Tumor Registrars Research Articles

Effect of streamlining prior authorization processes for systemic cancer treatments on efficiency and financial losses.

412 Background: Prior authorizations (PAs) for systemic cancer treatments are a barrier to timely quality cancer care delivery. The high administrative burden to complete peer to peer’s (P2P) and appeals leads to care delays and revenue losses. At our tertiary academic cancer center, in partnership with our cancer registry and Epic teams, we shifted the clerical work required for PAs from clinicians to certified tumor registrars (CTRs) to increase authorization efficiency and decrease financial losses from denials. Methods: Clinicians place the treatment plan order which goes to the CTR work queue. We leveraged the Epic staging smart form to consolidate and auto-populate common elements needed for PA. The CTRs then complete and validate the form prior to the authorization specialists obtaining the PA. We compared the pre (1/2022-9/2022) and post-implementation (11/2022-4/2024) periods on our primary outcomes: average monthly number of PAs pending review (including P2P, appeals and authorizations requiring clarifying clinical documentation), average monthly denial rate and average time spent on PAs by the authorization specialists. We compared the financial losses from denied PAs from calendar year 2022 to 2023. Analyses were conducted using descriptive statistics. Results: There was a 1.4% reduction in the average monthly number of PA’s pending review (350 vs 345). The average monthly authorized rate was steady from the beginning of the intervention (94%) to present time (95%). The average monthly denial rate was 2.9% at the beginning of the intervention and 0.9% at present time. There was a consistent decrease in the average time a PA was spent in pending review status: 118 hours in 2022, 102 hours in 2023 and 79 hours in quarter 1 of 2024. Despite the overall increase in volume of authorizations submitted, there was an overall cost savings of $27,392,894 from 2022 ($49,964,703) to 2023 ($22,571,809) from the decrease in denials. In 2022, 6.2% of total approved charges were denied and this decreased to 2.7% in 2023. Conclusions: By leveraging the EHR and optimizing existing non-clinical staff workflows, we maintained sustained average authorized and denial rates despite overall volume going up due to improved authorization efficiency and reduced financial losses due to denials. Additionally, there were overall cost savings for the institution due to the decreased percent of denials of total approved charges. Future work will include further refinement of treatment and disease specific optimizations.

Validation of Non-Small Cell Lung Cancer Clinical Insights Using a Generalized Oncology Natural Language Processing Model.

Limited studies have used natural language processing (NLP) in the context of non-small cell lung cancer (NSCLC). This study aimed to validate the application of an NLP model to an NSCLC cohort by extracting NSCLC concepts from free-text medical notes and converting them to structured, interpretable data. Patients with a lung neoplasm, NSCLC histology, and treatment information in their notes were selected from a repository of over 27 million patients. From these, 200 were randomly selected for this study with the longest and the most recent note included for each patient. An NLP model developed and validated on a large solid and blood cancer oncology cohort was applied to this NSCLC cohort. Two certified tumor registrars and a curator abstracted concepts from the notes: neoplasm, histology, stage, TNM values, and metastasis sites. This manually abstracted gold standard was compared with the NLP model output. Precision and recall scores were calculated. The NLP model extracted the NSCLC concepts with excellent precision and recall with the following scores, respectively: Lung neoplasm 100% and 100%, NSCLC histology 99% and 88%, histology correctly linked to neoplasm 98% and 79%, stage value 98.8% and 92%, stage TNM value 93% and 98%, and metastasis site 97% and 89%. High precision is related to a low number of false positives, and therefore, extracted concepts are likely accurate. High recall indicates that the model captured most of the desired concepts. This study validates that Optum's oncology NLP model has high precision and recall with clinical real-world data and is a reliable model to support research studies and clinical trials. This validation study shows that our nonspecific solid tumor and blood cancer oncology model is generalizable to successfully extract clinical information from specific cancer cohorts.

Abstract PO5-04-07: Characteristics of patients with metastatic HR+/HER2+ breast cancer in a Kaiser Permanente observational multi-site cohort

