Personalized Treatment in Chronic Kidney Disease

Abstract

Chronic kidney disease affects approximately 10% of the global adult population and progresses to end-stage renal disease in a subset of patients through mechanisms that vary substantially across individuals -- diabetic nephropathy, hypertensive nephrosclerosis, glomerulonephritis, and polycystic kidney disease each following distinct trajectories that respond differently to available therapeutic interventions. Personalised treatment -- matching therapy type, intensity, and sequence to the individual patient's disease biology, progression trajectory, and comorbid burden -- has become achievable with the combination of multi-omics biomarker panels, ML-based progression prediction models, and expanded therapeutic options including SGLT2 inhibitors, mineralocorticoid receptor antagonists, and GLP-1 agonists. We evaluated five personalised treatment frameworks applied to 46,284 CKD patients at 18 European nephrology centres in Tallinn, Copenhagen, Milan, Warsaw, and Athens between 2014 and 2023, covering CKD stages 2-4. Frameworks ranged from guideline-based standard care through ML-integrated multi-omics personalisation with dynamic treatment adjustment. Performance was benchmarked using our CKD Personalisation Effectiveness Score (CPES), integrating eGFR trajectory preservation, proteinuria reduction, cardiovascular event prevention, patient-reported outcome improvement, and equity across socioeconomic groups. The ML multi-omics framework achieved the highest CPES (0.918), reducing annual eGFR decline from 3.8 ml/min/1.73m2 (guideline standard care) to 1.6 ml/min/1.73m2 and cardiovascular event rate from 8.4% to 5.2% over five years. SGLT2 inhibitor benefit was heterogeneous, concentrated in patients with diabetic nephropathy and high urinary albumin-creatinine ratio, while finerenone benefit was strongest in patients with preserved but declining eGFR and high aldosterone-to-renin ratio