Identification of patients at high risk for preventable emergency department visits and inpatient admissions after starting chemotherapy: Machine learning applied to comprehensive electronic health record data

1511

Background: Acute care use is one of the largest drivers of cancer care costs. OP-35: Admissions and Emergency Department Visits for Patients Receiving Outpatient Chemotherapy is a CMS quality measure that will affect reimbursement based on unplanned inpatient admissions (IP) and emergency department (ED) visits. Targeted measures can reduce preventable acute care use but identifying which patients might benefit remains challenging. Prior predictive models have made use of a limited subset of the data available in the Electronic Health Record (EHR). We hypothesized dense, structured EHR data could be used to train machine learning algorithms to predict risk of preventable ED and IP visits. Methods: Patients treated at Stanford Health Care and affiliated community care sites between 2013 and 2015 who met inclusion criteria for OP-35 were selected from our EHR. Preventable ED or IP visits were identified using OP-35 criteria. Demographic, diagnosis, procedure, medication, laboratory, vital sign, and healthcare utilization data generated prior to chemotherapy treatment were obtained. A random split of 80% of the cohort was used to train a logistic regression with least absolute shrinkage and selection operator regularization (LASSO) model to predict risk for acute care events within the first 180 days of chemotherapy. The remaining 20% were used to measure model performance by the Area Under the Receiver Operator Curve (AUROC). Results: 8,439 patients were included, of whom 35% had one or more preventable event within 180 days of starting chemotherapy. Our LASSO model classified patients at risk for preventable ED or IP visits with an AUROC of 0.783 (95% CI: 0.761-0.806). Model performance was better for identifying risk for IP visits than ED visits. LASSO selected 125 of 760 possible features to use when classifying patients. These included prior acute care visits, cancer stage, race, laboratory values, and a diagnosis of depression. Key features for the model are shown in the table. Conclusions: Machine learning models trained on a large number of routinely collected clinical variables can identify patients at risk for acute care events with promising accuracy. These models have the potential to improve cancer care outcomes, patient experience, and costs by allowing for targeted preventative interventions. Future work will include prospective and external validation in other healthcare systems.[Table: see text]

Automated detection of aggressive and indolent prostate cancer on magnetic resonance imaging

PURPOSE

While multi-parametric magnetic resonance imaging (MRI) shows great promise in assisting with prostate cancer diagnosis and localization, subtle differences in appearance between cancer and normal tissue lead to many false positive and false negative interpretations by radiologists. We sought to automatically detect aggressive cancer (Gleason pattern ≥ 4) and indolent cancer (Gleason pattern 3) on a per-pixel basis on MRI to facilitate the targeting of aggressive cancer during biopsy.

METHODS

We created the Stanford Prostate Cancer Network (SPCNet), a convolutional neural network model, trained to distinguish between aggressive cancer, indolent cancer, and normal tissue on MRI. Ground truth cancer labels were obtained by registering MRI with whole-mount digital histopathology images from patients who underwent radical prostatectomy. Before registration, these histopathology images were automatically annotated to show Gleason patterns on a per-pixel basis. The model was trained on data from 78 patients who underwent radical prostatectomy and 24 patients without prostate cancer. The model was evaluated on a pixel and lesion level in 322 patients, including six patients with normal MRI and no cancer, 23 patients who underwent radical prostatectomy, and 293 patients who underwent biopsy. Moreover, we assessed the ability of our model to detect clinically significant cancer (lesions with an aggressive component) and compared it to the performance of radiologists.

RESULTS

Our model detected clinically significant lesions with an area under the receiver operator characteristics curve of 0.75 for radical prostatectomy patients and 0.80 for biopsy patients. Moreover, the model detected up to 18% of lesions missed by radiologists, and overall had a sensitivity and specificity that approached that of radiologists in detecting clinically significant cancer.

CONCLUSIONS

Our SPCNet model accurately detected aggressive prostate cancer. Its performance approached that of radiologists, and it helped identify lesions otherwise missed by radiologists. Our model has the potential to assist physicians in specifically targeting the aggressive component of prostate cancers during biopsy or focal treatment.

Discovery of CASP8 as a potential biomarker for high-risk prostate cancer through a high-multiplex immunoassay

Prostate cancer remains the most common non-cutaneous malignancy among men in the United States. To discover potential serum-based biomarkers for high-risk prostate cancer, we performed a high-multiplex immunoassay utilizing patient-matched pre-operative and post-operative serum samples from ten men with high-grade and high-volume prostate cancer. Our study identified six (CASP8, MSLN, FGFBP1, ICOSLG, TIE2 and S100A4) out of 174 proteins that were significantly decreased after radical prostatectomy. High levels of CASP8 were detected in pre-operative serum samples when compared to post-operative serum samples and serum samples from patients with benign prostate hyperplasia (BPH). By immunohistochemistry, CASP8 protein was expressed at higher levels in prostate cancer tissues compared to non-cancerous and BPH tissues. Likewise, CASP8 mRNA expression was significantly upregulated in prostate cancer when compared to benign prostate tissues in four independent clinical datasets. In addition, mRNA levels of CASP8 were higher in patients with recurrent prostate cancer when compared to patients with non-recurrent prostate cancer and high expression of CASP8 was associated with worse disease-free survival and overall survival in renal cancer. Together, our results suggest that CASP8 may potentially serve as a biomarker for high-risk prostate cancer and possibly renal cancer.

Ferroptosis Inducers Are a Novel Therapeutic Approach for Advanced Prostate Cancer

Ferroptosis is a type of programmed cell death induced by the accumulation of lipid peroxidation and lipid reactive oxygen species in cells. It has been recently demonstrated that cancer cells are vulnerable to ferroptosis inducers (FIN). However, the therapeutic potential of FINs in prostate cancer in preclinical settings has not been explored. In this study, we demonstrate that mediators of ferroptosis, solute carrier family 7 member 11, SLC3A2, and glutathione peroxidase, are expressed in treatment-resistant prostate cancer. We further demonstrate that treatment-resistant prostate cancer cells are sensitive to two FINs, erastin and RSL3. Treatment with erastin and RSL3 led to a significant decrease in prostate cancer cell growth and migration in vitro and significantly delayed the tumor growth of treatment-resistant prostate cancer in vivo, with no measurable side effects. Combination of erastin or RSL3 with standard-of-care second-generation antiandrogens for advanced prostate cancer halted prostate cancer cell growth and migration in vitro and tumor growth in vivo. These results demonstrate the potential of erastin or RSL3 independently and in combination with standard-of-care second-generation antiandrogens as novel therapeutic strategies for advanced prostate cancer. SIGNIFICANCE: These findings reveal that induction of ferroptosis is a new therapeutic strategy for advanced prostate cancer as a monotherapy and in combination with second-generation antiandrogens.

Discovery of PTN as a serum-based biomarker of pro-metastatic prostate cancer

Distinguishing clinically significant from indolent prostate cancer (PC) is a major clinical challenge. We utilised targeted protein biomarker discovery approach to identify biomarkers specific for pro-metastatic PC. Serum samples from the cancer-free group; Cambridge Prognostic Group 1 (CPG1, low risk); CPG5 (high risk) and metastatic disease were analysed using Olink Proteomics panels. Tissue validation was performed by immunohistochemistry in a radical prostatectomy cohort (n = 234). We discovered that nine proteins (pleiotrophin (PTN), MK, PVRL4, EPHA2, TFPI-2, hK11, SYND1, ANGPT2, and hK14) were elevated in metastatic PC patients when compared to other groups. PTN levels were increased in serum from men with CPG5 compared to benign and CPG1. High tissue PTN level was an independent predictor of biochemical recurrence and metastatic progression in low- and intermediate-grade disease. These findings suggest that PTN may represent a novel biomarker for the presence of poor prognosis local disease with the potential to metastasise warranting further investigation.

In Vivo Imaging of Methionine Aminopeptidase II for Prostate Cancer Risk Stratification

Prostate cancer is one of the most common malignancies worldwide, yet limited tools exist for prognostic risk stratification of the disease. Identification of new biomarkers representing intrinsic features of malignant transformation and development of prognostic imaging technologies are critical for improving treatment decisions and patient survival. In this study, we analyzed radical prostatectomy specimens from 422 patients with localized disease to define the expression pattern of methionine aminopeptidase II (MetAP2), a cytosolic metalloprotease that has been identified as a druggable target in cancer. MetAP2 was highly expressed in 54% of low-grade and 59% of high-grade cancers. Elevated levels of MetAP2 at diagnosis were associated with shorter time to recurrence. Controlled self-assembly of a synthetic small molecule enabled design of the first MetAP2-activated PET imaging tracer for monitoring MetAP2 activity in vivo. The nanoparticles assembled upon MetAP2 activation were imaged in single prostate cancer cells with post-click fluorescence labeling. The fluorine-18-labeled tracers successfully differentiated MetAP2 activity in both MetAP2-knockdown and inhibitor-treated human prostate cancer xenografts by micro-PET/CT scanning. This highly sensitive imaging technology may provide a new tool for noninvasive early-risk stratification of prostate cancer and monitoring the therapeutic effect of MetAP2 inhibitors as anticancer drugs. SIGNIFICANCE: This study defines MetAP2 as an early-risk stratifier for molecular imaging of aggressive prostate cancer and describes a MetAP2-activated self-assembly small-molecule PET tracer for imaging MetAP2 activity in vivo.

ProsRegNet: A deep learning framework for registration of MRI and histopathology images of the prostate

Magnetic resonance imaging (MRI) is an increasingly important tool for the diagnosis and treatment of prostate cancer. However, interpretation of MRI suffers from high inter-observer variability across radiologists, thereby contributing to missed clinically significant cancers, overdiagnosed low-risk cancers, and frequent false positives. Interpretation of MRI could be greatly improved by providing radiologists with an answer key that clearly shows cancer locations on MRI. Registration of histopathology images from patients who had radical prostatectomy to pre-operative MRI allows such mapping of ground truth cancer labels onto MRI. However, traditional MRI-histopathology registration approaches are computationally expensive and require careful choices of the cost function and registration hyperparameters. This paper presents ProsRegNet, a deep learning-based pipeline to accelerate and simplify MRI-histopathology image registration in prostate cancer. Our pipeline consists of image preprocessing, estimation of affine and deformable transformations by deep neural networks, and mapping cancer labels from histopathology images onto MRI using estimated transformations. We trained our neural network using MR and histopathology images of 99 patients from our internal cohort (Cohort 1) and evaluated its performance using 53 patients from three different cohorts (an additional 12 from Cohort 1 and 41 from two public cohorts). Results show that our deep learning pipeline has achieved more accurate registration results and is at least 20 times faster than a state-of-the-art registration algorithm. This important advance will provide radiologists with highly accurate prostate MRI answer keys, thereby facilitating improvements in the detection of prostate cancer on MRI. Our code is freely available at https://github.com/pimed//ProsRegNet.

