Background
Creatinine clearance (CrCl) has been used for decades as a proxy measure for glomerular filtration rate (GFR), and contemporary equations used to estimate GFR are based on serum creatinine concentration [
1,
2]. Besides creatinine production and GFR, another factor which can influence serum creatinine concentration is tubular secretion (i.e. non-filtration clearance) of creatinine. But there are surprising gaps in knowledge regarding the relation between CrCl and GFR.
First, prior papers have been inconsistent regarding the degree to which CrCl overestimates GFR, with estimates ranging from around 10% [
3‐
7] to over 60% [
8,
9].
Second, we recently questioned whether the observed larger ratio of CrCl to measured GFR among those with lower GFR is actually due to proportionally greater tubular creatinine secretion with chronic kidney disease (CKD) progression, as is commonly believed, or whether this can be entirely accounted for by (random) measurement error [
3,
10]. But our prior study was limited since it was based on only cross-sectional analysis of enrollees from one CKD study (Chronic Renal Insufficiency Cohort [CRIC]), and the CrCl and measured GFR were not obtained simultaneously [
3].
To address these two issues, we sought to analyze datasets where a larger number of patients had undergone measurement of both CrCl and GFR. The Modification of Diet in Renal Disease study (MDRD) and African American Study of Kidney Disease and Hypertension (AASK) are ideal data sources since they are both rigorously conducted cohorts and national in scope. They have the additional advantage of containing repeated measured of both CrCl and GFR, thus allowing us to perform longitudinal analyses and add a new dimension not present in our prior paper. We also conducted cross-sectional analysis of 4 different Mayo Clinic patient cohorts that underwent clinically-indicated CrCl and GFR measurements that were performed simultaneously (which should reduce dis-concordant measurement error).
Discussion
Although it has been known for many years that creatinine is cleared via both filtration and secretion, basic gaps remain in our knowledge regarding the physiology and pathophysiology of tubular secretion of creatinine in various states of health and disease. For example, exactly what fraction of creatinine is cleared via secretion vs. filtration? Is it possible that measurement error alone could entirely account for the longstanding observation that CrCl/GFR ratio is larger at lower GFR among patients with CKD? These questions are pertinent in the context of renewed recent interest in renal tubular function [
27‐
29]. Better understanding of tubular secretion as an independent marker of kidney function may provide insight into kidney disease pathophysiology and improve prediction of adverse outcomes [
27]. Others have emphasized that many drugs are cleared by tubular secretion, and drug-dose modification in CKD should not assume that renal excretory processes always decline in parallel with GFR as CKD progresses [
30].
In this study we showed the following. First, we are able to replicate the results of our original CRIC cross-sectional results in MDRD and AASK [
3]. Replicating our initial findings in two other large research cohorts underscores the idea that certain studies in the literature may have been biased by measurement error, which has been under-appreciated in the past. (In fact, measurement error has not been mentioned at all in the prior literature as far as we can tell.)
Second (and closely related conceptually), we demonstrate that measurement error likely played an under-appreciated role also in prior literature that describes longitudinal changes in the CrCl/GFR ratio. For example, in the much cited study by Shemesh et al. [
8], the authors concluded from their longitudinal analysis that “The opposite changes in fractional creatinine secretion and GFR as glomerular disease deteriorates or improves serve to blunt the magnitude of change when creatinine is used to monitor progression of the glomerular injury [
8].” Not considered by the authors is the fact that measurement error alone may explain much of this observation.
Third, since inaccurate collection of urine specimens is the biggest source of measurement error for directly measured GFR and CrCl by urinary clearance, when both CrCl and GFR were measured using the same timed urine collections, the role of measurement error should be considerably reduced. In this context, the findings in the Mayo Clinic cohort CKD patients (and kidney transplant recipients with CrCl < 75 ml/min/1.73 m
2) are of particular interest. In these cohorts that used the same urine collection to quantify iGFR and CrCl we no longer observed a decreasing CrCl/iGFR ratio at progressively lower CrCl. Instead, in these patients the CrCl/iGFR ratio goes up in cross-sectional analysis with progressively lower CrCl (and progressively lower iGFR). This—and the between-group comparisons shown in Table
1---supports the textbook teaching that there is indeed proportionally greater tubular creatinine secretion with worsening kidney function (since those with lower CrCl have worse kidney function). These data demonstrate that measurement error alone could not entirely account for the longstanding observation that CrCl/GFR ratio increases as GFR decreases in patients with CKD.
Fourth, we note that there is considerable variation in the average CrCl/iGFR ratio across studies. In CRIC study, the median ratio was 1.09 [
3]. In MDRD study it was 1.24 and in AASK study it was 1.13. One problem is that these and other older studies [
3‐
9] were done prior to the era of standardization of the IDMS-traceable reference calibrator. Another problem is that different studies used different methods to measure GFR and these are known to be inconsistent with each other [
31‐
33], thus rendering definitive conclusions difficult. Even within our study, serum creatinine was measured using the enzymatic method at the Mayo Clinic but not in MDRD or AASK. The alkaline picrate method for the measurement of creatinine can be influenced by non-creatinine chromogens (such as acetoacetate and some antibiotics), especially on older platforms [
34]. The so-called ‘Jaffe’ method has no standard recipe and much methodological variation has occurred over time [
35]. Interestingly in the Mayo Clinic data (Table
2), the CrCl/iGFR ratio appears to differ by type of clinical presentation. For example, for the category of iGFR 60–74 ml/min/1.73m
2, median CrCl/iGFR ratio was 1.18 in CKD patients but 1.04 in the post kidney donation subgroup. So perhaps it appears that despite decades of research and seemingly definitive statements in textbooks and review articles, we in fact do not know with certitude the degree by which CrCl overestimates GFR.
Strengths of our study include validation of our prior findings (in CRIC) [
3] from two other large CKD research studies (MDRD and AASK). We extended our cross-sectional analysis to longitudinal analyses to bring attention to the fact that measurement error could also have contributed to prior report of temporal changes in CrCl/GFR over time within a person [
8]. Finally, the Mayo Clinic data, based on using the same timed urine collection to measure CrCl and GFR, shed new light on the limited role of measurement error.
We also recognize several limitations. First, we did not measure iGFR and CrCl using the same blood samples in the Mayo Clinic patients which would have even further reduced measurement error. But previous Mayo Clinic studies have confirmed that a routine fasting morning serum creatinine drawn within 24-h of the study and a plasma creatinine obtained during the iothalamate study do not significantly differ (data not shown). Second, we did not have information on the concomitant use of drugs, such as trimethoprim or cimetidine, which could have affected tubular secretion of creatinine. Third, we did not perform interventional studies with agents such as cimetidine to examine the effect of blocking tubular secretion of Cr [
36‐
39]. However, these experiments are not straightforward since cimetidine is cleared by the kidneys, so in patients with more advanced CKD, less cimetidine is filtered and more cimetidine becomes available in the proximal tubular pericapillary circulation. Thus, more cimetidine enters the proximal tubular cells to compete with creatinine for the brush border (luminal) secretory transporter [
39]. It is also not clear how complete inhibition can be established. Interestingly, one older study which attempted to do that reported that “amount of overestimation of true GFR by creatinine clearance was not associated with extent of renal functional impairment [
37].” However, other studies suggested that the tubular secretion of cimetidine increases inversely with GFR [
36,
38,
39].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.