Adjuvant vemurafenib in resected, BRAFV600 mutation-positive melanoma (BRIM8): a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial

Articles

Adjuvant vemurafenib in resected, BRAFV600 mutation-positive melanoma (BRIM8): a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial Michele Maio, Karl Lewis, Lev Demidov, Mario Mandalà, Igor Bondarenko, Paolo A Ascierto, Christopher Herbert, Andrzej Mackiewicz, Piotr Rutkowski, Alexander Guminski, Grant R Goodman, Brian Simmons, Chenglin Ye, Yibing Yan, Dirk Schadendorf, and the BRIM8 Investigators*

Summary

Background Systemic adjuvant treatment might mitigate the high risk of disease recurrence in patients with resected stage IIC–III melanoma. The BRIM8 study evaluated adjuvant vemurafenib monotherapy in patients with resected, BRAFV600 mutation-positive melanoma. Methods BRIM8 was a phase 3, international, double-blind, randomised, placebo-controlled study that enrolled 498 adults (aged ≥18 years) with histologically confirmed stage IIC–IIIA–IIIB (cohort 1) or stage IIIC (cohort 2) BRAFV600 mutation-positive melanoma that was fully resected. Patients were randomly assigned (1:1) by an interactive voice or web response system to receive twice-daily adjuvant oral vemurafenib 960 mg tablets or matching placebo for 52 weeks (13 × 28-day cycles). Randomisation was done by permuted blocks (block size 6) and was stratified by pathological stage and region in cohort 1 and by region in cohort 2. The investigators, patients, and sponsor were masked to treatment assignment. The primary endpoint was disease-free survival in the intention-to-treat population, evaluated separately in each cohort. Hierarchical analysis of cohort 2 before cohort 1 was prespecified. This trial is registered with ClinicalTrials.gov, number NCT01667419. Findings The study enrolled 184 patients in cohort 2 (93 were assigned to vemurafenib and 91 to placebo) and 314 patients in cohort 1 (157 were assigned to vemurafenib and 157 to placebo). At the time of data cutoff (April 17, 2017), median study follow-up was 33·5 months (IQR 25·9–41·6) in cohort 2 and 30·8 months (25·5–40·7) in cohort 1. In cohort 2 (patients with stage IIIC disease), median disease-free survival was 23·1 months (95% CI 18·6–26·5) in the vemurafenib group versus 15·4 months (11·1–35·9) in the placebo group (hazard ratio [HR] 0·80, 95% CI 0·54–1·18; log-rank p=0·026). In cohort 1 (patients with stage IIC–IIIA–IIIB disease) median disease-free survival was not reached (95% CI not estimable) in the vemurafenib group versus 36·9 months (21·4–not estimable) in the placebo group (HR 0·54 [95% CI 0·37–0·78]; log-rank p=0·0010); however, the result was not significant because of the prespecified hierarchical prerequisite for the primary disease-free survival analysis of cohort 2 to show a significant disease-free survival benefit. Grade 3–4 adverse events occurred in 141 (57%) of 247 patients in the vemurafenib group and 37 (15%) of 247 patients in the placebo group. The most common grade 3–4 adverse events in the vemurafenib group were keratoacanthoma (24 [10%] of 247 patients), arthralgia (17 [7%]), squamous cell carcinoma (17 [7%]), rash (14 [6%]), and elevated alanine aminotransferase (14 [6%]), although all keratoacanthoma events and most squamous cell carcinoma events were by default graded as grade 3. In the placebo group, grade 3–4 adverse events did not exceed 2% for any of the reported terms. Serious adverse events were reported in 40 (16%) of 247 patients in the vemurafenib group and 25 (10%) of 247 patients in the placebo group. The most common serious adverse event was basal cell carcinoma, which was reported in eight (3%) patients in each group. One patient in the vemurafenib group of cohort 2 died 2 months after admission to hospital for grade 3 hypertension; however, this death was not considered to be related to the study drug. Interpretation The primary endpoint of disease-free survival was not met in cohort 2, and therefore the analysis of cohort 1 showing a numerical benefit in disease-free survival with vemurafenib versus placebo in patients with resected stage IIC–IIIA–IIIB BRAFV600 mutation-positive melanoma must be considered exploratory only. 1 year of adjuvant vemurafenib was well tolerated, but might not be an optimal treatment regimen in this patient population. Funding F Hoffman–La Roche Ltd.

Introduction Surgical resection of the primary tumour or affected lymph nodes is the standard of care in patients with stage II–III melanoma.1 Despite full resection, patients with stage IIC–III melanoma remain at high risk for

disease recurrence and death.2–4 This situation warrants the use of adjuvant approaches to improve clinical outcomes. Clinical studies of systemic adjuvant options that were approved at the time when this study was being designed (2011) had shown either limited effectiveness or

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Lancet Oncol 2018 Published Online February 21, 2018 http://dx.doi.org/10.1016/ S1470-2045(18)30106-2 See Online/Comment http://dx.doi.org/10.1016/ S1470-2045(18)30150-5 *Listed in the appendix Division of Medical Oncology and Immunotherapy, Center for Immuno-Oncology, University Hospital of Siena, Siena, Italy (M Maio MD); University of Colorado Comprehensive Cancer Center, Aurora, CO, USA (K Lewis MD); N N Blokhin Russian Cancer Research Center, Ministry of Health, Moscow, Russia (Prof L Demidov MD); Department of Oncology and Haematology, Papa Giovanni XXIII Cancer Center Hospital, Bergamo, Italy (M Mandalà MD); Dnipropetrovsk State Medical Academy, Dnipropetrovsk, Ukraine (I Bondarenko MD); Melanoma Unit, Cancer Immunotherapy and Innovative Therapies, Istituto Nazionale Tumori Fondazione Pascale, Naples, Italy (P A Ascierto MD); Bristol Haematology and Oncology Centre, Bristol, UK (C Herbert MD); Department of Cancer Immunology, Poznan University for Medical Sciences, Med-POLONIA, Poznan, Poland (Prof A Mackiewicz MD); Department of Soft Tissue/ Bone Sarcoma and Melanoma, Maria Sklodowska-Curie Institute—Oncology Center, Warsaw, Poland (Prof P Rutkowski MD); Melanoma Translational Research Group, Melanoma Institute Australia, Wollstonecraft, NSW, Australia (A Guminski PhD); Genentech, Inc, South San Francisco, CA, USA (G R Goodman MD, B Simmons PharmD, C Ye PhD,

1

Articles

Y Yan PhD); and Department of Dermatology, University Hospital Essen, Essen, Germany, and the German Cancer Consortium, Heidelberg, Germany (Prof D Schadendorf MD) Correspondence to: Dr Michele Maio, Division of Medical Oncology and Immunotherapy, Department of Oncology, University Hospital of Siena, Strada delle Scotte no. 14, 53100 Siena, Italy [email protected] See Online for appendix

Research in context Evidence before this study At the time this study was conceived (2011), patients with an initial diagnosis of stage IIC–III melanoma (per AJCC version 7 criteria) that was fully resected continued to have a clinically significant risk of recurrence and death, and interferons were the only approved adjuvant option. We searched PubMed from inception until Dec 31, 2011, for reports from large controlled clinical studies in this patient population with the search terms “melanoma” and “adjuvant” (sorted by publication date and clinical trial; abstracts were manually curated to include only studies with more than 100 patients). We also searched across major international oncology congresses. Studies showed that various interferon regimens, including interferon-α2a, interferon-α2b, and pegylated interferon-α2b, had been explored. A synthesis of data from these studies suggested that no clear evidence showed that interferon improved overall survival, and that there was no consensus regarding the optimal dosing intensity and duration of adjuvant interferon treatment. Moreover, the tolerability of these interferon regimens was not favourable, as evidenced by discontinuation rates ranging from around 30% to 57% and a decline in patient-reported health-related quality of life in some of these studies.