3D Registration of pre-surgical prostate MRI and histopathology images via super-resolution volume reconstruction

The use of MRI for prostate cancer diagnosis and treatment is increasing rapidly. However, identifying the presence and extent of cancer on MRI remains challenging, leading to high variability in detection even among expert radiologists. Improvement in cancer detection on MRI is essential to reducing this variability and maximizing the clinical utility of MRI. To date, such improvement has been limited by the lack of accurately labeled MRI datasets. Data from patients who underwent radical prostatectomy enables the spatial alignment of digitized histopathology images of the resected prostate with corresponding pre-surgical MRI. This alignment facilitates the delineation of detailed cancer labels on MRI via the projection of cancer from histopathology images onto MRI. We introduce a framework that performs 3D registration of whole-mount histopathology images to pre-surgical MRI in three steps. First, we developed a novel multi-image super-resolution generative adversarial network (miSRGAN), which learns information useful for 3D registration by producing a reconstructed 3D MRI. Second, we trained the network to learn information between histopathology slices to facilitate the application of 3D registration methods. Third, we registered the reconstructed 3D histopathology volumes to the reconstructed 3D MRI, mapping the extent of cancer from histopathology images onto MRI without the need for slice-to-slice correspondence. When compared to interpolation methods, our super-resolution reconstruction resulted in the highest PSNR relative to clinical 3D MRI (32.15 dB vs 30.16 dB for BSpline interpolation). Moreover, the registration of 3D volumes reconstructed via super-resolution for both MRI and histopathology images showed the best alignment of cancer regions when compared to (1) the state-of-the-art RAPSODI approach, (2) volumes that were not reconstructed, or (3) volumes that were reconstructed using nearest neighbor, linear, or BSpline interpolations. The improved 3D alignment of histopathology images and MRI facilitates the projection of accurate cancer labels on MRI, allowing for the development of improved MRI interpretation schemes and machine learning models to automatically detect cancer on MRI.

Assessment of a Clinical Trial-Derived Survival Model in Patients With Metastatic Castration-Resistant Prostate Cancer

IMPORTANCE

Randomized clinical trials (RCTs) are considered the criterion standard for clinical evidence. Despite their many benefits, RCTs have limitations, such as costliness, that may reduce the generalizability of their findings among diverse populations and routine care settings.

OBJECTIVE

To assess the performance of an RCT-derived prognostic model that predicts survival among patients with metastatic castration-resistant prostate cancer (CRPC) when the model is applied to real-world data from electronic health records (EHRs).

DESIGN, SETTING, AND PARTICIPANTS

The RCT-trained model and patient data from the RCTs were obtained from the Dialogue for Reverse Engineering Assessments and Methods (DREAM) challenge for prostate cancer, which occurred from March 16 to July 27, 2015. This challenge included 4 phase 3 clinical trials of patients with metastatic CRPC. Real-world data were obtained from the EHRs of a tertiary care academic medical center that includes a comprehensive cancer center. In this study, the DREAM challenge RCT-trained model was applied to real-world data from January 1, 2008, to December 31, 2019; the model was then retrained using EHR data with optimized feature selection. Patients with metastatic CRPC were divided into RCT and EHR cohorts based on data source. Data were analyzed from March 23, 2018, to October 22, 2020.

EXPOSURES

Patients who received treatment for metastatic CRPC.

MAIN OUTCOMES AND MEASURES

The primary outcome was the performance of an RCT-derived prognostic model that predicts survival among patients with metastatic CRPC when the model is applied to real-world data. Model performance was compared using 10-fold cross-validation according to time-dependent integrated area under the curve (iAUC) statistics.

RESULTS

Among 2113 participants with metastatic CRPC, 1600 participants were included in the RCT cohort, and 513 participants were included in the EHR cohort. The RCT cohort comprised a larger proportion of White participants (1390 patients [86.9%] vs 337 patients [65.7%]) and a smaller proportion of Hispanic participants (14 patients [0.9%] vs 42 patients [8.2%]), Asian participants (41 patients [2.6%] vs 88 patients [17.2%]), and participants older than 75 years (388 patients [24.3%] vs 191 patients [37.2%]) compared with the EHR cohort. Participants in the RCT cohort also had fewer comorbidities (mean [SD], 1.6 [1.8] comorbidities vs 2.5 [2.6] comorbidities, respectively) compared with those in the EHR cohort. Of the 101 variables used in the RCT-derived model, 10 were not available in the EHR data set, 3 of which were among the top 10 features in the DREAM challenge RCT model. The best-performing EHR-trained model included only 25 of the 101 variables included in the RCT-trained model. The performance of the RCT-trained and EHR-trained models was adequate in the EHR cohort (mean [SD] iAUC, 0.722 [0.118] and 0.762 [0.106], respectively); model optimization was associated with improved performance of the best-performing EHR model (mean [SD] iAUC, 0.792 [0.097]). The EHR-trained model classified 256 patients as having a high risk of mortality and 256 patients as having a low risk of mortality (hazard ratio, 2.7; 95% CI, 2.0-3.7; log-rank P < .001).

CONCLUSIONS AND RELEVANCE

In this study, although the RCT-trained models did not perform well when applied to real-world EHR data, retraining the models using real-world EHR data and optimizing variable selection was beneficial for model performance. As clinical evidence evolves to include more real-world data, both industry and academia will likely search for ways to balance model optimization with generalizability. This study provides a pragmatic approach to applying RCT-trained models to real-world data.

Real-world Evidence to Estimate Prostate Cancer Costs for First-line Treatment or Active Surveillance

Background

Prostate cancer is the most common cancer in men and second leading cause of cancer-related deaths. Changes in screening guidelines, adoption of active surveillance (AS), and implementation of high-cost technologies have changed treatment costs. Traditional cost-effectiveness studies rely on clinical trial protocols unlikely to capture actual practice behavior, and existing studies use data predating new technologies. Real-world evidence reflecting these changes is lacking.

Objective

To assess real-world costs of first-line prostate cancer management.

Design setting and participants

We used clinical electronic health records for 2008-2018 linked with the California Cancer Registry and the Medicare Fee Schedule to assess costs over 24 or 60 mo following diagnosis. We identified surgery or radiation treatments with structured methods, while we used both structured data and natural language processing to identify AS.

Outcome measurements and statistical analysis

Our results are risk-stratified calculated cost per day (CCPD) for first-line management, which are independent of treatment duration. We used the Kruskal-Wallis test to compare unadjusted CCPD while analysis of covariance log-linear models adjusted estimates for age and Charlson comorbidity.

Results and limitations

In 3433 patients, surgery (54.6%) was more common than radiation (22.3%) or AS (23.0%). Two years following diagnosis, AS ($2.97/d) was cheaper than surgery ($5.67/d) or radiation ($9.34/d) in favorable disease, while surgery ($7.17/d) was cheaper than radiation ($16.34/d) for unfavorable disease. At 5 yr, AS ($2.71/d) remained slightly cheaper than surgery ($2.87/d) and radiation ($4.36/d) in favorable disease, while for unfavorable disease surgery ($4.15/d) remained cheaper than radiation ($10.32/d). Study limitations include information derived from a single healthcare system and costs based on benchmark Medicare estimates rather than actual payment exchanges.

Patient summary

Active surveillance was cheaper than surgery (-47.6%) and radiation (-68.2%) at 2 yr for favorable-risk disease, which decreased by 5 yr (-5.6% and -37.8%, respectively). Surgery was less costly than radiation for unfavorable risk for both intervals (-56.1% and -59.8%, respectively).

Sudden PSA rise to ≥20 ng/ml and prostate cancer diagnosis in the United States: A population-based study

PURPOSE

While prostate-specific antigen (PSA) screening protocols vary, many clinicians have anecdotes of screened men with low PSA levels that rise significantly and are associated with high-risk prostate cancer (PC). We sought to better understand the frequency of high-risk cases that appear suddenly in a screened population.

METHODS

We utilized data from a Commercial and Medicare advantage claims database to identify all US men ages 50 and above undergoing PSA screening who then had a sudden interval rise in PSA (e.g., PSA ≥ 20) and diagnosis of PC. We determined associations with age, race, screening intensity, and baseline PSA levels.

RESULTS

In all, 526,120 men met entry criteria with an average age of 60.7 and follow-up of 5.6 years. As the baseline PSA increased, the rate of high-risk PC increased from 2/10,000 persons among men with the lowest baseline PSA (<1 ng/ml) to 14/10,000 person-years among men with a baseline PSA < 5 ng/ml. Moreover, as a man's age at baseline PSA increased, the rate of high-risk PC also increased. In contrast, the incidence of high-risk PC did not vary significantly by race/ethnicity. More screening PSAs and shorter intervals between PSA screenings were associated with a lower incidence of high-risk PC.

CONCLUSIONS

The incidence of high-risk PC in a screened population is low (<0.1%). Our findings suggest that systematic screening cannot eliminate all PC deaths and provide an estimate for the risk of the rapid development of high-risk cancers that is comparable to that observed in active surveillance populations.

Laboratory-wide association study of survival with prostate cancer

BACKGROUND

Estimates of overall patient health are essential to inform treatment decisions for patients diagnosed with cancer. The authors applied XWAS methods, herein referred to as "laboratory-wide association study (LWAS)", to evaluate associations between routinely collected laboratory tests and survival in veterans with prostate cancer.