Abstract Background: Approximately 10% of female breast cancer and 13% of male breast cancer tumors express hormone-receptor positive/human epidermal growth factor receptor 2 positive (HR+/HER2+) molecular subtype. Since HER2+ targeted agents were made available nearly 10 years ago, there has been limited innovation in treatment of patients with HR+/HER2+ metastatic breast cancer (mBC) to improve clinical outcomes. This study aims to characterize the HR+/HER2+ mBC patient population by describing demographic and clinical characteristics and treatment patterns in a racially diverse, insured, population-based cohort. This study includes data from 4 regions of the Kaiser Permanente (KP) health system – Northwest, Northern California, Southern California, and Colorado. Methods: Eligible study patients were adults age > 18 years with HR+/HER2+ invasive breast cancer who were either initially diagnosed with Stage IV disease or were diagnosed at an earlier stage and experienced recurrence with distant metastasis between 6/8/2012 and 12/31/2019 with data collected through 12/31/2021. Select information was extracted from electronic health records (EHR) of eligible patients and combined with manually abstracted data to create the analytic dataset. Breast cancer recurrence was identified by tumor registrars or obtained through a breast cancer recurrence detection algorithm and verified by chart review. Chart reviewers collected data on metastatic sites, disease progression, reasons for stopping systemic therapy medications, and verified use of medications. Preliminary results: During the study period, we identified 492 subjects with HR+/HER2+ mBC. Seven were men and 485 were women. 221 (45%) had de novo Stage IV breast cancer at diagnosis and 271 (55%) had recurrent mBC. Median age at metastatic diagnosis was 59 years for de novo cases and 62 years for recurrent cases. The cohort was racially and ethnically diverse, with 13% Asian subjects (11% de novo, 15% recurrent), 12% Black subjects (10% de novo, 14% recurrent), and 66% White subjects (71% de novo, 63% recurrent). 18% of cases were recorded as Hispanic (19% de novo, 18% recurrent) in the EHR. Major sites of metastases, including bone, liver, lung, brain, and distant lymph nodes, were similar in both the de novo and recurrent groups. Among women (see table below), 28 (6%) received no systemic treatment for their metastatic disease (9% de novo, 3% recurrent). Among women with systemic treatment, 80% received anti-HER2 agents (90% de novo, 73% recurrent). Of women with recurrent mBC who received systemic therapy, 56% had received anti-HER2 therapy in the adjuvant setting. Conclusion: This population-based study of a racially diverse cohort examined EHR-derived data to describe current demographics and treatment patterns for HR+/HER2+ mBC in an insured patient population. Over 90% of women received systemic therapy, and the majority of those received anti-HER2 therapy per current treatment guidelines. Future analyses will focus on clinical outcomes in this cohort with a goal of informing how prognosis can be improved. Table. Citation Format: Sheila Weinmann, Debra Ritzwoller, Marilyn Kwan, Reina Haque, Stella Stergiopoulos, Ashley Cha, Eric Gauthier, Daniel Sapp, John Dickerson, Kathryn Richert-Boe, Weiming Hu, Nikki Carroll, Lie Hong Chen, Valerie Lee, Ravi Zalavadia, Maureen O'Keeffe-Rosetti. Characteristics of patients with metastatic HR+/HER2+ breast cancer in a Kaiser Permanente observational multi-site cohort [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-04-07.

Streamlining chemotherapy prior authorizations: Leveraging resources to cut delays and boost revenue.

573 Background: Prior authorizations (PAs) for systemic cancer treatments are a barrier to timely quality cancer care delivery. The high administrative burden to complete peer to peer’s (P2P) and appeals leads to care delays and revenue losses. At our tertiary academic cancer center, in partnership with our cancer registry and Epic teams, we shifted the clerical work required for PAs from clinicians to certified tumor registrars (CTRs) to increase authorization efficiency and decrease financial losses from denials. Methods: Clinicians place the treatment plan order which goes to the CTR work queue. We leveraged the Epic staging smart form to consolidate and auto-populate common elements needed for PA. The CTRs then complete and validate the form prior to the authorization specialists obtaining the PA. We compared the pre (1/2022-9/2022) and post-implementation (11/2022-5/2023) periods on our primary outcomes: average monthly number of PAs pending review (including P2P, appeals and authorizations requiring clarifying clinical documentation) and average time spent on PAs by the authorization specialists. We compared the same post implementation period to the same time the year before on the financial loss from denied PAs. Analyses were conducted using Welch's t-test. Results: There was an 11% reduction in the average monthly number of PAs pending review (350 vs 313; p=0.27) from pre to post implementation. There was an average improvement of 6 hours spent obtaining a final approval/per authorization (100 vs 106 hours; p=0.64) from pre to post implementation. In addition, there was a trend toward improvement in financial losses from PA denials from Dec 2022 to March 2023 and a cost savings of ~ 7 million dollars from the same time period pre-intervention; p=0.01 (Table). Conclusions: By leveraging the EHR and optimizing existing non-clinical staff workflows, we demonstrated a sustained decrease in the number of ambulatory PAs pending review, time spent on the PAs and significantly decreased financial losses from denials. Future work will include treatment and disease specific optimizations and evaluating the impact on clinician well-being.[Table: see text]

Implementation of a streamlined prior authorization process to improve cancer care delivery.

