Development, validation, and head-to-head comparison of logistic regression-based nomograms and artificial neural network models predicting prostate cancer on initial extended biopsy

Development, Validation and Comparison of Artificial Neural Network and Logistic Regression Models Predicting Eosinophilic Chronic Rhinosinusitis With Nasal Polyps

PURPOSE

Chronic rhinosinusitis with nasal polyps (CRSwNP) can be classified into eosinophilic CRSwNP (eCRSwNP) and non-eosinophilic CRSwNP (non-eCRSwNP) by tissue biopsy, which is difficult to perform preoperatively. Clinical biomarkers have predictive value for the classification of CRSwNP. We aimed to evaluate the application of artificial neural network (ANN) modeling in distinguishing different endotypes of CRSwNP based on clinical biomarkers.

METHODS

Clinical parameters were collected from 109 CRSwNP patients, and their predictive ability was analyzed. ANN and logistic regression (LR) models were developed in the training group (72 patients) and further tested in the test group (37 patients). The output variable was the diagnosis of eCRSwNP, defined as tissue eosinophil count > 10 per high-power field. The receiver operating characteristics curve was used to assess model performance.

RESULTS

A total of 15 clinical features from 60 healthy controls, 60 eCRSwNP and 49 non-eCRSwNP were selected as candidate predictors. Nasal nitric oxide levels, peripheral eosinophil absolute count, total immunoglobulin E, and ratio of bilateral computed tomography scores for the ethmoid sinus and maxillary sinus were identified as important features for modeling. Two ANN models based on 4 and 15 clinical features were developed to predict eCRSwNP, which showed better performance, with the area under the receiver operator characteristics significantly higher than those from the respective LR models (0.976 vs. 0.902, P = 0.048; 0.970 vs. 0.845, P = 0.011). All ANN models had better fits than single variable prediction models (all P < 0.05), and ANN model 1 had the best predictive performance among all models.

CONCLUSIONS

Machine learning models assist clinicians in predicting endotypes of nasal polyps before invasive detection. The ANN model has the potential to predict eCRSwNP with high sensitivity and specificity, and is superior to the LR model. ANNs are valuable for optimizing personalized patient management.

Construction and Comparison of Different Models in Detecting Prostate Cancer and Clinically Significant Prostate Cancer

Background

With the widespread adoption of prostatic-specific antigen (PSA) screening, the detection rates of prostate cancer (PCa) have increased. Due to the low specificity and high false-positive rate of serum PSA levels, it was difficult to diagnose PCa accurately. To improve the diagnosis of PCa and clinically significant prostate cancer (CSPCa), we established novel models on the basis of the prostate health index (PHI) and multiparametric magnetic resonance imaging (mpMRI) in the Asian population.

Methods

We retrospectively collected the clinical indicators of patients with TPSA at 4–20 ng/ml. Furthermore, mpMRI was performed using a 3.0-T scanner and reported in the Prostate Imaging Reporting and Data System version 2.1 (PI-RADS). Univariable and multivariable logistic analyses were performed to construct the models. The performance of different models based on PSA derivatives, PHI derivatives, PI-RADS, and a combination of PHI derivatives and PI-RADS was evaluated.

Results

Among the 128 patients, 47 (36.72%) patients were diagnosed with CSPCa and 81 (63.28%) patients were diagnosed with non-CSPCa. Of the 81 (63.28%) patients, 8 (6.25%) patients were diagnosed with Gleason Grade 1 PCa and 73 (57.03%) patients were diagnosed with non-PCa. In the analysis of the receiver operator characteristic (ROC) curves in TPSA 4–20 ng/ml, the multivariable model for PCa was significantly larger than that for the model based on the PI-RADS (p = 0.004) and that for the model based on the PHI derivatives (p = 0.031) in diagnostic accuracy. The multivariable model for CSPCa was significantly larger than that for the model based on the PI-RADS (p = 0.003) and was non-significantly larger than that for the model based on the PHI derivatives (p = 0.061) in diagnostic accuracy. For PCa in TPSA 4–20 ng/ml, a multivariable model achieved the optimal diagnostic value at four levels of predictive variables. For CSPCa in TPSA 4–20 ng/ml, the multivariable model achieved the optimal diagnostic value at a sensitivity close to 90% and 80%.

Conclusions

The models combining PHI derivatives and PI-RADS performed better in detecting PCa and CSPCa than the models based on either PHI or PI-RADS.

The use of an artificial neural network in the evaluation of the extracorporeal shockwave lithotripsy as a treatment of choice for urinary lithiasis

Objective

Artificial neural networks (ANNs) are widely applied in medicine, since they substantially increase the sensitivity and specificity of the diagnosis, classification, and the prognosis of a medical condition. In this study, we constructed an ANN to evaluate several parameters of extracorporeal shockwave lithotripsy (ESWL), such as the outcome and safety of the procedure.

Methods

Patients with urinary lithiasis suitable for ESWL treatment were enrolled. An ANN was designed using MATLAB. Medical data were collected from all patients and 12 nodes were used as inputs. Conventional statistical analysis was also performed.

Results

Finally, 716 patients were included in our study. Univariate analysis revealed that diabetes and hydronephrosis were positively correlated with ESWL complications. Regarding efficacy, univariate analysis revealed that stone location, stone size, the number and density of shockwaves delivered, and the presence of a stent in the ureter were independent factors of the ESWL outcome. This was further confirmed when adjusted for sex and age in a multivariate analysis. The performance of the ANN at the end of the training state reached 98.72%. The four basic ratios (sensitivity, specificity, positive predictive value, and negative predictive value) were calculated for both training and evaluation data sets. The performance of the ANN at the end of the evaluation state was 81.43%.

Conclusion

Our ANN achieved high score in predicting the outcome and the side effects of the ESWL treatment for urinary stones.

A systematic review of the applications of Expert Systems (ES) and machine learning (ML) in clinical urology

BACKGROUND

Testing a hypothesis for 'factors-outcome effect' is a common quest, but standard statistical regression analysis tools are rendered ineffective by data contaminated with too many noisy variables. Expert Systems (ES) can provide an alternative methodology in analysing data to identify variables with the highest correlation to the outcome. By applying their effective machine learning (ML) abilities, significant research time and costs can be saved. The study aims to systematically review the applications of ES in urological research and their methodological models for effective multi-variate analysis. Their domains, development and validity will be identified.

METHODS

The PRISMA methodology was applied to formulate an effective method for data gathering and analysis. This study search included seven most relevant information sources: WEB OF SCIENCE, EMBASE, BIOSIS CITATION INDEX, SCOPUS, PUBMED, Google Scholar and MEDLINE. Eligible articles were included if they applied one of the known ML models for a clear urological research question involving multivariate analysis. Only articles with pertinent research methods in ES models were included. The analysed data included the system model, applications, input/output variables, target user, validation, and outcomes. Both ML models and the variable analysis were comparatively reported for each system.

RESULTS

The search identified n = 1087 articles from all databases and n = 712 were eligible for examination against inclusion criteria. A total of 168 systems were finally included and systematically analysed demonstrating a recent increase in uptake of ES in academic urology in particular artificial neural networks with 31 systems. Most of the systems were applied in urological oncology (prostate cancer = 15, bladder cancer = 13) where diagnostic, prognostic and survival predictor markers were investigated. Due to the heterogeneity of models and their statistical tests, a meta-analysis was not feasible.

CONCLUSION

ES utility offers an effective ML potential and their applications in research have demonstrated a valid model for multi-variate analysis. The complexity of their development can challenge their uptake in urological clinics whilst the limitation of the statistical tools in this domain has created a gap for further research studies. Integration of computer scientists in academic units has promoted the use of ES in clinical urological research.

Development and head-to-head comparison of machine-learning models to identify patients requiring prostate biopsy

Background

Machine learning has many attractive theoretic properties, specifically, the ability to handle non predefined relations. Additionally, studies have validated the clinical utility of mpMRI for the detection and localization of CSPCa (Gleason score ≥ 3 + 4). In this study, we sought to develop and compare machine-learning models incorporating mpMRI parameters with traditional logistic regression analysis for prediction of PCa (Gleason score ≥ 3 + 3) and CSPCa on initial biopsy.

Methods

A total of 688 patients with no prior prostate cancer diagnosis and tPSA ≤ 50 ng/ml, who underwent mpMRI and prostate biopsy were included between 2016 and 2020. We used four supervised machine-learning algorithms in a hypothesis-free manner to build models to predict PCa and CSPCa. The machine-learning models were compared to the logistic regression analysis using AUC, calibration plot, and decision curve analysis.

Results

The artificial neural network (ANN), support vector machine (SVM), and random forest (RF) yielded similar diagnostic accuracy with logistic regression, while classification and regression tree (CART, AUC = 0.834 and 0.867) had significantly lower diagnostic accuracy than logistic regression (AUC = 0.894 and 0.917) in prediction of PCa and CSPCa (all P < 0.05). However, the CART illustrated best calibration for PCa (SSR = 0.027) and CSPCa (SSR = 0.033). The ANN, SVM, RF, and LR for PCa had higher net benefit than CART across the threshold probabilities above 5%, and the five models for CSPCa displayed similar net benefit across the threshold probabilities below 40%. The RF (53% and 57%, respectively) and SVM (52% and 55%, respectively) for PCa and CSPCa spared more unnecessary biopsies than logistic regression (35% and 47%, respectively) at 95% sensitivity for detection of CSPCa.

