Construction of a nomogram predictive model for the accuracy of estimated glomerular filtration rate formulae in patients with renal cell carcinoma

1305 Objectives: To evaluate the applicability of several estimated glomerular filtration rate (eGFR) formulae in patients with renal cell carcinoma, and thereby build a nomogram to predict the reliability of eGFR. Methods: A total of 302 patients with renal cell carcinoma who underwent radionuclide renal dynamic imaging were collected in the first Hospital of Shanxi Medical University from June 2017 to August 2020. Serum creatinine (Scr), serum cystatin C (CysC), double plasma samples-GFR (tGFR) and the factors related to the accuracy of eGFR were recorded before operation. tGFR were compared with eGFR from the following nine formulae(Cockcroft-Gault, Modified MDRD formula for Chinese, Simplified MDRD formula, Chronic Kidney Disease Epidemiology (CKD-EPIScr, CKD-EPICysC, CKD-EPIScr-CysC), Wright and full age spectrum (FASScr, FASCysC, FASScr-CysC)). Bias, Accuracy, precision and agreement of eGFR formulae were examined. Univariate and Multivariate logistic regression were applied to identify characteristics associated with accurate eGFR and construct a predictive nomogram. Results: All eGFR formulae results tend to overestimate renal function. The FASScr-CysC and FASCysC exhibited lower deviation. Compared with other formulae, FASScr-CysC formula showed better concordance correlation coefficient (CCC=0.35), accuracy (P30=53.6) and precision (IQR=24.2). It was found that patients with complication had no significant impact on the results of filtered eGFR formula, and two group statistics analysis demonstrated that the applicability of the FASScr-CysC formula is better than other formulae. Univariate and Multivariate logistic regression showed that accuracy of FASScr-CysC was significantly correlated with eGFR, gender, BMI, CysC, tumor diameter and high-density lipoprotein cholesterol (P<0.05). A nomogram was constructed to help physicians to predict the reliability of accuracy of eGFR. The ROC curve (AUC=0.789), calibration curve and decision curve all indicated good accuracy and aggrement. Conclusions: Our results indicated that the error of eGFR by any formulae was common in evaluating the GFR of patients with renal cell carcinoma before operation. Moreover, FASScr-CysC was selected and suggested more suitable for GFR assessment. And our nomogram may assist urologists in decision-making when tGFR is needed.