Extracting Patient-Centered Outcomes from Clinical Notes in Electronic Health Records: Assessment of Urinary Incontinence After Radical Prostatectomy.

OBJECTIVE: To assess documentation of urinary incontinence (UI) in prostatectomy patients using unstructured clinical notes from Electronic Health Records (EHRs). METHODS: We developed a weakly-supervised natural language processing tool to extract assessments, as recorded in unstructured text notes, of UI before and after radical prostatectomy in a single academic practice across multiple clinicians. Validation was carried out using a subset of patients who completed EPIC-26 surveys before and after surgery. The prevalence of UI as assessed by EHR and EPIC-26 was compared using repeated-measures ANOVA. The agreement of reported UI between EHR and EPIC-26 was evaluated using Cohen's Kappa coefficient. RESULTS: A total of 4870 patients and 716 surveys were included. Preoperative prevalence of UI was 12.7 percent. Postoperative prevalence was 71.8 percent at 3 months, 50.2 percent at 6 months and 34.4 and 41.8 at 12 and 24 months, respectively. Similar rates were recorded by physicians in the EHR, particularly for early follow-up. For all time points, the agreement between EPIC-26 and the EHR was moderate (all p < 0.001) and ranged from 86.7 percent agreement at baseline (Kappa = 0.48) to 76.4 percent agreement at 24 months postoperative (Kappa = 0.047). CONCLUSIONS: We have developed a tool to assess documentation of UI after prostatectomy using EHR clinical notes. Our results suggest such a tool can facilitate unbiased measurement of important PCOs using real-word data, which are routinely recorded in EHR unstructured clinician notes. Integrating PCO information into clinical decision support can help guide shared treatment decisions and promote patient-valued care.

Is it possible to automatically assess pretreatment digital rectal examination documentation using natural language processing? A single-centre retrospective study.

OBJECTIVES: To develop and test a method for automatic assessment of a quality metric, provider-documented pretreatment digital rectal examination (DRE), using the outputs of a natural language processing (NLP) framework. SETTING: An electronic health records (EHR)-based prostate cancer data warehouse was used to identify patients and associated clinical notes from 1 January 2005 to 31 December 2017. Using a previously developed natural language processing pipeline, we classified DRE assessment as documented (currently or historically performed), deferred (or suggested as a future examination) and refused. PRIMARY AND SECONDARY OUTCOME MEASURES: We investigated the quality metric performance, documentation 6 months before treatment and identified patient and clinical factors associated with metric performance. RESULTS: The cohort included 7215 patients with prostate cancer and 426 227 unique clinical notes associated with pretreatment encounters. DREs of 5958 (82.6%) patients were documented and 1257 (17.4%) of patients did not have a DRE documented in the EHR. A total of 3742 (51.9%) patient DREs were documented within 6 months prior to treatment, meeting the quality metric. Patients with private insurance had a higher rate of DRE 6 months prior to starting treatment as compared with Medicaid-based or Medicare-based payors (77.3%vs69.5%, p=0.001). Patients undergoing chemotherapy, radiation therapy or surgery as the first line of treatment were more likely to have a documented DRE 6 months prior to treatment. CONCLUSION: EHRs contain valuable unstructured information and with NLP, it is feasible to accurately and efficiently identify quality metrics with current documentation clinician workflow.

Weakly supervised natural language processing for assessing patient-centered outcome following prostate cancer treatment.

Background: The population-based assessment of patient-centered outcomes (PCOs) has been limited by the efficient and accurate collection of these data. Natural language processing (NLP) pipelines can determine whether a clinical note within an electronic medical record contains evidence on these data. We present and demonstrate the accuracy of an NLP pipeline that targets to assess the presence, absence, or risk discussion of two important PCOs following prostate cancer treatment: urinary incontinence (UI) and bowel dysfunction (BD). Methods: We propose a weakly supervised NLP approach which annotates electronic medical record clinical notes without requiring manual chart review. A weighted function of neural word embedding was used to create a sentence-level vector representation of relevant expressions extracted from the clinical notes. Sentence vectors were used as input for a multinomial logistic model, with output being either presence, absence or risk discussion of UI/BD. The classifier was trained based on automated sentence annotation depending only on domain-specific dictionaries (weak supervision). Results: The model achieved an average F1 score of 0.86 for the sentence-level, three-tier classification task (presence/absence/risk) in both UI and BD. The model also outperformed a pre-existing rule-based model for note-level annotation of UI with significant margin. Conclusions: We demonstrate a machine learning method to categorize clinical notes based on important PCOs that trains a classifier on sentence vector representations labeled with a domain-specific dictionary, which eliminates the need for manual engineering of linguistic rules or manual chart review for extracting the PCOs. The weakly supervised NLP pipeline showed promising sensitivity and specificity for identifying important PCOs in unstructured clinical text notes compared to rule-based algorithms.

