A scoring system for predicting individual treatment effects of statins in type 2 diabetes patients on haemodialysis

Aims Haemodialysis patients have high cardiovascular disease risk. Although statins reduce this risk in chronic kidney disease, randomised trials in haemodialysis patients show no benefit. Post-hoc analyses of the German Diabetes Dialysis (4D) study identified patient-specific markers associated with heterogeneous treatment effects. We combined these markers to develop a score for predicting individual effects of statins in these patients. Methods and results We used data from the 4D study, enrolling 1255 haemodialysis patients with type 2 diabetes mellitus, randomised to atorvastatin or placebo and followed for a composite cardiovascular endpoint. We calculated two scores: score 1 based on all 23 predictive markers and score 2 based on 17 clinically accessible markers. Groups stratified by score 1 showed differential treatment effects: for score <26 (458 patients; 36%), the hazard ratio (95% confidence interval) was 1.54 (1.16-2.03), suggesting harm; for 26-31 (331 patients; 26%), it was 1.03 (0.72-1.48), suggesting a neutral effect; and for >31 (466 patients; 38%), it was 0.43 (0.30-0.60), suggesting a benefit. Statins also significantly reduced all-cause mortality in the benefit group. Stratification by score 2 yielded similar results but a smaller group gaining benefit (360 patients). Conclusion Statin effects in haemodialysis patients can be predicted by markers associated with plausible relevant mechanisms including cholesterol metabolism, atherosclerosis, protein energy wasting, or competing risks. In clinical practice, the score could aid in risk stratification, not only to select patients who benefit from statins but also to identify those whom treatment could harm.