The value of second-opinion pathology diagnoses on prostate biopsies from patients referred for management of prostate cancer.

Gleason score (GS) (sum of primary plus secondary grades) is used to predict patients' clinical outcome and to customize treatment strategies for prostate cancer (PC). However, due in part to pathologist misreading, there is significant discrepancy of GS between needle-core biopsies (NCB) and radical prostatectomy specimens. We assessed the requirement for re-evaluating NCB diagnosed by outside pathologists in patients referred to our institution for management of PC. In 100 patients, we reviewed both their original "outside" and second-opinion ("in-house") diagnoses of the same NCB specimens, and compared them with the diagnoses of the whole-mount radical prostatectomy (WMRP) specimens (gold standard for analysis). We found that both outside and in-house biopsy GS vary significantly from the WMRP diagnoses, with GS undergrading substantially predominating above overgrading. Statistical analysis demonstrated that the main diagnostic discrepancy was in the differentiation between primary and secondary Gleason grades (mainly 3 and 4) and that outside NCB GS was significantly less accurate with respect to the WMRP specimens than the in-house NCB GS. In addition, in a different cohort of 65 NCB cases, we found that in 5 out of 11 patients, outside pathologists failed to report the presence of extraprostatic extension, an important feature for diagnosis of a higher pathology stage (pT3a). Since histopathological evaluation is a critical factor for appropriate treatment selection, we recommend that a re-evaluation by in-house urologic pathologists should be performed in all outside NCB specimens before patients are admitted for treatment in any given institution.

The metabolites citrate, myo-inositol, and spermine are potential age-independent markers of prostate cancer in human expressed prostatic secretions.

OBJECTIVES: Due to specific physiological functions, prostatic tissues and fluids have unique metabolic profiles. In this study, proton nuclear magnetic resonance spectroscopy ((1)H-NMRS) is used to assess potential metabolic markers of prostate cancer (PCa) in human expressed prostatic secretions (EPS). METHODS: Metabolic profiles of EPS from 52 men with PCa and from 26 healthy controls were analyzed using quantitative (1)H-NMRS. The metabolites quantified included citrate, spermine, myo-inositol, lactate, alanine, phosphocholine, glutamine, acetate, and hydroxybutyrate. Logistic regression (LR) was used to model the risk of PCa based on metabolite concentrations while adjusting for age. RESULTS: The average age of the EPS donors with PCa was 58.0+/-7.0 years and 52.2+/-12.1 for the healthy donors. The median Gleason score for the men with PCa was 7 (range 5-9). The LR models indicated that the absolute concentrations of citrate, myo-inositol, and spermine were highly predictive of PCa and inversely related to the risk of PCa. The areas under the receiver operating characteristic curves (AUROC) for citrate, myo-inositol and spermine were 0.89, 0.87, and 0.79, respectively. At 90% sensitivity, these metabolites had specificities of 74%, 51%, and 34%, respectively. The LR analysis indicated that absolute levels of these three metabolites were independent of age. CONCLUSIONS: The results indicate that citrate, myo-inositol and spermine are potentially important markers of PCa in human EPS. Further, the absolute concentrations of these metabolites in EPS appear to be independent of age, increasing the potential utility of these markers due to elimination of age as a confounding variable.

Factors contributing to the racial differences in prostate cancer mortality.

OBJECTIVE: To analyse, in a retrospective cohort study, differences in rates of surgical treatment for prostate cancer between African-Americans and White Americans, and to evaluate the extent to which these differences are associated with disparities in survival rates between these groups. PATIENTS AND METHODS: Clinical, pathological, and demographic data from 4279 men diagnosed with clinically localized prostate cancer between 1980 and 1997 were used. The variables assessed included age, disease stage, tumour grade, comorbidities, treatment method, and socio-economic status (SES). Kaplan-Meier survival curves were generated and compared using log-rank tests. The Cox proportional hazards method was used for analyses involving adjustments for potential confounding factors. RESULTS: The surgical treatment rate was 17% for African-American and 28% for White patients (P < 0.001). In those patients treated conservatively or by radiation therapy, both crude and cancer-specific survival rates were lower for African-Americans than for Whites (P < 0.001). However, for patients undergoing surgery, differences in survival between African-Americans and Whites were not statistically significant. According to our models, SES explained 50% and surgical treatment rates approximately 34% of the differences in survival between African-Americans and Whites. CONCLUSIONS: This analysis suggests that the lower prostate cancer survival rates for the African-Americans in the present population can be largely explained by differences in SES and lower surgical treatment rates. Efforts to increase awareness of treatment options among African-American patients may be a way of improving survival in this group.

