Sodium-glucose cotransporter-2 inhibitors and gastrointestinal neoplasm risk in type 2 diabetes: a systematic review and meta-analysis of randomized controlled trials Authors: Yu-Fan Zhang, Jia-Jie Xie, Qi Wu, Xiao-Yu Zhou, Chen Zeng, Yan Zeng, Yi-Meng He, Man Guo, Fang-Yuan Teng, Yong Xu Published: 2026 ABSTRACT AbstractThe potential carcinogenic effects of sodium-glucose cotransporter 2 (SGLT2) inhibitors in patients with type 2 diabetes mellitus (T2DM) remain controversial, particularly regarding site-specific gastrointestinal (GI) neoplasms. This systematic review and meta-analysis aimed to determine the relationship between SGLT2 inhibitors and the risk of GI neoplasms in patients with T2DM. We searched PubMed, EMBASE, Cochrane CENTRAL, Scopus, and Web of Science through March 17, 2025, for RCTs in T2DM comparing SGLT2 inhibitors with placebo or active comparators. Two reviewers independently screened studies, extracted data, and assessed the risk of bias. The primary outcome was GI neoplasms reported in publications, supplementary materials, or trial registries, usually as adverse events rather than centrally adjudicated cancer endpoints. Pooled odds ratios (ORs) with 95% confidence intervals (CIs) were calculated in Stata 17.0. In 48 RCTs (n = 48,765), SGLT2 inhibitor therapy was not associated with overall GI neoplasm risk (OR = 1.10, 95% CI: 0.84–1.44; p = 0.46; I² = 0%). Site-specific analyses showed no statistically significant association for esophageal (OR = 1.12, 95% CI 0.37–3.45), gastric (1.20, 0.65–2.23), hepatic (0.62, 0.31–1.22), pancreatic (0.91, 0.51–1.64), colonic (1.28, 0.78–2.08), colorectal (0.76, 0.27–2.17), and rectal neoplasms (0.98, 0.49–1.97), with all p-values > 0.05. Subgroup analyses by agents (e.g., canagliflozin, dapagliflozin, empagliflozin), baseline age, body mass index (BMI), HbA1c, treatment duration, and dose were also non-significant (all p > 0.05). Approximately half of the trials had follow-up of one year or less, limiting our ability to evaluate long-term risk. Available RCT evidence does not show a clear increase in GI neoplasm risk with SGLT2 inhibitors in T2DM. However, limited follow-up, low event counts, and non-cancer-specific outcome ascertainment, the findings should be interpreted as reassuring but not definitive evidence of long-term oncologic safety.Systematic review registration: PROSPERO No. CRD42024619019.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s10238-026-02300-6.Keywords: SGLT2 inhibitors, Type 2 diabetes, Gastrointestinal neoplasms, Meta-analysis, Canagliflozin, Empagliflozin, Dapagliflozin Introduction The global prevalence of diabetes continues to escalate, affecting an estimated 589 million individuals in 2024, with projections reaching 853 million by 2050 [1]. Poor glycemic control worsens vascular complications and is increasingly implicated in cancer pathogenesis [2, 3]. Type 2 diabetes mellitus (T2DM) increases overall cancer risk by 10–20%, particularly for liver, pancreatic, and colorectal neoplasms, through mechanisms including hyperinsulinemia, insulin resistance, and chronic inflammation [4]. The long-term oncogenic implications of antihyperglycemic therapies remain controversial. Observational studies reveal divergent effects: insulin analogs and sulfonylureas may elevate gastrointestinal (GI) neoplasm risk [5], whereas metformin appears protective. In contrast, dipeptidyl peptidase-4 (DPP-4) inhibitors and glucagon-like peptide-1 (GLP-1) receptor agonists show neutral or inconsistent associations [6, 7]. Sodium–glucose cotransporter 2 (SGLT2) inhibitors—now a cornerstone for cardiorenal protection in T2DM [8]—lower glucose via insulin-independent blockade of renal glucose reabsorption, yet their potential carcinogenicity, particularly for GI neoplasms, remains uncertain. Several randomized controlled trials (RCTs) provide discordant signals: EMPA-REG OUTCOME reported numerically fewer colorectal cancers with empagliflozin, whereas the CANVAS Program observed more colorectal and pancreatic events with canagliflozin; both findings were statistically non-significant [9, 10]. Beyond these largely non-significant estimates, several large outcome trials have demonstrated statistically significant cardiorenal benefits with SGLT2 inhibitors (e.g., reduced heart failure hospitalization and slower chronic kidney disease progression), prompting guideline-level uptake and sharpening the need to clarify long-term oncologic safety under real-world exposure. Program-level safety summaries have also noted site-specific, sometimes nominally significant, imbalances (e.g., colorectal/pancreatic events in canagliflozin programs; fewer colorectal events with empagliflozin [11]) that have not been consistent across agents or replicated across programs [12]. Moreover, real-world cohorts have reported statistically significant associations—for example, a lower gastric cancer incidence versus DPP-4 inhibitors (adjusted HR ≈ 0.30) and reduced risks across several GI sites—that contrast with neutral RCT results [13]. Collectively, these mixed signals underscore both the clinical relevance and evidentiary uncertainty motivating our trial-based meta-analysis of GI neoplasms across agents, doses, and exposure windows. The relationship between SGLT2 inhibitor use and gastrointestinal neoplasm risk in T2DM remains inadequately defined, especially regarding dose-response, treatment duration, and effect modification by patient characteristics. Recent broader reviews have evaluated overall cancer risk with SGLT2 inhibitors, but fewer analyses have focused specifically on GI neoplasm subsites and prespecified clinical subgroups using RCT data. To address this gap, we conducted an RCT-based systematic review and meta-analysis to evaluate the association between SGLT2 inhibitor therapy and GI neoplasm incidence, including esophageal, gastric, pancreatic, hepatic, colonic, colorectal, rectal, and biliary tract