Comparative efficacy and safety of combination therapies for advanced melanoma: a network meta-analysis

Melanoma is an aggressive form of cancer, with a high mortality rate and low 5-year survival rate in patients with advanced-stage disease [1]. However, since the development of immune check point inhibitors and targeted therapies, melanoma treatment has remarkably improved patient survival [2]. Immune checkpoint inhibitors, a group of monoclonal antibodies, mainly block co-stimulators that down-regulate T-cell function to help tumour cells escape from immune attacks, such as by the T lymphocyte-associated antigen-4 (CTLA4) and programmed death-1/ligand-1 (PD1/L1) signalling molecules, to activate anti-tumour immune responses [3‐5]. Mitogen-activated protein kinase (MAPK) pathway blockades are another class of molecules that can be used to effectively treat advanced melanoma; mainly composing of BRAF and MEK inhibitors. BRAF inhibitors specifically target BRAF V600 mutations [6, 7], and MEK inhibitors block the downstream signal protein kinases of the MAPK pathway [8].

Recently, with the advancement of targeted therapy, more therapies have been combined, such as CTLA-4 or PD-1/L1 blockade plus chemotherapy, CTLA-4 blockade plus PD-1/L1 blockade, BRAF inhibitor plus MEK inhibitor, MEK inhibitor plus chemotherapy and other combination regimens, have been proven to show improvement in comparison with single-agent regimens [9‐11]. For example, the ipilimumab plus dacarbazine group showed a higher overall survival (OS) rate for 3 years than the dacarbazine group (20.8% vs. 12.2%, respectively). The nivolumab plus ipilimumab group showed better median progression-free survival (PFS) than the ipilimumab group (11.5 months vs. 2.9 months, respectively) [10, 11]. Meanwhile, BRAF and MEK inhibitors also significantly improved the effectiveness of treatment and reduced the incidence of secondary skin cancer [12].

However, the evidence from several trials does not offer a holistic view for these two categories of treatments, because head to head randomized controlled trials (RCTs) are still lacking among different implements (PD-1/L1 blockade plus chemotherapy, CTLA-4 blockade plus chemotherapy, PD-1/L1 blockade plus CTLA-4 blockade, PD-1/L1 blockade plus adjuvant therapy, BRAF inhibitor plus MEK inhibitor and MEK inhibitor plus chemotherapy). Network meta-analysis (NMA) can integrate direct and indirect evidence from RCTs and perform indirect comparisons through a common comparator [13‐16]. We used this tool to analyse the efficacy and toxicity of different combination regimens of immune check point inhibitors or MAPK pathway inhibitors by OS, PFS and serious adverse events (SAEs) in patients with advanced-stage melanoma.

Three hundred sixty-seven relevant references were identified for review of their titles and abstracts. Of these, 50 randomized controlled trials were retrieved for more details (19 lacked a control group and were 7 drug-dose comparisons); finally, 24 phase II or III randomized controlled trials were identified that met the eligibility criteria of this study. In total, 10,951 patients were included in this network meta-analysis. Figure 1 shows the flow diagram of the literature search and selection of clinical trials. The characteristics of the 24 included trials are summarized in Table 1 [10‐12, 19‐39]. Four were three-arm trials, and the others were two-arm trials in this analysis. Additional file 1 shows more detailed information of trials as PD-L1 and BRAFV600 expressions and strategies that were used in this study, such as the combination of PD-1/L1 blockade and CTLA-4 blockade (N = 1087), combination of CTLA-4 blockade and chemotherapy (N = 502), combination of CTLA-4 blockade and adjuvant (N = 921), combination of BRAF and MEK inhibition (N = 1622), combination of BRAF inhibition and chemotherapy (N = 925), combined MEK inhibition and chemotherapy (N = 1377) and others.