Conclusion

Machine-learning models (SVM and RF) yielded similar diagnostic accuracy and net benefit, while spared more biopsies at 95% sensitivity for detection of CSPCa, compared with logistic regression. However, no method achieved desired performance. All methods should continue to be explored and used in complementary ways.

An Artificial Neural Networks Model for Early Predicting In-Hospital Mortality in Acute Pancreatitis in MIMIC-III

Background

Early and accurate evaluation of severity and prognosis in acute pancreatitis (AP), especially at the time of admission is very significant. This study was aimed to develop an artificial neural networks (ANN) model for early prediction of in-hospital mortality in AP.

Methods

Patients with AP were identified from the Medical Information Mart for Intensive Care-III (MIMIC-III) database. Clinical and laboratory data were utilized to perform a predictive model by back propagation ANN approach.

Results

A total of 337 patients with AP were analyzed in the study, and the in-hospital mortality rate was 11.2%. A total of 12 variables that differed between patients in survivor group and nonsurvivor group were applied to construct ANN model. Three independent variables were identified as risk factors associated with in-hospital mortality by multivariate logistic regression analysis. The predictive performance based on the area under the receiver operating characteristic curve (AUC) was 0.769 for ANN model, 0.607 for logistic regression, 0.652 for Ranson score, and 0.401 for SOFA score.

Conclusion

An ANN predictive model for in-hospital mortality in patients with AP in MIMIC-III database was first performed. The patients with high risk of fatal outcome can be screened out easily in the early stage of AP by our model.

Multivariable Models Incorporating Multiparametric Magnetic Resonance Imaging Efficiently Predict Results of Prostate Biopsy and Reduce Unnecessary Biopsy

Purpose

We sought to develop diagnostic models incorporating mpMRI examination to identify PCa (Gleason score≥3+3) and CSPCa (Gleason score≥3+4) to reduce overdiagnosis and overtreatment.

Methods

We retrospectively identified 784 patients according to inclusion criteria between 2016 and 2020. The cohort was split into a training cohort of 548 (70%) patients and a validation cohort of 236 (30%) patients. Age, PSA derivatives, prostate volume, and mpMRI parameters were assessed as predictors for PCa and CSPCa. The multivariable models based on clinical parameters were evaluated using area under the curve (AUC), calibration plots, and decision curve analysis (DCA).

Results

Univariate analysis showed that age, tPSA, PSAD, prostate volume, MRI-PCa, MRI-seminal vesicle invasion, and MRI-lymph node invasion were significant predictors for both PCa and CSPCa (each p≤0.001). PSAD has the highest diagnostic accuracy in predicting PCa (AUC=0.79) and CSPCa (AUC=0.79). The multivariable models for PCa (AUC=0.92, 95% CI: 0.88–0.96) and CSPCa (AUC=0.95, 95% CI: 0.92–0.97) were significantly higher than the combination of derivatives for PSA (p=0.041 and 0.009 for PCa and CSPCa, respectively) or mpMRI (each p<0.001) in diagnostic accuracy. And the multivariable models for PCa and CSPCa illustrated better calibration and substantial improvement in DCA at threshold above 10%, compared with PSA or mpMRI derivatives. The PCa model with a 30% cutoff or CSPCa model with a 20% cutoff could spare the number of biopsies by 53%, and avoid the number of benign biopsies over 80%, while keeping a 95% sensitivity for detecting CSPCa.

Conclusion

Our multivariable models could reduce unnecessary biopsy without comprising the ability to diagnose CSPCa. Further prospective validation is required.

Who Can Avoid Biopsy of Magnetic Resonance Imaging-Negative Lobes without Compromising Significant Cancer Detection among Men with Unilateral Magnetic Resonance Imaging-Positive Lobes?

OBJECTIVES

To assess whether biopsy of multiparametric magnetic resonance imaging (MRI)-negative lobes can be avoided without compromising significant cancer (SC) detection among men with unilateral MRI-positive lobes.

METHODS

From April 2013 to April 2019, 322 men with elevated prostate-specific antigen (PSA <20 ng/mL) and unilateral MRI-positive lobes underwent targeted 4-core and systematic 14-core biopsy. MRI findings were prospectively collected and evaluated according to the Prostate Imaging-Reporting and Data System (PI-RADS) version 2, and scores ≥3 were considered positive. SC was defined as Gleason score ≥3 + 4 or maximal cancer length ≥5 mm. We developed predictive models of overall cancer and SC in MRI-negative lobes and evaluated the performance of these models.

RESULTS

Detection rates of overall cancer/SC were 69%/61% for the overall cohort, 58%/48% for MRI-positive lobes, and 36%/20% for MRI-negative lobes. Age ≥75 years, PSA density ≥0.3, and PI-RADS ≥4 were independently predictive of both overall cancer and SC in MRI-negative lobes; 1 point was assigned for each risk factor, and the predictive score was defined as the sum of points (0-3) for both overall cancer and SC. Areas under the curve of the model for overall cancer/SC were 0.67/0.71. In the decision curve analysis, the model was of value above the threshold probability of 13%/6% for detecting overall cancer/SC in MRI-negative lobes. Of 40 men with score 0, overall cancer/SC was detected in the MRI-negative lobe in 4 (10%)/1 (2.5%).

CONCLUSION

Biopsies of MRI-negative lobes may be avoided without compromising SC detection using our predictive model.

Development of the Smartphone Addiction Risk Rating Score for a Smartphone Addiction Management Application

Smartphone usage characteristics are useful for identification of the risk factors for smartphone addiction. Risk rating scores can be developed based on smartphone usage characteristics. This study aimed to investigate the smartphone addiction risk rating (SARR) score using smartphone usage characteristics. We evaluated 593 smartphone users using online surveys conducted between January 2 and January 31, 2019. We identified 102 smartphone users who were addicted to smartphones and 491 normal users based on the Korean Smartphone Addiction Proneness Scale for Adults. A multivariate logistic regression analysis was used to identify significant risk factors for smartphone addiction. The SARR score was calculated using a nomogram based on the significant risk factors. Weekend average usage time, habitual smartphone behavior, addictive smartphone behavior, social usage, and process usage were the significant risk factors associated with smartphone addiction. Furthermore, we developed the SARR score based on these factors. The SARR score ranged between 0 and 221 points, with the cut-off being 116.5 points. We developed a smartphone addiction management application using the SARR score. The SARR score provided insights for the development of monitoring, prevention, and prompt intervention services for smartphone addiction.

Prediction models for prostate cancer to be used in the primary care setting: a systematic review

OBJECTIVE

To identify risk prediction models for prostate cancer (PCa) that can be used in the primary care and community health settings.

DESIGN

Systematic review.

DATA SOURCES

MEDLINE and Embase databases combined from inception and up to the end of January 2019.

ELIGIBILITY

Studies were included based on satisfying all the following criteria: (i) presenting an evaluation of PCa risk at initial biopsy in patients with no history of PCa, (ii) studies not incorporating an invasive clinical assessment or expensive biomarker/genetic tests, (iii) inclusion of at least two variables with prostate-specific antigen (PSA) being one of them, and (iv) studies reporting a measure of predictive performance. The quality of the studies and risk of bias was assessed by using the Prediction model Risk Of Bias ASsessment Tool (PROBAST).

DATA EXTRACTION AND SYNTHESIS

Relevant information extracted for each model included: the year of publication, source of data, type of model, number of patients, country, age, PSA range, mean/median PSA, other variables included in the model, number of biopsy cores to assess outcomes, study endpoint(s), cancer detection, model validation and model performance.

RESULTS

An initial search yielded 109 potential studies, of which five met the set criteria. Four studies were cohort-based and one was a case-control study. PCa detection rate was between 20.6% and 55.8%. Area under the curve (AUC) was reported in four studies and ranged from 0.65 to 0.75. All models showed significant improvement in predicting PCa compared with being based on PSA alone. The difference in AUC between extended models and PSA alone was between 0.06 and 0.21.

CONCLUSION

Only a few PCa risk prediction models have the potential to be readily used in the primary healthcare or community health setting. Further studies are needed to investigate other potential variables that could be integrated into models to improve their clinical utility for PCa testing in a community setting.

Comparing a new risk prediction model with prostate cancer risk calculator apps in a Taiwanese population

PURPOSE

To develop a novel Taiwanese prostate cancer (PCa) risk model for predicting PCa, comparing its predictive performance with that of two well-established PCa risk calculator apps.

