Genomic and Socioeconomic Determinants of Racial Disparities in Breast Cancer Survival: Insights from the All of Us Program.

Background: Breast cancer outcomes are worse among Black women in the U.S. compared to White women. While extensive research has focused on risk factors contributing to breast cancer; the role of genomic elements in health disparities between these racial groups remains unclear. This study aims to identify genomic variants and socioeconomic status (SES) determinants influencing racial disparities in breast cancer survival through multiple mediation analyses. Methods: Our investigation is based on the NIH-supported All of Us (AoU) program and analyzes 7452 female participants with malignant tumors of breast, including 5073 with genomic data. A log-rank test reveals significant racial differences in overall survival time between Black and White participants (p-value = 0.04). Multiple mediation analysis examines the effects of 9481 genetic variables across 23 chromosomes in explaining the racial disparity in survival, adjusting for SES variables. Results: 15 gene mutations, in addition to age, general health, and general quality of life, have significant effects (p-values < 0.001) in explaining the observed racial disparity. Mutations in TMEM132B, NARFL, SALL1, PAD12, RIPK1, ASB14, DCX, GNB1L, ARHGAP32, AL135787.1, WBP11, SLC16A12AS1, AP000345.1, IKBKB, and SUPT20H have significantly different distributions between Black and White participants. The disparity is completely explained by the included variables as the direct effect is insignificant (p-value = 0.73). Conclusions: The combined impact of SES determinants and genetic mutations can explain the observed differences in breast cancer survival among Black and White participants. Future studies will explore pathways and design in vivo and in vitro experiments to validate the functions of these genes.

Racial Inequities in Receiving Neoadjuvant Treatment in Patients With Pancreatic Cancer Are Associated With Worse Outcomes in Louisiana.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related deaths. In recent years, the approach to managing this malignancy has evolved toward embracing neoadjuvant treatment (NAT), backed by studies reporting its survival benefit. This study aimed to identify factors that contribute to disparities in NAT utilization and their impact on outcomes in patients with PDAC who underwent resection in Louisiana. METHODS: Data on diagnosed PDAC cases were obtained from the Louisiana Tumor Registry between 2000 and 2020. We conducted multivariable logistic regression to adjust for potential confounding factors in assessing the covariate relationships with NAT use. Multivariate Cox regression analysis was performed to determine which factors were associated with survival. Chained multiple imputation was performed on covariates with missing data in multivariable regressions. RESULTS: The study encompassed 2121 patients who underwent resection for PDAC. Upon controlling for potential confounding variables, Black patients were on average 5.7% less likely to receive NAT than their White counterparts (ATE = 5.7, aOR= 0.56, 95% CI = 0.40-0.80, p = 0.001). After adjustment for confounding factors, there was a significant decrease in the risk of overall death for patients who received NAT (aHR = 0.82, 95% CI = 0.71-0.94, p = 0.006). There was no significant interaction between race and NAT for the risk of death. CONCLUSION: Black patients with PDAC were less likely to receive NAT before resection in Louisiana. Overall survival improved in patients who underwent NAT. These differences were independent of insurance status and poverty zip codes, and future investigations should identify modifiable barriers to access and receipt of NAT in patients with PDAC.

Development of message passing-based graph convolutional networks for classifying cancer pathology reports.

BACKGROUND: Applying graph convolutional networks (GCN) to the classification of free-form natural language texts leveraged by graph-of-words features (TextGCN) was studied and confirmed to be an effective means of describing complex natural language texts. However, the text classification models based on the TextGCN possess weaknesses in terms of memory consumption and model dissemination and distribution. In this paper, we present a fast message passing network (FastMPN), implementing a GCN with message passing architecture that provides versatility and flexibility by allowing trainable node embedding and edge weights, helping the GCN model find the better solution. We applied the FastMPN model to the task of clinical information extraction from cancer pathology reports, extracting the following six properties: main site, subsite, laterality, histology, behavior, and grade. RESULTS: We evaluated the clinical task performance of the FastMPN models in terms of micro- and macro-averaged F1 scores. A comparison was performed with the multi-task convolutional neural network (MT-CNN) model. Results show that the FastMPN model is equivalent to or better than the MT-CNN. CONCLUSIONS: Our implementation revealed that our FastMPN model, which is based on the PyTorch platform, can train a large corpus (667,290 training samples) with 202,373 unique words in less than 3 minutes per epoch using one NVIDIA V100 hardware accelerator. Our experiments demonstrated that using this implementation, the clinical task performance scores of information extraction related to tumors from cancer pathology reports were highly competitive.

