2015 Volume 3 Issue 1 Pages 50-57
Post-transplant renal graft function can be affected by a variety of recipient, donor, and immunological factors. Although the formula to calculate estimated glomerular filtration rate (eGFR) is well known, there are limited reports of formulas used to estimate post-transplant renal function. We have been conducting a clinical study to construct a prediction formula of postoperative Cr level based on pre/peri-operative clinical data of a donor and a recipient. Clinical data (training data) of 176 pairs of live kidney transplant (LKTx) donors and recipients who showed uneventful clinical courses, out of 470 pairs of LKTx from 2007 to 2012, were analyzed, and a prediction model of a recipient’s serum creatinine (Cr) level on postoperative day 7 (POD 7) was developed with symbolic regression. It was expressed with 10 variables including donor’s age, body weight, serum Cr level and 2 hours creatinine clearance before operation, and recipient’s age, body weight (or dry weight), gender, total ischemic time, warm ischemic time and graft weight. The formula was first applied to the training data and the average absolute difference between a predicted Cr (pCr) and an actual Cr (aCr) was 0.184±0.164 mg/dL. It was then applied to clinical data (future data) of 82 uneventful LKTx pairs, out of 89 LKTx pairts conducted in 2013. The average error of the absolute difference between pCr and aCr was 0.228±0.196 (range:0.0098 ~ 1.06, median 0.175) mg/dL. In most cases, the aCr was almost equal to or less than the pCr. For most cases, the prediction model showed a reasonable estimate or a slightly higher level of Cr at POD 7. Therefore, it could be used as an upper threshold of Cr level or “standard Cr level” for expected normal graft function. If aCr is much higher than pCr, it might be indicative of renal graft complications. We conclude that predicted Cr level at POD 7 might serve for early detection of rejection or other renal transplant complications.