Outcome of combination therapy using BRAF and MEK inhibitors among Asian patients with advanced melanoma: An analysis of 112 cases
Abstract Background: As most clinical trials evaluating BRAF and MEK inhibitor combi- nation therapy (B + Minh) have been conducted in Western countries, little is known about the effect of B + Minh among East Asian populations.
Material and methods: Data from patients with advanced melanoma treated using B + Minh (either dabrafenib + trametinib or encorafenib + binimetinib) were retrospectively collected from 16 institutes in Japan. Response rates, adverse events, patterns of failure and survival were analysed.
Results: We analysed 112 of 144 collected patient records and, of these, 14 had acral/mucosal melanoma. The response rate for the entire cohort was 75.0%. There were no statistical differ- ences in response rates between acral/mucosal and cutaneous melanomas (64.3% versus 76.5%), whereas previous treatment using immune checkpoint inhibitors (ICIs) did not affect response (72.7% versus 73.9%) to B + Minh, response to ICI after B + Minh was only 20%. Patients who achieved complete response had the best overall survival rates at 24 months (94.7%). Elevated serum lactate dehydrogenase levels and 3 or more metastatic sites were inde- pendently associated with survival. The most common relapse site was the brain (17.9%). More than half of the patients (58.8%) experienced grade III/IV pyrexia.
Conclusion: B + Minh was effective among Japanese patients with melanoma, including those with acral/mucosal melanoma. Factors associated with survival were similar to previous West- ern studies. B + Minh response was not affected by the previous use of ICI; however, vigilance against brain metastasis during B + Minh therapy is required as the brain was our most commonly encountered relapse site.
1. Introduction
Treatment of advanced melanoma dramatically changed with the introduction of immune checkpoint inhibitors (ICIs) and combination therapy using BRAF and MEK inhibitors (B + Minh) [1,2]. These new therapies have been proven effective in randomised clinical trials [3e7], but most of these studies were conducted in Western countries. As we have previously published, the pro- portion of cutaneous melanoma among Japanese melanoma cases is only one-third that of the West while acral/mucosal melanoma makes up more than half of all melanoma cases in this population [8]. The proportion of BRAF mutations greatly correlates with clinical type; approximately 50e70% of cutaneous melanomas harbour a BRAF mutation but only 10e20% of acral/ mucosal melanomas do [9]. Therefore, the applicability of Western trial results to Japanese patients is suspect.
With regard to ICIs, we previously reported that the overall survival (OS) of patients with acral melanoma was inferior to that of patients with superficial spreading melanoma [10,11], a result similar to other reports [12,13]. This indicates a lower susceptibility of acral/ mucosal melanoma to ICIs, an idea bolstered by the pooled analysis of the CheckMate 067/069 trial in which patients with/mucosal melanoma receiving nivolumab or in combination with ipilimumab experienced lower efficacy compared to cutaneous melanoma cases [14]. These results were recently mirrored in the JAMP study of 193 patients with acral melanoma in Japan [27]. Taken together, previous reports indicate that clinical type correlates directly with outcome when using ICIs.
Among Asian populations, Bai et al. [15] reported a response rate of acral/mucosal melanoma using B + Minh as 20%e38.1%, whereas others reported 61%e78.9% response rates [16e18]. However, all such reports included no more than 77 patients and definite conclusions are therefore difficult. We thus conducted a retrospective study to collect more data to evaluate the efficacy of B + Minh therapy and determine factors associated with outcome among Japanese patients with melanoma.
2. Material and methods
Patients with advanced melanoma who received B + Minh therapy (either dabrafenib + trametinib or encorafenib + binimetinib) from April 2010 to August 2019 at the 16 participating institutes were collected. Patient data, including age, sex, performance status, primary melanoma site, serum lactate dehydrogenase (LDH) level, metastatic sites, BRAF mutation test re- sults, previous treatments, outcome of B + Minh, adverse events, patterns of failure and treatments after B + Minh. This study was approved by the institutional review board of every participating institute. Tumour response was evaluated using Response Evaluation Criteria In Solid Tumours,version 1.1. The 95% confidence interval (95% CI) for response was based on binominal distribution.