Comparison of orthogonal NLP methods for clinical phenotyping and assessment of bone scan utilization among prostate cancer patients.

OBJECTIVE: Clinical care guidelines recommend that newly diagnosed prostate cancer patients at high risk for metastatic spread receive a bone scan prior to treatment and that low risk patients not receive it. The objective was to develop an automated pipeline to interrogate heterogeneous data to evaluate the use of bone scans using a two different Natural Language Processing (NLP) approaches. MATERIALS AND METHODS: Our cohort was divided into risk groups based on Electronic Health Records (EHR). Information on bone scan utilization was identified in both structured data and free text from clinical notes. Our pipeline annotated sentences with a combination of a rule-based method using the ConText algorithm (a generalization of NegEx) and a Convolutional Neural Network (CNN) method using word2vec to produce word embeddings. RESULTS: A total of 5500 patients and 369,764 notes were included in the study. A total of 39% of patients were high-risk and 73% of these received a bone scan; of the 18% low risk patients, 10% received one. The accuracy of CNN model outperformed the rule-based model one (F-measure = 0.918 and 0.897 respectively). We demonstrate a combination of both models could maximize precision or recall, based on the study question. CONCLUSION: Using structured data, we accurately classified patients' cancer risk group, identified bone scan documentation with two NLP methods, and evaluated guideline adherence. Our pipeline can be used to provide concrete feedback to clinicians and guide treatment decisions.

The CPC Risk Calculator: A New App to Predict Prostate-specific Antigen Recurrence During Follow-up After Radical Prostatectomy.

BACKGROUND: It can be challenging to predict the risk of biochemical recurrence (BR) during follow-up after radical prostatectomy (RP) in men who have undetectable prostate-specific antigen (PSA), even years after surgery. OBJECTIVE: To establish and validate a contemporary nomogram that predicts the absolute risk of BR every year after RP in men with undetectable PSA while accounting for competing risks of death. DESIGN, SETTING, AND PARTICIPANTS: A total of 3746 patients from Rigshospitalet (Copenhagen, Denmark) and Stanford Urology (Stanford, CA, USA) who underwent RP between 1995 and 2013 were included. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Time to BR was defined as the first PSA result ≥0.2 ng/ml. BR risk was computed using multiple cause-specific Cox regression including preoperative PSA, pT category, RP Gleason score (GS), and surgical margin (R) status. Death without BR was considered a competing event. The nomogram presents the future risk of BR for a man who is alive and without BR at the time of follow-up. Validation assessed the discrimination and accuracy using time-dependent area under the curve and Brier scores. RESULTS AND LIMITATIONS: The nomogram predicts risk of BR up to 12 yr after RP at an individual level. As example, the risk of BR for a man with pT3a, R-, GS 3 + 4, and preoperative PSA ≤10 ng/ml followed for 5 yr with undetectable PSA is 18% for the next 5 yr. External validation demonstrated both high accuracy and discrimination. The CPC Risk Calculator is available as a free Android and iOS App. Declining discrimination and accuracy after 7 yr of follow-up is the main limitation. CONCLUSIONS: This nomogram can be used as a tool to inform men with undetectable PSA during follow-up after RP about their future risk of BR, and may aid in decisions on the necessity for further follow-up. The nomogram is the first to be available as a free app. PATIENT SUMMARY: We developed an easily interpretable nomogram to evaluate the risk of prostate-specific antigen elevation (cancer recurrence) following complete removal of the prostate (radical prostatectomy). The tool can aid both physicians and patients in evaluating the future risk of cancer recurrence during follow-up after surgery. The model is available as a free mobile app that can be downloaded from the App Store.

