SEER Data Research Articles

Outcomes of Acute Myeloid Leukemia Patients Aged 60 Years and Above: An Analysis of SEER Data

Outcomes of Acute Myeloid Leukemia Patients Aged 60 Years and Above: An Analysis of SEER Data

Evaluating machine learning model bias and racial disparities in non-small cell lung cancer using SEER registry data.

Despite decades of pursuing health equity, racial and ethnic disparities persist in healthcare in America. For cancer specifically, one of the leading observed disparities is worse mortality among non-Hispanic Black patients compared to non-Hispanic White patients across the cancer care continuum. These real-world disparities are reflected in the data used to inform the decisions made to alleviate such inequities. Failing to account for inherently biased data underlying these observations could intensify racial cancer disparities and lead to misguided efforts that fail to appropriately address the real causes of health inequity. Estimate the racial/ethnic bias of machine learning models in predicting two-year survival and surgery treatment recommendation for non-small cell lung cancer (NSCLC) patients. A Cox survival model, and a LOGIT model as well as three other machine learning models for predicting surgery recommendation were trained using SEER data from NSCLC patients diagnosed from 2000-2018. Models were trained with a 70/30 train/test split (both including and excluding race/ethnicity) and evaluated using performance and fairness metrics. The effects of oversampling the training data were also evaluated. The survival models show disparate impact towards non-Hispanic Black patients regardless of whether race/ethnicity is used as a predictor. The models including race/ethnicity amplified the disparities observed in the data. The exclusion of race/ethnicity as a predictor in the survival and surgery recommendation models improved fairness metrics without degrading model performance. Stratified oversampling strategies reduced disparate impact while reducing the accuracy of the model. NSCLC disparities are complex and multifaceted. Yet, even when accounting for age and stage at diagnosis, non-Hispanic Black patients with NSCLC are less often recommended to have surgery than non-Hispanic White patients. Machine learning models amplified the racial/ethnic disparities across the cancer care continuum (which are reflected in the data used to make model decisions). Excluding race/ethnicity lowered the bias of the models but did not affect disparate impact. Developing analytical strategies to improve fairness would in turn improve the utility of machine learning approaches analyzing population-based cancer data.

Differential Analysis of Surgical Treatment Modalities in T2N0M0 Bladder Cancer Patients: A Novel Propensity Score-Based Cohort Study

ObjectiveThis study explored prognostic differences between radical cystectomy (RC), tri-modality treatment (TURBt combined with radiotherapy and chemotherapy, TMT) and electrocautery (EC) and partial cystectomy (PC) for T2N0M0 MIBC. Materials and MethodsIn the SEER database (2004-2015, 2018-2020), we identified patients with T2N0M0 bladder cancer receiving RC, TMT, EC or PC treatment. To minimize confounding factors, 1:1 propensity score matching (PSM) with a 0.1 caliper was conducted. Kaplan-Meier analysis and log-rank tests were used to compare survival rates. Cox analysis identified independent prognostic factors, tumor size, and age, stratified treatment groups, and further Kaplan-Meier analysis of CSS was performed in matched cohorts. ResultThis study included 6526 patients with T2N0M0 MIBC. Among them, 348 (5.33%) were in the PC group, 309 (4.73%) were in the EC group, 1833 (28.09%) were in the TMT group, and 4036 (61.84%) were in the RC group. After 1:1 propensity score matching, RC showed improved CSS (HR=0.67, 95%CI 0.47-0.95, and PC also benefited (HR=0.97, 95%CI 0.69-1.36) compared to EC. While TMT showed a worse end (HR=1.41, 95%Cl 1.03-1.92) compared to EC. Cox analysis was used to stratify tumor size and age for subgroup analysis. Results for tumor size subgroups were aligned with PSM findings. In the age-stratified subgroups, patients aged <67 years, both RC (HR=0.54, P=0.107) and TMT(HR=0.91, P=0.785) showed better prognoses compared to EC treatment, while PC treatment showed worse prognoses compared to EC treatment (HR=1.23, P=0.542).; for 68-77 years, RC (0.64, P=0.1436) and PC (HR=0.46, P=0.0283)had advantages, and PC is more recommended. For >78 years, RC had superior CSS over EC and PC, whereas TMT had the poorest prognosis. ConclusionIn clinical T2N0M0 MIBC, overall, RC outperformed focal-tumor therapy and PC, irrespective of tumor size. However, considering age, we recommend PC treatment for patients aged 68-77 and EC for those aged >78 years. Micro-AbstractThis study compared radical cystectomy (RC), focal-tumor therapy combined with radiotherapy and chemotherapy (TMT and EC), and partial cystectomy (PC) for non-metastatic muscle-invasive bladder cancer. Using SEER data and a series of statistical analyses, overall, RC outperformed focal-tumor therapy and PC, irrespective of tumor size. However, considering age, we recommend PC treatment for patients aged 68-77 and EC for those aged >78 years. In contrast, for patients younger than 67 years of age, we recommend RC for those who can tolerate surgical treatment and TMT for those who can't.