neoplasms. Stratified subgroup analyses were performed to explore potential effect modification by treatment dosage, duration, mean age, body mass index (BMI), and HbA1c. This synthesis was intended to clarify the available randomized evidence while recognizing that the included trials were not primarily designed to establish long-term cancer safety. Methods The protocol for this systematic review and meta-analysis was registered with PROSPERO (CRD42024619019). The review followed the PRISMA 2020 reporting guidelines [14], ensuring methodological transparency and adherence to contemporary standards for evidence synthesis. The completed PRISMA 2020 checklist is available in Supplementary Appendix 1. Search strategy We systematically searched PubMed, Embase, Web of Science, Cochrane Central Register of Controlled Trials (CENTRAL), and Scopus for randomized controlled trials (RCTs) of SGLT2 inhibitors in people with type 2 diabetes from database inception to March 17, 2025. Additionally, we searched for unpublished and ongoing trials in ClinicalTrials.gov, manually reviewed the reference lists of relevant systematic reviews and meta-analyses, and examined grey literature available in clinical trial registries, without language restrictions. After initial screening, potentially eligible records were assessed in full text, and relevant trial reports, supplementary materials, and registry records were reviewed for GI neoplasm outcomes. The search encompassed all SGLT2 inhibitors (e.g., dapagliflozin, empagliflozin) compared to placebo or active glucose-lowering agents (e.g., metformin, insulin) to assess the incidence of GI neoplasms during follow-up. The search strategy combined terms for SGLT2 inhibitors (e.g., *sodium-glucose cotransporter 2 inhibitor*, canagliflozin) and GI neoplasms (e.g., colorectal neoplasms, hepatocellular carcinoma), with full syntax provided in Supplementary Appendix 2. Two investigators (Yu-Fan Zhang and Jia-Jie Xie) independently screened titles/abstracts and assessed full texts for eligibility, with discrepancies resolved by consensus or consultation with a third reviewer (Qi Wu). Full texts, supplementary materials, and trial registry records were comprehensively reviewed to identify GI neoplasm outcomes that were not fully reported in the primary publications. Study selection Studies that fulfilled the following criteria were included in this meta-analysis: [1] participants were T2DM patients; [2] follow-up period ≥ 24 weeks; [3] the experimental group received SGLT2 inhibitor therapy (monotherapy or combination regimens), and the control group received placebo or non-SGLT2 glucose-lowering therapies; [4] reported GI neoplasm outcomes in publications, supplementary materials, or trial registries (e.g., ClinicalTrials.gov). Exclusion criteria were: [1] non-RCTs, including reviews, case-control studies, cross-sectional studies, and case reports; [2] patients with type 1 diabetes mellitus or non-diabetic individuals; [3] duplicate reports; and [4] reports for which outcome data could not be reliably extracted after review of all available sources (Supplementary Appendix 3). For studies with multiple publications, the version with the longest follow-up or most complete data was selected. Data extraction and quality assessment Data were extracted independently by two researchers (Yu-Fan Zhang and Jia-Jie Xie) and reviewed by a third researcher (Qi Wu). The extracted data comprised: [1] study characteristics, such as author, region, year of publication, and treatment duration; [2] participant characteristics, including age, sex, sample size, diabetes duration, body mass index, hemoglobin A1c (HbA1c) and fasting blood glucose concentrations; [3] the number of participants with any GI neoplasm and the number/proportion with each reported site-specific GI neoplasm in each study arm; and [4] drug name and dose used in the experimental and control groups (Supplementary Appendix 4 for details). GI neoplasm endpoints were extracted as reported by the original trials and registries. Because most trials did not prespecify cancer as a primary endpoint and commonly captured neoplasms as investigator-reported adverse events, central adjudication was not assumed unless explicitly stated. If multiple publications from the same study population were identified, the report with the longest follow-up duration and most complete data was selected. Risk of bias in the included RCTs was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool, covering the randomization process, deviations from intended interventions, missing outcome data, outcome measurement, selection of the reported result, and other potential sources of bias. Each domain was judged as low risk, some concerns, or high risk, as appropriate (Supplementary eFigure 1). Two reviewers (Yu-Fan Zhang and Jia-Jie Xie) independently performed the assessments, with disagreements resolved through discussion or adjudication by a third reviewer (Qi Wu). Detailed risk-of-bias judgments for all included studies are provided in the supplementary material. Statistical analysis Statistical analyses were performed using Stata 17.0 (StataCorp LLC). Odds ratios (ORs) with corresponding 95% confidence intervals (CIs) were calculated to quantify associations. Event counts and denominators were extracted for each trial arm, and study-specific proportions were tabulated to contextualize the rarity of GI neoplasm outcomes. Heterogeneity was quantified with I². When I² < 50%, fixed-effects models were applied; otherwise, random-effects models were used. Two-sided p < 0.05 indicated statistical significance; p < 0.10 was considered suggestive in prespecified subgroup analyses. Subgroup definitions Prespecified subgroups and their cut points were chosen a priori on biological and methodological grounds. Exposure duration was categorized as ≤ 52, 52–104, and > 104 weeks, consistent with prior SGLT2 meta-analyses evaluating malignancy