1531 Background: Prior authorizations (PAs) for systemic cancer treatments are increasingly becoming a barrier to timely quality cancer care delivery. There is an extraordinary administrative burden placed on clinicians to complete peer to peers (P2P) and appeals when authorization specialists are unable to identify appropriate clinical data in the EHRs. This leads to delays in care, staff burnout and loss of revenue. At our tertiary academic cancer center, in partnership with our cancer registry and Epic teams, we propose shifting the clerical work required for intravenous chemotherapy PAs from clinicians to certified tumor registrars (CTRs) to improve documentation needed for authorization specialists and decrease PA associated clinician burnout. Methods: Clinicians first place the treatment plan order which goes to the CTR work queue. We leveraged the Epic staging smart form to consolidate and auto-populate a list of common data elements needed for PA (performance status, cancer biomarkers, stage, line of treatment, and goals of treatment). The CTRs then complete and validate all elements necessary for authorization specialists to obtain the PA. Our primary outcomes include: average monthly number of intravenous chemotherapy authorizations pending review (including P2P, appeals and authorizations requiring other clinical interventions from the primary team) during a pre (9/1/2021-9/30/22) and post-implementation (11/14/22-1/30/23) period. Analyses were conducted using descriptive statistics. Results: The average monthly number of authorizations pending review was 332 pre-implementation and 227 post-implementation, which is a 32% reduction. To account for temporal changes, when comparing December 2021 to January 2022 vs December 2022 and January 2023, the average number of referrals pending review was 276 vs 227 which is an 18% reduction. First line treatment regimens accounted for most regimens requiring review; 28% and 27% pre-implementation and post-implementation, respectively. Non-chemotherapeutics like lanreotide and feraheme had the highest pending review status pre-and post-implementation respectively. Among chemo/immunotherapeutics, Nivolumab and Trastuzumab deruxtecan had the highest pending review status pre and post implementation, respectively. GI regimens had the highest number of regimens requiring review accounting for 17% vs 22% of all referrals pre and post implementation, respectively. Conclusions: By leveraging technology and increasing efficiencies of existing non-clinical staff workflows, we significantly decreased the number of ambulatory chemotherapy PAs that required further clinical review including P2P and appeals thereby improving efficient care delivery. Next steps include evaluation of sustained improvement, time to final authorization, revenue analysis and the impact on the well-being of clinicians.

Implementation of a streamlined prior authorization process to improve clinician wellness and cancer care delivery.

428 Background: Prior authorizations (PAs) for chemotherapy are increasingly becoming a barrier to timely and quality care delivery for cancer patients. There is an extraordinary administrative burden placed on clinical teams to participate in peer to peers (P2P) and file appeals when authorization specialists are unable to efficiently find the necessary clinical data in the EHR. 1 This leads to delays in care, staff burnout driven by decreased job satisfaction, loss of revenue from decreased clinician productivity and from chemotherapy given emergently but not retroactively reimbursed. 2 In partnership with the cancer registry team and Epic, we propose a workflow that shifts the clerical work required for PAs from the clinical teams to the certified tumor registrars (CTRs) using smart form optimization so the finance team has all the required information in one place to successfully and efficiently process the PAs. Methods: A list of common data elements including performance status, cancer biomarkers, line of treatment, goals of treatment and stage were identified. We are leveraging the staging smart form in EPIC to auto-populate these elements when available and then use CTRs to enter cancer stage, fill in any gaps and validate the data. Once approved by the CTR, the chemotherapy plan will be available in the authorization work queue for efficient processing. In 2020, we had 1,389 chemotherapy plans entered and about 38% of these required clinical intervention for financial clearance. We will track the number of P2Ps being conducted post-implementation and compare this to numbers prior to implementation. We will use descriptive statistics and t-test to compare these values. We are anticipating a 25% decrease in P2Ps. We will survey oncology physicians in June 2022 to measure baseline perceived physician impact of PAs and burnout and then again in January 2023 to measure the impact post-implementation. Results: A wellness survey was administered to Medical Oncology faculty within the Mount Sinai Health System pre-pandemic in 2020 and in 2021 and among respondents, approximately 1/3 have moderate to severe burnout risk pre- and post-pandemic. Coordination of care and PAs were rated as the largest barriers to wellbeing, and this increased from 43% pre- to 71% post-pandemic indicating a significant area of opportunity to improve wellbeing among medical oncology faculty. Conclusions: By leveraging technology and non-clinical staff workflows to improve the processing of insurance PAs of ambulatory intravenous chemotherapy, we aim to improve timeliness of quality care delivery to patients and clinical staff wellness. West CP et.al. Physician burnout: contributors, consequences and solutions. J Intern Med. Jun 2018;283(6):516-529. Lin NU et al. Increasing Burden of Prior Authorizations in the Delivery of Oncology Care in the United States. J Oncol Pract. Sep 2018;14(9):525-528.

1582P Real-world management of immune-related adverse events in the community setting

