Dr. Answer AI for Prostate Cancer: Predicting Biochemical Recurrence Following Radical Prostatectomy

Objectives: To develop a model to predict biochemical recurrence (BCR) after radical prostatectomy (RP), using artificial intelligence (AI) techniques. Patients and Methods: This study collected data from 7,128 patients with prostate cancer (PCa) who received RP at 3 tertiary hospitals. After preprocessing, we used the data of 6,755 cases to generate the BCR prediction model. There were 16 input variables with BCR as the outcome variable. We used a random forest to develop the model. Several sampling techniques were used to address class imbalances. Results: We achieved good performance using a random forest with synthetic minority oversampling technique (SMOTE) using Tomek links, edited nearest neighbors (ENN), and random oversampling: accuracy = 96.59%, recall = 95.49%, precision = 97.66%, F1 score = 96.59%, and ROC AUC = 98.83%. Conclusion: We developed a BCR prediction model for RP. The Dr. Answer AI project, which was developed based on our BCR prediction model, helps physicians and patients to make treatment decisions in the clinical follow-up process as a clinical decision support system.

Download Full-text

Defining predictors of early, intermediate, and late biochemical recurrence in men with clinically localized prostate cancer treated with minimally invasive radical prostatectomy.

Journal of Clinical Oncology ◽

10.1200/jco.2017.35.6_suppl.113 ◽

2017 ◽

Vol 35 (6_suppl) ◽

pp. 113-113

Author(s):

Silvia Garcia Barreras ◽

Igor Nunes-Silva ◽

Rafael Sanchez-Salas ◽

Fernando P. Secin ◽

Victor Srougi ◽

...

Keyword(s):

Prostate Cancer ◽

Logistic Regression ◽

Lymph Node ◽

Laparoscopic Surgery ◽

Radical Prostatectomy ◽

Minimally Invasive ◽

Biochemical Recurrence ◽

Surgical Margins ◽

Positive Surgical Margins

113 Background: Follow up after radical prostatectomy should be tailored to clinical and pathologic characteristics. To determine predictive factors for early, intermediate and late biochemical recurrence (BCR) after minimally invasive radical prostatectomy (MIRP: lap and robot) in patients with localized prostate cancer (PCa). Methods: Prospective clinical, pathologic, and outcome data were collected for 6195 patients with cT1-3N0M0 PCa treated with MIRP at our institution from 2000 to 2016. None of them received neoadjuvant therapy. BCR was defined as PSA level greater than 0.2 ng/ml. Time to BCR was divided in terciles to identify variables associated with early ( < 12 months), intermediate (12-36 months) and late BCR ( > 36 months). Comparisons among groups were performed using ANOVA or Chi square test. Logistic regression models were built to determine risk factors associated with BCR at each time interval. Results: We identified 1148 (19%) patients with BCR. Median time to BCR was 24 months. Statistically significant differences were found between the groups concerning PSA preoperative, D’Amico risk, type of surgery, pT stage, pathological Gleason, positive margins and extracapsular extension. Multivariable logistic regression analysis showed preoperative PSA, positive nodes, positive surgical margins and laparoscopic surgery were associated with early BCR. Laparoscopic surgery was the only risk factor associated with intermediate term BCR. Significant predictors of late BCR included Gleason ≥ 7, ≥ pT3, positive surgical margins, lymph node dissection performance and laparoscopic surgery. Conclusions: Patients with high risk features like Gleason ≥ 7, ≥ pT3 and or positive surgical margins may develop late recurrence and deserve long term follow up. Identify patients with higher PSA and lymph node invasion has an important predictive role due to the risk of BCR within the first year. The association between laparoscopic technique and late BCR deserves further evaluation.

Download Full-text

Statins Prevent Biochemical Recurrence of Prostate Cancer After Radical Prostatectomy: A Single-center Retrospective Study with a Median Follow-up of 51.20 Months

Research and Reports in Urology ◽

10.2147/rru.s258267 ◽

2020 ◽

Vol Volume 12 ◽

pp. 439-446

Author(s):

Roberto Jarimba ◽

João Pedroso Lima ◽

Miguel Eliseu ◽

João Carvalho ◽

Hugo Antunes ◽

...