METHODS

The authors identified 133,878 patients who were diagnosed with prostate cancer between 2000 and 2013 in the Veterans Health Administration using any laboratory tests collected within 6 months of diagnosis (3,345,083 results). Using the LWAS framework, the false-discovery rate was used to test the association between multiple laboratory tests and survival, and these results were validated using training, testing, and validation cohorts.

RESULTS

A total of 31 laboratory tests associated with survival met stringent LWAS criteria. LWAS confirmed markers of prostate cancer biology (prostate-specific antigen: hazard ratio [HR], 1.07 [95% confidence interval (95% CI), 1.06-1.08]; and alkaline phosphatase: HR, 1.22 [95% CI, 1.20-1.24]) as well laboratory tests of general health (eg, serum albumin: HR, 0.78 [95% CI, 0.76-0.80]; and creatinine: HR, 1.05 [95% CI, 1.03-1.07]) and inflammation (leukocyte count: HR, 1.23 [95% CI, 1.98-1.26]; and erythrocyte sedimentation rate: HR, 1.33 [95% CI, 1.09-1.61]). In addition, the authors derived and validated separate models for patients with localized and advanced disease, identifying 28 laboratory markers and 15 laboratory markers, respectively, in each cohort.

CONCLUSIONS

The authors identified routinely collected laboratory data associated with survival for patients with prostate cancer using LWAS methodologies, including markers of prostate cancer biology, overall health, and inflammation. Broadening consideration of determinants of survival beyond those related to cancer itself could help to inform the design of clinical trials and aid in shared decision making.

LAY SUMMARY

This article examined routine laboratory tests associated with survival among veterans with prostate cancer. Using laboratory-wide association studies, the authors identified 31 laboratory tests associated with survival that can be used to inform the design of clinical trials and aid patients in shared decision making.

Phenotyping severity of patient-centered outcomes using clinical notes: A prostate cancer use case

Introduction

A learning health system (LHS) must improve care in ways that are meaningful to patients, integrating patient‐centered outcomes (PCOs) into core infrastructure. PCOs are common following cancer treatment, such as urinary incontinence (UI) following prostatectomy. However, PCOs are not systematically recorded because they can only be described by the patient, are subjective and captured as unstructured text in the electronic health record (EHR). Therefore, PCOs pose significant challenges for phenotyping patients. Here, we present a natural language processing (NLP) approach for phenotyping patients with UI to classify their disease into severity subtypes, which can increase opportunities to provide precision‐based therapy and promote a value‐based delivery system.

Methods

Patients undergoing prostate cancer treatment from 2008 to 2018 were identified at an academic medical center. Using a hybrid NLP pipeline that combines rule‐based and deep learning methodologies, we classified positive UI cases as mild, moderate, and severe by mining clinical notes.

Results

The rule‐based model accurately classified UI into disease severity categories (accuracy: 0.86), which outperformed the deep learning model (accuracy: 0.73). In the deep learning model, the recall rates for mild and moderate group were higher than the precision rate (0.78 and 0.79, respectively). A hybrid model that combined both methods did not improve the accuracy of the rule‐based model but did outperform the deep learning model (accuracy: 0.75).

Conclusion

Phenotyping patients based on indication and severity of PCOs is essential to advance a patient centered LHS. EHRs contain valuable information on PCOs and by using NLP methods, it is feasible to accurately and efficiently phenotype PCO severity. Phenotyping must extend beyond the identification of disease to provide classification of disease severity that can be used to guide treatment and inform shared decision‐making. Our methods demonstrate a path to a patient centered LHS that could advance precision medicine.

Sprr2f protects against renal injury by decreasing the level of reactive oxygen species in female mice

Renal injury leads to chronic kidney disease, with which women are not only more likely to be diagnosed than men but have poorer outcomes as well. We have previously shown that expression of small proline-rich region 2f (Sprr2f), a member of the small proline-rich region (Sprr) gene family, is increased several hundredfold after renal injury using a unilateral ureteral obstruction (UUO) mouse model. To better understand the role of Sprr2f in renal injury, we generated a Sprr2f knockout (Sprr2f-KO) mouse model using CRISPR-Cas9 technology. Sprr2f-KO female mice showed greater renal damage after UUO compared with wild-type (Sprr2f-WT) animals, as evidenced by higher hydroxyproline levels and denser collagen staining, indicating a protective role of Sprr2f during renal injury. Gene expression profiling by RNA sequencing identified 162 genes whose expression levels were significantly different between day 0 and day 5 after UUO in Sprr2f-KO mice. Of the 162 genes, 121 genes were upregulated after UUO and enriched with those involved in oxidation-reduction, a phenomenon not observed in Sprr2f-WT animals, suggesting a protective role of Sprr2f in UUO through defense against oxidative damage. Consistently, bilateral ischemia-reperfusion injury resulted in higher serum blood urea nitrogen levels and higher tissue reactive oxygen species in Sprr2f-KO compared with Sprr2f-WT female mice. Moreover, cultured renal epithelial cells from Sprr2f-KO female mice showed lower viability after oxidative damage induced by menadione compared with Sprr2f-WT cells that could be rescued by supplementation with reduced glutathione, suggesting that Sprr2f induction after renal damage acts as a defense against reactive oxygen species.

Tailoring Intensity of Active Surveillance for Low-Risk Prostate Cancer Based on Individualized Prediction of Risk Stability

Importance

Active surveillance is increasingly recognized as the preferred standard of care for men with low-risk prostate cancer. However, active surveillance requires repeated assessments, including prostate-specific antigen tests and biopsies that may increase anxiety, risk of complications, and cost.

Objective

To identify and validate clinical parameters that can identify men who can safely defer follow-up prostate cancer assessments.

Design, Setting, and Participants

The Canary Prostate Active Surveillance Study (PASS) is a multicenter, prospective active surveillance cohort study initiated in July 2008, with ongoing accrual and a median follow-up period of 4.1 years. Men with prostate cancer managed with active surveillance from 9 North American academic medical centers were enrolled. Blood tests and biopsies were conducted on a defined schedule for least 5 years after enrollment. Model validation was performed among men at the University of California, San Francisco (UCSF) who did not enroll in PASS. Men with Gleason grade group 1 prostate cancer diagnosed since 2003 and enrolled in PASS before 2017 with at least 1 confirmatory biopsy after diagnosis were included. A total of 850 men met these criteria and had adequate follow-up. For the UCSF validation study, 533 active surveillance patients meeting the same criteria were identified. Exclusion criteria were treatment within 6 months of diagnosis, diagnosis before 2003, Gleason grade score of at least 2 at diagnosis or first surveillance biopsy, no surveillance biopsy, or missing data.

Exposures

Active surveillance for prostate cancer.

Main Outcomes and Measures

Time from confirmatory biopsy to reclassification, defined as Gleason grade group 2 or higher on subsequent biopsy.

Results

A total of 850 men (median [interquartile range] age, 64 [58-68] years; 774 [91%] White) were included in the PASS cohort. A total of 533 men (median [interquartile range] age, 61 [57-65] years; 422 [79%] White) were included in the UCSF cohort. Parameters predictive of reclassification on multivariable analysis included maximum percent positive cores (hazard ratio [HR], 1.30 [95% CI, 1.09-1.56]; P = .004), history of any negative biopsy after diagnosis (1 vs 0: HR, 0.52 [95% CI, 0.38-0.71]; P < .001 and ≥2 vs 0: HR, 0.18 [95% CI, 0.08-0.4]; P < .001), time since diagnosis (HR, 1.62 [95% CI, 1.28-2.05]; P < .001), body mass index (HR, 1.08 [95% CI, 1.05-1.12]; P < .001), prostate size (HR, 0.40 [95% CI, 0.25-0.62]; P < .001), prostate-specific antigen at diagnosis (HR, 1.51 [95% CI, 1.15-1.98]; P = .003), and prostate-specific antigen kinetics (HR, 1.46 [95% CI, 1.23-1.73]; P < .001). For prediction of nonreclassification at 4 years, the area under the receiver operating curve was 0.70 for the PASS cohort and 0.70 for the UCSF validation cohort. This model achieved a negative predictive value of 0.88 (95% CI, 0.83-0.94) for those in the bottom 25th percentile of risk and of 0.95 (95% CI, 0.89-1.00) for those in the bottom 10th percentile.

Conclusions and Relevance

In this study, among men with low-risk prostate cancer, heterogeneity prevailed in risk of subsequent disease reclassification. These findings suggest that active surveillance intensity can be modulated based on an individual's risk parameters and that many men may be safely monitored with a substantially less intensive surveillance regimen.

Association between patient-initiated emails and overall 2-year survival in cancer patients undergoing chemotherapy: Evidence from the real-world setting

PURPOSE

Prior studies suggest email communication between patients and providers may improve patient engagement and health outcomes. The purpose of this study was to determine whether patient-initiated emails are associated with overall survival benefits among cancer patients undergoing chemotherapy.

PATIENTS AND METHODS

We identified patient-initiated emails through the patient portal in electronic health records (EHR) among 9900 cancer patients receiving chemotherapy between 2013 and 2018. Email users were defined as patients who sent at least one email 12 months before to 2 months after chemotherapy started. A propensity score-matched cohort analysis was carried out to reduce bias due to confounding (age, primary cancer type, gender, insurance payor, ethnicity, race, stage, income, Charlson score, county of residence). The cohort included 3223 email users and 3223 non-email users. The primary outcome was overall 2-year survival stratified by email use. Secondary outcomes included number of face-to-face visits, prescriptions, and telephone calls. The healthcare teams' response to emails and other forms of communication was also investigated. Finally, a quality measure related to chemotherapy-related inpatient and emergency department visits was evaluated.

RESULTS

Overall 2-year survival was higher in patients who were email users, with an adjusted hazard ratio of 0.80 (95 CI 0.72-0.90; p < 0.001). Email users had higher rates of healthcare utilization, including face-to-face visits (63 vs. 50; p < 0.001), drug prescriptions (28 vs. 21; p < 0.001), and phone calls (18 vs. 16; p < 0.001). Clinical quality outcome measure of inpatient use was better among email users (p = 0.015).

