-stage melanoma (ie, in the adjuvant setting for resected stage III disease) could potentially prevent further development into stage IV melanoma. A phase 3 adjuvant trial of dabrafenib plus trametinib versus placebo is underway (NCT01682083).

Findings from a pooled retrospective analysis29 of 616 patients treated with pembrolizumab immunotherapy showed that baseline elevated LDH concentration and stage M1c disease were clinical predictors of inferior survival, with further refinement provided by the analysis of lymphocyte and eosinophil counts in peripheral blood. Although direct comparison of outcomes is not possible because of the use of different variables across studies, in our regression tree analysis model, 1-year overall survival was 90% in patients treated with dabrafenib plus trametinib with normal LDH concentration and fewer than three organ sites with metastasis at baseline, and was 76% in those with normal LDH and at least three organ sites with metastasis. The worst outcomes were in patients with baseline LDH at least two times the ULN. A potential caveat to comparing these studies is that the prevalence of patients with BRAFV600 mutations has consistently been only about 35% in most large immunotherapy trials,13, 14, and 30 because BRAF mutations have been associated with worse overall survival from diagnosis of stage IV disease before the development of contemporary treatments,2 and 3 and two previous studies31 and 32 identified associations of NRAS mutations (mutually exclusive to BRAF mutations) with improved outcomes with immunotherapy. Although the analysis of patients treated with pembrolizumab29 did not include data regarding BRAF mutation status, a study15 comparing 106 patients with BRAF-mutant metastatic melanoma with 334 patients with BRAF wild-type tumours reported that BRAF-mutant patients had a lower frequency of objective responses (BRAF mutant 30% vs wild-type 35%) and shorter median response duration (11·1 months vs 14·8 months). All BRAF-mutant patients had received previous treatments, and 76% had previous BRAF-inhibitor treatment and thus were likely to have worse prognostic baseline features (although these were not reported). In the initial report of the CheckMate 067 trial,14 which included 298 (32%) of 945 patients with treatment-naive melanoma with activating BRAF mutations, median progression-free survival for patients with BRAF mutations compared with patients with wild-type BRAF was shorter with nivolumab (BRAF-mutant 5·6 months vs wild-type 7·9 months), but similar with ipilimumab plus nivolumab combination treatment (11·7 months vs 11·2 months). Until the results of prospective trials comparing targeted therapy and immunotherapies are known, treatment decisions will be informed by additional analyses of the outcomes of patients with BRAF mutations in trials of immunotherapy stratified by factors identified as predictors of outcome in patients receiving targeted treatments.

In addition to assessing baseline predictors of outcomes, our analysis shows that progression patterns and survival after progression are heterogeneous after dabrafenib plus trametinib treatment. We noted large differences in survival after progression between patients with progression in baseline or new non-CNS lesions (median survival after progression 10·0 months) versus those with new CNS lesions or concurrent progression in baseline and new lesions (4·0 months). The reported differences in survival after progression in patients treated with dabrafenib plus trametinib support the notion that not all patients with disease that progresses on targeted treatment experience rapid deterioration and death; only 59 (16%) of 379 patients with progressive disease were in the group with the poorest survival after progression. Additionally, 33 (9%) of the 379 patients included in the survival after progression analysis went on to receive either nivolumab or pembrolizumab, and no patients went on to receive an anti-PD-L1 drug, suggesting that immunotherapy had little effect on survival in this cohort. The poor outcomes noted in patients with the CNS as the initial site of progression highlight the continued need to identify patients with melanoma who are at the highest risk of developing CNS involvement.

Although an interesting aspect for further exploration, this study did not include analyses to establish factors that predict toxicity with dabrafenib plus trametinib. However, a previous study reported that the most common toxicity, pyrexia, is not related to efficacy.33

We acknowledge that the retrospective approach we used was limited by an inability to identify new predictive markers of patient outcomes. Only classically recognised markers available for this analysis (ie, those deemed relevant before starting the dabrafenib plus trametinib trials) could be discarded or confirmed. Additionally, the application of these models to clinical practice will be strengthened by the availability of longer follow-up to further define the patients with the most favourable long-term outcomes.

In summary, our study provides a framework for the assessment of therapeutic development programmes in advanced melanoma to understand the heterogeneity of patients and to identify those with an unmet need for improved treatments. Furthermore, we show that a subgroup of patients have long-term disease control with the combination of dabrafenib plus trametinib, and for patients who received this combination rapid resistance is not inevitable. Comparison of these results with future datasets for other melanoma treatments are needed to establish whether factors associated with prolonged survival and response identified in this study apply to patients receiving other treatments, or if they are prognostic. These findings might also provide a framework to optimise translational studies of mechanisms behind treatment resistance, because the identification of targetable features that are associated with adverse characteristics and poor clinical outcomes might facilitate the development of rational approaches for clinical testing in those subgroups.34 Such studies are crucial for the design of future treatments and trials to further improve outcomes in patients with melanoma.

