AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Impact of Chronic Kidney Disease on the Predictive Value of Growth Differentiation Factor 15 for Cause-Specific Mortality in Patients with Diabetes and Coronary Artery Disease: The EXCEED-J Study
By Wada H, Shinozaki T, Suzuki M et al.
Background: Growth differentiation factor 15 (GDF-15), a stress-responsive transforming growth factor-β (TGF-β) superfamily cytokine, plays an important role in linking cardiovascular (CV) disease, renal disorders, and metabolic conditions. However, the impact of chronic kidney disease (CKD) on the predictive value of GDF-15 for cause-specific mortality in patients with diabetes and suspected or known coronary artery disease (CAD) remains unclear. Methods: Serum levels of GDF-15 were measured in 1,281 (566 CKD and 715 non-CKD) patients with diabetes and suspected or known CAD enrolled in the EXCEED-J study. The outcomes were all-cause, CV, non-CV, cancer, and non-CV/non-cancer deaths over a 6-year follow-up. Results: During follow-up, 214 all-cause, 68 CV, 120 non-CV, 51 cancer, 69 non-CV/non-cancer, and 26 undetermined deaths occurred in the entire cohort (128, 42, 70, 24, 46, and 16 in the CKD group; 86, 26, 50, 27, 23, and 10 in the non-CKD group, respectively). After adjustment for clinical confounders and established CV biomarkers (NT-proBNP, high-sensitivity [hs] cardiac troponin I, and hs-CRP), log-transformed (Ln-) GDF-15 levels (per 1-SD increase) were significantly associated with all-cause (adjusted hazard ratio [HR], 1.40; 95% confidence interval [CI], 1.17-1.69), non-CV (HR, 1.85; 95% CI, 1.44-2.38), cancer (HR, 1.53; 95% CI, 1.05-2.21), and non-CV/non-cancer (HR, 2.15; 95% CI, 1.53-3.02) deaths, but not with CV death, in the entire cohort. Similarly, in patients with CKD, they were significantly associated with all-cause (HR, 1.30; 95% CI, 1.03-1.64), non-CV (HR, 2.10; 95% CI, 1.52-2.90), cancer (HR, 2.16; 95% CI, 1.23-3.81), and non-CV/non-cancer (HR, 2.10; 95% CI, 1.40-3.17) deaths, but not with CV death. In contrast, in patients without CKD, they were significantly associated with all-cause (HR, 1.51; 95% CI, 1.17-1.96), non-CV (HR, 1.47; 95% CI, 1.04-2.08), and non-CV/non-cancer (HR, 1.83; 95% CI, 1.09-3.00) deaths, but not with CV or cancer death. The addition of Ln-GDF-15 to the model with clinical confounders and established CV biomarkers significantly improved the prediction of non-CV and cancer deaths in the entire cohort (for non-CV: continuous net reclassification improvement [cNRI] = 0.273, P = 0.030; integrated discrimination improvement [IDI] = 0.017, P = 0.005; for cancer: cNRI = 0.231, P = 0.016; IDI = 0.004, P = 0.029) and in patients with CKD (for non-CV: cNRI = 0.398, P = 0.030; IDI = 0.027, P = 0.045; for cancer: cNRI = 0.384, P = 0.002; IDI = 0.018, P = 0.005); however, it did not significantly improve the prediction of all-cause, CV, or non-CV/non-cancer deaths in the entire cohort or CKD group, nor any outcomes in patients without CKD. Conclusions: In patients with diabetes and suspected or known CAD, GDF-15 levels independently predicted non-CV mortality, particularly cancer mortality in the entire cohort and in the presence of CKD, but not in its absence.