Keyword(s):

Prostate Cancer ◽

Retrospective Study ◽

Radical Prostatectomy ◽

Biochemical Recurrence ◽

Single Center

Download Full-text

Tumor contact length used as a biomarker to predict extracapsular extension, lymph node involvement, and biochemical recurrence.

Journal of Clinical Oncology ◽

10.1200/jco.2019.37.15_suppl.e16588 ◽

2019 ◽

Vol 37 (15_suppl) ◽

pp. e16588-e16588

Author(s):

Públio Viana ◽

Thiana Rodrigues ◽

Davi alves martins Mota ◽

Giuliano Guglielmetti ◽

Diogo Assed Bastos ◽

...

Keyword(s):

Prostate Cancer ◽

Decision Making ◽

Magnetic Resonance ◽

Radical Prostatectomy ◽

Biochemical Recurrence ◽

Specific Antigen ◽

Contact Length ◽

Imaging Biomarker ◽

Exclusion Criterion

e16588 Background: Prostate cancer (PC) is the cancer most commonly diagnosed in men and the second leading cause of death. Multiparametric magnetic resonance imaging (mpMRI) is the benchmark imaging standard for local staging of PC. Patients with pathological extraprostatic tumor (pEPE) have a worse prognosis than those with confined organ disease with a higher risk of biochemical recurrence (BCR) after radical prostatectomy (RP). These factors, in addition to positive lymph nodes (PLN), are pivotal in the decision-making process regarding treatment. However, the subjective MRI analysis has low sensitivity in the detection of EPE and is plagued by the low interobserver agreement. Tumor contact length (TCL) is an objective parameter of mpMRI defined as the length of prostate cancer in contact with the prostate capsule, which may provide additional information about prostate cancer outcomes. We aimed to evaluate TCL as a predictor of pECE, PLN, and BCR in patients undergoing RP. Methods: In this Institutional Review Board approved single-institution prospective study, we update the follow-up of 148 patients and included 31 new patients, with prostate cancer who underwent prostate MRI before radical prostatectomy from March 2014 to november 2018. TCL was measured using T2-weighted magnetic resonance images.Postoperative prostate-specific antigen (PSA) values were obtained every 3 months in the first year, then biannually and annually thereafter. BCR was defined as PSA≥0.2ng/mL. The exclusion criterion was prior treatment. Results: The median follow-up was 44 months. 76/179 (42,4%) patients had pEPE, 12/179 (6,7%) had pLN, and 29/179 (16,2%) had BCR. TCL values were significantly higher in patients with pathological pECE ( 20,6 mm vs 9,9 mm, p < 0.001). Using a cut-off of 11,9 mm we found area under curve (AUC) of 0,71 to predict pathological ECE, with sensitivity 72,5% and specificity 67,5% . Using a cut-off of 16,5 mm we found AUC of 0,64 to predict BCR, with sensitivity 54,2% and specificity 67,8% and with a cut-off of 21,7 mm the AUC to predict PLN was 0,77, with sensitivity 66,7% and specificity 83,6%. Conclusions: We prospectively demonstrated in our population that TCL values were significantly higher in patients with pEPE , PLN and BCR. If validated, this imaging biomarker may facilitate and inform patient counseling and decision-making.

Download Full-text

Predictors of Follow-Up Visits Post Radical Prostatectomy

American Journal of Men s Health ◽

10.1177/1557988318762633 ◽

2018 ◽

Vol 12 (4) ◽

pp. 760-765 ◽

Cited By ~ 2

Author(s):

Saira Khan ◽

Veronica Hicks ◽

Danielle Rancilio ◽

Marvin Langston ◽

Katina Richardson ◽

...

