Skip to main content
NCBI home page
ESMO Real World Data and Digital Oncology logoLink to ESMO Real World Data and Digital Oncology
. 2024 Sep 18;6:100071. doi: 10.1016/j.esmorw.2024.100071

Comparative effectiveness among BRAF plus MEK inhibitors for patients with BRAF V600-mutant melanoma☆

GK In 1,∗, K Chen 2, G Sajeev 3, R Simpson 3, S Kalia 3, D Christensen 4, D Liu 3, N Rezai 5, A di Pietro 6, J Signorovitch 3
PMCID: PMC12836643  PMID: 41646100

Abstract

Background

Understanding of comparative efficacy of treatments can inform clinical decision-making. This study compared overall survival (OS) and progression-free survival (PFS) across patients with metastatic BRAFV600-mutant melanoma initiating encorafenib + binimetinib (ENCO + BINI), dabrafenib + trametinib (DAB + TRAM), and vemurafenib + cobimetinib (VEM + COBI).

Materials and methods

In this hybrid study, we contextualized OS and PFS between patients with metastatic BRAF V600E/K-mutant melanoma receiving ENCO + BINI in the phase III COLUMBUS trial (enrollment: December 2013 to April 2015) versus real-world data (RWD) from a nationwide electronic health record-derived deidentified database (treatment initiation: 2014-2021). After observing consistent outcomes across trial and RWD, we compared OS and PFS for a pooled ENCO + BINI cohort across these settings versus comparable DAB + TRAM and VEM + COBI cohorts from the real-world database.

Results

Of 716 patients [ENCO + BINI (n = 275; n = 192 from COLUMBUS, n = 83 from RWD), DAB + TRAM (n = 387), VEM + COBI (n = 54)], mean age was 56-60 years. OS and PFS were similar for ENCO + BINI-treated patients in COLUMBUS and RWD [adjusted hazard ratios: 1.03 (95% CI 0.62-1.72) for OS, 1.10 (0.69-1.75) for PFS]. Relative to the pooled ENCO + BINI group, adjusted OS and PFS were significantly worse for DAB + TRAM [OS: 1.32 (1.05-1.65), PFS: 1.49 (1.20-1.87)] and comparable for VEM + COBI [OS: 1.17 (0.76-1.79), PFS: 1.20 (0.79-1.82)]. Results were similar in comparisons based on the RWD groups alone, when trial data were excluded.

Conclusions

OS and PFS were longer with ENCO + BINI relative to DAB + TRAM and comparable to VEM + COBI, after adjusting for differences in patient profiles. These findings add to evidence informing the use of combination v-Raf murine sarcoma viral oncogene homolog B protein (BRAF)/mitogen-activated protein kinase kinase (MEK) inhibitors in metastatic BRAF V600-mutant melanoma.

Key words: encorafenib, binimetinib, electronic health record, melanoma, overall survival, progression-free survival

Highlights

  • •

    Real-world data and clinical trial data of ENCO + BINI were assessed for consistency.

  • •

    Pooled ENCO + BINI data were used to compare outcomes with other BRAF/MEK inhibitors.

  • •

    Survival was longer with ENCO + BINI relative to DAB + TRAM and comparable to VEM + COBI.

  • •

    Findings may help inform the use of BRAF/MEK inhibitors in metastatic melanoma.

Introduction

Melanoma is the fifth most common cancer in the USA and the deadliest form of skin cancer. Approximately 4% of melanoma patients are diagnosed with metastatic disease.1, 2, 3 The presence of v-Raf murine sarcoma viral oncogene homolog B protein (BRAF) V600 mutations, found in 40%-60% of melanoma cases, can stimulate sustained mitogen-activated protein kinase (MAPK) kinase (MEK) pathway signaling, leading to tumor growth and progression.4 Treatment options for metastatic melanoma include immunotherapy (IO) and combination BRAF and MEK targeted therapies.5,6

In 2018, the US Food and Drug Administration (FDA) approved the use of encorafenib in combination with binimetinib (ENCO + BINI) for the treatment of patients with unresectable or metastatic melanoma with a BRAF V600E or V600K mutation, based on the phase III COLUMBUS trial,7 which found that ENCO + BINI improved overall survival (OS) and progression-free survival (PFS) compared with vemurafenib monotherapy.8 There are, however, no head-to-head clinical trials comparing ENCO + BINI versus other FDA-approved BRAF/MEK inhibitor therapies, including dabrafenib plus trametinib (DAB + TRAM, approved in 2014) and vemurafenib plus cobimetinib (VEM + COBI, approved in 2015).9, 10, 11

Real-world data (RWD) are increasingly used to inform drug development, regulatory evaluations, health technology assessments, and clinical practice in oncology.12, 13, 14 RWD may help contextualize real-world effectiveness versus efficacy observed in trials, and importantly, help evaluate effectiveness in broader populations, including those with comorbidities or functional impairment who are ineligible for trials. RWD therefore offers a valuable source of data for assessment of comparative effectiveness. In addition, combining RWD and clinical trial data (CTD) may facilitate more robust evidence generation for newer treatments over using RWD alone, particularly when RWD is more limited; this approach may allow for increased sample sizes and greater precision among smaller treatment groups, potential to compare longer-term outcomes, and comparisons versus treatments not investigated in head-to-head trials. Combining RWD and CTD for ENCO + BINI, when suitable, would therefore facilitate more comprehensive comparative analyses versus other BRAF/MEK inhibitors than using either RWD alone, or indirect comparisons based on CTD alone.15

Differences between RWD and CTD in study populations, prognostic factors, outcome assessment, and background care, however, can bias comparisons across these settings.16 In practice, suitability of comparing and combining RWD and CTD should be determined case-by-case, considering whether the most important risks of bias can be addressed. Empirical evidence of consistency of outcomes for the same patient population across RWD and CTD settings17 can also provide an important check on risk of bias and increase confidence in comparisons across these settings.