risk and intended to approximate short-, intermediate-, and extended exposure windows relevant to oncologic latency [15]. Glycemic control was stratified by baseline HbA1c ≤ 8.0% vs. > 8.0%, consistent with the 2024 ADA Standards of Care, which endorse less stringent glycemic goals (< 8.0%) for many adults and that are widely used to demarcate poorer control in trials [16]. Adiposity followed established adult BMI categories (overweight: 25 ≤ BMI < 30 kg/m²; obesity: BMI ≥ 30 kg/m²) [17, 18]. Dose was classified as FDA-labeled low (starting) vs. high (maximum approved) for each agent (e.g., empagliflozin 10/25 mg; dapagliflozin 5/10 mg; canagliflozin 100/300 mg; ertugliflozin 5/15 mg); for agents approved outside the U.S., we used the corresponding national labels (e.g., tofogliflozin 20 mg; luseogliflozin 2.5–5 mg). Results Eligible studies and characteristics A total of 2,767 citations were initially identified, and 355 full-text articles were assessed for eligibility. Based on the search strategy, 48 RCTs [19–66] encompassing 41 articles involving 48,765 participants met the eligibility criteria (Fig. 1). Of these, 28,513 participants were randomized to intervention groups receiving SGLT2 inhibitors, including ertugliflozin, bexagliflozin, dapagliflozin, empagliflozin, canagliflozin, tofogliflozin, and luseogliflozin. The included trials spanned multiple countries, with sample sizes ranging from 138 to 6,609 participants. Intervention durations ranged from 24 to 338 weeks. The mean baseline age of participants was 61.14 years (standard deviation [SD] 10.61), with 59.9% male. At baseline, the mean BMI was 30.78 kg/m² (SD 6.11) and HbA1c was 8.18% (SD 0.94). Table 1 summarizes the detailed participant characteristics of each included study, while Fig. 2 presents the number and proportion of participants with GI neoplasms. Methodological quality assessment of the included studies, including descriptions of randomization and double-blinding procedures, is provided in Supplementary Appendix 5. Although most trials were methodologically rigorous for their primary efficacy and safety outcomes, cancer ascertainment was not the primary objective in most studies. PRISMA flow diagram Table 1Baseline characteristics of randomized controlled clinical trials and participants includedStudyInclusion PeriodFollow - upRandomized treatmentsDuration of diabetes. (Mean, SD). yearsTotal casesCompleted casesNo. (%) of patients with GI neoplasmsAge (mean, SD). yearsNo. (%) of maleWeight (mean, SD). kgBMI (mean, SD). kg/m2FPG (mmol/l)HbA1c (mean, SD). %No. (%) of WhiteNo. (%) of Black or African AmericanNo. (%) of AsianNo. (%) of Others Strojek K 2011 Apr 2008 to Nov 200924 weeksPlacebo Dapagliflozin 2.5 mg Dapagliflozin 5 mg Dapagliflozin 10 mg7.4 (5.7) 7.7 (6.0)7.4 (5.7) 7.2 (5.5)145 154 142151133140132141 1 (0.8)01 (0.8)060.3 (10.2)59.9 (10.1)60.2 (9.7)58.9 (8.3)71 (49.0) 77 (50.0) 71 (50.0) 66 (43.7)NR29.7 (4.6)30.0 (5.1)29.8 (5.2)29.8 (5.6)9.6 (2.1)9.6 (2.1) 9.7 (2.1) 9.6 (2.0)8.2 (0.7) 8.1 (0.8) 8.1 (0.8) 8.1 (0.7)101 (69.6)108 (70.1)96 (67.6)106 (70.2)045 (31.0)46 (29.9)46 (32.4)45 (29.8)0Lavalle-González FJ 2013 Apr 2010 to Aug 201252 weeksPlaceboSitagliptin 100 mgCanagliflozin 100 mgCanagliflozin 300 mg6.8 (5.3) 6.8 (5.2) 6.7 (5.4) 7.1 (5.4)183 366368 367155 319322 3231 (0.7)01 (0.3)055.3 (9.8) 55.5 (9.6)55.5 (9.4)55.3 (9.2)94 (51.4)172 (47.0) 174 (47.3) 165 (45.0)86.6 (22.4)87.7 (21.6) 88.8 (22.2) 85.4 (20.9)31.1 (6.1)32.0 (6.1)32.4 (6.4) 31.4 (6.3)9.1 (2.1) 9.4 (2.3)9.3 (2.3) 9.6 (2.5)8.0 (0.9)7.9 (0.9)7.9 (0.9)7.9 (0.9)129 (70.5) 264 (72.1) 252 (68.5) 256 (69.8)3 (1.6) 13 (3.6) 16 (4.3) 13 (3.5)30 (16.4) 41 (11.2)51 (13.9) 60 (16.3)21 (11.5) 48 (13.1) 49 (13.3) 38 (10.4)Barnett AH 2014 Sept 3, 2010. and Jul 26, 201252 weeksPlacebo Empagliflozin 10 mgEmpagliflozin 25 mgNR31998 321278 88 2801 (0.4)01 (0.4)64.1 (8.7) 63.2 (8.5) 63.9 (9.0)181 (56.7)60 (61.2) 189 (58.9)83.4 (19.0)92.1 (21.4)84.1 (20.1)30.6 (5.5) 32.4 (5.4)30.4 (5.4)8.0 (2.7)8.1 (1.9)8.1 (2.1)8.1 (0.8)8.0 (0.8)8.0 (0.8)191 (59.9) 69 (70.4)185 (57.6)6 (1.9)3 (3.1)13 (4.0)119 (37.3) 25 (25.5)122 (38.0)3 (0.9)1 (1.0)1 (0.4)Kaku K 2014Oct 30, 2010 and Feb 28, 201224 weeksPlaceboTofogliflozin 10 mgTofogliflozin 20 mgTofogliflozin 40 mg6.0 (6.1) 6.3 (7.1) 6.4 (5.1) 6.7 (5.5)5657 58 58485456542 (4.2)01 (1.8)056.8 (9.9)58.6 (9.8) 56.6 (10.2)57.0 (9.1)37 (66.1) 38 (66.7) 39 (67.2) 39 (67.2)71.2 (12.6) 67.3 (12.7) 68.1 (15.8) 68.7 (11.9)26.0 (4.1) 25.1 (3.5) 25.0 (4.6) 25.8 (4.1)9.4 (1.4) 9.5 (1.8) 9.4 (1.6) 9.3 (2.1)8.4 (0.8) 8.5 (0.8) 8.3 (0.8) 8.3 (0.8)NRNRNRNRLeiter LA 2014 Mar 15, 2010 to May 30, 201124 weeksPlaceboDapagliflozin 10 mg QD.PO13.0 (8.4) 13.5 (8.2)480482441 4282 (0.5)063.6 (7.0) 63.9 (7.6)323 (67.0)321 (66.9)93.2 (16.8) 94.5 (17.8)32.7 (5.7) 33.0 (5.3)9.2 (2.5)9.0 (2.6)8.1 (0.8) 8.0 (0.8)454 (94.6) 449 (93.2)9 (1.9) 10 (2.1)6 (1.3) 7 (1.5)11 (2.3)16 (3.3)Ji L 2014Jun 2010 to Mar 201224 weeksPlaceboDapagliflozin 5 mgDapagliflozin 10 mg1.3 (2.0)1.2 (2.3) 1.7 (2.8)132 128 133113 113 11701 (0.9)049.9 (10.9) 53.0 (11.1)51.2 (9.9)87 (65.9) 84 (65.6) 86 (64.7)72.2 (13.2) 68.9 (11.4) 70.9 (11.6)25.9 (3.6) 25.2 (3.3) 25.8 (3.4)9.3 (2.4) 8.6 (1.8) 9.0 (2.4)8.4 (1.0) 8.1 (0.7) 8.3 (1.0)00132 (100.0) 128 (100.0) 133 (100.0)0Rosenstock J 2014Mar 2011 to Apr 201352 weeksPlacebo Empagliflozin 10 mg Empagliflozin 25 mgNR189 187 190161 159 16801 (0.6)055.3 (10.1) 56.7 (8.7)58.0 (9.4)75 (39.9)97 (52.2)84 (44.4)95.5 (17.5)96.7 (17.9)95.9 (17.3)34.7 (4.3)34.7 (3.8) 35.0 (4.0)8.4 (2.5)8.8 (2.7) 8.3 (2.7)8.3 (0.1)8.4 (0.1)8.3 (0.1)174 (92.0) 175 (93.6) 182 (95.8)8 (4.2) 7 (3.7) 4 (2.1)0006 (3.2) 4 (2.1) 3 (1.6)Nauck MA 2014 Mar 31, 2008 to Dec 15, 200952 weeksDapagliflozin (2.5. 5. or 10 mg) Glipizide (5. 