Comprehensive evaluation of clinical outcomes in hepatic epithelioid hemangioendothelioma subsets: insights from SEER Database and departmental cohort analysis.

Epithelioid hemangioendothelioma (EHE), is an uncommon, intermediate-grade malignant vascular tumor that can manifest in diverse organs, including the liver, lungs, and bones. Given its unique malignancy profile and rarity, there lacks a consensus on a standardized treatment protocol for EHE, particularly for hepatic epithelioid hemangioendothelioma (HEHE). This study aims to elucidate factors influencing the clinical prognosis of EHE by analyzing data from the SEER database, complemented with insights from a departmental cohort of 9 HEHE cases. Through this, we hope to shed light on potential clinical outcomes and therapeutic strategies for HEHE. Using SEER data from 22 registries, we analyzed 313 liver cancer patients with ICD-O-3 9130 and 9133 histology. Twelve variables were examined using Cox regression and mlr3 machine learning. Significant variables were identified and compared. Clinical data, imaging characteristics, and treatment methods of nine patients from our cohort were also presented. In univariate and multivariate Cox regression analyses, Age, Sex, Year of diagnosis, Surgery of primary site, Chemotherapy, and Median household income were closely related to survival outcomes. Among the ten survival-related machine learning models, CoxPH, Flexible, Mboost, and Gamboost stood out based on Area Under the Curve(AUC), Decision Curve Analysis(DCA), and Calibration Curve Metrics. In the feature importance analysis of these four selected models, Age and Surgery of primary site were consistently identified as the most critical factors influencing prognosis. Additionally, the clinical data of nine patients from our cohort not only demonstrated unique imaging characteristics of HEHE but also underscored the importance of surgical intervention. For patients with resectable HEHE, surgical treatment is currently a highly important therapeutic approach.

Diversity in U.S. Clinical Trials for Head and Neck Cancer: Are We Improving?

This study assesses whether national initiatives undertaken to improve diversity in clinical trial enrollment have been successful within head and neck cancer (HNC) trials. A retrospective analysis was conducted of HNC trials published on clinicaltrials.gov with start dates between 2000 and 2023. Demographic data for 8998 HNC trial enrollees was abstracted and analyzed to investigate potential demographic shifts. In the past 20 years, the percentage of White patients increased 6.1%, Asian patient population decreased 3.1%, and Black patient population increased 0.8%. Compared with previously published SEER data, HNC trials have significantly more White patients, fewer Black patients, and fewer Asian/Native-Hawaiian patients than HNC patients at large. Despite efforts to increase diversity in HNC clinical trials in the United States, diversity has significantly decreased in the past 10 years. As current approaches are failing to show improvement, novel approaches to improving representation in clinical trials are necessitated.

Disparities in Stage at Diagnosis among Hispanic Patients with Gastric Cancer in the United States.

Racial disparities in gastric cancer outcomes, including stage at diagnosis and overall survival, continue to affect Hispanic and non-Hispanic populations. This study aims to evaluate these disparities across different racial groups. We conducted a retrospective cohort study using SEER data from 2018 to 2021, including 18,984 patients diagnosed with gastric cancer. Patients were selected based on ICD-O-3 codes for "stomach" with malignant behavior. Using ordered logistic regression, the association between race and stage at diagnosis was analyzed, while Cox proportional hazards models were used to assess OS and CSS. Hispanic patients were significantly more likely to be diagnosed at a later stage compared to non-Hispanic patients (OR: 1.19; 95% CI: 1.10-1.28). Both Hispanic and Black patients had worse OS compared to Non-Hispanic Whites (HR 1.10 CI 1.03-1.17, p = 0.003 and HR 1.13 CI 1.04-1.22, p = 0.002, respectively) as well as CSS. Hispanic patients are more likely to be diagnosed with advanced-stage gastric cancer and have poorer survival outcomes compared to non-Hispanic Whites. These disparities may be linked to differences in healthcare access, insurance, language barriers, and preventive care utilization.