In this study, we first contextualized real-world clinical effectiveness of ENCO + BINI through an empirical assessment of outcomes in RWD versus the COLUMBUS trial, after accounting for differences in patient profiles across these settings. Following this assessment of consistency, we pooled CTD and RWD for ENCO + BINI, and compared outcomes from this pooled group versus other BRAF/MEK inhibitors. We also conducted sensitivity analyses comparing BRAF/MEK inhibitors based on the available RWD alone, without CTD.

Materials and methods

Design and setting

This hybrid study used both CTD and RWD. The CTD were from the phase III COLUMBUS trial, a randomized, open-label, multinational study comparing ENCO + BINI versus vemurafenib or encorafenib monotherapy in patients with BRAF V600 E/K-mutant unresectable or metastatic melanoma.8 Trial participants were enrolled between December 2013 and April 2015, with follow-up data availability through 15 September 2020.

RWD for patients with advanced melanoma were drawn from the Flatiron Health electronic health record (EHR)-derived database, a longitudinal, deidentified real-world database derived from EHRs from approximately 250 to 280 US cancer clinics (approximately 800 sites) across the USA.18 Data included patient-level structured, EHR fields and data elements curated from unstructured sources (e.g. physician notes, radiology and pathology reports) curated via technology-enabled abstraction.19,20 Patient-level data from January 2011 to January 2022 were available for use in this study.

Independent ethics committees or review boards at each study site approved the COLUMBUS trial study protocol and amendments. Conduct of the study conformed with Good Clinical Practice guidelines and the ethical requirements outlined in the Declaration of Helsinki. Written informed consent was obtained from all patients before screening procedures were initiated. RWD drawn from Flatiron Health were deidentified, and as such institutional review board approval was not required for use in this study. This study was registered on clinicaltrials.gov (NCT05260684).

Participants

All patients randomized to the ENCO + BINI arm of the COLUMBUS trial were included. Patients from the real-world database meeting key selection criteria from the COLUMBUS trial who initiated ENCO + BINI, DAB + TRAM or VEM + COBI following approval of these therapies for metastatic melanoma in the USA, starting from January 2011 until 30 September 2021, were selected: eligible patients were adults aged 18 years and older with metastatic BRAF V600E/K-mutated melanoma, who had one or more prescription order for ENCO + BINI, DAB + TRAM or VEM + COBI after the date of metastatic diagnosis, had no history of leptomeningeal or central nervous system metastases, had no history of bone marrow or organ transplant before treatment initiation, were untreated or had prior expert clinician-defined first-line (1L) IO in the therapeutic sitting, and had Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1 within a 3-month window before treatment initiation.

Outcomes

OS and PFS from the time of treatment initiation (index date) through the end of available follow-up were analyzed. A side-by-side comparison of assessment and definitions of OS and PFS in COLUMBUS versus the real-world database is shown in Table 1.

Table 1.

Clinical trial versus real-world assessment of OS and PFS

Outcome COLUMBUS phase III trial (CTD)6 Real-world database (RWD)11
OS Data collection/assessment
  • -

    Patients were followed-up for survival during study visits while on treatment, and after treatment discontinuation every 12 weeks until death, loss to follow-up, withdrawal of consent for study participation or study closure.

  • -

    Survival follow-up was obtained in several ways, including via phone or study visits.

Analytic definition
  • -

    Time from the date of randomization (to ENCO + BINI) to the date of death due to any cause.

  • -

    If death was not observed, patients were censored at the date of last contact or the data analysis cut-off date, whichever occurred first.

 Data collection/assessment
  • -

    Data on death were drawn from structured or unstructured data fields in the EHR, and publicly available sources of mortality including the US Social Security Death Index, and commercial obituary data.

Analytic definition
  • -

    Time from index treatment initiation (i.e. ENCO + BINI, DAB + TRAM or VEM + COBI) to the date of death due to any cause.

  • -

    Patients without a recorded date of death were censored at their last known activity date (e.g. the last clinical note date) or the end of the follow-up period, whichever occurred first.
PFS Data collection/assessment
  • -

    Progression was based on regular imaging of tumors during the trial using RECIST version 1.1 criteria and confirmed by a blinded independent review committee.

  • -

    Assessments of progression occurred on a prespecified schedule: every 8 weeks from randomization during the first 24 months, and every 12 weeks after the first 24 months.

  • -

    Survival was assessed as outlined above for OS.

Analytic definition
  • -

    Time from the date of randomization (to ENCO + BINI) to the date of the first documented disease progression or death due to any cause while on ENCO + BINI, whichever occurred first.

  • -

    If a patient did not progress or die while on ENCO + BINI, or switched to another therapy before progression or death, they were censored at the date of the last adequate tumor assessment while on ENCO + BINI.

 Data collection/assessment
  • -

    Progression was based on a retrospective clinician interpretation of the entire patient chart, including evidence from radiographic scans, biopsy, physical examination, and/or clinical assessment.

  • -

    Mean interval between progression assessments during the study period was 11.1 weeks.

  • -

    Survival was assessed as outlined above for OS.

Analytic definition:
  • -

    Time from index treatment initiation to the earliest of date of first disease progression event on that treatment or death in the absence of progression.

  • -

    Patients without disease progression or death while on the index treatment, or who switch to another therapy were censored at the last date they could have been assessed for progression while on the index treatment (e.g. the last clinical note date).
Open in a new tab

CTD, clinical trial data; DAB + TRAM, dabrafenib plus trametinib; EHR, electronic health record; ENCO + BINI, encorafenib plus binimetinib; OS, overall survival; PFS, progression-free survival; RWD, real-world data; VEM + COBI, vemurafenib plus cobimetinib.

Comparator cohorts

Study outcomes were first compared between patients randomized to the ENCO + BINI arm in COLUMBUS (ENCO + BINI CTD) and patients initiating ENCO + BINI in RWD drawn from the real-world database (ENCO + BINI RWD), after applying key trial eligibility criteria to the RWD, adjusting for differences in baseline patient profiles.

Based on the empirical observation of consistent OS and PFS across ENCO + BINI groups, patients receiving ENCO + BINI were then pooled across these two groups, and compared collectively against patients initiating either DAB + TRAM or VEM + COBI identified from RWD.

