Risk models for recurrence and survival after kidney cancer: a systematic review

Development and External Validation of Nomograms for Predicting Disease-Free Survival and Overall Survival in Patients with cT1-ccRCC After Partial Nephrectomy: A Multicenter Retrospective Study

BACKGROUND

To develop a novel nomogram for predicting 2-year and 5-year disease-free survival (DFS) and overall survival (OS) in patients with cT1-clear cell renal cell carcinoma (ccRCC) undergoing partial nephrectomy (PN).

METHODS

A retrospective study was conducted across five urological centers, including 940 patients who underwent PN for cT1N0M0-ccRCC. Four centers were randomly selected to constitute the training group, while the remaining center served as the testing group. We employed the LASSO and multivariate Cox regression to develop new nomograms. The 1,000 bootstrap-corrected c-index, net reclassification improvement (NRI) and receiver operating characteristic curve were employed to compare the predictive abilities of new nomograms with the widely used UUIS and SSIGN models. Finally, the novel nomograms underwent external validation.

RESULTS

The training group included 714 patients, while the testing group consisted of 226 patients. The bootstrap-corrected c-indexes for the DFS and OS model were 0.870 and 0.902, respectively. In the training cohort, the AUC for the DFS and OS models at 2 years and 5 years were 0.953, 0.902, 0.988, and 0.911, respectively. These values were also assessed in the testing cohort. The predictive capabilities of the new nomograms surpassed those of the UUIS and SSIGN models (NRI > 0). Decision curve analysis demonstrated that the novel nomograms provide greater net benefits compared to the UUIS and SSIGN models.

CONCLUSIONS

Our novel nomograms demonstrated strong predictive ability for forecasting oncological outcomes in cT1-ccRCC patients after PN. These user-friendly nomograms are simple and convenient for clinical application, providing tangible clinical benefits.

Immunotherapy and stereotactic body radiotherapy for older patients with non-metastatic renal cancer unfit for surgery or decline nephrectomy: practical proposal by the International Geriatric Radiotherapy Group

The standard of care for non-metastatic renal cancer is surgical resection followed by adjuvant therapy for those at high risk for recurrences. However, for older patients, surgery may not be an option due to the high risk of complications which may result in death. In the past renal cancer was considered to be radio-resistant, and required a higher dose of radiation leading to excessive complications secondary to damage of the normal organs surrounding the cancer. Advances in radiotherapy technique such as stereotactic body radiotherapy (SBRT) has led to the delivery of a tumoricidal dose of radiation with minimal damage to the normal tissue. Excellent local control and survival have been reported for selective patients with small tumors following SBRT. However, for patients with poor prognostic factors such as large tumor size and aggressive histology, there was a higher rate of loco-regional recurrences and distant metastases. Those tumors frequently carry program death ligand 1 (PD-L1) which makes them an ideal target for immunotherapy with check point inhibitors (CPI). Given the synergy between radiotherapy and immunotherapy, we propose an algorithm combining CPI and SBRT for older patients with non-metastatic renal cancer who are not candidates for surgical resection or decline nephrectomy.

UroPredict: Machine learning model on real-world data for prediction of kidney cancer recurrence (UroCCR-120)

Renal cell carcinoma (RCC) is most often diagnosed at a localized stage, where surgery is the standard of care. Existing prognostic scores provide moderate predictive performance, leading to challenges in establishing follow-up recommendations after surgery and in selecting patients who could benefit from adjuvant therapy. In this study, we developed a model for individual postoperative disease-free survival (DFS) prediction using machine learning (ML) on real-world prospective data. Using the French kidney cancer research network database, UroCCR, we analyzed a cohort of surgically treated RCC patients. Participating sites were randomly assigned to either the training or testing cohort, and several ML models were trained on the training dataset. The predictive performance of the best ML model was then evaluated on the test dataset and compared with the usual risk scores. In total, 3372 patients were included, with a median follow-up of 30 months. The best results in predicting DFS were achieved using Cox PH models that included 24 variables, resulting in an iAUC of 0.81 [IC95% 0.77-0.85]. The ML model surpassed the predictive performance of the most commonly used risk scores while handling incomplete data in predictors. Lastly, patients were stratified into four prognostic groups with good discrimination (iAUC = 0.79 [IC95% 0.74-0.83]). Our study suggests that applying ML to real-world prospective data from patients undergoing surgery for localized or locally advanced RCC can provide accurate individual DFS prediction, outperforming traditional prognostic scores.

Predicting recurrence and survival in patients with non-metastatic renal-cell carcinoma after nephrectomy: a prospective population-based study with multicenter validation

BACKGROUND

Accurate prognostication of oncological outcomes is crucial for the optimal management of patients with renal cell carcinoma (RCC) after surgery. Previous prediction models were developed mainly based on retrospective data in the Western populations, and their predicting accuracy remains limited in contemporary, prospective validation. We aimed to develop contemporary RCC prognostic models for recurrence and overall survival (OS) using prospective population-based patient cohorts and compare their performance with existing, mostly utilized ones.

