INTRODUCTION AND OBJECTIVES: Few models predicting the presence of lymph node invasion (LNI) in patients with renal cell carcinoma (RCC) are available. In this study, we tested the ability of LNI risk estimation relying on clinically attainable variables. METHODS: Between 1987 and 2014, 4,948 RCC patients treated with either partial or radical nephrectomy within a multi-institutional cohort were identified. Multivariable logistic regression analyses were used to test the accuracy of all the available clinical characteristics in predicting LNI. A nomogram predicting the probability of LNI was constructed using the logistic regression-derived coefficients. Log transformation was applied for clinical tumor size after non-linearity analysis. Calibration plot and leave-one-out cross validation (LOOCV) were used for internal validation. RESULTS: Overall, 204 patients (4.1%) had LNI. In multivariable analyses, symptoms at diagnosis (OR: 1.64; p<0.006), clinical tumor size (OR: 1.11; p<0.001), non-organ confined (OR: 2.65; p<0.001), clinical LNI (OR: 15.6; p<0.001) and presence of clinical metastases (OR: 2.4; p<0.001) were each significantly associated with the risk of LNI. The curve depicting the relationship between predicted and observed LNI closely approximates the ideal predictions, which indicates excellent calibration. In LOOCV, the C-index of our model was 92.1%. Using a 5% nomogram cut-off, 4.346 of 4.948 patients (87.8%) would be spared lymph-node dissection (LND) and LNI would be missed in 39 patients (0.9%, 19.1% of all LNI). The sensitivity, specificity, and negative predictive value associated with the 5% cut-off were 80.9%, 90.8%, and 99.1%, respectively. To minimize the number of LNI patients missed, a 1% nomogram cut-off may be considered, allowing to spare 57% of LND and missing only 7 LNI patients (0.1%, 3% of all LNI cases). CONCLUSIONS: We developed and internally validated a tool capable of highly accurately predicting LNI in RCC patients. This accurate tool could be useful for patient counseling and risk stratification at medical decision-making. Based on our model, patients with a LNI risk < 1% may be safely spared LND. Given the number of LNI cases missed when higher cutoffs were considered, further studies aimed at identifying accurate biomarkers of hidden LNI are urgently needed, especially in low-stage disease.
PREDICTION OF LYMPH NODE INVASION IN PATIENTS WITH RENAL CELL CARCINOMA: RESULTS FROM A LARGE INTERNATIONAL CONSORTIUM / Paolo Dell’Oglio; Grant Stewart ; Tobias Klatte; Alessandro Volpe ; Bulent Akdogan; Marco Roscigno; Hans Langenhuijsen; Martin Marszalek ; Oscar Rodriguez Faba; Maciej Salagierski; Andrea Minervini, ; Sabine Brookman-May; Umberto Capitanio. - In: THE JOURNAL OF UROLOGY. - ISSN 0022-5347. - STAMPA. - 195:(2016), pp. 962-963.
PREDICTION OF LYMPH NODE INVASION IN PATIENTS WITH RENAL CELL CARCINOMA: RESULTS FROM A LARGE INTERNATIONAL CONSORTIUM
MINERVINI, ANDREA;
2016
Abstract
INTRODUCTION AND OBJECTIVES: Few models predicting the presence of lymph node invasion (LNI) in patients with renal cell carcinoma (RCC) are available. In this study, we tested the ability of LNI risk estimation relying on clinically attainable variables. METHODS: Between 1987 and 2014, 4,948 RCC patients treated with either partial or radical nephrectomy within a multi-institutional cohort were identified. Multivariable logistic regression analyses were used to test the accuracy of all the available clinical characteristics in predicting LNI. A nomogram predicting the probability of LNI was constructed using the logistic regression-derived coefficients. Log transformation was applied for clinical tumor size after non-linearity analysis. Calibration plot and leave-one-out cross validation (LOOCV) were used for internal validation. RESULTS: Overall, 204 patients (4.1%) had LNI. In multivariable analyses, symptoms at diagnosis (OR: 1.64; p<0.006), clinical tumor size (OR: 1.11; p<0.001), non-organ confined (OR: 2.65; p<0.001), clinical LNI (OR: 15.6; p<0.001) and presence of clinical metastases (OR: 2.4; p<0.001) were each significantly associated with the risk of LNI. The curve depicting the relationship between predicted and observed LNI closely approximates the ideal predictions, which indicates excellent calibration. In LOOCV, the C-index of our model was 92.1%. Using a 5% nomogram cut-off, 4.346 of 4.948 patients (87.8%) would be spared lymph-node dissection (LND) and LNI would be missed in 39 patients (0.9%, 19.1% of all LNI). The sensitivity, specificity, and negative predictive value associated with the 5% cut-off were 80.9%, 90.8%, and 99.1%, respectively. To minimize the number of LNI patients missed, a 1% nomogram cut-off may be considered, allowing to spare 57% of LND and missing only 7 LNI patients (0.1%, 3% of all LNI cases). CONCLUSIONS: We developed and internally validated a tool capable of highly accurately predicting LNI in RCC patients. This accurate tool could be useful for patient counseling and risk stratification at medical decision-making. Based on our model, patients with a LNI risk < 1% may be safely spared LND. Given the number of LNI cases missed when higher cutoffs were considered, further studies aimed at identifying accurate biomarkers of hidden LNI are urgently needed, especially in low-stage disease.File | Dimensione | Formato | |
---|---|---|---|
MP73-07 PREDICTION OF LYMPH NODE INVASION IN PATIENTS WITH RENAL CELL CARCINOMA RESULTS FROM A LARGE INTERNATIONAL CONSORTIUM.pdf
accesso aperto
Tipologia:
Pdf editoriale (Version of record)
Licenza:
Open Access
Dimensione
196.26 kB
Formato
Adobe PDF
|
196.26 kB | Adobe PDF |
I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.