Effectiveness of immune checkpoint inhibitors and other treatment modalities in patients with advanced mucosal melanomas: a systematic review and individual patient data meta-analysis

Background: Mucosal melanomas (MM) are an aggressive subtype of melanoma. Given the rarity of this disease, the conduct of clinical trials is challenging and has been limited. Current treatment options have been extrapolated from the more common cutaneous melanoma even though MM is distinct in pathogenesis, etiology and prognosis. This is the first meta-analysis to comprehensively assess the efficacy of immune checkpoint inhibitors (anti-PD1 and anti-CTLA4) and other treatment modalities (targeted therapy such as KIT inhibitors and VEGF inhibitors, as well as radiotherapy) on survival outcomes in MM to develop clinical guidelines for evidence-based management.

Methods: The protocol was prospectively registered on PROSPERO (PROSPERO ID: CRD42023411195). PubMed, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science and Google Scholar were searched from inception until 25 July 2024, for all cohort and observational studies. Eligible studies included those with five or more participants with locally advanced or metastatic MM treated with anti-PD1, anti-CTLA4, VEGF inhibitors and/or KIT inhibitors. Titles and abstracts of potential articles were screened and full texts of all potentially eligible studies were retrieved and reviewed by two independent reviewers. Individual patient data (IPD) from published Kaplan-Meier curves were reconstructed using a graphical reconstruction method and pooled as a one-stage meta-analysis. A sensitivity analysis using a two-stage meta-analysis approach was conducted. Extracted outcomes included overall survival (OS) and progression-free survival (PFS). For each treatment arm, median survival time and 12-month survival proportion were estimated. Data from double-arm trials was pooled to estimate hazard ratios (HRs), ratios of restricted mean time lost (RMTL) and restricted mean survival time (RMST).

Findings: From a total of 7402 studies, 35 eligible studies comprising a total of 2833 participants were included. Combined anti-PD1 and anti-CTLA4 therapy had the highest 12-month OS and 12-month PFS at 71.8% (95% CI: 67.6%, 76.2%, n = 476) and 35.1% (95% CI: 30.5%, 40.4%, n = 401) respectively, followed by anti-PD1 therapy alone (OS: 64.0% (95% CI: 61.4%, 66.7%, n = 1399); PFS: was 28.3% (95% CI: 25.8%, 31.2%, n = 1142), anti-PD1 and VEGF inhibitor combination therapy (OS: 57.1% (95% CI: 51.0%, 63.9%)), KIT inhibitors (OS: 48.2% (95% CI: 37.6%, 61.8%); PFS: 8.3% (95% CI: 3.7%, 18.7%)) and anti-CTLA4 therapy alone (OS: 33.3% (95% CI: 28.4%, 39.1%); PFS: 9.8% (95% CI: 5.9%, 16.5%)). In the double-arm studies, combination therapy with anti-PD1 and anti-CTLA4 had similar OS and PFS with anti-PD1 alone (OS: HR 0.856 (95% CI: 0.704, 1.04); RMTL ratio 0.932 (95% CI: 0.832, 1.044, P = 0.225); RMST ratio 1.102 (95% CI: 0.948, 1.281, P = 0.204); PFS: HR 0.919 (95% CI: 0.788, 1.07); RMTL ratio 0.936 (95% CI: 0.866, 1.013, P = 0.100); RMST ratio 1.21 (95% CI: 0.979, 1.496, P = 0.078)), however, anti-PD1 therapy alone had significantly better PFS than anti-CTLA4 alone (HR 0.548 (95% CI: 0.376, 0.799); RMTL ratio 0.715 (95% CI: 0.606, 0.844, P < 0.001); RMST ratio 1.659 (95% CI: 1.316, 2.092, P < 0.001)). Anti-PD1 therapy with radiotherapy versus anti-PD1 alone showed no significant difference (OS: HR 0.854 (95% CI: 0.567, 1.29); RMTL ratio 0.855 (95% CI: 0.675, 1.083, P = 0.193); RMST ratio 1.194 (95% CI: 0.928, 1.536, P = 0.168; PFS: HR 0.994 (95% CI: 0.710, 1.39); RMTL ratio 1.006 (95% CI: 0.87, 1.162, P = 0.939); RMST ratio 0.984 (95% CI: 0.658, 1.472, P = 0.939)).

Interpretation: For the systemic treatment of MM, anti-PD1 is the best monotherapy. While combining anti-PD1 with other treatment options such as anti-CTLA4, VEGF inhibitors or radiotherapy might achieve better outcomes, these improvements did not reach statistical significance when evaluated by HR, RMTL and RMST ratios.

Funding: This work was supported by the National Medical Research Council Transition Award (TA20nov-0020), SingHealth Duke-NUS Oncology Academic Clinical Programme (08/FY2020/EX/67-A143 and 08/FY2021/EX/17-A47), the Khoo Pilot Collaborative Award (Duke-NUS-KP(Coll)/2022/0020A), the National Medical Research Council Clinician Scientist-Individual Research Grant-New Investigator Grant (CNIGnov-0025), the Terry Fox Grant (I1056) and the Khoo Bridge Funding Award (Duke-NUS-KBrFA/2024/0083I).