Choosing Renal Function Metrics for Population PK

Why the Choice of Renal Measure Matters

Renal function is not a single directly observed quantity. In population pharmacokinetics, a renal measure is usually a surrogate for some aspect of kidney function that may influence drug clearance. Choosing a measure therefore requires asking two questions: what physiological quantity does it approximate, and is that quantity relevant to this drug’s renal elimination?

Creatinine clearance (CrCL), estimated glomerular filtration rate (eGFR), and cystatin C-based estimates are related but not interchangeable. They differ in their biomarkers, equations, units, intended populations, and susceptibility to nonrenal influences. The most familiar measure is not necessarily the most informative one for every drug or study population.

Mechanistic renal contributions to drug clearance include filtration and, for some drugs, tubular secretion or reabsorption. A filtration estimate can be informative when renal elimination depends substantially on filtration of unbound drug. It may be a weaker proxy when active transport, binding, or other processes make a major contribution. No general-purpose renal estimate directly measures the clearance of a particular drug.

What the Common Measures Represent

Creatinine is produced mainly by muscle metabolism and eliminated by the kidneys. Serum creatinine-based equations use the measured serum creatinine together with selected characteristics such as age and sex to estimate renal function. Because creatinine is also secreted by renal tubules, creatinine clearance can exceed GFR. A timed urine collection can estimate CrCL from urine creatinine concentration, urine flow, and serum creatinine, but collection errors and changing serum concentrations can undermine that estimate. A commonly used alternative in adult drug-dosing contexts is equation-based estimated CrCL. Such estimates are not direct measurements of filtration.

An eGFR equation estimates GFR rather than creatinine clearance. Creatinine-based eGFR equations are commonly used in clinical assessment, and their reported result is often indexed to BSA so that kidney function can be compared across people of different body sizes. This indexed value is not the same as an individual’s absolute filtration rate in mL/min. For some drug-clearance questions, the absolute rate may be more directly relevant; the indexed value may still be useful for describing or comparing renal function in a clinical context.

Cystatin C is a biomarker produced by nucleated cells and filtered at the glomerulus. Cystatin C-based equations provide another estimate of GFR, with potential value when creatinine production is atypical. They are not free of nonrenal influences: factors such as inflammation, thyroid status, corticosteroid exposure, and body composition can affect cystatin C. Equations that combine creatinine and cystatin C may reduce some limitations of relying on either marker alone, but remain estimates whose performance depends on the population and clinical setting.

Indexing, De-indexing, and Units

Indexing expresses estimated filtration relative to a standard BSA. If an indexed eGFR is \(GFR_{\mathrm{indexed}}\) and the person’s BSA is \(BSA\), the corresponding de-indexed estimate is calculated as follows:

The ratio \(BSA/(1.73\ \mathrm{m}^2)\) is dimensionless, so multiplying an indexed value in mL/min/1.73 m² by this ratio gives a value in mL/min. De-indexing changes the scale to an estimate for the person’s body size; it does not improve the underlying biomarker or make the estimate a direct measurement.

Population Fit and Biomarker Limitations

An equation is only as applicable as the population and conditions for which it was developed and validated. Adult equations should not be applied uncritically to children: growth and maturation affect both body size and biomarker interpretation, and pediatric renal estimates use approaches intended for pediatric populations. Similarly, results in people with extreme body composition, severe illness, or unusual biomarker production may be less reliable than in the equation’s typical population.

Creatinine-based estimates can be biased when muscle mass or creatinine generation differs substantially from what the equation expects. Frailty, amputation, neuromuscular disease, bodybuilding, and major changes in diet are examples that can make serum creatinine a misleading indicator of filtration. Some drugs inhibit tubular creatinine secretion and can raise serum creatinine without a corresponding reduction in GFR. In addition, creatinine-based equations generally assume a relatively stable serum creatinine; during rapidly changing kidney function, an equation based on a single value may lag behind the current state.

Cystatin C can be useful when low or unusually high muscle mass makes creatinine difficult to interpret, but it does not provide a universal correction. Nonrenal influences on cystatin C may create disagreement with creatinine-based estimates. When both biomarkers are available, a combined estimate can be considered, but the choice should still be justified by the population, assay and equation used, and clinical context.

Match the Measure to the Drug and Study

A renal metric should be selected in relation to the drug’s elimination pathway, not solely by convention. If a drug is largely eliminated by glomerular filtration, a measure of filtration may be mechanistically relevant, especially when the eliminated fraction is available for filtration. If renal clearance includes substantial active secretion or reabsorption, GFR alone may not track the full renal contribution. The earlier material on mechanistic renal contributions to drug clearance provides the relevant pathway context; here the key point is that a biomarker estimate is a proxy, not a pathway-specific assay.

The intended population also matters. For an adult dataset with a dosing history or external evidence tied to estimated CrCL, that measure may support comparability with the evidence base. For a study using standardized clinical eGFR reports, indexed eGFR may be readily available, but its units and purpose should be kept explicit. Where absolute filtration is more relevant, de-indexed eGFR can be considered. Cystatin C or a combined estimate may be informative if creatinine is expected to be unreliable, provided cystatin C’s own limitations are considered.

These choices are not a contest in which one metric is always best. The most defensible candidate is the one that is interpretable in the target population, sufficiently measured across the study, and biologically connected to the drug’s elimination. If multiple metrics are available, their agreement or disagreement is useful information rather than a reason to select one automatically.

Diagnostics, Alternatives, and Common Pitfalls

Before using a renal estimate, check how it was generated and recorded. Confirm the equation version, biomarker, units, indexing status, timing relative to concentration measurements, and whether the population is within the equation’s intended scope. Examine the distribution for implausible values and assess whether missingness or availability of cystatin C is concentrated in particular patient groups. When two estimates are available, plot or tabulate their paired differences and look for systematic patterns by age, body size, muscle mass, or clinical status.

Disagreement between CrCL and eGFR is not by itself evidence that one is wrong. They may differ because their equations target different quantities, use different inputs, or respond differently to tubular creatinine secretion and body size. A very high estimate in a person with low muscle mass, an abrupt change in serum creatinine, or a mismatch between reported units and values are reasons to investigate. Measurement error in the renal covariate can also weaken or distort an estimated covariate relationship in a population PK analysis.

The main alternatives are not interchangeable substitutes. Estimated CrCL may be preferable when consistency with a relevant dosing convention or prior evidence is important. Indexed eGFR is appropriate for standardized kidney-function reporting and comparisons on that scale. De-indexed eGFR expresses an estimated absolute filtration rate. Cystatin C-based and combined equations offer alternative biomarker information, especially when creatinine production is atypical. A measured clearance can be considered when the scientific question or clinical context requires it, but timed collections also have practical and measurement limitations.

A second common error is treating serum creatinine as a direct measure of kidney function, or treating any renal estimate as a direct measure of drug elimination. Both steps skip important assumptions. Biomarker production, equation validity, filtration, secretion, reabsorption, and the drug’s own disposition all affect how well a renal metric represents the process of interest. The metric-selection decision should therefore be recorded as a scientific choice with known limitations, not as a routine data-cleaning step.