Association between SGLT-2 inhibitor use and incident hepatocellular carcinoma in type 2 diabetes: a systematic review and meta-analysis

Type 2 diabetes mellitus (T2DM) is associated with an increased risk of hepatocellular carcinoma (HCC). Sodium-glucose cotransporter 2 inhibitors (SGLT-2i) are widely used for glycemic management, but their association with incident HCC remains uncertain. PubMed/MEDLINE, Embase, Web of Science, and the Cochrane Library were searched from inception through 18 May 2025, supplemented by reference-list screening. Eleven observational cohort reports were eligible. A prespecified hierarchy was used: compatible adjusted hazard ratios (HRs) were pooled for the HCC-specific primary synthesis, while reports without compatible HRs were retained for a separate unadjusted count-based OR sensitivity analysis when complete 2 × 2 counts were available. HR and OR estimates were never combined in one model, and one selected contrast per report was used within each synthesis. Random-effects models, leave-one-out analyses, comparator-, region-, and MASLD-based subgroup analyses, overlap sensitivity analyses, and GRADE certainty assessment were performed. Cirrhosis was the secondary outcome; liver-related mortality was not pooled because only one study reported an eligible estimate. Eleven reports contributed 2,597,368 analytical records across the selected study-level contrasts; this total is not a count of unique individuals because several reports used potentially overlapping databases. Six HCC-specific studies (424,892 analytical records) contributed compatible adjusted HRs. SGLT-2i use was associated with lower incident HCC (HR 0.60, 95% CI 0.48–0.75; I²=53.7%). Leave-one-out estimates ranged from HR 0.53 to 0.65 and remained statistically significant. Reports lacking compatible HRs were not discarded: a separate count-based sensitivity analysis across nine HCC-specific reports showed an association (unadjusted OR 0.49, 95% CI 0.39–0.63; I²=80.6%). HCC-specific comparator subgroup ORs were 0.37 for DPP-4i, 0.60 for other or mixed therapies, and 0.81 for GLP-1RA (interaction p < 0.001); these unadjusted, sparse analyses were exploratory and were not interpreted as effect modification of the adjusted-HR estimate. The pooled adjusted HR for incident cirrhosis was 0.84 (95% CI 0.70–1.00; I²=94.6%; p = 0.046), but this result became nonsignificant after omission of either of two studies. Certainty was very low for both outcomes. In observational data, SGLT-2i use was associated with a lower rate of incident HCC, but the certainty of evidence was very low because of residual confounding, heterogeneity, and possible population overlap. The cirrhosis finding was highly heterogeneous and not robust to leave-one-out analysis. These results do not establish a causal preventive effect or justify selecting SGLT-2i solely to prevent HCC. PROSPERO CRD420251058401.