Effect of Statin Therapy on the Progression of Autosomal Dominant Polycystic Kidney Disease. A Secondary Analysis of the HALT PKD Trials

Abstract

Background: Autosomal dominant polycystic kidney disease (ADPKD) commonly results in end-stage renal disease (ESRD), yet a long-term treatment that is well tolerated is still lacking. In a small randomized trial in children and adolescents pravastatin administration for 3 years was associated with reduced renal cyst growth, but no large trial has tested the effect of statins in adults.

Methods: We performed a post-hoc analysis of the HALT PKD trials to compare outcomes of participants who never used statins with those who used statin for at least 3 years. Because statins were not randomly allocated, we used propensity score models with inverse probability of treatment weighting to account for imbalances between the groups. For subjects in Study A (preserved renal function, n=438) relevant outcomes were percent change in total kidney and liver volume and the rate of decline in estimated glomerular filtration rate (eGFR); for those in Study B (reduced renal function, n=352) we compared time to the composite endpoint of death, ESRD or 50% decline in eGFR. Follow-up was 5-8 years.

Results: There was no difference in any outcome between the 2 groups. However, limitations of this analysis are the small number of statin users in Study A, different statin drugs and doses used, non-randomized allocation and advanced disease stage in Study B.

Conclusion: Although this post-hoc analysis of the HALT PKD trials does not demonstrate a benefit of statin therapy, conclusions remain preliminary. A larger randomized trial in young people with ADPKD is necessary to answer the question whether statins can slow renal cyst growth and preserve kidney function.

Keywords: Autosomal dominant polycystic kidney disease; HALT PKD trials; end-stage renal disease; glomerular filtration rate; hydroxymethylglutaryl-CoA reductase inhibitors; total kidney volume.

Figure 1

Standardized mean differences with (weighted differences) and without (unweighted differences) accounting for the inverse probability of treatment weight (IPTW). Baseline characteristics which were used in the propensity score models are shown for Study A (Fig. 1 a) and Study B (Fig. 1 b).

Figure 1

Standardized mean differences with (weighted differences) and without (unweighted differences) accounting for the inverse probability of treatment weight (IPTW). Baseline characteristics which were used in the propensity score models are shown for Study A (Fig. 1 a) and Study B (Fig. 1 b).

Figure 2

Model-based estimates of change in total kidney volume (TKV) from baseline over 60 months in Study A. Point estimates and 95% confidence intervals derived from linear mixed models accounting for the inverse probability of treatment weights and including predictors for month, statin use group, and their interaction. The difference between statin users (for at least 3 years) and never users was not significant (p=0.51).

Figure 3

Model-based estimates of change in height-adjusted total liver volume (htTLV) from baseline over 60 months in Study A. Point estimates and 95% confidence intervals derived from linear mixed models accounting for the inverse probability of treatment weights and including predictors for month, statin use group, and their interaction. The difference between statin users (for at least 3 years) and never users was not significant (p=0.54).

Figure 4

Model-based estimates of change in eGFR over 60 months in Study A. Point estimates and 95% confidence intervals derived from linear mixed models accounting for the inverse probability of treatment weights and including predictors for month, statin use group, and their interaction. The difference between statin users (for at least 3 years) and never users was not significant (p=0.57).

Figure 5

Probability of event-free survival from the composite outcome in Study B. Survival curves estimated from a Cox proportional hazards model accounting for the inverse probability of treatment weights and including statin use group as a predictor. Number of participants at risk are shown above the x-axis. There was no difference between statin users (for at least 3 years) and never users (p=0.96).