Keyword(s):

Prostate Cancer ◽

Radical Prostatectomy ◽

Biochemical Recurrence ◽

Rural Residence ◽

Psa Test ◽

Post Surgery ◽

Follow Up Care ◽

Long Term Follow Up

Long-term follow-up care among prostate cancer patients is important as biochemical recurrence can occur many years after diagnosis, with 20%–30% of men experiencing biochemical recurrence within 10 years of treatment. This study examined predictors of follow-up care among 1,158 radical prostatectomy patients, treated at the Washington University in St. Louis, within 6 months, 1 year, and 2 years post surgery. Predictors examined included age at surgery, race (Black vs. White), rural/urban status, education, marital status, and prostate cancer aggressiveness. Multivariable logistic regression was used to assess the association between the predictors and follow-up visits with a urologist in 6 months, the 1st year, and the 2nd year post surgery. In a secondary analysis, any follow-up visit with a prostate-specific antigen (PSA) test was included, regardless of provider type. Men that were Black ( 6 months OR: 0.60; 95% CI [0.36, 0.99], 1 year OR: 0.34; 95% CI [0.20, 0.59], 2 year OR: 0.41; 95% CI [0.25, 0.68]), resided in a rural residence ( 1 year OR: 0.61; 95% CI [0.44, 0.85], 2 year OR: 0.41; 95% CI [0.25, 0.68]), or were unmarried ( 2 year OR: 0.69; 95% CI [0.49, 0.97]) had a reduced odds of follow-up visits with a urologist. In models where any follow-up visit with a PSA test was examined, race remained a significant predictor of follow-up. The results indicate that Black men, men residing in a rural residence, and unmarried men may not receive adequate long-term follow-up care following radical prostatectomy. These men represent a high-risk group that could benefit from increased support post treatment.

Download Full-text

Lymphatic micrometastases predict biochemical recurrence in patients undergoing radical prostatectomy and pelvic lymph node dissection for prostate cancer

Aktuelle Urologie ◽

10.1055/a-0856-6545 ◽

2019 ◽

Vol 50 (06) ◽

pp. 612-618

Author(s):

Andreas Maxeiner ◽

Andreas Grevendieck ◽

Therese Pross ◽

Marc Rudl ◽

Alexander Arnold ◽

...

Keyword(s):

Prostate Cancer ◽

Radical Prostatectomy ◽

Biochemical Recurrence ◽

Diagnostic Value ◽

Nodal Metastasis ◽

Rank Test ◽

P Value ◽

Kaplan Meier ◽

Negative Findings

Abstract Background Nodal metastasis is a strong prognostic parameter in prostate cancer (PCa). We analysed the detection of micrometastases (miN + ) in initially nodal-negative (pN0) radical prostatectomy specimens from pT2a-c and pT3a PCa patients by immunohistochemistry (IHC). Material and Methods A total of 2352 lymph nodes of 193 PCa patients were centrally re-examined for miN + or miN- status using IHC. Results were correlated with clinical and follow-up data. Recurrence-free survival (RFS) was calculated with the log-rank test using the Kaplan-Meier method. In addition, a logistic regression analysis was performed. Results IHC detected miN + in a total of 17 patients (8.8 %). miN + seemed to be significantly associated with a higher Gleason score and was detected in more advanced pT stages. A total of 45 patients (23.1 %) had a biochemical recurrence (BCR). BCR was associated with miN +. Patients with miN + had a significantly shorter RFS (22.9 versus 58.7 months; p < 0.001). In the univariate (OR: 5.04; 95 % CI: 2.46 – 10.6; p-value: < 0.0001) and multivariate (OR: 3.29; 95 % CI: 1.54 – 7.08; p-value: 0.002) regression model, the miN + status was the strongest predictor of a BCR. Conclusions IHC seems to be of high diagnostic value for the detection of micrometastases in initially nodal-negative PCa patients. IHC should therefore be performed in PCa patients with nodal-negative findings.

Download Full-text

Extended versus limited pelvic lymphadenectomy during radical prostatectomy for intermediate- and high-risk prostate cancer: Early outcomes from a randomized controlled phase III study.

Journal of Clinical Oncology ◽

10.1200/jco.2017.35.15_suppl.5018 ◽

2017 ◽

Vol 35 (15_suppl) ◽

pp. 5018-5018 ◽

Cited By ~ 5

Author(s):

Jean Felipe Prodocimo Lestingi ◽

Giuliano Guglielmetti ◽

Jose Pontes Jr ◽

Anuar Ibrahim Mitre ◽

Alvaro Sarkis ◽

...

