From cancer safety signals to cardio-oncology translation: a dual-database bibliometric and BERTopic semantic analysis of SGLT2 inhibitor research in oncology (2014-2026).
文献摘要
Objective The clinical value of SGLT2 inhibitors has expanded from glycemic control to cardiorenal protection, while their oncology-related research encompasses long-term safety, tumor metabolism, real-world outcomes, and cardio-oncology. This study aimed to characterize the knowledge evolution of this field from cancer safety signals to testable clinical-translational questions and to distinguish research attention from evidence maturity. Methods English-language records were retrieved from Web of Science Core Collection (WoSCC) and PubMed from January 1, 2012, to July 26, 2026. The final audited corpus comprised 346 records (258 Articles and 88 Reviews). Reviews were retained as a separate descriptive stratum. The full corpus was used for publication, controlled-concept, and exploratory evidence-map analyses, whereas Article records were used for the primary BERTopic model and Article-specific inferential analyses. The WoSCC subset (n=283) was used for citation and co-citation analyses. Annual Article counts were modeled using Poisson regression with annual PubMed English-language journal-article volume as an offset to account for background growth in biomedical publishing; a sensitivity BERTopic model additionally included Reviews. Results A total of 346 records were included, of which 283 entered the citation analysis. Full-corpus publication output increased from 4 records in 2014 to 101 in 2025, while Article output increased from 3 to 67. After adjustment using annual PubMed English-language journal-article volume as the reference, the PubMed-referenced Article activity index showed an estimated annual IRR of 1.35 (95% CI, 1.28-1.43; P Conclusions SGLT2 inhibitor oncology research has expanded into parallel tracks of cancer safety, mechanistic antitumor hypotheses, real-world outcomes, and cardio-oncology supportive care. The bibliometric and semantic results describe indexed research attention rather than treatment efficacy or causal benefit. Available clinical evidence remains heterogeneous, and no overall class-wide cancer-risk increase has been confirmed without excluding rare, site-specific, agent-specific, or latency-related uncertainty. Prospective and randomized studies should address drug specificity, clinically achievable exposure, cancer-specific endpoints, treatment continuity, and patient-relevant outcomes.