GFR-Specific versus GFR-Agnostic Cutoffs for Parathyroid Hormone and Fibroblast Growth Factor-23 in Advanced Chronic Kidney Disease

Background: In the majority of patients with advanced chronic kidney disease (CKD), values of parathyroid hormone (PTH1–84) and fibroblast growth factor 23 (FGF-23) exceed the normal reference range, potentially as an appropriate adaptation to reduced glomerular filtration rate (GFR). We tested whether GFR-specific cutoffs for PTH1–84 and FGF-23 could better identify patients with inappropriately high PTH1–84 and FGF-23 for their degree of CKD and thereby improve prognostication of clinical outcomes compared to a uniform threshold. Methods: Prospective pan-Canadian cohort of 1,812 patients with mean estimated GFR (eGFR) 28.9 mL/min/1.73 m2 followed for a median of 52 months. Repeated log-rank tests were used to identify optimal cutoffs for PTH1–84 and FGF-23 within eGFR strata (<20, 20–29 and ≥30 mL/min/1.73 m2) that maximally differentiated high- and low-risk populations for (1) cardiovascular (CV) events (fatal or nonfatal myocardial infarction, coronary revascularization, stroke, heart failure) and (2) renal events (initiation of chronic renal replacement therapy). In multivariable models, we examined the association between ­GFR-specific cutoffs and outcomes and compared their added prognostic value to existing uniform thresholds. Results: Risk-based cutoffs for PTH1–84 and FGF-23 increased in a graded fashion with decreasing eGFR. Among patients with eGFR <20 mL/min/1.73 m2, CV risk-based cutoffs for PTH1–84 and FGF-23 were 3.4 and 5.5 times the upper limit of normal, respectively, and reclassified 31.9 and 35.1% of patients when added to a multivariable base model for CV events. In contrast, the addition of PTH1–84 and FGF-23 to the base model using uniform cutoffs failed to reclassify such patients. Similar findings were demonstrated for renal outcomes. Conclusion: GFR-specific risk-based cutoffs for PTH1–84 and FGF-23 may facilitate more meaningful risk stratification in advanced CKD than current GFR-agnostic reference ranges derived from healthy adults. This may be most applicable in those with severely reduced GFR.