Diabetes Duration, Severity, and the Risk of Gastrointestinal Bleeding: What the Gut Tells Us about Glucose

Gastrointestinal bleeding (GIB) remains a frequent medical emergency with in-hospital mortality of up to 10%, posing an enduring global burden.¹ Beyond classic risk factors such as advanced age, nonsteroidal anti-inflammatory drugs (NSAIDs), antithrombotics, chronic kidney disease, and cirrhosis, several studies have linked type 2 diabetes mellitus (T2DM) to GIB in recent decades.^2-5 T2DM was first identified as an independent risk factor for peptic ulcer bleeding (hazard ratio [HR], 1.44; 95% confidence interval [CI], 1.11 to 1.86) in a nationwide Taiwan database.² A subsequent analysis of aspirin users revealed that diabetes conferred a significant excess risk for major bleeding, including GIB (incidence rate ratio, 1.15; 95% CI, 1.03 to 1.29), acting additively with ulcerogenic medications.³ The UK Biobank cohort further confirmed this independent association with incident peptic ulcer disease (HR, 1.56; 95% CI, 1.43 to 1.71) over a 12-year follow-up,⁴ while an umbrella review of meta-analyses solidified T2DM as a consistent environmental risk factor.⁵ While these studies clearly establish an association between T2DM and GIB, they largely treated diabetes as a background cofactor rather than the primary variable of interest.^2,3

In this issue of Gut and Liver, Park et al.⁶ address this gap using the Korean National Health Insurance Service database. Among 522,399 participants who underwent a national health screening program and followed over a decade, they evaluated GIB risk across a spectrum of glycemic states: normal glucose, impaired fasting glucose, new-onset diabetes, diabetes duration <5 years, and diabetes duration ≥5 years. Using a stringent composite outcome, they demonstrated a clear duration- and severity-dependent gradient for upper GIB. Compared to normoglycemic individuals, adjusted HRs rose steadily: 1.08 for impaired fasting glucose, 1.27 for new-onset T2DM, 1.56 for T2DM <5 years, and 1.74 for T2DM ≥5 years. Furthermore, insulin use was associated with a 4- to 5-fold increased risk across all subgroups. By contrast, no statistically significant association was identified between diabetes status and lower GIB in any duration category. This measurable, dose-dependent gradient strengthens causal inference. Crucially, this risk was most pronounced in younger individuals without typical vascular comorbidities (e.g., hypertension, dyslipidemia). This suggests that T2DM itself is a dominant driver of upper GIB before age-related vascular burden accumulates, highlighting the critical need for early preventive vigilance.

The biological rationale for these findings is robust. Prolonged hyperglycemia promotes microvascular injury via endothelial dysfunction, oxidative stress, and advanced glycation end-products, severely impairing tissue healing. Consequently, gastrointestinal mucosa, which is constantly exposed to mechanical and chemical insults, fails to repair ordinary microinjuries, allowing erosions and ulcers to progress to frank hemorrhage. In streptozotocin-induced diabetic rat models, sustained hyperglycemia produces progressive gastric mucosal lesions that worsen in a time-dependent fashion.⁷ Clinically, the CORONOR registry of 4,184 outpatients demonstrated that diabetic patients faced a significantly higher major bleeding risk (primarily GIB) regardless of whether their glycated hemoglobin (HbA1c) was below or above 7% (HR: 1.66 and 1.75, respectively, versus non-diabetic individuals).⁸ This indicates that the diabetic state itself, not just the degree of hyperglycemia, contributes to hemorrhagic vulnerability. Furthermore, longitudinal endoscopic cohorts link higher HbA1c with an increased incidence of acid-related upper gastrointestinal disorders,⁹ suggesting that glycemic burden operates through multiple converging pathways to increase upper GIB risk.

However, the optimal preventive strategy remains unclear. In the present study, >30 days of proton pump inhibitor (PPI) use did not significantly reduce upper GIB risk among diabetic patients. While this diverges from the established paradigm of PPI prophylaxis for high-risk NSAID or antithrombotic users, this null finding may be partly attributable to confounding by indication and the exclusion of patients with prior GIB. Rather than contradicting current guidelines, these findings underscore the heterogeneity of upper GIB etiologies in patients with diabetes and the need for tailored gastroprotection. Notably, the present analysis relates diabetes burden to a claims-defined clinically significant upper GIB outcome, rather than to lesion-specific endoscopic diagnoses. Thus, although an acid-related or ulcerative mucosal pathway is plausible, the present data cannot determine which lesions actually accounted for the observed excess risk. The neutral association with PPI use should therefore be interpreted cautiously, as it may reflect mismatch between a lesion-directed preventive strategy and a heterogeneous bleeding endpoint, rather than absence of benefit from gastroprotection in appropriately selected patients. Furthermore, given that long-term PPI use in T2DM is associated with increased cardiovascular events and mortality,¹⁰ these findings do not justify routine prophylaxis and instead support individualized risk assessment, with regular reassessment in accordance with deprescribing guidelines.

The etiologies driving lower GIB differ fundamentally from those underlying upper GIB, primarily comprising diverticular bleeding and angiodysplasia. The underlying pathophysiological mechanisms of these conditions are distinct; for instance, angiodysplasia is heavily associated with aortic stenosis and chronic kidney disease. This mechanistic divergence plausibly explains the different association observed in lower GIB. Whether T2DM modulates the risk of other etiologies sharing microvascular pathways, such as ischemic colitis, remains an open question.

Looking ahead, several research priorities exist. First, while experimental models indicate that insulin-mediated glucose normalization restores mucosal healing,⁷ direct clinical evidence confirming that tight glycemic control lowers GIB incidence remains lacking. This strongly argues for prospective trials examining whether intensive glycemic management prevents incident or recurrent upper GIB, especially in high-risk populations. Second, the effect of modern antidiabetic pharmacotherapies, such as glucagon-like peptide-1 receptor agonists and sodium-glucose cotransporter 2 inhibitors on GIB risk remains poorly understood, given their distinct gastrointestinal profiles. Finally, future studies incorporating continuous glucose monitoring and serial HbA1c metrics will be important to clarify whether glycemic variability independently predicts GIB risk beyond mean glucose exposure.

Ultimately, convergent clinical and experimental evidence supports upper GIB as a clinically relevant complication of diabetes. This risk escalates in a dose-dependent manner with disease chronicity and glycemic burden, even after adjustment for the medications and comorbidities that typically accompany longstanding T2DM. For clinicians, these findings mandate heightened vigilance for upper GIB in patients with prolonged or poorly controlled diabetes, particularly when co-prescribing NSAIDs or antithrombotic agents. Most fundamentally, these findings reinforce a critical truth: effective, sustained glycemic management matters not only for the retina, kidney, and heart—but also for the gut.