Clustering-derived clinical subtypes and outcomes of chronic kidney disease in patients with diabetes: a multicenter cohort study.

BACKGROUND

Chronic kidney disease (CKD) with diabetes is a heterogeneous condition with multiple underlying causes. However, no classification system appropriately captures the diversity of patients with CKD and diabetes or provides prognostic insights.

METHODS

We included 11,925 patients with CKD and diabetes from the Multicenter Cohort of Diabetic Kidney Disease Study (MEET-DKD) between 2000-2023. K-means clustering was implemented using thirty-two variables to identify the subtypes of CKD with diabetes. The associations between subtypes and cardiovascular events, progression to end-stage kidney disease (ESKD), and mortality were evaluated.

RESULTS

Three distinct diabetic CKD subgroups were identified; cluster 1 (n=4,494 [37.7%]) included predominantly older patients with multiple comorbidities and the highest medication use (comorbidity-dominant subtype); cluster 2 (n = 2,240 [18.8%]) included patients with lower renal function, and a moderate burden of comorbidities and medications (advanced CKD subtype); and cluster 3 (n = 5,191 [43.5%]) comprised patients with a higher prevalence of obesity, fewer comorbidities, and comparatively favorable laboratory profiles (obesity-dominant subtype). Compared with the obesity-dominant subtype, the comorbidity-dominant subtype exhibited the highest mortality risk (hazard ratio, 1.48; 95% confidence interval, 1.37-1.59). For progression to ESKD, using the comorbidity-dominant subtype as the reference, the advanced CKD subtype showed the highest risk (hazard ratio, 1.46; 95% confidence interval, 1.31-1.64). Similarly, compared with the comorbidity-dominant subtype, the obesity-dominant subtype showed the greatest risk of cardiovascular events (hazard ratio, 1.47; 95% confidence interval, 1.32-1.64).

CONCLUSION

Our study identified three distinct subgroups of diabetic patients with CKD with differential risks for major outcomes. This highlights the heterogeneity of CKD in patients with diabetes, and suggest that identified subtypes may provide additional prognostic information for risk stratification. Further external validation is required before clinical implementation.