The Chi-square test was used for analysis of cate- gorical variables. Progression-free survival (PFS) and OS were defined as the time from the initial B + Minh treatment date to the date of either progression or death, as appropriate. PFS and OS were estimated using Kaplan-Meier curves and the log-rank test was used to compare differences in OS. For factors shown to be significant by the univariate analysis, multivariate Cox regression analyses in a stepwise procedure were performed to determine the independent factors associated with survival. In this procedure, a 0.1 significance level for entering and eliminating explanatory variables was used. P-values of less than 0.05 were considered signif- icant. All statistical analyses were performed using Stat Flex, version 6.0 (Artech, Osaka, Japan).
3. Results
3.1. Patient characteristics
In total, data from 144 patients were collected, but 32 patients were excluded for the following reasons: 12 received BRAF inhibitor only, 10 used the therapy in an adjuvant setting and 10 lacked clinical information. Finally, 112 patients with advanced melanoma treated with B + Minh were enrolled in this study. Patient characteristics are shown in Table 1. There were 59 men (52%) and ages ranged from 20 to 90 years (mean age: 60.5 years). Most of the study cohort had cutaneous melanoma (85 patients) but 11 patients with acral and 3 patients with mucosal melanoma were also enrolled. There were 11 patients with unresectable stage III dis- ease, but the rest (101 patients) had distant metastases. Of these, 43 patients had 3 or more metastatic sites and 45 patients had elevated LDH levels. Thirty-four pa- tients had been previously treated with ICIs and 15 patients with vemurafenib. A majority of the patients (110 patients) were treated with dabrafenib plus trame- tinib, and 2 patients were treated with encorafenib plus binimetinib. No patient was treated with vemurafenib plus cobimetinib since this combination is not approved in Japan. Of 112 patients, 31 patients are still receiving treatment, whereas 81 patients have had their treatment terminated due to progression of disease (49 patients), adverse events (20 patients), patient decision (7 patients) or other unspecified reasons (5 patients).
3.2. Overall response rate
The result of the response evaluation is shown in Table 2. Overall, 23 patients achieved complete response (CR), 61 had a partial response (PR), 24 maintained a stable disease (SD) state, and 4 experienced progressive disease (PD). The response rate was 75.0% (95% CI: 67.0e82.1) and the disease control rate (CR + PR + SD) was 96.4%. The percentage of patients who achieved CR among patients without previous use of ICI or vemur- afenib was higher than those with treatment (25.6% versus 8.8% and 23.7% versus 0%, respectively) but there were no significant differences. Response rates between patients with or without previous ICI treatment were similar (73.5% versus 75.6%). In line with this, response rates between patients with or without previous vemurafenib were similar (66.7% versus 76.3%). Inter- estingly, the percentage of patients who achieved CR and response rate among acral/mucosal melanoma cases was similar to cutaneous melanoma (28.6% versus 21.2% and 64.3% versus 76.5, respectively).
3.3. Survival
PFS and OS are shown in Fig. 1. The median PFS of the entire cohort was 13.0 months (95% CI: 6.0e21.8) and the median OS of the entire cohort was 35.0 months (95% CI: 16.0 to not reached).
3.4. Adverse event
Patients who experienced grade III or IV adverse events (G3/4 AE) are listed in Table 3. Since only 2 patients received encorafenib plus binimetinib, we excluded their data from our adverse event analyses. The total number of GIII/IV AEs were 52 and the most common GIII/IV AE was pyrexia (23 patients), followed by elevated serum liver enzyme levels (5 patients) cutaneous reaction (5 patients), and elevated creatinine kinase levels (5 pa- tients). Most patients with GIII/IV AEs had their treatments interrupted. Of the GIII/IV pyrexia sufferers, 18 required steroids. There were no treatment-related deaths in this cohort.
3.5. Patterns of recurrence
During the B + Minh therapy, there were 68 events in 58 patients (Table 4). Notably, the brain was the most common site of recurrence with development in 20 pa- tients (17.9%). The lymph nodes were the next most common (11 patients), followed by skin (9 patients), liver (7 patients), bone (7 patients) and lungs (6 patients).