Complex Relationships between Diagnostics and Survival in Chronic Lymphocytic Leukemia in Denmark, Finland, Norway, and Sweden.

Chronic lymphocytic leukemia (CLL) is a common hematological malignancy with highly variable clinical presentation. Many patients never require any treatment but for the others, chemotherapy, immunochemotherapy, and newer targeted therapies have changed the treatment landscape. Diagnostic age influences the applied treatment, and we thus wanted to analyze age-specific survival trends through 50 years up to 2020s. We used 1- and 5-year relative survival from the NORDCAN database, with data from Denmark (DK), Finland (FI), Norway (NO), and Sweden (SE). Because of the variable presentation of CLL, we also considered incidence and mortality trends. For comparison, US SEER data were used. The large age-specific survival differences in 1972-76 almost disappeared by 2017-21. While 5-year survival in younger patients exceeded 90%, for those diagnosed at age 80-89 years, survival reached 90% in DK and SE women, 80% in NO and SE men, but only 50% in FI. DK 5-year overall survival for men was 92.4%, and for women, it was 96.3%. These survival figures were higher than age-group-specific US survival data. The DK data are probably global top figures for national survival which could be achieved by boosting survival even among the oldest patients. The qualification to these figures and international comparisons is that survival needs to be considered in terms of incidence, which is high in DK and NO. Low survival of the FI 80-89-year-old patients, even in the first year after diagnosis, may suggest delayed diagnosis, which should call for a closer national scrutiny.

54853 Trends in Incidence and Mortality of Pleomorphic Dermal Sarcoma: A Retrospective Cohort Study Using SEER Data

54853 Trends in Incidence and Mortality of Pleomorphic Dermal Sarcoma: A Retrospective Cohort Study Using SEER Data

Pediatric mandibular malignancies: a comprehensive analysis of SEER data

Mandibular malignancies are rare in the pediatric population and subsequently not well characterized. SEER 18 registry data was collected, applying age 0–18 years and ICD-O-3 code C41.1 (‘mandible’). Univariate Cox regression analysis was conducted, and hazard ratios (HR) were calculated for overall survival (OS) and disease-specific survival (DSS) according to patient demographics, tumor characteristics, and treatment. Kaplan–Meier survival curves were generated for OS and DSS. Sixty-four patients met the inclusion criteria. The median age at diagnosis was 13.0 years, and median survival was 8.7 years. Osteosarcoma was the most common histological diagnosis (n = 22). Sex, race, age (<13 vs ≥13 years), histological type, odontogenic origin, and treatment modality were found not to be associated with OS or DSS. The SEER stage ‘distant’ was significantly associated with an elevated HR of 6.28 for DSS (P = 0.027) and 5.29 for OS (P = 0.025). Kaplan–Meier survival curves demonstrated significantly lower 5-year DSS (P < 0.001) and OS (P < 0.001) for SEER ‘distant’ stage. This study includes the analysis of a large number of pediatric mandibular malignancies when compared to previous studies. ‘Distant’ stage was associated with decreased survival. Early clinical suspicion and diagnosis are paramount for improved survival.

Lymph node involvement is associated with overall survival for elderly patients with non-metastatic gallbladder adenocarcinoma.

Lymph node involvement is recognized as a prognostic factor for patients with gallbladder cancer. However, the N stage varied from different editions of the American Joint Committee on Cancer (AJCC) TNM Classification. Our objective was to investigate the impact of lymph node involvement on overall survival in elderly patients with non-metastatic gallbladder adenocarcinoma. Patients older than 65 years with non-metastatic gallbladder adenocarcinoma were identified from the SEER data. We used Cox proportional hazard regression analysis to select the independent risk factor. A nomogram was built to identify the 1-, 3-, and 5-years' prognostic impact. Univariate and multivariate models were used to examine the correlation of overall survival (OS) with the number of metastatic nodes. A total of 1,654 patients (706 with and 948 without nodal involvement) were included. Cox proportional hazard regression analyses found that age, gender, tumor size, lymph node involvement, and surgical options were risk factors for the prognosis and were integrated into the nomogram. After adjustment, OS was compromised for patients who receive surgery with nodal involvement [hazard ratio (HR), 2.238; P < 0.01]. Furthermore, after adjustment the presence of more than two metastatic lymph nodes was associated with decreased OS (HR, 3.305; P < 0.01). Our results suggest that lymph node involvement is associated with compromised survival in elderly patients. A nomogram was developed to predict the prognosis of gallbladder cancer. A change point of more than two metastatic lymph nodes seems to carry prognostic significance, calling for closer monitoring of elderly patients with gallbladder cancer with involvement of increased number of lymph nodes.

