Faculty Bibliography

INTRODUCTION/BACKGROUND:Penectomy for PC is useful in staging, disease prognosis, and treatment. Limited studies have evaluated its surgical complications. We sought to assess these complications and determine predictive models to create a novel risk score for penectomy complications. PATIENTS AND METHODS/METHODS:A retrospective review of patients undergoing PC surgical management from the 2005-2016 American College of Surgeons National Surgical Quality Improvement Program was performed. Data were queried for partial and total penectomy among those with PC. To develop predictive models of complications, we fit LASSO logistic, random forest, and stepwise logistic models to training data using cross-validation, demographic, comorbidity, laboratory, and wound characteristics as candidate predictors. Each model was evaluated on the test data using receiver operating characteristic curves. A novel risk score was created by rounding coefficients from the LASSO logistic model. RESULTS:A total of 304 cases met the inclusion criteria. Overall incidence of penectomy complications was 19.7%, where urinary tract infection (3.0%), superficial surgical site infection (3.0%), and bleeding requiring transfusion (3.9%) were most common. LASSO logistic, random forest, and stepwise logistic models for predicting complications had area under the curve (AUC) [95% confidence interval] values of 0.66 [0.52-0.81], 0.73 [0.63-0.83], and 0.59 [0.45-0.74], respectively. Eleven variables were included in the risk score. The LASSO model-derived risk score had moderately good performance (area under the curve [95% confidence interval] 0.74 [0.66-0.82]). Using a cutoff point of 6, the score attains sensitivity 0.58, specificity 0.74, and kappa 0.26. CONCLUSION/CONCLUSIONS:PC management through penectomy is associated with appreciable complications rates. Predictive models of penectomy complications performed moderately well. Our novel prognostic risk score may allow for improved preoperative counseling and risk stratification of men undergoing surgical management of PC.

Background/UNASSIGNED:Radical cystectomy for bladder cancer has one of the highest rates of morbidity among urologic surgery, but the ability to predict postoperative complications remains poor. Our study objective was to create machine learning models to predict complications and factors leading to extended length of hospital stay and discharge to a higher level of care after radical cystectomy. Methods/UNASSIGNED:Using the American College of Surgeons National Surgical Quality Improvement Program, peri-operative adverse outcome variables for patients undergoing elective radical cystectomy for bladder cancer from 2005 to 2016 were extracted. Variables assessed include occurrence of minor, infectious, serious, or any adverse events, extended length of hospital stay, and discharge to higher-level care. To develop predictive models of radical cystectomy complications, we fit generalized additive model (GAM), least absolute shrinkage and selection operator (LASSO) logistic, neural network, and random forest models to training data using various candidate predictor variables. Each model was evaluated on the test data using receiver operating characteristic curves. Results/UNASSIGNED:A total of 7557 patients were identified who met the inclusion criteria, and 2221 complications occurred. LASSO logistic models demonstrated the highest area under curve for predicting any complications (0.63), discharge to a higher level of care (0.75), extended length of stay (0.68), and infectious (0.62) adverse events. This was comparable with random forest in predicting minor (0.60) and serious (0.63) adverse events. Conclusions/UNASSIGNED:Our models perform modestly in predicting radical cystectomy complications, highlighting both the complex cystectomy process and the limitations of large healthcare datasets. Identifying the most important variable leading to each type of adverse event may allow for further strategies to model cystectomy complications and target optimization of modifiable variables pre-operative to reduce postoperative adverse events.

CONTEXT/BACKGROUND:- In Gleason score GS (7) prostate cancers, the quantity of Gleason pattern 4 (GP 4) is an important prognostic factor and influences treatment decisions. Magnetic resonance imaging (MRI)-targeted biopsy has been increasingly used in clinical practice. OBJECTIVE:- To investigate whether MRI-targeted biopsy may detect GS 7 prostate cancer with greater GP 4 quantity, and whether it improves biopsy/radical prostatectomy GS concordance. DESIGN/METHODS:- A total of 243 paired standard and MRI-targeted biopsies with cancer in either standard or targeted or both were studied, 65 of which had subsequent radical prostatectomy. The biopsy findings, including GS and tumor volume, were correlated with the radical prostatectomy findings. RESULTS:- More prostate cancers detected by MRI-targeted biopsy were GS 7 or higher. Mean GP 4 percentage in GS 7 cancers was 31.0% ± 29.3% by MRI-targeted biopsy versus 25.1% ± 29.5% by standard biopsy. A total of 122 of 218 (56.0%) and 96 of 217 (44.2%) prostate cancers diagnosed on targeted biopsy and standard biopsy, respectively, had a GP 4 of 10% or greater ( P = .01). Gleason upgrading was seen in 12 of 59 cases (20.3%) from MRI-targeted biopsy and in 24 of 57 cases (42.1%) from standard biopsy ( P = .01). Gleason upgrading correlated with the biopsy cancer volume inversely and GP 4 of 30% or less in standard biopsy. Such correlation was not found in MRI-targeted biopsy. CONCLUSIONS:- Magnetic resonance imaging-targeted biopsy may detect more aggressive prostate cancers and reduce the risk of Gleason upgrading in radical prostatectomy. This study supports a potential role for MRI-targeted biopsy in the workup of prostate cancer and inclusion of percentage of GP 4 in the prostate biopsy reports.