disease. In BRIM8, a two-cohort design was used to separately evaluate efficacy in patients with fully resected, BRAFV600 mutation-positive disease that was either stage IIC–IIIB or stage IIIC at diagnosis, because the risk of recurrence or death is quite different in these populations. This design allowed for differentiation of clinical benefit in these patient populations, with prespecified and distinct statistical considerations in each cohort. The study showed that although vemurafenib monotherapy was biologically active in patients with stage IIIC disease, a significant improvement in disease-free survival was not observed and the study did not achieve the primary endpoint. Vemurafenib monotherapy did provide a risk reduction of disease-free survival events in patients with resected stage IIC–IIIB disease, but the result could not be considered significant because of the protocol-defined prerequisite for the disease-free survival benefit in patients with stage IIIC disease to be significant. Implications of all the available evidence The study did not meet its primary endpoint. The data from the BRIM8 study suggest that the optimal dosing and intensity of adjuvant treatment might vary by the magnitude of recurrence risk.

Added value of this study Most controlled studies of adjuvant treatment so far have been single-cohort studies in patients with resected stage III–IV

unfavourable toxicity profiles. Data for overall survival benefit in individual studies of adjuvant interferon-α are conflicting,2,5,6 and although ipilimumab monotherapy significantly improved outcomes versus placebo (recurrence-free survival hazard ratio [HR] 0·76 [95% CI 0·64–0·89]; p<0·001), it was asso­ciated with substantial toxicity (53% of patients discontinued treat­ment because of adverse events, 42% had grade 3–4 immune-related adverse events, and 1% had fatal immune-related adverse events).7 More recently, the COMBI-AD and CheckMate 238 studies showed improved survival in patients with resected stage III or stage IIIB–IV melanoma, respectively.8,9 In patients with melanoma, 5-year survival rates vary quite broadly by stage of disease at diagnosis. In patients with stage IIC–IIIB melanoma, 5-year survival ranges from approximately 55% to 78%, whereas 5-year-survival for patients with stage IIIC disease is around 40%.2 Approximately 50% of patients with melanoma have tumours that harbour the BRAFV600 mutation,10 and the orally administered BRAF inhibitor vemurafenib has been shown to be an effective and safe treatment for patients with BRAFV600 mutation-positive advanced or metastatic melanoma.11,12 Accordingly, we did a placebocontrolled study to evaluate the efficacy and safety of 1 year of adjuvant vemurafenib in two cohorts of patients 2

with resected, BRAFV600 mutation-positive melanoma: one cohort of patients with stage IIC–IIIB disease and the other with patients with stage IIIC disease only. The planned treatment duration of 1 year of adjuvant vemurafenib was based on previous studies of various adjuvant interferon regimens for melanoma.5

Methods

Study design and participants BRIM8 was a phase 3, international, double-blind, randomised, placebo-controlled study that was done across 124 cancer treatment centres and hospitals in North and South America (USA, Canada, Brazil, and Argentina), Australia, Europe (Austria, Belgium, Croatia, Czech Republic, Estonia, France, Germany, Ireland, Italy, Netherlands, Poland, Portugal, Russia, Serbia, Spain, Sweden, Switzerland, Ukraine, and the UK), and South Africa. The study enrolled patients based on disease stage at diagnosis in two cohorts: patients with stage IIC–IIIB disease were enrolled in cohort 1, and patients with stage IIIC disease were enrolled in cohort 2. Adults (aged ≥18 years) with histologically confirmed cutaneous melanoma (pathological stage IIC or stage III per the American Joint Committee on Cancer [AJCC] Staging Criteria, version 72), completely resected per evaluation by local investigators, and confirmed as BRAFV600

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Articles

mutation positive by the Cobas BRAFV600 Mutation Test (Roche Molecular Diagnostics, Pleasanton, CA, USA), were eligible. This included patients with stage IIIA melanoma who had one or more nodal metastases greater than 1 mm in diameter and patients with lymph node involvement at initial presentation or a first metachronous nodal recurrence. Patients in whom a sentinel lymph node biopsy procedure could not be done or a sentinel lymph node was not detected were required to undergo complete regional lymphadenectomy. All patients with either clinical or radiographic evidence of regional lymph node involvement or evidence of melanoma involvement in the sentinel lymph node were required to undergo a complete regional lymphadenectomy. Additional eligibility criteria included an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1; adequate haematological, liver, and renal function; a full recovery from the effects of any major surgery or any previous substantial traumatic injury; and a life expectancy of at least 5 years. Detailed inclusion and exclusion criteria are in the appendix (pp 76–80). Patients with either a history of, or current, clinical, radiographic, or pathological evidence of in-transit metastases, satellite, or microsatellite lesions were excluded, as were those with a history of any systemic, local, or radiotherapy for cancer. Additional exclusion criteria are noted in the appendix (p 78). In brief, patients with major surgical procedures within 4 weeks of study entry, active or chronic infection, autoimmune disease, history of malabsorption, or unwillingness or inability to comply with study and follow-up procedures were excluded. Study oversight was provided by Genentech/Roche personnel for this trial and an independent Data Safety Monitoring Board (DSMB). This committee did periodic reviews of selected safety data according to procedures outlined in the DSMB charter. The study conduct conformed with the International Conference on Harmonisation E6 guideline for Good Clinical Practice and the principles of the Declaration of Helsinki or the laws and regulations of the country in which the research was done, whichever afforded the greater protection to the patient. All patients provided written informed consent before the conduct of any study procedures.

Randomisation and masking Eligible patients were randomised (1:1) by an interactive voice or web response system to receive oral vemurafenib or matching placebo. Randomisation was stratified by pathological stage (IIC vs IIIA vs IIIB) and region (North America vs Australia, New Zealand, South Africa and Latin American countries vs the rest of the world) in cohort 1 and by region in cohort 2. Regions were grouped based on geographical proximity. A stratified, permuted, block randomisation scheme (block size 6) was used to

obtain an approximate 1:1 ratio between the two treatment groups. The investigators, patients, and sponsor were masked to treatment assignment. Patients were randomly assigned to receive placebo or vemurafenib through the use of an interactive voice or web response system. Placebo tablets and packaging did not permit their identification as distinct from those of the active comparator, vemurafenib.

