Tolerance and Efficacy of BRAF Plus MEK Inhibition in Patients With Melanoma Who Previously Have Received Programmed Cell Death Protein 1-Based Therapy

Abstract

BACKGROUND:

Combined <zaq;4>BRAF and MEK inhibition (BRAF-MEK) is a standard therapy for patients with BRAF V600-mutant melanoma, but to the authors’ knowledge, the tolerance, adverse event (AE) profile, and efficacy have not been well defined in the post-programmed cell death protein 1 (PD-1) setting.

METHODS:

Patients with BRAF V600-mutant melanoma who received combined BRAF-MEK after prior PD-1-based therapy were assembled from 4 tertiary care centers in the United States and Australia. Dose modification was defined as a treatment break, dose reduction, or intermittent dosing. Rates of hospitalization and discontinuation due to AEs were collected, and overall survival (OS) was calculated using Kaplan-Meier methods from the time of the initiation of BRAF-MEK therapy.

RESULTS:

A total of 78 patients were identified as having received a BRAF-MEK regimen at a median of 34 days after the last dose of PD-1-based therapy. The majority of patients (86%) received the combination of dabrafenib and trametinib. Approximately 80% of patients had American Joint Committee on Cancer M1c or M1d disease. Sixty-five regimens (83%) had ≥1 dose modification. The median time to the first dose modification was 14 days; 86% occurred within 90 days and 71% involved pyrexia. Dose modifications were more common in patients receiving BRAF-MEK <90 days after the last dose of PD-1 and who were not receiving steroids. Of the dose modifications, 25 (31%) led to an AE-related hospitalization. Among 55 BRAF-naive patients, the median time receiving BRAF-MEK therapy was 5.8 months and the median OS was 15.6 months.

CONCLUSIONS:

The majority of patients receiving BRAF-MEK inhibition after PD-1 therapy require dose interruptions, and a significant minority require hospitalization for AEs. In this higher risk population, the median time receiving therapy and OS may be inferior to those presented in published phase 3 trials.

INTRODUCTION

Standard frontline therapies for patients with advanced BRAF V600-mutant melanoma include combined BRAF-MEK inhibition with 1 of 3 approved agents (vemurafenib plus cobimetinib [VC], dabrafenib plus trametinib [DT], or encorafenib plus binimetinib)^1–3 as well as programmed cell death protein 1 (PD-1)-based therapy with either pembrolizumab⁴ or nivolumab as monotherapy⁵ or combined nivolumab plus ipilimumab.⁶ These therapies all lead to improved prolonged progression-free survival (PFS) and overall survival (OS), but to the best of our knowledge there are no randomized data to inform frontline decision making regarding the use of BRAF-MEK versus PD-1-based therapy for patients with BRAF V600-mutant melanoma. Nonetheless, based on the perceived higher possibility of long-term disease control, anti-PD-1 therapy often is favored as first-line therapy.

Clinical and pharmacodynamic data have suggested that immune-related adverse events (AEs) can occur months after the last dose of PD-1-based therapy, presumably via prolonged PD-1 receptor occupancy on immune cells that lasts 3 to 6 months.^7,8 This is in contrast to cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) receptor occupancy, which is believed to last on the order of several weeks.⁹ To our knowledge to date, the data combining checkpoint inhibitors with BRAF-MEK inhibitors suggest a higher rate of AEs.^10–12 For example, the rate of grade 3 to grade 4 AEs in a phase 1 study of dabrafenib, trametinib, and pembrolizumab was 73%,¹¹ which appears higher than the rate of 32% noted with the use of DT.¹ An ongoing trial of vemurafenib, cobimetinib, and atezolizumab modified the original dosing schedule due to pyrexia and rash to allow a 4-week run-in of targeted therapy before therapy with programmed death-ligand 1 (PD-L1) was initiated.¹³

Although BRAF-MEK inhibition often is used in the second-line setting, the prospective trials reporting safety and efficacy to our knowledge have enrolled patients with essentially no exposure to prior PD-1 blockade; only the trial of encorafenib plus binimetinib reported any prior therapy with PD-1, and this was only in 1% of patients.^1–3 More recently, case reports^14–17 and a small series¹⁸ have described severe reactions to BRAF-MEK inhibition after PD-1 blockade, including systemic inflammatory response syndrome, hemophagocytic lymphohistiocytosis, and cutaneous toxicities. Further data are required to describe the safety, patient tolerance, and efficacy of these agents in the post-PD-1 setting. Therefore, we assembled a retrospective, multi-institutional cohort of patients receiving BRAF-MEK inhibitors at any time after exposure to PD-1-based therapy. We hypothesized that rates of dose modification, discontinuation, and hospitalization would be relatively frequent and might vary by age, stage of disease, and the interval between the last PD-1 dose and the onset of targeted therapy. Furthermore, patients who require second-line therapy with BRAF-MEK inhibitors may have reduced efficacy compared with patients in prospective trials who are treatment-naive or experienced disease progression while receiving treatment with ipilimumab alone.

