Surveillance, Epidemiology, And End Results Registry Research Articles

Evaluating population-level outcomes in Chronic Lymphocytic leukemia in the era of novel therapies using the SEER registry

In the last decade, novel agents such as BTK and BCL-2 inhibitors have revolutionized treatment of CLL/SLL, with clinical trials showing improved overall survival compared to chemotherapeutic agents. However, studies examining whether they have improved overall survival at the population level are lacking. We evaluated this by conducting a retrospective analysis of CLL/SLL patients registered in the National Cancer Institute’s surveillance epidemiology and end results (SEER) database, analyzing overall survival (OS) in periods pre- and post-availability of novel agents, along with demographic information. Our results showed that median OS significantly improved over time [7.8 years (2000–2005), 9.1 years (2006–2013), and not reached (2014–2018) (p < 0.001)]. Compared to diagnosis in 2014–2018, diagnosis in earlier periods was associated with higher mortality risk (2000–2005-HR 1.32, 95 % CI 1.28–1.37, p < 0.001: 2006–2013-HR 1.09, 95 % CI 1.06–1.13, p < 0.001). Lower mortality risk was seen in patients age < 85 years whereas median household income of <$75000 was associated with higher mortality. Our study provides real-world data suggesting a possible multifactorial contribution to improvement in survival, including availability of novel agents, better monitoring, and supportive care. They also show discrepancies in overall survival for CLL/SLL patients due to socioeconomic status and demographic factors.

Abstract 2318: Multimodal machine learning for the automatic classification of recurrent cancers

Abstract The National Cancer Institute’s (NCI) Surveillance, Epidemiology, and End Results (SEER) registries maintain and organize cancer incidence information allowing researchers to derive valuable insights into cancer epidemiology. While significant attention has been devoted to identifying cancers either from clinical text or through tabular data collected by SEER registries, there has been less emphasis on integrating these distinct modes of data. In our multimodal deep learning approach, we use longitudinal tabular data from the Consolidated Tumor Case (CTC) database that encompass a patient’s past diagnoses. This tabular information can augment clinical text to aid in the classification of pathology reports indicative of recurrent cancers. Four NCI SEER registries (Louisiana, New Jersey, Seattle and Utah) have manually labeled 61,150 pathology reports with one of six categories, which we refine into a four-class classification problem. Each pathology report is identified as either positive for recurrence, negative for recurrence/not disease free, new tumor, or an “other” (no malignancy/uncertain) class. Natural Language Processing techniques can extract meaningful information from clinical pathology reports, aiding in the identification of subtle indicators of recurrence by using relevant context. We use a hierarchical self-attention model (HiSAN) to construct document embeddings and classify the pathology report. To further enhance the predictive accuracy of our modeling approach we fuse the textual information from a pathology report with categorical data about patient’s cancer history. For each report, we create a patient context vector that encapsulates tumor-level information from patient’s previous cancer(s). The selected CTC records are associated with cancers diagnosed more than 120 days before the date of biospecimen collection stated in the pathology report. The patient context vector is crafted based on diverse categorical features; including cancer staging, patient age, treatment and sites of metastasis at the time of diagnosis. Features are represented using a combination of one-hot encoding and binning. Additionally, we employ patient and feature-level normalization to maintain proportional significance of features for individuals with multiple past diagnoses. We present preliminary results corresponding to different approaches for classifying cancer recurrence; first, we observe that using only the pathology reports as input yields an accuracy of 68%. Secondly, when using only CTC features with an XGBoost model, we achieve an accuracy of 49%. Finally we show that leveraging multiple data modalities, i.e. HiSAN generated pathology report embeddings and CTC data, significantly improves the model’s predictive accuracy to 76%. This research demonstrates a promising path forward in enhancing classification of clinical text by incorporating longitudinal patient history data. Citation Format: Patrycja Krawczuk, Zachary Fox, Dakota Murdock, Jennifer Doherty, Antoinette Stroupe, Stephen M. Schwartz, Lynne Penberthy, Elizabeth Hsu, Serban Negoita, Valentina Petkov, Heidi Hanson. Multimodal machine learning for the automatic classification of recurrent cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2318.

Accessibility to ERCP-performing hospitals among patients with pancreatic cancer living in SEER regions.

The two most common interventions used to treat painless jaundice from pancreatic cancer are endoscopic retrograde cholangiopancreatography (ERCP) and percutaneous transhepatic biliary drainage (PTBD). Our study aimed to characterize the geographic distribution of ERCP-performing hospitals among patients with pancreatic cancer in the United States and the association between geographic accessibility to ERCP-performing hospitals and biliary interventions patients receive. This is a retrospective cohort study using the Surveillance, Epidemiology, and End Results (SEER)-Medicare database for pancreatic cancer from 2005 to 2013. Multilevel models were used to examine the association between accessibility to ERCP hospitals within a 30- and 45-min drive from the patient's residential ZIP Code and the receipt of ERCP treatment. A two-step floating catchment area model was used to calculate the measure of accessibility based on the distribution across SEER regions. 7464 and 782 patients underwent ERCP and PTBD, respectively, over the study period. There were 808 hospitals in which 8246 patients diagnosed with pancreatic cancer in SEER regions from 2005 to 2013 received a procedure. Patients with high accessibility within both 30- and 45-min drive to an ERCP-performing hospital were more likely to receive an ERCP (30-min adjusted odds ratio [aOR]: 1.53, 95% confidence interval [CI]: 1.17-2.01; 45-min aOR: 1.31, 95% CI: 1.01-1.70). Furthermore, in the adjusted model, Black patients (vs. White) and patients with stage IV disease were less likely to receive ERCP than PTBD. Patients with pancreatic cancer and high accessibility to an ERCP-performing hospital were more likely to receive ERCP. Disparities in the receipt of ERCP persisted for Black patients regardless of their access to ERCP-performing hospitals.