Dilution Factor of Quantitative Bacterial Cultures Obtained by Bronchoalveolar Lavage in Patients with Ventilator-Associated Bacterial Pneumonia.

Ventilator-associated bacterial pneumonia (VABP) is a difficult therapeutic problem. Considerable controversy exists regarding the optimal chemotherapy for this entity. The recent guidelines of the Infectious Diseases Society of America and the American Thoracic Society recommend a 7-day therapeutic course for VABP based on the balance of no negative impact on all-cause mortality, less resistance emergence, and fewer antibiotic treatment days, counterbalanced with a higher relapse rate for patients whose pathogen is a nonfermenter. The bacterial burden causing an infection has a substantial impact on treatment outcome and resistance selection. We describe the baseline bronchoalveolar lavage (BAL) fluid burden of organisms in suspected VABP patients screened for inclusion in a clinical trial. We measured the urea concentrations in plasma and BAL fluid to provide an index of the dilution of the bacterial and drug concentrations in the lung epithelial lining fluid introduced by the BAL procedure. We were then able to calculate the true bacterial burden as the diluted colony count times the dilution factor. The median dilution factor was 28.7, with the interquartile range (IQR) being 11.9 to 53.2. Median dilution factor-corrected colony counts were 6.18 log10(CFU/ml) [IQR, 5.43 to 6.46 log10(CFU/ml)]. In a subset of patients, repeat BAL on day 5 showed a good stability of the dilution factor. We previously showed that large bacterial burdens reduce or stop bacterial killing by granulocytes. (This study has been registered at ClinicalTrials.gov under registration no. NCT01570192.).

An Automated Feature Engineering for Digital Rectal Examination Documentation using Natural Language Processing.

Digital rectal examination (DRE) is considered a quality metric for prostate cancer care. However, much of the DRE related rich information is documented as free-text in clinical narratives. Therefore, we aimed to develop a natural language processing (NLP) pipeline for automatic documentation of DRE in clinical notes using a domain-specific dictionary created by clinical experts and an extended version of the same dictionary learned by clinical notes using distributional semantics algorithms. The proposed pipeline was compared to a baseline NLP algorithm and the results of the proposed pipeline were found superior in terms of precision (0.95) and recall (0.90) for documentation of DRE. We believe the rule-based NLP pipeline enriched with terms learned from the whole corpus can provide accurate and efficient identification of this quality metric.

Comparative rates of upstaging and upgrading in Caucasian and Korean prostate cancer patients eligible for active surveillance.

PURPOSE: To investigate the impact of race on the risk of pathological upgrading and upstaging at radical prostatectomy (RP) in an Asian (Korean) and Western (Caucasian) cohort eligible for active surveillance (AS). MATERIALS AND METHODS: We performed a retrospective cohort study of 854 patients eligible for AS who underwent RP in United States (n = 261) and Korea (n = 593) between 2006 and 2015. After adjusting for age, PSA level, and prostate volume, we utilized multivariate logistic regression analysis to assess the effect of race on upgrading or upstaging. RESULTS: There were significant differences between Caucasian and Korean patients in terms of age at surgery (60.2 yr. vs. 64.1 yr.), PSA density (0.115 ng/mL/mL vs. 0.165 ng/mL/mL) and mean number of positive cores (3.5 vs. 2.4), but not in preoperative PSA values (5.11 ng/mL vs. 5.05 ng/mL). The rate of upstaging from cT1 or cT2 to pT3 or higher was not significantly different between the two cohorts (8.8% vs. 11.0%, P = 0.341). However, there were higher rates of upgrading to high-grade cancer (Gleason 4+3 or higher) in Korean patients (9.1%) when compared to Caucasian counterparts (2.7%) (P = 0.003). Multivariate logistic regression analysis showed that age (OR 1.07, P < 0.001) and smaller prostate volume (OR 0.97, P < 0.001), but not race, were significantly associated with upstaging or upgrading. CONCLUSIONS: There were no differences in rates of upgrading or upstaging between Caucasian and Korean men eligible for active surveillance.

Temporal Trends in Clinical and Pathological Characteristics for Men Undergoing Radical Prostatectomy Between 1995 and 2013 at Rigshospitalet, Copenhagen, Denmark, and Stanford University Hospital, United States.