Comparisons of DAB + TRAM and VEM + COBI versus the ENCO + BINI CTD and ENCO + BINI RWD groups individually were also conducted in sensitivity analyses.

Statistical analyses

Baseline demographics, clinical and treatment characteristics were summarized for all treatment groups. For the RWD groups, patient characteristics were taken from the index date or the most recent visit within 3 months before the index date. To facilitate baseline adjustment, missing baseline data on ECOG PS (1%-40% missing across groups), lactate dehydrogenase (LDH; 54%-60%), body mass index (BMI; 0%-20%), race (0%-12%), time from metastasis to treatment initiation (0%-1%), and time from initial diagnosis to metastasis (0%-2%) were imputed using multiple imputation by chained equations.21

OS and PFS were first compared across ENCO + BINI CTD and ENCO + BINI RWD groups using Kaplan–Meier analyses and unadjusted and adjusted Cox proportional hazards models. Adjustment factors were sex, age, race, BMI, ECOG PS, LDH, prior treatment experience, prior medication, prior surgery, time from metastasis to treatment initiation, and time from initial diagnosis to metastasis. Given consistent OS and PFS outcomes (see results), the two ENCO + BINI groups were pooled and compared versus the DAB + TRAM and VEM + COBI groups, using similar unadjusted and adjusted analyses. Models were fit in each imputed dataset and effect estimates were pooled across datasets using Rubin’s rules.21,22

Sensitivity analyses

Several sensitivity analyses were conducted. First, unadjusted and adjusted analyses were repeated with the ENCO + BINI CTD and ENCO + BINI RWD treated as separate groups in statistical models for comparison with DAB + TRAM and VEM + COBI. Second, analyses excluding the COLUMBUS CTD entirely were conducted; for this analysis, all multiple imputation and statistical modeling steps were conducted based on the RWD alone. Additionally, as changes in the treatment landscape over the study period may have affected OS outcomes, we conducted Kaplan–Meier analyses investigating OS by year of treatment initiation (2014-2021) in the DAB + TRAM group. Similar analyses by year of treatment initiation were not feasible for VEM + COBI due to small sample size. We also repeated our unadjusted and adjusted analyses of OS restricting the analytic sample to RWD patients with index dates following the approval of ENCO + BINI on 27 June 2018, to remove any impact of differences between groups due to the time period of treatment initiation.

Results

The selection criteria for patients identified from real-world EHR are shown in Supplementary Table S1, available at https://doi.org/10.1016/j.esmorw.2024.100071. Patient characteristics at baseline are summarized in Table 2. Age at treatment initiation was similar across all treatment groups. Patients in the ENCO + BINI RWD group (n = 83) had worse LDH, worse ECOG PS, and were more likely to have had prior 1L IO, compared with the ENCO + BINI CTD group (n = 192). When these two groups were pooled together, patients in the pooled ENCO + BINI group (n = 275) had better LDH [68.4% ≤upper limit of normal (ULN) versus 49.4% and 64.8%] and better ECOG PS (33.1% with ECOG PS = 1 versus 45.7% and 35.2%), relative to the DAB + TRAM (n = 387) or VEM + COBI groups (n = 54); the pooled ENCO + BINI group also had a lower likelihood of having had prior 1L IO in therapeutic setting (17.5% versus 23.0% and 24.1%), and higher likelihood of therapies in the adjuvant setting (81.5% versus 64.3% and 66.7%).

Table 2.

Baseline characteristics

Criteria ENCO + BINI CTD (N = 192) ENCO + BINI RWD (N = 83) ENCO + BINI pooled (N = 275) DAB + TRAM RWD (N = 387) VEM + COBI RWD (N = 54)
Age, mean ± SD (median) 56.2 ± 13.6 (57.0) 60.4 ± 14.2 (62.0) 57.5 ± 13.9 (59.0) 59.3 ± 13.5 (61.0) 56.4 ± 14.6 (56.0)
Males, n (%) 115 (59.9) 47 (56.6) 162 (58.9) 236 (61.0) 37 (68.5)
White/Caucasian, n (%) [after multiple imputation] 181 (94.3) 71 (85.5) 253 (92.0) 353 (91.2) 48 (88.9)
BMI, mean ± SD (median) [after multiple imputation] 26.8 ± 5.0 (26.0) 28.3 ± 7.0 (26.6) 27.2 ± 5.7 (26.3) 29.3 ± 6.2 (29.0) 27.9 ± 6.0 (27.6)
Year of treatment initiation, n (%)
 2014-2015 192 (100.0) 0 (0.0) 192 (69.8) 115 (29.7) 2 (3.7)
 2016-2017 0 (0.0) 0 (0.0) 0 (0.0) 133 (34.4) 33 (61.1)
 2018-2019 0 (0.0) 41 (49.4) 41 (14.9) 99 (25.6) 12 (22.2)
 2020-2021 0 (0.0) 42 (50.6) 42 (15.3) 40 (10.3) 7 (13.0)
Geographic regions, n (%)
 Europe 156 (81.3) 0 (0.0) 156 (56.7) 0 (0.0) 0 (0.0)
 North America 17 (8.9) 83 (100.0) 100 (36.4) 387 (100.0) 54 (100.0)
 Other 19 (9.9) 0 (0.0) 19 (6.9) 0 (0.0) 0 (0.0)
BRAF mutation status, n (%)
 V600E 170 (88.5) 68 (81.9) 238 (86.5) 308 (79.6) 43 (79.6)
 V600K 22 (11.5) 15 (18.1) 37 (13.5) 79 (20.4) 11 (20.4)
ECOG performance status, n (%)
 0 136 (71.2) 36 (54.5) 172 (66.9) 125 (53.9) 22 (62.9)
 1 55 (28.8) 30 (45.) 85 (33.1) 107 (46.1) 13 (37.1)
 Missing 1 (0.5) 17 (20.5) 18 (6.5) 155 (40.1) 19 (35.2)
ECOG performance status, n (%) [after multiple imputation]
 0 137 (71.4) 46 (55.4) 184 (66.9) 210 (54.3) 35 (64.8)
 1 55 (28.6) 37 (44.6) 91 (33.1) 177 (45.7) 19 (35.2)
LDH, n (%)
 ≤ULN 137 (71.4) 22 (57.9) 159 (69.1) 72 (48.3) 16 (66.7)
 >ULN 55 (28.6) 16 (42.1) 71 (30.9) 77 (51.7) 8 (33.3)
 Missing 0 (0.0) 45 (54.2) 45 (16.4) 238 (61.5) 30 (55.6)
LDH, n (%) [after multiple imputation]
 ≤ULN 137 (71.4) 50 (60.2) 188 (68.4) 191 (49.4) 35 (64.8)
 >ULN 55 (28.6) 33 (39.8) 87 (31.6) 196 (50.6) 19 (35.2)
AJCC stage at initial diagnosis, n (%)
 0-II 80 (42.1) 21 (25.3) 101 (36.8) 85 (22.3) 7 (13.0)
 III-IV 107 (56.3) 47 (56.6) 154 (56.0) 225 (58.9) 32 (59.2)
 Unknown 5 (2.6) 15 (18.1) 20 (7.3) 77 (20.1) 15 (27.8)
Months from melanoma diagnosis to metastatic disease, mean ± SD (median) [after multiple imputation] 36.6 ± 60.6 (14.8) 49.3 ± 78.9 (14.8) 40.5 ± 66.7 (14.8) 33.1 ± 56.8 (13.5) 38.0 ± 56.2 (14.5)
Months from metastatic disease to treatment initiation, mean ± SD (median) [after multiple imputation] 9.2 ± 14.7 (3.4) 6.3 ± 10.5 (2.3) 8.3 ± 13.6 (3.0) 3.5 ± 5.7 (1.3) 3.6 ± 7.5 (1.2)
Treatment experience in therapeutic setting, n (%)
 Prior 1L IO therapy 18 (9.4) 30 (36.1) 48 (17.5) 89 (23.0) 13 (24.1)
 Treatment naive 174 (90.6) 53 (63.9) 227 (82.5) 298 (77.0) 41 (75.9)
Prior systemic therapy in adjuvant setting, n (%) 158 (82.3) 66 (79.5) 224 (81.5) 249 (64.3) 36 (66.7)
Prior IO/chemotherapy in adjuvant setting, n (%) 62 (32.3) 43 (51.8) 105 (38.2) 117 (30.2) 18 (33.3)
Prior surgery in adjuvant setting, n (%) 146 (76.0) 52 (62.7) 198 (72.0) 209 (54.0) 30 (55.6)
Open in a new tab