Abstract B017: Adverse events following immune checkpoint inhibitors by pre-existing comorbidities and polypharmacy among patients with lung cancer: A population-based study

Abstract Introduction: Pre-existing comorbidities and polypharmacy (PP) are significant issues among older lung cancer patients leading to complex treatment decisions. Immune check inhibitors (ICIs) have significantly improved lung cancer outcomes; however, most patients with comorbidities are excluded from ICI trials. Moreover, medication management in older adults with cancer is complex due to the volume of medications used and existing comorbidities requiring a high degree of coordination among multiple specialists. This study aims to understand the impact of pre-existing comorbidities and PP among older lung cancer patients using ICI. Methods: We used Surveillance, Epidemiology, and End Results (SEER)-Medicare-linked data to identify patients aged ≥66 diagnosed with lung cancer (1999-2017) having continuous Medicare coverage for up to 2 years or until death. Immune-related adverse events (irAEs) grade 3/4 were defined as immune-suppressive medication claims during inpatient hospitalization 2 years after ICI, excluding the 14 days before the hospitalization. Comorbidities were calculated using a modified Charlson comorbidity index (CCI) and stratified into 3 groups: 0, 1, and more than 2 conditions. PP was stratified into 3 groups: No PP (&amp;lt;5 medications), PP (5-9 medications), and excessive PP (&amp;gt;10 medications). Incidence rates (IRs) and adjusted IR ratios (IRRs) were determined using negative binomial regression. The IRR model was adjusted for age, race, marital status, SEER region, state buy-in, education, and income. Results: We identified 2,461 lung cancer patients treated with ICIs. The IR for grade 3/4 irAEs was 27.59 per 100 person-years (PY) for CCI³2, 20.25 per 100 PY among 1 CCI, and 14.99 per 100 PY among no CCI. Overall IR of grade 3/4 irAEs was 29.29 per 100 PY among patients with excessive PP vs. 19.25 per 100 PY among PP and 13.99 per 100 PY among no PP. Patients with CCI more than 2 are 1.53 times more likely to develop a grade 3/4 irAE than those without CCI (IRR: 1.53; 95% CI: 1.09-2.14). Patients with excessive PP were 1.92 times more likely to develop a grade 3/4 irAE than those using &amp;lt;5 medications (IRR: 1.92; 95% CI: 1.31-2.83). Conclusion: Older adults with lung cancer usually have pre-existing comorbidities and, on average, use at least 11 concurrent medications. Multi-morbidity and polypharmacy are associated with a higher rate of irAEs among these patients; therefore, it’s critical to pay close attention to this population while on ICIs. Further studies are warranted to confirm these findings. Citation Format: Nikita Nikita, Swapnil Sharma, Amy Shaver, Krupa Gandhi, Scott Keith, Christopher Yang, Sarah Gordon, Grace Lu-Yao. Adverse events following immune checkpoint inhibitors by pre-existing comorbidities and polypharmacy among patients with lung cancer: A population-based study [abstract]. In: Proceedings of the 16th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2023 Sep 29-Oct 2;Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(12 Suppl):Abstract nr B017.

Health Insurance Continuity Is Associated with Stage at Diagnosis Among Children, Adolescents, and Young Adults Newly Diagnosed with Lymphoma

Background: Many uninsured patients do not gain Medicaid coverage until the point of, or after, their cancer diagnosis, which can limit access to early cancer detection and delay diagnosis and treatment, leading to poorer survival outcomes. Our previous research found that gaining Medicaid at, or after, diagnosis (vs. continuous Medicaid coverage) is significantly associated with inferior survival among children and adolescents/young adults (AYAs) with lymphoma, a leading cause of pediatric and AYA cancers. However, whether this relationship is mediated by lymphoma stage at diagnosis is unknown. To fill this knowledge gap, the current study examines the association between Medicaid coverage continuity and disease stage among this vulnerable population. Methods: We identified 13,655 children and AYAs (ages 0-39 years) newly diagnosed with lymphoma between 2007-2012, using the linked Surveillance, Epidemiology, and End Results (SEER) registries and Medicaid enrollment data. To measure coverage continuity, we categorized patients into those with 1) continuous Medicaid (enrolled in Medicaid for ≥12 months prior to and through diagnosis), 2) newly gained Medicaid (enrolled in the month of or within 12 months after diagnosis), 3) other patterns of noncontinuous Medicaid, 4) private insurance at diagnosis, 5) other insurance types at diagnosis (Medicare, TRICARE, Military, VA, or Indian/Public Health Service), and 6) unknown or no insurance at diagnosis. The last 3 groups comprised patients not linked to Medicaid enrollment data; thus their insurance continuity patterns were unknown, and their insurance status at the time of diagnosis was extracted from SEER registries. The association between coverage continuity and late-stage diagnosis was examined with separate multivariable logistic regression models among all patients and by cancer subtypes (Hodgkin lymphoma, non-Hodgkin lymphoma). Late-stage was defined as stage IV (vs. stage I-III) using the Ann-Arbor staging classification. Regressions also adjusted for sex, race/ethnicity, age at diagnosis, diagnosis year, rurality, neighborhood socioeconomic status (SES), and state of residence. Results: Half of our sample had private insurance (50.3%), followed by Medicaid insurance (26.9%), other insurance type (10.3%), and unknown (7.9%) or no insurance (4.7%). Among Medicaid-insured patients, 40.1% had newly gained Medicaid, followed by continuous Medicaid (32.5%) and other patterns of noncontinuous Medicaid (27.3%). Overall, one-fourth (24.8%) of the sample had stage IV lymphoma (Figure 1). Compared with privately insured patients, the adjusted percentage of stage IV diagnoses was 16.8 percentage points (ppt; 95% CI=14.6-19.0, p&amp;lt;0.001), 10.6 ppt (95% CI=8.2-13.0, p&amp;lt;0.001), and 4.9 ppt (95% CI=2.7-7.1, p&amp;lt;0.001) higher in patients with newly gained Medicaid, other patterns of noncontinuous Medicaid, and continuous Medicaid, respectively. Among the subset of Medicaid-insured patients, the adjusted percentage of stage IV diagnoses was 10.7 ppt (95% CI=7.5-13.8, p&amp;lt;0.001) and 5.4 ppt (95% CI=3.1-7.6, p&amp;lt;0.001) higher among patients with newly gained Medicaid and other patterns of noncontinuous Medicaid, respectively, compared to those with continuous Medicaid. These patterns persisted across lymphoma subtypes. Notably, coverage continuity patterns varied by sociodemographic factors, with non-Hispanic Black (vs. non-Hispanic White) patients, AYAs ages 15-39 years (vs. pediatric patients ages ≤14), and residents of non-metropolitan (vs. metropolitan) areas or lower SES neighborhoods significantly more likely to have newly gained Medicaid. Conclusions: Less than one-third of Medicaid-insured children and AYAs with lymphoma had continuous insurance coverage prior to cancer diagnosis. Lacking continuous Medicaid coverage was significantly associated with advanced-stage lymphoma diagnosis.