With the emergence of targeted therapies, the treatments of advanced melanoma have significantly improved, and recently, MAPK pathway inhibitors and immune checkpoint inhibitors have been shown to be effective therapeutic choices. However, comparative trials between these two categories are lacking, especially regarding the different combinations based on these two categories. We used network meta-analysis to explain the efficacy and toxicity of all available combinations and provide suggestions for patients and clinicians. In this study, we found that the combination of BRAF and MEK inhibitors showed a significant advantage in PFS compared with all of the other implements except for the combination of anti-PD-1/L1 inhibitors and anti-CTLA-4 inhibitors (HR:0.61; 95% CrI: 0.30, 1.25). Regarding OS, this combination was also superior to the other modalities except for PD-1/L1 inhibitors (HR: 0.85; 95% CrI: 0.59, 1.21). Because the difference between the combination of BRAF and MEK inhibitors and combination of PD-1/L1 and CTLA-4 inhibitors showed no significance, we deduced that these two treatments showed non-inferiority in PFS. Similarly, regarding OS, we inferred that the combination of BRAF and MEK inhibitors showed a non-inferiority relationship with PD-1/L1 inhibitors. Because the efficiency of treatment was also associated with the risk of toxic effects, we analysed the severe adverse event rates among different implements; the results showed that there was no modality with a lower rate than chemotherapy. The combination of a MEK inhibitor and chemotherapy was the only implement with a higher rate than chemotherapy (RR: 1.76; 95% CrI: 1.21, 2.48).

From our results, we determined that treatment with the combination of BRAF and MEK inhibitors was superior to each modality except the combination of PD-1/L1 blockade and CTLA-4 blockade in PFS, with no increased risk of toxicity than any of the other treatments. Additionally, the combination of BRAF and MEK inhibitors showed a greater benefit than any implement except for the PD-1/L1 blockade in OS. However, compared with the combination of PD-1/L1 and CTLA-4 inhibitors in PFS, the combination of BRAF and MEK inhibitors showed no significant difference, and indicated that these two categories are non-inferior. Again, in OS, the combination of BRAF and MEK inhibitors was also not better than PD-1/L1 blockade, with no increased severe adverse event rate. Our findings were supported by two similar network analyses that also demonstrated the efficacy of the combination of BRAF and MEK inhibitors [40, 41]. However, in contrast to the OS data, our results showed that the combination of BRAF and MEK inhibitors showed no significant difference with CTLA-4 and GM-CSF. With the PD-1/L1 blockade, this condition may have been due to the limited number of patients who had a BRAF mutation. Thus, only one trial of CTLA-4 and GM-CSF was included in their study and caused heterogeneity. In our study, we included more than one trial and grouped them according to their properties [41]. Additionally, within the toxicity data, our results showed no evidence of a higher risk of toxicity with the PD-1/L1 and CTLA-4 combination, not as reported in the previous study. In our analysis, the combination of a MEK inhibitor and chemotherapy was the only treatment with an increased risk of toxicity. This difference might be because we used only severe adverse events as an estimate of toxicity.

There are several advantages in our study. We considered all available comparisons based on immune check point inhibitors and MAPK pathway inhibitors, such as MEK inhibitors, and their combination with chemotherapy, which were unavailable in prior studies [40, 41]. These results could provide an overall perspective of the efficacy and toxicity of different combination regimens in patients with advanced-stage melanoma. We did not limit the population of patients and included trials of patients with BRAF-mutant and PD-L1 expression [11, 35, 38, 42] because the BRAFV600E mutation was present in almost 40–60% of all patients with advanced melanoma, and BRAF-mutant patients could benefit from both MAPK pathway inhibitors and immune checkpoint inhibitors [43, 44]. We also included patients who received prior treatments because, in clinical practice, patients with treatment-naïve and prior treatments both exist. Thus, we believed the analysis of patients without limitations was more proper. Besides PFS, we also analyzed OS to evaluate the efficacy of different combination regimens for treatment of advanced-stage melanoma, which was not provided in the previous study [40]. We performed safety/toxicity analysis according to the rates of any SAEs because these results more realistic and practical than those of treatment-related adverse event rate and provided comprehensive insights into comparisons of the crossover for each treatment.

Our network analysis offers the most comprehensive comparisons based on targeted and immune check point inhibitor therapy in patients with advanced melanoma without a mutant-status limitation, which is convincing than imposing the limitation. As in the absence of a direct comparison among different treatments, our results suggest that PFS is best in patients treated with the combination of BRAF and MEK inhibitors or the combination of PD-1/L1 and CTLA-4 blockade, the efficacy of these two treatments shows no significant difference. Meanwhile, OS is best with the BRAF and MEK inhibitors combination or PD-1/L1 inhibitor, with no significant difference between these two treatments. Additionally, because of heterogeneity and the limitations, this conclusion should be interpreted very cautiously. Furthermore, several direct comparisons are ongoing, such as BRAF-MEK inhibitors compared with PD-1/L1 or CTLA-4 blockade or PD-1/L1 in combination with a MEK inhibitor or BRAF inhibitor. We believe our results will be confirmed in future trials.