PURPOSE: To analyze how prostate-specific antigen (PSA) screening and practice patterns has affected trends in tumor characteristics in men undergoing radical prostatectomy (RP) in the United States and Denmark. Unlike in the United States, PSA screening has not been recommended in Denmark. PATIENTS AND METHODS: We performed an observational register study using pre- and postoperative data on 2168 Danish patients from Rigshospitalet, Copenhagen, Denmark, and 2236 patients from Stanford University Hospital, Stanford, CA, who underwent RP between 1995 and 2013. Patients were stratified according to Cancer of the Prostate Risk Assessment-Postsurgical (CAPRA-S) risk groups and D'Amico risk classification and were clustered into 4 time periods (1995-1999, 2000-2004, 2005-2009, and 2010-2013). Temporal trends in the proportions of patients of a given variable at the 2 institutions were evaluated with Cochran-Armitage test for trends and chi-square testing. RESULTS: A total of 4404 patients were included. Temporal changes in preoperative PSA, age, grade, and stage was found in both cohorts. Median preoperative PSA declined in both cohorts, while median age increased, with the Danish cohort showing the greatest changes in both PSA and age. In both cohorts, there was a trend for higher-risk preoperative features before RP over time. In 2010-2013, 27.7% and 21.8% of the patients were in the D'Amico high-risk group at Copenhagen and Stanford, respectively. CONCLUSION: Despite recommendation against PSA screening in Denmark, Danish men undergoing RP at Rigshospitalet to a considerable extent now resemble American men undergoing RP at Stanford. At both sites, there is continued trend to reduce the number of men undergoing RP for low-risk prostate cancer.

Mining Electronic Health Records to Extract Patient-Centered Outcomes Following Prostate Cancer Treatment.

The clinical, granular data in electronic health record (EHR) systems provide opportunities to improve patient care using informatics retrieval methods. However, it is well known that many methodological obstacles exist in accessing data within EHRs. In particular, clinical notes routinely stored in EHR are composed from narrative, highly unstructured and heterogeneous biomedical text. This inherent complexity hinders the ability to perform automated large-scale medical knowledge extraction tasks without the use of computational linguistics methods. The aim of this work was to develop and validate a Natural Language Processing (NLP) pipeline to detect important patient-centered outcomes (PCOs) as interpreted and documented by clinicians in their dictated notes for male patients receiving treatment for localized prostate cancer at an academic medical center.

Biologic differences between peripheral and transition zone prostate cancer.

BACKGROUND: Prostate cancer arises in the transition zone (TZ) in approximately 20-25% of cases. Modern biopsy and surveillance protocols, and advances in prostate cancer imaging, have renewed interest in TZ prostate cancers. We compared TZ and PZ prostate cancer to determine if cancer location is independently associated with better outcomes. METHODS: We evaluated an expanded cohort of 1354 men who underwent radical prostatectomy between 1983 and 2003 with updated long-term clinical follow-up. Regression models were used to compare the volume of high-grade (Gleason 4 or 5) cancer and total cancer volume by location. Uni- and multi-variable logistic regression models tested the associations between cancer location and adverse pathologic features. Multivariable proportional hazard models were fit to examine cancer recurrence. RESULTS: Patients with TZ cancer presented with higher pre-operative serum PSA values (11.07 vs. 7.86 ng/ml) and larger total cancer volume (7.1 vs. 3.8 cc). Patients with TZ cancer had decreased odds of seminal vesicle invasion (OR 0.08, 95% CI 0.03, 0.21), extra-capsular extension (OR 0.56, 95% CI 0.35, 0.92), and lymphovascular invasion (OR 0.48, 95% CI 0.27, 0.87) in multivariable models. TZ cancers were independently associated with decreased hazard of tumor recurrence (HR 0.62, 95% CI 0.43, 0.90). CONCLUSIONS: TZ cancer prostate is associated with favorable pathologic features and better recurrence-free survival despite being diagnosed with larger cancers and higher PSA values. Tumor location should be taken into account when stratifying patient risk before and after prostatectomy, particularly with the evolving role of imaging in prostate cancer management.

Experiences of diabetes self-management: a focus group study among Australians with type 2 diabetes.

