The Validity of Left Ventricular Mass as a Surrogate End Point for All-Cause and Cardiovascular Mortality Outcomes in People With CKD: A Systematic Review and Meta-analysis

Background: Left ventricular mass (LVM) is a widely used surrogate end point in randomized trials involving people with chronic kidney disease (CKD) because treatment-induced LVM reductions are assumed to lower cardiovascular risk. The aim of this study was to assess the validity of LVM as a surrogate end point for all-cause and cardiovascular mortality in CKD. Study Design: Systematic review and meta-analysis. Setting & Population: Participants with any stages of CKD. Selection Criteria for Studies: Randomized controlled trials with 3 or more months' follow-up that reported LVM data. Intervention: Any pharmacologic or nonpharmacologic intervention. Outcomes: The surrogate outcome of interest was LVM change from baseline to last measurement, and clinical outcomes of interest were all-cause and cardiovascular mortality. Standardized mean differences (SMDs) of LVM change and relative risk for mortality were estimated using pairwise random-effects metaanalysis. Correlations between surrogate and clinical outcomes were summarized across all interventions combined using bivariate random-effects Bayesian models, and 95% credible intervals were computed. Results: 73 trials (6,732 participants) covering 25 intervention classes were included in the meta-analysis. Overall, risk of bias was uncertain or high. Only 3 interventions reduced LVM: erythropoiesis-stimulating agents (9 trials; SMD, -0.13; 95% CI, -0.23 to -0.03), renin-angiotensin-aldosterone system inhibitors (13 trials; SMD, -0.28; 95% CI, -0.45 to -0.12), and isosorbide mononitrate (2 trials; SMD, -0.43; 95% CI, -0.72 to -0.14). All interventions had uncertain effects on all-cause and cardiovascular mortality. There were weak and imprecise associations between the effects of interventions on LVM change and allcause (32 trials; 5,044 participants; correlation coefficient, 0.28; 95% credible interval, -0.13 to 0.59) and cardiovascular mortality (13 trials; 2,327 participants; correlation coefficient, 0.30; 95% credible interval, -20.54 to 0.76). Limitations: Limited long-term data, suboptimal quality of included studies. Conclusions: There was no clear and consistent association between intervention-induced LVM change and mortality. Evidence for LVM as a valid surrogate end point in CKD is currently lacking. Crown Copyright (C) 2016 Published by Elsevier Inc. on behalf of the National Kidney Foundation, Inc. All rights reserved.