METHODS

1545 men undergoing prostate biopsies in a Taiwanese tertiary medical center between 2012 and 2019 were identified retrospectively. A five-fold cross-validated logistic regression risk model was created to calculate the probabilities of PCa and high-grade PCa (Gleason score ≧ 7), to compare those of the Rotterdam and Coral apps. Discrimination was analyzed using the area under the receiver operator characteristic curve (AUC). Calibration was graphically evaluated with the goodness-of-fit test. Decision-curve analysis was performed for clinical utility. At different risk thresholds to biopsy, the proportion of biopsies saved versus low- and high-grade PCa missed were presented.

RESULTS

Overall, 278/1309 (21.2%) patients were diagnosed with PCa, and 181 out of 278 (65.1%) patients had high-grade PCa. Both our model and the Rotterdam app demonstrated better discriminative ability than the Coral app for detection of PCa (AUC: 0.795 vs 0.792 vs 0.697, DeLong's method: P < 0.001) and high-grade PCa (AUC: 0.869 vs 0.873 vs 0.767, P < 0.001). Using a ≥ 10% risk threshold for high-grade PCa to biopsy, our model could save 67.2% of total biopsies; among these saved biopsies, only 3.4% high-grade PCa would be missed.

CONCLUSION

Our new logistic regression model, similar to the Rotterdam app, outperformed the Coral app in the prediction of PCa and high-grade PCa. Additionally, our model could save unnecessary biopsies and avoid missing clinically significant PCa in the Taiwanese population.

Development, Validation and Comparison of Artificial Neural Network Models and Logistic Regression Models Predicting Survival of Unresectable Pancreatic Cancer

Background: Prediction models for the overall survival of pancreatic cancer remain unsatisfactory. We aimed to explore artificial neural networks (ANNs) modeling to predict the survival of unresectable pancreatic cancer patients.

Methods: Thirty-two clinical parameters were collected from 221 unresectable pancreatic cancer patients, and their prognostic ability was evaluated using univariate and multivariate logistic regression. ANN and logistic regression (LR) models were developed on a training group (168 patients), and the area under the ROC curve (AUC) was used for comparison of the ANN and LR models. The models were further tested on the testing group (53 patients), and k-statistics were used for accuracy comparison.

Results: We built three ANN models, based on 3, 7, and 32 basic features, to predict 8 month survival. All 3 ANN models showed better performance, with AUCs significantly higher than those from the respective LR models (0.811 vs. 0.680, 0.844 vs. 0.722, 0.921 vs. 0.849, all p < 0.05). The ability of the ANN models to discriminate 8 month survival with higher accuracy than the respective LR models was further confirmed in 53 consecutive patients.

Conclusion: We developed ANN models predicting the 8 month survival of unresectable pancreatic cancer patients. These models may help to optimize personalized patient management.

Who Can Avoid Systematic Biopsy Without Missing Clinically Significant Prostate Cancer in Men Who Undergo Magnetic Resonance Imaging-Targeted Biopsy?

BACKGROUND

The objective of the study was to identify a subset of men who can avoid systematic multisite biopsy (SyB) among those undergoing magnetic resonance imaging (MRI)-targeted transperineal 4-core biopsy (TgB) without missing clinically significant cancer (SC).

PATIENTS AND METHODS

From April 2013 to December 2017, 304 men with elevated prostate-specific antigen levels (< 20 ng/mL) or abnormal digital rectal examination and positive MRI findings underwent transrecta ultrasound and MRI-targeted transperineal 4-core with 14-core systematic biopsy. MRI findings were prospectively collected and evaluated using Prostate Imaging-Reporting and Data System version 2 (PI-RADS), and scores ≥3 were considered positive. SC was defined as Gleason score ≥3 + 4 or maximum cancer length ≥5 mm. We evaluated the diagnostic performance of TgB and SyB to detect SC and characterized men who could avoid SyB without missing SC.

RESULTS

Detection rates of any cancer and SC for TgB/SyB/their combination were 59%/63%/68% and 51%/52%/61%, respectively. TgB alone missed 14% (29/207) of any cancer and 16% (29/184) of SC detected using TgB with SyB. In uni- and multivariable analyses, PI-RADS scores of 3 to 4 were independent predictors for missing SC using TgB alone. When restricted to 81 men with PI-RADS scores of 5 (27%), SC was missed using TgB alone only in 3 (4.6% vs. 22% for the remaining 223 men; P = .007).

CONCLUSION

SC was missed using TgB alone in a non-negligible proportion of men who underwent TgB and SyB. SyB might be safely avoided in men with PI-RADS score 5 lesions with reduced risks of missing SC.

Comparison of cancer detection rates by transrectal prostate biopsy for prostate cancer using two different nomograms based on patient's age and prostate volume

Background

The aim of this study is to evaluate the efficacy of two different Nara Urological Research and Treatment Group (NURTG) nomograms allocating 6–12 biopsy cores based on age and prostate volume.

Materials and methods

From April 2006 to July 2014, a total of 1,605 patients who underwent initial prostate biopsy were enrolled. Based on a nomogram taking the patient’s age and prostate volume into consideration, 6–12 biopsy cores were allocated. Two types of nomogram were used, for the former group (before March 2009) and latter group (March 2009 onward). Cancer detection rates in all patients and those with prostate-specific antigen values in the gray zone (4.0–10 ng/mL) were compared. Predictive parameters for detection of prostate cancer in gray-zone patients were also investigated.

Results

The cancer detection rates in all patients and those in the gray zone were 48% and 38% in the former group and 54% and 41% in the latter group, respectively. The cancer detection rate in all patients was significantly higher in the latter group compared with the former group, but detection in gray-zone patients did not show a significant difference between the two groups (P=0.011 and P=0.37, respectively). Multivariate analysis indicated that age, digital rectal examination, prostate volume, transrectal ultrasonography findings, and volume/biopsy ratio were significant predictive parameters in gray-zone patients. The clinically insignificant cancer detection rate was significantly lower in the latter group compared with the former group (P=0.0008).

Conclusion

The latter nomogram provided more acceptable detection rates of clinically significant and insignificant cancer than the former one, and we consider that an initial maximum 12-core transrectal ultrasound-guided needle biopsy may be sufficient for prostate cancer diagnosis.

Factors associated with de novo metastatic disease in invasive breast cancer: comparison of artificial neural network and logistic regression models

Background

De novo metastasis of breast cancer is a complex clinical issue to be identified. This study was the first to construct artificial neural networks (ANN) and logistic regression (LR) models with comparison to find out important factors associated with occurrence of de novo metastasis in invasive breast cancer.

Methods

A total of 40,899 patients diagnosed with de novo metastatic breast cancer in 2010 from Surveillance, Epidemiology and End Results (SEER) Cancer database were enrolled. ANN models and LR models were constructed based on thirteen relevant factors by 10-fold cross-validation approach respectively. Evaluation indexes as well as processing time were compared.

Results

Overall area under ROC curve (AUC) value of ANN models was significantly higher than that of LR models (0.917±0.01 vs. 0.844±0.011, P<0.001). In ANN models, number of positive ipsilateral axillary lymph nodes, tumor size, lymph node ratio (LNR) and regional lymph nodes status were important associated factors. While under the same experiment environment, ANN models obviously took much more processing time than LR models did (14,400 vs. 15 minutes for 10-fold cross-validation).

Conclusions

ANN models outperformed traditional LR models in identifying de novo metastasis of breast cancer. On the other hand, the much longer processing time of ANN models should also be considered.

A Nomogram Based on a TRUS Five-Grade Scoring System for the Prediction of Prostate Cancer and High Grade Prostate Cancer at Initial TRUS-Guided Biopsy

Purpose: To evaluate the efficacy of transrectal ultrasound five-grade scoring system (TRUS-5) in predicting prostate cancer (PCa) and high grade PCa (HGPCa), compared with TRUS two-grade scoring system (TRUS-2), and establish a TRUS-5 based nomogram for the prediction of PCa and HGPCa at initial biopsy (IPBx).

Methods: Data were collected from 862 men who underwent initial TRUS-guided 12-core prostate biopsy. Age, prostate-specific antigen (PSA), percent free PSA, digital rectal examination (DRE), prostate volume (PV), PSA density (PSAD) and TRUS findings were included in the analysis. For TRUS-5, the probability of PCa was quantified on a scale from 1 (benign) to 5 (malignant). TRUS-2 used the grades “normal” and “suspicious”. After univariate and multivariate logistic regression analyses, nomogram models were developed and internally validated based on independent predictors to predict the probability of PCa and HGPCa.

Results: Overall PCa was detected in 42% (362/862) with 26.22% (226/862) showing HGPCa. TRUS-5 significantly outperformed TRUS-2 for the risk prediction of PCa and HGPCa (area under the receiver operating characteristic curve [AUC]: 0.787 vs. 0.694 for PCa, 0.841 vs. 0.713 for HGPCa, P<0.05). The TRUS-5 based nomogram showed higher AUCs (0.905 for PCa, 0.903 for HGPCa) than PSA alone, clinical base model, the TRUS-2 based model, and other predictive models (P<0.05).

Conclusions: TRUS-5 represents a better imaging predictor than TRUS-2 for PCa and HGPCa. Our TRUS-5 based nomogram models performed well for the prediction of PCa and HGPCa at IPBx, which may help to make the decision to biopsy.