MATERIALS AND METHODS

After institutional review board approval, a sequential cohort of patients with BRAF V600-mutant melanoma who received combined BRAF-MEK inhibition after prior PD-1-based therapy was assembled from 4 tertiary care centers in the United States and Australia. Dose modification was defined as a treatment break, dose reduction, or planned intermittent dosing due to AEs. Rates of hospitalization, emergency department (ED) visits, and discontinuation due to AEs were collected.

AEs deemed to be at least possibly related to BRAF-MEK therapy were collected retrospectively. Due to the retrospective nature of the review, toxicity grades were not available for fatigue and asthenia, myalgias and myopathy, and rash. Rashes were grouped because they could not be subtyped reliably retrospectively. Fever; cytopenias; elevated aspartate aminotransferase (AST), alanine aminotransferase (ALT), or creatinine; hypertension; reduced ejection fraction; and select neurologic abnormalities were graded using Common Terminology Criteria for Adverse Events (version 4.03).

Statistical Analysis

OS and time receiving BRAF-MEK therapy were calculated using Kaplan-Meier methods from the time of the initiation of BRAF-MEK therapy until the earliest next systemic therapy, clinical deterioration without further therapy, or death. The log-rank test was used to compare time on therapy between different subgroups. The association between the incidence of dose modification, discontinuation, and hospitalization with categorical variables was analyzed using the Fisher exact test.

RESULTS

Demographics

A total of 78 patients were identified (Table 1), 48 of whom (62%) were male; the median age of the patients was 58 years. The majority of primary tumors were cutaneous (82%) or unknown (11%). As expected, the most common V600 mutation was V600E (69 patients; 88%). The majority of patients had American Joint Committee on Cancer M1c or M1d disease (80%). The Eastern Cooperative Oncology Group (ECOG) performance status was 0 in 21% of patients, 1 in 57% of patients, and 2 to 3 in 22% of patients. Lactate dehydrogenase (LDH) was high in 43 patients (55%). Fifty-five patients (71%) were BRAF therapy-naive. A total of 23 patients (29%) had previously received either BRAF monotherapy or BRAF-MEK combination therapy prior to the initiation of PD-1-based therapy.

TABLE 1.

Patient Characteristics

Characteristic	No. of Patients N = 78
Median age (range), y	58 (26–88)
Male sex, no. (%)	48 (62)
Primary tumor, no. (%)
Cutaneous	64 (82)
Unknown	9 (11)
Acral	3 (4)
Mucosal	2 (3)
V600 mutation, no. (%)
E	69 (88)
K	9 (12)
AJCC stage of disease
IIIC/M0	3 (4)
M1a	2 (3)
M1b	10 (13)
M1c	32 (41)
M1d	31 (39)
ECOG performance status, no. (%)
0	16 (21)
1	45 (57)
2–3	17 (22)
Baseline LDH, no. (%)
Normal	23 (30)
Elevated	43 (55)
Unknown	12 (15)
Previous BRAF therapy, no. (%)a
No	55 (71)
Yes	23 (29)
BRAF monotherapy	10 (13)
BRAF-MEK	13 (16)
PD-1 regimen
PD-1 monotherapy, never ipilimumab	25 (32)
Nivolumab plus ipilimumab	31 (40)
PD-1 monotherapy with prior ipilimumab	20 (25)
PD-1 monotherapy with subsequent ipilimumab	2 (3)
BRAF-MEK regimens used, no. (%)
Dabrafenib and trametinib	67 (86)
Vemurafenib and cobimetinib	10 (13)
Encorafenib and binimetinib	1 (1)
Median time from last PD-1 dose to BRAF-MEK start (range), d	35 (1–410)
IQR, d	15–69

Abbreviations: AJCC, American Joint Committee on Cancer; ECOG, Eastern Cooperative Oncology Group; IQR, interquartile range; LDH, lactate dehydrogenase; PD-1, programmed cell death protein 1.

^aPrior to PD-1 therapy.