Risk of Therapy-Related Acute Myeloid Leukemia (t-AML) in Adolescent and Young Adults Cancer Patients Treated with Chemotherapy with or without Radiotherapy: A National Population-Based Cohort Study Using SEER Registry

INTRODUCTION The risk of treatment-related acute myeloid leukemia (t-AML) in adolescents and young adults (AYA) with first cancers treated with chemotherapy with or without radiation is not well characterized. We analyzed Surveillance, Epidemiology and End Result (SEER) data for the past 45-years (1975-2020) to identify the overall risk of t-AML in treated AYA cancer patients (pts) and by index cancer sites as well as latency periods. METHOD We analyzed NCI's SEER-8 registries. ICD-O-3/WHO-2008 codes were used to identify pts with index cancer types between the ages of 15-39 years (AYA), treated with either chemotherapy or radiotherapy, and with a subsequent diagnosis of acute myeloid leukemia (AML) at least one year after index cancer, between January 1975 and December 2020. The risk of t-AML was estimated by standardized incidence ratio (SIR), defined as the ratio of observed t-AML-cases in the AYA-with-cancer, to the expected number of AML-cases in the matched general US population. A reference rate file for AML incidence in age, gender, race and calendar-year matched US population is available in SEERStat software. A Poisson distribution of observed t-AML was assumed for the calculation of 95% confidence intervals (CIs). A P-value &amp;lt;.05 was considered statistically significant. We stratified the risk of t-AML for index cancer types with ten or more pts, by attained-calendar-years, and latency at multiple time intervals (Table-1) from index cancer diagnosis. RESULT Among 102,879 AYA cancer pts eligible for analysis, 236 pts developed t-AML with SIR of 6.19 (95%CI: 5.43-7.04, P &amp;lt;.05) with excess risk of 1.46 per-10000 person-years (pyrs). (Table-1). We stratified risk for index cancer types with at least ten patients developing t-AML, these included breast, Hodgkin lymphoma (HL), non-Hodgkin lymphoma (NHL), testicular, thyroid, bone and joint, and soft tissue including heart. As a general trend, risk of t-AML was highest during 1-3 years after index cancer diagnosis compared to subsequent latency periods. However, risk for t-AML was significantly greater than the matched-general US population even after 20-40 years of breast cancer, HL, and NHL (Table-1). When stratified by attained calendar-year periods (1975-1985, 1986-1995, 1996-2005, 2006-2015, 2016-2020), risk for t-AML was higher than general population in all time periods. However, overall SIR was comparatively lower for recent time periods, particularly lowest during 2016-2020 [SIR: 3.49 (2.51-4.72, P &amp;lt;.05)]. Notably, the highest risk for t-AML in 2016-2020 was observed with index cancer of soft tissue including heart [SIR: 10.12 (1.23-36.57), P &amp;lt;.05]. CONCLUSION In this study spanning between 1975-2020, we observed the risk of t-AML trending down among AYA in recent years, albeit higher than the age matched general population. The risk of t-AML remained significantly elevated even after 20-40 from diagnosis of breast cancer, HL and NHL. Since these pts likely received radiation and older (higher intensity) chemotherapy based-regimens, it is essential to analyze the long-term impact of newer therapies for these cancers and assess the risk of t-AML in future studies.

Updated Trends in the Utilization of Brachytherapy in Cervical Cancer in the United States: A Surveillance, Epidemiology, and End-Results Study

BackgroundA previous Surveillance, Epidemiology, and End Results (SEER) study revealed a concerning decline in brachytherapy utilization in the United States between 1988-2009. This study evaluates recent trends in brachytherapy utilization in cervical cancer, and identifies factors and survival benefit associated with the use of brachytherapy treatment. Methods and MaterialsUsing SEER data, 8500 patients with FIGO 2009 stage IB2-IVA cervical cancer treated with external beam radiotherapy (EBRT) between 2000-2020 were identified. Logistic regression analysis was performed on potential factors associated with brachytherapy use: age, marital status, race, ethnicity, income, metropolitan status, year of diagnosis, SEER region, histology, grade and stage. To adjust for differences between patients who received brachytherapy and those who did not, propensity-score matching was used. Multivariable Cox regression analysis assessed the association of brachytherapy use with cervical cancer-specific mortality (CSM) and all-cause mortality (ACM) in the matched cohort. ResultsSixty-four percent of the 8500 women received brachytherapy in combination with EBRT; 36% received EBRT alone. The brachytherapy utilization rate declined sharply in 2003/2004 (lowest rate 44% in 2003), and then gradually improved especially in 2018-2020 (76%). Factors associated with higher odds of brachytherapy use included younger age, married (versus single), later years of diagnosis, certain SEER regions and earlier stage. In the propensity-score matched cohort, brachytherapy treatment was associated with lower 4-year cumulative incidence of cancer death (32.1% vs 43.4%, P<.001) and better overall survival (OS, 64.0% vs 51.4%, P<.001). Brachytherapy treatment was independently associated with lower CSM (hazard ratio [HR] 0.70, 95% CI 0.64-0.76; P<.001) and ACM (HR 0.72, 95% CI 0.67-0.78; P<.001). ConclusionBrachytherapy utilization among SEER regions has improved since 2004 in patients with stage IB2-IVA cervical cancer. Brachytherapy use remains independently associated with significantly lower CSM and ACM, and an essential component of treatment for patients with locally advanced cervical cancer.