Contributors

GVL, J-JG, SRL, PL, KTF, and MAD conceived and designed the study. SRL and CM did the statistical analyses. All authors analysed and interpreted the data, drafted the manuscript, and approved the final version.

Declaration of interests

GVL is a consultant advisor to Amgen, Bristol-Myers Squibb, Merck/MSD, Novartis, and Roche, and has received honoraria from Merck, Bristol-Myers Squibb, and Novartis. J-JG has consulted for Novartis, Roche, Merck, Bristol-Myers Squibb, and Amgen. PN has received honoraria from Novartis. AR has consulted for Pfizer, Roche, Merck, and Amgen, and owns shares in Kite Pharma. CR has consulted for and received honoraria from Bristol-Myers Squibb, GlaxoSmithKline, Novartis, Amgen, Merck, and Roche. DS has consulted for and received honoraria from Novartis, Roche, and Amgen, and has received honoraria and grant support from Merck/MSD. SRL and CM are employees of Novartis. PL is a former employee of Novartis and owns shares in the company. KTF has consulted for and received honoraria from Novartis, Roche, and Array, and has received grant support from Novartis. MAD has received honoraria from Novartis, Roche/Genentech, Sanofi, and Vaccinex, and is the principal investigator of grants to The University of Texas MD Anderson Cancer Center from Roche/Genentech, Sanofi, AstraZeneca, Myriad, and Merck.

Acknowledgments

Medical writing assistance in the form of collating author comments, copyediting, and editorial assistance was provided by Amanda L Kauffman of Articulate Science and funded by Novartis.

