Development of a Nomogram to Predict Progression-Free Survival in Patients with Locally Advanced Renal Cell Carcinoma
Ruo-Tao Xiao,Cheng Liu,Bin Yang,Wei He,Chu-Xiao Xu,Zhi-Gang Chen,Lu-Lin Ma,Yuan-Yuan Ji
DOI: https://doi.org/10.1097/cm9.0000000000001833
IF: 6.133
2021-01-01
Chinese Medical Journal
Abstract:To the Editor: Renal cell carcinoma (RCC) accounts for approximately 3% of all malignancies in adults.[1] Locally advanced RCC has a relatively distinct and adverse prognosis with a 5-year cancer specific survival (CSS) rate ranging from 28% to 67% after curative surgery.[2] It is important to accurately stratify the risk of disease recurrence for locally advanced RCC. Several prognostic models like Stage, Size, Grade, and Necrosis (SSIGN), Leibovich, Cindolo, Yaycioglu, Memorial Sloan Kettering Cancer Center (MSKCC), Kattan, and Karakiewicz have been developed to predict the prognosis of localized RCC and the concordance index (C-index) has been found to range from 0.65 to 0.84.[3,4] To our knowledge, these models have mainly been developed based on the data from low-risk patients with RCC. In an external prospective cohort comprising patients with intermediate or high-risk RCC, the performances of these models were found to be sharply decreased (C-index range from 0.587 to 0.69).[4] Thus, in this study, we aimed to develop a more accurate prognostic model for better risk stratification in locally advanced RCC. Patients were identified from a retrospective, single-center database consisting of 759 patients who had undergone nephrectomy for RCC from January 2015 to December 2017 at Peking University Third Hospital. The study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee. According to the 8th Tumor-Node-Metastasis (TNM) classification, patients with T3-4N0M0 or T1-4N1M0 RCC were defined as having locally advanced RCC. Thus, patients who were diagnosed with locally advanced RCC and underwent curative-intent surgery were included in our study cohort. Patients with bilateral or recurrent tumors, severe comorbidities, and incomplete data, and those who had undergone cytoreductive surgery were excluded. Based on the inclusion and exclusion criteria, a total of 215 patients were enrolled in our study and formed the analytical cohort. Patient demographic and clinicopathological data were collected, including sex, age, symptoms at presentation, body mass index, medical comorbidities, surgical approach, surgical time, inter-operative blood loss, tumor side, and tumor size, and pathologic data including histologic subtype, nuclear grade, necrosis, sarcomatoid and rheumatoid differentiation, lymphovascular invasion, renal sinus invasion, perirenal fat invasion, urinary collecting system invasion, venous tumor thrombus, lymph node invasion, and adrenal invasion. The total points of the SSIGN, Leibovich, Cindolo, Yaycioglu, MSKCC, Kattan, and Karakiewicz models were calculated for each patient. Disease progression was defined as any evidence of recurrence or metastasis, or tumor progression in the pre-existing metastatic site. Progression-free survival (PFS) was defined as the time from the date of surgery to disease progression. Patients were advised to not receive any post-operative adjuvant therapy until disease progression. Categorical variables are reported as numbers and proportions, and continuous variables are reported as mean ± standard deviations or median and interquartile range (IQR). The associations of relevant clinicopathological data with PFS were assessed using Cox proportional hazards regression models. The backward stepwise selection was used to identify the independent factors for multivariable Cox proportional hazards regression models. Selected independent factors and important clinical factors were incorporated in the nomograms to predict the probability of 2-, 3-, and 4-year PFS. The performance of the nomogram was measured using the C-index, which ranges from 0.5 (no predictive power) to 1 (perfect prediction). Calibration was evaluated using a calibration curve, which assessed the outcomes between the observed outcome probabilities and the nomogram-predicted probabilities with a bootstrapped resampling of 1000 times. Nomogram establishment and calibration were conducted using R software (Version 4.0.3) using the “rms” package, and other statistical analyses were performed using SPSS (Version 26, IBM, Armonk, NY, USA). All tests were two sided, and P < 0.05 was considered statistically significant. Patients’ characteristics are shown in [Supplementary Table 1, https://links.lww.com/CM9/A807]. The mean age of the 215 patients in this study was 59.33 ± 11.42 years and most patients were male (N = 151, 70.2%). The median time of follow-up was 36 months (IQR: 27–40 months). Sixty-four patients (29.8%) were found to have disease progression during follow-up. The median PFS of the study cohort was 46 months (IQR: 38–57 months). Univariable and multivariable Cox regression analysis is shown in