The PD-1-based regimens included combined nivolumab plus ipilimumab in 31 patients (40%), PD-1 monotherapy with no ipilimumab exposure in 25 patients (32%), or sequential PD-1 monotherapy and ipilimumab in 22 patients (28%). Sixty-seven patients (86%) received DT, 10 patients (13%) received VC, and 1 patient received encorafenib and binimetinib. The median interval between the last PD-1 dose and the initiation of therapy with BRAF-MEK was 34 days (range, 1–410 days), and approximately 75% of patients received BRAF-MEK within 74 days of the last PD-1 dose.

Adverse Events

The median time to the onset of AEs ranged from 2 weeks to 5 weeks for the majority of AEs (Fig. 1A). Overall, the most common AE was fever, which occurred in 73% of regimens (76% of DT regimens and 60% of VC regimens); the median maximum temperature was 39.3^oC. Rash was reported in 36% of regimens and appeared to be more common with VC compared with the DT regimen (80% vs 28%; P=.001). Other AEs commonly observed included fatigue and weakness (32%), AST and ALT elevation (29%; median grade of 2), nausea and vomiting (27%), and myopathy and myalgias (19%). Less common AEs included cytopenias (13%), cardiac toxicity (9%), central nervous system events (8%), and acute kidney injury (5%; grades 1–2) (Fig. 1B). The most common cytopenias were thrombocytopenia (8%) or involved multiple cell lines (4%; grades 2–3). Cardiac toxicities included grade 3 hypertension (4%) or reduced ejection fraction (3%) and single cases of grade 3 atrial fibrillation and mitral regurgitation. Central nervous system toxicities included 2 cases of grade 2 sensorineural hearing loss and single cases of seizure, aseptic meningitis, and intraparenchymal hemorrhage.

Figure 1.

(A) Time to onset of the most common adverse events (AEs) shown in a box-and-whisker plot of the median, interquartile range (IQR) (colored box), and range (outer range). (B) The most common AEs are shown. Fever was the most common AE, followed by rash, fatigue/weakness, and gastrointestinal toxicities. ALT indicates alanine aminotransferase; AST, aspartate aminotransferase; CNS, central nervous system; min, minimum; max, maximum; tox, toxicity.

Dose Modifications and Hospitalizations

A total of 65 patients (83%) experienced ≥1 dose modifications. The first dose modification consisted of a dose reduction and/or schedule change to intermittent dosing in 29 regimens (45%), and dose withholding only in 36 regimens (55%). Of the 36 cases in which the dose was withheld and restarted at the same dose, 21 patients (58%) eventually required intermittent dosing or a dose reduction. Forty-three regimens (55%) required a second dose modification and 25 regimens (32%) required ≥3 dose modifications. Among the AEs leading to a first dose modification, fever was most often implicated (71% of modifications), followed by nausea and vomiting (15% of modifications), AST and ALT elevation (11% of modifications), and rash (11% of modifications) (Fig. 2).

Figure 2.

Adverse events (AEs) associated with dose modifications. Fever was the most common AE associated with dose modification, followed by gastrointestinal AEs (nausea/vomiting, diarrhea), constitutional symptoms (fatigue/weakness, myopathy/myalgias), rash, and laboratory abnormalities (aspartate aminotransferase [AST]/alanine aminotransferase [ALT] elevation, cytopenias). CNS indicates central nervous system.

The median time to the first dose modification was 14 days, with 56 dose modifications (86%) occurring within the first 90 days of BRAF-MEK therapy. Factors found to be associated with dose modification included normal LDH (vs elevated; P=.011), systemic steroid use at the time of initiation of BRAF-MEK therapy (no vs yes; P=.002), and time from last PD-1 dose of <90 days (vs ≥90 days; P=.030) (Table 2). There was a nonsignificant trend toward modification by M1c to M1d disease versus M0 to M1b disease (P=.059). Age, ECOG performance status, prior BRAF inhibition, type of prior PD-1-based therapy, and type of BRAF-MEK therapy (DT vs VC) were not found to be significantly associated with dose modification.

TABLE 2.

Univariate Analysis of Factors Associated With Dose Modification

Factor	Pa
AJCC M1c/d disease (vs M0-M1b)	.059
Normal LDH (vs elevated)	.011
Steroids at time of treatment initiation (no vs yes)	.002
Time from PD-1 <90 d (vs ≥90 d)	.030
Age (<65 y vs ≥65 y)	.547
ECOG performance status (0/1 vs 2/3)	.464
Prior BRAF therapy (yes vs no)b	.745
Type of prior PD-1-based therapy (PD-1 monotherapy vs PD-1 plus ipilimumab (combination or sequential)	.746
Type of BRAF-MEK therapy (DT vs VC)	.199

Abbreviations: AJCC, American Joint Committee on Cancer; DT, dabrafenib and trametinib; ECOG, Eastern Cooperative Oncology Group; LDH, lactate dehydrogenase; PD-1, programmed cell death protein 1; VC, vemurafenib and cobimetinib.