Supplementary Material

Download file

Supplementary appendix

References

  • 1 H Davies, GR Bignell, C Cox, et al. Mutations of the BRAF gene in human cancer. Nature. 2002;417:949-954 Crossref
  • 2 JA Jakob, RL Bassett Jr, CS Ng, et al. NRAS mutation status is an independent prognostic factor in metastatic melanoma. Cancer. 2012;118:4014-4023 Crossref
  • 3 GV Long, AM Menzies, AM Nagrial, et al. Prognostic and clinicopathologic associations of oncogenic BRAF in metastatic melanoma. J Clin Oncol. 2011;29:1239-1246 Crossref
  • 4 KT Flaherty, JR Infante, A Daud, et al. Combined BRAF and MEK inhibition in melanoma with BRAF V600 mutations. N Engl J Med. 2012;367:1694-1703 Crossref
  • 5 GV Long, JS Weber, JR Infante, et al. Overall survival and durable responses in patients with BRAF V600-mutant metastatic melanoma receiving dabrafenib combined with trametinib. J Clin Oncol. 2016;34:871-878
  • 6 GV Long, D Stroyakovskiy, H Gogas, et al. Dabrafenib and trametinib versus dabrafenib and placebo for Val600 BRAF-mutant melanoma: a multicentre, double-blind, phase 3 randomised controlled trial. Lancet. 2015;386:444-451 Crossref
  • 7 GV Long, D Stroyakovskiy, H Gogas, et al. Combined BRAF and MEK inhibition versus BRAF inhibition alone in melanoma. N Engl J Med. 2014;371:1877-1888 Crossref
  • 8 C Robert, B Karaszewska, J Schachter, et al. Improved overall survival in melanoma with combined dabrafenib and trametinib. N Engl J Med. 2015;372:30-39 Crossref
  • 9 C Robert, B Karaszewska, J Schachter, et al. Two year estimate of overall survival in COMBI-v, a randomized, open-label, phase III study comparing the combination of dabrafenib (D) and trametinib (T) with vemurafenib (vem) as first-line therapy in patients (pts) with unresectable or metastatic BRAF V600E/K mutation-positive cutaneous melanoma. Ann Oncol. 2015;26:3301 (abstr).
  • 10 C Robert, B Karaszewska, J Schachter, et al. Three-year estimate of overall survival in COMBI-v, a randomized phase 3 study evaluating first-line dabrafenib (D) + trametinib (T) in patients (pts) with unresectable or metastatic BRAF V600E/K–mutant cutaneous melanoma. Ann Oncol. 2016;27(suppl 6):LBA40 (abstr).
  • 11 J Larkin, PA Ascierto, B Dréno, et al. Combined vemurafenib and cobimetinib in BRAF-mutated melanoma. N Engl J Med. 2014;371:1867-1876 Crossref
  • 12 C Robert, GV Long, B Brady, et al. Nivolumab in previously untreated melanoma without BRAF mutation. N Engl J Med. 2015;372:320-330 Crossref
  • 13 C Robert, J Schachter, GV Long, et al. Pembrolizumab versus ipilimumab in advanced melanoma. N Engl J Med. 2015;372:2521-2532 Crossref
  • 14 J Larkin, V Chiarion-Sileni, R Gonzalez, et al. Combined nivolumab and ipilimumab or monotherapy in untreated melanoma. N Engl J Med. 2015;373:23-34 Crossref
  • 15 J Larkin, CD Lao, WJ Urba, et al. Efficacy and safety of nivolumab in patients with BRAF V600 mutant and BRAF wild-type advanced melanoma: a pooled analysis of 4 clinical trials. JAMA Oncol. 2015;1:433-440 Crossref
  • 16 CM Balch, JE Gershenwald, SJ Soong, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol. 2009;27:6199-6206 Crossref
  • 17 DR Cox, D Oakes. Analysis of survival data (Chapman and Hall/CRC, Boca Raton, FL, 1984)
  • 18 TM Therneau, PM Grambsch. Modeling survival data: extending the Cox model (Springer, New York, 2001)
  • 19 EJ Atkinson, TM Therneau. An introduction to recursive partitioning using the RPART routines. Mayo Foundation. http://www.mayo.edu/research/documents/rpartminipdf/doc-10027257 (2000) (accessed Nov 1, 2016).
  • 20 LA Clark, D Pregibon. Tree-based models. JM Chambers, TJ Hastie (Eds.) Statistical models in S (Waswroth and Brooks, Pacific Grove, CA, 1992) 377-420
  • 21 A Daud, J Weber, J Sosman, et al. Updated overall survival for BRF113220: a phase 1–2 study of dabrafenib alone vs combined dabrafenib and trametinib in patients with BRAF V600 mutation-positive metastatic melanoma. Proc Am Soc Clin Oncol. 2015;33(suppl):9036 (abstr).
  • 22 B Weide, S Richter, P Buttner, et al. Serum S100B, lactate dehydrogenase and brain metastasis are prognostic factors in patients with distant melanoma metastasis and systemic therapy. PLoS One. 2013;8:e81624 Crossref
  • 23 MM Chan, LE Haydu, AM Menzies, et al. The nature and management of metastatic melanoma after progression on BRAF inhibitors: effects of extended BRAF inhibition. Cancer. 2014;120:3142-3153 Crossref
  • 24 TK Eigentler, A Figl, D Krex, et al. Number of metastases, serum lactate dehydrogenase level, and type of treatment are prognostic factors in patients with brain metastases of malignant melanoma. Cancer. 2011;117:1697-1703 Crossref
  • 25 S Kelderman, B Heemskerk, H van Tinteren, et al. Lactate dehydrogenase as a selection criterion for ipilimumab treatment in metastatic melanoma. Cancer Immunol Immunother. 2014;63:449-458
  • 26 RW Joseph, J Elassaiss-Schaap, JD Wolchok, et al. Baseline tumor size as an independent prognostic factor for overall survival in patients with metastatic melanoma treated with the anti-PD-1 monoclonal antibody MK-3475. Proc Am Soc Clin Oncol. 2014;32(suppl 5):3015 (abstr).
  • 27 A Daud, A Ribas, C Robert, et al. Long-term efficacy of pembrolizumab (pembro; MK-3475) in a pooled analysis of 655 patients (pts) with advanced melanoma (MEL) enrolled in KEYNOTE-001. Proc Am Soc Clin Oncol. 2015;33(suppl):9005 (abstr).
  • 28 J Larkin, V Chiarion-Sileni, R Gonzalez, et al. Efficacy and safety in key patient subgroups of nivolumab alone or combined with ipilimumab versus ipilimumab alone in treatment-naive patients with advanced melanoma (CheckMate 067). Ann Oncol. 2015;26(suppl 3):S664-S665 (abstr 3303).
  • 29 B Weide, A Martens, JC Hassel, et al. Baseline biomarkers for outcome of melanoma patients treated with pembrolizumab. Clin Cancer Res. 2016; 10.1158/1078-0432.CCR-16-0127 published online May 16.
  • 30 JS Weber, SP D'Angelo, D Minor, et al. Nivolumab versus chemotherapy in patients with advanced melanoma who progressed after anti-CTLA-4 treatment (CheckMate 037): a randomised, controlled, open-label, phase 3 trial. Lancet Oncol. 2015;16:375-384 Crossref
  • 31 DB Johnson, CM Lovly, M Flavin, et al. Impact of NRAS mutations for patients with advanced melanoma treated with immune therapies. Cancer Immunol Res. 2015;3:288-295 Crossref
  • 32 RW Joseph, RJ Sullivan, R Harrell. Correlation of NRAS mutations with clinical response to high-dose IL-2 in patients with advanced melanoma. J Immunother. 2012;35:66-72 Crossref
  • 33 AM Menzies, MT Ashworth, S Swann, et al. Characteristics of pyrexia in BRAFV600E/K metastatic melanoma patients treated with combined dabrafenib and trametinib in a phase I/II clinical trial. Ann Oncol. 2015;26:415-421 Crossref
  • 34 JJ Grob, GV Long, D Schadendorf, K Flaherty. Disease kinetics for decision-making in advanced melanoma: a call for scenario-driven strategy trials. Lancet Oncol. 2015;16:e522-e526