Regional differences in clear cell metastatic renal cell carcinoma patients across the USA

PurposeTo test for regional differences in clear cell metastatic renal cell carcinoma (ccmRCC) patients across the USA.MethodsThe Surveillance, Epidemiology, and End Results (SEER) database (2000–2018) was used to tabulate patient (age at diagnosis, sex, race/ethnicity), tumor (N stage, sites of metastasis) and treatment characteristics (proportions of nephrectomy and systemic therapy), according to 12 SEER registries. Multinomial regression models, as well as multivariable Cox regression models, tested the overall mortality (OM) adjusting for those patient, tumor and treatment characteristics.ResultsIn 9882 ccmRCC patients, registry-specific patient counts ranged from 4025 (41%) to 189 (2%). Differences across registries existed for sex (24–36% female), race/ethnicity (1–75% non-Caucasian), N stage (N1 25–35%, NX 3–13%), proportions of nephrectomy (44–63%) and systemic therapy (41–56%). Significant inter-registry differences remained after adjustment for proportions of nephrectomy (46–63%) and systemic therapy (35–56%). Unadjusted 5-year OM ranged from 73 to 85%. In multivariable analyses, three registries exhibited significantly higher OM (SEER registry 5: hazard ratio (HR) 1.20, p = 0.0001; SEER registry 7:HR 1.15, p = 0.008M SEER registry 10: HR 1.15, p = 0.04), relative to the largest reference registry (n = 4025).ConclusionImportant regional differences including patient, tumor and treatment characteristics exist, when ccmRCC patients included in the SEER database are studied. Even after adjustment for these characteristics, important OM differences persisted, which may require more detailed analyses to further investigate these unexpected differences.

Regional differences in metastatic urothelial carcinoma of the urinary bladder patients across the United States SEER registries.

Despite advances in treatment, metastatic urothelial carcinoma of the urinary bladder (mUCUB) is associated with high mortality and treatment risk. We tested for regional differences in mUCUB within a large-scale, population-based database. Using the Surveillance, Epidemiology and End Results (SEER) database (2010-2018), patient (age, sex, race/ethnicity), tumor (T-stage, N-stage, number of metastatic sites), and treatment (systemic therapy, radical cystectomy) characteristics were tabulated for mUCUB patients according to 11 SEER registries. Multinomial regression models and multivariable Cox regression models tested overall mortality (OM), adjusting for patient, tumor and treatment characteristics. In 4817 mUCUB patients, registry-specific patient counts ranged from 1855 (38.5%) to 105 (2.2%). Important inter-regional differences existed for race/ethnicity (3-36% for others than non-Hispanic Whites), N-stage (28-39% for N1-3, 44-58% in N0, 8-22% for unknown N-stage), systemic therapy (38-54%) and radical cystectomy (3-11%). In multivariable analyses adjusting for these patient, tumor, and treatment characteristics, one registry exhibited significantly lower OM (SEER registry 10: hazard ratio [HR] 0.83) and two other registries exhibited significantly higher OM (SEER registries 9: HR 1.13; SEER registry 8: HR 1.24) relative to the largest reference registry (n=1855). We identified important regional differences that included patient, tumor, and treatment characteristics. Even after adjustment for these characteristics, important OM differences persisted, which may warrant more detailed investigation.

Radiation Therapy Changed the Second Malignancy Pattern in Rectal Cancer Survivors.

Background and Objectives: Radiotherapy (RT) plays an important role in the treatment for locally advanced rectal cancer patients. It can bring radio exposure together with the survival benefit. Cancer survivors are generally at an increased risk for second malignancies, and survivors receiving RT may have higher risks than survivors not receiving RT. Whether the risk of an all-site second malignancy may increase after RT is still debated. This study aims to compare the second malignancy pattern in rectal cancer survivors after RT. Materials and Methods: The Surveillance, Epidemiology, and End Results (SEER) database was used for analysis. In total, 49,961 rectal cancer patients (20-84 years of age) were identified between 2000 and 2012 from 18 SEER registries. All patients underwent surgery. The occurrence of second malignancies diagnosed after rectal cancer diagnosis was compared in patients who received and did not receive RT. The standardized incidence ratios (SIRs) with 95% confidence intervals (CIs) were used. SEER*Stat was used to generate the 95% CIs for the SIR statistics using the exact method. Results: Of the total 49,961 patients, 5582 developed second malignancies. For all-site second primary malignancies, the age-adjusted SIRs were 1.14 (95% CI 1.1-1.18) and 1.00 (95% CI 0.96-1.04) in the no RT and RT groups, respectively. In 23,192 patients from the surgery-only group, 2604 had second malignancies, and in 26,769 patients who received RT, 2978 developed second malignancies. With respect to every site, the risk of secondary prostate cancer was significantly lower in the RT group (SIR = 0.39, 95% CI 0.33-0.46) than that in the surgery-only group (SIR = 1.04, 95% CI 0.96-1.12). Moreover, the risk of thyroid cancer was significantly higher in the RT group (SIR = 2.80, 95% CI 2.2-3.51) than that in the surgery-only group (SIR = 1.29, 95% CI 0.99-1.66). Conclusions: RT may change the second malignancy pattern in rectal cancer survivors; the risk of prostate cancer decreased, and the risk of thyroid cancer increased most significantly.

