Correct estimation of GFR is essential to diagnose and stage CKD, to assess individual risks, and to allowing comparisons between different populations in epidemiologic studies. The KDIGO analyses included some patients from China and Japan, but most of the data available for analysis is of mainly Caucasian or US black subjects with eGFR calculated using S-MDRD formula . This study found that GFR estimates, CKD prevalence, proportions with increased risks, and gender ratios in this Asian cohort varied widely depending on the equation used. To our knowledge this is the first study to evaluate the use of J-EPI and T-GFR equations on the prevalence of CKD, and the concordance of these new Asian formulae on CKD staging using the new KDIGO guidelines.
In this study, CKD 3-5 prevalence was as high as 37.9% with J-EPI whereas the prevalence was only 5-6% with T-GFR or C-MDRD, 20.2% with S-MDRD and 12.1% with CKD-EPI. The importance of choice of eGFR equations on CKD prevalence have been shown previously in studies from the US and Europe [8, 11, 22]. In these studies, changing from S-MDRD to CKD-EPI resulted in lower CKD prevalence estimates. In US white and black subjects, CKD-EPI has emerged as the preferred equation for population studies as it has been shown to be more accurate in CKD stage classification than the MDRD study equation [11, 12]. At present, it remains unclear which is the best formula for Asian subjects. CKD-EPI has recently been shown not to provide consistent prediction of true GFR when applied to Asian subjects .
In diagnosing moderate CKD (eGFR < 60) in a general population, the concordance of T-GFR or C-MDRD with CKD-EPI was over 90% compared to only 80% with S-MDRD. In assigning stage for all subjects with CKD (from stages 1 to 5), the concordance of T-GFR or C-MDTRD equations was only about 50-60% with CKD-EPI, and less than 40% with S-MDRD. The concordance between T-GFR and C-MDRD was high, but both had very poor concordance with J-EPI. Therefore, it appears that T-GFR, C-MDRD and CKD-EPI may produce comparable results in identifying individuals with moderate CKD, but the CKD stage assigned, if all stages of CKD were assessed, may be different in half the time.
Using the 2009 KDIGO guidelines to subdivide stage 3, we showed that the CKD ALL, stage concordance between the Asian and Caucasian equations decreased by about 5-9% when compared to the KDOQI classification in which stage 3 was considered as one stage. KDIGO also recommended the use of composite rankings of relative risks according to GFR and proteinuria stages to assess individuals. In our analysis, the choice of equation affected the numbers at risk quite markedly especially those with mildly increased risks. Taken together this data suggests that the choice of eGFR equation has very considerable impact in diagnosing CKD, CKD staging, individual risk assessment.
This study demonstrated that changing eGFR equations could affect CKD gender ratios quite dramatically. Similar to our study, the male to female CKD ratio among US NHANES subjects increased slightly when the GFR equation was changed from S-MDRD to CKD-EPI . However, the largest difference in our study was found when T-GFR was used, since this led to a reversal in the male to female ratio even when compared to C-MDRD. This finding, however, is not consistent with the slightly higher male to female ratios among those receiving renal replacement therapy in Thailand and will require further analysis.
We showed that both stage G3aA1 and higher stages of CKD increased with age, but the absolute numbers and relative proportions depend on the equations used. Young and middle age subjects with stage 3aA1 have been shown to be at increased risks of renal events, cardiac deaths and all cause mortality [4, 24]. Among older patients, similar increases in risk for renal events were found, but the effects on cardiovascular deaths were less consistent. In the oldest age group, nearly all the differences CKD 3-5 prevalence between S-MDRD, CKD-EPI and C-MDRD could be accounted for by differences in stage 3aA1. Clearly the magnitude of attributed risks for developing cardio-renal complications for this stage will depend on the equation used to stage CKD.
Differences between the Asian equations may be due to differences in GFR measurement methods, creatinine calibration or true differences in the study populations [25, 26]. J-EPI is now the equation of choice in the Japanese population . Muscle mass is a major determinant in the relationship between serum creatinine and true GFR . Muscle mass is lower in Japanese compared to American subjects [26, 27]. This is consistent with a coefficient of < 1 for the Japanese equation, but still cannot account for the very high prevalence of CKD when this equation is applied to other Asian populations. Indeed, when applied to the Thai population as in this study, the prevalence of CKD is far too high for J-EPI equation to be accurate.
This study employed IDMS-standardized enzymatic method to measure serum creatinine similar to S-MDRD, CKD-EPI, T-GFR and J-EPI. On the other hand, C-MDRD was derived from Jaffe method after calibration to the Cleveland clinic laboratory, which was used to develop the MDRD equation . By applying a correction factor to adjust for standardized enzymatic creatinine and Cleveland clinic Jaffe creatinine, systematic differences between the two methods can be minimized . Nonetheless, creatinine method differences could account for some differences between C-MDRD and other equations .
Methods to measure the reference GFR may contribute to the differences between equations [28, 29]. Iothalamate clearance, used in the S-MDRD study, has been shown to overestimate GFR when compared to standard inulin clearance [30, 31] and this could contribute to the lower Japanese correction coefficient. Plasma clearance of 99mTc-DTPA was used as a reference GFR method in C-MDRD and T-GFR [9, 14]. DTPA could overestimate GFR when compared to inulin clearance . In addition, Chinese and Thai studies employed quite short clearance protocols which could further contribute to overestimation of GFR compared to inulin  The similarities for the T-GFR and C-MDRD equations may in part reflect their common use of DTPA as the reference method [9, 14]. Slight differences in protocol may contribute to the different coefficients. Both Chinese, Japanese, and Thai studies included only CKD patients, and hence it is uncertain how well these formulae to can be applied to normal subjects.
The limitations of this study include the fact that proteinuria was defined by dipstick. Although previous studies have shown that dipstick proteinuria provide similar risk prediction as albumin excretion , the use of albumin to creatinine ratio would have allowed the inclusion of those with microalbuminuria, and lead to more accurate staging and risk assessment for our subjects. Secondly, this study was designed to collect cardiovascular risk factors, and hence there is limited data on subjects with known kidney diseases beyond the presence of diabetes and hypertension.
Asians have among the highest rates of ESRD in the world [6, 7]. The Thai Renal Replacement Registry data showed the steep increase of renal replacement prevalence from 302.6 per million populations in 2006 to 496.9 in 2008. Nonetheless, accurate estimation of CKD prevalence remains a problem for Asian populations. The prevalence of CKD3-5 observed in our study using S-MDRD was fairly high, but is comparable to the 15% observed in other population surveys from younger subjects in Thailand, in which S-MDRD was used to classify CKD [20, 32]. Such high rate of CKD is a concern. It is uncertain if the high rates reflect inappropriate application of the Caucasian equations to the Thai population or a genuine increase in CKD in our population. It remains unclear which equation should be used to classify CKD in Thai or other Asian subjects. Although T-GFR was developed in Thai CKD subjects, T-GFR equation may not be the ideal equation in our population since there may be a bias especially among those with lower GFR. For example, the creatinine value consistent with a GFR of 5 ml/min/1.73 m2 in a 50 year old male would be 11 mg/dl and 30 mg/dl by S-MDRD and T-GFR studies . Nonetheless, by reclassifying our patients with the T-GFR or C-MDRD will lead to a reduction of those with CKD 3-5 by 2 to 3 folds with the impact greatest among the elderly group.