3.6. Factors associated with progression-free survival
Similar to a previous analysis using clinical trial data [19], patients who achieved CR had significantly better PFS compared with patients with PR and SD (Fig. 2A). Patients with metastases to less than 3 sites had better PFS compared with those with 3 or more metastatic sites (Fig. 2B). However, the study cohort did not reach significance when grouped by normal and elevated LDH (Fig. 2C). There were significant differences between M1b and M1d patients, whereas others showed no dif- ferences (Fig. 2D). Interestingly, there were no signifi- cant differences by clinical types (Fig. 2E), previous use of vemurafenib (Fig. 3A), and previous use of ICI (Fig. 3B). Similarly, no differences were seen in GIII/IV AE development, use of steroids, dose reduction, or dose interruption (Fig. 3CeF).
3.7. Factors associated with OS
Patients who achieved CR had significantly better OS compared with PR and SD (Fig. 4A). Patients with less than 3 metastatic sites and normal LDH had better OS (Fig. 4B and C). However, there were no significant differences between stages (Fig. 4D). Similar to that of PFS, clinical subtype was not associated with OS (Fig. 4E). As expected, patients previously treated with vemurafenib or ICI had worse OS (Fig. 5A and B) Although the development of GIII/IV AE, dose reduc- tion and interruption did not affect OS, patients receiving steroids showed a tendency toward worse OS (Fig. 5CeF).
Univariate analysis using Cox regression analysis showed similar results as determined by log-rank testing (Table 5). By multivariate analysis, previous use of vemurafenib (hazard ratio: 2.28, P Z 0.023), serum LDH levels (hazard ratio: 2.18, P Z 0.013) and the number of metastatic sites (hazard ratio: 1.99, P Z 0.033) were independently associated with OS.
3.8. Therapy after B + Minh
Fifty-three patients received subsequent therapy after B + Minh as shown in Table 6. The most common therapy was ICI (50 patients) followed by B + Minh switching (4 patients), chemotherapy (3 patients), oper- ation (2 patients), and radiation (2 patients). The response rate among patients who received ICI was 20% (3 CR and 7 PR). As for patients receiving ICI, received ipilimumab monotherapy (no responder), 34 received nivolumab or pembrolizumab monotherapy (2 CR and 6 PR), and 12 received nivolumab plus ipilimumab (1 CR and 1 PR).
4. Discussion
BRAF inhibitor therapy for patients with BRAF mu- tations was started in 2011 with the approval of vemurafenib. To improve response and better prevent acquired resistance to BRAF inhibitors, combined use of MEK inhibitors became the standard of care [3e5]. Currently, there are 3 combinations available: vemurafenib + cobimetinib, dabrafenib + trametinib, and encorafenib + binimetinib. Each combination has been reported to be effective, with a 68% response rate for vemurafenib + cobimetinib [5], 64% rate for dabrafenib + trametinib [4] and 63% rate for encorafenib + binimetinib [3]. Therefore, B + Minh therapy for BRAF mutated melanoma carries a high expectation for favourable response. However, all the previous phase II/III trials were primarily composed of Caucasian patients with melanoma and only a very small proportion of Asian patients were included. As we previously reported, patient backgrounds between Asians and Caucasians differ [8], and the acral/mucosal melanoma more prevalent in Asia has been shown to be less susceptible to ICI [11,13,20]. Therefore, response and safety data for B + Minh therapy generated solely from an Asian population was required.
As 98% of the patients in our cohort received dabrafenib + trametinib, we compared our results to pivotal trials using dabrafenib + trametinib, namely COMBI-d [21] and COMBI-v [4]. The best overall response of our entire cohort was 73.5% (95% CI: 64.7e81.4) which is comparable to that of COMBI- d and COMBI-v (69% and 64%). The median PFS of our cohort was 13.0 months (95% CI: 8.0e19.0) which was similarly comparable to that of COMBI-d and COMBI-v (11.0 months and 11.4 months). As for OS, the median OS of our cohort was 35.0 months (95% CI: 15.0 e not reached) compared with 25.1 months (95% CI: 19.1 e not reached) in COMBI-d and not reached in COMBI-v. Although OS may not be optimal for com- parison because it might be greatly affected by subse- quent therapy after progression and/or length of follow- up, we believe that the OS in our cohort is comparable to those of COMBI-d and COMBI-v.