CONCLUSION

Patient-initiated emails are associated with a survival benefit among cancer patients receiving chemotherapy and may be a proxy for patient engagement. As value-based payment models emphasize incorporating the patients' voice into their care, email communications could serve as a novel source of patient-generated data.

Development of a DNA Methylation-Based Diagnostic Signature to Distinguish Benign Oncocytoma From Renal Cell Carcinoma

PURPOSE

A challenge in the diagnosis of renal cell carcinoma (RCC) is to distinguish chromophobe RCC (chRCC) from benign renal oncocytoma, because these tumor types are histologically and morphologically similar, yet they require different clinical management. Molecular biomarkers could provide a way of distinguishing oncocytoma from chRCC, which could prevent unnecessary treatment of oncocytoma. Such biomarkers could also be applied to preoperative biopsy specimens such as needle core biopsy specimens, to avoid unnecessary surgery of oncocytoma.

METHODS

We profiled DNA methylation in fresh-frozen oncocytoma and chRCC tumors and adjacent normal tissue and used machine learning to identify a signature of differentially methylated cytosine-phosphate-guanine sites (CpGs) that robustly distinguish oncocytoma from chRCC.

RESULTS

Unsupervised clustering of Stanford and preexisting RCC data from The Cancer Genome Atlas (TCGA) revealed that of all RCC subtypes, oncocytoma is most similar to chRCC. Unexpectedly, however, oncocytoma features more extensive, overall abnormal methylation than does chRCC. We identified 79 CpGs with large methylation differences between oncocytoma and chRCC. A diagnostic model trained on 30 CpGs could distinguish oncocytoma from chRCC in 10-fold cross-validation (area under the receiver operating curve [AUC], 0.96 (95% CI, 0.88 to 1.00)) and could distinguish TCGA chRCCs from an independent set of oncocytomas from a previous study (AUC, 0.87). This signature also separated oncocytoma from other RCC subtypes and normal tissue, revealing it as a standalone diagnostic biomarker for oncocytoma.

CONCLUSION

This CpG signature could be developed as a clinical biomarker to support differential diagnosis of oncocytoma and chRCC in surgical samples. With improved biopsy techniques, this signature could be applied to preoperative biopsy specimens.

The m6A RNA demethylase FTO is a HIF-independent synthetic lethal partner with the VHL tumor suppressor

Loss of the von Hippel-Lindau (VHL) tumor suppressor is a hallmark feature of renal clear cell carcinoma. VHL inactivation results in the constitutive activation of the hypoxia-inducible factors (HIFs) HIF-1 and HIF-2 and their downstream targets, including the proangiogenic factors VEGF and PDGF. However, antiangiogenic agents and HIF-2 inhibitors have limited efficacy in cancer therapy due to the development of resistance. Here we employed an innovative computational platform, Mining of Synthetic Lethals (MiSL), to identify synthetic lethal interactions with the loss of VHL through analysis of primary tumor genomic and transcriptomic data. Using this approach, we identified a synthetic lethal interaction between VHL and the m⁶A RNA demethylase FTO in renal cell carcinoma. MiSL identified FTO as a synthetic lethal partner of VHL because deletions of FTO are mutually exclusive with VHL loss in pan cancer datasets. Moreover, FTO expression is increased in VHL-deficient ccRCC tumors compared to normal adjacent tissue. Genetic inactivation of FTO using multiple orthogonal approaches revealed that FTO inhibition selectively reduces the growth and survival of VHL-deficient cells in vitro and in vivo. Notably, FTO inhibition reduced the survival of both HIF wild type and HIF-deficient tumors, identifying FTO as an HIF-independent vulnerability of VHL-deficient cancers. Integrated analysis of transcriptome-wide m⁶A-seq and mRNA-seq analysis identified the glutamine transporter SLC1A5 as an FTO target that promotes metabolic reprogramming and survival of VHL-deficient ccRCC cells. These findings identify FTO as a potential HIF-independent therapeutic target for the treatment of VHL-deficient renal cell carcinoma.

Registration of presurgical MRI and histopathology images from radical prostatectomy via RAPSODI

PURPOSE

Magnetic resonance imaging (MRI) has great potential to improve prostate cancer diagnosis; however, subtle differences between cancer and confounding conditions render prostate MRI interpretation challenging. The tissue collected from patients who undergo radical prostatectomy provides a unique opportunity to correlate histopathology images of the prostate with preoperative MRI to accurately map the extent of cancer from histopathology images onto MRI. We seek to develop an open-source, easy-to-use platform to align presurgical MRI and histopathology images of resected prostates in patients who underwent radical prostatectomy to create accurate cancer labels on MRI.

METHODS

Here, we introduce RAdiology Pathology Spatial Open-Source multi-Dimensional Integration (RAPSODI), the first open-source framework for the registration of radiology and pathology images. RAPSODI relies on three steps. First, it creates a three-dimensional (3D) reconstruction of the histopathology specimen as a digital representation of the tissue before gross sectioning. Second, RAPSODI registers corresponding histopathology and MRI slices. Third, the optimized transforms are applied to the cancer regions outlined on the histopathology images to project those labels onto the preoperative MRI.

RESULTS

We tested RAPSODI in a phantom study where we simulated various conditions, for example, tissue shrinkage during fixation. Our experiments showed that RAPSODI can reliably correct multiple artifacts. We also evaluated RAPSODI in 157 patients from three institutions that underwent radical prostatectomy and have very different pathology processing and scanning. RAPSODI was evaluated in 907 corresponding histpathology-MRI slices and achieved a Dice coefficient of 0.97 ± 0.01 for the prostate, a Hausdorff distance of 1.99 ± 0.70 mm for the prostate boundary, a urethra deviation of 3.09 ± 1.45 mm, and a landmark deviation of 2.80 ± 0.59 mm between registered histopathology images and MRI.

CONCLUSION

Our robust framework successfully mapped the extent of cancer from histopathology slices onto MRI providing labels from training machine learning methods to detect cancer on MRI.

Early-Life Cardiorespiratory Fitness and Long-term Risk of Prostate Cancer

BACKGROUND

Adolescence is a period of rapid prostatic growth, yet is understudied for susceptibility for future risk of prostate cancer. We examined cardiorespiratory fitness (CRF) in late adolescence in relation to long-term prostate cancer risk.

METHODS

A population-based cohort study was conducted of all 699,125 Swedish military conscripts during 1972-1985 (97%-98% of 18-year-old men) in relation to risk of prostate cancer overall, aggressive prostate cancer, and prostate cancer mortality during 1998-2017 (ages 50-65 years). CRF was measured by maximal aerobic workload, and prostate cancer was ascertained using the National Prostate Cancer Register. Muscle strength was examined as a secondary predictor.

RESULTS

In 38.8 million person-years of follow-up, 10,782 (1.5%) men were diagnosed with prostate cancer. Adjusting for sociodemographic factors, height, weight, and family history of prostate cancer, high CRF was associated with a slightly increased risk of any prostate cancer [highest vs. lowest quintile: incidence rate ratio (IRR), 1.10; 95% CI, 1.03-1.19; P = 0.008], but was neither significantly associated with aggressive prostate cancer (1.01; 0.85-1.21; P = 0.90) nor prostate cancer mortality (1.24; 0.73-2.13; P = 0.42). High muscle strength also was associated with a modestly increased risk of any prostate cancer (highest vs. lowest quintile: IRR, 1.14; 95% CI, 1.07-1.23; P < 0.001), but neither with aggressive prostate cancer (0.88; 0.74-1.04; P = 0.14) nor prostate cancer mortality (0.81; 0.48-1.37; P = 0.43).

CONCLUSIONS

High CRF or muscle strength in late adolescence was associated with slightly increased future risk of prostate cancer, possibly related to increased screening, but neither with risk of aggressive prostate cancer nor prostate cancer mortality.

IMPACT

These findings illustrate the importance of distinguishing aggressive from indolent prostate cancer and assessing for potential detection bias.

PSA Testing Use and Prostate Cancer Diagnostic Stage After the 2012 U.S. Preventive Services Task Force Guideline Changes

BACKGROUND

Most patients with prostate cancer are diagnosed with low-grade, localized disease and may not require definitive treatment. In 2012, the U.S. Preventive Services Task Force (USPSTF) recommended against prostate cancer screening to address overdetection and overtreatment. This study sought to determine the effect of guideline changes on prostate-specific antigen (PSA) screening and initial diagnostic stage for prostate cancer.

PATIENTS AND METHODS

A difference-in-differences analysis was conducted to compare changes in PSA screening (exposure) relative to cholesterol testing (control) after the 2012 USPSTF guideline changes, and chi-square test was used to determine whether there was a subsequent decrease in early-stage, low-risk prostate cancer diagnoses. Data were derived from a tertiary academic medical center's electronic health records, a national commercial insurance database (OptumLabs), and the SEER database for men aged ≥35 years before (2008-2011) and after (2013-2016) the guideline changes.

RESULTS

In both the academic center and insurance databases, PSA testing significantly decreased for all men compared with the control. The greatest decrease was among men aged 55 to 74 years at the academic center and among those aged ≥75 years in the commercial database. The proportion of early-stage prostate cancer diagnoses (<T2) decreased across age groups at the academic center and in the SEER database.

CONCLUSIONS

In primary care, PSA testing decreased significantly and fewer prostate cancers were diagnosed at an early stage, suggesting provider adherence to the 2012 USPSTF guideline changes. Long-term follow-up is needed to understand the effect of decreased screening on prostate cancer survival.

Simultaneous transrectal ultrasound and photoacoustic human prostate imaging

Imaging technologies that simultaneously provide anatomical, functional, and molecular information are emerging as an attractive choice for disease screening and management. Since the 1980s, transrectal ultrasound (TRUS) has been routinely used to visualize prostatic anatomy and guide needle biopsy, despite limited specificity. Photoacoustic imaging (PAI) provides functional and molecular information at ultrasonic resolution based on optical absorption. Combining the strengths of TRUS and PAI approaches, we report the development and bench-to-bedside translation of an integrated TRUS and photoacoustic (TRUSPA) device. TRUSPA uses a miniaturized capacitive micromachined ultrasonic transducer array for simultaneous imaging of anatomical and molecular optical contrasts [intrinsic: hemoglobin; extrinsic: intravenous indocyanine green (ICG)] of the human prostate. Hemoglobin absorption mapped vascularity of the prostate and surroundings, whereas ICG absorption enhanced the intraprostatic photoacoustic contrast. Future work using the TRUSPA device for biomarker-specific molecular imaging may enable a fundamentally new approach to prostate cancer diagnosis, prognostication, and therapeutic monitoring.