AIMS AND OBJECTIVES: The aim of this study was to explore the experiences and concerns of individuals with type 2 diabetes mellitus, in a predominantly low socio-economic setting. BACKGROUND: Currently, approximately 1 million Australians have diabetes and rates have more than doubled since 1989. Type 2 diabetes mellitus accounts for approximately 85% of diabetes cases. Risk factors include obesity, older age, low socio-economic status, sedentary lifestyle and ethnicity. Older individuals from low socio-economic backgrounds are particularly at risk of both developing and of mismanaging their condition. DESIGN: Exploratory qualitative design. METHODS: Focus groups were used to collect data from 22 individuals, aged 40 to more than 70 years, with type 2 diabetes mellitus, who were attending local health services for their diabetes care. Focus groups ranged in size from four to eight individuals and all were recorded, transcribed and analysed. Data were analysed using a thematic analysis approach. RESULTS: Participants described their experiences of managing their diabetes as emotionally, physically and socially challenging. Data analysis revealed four main themes including: (1) diabetes the silent disease; (2) a personal journey (3) the work of managing diabetes; and (4) access to resources and services. Throughout, participants highlighted the impact of diabetes on the family, and the importance of family members in providing support and encouragement to assist their self-management efforts. CONCLUSIONS: Participants in this study were generally satisfied with their diabetes care but identified a need for clear simple instruction immediately post-diagnosis, followed by a need for additional informal information when they had gained some understanding of their condition. RELEVANCE TO CLINICAL PRACTICE: Findings reveal a number of unmet information and support needs for individuals with type 2 diabetes mellitus. In particular, it is important for healthcare professionals and family members to recognise the significant emotional burden that diabetes imposes, and the type and quantity of information individuals with diabetes prefer. It is also important to consider levels of health literacy in the community when developing diabetes-related information or programmes.

Matched comparison of robot-assisted, laparoscopic and open radical prostatectomy regarding pathologic and oncologic outcomes in obese patients.

OBJECTIVES: To investigate pathological and oncological outcomes of obese patients who underwent robot-assisted radical prostatectomy (RARP) compared with laparoscopic radical prostatectomy (LRP) or open retropubic radical prostatectomy (RRP) since limited comparative data exist with regard to oncological and survival outcomes. METHODS: A total of 869 patients with body mass index ≥ 30 from two academic centers were identified. A total of 194 patients who underwent RARP were propensity score (PS) matched 1:1 to LRP or RRP cases. PS-matching variables included prostate-specific antigen (PSA), biopsy Gleason score, clinical stage, surgeon experience, and nerve-sparing technique. Predictors of positive surgical margins (PSMs) were analyzed using logistic regression. Predictors of recurrence-free survival (RFS) were analyzed within Cox regression models. Overall survival was compared with RFS using the log-rank test. RESULTS: Pathologic Gleason scores <7, =7, and >7 were found in 24.2, 63.6, and 11.7 % of patients, respectively. There were no statistically significant differences related to pathologic stage or lymph node metastases between surgical techniques. PSM for pT2 disease were observed in 22.9, 17.4, and 19.3 % of patients undergoing RARP, LRP, and RRP, respectively (not significantly different). Preoperative PSA and clinical stage cT2 disease were independently associated with PSM. There were no significant differences in mean 3-year RFS for RARP, LRP, and RRP (87.4, 91.0, and 85.7 %). Biopsy Gleason score >7, PSM, and clinical stage two were independent predictors of decreased RFS. CONCLUSIONS: RARP demonstrates similar pathological and oncological results compared with LRP or RRP for obese patients.

Matched comparison of outcomes following open and minimally invasive radical prostatectomy for high-risk patients.

