Early magnesium changes may contribute to SGLT2 inhibitor–induced left ventricular mass regression in CKD

Background Disturbances in the bone and mineral homeostasis are central drivers of cardiovascular disease in chronic kidney disease (CKD), contributing to vascular calcification and left ventricular hypertrophy. In DECODE-CKD, the sodium-glucose cotransporter 2 inhibitor (SGLT2i) dapagliflozin induced significant left ventricular mass index (LVMI) regression in patients with CKD, representing a novel structural cardiac benefit in this population. The biological pathways underlying this reverse remodeling remain incompletely defined. We investigated whether early changes in mineral metabolism contribute to LV mass regression among CKD patients undergoing dapagliflozin treatment. Methods DECODE-CKD was a 6-months randomized, placebo-controlled trial evaluating the effect of dapagliflozin on LVMI among 222 patients with CKD. Mediation analyses using structural equation modelling assessed whether 2-week changes in magnesium, phosphate, parathyroid hormone, and ionised calcium mediated dapagliflozin-induced LV mass regression. Treatment predicted mineral change (adjusted for baseline levels), and follow-up LVMI was regressed on treatment, mineral change, baseline LVMI, and baseline mineral level. Analyses were restricted to a complete-case sample. Results Among 200 complete cases (mean age 66.5±13.3 years; 29% female) mean eGFR was 43.6 ± 12.8 mL/min/1.73m² in the dapagliflozin group and 42.7 ± 14.6 mL/min/1.73m² in the placebo group. Baseline ionized calcium, phosphate, and magnesium concentrations were within the normal range (Figure 1). Dapagliflozin reduced LVMI by approximately 8 g/m² across models (all p<0.001). Treatment significantly increased 2-week magnesium levels, and increases in magnesium were associated with greater LVMI regression (β=-32.5 g/m², p=0.039). The indirect effect through magnesium was -1.47 g/m² (p=0.052), corresponding to 18% of the total treatment effect. Dapagliflozin showed a borderline increase in 2-week phosphate levels (p=0.068), however phosphate change was not significantly associated with LVMI and did not mediate the treatment effect. Treatment did not significantly alter parathyroid hormone or ionised calcium levels, and neither marker was associated with LVMI. Conclusions Early changes in magnesium may account for a modest proportion of dapagliflozin-induced LV mass regression in CKD, identifying magnesium as a previously underappreciated contributor to SGLT2i-mediated reverse remodeling in CKD. However, the majority of the remodeling effect remains independent of short-term mineral changes, suggesting additional direct or indirect cardiac mechanisms.