Prospective, longitudinal study of cancer predictors and rates in a New York City cohort of 598 patients with acromegaly.

Long-term GH/IGF-1 excess could increase risk of cancer in acromegaly, but individual levels of these hormones do not relate to this risk. Therefore, we newly investigated longitudinally-measured IGF-1 levels as a potential predictor of cancer in a large NYC acromegaly cohort. We conducted a prospective, longitudinal study of 598 acromegaly (309 men, 289 women) and 292 clinically nonfunctioning pituitary adenoma (CNFPA)(140 women, 152 men) patients from the same underlying population. GH and IGF-1 levels were measured longitudinally and outcomes were observed during long-term follow-up. Cumulative exposure to IGF-1 excess was tested as a predictor of cancer. We compared cancer prevalence in acromegaly and CNFPA cohorts and incidence in each to that expected from SEER data. Cancer prevalence by last follow up was 22.6% in acromegaly and 12.7% in CNFPAs (OR = 1.99 (95% CI, 1.34, 2.97)(P=0.0005). Overall SIR for cancer was 1.78 (1.51, 1.81) in the acromegaly and 1.26 (0.89, 1.70) in the CNFPA cohorts. Cumulative exposure to IGF-1 excess, OR=1.278 (1.060, 1.541)(P = 0.01), years from acromegaly diagnosis to cancer or last follow up, OR= 1.03 (1.004, 1.057)(P=0.024), and age at follow up, OR =1.064 (1.047, 1.082)(P<0.001), were predictors of cancer. Cancer risk is increased in acromegaly, but not in CNFPA patients. Cumulative exposure to IGF-1 excess is a predictor of cancer in acromegaly. Our data suggest that cancer risk in acromegaly relates to the degree and duration of IGF-1 excess and that full appreciation of this risk requires long-term follow up.

Right Hemicolectomy and Appendicectomy as Treatments for Goblet Cell Adenocarcinoma: A Comparative Analysis of Two Large National Databases.

Right hemicolectomy (RHC) remains the treatment standard for goblet cell adenocarcinoma (GCA), despite limited evidence supporting survival benefit. This study aims to explore factors influencing surgical management and survival outcomes among patients treated with RHC or appendicectomy using NCRAS (UK) and SEER (USA) data. A retrospective analysis was conducted using 998 (NCRAS) and 1703 (SEER) cases. Factors influencing procedure type were explored using logistic regression analyses. Overall survival (OS) probabilities and Kaplan-Meier (KM) plots were generated using KM analysis and the log-rank test compared survival between groups. Cox regression analyses were performed to assess hazard ratios. The NCRAS analysis revealed that age and regional stage disease were determinants of undergoing RHC, with all age groups showing similar odds of receiving RHC, excluding the 75+ age group. The SEER analysis revealed tumour size > 2 cm, and receipt of chemotherapy were determinants of undergoing RHC, unlike the distant stage, which was associated with appendicectomy. Surgery type was not a significant predictor of OS in both analyses. In NCRAS, age and stage were significant predictors of OS. In SEER, age, stage, and Black race were significant predictors of worse OS. The study shows variations in the surgical management of GCA, with limited evidence to support a widespread recommendation for RHC.