Keyword(s):

Prostate Cancer ◽

Lymph Node ◽

High Risk ◽

Radical Prostatectomy ◽

Biochemical Recurrence ◽

Phase Iii ◽

Transfusion Rate ◽

High Risk Prostate Cancer ◽

Risk Prostate Cancer

5018 Background: The role of extended pelvic lymph node dissection (ePLND) in treating prostate cancer (PCa) patients remains controversial, mainly by the lack of RCTs. Methods: Patients with D'Amico intermediate or high risk PCa, absence of bone metastasis and no previous treatment were prospectively computer randomised to undergo extended or limited PLND (1:1) during radical prostatectomy. Limited PLND (lPLND) included the obturator chain bilaterally; ePLND involved bilaterally chains: obturator, external-, internal-, common-iliac and pre-sacral. Surgical specimens and each chain were analyzed separately, according to College of American Pathologists. All patients signed a free and informed consent and local ethics committee approved the study. The primary endpoint was biochemical recurrence-free survival, analysed in the intention-to-treat population. This trial is registered with ClinicalTrials.gov, number NCT01812902. Results: Since May 2012 until August 2016, 291 patients were randomly assigned, 145 to ePLND and 146 to lPLND. Preoperative data were comparable between groups. Median follow-up was 35.2 months. EPLND increased significantly operative time (54 minutes), estimated blood loss (100 mL), length of hospital stays (1 day) [p≤0.001], transfusion rate [p = 0.05] and postoperative complications according to Clavien scale [p = 0.03]. There was no difference in Pathologic Gleason grade, T stage or positive surgical margin. On ePLND and lPLND groups, 59.3% and 61.7% were staged ≥ pT3a, respectively. EPLND and lPLND yielded median (mean) 17 (19.8) and 3 (4.1) nodes, respectively (p < 0.001). EPLND showed 6.3 times more lymph node metastases (p < 0.001) and only it was able to show positive nodes in intermediate risk. There were no difference in biochemical recurrence (PSA ≥ 0.2 ng/mL) using Kaplan-Meyer method (p = 0.4), Radiotherapy, Androgen Deprivation Therapy, bone metastases or death. Conclusions: Extended lymphadenectomy in intermediate- and high-risk prostate cancer patients is associated with better tumor staging, increased morbidity and no oncological benefits in this initial short follow-up time. Clinical trial information: NCT01812902.

Download Full-text

A Machine Learning Approach to Predict an Early Biochemical Recurrence after a Radical Prostatectomy

Applied Sciences ◽

10.3390/app10113854 ◽

2020 ◽

Vol 10 (11) ◽

pp. 3854

Author(s):

Seongkeun Park ◽

Jieun Byun ◽

Ji young Woo

Keyword(s):

Prostate Cancer ◽

Machine Learning ◽

Radical Prostatectomy ◽

Prediction Model ◽

Cancer Patients ◽

Biochemical Recurrence ◽

Learning Algorithms ◽

Machine Learning Algorithms ◽

Operating Characteristics ◽

Prediction Ability

Background: Approximately 20–50% of prostate cancer patients experience biochemical recurrence (BCR) after radical prostatectomy (RP). Among them, cancer recurrence occurs in about 20–30%. Thus, we aim to reveal the utility of machine learning algorithms for the prediction of early BCR after RP. Methods: A total of 104 prostate cancer patients who underwent magnetic resonance imaging and RP were evaluated. Four well-known machine learning algorithms (i.e., k-nearest neighbors (KNN), multilayer perceptron (MLP), decision tree (DT), and auto-encoder) were applied to build a prediction model for early BCR using preoperative clinical and imaging and postoperative pathologic data. The sensitivity, specificity, and accuracy for detection of early BCR of each algorithm were evaluated. Area under the receiver operating characteristics (AUROC) analyses were conducted. Results: A prediction model using an auto-encoder showed the highest prediction ability of early BCR after RP using all data as input (AUC = 0.638) and only preoperative clinical and imaging data (AUC = 0.656), followed by MLP (AUC = 0.607 and 0.598), KNN (AUC = 0.596 and 0.571), and DT (AUC = 0.534 and 0.495). Conclusion: The auto-encoder-based prediction system has the potential for accurate detection of early BCR and could be useful for long-term follow-up planning in prostate cancer patients after RP.