Procedures Eligible patients were to receive oral vemurafenib tablets (960 mg twice daily for 52 weeks [13 × 28-day cycles]) or matching placebo. These were dispensed by the trial nurse or practitioner at the study centres to patients and from placebo to self-administered. Crossover vemurafenib treatment was not allowed. Surveillance for tumour recurrence, including contrast-enhanced CT or MRI of the chest, abdomen, and pelvis (every 13 weeks for the first 2 years and then every 26 weeks for years 3–5), and physical examination were done (see appendix pp 6–11 for the schedule of assessments). For the primary endpoint, melanoma recurrence was assessed by the investigator, whereas the occurrence of new primary melanoma was confirmed histopathologically by a blinded review at a central pathology laboratory (coordinated by Dermpath Diagnostics, WI, USA). Details of adverse event and laboratory test monitoring are provided in the appendix (pp 8–11). Safety was monitored continuously during each treatment cycle and up to and including 28 days after the last dose of study drug. Per regulatory requirements, the treatment code was available through an interactive voice or web response system to the Roche Drug Safety Group for all unexpected serious adverse events that were considered by the investigator to be related to the study drug. Severity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) version 4.0. In patients with grade 2 (intolerable) or grade 3 adverse events, study drug was withheld until recovery to grade 0 or 1, and thereafter resumed with a 25% dose reduction (to three tablets twice daily) for the first occurrence, or a 50% reduction (to two tablets twice daily) for the second occurrence. In patients with intolerable grade 4 adverse events, study drug was withheld until recovery to grade 0 or 1, and thereafter resumed with a 50% dose reduction (two tablets twice daily) for the first occurrence. The study drug was discontinued for a third occurrence of grade 2 (intolerable) or grade 3 events or the second occurrence of a grade 4 event (appendix p 12). Any patients with a break in dosing in excess of 28 consecutive days were permanently discontinued from study treatment.

Outcomes The primary endpoint was disease-free survival, defined as the time from randomisation until the date of the first

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

3

Articles

498 patients were enrolled

314 were assigned to cohort 1 (stage IIC, IIIA [>1 mm], and IIIB disease)

157 randomly assigned to vemurafenib 154 received vemurafenib 3 did not receive vemurafenib due to protocol violations

184 were assigned to cohort 2 (stage IIIC disease)

157 randomly assigned to placebo 156 received vemurafenib 1 did not receive placebo because of withdrawn consent

67 discontinued treatment prematurely 34 had adverse events 3 had a protocol violation 2 were non-compliant with study drug 9 withdrew from study 2 had physicians decide to discontinue their patients 14 had recurrence of disease 3 for other reasons

90 completed treatment

66 discontinued treatment prematurely 5 had adverse events 1 had a protocol violation 1 was non-compliant with study drug 6 withdrew from study 1 had physician decide to discontinue their patient 51 had recurrence of disease 1 for other reasons

91 completed treatment

154 included in the safety analysis population 157 included in the efficacy analysis population

93 randomly assigned and received vemurafenib

156 included in the safety analysis population 157 included in the efficacy analysis population

91 randomly assigned and received placebo

35 discontinued treatment prematurely 13 had adverse events 1 was non-compliant with study drug 1 was non-compliant 5 withdrew from study 12 had recurrence of disease 3 for other reasons

38 discontinued treatment prematurely 1 had protocol violation 1 withdrew from study 36 had recurrence of disease

58 completed treatment

53 completed treatment

93 included in the safety analysis population 93 included in the efficacy analysis population

91 included in the safety analysis population 91 included in the efficacy analysis population

Figure 1: Trial profile

local, regional, or distant melanoma recurrence, occurrence of new primary melanoma, or death from any cause, whichever occurred first. Secondary endpoints were distant metastasis-free survival (time from randomisation until the date of diagnosis of distant metastases or death from any cause), overall survival (time from random­isation to the date of death from any cause), safety and tolerability, health-related quality of life, and pharma­cokinetics. Health-related quality-of-life and pharmacokinetic data will be presented in a separate report. Exploratory objectives were to assess outcomes in patients whose tumours harboured BRAF mutations other than V600E, the exploration of molecular markers in plasma, serum, and tumour tissue to predict recurrence and resistance to vemurafenib, and the molecular characterisation of squamous cell carcinoma. These analyses will be done and reported separately.

Statistical analysis For cohort 1, a sample size of 300 patients was estimated to be necessary to achieve 80% power to detect an HR of 0·60 with a two-sided, stratified log-rank test at the 0·05 significance level. The underlying assumptions for this estimate were a median disease-free survival of 24 months for the placebo control group, a median disease-free survival of 40 months for the vemurafenib 4

treatment group, and a 5% annual loss to follow-up. The primary analysis was originally planned for when about 120 disease-free survival events in both cohorts had accrued. For cohort 2, a sample size of 175 patients was estimated to be necessary to achieve 80% power to detect an HR of 0·58 with a two-sided, stratified logrank test at the 0·05 significance level. The underlying assumptions for this estimate were a median diseasefree survival of 7·7 months for the placebo control group, a median disease-free survival of 13·3 months for the vemurafenib treatment group, and a 5% annual loss to follow-up. The assumptions for median diseasefree survival were based on extrapolations from historical data reported for the control groups of adjuvant interferon studies.13,14 Owing to a slower than expected disease-free survival event rate in cohort 2 than anticipated, a protocol amendment was implemented on March 14, 2017. In this amendment, the number of disease-free survival events required for the primary disease-free survival analysis in cohort 2 were reduced from 120 to 105. Hierarchical testing of disease-free survival in cohort 2 before cohort 1 was prespecified to maintain an overall type I error rate of 0·05 (two-sided), with the assumption being that observing a biological effect in higher risk disease (cohort 2) would suggest a treatment effect across

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Articles

the continuum of melanoma (from metastatic to stage IIC patients). Only a p value for cohort 2 of 0·05 or less would allow for the analysis of cohort 1 to be considered significant. Two overall survival analyses were planned for each cohort. The interim analysis of overall survival in each cohort was to be done at the time of the final disease-free survival analysis. The final overall survival analyses for cohorts 1 and 2 were to be done after the occurrence of about 107 and 118 deaths, respectively (projected to occur at around month 72 in each cohort) or at month 72, whichever occurred first. The efficacy analyses included all randomly assigned patients (intention-to-treat analysis), and patients were analysed according to the treatment assigned at randomisation. The primary objective and all secondary objectives of this study were evaluated separately for each cohort. The analysis of survival was done using the Kaplan-Meier method and the 95% CI for the medians were computed using the method of Brookmeyer and Crowley.15 The HR of disease-free survival and the associated two-sided 95% CIs were estimated by using a stratified Cox proportional hazards model. An exploratory analysis of disease-free survival based on the pooled data from both cohorts was also done for descriptive purposes, to put these data in context with other studies wherein the populations of these two cohorts were combined a priori. The safety analyses included all patients who received at least one dose of study drug (vemurafenib or placebo), with patients analysed according to the treat­ ment received. Safety was assessed through sum­ maries of all adverse events and their severity. All verbatim descriptions of treatment-emergent adverse events were summarised using Medical Dictionary for Regulatory Activities-preferred terms. Adverse events were graded by the investigator according to NCI CTCAE, version 4.0. The statistical software used in this study was SAS, version 9.2. This trial is registered with ClinicalTrials.gov, number NCT01667419.

