The Relative Importance of Race Compared to Health Care and Social Factors in Predicting Prostate Cancer Mortality: A Random Forest Approach

TCGA Expression Analyses of 10 Carcinoma Types Reveal Clinically Significant Racial Differences

Epidemiological studies reveal disparities in cancer incidence and outcome rates between racial groups in the United States. In our study, we investigated molecular differences between racial groups in 10 carcinoma types. We used publicly available data from The Cancer Genome Atlas to identify patterns of differential gene expression in tumor samples obtained from 4112 White, Black/African American, and Asian patients. We identified race-dependent expression of numerous genes whose mRNA transcript levels were significantly correlated with patients' survival. Only a small subset of these genes was differentially expressed in multiple carcinomas, including genes involved in cell cycle progression such as CCNB1, CCNE1, CCNE2, and FOXM1. In contrast, most other genes, such as transcriptional factor ETS1 and apoptotic gene BAK1, were differentially expressed and clinically significant only in specific cancer types. Our analyses also revealed race-dependent, cancer-specific regulation of biological pathways. Importantly, homology-directed repair and ERBB4-mediated nuclear signaling were both upregulated in Black samples compared to White samples in four carcinoma types. This large-scale pan-cancer study refines our understanding of the cancer health disparity and can help inform the use of novel biomarkers in clinical settings and the future development of precision therapies.

Investigating the racial gap in prostate cancer screening with prostate-specific antigen among younger men from 2012 to 2020

BACKGROUND

The United States Preventive Services Task Force recommended against prostate-specific antigen (PSA) screening in 2012, which was modified in 2018 into shared decision making for men aged 55-70 years with a life expectancy over 10 years. We studied the trends in PSA screening in younger Black and White men with the implementation of the 2012 and 2018 guidelines.

METHODS

Younger Black and White men (aged 40-54 years) were identified using the Behavioral Risk Factor Surveillance System database biennially from 2012 to 2020. Our primary outcome was PSA screening within 2 years of the survey. An adjusted logistic regression model with 2-way interaction assessment between race and survey year was used to investigate the temporal trend of PSA screening in younger Black and White men.

RESULTS

A total of 142 892 men were included. We saw steadily decreasing odds of PSA screening among both younger Black and White men in 2014, 2016, 2018, and 2020 compared with 2012 (for younger Black men: odds ratio [OR]2014 = 0.77, 95% confidence interval [CI] = 0.62 to 0.96, OR2016 = 0.51, 95% CI = 0.41 to 0.63, OR2018 = 0.33, 95%CI = 0.27 to 0.42, OR2020 = 0.25, 95% CI = 0.18 to 0.32; and for younger White men: OR2014 = 0.81, 95% CI = 0.76 to 0.87, OR2016 = 0.66, 95% CI = 0.61 to 0.71, OR2018 = 0.41, 95%CI = 0.37 to 0.44, OR2020 = 0.36, 95% CI = 0.33 to 0.39). Younger Black men showed a brisker decrease in PSA screening in 2016, 2018, and 2020 compared with younger White men (all P < .05).

CONCLUSIONS

PSA screening among younger men steadily decreased over the past decade since the 2012 United States Preventive Services Task Force guidelines, demonstrating a narrowing racial gap. How such an observed trend translates to long-term clinical outcomes for younger Black men remains to be seen.

Exploring the Use of Artificial Intelligence in the Management of Prostate Cancer

PURPOSE OF REVIEW

This review aims to explore the current state of research on the use of artificial intelligence (AI) in the management of prostate cancer. We examine the various applications of AI in prostate cancer, including image analysis, prediction of treatment outcomes, and patient stratification. Additionally, the review will evaluate the current limitations and challenges faced in the implementation of AI in prostate cancer management.

RECENT FINDINGS

Recent literature has focused particularly on the use of AI in radiomics, pathomics, the evaluation of surgical skills, and patient outcomes. AI has the potential to revolutionize the future of prostate cancer management by improving diagnostic accuracy, treatment planning, and patient outcomes. Studies have shown improved accuracy and efficiency of AI models in the detection and treatment of prostate cancer, but further research is needed to understand its full potential as well as limitations.

