AUTHORS
Julia Z Xu, MD, MScGH, Michele Ly, MD2, Rashmi Sahay3, Shiran Chen, MS4, Nicole Herrera, MPH5, Santosh L Saraf, MD6
AFFILIATIONS
1 Division of Classical Hematology, University of Pittsburgh, Pittsburgh, Pennsylvania, United States, Pittsburgh Heart, Lung and Blood Vascular Medicine Institute, Pittsburgh, Pennsylvania, United States
2 University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, United States
3 Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States
4 Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States
5 ASH Research Collaborative, Washington DC, District of Columbia, United States
6 University of Illinois College of Medicine, Chicago, Illinois, United States
ABSTRACT
Background:
Kidney disease affects individuals with sickle cell disease (SCD) across the lifespan. Renal manifestations begin in childhood, with some experiencing rapid decline leading to end-stage kidney disease (ESKD). We aim to use latent class trajectory modeling to study renal function decline using the ASH Research Collaborative (ASH RC) Data Hub, a multicenter, real-world dataset of pediatric and adult patients with SCD. Modeling renal function across the lifespan is complex, as different eGFR equations are used (Schwartz for children; CKD-EPI for adults). The Full Age Spectrum (FAS) equation can estimate eGFR at all ages (Pottel et al. 2016) but is not widely used for SCD. We evaluated baseline data and compared the three eGFR equations to guide future trajectory analysis to identify high-risk subgroups for targeted intervention.
Method:
We abstracted demographic, clinical, and longitudinal eGFR data from the Data Hub spanning 10 years (2015-2025). Individuals with a physician-attested SCD diagnosis and ≥2 outpatient eGFRs separated by ≥6 months were included. Patients on dialysis, with eGFR <15 mL/min, or with a history of kidney or bone marrow transplant or gene therapy were excluded. eGFRs from acute care encounters were also excluded. We used linear mixed-effects models and Bland-Altman analyses to assess agreement among eGFR equations (Schwartz, CKD-EPI, and FAS) within pediatric (<18 years) and adult (≥18 years) subgroups. For each pair, we plotted the mean difference against the average eGFR and computed 95% limits of agreement. Proportional bias was assessed by regressing the difference on the mean eGFR; a significant slope indicated that disagreement varied across the eGFR range.
Results:
A total of 4,837 individuals with SCD (1,641 pediatric and 3,196 adult) were examined, contributing 103,182 eGFR assessments. The overall mean age was 25 ±16 years; 54.9% female, 89.4% Black or African American, 8.2% Other/Unknown, and 10.4% Hispanic/Latino. SCD subtypes included 65.6% HbSS, 24.9% HbSC, 5.7% HbSβ+, and 2.5% HbSβ0. Mean baseline creatinine was 0.67 ±0.4 mg/dL, and eGFRs were 106.9, 129.6, and 120.7 mL/min by Schwartz, CKD-EPI, and FAS equations, respectively. Pediatric and adult cohorts were similar, aside from age (8.8 ±5.3 vs. 33.4 ±13.1 years). The adult group was more female (57.7% vs. 49.5%) and had fewer Other/Unknown races (4.9% vs. 14.7%). Pediatric baseline creatinine was 0.46 ±0.16 mg/dL vs. 0.77 ±0.38 mg/dL in adults. Mean baseline eGFR was lower in adults vs children by Schwartz (99.8 vs. 120.8 mL/min) and CKD-EPI (113.5 vs. 160.9 mL/min), but similar by FAS (120.6 vs. 121.0 mL/min).
In the pediatric cohort, accounting for repeated measures, the mean Schwartz eGFR was lower than CKD-EPI (-34.3 mL/min, p<0.001), especially in the range of 100-175 mL/min. Proportional bias was significantly positive (slope 1.39; p<0.0001), indicating greater differences between equations at higher eGFRs. Above 200 mL/min, a subset of Schwartz eGFRs were 200-400 mL/min above CKD-EPI, consistent with known overestimation by the Schwartz equation (Larché et al. 2026). The mean difference between Schwartz and FAS eGFRs was small and not significant (-0.86 mL/min; p=0.06), with no proportional bias (slope 0.018; p=0.098).
In adults, mean Schwartz eGFRs were lower than CKD-EPI (-9.6 mL/min, p<0.0001), especially in 50-100 mL/min range, with a significant positive proportional bias (slope 0.80; p<0.0001), causing similarly large overestimations of eGFR by Schwartz above 175 mL/min. The mean eGFR difference between CKD-EPI and FAS was also significant but smaller (-5.74 mL/min; p<0.0001), and proportional bias was significantly negative (slope -0.68; p<0.0001), showing larger differences at low eGFRs. None of the compared equations showed sufficient agreement for interchangeability across the full eGFR range or age spectrum.
Conclusion:
This preliminary evaluation of eGFR data from a large, prospective multicenter SCD cohort shows that agreement among the Schwartz, CKD-EPI, and FAS eGFR equations varies by age group and eGFR range. The Schwartz and FAS equations show the most agreement when age < 18, and the Schwartz and CKD-EPI equations show agreement at lower eGFRs. Findings highlight the importance of choosing the right equation when studying renal function in SCD. Ongoing analysis will use latent class mixed modeling (LCMM) to identify eGFR decline patterns and their links to ESKD and other adverse renal outcomes.
DISCLOSURES
Julia Z Xu, MD, MScGH: GSK - Research Funding; Bain Capital Life Sciences - Consultancy (Includes expert testimony); Tarus Therapeutics - Consultancy (Includes expert testimony); Agios Pharmaceuticals, Inc. - Consultancy (Includes expert testimony); Novo Nordisk - Consultancy (Includes expert testimony); Akebia - Consultancy (Includes expert testimony); Vascarta - Consultancy (Includes expert testimony)