Abstract
The overall survival (OS) of patients with metastatic uveal melanoma is short, the evidence for effectiveness of treatments is limited, and no consensus on the choice of treatment exists. We aimed to advance interpretation of OS as an outcome by pooling peer-reviewed data. The design is a systematic review and meta-analysis. We searched PubMed from 1 January 1980, to 29 March 2017, for articles reporting patient-level survival in Kaplan-Meier or numerical form. We digitized survival graphs, pooled individual survival times, calculated median OS by treatment modality, and compared each modality by the log-rank test and Cox regression using conventional chemotherapy (CHT) as a reference. Individual-level data were obtained from 78 articles with 2494 patients. The median OS across all treatment modalities was 1.07 years (range: 0.59-2.50 years). Pooled OS reported after isolated hepatic perfusion [median OS: 1.34 years; hazard ratio (HR): 0.92, 95% confidence interval (CI): 0.87-0.97, P = 0.0040], immunoembolization (median OS: 1.63; HR: 0.97, 95% CI: 0.95-1.00, P = 0.0080), and surgery (median OS: 1.43; HR: 0.94, 95% CI: 0.92-0.96, P < 0.0001) was longer, and after checkpoint inhibitor shorter (median OS: 0.59; HR: 1.13, 95% CI: 1.06-1.20, P < 0.0001) than after CHT (median OS: 0.91 years), but subject to identifiable confounding factors. OS following other modalities did not differ from CHT. Reported OS was unassociated with the decade of publication, but depended on the percentage of first-line treated patients. Our results suggest no clinically significant difference in OS by treatment modality or decade. Most of the difference in reported OS likely is attributable to surveillance, selection, and publication bias rather than treatment-related prolongation. Our pooled data provide benchmarks for future trials.