Abstract
BACKGROUND Serum parathyroid hormone (PTH), phosphorus, and calcium levels are surrogate outcomes that are central to the evaluation of drug treatments in chronic kidney disease (CKD). This systematic review evaluates the evidence for the correlation between drug effects on biochemical (PTH, phosphorus, and calcium) and all-cause and cardiovascular mortality end points in adults with CKD. STUDY DESIGN Systematic review and meta-analysis. SETTING & POPULATION Adults with CKD. SELECTION CRITERIA FOR STUDIES Randomized trials reporting drug effects on biochemical and mortality end points. INTERVENTION Drug interventions with effects on serum PTH, phosphorus, and calcium levels, including vitamin D compounds, phosphate binders, cinacalcet, bisphosphonates, and calcitonin. OUTCOMES Correlation between drug effects on biochemical and all-cause and cardiovascular mortality. RESULTS 28 studies (6,999 participants) reported both biochemical and mortality outcomes and were eligible for analysis. Associations between drug effects on surrogate biochemical end points and corresponding effects on mortality were weak and imprecise. All correlation coefficients were less than 0.70, and 95% credible intervals were generally wide and overlapped with zero, consistent with the possibility of no association. The exception was an inverse correlation between drug effects on serum PTH levels and all-cause mortality, which was nominally significant (-0.64; 95% credible interval, -0.85 to -0.15), but the strength of this association was very imprecise. Risk of bias within available trials was generally high, further reducing confidence in the summary correlations. Findings were robust to adjustment for age, baseline serum PTH level, allocation concealment, CKD stage, and drug class. LIMITATIONS Low power in analyses and combining evidence from many different drug comparisons with incomplete data across studies. CONCLUSIONS Drug effects on serum PTH, phosphorus, and calcium levels are weakly and imprecisely correlated with all-cause and cardiovascular death in the setting of CKD. Risks of mortality (patient-level outcome) cannot be inferred from treatment-induced changes in biochemical outcomes in people with CKD. Similarly, existing data do not exclude a mortality benefit with treatment. Trials need to address patient-centered outcomes to evaluate drug effectiveness in this setting.
