To develop a machine learning (ML)-assisted model for identifying the candidates for extended pelvic lymph node dissection (ePLND) in prostate cancer by integrating clinical, biopsy and precisely defined MRI findings.
248 patients treated with radical prostatectomy and ePLND or PLND were included. ML-assisted models were developed from 18 integrated features using a logistic regression (LR), support vector machine (SVM) and random forests (RFs), respectively. The models were compared to a MSKCC nomogram using the receiver operating characteristic-derived area under the curve (AUC), calibration plot and decision-curve analysis (DCA).
Total 59/248 (23.8%) lymph node invasion (LNIs) were identified at surgery. The predictive accuracy of ML-based models, with (+) or without (-) MRI-reported LNI, yielded similar AUCs (RFs+ /RFs- : 0.906/0.885; SVM+ /SVM- : 0.891/0.868; LR+ /LR- : 0.886/0.882), while were higher than MSKCC nomogram (0.816; p-value < 0.001). The calibration of MSKCC tended to underestimate LNI risk across the entire range of predicted probabilities compared to ML-assisted models. The DCA demonstrated that the ML-assisted models significantly improved risk prediction at risk threshold ≤ 80% compared to MSKCC. If ePLNDs missed was controlled < 3%, both RFs+ and RFs- resulted in higher positive predictive value (51.4%/49.6% vs 40.3%), similar negative predictive value (97.2%/97.8% vs 97.2%), and higher No. of ePLNDs spared (56.9%/54.4% vs 43.9%) compared to MSKCC.
Our ML-based model below 15% cutoff, superior to MSKCC nomogram, allows to 50% or more ePLNDs spared at the cost of missing < 3% LNIs. This article is protected by copyright. All rights reserved.
BJU international. 2019 Aug 07 [Epub ahead of print]
Ying Hou, Mei-Ling Bao, Chen-Jiang Wu, Jing Zhang, Yu-Dong Zhang, Hai-Bin Shi
Department of Radiology, the First Affiliated Hospital with Nanjing Medical University, No. 300, Guangzhou Road, Nanjing, Jiangsu Province, China, 210009., Department of Pathology, the First Affiliated Hospital with Nanjing Medical University, No. 300, Guangzhou Road, Nanjing, Jiangsu Province, China, 210009.