Racial differences in serum prostate-specific antigen (PSA) doubling time, histopathological variables and long-term PSA recurrence between African-American and white American men undergoing radical prostatectomy for clinically localized prostate cancer.

OBJECTIVE: To determine if there are significant differences in biochemical characteristics, biopsy variables, histopathological data, and rates of prostate-specific antigen (PSA) recurrence between African-American (AA) and white American (WA) men undergoing radical prostatectomy (RP), as AA men are twice as likely to die from prostate cancer than their white counterparts. PATIENTS AND METHODS: We established a cohort of 1058 patients (402 AA, 646 WA) who had RP and were followed for PSA recurrence. Age, race, serum PSA, biopsy Gleason score, clinical stage, pathological stage, and PSA recurrence data were available for the cohort. The chi-square test of proportions and t-tests were used to assess basic associations with race, and log-rank tests and Cox regression models for time to PSA recurrence. Forward stepwise variable selection was used to assess the effect on the risk of PSA recurrence for race, adjusted by the other variables added one at a time. RESULTS: The AA men had higher baseline PSA levels, more high-grade prostatic intraepithelial neoplasia (HGPIN) in the biopsy, and more HGPIN in the pathology specimen than WA men. The AA men also had a shorter mean (sd) PSA doubling time before RP, at 4.2 (4.7) vs 5.2 (5.9) years. However, race was not an independent predictor of PSA recurrence (P = 0.225). Important predictors for PSA recurrence in a multivariable model were biopsy HGPIN (P < 0.014), unilateral vs bilateral cancer (P < 0.006), pathology Gleason score and positive margin status (both P < 0.001). CONCLUSIONS: This study indicates that while there are racial differences in baseline serum PSA and incidence of HGPIN, race is not an independent risk factor for PSA recurrence. Rather, other variables such as pathology Gleason score, bilateral cancers, HGPIN and margin positivity are independently associated with PSA recurrence. The PSA doubling time after recurrence may also be important, leading to the increased mortality of AA men with prostate cancer.

The use of an electronic audience response system for data collection.

BACKGROUND: Gathering complete and accurate data from community groups, particularly medically underserved populations, is challenging. METHODS: An electronic audience response system (ARS) is a novel method for the efficient collection of data while maintaining participant confidentiality in group settings. RESULTS: Because data are captured electronically, an ARS eliminates the need to transfer data from paper forms, reducing errors and the amount of time required for data management. CONCLUSIONS: ARS is a useful data collection tool that works well with diverse populations and greatly increases data accuracy and completeness while maintaining participant confidentiality.

Model to predict prostate biopsy outcome in large screening population with independent validation in referral setting.

OBJECTIVES: To develop a model capable of predicting prostate biopsy outcomes in a large screening population, with independent validation in the referral setting. METHODS: Data from 3814 men participating in the Tyrol screening project were used to develop the model. Prospectively collected data from two independent sites in the United States (Virginia Mason Clinic, Seattle, Wash and Stanford University, Stanford, Calif) were used to validate the model independently. The Tyrol data was split randomly into three cross-validation sets, and a feed-forward, back error-propagation artificial neural network (ANN) was alternately trained on a combination of two of these data sets and validated on the remaining data set. Similarly, three logistic regression (LR) models were produced and validated using identical cross-validation data sets. The Tyrol model with the median area under receiver operating characteristic curve (AUROC) was then validated against the Virginia Mason (n = 491) and Stanford University (n = 483) data sets. RESULTS: The AUROCs for the three cross-validations were 0.74, 0.76, and 0.75 for the ANN and 0.75, 0.76, and 0.75 for the LR models. The mean AUROC for both ANN and LR was 0.75 with a standard deviation of 0.009 for ANN and 0.006 for LR. The AUROCs for the Virginia Mason and Stanford University data were 0.74 (both ANN and LR) and 0.73 (ANN) and 0.72 (LR), respectively. CONCLUSIONS: This model, designed to predict the prostate biopsy outcome, performed accurately and consistently when validated with data from two independent referral centers in the United States, suggesting that it generalizes well and may be of clinical utility to a broad range of patients.

Biochemical recurrence and survival prediction models for the management of clinically localized prostate cancer.