Health disparities in acute myeloid leukemia: A decade-long analysis of Afro-Caribbean patients in a safety net New York City hospital.

e18515 Background: Acute Myeloid Leukemia (AML) in the Afro-Caribbean population is associated with poor outcomes due to disparities in health insurance and access to healthcare resources, among other factors, when compared to survival data in the national SEER database. This study reviews clinical characteristics and treatment outcomes of a cohort of 51 Afro-Caribbean patients who were diagnosed with AML and treated in the past 10 years in a urban safety net hospital in Brooklyn, New York. Methods: This is a retrospective review of patients diagnosed and treated for AML at New York City Health and Hospitals/Kings County from 2012 to 2022. Electronic Health Data was reviewed for clinicopathologic data and treatment outcomes. Analysis was performed through descriptive statistics with specific focus on insurance status and survival outcomes. Results: Our AML patient cohort had a mean age of 66, with 84% of patients identifying as Afro-Caribbean, and 51% being female. In this cohort, the AML risk stratification distribution based on 2022 European LeukemiaNet Risk classification includes 21 (41%) patients in the adverse-risk group, 23 (45%) patients in the intermediate-risk group, and 7 (14%) patients in the favorable-risk group. Of the 36 patients with molecular analysis data, 9 (25%) harbored a mutation in FLT3-ITD. Healthcare disparities are evident in insurance coverage, as 58% of patients lacked medical insurance, 19% had Medicare/Medicaid, and 23% had private insurance. At the time of this retrospective chart review, 35% of patients lacked any follow-up information, including appointments or status updates regarding their condition, vital status, or engagement with healthcare, possibly due to missed appointments, transfers, mortality, or other reasons. Of the 27 patients who were deemed transplant eligible, three (11%) successfully completed allogeneic bone marrow transplant. Reasons for non-transplantation include lack of a compatible donor, inadequate social support, insurance-related challenges, or patient refusal. Patients who completed induction and consolidation treatment with documented follow-up had a 23% survival rate at the five-year mark, compared to 31.7% overall survival rate reported in SEER data from 2013 to 2019. Conclusions: Our study highlights persistent healthcare disparities in Afro-Caribbean AML patients leading to poor survival outcomes. Despite treatment advancements, challenges related to insurance coverage, social support, follow-up care, and access to restricted, advanced care remain. Addressing these disparities calls for targeted interventions to enhance access and support systems for this vulnerable population.

Cancer statistics: The United States vs. worldwide.

e23276 Background: Cancer is now a pandemic worldwide. Although the mortality from cancer is downtrending due to the advent of newer treatment modalities, the global burden of cancer is exponentially increasing due to early detection.Understanding the current cancer demographics can aid in developing strategies to curb the morbidity and mortality from these malignancies. Methods: The data was obtained using 2 databases: GLOBOCAN for global trends, &amp; Surveillance, Epidemiology and End Result database Research Plus Data, 17 Registries, Nov 2022 Sub (2000-2020) for the data in the United States (US). We further stratified SEER data based on incidence, gender &amp; median months of survival (mOS), using the Log-rank test (GraphPad Prism). Results: As per the latest GLOBOCAN analysis, 18,094,716 new cancers (51.63% males,48.36% females) &amp; 9,894,402 deaths (55.49% males, 44.51% females) from cancer occurred, worldwide. The 5 most commonly diagnosed cancers globally were, breast (12.5%), lung (12.2%), colorectal cancer (CRC) (10.7%), prostate (7.8%), &amp; stomach (6%). However, the 5 most common cancers causing mortality globally, were lung (18.2%), CRC (9.5%), liver (8.4%) stomach (7.8%), &amp; breast (6.9%). Whereas, in the US, 2,281,658 new cancers (males 53.75%, females 46.24%) were detected in 2020, &amp; 612,390 deaths (males 52.58%, females 47.41%) occurred from it. The 5 most commonly diagnosed cancers were breast (11.1%), lung (10%), prostate (9.2%), CRC (6.8%), &amp; melanoma (4.2%). Cancer-related mortality was attributed to lung (22.6%), CRC (8.9%), pancreas (7.8%), breast (7%), &amp; prostate (5.3%). SEER data analysis for cancer mortality in the US is listed below in Table 1. Conclusions: Analysis of 2 large databases show that the most common cancers around the world, including the US, are breast, lung, colon &amp; prostate cancers. Breast cancer has the highest incidence, but lung cancer has the highest mortality worldwide. In the US, Whites are more affected &amp; except breast cancer, males have a greater predeliction for the common cancers. mOS was greater for females over all. This study emphasizes the need for further analysis to introspect if modalities such as next generation sequencing-directed immunotherapy would prove beneficial in increasing overall survival. [Table: see text]

Clinical characteristics and independent prognostic factors of patients with mantle cell lymphoma over last two decades with a focus on the interaction between tumor stage and age.