AZGP1 Protein Expression in Hormone-Naïve Advanced Prostate Cancer Treated with Primary Androgen Deprivation Therapy

Biomarkers for predicting the risk of castration-resistant prostate cancer (CRPC) in men treated with primary androgen deprivation therapy (ADT) are lacking. We investigated whether Zinc-alpha 2 glycoprotein (AZGP1) expression in the diagnostic biopsies of men with hormone-naïve prostate cancer (PCa) undergoing primary ADT was predictive of the development of CRPC and PCa-specific mortality. The study included 191 patients who commenced ADT from 2000 to 2011. The AZGP1 expression was evaluated using immunohistochemistry and scored as high or low expression. The risks of CRPC and PCa-specific mortality were analyzed using stratified cumulative incidences and a cause-specific COX regression analysis for competing risk assessment. The median follow-up time was 9.8 (IQR: 6.1–12.7) years. In total, 94 and 97 patients presented with low and high AZGP1 expression, respectively. A low AZGP1 expression was found to be associated with a shorter time to CRPC when compared to patients with a high AZGP1 expression (HR: 1.5; 95% CI: 1.0–2.1; p = 0.03). However, the multivariable analysis demonstrated no added benefit by adding the AZGP1 expression to prediction models for CRPC. No differences for PCa-specific mortality between the AZGP1 groups were observed. In conclusion, a low AZGP1 expression was associated with a shorter time to CRPC for PCa patients treated with first-line ADT but did not add any predictive information besides well-established clinicopathological variables.

Life expectancy estimates for patients diagnosed with prostate cancer in the Veterans Health Administration

PURPOSE

Accurate life expectancy estimates are required to inform prostate cancer treatment decisions. However, few models are specific to the population served or easily implemented in a clinical setting. We sought to create life expectancy estimates specific to Veterans diagnosed with prostate cancer.

MATERIALS AND METHODS

Using national Veterans Health Administration electronic health records, we identified Veterans diagnosed with prostate cancer between 2000 and 2015. We abstracted demographics, comorbidities, oncologic staging, and treatment information. We fit Cox Proportional Hazards models to determine the impact of age, comorbidity, cancer risk, and race on survival. We stratified life expectancy estimates by age, comorbidity and cancer stage.

RESULTS

Our analytic cohort included 145,678 patients. Survival modeling demonstrated the importance of age and comorbidity across all cancer risk categories. Life expectancy estimates generated from age and comorbidity data were predictive of overall survival (C-index 0.676, 95% CI 0.674-0.679) and visualized using Kaplan-Meier plots and heatmaps stratified by age and comorbidity. Separate life expectancy estimates were generated for patients with localized or advanced disease. These life expectancy estimates calibrate well across prostate cancer risk categories.

CONCLUSIONS

Life expectancy estimates are essential to providing patient-centered prostate cancer care. We developed accessible life expectancy estimation tools for Veterans diagnosed with prostate cancer that can be used in routine clinical practice to inform medical-decision making.

Identification of diagnostic metabolic signatures in clear cell renal cell carcinoma using mass spectrometry imaging

Clear cell renal cell carcinoma (ccRCC) is the most common and lethal subtype of kidney cancer. Intraoperative frozen section (IFS) analysis is used to confirm the diagnosis during partial nephrectomy. However, surgical margin evaluation using IFS analysis is time consuming and unreliable, leading to relatively low utilization. In our study, we demonstrated the use of desorption electrospray ionization mass spectrometry imaging (DESI-MSI) as a molecular diagnostic and prognostic tool for ccRCC. DESI-MSI was conducted on fresh-frozen 23 normal tumor paired nephrectomy specimens of ccRCC. An independent validation cohort of 17 normal tumor pairs was analyzed. DESI-MSI provides two-dimensional molecular images of tissues with mass spectra representing small metabolites, fatty acids and lipids. These tissues were subjected to histopathologic evaluation. A set of metabolites that distinguish ccRCC from normal kidney were identified by performing least absolute shrinkage and selection operator (Lasso) and log-ratio Lasso analysis. Lasso analysis with leave-one-patient-out cross-validation selected 57 peaks from over 27,000 metabolic features across 37,608 pixels obtained using DESI-MSI of ccRCC and normal tissues. Baseline Lasso of metabolites predicted the class of each tissue to be normal or cancerous tissue with an accuracy of 94 and 76%, respectively. Combining the baseline Lasso with the ratio of glucose to arachidonic acid could potentially reduce scan time and improve accuracy to identify normal (82%) and ccRCC (88%) tissue. DESI-MSI allows rapid detection of metabolites associated with normal and ccRCC with high accuracy. As this technology advances, it could be used for rapid intraoperative assessment of surgical margin status.

Abstract C053: Testicular cancer in Hispanics: Incidence of subtypes over time according to neighborhood sociodemographic factors in California

Background: Hispanic men in the U.S. experience the second-highest incidence rate of testicular cancer, behind non-Hispanic (NH) White men. Incidence of testicular cancer is increasing in the U.S., despite reports of a plateau during the 1990’s, and increases are especially steep in the Hispanic population. To date, the literature does not address whether the incidence of testicular cancer or the observed increases in incidence differ according to neighborhood factors. Purpose: We examined incidence rates and changes in incidence over time for testicular cancer histologic subtypes (i.e., seminoma and nonseminoma) according to neighborhood socioeconomic status (nSES) among Hispanic and, for comparison, NH White men, and according to neighborhood ethnic enclave among Hispanic men, using California Cancer Registry Data. Methods: We conducted a population-based study of 12,228 Hispanic and NH White men diagnosed with testicular cancer in California during three pericensal periods 1988-1992, 1998-2002, and 2008-2012. We calculated incidence rates according to nSES and, among Hispanics, according to ethnic enclave. Incidence rate ratios were calculated to compare incidence rates across nSES and ethnic enclave and to examine changes in incidence rates over time. Results: Hispanic men residing in high SES neighborhoods, compared to low SES neighborhoods, had greater incidence of both seminoma and nonseminoma testicular cancer across pericensal periods (2008-2012, high to low nSES, seminoma IRR, 1.67; 95% CI, 1.38-2.02 and nonseminoma IRR, 1.22; 95% CI, 1.00-1.48). Hispanic men residing in low ethnic enclave neighborhoods also had higher incidence of both seminoma and nonseminoma across pericensal periods. Between the periods 1998-2002 and 2008-2012, Hispanic men residing in low SES neighborhoods experienced increased incidence of seminoma (IRR, 2008-2012 compared to 1998-2002, 1.39; 95% CI, 1.17-1.65) while those residing in both low and middle SES neighborhood experienced increased incidence of nonseminoma (IRR, 2008-2012 compared to 1998-2002 for low nSES, 1.87; 95% CI, 1.57-2.20 and for middle nSES, 1.48; 95% CI, 1.21-1.79). Conclusions: While Hispanic men residing in neighborhoods with higher SES and lower enclave status have greater incidence of both seminoma and nonseminoma testicular cancer, recent increases in incidence are driven by Hispanic men residing in lower SES neighborhoods, particularly for the nonseminoma histologic subtype.

Citation Format: Mindy C DeRouen, Meg McKinley, Sumit A Shah, Hala T Borno, Rhonda Aoki, Daphne Y Lichtensztajn, John T Leppert, James D Brooks, Benjamin Chung, Scarlett L Gomez, Iona Cheng. Testicular cancer in Hispanics: Incidence of subtypes over time according to neighborhood sociodemographic factors in California [abstract]. In: Proceedings of the Twelfth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2019 Sep 20-23; San Francisco, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl_2):Abstract nr C053.

Abstract PR16: Association of renal cell carcinoma subtypes with race/ethnicity and comorbid medical conditions

Background: Renal cell carcinomas (RCC) comprise distinct subtypes that differ in molecular characteristics and prognosis. The distribution of these subtypes varies by race/ethnicity. Hypertension, obesity, chronic kidney disease, and diabetes have been associated with increased risk of RCC, and emerging evidence suggests that the risk may be subtype specific. We assessed whether race/ethnicity and comorbidities were independently associated with RCC subtypes.

Methods: Using population-based data from the California Cancer Registry linked to the Office of Statewide Health Planning and Development, we identified non-Latino White, non-Latino Black, Latino, and Asian/ Pacific Islander adults diagnosed with their first microscopically confirmed RCC between 2005 and 2015. Diagnosis of hypertension, diabetes, and kidney disease was defined by ICD-9 and ICD-10 codes present prior to RCC diagnosis. We used multivariable logistic regression to model the association of the three main RCC subtypes (clear cell, papillary, and chromophobe) with race/ethnicity adjusting for comorbidity, sex, neighborhood socioeconomic status, age, and year of diagnosis.

Results: Of the 40,016 cases of RCC included, 62.6% were clear cell, 10.9% papillary, and 6.0% chromophobe. There were striking differences in the proportion of clear cell and papillary subtypes by race/ethnicity, ranging from 40.4% clear cell and 30.4% papillary in non-Latino Black adults to 70.7% clear cell and 4.5% papillary in Latino adults. The prevalence of comorbid conditions also varied by race/ethnicity—most notably the greater prevalence of kidney disease in the non-Latino Black group. In multivariable analysis, non-Latino Black individuals had a higher likelihood of presenting with papillary (odds ratio (OR) 3.35, 95% confidence interval (CI) 3.05-3.68) and chromophobe (OR 1.23, 95% CI 1.06-1.44) subtype compared to those identified as non-Latino White. In contrast, both Latino and Asian/Pacific Islander individuals were more likely than those of non-Latino White race/ethnicity to present with clear-cell subtype (OR 1.48, 95% CI 1.41-1.56 and OR 1.30, 95% CI 1.20-1.40, respectively). Clear-cell subtype was associated with diabetic renal disease (OR 1.39, 95%CI 1.23-1.58) and uncomplicated diabetes (OR 1.29, 95% CI 1.22-1.37), while papillary subtype was associated with hypertension (OR 1.22, 95% CI 1.13-1.32), hypertensive renal disease (OR 1.53, 95% CI 1.34-1.75), and end-stage renal disease (OR 1.55, 95% CI 1.31-1.84).