Immune checkpoint inhibitors (ICIs) are integral to the treatment of many cancers, often leading to improved survival outcomes and durable response. Immune-related adverse events (irAEs) differ in presentation and management from adverse events associated with other cancer therapies. This study describes the characteristics and management of irAEs in the community setting. This retrospective cohort study used electronic health record data from 2 community health systems (30+ hospitals, 500+ sites) from the Syapse Learning Health Network. Patients (pts) first exposed to ICIs between 1/1/2015 and 2/28/2019 were included with data cutoff of 3/1/2020. Pts with irAEs were identified using a 2-step screening process with a target cohort of ∼400 pts. Step 1: Technology-enabled screening using medication names, ICD codes, ER/hospitalization records and abnormal clinical labs. Step 2: Chart review by certified tumor registrars to confirm and characterize irAEs (attribution, severity, management and outcomes), with adjudication by a medical oncologist. 1382 pts were reviewed from a pool of 2129 pts with suspected irAEs, resulting in a cohort of 403 pts with irAEs. Lung cancer (60%) and melanoma (16%) were the most common histologies. The median time to irAE onset was 77 days. 91 pts (23%) had an ER/hospitalization visit due to irAE, with 43% of visits due to gastrointestinal irAE. irAE-related deaths occurred in 7 pts (1.7% of irAE cases). Systemic steroid use was documented in 236 pts (59%; duration of use 31 days).) Pts treated with steroids ≤24 hours of irAE onset had a higher resolution proportion at 3 months (0.7; 95% CI 0.6-0.8) than pts treated >72 hours (0.5; 95% CI 0.5-0.6). Of 78 pts with irAE related treatment discontinuation who were retreated with ICIs, half developed irAE recurrence.Table: 1582PICI Monotherapy (N = 328)ICI + Chemotherapy (N = 29)ICI + ICI (N = 46)All (N = 403)ICI initiation to irAE onset, days, median (IQR)84 (41-170)99 (13-140)24 (11-60)77 (33-150)Most common irAEsEndocrine115 (35%)6 (21%)16 (35%)137 (34%)Dermatologic77 (23%)7 (24%)21 (46%)105 (26%)GI81 (25%)11 (38%)25 (54%)117 (29%)Pulmonary45 (14%)3 (10%)1 (2%)49 (12%)ER/hospitalization68 (21%)10 (34%)13 (28%)91 (23%)Most common irAEs (ER/hospitalization patients)GI27 (40%)4 (40%)8 (62%)39 (43%)Pulmonary23 (34%)2 (20%)0 (0%)25 (27%)Treated with steroids188 (57%)16 (55%)32 (70%)236 (59%)Retreated with ICI62 (19%)2 (7%)14 (30%)78 (19%)Death (among patients with irAE)6 (2%)0 (0%)1 (2%)7 (2%) Open table in a new tab Pts with prompt recognition and management of irAEs experienced superior irAE resolution in a community setting.

Using machine learning on real-world data to predict metastatic status.

1550 Background: Real world data (RWD) is increasingly used to inform research, patient care, and population health in oncology; however, using RWD at scale requires accurate methods to identify clinically-relevant attributes. Metastatic status is a highly relevant clinical attribute in cancer patients but it is not routinely captured in structured formats and its determination conventionally requires review and interpretation by certified tumor registrars (CTRs). Clinical diagnoses, treatments, imaging procedures and other clinical variables documented in electronic health records (EHRs) can be used to differentiate metastatic from non-metastatic patients. This study describes an effective machine learning approach in utilizing prevalent and standardized data elements from EHRs across multiple health systems. Methods: 28,043 lung cancer and breast cancer patients from two large health systems within the Syapse Learning Health Network with data sources from CTR abstraction and EHRs were analyzed. Patients were labeled for reference metastatic status by CTRs and split into training (n = 22,434) and testing (n = 5,609) cohorts, with proportionate distribution of cancer type and metastatic status between cohorts. A regularized gradient boosting algorithm, XGBoost, was trained using over 750 variables from the patient records collected at the time of or after the initial cancer diagnosis. Results: Integration of ICD-10-CM codes with antineoplastic treatment history and radiologic imaging procedure orders achieved metastatic status prediction with increases to precision and recall in lung cancer (21% and 32% respectively) and breast cancer (39% and 9% respectively), when compared to the use of only ICD-10-CM diagnosis codes for secondary malignant neoplasms (Table). The addition of treatment and procedure data from different cancer types improved the model classification within individual cancer types. Conclusions: One of the biggest challenges in using RWD for precision oncology is identification of clinically-relevant phenotypes at scale. Here we demonstrate a scalable evidence-based method utilizing structured data for imputing metastatic status with high predictive power from two separate health systems. With further validation, this approach may be generalized to other cancer types, applied to temporal slices of data to identify changes in metastatic status, as well as provide a high-confidence designation of metastatic status for other use cases such as staging.[Table: see text]

A Question-and-Answer System to Extract Data From Free-Text Oncological Pathology Reports (CancerBERT Network): Development Study.

BackgroundInformation in pathology reports is critical for cancer care. Natural language processing (NLP) systems used to extract information from pathology reports are often narrow in scope or require extensive tuning. Consequently, there is growing interest in automated deep learning approaches. A powerful new NLP algorithm, bidirectional encoder representations from transformers (BERT), was published in late 2018. BERT set new performance standards on tasks as diverse as question answering, named entity recognition, speech recognition, and more.ObjectiveThe aim of this study is to develop a BERT-based system to automatically extract detailed tumor site and histology information from free-text oncological pathology reports.MethodsWe pursued three specific aims: extract accurate tumor site and histology descriptions from free-text pathology reports, accommodate the diverse terminology used to indicate the same pathology, and provide accurate standardized tumor site and histology codes for use by downstream applications. We first trained a base language model to comprehend the technical language in pathology reports. This involved unsupervised learning on a training corpus of 275,605 electronic pathology reports from 164,531 unique patients that included 121 million words. Next, we trained a question-and-answer (Q&A) model that connects a Q&A layer to the base pathology language model to answer pathology questions. Our Q&A system was designed to search for the answers to two predefined questions in each pathology report: What organ contains the tumor? and What is the kind of tumor or carcinoma? This involved supervised training on 8197 pathology reports, each with ground truth answers to these 2 questions determined by certified tumor registrars. The data set included 214 tumor sites and 193 histologies. The tumor site and histology phrases extracted by the Q&A model were used to predict International Classification of Diseases for Oncology, Third Edition (ICD-O-3), site and histology codes. This involved fine-tuning two additional BERT models: one to predict site codes and another to predict histology codes. Our final system includes a network of 3 BERT-based models. We call this CancerBERT network (caBERTnet). We evaluated caBERTnet using a sequestered test data set of 2050 pathology reports with ground truth answers determined by certified tumor registrars.ResultscaBERTnet’s accuracies for predicting group-level site and histology codes were 93.53% (1895/2026) and 97.6% (1993/2042), respectively. The top 5 accuracies for predicting fine-grained ICD-O-3 site and histology codes with 5 or more samples each in the training data set were 92.95% (1794/1930) and 96.01% (1853/1930), respectively.ConclusionsWe have developed an NLP system that outperforms existing algorithms at predicting ICD-O-3 codes across an extensive range of tumor sites and histologies. Our new system could help reduce treatment delays, increase enrollment in clinical trials of new therapies, and improve patient outcomes.

