A prospective, longitudinal cohort study published in The BMJ reports that glomerular filtration rate (GFR) estimating equations incorporating both creatinine and cystatin C demonstrate superior accuracy in monitoring moderate chronic kidney disease (CKD) compared with creatinine-based equations alone, though all tested equations consistently underestimated the true rate of GFR decline over a three-year observation period.

The study enrolled 1,229 adults across six centers in England between April 2014 and December 2017, drawing participants from primary, secondary, and tertiary care settings. Eligible participants were aged 18 years or older and carried a confirmed diagnosis of moderate CKD, defined as a creatinine-estimated GFR of 30 to 59 mL/min/1.73 m² sustained across at least three successive months prior to recruitment. After accounting for attrition and incomplete data collection, 875 participants contributed measured and estimated GFR data at both the start and end of the study period, constituting the analytic cohort.

Reference GFR was established through iohexol clearance, a measured GFR (mGFR) technique regarded as a reliable reference standard in CKD research. Median mGFR declined from 48.1 mL/min/1.73 m² at baseline to 43.6 mL/min/1.73 m² at three years, representing a clinically meaningful deterioration in renal function across the cohort.

Estimating Equations Under Evaluation

The investigators assessed several GFR estimating equations derived from the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) framework and the European Kidney Function Consortium (EKFC). These included single-biomarker equations relying on creatinine alone and dual-biomarker equations incorporating both creatinine and cystatin C. The primary outcome measure was the ability of each equation to monitor GFR change over three years, with agreement defined as a slope deviation from mGFR within plus or minus 3 mL/min/1.73 m² per year. A secondary outcome assessed the capacity of each equation to detect disease progression, operationalized as a reduction in mGFR of 25% or greater accompanied by a downward shift in CKD category.

Dual Biomarker Equations Demonstrate Superior Agreement

All equations achieved agreement with change in mGFR in more than 72.5% of participants, a finding that establishes a reasonable baseline performance across the field. However, dual biomarker equations consistently outperformed the creatinine-only CKD-EPI equation on the primary outcome measure.

The CKD-EPIcreatinine-cystatin equation achieved agreement in 78.6% of participants (95% confidence interval, 75.8% to 81.3%). The CKD-EPI(2021)creatinine-cystatin equation demonstrated agreement in 78.1% (95% CI, 75.2% to 80.8%). The EKFCcreatinine-cystatin equation performed best among the dual biomarker approaches, reaching agreement in 80.2% of participants (95% CI, 77.4% to 82.8%). By contrast, the CKD-EPIcreatinine equation achieved agreement in 73.1% of participants (95% CI, 70.1% to 76.1%). The differences between single and dual biomarker equations reached statistical significance (all P values below 0.01).

The progressive detection analysis yielded notably high specificity across all equations, with values exceeding 90.4% for identifying CKD progression. This finding suggests that false positive identification of disease progression represents a limited concern in clinical application, though the investigators did not report sensitivity values in the available summary, warranting caution in interpreting the full diagnostic performance profile.

Systematic Underestimation Across All Equations

A finding of particular interest across all evaluated equations was the consistent underestimation of GFR decline. Median change in mGFR exceeded median change in estimated GFR for every equation examined. The investigators describe this pattern as meriting further investigation, and the conclusion is well-supported by the longitudinal design of the study.

The clinical implications of this systematic underestimation are consequential. Patients whose renal function is declining more rapidly than estimated GFR equations suggest may receive inadequate clinical attention, delayed referral to nephrology services, or suboptimal medication dosing adjustments. In moderate CKD, where disease trajectory directly informs decisions about cardiovascular risk management, dialysis planning, and transplant listing, the accuracy of monitoring tools carries direct bearing on patient outcomes.

The mechanisms underlying this underestimation are not fully elucidated in the current study. Plausible contributors include non-GFR determinants of creatinine and cystatin C metabolism, such as muscle mass, diet, inflammation, and thyroid function, each of which may shift across a three-year period in ways that partially mask true GFR decline. The systematic nature of the underestimation across both biomarkers and multiple equation frameworks suggests that the issue is not unique to any single algorithmic approach but may reflect a broader structural limitation in biomarker-based GFR estimation during longitudinal monitoring.

Study Design Strengths and Limitations

The prospective, longitudinal cohort design represents a methodological strength. Longitudinal accuracy data on GFR estimating equations are substantially less common in the literature than cross-sectional validation studies, and the three-year follow-up period provides a meaningful window for observing disease trajectory in a population with established moderate CKD.

The use of iohexol clearance as the reference standard adds further credibility. Iohexol clearance is a well-characterized exogenous filtration marker that does not require radioactive administration, making it practically suitable for multicenter deployment across diverse care settings.

Several limitations warrant acknowledgment. The cohort was recruited exclusively from centers in England, which constrains direct generalizability to populations with different demographic compositions, dietary patterns, or genetic backgrounds. Ethnic diversity within the cohort is not detailed in the available summary, a consideration of particular relevance given that prior CKD-EPI equation iterations generated substantial discussion regarding race-based adjustment terms and their appropriateness. The 2021 revision of the CKD-EPI equation removed race as a variable, and its inclusion in this study provides partial insight into its longitudinal performance, though the analytic cohort’s composition limits interpretation.

Additionally, the study period concluded in December 2017, meaning the data represent a retrospective window by current standards. Nephrology practice, CKD staging criteria, and biomarker assay standardization have continued to evolve. Whether the findings translate directly to contemporary clinical practice merits consideration, though the fundamental physiology underlying GFR estimation has not changed substantially.

The 354 participants who did not contribute complete longitudinal data introduce the possibility of attrition bias. Patients lost to follow-up in longitudinal CKD research frequently differ systematically from completers, often exhibiting either more rapid disease progression or competing comorbidities that preclude sustained study participation. The investigators’ analytic approach to this limitation is not fully detailed in the available summary.

Clinical and Policy Implications

Current clinical practice in CKD monitoring relies predominantly on creatinine-based eGFR, partly due to cost considerations and the widespread availability of serum creatinine assays relative to cystatin C. Cystatin C measurement remains less uniformly available across health systems, and its cost presents a barrier to routine incorporation in some resource-constrained settings.

The present findings provide substantive longitudinal evidence supporting broader clinical adoption of dual biomarker equations where cystatin C measurement is feasible. National guidelines from bodies including the Kidney Disease Improving Global Outcomes (KDIGO) consortium have increasingly acknowledged the potential of cystatin C as a confirmatory biomarker, and this study contributes to the evidentiary foundation that may support future guideline revisions favoring dual biomarker use in routine monitoring rather than confirmatory testing alone.

For health systems considering implementation, the operational question centers on the incremental cost of cystatin C assays relative to the clinical benefit of more accurate disease monitoring. A reduction in missed or delayed progression events carries downstream cost implications in terms of preventable hospitalizations, unplanned dialysis initiation, and suboptimal management of CKD-related cardiovascular risk. Formal health economic modeling informed by the accuracy differences reported in this study would provide a more complete basis for policy recommendations.

Broader Context

The study addresses a gap that has persisted in the nephrology literature for some time. Cross-sectional validation of GFR estimating equations has accumulated substantially over the past two decades, yet the longitudinal performance of these tools, specifically their capacity to track individual patients’ disease trajectory over time rather than estimate a single-point GFR value, has received comparatively limited rigorous examination. The distinction matters clinically: a clinician managing a patient with moderate CKD needs to know not only where that patient’s GFR stands today but whether it is stable, slowly declining,