Cohort 2 (n=184)

Cohort 1 (n=314)

Vemurafenib group (n=93)

Placebo group (n=91)

Vemurafenib group (n=157)

Placebo group (n=157)

55 (40–61)

50 (38–58)

51 (43–60)

49 (40–59) 85 (54%)

Age (years) Age group (years) ≤50

40 (43%)

48 (53%)

78 (50%)

>50

53 (57%)

43 (47%)

79 (50%)

72 (46%)

≤65

77 (83%)

80 (88%)

136 (87%)

137 (87%)

>65

16 (17%)

11 (12%)

21 (13%)

20 (13%)

Sex Female

41 (44%)

32 (35%)

73 (46%)

69 (44%)

Male

52 (56%)

59 (65%)

84 (54%)

88 (56%) 150 (96%)

Race White

84 (90%)

81 (89%)

150 (96%)

Other

1 (1%)

2 (2%)

1 (1%)

2 (1%)

Mixed

1 (1%)

0

0

0

Unknown

7 (8%)

8 (9%)

6 (4%)

5 (3%)

Pathological stage IIC

··

··

15 (10%)

12 (8%)

IIIA

··

··

36 (23%)

39 (25%)

··

106 (68%)

106 (68%)

IIIB IIIC

·· 93 (100%)

91 (100%)

··

Primary tumour ulceration or mitosis present

57/85 (67%)

49/78 (63%)

63/144 (44%)

49/143 (34%)

··

First metachronous nodal recurrence

29/93 (31%)

32/90 (36%)

26/157 (17%)

36/157 (23%)

Lymph node type at baseline Micrometastasis

··

··

84/142 (59%)

76/145 (52%)

58/142 (41%)

69/145 (48%)

Macrometastasis

30/93 (32%)

27/91 (30%)

N3

63/93 (68%)

64/91 (70%)

0

81/92 (88%)

78/91 (86%)

143/155 (92%)

136/157 (87%)

1

11/92 (12%)

13/91 (14%)

12/155 (8%)

21/157 (13%)

77/85 (91%)

67/75 (89%)

123/136 (90%)

129/138 (93%)

8/85 (9%)

8/75 (11%)

13/136 (10%)

9/138 (7%)

··

··

ECOG performance status

BRAF mutation status at baseline V600E Non-V600E

Data are n (%), n/N (%), or median (IQR). ECOG=Eastern Cooperative Oncology Group.

Table 1: Baseline characteristics

Role of the funding source This trial was designed and funded by the sponsor (F Hoffmann–La Roche Ltd). The sponsor and their representatives collected and analysed the data. All authors had full access to study data, and the corresponding author had final responsibility for the decision to submit for publication.

Results Between Sept 10, 2012, and Aug 10, 2015, 498 patients were enrolled and randomly assigned to receive treatment with adjuvant vemurafenib or matching placebo: 184 patients in cohort 2 and 314 patients in cohort 1 (figure 1). Of the 184 patients enrolled in cohort 2, 93 were randomly assigned to vemurafenib and 91 to placebo (the intention-to-treat population). In cohort 2,

all patients received their allocated treatment; thus, the safety population was identical to the intention-to-treat population. Of the 314 patients enrolled in cohort 1, 157 were randomly assigned to vemurafenib and 157 to placebo (the intention-to-treat population). In cohort 1, three patients in the vemurafenib group and one patient in the placebo group did not receive their assigned treatment (three patients in the vemurafenib group had protocol violations and one patient in the placebo group withdrew consent); these four patients were therefore excluded from the safety population. At the time of data cutoff (April 17, 2017), median study follow-up was 33·5 months (IQR 25·9–41·6) in cohort 2 and 30·8 months (25·5–40·7) in cohort 1. In cohort 2, 28 (30%) of 93 patients in the vemurafenib group and

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

5

Articles

A 100 +

+

Hazard ratio 0·80 (95% CI 0·54–1·18) Log-rank p=0·26

+ +

80 Disease-free survival (%)

+ +

+

60

++ ++ ++ +++ +++ ++ + +++++++ +++ +++ ++ +++

40

20

++++

+

++ ++

+

+ ++ + + +++ + + + +

Vemurafenib group Placebo group + Censored

0 Number at risk (number censored) Vemurafenib group Placebo group

++

0

3

6

9

12

15

18

21

24

27

30

33

36

39

42

45

93 (0) 87 (2) 85 (3) 76 (3) 70 (4) 61 (4) 57 (5) 44 (9) 29 (18) 16 (27) 15 (28) 13 (30) 11 (31) 7 (34) 5 (36) 1 (40) 91 (0) 71 (0) 59 (1) 54 (2) 51 (2) 45 (2) 43 (2) 39 (6) 31 (13) 21 (20) 16 (24) 13 (27) 11 (28) 8 (31) 7 (32) 5 (34)

48

51

·· 1 (37)

·· ··

B 100 ++++ +

Disease-free survival (%)

80

+ +

Hazard ratio 0·54 (95% CI 0·37–0·78) Log-rank p=0·0010

++ ++

++

+ +

60

+

+

++

+++++ ++ +

++

+++++++ ++++++++++++ ++++++ + ++ ++++ ++++ ++

+++++++++++

++++++

+++ + ++++++++ ++++

++++ + + + +

+++++

++ +

++++++++ ++ +++++++

++++

++ +

+

40

20

0 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 Number at risk (number censored) Vemurafenib group 157 (0) 146 (7) 137 (9) 129 (12) 120 (14) 115 (15) 107 (19) 94 (27) 72 (46) 49 (67) 38 (76) 31 (82) 26 (86) 18 (94) 15 (97) 4 (108) 2 (110) Placebo group 157 (0) 129 (3) 118 (4) 106 (4) 100 (5) 94 (6) 90 (7) 79 (15) 65 (26) 43 (44) 35 (52) 31 (56) 28 (59) 22 (64) 12 (73) 3 (82) 1 (84)

51 ·· ··

C 100 + ++++ + +

Disease-free survival (%)

80

Figure 2: Disease-free survival in (A) cohort 2 (patients with fully resected stage IIIC disease), (B) cohort 1 (patients with fully resected stage IIC–IIIB disease), and (C) the prespecified pooled intention-to-treat population (cohorts 1 and 2 combined)

6

+ +

+++

Hazard ratio 0·65 (95% CI 0·50–0·85) Log-rank p=0·0013

+ +++

++ +

60

40

+

+

+

++

+++ ++++ + ++++++ +++++++++ ++++++ +++ ++++ ++++ +++ +++++ + ++ +++ +++++ +++ ++ ++++++++++ + +++++++ + +++++++ +++ +++++ ++++++++++ ++ + ++++ ++++++ + +++ +++++ +++ + ++ +++++ ++++ ++ + +

+

20

0

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 Time since randomisation (months) Number at risk (number censored) Vemurafenib group 250 (0) 233 (9) 222 (12) 205 (15) 190 (18) 176 (19) 164 (24) 138 (36) 101 (64) 65 (94) 53 (104) 44 (112) 37 (117) 25 (128) 20 (133) 5 (148) 2 (151) Placebo group 248 (0) 200 (3) 117 (5) 160 (6) 151 (7) 139 (8) 133 (9) 118 (21) 96 (39) 64 (64) 51 (76) 44 (83) 39 (87) 30 (95) 19 (105) 8 (116) 2 (121)