Predicting COVID-19 community infection relative risk with a Dynamic Bayesian Network

As COVID-19 continues to impact the United States and the world at large it is becoming increasingly necessary to develop methods which predict local scale spread of the disease. This is especially important as newer variants of the virus are likely to emerge and threaten community spread. We develop a Dynamic Bayesian Network (DBN) to predict community-level relative risk of COVID-19 infection at the census tract scale in the U.S. state of Indiana. The model incorporates measures of social and environmental vulnerability-including environmental determinants of COVID-19 infection-into a spatial temporal prediction of infection relative risk 1-month into the future. The DBN significantly outperforms five other modeling techniques used for comparison and which are typically applied in spatial epidemiological applications. The logic behind the DBN also makes it very well-suited for spatial-temporal prediction and for "what-if" analysis. The research results also highlight the need for further research using DBN-type approaches that incorporate methods of artificial intelligence into modeling dynamic processes, especially prominent within spatial epidemiologic applications.

Variation in and Factors Associated With US County-Level Cancer Mortality, 2008-2019

IMPORTANCE

The association between cancer mortality and risk factors may vary by geography. However, conventional methodological approaches rarely account for this variation.

OBJECTIVE

To identify geographic variations in the association between risk factors and cancer mortality.

DESIGN, SETTING, AND PARTICIPANTS

This geospatial cross-sectional study used county-level data from the National Center for Health Statistics for individuals who died of cancer from 2008 to 2019. Risk factor data were obtained from County Health Rankings & Roadmaps, Health Resources and Services Administration, and Centers for Disease Control and Prevention. Analyses were conducted from October 2021 to July 2022.

MAIN OUTCOMES AND MEASURES

Conventional random forest models were applied nationwide and by US region, and the geographical random forest model (accounting for local variation of association) was applied to assess associations between a wide range of risk factors and cancer mortality.

RESULTS

The study included 7 179 201 individuals (median age, 70-74 years; 3 409 508 women [47.5%]) who died from cancer in 3108 contiguous US counties during 2008 to 2019. The mean (SD) county-level cancer mortality rate was 177.0 (26.4) deaths per 100 000 people. On the basis of the variable importance measure, the random forest models identified multiple risk factors associated with cancer mortality, including smoking, receipt of Supplemental Nutrition Assistance Program (SNAP) benefits, and obesity. The geographical random forest model further identified risk factors that varied at the county level. For example, receipt of SNAP benefits was a high-importance factor in the Appalachian region, North and South Dakota, and Northern California; smoking was of high importance in Kentucky and Tennessee; and female-headed households were high-importance factors in North and South Dakota. Geographic areas with certain high-importance risk factors did not consistently have a corresponding high prevalence of the same risk factors.

CONCLUSIONS AND RELEVANCE

In this cross-sectional study, the associations between cancer mortality and risk factors varied by geography in a way that did not correspond strictly to risk factor prevalence. The degree to which other place-specific characteristics, observed and unobserved, modify risk factor effects should be further explored, and this work suggests that risk factor importance may be a preferable paradigm for selecting cancer control interventions compared with risk factor prevalence.

Mediators of Racial Disparity in the Use of Prostate Magnetic Resonance Imaging Among Patients With Prostate Cancer

Importance

Racial disparity in the use of prostate magnetic resonance imaging (MRI) presents obstacles to closing gaps in prostate cancer diagnosis, treatment, and outcome.

Objective

To identify clinical, sociodemographic, and structural processes underlying racial disparity in the use of prostate MRI among men with a new diagnosis of prostate cancer.

Design, Setting, and Participants

This population-based cohort study used mediation analysis to assess claims in the US Surveillance, Epidemiology, and End Results (SEER)-Medicare database for prostate MRI among 39 534 patients with a diagnosis of localized prostate cancer from January 1, 2011, to December 31, 2015. Statistical analysis was performed from April 1, 2020, to September 1, 2021.

Exposure

Diagnosis of prostate cancer.

Main Outcomes and Measures

Claims for prostate MRI within 6 months before or after diagnosis of prostate cancer were assessed. Candidate clinical and sociodemographic meditators were identified based on their association with both race and prostate MRI, including the Index of Concentration at the Extremes (ICE), as specified to measure racialized residential segregation. Mediation analysis was performed using nonlinear multiple additive regression trees models to estimate the direct and indirect effects of mediators.