^aBold type indicates statistical significance.

^bPrior to PD-1-based therapy.

Twenty-five regimens (32%) resulted in at least 1 AE-related hospitalization (range, 1–4 hospitalizations) and another 18 regimens (23%) resulted in ≥1 ED visits without admission. The most common cause for an initial hospitalization while receiving therapy was a sepsis-like syndrome with high fever with or without hypotension (20 patients; 80% of hospitalizations) followed by asthenia and fatigue (5 patients; 20%) (Table 3). Twelve patients (15%) withdrew from BRAF-MEK therapy because of AEs, most commonly fever, nausea and vomiting, and fatigue and weakness. No significant associations were observed between age, ECOG performance status, stage of disease, LDH, use of steroids at the time of targeted therapy initiation, prior BRAF inhibitor therapy, type of prior PD-1-based therapy, or type of BRAF-MEK therapy (DT vs VC) and AE-related hospitalizations or AE-related discontinuation of BRAF-MEK therapy.

TABLE 3.

AEs Associated With Initial Hospitalization

AE	Incidence (Percentage of Hospitalizations)a
Fever	20 (80)
Fatigue/weakness	5 (20)
Rash/photosensitivity	5 (20)
Nausea/vomiting	4 (16)
Diarrhea	4 (16)
Cardiac toxicity	3 (12)
Cytopenias	3 (12)
AST/ALT elevation	2 (8)
CNS toxicity	2 (8)
Renal toxicity	2 (8)

Abbreviations: AE, adverse event; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CNS, central nervous system.

^aIncidence refers to the number of regimens in which the AE was associated with the initial hospitalization. Greater than 1 adverse event could be associated with a hospitalization.

Efficacy

After a median follow-up of 14 months for patients who were BRAF therapy-naive, the median time receiving BRAF-MEK was 5.8 months and the median OS was 15.6 months (Fig. 3). Patients with elevated LDH levels were found to have a shorter median time on therapy compared with those with normal LDH values (4.3 months vs 11 months; P=.004). Patients with M1c to M1d disease versus M0 to M1b disease also had a shorter median time on therapy (5.8 months vs 11 months), although this difference was not found to be statistically significant (P=.399).

Figure 3.

Time receiving BRAF-MEK therapy and overall survival. (A) Time receiving BRAF-MEK therapy among previously BRAF therapy-naive patients. (B) Overall survival for this same population. The median follow-up was 14 months. The median time receiving BRAF-MEK therapy was 5.8 months, and the median overall survival was 15.6 months.

DISCUSSION

To the best of our knowledge, the current study cohort represents the largest cohort to date of patients with advanced melanomas receiving BRAF-MEK inhibition after anti-PD-1-based checkpoint inhibition. The most important finding is that BRAF-MEK therapy with DT or VC at the standard dose and schedule does not appear to be tolerated as well after receipt of anti-PD-1 treatment because it is in the upfront setting. The vast majority of regimens (83%) required dose modifications; this appears higher than the dose modification rates of 47% to 61% observed for vemurafenib, cobimetinib, or both,¹⁹ and the dose modification rates of 33% to 55% for DT²⁰ administered in PD-1-naive settings. Even when using a narrower definition of dose modification that ignores dose withholdings, approximately 64% of regimens in the post-PD-1 setting required a dose reduction or intermittent dosing. It is possible that the current doses for DT or VC may not have been considered tolerable if the dose escalation trials had been conducted in the post-PD-1 setting. It is interesting to note that this cohort was not helpful for examining encorafenib plus binimetinib tolerance in the post-PD-1 setting.

Factors found to be associated with dose modifications included the initiation of targeted therapy within 90 days of the last anti-PD-1 dose and a lack of steroid use at the onset of BRAF-MEK inhibition. The presence or absence of ipilimumab was not found to be associated with dose modification, although this comparison is likely underpowered. This finding supports the theory that T-cell disinhibition via PD-1 alone or in combination with CTLA-4 is mediating the increased rate of toxicity. This also is consistent with prior data linking steroid use to lower rates of recurrent pyrexia in patients enrolled on a phase 2 study of DT²¹ and suggests that prophylactic steroids may be a useful strategy for preventing pyrexia with these agents in the second-line setting. The rapid onset of symptoms such as pyrexia also suggests that short-term MAPK inhibition may exacerbate many of these AEs. Further translational work is required to understand these potential mechanisms of toxicity. For example, it is not clear why the rate of modifications was higher in patients with normal LDH versus elevated LDH; it is unlikely to be related solely to the significantly longer time receiving therapy because the majority of dose modifications began during the first 90 days of therapy.