Higher 10-Year Survival with Breast-Conserving Therapy over Mastectomy for Women with Early-Stage (I-II) Breast Cancer: Analysis of the CDC Patterns of Care Data Base.

Background: Studies in the United States are scarce that assess the survival differences between breast-conserving surgery plus radiation (Breast-Conserving Therapy; BCT) and mastectomy groups using population-based data while accounting for sociodemographic and clinical factors that affect the survival of women with early-stage breast cancer (ESBC). Objective: To assess whether BCT provides superior long-term overall survival (OS) and breast cancer-specific survival (BCSS) compared with mastectomy in women with ESBC, while considering key factors that impact survival. Design: Cohort study. Methods: We analyzed data on women aged 20 years and older diagnosed with stage I-II breast cancer (BC) in 2004 who received either BCT or mastectomy. The data were collected by 5 state cancer registries through the Centers for Disease Control and Prevention-funded Patterns of Care study. Multivariable Cox proportional hazard models, accounting for sociodemographic and clinical factors, were used to calculate hazard ratios (HRs) with 95% confidence intervals (CI). Sensitivity analysis involved optimal caliper propensity score (PS) matching to address residual confounding. Results: Of the 3495 women, 41.5% underwent mastectomy. The 10-year OS and BCSS were 82.7% and 91.1% for BCT and 72.3% and 85.7% for mastectomy, respectively. Adjusted models showed that mastectomy recipients had a 22% higher risk of all-cause deaths (ACD) (HR = 1.22, 95% CI = [1.06, 1.41]) and a 26% higher risk of breast cancer-specific deaths (BCD) (HR = 1.26, 95% CI = [1.02, 1.55]) than BCT recipients. Sensitivity analysis demonstrated that mastectomy was associated with a higher risk of ACD (P < .05) but did not exhibit a statistically significant risk for BCD. Women with HR+/HER2+ (luminal B) or invasive ductal carcinoma BC who underwent mastectomy had higher risks of ACD and BCD compared with BCT recipients, while the hazards for ACD in triple-negative BC did not remain significant after adjusting for covariates. Conclusion: ESBC BCT recipients demonstrate superior OS and BCSS compared with mastectomy recipients.

Breast-preserving treatment leads to higher 10-year survival in early-stage breast cancer This study compared the long-term survival outcomes of breast-conserving therapy (BCT) and mastectomy for early-stage breast cancer. Analyzing data from over 3400 women diagnosed in 2004, researchers found that BCT recipients had higher 10-year overall and breast cancer-specific survival rates compared with mastectomy recipients. Adjusted models showed a 22% higher risk of all-cause death.

Glycemic control in diabetic patients improved overall lung cancer survival across diverse populations.

BACKGROUND: The consequence of diabetes on lung cancer overall survival (OS) is debated. This retrospective study used 2 large lung cancer databases to assess comprehensively diabetes effects on lung cancer OS in diverse demographic populations, including health disparity. METHODS: The University of Texas MD Anderson Cancer Center database (32 643 lung cancer patients with 11 973 patients with diabetes) was extracted from electronic health records (EHRs) using natural language processing (NLP). Associations were between diabetes and lung cancer prognostic features (age, sex, race, body mass index [BMI], insurance status, smoking, stage, and histopathology). Hemoglobin A1C (HgbA1c) and glucose levels assessed glycemic control. Validation was with a Louisiana cohort (17 768 lung cancer patients with 5402 patients with diabetes) enriched for health disparity cases. Kaplan-Meier analysis, log-rank test, multivariable Cox proportional hazard models, and survival tree analyses were employed. RESULTS: Lung cancer patients with diabetes exhibited marginally elevated OS or no statistically significant difference versus nondiabetic patients. When examining OS for 2 glycemic levels (HgbA1c > 7.0 or glucose > 154 mg/dL vs HgbA1c > 9.0 or glucose > 215 mg/dL), a statistically significant improvement in OS occurred in lung cancer patients with controlled versus uncontrolled glycemia (P < .0001). This improvement spanned sex, age, smoking status, insurance status, stage, race, BMI, histopathology, and therapy. Survival tree analysis revealed that obese and morbidly obese patients with controlled glycemia had higher lung cancer OS than comparison groups. CONCLUSION: These findings indicate a need for optimal glycemic control to improve lung cancer OS in diverse populations with diabetes.

NCI SEER-Linked Virtual Tissue Repository Pilot.