Note: table reports summary statistics after multiple imputation unless specified. Multiple imputation model included all baseline characteristics as well as treatment group. A total of 100 imputed datasets were created, and baseline characteristics were summarized for each imputed dataset, and averaged across all datasets. Observed characteristics were obtained from visits ≤3 months before and closest in time to the index date.

1L, first-line; AJCC, American Joint Committee on Cancer; BMI, body mass index; CTD, clinical trial data; DAB + TRAM, dabrafenib plus trametinib; ECOG, Eastern Cooperative Oncology Group; ENCO + BINI, encorafenib plus binimetinib; IO, immunotherapy; LDH, lactate dehydrogenase; RWD, real-world data; SD, standard deviation; ULN, upper limit of normal; VEM + COBI, vemurafenib plus cobimetinib.

There were differences in geography across treatment groups, with the ENCO + BINI CTD including patients in both Europe and the USA, while the RWD cohorts consisted entirely of US-based patient data. There were also differences in the year of treatment initiation, reflective of the timing of patient enrolment in the COLUMBUS study and FDA approvals of ENCO + BINI, DAB + TRAM, and VEM + COBI (2018, 2014, and 2015, respectively).

OS analyses

There were 131 deaths in the ENCO + BINI CTD and 28 deaths in the ENCO + BINI RWD over the available follow-up after treatment initiation. OS over time was not significantly different between these two groups (log-rank test P = 0.35) (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmorw.2024.100071). Median OS was numerically shorter in ENCO + BINI RWD versus ENCO + BINI CTD (24.0 versus 33.6 months). OS at 1 year (75.4% versus 75.5%) and 2 years (51.8% versus 57.7%) after treatment initiation was similar between the groups. OS was not significantly different between groups in either unadjusted [hazard ratio (HR): 1.22, (95% confidence interval, CI 0.80-1.87)] or adjusted [HR: 1.03 (95% CI 0.62-1.72)] analyses (Supplementary Table S2, available at https://doi.org/10.1016/j.esmorw.2024.100071).

There were 159 deaths in the pooled ENCO + BINI cohort, and 259 and 30 deaths in the DAB + TRAM and VEM + COBI groups, respectively. Median OS was longer for the pooled ENCO + BINI group compared with DAB + TRAM and VEM + COBI (30.0 versus 16.1 and 25.0 months, respectively) as was survival at 1 year (75.4% versus 59.6% and 66.5%, respectively) and at 2 years (56.6% versus 37.9% and 52.2%, respectively) (Figure 1). In adjusted analyses, relative to the pooled ENCO + BINI group, OS was significantly worse for DAB + TRAM [HR: 1.32 (95% CI 1.05-1.65)], and not significantly different for VEM + COBI [HR: 1.17 (95% CI 0.76-1.79)] (Table 3, top panel; Supplementary Table S3, available at https://doi.org/10.1016/j.esmorw.2024.100071).

Figure 1.

Figure 1

Open in a new tab

Kaplan–Meier curve for OS. CTD, clinical trial data; DAB + TRAM, dabrafenib plus trametinib; ENCO + BINI, encorafenib plus binimetinib; OS, overall survival; RWD, real-world data; VEM + COBI, vemurafenib plus cobimetinib.

Table 3.

Unadjusted and adjusted OS and PFS results for primary and sensitivity analyses.

