New study shows that existing serum creatinine-based equations are inaccurate for estimating GFR in Indian subjects

One of the most influential advance in the management of patients with kidney disease in the last 20 years has been mathematical – the derivation of formulae to estimate glomerular filtration rate (GFR) using serum level of the filtration marker creatinine, and subject characteristics (age and sex). This big advance allowed accurate diagnosis, classification and tailored treatment planning for chronic kidney disease (CKD). Online calculators allow us to quickly get a number and laboratories are using them to report eGFR.

This equation was developed in largely Caucasian North America in population – and it became evident during the initial derivation that the same equation did not accurately predict true GFR in blacks and whites, leading to incorporation of an ethnicity specific correction factor for blacks. Subsequently, based on the same principle, correction factors were developed for the Japanese and Chinese populations. This means that the same serum creatinine, age and sex combination that gives a GFR value of 100 ml/min in Caucasians would give a value of 112 ml/min in blacks, 123 ml/min in the Chinese and 81 ml/min in the Japanese.

What does it all mean for us in India? Since serum creatinine depends upon muscle mass and dietary intake of meat, it is not hard to hypothesize that the original equation would not work well in Indian population. Moreover, given the difference in dietary habits, GFR in normal Indian population might well be lower than that described in the western population. A couple of studies (here and here)  indeed suggested that the normal distribution of GFR in the Indian population is different than that described in the literature.

In that case, do these people have CKD or not? It is worth noting that the CKD definition says that abnormalities are of relevance only when they have consequences for health, which we seem to overlook. So if a person has a physiological GFR of 80 ml/min but otherwise is not at increased risk of any adverse consequences it would be improper to classify that person as CKD.

Now we have a new study* that lends substantial support to this hypothesis. In this study, GFR was measured in 130 subjects – healthy individuals as well as those with CKD, using inulin clearance, the gold standard for GFR measurement. Serum creatinine and cystatin C were measured using validated methods. The study showed that the true (measured) GFR was lower than the estimated GFR ursine serum creatinine throughout the range of GFR values. Only 22% of subjects showed eGFR values (using the CKD-EPI equation which is used by clinical laboratories to report eGFR in India) that fell within plus or minus 30% of true GFR, which is terrible!

Unsurprisingly, creatinine excretion and urinary urea nitrogen appearance data in the study showed low values – providing a possible reason for this finding.

Interestingly, when GFR was estimated using serum cystatin C, about 75% values fell within the same 30% range, suggesting that a marker that does not depend upon the muscle mass is more accurate in estimating the truce GFR.

Since serum creatinine is widely available and cheap, it will remain the molecule that is used to estimate GFR in the foreseeable future, so it is imperative that a correction factor is developed, and validated for populations in all parts of India and across all GFR values. This requires simultaneous measurement of GFR using a reference method (inulin, iohexol or DTPA clearance) and serum creatinine testing performed using validated methodology that can be traced back against IDMS standard.

This finding raises a lot of important questions relevant to clinical practice. The most uncomfortable of which is- do we even know how to tell anyone whether they do or do not have healthy kidneys, and make an accurate diagnosis of CKD?

Until we have an accurate measure, we will continue to misdiagnose, misclassify and perhaps wrongly treat our patients . This is really important in the context of public health measures that include community screening for CKD. Another important consideration relates to accurately identifying kidney donors who will not be subjected to additional kidney risk by the act of donation. It is worth noting that 3 out of 63 healthy subjects were found to have measured GFR of <60 ml/min. How do we determine the significance low this “low” GFR? By following these people up and finding out what happens!!

This is a challenge to the Indian nephrology research community and funding agencies to support measures that answer this question and dance clinical care and research in kidney disease.

*The study was funded by the Department of Biotechnology, Govt of India and was done in collaboration with the Japanese Society of Nephrology

3 thoughts on “New study shows that existing serum creatinine-based equations are inaccurate for estimating GFR in Indian subjects

  1. Dr. Hasit Patel

    What is taught to us in nephrology is actually now proven by research. Our teachers always taught us that this formula are different in Indian population.
    Even dietary patterns are different in different regions of India….

    Only option at present seems feasible as per my small knowledge is conducting a research as per zones of India with 24 hour creatinine clearance, either inulin clearance or DTPA….. estimating correction factor in Indian population and if required zonal population……

    Further in India we have different labs doing by different methods… as already suggested by few international bodies…like INR…we should have some standard with which it will compare the results given by particular method.

  2. PDR

    Well done

    The answer however seems to be simple and the work already done. It’s also staring in the face.

    Thereby the need/romantic n sentimental urge for an equation, that doesn’t seem to be doing its job.

    CYS does better
    in aki (rise early)
    ckd (staging) n
    higher gfrs (donation n screening)
    Lower gfr ( liver disease )

    Validate in bigger nos.
    Numbers , ofcourse centres n directions
    Way to go cys –

    RIP creat ?

    1. vjha Post author

      Yes – cystatin seems to work better but costs much more, is technologically challenging and it would be hell of a job to change all labs right away. So we have creatinine for the foreseeable feature and need to make it work better.


Leave a Reply

Your email address will not be published. Required fields are marked *