10. or 20 mg)NR40640832231401 (0.3)58.1 (9.4) 58.6 (9.8)221 (55.3)220 (54.9)NR31.7 (5.1)31.2 (5.1)9.0 (2.1)9.1 (2.3)7.7 (0.9)7.7 (0.9)327 (81.8)323 (80.6)26 (6.5)24 (6.0)27 (6.8)34 (8.5)20 (5.0)20 (5.0) Ridderstråle M 2014Aug 2010 to Jun 2011104 weeksEmpagliflozin 25mgglimepiride 1–4 mgNR769 780610 5892 (0.3)5 (0.9)55.7 (10.4)56.2 (10.3)421 (54.0)432 (56.5)83.0 (19.2) 82.5 (19.2)30.3 (5.3) 30.0 (5.3)8.32 (1.9) 8.32 (1.7)7.92 (0.8)7.92 (0.8)519 (66.5) 498 (65.1)8 (1.0) 12 (1.6)253 (32.4) 254 (33.2)0 1 (<1%)Inagaki N 2014Jul 2011 to Aug 201224 weeksPlaceboCanagliflozin 100 mg Canagliflozin 200 mg5.6 (5.8) 4.7 (4.6) 5.9 (5.9)93 90 897484832 (2.7)0058.2 (11.0) 58.4 (10.4) 57.4 (11.1)60 (64.5) 59 (65.6) 72 (81.8)68.6 (15.2) 69.1 (14.5) 69.9 (14.2)25.9 (4.4) 25.6 (4.2) 25.4 (4.2)9.1 (1.8) 8.8 (2.0) 9.2 (1.9)8.0 (0.7)8.0 (0.7) 8.0 (0.8)0093 (100.0)90 (100.0) 88 (100.0)0Wilding JP 2014Apr 30, 2008 to May 19, 200924 weeksPlacebo Dapagliflozin2. 5mg Dapagliflozin5/10 mg Dapagliflozin10 mg13.5 (7.3) 13.6 (6.6) 13.1 (7.8) 14.2 (7.3)193202211194168180186178001 (0.5)2 (1.1)58.8 (8.6) 59.8 (7.6) 59.3 (7.9) 59.3 (8.8)98 (50.8) 102 (50.5) 111 (52.6) 107 (55.2)94.5 (19.8) 93.0 (16.7) 93.3 (17.4) 94.5 (16.8)33.1 (5.9) 33.0 (5.0) 33.0 (5.3) 33.4 (5.1)9.5 (3.2) 10.0 (3.3) 10.3 (3.3) 9.6 (3.0)8.5 (0.8)8.5 (0.8) 8.6 (0.9) 8.6 (0.8)186 (96.4)190 (94.1) 200 (94.8) 184 (94.8)6 (3.1) 3 (1.5) 5 (2.4) 5 (2.6)0 7 (3.5) 3 (1.4) 3 (1.5)1 (0.5) 2 (1.0) 3 (1.4) 2 (1.0)Bailey CJ 2015Jun 7, 2010 to May 23, 201326 weeksPlaceboCanagliflozin 100 mg Canagliflozin 300 mg11.4 (7.3)12.3 (7.8)11.3 (7.2)237 241 236158 184 1781 (0.6)0063.2 (6.2)64.3 (6.5)63.4 (6.0)143 (60.3) 124 (51.5) 129 (54.7)91.1 (17.5) 88.4 (15.6) 88.8 (17.1)31.8 (4.8) 31.4 (4.4)31.5 (4.6)8.7 (2.2)8.9 (2.2)8.5 (2.0)7.8 (0.8)7.8 (0.8)7.7 (0.8)185 (78.1) 194 (80.5) 173 (73.3)20 (8.4) 18 (7.5) 19 (8.1)21 (8.9) 15 (6.2) 25 (10.6)11 (4.6) 14 (5.8) 19 (8.1)J Rosenstock 2015Nov 2009 to May 201278 weeksPlaceboEmpagliflozin 10 mg Empagliflozin 25 mgNR170 169 155147 153 1291 (0.7)02 (1.6)58.1 (9.4) 58.6 (9.8) 59.9 (10.5)90 (53) 93 (55) 93 (60)90.5 (1.7) 91.6 (1.5) 94.7 (1.7)31.8 (6.0) 32.1 (5.8) 32.7 (5.9)7.9 (0.2) 7.7 (0.2) 8.1 (0.2)8.1 (0.1) 8.3 (0.1)8.3 (0.1)113 (66.4) 119 (70.4) 111 (71.6)21 (12.4) 12 (7.1) 15 (9.7)33 (19.4) 37 (21.9) 28 (18.1)3 (1.8) 1 (0.6) 1 (0.6)Roden M 2015Feb 2011 to May 201324 weeksPlaceboEmpagliflozin 10 mgEmpagliflozin 25 mgSitagliptin 100 mgNR228224 224223119147 14313601 (0.7)01 (0.7)54.9 (10.9)56.2 (11.6)53.8 (11.6)55.1 (9.9)123 (53.9) 142 (63.4) 145 (64.7) 141 (63.2)78.2 (19.9)78.4 (18.7) 77.8 (18.0) 79.3 (20.4)28.7 (6.2) 28.3 (5.5)28.2 (5.5)28.2 (5.2)8.6 (2.0)8.5 (1.8)8.5 (1.9) 8.2 (1.6)7.9 (0.8) 7.9 (0.9) 7.9 (0.9) 7.9 (0.8)76 (33.3) 77 (34.4) 73 (32.6) 76 (34.1)6 (2.6) 3 (1.3) 7 (3.1) 3 (1.3)146 (64.0) 143 (63.8) 144 (64.3) 143 (64.1)0 1 (0.4) 0 1 (0.4)Rosenstock Julio 2015Jul 2012 to Jan 201424 weeksSaxagliptin 5 mg and Dapagliflozin 10 mg plus MET Saxagliptin 5 mg and Placebo plus MET Dapagliflozin 10 mg and Placebo plus MET7.6 (5.3)8.2(5.5)7.4(5.4)179176179161 160 1691 (0.6)0053 (10)55 (10)54 (10)94 (53.4) 89 (49.7) 85 (47.5)87.1 (18.0)88.0 (18.7)86.3 (18.6)31.8 (4.8)31.8 (5.1)31.5 (5.3)10 (2.53)10.7 (2.5)10.3 (2.7)8.92 (1.18)9.03 (1.05)8.87 (1.16)120 (67.0)121 (68.7)131 (73.2)22 (12.3)22 (12.5)16 (8.9)12 (6.7)11 (6.3)10 (5.6)25 (14.0)22 (12.5)22 (12.3)Rosenstock J 2016May 16, 2013 to Dec 1,201426 weeksMetformin XR Canagliflozin 100 mgCanagliflozin 300 mg Canagliflozin 100 mg + Metformin XRCanagliflozin 300 mg + Metformin XR3.3 (4.5)3.5 (4.4)3.3 (4.4)2.9 (3.3)3.3 (3.9)2372372382372372052112162252120001 (0.4)055.2 (9.8)54.0 (10.7)55.8 (9.6)54.2 (9.7)55.4 (9.8)116 (48.9)105 (44.3)125 (52.5)108 (45.6)115 (48.5)92.1 (20.1)90.2 (18.6) 93.0 (19.9)88.3 (17.6) 91.4 (21.4)33.0 (6.0)32.4 (5.4) 32.6 (5.8)31.9 (5.3) 32.8 (6.5)10.6 (2.7)10.9 (3.0) 10.7 (2.9)10.6 (2.8) 11.2 (3.1)8.8 (1.2)8.8 (1.2)8.8 (1.2)8.8 (1.1)8.8 (1.2)192 (81.0)192 (81.0) 208 (87.4)189 (79.7) 187 (78.9)9 (3.8)12 (5.1)8 (3.4) 6 (2.5) 8 (3.4)9 (3.8)4 (1.7) 4 (1.7)6 (2.5) 5 (2.1)27 (11.4)29 (12.2) 18 (7.6)36 (15.2) 37 (15.6)Haneda M 2016May 2011 to October 201252 weeksPlaceboLuseogliflozin 2.5 or 5 mg12.6 (10.4) 10.4 (6.9)5095458301 (1.2)68.4 (8.9)67.9 (8.9)39 (78.0)72 (75.8)67.8 (11.8) 66.9 (13.6)25.8 (4.0) 25.5 (4.2)7.9 (1.5) 7.8 (1.7)7.7 (0.7) 7.7 (0.7)NRNRNRNRKadowaki T 2017Jan 2015 to Jan 201624 weeksPlacebo + Teneligliptin 20 mgCanagliflozin 100 mg + Teneligliptin 20 mg6.5 (3.9)8.3 (7.7)6870616701 (1.5)56.0 (9.5)58.4 (8.9)53 (77.9)54 (77.1)73.3 (12.9)71.3 (15.9)26.4 (3.9)25.5 (4.2)9.2 (1.9)9.7 (1.7)7.9 (0.8)8.2 (0.9)NRNRNRNRPratley RE 2017Apr 22,2014 toMay 26,201652 weeksErtugliflozin 5 mg Ertugliflozin 15 mgSitagliptin 100 mg Ertugliflozin 5 mg + Sitagliptin 100 mg Ertugliflozin 15 mg + Sitagliptin 100 mg7.1 (5.4) 7.3 (5.4)6.2 (5.2) 7.0 (5.6) 6.9 (5.2)25024824724324522621721922121800002 (0.9)55.1 (10.1) 55.3 (9.5) 54.8 (10.7) 55.2 (10.4) 55.1 (9.8)127 (50.8) 134 (54.0) 154 (62.3) 123 (50.6) 126 (51.6)88.6 (22.2) 88.0 (20.3) 89.8 (23.5) 89.5 (20.8) 87.5 (20.5)31.8 (6.2) 31.5 (5.8) 31.7 (6.5) 32.5 (6.7) 31.8 (6.5)10.23 (2.9) 9.97 (2.54) 9.69 (2.59)10.21(2.46) 9.84 (2.74)8.6 (1.0) 8.6 (1.0) 8.5 (1.0) 8.6 (1.0) 8.6 (1.0)7 (2.8) 4 (1.6) 4 (1.6) 2 (0.8) 4 (1.6)7 (2.8) 6 (2.4) 11 (4.5) 12 (4.9) 10 (4.1)22 (8.8) 22 (8.9) 29 (11.7) 22 (9.1) 36 (14.8)8 (3.2) 11 (4.4) 10 (3.6) 10 (4.1) 6 (2.5)Terra SG 2017Oct 9, 2013 to Jul 28, 201652 weeksPlacebo Ertugliflozin 5 mg Ertugliflozin 15 mg4.6 (4.5) 5.1 (5.1) 5.2 (5.6)153156 152121137 128001 (0.8)56.1 (10.9) 56.8 (11.4) 56.2 (10.8)82 (53.6) 89 (57.1) 90 (59.2)94.2 (25.2) 94.0 (25.4) 90.6 (18.3)33.3 (6.8) 33.2 (7.4) 32.5 (5.7)10.0 (2.5) 10.0 (2.7) 9.9 (2.7)8.1 (0.9) 8.2 (0.9) 8.4 (1.1)126 (82.4) 134 (85.9) 126 (82.9)9 (5.9)10 (6.4) 10 (6.6)15 (9.8) 10 (6.4) 14 (9.2)3 (2.0)2 (1.3)2 (1.3)CANVAS-R 2018Jan 16, 2014 to Feb 23,2017156 weeksPlaceboCanagliflozin 100 and 300 mgNR2905 29072866287210 (0.3)9 (0.3)64.0 (8.3) 63.9 (8.4)1794 (61.8)1854 (63.8)NRNRNRNR2372 (81.7)2393 (82.3)125 (4.3)106 (3.6)245 (8.4)244 (8.4)245 (5.7)164 (5.6)CANVAS 2018Dec 9, 2009 to Feb 22, 2017338 weeksPlaceboCanagliflozin 100 mg Canagliflozin 300 mgNR1442 1445144312971344144113 (1.0)25 (1.9)19 (1.3)62.3 (8.0) 62.2 (8.0) 62.8 (8.1)956 (66.3)961 (66.5) 944 (65.4)NRNRNRNR1064 (73.8)1060 (73.4)1055 (73.1)35 (2.4)32 (2.2)38 (2.6)262 (18.2)270 (18.7)263 (18.2)81 (5.6)83 (5.7)87 (8.2)CREDENCE 2018Feb 17, 2014 to Oct 30, 2018240 weeksPlacebo Canagliflozin 100 mgNR21992202217421877 (0.3)14 (0.6)63.2 (9.23)62.9 (9.17)1467 (66.7)1440 (65.4)NRNRNRNR1436 (65.3)1476 (67.0)112 (5.1)112 (5.1)452 (20.6)425 (19.3)199 (9.0)189 (8.6)Rosenstock J 2018Dec 13, 2013 to Aug 3, 201726 weeksPlacebo Ertugliflozin 5 mgErtugliflozin 15 mg8.0 (6.3)7.9 (6.1)8.1 (5.5)209 207 205175 187 1801 (0.6)0056.5 (8.7)56.6 (8.1)56.9 (9.4)98 (46.9) 97 (46.9) 93 (45.4)84.5(17.1)84.8(17.2)85.3(16.5)30.7 (4.7) 30.8 (4.8) 31.1 (4.5)9.4 (2.3) 9.3 (2.5) 9.3 (2.5)8.2 (0.9) 8.1 (0.9)8.1 (0.9)144 (68.9)134 (64.7)133 (64.9)19 (9.1) 22 (10.6)23 (11.2)31 (14.8) 34 (16.4)35 (17.1)15 (7.2)17 (8.2)14 (6.8)Scott R 2018 Oct 21, 2015 to Oct 10, 201724 weeksSitagliptin 100 mg Dapagliflozin 5-10 mg10.5 (7.0)10.7 (7.4)307306299 29601 (0.3)67.7 (8.5)66.6 (8.6)169 (55.0)186 (60.8)87.4 (20.2) 88.7 (18.0)31.8 (5.7)31.5 (5.3)9.0 (2.2)9.2 (2.3)7.7 (0.7)7.8 (0.7)240 (78.2) 234 (76.5)8 (2.6) 11 (3.6)11 (3.6) 7 (2.3)48 (15.7)54 (17.6)Yang W 2018Mar 31,2014 to Jan 28,201624 weeksPlacebo + InsulinDapagliflozin 10 mg QD.PO + Insulin12.2 (6.7)12.7 (7.2)1291391251351 (0.8)1 (0.7)58.6 (8.9)56.5 (8.4)64 (48.1) 66 (47.5)72.4 (13.1) 71.1 (12.0)26.7 (3.3)26.4 (3.8)9.3 (2.2) 10.0 (2.6)8.6 (0.8)8.5 (0.8)1 (0.8) 000132 (99.2) 137 (98.5)0 2 (1.4)Hollander P 2019Dec 16, 2013 to Apr 18, 2017104 weeksErtugliflozin 5 mgErtugliflozin 15 mgGlimepiride7.3 (5.7) 7.5 (5.7) 7.6 (5.6)4454354353393403271 (0.3)1 (0.3)058.7 (9.8) 58.0 (9.9) 57.9 (9.1)227 (51.0) 191 (43.9) 224 (51.5)88.0 (19.0)85.7 (19.1) 86.9 (20.8)31.7 (5.6) 31.3 (6.2) 31.2 (6.4)9.0 (1.9) 9.0 (2.0) 8.8 (1.9)7.8 (0.6) 7.8 (0.6) 7.8 (0.6)329 (73.9) 311 (71.5) 317 (72.9)17 (3.8) 19 (4.4) 25 (5.7)81 (18.2) 85 (19.5) 72 (16.6)18 (4.0) 20 (4.6) 21 (4.8)Rodbard HW 2019Aug 10, 2016 to Mar 8, 201826 weeksSemaglutide 14 mgEmpagliflozin 25 mg7.2 (5.8)7.7 (6.3) 41141040038701 (0.3)57.0 (10.0)58.0 (10.0)205 (49.9) 201 (49.0)91.9 (20.5)91.3 (20.1)32.9 (6.3) 32.8 (5.9)9.5 (2.3) 9.7 (2.5)8.1 (0.9) 8.1 (0.9)355 (86.4) 353 (86.1)26 (6.3) 33 (8.0)28 (6.8) 21 (5.1)2 (0.5)3 (0.7)Ji L 2019 Dec 16, 2015 to Dec 27, 201726 weeksPlaceboErtugliflozin 5 mgErtugliflozin 15 mg6.4 (5.1) 7.0 (5.0) 7.5 (5.1)167170169157163160001 (0.6)56.9 (9.0) 56.1 (9.0) 56.3 (9.3)88 (52.7) 95 (55.9)98 (58.0)70.1 (12.4) 71.4 (11.1) 69.5 (10.9)26.1 (3.4) 26.0 (2.8) 25.7 (3.2)9.2 (2.1) 9.4 (2.0) 9.3 (2.3)8.1 (1.0) 8.1 (0.9) 8.1 (0.9)000000167 (100.0)170 (100.0)169 (100.0)000Halvorsen YD 2019Oct 12, 2017 to Oct 31, 201824 weeksBexagliflozin 20 mg Sitagliptin 100 mg8.22 (5.7) 9.36 (5.7)19119318018901 (0.5)59.3 (9.7) 59.6 (9.8)120 (62.8) 126 (65.3)90.3 (20.7) 89.4 (19.0)32.1 (6.1) 31.4 (5.3)9.8 (2.3) 10.0 (2.6)7.9 (0.8) 8.0 (0.9)158 (82.7)156 (80.8)6 (3.1) 2 (1.0)27 (14.1) 35 (18.1)00Fuchigami A 2020July 2017 to June 201824 weeksDapaglifozin 5.0–10mg Sitagliptin 50–100 mg6.0 (6.4)5.6 (5.8)17017016816301 (0.6)58.3 (12.4)57.9 (12.1)104 (61.9)95 (58.3)74.5 (13.4)74.9 (15.0)27.8 (4.0) 27.9 (4.2)8.4 (1.86) 8.4 (1.71)7.8 (0.8) 7.8 (0.8)0000170 (100.0)170 (100.0)00Katakami N 2020Jan 12, 2016. and Nov 25, 2016.104 weeksTofoglifozin 20 mg Use drugs other than SGLT2 inhibitors12.1 (8.4)12.5 (8.3)16917014014704 (2.7)61.3 (9.3)60.9 (9.7)99 (58.6) 99 (58.2)NR27.0 (5.8)27.0 (4.6)7.8 (1.7)7.8 (1.8)7.4 (0.7)7.3 (0.7)NRNRNRNRJabbour SA 2020Sept 4, 2014 to Dec 28, 2017104 weeksExenatide 2 mg plus Dapagliflozin 10 mg exenatide 2 mg plus PlaceboDapagliflozin 10 mg7.6 (6.0)7.4 (5.5)7.1 (5.5)233 231 231155 154 13601 (0.6)1 (0.7)53.8 (9.8)54.2 (9.6)54.5 (9.2)102 (44.7)116 (51.1) 110 (47.8)91.8 (22.2)89.8 (20.2)91.1 (19.7)33.2 (6.8)32.0 (5.9)33.0 (6.1)11.0 (3.0)10.7 (2.8)10.6 (2.6)9.3 (1.1)9.3 (1.1)9.3 (1.0)190 (83.3)194 (85.5)189 (82.2)34 (14.9)27 (11.9)33 (14.3)3 (1.3)1 (0.4)1 (0.4)96 (42.1)96 (42.3)92 (42.0)Allegretti AS 2021Sept 23, 2016 to Jan 11, 201824 weeksPlaceboBexagliflozin 20 mgNR1551571441523 (2.7)1(0.7)69.9 (8.3)69.3 (8.4)104 (67.1)92 (58.6)NRNRNRNR88 (56.8)83 (52.9)6 (3.9)9 (5.7)59 (38.1)61 (38.9)2 (1.3)4 (2.5)BEST 2021Oct 12, 2015 to Oct 23, 201952 weeksPlaceboBexagliflozin 20 mg15.2 (9.3)14.9 (8.6)56711334829873 (0.8)3 (0.3)64.6 (8.0)64.4 (8.0)390 (68.8)792 (69.9)92.6 (20.0)94.6 (21.9)32.2 (5.8)32.8 (6.1)NR8.3 (1.0)8.3 (0.9)430 (75.8)879 (77.6)28 (4.9)46 (4.1)55 (9.7)108 (9.5)54 (9.5)100 (8.8)SOTA-BONE 2021 Feb 19, 2018 to May 30, 20202 weeksPlacebo Sotagliflozin 200 mg Sotagliflozin 400 mgNR126 125 1259 9 9002 (1.6)66.3 (5.7) 66.1 (6.7) 66.5 (6.9)70 (55.6) 69 (55.2) 70 (56.0)NRNRNR8.4 (0.9) 8.3 (0.9) 8.3 (0.9)88 (69.8)87 (69.6)88 (70.4)8 (6.3)8 (6.4)6 (4.8)26 (20.6)25 (20.0)24 (19.2)4 (3.1)5 (4.0)7 (5.6)NCT02769481 2021Aug 15, 2016 to Jun 14, 2019NRBexagliflozin 20 mg Glimepiride 2, 4 or 6 mgNR21321318017701 (0.6)59.5 (9.1) 59.7 (10.4)118 (55.4)130 (61.0)88.0 (19.1)90.2 (17.6)31.5 (4.9)32.2 (5.2)NRNR198 (93.0)204 (95.8)5 (2.3)4 (1.9)9 (4.2)4 (1.9)1 (0.5)1 (0.5)NCT02926950 2021Nov 1, 2016 to Mar 22, 20194 weeksPlacebo + Metformin Sotagliflozin 400 mg + MetforminNR259 259210 21101 (0.5)59.9 (9.4) 60.0 (10.1)146 (56.4) 142 (54.8)NRNRNR8.2 (0.8) 8.2 (0.8)197 (76.1)223 (86.1)40 (15.4)28 (10.8)19 (7.3)6 (2.3)3 (1.2)2 (0.8)NCT03066830 2021Feb 24, 2017 to Apr 30, 20194 weeksPlacebo Sotagliflozin 400 mgNR254 253221 2311 (0.5)1 (0.4)63.0 (9.9) 63.3 (8.8)130 (51.2) 149 (58.9)NRNRNR8.18 (0.83) 8.20 (0.83)211 (83.1)215 (85.0)8 (3.1)8 (3.2)34 (13.4)28 (11.1)1 (0.4)2 (0.8)SOTA-GLIM 2021Dec 1, 2017 to Sept 6, 20192 weeksPlacebo Sotagliflozin 400 mg Sotagliflozin 200 mg GlimepirideNR159 317 160 318125 266 128 2701 (0.8)00058.8 (11.2) 59.7 (10.4) 58.6 (9.9) 59.8 (9.6)82 (51.6)160 (50.5)85 (53.1)175 (55.0)NRNRNR8.1 (0.7) 8.1 (0.8)8.1 (0.8)8.1 (0.8)141 (88.7)269 (84.9)136 (85.0)277 (87.1)13 (8.2)34 (10.7)19 (11.9)36 (11.3)3 (1.9)5 (1.6)3 (1.9)4 (1.3)2 (1.2)9 (2.8)2 (1.3)1 (0.3)SOTA-EMPA 2021Nov 27, 2017 to May 16, 2019NRPlaceboSotagliflozin 400 mg Empagliflozin 25 mgNR154307 309147293 292001 (0.3)59.8 (9.6)58.9 (9.7) 59.7 (9.6)97 (51.3)166 (54.1)151 (48.9)NRNRNR8.2 (0.8) 8.2 (0.9)8.2 (0.9)137 (89.0)256 (83.4)257 (83.2)5 (3.2)13 (4.2)13 (4.2)4 (2.6)17 (5.5)11 (3.6)8 (5.2)21 (6.9)28 (9.0)SOTA-INS 2021Sept 15, 2017 to Sept 17, 20192 weeksPlacebo Sotagliflozin 200 mg Sotagliflozin 400 mgNR144 141 286129 130 24901 (0.8)062.2 (8.9) 62.1 (10.2)62.7 (9.1)86 (59.7)76 (53.9)154 (53.8)NRNRNR8.8 (0.8)8.8 (0.8)8.7 (0.8)117 (81.3)124 (87.9)234 (81.8)10 (6.9)10 (7.1)27 (9.4)9 (6.3)1 (0.7)6 (2.1)8 (5.5)6 (4.3)19 (6.6)Cherney DZI 2021Aug 16, 2017 to Dec 11, 20194 weeksPlaceboSotagliflozin 200 mgSotagliflozin 400 mg20.7 (8.9)19.6 (9.3)18.5 (9.2)93929275 78 7801 (1.3)068.0 (8.3) 66.8 (10.0) 67.3 (9.6)44 (47.8) 49 (53.3) 35 (48.7)NR31.7 (5.7) 30.9 (5.4) 32.1 (6.2)8.7 (3.2) 8.8 (3.3) 8.2 (2.6)8.4 (1.1) 8.3 (1.0) 8.3 (0.9)76 (81.7) 73 (79.3) 78 (84.8)6 (6.5) 7 (7.6) 3 (3.3)1 (1.1) 2 (2.2) 2 (2.2)51 (58.4)43 (46.8)42 (45.7)Wada T 2022Feb 15, 2018 to Jan 21, 2021104 weeksCanagliflozin 100 mg Placebo 15.4 (8.6)16.5 (9.0)1541541241275 (4.0)5 (3.9)62.5 (10.5)62.4 (11.1)115 (74.7)129 (83.8)NR26.7 (4.4) 27.1 (4.5)NR7.8 (1.1)7.8 (1.1)00154 (100.0) 154 (100.0)0Takeshita Y 2022Mar 2016 to Dec 201948 weeksTofogliflozin 20 mgGlimepiride 0.8 mgNR202020201 (5.0)059.0 (16.1)50.5 (19.8)7 (35,0)14 (70.0)79.3 (18.2)84.7 (25.4)31.0 (6.7)32.0 (8.8)8.0 (2.3)7.8 (2.0)7.9 (1.1)8.2 (2.2)0020 (100.0)20 (100.0)0Cherney DZI 2023Aug 16, 2017 to Oct 25, 201952 weeksPlacebo Sotagliflozin 200 mg Sotagliflozin 400 mg17.7 (9.5)17.2 (9.3)16.3 (8.2)260 263 264224 240 232001 (0.4)69.3 (8.1) 69.6 (7.5) 69.5 (8.2)149 (57.3)143 (54.4)152 (57.6)NR32.5 (5.2) 32.3 (5.7) 32.4 (5.2)9.2 (3.3) 9.1 (2.7)9.0 (2.9)8.3 (1.0) 8.3 (0.9) 8.3 (0.9)220 (84.6) 231 (87.8) 215 (81.4)12 (4.6) 14 (5.3) 15 (5.7)5 (1.9) 7 (2.7) 8 (3.0)23 (8.8)11 (4.2)26 (9.8)Herrington WG 2023Jan 31, 2019 to Jul 2, 2024162 weeksPlaceboEmpagliflozin 10 mgNR3305 33042457 254913 (0.5)16 (0.6)63.8 (13.9)63.9 (13.9)2207 (66.8)2210 (66.9)NR29.8 (6.8)29.7 (6.7)NRNR1920 (58.1)1939 (58.7)134 (4.1)128 (3.9)1199 (36.3)1194 (36.1)52 (1.6)43 (1.3)Shimoda M 2024Apr 1, 2019 to Dec 31,202224 weeksLuseogliflozin 2.5 mg Teneligliptin 20 mg10.1 (7.9) 9.2 (7.6)4954445201 (1.9)60.8 (11.1)62.6 (11.2)26 (53.1)31 (57.4)69.9 (12.3) 71.9 (19.1)27.0 (4.2) 27.2 (5.4)8.3 (1.3)8.2 (1.0)7.6 (0.4) 7.5 (0.5)0049 (100.0)54 (100.0)0 Baseline characteristics of randomized controlled clinical trials and participants included Risk of GI neoplasms among patients with T2DM treated with SGLT2 inhibitors GI neoplasm risk and SGLT2 inhibitors Control groups received either placebo or alternative antihyperglycemic agents. GI neoplasm events were rare, and pooled results did not show a statistically significant elevation in GI neoplasm risk among SGLT2 inhibitor users compared with non-SGLT2 treatments (OR = 1.10, 95% CI 0.85–1.43; p = 0.46; Fig. 2). Site-specific outcomes Site-specific analyses revealed no statistically significant association between SGLT2 inhibitor use and any individual GI neoplasm outcome. The odds ratios with 95% confidence intervals were as follows: esophageal (OR = 1.12, 95% CI 0.37–3.45), gastric (1.20, 0.65–2.23), hepatic (0.62, 0.31–1.22), pancreatic (0.91, 0.51–1.64), colonic (1.28, 0.78–2.08), colorectal (0.76, 0.27–2.17), and rectal neoplasms (0.98, 0.49–1.97) (all p > 0.05; Supplementary eFigure 2). In all cases, the confidence intervals crossed the null value. Because site-specific event counts were low and several intervals were wide, these analyses should be interpreted as showing no clear signal rather than definitively excluding modest site-specific effects. Drug-specific analyses Subgroup analyses by individual SGLT2 inhibitor identified no statistically significant association with gastrointestinal neoplasm risk for any agent. The odds ratios (95% CIs) were as follows: canagliflozin, 1.02 (0.68–1.52); dapagliflozin, 1.27 (0.53–3.03); empagliflozin, 1.23 (0.70–2.19); ertugliflozin, 1.32 (0.39–4.46); tofogliflozin, 1.10 (0.32–3.72); and bexagliflozin, 0.85 (0.27–2.62) (all p > 0.05; Supplementary eFigure 3). These findings do not demonstrate a distinct drug-specific signal within the available trial data, but they should be interpreted cautiously because several drug-specific estimates were based on few events. Clinical subgroups Subgroup analyses were conducted according to baseline age, BMI, treatment duration, baseline HbA1c, and drug dosage. The pooled odds ratios with 95% confidence intervals across these subgroups were generally centered around the null value, with no reproducible effect modification observed (Supplementary eFigs. 4, 5, 6, 7, 8, 9 and 10). Across prespecified subgroups (agent, dose, exposure duration, age, BMI, baseline HbA1c), pooled ORs clustered around unity, and interaction p-values were non-significant, indicating no reproducible effect modification (Fig. 3). Event counts remained low in several strata, so the absence of statistically significant interactions should not be interpreted as proof of equivalence across all patient groups. Future studies with longer follow-up and a greater number of endpoint events will be required to detect potential subtle effect modifiers. Subgroup analyses of gastrointestinal neoplasm risk with interaction p-values Publication bias Egger’s test showed no evidence of publication bias (p = 0.15; t = 1.46). The funnel plot was largely symmetrical, indicating little evidence of publication bias (Supplementary eFigures 11–28). Discussion This meta-analysis of 48 RCTs involving 48,765 patients with T2DM found no clear evidence that SGLT2 inhibitor therapy increases the risk of gastrointestinal neoplasms (OR 1.10, 95% CI 0.85–1.43) over the follow-up durations captured in the available trials. The overall estimate and most subgroup estimates were centered near