Footnotes

a Melanoma Institute Australia, The University of Sydney, Sydney, NSW, Australia

b Royal North Shore and Mater Hospitals, Sydney, Australia

c Service de Dermatologie, Centre Hospitalo-Universitaire Timone, Aix Marseille Université, Marseille CEDEX 05, France

d Mount Vernon Cancer Centre, Northwood, UK

e Department of Medicine, Hematology/Oncology, UCLA Jonsson Comprehensive Cancer Center, Los Angeles, CA, USA

f Gustave Roussy, Département de Médecine Oncologique, Service de Dermatologie et Université Paris-Sud, Faculté de Médecine, Villejuif, France

g Department of Dermatology, University Hospital Essen, Essen, Germany

h German Cancer Consortium, Heidelberg, Germany

i Biostatistics, Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA

j Global Oncology, Novartis Pharma AG, Basel, Switzerland

k Developmental Therapeutics and Melanoma Programs, Massachusetts General Hospital Cancer Center, Boston, MA, USA

l Melanoma Medical Oncology and Systems Biology, The University of Texas MDAnderson Cancer Center, Houston, TX, USA

* Correspondence to: Prof Georgina V Long, Melanoma Institute Australia, The University of Sydney, North Sydney, NSW 2060, Australia

Reprints Account Manager
Emma Steel
T: +44 (0)20 7424 4221
e.steel@elsevier.com
The Lancet Oncology is a monthly
subscription journal. For further
information on how to subscribe
please contact our
Subscription Department
T: +44 (0) 1865 843077
F: +44 (0) 1865 843970
custserv@lancet.com
(North America)
T: +1 (800) 462 6198
F: +1 (800) 327 9021
USLancetCS@elsevier.com

Printed by
 

Elsevier España, S.L.U.
(A member of Elsevier)
Av. Josep Tarradellas, 20-30
08029 Barcelona
Tel.: 932 000 711
Fax: 932 091 136

© 2016 Elsevier Ltd. All rights reserved. This journal and the individual contributions contained in it are protected under copyright by Elsevier Ltd, and the following terms and conditions apply to their use. The Lancet is a trade mark of RELX Intellectual Properties SA, used under license.
Publication information The Lancet Oncology (ISSN 1470-2045) is published monthly by Elsevier (The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK). Periodicals postage paid at Rahway, NJ, USA. POSTMASTER: send address corrections to The Lancet Oncology c/o Mercury International, 365 Blair Road, Avenel, NJ 07001, USA.
Photocopying Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Global Rights Department, The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK, tel: +1 215 239 3804 or +44 (0)1865 843830, fax: +44 (0)1865 853333, email healthpermissions@elsevier.com In the USA, users may clear permissions and make payments through the Copyright Clearance Center Inc, 222 Rosewood Drive, Danvers, MA 01923, USA; tel: +1 (978) 7508400, fax: +1 (978) 646-8600. Other countries may have a local reprographic rights agency for payments.
Derivative works Subscribers may reproduce tables of contents or prepare lists of articles including abstracts for internal circulation within their institutions. Permission of the Publisher is required for resale or distribution outside the institution. Permission of the Publisher is required for all other derivative works, including compilations and translations.
Electronic storage or usage Permission of the Publisher is required to store or use electronically any material contained in this journal, including any article or part of an article. Except as outlined above, no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Health Sciences Rights Department, at the mail, fax, and email addresses noted above.
Notice No responsibility is assumed by Elsevier for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation
of any methods, products, instructions, or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.
The views expressed in this journal are not necessarily those of the Editor, the International Advisory Board, or Elsevier Ltd.

1612045617