Development of a Nomogram to Predict Progression-Free Survival in Patients with Locally Advanced Renal Cell Carcinoma
Renal cell carcinoma (RCC) is a significant malignancy, accounting for approximately 3% of all adult cancers. Among its various forms, locally advanced RCC presents a particularly challenging prognosis. Patients with this condition have a 5-year cancer-specific survival (CSS) rate ranging from 28% to 67% following curative surgery. Given the adverse outcomes associated with locally advanced RCC, accurate risk stratification for disease recurrence is crucial for optimizing patient management and treatment strategies.
Historically, several prognostic models have been developed to predict outcomes in patients with localized RCC. These models include the Stage, Size, Grade, and Necrosis (SSIGN) score, Leibovich score, Cindolo score, Yaycioglu score, Memorial Sloan Kettering Cancer Center (MSKCC) model, Kattan model, and Karakiewicz model. These models have demonstrated varying levels of predictive accuracy, with concordance indices (C-indices) ranging from 0.65 to 0.84. However, these models were primarily developed using data from low-risk RCC patients. When applied to intermediate or high-risk RCC patients in external prospective cohorts, their predictive performance significantly declined, with C-indices falling to a range of 0.587 to 0.69. This limitation underscores the need for a more accurate prognostic model tailored specifically to locally advanced RCC patients.
To address this gap, a retrospective study was conducted at Peking University Third Hospital, involving 759 patients who underwent nephrectomy for RCC between January 2015 and December 2017. After applying inclusion and exclusion criteria, 215 patients with locally advanced RCC were included in the final analytical cohort. These patients were classified according to the 8th edition of the Tumor-Node-Metastasis (TNM) classification, specifically those with T3-4N0M0 or T1-4N1M0 disease. Patients with bilateral or recurrent tumors, severe comorbidities, incomplete data, or those who underwent cytoreductive surgery were excluded.
The study collected comprehensive demographic and clinicopathological data, including sex, age, symptoms at presentation, body mass index, medical comorbidities, surgical approach, operative time, intraoperative blood loss, tumor side, tumor size, histologic subtype, nuclear grade, necrosis, sarcomatoid and rhabdoid 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 for the SSIGN, Leibovich, Cindolo, Yaycioglu, MSKCC, Kattan, and Karakiewicz models were calculated for each patient.
Disease progression was defined as any evidence of recurrence, metastasis, or tumor progression in pre-existing metastatic sites. Progression-free survival (PFS) was measured from the date of surgery to disease progression. Patients were advised against receiving postoperative adjuvant therapy until disease progression occurred.
The study cohort had a mean age of 59.33 ± 11.42 years, with a male predominance (70.2%). The median follow-up time was 36 months, during which 29.8% of patients experienced disease progression. The median PFS was 46 months. Univariable and multivariable Cox regression analyses were performed to identify independent prognostic factors. In the multivariable analysis, nuclear grade (hazard ratio [HR]: 1.892 for grades III–IV, P = 0.019), lymph node invasion (HR: 3.817, P = 0.004), and venous tumor thrombus (HR: 1.809, P = 0.036) emerged as significant independent predictors of PFS. Although symptoms at presentation (HR: 1.622, P = 0.080) did not reach statistical significance, they were retained in the model due to their clinical relevance.
Based on these findings, a nomogram was developed to predict the probability of 2-, 3-, and 4-year PFS. The nomogram incorporated four variables: symptoms at presentation, nuclear grade, venous tumor thrombus, and lymph node invasion. The performance of the nomogram was evaluated using the C-index, which ranged from 0.751 to 0.783, indicating good discrimination. Calibration curves, generated using a 1000-bootstrap resampling method, demonstrated strong agreement between predicted and observed outcomes.
When compared to existing prognostic models, the newly developed nomogram exhibited superior predictive accuracy. The Karakiewicz model performed the best among the conventional models (C-index: 0.673–0.781), while the Kattan model performed the worst (C-index: 0.566–0.624). The nomogram’s C-index was consistently higher than those of the SSIGN, Leibovich, Cindolo, Yaycioglu, MSKCC, Kattan, and Karakiewicz models, highlighting its potential for improved risk stratification in locally advanced RCC patients.
The study also highlighted the limitations of the current TNM staging system in providing satisfactory risk stratification for locally advanced RCC. The pathologic heterogeneity of locally advanced RCC, which includes perirenal fat invasion, renal sinus invasion, urinary collecting system invasion, and segmental renal vein invasion (T3 stage), tumor invasion beyond Gerota fascia (T4 stage), and lymph node invasion (N1 stage), contributes to the variability in prognosis even within the same stage. The identification of nuclear grade, venous tumor thrombus, and lymph node invasion as independent prognostic factors underscores the need for more nuanced risk stratification tools.
Despite its strengths, the study has several limitations. First, the C-index of the nomogram, while higher than existing models, was only moderately predictive (0.751–0.783). This suggests that the complexity of prognostic factors in locally advanced RCC may not be fully captured by the four variables included in the nomogram. Future studies incorporating additional clinical and genetic parameters may further enhance predictive accuracy. Second, the nomogram was developed and validated using data from a single-center cohort. External validation in multicenter studies is necessary to confirm its generalizability.
In conclusion, this study developed a nomogram with good discrimination and calibration for predicting PFS in patients with locally advanced RCC. The nomogram, which incorporates symptoms at presentation, nuclear grade, venous tumor thrombus, and lymph node invasion, demonstrated superior predictive accuracy compared to existing models. This tool has the potential to improve risk stratification and guide clinical decision-making for patients with locally advanced RCC. Further research is needed to refine the nomogram and validate its utility in diverse patient populations.
doi.org/10.1097/CM9.0000000000001833
Was this helpful?
0 / 0