Conclusion: In addition to race/ethnicity, specific comorbidities are associated with RCC subtype. The association of diabetes, hypertension, and end-stage renal disease with RCC subtype may provide clues to disease etiology as well as avenues for disease prevention.

This abstract is also being presented as Poster D119.

Citation Format: Daphne Y. Lichtensztajn, Brenda M. Hofer, John T. Leppert, James D. Brooks, Benjamin I. Chung, Sumit A. Shah, Mindy C. DeRouen, Scarlett L. Gomez, Iona Cheng. Association of renal cell carcinoma subtypes with race/ethnicity and comorbid medical conditions [abstract]. In: Proceedings of the Twelfth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2019 Sep 20-23; San Francisco, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl_2):Abstract nr PR16.

Testicular cancer in Hispanics: incidence of subtypes over time according to neighborhood sociodemographic factors in California

PURPOSE

Hispanic men in the USA experience the second-highest incidence rate of testicular germ cell tumors (TGCTs), behind non-Hispanic (NH) White men, and have experienced steep increases in TGCT in recent decades. It is unknown whether increases in incidence differ according to neighborhood sociodemographic factors.

METHODS

We conducted a population-based study of n = 3759 Hispanic and n = 8469 NH White men (n = 12,228 total) diagnosed with TGCT in California during the three most recent pericensal periods. We calculated incidence rates according to neighborhood socioeconomic status (nSES) and among Hispanics, according to ethnic enclave. We calculated incidence rate ratios to compare rates across nSES and ethnic enclave and to examine changes in rates over pericensal time periods according to these neighborhood factors for major histologic types (i.e., seminoma and nonseminoma).

RESULTS

Hispanic men residing in high SES, compared to low SES, neighborhoods had greater incidence of seminoma and nonseminoma testicular cancer across pericensal periods, as did Hispanic men in low enclave (less ethnic), compared to high enclave, neighborhoods. Between the periods 1998-2002 and 2008-2012, Hispanic men residing in low SES neighborhoods experienced a 39% increased incidence of seminoma, while those residing in low and middle SES neighborhoods experienced 87% and 48% increased incidence of nonseminoma, respectively.

CONCLUSION

While TGCT incidence has increased among all Hispanic men, incidence increases appear to be driven disproportionately by those residing in lower SES and lower enclave neighborhoods, particularly for nonseminoma.

Detection of prostate cancer and determination of its significance using explainable artificial intelligence

5555

Background: The variation of the human perception has limited the potential of multi-parametric magnetic resonance imaging (mpMRI) of the prostate in determining prostate cancer and identifying significant prostate cancer. The current study aims to overcome this limitation and utilizes an explainable artificial intelligence to leverage the diagnostic potential of mpMRI in detecting prostate cancer (PCa) and determining its significance. Methods: A total of 6,020 MR images from 1,498 cases were considered (1,785 T2 images, 2,719 DWI images, and 1,516 ADC maps). The treatment determined the significance of PCa. Cases who received radical prostatectomy were considered significant, whereas cases with active surveillance and followed for at least two years were considered insignificant. The negative biopsy cases have either a single biopsy setting or multiple biopsy settings with the PCa exclusion. The images were randomly divided into development (80%) and test sets (20%) after stratifying according to the case in each image type. The development set was then divided into a training set (90%) and a validation set (10%). We developed deep learning models for PCa detection and the determination of significant PCa based on the PlexusNet architecture that supports explainable deep learning and volumetric input data. The input data for PCa detection was T2-weighted images, whereas the input data for determining significant PCa include all images types. The performance of PCa detection and determination of significant PCa was measured using the area under receiving characteristic operating curve (AUROC) and compared to the maximum PiRAD score (version 2) at the case level. The 10,000 times bootstrapping resampling was applied to measure the 95% confidence interval (CI) of AUROC. Results: The AUROC for the PCa detection was 0.833 (95% CI: 0.788-0.879) compared to the PiRAD score with 0.75 (0.718-0.764). The DL models to detect significant PCa using the ADC map or DWI images achieved the highest AUROC [ADC: 0.945 (95% CI: 0.913-0.982; DWI: 0.912 (95% CI: 0.871-0.954)] compared to a DL model using T2 weighted (0.850; 95% CI: 0.791-0.908) or PiRAD scores (0.604; 95% CI: 0.544-0.663). Finally, the attention map of PlexusNet from mpMRI with PCa correctly showed areas that contain PCa after matching with corresponding prostatectomy slice. Conclusions: We found that explainable deep learning is feasible on mpMRI and achieves high accuracy in determining cases with PCa and identifying cases with significant PCa.

Determination of biologic and prognostic feature scores from whole slide histology images using deep learning

e17527

Background: In cancer, histopathology is a reflection of the underlying molecular changes in the cancer cells and provides prognostic information on the risk of disease progression. Therefore, whole slide images may harbor histopathological features that have a biological association and are prognostic. Methods: This study has extracted histopathological feature scores generated from hematoxylin and eosin (HE) histology images based on deep learning models developed for the detection of pathological findings related to prostate cancer (PCa). Correlation analyses between the histopathological feature scores and the most relevant genomic alterations related to PCa were performed based on the original results and diagnostic histology images from TCGA PRAD study (n = 251). We extracted feature scores from tumor lesions after applying tumor segmentation and several data transformation using five models developed for detection of cribriform or ductal morphologies, Gleason patterns 3 and 4, and the presumed tumor precursor. For prognostic evaluation, we performed survival analyses of 371 patients from the TCGA PRAD dataset with biochemical recurrence (BCR) using a Cox regression model, Kaplan Meier (KM) curves. We applied the bootstrapping resampling for the uncertainty evaluation and C-statistics for the randomness measurement. Results: The feature scores were significantly correlated with the androgen receptor protein expression, an androgen-signaling score, mRNA expression, and androgen receptor splice variant 7. In addition, feature scores were associated with SPINK1 overexpression, the heterozygous loss of TP53, and SPOP mutations. Additionally, the mRNA and miRNA clusters identified by the TCGA research team for PCa. These features were independent of Gleason grade and were non-random. The survival analyses revealed that a model, including three of five feature scores, achieved a c-index of 0.706 (95% CI: 0.606-0.779). The KM curve showed that these risk groups based on the Cox regression model are significantly discriminative (Log-rank P-value < 0.0001). The low-risk group (n = 177) achieved a 2-year BCR-free survival rate (BFS) of 97.4% (95% CI: 94.9 - 100.0%) and a 5-year PFS of 88.3% (95% CI: 80.6 - 96.7%). In contrast, the high-risk group (n = 194) showed a 2-year PFS of 86.3% (95% CI: 81.1 - 91.8%) and a 5-year BFS of 66.9% (95% CI: 54.6 - 0.82.1%). Conclusions: Our findings uncover the potential of feature scores from histology images as digital biomarkers in precision medicine and as an expanding utility for digital pathology.

Four distinct patient-reported outcome (PRO) trajectories in longitudinal responses collected before, during, and after chemotherapy

2012

Background: Cancer chemotherapy, whether given with curative or palliative intent, is toxic. Toxicity is routinely captured in clinical trials by investigator observation and increasingly by PRO. The ability to capture PRO in the routine treatment workflow has been standard at Stanford since 2015 (Roy et al ASCO 2020). Analysis of longitudinally captured, real world PRO and prospectively identifying patients (pts) whose quality of life (QOL) is at risk of deteriorating either permanently or temporarily is needed. Routine serial PRO measurement should enhance precision care delivery, precision toxicity detection and management. Methods: We identified patients undergoing chemotherapy at Stanford and analyzed PROMIS (PRO Measurement Information System) responses. Pts with PROMIS survey information at three intervals—pre-treatment, during chemotherapy and post chemotherapy—were identified. We evaluated global physical health (GPH) and global mental health (GMH). Pts with a clinically significant decrease (CSD) in GPH or GMH scores were identified. A k-median cluster analysis was used to identify patient trajectory clusters and a machine-learning model was applied to identify risk factors for CSD and predict CSD. Results: We identified 670 adult oncology patients undergoing chemotherapy who completed at least one PROMIS survey in each interval. GPH scores were 48.4 ± 9.1 before, 47.1 ± 8.5 during, and 48.5 ± 8.9 after chemotherapy and GMH scores were 50.5 ± 8.2, 49.1 ± 8.5, and 50.7 ± 9.0, respectively. The majority of patients did not have a CSD in GPH or GMH post treatment compared to pretreatment scores. Pretreatment scores were the strongest predictor of a CSD in GPH and GMH. Trajectory clustering identified four distinct trajectories: Temporary Improver, Temporary Deteriorator, Improver, Inexorable Deteriorators. We were not able to predict any cluster based on pre-treatment features. Conclusions: Using routinely collected PROMIS surveys in a real-world setting, we are able to predict patients with post-treatment decreases in their physical and mental well-being. We further defined four novel patient trajectories during chemotherapy, which could guide personalized supportive interventions to improve patient’s chemotherapy experience. Identification of patients at risk for deterioration and the patterns of deterioration could help guide efficient deployment of toxicity mitigating and supportive care interventions to patients most in need.

17-Gene Genomic Prostate Score Test Results in the Canary Prostate Active Surveillance Study (PASS) Cohort

PURPOSE

The 17-gene Oncotype DX Genomic Prostate Score (GPS) test predicts adverse pathology (AP) in patients with low-risk prostate cancer treated with immediate surgery. We evaluated the GPS test as a predictor of outcomes in a multicenter active surveillance cohort.