Abstract P2-09-05: Germline BRCA1/2 mutation testing in human epidermal growth factor receptor 2 (HER2)-negative advanced breast cancer (ABC): A real-world study in the Syapse Learning Health Network (LHN)

Abstract Background: Germline (g)BRCA1/2 mutations represent approximately 5% of metastatic breast cancer. Poly ADP-ribose polymerase inhibitors (PARPi) have shown improved clinical outcomes, a manageable toxicity profile, and favorable patient (pt)-reported outcomes versus chemotherapy in pts with gBRCA1/2 mutated HER2- locally advanced or metastatic breast cancer. With the advent of PARPi, clinical guidelines have broadened eligibility criteria for gBRCA1/2 testing. However, limited information is available on the impact of the COVID-19 pandemic on gBRCA1/2 testing rates. We assessed trends and factors associated with gBRCA1/2 testing in pts with HER2- ABC before and during the COVID-19 pandemic. Methods: This retrospective study included pts from the Syapse LHN, a longitudinal database of pts with cancer cared for in community-based, integrated care delivery networks in 25 states in the United States. Pts were eligible for gBRCA1/2 testing from initial ABC diagnosis until death or date of last contact with the participating health system. Information on gBRCA1/2 testing was obtained from scaled sources and further curated by Certified Tumor Registrars. Logistic regression evaluated the associations between age at diagnosis, family history of relevant cancer, race/ethnicity, median household income, health system, and diagnosis year with gBRCA1/2 testing among HER2- ABCs; models included hormone receptor status. Results: The study population included 1769 pts with HER2- ABC, including 577 pts with triple negative ABC initially diagnosed from 2010: 96% were women, 69% were non-Hispanic White, and 94% had an estimated median household income >$30,000 USD; median age at initial diagnosis was 61 years. The percentage of pts ever gBRCA1/2-tested among those eligible increased over time: 26%, 28%, and 31% by end of 2018, 2019, and 2020, respectively. Similarly, the percentages of new testing among eligible but not previously tested pts increased from 2018-March 2020, decreased from April-September 2020, and trended upwards thereafter (Table 1). In logistic regression models combining data from pre- and post-COVID-19 periods, family history of relevant cancer (odds ratio [OR]=1.9; 95% CI, 1.5-2.4), younger age at diagnosis (>65 reference; <45: OR=12.8, 95% CI, 8.9-18.3; 45-54: OR=6.7, 95% CI, 4.9-9.3; 55-64: OR=2.0, 95% CI, 1.5-2.8), and diagnosis year of 2013 or later (OR=1.9, 95% CI, 1.4-2.6) were significantly associated with increased odds of gBRCA1/2 testing. Positive hormone receptor status (OR=0.5; 95% CI, 0.4-0.6) and Hispanic ethnicity (OR=0.5; 95% CI, 0.3-0.9) were significantly associated with reduced odds; associations with non-Hispanic Black ethnicity did not reach statistical significance (OR=0.8; 95% CI, 0.6-1.1). Conclusion: Following the expanded eligibility criteria for gBRCA1/2 testing, testing rates increased from 2018 to 2019 and decreased only slightly during the national COVID-19 lockdown. Age at diagnosis, family history, diagnosis year, ethnicity, and hormone receptor status impacted the odds of testing. Given that gBRCA1/2 mutations are actionable, focused efforts should be developed to resume the pre-pandemic trajectory of gBRCA1/2 mutation testing. Funding: Pfizer Inc Table 1.Percentages of Newly gBRCA1/2-Tested Pts with HER2- ABC Who Were Eligible for Testing Anytime During the Time Period and Not Previously TestedTime Period2018, % (n/N)2019, % (n/N)2020, % (n/NJanuary-March2.3 (10/427)2.7 (14/512)4.3 (23/537)April-June2.9 (13/447)3.4 (18/536)2.8 (14/495)July-September4.3 (20/468)4.7 (27/569)2.5 (13/517)October-December1.9 (9/478)3.1 (17/548)3.5 (16/456)Total8.4 (52/622)10.0 (76/757)8.8 (66/747)A pt diagnosed in February and tested in April of 2018 would be included in the denominator (N) for January-March and April-June of 2018 and in the numerator (n) for April-June of 2018. Citation Format: Alexander Niyazov, Monika Izano, Colden Johanson, Sheetal Walters, Anna Berry, Bhakti Arondekar, A. Douglas Laird, Lillian Shahied Arruda, Henry Kaplan. Germline BRCA1/2 mutation testing in human epidermal growth factor receptor 2 (HER2)-negative advanced breast cancer (ABC): A real-world study in the Syapse Learning Health Network (LHN) [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-09-05.