BACKGROUND: The Surveillance, Epidemiology, and End Results (SEER) Program with the National Cancer Institute tested whether population-based cancer registries can serve as honest brokers to acquire tissue and data in the SEER-Linked Virtual Tissue Repository (VTR) Pilot. METHODS: We collected formalin-fixed, paraffin-embedded tissue and clinical data from patients with pancreatic ductal adenocarcinoma (PDAC) and breast cancer (BC) for two studies comparing cancer cases with highly unusual survival (≥5 years for PDAC and ≤30 months for BC) to pair-matched controls with usual survival (≤2 years for PDAC and ≥5 years for BC). Success was defined as the ability for registries to acquire tissue and data on cancer cases with highly unusual outcomes. RESULTS: Of 98 PDAC and 103 BC matched cases eligible for tissue collection, sources of attrition for tissue collection were tissue being unavailable, control paired with failed case, second control that was not requested, tumor necrosis ≥20%, and low tumor cellularity. In total, tissue meeting the study criteria was obtained for 70 (71%) PDAC and 74 (72%) BC matched cases. For patients with tissue received, clinical data completeness ranged from 59% for CA-19-9 after treatment to >95% for margin status, whether radiation therapy and chemotherapy were administered, and comorbidities. CONCLUSIONS: The VTR Pilot demonstrated the feasibility of using SEER cancer registries as honest brokers to provide tissue and clinical data for secondary use in research. Studies using this program should oversample by 45% to 50% to obtain sufficient sample size and targeted population representation and involve subspecialty matter expert pathologists for tissue selection.

Reporting tumor genomic test results to SEER registries via linkages.

BACKGROUND: Precision medicine has become a mainstay of cancer care in recent years. The National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) Program has been an authoritative source of cancer statistics and data since 1973. However, tumor genomic information has not been adequately captured in the cancer surveillance data, which impedes population-based research on molecular subtypes. To address this, the SEER Program has developed and implemented a centralized process to link SEER registries' tumor cases with genomic test results that are provided by molecular laboratories to the registries. METHODS: Data linkages were carried out following operating procedures for centralized linkages established by the SEER Program. The linkages used Match*Pro, a probabilistic linkage software, and were facilitated by the registries' trusted third party (an honest broker). The SEER registries provide to NCI limited datasets that undergo preliminary evaluation prior to their release to the research community. RESULTS: Recently conducted genomic linkages included OncotypeDX Breast Recurrence Score, OncotypeDX Breast Ductal Carcinoma in Situ, OncotypeDX Genomic Prostate Score, Decipher Prostate Genomic Classifier, DecisionDX Uveal Melanoma, DecisionDX Preferentially Expressed Antigen in Melanoma, DecisionDX Melanoma, and germline tests results in Georgia and California SEER registries. CONCLUSIONS: The linkages of cancer cases from SEER registries with genomic test results obtained from molecular laboratories offer an effective approach for data collection in cancer surveillance. By providing de-identified data to the research community, the NCI's SEER Program enables scientists to investigate numerous research inquiries.

Assessing sociodemographic and regional disparities in Oncotype DX Genomic Prostate Score uptake.

BACKGROUND: The Oncotype DX Genomic Prostate Score (ODX-GPS) is a gene expression assay that predicts disease aggressiveness. The objective of this study was to identify sociodemographic and regional factors associated with ODX-GPS uptake. METHODS: Data from Surveillance Epidemiology and End Results registries on men with localized prostate cancer with a Gleason score of 3 + 3 or 3 + 4, PSA ≤20 ng/mL, and stage T1c to T2c disease from 2013 through 2017 were linked with ODX-GPS data. Census-tract level neighborhood socioeconomic status (nSES) quintiles were constructed using a composite socioeconomic score. Multivariable logistic regression was used to estimate the associations of ODX-GPS uptake with age at diagnosis, race and ethnicity, nSES, geographic region, insurance type, and marital status, accounting for National Comprehensive Cancer Network risk group, year of diagnosis, and clustering by census tract. RESULTS: Among 111,434 eligible men, 5.5% had ODX-GPS test uptake. Of these, 78.3% were non-Hispanic White, 9.6% were Black, 6.7% were Hispanic, and 3.6% were Asian American. Black men had the lowest odds of ODX-GPS uptake (odds ratio, 0.70; 95% confidence interval [CI], 0.63-0.76). Those in the highest versus lowest quintile of nSES were 1.64 times more likely (95% CI, 1.38-2.94) to have ODX-GPS uptake. The odds of ODX-GPS uptake were statistically significantly higher among men residing in the Northeast, West, and Midwest compared to the South. CONCLUSIONS: Disparities in ODX-GPS uptake by race, ethnicity, nSES, and geographical region were identified. Concerted efforts should be made to ensure that this clinical test is equitably available.