e19065 Background: Mantle Cell Lymphoma (MCL) poses challenges in clinical management due to its rarity and aggressiveness. Understanding its epidemiology and outcomes is crucial. This study examines demographic and clinicopathologic profiles of 746 MCL patients in the USA from 2000-2015, focusing on the interplay between age and tumor stage. Methods: Using SEER data, this retrospective study analyzes demographic and clinical variables to assess mortality risks. Univariate and multivariate Cox regression models evaluate overall mortality (OM) and MCL-specific mortality (CSM), identifying independent prognostic factors (HR &gt;1). The study explores age-tumor stage interaction. Results: Multivariate Cox proportional hazard regression analysis revealed age as pivotal, with those ≥80 facing higher risks for overall mortality (HR=3.93) and MCL-related mortality (HR=4.02) . Advanced Ann Arbor staging (II, III, IV) predicts heightened risks for both mortality types. Widowed status significantly influences outcomes ,with widowed individuals facing notably higher risks of overall mortality (HR=1.85) and MCL-related mortality (HR=1.81).Stage II and age groups 60-79 and ≥80 show substantial mortality risks, with similar trends in Stage IV combined with these age groups. Intriguingly, Stage III lacks significant age interactions, suggesting heterogeneous clinical profiles and treatment responses compared to patients in Stages II and IV. Therefore, the influence of age on mortality outcomes may be obscured by the complexities inherent in Stage III disease presentation and management. Conclusions: This study underscores age, staging, and marital status as key factors influencing MCL mortality. Widowed status emerges as notable, emphasizing social support's importance. Integrating these insights into clinical practice can enhance risk stratification and guide tailored interventions for improved outcomes. Our findings, particularly the lack of significant interactions between Stage III and age groups, highlight the intricate interplay between age and disease stage in MCL patients. This observation prompts us to advocate for further research aimed at elucidating the underlying mechanisms driving these complex interactions. Understanding the biological and clinical factors that influence the relationship between age and disease stage could provide valuable insights into MCL pathogenesis and progression. Further research in this area is warranted to inform personalized therapeutic approaches and ultimately enhance patient care and outcomes.

Clinical large language model to predict loss to follow up for oncology patients discharged to skilled nursing facilities.

e13602 Background: The prognosis of cancer patients is heavily influenced by performance status which also informs an oncologist’s decision for initiating or continuing with cancer treatment. From SEER data, there is evidence that oncology patients who are discharged to a skilled nursing facility are less likely to receive chemotherapy and radiation. Patients with advanced cancer discharged from hospital also have a readmission rate as high as 34 percent. Those that are unlikely to follow up are a vulnerable population that should be identified to discuss how to best optimize post hospitalization care. Methods: University of Florida's Integrated Data Repository (IDR) was queried for established oncology patients at University of Florida Health from 2012 – 2023. The cohort was further narrowed down to patients who had exposures to antineoplastics within 6 months and loss to follow up was defined as not having a clinic visit within 6 months of hospital discharge. The study cohort included 194 patients and 97 patients were not lost to follow up and 97 patients were lost to follow up. Structured health data including ICD-9 and 10 codes, antineoplastic drug exposures, Area Deprivation Index (ADI), discharge vitals were extracted. Dataset was split into 70% training and 30% testing. Using logistic regression, random forest and XGBoost, loss to follow up was predicted with 5 fold cross validation and hyperparameter tuning with GridSearchCV for the best ROC AUC. Results: The median Charlson score for cohort was 11, median length of stay in the hospital was 7 days, median ED visits within 6 months was 8, median LACE score of 21. 86% of the cohort had Medicare, 6% with Medicaid, 6% Blue Cross, 2% Managed Care and 0.5% Other. There were 96 females and 98 males. 66% of the cohort was Caucasian and 31% Black, 3.9% Hispanic, 1% Asian, 0.5% multiracial. XGBoost and random forest performed better than logistic regression, both achieving a AUC ROC of 0.83 vs. 0.78. XGBoost had a higher F1 score of 0.75, sensitivity of 0.71, compared to 0.68 and 0.55 respectively for both Random Forest and logistic regression. The most important features for Random Forest were ED visits within 6 months, followed by diastolic blood pressure, heart rate, temperature then systolic blood pressure on day of discharge. The most important features for XGBoost were diagnoses of pulmonary congestion, pulmonary embolism, pleural effusion and type 2 diabetes with complications. Conclusions: Our XGBoost model achieved metrics which are in line with the higher performances seen with machine learning models predicting readmissions following hospital discharge. This serves as a good baseline model to compare against new state of the art models that incorporate unstructured data such as clinical notes and radiology image reports. These models are currently under development by the same authors using the clinical large language model GatorTron.