the null. Dose-stratified analyses did not suggest a discernible dose–response pattern within approved dosing ranges. These findings are reassuring for currently recommended dosing, but they should not be interpreted as definitive proof of long-term oncologic safety because many trials were not designed to assess cancer outcomes and the number of site-specific events was limited. Subgroups were prespecified on biological and methodological grounds—BMI (adiposity-related inflammation); HbA1c (glycemic/hyperinsulinemic burden); exposure duration (latency; ≤52, 52–104, > 104 weeks); dose (FDA-labeled low vs. high; potential dose–response); and agent (drug-specific selectivity or off-target effects). Across strata, pooled ORs generally centered on 1.0 with 95% CIs spanning unity, indicating no reproducible effect modification within the limits of the available data. Mechanistically, SGLT2 inhibitors are proposed to engage several anticancer pathways, such as reducing hyperinsulinemia, attenuating systemic inflammation, and limiting glucose availability for tumors through urinary glucose excretion [67]. Indeed, improvements in insulin sensitivity have been documented in meta-analyses, with more consistent effects observed for agents like dapagliflozin [68]. Preclinical models further suggest that certain SGLT2 inhibitors may directly inhibit cancer cell glucose uptake [69]. However, despite these plausible mechanisms, our synthesis of RCT data—under current clinical dosing and follow-up durations—did not show a measurable reduction in GI cancer incidence. This disconnect likely stems from several factors: (i) the drug concentrations required for direct antitumor effects in vitro often exceed plasma levels achieved with standard human dosing [70]; (ii) the complex, multifactorial nature of human carcinogenesis—shaped by immune surveillance, the tumor microenvironment, and long-term adaptation—is not fully captured by animal or cell models [71]; and (iii) the insufficient duration of RCTs, which may preclude the detection of slow, latency-dependent effects. Thus, while preclinical data outline plausible pathways, they have not translated into a clinically meaningful reduction in GI cancer risk, underscoring the need for longer, cancer-endpoint trials. The neutral findings from our RCT-based meta-analysis contrast with results from several large real-world observational studies suggesting lower risks for certain GI neoplasms with SGLT2 inhibitors. In a population-based study in Gastric Cancer, initiation of SGLT2 inhibitors was associated with a substantially reduced incidence of gastric cancer versus DPP-4 inhibitors (adjusted HR = 0.30) [13]. A cohort from China’s Taiwan region likewise reported a lower overall cancer risk (HR 0.90, 95% CI 0.87–0.93), with favorable estimates for GI neoplasms (esophageal HR 0.57; colorectal/liver HR 0.90) [72]. However, other real-world evidence—including quantitative overviews of reviews—has not shown a class-wide reduction in total or most site-specific cancers, and several population-based analyses report null associations after extensive adjustment. Taken together, observational results are heterogeneous: some signals suggest site-specific benefit, while others are neutral. Given the inherent limitations of nonrandomized designs (residual confounding by indication, immortal-time bias, differential screening, and unmeasured lifestyle factors), our RCT synthesis—though limited by relatively short follow-up—provides the highest internal validity and does not support a clinically meaningful reduction in GI cancer incidence within current trial durations. Longer RCTs with cancer endpoints, or rigorously specified target-trial emulations, are needed to verify or refute these real-world signals [73]. From a clinical safety perspective, our findings provide reassurance that a material increase in GI neoplasm risk was not observed with SGLT2 inhibitor use in RCTs. It is well established that SGLT2 inhibitors are associated with an increased risk of genital mycotic infections, though these are typically manageable [74]. Other risks, such as volume depletion and rare diabetic ketoacidosis events, are also recognized and can be mitigated with appropriate patient selection and monitoring [75, 76]. Large outcome trials have consistently demonstrated significant cardiorenal benefits that, for many patients with T2DM, outweigh these known risks [77]. Nevertheless, because cancer endpoints were generally not primary or centrally adjudicated outcomes, the present results should be viewed as supportive short- to intermediate-term evidence rather than a definitive assessment of long-term cancer safety. Our findings are broadly consistent with recent systematic reviews and meta-analyses reporting no material increase in overall or digestive-system neoplasms with SGLT2 inhibitors. Wang et al., for example, found no significant association between SGLT2 inhibitor use and overall neoplasm risk in a meta-analysis of 53 RCTs involving 126,232 participants, whereas Obaid et al. reported a neutral association for digestive-system malignancies [78, 79]. However, these studies primarily evaluated total cancer risk or broad organ-system categories, with gastrointestinal neoplasms treated as one of several secondary outcomes. Similarly, Xu et al. focused on cancer risk in randomized active-controlled trials, while Pelletier et al. summarized evidence at the review level and identified substantial variation in the scope, methodology, endpoint definitions, and quality of previous quantitative reviews [80, 81]. In contrast, the present study addresses a narrower and more anatomically resolved clinical question by focusing specifically on gastrointestinal neoplasms in patients with type 2 diabetes. Our analysis was restricted to randomized controlled trials and evaluated esophageal, gastric, hepatic, pancreatic, colonic, colorectal, rectal, and, where data were