51

·· ··

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Articles

31 (34%) of 91 in the placebo group had discontinued from the study. In cohort 1, 36 (23%) of 157 patients in the vemurafenib group and 41 (26%) of 157 in the placebo group had discontinued from the study (appendix p 13). In both cohorts, death and withdrawal of consent were the most common reasons for study discontinuation. Most patients were white (465 [93%] of 498 patients), aged 65 years or younger (430 [86%] of 498), and had an ECOG performance status of 0 (438 [88%] of 495). Disease and baseline characteristics were well balanced between treatment groups in both cohorts (table 1). Most patients in cohort 1 had a diagnosis of stage IIIB disease (212 [68%] of 314). At data cutoff, 105 disease-free survival events had occurred in cohort 2 (52 in the vemurafenib group and 53 in the placebo group) and 117 disease-free survival events had occurred in cohort 1 (45 in the vemurafenib group and 72 in the placebo group). In cohort 2, median disease-free survival was 23·1 months (95% CI 18·6–26·5) in the vemurafenib group versus 15·4 months (11·1–35·9) in the placebo group (HR 0·80 [95% CI 0·54–1·18], log-rank p=0·26; figure 2A). 1-year disease-free survival was 78·9% (95% CI 70·5–87·3) in the vemurafenib group versus 58·0% (47·8–68·1) in the placebo group and 2-year disease-free survival was 46·3% (95% CI 35·4–57·1) in the vemurafenib group versus 47·5% (37·1–57·9) in the placebo group. Because of the prespecified hierarchical testing of cohort 2 before cohort 1, and because the primary endpoint was not met in cohort 2, the analyses in cohort 1 cannot be regarded as significant (p values provided hereafter are for descriptive purposes only). In cohort 1, median disease-free survival was not reached (95% CI not estimable [NE]) in the vemurafenib group versus 36·9 months (21·4–NE) in the placebo group (HR 0·54, 95% CI 0·37–0·78, log-rank p=0·0010; figure 2B). 1-year disease-free survival was 84·3% (95% CI 78·5–90·2) in the vemurafenib group versus 66·2% (58·7–73·7) in the placebo group and 2-year disease-free survival was 72·3% (95% CI 64·9–79·8) in the vemurafenib group versus 56·5% (48·5–64·4) in the placebo group. In the prespecified exploratory pooled intention-totreat analysis of both cohorts, median disease-free survival was not estimable in the vemurafenib group versus 25·8 months (95% CI 20·5–NE) in the placebo group (HR 0·65 [95% CI 0·50–0·85], log-rank p=0·0013; figure 2C). In this pooled analysis, 1-year disease-free survival was 82·2% (95% CI 77·4–87·1) in the vemurafenib group versus 63·1% (57·1–69·2) in the placebo group and 2-year disease-free survival was 62·2% (95% CI 55·8–68·6) in the vemurafenib group versus 53·1% (46·8–59·5) in the placebo group. Subgroup analyses defined by common prognostic covariates were generally consistent with the overall analysis of disease-free survival (figure 3). In cohort 2, median distant metastasis-free survival was 37·2 months (95% CI 22·1–NE) in patients receiving

adjuvant vemurafenib versus 30·7 months (24·5–NE) in those receiving placebo (HR 0·91, 95% CI 0·57–1·44, log-rank p=0·68; appendix pp 14, 15, 26–28). In cohort 1, median distant metastasis-free survival was not reached in either treatment group (HR for vemurafenib versus placebo 0·58, 95% CI 0·37–0·90, log-rank p=0·013; appendix p 29). In the prespecified exploratory pooled intention-to-treat analysis, median distant metastasis-free survival was not estimable in the vemurafenib group versus 47·8 months (95% CI 30·7–NE) in the placebo group (HR 0·70 [95% CI 0·52–0·96], log-rank p=0·027; appendix p 29). Sites of distant metastases are shown in the appendix (p 16). At the time of data cutoff, a total of 83 deaths had occurred in the overall study population. The exact date of death was not available for one patient in the placebo group of cohort 2; this patient was censored from the overall survival analysis on the date they were last confirmed to be alive. Thus, the analysis included 82 deaths (35 in the vemurafenib group [16 in cohort 1, 19 in cohort 2] and 47 in the placebo group [28 in cohort 1, 19 in cohort 2]). Overall survival data are immature (appendix p 14). The 2-year landmark overall survival was 93·4% (95% CI 89·2–97·6) in the vemurafenib group and 86·8% (81·3–92·4) in the placebo group in cohort 1, and 83·7% (75·8–91·5) in the vemurafenib group and 85·4% (77·7–93·1) in the placebo group in cohort 2. Most patients in both cohorts received their planned treatment. In the combined safety population (both cohorts), the median dose intensity in the vemurafenib group was 82·1% (IQR 67·5–98·6) versus 99·0% (97·1–99·7) in the placebo group, and the median treatment duration was 364 days (IQR 168–365) in the vemurafenib group versus 364 days (143–364) in the placebo group. Consistent with the known safety profile of vemurafenib, arthralgia, skin-related events (eg, alopecia and rash), nausea, and fatigue were the most frequently reported adverse events (table 2). Most adverse events were grade 1 or 2 in severity and were manageable. Grade 3–4 adverse events occurred in 141 (57%) of 247 patients in the vemurafenib group and 37 (15%) of 247 patients in the placebo group. The most common grade 3–4 adverse events in the vemurafenib group were keratoacanthoma (24 [10%] of 247 patients), arthralgia (17 [7%]), squamous cell carcinoma (17 [7%]), rash (14 [6%]), and elevated alanine aminotransferase (14 [6%]). In the placebo group, grade 3–4 adverse events did not exceed 2% for any of the reported terms. A complete list of all grade 3–4 events is provided in the appendix (pp 17–19). Serious adverse events were reported in 40 (16%) of 247 patients in the vemurafenib group and 25 (10%) of 247 patients in the placebo group. The most common serious adverse event was basal cell carcinoma, which occurred in eight (3%) patients in each group (appendix pp 21, 22). In the overall safety population (both cohorts combined; n=494), 49 (20%) of 247 patients receiving vemurafenib

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

7

Articles

A

Placebo group (n=91)

Vemurafenib group (n=93)

Events (n)/ Median disease-free survival, patients (N) months (95% CI)

Events (n)/ Median disease-free survival, patients (N) months (95% CI)

Hazard ratio (95% Wald CI)

53/91

15·4 (11·1–35·9)

52/93

23·1 (18·6–26·5)

0·81 (0·55–1·19)

≤65

47/80

18·0 (12·0–35·9)

42/77

23·6 (18·6–37·2)

0·80 (0·53–1·22)

>65

6/11

11·1 (2·9–NE)

10/16

20·5 (13·6–NE)

0·84 (0·30–2·34)

AUS–NZL–ZAF–LCR

5/9

15·4 (3·0–NE)

4/10

24·4 (22·1–NE)

0·63 (0·17–2·34)

North America

3/6

47·8 (5·7–47·8)

1/6

NE (26·5–NE)

0·51 (0·05–5·73)

45/76

14·6 (7·5–28·3)

47/77

20·7 (17·7–24·2)

0·83 (0·55–1·25)

Male

36/59

14·6 (6·4–47·8)

33/52

20·1 (16·5–23·6)

0·94 (0·58–1·53)

Female

17/32

25·1 (12·3–NE)

19/41

26·5 (22·1–NE)

0·71 (0·37–1·37)

Present

31/49

15·4 (7·0–35·9)

39/57

18·3 (14·7–23·6)

0·94 (0·58–1·52)

Not present

14/29

28·3 (5·7–NE)

11/28

24·2 (20·1–NE)

0·68 (0·31–1·51)

0

45/78

21·4 (12·0–35·9)

44/81

23·6 (18·3–37·2)

0·82 (0·54–1·25)

1

8/13

7/11

22·1 (18·2–23·5)

0·66 (0·23–1·88)

All patients in cohort 2 Age group (years)

Region

Rest of world Sex

Tumor ulceration status–mitosis at baseline

ECOG performance status at baseline 11·1 (2·8–NE)

