Radiographic size of retroperitoneal lymph nodes predicts pathological nodal involvement for patients with renal cell carcinoma: development of a risk prediction model

Boris Gershman, Naoki Takahashi, Daniel M. Moreira, Robert H. Thompson, Stephen A. Boorjian, Christine M. Lohse, Brian A. Costello, John C. Cheville, Bradley C. Leibovich

Research output: Contribution to journalArticlepeer-review

18 Scopus citations

Abstract

Objectives: To evaluate the ability of clinical and radiographic features to predict lymph node (pN1) disease among patients with renal cell carcinoma undergoing radical nephrectomy (RN), and to develop a preoperative risk prediction model. Patients and Methods: In all, 220 patients with preoperative computed tomography scans available for review underwent RN with lymph node dissection (LND) from 2000 to 2010. Radiographic features were assessed by one genitourinary radiologist blinded to pN status. Associations of features with pN1 disease were evaluated using logistic regression to develop predictive models. Model performance was assessed using area under the receiver operating characteristic curve (AUC) and decision curve analysis. Results: The median (interquartile range) lymph node yield was 10 (5–18). In all, 55 patients (25%) had pN1 disease at RN. On univariable analysis, the maximum lymph node (LN) short axis diameter [odds ratio (OR) 1.17; P < 0.001] predicted pN1 disease with an AUC of 0.84. Although several clinical and radiographic features were associated with pN1 disease, only two were retained in the multivariable model: maximum LN short axis diameter (OR 1.19; P <0.001) and radiographic perinephric/sinus fat invasion (OR 44.64; P = 0.01), with an AUC of 0.85. On decision curve analysis, the single variable and multivariable models showed similar net benefit. Conclusion: Two radiographic features, maximum LN short axis diameter and perinephric/sinus fat invasion, outperformed traditional clinical variables in predicting pN1 disease. Maximum LN short axis diameter alone showed excellent predictive performance, and, if validated externally, would provide for a simple model to guide patient selection for LND.

Original languageEnglish (US)
Pages (from-to)742-749
Number of pages8
JournalBJU international
Volume118
Issue number5
DOIs
StatePublished - Nov 1 2016

Keywords

  • lymph node dissection
  • lymphadenopathy
  • nephrectomy
  • node positive
  • renal cell carcinoma

ASJC Scopus subject areas

  • Urology

Fingerprint Dive into the research topics of 'Radiographic size of retroperitoneal lymph nodes predicts pathological nodal involvement for patients with renal cell carcinoma: development of a risk prediction model'. Together they form a unique fingerprint.

Cite this