Advanced

Risk prediction for renal cell Carcinoma : Results from the European Prospective Investigation into Cancer and nutrition (EPIC) prospective cohort study

Singleton, Rosie K. ; Heath, Alicia K. ; Clasen, Joanna L. ; Scelo, Ghislaine ; Johansson, Mattias ; Le Calvez-Kelm, Florence ; Weiderpass, Elisabete ; Liedberg, Fredrik LU ; Ljungberg, Borje and Harbs, Justin , et al. (2021) In Cancer Epidemiology Biomarkers and Prevention 30(3). p.507-512
Abstract

Background: Early detection of renal cell carcinoma (RCC) has the potential to improve disease outcomes. No screening program for sporadic RCC is in place. Given relatively low incidence, screening would need to focus on people at high risk of clinically meaningful disease so as to limit overdiagnosis and screen-detected false positives. Methods: Among 192,172 participants from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (including 588 incident RCC cases), we evaluated a published RCC risk prediction model (including age, sex, BMI, and smoking status) in terms of discrimination (C-statistic) and calibration (observed probability as a function of predicted probability). We used a flexible parametric... (More)

Background: Early detection of renal cell carcinoma (RCC) has the potential to improve disease outcomes. No screening program for sporadic RCC is in place. Given relatively low incidence, screening would need to focus on people at high risk of clinically meaningful disease so as to limit overdiagnosis and screen-detected false positives. Methods: Among 192,172 participants from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (including 588 incident RCC cases), we evaluated a published RCC risk prediction model (including age, sex, BMI, and smoking status) in terms of discrimination (C-statistic) and calibration (observed probability as a function of predicted probability). We used a flexible parametric survival model to develop an expanded model including age, sex, BMI, and smoking status, with the addition of self-reported history of hypertension and measured blood pressure. Results: The previously published model yielded well-calibrated probabilities and good discrimination (C-statistic [95% CI]: 0.699 [0.679–0.721]). Our model had slightly improved discrimination (0.714 [0.694–0.735], bootstrap optimism-corrected C-statistic: 0.709). Despite this good performance, predicted risk was low for the vast majority of participants, with 70% of participants having 10-year risk less than 0.0025. Conclusions: Although the models performed well for the prediction of incident RCC, they are currently insufficiently powerful to identify individuals at substantial risk of RCC in a general population. Impact: Despite the promising performance of the EPIC RCC risk prediction model, further development of the model, possibly including biomarkers of risk, is required to enable risk stratification of RCC.

(Less)
Please use this url to cite or link to this publication:
@article{f7f3c57b-82e0-4f47-98fa-269a14905f64,
  abstract     = {<p>Background: Early detection of renal cell carcinoma (RCC) has the potential to improve disease outcomes. No screening program for sporadic RCC is in place. Given relatively low incidence, screening would need to focus on people at high risk of clinically meaningful disease so as to limit overdiagnosis and screen-detected false positives. Methods: Among 192,172 participants from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (including 588 incident RCC cases), we evaluated a published RCC risk prediction model (including age, sex, BMI, and smoking status) in terms of discrimination (C-statistic) and calibration (observed probability as a function of predicted probability). We used a flexible parametric survival model to develop an expanded model including age, sex, BMI, and smoking status, with the addition of self-reported history of hypertension and measured blood pressure. Results: The previously published model yielded well-calibrated probabilities and good discrimination (C-statistic [95% CI]: 0.699 [0.679–0.721]). Our model had slightly improved discrimination (0.714 [0.694–0.735], bootstrap optimism-corrected C-statistic: 0.709). Despite this good performance, predicted risk was low for the vast majority of participants, with 70% of participants having 10-year risk less than 0.0025. Conclusions: Although the models performed well for the prediction of incident RCC, they are currently insufficiently powerful to identify individuals at substantial risk of RCC in a general population. Impact: Despite the promising performance of the EPIC RCC risk prediction model, further development of the model, possibly including biomarkers of risk, is required to enable risk stratification of RCC.</p>},
  author       = {Singleton, Rosie K. and Heath, Alicia K. and Clasen, Joanna L. and Scelo, Ghislaine and Johansson, Mattias and Le Calvez-Kelm, Florence and Weiderpass, Elisabete and Liedberg, Fredrik and Ljungberg, Borje and Harbs, Justin and Olsen, Anja and Tjønneland, Anne and Dahm, Christina C. and Kaaks, Rudolf and Fortner, Renee T. and Panico, Salvatore and Tagliabue, Giovanna and Masala, Giovanna and Tumino, Rosario and Ricceri, Fulvio and Gram, Inger T. and Santiuste, Carmen and Bonet, Catalina and Rodriguez-Barranco, Miguel and Schulze, Mattias B. and Bergmann, Manuela M. and Travis, Ruth C. and Tzoulaki, Ioanna and Riboli, Elio and Muller, David C.},
  issn         = {1055-9965},
  language     = {eng},
  number       = {3},
  pages        = {507--512},
  publisher    = {American Association for Cancer Research},
  series       = {Cancer Epidemiology Biomarkers and Prevention},
  title        = {Risk prediction for renal cell Carcinoma : Results from the European Prospective Investigation into Cancer and nutrition (EPIC) prospective cohort study},
  url          = {http://dx.doi.org/10.1158/1055-9965.EPI-20-1438},
  doi          = {10.1158/1055-9965.EPI-20-1438},
  volume       = {30},
  year         = {2021},
}