Development of a Ready-to-Use Graphical Tool Based on Artificial Neural Network Classification: Application for the Prediction of Late Fecal Incontinence After Prostate Cancer Radiation Therapy

PURPOSE

This study was designed to apply artificial neural network (ANN) classification methods for the prediction of late fecal incontinence (LFI) after high-dose prostate cancer radiation therapy and to develop a ready-to-use graphical tool.

MATERIALS AND METHODS

In this study, 598 men recruited in 2 national multicenter trials were analyzed. Information was recorded on comorbidity, previous abdominal surgery, use of drugs, and dose distribution. Fecal incontinence was prospectively evaluated through self-reported questionnaires. To develop the ANN, the study population was randomly split into training (n = 300), validation (n = 149), and test (n = 149) sets. Mean grade of longitudinal LFI (ie, expressed as the average incontinence grade over the first 3 years after radiation therapy) ≥1 was considered the endpoint. A suitable subset of variables able to better predict LFI was selected by simulating 100,000 ANN configurations. The search for the definitive ANN was then performed by varying the number of inputs and hidden neurons from 4 to 5 and from 1 to 9, respectively. A final classification model was established as the average of the best 5 among 500 ANNs with the same architecture. An ANN-based graphical method to compute LFI prediction was developed to include one continuous and n dichotomous variables.

RESULTS

An ANN architecture was selected, with 5 input variables (mean dose, previous abdominal surgery, use of anticoagulants, use of antihypertensive drugs, and use of neoadjuvant and adjuvant hormone therapy) and 4 hidden neurons. The developed classification model correctly identified patients with LFI with 80.8% sensitivity and 63.7% ± 1.0% specificity and an area under the curve of 0.78. The developed graphical tool may efficiently classify patients in low, intermediate, and high LFI risk classes.

CONCLUSIONS

An ANN-based model was developed to predict LFI. The model was translated in a ready-to-use graphical tool for LFI risk classification, with direct interpretation of the role of the predictors.

Could Magnetic Resonance Imaging Help to Identify the Presence of Prostate Cancer Before Initial Biopsy? The Development of Nomogram Predicting the Outcomes of Prostate Biopsy in the Chinese Population

PURPOSE

This study was designed to investigate the effectiveness of magnetic resonance imaging (MRI) in diagnosing prostate cancer (PCa) and high-grade prostate cancer (HGPCa) before transrectal ultrasound (TRUS)-guided biopsy.

METHODS

The clinical data of 894 patients who received TRUS-guided biopsy and prior MRI test from a large Chinese center was reviewed. Based on Prostate Imaging Reporting and Data System (PI-RADS) scoring, all MRIs were re-reviewed and assigned as Grade 0-2 (PI-RADS 1-2; PI-RADS 3; PI-RADS 4-5). We constructed two models both in predicting PCa and HGPCa (Gleason score ≥ 4 + 3): Model 1 with MRI and Model 2 without MRI. Other clinical factors include age, digital rectal examination, PSA, free-PSA, volume, and TRUS.

RESULTS

PCa and HGPCa were present in 434 (48.5 %) and 218 (24.4 %) patients. An MRI Grade 0, 1, and 2 were assigned in 324 (36.2 %), 193 (21.6 %) and 377 (42.2 %) patients, which was associated with the presence of PCa (p < 0.001) and HGPCa (p < 0.001). Particularly in patients aged ≤55 years, the assignment of MRI Grade 0 was correlated with extremely low rate of PCa (1/27) and no HGPCa. The c-statistic of Model 1 and Model 2 for predicting PCa was 0.875 and 0.841 (Z = 4.2302, p < 0.001), whereas for predicting HGPCa was 0.872 and 0.850 (Z = 3.265, p = 0.001). Model 1 exhibited higher sensitivity and specificity at same cutoffs, and decision-curve analysis also suggested the favorable clinical utility of Model 1.

CONCLUSIONS

Prostate MRI before biopsy could predict the presence of PCa and HGPCa, especially in younger patients. The incorporation of MRI in nomograms could increase predictive accuracy.

Development and external multicenter validation of Chinese Prostate Cancer Consortium prostate cancer risk calculator for initial prostate biopsy

OBJECTIVE

Substantial differences exist in the relationship of prostate cancer (PCa) detection rate and prostate-specific antigen (PSA) level between Western and Asian populations. Classic Western risk calculators, European Randomized Study for Screening of Prostate Cancer Risk Calculator, and Prostate Cancer Prevention Trial Risk Calculator, were shown to be not applicable in Asian populations. We aimed to develop and validate a risk calculator for predicting the probability of PCa and high-grade PCa (defined as Gleason Score sum 7 or higher) at initial prostate biopsy in Chinese men.

MATERIALS AND METHODS

Urology outpatients who underwent initial prostate biopsy according to the inclusion criteria were included. The multivariate logistic regression-based Chinese Prostate Cancer Consortium Risk Calculator (CPCC-RC) was constructed with cases from 2 hospitals in Shanghai. Discriminative ability, calibration and decision curve analysis were externally validated in 3 CPCC member hospitals.

RESULTS

Of the 1,835 patients involved, PCa was identified in 338/924 (36.6%) and 294/911 (32.3%) men in the development and validation cohort, respectively. Multivariate logistic regression analyses showed that 5 predictors (age, logPSA, logPV, free PSA ratio, and digital rectal examination) were associated with PCa (Model 1) or high-grade PCa (Model 2), respectively. The area under the curve of Model 1 and Model 2 was 0.801 (95% CI: 0.771-0.831) and 0.826 (95% CI: 0.796-0.857), respectively. Both models illustrated good calibration and substantial improvement in decision curve analyses than any single predictors at all threshold probabilities. Higher predicting accuracy, better calibration, and greater clinical benefit were achieved by CPCC-RC, compared with European Randomized Study for Screening of Prostate Cancer Risk Calculator and Prostate Cancer Prevention Trial Risk Calculator in predicting PCa.

CONCLUSIONS

CPCC-RC performed well in discrimination and calibration and decision curve analysis in external validation compared with Western risk calculators. CPCC-RC may aid in decision-making of prostate biopsy in Chinese or in other Asian populations with similar genetic and environmental backgrounds.

The optimal number of initial prostate biopsy cores in daily practice: a prospective study using the Nara Urological Research and Treatment Group nomogram

Background

To elucidate the optimal number of prostate biopsy cores using a nomogram allocating 6–12 biopsy cores, the number generally used in daily practice, based on age and prostate volume (PV).

Methods

We enrolled 936 patients who received an initial prostate biopsy from April 2006 to January 2009. A number of 6–12 biopsy cores was allocated based on age and PV Nara Urological Research and Treatment Group (NURTG) nomogram. To elucidate the predictive parameters of cancer detection in patients with a prostate specific antigen (PSA) value in the gray zone, univariate and multivariate logistic regression analysis were carried out.

Results

The total cancer detection rate and the cancer detection rate in the PSA gray zone (4.1–10.0 ng/mL) were 48.0 and 37.6 %, respectively. The cancer detection rates in the gray zone stratified by patient age of ≤59, 60–64, 65–69, 70–74, 75–79, and ≥80 years were 28.4, 35.0, 26.9, 37.9, 45.7, and 54.8 %, respectively. The significant predictive parameters of cancer detection in the gray zone were age, volume biopsy ratio (VBR: PV divided by number of biopsy cores), PSA density (PSAD), digital rectal examination findings, and transrectal ultrasound findings in univariate analyses. Finally, age, VBR, and PSAD were independent parameters to predict cancer detection in the gray zone. The adverse event profile was acceptable.

Conclusions

Our present study revealed that the cancer detection rate using the NURTG nomogram allocating 6–12 biopsy cores, the number generally used in daily practice, based on age and PV, could provide similar efficacy as previous studies involving more biopsy cores. In older patients the number of biopsy cores could be reduced.

Do prostate cancer risk models improve the predictive accuracy of PSA screening? A meta-analysis

BACKGROUND

Despite the extensive development of risk prediction models to aid patient decision-making on prostate screening, it is unknown whether these models could improve predictive accuracy of PSA testing to detect prostate cancer (PCa). The objective of this study was to perform a systematic review to identify PCa risk models and to assess the model's performance to predict PCa by conducting a meta-analysis.

DESIGN

A systematic literature search of Medline was conducted to identify PCa predictive risk models that used at least two variables, of which one of the variables was prostate-specific antigen (PSA) level. Model performance (discrimination and calibration) was assessed. Prediction models validated in ≥5 study populations and reported area under the curve (AUC) for prediction of any or clinically significant PCa were eligible for meta-analysis. Summary AUC and 95% CIs were calculated using a random-effects model.