Results

A total of 39 534 eligible male patients (3979 Black patients [10.1%] and 32 585 White patients [82.4%]; mean [SD] age, 72.8 [5.3] years) were identified. Black patients with prostate cancer were less likely than White patients to receive a prostate MRI (6.3% vs 9.9%; unadjusted odds ratio, 0.62, 95% CI, 0.54-0.70). Approximately 24% (95% CI, 14%-32%) of the racial disparity in prostate MRI use between Black and White patients was attributable to geographic differences (SEER registry), 19% (95% CI, 11%-28%) was attributable to neighborhood-level socioeconomic status (residence in a high-poverty area), 19% (95% CI, 10%-29%) was attributable to racialized residential segregation (ICE quintile), and 11% (95% CI, 7%-16%) was attributable to a marker of individual-level socioeconomic status (dual eligibility for Medicare and Medicaid). Clinical and pathologic factors were not significant mediators. In this model, the identified mediators accounted for 81% (95% CI, 64%-98%) of the observed racial disparity in prostate MRI use between Black and White patients.

Conclusions and Relevance

In this this population-based cohort study of US adults, mediation analysis revealed that sociodemographic factors and manifestations of structural racism, including poverty and residential segregation, explained most of the racial disparity in the use of prostate MRI among older Black and White men with prostate cancer. These findings can be applied to develop targeted strategies to improve cancer care equity.

Association of Race, Ethnicity, and Socioeconomic Status With Esthesioneuroblastoma Presentation, Treatment, and Survival

Objective

Socioeconomic and other demographic factors are associated with outcomes in head and neck cancer. This study uses a national cancer database to explore how patient race, ethnicity, and socioeconomic status (SES) are associated with esthesioneuroblastoma outcomes, including 5-year disease-specific survival (DSS), conditional DSS, stage at diagnosis, and treatment.

Study Design

Retrospective cohort analysis.

Setting

Patients with esthesioneuroblastomas between 1973 and 2015 from the SEER registry (Surveillance, Epidemiology, and End Results).

Methods

The National Cancer Institute Yost Index, a census tract–level composite score composed of 7 parameters, was used to categorize the SES of patients. Kaplan-Meier analysis and Cox regression were conducted to assess DSS. Conditional DSS was calculated per estimates from simplified Cox models. Logistic regression was conducted to identify risk factors for advanced cancer stage at diagnosis and the likelihood of receiving multimodal therapy.

Results

Complete data were included for 561 patients. DSS was significantly associated with SES (log-rank, P < .01) but not race. According to Cox regression, DSS was worse for the lowest SES tertile vs the highest (hazard ratio, 1.70 [95% CI, 1.05-2.75]; P = .03). Patients of the lowest SES tertile exhibited an increased risk of advanced cancer stage at diagnosis as compared with the highest SES tertile (odds ratio, 1.84 [95% CI, 1.06-3.30]; P = .035). Black patients (odds ratio, 0.44 [95% CI, 0.24-0.84]; P = .011) were less likely than other patients to receive multimodal therapy. SES alone was not associated with receiving multimodal therapy.

Conclusion

SES is significantly associated with DSS and conditional DSS for patients with esthesioneuroblastomas. Inequalities in access to care and treatment likely contribute to these disparities.

Racial and ethnic disparities in paranasal sinus malignancies

BACKGROUND

Racial and ethnic disparities in cancer outcomes have been demonstrated for several different malignancies. In this study we aimed to quantify disease-specific survival (DSS) and the 5-year conditional disease-specific survival (CDSS, the change in life expectancy with increasing survivorship) for paranasal sinus cancer by race and ethnicity.

METHODS

Patients with sinus cancer between 1973 and 2015 were extracted from the Surveillance, Epidemiology, End Results (SEER) registry. Kaplan-Meier analysis for DSS was stratified by race and ethnicity. Cox regression models of DSS were generated controlling for stage, age, race, and ethnicity. CDSS was calculated using Cox models. Logistic regression was conducted to identify risk factors for younger age at diagnosis, late-stage at diagnosis, and likelihood of receiving surgical intervention when recommended.