First metachronous nodal recurrence at baseline Yes

22/32

15·4 (5·7–35·9)

17/29

22·1 (18·3–37·2)

0·75 (0·39–1·42)

No

30/58

25·1 (7·0–NE)

35/64

23·5 (17·7–NE)

0·86 (0·53–1·41)

Macrometastasis

20/27

12·0 (3·6–18·0)

19/30

22·1 (18·2–37·2)

0·52 (0·27–0·98)

N3

33/64

26·1 (12·1–NE)

33/63

23·5 (18·2–NE)

0·95 (0·58–1·55)

Lymph node type at baseline

0·1

B

1

10 Hazard ratio (95% Wald CI)

Placebo group (n=157)

Vemurafenib group (n=157)

Events (n)/ Median disease-free survival, patients (N) months (95% CI)

Events (n)/ Median disease-free survival, patients (N) months (95% CI)

72/157

36·9 (21·4–NE)

45/157

NE (NE)

0·55 (0·38–0·80)

≤65

64/137

39·6 (20·8–NE)

39/136

NE (NE)

0·53 (0·36–0·79)

>65

8/20

36·9 (7·4–NE)

6/21

NE (14·6–NE)

0·69 (0·24–1·99)

AUS–NZL–ZAF–LCR

6/22

NE (24·4–NE)

4/21

NE (NE)

0·67 (0·19–2·40)

North America

9/29

NE (25·8–NE)

7/30

NE (29·8–NE)

0·73 (0·27–1·96)

57/106

21·4 (13·3–NE)

34/106

NE (32·9–NE)

0·50 (0·32–0·76)

Male

43/88

27·0 (15·2–NE)

28/84

NE (29·8–NE)

0·59 (0·37–0·96)

Female

29/69

NE (20·8–NE)

17/73

NE (NE)

0·50 (0·27–0·91)

Stage IIC

6/12

36·9 (15·2–NE)

0/15

NE (NE)

Stage IIIA

15/39

NE (24·4–NE)

8/36

NE (29·8–NE)

0·52 (0·22–1·23)

Stage IIIB

51/106

25·8 (14·5–NE)

37/106

NE (32·9–NE)

0·63 (0·41–0·96)

Present

25/49

25·8 (14·5–NE)

20/63

NE (33·3–NE)

0·57 (0·32–1·03)

Not present

45/94

27·0 (15·2–NE)

21/81

NE (32·9–NE)

0·45 (0·27–0·76)

0

62/136

39·6 (21·4–NE)

42/143

NE (NE)

0·58 (0·39–0·86)

1

10/21

36·9 (7·1–NE)

3/12

NE (32·9–NE)

0·39 (0·11–1·42)

Yes

16/36

NE (12·1–NE)

8/26

No

56/121

39·6 (21·4–NE)

Macrometastasis

32/69

Micrometastasis

34/76

All patients in cohort 1 Age group (years)

Region

Rest of world Sex

Pathological stage <0·01 (0·00–NE)

Tumor ulceration status–mitosis at baseline

ECOG performance status at baseline

First metachronous nodal recurrence at baseline NE (32·9–NE)

0·59 (0·25–1·38)

37/131

NE (NE)

0·55 (0·36–0·83)

27·0 (12·1–NE)

17/58

NE (32·9–NE)

0·53 (0·29–0·96)

39·6 (21·4–NE)

28/84

NE (28·5–NE)

0·69 (0·42–1·13)

Lymph node type at baseline

0·1 Favours vemurafenib

8

1

10 Favours placebo

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Articles

and five (2%) of 247 receiving placebo discontinued the study treatment because of an adverse event. The most common adverse events leading to treatment discontinuation in the vemurafenib group were those related to the skin or musculoskeletal tissue (appendix p 20). In the placebo group, no single preferred adverse event term was a reason for treatment discontinuation in more than 1% of patients. In the vemurafenib group, 155 (63%) of 247 patients had a dose modification (dose reduction or treatment interruption) because of adverse events. The most common adverse events leading to dose modification were dermatological events, including rash (30 [12%] of 247 patients), photosensitivity reaction (12 [5%]), maculopapular rash (12 [5%]), pruritus (seven [3%]), hyperkeratosis (six [2%]), and palma–plantar erythrodysaesthesia syndrome (six [2%]). In the placebo group, 38 (15%) of 247 patients had a dose modification; no single preferred adverse event term was a reason for dose modification in more than 2% of patients. As expected, more vemurafenib-treated patients developed secondary skin malignancies than their placebo-treated counterparts (appendix p 23); new primary melanomas were regarded as disease-free survival events and are included in efficacy rather than safety analyses. Patients given vemurafenib had a higher incidence of squamous cell carcinoma of the skin (18 [7%] of 247 patients) than those given placebo (three [1%] of 247). There was no difference in the incidence of basal cell carcinoma between the treatment groups. The incidence of keratoacanthoma was also higher in vemurafenib-treated patients (24 [10%] of 247 patients) than placebo-treated patients (two [1%] of 247). Overall, 39 [16%] of 247 patients treated with vemurafenib and six [2%] of 247 treated with placebo had squamous cell carcinoma, keratoacanthoma, Bowen’s disease, or a combination of two or three of these. One fatal adverse event (grade 5) was reported in a patient in the vemurafenib group of cohort 2, who died 2 months after admission to hospital for grade 3 hypertension. Brain imaging revealed haemorrhage in a cerebral lesion consistent with metastasis; the patient had surgery for the cerebral haemorrhage and was subsequently discharged. The death was not considered to be related to the study drug.

Discussion The BRIM8 study did not meet its primary disease-free survival endpoint in patients with stage IIIC disease (cohort 2). However, in patients with resected stage IIC–IIIA–IIIB BRAFV600 mutation-positive melanoma (cohort 1), adjuvant vemurafenib therapy did reduce the risk of a disease-free survival event versus placebo. Although the study design rendered this benefit Figure 3: Forest plot of disease-free survival in (A) cohort 2 and (B) cohort 1 AUS=Australia. NZL=New Zealand. ZAF=South Africa. LCR=Latin America–Caribbean. ECOG=Eastern Cooperative Oncology Group. NE=not estimable.

Vemurafenib (n=247)

Placebo (n=247)

Grade 1–2

Grade 3

Grade 4

Grade 1–2

Grade 3

Grade 4

Any

104 (42%)

130 (53%)

11 (4%)

182 (74%)

33 (13%)

4 (2%)

Arthralgia Alopecia

134 (54%)

17 (7%)

93 (38%)

1 (<1%)

0

54 (22%)

0

0

14 (6%)

0

0 0

1 (<1%)

27 (11%)

3 (1%)

0

Rash

78 (32%)

13 (5%)

Hyperkeratosis

85 (34%)

2 (1%)

0

6 (2%)

0

0

Nausea

85 (34%)

1 (<1%)

0

45 (18%)

0

0

Photosensitivity reaction

80 (32%)

4 (2%)

0

9 (4%)

0

0

Fatigue

71 (29%)

7 (3%)

0

69 (28%)

1 (<1%)

0

Pruritus

70 (28%)

2 (1%)

0

30 (12%)

0

0

Diarrhoea

60 (24%)

5 (2%)

0

39 (16%)

2 (1%)

0

Headache

49 (20%)

0

0

43 (18%)

0

0

Dry skin

48 (19%)

0

0

18 (7%)

