Creatinine and GFR: an imperfect marriage of convenience

Sir, While I would like to congratulate Fontsere [ 1 ] for his recent work that further elucidates the relationship between glomerular filtration rate (GFR) and serum creatinine, [ 1 ] it is not without some consternation that I watch the evolution of the use of creatinine as a marker for GFR. Since it has always been a marriage of convenience at the expense of accuracy, I think that no one should be shocked that Fontsere, as have others before him [ 2 ], found that such equations are not an accurate measure of GFR. Researchers in the early 20th century considered inulin to be the best marker for GFR, but it was inconvenient because it required a timed urine collection and the administration of an exogenous substance [ 3 ]. Oddly enough, when Rehberg [ 4 ] first proposed the use of creatinine, it was that as an exogenous administration; because there were many substances in the serum that were not really creatinine but gave Jaffe's reaction, one could not rely on endogenous creatinine. However, since tubular secretion of creatinine often counterbalanced the overestimation in the serum by non-creatinine chromogens, a marriage of convenience was born, because the measurement of an endogenous creatinine clearance did not require the inconvenience of the administration of an exogenous substance. By the time micropuncture studies confirmed that the advantage of inulin over creatinine, we knew that creatinine clearance was indeed an inadequate measure of GFR [ 5 ], yet we continued to use creatinine because it was convenient, and by the late 20th century we had had so much experience with creatinine clearance that we really knew more about human diseases and symptoms at any given creatinine clearance than at the true GFR. In reality, we do not use the creatinine clearance to estimate GFR since we have so little experience with real inulin clearances, but we use it because we have become comfortable with creatinine and have such a vast experience with creatinine clearances that we can better predict symptoms at any known creatinine clearance than we could if we had an inulin clearance or an actual GFR. Most equations are designed to fit well with mean measured GFRs [ 6 ], but practicing physicians, unlike administrators and epidemiologists, do not deal with means but individual values. Similarly, as Fontsere correctly notes, anything that upsets that tenuous balance and interferes with the production, (including diurnal variations) [ 7–9 ] measurement [ 10–13 ] or secretion [ 14–16 ] of endogenous creatinine will further alter those estimations of GFR from equations using a solitary serum creatinine. Therefore, the problem is not limited to one particular equation. The problem is and has always been with creatinine itself and our own search for convenience. Creatinine clearance is not and has never been synonymous with GFR, and all of the regression analysis will not make it so because the serum creatinine depends upon many factors other than filtration. We should not be surprised that the more approximations that we make, the less accurate our data becomes. The problems come when we actually delude ourselves (and others) into thinking that these equations actually represent an actual GFR. Conflict of interest statement . None declared.