RESULTS

The analysis included a total of 5202 patients. DSS was significantly different when stratified by race (p < 0.01). Compared with White patients, Black patients (hazard ratio [HR], 1.29; 95% confidence interval [CI], 1.13-1.45; p < 0.001) and American Indian/Alaskan Natives (HR, 1.94; 95% CI, 1.37-2.74, p < 0.001) exhibited increased mortality when controlling for other factors. Black patients had worse CDSS for regional and distant staged cancer compared with other races; American Indian/Alaskan Native patients had worse CDSS for cancers of all stages. Hispanic patients were more likely to present with advanced disease (odds ratio [OR], 1.47; 95% CI, 1.07-2.07; p = 0.020). American Indian/Alaskan Native patients were less likely than White patients to receive surgical intervention when recommended (OR, 0.42; 95% CI, 0.21-0.04; p = 0.024). Nonwhite patients were more likely to be diagnosed at a younger age. Variations in racial and ethnic disparities were observed over time.

CONCLUSION

Race and ethnicity significantly impact paranasal sinus cancer outcome metrics. Disparities in outcomes are likely multifactorial.

A scoping review on the use of machine learning in research on social determinants of health: Trends and research prospects

Background

Machine learning (ML) has spread rapidly from computer science to several disciplines. Given the predictive capacity of ML, it offers new opportunities for health, behavioral, and social scientists. However, it remains unclear how and to what extent ML is being used in studies of social determinants of health (SDH).

Methods

Using four search engines, we conducted a scoping review of studies that used ML to study SDH (published before May 1, 2020). Two independent reviewers analyzed the relevant studies. For each study, we identified the research questions, Results, data, and algorithms. We synthesized our findings in a narrative report.

Results

Of the initial 8097 hits, we identified 82 relevant studies. The number of publications has risen during the past decade. More than half of the studies (n = 46) used US data. About 80% (n = 66) utilized surveys, and 70% (n = 57) employed ML for common prediction tasks. Although the number of studies in ML and SDH is growing rapidly, only a few studies used ML to improve causal inference, curate data, or identify social bias in predictions (i.e., algorithmic fairness).

Conclusions

While ML equips researchers with new ways to measure health outcomes and their determinants from non-conventional sources such as text, audio, and image data, most studies still rely on traditional surveys. Although there are no guarantees that ML will lead to better social epidemiological research, the potential for innovation in SDH research is evident as a result of harnessing the predictive power of ML for causality, data curation, or algorithmic fairness.

Artificial intelligence in oncology: From bench to clinic

In the past few years, Artificial Intelligence (AI) techniques have been applied to almost every facet of oncology, from basic research to drug development and clinical care. In the clinical arena where AI has perhaps received the most attention, AI is showing promise in enhancing and automating image-based diagnostic approaches in fields such as radiology and pathology. Robust AI applications, which retain high performance and reproducibility over multiple datasets, extend from predicting indications for drug development to improving clinical decision support using electronic health record data. In this article, we review some of these advances. We also introduce common concepts and fundamentals of AI and its various uses, along with its caveats, to provide an overview of the opportunities and challenges in the field of oncology. Leveraging AI techniques productively to provide better care throughout a patient's medical journey can fuel the predictive promise of precision medicine.

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

OBJECTIVES

To develop a model to predict biochemical recurrence (BCR) after radical prostatectomy (RP), using artificial intelligence (AI) techniques.

PATIENTS AND METHODS

This study collected data from 7,128 patients with prostate cancer (PCa) who received RP at 3 tertiary hospitals. After preprocessing, we used the data of 6,755 cases to generate the BCR prediction model. There were 16 input variables with BCR as the outcome variable. We used a random forest to develop the model. Several sampling techniques were used to address class imbalances.

RESULTS

We achieved good performance using a random forest with synthetic minority oversampling technique (SMOTE) using Tomek links, edited nearest neighbors (ENN), and random oversampling: accuracy = 96.59%, recall = 95.49%, precision = 97.66%, F1 score = 96.59%, and ROC AUC = 98.83%.

CONCLUSION

We developed a BCR prediction model for RP. The Dr. Answer AI project, which was developed based on our BCR prediction model, helps physicians and patients to make treatment decisions in the clinical follow-up process as a clinical decision support system.