POSA409 Somatic Gene Alteration Testing and Systemic Antineoplastic Treatment Utilization Among Patients with Advanced Non-Small Cell Lung Cancer across Large Community Health Systems in the US

Community health systems (CHSs) treat over 50% of patients with cancer across the US. The somatic gene alteration testing and systemic treatments used among patients with advanced Non-Small Cell Lung Cancer (aNSCLC) seen in these CHSs are not widely published. This analysis describes the most common gene alteration testing and systemic treatment utilization of these patients. A retrospective analysis was performed utilizing the Syapse Learning Health Network (SLHN), electronic medical record (EMR) derived database that collects data from multiple care settings that make up the US CHSs across 33 states, 450+ hospitals and 1,900+ community employed oncologists. Patients >18 yo at the time of aNSCLC diagnosis from January 1, 2017 - June 22, 2021 were identified from the SLHN using ICD-10 codes. Data for this analysis included both structured and unstructured EMR data validated by Syapse’s Certified Tumor Registrars and then descriptively summarized. 4,841 patients from the SLHN with aNSCLC (stage IIIB+) were included in the analysis. The majority were tested for PD-L1, EGFR, ALK, and ROS1 (73%, 63%, 59%, 51% of patients, respectively). BRAF, KRAS, MET and ERBB2 were tested in 50%, 45%, 43% and 41% of patients, respectively. 3,119 patients received antineoplastic systemic treatment; in the 1L setting 69% received chemotherapy, 50% received immunotherapy and 16% received targeted therapy. 61% received a carboplatin containing regimen, 47% received a pembrolizumab containing regimen, 34% received a pemetrexed containing regimen and 25% received a paclitaxel containing regimen. Broad genetic profiling is recommended for aNSCLC according to NCCN guidelines and within community health systems many patients had evidence of broad testing. Patients were most likely to have received chemotherapy, most likely a carboplatin containing regimen. Further analysis is planned to understand the correlation between testing conducted and treatment utilization within CHSs.

POSA410 Cancer Subtypes and Systemic Antineoplastic Treatment Utilization Among Patients with Advanced Breast Cancer Across Large Community Health Systems in the US

Community health systems (CHSs) play a large role in the care of patients with cancer with these systems treating over 50% of patients with cancer across the US. The cancer subtypes observed and systemic treatments used among patients with advanced breast cancer seen in these CHSs are not widely published. This analysis describes the cancer subtypes and systemic treatment utilization of these patients treated across a sample of CHSs. A retrospective analysis was performed utilizing the Syapse Learning Health Network (SLHN), electronic medical record (EMR) derived database that collects data from multiple care settings that make up the US CHSs across 33 states, 450+ hospitals and 1,900+ community employed oncologists. Patients >18 yo at advanced breast cancer diagnosis (IIIB+) from January 1, 2017 through June 22, 2021 were identified from the SLHN using ICD-10 codes. Data utilized for this analysis included both structured (set EMR fields like sex, birth date) and unstructured EMR data (physician notes) validated by Syapse’s Certified Tumor Registrars and then descriptively summarized. 1,763 patients from the SLHN with advanced disease (stage IIIB+) were included in the analysis. 54% of these patients were HR+/HER2-, 24% HR-/HER2-,12% HR+/HER2+, 7% HR-/HER2+ and 3% were unknown. 1,445 patients received systemic treatment; in the 1L setting 31% received a cyclophosphamide containing regimen, 29% a doxorubicin containing regimen, 21% for paclitaxel, 17% for letrozole, 15% for anastrozole, 15% for trastuzumab, 15% for pertuzumab, 14% for docetaxel, 14% for carboplatin and 12% for a palbociclib containing regimen. HR+/HER2- patients were the most common breast cancer subtype within this sample of CHSs patients which roughly correlates to SEER statistics of breast cancer subtypes. Patients with advanced breast cancer treated in the SLHN were most likely to receive a cyclophosphamide or a doxorubicin-containing regimen. Further analysis is planned to understand treatment patterns by cancer subtype.

Identifying Core Functions of an Evidence-Based Intervention to Improve Cancer Care Quality in Rural Hospitals.