Regional differences in penile cancer patient characteristics and treatment rates across the United States

IntroductionWe tested for regional-specific differences in patient, tumor and treatment characteristics as well as cancer-specific mortality (CSM) of squamous cell carcinoma of the penis (SCCP) patients, across the Surveillance, Epidemiology, and End Results (SEER) registries. MethodsThe SEER database (2000–2018) was used to tabulate patient (age at diagnosis, race/ethnicity), tumor (stage, grade, N-stage) and treatment characteristics (proportions of primary tumor surgery, local lymph node surgery, systemic therapy), according to 12 SEER registries. Multinomial regression models, as well as multivariable Cox regression models tested for CSM differences, adjusting for patient, tumor and treatment characteristics. ResultsIn 5395 SCCP patients, registry-specific patient counts ranged from 2060 (38 %) to 64 (1 %). Differences across registries existed for race/ethnicity, stage, grade and N-stage. Additionally, in stage I-II SCCP patients, proportions of local tumor destruction (LTD) ranged from 19 % to 39 % and from 33 % to 61 % for partial penectomy. In stage III-IV SCCP patients, proportions of partial penectomy ranged from 40 % to 59 % and from 17 % to 50 % for radical penectomy. Local lymph node surgery ranged from 8 % to 24 % and proportions of systemic therapy ranged from 3 % to 14 %. Significant inter-registry differences remained, after adjustment for treatment proportions. Unadjusted five-year CSM ranged from 19 % to 32 %. In multivariable analyses, one registry exhibited significantly higher CSM (SEER registry 10, Hazard Ratio [HR] 1.48), relative to the largest reference registry (SEER registry 1, n = 2060). ConclusionImportant regional differences including patient, tumor and treatment characteristics exist for SCCP patients across SEER registries. After multivariable adjustment, no differences in CSM were recorded, with the exception of one registry.

Applying computer text mining algorithms for oversampling tumor mutation status in medical records for the NCI Patterns of Care studies

Backgrounds: The National Cancer Institute (NCI) conducts Patterns of Care (POC) studies for selected cancer sites under a Congressional Mandate. These studies aim to collect treatment information beyond what is typically collected by the NCI’s Surveillance, Epidemiology, and End Results (SEER) Program. The 2019 POC study focused on non-small cell lung cancer (NSCLC) and melanoma cancer sites. For the NSCLC cases, one of the primary sampling objectives was to oversample patients who tested positive for EGFR/ALK mutations, but initial information on mutation test results was unavailable prior to selecting the study sample. Methods: To address this, text mining algorithms were developed to screen all eligible NSCLC cases from the SEER database. These algorithms were designed to identify the mutation test status, allowing for stratified sampling based on SEER registry, sex, race/ethnicity, and tumor mutation test results. Results: The final NSCLC sample included 2,434 patients aged 20+ with advanced stage (IIIB-IVB) NSCLC diagnosed in 2017 and 2018. Among this sample, 692 cases (13.2%) tested positive for EGFR/ALK mutations. An evaluation of the text mining algorithms performance, based on cases where both algorithm results and known EGFR/ALK status from medical chart abstraction were available, showed good results: sensitivity of 77.6%, specificity of 90.8%, and an overall accuracy 84.8%. Conclusions: The adaption of text mining algorithm proved effective in oversample patients with uncommon conditions in studies where electronic medical records are accessible. The 2019 POC study provides valuable data for researchers to evaluate cancer therapy details and patient characteristics, particularly among those with EGFR/ALK test positive cases.

Combining the 31-gene expression profile test for cutaneous melanoma with the American Joint Committee on Cancer staging identifies the highest-risk patients with stage I-II disease

Introduction: Management guidelines for cutaneous melanoma (CM) are based on patients’ recurrence risk by stage. Most newly diagnosed patients (88%) will be categorized as node-negative (stage I-II) and will be considered low-risk. However, because of the size of this group, the small percentage of stage I-II individuals who do die of melanoma account for the majority of melanoma-associated deaths. In collaboration with the Surveillance, Epidemiology, and End Results (SEER) program, we demonstrated, in a large, unselected cohort of clinically tested patients, that the 31-GEP test identifies those individuals with stage I-II disease, who have higher risk of melanoma-specific death and who may benefit from more aggressive management. &#x0D; Methods: SEER registries linked individuals diagnosed with CM between 2013-2018 were linked to data for 31-GEP-tested patients (N=9,207 after exclusions). For this study, analysis focused on the subset reported as node-negative (N=6,301). Patient 5-year melanoma-specific survival (MSS) was estimated using Kaplan-Meier analysis and the log-rank test for patients considered by the 31-GEP to be low-risk (Class 1A), intermediate-risk (Class 1B/2A), or high-risk (Class 2B). Multivariable Cox regression analysis was used to evaluate significant predictors of melanoma-specific death. Notably, at the time of diagnosis, none of the immune checkpoint inhibitor or targeted signal transduction therapies were approved for use in node-negative patients, providing a largely contemporary therapy naïve cohort. &#x0D; Results: Patients with a Class 2B 31-GEP result had significantly lower 5-year MSS than patients with Class 1B/2A or Class 1A results (85.0% vs. 95.6% vs. 97.9%, p&lt;0.001). The 31-GEP Class 2B result (HR=4.08, p&lt;0.001) was the strongest predictor of melanoma-specific death. Breslow thickness (HR=1.16, p=0.002), presence of ulceration (HR=2.10, p=0.006), and age (HR=1.05, p&lt;0.001) were also significant predictors of melanoma-specific death. The 31-GEP had a sensitivity of 78.4% and a negative predictive value (NPV) of 99.4%. Combining the 31-GEP Class result with AJCC staging could increase the sensitivity to 82.0% while maintaining a high NPV (99.4%). &#x0D; Conclusion: In a large, unselected cohort of patients with stage I-II CM, the 31-GEP Class 2B identified patients with a high risk of progression and death from melanoma who should be considered for more aggressive management. Conversely, the high NPV suggests that the 31-GEP reliably identifies patients at low risk of tumor progression who could safely avoid intensive surveillance and intervention.

