Type-2 Diabetes, medications, and risk of multiple colorectal polyps: A colonoscopy-based study using natural language processing

Introduction : Colorectal polyps are known precursors to colorectal cancer (CRC), and a higher polyp count is associated with an increased CRC risk. This study aimed to investigate the impact of type-2 diabetes (T2D) and its treatments on polyp count. Methods: We leveraged pathology reports from the University of Utah (UofU) Enterprise Data Warehouse (EDW) to develop a rule-based natural language processing pipeline, to extract polyp diagnoses and features (site, count) for 38,038 patients who underwent colonoscopy at the UofU Gastroenterology clinic from 2011-2020. We identified 6,556 patients with T2D via ICD codes and anti-diabetes medication prescriptions. Patient characteristics were extracted from the EDW, including age, sex, race, smoking, BMI, anti-inflammatory medication use, and active prescriptions for insulin, metformin, sulfonylureas, GLP-1 agonists, and DPP-4 inhibitors. Polyp count was categorized as none, one, or multiple. Adjusted odds ratios (OR) and 95% confidence intervals (CI) for polyp counts were calculated using multinomial logistic regression. Results: Patients were on average 56 years old, 84% White, 51% female, with a mean BMI of 30 kg/m 2 . 73% of patients with T2D used anti-diabetes medications. T2D alone was not associated with multiple polyps, but patients with T2D taking anti-diabetes medications had a lower risk of polyps overall, irrespective of polyp count. Insulin use was associated with an increased risk of one or multiple polyps [OR(95%CI)=1.29(1.08-1.54) and 1.40(1.18-1.66), respectively], compared to other medications. Metformin and GLP-1 agonists were both associated with a decreased risk of one [Metformin OR(95% CI)=0.71(0.59-0.85), GLP-1=0.79(0.64-0.99)] or multiple polyps [Metformin OR(95% CI)=0.71(0.52-0.73), GLP-1=0.63(0.51-0.77)], compared to patients taking medications other than metformin or GLP-1 agonists, respectively. Compared to patients taking no anti-diabetes medications, those on insulin, metformin, DPP-4 inhibitors, or sulfonylureas had an increased risk of at least one polyp, while patients taking GLP-1 agonists had a lower risk of multiple polyps [OR(95%CI)=0.81(0.63-1.03)]. DPP-4 inhibitors and sulfonylureas also had a lower risk of multiple polyps [OR(95%CI)=0.79(0.64-0.97) and 0.80(0.68, 0.96), respectively]. Conclusion: Overall, T2D alone was not associated with polyp count, but medication use modified this relationship. Insulin use was associated with increased risk, whereas the use of metformin, GLP-1 agonists, DPP-4 inhibitors, and sulfonylureas was associated with decreased risk of one or multiple polyps, compared to patients taking other anti-diabetes medications. These results highlight the potential role of T2D medication choice in altering colorectal polyp risk, underscoring the need for more research into underlying mechanisms to guide potential interventions.