Impact of rural location on receipt of standard of care treatment and survival for locally advanced bladder cancer in Louisiana.

OBJECTIVE: We aim to determine the effect of region of residence (urban vs. rural) on the odds of receiving standard of care treatment for locally advanced BCa in Louisiana and its impact on survival outcomes. METHODS: Using the Louisiana Tumor Registry, we identified American Joint Committee on Cancer (AJCC) stage II or III, BCa diagnoses in Louisiana residents between 2010 and 2020. Treatment received was classified as standard or non-standard of care according to American Urological Association (AUA) guidelines and location of residence was determined using Rural Urban Commuting Area-Tract-level 2010 (RUCA). Multivariable logistic regression analyses and multivariate cox proportional hazard analyses were performed. RESULTS: Of 983 eligible patients, 85.6% (841/983) lived in urban areas. Overall, only 37.5% received standard-of-care (SOC) for the definitive management of locally advanced bladder cancer. Individuals living in rural areas (OR 0.53, 95% CI: 0.31-0.91, p = 0.02) were less likely to receive standard of care treatment. Both rural residence and receipt of non-standard of care therapy were associated with an increased risk of bladder cancer-specific (adj HR 1.53, 95% CI: 1.09-2.14, p = 0.01 and adj HR: 1.85, 95% CI: 1.43-2.39, <0.0001) and overall mortality (adj HR: 1.28, 95% CI: 1.01-1.61, p = 0.04 and adj HR: 1.73 95% CI: 1.44-2.07, p < 0.0001). CONCLUSIONS: Most patients with locally advanced bladder cancer in Louisiana do not receive SOC therapy. Individuals living in rural locations are more likely to receive non-standard of care therapy than individuals in urban areas. Nonstandard of care treatment and rural residence are both associated with worse survival outcomes for Louisiana residents with locally advanced bladder cancer.

Survival paradox between stage IIB/C and stage IIIA colon cancer: is it time to revise the American Joint Committee on Cancer TNM system?

INTRODUCTION: A survival paradox between T4N0 (Stage IIB/IIC) and Stage IIIA colon cancer exists, even after adjusting for adequate lymph node (LN) retrieval and receipt of adjuvant chemotherapy (C). We conducted a large hospital-based study to re-evaluate this survival paradox based on the newest 8th edition staging system. METHODS: The National Cancer Data Base was queried to evaluate 35,606 patients diagnosed with Stage IIB, IIC, and IIIA colon cancer between 2010 and 2017. The Kaplan-Meier method and log-rank test were used to compare unadjusted overall survival (OS). Multivariable Cox proportional hazards model was used to determine the association of stage with hazard ratios adjusted for relevant demographic and clinical variables including ≥ 12 LNs retrieved and receipt of adjuvant chemotherapy. P value < 0.05 was considered statistically significant. RESULTS: The 5-year OS for optimally treated stage IIIA colon cancer (receipt of C) was 84.3%, which was significantly higher than stage IIB/C (≥ 12 LNs retrieved + C) (72.8%; P < 0.0001). Stage was an independent predictor of OS. Among optimally treated Stage IIIA patients, T1N1 had the best survival (90.6%) while stage T4bN0 (stage IIC) had the worst (70.9%) (P < 0.0001). Compared to stage IIB, stage IIC had a 17% increased risk of overall death while stage IIIA had a 21% reduction in death (P < 0.0001). CONCLUSION: Stage IIB/C and Stage IIIA survival paradox persists even after accounting for receipt of adjuvant chemotherapy and adequate lymph node retrieval. Future iteration of the TNM system should take this paradox into consideration.

Demographic and Clinical Factors Associated With Health-Related Quality-of-Life Profiles Among Prostate Cancer Survivors.

PURPOSE: Our purpose was to describe the prevalence and predictors of symptom and function clusters related to physical, emotional, and social components of general health-related quality of life (HRQOL) in a population-based sample of prostate cancer (PCa) survivors. METHODS: Participants (N = 1,162) completed a baseline survey at a median of 9 months after diagnosis to ascertain the co-occurrence of eight symptom and functional domains that are common across all cancers and not treatment-specific. We used latent profile analysis (LPA) to identify subgroup profiles of survivors with low, moderate, or high HRQOL levels. Multinomial logistic regression models were used to identify clinical and sociodemographic factors associated with survivors' membership in the low versus moderate or high HRQOL profile. RESULTS: The LPA identified 16% of survivors who were categorized in the low HRQOL profile at baseline, indicative of the highest symptom burden and lowest functioning. Factors related to survivors' membership in the low versus higher HRQOL profile groups included less than age 65 years at diagnosis, identifying as non-Hispanic Black race, not working, being a former versus never smoker, systemic therapy, less companionship, more comorbidities, lower health care financial well-being, or less spirituality. Several factors remained associated with remaining in the low versus higher HRQOL profiles on the follow-up survey (n = 699), including younger age, Black race, comorbidity, and lower financial and spiritual well-being. CONCLUSION: About one of six PCa survivors experienced elevated physical and psychosocial symptoms that were independent of local curative therapy, but with younger age, race, comorbidity, and lower financial and spiritual well-being as stable risk factors for poor HRQOL over time.