A Data Science Approach to Analyze the Association of Socioeconomic and Environmental Conditions With Disparities in Pediatric Surgery

Background: Scientific evidence confirm that significant racial disparities exist in healthcare, including surgery outcomes. However, the causal pathway underlying disparities at preoperative physical condition of children is not well-understood. Objectives: This research aims to uncover the role of socioeconomic and environmental factors in racial disparities at the preoperative physical condition of children through multidimensional integration of several data sources at the patient and population level. Methods: After the data integration process an unsupervised k-means algorithm on neighborhood quality metrics was developed to split 29 zip-codes from Memphis, TN into good and poor-quality neighborhoods. Results: An unadjusted comparison of African Americans and white children showed that the prevalence of poor preoperative condition is significantly higher among African Americans compared to whites. No statistically significant difference in surgery outcome was present when adjusted by surgical severity and neighborhood quality. Conclusions: The socioenvironmental factors affect the preoperative clinical condition of children and their surgical outcomes.

The Impact of African Ancestry on Prostate Cancer Disparities in the Era of Precision Medicine

Prostate cancer disproportionately affects men of African ancestry at nearly twice the rate of men of European ancestry despite the advancement of treatment strategies and prevention. In this review, we discuss the underlying causes of these disparities including genetics, environmental/behavioral, and social determinants of health while highlighting the implications and challenges that contribute to the stark underrepresentation of men of African ancestry in clinical trials and genetic research studies. Reducing prostate cancer disparities through the development of personalized medicine approaches based on genetics will require a holistic understanding of the complex interplay of non-genetic factors that disproportionately exacerbate the observed disparity between men of African and European ancestries.

Classify multicategory outcome in patients with lung adenocarcinoma using clinical, transcriptomic and clinico-transcriptomic data: machine learning versus multinomial models

Classification of multicategory survival-outcome is important for precision oncology. Machine learning (ML) algorithms have been used to accurately classify multi-category survival-outcome of some cancer-types, but not yet that of lung adenocarcinoma. Therefore, we compared the performances of 3 ML models (random forests, support vector machine [SVM], multilayer perceptron) and multinomial logistic regression (Mlogit) models for classifying 4-category survival-outcome of lung adenocarcinoma using the TCGA. Mlogit model overall performed similar to SVM and multilayer perceptron models (micro-average area under curve=0.82), while random forests model was inferior. Surprisingly, transcriptomic data alone and clinico-transcriptomic data appeared sufficient to accurately classify the 4-category survival-outcome in these patients, but no models using clinical data alone performed well. Notably, NDUFS5, P2RY2, PRPF18, CCL24, ZNF813, MYL6, FLJ41941, POU5F1B, and SUV420H1 were the top-ranked genes that were associated with alive without disease and inversely linked to other outcomes. Similarly, BDKRB2, TERC, DNAJA3, MRPL15, SLC16A13, CRHBP and ACSBG2 were associated with alive with progression and GAL3ST3, AD2, RAB41, HDC, and PLEKHG1 associated with dead with disease, respectively, while also inversely linked other outcomes. These cross-linked genes may be used for risk-stratification and future treatment development.

Lack of Racial Survival Differences in Metastatic Prostate Cancer in National Cancer Data Base (NCDB): A Different Finding Compared to Non-metastatic Disease

Background: Inconsistent findings have been reported in the literature regarding racial differences in survival outcomes between African American and white patients with metastatic prostate cancer (mPCa). The current study utilized a national database to determine whether racial differences exist among the target population to address this inconsistency.

Methods: This study retrospectively reviewed prostate cancer (PCa) patient data (N = 1,319,225) from the National Cancer Database (NCDB). The data were divided into three groupings based on the metastatic status: (1) no metastasis (N = 318,291), (2) bone metastasis (N = 29,639), and (3) metastases to locations other than bone, such as brain, liver, or lung (N = 952). Survival probabilities of African American and white PCa patients with bone metastasis were examined through parametric proportional hazards Weibull models and Bayesian survival analysis. These results were compared to patients with no metastasis or other types of metastases.