Age-specific cardiovascular disease-related mortality among patients with major gastrointestinal cancers: A SEER population-based study.

Studies have reported age as a risk factor for cardiovascular disease (CVD)-related mortality; however, only a few studies have focused on the relationship between age and CVD-related mortality, especially among major gastrointestinal cancers. The present retrospective cohort enrolled patients with colorectal, pancreatic, hepatocellular, gastric, and esophageal cancer between 2000 to 2015 from the Surveillance, Epidemiology and End Results Registry (SEER). Standardized mortality ratio (SMR), competing risk regression, and restricted cubic spline (RCS) analyses were used in our study. We analyzed 576,713 patients with major gastrointestinal cancers (327,800 patients with colorectal cancer, 93,310 with pancreatic cancer, 69,757 with hepatocellular cancer, 52,024 with gastric cancer, and 33,822 with esophageal cancer). Overall, CVD-related mortality gradually decreased every year, and the majority were older patients. All cancer patients had a higher CVD-related mortality rate than the general U.S. The adjusted sub-hazard ratios for middle-aged with colorectal cancer, pancreatic cancer, hepatocellular cancer, gastric cancer, and esophageal cancer were 2.55 (95% CI: 2.15-3.03), 1.77 (95% CI: 1.06-2.97), 2.64 (95% CI: 1.60-4.36), 2.15 (95% CI: 1.32-3.51), and 2.28 (95% CI: 1.17-4.44), respectively. The adjusted sub-hazard ratios for older patients with colorectal cancer, pancreatic cancer, hepatocellular cancer, gastric cancer, and esophageal cancer were 11.23 (95% CI: 9.50-13.27), 4.05 (95% CI: 2.46-6.66), 4.47 (95% CI: 2.72-7.35), 7.16 (95% CI: 4.49-11.41), and 4.40 (95% CI: 2.28-8.48), respectively. A non-linear relationship between age at diagnosis and CVD-related mortality was found in colorectal cancer, pancreatic cancer, and esophageal cancer; their reference ages were 67, 69, and 66 years old, respectively. This study demonstrated that age was a risk factor for CVD-related mortality among major gastrointestinal cancers.

Long-term outcomes in a population-based cohort of 2,967 uveal melanoma patients clinically tested with the 15-gene expression profile: A collaborative study with the National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) Program Registries.

e21574 Background: Metastatic risk prediction in uveal melanoma is commonly performed using a commercially available 15-gene expression profile (15-GEP) test. A recent collaborative effort with the NCI’s SEER program registries allowed for serial linkage of UM patients captured by SEER registries with clinical 15-GEP test results. The objective of this study was to confirm the 15-GEP test’s performance in clinically tested uveal melanoma patients with long-term outcomes at the population level. Methods: Following established operating procedures, SEER registries linked UM cases diagnosed from 2010–2019 to 15-GEP test results provided by Castle Biosciences using a trusted third-party (honest broker). The patients were clinically tested as recommended by guidelines (testing was performed either on fine needle aspiration biopsy tissue or formalin-fixed, paraffin-embedded tissue from enucleations). De-identified data were provided to NCI and Castle Biosciences scientists for this study. Kaplan-Meier analysis with log-rank test was used to evaluate the available endpoints of melanoma-specific survival (MSS) and overall survival (OS). Multivariable analysis was performed with Cox proportional hazards modeling. Results: A total of 2,967 UM patients ≥ 18 years of age with localized disease (excluding Stage IV) and 15-GEP test result were included in this study (Class 1, 1904; Class 2, 1063). The cohort was 51.7% male, with a median age of 63 years. Median follow-up time was 2.2 years. Patients with a low-risk (Class 1) vs. high-risk (Class 2) result had significantly higher 7-year survival outcomes (OS, 78.9 vs. 31.8%, p &lt; 0.001; MSS, 90.4 vs. 41.6%, p &lt; 0.001, respectively). Patients with a Class 2 result were over six times more likely to experience a UM-related death than those with a Class 1 result (31.8 [338/1063] vs. 5.0% [96/1904]; p &lt; 0.001). Multivariable analysis showed that the 15-GEP was the strongest independent predictor of MSS when compared with clinicopathologic features (tumor diameter or thickness) and age, with a Class 2 hazard ratio of 6.2 (95% CI, 3.9–9.7; p &lt; 0.001). Notably, there was no association between GEP class call distribution and socioeconomic status as measured by the Yost index (p = 0.79). Conclusions: In this analysis, the 15-GEP accurately stratified risk of death from UM in a clinically tested, population-based cohort of patients, confirming previous studies. Socioeconomic status did not impact the likelihood of getting a high-risk result, suggesting equitable access to care for patients who receive 15-GEP testing. Ongoing analyses of SEER-linked UM data will include patients who received a PRAME test result, allowing for assessment of this adjunctive biomarker in combination with the 15-GEP result.

Enzalutamide vs abiraterone in patients with prostate cancer with and without type 2 diabetes mellitus.