Medicaid Expansion Increases Access for Rural and Impoverished Patients with Pancreatic Ductal Adenocarcinoma in Southern States.

INTRODUCTION: Medicaid expansion (ME) impacted patients when assessed at a national level. However, of the 32 states in which Medicaid expansion occurred, only 3 were Southern states. Whether results apply to Southern states that share similar geopolitical perspectives remains elusive. We aimed to assess the impact of ME on pancreatic ductal adenocarcinoma (PDAC) treatment in eight Southern states in the USA. PATIENTS AND METHODS: We identified uninsured or Medicaid patients (age 40-64 years) diagnosed with PDAC between 2011 and 2018 in Southern states from the North American Association of Central Cancer Registries-Cancer in North America (NAACCR-CiNA) research dataset. Medicaid-expanded states (MES; Louisiana, Kentucky, and Arkansas) were compared with non-MES (NMES; Tennessee, Alabama, Mississippi, Texas, and Oklahoma) using multivariate logistic regression. P < 0.05 was considered statistically significant. RESULTS: Among 3036 patients, MES significantly increased odds of Medicaid insurance by 36%, and increased proportions of insured Black patients by 3.7%, rural patients by 3.8%, and impoverished patients by 18.4%. After adjusting for age, race, rural-urban status, poverty status, and summary stage, the odds of receiving radiation therapy decreased by 26% for each year of expansion in expanded states (P = 0.01). Last, ME did not result in a significant difference between MES and NMES in diagnosing early stage disease (P = 0.98) nor in receipt of chemotherapy or surgery (P = 0.23 and P = 0.63, respectively). CONCLUSIONS: ME in Southern states increased insurance access to traditionally underserved groups. Interestingly, ME decreased the odds of receiving radiation therapy yearly and had no significant impact on receipt of chemotherapy or surgery.

Path-BigBird: An AI-Driven Transformer Approach to Classification of Cancer Pathology Reports.

PURPOSE: Surgical pathology reports are critical for cancer diagnosis and management. To accurately extract information about tumor characteristics from pathology reports in near real time, we explore the impact of using domain-specific transformer models that understand cancer pathology reports. METHODS: We built a pathology transformer model, Path-BigBird, by using 2.7 million pathology reports from six SEER cancer registries. We then compare different variations of Path-BigBird with two less computationally intensive methods: Hierarchical Self-Attention Network (HiSAN) classification model and an off-the-shelf clinical transformer model (Clinical BigBird). We use five pathology information extraction tasks for evaluation: site, subsite, laterality, histology, and behavior. Model performance is evaluated by using macro and micro F1 scores. RESULTS: We found that Path-BigBird and Clinical BigBird outperformed the HiSAN in all tasks. Clinical BigBird performed better on the site and laterality tasks. Versions of the Path-BigBird model performed best on the two most difficult tasks: subsite (micro F1 score of 72.53, macro F1 score of 35.76) and histology (micro F1 score of 80.96, macro F1 score of 37.94). The largest performance gains over the HiSAN model were for histology, for which a Path-BigBird model increased the micro F1 score by 1.44 points and the macro F1 score by 3.55 points. Overall, the results suggest that a Path-BigBird model with a vocabulary derived from well-curated and deidentified data is the best-performing model. CONCLUSION: The Path-BigBird pathology transformer model improves automated information extraction from pathology reports. Although Path-BigBird outperforms Clinical BigBird and HiSAN, these less computationally expensive models still have utility when resources are constrained.

Association of Major Adverse Financial Events and Later-Stage Cancer Diagnosis in the United States.