METHODS

In this prospective analysis and external validation study, the development set included 11  128 consecutive patients with non-metastatic RCC treated at a tertiary urology center in China between 2006 and 2022, and the validation set included 853 patients treated at 13 medical centers in the USA between 1996 and 2013. The primary outcome was progression-free survival (PFS), and the secondary outcome was OS. Multivariable Cox regression was used for variable selection and model development. Model performance was assessed by discrimination [Harrell's C-index and time-dependent areas under the curve (AUC)] and calibration (calibration plots). Models were validated internally by bootstrapping and externally by examining their performance in the validation set. The predictive accuracy of the models was compared with validated models commonly used in clinical trial designs and with recently developed models without extensive validation.

RESULTS

Of the 11  128 patients included in the development set, 633 PFS and 588 OS events occurred over a median follow-up of 4.3 years [interquartile range (IQR) 1.7-7.8]. Six common clinicopathologic variables (tumor necrosis, size, grade, thrombus, nodal involvement, and perinephric or renal sinus fat invasion) were included in each model. The models demonstrated similar C-indices in the development set (0.790 [95% CI 0.773-0.806] for PFS and 0.793 [95% CI 0.773-0.811] for OS) and in the external validation set (0.773 [0.731-0.816] and 0.723 [0.731-0.816]). A relatively stable predictive ability of the models was observed in the development set (PFS: time-dependent AUC 0.832 at 1 year to 0.760 at 9 years; OS: 0.828 at 1 year to 0.794 at 9 years). The models were well calibrated and their predictions correlated with the observed outcome at 3, 5, and 7 years in both development and validation sets. In comparison to existing prognostic models, the present models showed superior performance, as indicated by C-indices ranging from 0.722 to 0.755 (all P <0.0001) for PFS and from 0.680 to 0.744 (all P <0.0001) for OS. The predictive accuracy of the current models was robust in patients with clear-cell and non-clear-cell RCC.

CONCLUSIONS

Based on a prospective population-based patient cohort, the newly developed prognostic models were externally validated and outperformed the currently available models for predicting recurrence and survival in patients with non-metastatic RCC after surgery. The current models have the potential to aid in clinical trial design and facilitate clinical decision-making for both clear-cell and non-clear-cell RCC patients at varying risk of recurrence and survival.

A retrospective cohort study of the impact of peripheral blood gamma- delta T cells to prognosis of nonmetastatic renal cell cancer after curative resection

AIM

Gamma-delta-T cells (γδT) have potential antitumor roles and have recently been applied in adoptive immunotherapy. In the present study, we focused on the proportion of γδT cells in the peripheral blood just before surgery for renal cell cancer (RCC) and investigated whether their proportion affected recurrence-free survival (RFS) and overall survival (OS) retrospectively.

PATIENTS AND METHODS

A total of 137 patients with localized, non-metastatic RCC who received surgery at our institutes were analyzed retrospectively. The patients were divided into 2 groups: normal and low γδT cell groups based on the proportion of peripheral blood γδT cells. Kaplan-Meier curves were constructed to access the association of the proportion of peripheral blood γδT cells to RFS and OS. Cox regression were also constructed to access the risks to prognosis. Uni- and multivariate logistic regressions were performed to access associations between risk factors and, RFS and OS.

RESULTS

Among 137 patients, 40 had a proportion of γδT cells in peripheral blood of less than 1%, which was below the normal range. The remaining 97 patients had these cells in peripheral blood at 1% or higher. In the groups with low γδT cells, 13 patients had recurrences, and 9 patients dies during the observation period. In the groups with normal γδT cells, 16 patients had recurrences, and 8 patients died. The normal γδT cell group demonstrated significantly better prognosis in terms of RFS and OS. Multivariate analysis revealed that a low hemoglobin level, a low proportion of γδT cells, and a high pathological T stage (pT) were statistically independent risk factors for RFS. Age, albumin, C-reactive protein (CRP), % γδT cells, and pT were statistically significant factors affecting OS and only pT was an independent risk factor by multivariate analysis.

CONCLUSION

A low proportion of γδT cells was identified as one of the risk factors for RFS. Our findings will provide clues to develop strategies for early intervention in preventing recurrence even after complete resection of RCC and, such as adoptive immunotherapy using autologous γδT cells in patients with a low proportion of γδT cells.

Adjuvant immunotherapy in renal cell carcinoma: a systematic review and meta-analysis

OBJECTIVES

To synthesise available data regarding the disease-free survival (DFS) benefit of adjuvant immune checkpoint inhibitors (ICIs) for patients with renal cell carcinoma (RCC) and evaluate the overall safety profile of ICIs in this setting.