5066 Background: Estimations are that 8-18% of newly diagnosed cancer patients have Type 2 Diabetes Mellitus (T2DM). Current guidelines offer no clarification on the choice of therapy for prostate cancer (PCa) patients with T2DM. Since the comorbid PCa—T2DM population is increasing and these patients face unique challenges from adverse events requiring acute care utilization (ACU) it is important to ascertain which medication poses less of a threat of adverse events. The objective of this study was to describe the SEER-Medicare T2DM-PCa population in terms of those with and without complications and determine if T2DM influences ACU and PCa medication choice. Methods: We identified men with primary PCa treated with abiraterone (ABI, 2011) or enzalutamide (ENZ, 2012) with comorbid T2DM using the Surveillance, Epidemiology, and End Results (SEER)-Medicare database (1999-2019). Patients with a missing date of diagnosis, those diagnosed at death, having HMO or no continuous A and B coverage 12 months before diagnosis and 6 months continuous part D coverage before starting treatment as well as 6 months following treatment or until death were excluded. T2DM was identified using ICD-9 and ICD-10 codes. ACU was calculated utilizing the CMS Research Data Assistance Center definition focusing on emergency room visits resulting in in-patent hospitalization during the 6 months following treatment initiation. Differences in emergency room in-patient visits were evaluated using negative binomial models adjusted for age, race, SEER region, marital status, state-buyin, education, income, and modified Charlson Comorbidity Index (excluding T2DM). Results: The sample of 11,163 men included 61.8% treated with ABI and 38.2% with ENZ. We identified 3,044 (27.3%) men with T2DM, of which, 59.1% were treated with ABI and 40.8% with ENZ. Compared to those without T2DM those with T2DM had more ACU, regardless of medication, with a higher rate in the ABI than ENZ group, see table. Among diabetics, those with complications related to T2DM compared to those without also experienced a higher rate of ACU. The rate of ACU was 43% higher among those treated with ABI compared to ENZ (IRR 1.43; 95% CI 1.28, 1.61). Overall mortality (15.7% vs 10.1%, p&lt;0.01) and PCa-specific mortality (8.2% vs 5.3%, p&lt;0.01) were worse in those with T2DM treated with ABI compared to ENZ. Conclusions: Outcomes for PCA patients with T2DM are worse than for those without. ACU is higher in men treated with ABI compared to ENZ regardless of T2DM status. Given the worse outcomes for comorbid T2DM patients, further study to evaluate comorbidity management in this population is warranted. [Table: see text]

Functional status and therapy receipt among older adults with multiple myeloma.

e18770 Background: Multiple myeloma (MM) disproportionately affects older adults, among whom aging-related impairments in activities of daily living (ADLs) are common. We aimed to evaluate the association between functional disability and receipt of myeloma therapy in a population of older MM patients who had received home health (HH) services in the year prior to MM diagnosis. Methods: We identified a cohort of adults age ≥66 diagnosed with MM from 2010-2017 who received HH services prior to diagnosis from the Surveillance, Epidemiology, and End Results (SEER) database linked to Medicare claims and the Outcome and Assessment Information Set (OASIS). OASIS captures functional assessments for Medicare beneficiaries receiving HH. Continuous Medicare enrollment for 12 months before and after cancer diagnosis was required. Primary exposure was disability, a composite score of OASIS ADL items, categorized by distributional breakpoints as mild (0-9), moderate (10-25) or severe (26-40). Primary outcome was receipt of MM therapy in the year following diagnosis. Secondary outcomes were treatment type (triplet therapy, stem cell transplant) and healthcare utilization (emergency department [ED] visits, hospitalizations). Associations between disability and outcomes were assessed via hazard ratios, adjusted for age, sex, race, comorbidity, and SEER region with death as a competing risk. For utilization outcomes, we examined differences in mean cumulative counts (MCC) for each event between disability strata. Results: Among 4,108 adults, those with severe disability were less likely to receive MM therapy (vs mild, HR = 0.70 (95% CI: 0.56-0.88); vs. moderate, HR = 0.63 (0.51-0.79). Individuals with moderate disability were more likely to receive treatment than those with mild disability (HR = 1.10 (1.02, 1.20)). Individuals with severe disability were less likely to receive triplet therapy (vs. mild, HR = 0.57 (0.35, 0.93)). Rates of triplet therapy did not differ significantly between moderate and mild disability groups (HR = 1.11 (0.94, 1.32)). Transplant was rare in all groups ( &lt; 5%). Individuals with mild disability had fewer ED visits (MCC difference vs. moderate = -0.34 (-0.50, -0.18); vs. severe = -0.57 (-1.05, -0.10)) and fewer hospitalizations (MCC difference vs. moderate = -0.42 (-0.54, -0.30); vs. severe = -0.59 (-0.93, -0.26)) in the year following diagnosis. Both measures did not differ significantly between moderate and severe groups. Conclusions: In this large, nationwide study of older adults with MM and HH use, we identified differences in treatment receipt and healthcare utilization by degree of pre-diagnosis functional disability. Individuals with moderate disability had comparable or more treatment than those with mild disability, though they were similar to the severe group in healthcare utilization. Individuals with moderate disability receiving HH may need supportive care resources when initiating therapy.

Adjuvant therapy in localized vs regional biliary tract cancers: A SEER analysis.