0

0

Pain in extremity

47 (19%)

1 (<1%)

0

19 (8%)

0

0

Pyrexia

44 (18%)

0

0

18 (7%)

0

0

Sunburn

44 (18%)

0

0

4 (2%)

0

0

Asthenia

38 (15%)

3 (1%)

0

23 (9%)

0

0

Skin papilloma

40 (16%)

0

0

4 (2%)

0

0

Erythema

37 (15%)

0

0

14 (6%)

0

0

Decreased appetite

33 (13%)

0

0

13 (5%)

0

0

Vomiting

33 (13%)

0

0

13 (5%)

0

0

Myalgia

29 (12%)

3 (1%)

0

12 (5%)

0

0

Alanine aminotransferase increased

28 (11%)

13 (5%)

Dysgeusia

24 (10%)

Aspartate aminotransferase increased

21 (9%)

Maculopapular rash

16 (6%)

Hypertension

14 (6%)

Lipase increased

3 (1%)

Basal cell carcinoma

1 (<1%)

8 (3%)

1 (<1%)

0

1 (<1%)

0

7 (3%)

0

0

8 (3%)

0

7 (3%)

1 (<1%)

0

5 (2%)

0

2 (1%)

0

0

6 (2%)

0

2 (1%)

2 (1%)

0

11 (4%)

0

1 (<1%)

1 (<1%)

0

2 (1%)

6 (2%)

0

3 (1%)

5 (2%)

0

Gammaglutamyltransferase increased

1 (<1%)

7 (3%)

1 (<1%)

0

0

0

Squamous cell carcinoma of the skin

1 (<1%)

17 (7%)

0

0

3 (1%)

0

Keratoacanthoma

0

24 (10%)

0

1 (<1%)

1 (<1%)

0

Data are n (%). A full table of all grade 3–4 adverse events is in the appendix pp 17–19. Only one grade 5 adverse event occurred in this study (described in the Results); this death was not considered to be related to study drug.

Table 2: Summary of common (any grade in ≥10% and grade 3–4 in ≥2%) adverse events reported, regardless of causality

statistically non-significant and these findings should therefore be regarded as exploratory only, the risk reduction for disease-free survival events in cohort 1 patients is greater than that reported in studies of adjuvant interferon or ipililumab.5,7 A two-cohort design was chosen for this trial so that disease-free survival events from stage IIIC patients would not mask any clinical benefit in those with a lower recurrence risk, since available historical data show that the risk of recurrence and death is higher for resected stage IIIC disease than for resected stage IIC–IIIB disease. However, the rate of disease-free survival events in

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

9

Articles

cohort 2 was slower than anticipated and necessitated an amendment to reduce the number of disease-free survival events from 120 to 105 to trigger the primary disease-free survival analysis. Cohort 1 of the current study included a small number (n=27 [9%]) of stage IIC patients whose survival and recurrence risk prognosis is very similar to that of patients with stage IIIB disease, who constituted the majority of patients in this cohort (n=212 [68%]).2,14 Thus, it is unlikely that the observed disease-free survival benefit in cohort 1 was driven by the small number of patients with stage IIC disease at diagnosis. Future studies of adjuvant therapy in melanoma should include patients with stage IIC disease, because their recurrence risk profile is similar to that of patients with stage IIIB disease. Although adjuvant interferon was widely approved for the treatment of melanoma at the time BRIM8 was designed, the trial sponsors chose to use placebo as the control group in this study for several reasons: first, because a placebo comparator would minimise bias in assessments of safety, efficacy, and quality of life; and second, because interferon treatment was associated with modest efficacy and poor safety and tolerability13,16,17 and was not considered a widely used global standard of care. Since most clinical trials of various interferon regimens in patients with melanoma had evaluated 1 year of systemic adjuvant treatment,5 a planned treatment duration of 1 year of adjuvant vemurafenib was chosen. Recently reported contemporary studies exploring dabrafenib and trametinib or immune checkpoint inhibitors also assessed a 1-year adjuvant regimen.8,9 As noted previously, the observed median disease-free survival in both groups of both cohorts was much longer (approximately 1·5 to two times) than we had initially assumed in the sample size considerations for this study. At the time of the study design (2010–11), there were scarce data to reliably describe disease-free survival or recurrence-free survival outcomes in this population, and available data were additionally confounded by non-alignment of disease subclassifications with the most recent AJCC classification;2 it was therefore necessary to extrapolate from various pooled analyses of interferon trials.13,14 The observed median disease-free survival for the placebo group of cohort 2 was consistent with recent analyses in patients with resected stage IIIC melanoma not receiving adjuvant therapy (11·0 months),18 and might be attributed to improvements in surgical outcomes and standards of care over time. In patients with resected stage IIIC disease (cohort 2), although disease-free survival was not significantly improved with vemurafenib, biological activity was apparent, since the disease-free survival curves exhibited a sizeable and early separation favouring vemurafenib over placebo. However, the disease-free survival curves converged at around 2 years. Given that prognostic factors were well balanced between study cohorts and groups, it is reasonable to conclude that patients with stage IIIC disease might be biologically distinct from patients with 10

stage IIC–IIIA–IIIB disease. The fact that the disease-free survival curves in cohort 2 patients (stage IIIC) begin to converge at the time when patients had completed their planned adjuvant treatment might also mean that prolonged treatment duration or treatment intensity might have mitigated the observed pattern of recurrence. In a recent study, 4 months of adjuvant dabrafenib monotherapy did not result in a detectable improvement in the 2-year recurrence-free survival rate in the 21 evaluable patients with resected stage IIIC BRAFV600E/K mutation-positive melanoma.19 Any differences in the relative clinical benefit with regard to locoregional and distant relapse between patients with stage IIIC disease and those with stage IIC–IIIA–IIIB disease are difficult to meaningfully dissect because of the small number of events and the fact that the disease-free survival and distant metastasis-free survival curves trend very similarly. The safety profile of adjuvant vemurafenib in this study was consistent with that previously observed in patients with metastatic disease in the pivotal BRIM3 study,11,12 and also in a more recent, large (n=3219), open-label study with similar follow-up in the setting of metastatic disease.20 Most commonly reported adverse events were grade 1 or 2 and were manageable. Treatment was discontinued for adverse events in 20% of patients, which compares favourably with discontinuation rates reported with other approved adjuvant treatment options7,13,17 and was in a similar range to that of emerging options (nivolumab [9·7%]9 and dabrafenib plus trametinib [26·0%]8). The incidence of secondary malignancies, including squamous cell carcinoma of the skin, was similar to that previously observed with vemurafenib.12,20 More recent adjuvant studies have shown clinically meaningful benefit in patients with resected stage III–IV disease.8,9 The placebo-controlled COMBI-AD study8 recently showed that combination adjuvant treatment with a MEK and a BRAF inhibitor reduced the risk of recurrence in patients with resected stage III BRAFV600 mutationpositive melanoma (HR 0·47, 95% CI 0·39–0·58, p<0·001). In addition, a recent head-to-head adjuvant study (CheckMate 238)9 comparing immunotherapy regimens of nivolumab versus ipilimumab in patients (regardless of BRAF mutation status) with resected stage IIIB–IV melanoma showed that nivolumab significantly reduced the rate of recurrence or death (HR 0·65, 95% CI 0·51–0·83, p<0·001) with a lower incidence of grade 3–4 events (25·4% vs 55·2%) and discontinuation rate related to adverse events (9·7% vs 42·6%). Based on these results, it is apparent that combination adjuvant treatment with BRAF and MEK inhibitors or single-agent nivolumab provide more favourable disease-free survival and overall survival outcomes in patients with melanoma who are at high risk of recurrence. Given the extent of the heterogeneity of recurrence risk, we cannot exclude a role for single-agent BRAF inhibitors within certain disease substages of this patient population, although there are no ongoing or planned studies to explore this question.