In the pooled analysis of COMBI-d and COMBI-v [19], normal LDH levels and less than 3 metastatic sites were associated with better survival. In our study cohort, both factors were similarly associated with OS (Fig. 4B and C) and multivariate analysis also revealed that these factors were independently associated with survival (Table 5). On the other hand, only the number of metastatic sites was associated with PFS (Fig. 2B,C),a result attributable to our small study cohort. Response to the treatment has also been reported to be associated with survival in the pooled analysis [19] because patients with CR generally achieve excellent outcomes. In our study cohort, patients with CR (23 patients) had significantly better PFS and OS (Figs. 2A and 4A). Taken together, these factors should be considered when using B + Minh.
Nearly half of Japanese patients with melanoma develop the acral or mucosal varieties [8] but only 3 studies from Asian countries using B + Minh in these types have been published [16e18]. Kim et al. {Kim, 2016 #1650} reported 19 cases of acral/ mucosal melanoma but only 6 were treated with B + Minh. Takahashi et al. {Takahashi, 2020 #1701} reported 50 Japanese patients treated with B + Minh and only 6 acral/mucosal melanoma patients were included. Even a larger study {Si, 2020 #1716} on 77 patients only examined 3 patients with acral/mucosal melanoma. Importantly, while none of these reports included efficacy data specific to acral/mucosal mela- noma, the 14 patients with acral/mucosal melanoma in our study had both responses (Table 2) and survival (Figs. 2E and 4E) similar to cutaneous melanoma cases. To the best of our knowledge, this is the largest study report on the efficacy of acral/mucosal melanoma treated with B + Minh.
We have previously published a single-institute analysis showing that the occurrence of brain metas- tasis during B + Minh therapy was statistically higher compared with ICI (40% versus 5%, respectively) [22]. In the pooled analysis of COMBI-d and COMBI-v [23], which included 617 patients, there were 113 patients (18.3%) who developed brain metastases while, in our study cohort, 20 patients (17.9%) developed the brain metastases (Table 4) most commonly seen during relapse. On the other hand, in the CheckMate 067 trial for nivolumab, ipilimumab and combination therapy for
patients with advanced melanoma [24], brain metastasis was seen in only 52 patients (5.5%), which was consis- tent with our previous analysis [22]. Taken together, vigilance against brain metastasis during B + Minh therapy is recommended to enable discovery at the early stages of this complication.
As for the safety profile, 23 patients (20.9%) of our cohort experienced GIII/IV pyrexia compared to 7% and 4% in COMBI-d and COMBI-v, respectively. Although unclear, the following two factors might have contributed: first, pyrexia occurred when the patients were at home and could not be directly confirmed. Second, systemic drug dosages for Japanese patients might be higher than the Caucasians as the average height and weight of Japanese males are reported as 1.71 m and 69.0 kg in contrast to 1.77 m and 90.9 kg for United States males [25]. According to this data, the average body surface area for Japanese males as calcu- lated by the DuBois method [26] is 1.806 square metres compared with 2.082 square metres for United States males. This translates into Japanese males receiving a 15% higher effective dose per square metre because the measured dose of BRAF and MEK inhibitors were the same. However, other GIII/IV AEs were not higher in our cohort than previous reports, which complicates efforts to directly correlate GIII/IVAEs with dosage in Japanese patients.
There are several limitations in our study. First, this study was retrospective and therefore we could not exclude possible bias. Second, the quality threshold of the data collected did not reach clinical trial levels. Third, the number of enrolled patients was much smaller than previous trials. Moreover, our study was able to enrol only 14 cases of acral/mucosal melanoma. How- ever, we believe our data might provide foundational evidence to the differential response of BRAF inhibitors in Asian patients with melanoma with BRAF mutations. An international collaborative study should be con- ducted to generate high-quality evidence based on Asian populations.