OS
HR (95% CI); P value		 PFS
HR (95% CI); P value
Primary analysis: using trial and RWD, and pooling ENCO + BINI groups
DAB + TRAM versus ENCO + BINI pooled (ref.)
 Unadjusted 1.58 (1.30-1.93); P < 0.001 1.80 (1.48-2.19); P < 0.001
 Adjusteda 1.32 (1.05-1.65); P = 0.02 1.49 (1.20-1.87); P < 0.001
VEM + COBI versus ENCO + BINI pooled (ref.)
 Unadjusted 1.22 (0.83-1.80); P = 0.32 1.36 (0.92-2.01); P = 0.12
 Adjusteda 1.17 (0.76-1.79); P = 0.47 1.20 (0.79-1.82); P = 0.40
Sensitivity analysis 1: using trial and RWD, without pooling ENCO + BINI groups
DAB + TRAM versus ENCO + BINI RWD (ref.)
 Unadjusted 1.48 (1.00-2.19); P = 0.05 1.58 (1.13-2.20); P < 0.01
 Adjusteda 1.44 (0.96-2.16); P = 0.08 1.43 (1.01-2.03); P < 0.05
VEM + COBI versus ENCO + BINI RWD (ref.)
 Unadjusted 1.14 (0.68-1.91); P = 0.62 1.20 (0.74-1.92); P = 0.46
 Adjusteda 1.29 (0.74-2.23); P = 0.37 1.14 (0.69-1.90); P = 0.61
ENCO + BINI CTD versus ENCO + BINI RWD (ref.)
 Unadjusted 0.92 (0.61-1.39); P = 0.69 0.84 (0.59-1.21); P = 0.35
 Adjusteda 1.12 (0.72-1.74); P = 0.62 0.93 (0.63-1.37); P = 0.71
Sensitivity analysis 2: using RWD only
DAB + TRAM versus ENCO + BINI (ref.)
 Unadjusted 1.52 (1.03-2.25); P = 0.04 1.59 (1.14-2.22); P < 0.01
 Adjusteda 1.47 (0.98-2.20); P = 0.06 1.36 (0.96-1.93); P = 0.08
VEM + COBI versus ENCO + BINI (ref.)
 Unadjusted 1.17 (0.70-1.97); P = 0.55 1.21 (0.75-1.95); P = 0.43
 Adjusteda 1.28 (0.75-2.21); P = 0.37 1.06 (0.64-1.74); P = 0.82
Open in a new tab

1L, first-line; BMI, body mass index; CI, confidence interval; CTD, clinical trial data; DAB + TRAM, dabrafenib plus trametinib; ECOG, Eastern Cooperative Oncology Group; ENCO + BINI, encorafenib plus binimetinib; HR, hazard ratio; IO, immunotherapy; LDH, lactate dehydrogenase; OS, overall survival; PFS, progression-free survival; ref., reference; RWD, real-world data; VEM + COBI, vemurafenib plus cobimetinib.

a

Adjusted for age, sex, race, BMI, ECOG performance status, LDH, prior 1L IO, prior medication/surgery, time from metastasis to treatment initiation, and time from initial diagnosis to metastasis.

PFS analyses

A similar pattern of results was observed for PFS. There were 122 and 40 PFS events in the ENCO + BINI CTD and ENCO + BINI RWD groups, respectively. PFS over time was not significantly different between these groups (log-rank test P = 0.24) (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmorw.2024.100071). Median PFS was numerically shorter in ENCO + BINI RWD (12.4 versus 14.9 months). PFS was not significantly different between ENCO + BINI CTD and RWD groups in either unadjusted analyses [HR: 1.24 (95% CI 0.86-1.80)] or adjusted analyses [HR: 1.10 (95% CI 0.69-1.75)] (Supplementary Table S4, available at https://doi.org/10.1016/j.esmorw.2024.100071).

There were 162, 276, and 30 PFS events in the pooled ENCO + BINI group, DAB + TRAM, and VEM + COBI groups, respectively. Median PFS was numerically longer for the pooled ENCO + BINI group compared with both DAB + TRAM (14.1 versus 7.4 months) and VEM + COBI (14.1 versus 7.9 months) (Figure 2). PFS rates were higher for the pooled ENCO + BINI group compared with DAB + TRAM and VEM + COBI at 1 year (54.7% versus 32.6% and 46.9%, respectively) and at 2 years (34.4% versus 17.0% and 30.3%, respectively). In adjusted analyses, compared with the pooled ENCO + BINI groups, PFS was significantly worse for DAB + TRAM [HR: 1.49 (95% CI 1.20-1.87)] and not significantly different for VEM + COBI [HR: 1.20 (95% CI 0.79-1.82)] (Table 3, Primary Analysis; Supplementary Table S5, available at https://doi.org/10.1016/j.esmorw.2024.100071).

Figure 2.

Figure 2

Open in a new tab

Kaplan–Meier curve for PFS. CTD, clinical trial data; DAB + TRAM, dabrafenib plus trametinib; ENCO + BINI, encorafenib plus binimetinib; PFS, progression-free survival; RWD, real-world data; VEM + COBI, vemurafenib plus cobimetinib.

Sensitivity analyses

Sensitivity analyses treating the ENCO + BINI CTD and ENCO + BINI RWD as separate groups yielded similar results (Table 3, Sensitivity Analysis 1; Supplementary Tables S6 and S7, available at https://doi.org/10.1016/j.esmorw.2024.100071). After adjustment, the DAB + TRAM RWD group still had numerically worse OS [HR: 1.44 (95% CI 0.96-2.16)] and significantly longer PFS [HR: 1.43 (95% CI 1.01-2.03)] than the ENCO + BINI group. There was no significant difference between the ENCO + BINI RWD group and the VEM + COBI group in adjusted analyses of OS [HR: 1.29 (95% CI 0.74-2.23)] or PFS [HR: 1.14 (95% CI 0.69-1.90)].

Sensitivity analyses based on the RWD alone (n = 524) yielded estimated effects relative to ENCO + BINI of similar magnitude as observed in the pooled analysis (Table 3, Sensitivity Analysis 2). After adjustment, the DAB + TRAM group still had numerically worse OS [HR: 1.47 (95% CI 0.98-2.20)] and PFS [HR: 1.36 (95% CI 0.96-1.93)] than the ENCO + BINI group. There was no significant difference between the VEM + COBI and the ENCO + BINI groups in adjusted analyses of OS [HR: 1.28 (95% CI 0.75-2.21)] or PFS [HR: 1.06 (95% CI 0.64-1.74)].

