435 Background: Current American Urologic Association (AUA) and National Comprehensive Cancer Network (NCCN) guidelines do not differentiate between clear cell RCC and non–clear cell RCC (nccRCC), while European Urologic Association (EAU) does. Long-term recurrence data for nccRCC remains limited. We analyzed recurrence patterns after surgery for nccRCC and compared the prognostic performance of the guidelines. Methods: We identified 712 adults treated surgically for non-metastatic, non-hereditary nccRCC (2003-2015). Eligible histologies included papillary, chromophobe, translocation, mucinous tubular/spindle cell carcinoma, and unclassified RCC. Clinicopathologic factors, recurrence-free survival (RFS) and first-recurrence were recorded. Patients were classified by guideline risk categories. Model discrimination was assessed by concordance index (C-index). Model calibration was assessed by integrated Brier score (IBS), where lower values indicated better calibration and predictive accuracy. Results: Papillary RCC was the most common (420/712, 59%). Median follow-up was 120 months (IQR 108, 126). Recurrences were seen in 116 patients (16%) at a median of 26 months (IQR 9, 63). Intra-abdominal visceral organs or retroperitoneal lymph nodes were the most common sites of extra-renal recurrence (44/116, 38%), followed by lung (22/116, 19%). 5-year RFS was 85.8% (95% CI 83-89). When adjusted for stage, chromophobe was associated with lowest risk of recurrence (HR 0.20, 95% CI 0.11, 0.37). Most patients were AUA/EAU low risk (Table). RFS was significantly different by risk groups across all guidelines (Log-rank p5 years post-op) were common (31/116, 27%), particularly among intermediate risk patients (Table). Conclusions: Recurrence after surgery for nccRCC is uncommon but its patterns of recurrence are unique. Contrary to ccRCC, nccRCC tended to recur in the abdomen more often than in the chest, which should guide imaging modality choice for surveillance. Late recurrences were common, highlighting the need for careful long-term surveillance. All three national guidelines stratify risk, but the AUA model may provide the most accurate discrimination and calibration. N = 712 (%) Recurrences (%) Late Recurrences (%) Time to 95% of Observed Recurrences (months) AUA Low Risk 303 (43) 27/303 (9) 26 144 AUA Intermediate Risk 241 (34) 27/241 (11) 59 159 AUA High Risk 115 (16) 25/115 (22) 20 113 AUA Very High Risk 53 (7) 37/53 (70) 8 90 EAU Low Risk 307 (43) 27/307 (9) 26 144 EAU Intermediate Risk 173 (24) 17/173 (10) 71 226 EAU High Risk 232 (33) 72/232 (31) 17 111 NCCN Stage 1 476 (67) 44/476 (9) 43 152 NCCN Stage 2 74 (10) 13/74 (18) 39 159 NCCN Stage 3/4 162 (23) 59/162 (36) 12 111
Das et al. (Sun,) studied this question.