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

Articles

The BRIM8 study had several limitations. First, few data were available to define recurrence risk in the subsets of patients with resected disease included in the two cohorts in this study, which could have affected the overall design and powering of the study. Second, the small sample size resulted in a low number of patients in some of the patient subgroups to provide useful clinical insights. Finally, future interpretation of the data in the context of the upcoming revised AJCC classification of melanoma might warrant further evaluation. In conclusion, the primary disease-free survival end­ point was not met in patients with resected stage IIIC BRAFV600 mutation-positive melanoma, although 1 year of adjuvant vemurafenib showed a numerical benefit in disease-free survival for patients with resected stage IIC–IIIA–IIIB disease.

3

Contributors MMai, KL, DS, GRG, BS, and CY conceived and designed the study. GRG, BS, CY, and YY facilitated data collation and analysis. All authors wrote and reviewed the manuscript and provided final approval on the version to be submitted.

9

Declaration of interests MMai reports serving on advisory boards for Roche, Bristol-Myers Squibb, AstraZeneca, and Incyte. KL reports grants and personal fees from Roche/ Genentech. LD reports personal fees from Roche, Bristol-Myers Squibb, Merck Sharp & Dohme, Novartis, GlaxoSmithKline, and Amgen. MMan reports grants from Roche and personal fees from Roche, Novartis, Bristol-Myers Squibb, and Merck Sharp & Dohme. PAA reports serving on advisory boards for Roche/Genentech, Bristol-Myers Squibb, Array, Merck Sharp & Dohme, Novartis, and Amgen, and receiving grants from Roche/Genentech, Bristol-Myers Squibb, and Array. PR reports serving on advisory boards for Roche, Novartis, Merck Sharp & Dohme, Bristol-Myers Squibb, Blueprint, and Amgen; receiving lecture fees from Roche, Novartis, Merck Sharp & Dohme, Bristol-Myers Squibb, and Pfizer; and receiving grants from Novartis. AG reports personal fees from Roche, Merck Sharpe & Dohme, Merck, Bristol-Myers Squibb, and Eisai, and travel support from Bristol-Myers Squibb and Astellas. GRG, BS, CY, and YY are employees of, and hold stock in, Roche/Genentech. DS reports patients’ fees from Roche, Novartis, Merck, Merck Sharp & Dohme, Bristol-Myers Squibb, and GlaxoSmithKline, and personal fees from Amgen, Boehringer Ingelheim, and Leo Pharma. All other authors declare no competing interests. Acknowledgments We thank all the patients who agreed to take part in the trial. We also thank the investigators who participated (appendix pp 2–5) and the Genentech and Roche teams. The study was funded and conducted by F Hoffman–La Roche Ltd. Medical writing and editorial support for this manuscript was provided by Melanie Sweetlove and Jerome F Sah of ApotheCom, and was funded by F Hoffmann–La Roche Ltd. References 1 Coit DG, Thompson JA, Algazi A, et al. Melanoma, version 2.2016, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2016; 14: 450–73. 2 Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 2009; 27: 6199–206.

4

5 6

7 8

10 11 12

13

14

15 16

17 18 19

20

Balch CM, Gershenwald JE, Soong SJ, et al. Multivariate analysis of prognostic factors among 2,313 patients with stage III melanoma: comparison of nodal micrometastases versus macrometastases. J Clin Oncol 2010; 28: 2452–59. Sivendran S, Chang R, Pham L, et al. Dissection of immune gene networks in primary melanoma tumors critical for antitumor surveillance of patients with stage II–III resectable disease. J Invest Dermatol 2014; 134: 2202–11. Ives NJ, Suciu S, Eggermont AMM, et al. Adjuvant interferon-alpha for the treatment of high-risk melanoma: an individual patient data meta-analysis. Eur J Cancer 2017; 82: 171–83. Malczewski A, Marshall A, Payne MJ, et al. Intravenous high-dose interferon with or without maintenance treatment in melanoma at high risk of recurrence: meta-analysis of three trials. Cancer Med 2016; 5: 17–23. Eggermont AMM, Chiarion-Sileni V, Grob J-J, et al. Prolonged survival in stage III melanoma with ipilimumab adjuvant therapy. N Engl J Med 2016; 375: 1845–55. Long GV, Hauschild A, Santinami M, et al. Adjuvant dabrafenib plus trametinib in stage III BRAF-mutated melanoma. N Engl J Med 2017; 377: 1813–23. Weber J, Mandala M, Del Vecchio M, et al. Adjuvant nivolumab versus ipilimumab in resected stage III or IV melanoma. N Engl J Med 2017; 377: 1824–35. Tsao H, Chin L, Garraway LA, Fisher DE. Melanoma: from mutations to medicine. Genes Dev 2012; 26: 1131–55. Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011; 364: 2507–16. McArthur GA, Chapman PB, Robert C, et al. Safety and efficacy of vemurafenib in BRAFV600E and BRAFV600K mutation-positive melanoma (BRIM-3): extended follow-up of a phase 3, randomised, open-label study. Lancet Oncol 2014; 15: 323–32. Eggermont AMM, Suciu S, Santinami M, et al. Adjuvant therapy with pegylated interferon alfa-2b versus observation alone in resected stage III melanoma: final results of EORTC 18991, a randomised phase III trial. Lancet 2008; 372: 117–26. Kirkwood JM, Manola J, Ibrahim J, et al. A pooled analysis of Eastern Cooperative Oncology Group and intergroup trials of adjuvant high-dose interferon for melanoma. Clin Cancer Res 2004; 10: 1670–07. Brookmeyer R, Crowley J. A confidence interval for the median survival time. Biometrics 1982; 38: 29–41. Bottomley A, Coens C, Suciu S, et al. Adjuvant therapy with pegylated interferon alfa-2b versus observation in resected stage III melanoma: a phase III randomized controlled trial of health-related quality of life and symptoms by the European Organisation for Research and Treatment of Cancer Melanoma Group. J Clin Oncol 2009; 27: 2916–23. Mocellin S, Pasquali S, Rossi CR, Nitti D. Interferon alpha adjuvant therapy in patients with high-risk melanoma: a systematic review and meta-analysis. J Natl Cancer Inst 2010; 102: 493–501. Madu MF, Schopman JHH, Berger DMS, et al. Clinical prognostic markers in stage IIIC melanoma. J Surg Oncol 2017; 116: 244–51. Momtaz P, Harding JJ, Ariyan C, et al. Four-month course of adjuvant dabrafenib in patients with surgically resected stage IIIC melanoma characterized by a BRAFV600E/K mutation. Oncotarget 2017; 8: 105000–10. Blank CU, Larkin J, Arance AM, et al. Open-label, multicentre safety study of vemurafenib in 3219 patients with BRAFV600 mutation-positive metastatic melanoma: 2-year follow-up data and long-term responders’ analysis. Eur J Cancer 2017; 79: 176–84.

www.thelancet.com/oncology Published online February 21, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30106-2

11