Finally, median OS for DAB + TRAM patients ranged from 11.5 to 18.5 months across 2014-2021 (Supplementary Figure S3 and Table S8, available at https://doi.org/10.1016/j.esmorw.2024.100071). There were no clear trends indicating longer OS for DAB + TRAM patients initiating treatment in later years. Sensitivity analyses comparing OS in the subset of patients initiating treatment after FDA approval of ENCO + BINI (sample sizes of 83, 106, and 10 for ENCO + BINI, DAB + TRAM, and VEM + COBI groups, respectively) yielded similar results. In this subset of patients, there remained a significant difference in OS between DAB + TRAM and ENCO + BINI [HR of 1.65 (95% CI 1.03-2.66)], and no difference between VEM + COBI and ENCO + BINI [HR of 0.70 (95% CI 0.16-3.07)].

Discussion

We found consistent OS and PFS for patients with metastatic melanoma treated with ENCO + BINI across clinical trial and real-world settings. This consistency indicates that benefits experienced by patients receiving ENCO + BINI in real-world clinical practice are consistent with those observed in the COLUMBUS trial. Based on this empirical finding and supportive evidence from previous research,17 we pooled evidence for ENCO + BINI from CTD and RWD and found that OS and PFS were prolonged in patients treated with ENCO + BINI relative to DAB + TRAM, and comparable to patients treated with VEM + COBI.

These findings are in line with a recent Bayesian network meta-analysis of CTD, which reported numerically longer OS and PFS for ENCO + BINI relative to DAB + TRAM and VEM + COBI (probability of ENCO + BINI being better was 80% and 75%, respectively, for OS and 88% and 73%, respectively, for PFS).15 Side-by-side comparisons of PFS and OS rates in the underlying trials also suggest a small but consistent numerical advantage in OS and PFS for ENCO + BINI relative to DAB + TRAM and VEM + COBI.8,23,24 While the network meta-analysis also suggested higher overall response rate (ORR) for ENCO + BINI (probability of being better than DAB + TRAM and VEM + COBI was 99% and 87%, respectively), we did not have data to investigate ORR in this analysis.

OS and PFS for DAB + TRAM were shorter in our real-world sample than reported in the clinical trials of DAB + TRAM (median OS and PFS of 16.1 and 7.4 months, respectively in our unadjusted analysis versus 25.9 and 11.1 months in a pooled analysis of COMBI-v and COMBI-d trials).24 These differences may be explained by a worse baseline prognosis overall for DAB + TRAM real-world patients included here, relative to those enrolled in the COMBI-v and COMBI-d trials. The application of key ENCO + BINI COLUMBUS trial eligibility criteria to our cohort allowed inclusion of DAB + TRAM patients who had received prior 1L IO, whereas the COMBI-v and COMBI-d trials included only previously untreated patients. Compared with COMBI-v/COMBI-d trial patients, DAB + TRAM patients here were older, less likely to have ECOG PS of 0 and more likely to have LDH > ULN, all of which are associated with worse OS.25, 26, 27 Notably, the comparative analyses done in this study adjusted for these characteristics, which suggests that the longer OS for ENCO + BINI versus the DAB + TRAM real-world patients observed here is not likely explained by these differences in baseline prognostic factors.

Our comparative analyses of OS and PFS were facilitated by availability of well-validated real-world mortality and PFS endpoints.18,28 Comparability of OS outcomes between clinical trials and real-world patient populations from the deidentified database meeting key trial criteria has also been reported in several previous studies,29 including in advanced melanoma.17 Progression was determined based on RECIST-based criteria30 in COLUMBUS, versus retrospective clinician interpretation of evidence from radiographic scans, biopsy, physical examination, and/or clinical assessments in the real-world database. Nonetheless, we observed similar PFS across these settings after adjustment for baseline characteristics. This may partially be due to radiographic scans being used in >85% of assessments, assessments occurring at a similar frequency (median of 11 weeks apart in RWD versus 8-12-week intervals in the trial), and harmonization of analytic and censoring definitions of PFS across the data sources. Consistency of PFS outcomes between the real-world database and clinical trials after accounting for population differences has also been reported in other studies.31

Safety outcomes were not analyzed in this study. Indirect comparisons of safety outcomes from CTD have reported lower odds of serious adverse events (AEs) and discontinuation due to AEs with ENCO + BINI relative to DAB + TRAM and VEM + COBI.15 The impact of treatment-related toxicities on real-world treatment patterns and effectiveness of different BRAF/MEK inhibitor therapies is worthy of future investigation, as the need for dose interruptions treatment may limit effectiveness. Management of treatment-related toxicities for BRAF/MEK inhibitors has also evolved with greater prescriber experience and should be considered when comparing their safety profiles.

Patients with brain metastases, who have markedly worse prognosis,25,32 were excluded from this study, to be consistent with COLUMBUS trial criteria. About 20% of patients with confirmed BRAF+ status in our database had nervous system metastases (determined based on ICD diagnosis codes) before treatment initiation. Real-world safety outcomes and effectiveness of targeted therapies, specifically in patients with brain metastases, could also be investigated in future studies.

This study has some important strengths. Key inclusion and exclusion criteria from the COLUMBUS trial were applied to the RWD, with adjustment for key prognostic factors used to reduce risk of bias. While risk of bias cannot be ruled out entirely, the empirical observation of highly consistent OS and PFS between the ENCO + BINI trial and RWD arms indicates that possible sources of bias have a minimal net impact on these outcomes, and support combining data across these sources. In line with this consistency, our two sets of sensitivity analyses comparing DAB + TRAM and VEM + COBI to the ENCO + BINI RWD alone showed treatment differences of similar magnitude, with, as expected, slightly wider confidence intervals due to the lower precision when analyzing the RWD alone. Finally, with the emergence of IO therapies and greater accrued experience with BRAF + MEK inhibitors in the last decade, patients receiving targeted therapies later in our assessment period (2011-2021) face different considerations and options when choosing first and later-line therapies compared with patients treated earlier, which may also contribute to the differences in OS observed across treatment groups. We did not observe differences in OS by year of DAB + TRAM treatment initiation, however, and we observed similar results when restricting analyses only to RWD patients who initiated any of the three therapies after approval of ENCO + BINI, which suggests these differences are unlikely to explain our findings.