RESULTS

The systematic review identified 127 unique PCa prediction models; however, only six models met study criteria for meta-analysis for predicting any PCa: Prostataclass, Finne, Karakiewcz, Prostate Cancer Prevention Trial (PCPT), Chun, and the European Randomized Study of Screening for Prostate Cancer Risk Calculator 3 (ERSPC RC3). Summary AUC estimates show that PCPT does not differ from PSA testing (0.66) despite performing better in studies validating both PSA and PCPT. Predictive accuracy to discriminate PCa increases with Finne (AUC = 0.74), Karakiewcz (AUC = 0.74), Chun (AUC = 0.76) and ERSPC RC3 and Prostataclass have the highest discriminative value (AUC = 0.79), which is equivalent to doubling the sensitivity of PSA testing (44% versus 21%) without loss of specificity. The discriminative accuracy of PCPT to detect clinically significant PCa was AUC = 0.71. Calibration measures of the models were poorly reported.

CONCLUSIONS

Risk prediction models improve the predictive accuracy of PSA testing to detect PCa. Future developments in the use of PCa risk models should evaluate its clinical effectiveness in practice.

Configuration and validation of a novel prostate disease nomogram predicting prostate biopsy outcome: A prospective study correlating clinical indicators among Filipino adult males with elevated PSA level

Objective

To configure and validate a novel prostate disease nomogram providing prostate biopsy outcome probabilities from a prospective study correlating clinical indicators and diagnostic parameters among Filipino adult male with elevated serum total prostate specific antigen (PSA) level.

Methods

All men with an elevated serum total PSA underwent initial prostate biopsy at our institution from January 2011 to August 2014 were included. Clinical indicators, diagnostic parameters, which include PSA level and PSA-derivatives, were collected as predictive factors for biopsy outcome. Multiple logistic-regression analysis involving a backward elimination selection procedure was used to select independent predictors. A nomogram was developed to calculate the probability of the biopsy outcomes. External validation of the nomogram was performed using separate data set from another center for determination of sensitivity and specificity. A receiver-operating characteristic (ROC) curve was used to assess the accuracy in predicting differential biopsy outcome.

Results

Total of 552 patients was included. One hundred and ninety-one (34.6%) patients had benign prostatic hyperplasia, and 165 (29.9%) had chronic prostatitis. The remaining 196 (35.5%) patients had prostate adenocarcinoma. The significant independent variables used to predict biopsy outcome were age, family history of prostate cancer, prior antibiotic intake, PSA level, PSA-density, PSA-velocity, echogenic findings on ultrasound, and DRE status. The areas under the receiver-operating characteristic curve for prostate cancer using PSA alone and the nomogram were 0.688 and 0.804, respectively.

Conclusion

The nomogram configured based on routinely available clinical parameters, provides high predictive accuracy with good performance characteristics in predicting the prostate biopsy outcome such as presence of prostate cancer, high Gleason prostate cancer, benign prostatic hyperplasia, and chronic prostatitis.

Random biopsy: when, how many and where to take the cores?

PURPOSE

The optimal random prostate biopsy scheme (PBx) in the initial and repeated setting is still an issue of controversy. We performed an analysis of the recent literature about the prostate biopsy techniques.

METHODS

We performed a clinical and critical literature review by searching MEDLINE database from January 2005 up to January 2014. Electronic searches were limited to the English language, and the keywords prostate cancer, prostate biopsy, transrectal ultrasound, transperineal prostate biopsy were used.

RESULTS

Prostate biopsy strategy in initial setting. According to the literature and the major international guidelines, the recommended approach in initial setting is still the extended scheme (EPBx) (12 cores). However, there is now a growing evidence in the literature that (a) saturation PBx (>20 cores) (SPBx) might be indicated in patients with PSA <10 ng/ml or low PSA density or large prostate and (b) an individualized approach with more than 12 cores according to the clinical characteristics of the patients may optimize cancer detection in the single patient. Moreover, in the era of multi-parametric MRI (mpMRI), EPBx or SPBX may be substituted by mpMRI-targeted biopsies that have demonstrated superiority over systematic random biopsies for the detection of clinically significant disease and representation of disease burden, while deploying fewer cores. Prostate biopsy strategy in repeat setting. How and how many cores should be taken in the different scenarios in the repeated setting is still unclear. SPBx clearly improves cancer detection if clinical suspicion persists after previous biopsy with negative findings and is able to provide an accurate prediction of prostate tumour volume and grade. Nevertheless, international guidelines do not strongly recommended SPBx in all situations of repeated setting. In the active surveillance and in focal therapy protocols, the optimal schemes have to be defined.

CONCLUSIONS

The course of PBx has changed significantly from sextant biopsies to systematic and from extended to SPBx schemes. The issue about the number and location of the cores is still a matter of debate both in initial and in repeat setting. At present, EPBx is sufficient in most of the cases to provide adequate diagnosis and prostate cancer characterization in the initial setting, while SPBx seems to be necessary in repeat setting. The PBx schemes are evolving also because the scenario in which a PBx is necessary is changing. Random prostate PBx do not represent the future, while imaging target biopsy are becoming more popular.

Development, validation and comparison of two nomograms predicting prostate cancer at initial 12-core biopsy

AIM

Our aim was to establish, validate and compare two nomograms in an Iranian population for the first time using clinical, laboratory and transrectal ultrasonography (TRUS) findings for predicting prostate cancer at initial biopsy.

METHODS

Data were collected on a total of 916 men referred for an initial prostate biopsy in our center in a 7-year period. Variables analyzed included age, prostate-specific antigen (PSA), free/total PSA (%fPSA), digital rectal examination (DRE) findings, prostate volume (PV) and presence of hypoechoic lesion on TRUS. Univariate logistic regression models were fitted to test cancer predictors. Two multivariate logistic regression models were fitted to create nomograms. Both models were internally validated. Calibration of nomograms was assessed graphically. The area under the receiver operating characteristic curve (AUC) was calculated as a scale of discrimination and predictive accuracy and also used to compare models.

RESULTS

Prostate cancer was detected in 221/669 (33%) men. Based on univariate logistic regression, all of variables except DRE were significant predictors of prostate cancer, with highest AUC for PV (AUC 0.696, 95% CI 0.653-0.738).AUC of nomogram with and without TRUS findings and PSA alone were 0.791, 0.721 and 0.624, respectively. In internal validation, both nomograms had acceptable calibration plots.

CONCLUSION

Our nomogram based on age, DRE, PSA, %fPSA and TRUS finding was significantly more accurate in predicting initial prostate biopsy outcome in men.

Management of men with a suspicion of prostate cancer after negative initial prostate biopsy results

INTRODUCTION

For men with elevated prostate-specific antigen (PSA), appropriate management after negative prostate biopsy remains controversial. After determining PSA kinetics, subsequent follow-up was considered.

PATIENTS AND METHODS

A total of 115 cases with negative repeat biopsy were followed by evaluating PSA kinetics and ratio of percent free PSA (F/T) and by performing second repeat biopsy.

RESULTS

Eighteen cancer cases were diagnosed. Shorter PSA doubling times and faster velocities were found in cancer cases compared with cases without cancer. We observed a clear decrease in F/T among cancer cases.

CONCLUSIONS

To avoid unnecessary repeat biopsies, cases with a suspicion of cancer after negative biopsy can be divided into two groups: one that requires additional biopsies and one with an average change in PSA of <1 ng/ml/year and no change in F/T, which is recommended for surveillance as stable disease without biopsy over a specified time period.

Anatomic segmentation improves prostate cancer detection with artificial neural networks analysis of 1H magnetic resonance spectroscopic imaging

PURPOSE

To assess whether an artificial neural network (ANN) model is a useful tool for automatic detection of cancerous voxels in the prostate from (1)H-MRSI datasets and whether the addition of information about anatomical segmentation improves the detection of cancer.

MATERIALS AND METHODS

The Institutional Review Board approved this HIPAA-compliant study and waived informed consent. Eighteen men with prostate cancer (median age, 55 years; range, 36-71 years) who underwent endorectal MRI/MRSI before radical prostatectomy were included in this study. These patients had at least one cancer area on whole-mount histopathological map and at least one matching MRSI voxel suspicious for cancer detected. Two ANN models for automatic classification of MRSI voxels in the prostate were implemented and compared: model 1, which used only spectra as input, and model 2, which used the spectra plus information from anatomical segmentation. The models were trained, tested and validated using spectra from voxels that the spectroscopist had designated as cancer and that were verified on histopathological maps.

RESULTS

At ROC analysis, model 2 (AUC = 0.968) provided significantly better (P = 0.03) classification of cancerous voxels than did model 1 (AUC = 0.949).

CONCLUSION

Automatic analysis of prostate MRSI to detect cancer using ANN model is feasible. Application of anatomical segmentation from MRI as an additional input to ANN improves the accuracy of detecting cancerous voxels from MRSI.

Nomogram using transrectal ultrasound-derived information predicting the detection of high grade prostate cancer on initial biopsy

PURPOSE

To develop a nomogram using transrectal ultrasound (TRUS)-derived information for predicting high grade (HG) prostate cancer (PCa) on initial biopsy.