Download Full-text

Effect of open versus robot-assisted laparoscopic radical prostatectomy on cancer control in patients with clinically localized prostate cancer: Prospective analysis of 1,014 consecutive patients.

Journal of Clinical Oncology ◽

10.1200/jco.2012.30.5_suppl.80 ◽

2012 ◽

Vol 30 (5_suppl) ◽

pp. 80-80

Author(s):

Adrian Stuart Fairey ◽

Niels Jacobsen ◽

Don Voaklander ◽

Eric Estey

Keyword(s):

Prostate Cancer ◽

Radical Prostatectomy ◽

Surgical Approach ◽

Biochemical Recurrence ◽

Cancer Control ◽

Laparoscopic Radical Prostatectomy ◽

Localized Prostate Cancer ◽

Prospective Analysis ◽

Robot Assisted

80 Background: There are limited prospective data comparing outcomes of Open Radical Prostatectomy (ORP) and Robot-Assisted Laparoscopic Radical Prostatectomy (RALRP) for clinically localized prostate cancer. Our aim was to compare ORP and RALRP with respect to cancer control outcomes. Methods: A prospective analysis of data from the University of Alberta Radical Prostatectomy Database was performed. Between September 2007 and August 2010, 1019 consecutive men underwent radical prostatectomy for clinically localized prostate cancer. The surgical approach was selected by the surgeon. The outcomes were biochemical recurrence (BCR) and positive surgical margins (PSM). BCR was defined as a PSA ≥ 0.1 ng/ml followed by a subsequent confirmatory value or initiation of salvage therapy. PSM was defined as the presence of cancer at the inked margin in the radical prostatectomy specimen. The Kaplan-Meier method was used to estimate biochemical recurrence free survival (BCRFS). Univariable and multivariable analyses were used to determine the association between surgical approach and outcomes. Results: Data were evaluable for 1014 out of 1019 patients. 204 patients underwent ORP and 810 patients underwent RALRP. The median follow-up duration was 21 months (IQR 12 to 29). Baseline characteristics were similar between the groups. In univariable analysis, 3-year BCRFS (90.6% versus 88.9%), overall PSM (26.5% versus 28.8%), and stage-stratified PSM (pT2: 19.9% versus 21.8%; pT3: 40.6% versus 49.1%) did not differ between the groups (all comparisons p>0.05). In multivariable analysis, surgical approach was not independently associated with BCR (HR 0.77, 95% CI 0.43 to 1.37, p=0.37) or PSM (OR 1.2, 95% CI 0.80 to 1.67, p=0.44). Conclusions: ORP and RALRP provided comparable short-term oncologic efficacy. Extended follow-up of the prospective cohort is needed to confirm these preliminary findings.

Download Full-text

Impact of testosterone replacement therapy after radical prostatectomy on prostate cancer outcomes.

Journal of Clinical Oncology ◽

10.1200/jco.2019.37.7_suppl.100 ◽

2019 ◽

Vol 37 (7_suppl) ◽

pp. 100-100

Author(s):

Reith Sarkar ◽

J Kellogg Parsons ◽

John Paul Einck ◽

Arno James Mundt ◽

A. Karim Kader ◽

...

Keyword(s):