PURPOSE: Comparative data related to the use of open and minimally invasive surgical approaches for the treatment of high-risk prostate cancer (PCa) remain limited. We determined outcomes of open radical prostatectomy (RRP), laparoscopic RP (LRP), and robot-assisted RP (RARP) in matched cohorts of patients with high-risk prostate cancer. MATERIALS AND METHODS: A total of 805 patients with high-risk PCa [prostate-specific antigen (PSA) >20 ng/mL, Gleason score ≥8, or clinical stage ≥cT2c] were identified. A total of 407 RRP cases were propensity score (PS) matched 1:1 to 398 LRP or RARP cases to yield 3 cohorts (RARP, LRP, and RRP) of 110 patients each for analysis. PS matching variables included the following: age, clinical stage, preoperative PSA, biopsy Gleason score, surgeon experience, and nerve-sparing technique. Overall survival (OS) and recurrence-free survival (RFS) were compared with log-rank test. RFS predictor analysis was calculated within Cox regression models. RESULTS: Pathological Gleason scores <7, =7, and >7 were found in 3.3, 50.9, and 45.8 % of patients. There were no statistically significant differences for pathological stage and positive surgical margins between surgical techniques. Mean 3-year RFS was 41.4, 77.9, and 54.1 %, for RARP, LRP, and RRP, respectively (p < 0.0001 for RARP vs. LRP). There were no significant differences for mean estimated 3-year OS for patients treated with RARP, LRP, or RRP (95.4, 98.1, and 100 %). CONCLUSIONS: RARP demonstrated similar oncologic outcomes compared to RRP and LRP in a PS-matched cohort of patients with high-risk prostate cancer.

Higher rates of upgrading and upstaging in older patients undergoing radical prostatectomy and qualifying for active surveillance.

OBJECTIVE: To determine pathological and oncological outcomes of patients diagnosed with low-risk prostate cancer in two age cohorts who underwent radical prostatectomy (RP) and qualified for active surveillance (AS) according to Prostate Cancer Research International: Active Surveillance (PRIAS) criteria, as AS for low-risk prostate cancer represents an acceptable management strategy especially for older patients. PATIENTS AND METHODS: In all, 320 patients aged ≥65 years who underwent RP and were eligible for AS according to PRIAS criteria were propensity score matched 1:1 to patients aged <65 years. Patient characteristics were compared with chi-square, Kruskal-Wallis, and one-way anova tests. Predictors of RP pathological upgrading or upstaging were analysed using logistic regression. Recurrence-free survival (RFS) and overall survival (OS) were calculated using the Kaplan-Meier method. Predictors of RFS were analysed within Cox regression models. RESULTS: Pathological upgrading and upstaging were significantly higher among older (≥65 years) vs younger (<65 years) patients (53.1% vs 44.1% and 12.2% vs 7.2%, respectively). Higher prostate-specific antigen levels and increasing age were independent predictors of upgrading among patients aged <65 years. There were no differences in RFS or OS between the two age groups. Positive surgical margin status was the only independent predictor of shorter RFS. CONCLUSIONS: Patients aged ≥65 years who are eligible for AS by PRIAS criteria have a higher risk of being upgraded and upstaged at RP than those aged <65 years. These findings should be taken into consideration when discussing treatment options for patients diagnosed with prostate cancer.

A model for the design and construction of a resource for the validation of prognostic prostate cancer biomarkers: the Canary Prostate Cancer Tissue Microarray.

Tissue microarrays (TMAs) provide unique resources for rapid evaluation and validation of tissue biomarkers. The Canary Foundation Retrospective Prostate Tissue Microarray Resource used a rigorous statistical design, quota sampling, a variation of the case-cohort study, to select patients for inclusion in a multicenter, retrospective prostate cancer TMA cohort. The study is designed to definitively validate tissue biomarkers of prostate cancer recurrence after radical prostatectomy. Tissue samples from over 1000 participants treated for prostate cancer with radical prostatectomy between 1995 and 2004 were selected at 6 participating institutions in the United States and Canada. This design captured the heterogeneity of screening and clinical practices in the contemporary North American population. Standardized clinical data were collected in a centralized database. The project has been informative in several respects. The scale and complexity of assembling TMAs with over 200 cases at each of 6 sites involved unanticipated levels of effort and time. Our statistical design promises to provide a model for outcome-based studies where tissue localization methods are applied to high-density TMAs.

Smoking and adverse outcomes at radical prostatectomy.

