Introduction:

Radical and partial nephrectomy (RN/PN) remain essential for the treatment of localized renal cell carcinoma (RCC). PN is generally preferred due to better functional outcomes, but RN is particularly relevant when increased oncologic potential is suggested based on large tumor size, concerning imaging findings, or aggressive histology. Recent studies suggest that patients with new baseline GFR (NBGFR)>45ml/min/1.73m² after RN/PN have strong survival characteristics, similar to patients without chronic kidney disease. Accordingly, accurate prediction of NBGFR can have significant implications for RCC management, including challenging decisions about RN vs. PN and patient counseling in terms of expected long-term prognosis. Current models to predict NBGFR post-RN are methodologically complex, demonstrate only moderate predictive ability, and omit two pertinent functional parameters: split-renal-function (SRF) and renal-functional-compensation (RFC). We propose a conceptually simple SRF-based model that predicts NBGFR based on these parameters, and compare its predictive accuracy with previously published non-SRF-based models.

Methods:

All 236 RCC patients who underwent RN(2010-2012) at Cleveland Clinic with available preoperative CT/MRI imaging and preoperative/postoperative measures of renal function were analyzed. NBGFR was defined as GFR 3-12 months post-RN, and RFC (% change in renal function of the preserved kidney) was estimated to be 24% based on preliminary analyses from independent datasets. SRF was determined using semi-automated software that provides differential parenchymal-volume-analysis (PVA) from preoperative imaging(FUJIFILM-Medical-Systems). Our proposed SRF-based model was: Predicted NBGFR=Global GFR_Pre-RN×SRF×1.24¹. The previously published non-SRF-based models for comparison were: 1)Predicted NBGFR=17+preoperative GFR(×0.65)–age(×0.25)+3(tumor>7cm)–2(diabetes)²; 2)Predicted NBGFR=preoperative GFR–(18.60–0.38×age+0.30×weight(kg)³; and 3)Predicted NBGFR=34.46–0.22×age+0.45×preoperative GFR–8.69(if diabetes)⁴. Predictive accuracy for each model was assessed based on alignment between predicted and observed NBGFR. The correlation coefficients (r) were obtained from linear regression analyses, and bootstrapping was conducted to compare r for the proposed SRF-based model with each of the non-SRF-based models.

Results:

The r value for the SRF-based model was 0.86 (95% CI: 0.82-0.90; Figure 1A). For the non-SRF-based models, the r values were 0.71 (95% CI: 0.65-0.77; Model 1), 0.55 (95% CI: 0.44-0.64; Model 2), and 0.68 (95% CI: 0.61-0.75; Model 3) (Figure 1B-D, respectively). The differences in r values between the SRF-based model and the non-SRF-based models were found to be statistically significant (all p≤0.003), as detailed in the legend to Figure 1.

Conclusion:

We propose a novel model based on SRF and RFC to predict NBGFR after RN that can facilitate counselling about expected postoperative renal functional outcomes and to facilitate difficult decisions regarding RN vs. PN for localized RCC. This SRF-based model provides a conceptually simple, accurate, and clinically implementable approach to predict NBGFR after RN. SRF can now be determined at point-of-care using PVA software that is affordable, readily available, and more accurate than nuclear renal scans (FUJIFILM-Medical-Systems). In addition to simplicity, this method of calculating NBGFR depends only on preoperative imaging and labs that patients routinely receive, and does not require chart review. The SRF-based model demonstrates greater predictive accuracy for NBGFR after RN than each of the three published non-SRF-based models.

Funding: N/A