MATERIALS AND METHODS

Diagnostic biopsy tissue was obtained from men enrolled at 8 sites in the Canary Prostate Active Surveillance Study. The primary endpoint was AP (Gleason Grade Group [GG] ≥ 3, ≥ pT3a) in men who underwent radical prostatectomy (RP) after initial surveillance. Multivariable regression models for interval-censored data were used to evaluate the association between AP and GPS. Inverse probability of censoring weighting was applied to adjust for informative censoring. Predictiveness curves were used to evaluate how models stratified risk of AP. Association between GPS and time to upgrade on surveillance biopsy was evaluated using Cox proportional hazards models.

RESULTS

GPS results were obtained for 432 men (median follow-up, 4.6 years); 101 underwent RP after a median 2.1 years of surveillance, and 52 had AP. A total of 167 men (39%) upgraded at a subsequent biopsy. GPS was significantly associated with AP when adjusted for diagnostic GG (hazards ratio [HR]/5 GPS units, 1.18; 95% CI, 1.04 to 1.44; P = .030), but not when also adjusted for prostate-specific antigen density (PSAD; HR, 1.85; 95% CI, 0.99 to 4.19; P = .066). Models containing PSAD and GG, or PSAD, GG, and GPS may stratify risk better than a model with GPS and GG. No association was observed between GPS and subsequent biopsy upgrade (P = .48).

CONCLUSION

In our study, the independent association of GPS with AP after initial active surveillance was not statistically significant, and there was no association with upgrading in surveillance biopsy. Adding GPS to a model containing PSAD and diagnostic GG did not significantly improve stratification of risk for AP over the clinical variables alone.

Leveraging Digital Data to Inform and Improve Quality Cancer Care

BACKGROUND

Efficient capture of routine clinical care and patient outcomes is needed at a population-level, as is evidence on important treatment-related side effects and their effect on well-being and clinical outcomes. The increasing availability of electronic health records (EHR) offers new opportunities to generate population-level patient-centered evidence on oncologic care that can better guide treatment decisions and patient-valued care.

METHODS

This study includes patients seeking care at an academic medical center, 2008 to 2018. Digital data sources are combined to address missingness, inaccuracy, and noise common to EHR data. Clinical concepts were identified and extracted from EHR unstructured data using natural language processing (NLP) and machine/deep learning techniques. All models are trained, tested, and validated on independent data samples using standard metrics.

RESULTS

We provide use cases for using EHR data to assess guideline adherence and quality measurements among patients with cancer. Pretreatment assessment was evaluated by guideline adherence and quality metrics for cancer staging metrics. Our studies in perioperative quality focused on medications administered and guideline adherence. Patient outcomes included treatment-related side effects and patient-reported outcomes.

CONCLUSIONS

Advanced technologies applied to EHRs present opportunities to advance population-level quality assessment, to learn from routinely collected clinical data for personalized treatment guidelines, and to augment epidemiologic and population health studies. The effective use of digital data can inform patient-valued care, quality initiatives, and policy guidelines.

IMPACT

A comprehensive set of health data analyzed with advanced technologies results in a unique resource that facilitates wide-ranging, innovative, and impactful research on prostate cancer. This work demonstrates new ways to use the EHRs and technology to advance epidemiologic studies and benefit oncologic care.See all articles in this CEBP Focus section, "Modernizing Population Science."

Clinical Documentation to Predict Factors Associated with Urinary Incontinence Following Prostatectomy for Prostate Cancer

Background

Advances in data collection provide opportunities to use population samples in identifying risk factors for urinary incontinence (UI), which occurs in up to 71% of men with prostate cancer following prostatectomy. Most studies on patient-centered outcomes use surveys or manual chart abstraction for data collection, which can be costly and difficult to scale. We sought to evaluate rates of and risk factors for UI following prostatectomy using natural language processing on electronic health record (EHR) data.

Methods

We conducted a retrospective analysis of patients undergoing prostatectomy for prostate cancer between January 2008 and August 2018 using EHR data from an academic medical center. UI incidence for each patient in the cohort was assessed using natural language processing from clinical notes generated pre- and postoperatively. Multivariable logistic regression was used to evaluate potential risk factors for postoperative UI at various time points within 2 years following surgery.

Results

We identified 3792 patients who underwent prostatectomy for prostate cancer. We found a significant association between preoperative UI and UI in the first (odds ratio [OR], 2.30; 95% confidence interval [CI], 1.24–4.28) and second (OR 2.24, 95% CI 1.04–4.83) years following surgery. Preoperative body mass index was also associated with UI in the second postoperative year (OR 1.11, 95% CI 1.02–1.21).

Conclusion

We show that a natural language processing approach using clinical narratives can be used to assess risk for UI in prostate cancer patients. Unstructured clinical narrative text can help advance future population-level research in patient-centered outcomes and quality of care.

S100A10 Is a Critical Mediator of GAS6/AXL-Induced Angiogenesis in Renal Cell Carcinoma

Angiogenesis is a hallmark of cancer that promotes tumor progression and metastasis. However, antiangiogenic agents have limited efficacy in cancer therapy due to the development of resistance. In clear cell renal cell carcinoma (ccRCC), AXL expression is associated with antiangiogenic resistance and poor survival. Here, we establish a role for GAS6/AXL signaling in promoting the angiogenic potential of ccRCC cells through the regulation of the plasminogen receptor S100A10. Genetic and therapeutic inhibition of AXL signaling in ccRCC tumor xenografts reduced tumor vessel density and growth under the renal capsule. GAS6/AXL signaling activated the expression of S100A10 through SRC to promote plasmin production, endothelial cell invasion, and angiogenesis. Importantly, treatment with the small molecule AXL inhibitor cabozantinib or an ultra-high affinity soluble AXL Fc fusion decoy receptor (sAXL) reduced the growth of a pazopanib-resistant ccRCC patient-derived xenograft. Moreover, the combination of sAXL synergized with pazopanib and axitinib to reduce ccRCC patient-derived xenograft growth and vessel density. These findings highlight a role for AXL/S100A10 signaling in mediating the angiogenic potential of ccRCC cells and support the combination of AXL inhibitors with antiangiogenic agents for advanced ccRCC. SIGNIFICANCE: These findings show that angiogenesis in renal cell carcinoma (RCC) is regulated through AXL/S100A10 signaling and support the combination of AXL inhibitors with antiangiogenic agents for the treatment of RCC.

Machine Learning Approaches for Extracting Stage from Pathology Reports in Prostate Cancer

Clinical and pathological stage are defining parameters in oncology, which direct a patient's treatment options and prognosis. Pathology reports contain a wealth of staging information that is not stored in structured form in most electronic health records (EHRs). Therefore, we evaluated three supervised machine learning methods (Support Vector Machine, Decision Trees, Gradient Boosting) to classify free-text pathology reports for prostate cancer into T, N and M stage groups.

Improved detection of prostate cancer using a magneto-nanosensor assay for serum circulating autoantibodies

Purpose

To develop a magneto-nanosensor (MNS) based multiplex assay to measure protein and autoantibody biomarkers from human serum for prostate cancer (CaP) diagnosis.

Materials and methods

A 4-panel MNS autoantibody assay and a MNS protein assay were developed and optimized in our labs. Using these assays, serum concentration of six biomarkers including prostate-specific antigen (PSA) protein, free/total PSA ratio, as well as four autoantibodies against Parkinson disease 7 (PARK7), TAR DNA-binding protein 43 (TARDBP), Talin 1 (TLN1), and Caldesmon 1 (CALD1) and were analyzed. Human serum samples from 99 patients (50 with non-cancer and 49 with clinically localized CaP) were evaluated.

Results

The MNS assay showed excellent performance characteristics and no cross-reactivity. All autoantibody assays showed a statistically significant difference between CaP and non-cancer samples except for PARK7. The most significant difference was the combination of the four autoantibodies as a panel in addition to the free/total PSA ratio. This combination had the highest area under the curve (AUC)– 0.916 in ROC analysis.

Conclusions

Our results suggest that this autoantibody panel along with PSA and free PSA have potential to segregate patients without cancer from those with prostate cancer with higher sensitivity and specificity than PSA alone.

miR-22 Regulates Invasion, Gene Expression and Predicts Overall Survival in Patients with Clear Cell Renal Cell Carcinoma

Background:

Clear cell renal cell carcinoma (ccRCC) is molecularly diverse and distinct molecular subtypes show different clinical outcomes. MicroRNAs (miRNAs) are essential components of gene regulatory networks and play a crucial role in progression of many cancer types including ccRCC.

Objective:

Identify prognostic miRNAs and determine the role of miR-22 in ccRCC.

Methods:

Hierarchical clustering was done in R using gene expression profiles of over 450 ccRCC cases in The Cancer Genome Atlas (TCGA). Kaplan-Meier analysis was performed to identify prognostic miRNAs in the TCGA dataset. RNA-Seq was performed to identify miR-22 target genes in primary ccRCC cells and Matrigel invasion assay was performed to assess the effects of miR-22 overexpression on cell invasion.

Results:

Hierarchical clustering analysis using 2,621 prognostic genes previously identified by our group demonstrated that ccRCC patients with longer overall survival expressed lower levels of genes promoting proliferation or immune responses, while better maintaining gene expression associated with cortical differentiation and cell adhesion. Targets of 26 miRNAs were significantly enriched in the 2,621 prognostic genes and these miRNAs were prognostic by themselves. MiR-22 was associated with poor overall survival in the TCGA dataset. Overexpression of miR-22 promoted invasion of primary ccRCC cells in vitro and modulated transcriptional programs implicated in cancer progression including DNA repair, cell proliferation and invasion.

Conclusions:

Our results suggest that ccRCCs with differential clinical outcomes have distinct transcriptomes for which miRNAs could serve as master regulators. MiR-22, as a master regulator, promotes ccRCC progression at least in part by enhancing cell invasion.