available, biliary tract neoplasms separately, thereby reducing the possibility that associations differing across anatomical sites would be obscured within a single digestive-system composite. We also extended previous evidence by searching trial publications, supplementary materials, and registry records for rarely reported neoplasm events and by conducting prespecified subgroup analyses according to individual SGLT2 inhibitor, FDA-labeled dose, exposure duration (≤ 52, 52–104, and > 104 weeks), age, BMI, and HbA1c, with interaction tests where appropriate. Thus, the principal contribution of our study is not a larger all-cancer evidence base, but a more focused, anatomically detailed, and clinically stratified synthesis of randomized evidence concerning gastrointestinal neoplasms. Nevertheless, the limited follow-up, sparse site-specific events, and predominantly non-adjudicated cancer ascertainment in the underlying trials preclude definitive conclusions regarding long-term oncologic safety. This study has limitations. First, although randomization is intended to balance confounding, cancer was captured as a secondary or adverse-event outcome in most included RCTs; key baseline oncogenic risk factors (e.g., smoking, alcohol use, family history, prior screening) were variably measured and inconsistently reported across groups, leaving the possibility of residual imbalance in cancer-related risk factors despite randomization. Second, the evaluation of long-term carcinogenicity is constrained by short follow-up: nearly half of the trials (46%) followed participants for ≤ 1 year, which may be insufficient to detect latency-dependent or cumulative exposure effects on GI neoplasms. Third, GI neoplasms were commonly identified from investigator-reported adverse events, publications, supplementary materials, or trial registries rather than from uniform, centrally adjudicated cancer surveillance. This may introduce outcome misclassification, underascertainment of asymptomatic or incompletely investigated neoplasms, and incomplete information on disease severity, histology, stage, or screening context. Fourth, site-specific analyses and several subgroup comparisons were limited by low event counts, yielding wide confidence intervals and limited power to detect small but clinically relevant effects. Finally, to maintain currency, we included recent RCTs and registered clinical studies with results in the meta-analysis; while this enhances timeliness, some estimates may be based on immature follow-up and require confirmation with longer observation. These considerations temper the interpretation of the neutral findings and underscore the need for longer RCTs with prespecified cancer endpoints, standardized cancer ascertainment, and systematic collection of baseline cancer risk factors. In conclusion, SGLT2 inhibitor treatment in adults with T2DM was not associated with a clear increase in gastrointestinal neoplasm incidence over the durations captured by the available randomized trials, with generally consistent results across agents and anatomical subsites. These findings are reassuring, but they do not establish definitive long-term oncologic safety because cancer events were rare, many trials had limited follow-up, and most outcomes were not cancer-specific primary endpoints. Clarification of long-term risk should come from extended surveillance cohorts, target-trial emulations with careful control of bias, and dedicated trials or post-trial follow-up studies with standardized cancer ascertainment. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements None. Abbreviations Sodium-glucose cotransporter 2 Glucagon-like peptide-1 Dipeptidyl peptidase-4 Gastrointestinal Odds ratios Confidence intervals Hazard Ratio I-squared statistic Standard deviation Risk of Bias 2 tool Randomized controlled trials Type 2 diabetes mellitus Body mass index Glycated hemoglobin Cochrane Central Register of Controlled Trials Author contributions Yu-Fan Zhang : investigation, software, data curation, writing-original draft, writing-review & editing; Jia-Jie Xie : data curation, methodology, software; Qi Wu : investigation, methodology; Xiao-Yu Zhou : methodology; Chen Zeng : software; Yan Zeng : validation; Yi-Meng He : software; Man Guo : conceptualization, resources, funding acquisition, supervision, writing-review & editing; Fang-Yuan Teng : funding acquisition, resources, writing-review & editing; Yong Xu : funding acquisition, supervision, resources, writing-review & editing. Funding This work was supported by the National Natural Science Foundation of China (grant numbers: U22A20286, 82470854), Noncommunicable Chronic Diseases-National Scienceand Technology Major Project (grant number: 2024ZD0531300), the project of the Sichuan Science and Technology Department (grant number: 2026NSFSCZY0052), the Science and Technology Strategic Cooperation Programs of Luzhou Municipal People’s Government and Southwest Medical University (grant number: 2025LZXNYDPD04), the Scientific research project of Southwest Medical University (grant numbers: 2024LCYXZX02, 2025LCYXZX01). Data availability The authors confirm that the data supporting the findings of this study are available within the article and the supplementary material. Declarations Competing interests. The authors declare no competing interests. Ethics approval and consent to participate Not applicable. Consent for publication All the authors have provided their consent for publication of the manuscript. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Yu-Fan Zhang, Jia-Jie Xie and Qi Wu contributed equally to this work. Contributor Information Man Guo, Email: guoman613@swmu.edu.cn. Fang-Yuan Teng, Email: tengfangyuan383@swmu.edu.cn. Yong Xu, Email: xywyll@swmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The authors confirm that the data supporting the findings of this study are available within the article and the supplementary material.