PURPOSE: This study assessed the prevalence of specific major adverse financial events (AFEs)-bankruptcies, liens, and evictions-before a cancer diagnosis and their association with later-stage cancer at diagnosis. METHODS: Patients age 20-69 years diagnosed with cancer during 2014-2015 were identified from the Seattle, Louisiana, and Georgia SEER population-based cancer registries. Registry data were linked with LexisNexis consumer data to identify patients with a history of court-documented AFEs before cancer diagnosis. The association of AFEs and later-stage cancer diagnoses (stages III/IV) was assessed using separate sex-specific multivariable logistic regression. RESULTS: Among 101,649 patients with cancer linked to LexisNexis data, 36,791 (36.2%) had a major AFE reported before diagnosis. The mean and median timing of the AFE closest to diagnosis were 93 and 77 months, respectively. AFEs were most common among non-Hispanic Black, unmarried, and low-income patients. Individuals with previous AFEs were more likely to be diagnosed with later-stage cancer than individuals with no AFE (males-odds ratio [OR], 1.09 [95% CI, 1.03 to 1.14]; P < .001; females-OR, 1.18 [95% CI, 1.13 to 1.24]; P < .0001) after adjusting for age, race, marital status, income, registry, and cancer type. Associations between AFEs prediagnosis and later-stage disease did not vary by AFE timing. CONCLUSION: One third of newly diagnosed patients with cancer had a major AFE before their diagnosis. Patients with AFEs were more likely to have later-stage diagnosis, even accounting for traditional measures of socioeconomic status that influence the stage at diagnosis. The prevalence of prediagnosis AFEs underscores financial vulnerability of patients with cancer before their diagnosis, before any subsequent financial burden associated with cancer treatment.

Adherence Rate to Alliance for Clinical Trials in Oncology Z0011 Trial Based on Breast Cancer Subtype.

BACKGROUND: The American College of Surgeons Oncology Group Z0011 (ACOSOG Z0011 or Z11) trial demonstrated no survival advantage with completion axillary lymph node dissection (ALND) for patients with T1-2 breast cancer, 1 to 2 positive SLNs who received adjuvant chemoradiation therapy. More than 70% of the cohort had estrogen receptor (ER)+ tumors. There is paucity of data on the adherence rate to Z11, as well as a dearth of data on the applicability of Z11 for the different subtypes. We conducted a large hospital-based study to evaluate the adherence rate to Z11 based on subtypes. STUDY DESIGN: The National Cancer Database was queried to evaluate 33,859 patients diagnosed with T1-2, N1, and M0 breast cancer treated with lumpectomy with negative margins, and adjuvant chemoradiation therapy between 2012 and 2018. Patients were classified into 3 groups: (1) ER+/HER2-, (2) ER-/HER2-, and (3) HER2+ regardless of ER status. The revised Scope of the Regional Lymph Node Surgery 2012 was used to classify patients into those who underwent an SLN or ALND. Differences in use of ALND by subtypes were compared. The Kaplan-Meier method and log-rank test were used to compare overall survival (OS). A p value of <0.05 was considered statistically significant. RESULTS: For ER+/human epidermal growth factor receptor 2 (HER2)-, ER-/HER2-, and HER2+ tumors, the rate of ALND was 43.6%, 50.2%, and 47.8%, respectively. The 5-year OS for SLN and ALND for the entire cohort was 94.0% and 93.1% (p = 0.0004); for ER+/HER2-, it was 95.4% and 94.7% (p = 0.04); for ER-/HER2-, it was 84.1% and 84.3% (p = 0.41); for HER2+, it was 94.2% and 93.2% (p = 0.20). Multivariable cox proportional hazard regression analysis demonstrated no significant survival differences between SLN and ALND (p = 0.776). CONCLUSIONS: Z11 is applicable for women with early N1 disease, regardless of subtypes. ALND did not confer a survival advantage over SLN. Despite this, up to 50% of patients who fit Z11 criteria continue to undergo ALND.

Functional Outcomes After Localized Prostate Cancer Treatment.