Background:Rural patients experience worse cancer survival outcomes than urban patients despite similar incidence rates, due in part to significant barriers to accessing quality cancer care. Community hospitals in non-metropolitan/rural areas play a crucial role in providing care to patients who desire and are able to receive care locally. However, rural community hospitals typically face challenges to providing comprehensive care due to lack of resources. The University of Kentucky’s Markey Cancer Center Affiliate Network (MCCAN) is an effective complex, multi-level intervention, improving cancer care in rural/under-resourced hospitals by supporting them in achieving American College of Surgeons Commission on Cancer (CoC) standards. With the long-term goal of adapting MCCAN for other rural contexts, we aimed to identify MCCAN’s core functions (i.e., the components key to the intervention’s effectiveness/implementation) using theory-driven qualitative data research methods.Methods:We conducted eight semi-structured virtual interviews with administrators, coordinators, clinicians, and certified tumor registrars from five MCCAN affiliate hospitals that were not CoC-accredited prior to joining MCCAN. Study team members coded interview transcripts and identified themes related to how MCCAN engaged affiliate sites in improving care quality (intervention functions) and implementing CoC standards (implementation functions) and analyzed themes to identify core functions. We then mapped core functions onto existing theories of change and presented the functions to MCCAN leadership to confirm validity and completeness of the functions.Results:Intervention core functions included: providing expertise and templates for achieving accreditation, establishing a culture of quality-improvement among affiliates, and fostering a shared goal of quality care. Implementation core functions included: fostering a sense of community and partnership, building trust between affiliates and Markey, providing information and resources to increase feasibility and acceptability of meeting CoC standards, and mentoring and empowering administrators and clinicians to champion implementation.Conclusion:The MCCAN intervention presents a more equitable strategy of extending the resources and expertise of large cancer centers to assist smaller community hospitals in achieving evidence-based standards for cancer care. Using rigorous qualitative methods, we distilled this intervention into its core functions, positioning us (and others) to adapt the MCCAN intervention to address cancer disparities in other rural contexts.

POSB413 Demographic Characteristics and Disease Subtypes of Patients with Acute Myeloid Leukemia across Large Community Health Systems in the US

Community health systems (CHSs) treat over 50% of patients with cancer across the US. The profile of patients with Acute Myeloid Leukemia (AML) seen in CHSs within the US are not widely published. This analysis describes the demographic characteristics and disease subtypes of patients with AML treated across a sample of CHSs. A retrospective analysis was performed utilizing the Syapse Learning Health Network (SLHN), an electronic medical record (EMR) derived database that collects cancer care data from multiple care settings that make up the US CHSs across 33 states, 450+ hospitals and 1,900+ community employed oncologists. Patients aged >18 yo at AML diagnosis from January 1, 2010 - June 22, 2021 were identified from the SLHN using ICD-10 codes. APL was deliberately excluded. Data elements were derived from both structured and unstructured sources, via direct integrations and Certified Tumor Registrars, respectively, and are descriptively summarized. 1,001 patients from the SLHN were included in the analysis. Patient demographics included: median age of 69 years at AML diagnosis; 46% female and 54% male; 78% White, 9% Black, 2% Asian, 11% other/unknown; 4% of Hispanic/Latino ethnicity. According to WHO Classification, 36% of patients were classified at diagnosis as AML with myelodysplasia related changes, 21% were AML Not Otherwise Specified (NOS), 18% were classified as therapy-related myeloid neoplasms, and 8% were AML with recurrent genetic mutations. Patients with AML treated in CHSs within the SLHN were elderly, white and skewed slightly more male, with a relatively high percentage of patients having myelodysplastic features. Further evaluation is needed to assess whether the real world group of patients with AML and myelodysplasia has a worse outcome.

Customizable Natural Language Processing Biomarker Extraction Tool.

Natural language processing (NLP) in pathology reports to extract biomarker information is an ongoing area of research. MetaMap is a natural language processing tool developed and funded by the National Library of Medicine to map biomedical text to the Unified Medical Language System Metathesaurus by applying specific tags to clinically relevant terms. Although results are useful without additional postprocessing, these tags lack important contextual information. Our novel method takes terminology-driven semantic tags and incorporates those into a semantic frame that is task-specific to add necessary context to MetaMap. We use important contextual information to capture biomarker results to support Community Health System's use of Precision Medicine treatments for patients with cancer. For each biomarker, the name, type, numeric quantifiers, non-numeric qualifiers, and the time frame are extracted. These fields then associate biomarkers with their context in the pathology report such as test type, probe intensity, copy-number changes, and even failed results. A selection of 6,713 relevant reports contained the following standard-of-care biomarkers for metastatic breast cancer: breast cancer gene 1 and 2, estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2, and programmed death-ligand 1. The method was tested on pathology reports from the internal pathology laboratory at Henry Ford Health System. A certified tumor registrar reviewed 400 tests, which showed > 95% accuracy for all extracted biomarker types. Using this new method, it is possible to extract high-quality, contextual biomarker information, and this represents a significant advance in biomarker extraction.