METHODS

Data were collected on 1,048 men with serum prostate-specific antigen (PSA) levels 4.0 to 9.9 ng/mL who underwent an initial prostate biopsy. Two logistic regression-based nomograms were constructed to predict the detection of PCa. Nomogram-1 incorporated age, digital rectal examination, PSA and percent free PSA data, whereas nomogram-2 incorporated those factors plus TRUS-derived information (i.e., prostate volume and the presence of hypoechoic lesions). The prediction of any PCa and HGPCa (Gleason score≥7) were determined. Twenty percent of the data were randomly reserved for study validation, and the predictive accuracies of the two nomograms were directly compared.

RESULTS

Of the 1,048 men who underwent biopsy, 216 (20.6%) were found to have any PCa, and 97 (9.3%) were found to have HGPCa. All six risk factors were found to be independent predictors for both any PCa and HGPCa. The area under curve (AUC) for nomogram-2 was 0.76 (95% confidence interval [CI], 0.72 to 0.81) for predicting any PCa, and 0.83 (95% CI, 0.79 to 0.88) for predicting HGPCa. These AUCs were greater than those for nomogram-1 (0.72 [95% CI, 0.68 to 0.76 for any PCa; P<0.001], 0.78 [95% CI, 0.72 to 0.83 for HGPCa; P<0.001]). Removing the TRUS-derived information from nomogram-2 resulted in an incremental AUC decrease of 0.052 for any PCa and 0.063 for HGPCa.

CONCLUSIONS

The nomogram using TRUS-derived information had a high predictive accuracy for HGPCa on initial prostate biopsy.

Unmet needs in the prediction and detection of metastases in prostate cancer

The therapeutic landscape for the treatment of advanced prostate cancer is rapidly evolving, especially for those patients with metastatic castration-resistant prostate cancer (CPRC). Despite advances in therapy options, the diagnostic landscape has remained relatively static, with few guidelines or reviews addressing the optimal timing or methodology for the radiographic detection of metastatic disease. Given recent reports indicating a substantial proportion of patients with CRPC thought to be nonmetastatic (M0) are in fact metastatic (M1), there is now a clear opportunity and need for improvement in detection practices. Herein, we discuss the current status of predicting the presence of metastatic disease, with a particular emphasis on the detection of the M0 to M1 transition. In addition, we review current data on newer imaging technologies that are changing the way metastases are detected. Whether earlier detection of metastatic disease will ultimately improve patient outcomes is unknown, but given that the therapeutic options for those with metastatic and nonmetastatic CPRC vary, there are considerable implications of how and when metastases are detected.

Impact of lower urinary tract symptoms on prostate cancer risk among Japanese men with prostate-specific antigen <10 ng/mL and non-suspicious digital rectal examination

OBJECTIVE

To investigate the association between lower urinary tract symptoms status and prostate cancer risk at initial extended biopsy.

METHODS

Between 2005 and 2011, the International Prostate Symptom Score was completed on 1467 consecutive men with prostate-specific antigen <10 ng/mL and non-suspicious digital rectal examination. After excluding 308 men treated with alpha-blockers, the remaining 1159 men were enrolled in the present study. Lower urinary tract symptoms status was divided into absent or mild (International Prostate Symptom Score scores of 0-7) and moderate or severe lower urinary tract symptoms (International Prostate Symptom Score scores of 8-35). The risks of prostate cancer diagnosis and high-grade (Gleason score ≥4 + 3) prostate cancer diagnosis in relation to lower urinary tract symptoms status was evaluated using logistic regression. A stratified analysis based on prostate volume (<30 cc, 30-50 cc and >50 cc) was also carried out.

RESULTS

Of 1159 patients, 421 (36.3%) had a positive biopsy and 590 (51.0%) had moderate or severe lower urinary tract symptoms. On multivariate analysis, absent or mild lower urinary tract symptoms had a significant and positive impact on the risk of prostate cancer and high-grade disease (odds ratio 1.64 and 1.70, P = 0.0007 and 0.0121, respectively). Furthermore, the aforementioned findings for prostate cancer detection did not change throughout every prostate volume subgroup. In contrast, in men with prostate volume ≤50 cc, but not in those with prostate volume >50 cc, prostate-specific antigen or %free prostate-specific antigen remained as a significant predictor of prostate cancer.

CONCLUSION

In men with elevated prostate-specific antigen, absent or mild lower urinary tract symptoms are positively associated with prostate cancer and high-grade disease regardless of the prostate volume. This finding is especially useful in men with enlarged prostates.

Usefulness of pre-biopsy multiparametric magnetic resonance imaging and clinical variables to reduce initial prostate biopsy in men with suspected clinically localized prostate cancer

PURPOSE

We evaluated the usefulness of pre-biopsy multiparametric magnetic resonance imaging and clinical variables to decrease initial prostate biopsies.

MATERIALS AND METHODS

We prospectively evaluated 351 consecutive men with prostate specific antigen between 2.5 and 20 ng/ml, and/or digital rectal examination suspicious for clinically localized disease. All men underwent pre-biopsy multiparametric magnetic resonance imaging and initial 14 to 29-core biopsy, including anterior sampling. Three definitions of significant cancer were defined based on Gleason score and cancer volume (percent positive core and/or maximum cancer length). The overall cohort was divided into men at low risk-prostate specific antigen less than 10 ng/ml and normal digital rectal examination, and high risk-prostate specific antigen 10 ng/ml or greater and/or abnormal digital rectal examination. We evaluated the frequency of significant cancer according to magnetic resonance imaging and risk categories. Clinical variables as significant cancer predictors were analyzed using logistic regression. The sensitivity, specificity, and positive and negative predictive values of magnetic resonance imaging were calculated with or without clinical variables for significant cancer.

RESULTS

The frequency of significant cancer in men with negative vs positive magnetic resonance imaging was 9% to 13% vs 43% to 50% in the low risk group and 47% to 51% vs 68% to 71% in the high risk group. In men at low risk with negative magnetic resonance imaging prostate volume was the only significant predictor of significant cancer. In the low risk group the negative predictive value for significant cancer of a combination of positive magnetic resonance imaging and lower prostate volume (less than 33 ml) was 93.7% to 97.5%.

CONCLUSIONS

Pre-biopsy multiparametric magnetic resonance imaging along with prostate volume decreases the number of initial prostate biopsies by discriminating between significant cancer and other cancer in men with prostate specific antigen less than 10 ng/ml and normal digital rectal examination.

Artificial neural networks and prostate cancer--tools for diagnosis and management

Artificial neural networks (ANNs) are mathematical models that are based on biological neural networks and are composed of interconnected groups of artificial neurons. ANNs are used to map and predict outcomes in complex relationships between given 'inputs' and sought-after 'outputs' and can also be used find patterns in datasets. In medicine, ANN applications have been used in cancer diagnosis, staging and recurrence prediction since the mid-1990s, when an enormous effort was initiated, especially in prostate cancer detection. Modern ANNs can incorporate new biomarkers and imaging data to improve their predictive power and can offer a number of advantages as clinical decision making tools, such as easy handling of distribution-free input parameters. Most importantly, ANNs consider nonlinear relationships among input data that cannot always be recognized by conventional analyses. In the future, complex medical diagnostic and treatment decisions will be increasingly based on ANNs and other multivariate models.

A nomogram based on age, prostate-specific antigen level, prostate volume and digital rectal examination for predicting risk of prostate cancer

Nomograms for predicting the risk of prostate cancer developed using other populations may introduce sizable bias when applied to a Chinese cohort. In the present study, we sought to develop a nomogram for predicting the probability of a positive initial prostate biopsy in a Chinese population. A total of 535 Chinese men who underwent a prostatic biopsy for the detection of prostate cancer in the past decade with complete biopsy data were included. Stepwise logistic regression was used to determine the independent predictors of a positive initial biopsy. Age, prostate-specific antigen (PSA), prostate volume (PV), digital rectal examination (DRE) status, % free PSA and transrectal ultrasound (TRUS) findings were included in the analysis. A nomogram model was developed that was based on these independent predictors to calculate the probability of a positive initial prostate biopsy. A receiver-operating characteristic curve was used to assess the accuracy of using the nomogram and PSA levels alone for predicting positive prostate biopsy. The rate for positive initial prostate biopsy was 41.7% (223/535). The independent variables used to predict a positive initial prostate biopsy were age, PSA, PV and DRE status. The areas under the receiver-operating characteristic curve for a positive initial prostate biopsy for PSA alone and the nomogram were 79.7% and 84.8%, respectively. Our results indicate that the risk of a positive initial prostate biopsy can be predicted to a satisfactory level in a Chinese population using our nomogram. The nomogram can be used to identify and counsel patients who should consider a prostate biopsy, ultimately enhancing accuracy in diagnosing prostate cancer.

Investigation of optimal prostate biopsy schemes for Chinese patients with different clinical characteristics

PURPOSE

To investigate the optimal schemes of prostate biopsy according to prostate volume (PV), age and transrectal ultrasound (TRUS) status in Chinese men.

METHODS

923 consecutive patients who underwent initial TRUS-guided systematic 12-core prostate biopsy (12PBx) were enrolled in this study. The 12PBx was obtained by overlapping of conventional sextant, lateral base, mid-gland of peripheral zone and apex. Each sample was individually marked and inked before fixation. Patients were divided into 8 subgroups on the basis of independent risk factors investigated using logistic regression model. Subsequently, 12PBx was defined as self-control for the analysis of biopsy schemes (6-, 8- and 10PBx) on individual core basis. The prostate cancer detection rates (CDRs) of 6-, 8-, 10- and 12PBx were compared for each individual subgroup.