Importance: Adverse outcomes associated with treatments for localized prostate cancer remain unclear. Objective: To compare rates of adverse functional outcomes between specific treatments for localized prostate cancer. Design, Setting, and Participants: An observational cohort study using data from 5 US Surveillance, Epidemiology, and End Results Program registries. Participants were treated for localized prostate cancer between 2011 and 2012. At baseline, 1877 had favorable-prognosis prostate cancer (defined as cT1-cT2bN0M0, prostate-specific antigen level <20 ng/mL, and grade group 1-2) and 568 had unfavorable-prognosis prostate cancer (defined as cT2cN0M0, prostate-specific antigen level of 20-50 ng/mL, or grade group 3-5). Follow-up data were collected by questionnaire through February 1, 2022. Exposures: Radical prostatectomy (n = 1043), external beam radiotherapy (n = 359), brachytherapy (n = 96), or active surveillance (n = 379) for favorable-prognosis disease and radical prostatectomy (n = 362) or external beam radiotherapy with androgen deprivation therapy (n = 206) for unfavorable-prognosis disease. Main Outcomes and Measures: Outcomes were patient-reported sexual, urinary, bowel, and hormone function measured using the 26-item Expanded Prostate Cancer Index Composite (range, 0-100; 100 = best). Associations of specific therapies with each outcome were estimated and compared at 10 years after treatment, adjusting for corresponding baseline scores, and patient and tumor characteristics. Minimum clinically important differences were 10 to 12 for sexual function, 6 to 9 for urinary incontinence, 5 to 7 for urinary irritation, and 4 to 6 for bowel and hormone function. Results: A total of 2445 patients with localized prostate cancer (median age, 64 years; 14% Black, 8% Hispanic) were included and followed up for a median of 9.5 years. Among 1877 patients with favorable prognosis, radical prostatectomy was associated with worse urinary incontinence (adjusted mean difference, -12.1 [95% CI, -16.2 to -8.0]), but not worse sexual function (adjusted mean difference, -7.2 [95% CI, -12.3 to -2.0]), compared with active surveillance. Among 568 patients with unfavorable prognosis, radical prostatectomy was associated with worse urinary incontinence (adjusted mean difference, -26.6 [95% CI, -35.0 to -18.2]), but not worse sexual function (adjusted mean difference, -1.4 [95% CI, -11.1 to 8.3), compared with external beam radiotherapy with androgen deprivation therapy. Among patients with unfavorable prognosis, external beam radiotherapy with androgen deprivation therapy was associated with worse bowel (adjusted mean difference, -4.9 [95% CI, -9.2 to -0.7]) and hormone (adjusted mean difference, -4.9 [95% CI, -9.5 to -0.3]) function compared with radical prostatectomy. Conclusions and Relevance: Among patients treated for localized prostate cancer, radical prostatectomy was associated with worse urinary incontinence but not worse sexual function at 10-year follow-up compared with radiotherapy or surveillance among people with more favorable prognosis and compared with radiotherapy for those with unfavorable prognosis. Among men with unfavorable-prognosis disease, external beam radiotherapy with androgen deprivation therapy was associated with worse bowel and hormone function at 10-year follow-up compared with radical prostatectomy.

Deep learning uncertainty quantification for clinical text classification.

INTRODUCTION: Machine learning algorithms are expected to work side-by-side with humans in decision-making pipelines. Thus, the ability of classifiers to make reliable decisions is of paramount importance. Deep neural networks (DNNs) represent the state-of-the-art models to address real-world classification. Although the strength of activation in DNNs is often correlated with the network's confidence, in-depth analyses are needed to establish whether they are well calibrated. METHOD: In this paper, we demonstrate the use of DNN-based classification tools to benefit cancer registries by automating information extraction of disease at diagnosis and at surgery from electronic text pathology reports from the US National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) population-based cancer registries. In particular, we introduce multiple methods for selective classification to achieve a target level of accuracy on multiple classification tasks while minimizing the rejection amount-that is, the number of electronic pathology reports for which the model's predictions are unreliable. We evaluate the proposed methods by comparing our approach with the current in-house deep learning-based abstaining classifier. RESULTS: Overall, all the proposed selective classification methods effectively allow for achieving the targeted level of accuracy or higher in a trade-off analysis aimed to minimize the rejection rate. On in-distribution validation and holdout test data, with all the proposed methods, we achieve on all tasks the required target level of accuracy with a lower rejection rate than the deep abstaining classifier (DAC). Interpreting the results for the out-of-distribution test data is more complex; nevertheless, in this case as well, the rejection rate from the best among the proposed methods achieving 97% accuracy or higher is lower than the rejection rate based on the DAC. CONCLUSIONS: We show that although both approaches can flag those samples that should be manually reviewed and labeled by human annotators, the newly proposed methods retain a larger fraction and do so without retraining-thus offering a reduced computational cost compared with the in-house deep learning-based abstaining classifier.

DeepPhe-CR: Natural Language Processing Software Services for Cancer Registrar Case Abstraction.