e16221 Background: Biliary Tract Cancers (BTC) are aggressive malignancies. Treatment of early-stage disease involves surgical resection, but whether adjuvant therapy offers a clear survival benefit remains inconclusive. This study aims to explore the impact of demographic factors and adjuvant therapy on survival outcomes at a population level through the lens of the latest SEER (Surveillance, Epidemiology, and End Results) registries. Methods: Cases were filtered by the following criteria: histologic type (adenocarcinoma), anatomic location (gallbladder, intrahepatic, extrahepatic), stage (local and regional), microscopically confirmed diagnosis, and confirmed surgical resection. Cases with incomplete entries were excluded. A Cox proportional hazard model was used to analyze survival in the context of race, gender, income, and type of adjuvant therapy (none, chemotherapy (AC), chemo-radiotherapy (CRT), and radiation (RT)). Data analysis was performed in RStudio. Results: Between 2005 and 2019, 3565 patients were identified. Patients with regional BTC demonstrated improved median survival with AC (27 vs 23 months, HR 0.72, CI 0.63-0.82, p ≤.05) and CRT (28 vs 23 months, HR 0.63, CI 0.56-0.72, p ≤.05). Median survival in localized BTC was 52 months. Adjuvant therapy of any type did not impact survival for localized BTC. Analysis of race as a covariate in localized BTC showed that all minority groups had improved survival compared to non-Hispanic whites (Asian, HR 0.33; Black, HR 0.50; Hispanic, HR 0.38, p≤.05). In regional gallbladder BTC, this survival benefit was limited to Asians only (HR 0.52, p≤.05). Conclusions: Our data show that AC and CRT may improve survival in patients with regional BTC across all anatomic subtypes. In our cohort, adjuvant therapy had no impact on survival in localized BTC. Differences in outcomes may exist among racial groups for different subtypes of BTC. Future adjuvant clinical trials for BTC should consider stratifying for regional and localized disease. [Table: see text]

Integrating the 31-gene expression profile test into clinical decision-making to guide risk-aligned care decisions for patients with stage I-III cutaneous melanoma: NCI-SEER Analysis.

6601 Background: Cutaneous melanoma (CM) management depends on an estimate of recurrence risk or death. Because of recognized disease risk and drug efficacy, pembrolizumab immunotherapy has been approved for stage IIB-C patients. However, it is a treatment with potential financial and biological toxicities with 16% of patients experiencing severe treatment related adverse events. Improved risk stratification could identify individuals for whom more selective management would be appropriate. In collaboration with the NCI Surveillance, Epidemiology, and End Results (SEER) program, which includes information on patients tested with a 31-gene expression profile (31-GEP) test in an unbiased and unselected manner, we confirm the 31-GEP adds independent prognostic value in patients with stage I-III CM, and investigated whether the 31-GEP could identify stage IIB-C patients who might reasonably avoid adjuvant therapy while also identifying patients with stage I disease who might benefit from more aggressive diagnostic and/or therapeutic intervention. Methods: SEER registries linked CM cases diagnosed 2013-2018 to 31-GEP test results provided by Castle Biosciences (9,207 matched) using a registry trusted third party (honest broker), including subsets of stage IIB-IIC (n=775) and stage I (n=5,651) melanoma. The primary endpoint was melanoma-specific survival (MSS). The 31-GEP risk-groups included Class 1A (low), Classes 1B/2A (intermediate), and Class 2B (high). Each was compared using Kaplan-Meier analysis and the log-rank test. Results: In the overall cohort, patients with a Class 1A result had a significantly higher 5-year MSS than a Class 1B/2A or Class 2B result (97.4% vs. 93.5% vs. 83.3%, p&lt;0.001). In multivariable analysis, a Class 2B result (HR=3.26, p&lt;0.001) and positive lymph node status (HR=3.41, p&lt;0.001) were the strongest predictors of diminished MSS. In stage IIB-C, 18.8% (146/775) of patients received a Class 1A result with an associated 5-year MSS of 89.1% (vs. Class 2B: 76.6%, p=0.001), despite the absence of pembrolizumab approval for the evaluated SEER population. Alternatively, 4.6% (260/5,651) of patients with stage I disease had a Class 2B result, with an associated 5-year MSS of 92.3% (vs. Class 1A: 98.0%, p&lt;0.001). Conclusions: The SEER database allowed a non-biased analysis of a large cohort of patients with stage I-III CM who were tested with the 31-GEP. The 31-GEP identifies patients, including those with stage IIB-C disease, with favorable outcomes and could help physicians and patients decide whether to pursue adjuvant therapy. Moreover, the 31-GEP identities high-risk patients, including those with stage I disease, for whom more aggressive follow-up and early intervention should be considered.

Comparison of diversity across real-world evidence, randomized clinical trial, and Surveillance, Epidemiology, and End Results (SEER) data in patients with metastatic breast cancer.

e18670 Background: Compared with patients in typical care settings, participants in randomized clinical trials (RCTs) tend to be younger and less diverse. We compared representation of age, race, and ethnicity at initiation of first-line (1L) therapy for metastatic breast cancer across real-world evidence (RWE), RCT, and Surveillance, Epidemiology, and End Results (SEER) registry data. Methods: RWE and SEER inclusion criteria were based on corresponding RCTs. We compared age, race, and ethnicity at initiation of 1L therapy among female patients with metastatic breast cancer across 2 RWE chart review studies (completed between 2019-2021), 4 RCTs with similar eligibility (reported between 2014–2021), and the latest SEER registry data (2017–2019) using t-tests and 2-sided chi-square tests. Results: We collected demographics of 57,479 patients (RWE: n = 860, 1.5%; RCT: n = 2,707, 4.7%; SEER: n = 53,912, 93.8%). Mean age at 1L therapy initiation was significantly higher in RWE than RCT studies (61.8 vs 56.0 years) and representation of Black patients was significantly higher in RWE than RCT studies (25.3% vs 2.9%), both P&lt; .001. Mean age at metastatic diagnosis in SEER was 59.0 years; age at 1L initiation was not reported. SEER included 13.4% Non-Hispanic Black patients. Hispanic representation was 11.2% in RWE, not reported in RCT, and 15.9% in SEER studies (Table). Conclusions: Decisions about treatment choices among clinicians and payers are primarily driven by RCT results, yet they may not be representative of real-world populations. We found that RCTs had lower representation of older and Black patients than both SEER and RWE studies. Limitations included varying study periods, lack of Hispanic ethnicity reporting in RCTs, differential SEER participation by US region, and potential patient overlap. Well-conducted RWE studies may fill gaps left by RCTs for improving representation and generalizability to patients with metastatic breast cancer. [Table: see text]