This study also has the following limitations. We were unable to apply the COLUMBUS trial requirement of having one or more measurable lesions based on RECIST, which may contribute to differences in the populations compared. COLUMBUS was a multinational trial, whereas data from the real-world database is from the USA; differences in background care and prescriber experience across geographies and care settings may also contribute to observed differences in OS and PFS. We were not able to adjust for number and sites of metastases, American Joint Committee on Cancer (AJCC) stage at treatment initiation or for potential differences in adherence and post-baseline treatments between trial and real-world settings; differences across groups in these or other unadjusted factors may also contribute to the findings. Finally, caution is warranted when interpreting differences between ENCO + BINI and other targeted treatments beyond the range of overlapping follow-up times here (∼1.5-2 years), given the smaller sample size of ENCO + BINI patients from RWD and consequently lower power to detect longer-term differences in OS. Updating this analysis as more ENCO + BINI RWD accrues will be important to assess comparability over longer time periods.

Conclusions

OS and PFS were longer in patients initiating ENCO + BINI relative to DAB + TRAM and comparable to VEM + COBI, after adjusting for differences in patient profiles. These findings add to evidence informing the use of combination BRAF/MEK inhibitor targeted therapies in metastatic BRAF V600-mutant melanoma. Comparative effectiveness of BRAF + MEK inhibitor therapies should continue to be evaluated as more RWD become available.

Acknowledgments

Funding

This work was supported by Pfizer (no grant number). The study sponsor was involved in several aspects of the research, including the study design, interpretation of data, writing of the manuscript, and decision to submit the manuscript for publication.

Disclosure

GKI’s personal and institutional financial relationships include the following: principal investigator (Pfizer), speaker’s bureau (Merck), advisory board (Novartis, Bristol Myers Squibb, Regeneron, Sanofi, Castle Biosciences, Array). Institutional financial relationships include principal investigator (Regeneron, Idera, Array Biosciences, Xencor, Checkmate Pharmaceuticals, Iovance, Bicara, and Instil Bio).

KC, NR, and ADP are full-time employees of Pfizer, Inc.

GS, RS, SK, DC, DL, and JS are employees of Analysis Group, Inc. which received consulting fees from Pfizer for this research.

Data sharing

The clinical trial data used in this study from the phase III COLUMBUS trial. Upon request, and subject to review, Pfizer will provide the data that support the findings of this study. Subject to certain criteria, conditions, and exceptions, Pfizer may also provide access to the related individual deidentified participant data. See https://www.pfizer.com/science/clinical-trials/trial-data-and-results for more information.

The data on the real-world cohort used in this study have been originated by Flatiron Health, Inc. and were purchased by Pfizer from Flatiron Health Inc. for the purpose of this research. Access to the deidentified dataset is subject to a contractual agreement with Flatiron Health; for data access, interested researchers should contact Flatiron Health. A licensing agreement is legally required before sharing these data in order to safeguard sensitive patient information, and to ensure proper deidentification and compliance with applicable restrictions and requirements under Health Insurance Portability and Accountability Act.

Supplementary data

Supplementary Tables and Figures
mmc1.docx (174.8KB, docx)
Supplementary data
mmc2.pdf (1.9MB, pdf)
Supplementary Material
mmc3.docx (532KB, docx)