Use of surgery for de novo metastatic breast cancer (mBC).

1590 Background: In the U.S., locoregional surgery for de novo mBC was common in the 1990s and 2000s, but subsequent prospective data did not show a survival advantage. Current guidelines recommend surgery only for symptom control. We investigated trends in the use of surgery (lumpectomy or mastectomy) for mBC from 2010 to 2019 using SEER and institutional data, aiming to identify factors that correlate with variability in surgery use. Methods: We included all de novo mBC diagnoses from SEER-17 from 2010-2019. The primary outcome was receipt of surgery, and we included year of diagnosis, demographic (race/ethnicity, age, marital status, median household income in the patient’s county, and urban/rural residence), and disease factors (estrogen receptor (ER) and HER2 status, tumor size, and presence of visceral metastases) in the multivariable logistic regression. We similarly identified all de novo mBC diagnoses from Oncoshare, which merges EMR and California Cancer Registry (CCR) data for patients treated in the Stanford Health Care Alliance. For Oncoshare inclusion a patient must have an encounter with Stanford and be in CCR; however, a patient does not have to receive all cancer care at Stanford. In this analysis, we included distance to the cancer center and insurance status. Results are presented as unadjusted percentages or odds ratios with 95% confidence intervals. Results: In SEER (n = 24,146), the use of surgery for mBC declined from 41% in 2010 to 22% in 2019, a trend also observed in the institutional data (n = 785). Surgery rates were highest in younger patients and declined with each decade of life (44% for those in their 30s vs 19% for those &gt; 80). Surgery was more common in tumors &gt; 2 cm, ER-negative disease, and in patients without visceral metastases. In SEER, patients from counties with median incomes &lt; $50,000 were more likely to have surgery than those from counties with incomes &gt; $75,000 (36% vs 27%; OR: 1.3 [1.2-1.5]). Rural patients were more likely to have surgery than urban patients (34% vs 30%; OR: 1.1 [1.0-1.3]). Unmarried persons had lower odds of surgery than married persons (OR: 0.8 [0.7-0.9]). The institutional data reproduced the findings of the SEER analysis, except that a relationship between surgery and marital status and income was not observed. In the institutional data, patients living farther from the Stanford Cancer Center ( &gt; 100 km) were more likely to have surgery (40% vs 32%; OR 2.1 [1.2-3.8]). Conclusions: The use of surgery for de novomBC decreased dramatically in the 2010s from 41% to 22%. All demographics saw a reduction; however, younger, and married patients underwent surgery at higher rates. Patients from lower-income and rural counties—areas typically with less access to care—were also more likely to have surgery. The institutional analysis showed that a greater distance from an urban hospital network correlated with increased odds of surgery. These patterns suggest differential care provision in rural areas for patients with mBC.

Enrollment of older adults in clinical trials for oncology indications, 2010-2020: An evaluation by the FDA.

e23029 Background: Underrepresentation of older adults (≥65 years) in clinical trials may limit important evidence for regulatory and clinical decision-making. Prior FDA analysis1 demonstrated underrepresentation of older adults in oncology clinical trials. With recent efforts to encourage inclusion of older adults, we evaluated representation in clinical trials using updated data. Methods: This analysis used patient-level data from pivotal trials supporting approvals of new drugs and biologics, and expanded indications from 2010-2020 for cancers with high incidence in the U.S. population and older adults: breast (BC), colorectal (CRC), non-small cell lung (NSCLC), prostate (PC), chronic myeloid leukemia (CML), and multiple myeloma (MM). Age distribution of trial participants compared to incident cases from the NCI SEER Program was analyzed using the Enrollment to Incidence Ratio (EIR). For each cancer-age group, EIR is the percentage of trial participants divided by percentage of incident cases. Previously established EIR categories of &lt; 0.8, 0.80-1.20, and &gt; 1.2 indicate under-, adequate, and over-representation, respectively. Results: The analysis included 86000 participants in 152 trials. Adults &gt;65 yrs were underrepresented in BC trials. Adults &gt;75 yrs were underrepresented in BC, CRC, NSCLC, CML, and MM trials (Table 1). Conclusions: Despite slight variation across cancers, adults &gt;75 yrs appeared to be underrepresented in clinical trials, excepting PC. Given heterogeneity in the natural history by cancer site, incidence may not best represent age distribution of the intended use population for all trials. Further evaluation using prevalence as a comparator may provide additional insights. Additionally, results may be affected by clinical factors such as stage; however, comparison is limited by SEER data element availability. Planned sensitivity analyses such as stratification by drug class and line of therapy may further elucidate enrollment patterns.

