ABSTRACT
Objective:To explore the predictors of upper tract urothelial carcinoma (UTUC) invading muscle.Methods:The preoperative clinical data of 169 patients with UTUC who underwent radical nephroureterectomy (RNU) in the Department of Urology, Second Hospital of Tianjin Medical University from August 2019 to September 2021 were retrospectively collected. The average age of the patients was (67.7±8.3)years. There were 95 male cases (56.2%).15 cases (15.8%) were with diabetes, 46 cases (48.4%) with hypertension, 68 cases (71.6%) with hematuria, 44 cases (46.3%) with renal pelvis tumor, 74 cases (77.9%) with hydronephrosis and 8 cases (8.4%) with multifocality. There were 15 (20.3%), with diabetes, 40(54.1%) with hypertension, 58(78.4%)with hematuria, 32(43.2%) with renal pelvis tumor, 60(81.1%) with hydronephrosis and 2(2.7%) with multifocality among 74 female patients(43.8%), respectively. Patients’ body mass index (BMI) was (24.2±3.2) kg/m 2.The tumor diameter was (3.4±1.8) cm. The average count of neutrophil (NEU) was (4.2±1.7) ×10 9/L, with monocyte (MON) (0.4±0.2) ×10 9/L, platelet (PLT) (237.8±75.3) ×10 9/L and lymphocyte (LYM) (1.7±0.5) ×10 9/L. Neutrophil-to-lymphocyte ratio (NLR) was 2.7±1.4. Mmonocyte-to-lymphocyte ratio (MLR) was 0.3±0.1 and platelet-to-lymphocyte ratio (PLR) was 151.1±59.6. The concentrations of hemoglobin (Hb) and fibrinogen (FIB) were (127.1±18.3) g/L and(3.5±1.0) g/L. Univariate logistic regression analysis was performed on all preoperative data, and a nomogram was established to predict UTUC invading muscle by stepwise regression combined with multivariate logistic analysis. Concordance index (C-index) was used to evaluate the accuracy of the model, and Hosmer-Lemeshow test was used to check the fit. Using 1000 bootstrap resampling to validate the model and draw calibration plot. Receiver operating characteristics (ROC) curves were used to determine the most appropriate cut-off point of the model to predict tumor invasion and evaluate its sensitivity and specificity. Results:In this study, there were 169 cases. Postoperative pathological results showed that 18 cases (10.6%) were of low grade and 151 cases (89.4%) were of high grade. There were 2 patients (1.2%) with pT a stage, 44 patients (26.0%) with pT 1 stage, 37 patients (21.9%) with pT 2 stage, 81 patients (47.9%) with pT 3 stage and 5 patients (3.0%) with pT 4 stage. According to tumor stage, 46 patients were included into non-muscle invasive disease (NMID, pT a/T is/T 1) group and 123 patients were included into muscle invasive disease (MID, pT 2/T 3/T 4) group. Univariate analysis showed that hydronephrosis ( OR=2.919, 95% CI 1.332-6.387, P=0.007) was significantly correlated with MID. Tumor location in ureter ( OR=1.898, 95% CI 0.960-3.800, P=0.067), higher NLR value ( OR=1.313, 95% CI 0.9901.832, P=0.082) and FIB concentration ( OR=1.436, 95% CI 1.008-2.125, P=0.056) might have potential association with MID. Stepwise regression method was used to screen out PLT, FIB and hydronefrosis as the optimal combination for MID prediction. multivariate analysis suggested that PLT decreasing ( OR=0.727, 95% CI 0.548-0.955, P=0.023), FIB increasing ( OR=1.629, 95% CI 1.084-2.552, P=0.025) and hydronephrosis ( OR=2.500, 95% CI 1.111-5.601, P=0.026) were independent predictors of MID.The C-index of the model constructed by the three factors was 0.682, and the P-value of the goodness of fit was 0.778. The ROC curve showed that the optimal cut-off point for the prediction of high-risk MID by the nomogram was 0.627, achieving 39.1% specificity, 88.6% sensitivity and 0.682 area under the curve (AUC). Conclusions:Preoperative decrease of PLT, increase of FIB and hydronephrosis were independent factors for predicting UTUC invading muscle.