RESULTS

The 12PBx detected 253 (27.4%) cases of prostate cancer (PCa), of which 67.2, 47.1 and 61.3% were located in the base, mid-gland and apex, respectively. Multivariate analysis indicated that age, TRUS status and PV were independent risk factors for PCa detection. CDR increased with increasing biopsy cores. However, for patients with age ≥65 years, positive TRUS and PV <38.5 cm(3), CDR of 8PBx (30.6%) was similar to 10PBx (32.2%) and 12PBx (32.2%); for patients with age ≥65 years, negative TRUS and PV <38.5 cm(3) or ones with age ≥65 years, positive TRUS and PV ≥38.5 cm(3), 10PBx was as effective as 12PBx in detecting PCa (27.8, 27.5 vs. 28.9, 29.3%, respectively).

CONCLUSION

Age, TRUS status and PV were independent risk factors for PCa detection. Traditional sextant biopsy is not recommended. 8-, 10-, or 12PBx as an individual biopsy scheme might be adopted according to these risk factors for Chinese patients.

A novel equation and nomogram including body weight for estimating prostate volumes in men with biopsy-proven benign prostatic hyperplasia

Anthropometric measurements, e.g., body weight (BW), body mass index (BMI), as well as serum prostate-specific antigen (PSA) and percent-free PSA (%fPSA) have been shown to have positive correlations with total prostate volume (TPV). We developed an equation and nomogram for estimating TPV, incorporating these predictors in men with benign prostatic hyperplasia (BPH). A total of 1852 men, including 1113 at Tokyo Medical and Dental University (TMDU) Hospital as a training set and 739 at Cancer Institute Hospital (CIH) as a validation set, with PSA levels of up to 20 ng ml(-1), who underwent extended prostate biopsy and were proved to have BPH, were enrolled in this study. We developed an equation for continuously coded TPV and a logistic regression-based nomogram for estimating a TPV greater than 40 ml. Predictive accuracy and performance characteristics were assessed using an area under the receiver operating characteristics curve (AUC) and calibration plots. The final linear regression model indicated age, PSA, %fPSA and BW as independent predictors of continuously coded TPV. For predictions in the training set, the multiple correlation coefficient was increased from 0.38 for PSA alone to 0.60 in the final model. We developed a novel nomogram incorporating age, PSA, %fPSA and BW for estimating TPV greater than 40 ml. External validation confirmed its predictive accuracy, with AUC value of 0.764. Calibration plots showed good agreement between predicted probability and observed proportion. In conclusion, TPV can be easily estimated using these four independent predictors.

External validation of an artificial neural network and two nomograms for prostate cancer detection

Background. Multivariate models are used to increase prostate cancer (PCa) detection rate and to reduce unnecessary biopsies. An external validation of the artificial neural network (ANN) "ProstataClass" (ANN-Charité) was performed with daily routine data. Materials and Methods. The individual ANN predictions were generated with the use of the ANN application for PSA and free PSA assays, which rely on age, tPSA, %fPSA, prostate volume, and DRE (ANN-Charité). Diagnostic validity of tPSA, %fPSA, and the ANN was evaluated by ROC curve analysis and comparisons of observed versus predicted probabilities. Results. Overall, 101 (35.8%) PCa were detected. The areas under the ROC curve (AUCs) were 0.501 for tPSA, 0.669 for %fPSA, 0.694 for ANN-Charité, 0.713 for nomogram I, and 0.742 for nomogram II, showing a significant advantage for nomogram II (P = 0.009) compared with %fPSA while the other model did not differ from %fPSA (P = 0.15 and P = 0.41). All models overestimated the predicted PCa probability. Conclusions. Beside ROC analysis, calibration is an important tool to determine the true value of using a model in clinical practice. The worth of multivariate models is limited when external validations were performed without knowledge of the circumstances of the model's development.

Performance of prostate-specific antigen mass in estimation of prostate volume in Japanese men with benign prostate hyperplasia

OBJECTIVES

Obese men with benign prostate hyperplasia might have lower serum prostate-specific antigen because of hemodilution, resulting in underestimation of total prostate volume by serum prostate-specific antigen. The aim of this study was to compare the performance of prostate-specific antigen mass as the absolute amount of prostate-specific antigen protein secreted into circulation with that of serum prostate-specific antigen in the prediction of total prostate volume.

METHODS

A total of 1517 men with serum prostate-specific antigen up to 10 ng/mL, including 1425 with biopsy-proven benign prostate hyperplasia, were enrolled in this study. Height and weight were used to estimate body mass index, body surface area and plasma volume. Prostate-specific antigen mass was calculated as serum prostate-specific antigen multiplied by plasma volume. The association between serum prostate-specific antigen or prostate-specific antigen mass and transrectal ultrasound-measured total prostate volume were evaluated by Pearson's correlation coefficient (Υ), linear regression analyses and receiver operating characteristic curves.

RESULTS

Serum prostate-specific antigen had an inverse relationship with plasma volume, decreasing as plasma volume increased, after adjustment of total prostate volume. Larger total prostate volume per serum prostate-specific antigen was found in men with higher body mass index or plasma volume. Among all participants, the correlation (Υ = 0.456) between prostate-specific antigen mass and total prostate volume was apparently stronger than that (Υ = 0.442) between serum prostate-specific antigen and total prostate volume. Prostate-specific antigen mass outperformed serum prostate-specific antigen at estimating total prostate volume cut-off values of 30 and 40 mL. These findings were more significant in men aged ≥60 years.

CONCLUSIONS

Prostate-specific antigen mass performs better than serum prostate-specific antigen in estimating TPV, especially in men aged ≥60 years.

An overview of prostate diseases and their characteristics specific to Asian men

In this paper, we reviewed the features of common prostate diseases, such as benign prostatic hyperplasia (BPH), prostate cancer (PCa) and chronic prostatitis (CP) that are specific to Asian men. Compared to the Westerners, Asians exhibit particular characteristics of prostate diseases. Through summarizing the epidemiology, symptomatology, diagnostics and therapeutics of these diseases, we find that Asians have a lower incidence of PCa than whites, but the incidences of BPH and CP are similar. Asian men with CP often suffer from fewer disease sites, but have a higher frequency of pain during urination rather than after sexual climax. Prostate-specific antigen (PSA) is a widely used marker for the diagnosis of PCa in both Asian and Western countries. Although the PSA level may be lower in Asians, the threshold used is based on whites. After reviewing the treatments available for these diseases, we did not find a fundamental difference between Asians and whites. Furthermore, the selection for the most appropriate treatment based on the individual needs of patients remains a challenge to urologists in Asia. After considering the traits of prostate diseases that are specific to Asian men, we hope to pave the way for the development of specific diagnostic and therapeutic strategies targeted specifically to Asian men.

Machine learning for improved pathological staging of prostate cancer: a performance comparison on a range of classifiers

OBJECTIVES

Prediction of prostate cancer pathological stage is an essential step in a patient's pathway. It determines the treatment that will be applied further. In current practice, urologists use the pathological stage predictions provided in Partin tables to support their decisions. However, Partin tables are based on logistic regression (LR) and built from US data. Our objective is to investigate a range of both predictive methods and of predictive variables for pathological stage prediction and assess them with respect to their predictive quality based on U.K. data.

METHODS AND MATERIAL

The latest version of Partin tables was applied to a large scale British dataset in order to measure their performances by mean of concordance index (c-index). The data was collected by the British Association of Urological Surgeons (BAUS) and gathered records from over 1700 patients treated with prostatectomy in 57 centers across UK. The original methodology was replicated using the BAUS dataset and evaluated using concordance index. In addition, a selection of classifiers, including, among others, LR, artificial neural networks and Bayesian networks (BNs) was applied to the same data and compared with each other using the area under the ROC curve (AUC). Subsets of the data were created in order to observe how classifiers perform with the inclusion of extra variables. Finally a local dataset prepared by the Aberdeen Royal Infirmary was used to study the effect on predictive performance of using different variables.

RESULTS

Partin tables have low predictive quality (c-index=0.602) when applied on UK data for comparison on patients with organ confined and extra prostatic extension conditions, patients at the two most frequently observed pathological stages. The use of replicate lookup tables built from British data shows an improvement in the classification, but the overall predictive quality remains low (c-index=0.610). Comparing a range of classifiers shows that BNs generally outperform other methods. Using the four variables from Partin tables, naive Bayes is the best classifier for the prediction of each class label (AUC=0.662 for OC). When two additional variables are added, the results of LR (0.675), artificial neural networks (0.656) and BN methods (0.679) are overall improved. BNs show higher AUCs than the other methods when the number of variables raises

CONCLUSION

The predictive quality of Partin tables can be described as low to moderate on U.K. data. This means that following the predictions generated by Partin tables, many patients would received an inappropriate treatment, generally associated with a deterioration of their quality of life. In addition to demographic differences between U.K. and the original U.S. population, the methodology and in particular LR present limitations. BN represents a promising alternative to LR from which prostate cancer staging can benefit. Heuristic search for structure learning and the inclusion of more variables are elements that further improve BN models quality.