Small bowel adenocarcinoma: A SEER database analysis.

e16383 Background: Small bowel malignancies (SBM) account for about 3% of all gastrointestinal malignancies. Small bowel adenocarcinoma (SBA) accounts for the majority of the SBM. SBA is associated with poor prognosis and treatment options are limited in patients with advanced cancer. We analyzed SEER data to study the factors that impacted survival in patients with SBA. Methods: We reviewed data on the SEER database (version Stat 8.4). We extracted data to include patients with adenocarcinoma with labeled primary site as small bowel diagnosed between 2000-2019. We excluded other histology. Variables included were age at diagnosis, race, gender, stage at diagnosis, and grade of malignancy. Chi-square test was used to analyze the association between each variable and survival. Results: We identified 3415 cases of SBA that met our inclusion criteria. The number of annual cases increased from 2000-2019. The mean age was 65.3 years (range 57-75 years). The Male to Female ratio was 11:9. 76% were white, 13.7% were black, 13% were others. Duodenum was the most common primary site with 58%, jejunum at 15.5%, ileum at 13%, and 13.5% overlapped sites or unknown. SEER combined summary stage at diagnosis was 32% regional, 18% localized, 29% distant, and 23% unknown. 6% was grade I, 37% grade II, 27% grade III, 1% grade IV, and 28% unknown. Median survival in patients with SBA was 13 months. The median survival for patients less than or equal to 60 years was 24 months compared to 11 months in patients over 60 years of age (p &lt; 0.0001). Median survival in males was 13 months vs 12 months in females, (p = 0.548). There was no significant difference in survival per race (p = 0.5849). In this cohort, jejunal adenocarcinoma had better outcomes with a survival of 27 months (p &lt; 0.0001). Grade I and II were associated with better outcomes with survival of 21.5 months and 24 months (p &lt; 0.0001). Multivariate analysis showed that age less than 60, early-stage disease, jejunal primary site, and low-grade disease were associated with a better outcome. Conclusions: The number of cases of SBA in the SEER database has increased over the last few decades. SBA is associated with poor outcomes and our analysis shows that patients with early-stage disease, and younger age have better outcomes. Management of SBA is based on retrospective data and data extrapolated from management of colon adenocarcinoma, due to sparsity of prospective data. Clinical trials are needed to identify appropriate therapeutic options for the management of SBA.

ОЦЕНКА ПОТЕНЦИАЛА ЭНЕРГОСБЕРЕЖЕНИЯ В БЮДЖЕТНЫХ УЧРЕЖДЕНИЯХ В СИСТЕМЕ УПРАВЛЕНИЯ ЭНЕРГОРЕСУРСАМИ (СУЭР)

Developed and in operation since 2019, the information and analytical Energy Resources Management System of the Belgorod Region (ERMS) ensures the digitalization of processes related to the provision of measures to improve energy efficiency related to the powers of state authorities of the constituent entities of the Russian Federation and local governments. The SEER uses methods for assessing the energy efficiency of large geographically distributed consumers of utility resources based on an individual analysis of the characteristics and heat consumption of individual buildings. Currently, the SEER includes information for 2,144 municipal and regional institutions in the Belgorod region, which operate in 5,332 buildings. Costs for energy resources and water in 2023 in these institutions amounted to 4.8 billion rubles; over the five years of operation of the SUER, they decreased by 8% under comparable conditions. Based on SEER data, the total savings potential that can be achieved in regional and municipal institutions of the Belgorod region is 14.3% of annual costs for energy resources and water, including 0.8% from organizational activities and 13.5% from technical measures with an average payback period of 6 years. During 2021–2024, pilot projects for the implementation of SEMS were implemented in 9 Russian regions and one organization, which showed savings potential ranging from 4.4 to 29.5%.