Prognostic Factors Associated With Mortality in Adult Hemophagocytic Lymphohistiocytosis: A Systematic Review and Meta-Analysis.

Adult hemophagocytic lymphohistiocytosis (HLH) carries a high case-fatality rate, yet reported prognostic factors are discordant and have never been pooled. The aim of this study was to perform the first systematic review and meta-analysis of prognostic factors for mortality in adult HLH. PubMed/MEDLINE, Embase, Web of Science, and the Cochrane Library were searched (from database inception to February 15, 2026) for observational cohort studies of adults (≥18 years) with HLH diagnosed by the HLH-2004 criteria and/or the HScore, reporting mortality against at least one candidate prognostic factor. Risk of bias was appraised using the Quality In Prognosis Studies (QUIPS) tool, and certainty using GRADE. Random-effects meta-analyses (DerSimonian-Laird with Hartung-Knapp correction) were pre-specified for factors reported in at least three cohorts. Reporting followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020. A total of 18 cohorts (3,570 adults) were included. Pooled mortality was 49.5% (95% CI: 41.1-57.8), rising from 34.4% (29.0-40.0) within 30 days to 57.3% (47.9-66.4) at longest follow-up (p<0.001). Malignancy-associated HLH carried higher mortality than other triggers (47.1% vs. 32.2%; 12 cohorts, n=2,373; pooled odds ratio 2.53, 95% CI: 1.55-4.12; I²=66.7%). Older age (2.23, 95% CI: 1.57-3.15; I²=0%), hyperferritinemia (3.07, 95% CI: 1.07-8.82), and elevated lactate dehydrogenase (2.02, 95% CI: 1.05-3.88) independently predicted death; thrombocytopenia and prolonged activated partial thromboplastin time were directionally harmful but imprecise. Egger's test indicated small-study effects (p=0.014). Certainty was low to moderate. Approximately half of adults with HLH die following diagnosis, with nearly one-third of deaths occurring within 30 days. A malignant trigger approximately doubles the odds of death; age, ferritin, and lactate dehydrogenase add independent information. Prospective, multinational registry data with pre-specified thresholds are needed.