Among 575,139 newly diagnosed Type 2 diabetes patients matched against 689,719 non-diabetic adults in the Epic Cosmos EHR platform (2012–2025), cancer risk differed dramatically by T2D subtype. The severe insulin-deficient subtype (SIDD) carried the steepest overall hazard (HR=3.87 for combined cancers), while the mild obesity-related subtype (MOD) showed paradoxically lower breast and prostate cancer rates — MOD patients had a 40% reduced prostate cancer hazard (HR=0.60) compared to non-diabetics. Colorectal, pancreatic, liver, endometrial, and ovarian cancers were elevated across all subtypes, yet cancer screening uptake was lower in every T2D group.
This large-scale analysis challenges the conventional practice of treating Type 2 diabetes as a monolithic cancer risk factor. The field has long known that diabetes elevates risk for certain malignancies — likely through hyperinsulinemia, chronic inflammation, and IGF-1 signaling — but subtype-stratified data at this scale is genuinely novel. The inverse associations for prostate and, in MOD patients, breast cancer deserve scrutiny: lower androgen levels in obese metabolic profiles may partly explain the prostate finding, though confounding by PSA screening disparities (confirmed here) complicates interpretation. Critical limitations include the retrospective EHR design, median follow-up of only 3.8 years (too short to capture many solid tumors), and potential subtype misclassification. As a preprint not yet peer-reviewed, these hazard estimates should be treated cautiously. If validated, subtype-informed cancer surveillance protocols could meaningfully shift preventive oncology practice.