MATERIALS AND METHODS

We utilised PubMed, Embase, and relevant conference proceedings to identify phase III randomised controlled trials comparing adjuvant ICIs vs placebo/observation for RCC. The primary outcome of interest was DFS. Variables for subgroup analyses were programmed death-ligand 1 (PD-L1) expression, sarcomatoid features, nephrectomy type, and disease-risk category. Secondary outcomes included Grade ≥3 adverse events (AEs), immune-related AEs, and treatment discontinuation due to AEs. All outcomes were analysed using random-effects models owing to inter-study heterogeneity.

RESULTS

Among the four included studies, one demonstrated a significant DFS benefit. There was considerable clinical and statistical heterogeneity (I2  = 64%) due to differences in inclusion criteria and interventions. While pooled results across the four studies did not demonstrate a significant benefit in DFS overall (hazard ratio [HR] 0.85, 95% confidence interval [CI] 0.69-1.04) there was significant benefit among patients with positive PD-L1 expression (HR 0.72, 95% CI 0.55-0.94) and sarcomatoid features (HR 0.59, 95% CI 0.38-0.91).

CONCLUSION

The evidence base to date regarding ICIs as adjuvant therapy in RCC is mixed - conclusions are limited by considerable heterogeneity between studies. However, pooled analyses suggest that patients with positive PD-L1 expression or sarcomatoid features are most likely to benefit from adjuvant immunotherapy.

Critical Appraisal of Leibovich 2018 and GRANT Models for Prediction of Cancer-Specific Survival in Non-Metastatic Chromophobe Renal Cell Carcinoma

Within the Surveillance, Epidemiology, and End Results database (2000–2019), we identified 5522 unilateral surgically treated non-metastatic chromophobe kidney cancer (chRCC) patients. This population was randomly divided into development vs. external validation cohorts. In the development cohort, the original Leibovich 2018 and GRANT categories were applied to predict 5- and 10-year cancer-specific survival (CSS). Subsequently, a novel multivariable nomogram was developed. Accuracy, calibration and decision curve analyses (DCA) tested the Cox regression-based nomogram as well as the Leibovich 2018 and GRANT risk categories in the external validation cohort. The accuracy of the Leibovich 2018 and GRANT models was 0.65 and 0.64 at ten years, respectively. The novel prognostic nomogram had an accuracy of 0.78 at ten years. All models exhibited good calibration. In DCA, Leibovich 2018 outperformed the novel nomogram within selected ranges of threshold probabilities at ten years. Conversely, the novel nomogram outperformed Leibovich 2018 for other values of threshold probabilities. In summary, Leibovich 2018 and GRANT risk categories exhibited borderline low accuracy in predicting CSS in North American non-metastatic chRCC patients. Conversely, the novel nomogram exhibited higher accuracy. However, in DCA, all examined models exhibited limitations within specific threshold probability intervals. In consequence, all three examined models provide individual predictions that might be suboptimal and be affected by limitations determined by the natural history of chRCC, where few deaths occur within ten years from surgery. Further investigations regarding established and novel predictors of CSS and relying on large sample sizes with longer follow-up are needed to better stratify CSS in chRCC.

Local Treatment of Recurrent Renal Cell Carcinoma May Have a Significant Survival Effect Across All Risk-of-recurrence Groups

Background

Retrospective comparative studies suggest a survival benefit after complete local treatment of recurrence (LTR) in renal cell carcinoma (RCC), which may be largely due to an indication bias.

Objective

To determine the role of LTR in a homogeneous population characterised by limited and potentially resectable recurrence.

Design setting and participants

RECUR is a protocol-based multicentre European registry capturing patient and tumour characteristics, risk of recurrence (RoR), recurrence patterns, and survival of those curatively treated for nonmetastatic RCC from 2006 to 2011. Per-protocol resectable disease (RD) recurrence was defined as (1) solitary metastases, (2) oligometastases, or (3) renal fossa or renal recurrence after radical or partial nephrectomy, respectively.

Intervention

Local treatment of recurrence.

Outcome measurements and statistical analysis

Overall survival (OS) and cancer-specific survival was compared in the RD population that underwent LTR versus no LTR. We constructed a multivariate model to predict risk factors for overall mortality and analysed the effect of LTR across RoR groups.