References

  • 1.SEER. Cancer stat facts melanoma of the skin. National Cancer Institute. https://seer.cancer.gov/statfacts/html/melan.html Available at. Updated 2023.
  • 2.American Cancer Society Cancer Facts & Figures 2021. 2021. https://www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-figures/cancer-facts-figures-2021.html Available at.
  • 3.American Cancer Society Cancer Facts & Figures 2022. 2022. https://www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-figures/cancer-facts-figures-2022.html Available at.
  • 4.Ny L., Hernberg M., Nyakas M., et al. BRAF mutational status as a prognostic marker for survival in malignant melanoma: a systematic review and meta-analysis. Acta Oncol. 2020;59(7):833–844. doi: 10.1080/0284186X.2020.1747636. [DOI] [PubMed] [Google Scholar]
  • 5.Switzer B., Puzanov I., Skitzki J.J., Hamad L., Ernstoff M.S. Managing metastatic melanoma in 2022: a clinical review. JCO Oncol Pract. 2022;18(5):335–351. doi: 10.1200/OP.21.00686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Vanneman M., Dranoff G. Combining immunotherapy and targeted therapies in cancer treatment. Nat Rev Cancer. 2012;12(4):237–251. doi: 10.1038/nrc3237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.BRAFTOVI® (encorafenib) Prescribing information. Array BioPharma Inc. https://labeling.pfizer.com/ShowLabeling.aspx?format=PDF&id=12990 Available at.
  • 8.Dummer R., Ascierto P.A., Gogas H.J., et al. Encorafenib plus binimetinib versus vemurafenib or encorafenib in patients with BRAF-mutant melanoma (COLUMBUS): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol. 2018;19(5):603–615. doi: 10.1016/S1470-2045(18)30142-6. [DOI] [PubMed] [Google Scholar]
  • 9.National Cancer Institute Combining two types of targeted therapy improves survival in some patients with advanced melanoma (updated) Updated June 12, 2015. https://www.cancer.gov/types/skin/research/dabrafenib-trametinib Available at.
  • 10.TAFINLAR® (dabrafenib) Prescribing information. August 2023. https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/202806s027lbl.pdf Available at.
  • 11.ZELBORAF® (vemurafenib) Prescribing information. May 2020. https://www.accessdata.fda.gov/drugsatfda_docs/label/2020/202429s019lbl.pdf Available at.
  • 12.Burcu M., Dreyer N.A., Franklin J.M., et al. Real-world evidence to support regulatory decision-making for medicines: considerations for external control arms. Pharmacoepidemiol Drug Saf. 2020;29(10):1228–1235. doi: 10.1002/pds.4975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Mishra-Kalyani P., Kordestani L.A., Rivera D., et al. External control arms in oncology: current use and future directions. Ann Oncol. 2022;33(4):376–383. doi: 10.1016/j.annonc.2021.12.015. [DOI] [PubMed] [Google Scholar]
  • 14.Yap T.A., Jacobs I., Baumfeld Andre E., Lee L.J., Beaupre D., Azoulay L. Application of real-world data to external control groups in oncology clinical trial drug development. Front Oncol. 2022;11:695936. doi: 10.3389/fonc.2021.695936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Corrie P., Meyer N., Berardi R., et al. Comparative efficacy and safety of targeted therapies for BRAF-mutant unresectable or metastatic melanoma: results from a systematic literature review and a network meta-analysis. Cancer Treat Rev. 2022;110:102463. doi: 10.1016/j.ctrv.2022.102463. [DOI] [PubMed] [Google Scholar]
  • 16.Pocock S.J. The combination of randomized and historical controls in clinical trials. J Chronic Dis. 1976;29(3):175–188. doi: 10.1016/0021-9681(76)90044-8. [DOI] [PubMed] [Google Scholar]
  • 17.Signorovitch J., Moshyk A., Zhao J., et al. Overall survival in the real-world and clinical trials: a case study validating external controls in advanced melanoma. Future Oncol. 2022;18(11):1321–1331. doi: 10.2217/fon-2021-1054. [DOI] [PubMed] [Google Scholar]
  • 18.Torres A.Z., Nussbaum N.C., Parrinello C.M., et al. Analysis of a real-world progression variable and related endpoints for patients with five different cancer types. Adv Ther. 2022;39(6):2831–2849. doi: 10.1007/s12325-022-02091-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Ma X, Long L, Moon S, Adamson BJS, Baxi SS. Comparison of population characteristics in real-world clinical oncology databases in the US: Flatiron Health, SEER, and NPCR. [Preprint.] medRxiv. Advance Access published on June 7, 2023, doi: 10.1101/2020.03.16.20037143. [DOI]
  • 20.Birnbaum B., Nussbaum N., Seidl-Rathkopf K., et al. Model-assisted cohort selection with bias analysis for generating large-scale cohorts from the EHR for oncology research. [Preprint] arXiv. Advanced. Access published on January 13, 2020. [DOI]
  • 21.White I.R., Royston P., Wood A.M. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377–399. doi: 10.1002/sim.4067. [DOI] [PubMed] [Google Scholar]
  • 22.Rubin D.B. Vol. 81. John Wiley & Sons; Hoboken, NJ: 2004. (Multiple Imputation for Nonresponse in Surveys). [Google Scholar]
  • 23.Ascierto P.A., Dreno B., Larkin J., et al. 5-year outcomes with cobimetinib plus vemurafenib in BRAFV600 mutation-positive advanced melanoma: extended follow-up of the coBRIM study. Clin Cancer Res. 2021;27(19):5225–5235. doi: 10.1158/1078-0432.CCR-21-0809. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Robert C., Grob J.J., Stroyakovskiy D., et al. Five-year outcomes with dabrafenib plus trametinib in metastatic melanoma. N Engl J Med. 2019;381(7):626–636. doi: 10.1056/NEJMoa1904059. [DOI] [PubMed] [Google Scholar]
  • 25.Hauschild A., Larkin J., Ribas A., et al. Modeled prognostic subgroups for survival and treatment outcomes in BRAF v600-mutated metastatic melanoma: pooled analysis of 4 randomized clinical trials. JAMA Oncol. 2018;4(10):1382–1388. doi: 10.1001/jamaoncol.2018.2668. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Manola J., Atkins M., Ibrahim J., Kirkwood J. Prognostic factors in metastatic melanoma: a pooled analysis of Eastern Cooperative Oncology Group trials. J Clin Oncol. 2000;18(22):3782–3793. doi: 10.1200/JCO.2000.18.22.3782. [DOI] [PubMed] [Google Scholar]
  • 27.Petrelli F., Ardito R., Merelli B., et al. Prognostic and predictive role of elevated lactate dehydrogenase in patients with melanoma treated with immunotherapy and BRAF inhibitors: a systematic review and meta-analysis. Melanoma Res. 2019;29(1):1–12. doi: 10.1097/CMR.0000000000000520. [DOI] [PubMed] [Google Scholar]
  • 28.Zhang Q., Gossai A., Monroe S., Nussbaum N.C., Parrinello C.M. Validation analysis of a composite real-world mortality endpoint for patients with cancer in the United States. Health Serv Res. 2021;56(6):1281–1287. doi: 10.1111/1475-6773.13669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Ton T.G.N., Pal N., Trinh H., et al. Replication of overall survival, progression-free survival, and overall response in chemotherapy arms of non-small cell lung cancer trials using real-world data. Clin Cancer Res. 2022;28(13):2844–2853. doi: 10.1158/1078-0432.CCR-22-0471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Eisenhauer E.A., Therasse P., Bogaerts J., et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1) Eur J Cancer. 2009;45(2):228–247. doi: 10.1016/j.ejca.2008.10.026. [DOI] [PubMed] [Google Scholar]
  • 31.Huang Bartlett C., Mardekian J., Cotter M.J., et al. Concordance of real-world versus conventional progression-free survival from a phase 3 trial of endocrine therapy as first-line treatment for metastatic breast cancer. PLoS One. 2020;15(4) doi: 10.1371/journal.pone.0227256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Diaz M.J., Mark I., Rodriguez D., et al. Melanoma brain metastases: a systematic review of opportunities for earlier detection, diagnosis, and treatment. Life (Basel) 2023;13(3):828. doi: 10.3390/life13030828. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Tables and Figures
mmc1.docx (174.8KB, docx)
Supplementary data
mmc2.pdf (1.9MB, pdf)
Supplementary Material
mmc3.docx (532KB, docx)

Articles from ESMO Real World Data and Digital Oncology are provided here courtesy of Elsevier

RESOURCES