Prostate Cancer ◽

Overall Survival ◽

Radical Prostatectomy ◽

Cancer Patients ◽

Biochemical Recurrence ◽

Testosterone Replacement ◽

Testosterone Replacement Therapy ◽

Recurrence Free Survival ◽

Free Survival

100 Background: Currently there is little data to guide the use of post-radical prostatectomy (RP) testosterone replacement therapy in prostate cancer. We sought to evaluate the impact of post-RP testosterone replacement on prostate cancer outcomes in a large national cohort. Methods: We conducted a population-based cohort study using the Veterans Affairs Informatics and Computing Infrastructure. We identified node-negative and non-metastatic prostate cancer patients diagnosed between 2001-2015 treated with RP. We excluded patients for missing covariate and follow-up data. We then coded receipt of testosterone replacement after RP as a time-dependent covariate. Other covariates included: age, Charlson Comorbidity index, diagnosis year, body mass index, race, PSA, clinical T/N/M stage, Gleason score, and receipt of hormone therapy. Biochemical recurrence was defined as a post-RP PSA≥0.2. We evaluated prostate cancer-specific survival, overall survival, and biochemical recurrence free survival using multivariable Cox regression. Results: Our cohort included 28,651 patients, of whom 469 (1.6%) received testosterone replacement after RP. Median follow up was 7.4 years. There were no differences in clinical T stage, median post-RP PSA (testosterone: 0 non-testosterone: 0; p = 0.18), or hormone therapy use between treatment groups. Testosterone patients were more likely to be of younger age, have higher comorbidity, non-black, have a lower median pre-treatment PSA (5.0 vs 5.8; p < 0.001), and have higher BMI. The median time from RP to TRT was 3.0 years. After controlling for potential confounders, we found no difference in prostate cancer specific mortality (HR 0.73; 95% CI 0.32-1.62; p = 0.43), overall survival (HR 1.11; 95% CI 0.86-1.44; p = 0.43), non-cancer mortality (HR 1.17; 95% CI 0.89-1.55; p = 0.26) biochemical recurrence free survival (HR 1.07; 95% CI 0.84-1.36; p = 0.59) between testosterone users and non-users. Conclusions: Our results suggest that testosterone replacement is safe in prostate cancer patients who have undergone RP, though prospective data is necessary to confirm this finding.

Download Full-text

Contrast-enhanced transrectal ultrasound can reduce collection of unnecessary biopsies when diagnosing prostate cancer and is predictive of biochemical recurrence following a radical prostatectomy in patients with localized prostate cancer

10.21203/rs.2.12072/v2 ◽

2020 ◽

Author(s):

Hong-wei Zhao ◽

Jian Li ◽

Jia-Zheng Cao ◽

Juan Lin ◽

Zhu Wang ◽

...

Keyword(s):

Prostate Cancer ◽

Radical Prostatectomy ◽

Cancer Diagnosis ◽

Biochemical Recurrence ◽

Transrectal Ultrasound ◽

Localized Prostate Cancer ◽

Prostate Cancer Diagnosis ◽

Cox Regression Analysis ◽

Contrast Enhanced

Abstract Background: To investigate the value of using contrast-enhanced transrectal ultrasound (CETRUS) to reduce unnecessary collection of biopsies during prostate cancer diagnosis and its utility in predicting biochemical recurrence in patients with localized prostate cancer. Methods: This was a prospective study of suspected prostate cancer patients who were evaluated with CETRUS followed by a prostate biopsy. Prostate blood flow via CETRUS was graded using a 5-point scale. The relationship between CETRUS score and biopsy outcome was then analyzed for all patients; univariate and multi-variate analyses were used to determine the probable prognostic factors for biochemical recurrence in patients with localized prostate cancer that underwent a radical prostatectomy. Results: A total of 347 patients were enrolled in the study. Prostate cancer was found in 164 patients. A significant positive correlation (r = 0.69, p < 0.001) was found between CETRUS scores and prostate cancer incidence. Using CETRUS scores ≥ 2 as the threshold for when to biopsy could have safely reduced the number of biopsies taken overall by 12.1% (42/347) and spared 23.0% (42/183) of patients from undergoing an unnecessary biopsy. 77 patients with localized prostate cancer underwent a radical prostatectomy. The median follow-up time was 30 months (range: 8-56 months) and 17 of these 77 patients exhibited biochemical recurrence during the follow-up period. 3-year biochemical recurrence-free survival rates were 86% for patients with low CETRUS scores (≤ 3) and 59% for patients with high scores (> 3; p = 0.015). Multivariate Cox regression analysis indicated that CETRUS score was an independent predictor of biochemical recurrence (HR: 7.02; 95% CI: 2.00-24.69; p = 0.002). Conclusions: CETRUS scores may be a useful tool for reducing the collection unnecessary biopsy samples during prostate cancer diagnosis and are predictive of biochemical recurrence in patients with localized prostate cancer following a radical prostatectomy.

Download Full-text