Correspondence-In reply-Accounting for body composition does not improve cystatin c estimation of GFR in diabetic subjects with CKD

We thank Rigalleau et al for their interest in our report and are encouraged that other researchers have begun applying our methods. As originally stated, our intention is to test the hypothesis that glomerular filtration rate (GFR) estimation using cystatin C (CysC) level is not independent of body composition, not to generate estimation equations for use in clinical practice. Consequently, it is not surprising that our equations, developed in 77 patients, perform poorly against the Modification of Diet in Renal Disease equation.1 The observed bias also may be caused by differences in methods, including different assays for CysC and creatinine,2 filtration markers,3 and different dual-energy X-ray absorptiometry scanners and/or software to measure lean body mass (LM).4 Thus, a fairer analysis of their data would be to generate equations similar to ours, but using their own constants. Furthermore, Rigalleau et al concluded that our equations performed poorly because, in their sample, serum CysC level did not correlate with LM. However, it is not clear whether Rigalleau et al first accounted for GFR. In our study, serum CysC level did not correlate with LM unless entered after GFR in multiple regression. Unaccountably, serum creatinine level did not correlate with LM in their data set either (possibly indicative of a type II error). We agree that in patients with diabetes, other factors may influence estimated GFR5 and LM may not accurately reflect body cell mass in patients with fluid overload (as originally stated by us). Models to determine body cell mass are being developed, but remain experimental6 or for research use only.7 Development of these models will enable more accurate evaluation of the utility of body composition for estimated GFR in patients with diabetes.