Results: No statistically supported racial disparities were observed for African American and white men with bone metastasis (p = 0.885). Similarly, there were no racial disparities in survival for those men suffering from other metastases (liver, lung, or brain). However, racial disparities in survival were observed among the two racial groups with non-metastatic PCa (p < 0.001) or when metastasis status was not taken into account (p < 0.001). The Bayesian analysis corroborates the finding.

Conclusion: This research supports our previous findings and shows that there are no racial differences in survival outcomes between African American and white patients with mPCa. In contrast, racial disparities in the survival outcome continue to exist among non-metastatic PCa patients. Further research is warranted to explain this difference.

Factors associated with appropriate and low-value PSA testing

BACKGROUND

Prostate-specific antigen (PSA) testing for early detection of prostate cancer is low-value when it is not indicated by guidelines and the harms outweigh the benefits. In this retrospective cohort study, we identify provider and patient factors associated with PSA testing, particularly in situations where testing would be low-value.

METHODS

We used electronic health record data from 2011 to 2018 representing 1,738,021 health system encounters in the United States. Using logistic generalized estimating equation models, we examined patient factors (age, comorbid illness, family history, race and prior PSA results), provider factors (gender, specialty, graduation year and medical school rank), and overall time trends associated with PSA testing in low-value and appropriate settings.

RESULTS

Comorbid illness (odds ratio (OR) 0.0 for 3+ conditions vs none) and no prior PSA testing (OR 0.2) were associated with a lower likelihood of PSA testing in low-value situations, while family history of prostate cancer (OR 1.6) and high prior PSA test results (OR 2.2 for PSA > 6 vs 0-1) were associated with a greater likelihood. Men aged 55-65 years were at greatest risk for PSA testing in low-value situations. The provider factor associated with PSA testing in low-value situations was specialty, with urologists being most likely (OR 2.3 versus advanced practice providers). Internal medicine physicians were more likely to perform PSA testing during low-value situations (OR 1.3 versus advanced practice providers) but much more likely to order a PSA test where appropriate (OR 2.2). All PSA testing decreased since 2011.

CONCLUSION

We identified several patient and provider factors associated with PSA testing in low-value settings. Some aspects suggest attention to relevant factors for PSA testing in low-value settings (e.g. comorbid illness), while others may encourage PSA testing in low-value settings (e.g. family history). The greatest likelihood of PSA testing in low-value settings is among men within the age range most commonly recommended by guidelines.

Predicting long-term multicategory cause of death in patients with prostate cancer: random forest versus multinomial model

The majority of patients with prostate cancer die of non-cancer causes of death (COD). It is thus important to accurately predict multi-category COD in these patients. Random forest (RF), a popular machine learning model, has been shown useful for predicting binary cancer-specific deaths. However, its accuracy for predicting multi-category COD in cancer patients is unclear. We included patients in Surveillance, Epidemiology, and End Results-18 cancer registry-program with prostate cancer diagnosed in 2004 (followed-up through 2016). They were randomly divided into training and testing sets with equal sizes. We evaluated prediction accuracies of RF and conventional statistical/multinomial models for 6-category COD by data-encoding types using the 2-fold cross-validation approach. Among 49,864 prostate cancer patients, 29,611 (59.4%) were alive at the end of follow-up, and 5,448 (10.9%) died of cardiovascular disease, 4,607 (9.2%) of prostate cancer, 3,681 (7.4%) of non-prostate cancer, 717 (1.4%) of infection, and 5,800 (11.6%) of other causes. We predicted 6-category COD among these patients with a mean accuracy of 59.1% (n=240, 95% CI, 58.7%-59.4%) in RF models with one-hot encoding, and 50.4% (95% CI, 49.7%-51.0%) in multinomial models. Tumor characteristics, prostate-specific antigen level, and diagnosis confirmation-method were important in RF and multinomial models. In RF models, no statistical differences were found between the accuracies of training versus cross-validation phases, and those of categorical versus one-hot encoding. We here report that RF models can outperform multinomial logistic models (absolute accuracy-difference, 8.7%) in predicting long-term 6-category COD among prostate cancer patients, while pathology diagnosis itself and tumor pathology remain important factors.

Association of Treatment With 5α-Reductase Inhibitors and Prostate Cancer Mortality Among Older Adults

This cohort study examines the association between use of 5α-reductase inhibitors and prostate cancer mortality among US Medicare beneficiaries.