Results and limitations

Of 3039 patients with localised RCC treated with curative intent, 505 presented with recurrence, including 176 with RD. Of these patients, 97 underwent LTR and 79 no LTR. Patients in the LTR group were younger (64.3 [40-80] vs 69.2 [45-87] yr; p = 0.001). The median OS was 70.3 mo (95% confidence interval [CI] 58-82.6) versus 27.4 mo (95% CI 23.6-31.15) in the LTR versus no-LTR group (p < 0.001). After a multivariate analysis, having LTR (hazard ratio [HR] 0.37 [95% CI 0.2-0.6]), having low- versus high-risk RoR (HR 0.42 [95% CI [0.20-0.83]), and not having extra-abdominal/thoracic metastasis (HR 1.96 [95% CI 1.02-3.77]) were prognostic factors of longer OS. The LTR effect on survival was consistent across risk groups. OS HR for high, intermediate, and low risks were 0.36 (0.2-0.64), 0.27 (0.11-0.65), and 0.26 (0.08-0.8), respectively. Limitations include retrospective design.

Conclusions

This is the first study assessing the effectiveness of LTR in RCC in a comparable population with RD. This study supports the role of LTR across all RoR groups.

Patient summary

We assessed the effectiveness of local treatment of resectable recurrent renal cell carcinoma after surgical treatment of the primary kidney tumour. Local treatment of recurrence was associated with longer survival across groups with a risk of recurrence.

Application of Machine Learning Models to Predict Recurrence After Surgical Resection of Nonmetastatic Renal Cell Carcinoma

BACKGROUND

Predictive tools can be useful for adapting surveillance or including patients in adjuvant trials after surgical resection of nonmetastatic renal cell carcinoma (RCC). Current models have been built using traditional statistical modelling and prespecified variables, which limits their performance.

OBJECTIVE

To investigate the performance of machine learning (ML) framework to predict recurrence after RCC surgery and compare them with current validated models.

DESIGN, SETTING, AND PARTICIPANTS

In this observational study, we derived and tested several ML-based models (Random Survival Forests [RSF], Survival Support Vector Machines [S-SVM], and Extreme Gradient Boosting [XG boost]) to predict recurrence of patients who underwent radical or partial nephrectomy for a nonmetastatic RCC, between 2013 and 2020, at 21 French medical centres.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

The primary end point was disease-free survival. Model discrimination was assessed using the concordance index (c-index), and calibration was assessed using the Brier score. ML models were compared with four conventional prognostic models, using decision curve analysis (DCA).

RESULTS AND LIMITATIONS

A total of 4067 patients were included in this study (3253 in the development cohort and 814 in the validation cohort). Most tumours (69%) were clear cell RCC, 40% were of high grade (nuclear International Society of Urological Pathology grade 3 or 4), and 24% had necrosis. Of the patients, 4% had nodal involvement. After a median follow-up of 57 mo (interquartile range 29-76), 523 (13%) patients recurred. ML models obtained higher c-index values than conventional models. The RSF yielded the highest c-index values (0.794), followed by S-SVM (c-index 0.784) and XG boost (c-index 0.782). In addition, all models showed good calibration with low integrated Brier scores (all integrated brier scores <0.1). However, we found calibration drift over time for all models, albeit with a smaller magnitude for ML models. Finally, DCA showed an incremental net benefit from all ML models compared with conventional models currently used in practice.

CONCLUSIONS

Applying ML approaches to predict recurrence following surgical resection of RCC resulted in better prediction than that of current validated models available in clinical practice. However, there is still room for improvement, which may come from the integration of novel biological and/or imaging biomarkers.

PATIENT SUMMARY

We found that artificial intelligence algorithms could better predict the risk of recurrence after surgery for a localised kidney cancer. These algorithms may help better select patients who will benefit from medical treatment after surgery.

The Four-Feature Prognostic Models for Cancer-Specific and Overall Survival after Surgery for Localized Clear Cell Renal Cancer: Is There a Place for Inflammatory Markers?

We aimed at a determination of the relevance of comorbidities and selected inflammatory markers to the survival of patients with primary non-metastatic localized clear cell renal cancer (RCC). We retrospectively analyzed data from a single tertiary center on 294 patients who underwent a partial or radical nephrectomy in the years 2012–2018. The following parameters were incorporated in the risk score: tumor stage, grade, size, selected hematological markers (SIRI—systemic inflammatory response index; SII—systemic immune-inflammation index) and a comorbidities assessment tool (CCI—Charlson Comorbidity Index). For further analysis we compared our model with existing prognostic tools. In a multivariate analysis, tumor stage (p = 0.01), tumor grade (p = 0.03), tumor size (p = 0.006) and SII (p = 0.02) were significant predictors of CSS, while tumor grade (p = 0.02), CCI (p = 0.02), tumor size (p = 0.01) and SIRI (p = 0.03) were significant predictors of OS. We demonstrated that our model was characterized by higher accuracy in terms of OS prediction compared to the Leibovich and GRANT models and outperformed the GRANT model in terms of CSS prediction, while non-inferiority to the VENUSS model was revealed. Four different features were included in the predictive models for CSS (grade, size, stage and SII) and OS (grade, size, CCI and SIRI) and were characterized by adequate or even superior accuracy when compared with existing prognostic tools.