Supplementary Table 2, https://links.lww.com/CM9/A807. In the multivariable analysis, nuclear grade (HR: 1.892 for III–IV, P = 0.019), lymph node invasion (HR: 3.817, P = 0.004), and venous tumor thrombus (HR: 1.809, P = 0.036) were independent factors. Symptoms at presentation (HR: 1.622, P = 0.080) remained in the multivariable model, although it was not statistically significant. Considering the clinical importance of the symptoms at presentation, we established a nomogram based on nuclear grade, lymph node invasion, venous tumor thrombus, and symptoms at presentation [Figure 1].Figure 1: The nomogram to predict PFS was created based on symptoms at present, nuclear grade, venous tumor thrombus, and lymph node invasion. PFS: Progression-free survival.The discrimination of the nomogram was relatively high and had a C-index of 0.751 to 0.783 for our cohort. Calibration curves were developed using a 1000 bootstrap resampling method and good agreement was found between the predicted and actual probabilities [Supplementary Figure 1, https://links.lww.com/CM9/A806]. The 2-, 3-, and 4-year predicted accuracy of our nomogram, SSIGN, Leibovich, Cindolo, Yaycioglu, MSKCC, Kattan, and Karakiewicz models is shown in Supplementary Table 3, https://links.lww.com/CM9/A807. The Karakiewicz model was found to perform the best (C-index: 0.673–0.781) and the Kattan model was determined to perform the worst (C-index: 0.566–0.624). Our nomogram exhibited better accuracy than the SSIGN, Leibovich, Cindolo, Yaycioglu, MSKCC, Kattan, and Karakiewicz models. In this study, we analyzed 215 consecutive patients diagnosed with locally advanced RCC and reported several noteworthy findings. First, we found that nuclear grade, venous tumor thrombus, and lymph node invasion were independent prognostic factors among these patients. Second, we established a nomogram that exhibited good discrimination and calibration when used in our study cohort. Third, compared with other conventional models, our nomogram was found to be slightly but significantly superior for risk stratification of locally advanced RCC. To the best of our knowledge, locally advanced RCC is pathologically heterogeneous, which groups perirenal fat invasion, renal sinus invasion, urinary collecting system invasion, and segmental renal vein invasion as T3 stage; tumor invasion beyond Gerota fascia as T4 stage; and lymph node invasion as N1 stage. Previous studies report that the prognosis varies depending on the pathologic features even if it is the same stage.[5] We found that among these, venous tumor thrombus and lymph node invasion were independent prognostic factors that correlated with PFS in locally advanced RCC. Besides, the nuclear grade was also an independent prognostic factor in our study cohort, which suggested that the current TNM staging system may not be sufficient to provide satisfactory risk stratification in patients with locally advanced RCC. Owing to the shortcomings of the current TNM staging system, several models have been reported to predict prognosis in patients with RCC. However, the performances of current models were found to sharply decrease in the prospective clinical trial cohort comprising patients with intermediate or high-risk localized RCC.[4] We speculated that in the development cohort, the patients were mainly grouped as low-risk RCC. Thus, in the intermediate or high-risk RCC cohorts, the predicted accuracy of these models was limited. Considering this speculation, we developed a nomogram based on a group of patients with locally advanced RCC. Notably, our nomogram was found to be superior compared with other conventional models. Besides, our nomogram only incorporated four simple variables including the symptoms at presentation, nuclear grade, venous tumor thrombus, and lymph node invasion, all of which were easy to obtain post-operatively. Our study has several limitations. First, the C-index of our nomogram was 0.751 to 0.783 and slightly higher than that of the other models, which suggested moderate, but not optimal, predicted accuracy of the model. Thus, the potential prognostic factors of locally advanced RCC were extremely complex and could not be thoroughly revealed based on these four simple factors. A multi-parameter model coupled with clinical and gene information should be investigated in future studies to improve the predicted accuracy of prognosis in these patients. Second, our nomogram was developed, and internal validation was performed based on data from a single-center cohort. Further studies are needed to externally validate the proposed nomograms. To summarize, we developed a nomogram with good discrimination and calibration to predict the prognosis of locally advanced RCC. Our nomogram was superior to other conventional models in predicting the 2-, 3-, and 4-year PFS, indicating its ability of better risk stratification in these patients. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed. Conflicts of interest None.