PURPOSE: Manual extraction of case details from patient records for cancer surveillance is a resource-intensive task. Natural Language Processing (NLP) techniques have been proposed for automating the identification of key details in clinical notes. Our goal was to develop NLP application programming interfaces (APIs) for integration into cancer registry data abstraction tools in a computer-assisted abstraction setting. METHODS: We used cancer registry manual abstraction processes to guide the design of DeepPhe-CR, a web-based NLP service API. The coding of key variables was performed through NLP methods validated using established workflows. A container-based implementation of the NLP methods and the supporting infrastructure was developed. Existing registry data abstraction software was modified to include results from DeepPhe-CR. An initial usability study with data registrars provided early validation of the feasibility of the DeepPhe-CR tools. RESULTS: API calls support submission of single documents and summarization of cases across one or more documents. The container-based implementation uses a REST router to handle requests and support a graph database for storing results. NLP modules extract topography, histology, behavior, laterality, and grade at 0.79-1.00 F1 across multiple cancer types (breast, prostate, lung, colorectal, ovary, and pediatric brain) from data of two population-based cancer registries. Usability study participants were able to use the tool effectively and expressed interest in the tool. CONCLUSION: The DeepPhe-CR system provides an architecture for building cancer-specific NLP tools directly into registrar workflows in a computer-assisted abstraction setting. Improved user interactions in client tools may be needed to realize the potential of these approaches.

Rural versus urban commuting patients with stage III colon cancer: is there a difference in treatment and outcome?

BACKGROUND: To evaluate if there are differences in outcomes for patients with stage III colon cancer in those from urban vs. rural commuting areas. METHODS: Data were evaluated on patients diagnosed with stage III colon cancer between 2012 and2018 from the Louisiana Tumor Registry. Patients were classified into rural and urban groups. Data on overall survival, time from diagnosis to surgery and time from surgery to chemotherapy, and sociodemographic factors (including race, age, and poverty level) were recorded. RESULTS: Of 2652 patients identified, 2159 were urban (81.4%) and 493 rural (18.6%). No age difference between rural and urban patients (p = 0.56). Stage IIIB accounted for 66.7%, followed by IIIC (21.6%) and IIIA (11%), with a significant difference between rural and urban patients based on stage (p = 0.02). There was no difference in the extent of surgery (p = 0.34) or tumor size (p = 0.72) between urban and rural settings. No difference in undergoing chemotherapy (p = 0.12). There was a statistically significant difference in receiving timely treatment for hospital volume (p < 0.0001) and poverty level (p < 0.0001), but no difference in time from diagnosis to surgery (p = 0.48), and time from surgery to chemotherapy (p = 0.27). Non-Hispanic Blacks were less likely to receive timely treatment when compared with non-Hispanic Whites for both surgery and adjuvant chemotherapy, (aHR 0.91, 95% CI 0.83-0.99) and (aHR 0.86, 95% CI 0.77-0.97), respectively. There was no difference in Kaplan-Meier overall survival curves comparing rural vs. urban patients (p = 0.77). CONCLUSIONS: There was no statistical difference in overall survival, time to surgery, and time to adjuvant chemotherapy between rural and urban patients with Stage III colon cancer.

Effect of chronic disease on racial difference in COVID-19-associated hospitalization among cancer patients.

BACKGROUND: Research indicates that Black cancer patients have higher rates of COVID-19 hospitalization than their White counterparts. However, the extent to which chronic diseases contribute to racial disparities remains uncertain. We aimed to quantify the effect of chronic diseases on racial disparity in COVID-19-associated hospitalization among cancer patients. METHODS: We linked Louisiana Tumor Registry's data with statewide COVID-19 data and hospital in-patient discharge data to identify patients diagnosed with cancer in 2015-2019 who tested positive for COVID-19 in 2020 and those with COVID-19-associated hospitalization. Multivariable logistic regression and mediation methods based on linear structural equations were employed to assess the effects of the number of chronic diseases (0, 1-2, ≥3) and individual chronic diseases. RESULTS: Of 6381 cancer patients who tested positive for COVID-19, 31.6% were non-Hispanic Black cancer patients. Compared with non-Hispanic White cancer patients, non-Hispanic Black cancer patients had a higher prevalence of chronic diseases (79.5% vs 66.0%) and higher COVID-19-associated hospitalization (27.2% vs 17.2%). The odds of COVID-19-associated hospitalization were 80% higher for non-Hispanic Black cancer patients than non-Hispanic White cancer patients (odds ratio = 1.80, 95% confidence interval = 1.59 to 2.04). After adjusting for age, sex, insurance, poverty, obesity, and cancer type, number of chronic diseases explained 37.8% of the racial disparity in COVID-19-associated hospitalization, and hypertension, diabetes, and chronic renal disease were the top 3 chronic diseases explaining 9.6%, 8.9%, and 7.3% of the racial disparity, respectively. CONCLUSION: Chronic diseases played a substantial role in the racial disparity in COVID-19-associated hospitalization among cancer patients, especially hypertension, diabetes, and renal disease. Understanding and addressing the root causes are crucial for targeted interventions, policies, and health-care strategies to reduce racial disparity.