Early detection of unilateral ureteral obstruction by desorption electrospray ionization mass spectrometry

Desorption electrospray ionization mass spectrometry (DESI-MS) is an emerging analytical tool for rapid in situ assessment of metabolomic profiles on tissue sections without tissue pretreatment or labeling. We applied DESI-MS to identify candidate metabolic biomarkers associated with kidney injury at the early stage. DESI-MS was performed on sections of kidneys from 80 mice over a time course following unilateral ureteral obstruction (UUO) and compared to sham controls. A predictive model of renal damage was constructed using the LASSO (least absolute shrinkage and selection operator) method. Levels of lipid and small metabolites were significantly altered and glycerophospholipids comprised a significant fraction of altered species. These changes correlate with altered expression of lipid metabolic genes, with most genes showing decreased expression. However, rapid upregulation of PG(22:6/22:6) level appeared to be a hitherto unknown feature of the metabolic shift observed in UUO. Using LASSO and SAM (significance analysis of microarrays), we identified a set of well-measured metabolites that accurately predicted UUO-induced renal damage that was detectable by 12 h after UUO, prior to apparent histological changes. Thus, DESI-MS could serve as a useful adjunct to histology in identifying renal damage and demonstrates early and broad changes in membrane associated lipids.

Predictive value of AZGP1 following radical prostatectomy for prostate cancer: a cohort study and meta-analysis

AIMS

Zinc-alpha 2-glycoprotein (AZGP1) is a promising tissue biomarker to predict outcomes in men undergoing treatment for localised prostate cancer (PCa). We aimed to examine the association between AZGP1 expression and the endpoints: risk of biochemical failure (BF), initiating castration-based treatment, developing castration-resistant PCa (CRPC) and PCa-specific mortality following radical prostatectomy (RP).

METHODS

The study included a prospective cohort of 302 patients who underwent RP for PCa from 2002 to 2005. AZGP1 expression was analysed using immunohistochemistry on tissue microarray RP specimens and was scored semiquantitively as low or high expression. Risk of all endpoints was analysed using stratified cumulative incidences and cause-specific Cox regression, and validated with receiver operating curves, calibration and discrimination in competing-risk analyses. A meta-analysis was performed including previous studies investigating AZGP1 expression and risk of BF following RP.

RESULTS

Median time of follow-up was 14.0 years. The cumulative incidence of all endpoints was significantly higher in patients with low AZGP1 expression compared with patients with high AZGP1 expression (p<0.001). In a multivariate analysis, low AZGP1 expression increases the risk of BF (HR 2.7; 95% CI 1.9 to 3.8; p<0.0001), castration-based treatment (HR 2.2; 95% CI 1.2 to 4.2; p=0.01) and CRPC (HR 2.3; 95% CI 1.1 to 5.0; p=0.03). Validation showed a low risk of prediction error and a high model performance for all endpoints. In a meta-analysis, low AZGP1 was associated with BF (HR 1.7; 95% CI 1.2 to 2.5).

CONCLUSIONS

Low AZGP1 expression is associated with the risk of aggressive time-dependent outcomes in men undergoing RP for localised PCa.

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.

Applying the PRECISION approach in biopsy naïve and previously negative prostate biopsy patients

OBJECTIVES

The PRECISION trial provides level 1 evidence supporting prebiopsy multiparametric magnetic resonance imaging (mpMRI) followed by targeted biopsy only when mpMRI is abnormal [1]. This approach reduced over-detection of low-grade cancer while increasing detection of clinically significant cancer (CSC). Still, important questions remain regarding the reproducibility of these findings outside of a clinical trial and quantifying missed CSC diagnoses using this approach. To address these issues, we retrospectively applied the PRECISION strategy in men who each underwent prebiopsy mpMRI followed by systematic and targeted biopsy.

METHODS AND MATERIALS

Clinical, imaging, and pathology data were prospectively collected from 358 biopsy naïve men and 202 men with previous negative biopsies. To apply the PRECISION approach, a retrospective analysis was done comparing the cancer yield from 2 diagnostic strategies: (1) mpMRI followed by targeted biopsy alone for men with Prostate Imaging Reporting and Data System ≥ 3 lesions and (2) systematic biopsy alone for all men. Primary outcomes were biopsies avoided and the proportion of CSC cancer (Grade Group 2-5) and non-CSC (Grade Group 1).

RESULTS

In biopsy naïve patients, the mpMRI diagnostic strategy would have avoided 19% of biopsies while detecting 2.5% more CSC (P= 0.480) and 12% less non-CSC (P< 0.001). Thirteen percent (n= 9) of men with normal mpMRI had CSC on systematic biopsy. For previous negative biopsy patients, the mpMRI diagnostic strategy avoided 21% of biopsies, while detecting 1.5% more CSC (P= 0.737) and 13% less non-CSC (P< 0.001). Seven percent (n= 3) of men with normal mpMRI had CSC on systematic biopsy.

CONCLUSIONS

Our results provide external validation of the PRECISION finding that mpMRI followed by targeted biopsy of suspicious lesions reduces biopsies and over-diagnosis of low-grade cancer. Unlike PRECISION, we did not find increased diagnosis of CSC. This was true in both biopsy naïve and previously negative biopsy cohorts. We have incorporated this information into shared decision making, which has led some men to choose to avoid biopsy. However, we continue to recommend targeted and systematic biopsy in men with abnormal MRI.

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.

Genomic analysis of benign prostatic hyperplasia implicates cellular re-landscaping in disease pathogenesis

Benign prostatic hyperplasia (BPH) is the most common cause of lower urinary tract symptoms in men. Current treatments target prostate physiology rather than BPH pathophysiology and are only partially effective. Here, we applied next-generation sequencing to gain new insight into BPH. By RNAseq, we uncovered transcriptional heterogeneity among BPH cases, where a 65-gene BPH stromal signature correlated with symptom severity. Stromal signaling molecules BMP5 and CXCL13 were enriched in BPH while estrogen regulated pathways were depleted. Notably, BMP5 addition to cultured prostatic myofibroblasts altered their expression profile towards a BPH profile that included the BPH stromal signature. RNAseq also suggested an altered cellular milieu in BPH, which we verified by immunohistochemistry and single-cell RNAseq. In particular, BPH tissues exhibited enrichment of myofibroblast subsets, whilst depletion of neuroendocrine cells and an estrogen receptor (ESR1)-positive fibroblast cell type residing near epithelium. By whole-exome sequencing, we uncovered somatic single-nucleotide variants (SNVs) in BPH, of uncertain pathogenic significance but indicative of clonal cell expansions. Thus, genomic characterization of BPH has identified a clinically-relevant stromal signature and new candidate disease pathways (including a likely role for BMP5 signaling), and reveals BPH to be not merely a hyperplasia, but rather a fundamental re-landscaping of cell types.

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.

Distribution of global health measures from routinely collected PROMIS surveys in patients with breast cancer or prostate cancer

BACKGROUND

The collection of patient-reported outcomes (PROs) is an emerging priority internationally, guiding clinical care, quality improvement projects and research studies. After the deployment of Patient-Reported Outcomes Measurement Information System (PROMIS) surveys in routine outpatient workflows at an academic cancer center, electronic health record data were used to evaluate survey completion rates and self-reported global health measures across 2 tumor types: breast and prostate cancer.

METHODS

This study retrospectively analyzed 11,657 PROMIS surveys from patients with breast cancer and 4411 surveys from patients with prostate cancer, and it calculated survey completion rates and global physical health (GPH) and global mental health (GMH) scores between 2013 and 2018.

RESULTS

A total of 36.6% of eligible patients with breast cancer and 23.7% of patients with prostate cancer completed at least 1 survey, with completion rates lower among black patients for both tumor types (P < .05). The mean T scores (calibrated to a general population mean of 50) for GPH were 48.4 ± 9 for breast cancer and 50.6 ± 9 for prostate cancer, and the GMH scores were 52.7 ± 8 and 52.1 ± 9, respectively. GPH and GMH were frequently lower among ethnic minorities, patients without private health insurance, and those with advanced disease.

CONCLUSIONS

This analysis provides important baseline data on patient-reported global health in breast and prostate cancer. Demonstrating that PROs can be integrated into clinical workflows, this study shows that supportive efforts may be needed to improve PRO collection and global health endpoints in vulnerable populations.

Performance of the 17-gene genomic prostate score test in men with prostate cancer (PCa) managed with active surveillance (AS): Results from the Canary Prostate Active Surveillance Study (PASS)

262

Background: The 17-gene Genomic Prostate Score (GPS) test (scale 0-100) predicts adverse surgical pathology (AP) and recurrence in newly diagnosed low- and intermediate-risk PCa. Studies of the predictive value of the GPS test in men initially managed with AS are limited. Methods: Diagnostic biopsy tissue was obtained from 634 men enrolled at 8 sites in PASS. Time to AP (Gleason Grade Group (GG) ≥3, ≥pT3a, or N1) in men who underwent radical prostatectomy (RP) was the primary endpoint. All diagnostic biopsies and RP specimens were centrally reviewed. Multivariate regression models for interval censored data were used to evaluate the association between time to AP and GPS. Inverse probability of censoring weighting was applied to adjust for informative censoring. Association between GPS and time to Gleason score upgrade on surveillance biopsy was also evaluated using a Cox Proportional Hazards model. Results: GPS results were obtained for 432 men (median follow-up 4.6 [IQR: 2.9-6.2] years); 374 and 58 with GG 1 or 2 cancer, respectively; median PSA density (PSAD) was 0.11 [IQR: 0.08-0.15]; 101 men underwent RP with central pathology after a median of 2.1 [IQR: 1.3-4.3] years surveillance, and 52 (52%) men undergoing RP had AP. 167 men upgraded at a subsequent biopsy. No clinico-pathologic covariates were significantly associated with AP other than PSAD. GPS was significantly associated with time to AP (hazards ratio [HR]/20 GPS units: 1.96 [95% CI = 1.17-4.28]; p = 0.030), when adjusted for diagnostic GG, or for dichotomous PSAD ( < vs ≥ 0.15; HR: 1.83, 95% CI = 1.04-3.62; p = 0.046). GPS was not significantly associated with AP (HR: 1.61, 95% CI = 0.87-2.98; p = 0.12) when adjusted for continuous PSAD. No association, either univariable or multivariable, was observed between GPS and subsequent biopsy upgrade. Conclusions: In a cohort of men on AS, GPS was associated with time to AP when adjusted for diagnostic GG or dichotomous PSAD. GPS was not associated with surveillance biopsy GG upgrading or AP at surgery after adjustment for continuous PSAD, although a trend was seen for AP, suggesting an association may be seen in a larger study.