A machine learning‐assisted decision‐support model to better identify patients with prostate cancer requiring an extended pelvic lymph node dissection

Abstract

ObjectivesTo develop a machine learning (ML)-assisted model to identify candidates for extended pelvic lymph node dissection (ePLND) in prostate cancer by integrating clinical, biopsy, and precisely defined magnetic resonance imaging (MRI) findings. Patients and MethodsIn all, 248 patients treated with radical prostatectomy and ePLND or PLND were included. ML-assisted models were developed from 18 integrated features using logistic regression (LR), support vector machine (SVM), and random forests (RFs). The models were compared to the Memorial SloanKettering Cancer Center (MSKCC) nomogram using receiver operating characteristic-derived area under the curve (AUC) calibration plots and decision curve analysis (DCA). ResultsA total of 59/248 (23.8%) lymph node invasions (LNIs) were identified at surgery. The predictive accuracy of the ML-based models, with (+) or without (À) MRI-reported LNI, yielded similar AUCs (RFs + /RFs À : 0.906/0.885; SVM + /SVM À : 0.891/ 0.868; LR + /LR À : 0.886/0.882) and were higher than the MSKCC nomogram (0.816; P < 0.001). The calibration of the MSKCC nomogram tended to underestimate LNI risk across the entire range of predicted probabilities compared to the ML-assisted models. The DCA showed that the ML-assisted models significantly improved risk prediction at a risk threshold of ≤80% compared to the MSKCC nomogram. If ePLNDs missed was controlled at <3%, both RFs + and RFs À resulted in a higher positive predictive value (51.4%/49.6% vs 40.3%), similar negative predictive value (97.2%/97.8% vs 97.2%), and higher number of ePLNDs spared (56.9%/54.4% vs 43.9%) compared to the MSKCC nomogram. ConclusionsOur ML-based model, with a 5-15% cutoff, is superior to the MSKCC nomogram, sparing ≥50% of ePLNDs with a risk of missing <3% of LNIs.Keywords pelvic lymph node invasion, logistic regression, support vector machine, random forests, machine learning, #ProstateCancer, #PCSM

An Artificial Intelligence-based Multimodality-Integration for Precise Prediction of Prostate Cancer Lymph Node Metastasis: A Retrospective Two-center Study

“…First, our results show that the newly built imaging hallmarks, i.e., RML score and DL score, are workable for the nodal staging. In the previous studies, attempts to adding radiological evaluations to clinical nomograms have shown promising results, with the AUC of 0.89-0.95 16,20,32 . In current work, when the new imaging hallmarks were integrated with clinical factors simultaneously, the predictive performance of PRISK was signi cantly higher than that of any other method.…”

“…In each patient, the radiologists identi ed the leading cancer lesion, referring to those with the higher Prostate Imaging and Reporting and Data System (PI-RADS) version 2.1 (v2.1) score or larger diameter if the score was the same, and the following imaging features were recorded: (i) prostate volume; (ii) zone of lesion origin (peripheral zone (PZ) or transitional zone(TZ)); (iii) shape of the lesion (regular or irregular); (iv) margin of the lesion (well-de ned or ill-de ned); (v) tumor max diameter; (vi) volumetric mean ADC value; (vii) PI-RADS score (PI-RADS 1-5) 24,25 ; (viii) MRI T-stage (≤ T1c, T2a, T2b, ≥ T2c); (ix) MRI-based ECE, SVI, and PLNM (absent or present). The de nition of MRI-based imaging features was described in Supplementary Data 1, referring to criteria previously reported 20,26 . All cases were interpreted individually rst and then re-reviewed in tandem by the two readers 4 weeks after the individual evaluation.…”

“…Several studies had attempted to develop models for the prediction of PLNM from imaging ndings and demographic factors [17][18][19] . Others have carried out on the application of arti cial intelligence (AI) such as machine learning or deep learning methods, yielding higher accuracy than the Briganti and MSKCC nomograms 20 . Transforming full imaging ndings, especially these high-dimensional machine-learning radiomics features and/or deep learning-generative features, into a clinically interpretable signature is a potential alternative for improved prognostic and predictive accuracy in patients with localized PCa.…”

Predicting Prostate Cancer Upgrading of Biopsy Gleason Grade Group at Radical Prostatectomy Using Machine Learning-Assisted Decision-support Models

“…where Y is the output, is the nonzero coe cient, and is the selected clinical features based on results of multivariable logistic regression analysis [14].…”

“…where is the output value of predictive models, and indicates the probabilities of harboring upgrading at RP [14].…”