Avoiding pitfalls in applying prediction models, as illustrated by the example of prostate cancer diagnosis

BACKGROUND

The use of different mathematical models to support medical decisions is accompanied by increasing uncertainties when they are applied in practice. Using prostate cancer (PCa) risk models as an example, we recommend requirements for model development and draw attention to possible pitfalls so as to avoid the uncritical use of these models.

CONTENT

We conducted MEDLINE searches for applications of multivariate models supporting the prediction of PCa risk. We critically reviewed the methodological aspects of model development and the biological and analytical variability of the parameters used for model development. In addition, we reviewed the role of prostate biopsy as the gold standard for confirming diagnoses. In addition, we analyzed different methods of model evaluation with respect to their application to different populations. When using models in clinical practice, one must validate the results with a population from the application field. Typical model characteristics (such as discrimination performance and calibration) and methods for assessing the risk of a decision should be used when evaluating a model's output. The choice of a model should be based on these results and on the practicality of its use.

SUMMARY

To avoid possible errors in applying prediction models (the risk of PCa, for example) requires examining the possible pitfalls of the underlying mathematical models in the context of the individual case. The main tools for this purpose are discrimination, calibration, and decision curve analysis.

Use of machine learning methods for prediction of acute toxicity in organs at risk following prostate radiotherapy

PURPOSE

The goal of this study is to investigate the advantages of large scale optimization methods vs conventional classification techniques in predicting acute toxicity for urinary bladder and rectum due to prostate irradiation.

METHODS

Clinical and dosimetric data of 321 patients undergoing prostate conformal radiotherapy were recorded. Gastro-intestinal and genito-urinary acute toxicities were scored according to the Radiation Therapy Oncology Group/European Organization for Research and Treatment of Cancer (RTOG/EORTC) scale. Patients were classified in two categories to separate mild (Grade < 2) from severe toxicity levels (Grade > 2). Machine learning methods at different complexity were implemented to predict toxicity as a function of multiple variables. The first approach consisted of a large scale optimization method, based on genetic algorithms (GAs) and artificial neural networks (ANN). The second approach was a binary classifier based on support vector machines (SVM).

RESULTS

The ANN and SVM-based solutions showed comparable prediction accuracy, exhibiting an area under the receiver operating characteristic (ROC) curve of 0.7. Different sensitivity and specificity features were measured for the two approaches. The ANN algorithm showed enhanced sensitivity if combined with appropriate classification criteria.

CONCLUSIONS

The results demonstrate that high sensitivity in toxicity prediction can be achieved with optimized ANNs, that are put forward to represent a valuable support in medical decisions. Future studies will be focused on enlarging the available patient database to increase the reliability of toxicity prediction algorithms and to define optimal classification criteria.

Between-method differences in prostate-specific antigen assays affect prostate cancer risk prediction by nomograms

BACKGROUND

To date, no published nomogram for prostate cancer (PCa) risk prediction has considered the between-method differences associated with estimating concentrations of prostate-specific antigen (PSA).

METHODS

Total PSA (tPSA) and free PSA were measured in 780 biopsy-referred men with 5 different assays. These data, together with other clinical parameters, were applied to 5 published nomograms that are used for PCa detection. Discrimination and calibration criteria were used to characterize the accuracy of the nomogram models under these conditions.

RESULTS

PCa was found in 455 men (58.3%), and 325 men had no evidence of malignancy. Median tPSA concentrations ranged from 5.5 μg/L to 7.04 μg/L, whereas the median percentage of free PSA ranged from 10.6% to 16.4%. Both the calibration and discrimination of the nomograms varied significantly across different types of PSA assays. Median PCa probabilities, which indicate PCa risk, ranged from 0.59 to 0.76 when different PSA assays were used within the same nomogram. On the other hand, various nomograms produced different PCa probabilities when the same PSA assay was used. Although the ROC curves had comparable areas under the ROC curve, considerable differences were observed among the 5 assays when the sensitivities and specificities at various PCa probability cutoffs were analyzed.

CONCLUSIONS

The accuracy of the PCa probabilities predicted according to different nomograms is limited by the lack of agreement between the different PSA assays. This difference between methods may lead to unacceptable variation in PCa risk prediction. A more cautious application of nomograms is recommended.

Initial biopsy outcome prediction in Korean patients-comparison of a noble web-based Korean prostate cancer risk calculator versus prostate-specific antigen testing

We developed and validated a novel Korean prostate cancer risk calculator (KPCRC) for predicting the probability of a positive initial prostate biopsy in a Korean population. Data were collected from 602 Koreans who underwent initial prostate biopsies due to an increased level of prostate-specific antigen (PSA), a palpable nodule upon digital rectal examination (DRE), or a hypoechoic lesion upon transrectal ultrasound (TRUS). The clinical and laboratory variables were analyzed by simple and multiple logistic regression analysis. The area under the receiver operating characteristic curve (AUC) was computed to compare its performance to PSA testing alone. Prostate cancer was detected in 172 (28.6%) men. Independent predictors included age, DRE findings, PSA level, and prostate transitional zone volume. We developed the KPCRC using these variables. The AUC for the selected model was 0.91, and that of PSA testing alone was 0.83 (P < 0.001). The AUC for the selected model with an additional dataset was 0.79, and that of PSA testing alone was 0.73 (P = 0.004). The calculator is available on the website: http://pcrc.korea.ac.kr. The KPCRC improved the performance of PSA testing alone in predicting the risk of prostate cancer in a Korean population. This calculator would be a practical tool for physicians and patients.

A novel repeat biopsy nomogram based on three-dimensional extended biopsy

OBJECTIVES

To develop a nomogram based on a cohort examined with 3-dimensional (3D) protocol for diagnosis of prostate cancer on repeat biopsy.

METHODS

Of 4074 consecutive men undergoing prostate biopsy at our institutions between 2000 and 2009, 775 men with at least 1 previous negative biopsy underwent repeat biopsy with a 3D protocol. Men with previous atypical glands or atypical small acinar proliferation and/or without available prostate-specific antigen (PSA) kinetics information were excluded. The remaining 515 men constituted the study cohort. We developed a logistic regression-based nomogram with 70% of the cohort selected randomly; we validated it with the remaining 30%. Predictive accuracy and performance characteristics were assessed using the area under the receiver operating characteristic curve (AUC) and calibration plots, respectively. The threshold probability was evaluated with decision curve analysis.

RESULTS

We developed a novel repeat biopsy nomogram incorporating age, free to total PSA ratio, prostate volume, history of previous extended biopsy, and PSA doubling time. Validation confirmed predictive accuracy with an AUC value of 0.791. Calibration plots showed good agreement. The decision curve of the nomogram was superior to the decision curve of biopsying all men in a range of threshold probability over 0.15. At the threshold of 0.2, the number of unnecessary biopsies could be reduced by 10 per 100, without missing prostate cancer.

CONCLUSIONS

We developed a novel repeat biopsy nomogram based on a cohort examined with 3D protocol. This repeat biopsy nomogram is clinically beneficial, preventing a substantial number of unnecessary biopsies.

Optimizing performance and interpretation of prostate biopsy: a critical analysis of the literature

CONTEXT

The number and location of biopsy cores and the interpretation of prostate biopsy in different clinical settings remain the subjects of continuing debate.

OBJECTIVE

Our aim was to review the current evidence regarding the performance and interpretation of initial, repeat, and saturation prostatic biopsy.

EVIDENCE ACQUISITION

A comprehensive Medline search was performed using the Medical Subject Heading search terms prostate biopsy, prostate cancer, detection, transrectal ultrasound (TRUS), nomogram, and diagnosis. Results were restricted to the English language, with preference given to those published within the last 3 yr.

EVIDENCE SYNTHESIS

At initial biopsy, a minimum of 10 but not >18 systematic cores are recommended, with 14-18 cores in glands ≥ 50 cm³. Biopsies should be directed laterally, and transition zone (TZ) cores are not recommended in the initial biopsy setting. Further biopsy sets, either as an extended repeat or as a saturation biopsy (≥ 20 cores) including the TZ, are warranted in young and fit men with a persistent suspicion of prostate cancer. An immediate repeat biopsy is not indicated for prior high-grade prostatic intraepithelial neoplasia diagnosis given an adequate extended initial biopsy. Conversely, biopsies with atypical glands that are suspicious but not diagnostic of cancer should be repeated within 3-6 mo. Overall recommendations for further biopsy sets (a third set or more) cannot be made. Transrectal ultrasound-guided systematic biopsies represent the standard-of-care method of prostate sampling. However, transperineal biopsies are an up-to-standard alternative.

CONCLUSIONS

The optimal prostatic biopsy regimen should be based on the individualized clinical setting of the patient and should follow the minimum standard requirements reported in this paper.