Patients should be counseled that it often will be difficult to tolerate BRAF-MEK inhibition after anti-PD-1-based therapy. The median time to dose modification was 14 days; this most often was for fevers and chills. Pyrexia often was severe (median temperature of 39.4^oC) and mimicked sepsis in the majority of patients. Other symptoms frequently associated with dose interruptions included nausea and vomiting, rash, and diarrhea. These agents appear to cause AEs not described in the prospective trials. For example, cytopenias were noted in 13% of regimens, and 2 cases of sensorineural hearing loss were described. Approximately 1 in 3 patients required hospitalization, most often for a sepsis-like syndrome or significant weakness and fatigue. When incorporating ED visits, approximately one-half of patients will require acute evaluation for symptoms while receiving these therapies.

A major limitation of the current retrospective analysis is the lack of objective, uniform data collection for AEs, particularly for subjective ones such as nausea and fatigue. This limitation is mitigated by focusing on key clinical events, such as an initial dose modification or hospitalization, which are more easily assessed retrospectively. If anything, the current analysis may underestimate the rate of subjective AEs. Taken together, these findings inform routine clinical practice for patients who have received anti-PD-1 therapy. Given the potential severity and rapid onset of pyrexia and the potential for atypical AEs, physicians should counsel their patients to withdraw from treatment with BRAF-MEK inhibitors as soon as they experience a prodrome. Physicians can offer anticipatory guidance for these symptoms, which could lower the rates of ED use and hospitalization. With the growing use of frontline anti-PD-1-containing regimens across oncology, including in the adjuvant setting, clinical trials of subsequent-line therapies should track the last dose of anti-PD-1-based therapy to test whether those patients with more recent anti-PD-1 exposure may have a different AE profile from those with more remote anti-PD-1 exposure.

To the best of our knowledge, the large, prospective trials of BRAF-MEK inhibitors enrolled patients with relatively lower risk features: nearly all lacked brain metastases, approximately 75% had an ECOG performance status of 0, and 36% to 46% had an elevated LDH level.^1,22 In contrast, the current retrospective study cohort represents a higher risk, “real-world” population of patients with melanoma: 80% had M1c or M1d disease and 55% had an elevated LDH at the initiation of targeted therapy. The efficacy in this cohort likely reflects this higher risk disease status. In the subset of patients who were BRAF-MEK-naive after treatment with PD-1, the median time receiving targeted therapy was only 5.8 months, and the median OS was 15.6 months. Recognizing the limits of cross-trial comparisons, this appears to be lower than the median PFS of 11 to 12 months and the median OS of 22 to 25 months observed in the phase 3 trials of VC and DT.^1,22 The median time receiving treatment in this higher risk cohort was closer to the median PFS of 5.6 months noted in patients with untreated, asymptomatic brain metastases who were treated with DT.²³ It also is consistent with other retrospective analyses examining the efficacy of BRAF targeted therapy before or after anti-PD-1/PD-L1 therapy.²⁴ It is not clear whether the increased dose interruptions and reductions underlie this decrease in efficacy or whether microenvironmental features associated with PD-1 resistance also promote cross-resistance to BRAF-MEK inhibition.²⁵

The safety and efficacy data from the cohort of patients in the current study suggest that anti-PD-1 exposure may influence the frequency and magnitude of AEs associated with subsequent BRAF-MEK inhibition in ways that may impact efficacy outcomes. This will prove relevant for clinical trials that enroll patients with anti-PD-1-resistant, BRAF V600-mutant melanoma and more broadly for patients with other malignancies such as renal cell carcinoma and non-small cell lung carcinoma, in whom targeted therapy may be used after frontline anti-PD-1-containing regimens.

The efficacy and tolerance of BRAF-MEK inhibition in patients with melanoma has been defined in trials conducted prior to the advent of frontline programmed cell death protein 1 (PD-1) blockade. In the current retrospective analysis of patients who experienced disease progression while receiving PD-1-based therapy, dose modifications are frequent and efficacy appears to be lower compared with previously published prospective trials.