A number of new predictive modeling techniques have emerged in the past several years. These methods, which have been developed in fields such as artificial intelligence research, engineering, and meteorology, are now being applied to problems in medicine with promising results. This review outlines our recent work with use of selected advanced techniques such as artificial neural networks, genetic algorithms, and propensity scoring to develop useful models for estimating the risk of biochemical recurrence and long-term survival in men with clinically localized prostate cancer. In addition, we include a description of our efforts to develop a comprehensive prostate cancer database that, along with these novel modeling techniques, provides a powerful research tool that allows for the stratification of risk for treatment failure and survival by such factors as age, race, and comorbidities. Clinical and pathologic data from 1400 patients were used to develop the biochemical recurrence model. The area under the receiver operating characteristic curve for this model was 0.83, with a sensitivity of 85% and specificity of 74%. For the survival model, data from 6149 men were used. Our analysis indicated that age, income, and comorbidities had a statistically significant impact on survival. The effect of race did not reach statistical significance in this regard. The C index value for the model was 0.69 for overall survival. We conclude that these methods, along with a comprehensive database, allow for the development of models that provide estimates of treatment failure risk and survival probability that are more meaningful and clinically useful than those previously developed.

Artificial neural networks for predictive modeling in prostate cancer.

Artificial neural networks (ANNs) represent a relatively new methodology for predictive modeling in medicine. ANNs, a form of artificial intelligence loosely based on the brain, have a demonstrated ability to learn complex and subtle relationships between variables in medical applications. In contrast with traditional statistical techniques, ANNs are capable of automatically resolving these relationships without the need for a priori assumptions about the nature of the interactions between variables. As with any technique, ANNs have limitations and potential drawbacks. This article provides an overview of the theoretical basis of ANNs, how they function, their strengths and limitations, and examples of how ANNs have been used to develop predictive models for the management of prostate cancer.

Long-term survival probability in men with clinically localized prostate cancer: a case-control, propensity modeling study stratified by race, age, treatment and comorbidities.

PURPOSE: We used a propensity risk scoring approach to model long-term survival for men with clinically localized prostate cancer. We developed comprehensive lookup tables for estimating survival probability stratified by patient age, race, readily available clinical variables, comorbidities and treatment type. MATERIALS AND METHODS: We retrospectively identified a cohort of 1611 men with clinically localized prostate cancer (patients) and 4538 age, race and comorbidity matched controls. Based on demographic and clinical variables propensity risk scoring was used to develop smoothed survival prediction models for patients and controls. Based on these models tables were created to provide 10-year overall survival estimates. The tables are stratified by patient age, race, comorbidity, prostate specific antigen, cancer grade, and treatment type when applicable. RESULTS: Mean patient age was 67.0 years and median baseline prostate specific antigen was 8.5 ng/ml. Of the patients 68% had biopsy cancer grade 2, 39% were black, 29% received conservative treatment, 43% underwent radical prostatectomy and 27% were treated with radiation therapy. Crude and cancer specific 10-year survival rates were 67% and 93%, respectively. Validation C-index values were 0.63 for the cancer specific model and 0.69 for the overall survival model. CONCLUSIONS: These lookup tables provide physicians and patients with realistic estimates of 10-year survival and allow them to compare the impact of cancer vs noncancer factors on patient mortality.

Predicting the outcome of prostate biopsy in a racially diverse population: a prospective study.

OBJECTIVES: To develop a mathematical model to predict prostate biopsy outcome using readily available clinical variables. METHODS: A total of 319 men (78% African American) undergoing transrectal ultrasound-guided prostate biopsy were prospectively studied. The parameters collected included age, race, prostate-specific antigen (PSA) level, PSA density (PSAD), digital rectal examination findings, biopsy history, prostate volume (by transrectal ultrasound), and ultrasound findings. Models were constructed using multivariate logistic regression (LR) analysis and back-propagation artificial neural networks (ANNs). Patient data were randomly split into five cross-validation sets and used to develop and validate the LR and ANN models. RESULTS: Of the 319 men, 39% had a positive biopsy. The mean patient age was 65.1 +/- 8.3 years, with a mean PSA level of 12.6 +/- 24.9 ng/mL and a mean PSAD of 0.31 +/- 0.66 ng/mL/cm(3). Univariate analysis indicated a significant difference in age, PSA level, PSAD, free PSA, digital rectal examination findings, TRUS lesion, and biopsy history between the positive and negative biopsy groups (P <0.01). The mean area under the receiver operating characteristic curve (AUROC) for the five LR models was 0.76 +/- 0.04 (range 0.71 to 0.81). The median LR AUROC was 0.76, with a corresponding specificity of 0.13 at a sensitivity of 0.95. The mean AUROC for the five ANN models was 0.76 +/- 0.04 (range 0.71 to 0.83). The median ANN AUROC was 0.76, with a corresponding specificity of 0.21 at a sensitivity of 0.95. CONCLUSIONS: Two models (LR and ANN) that predict outcome with high efficiency (AUROC = 0.76) were constructed from a contemporary, prospective database. Such models may be useful to patients and physicians alike when assessing the diagnostic strategies available to detect prostate cancer.