BACKGROUND: Multiple large epidemiologic studies have examined the relationship between smoking and prostate cancer incidence and mortality only to arrive at contradictory results. In this series, we studied the effect of smoking on pathologic outcomes and biochemical recurrence in a cohort of men undergoing radical prostatectomy. METHODS: We identified 630 men who underwent radical prostatectomy between 1989 and 2005 who had detailed smoking histories. There were 321 smokers and 309 nonsmokers. Pathologic outcomes included prostate weight, volume of cancer, volume of high grade cancer, margin status, seminal vesicle involvement, extraprostatic extension, perineural invasion, angiolymphatic invasion, and the presence of nodal metastasis. Biochemical recurrence was defined as a postoperative PSA ≥ 0.1 ng/ml. Univariate analysis and multivariate linear and Cox regression were used to study the impact of smoking on these outcomes. RESULTS: The volume of cancer (2.54 vs. 2.16 ml, P = 0.016) and the volume of high grade cancer (0.58 vs. 0.28 ml, P = 0.004) were greater in smokers compared with nonsmokers. Smoking independently predicted greater volumes of cancer and high grade cancer in multivariate analysis. Heavy smokers (≥20 pack-year history) had a greater risk of biochemical recurrence on univariate survival analysis. Smoking also predicted a greater risk of biochemical recurrence on Cox regression, the magnitude of which was approximately 1% per pack-year smoked. CONCLUSIONS: Smoking is associated with adverse pathologic features and a higher risk of biochemical recurrence in men undergoing radical prostatectomy. If confirmed by additional studies, smoking history may need to be included into risk assessment models.

Performance characteristics of prostate-specific antigen in patients undergoing radical prostatectomy.

OBJECTIVE: To assess the performance characteristics of prostate-specific antigen (PSA) for predicting the volume of total or high-grade cancer in men undergoing radical prostatectomy. It is known that the performance characteristics of PSA are improved for predicting the presence of high-grade prostate cancer. METHODS: We identified 1459 patients from the Stanford Radical Prostatectomy Database with clinical Stage T1c (n = 783) and T2 (n = 676) disease who underwent surgery from 1988 to 2003 with detailed morphometric mapping. We generated receiver operating characteristic curves for PSA levels according to the total and high-grade (Gleason score 4 or 5) cancer volume and compared the areas under the curve (AUC) for the various total and high-grade cancer volumes. RESULTS: For patients with Stage T1c disease, the AUC for the PSA ROC curve increased in a stepwise fashion as both the total cancer volume and the high-grade cancer volume increased. Significant differences between the AUCs for low and high volumes of total and high-grade disease were observed. For T2 disease, the AUCs for predicting high-grade cancer volume were generally greater than the corresponding AUCs for T1c disease, although no incremental increase was observed. CONCLUSION: In patients with Stage T1c disease, in whom the PSA level was the driving force for biopsy, the PSA performance improved in a stepwise fashion with greater total and high-grade cancer volumes as evidenced by improved ROC. Previous studies have shown that PSA performs better for detecting the presence of high-grade disease. We have shown that PSA performs better in predicting greater volumes of high-grade disease in radical prostatectomy specimens.

A tri-marker proliferation index predicts biochemical recurrence after surgery for prostate cancer.

Prostate cancer exhibits tremendous variability in clinical behavior, ranging from indolent to lethal disease. Better prognostic markers are needed to stratify patients for appropriately aggressive therapy. By expression profiling, we can identify a proliferation signature variably expressed in prostate cancers. Here, we asked whether one or more tissue biomarkers might capture that information, and provide prognostic utility. We assayed three proliferation signature genes: MKI67 (Ki-67; also a classic proliferation biomarker), TOP2A (DNA topoisomerase II, alpha), and E2F1 (E2F transcription factor 1). Immunohistochemical staining was evaluable on 139 radical prostatectomy cases (in tissue microarray format), with a median clinical follow-up of eight years. Each of the three proliferation markers was by itself prognostic. Notably, combining the three markers together as a "proliferation index" (0 or 1, vs. 2 or 3 positive markers) provided superior prognostic performance (hazard ratio = 2.6 (95% CI: 1.4-4.9); P = 0.001). In a multivariate analysis that included preoperative serum prostate specific antigen (PSA) levels, Gleason grade and pathologic tumor stage, the composite proliferation index remained a significant predictor (P = 0.005). Analysis of receiver-operating characteristic (ROC) curves confirmed the improved prognostication afforded by incorporating the proliferation index (compared to the clinicopathologic data alone). Our findings highlight the potential value of a multi-gene signature-based diagnostic, and define a tri-marker proliferation index with possible utility for improved prognostication and treatment stratification in prostate cancer.