FC029: A Multivariate Model Identifies Genotype, Hypertension and Kidney Length as Independent Baseline Predictors of Disease Progression in a Longitudinal Autosomal Dominant Polycystic Kidney Disease Patient Cohort

Abstract BACKGROUND AND AIMS Patients with autosomal dominant polycystic kidney disease (ADPKD) and evidence of rapid kidney function decline (∆eGFR > 2.5 mL/min/year) are eligible for disease modifying therapy. Predictors of ∆eGFR are important to identify rapid progressors at an early stage before significant loss of kidney function. Genotype and baseline total kidney volume (TKV) are considered the most accurate individual prognostic factors though neither is widely performed in clinical practice. The major objective of this study was to assess what baseline clinical factors could most accurately predict ∆eGFR in an unselected ADPKD cohort (n = 625) followed-up through a specialist PKD clinic between 2010 and 2021. METHOD A cross-sectional retrospective analysis was performed in all ADPKD patients with sufficient data. The mean follow-up duration was 11 (±11.2) years. The rate of estimated glomerular filtration rate (eGFR) decline was determined by linear regression, based on a minimum of 5 years of renal function prior to the study or renal replacement therapy. Patients who were genotyped (n = 390) were classified into those with PKD1 (T, truncating or NT, non-truncating), PKD2, other genes (non-PKD1 or PKD2) or NMD (no mutation detected). Genotype combined with baseline imaging (ultrasound) was available in a subset of patients (n = 138). There was insufficient TKV information in this cohort to include it as an independent variable. RESULTS Kaplan–Meier plots (probability of eGFR > 60 mL/min/1.73 m2 v age, n = 368) showed a significant genotype–phenotype correlation with a median age of 54, 61, 69 and 73 years for PKD1-T, PKD1-NT, PKD2, NMD/other genes, respectively. Dividing the cohort into 3 ∆eGFR groups (<2.5, 2.5–5, >5 mL/min/year,  n = 316) confirmed significant differences in genotype (PKD1-T), baseline hypertension, mean kidney length and proteinuria (PCR > 50 mg/mmol) between patients in the low (<2.5) and very rapid (>5) decline groups. A multivariate linear regression model was developed with all potential baseline factors significant on univariate analysis. We found that a PKD1 truncating mutation (PKD1-T), mean kidney length and baseline hypertension were significant independent predictors for ∆eGFR. This model accounted for 27% of the variability in ∆eGFR (r2 = 0.2668; P < 0.05). CONCLUSION Our multivariate linear regression model confirms the importance of genotype, phenotype and imaging factors as independent prognostic variables determining ∆eGFR over a minimum period of 5 years follow-up in a longitudinal ADPKD cohort. Since the current model accounts for 27% of variability, it will be important to identify other relevant clinical, genetic or epigenetic factors that determine ∆eGFR.