Abstract P159: Molecular tumor board impact at two large health systems

Abstract Background: Molecular tumor boards (MTB) are a key component of most precision oncology programs, designed to provide a structured multidisciplinary approach to evaluating cancer patients (pts) for therapy (Tx), clinical trial (CT) enrollment, and genetic counseling (GC) services. Two health systems separately instituted MTB in 2017, and through a bio-informatics platform have been tracking the impact of these on pt care. We describe the collective experience in tracking MTB impact. Methods: This is a retrospective cohort study of pts reviewed for the first time by MTB between September 1, 2017 and September 30, 2020 at either of the two health systems (HS). Follow up data were obtained by a certified tumor registrar at a median time of 109 days after presentation. Demographics and clinical characteristics of MTB pts were obtained. Pts were identified for whom Tx or CTs were recommended and subsequently administered after MTB review, along with pts for whom GC was recommended and subsequently had genetic testing performed. Results: 351 pts were evaluated. Pt demographics included: median age of 65 years; 61% female; 83% White, 13% Black, 3% Asian, with 2% of Hispanic/Latino ethnicity. The most common primary sites observed were lung (75 pts; 21%), breast (44 pts; 13%) and central nervous system (26 pts, 7%). 334 pts (95%) had data regarding Tx and CT recommendations. 124 pts (37%) received a Tx recommendation and 36 (29%) of these received Tx. 73 pts (22%) received a CT recommendation and 4 (6%) of these pts were enrolled on CTs. 340 pts (97%) had data regarding GC recommendations. 94 pts (28%) received a GC referral and 28 (30%) of these received GC. 25 (89%) of these received a molecular test for germline cancer risk, and 10 (40%) of these had a risk mutation identified in one of the following: ATM (4), BRCA2 (2), BARD1 (1), CHEK2 (1), RAD51C (1), and RET (1). Conclusions: MTB was successful in matching pts to Tx and CT, and in providing appropriate referrals to GC and identifying germline risk mutations. The bio-informatics platform provided a uniform format and structure to collate and analyze data between programs. These parameters will serve as a baseline to monitor future MTB impact and trends as these precision oncology programs continue to grow. The relatively low percentage of patients of color whose cases were reviewed in this MTB analysis merits further understanding. Citation Format: Igor I. Rybkin, Michael A. Thompson, Frank M. Wolf, Kristen Collins, Louisa Laidlaw, Tom Mikkelsen, Jennifer Godden, Mary Walters, James L. Weese, Ronda Broome, Joe Burkhart, Veronica Jones, Chenan Zhang, Thomas D. Brown, Anna Berry. Molecular tumor board impact at two large health systems [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2021 Oct 7-10. Philadelphia (PA): AACR; Mol Cancer Ther 2021;20(12 Suppl):Abstract nr P159.

Next Generation of Central Cancer Registries.

For central cancer registries to become a more significant public health resource, they must evolve to capture more timely, accurate, and extensive data. Key stakeholders have called for a faster time to deliver work products, data extensions such as social determinants of health, and more relevant information for cancer control programs at the local level. The proposed model consists of near real-time reporting stages to replace the current time and labor-intensive efforts to populate a complete cancer case abstract on the basis of the 12- and 24-month data submission timelines. The first stage collects a cancer diagnosis minimum data set sufficient to describe population incidence and prevalence, which is then followed by a second stage capturing subsequent case updates and treatment data. A third stage procures targeted information in response to identified research projects' needs. The model also provides for further supplemental reports as may be defined to gather additional data. All stages leverage electronic health records' widespread development and the many emerging standards for data content, including national policies related to healthcare and technical standards for interoperability, such as the Fast Healthcare Interoperability Resources specifications to automate and accelerate reporting to central cancer registries. The emergence of application programming interfaces that allow for more interoperability among systems would be leveraged, leading to more efficient information sharing. Adopting this model will expedite cancer data availability to improve cancer control while supporting data integrity and flexibility in data items. It presents a long-term and feasible solution that addresses the extensive burden and unsustainable manual data collection requirements placed on Certified Tumor Registrars at disease reporting entities nationally.

Twenty-Five Years of Cancer Follow-Up; Is the Data Worth the Effort?

Substantial resources are dedicated to long-term follow-up within cancer registries; however, the completeness of these data is poorly characterized. Our objectives were to quantify long-term cancer follow-up data completeness and the effort required to collect these data using the National Cancer Database (NCDB). To quantify data completeness, patients diagnosed with cancer in 1989 were identified in the NCDB and loss to follow-up rates were assessed for 25 years after diagnosis. To quantify data collection effort, patients diagnosed from 1989 to 2014 who were alive and eligible for follow-up in 2014 were identified and the effort to perform patient follow-up was obtained via a survey of tumor registrars. The effort to perform follow-up beyond various intervals after diagnosis was calculated. In total, 484,201 patients at 958 hospitals were diagnosed with cancer in 1989. After 5 years, 6.5% of patients were lost to follow-up (13.1% of living patients), 50.3% were deceased, and 43.2% had ongoing follow-up. After 15 years, 22.9% were lost to follow-up (68.7% of living patients), 66.7% were deceased, and 10.5% had ongoing follow-up. By 25 years, loss to follow-up increased to 28.6% (93.7% of living patients), 69.5% were deceased, and 1.9% had ongoing follow-up. In 2014, 522,838h were spent performing follow-up for 2,091,353 patients at 1456 hospitals who were >15 years from their initial cancer diagnosis. While 5-year follow-up is excellent in the NCDB, loss to follow-up increases over time. The impact of curtailing data collection is under investigation and follow-up duration requirements will be re-evaluated.

More than treatment refusal: an NCDB analysis of the impact of endometrial cancer treatment refusal on racial survival disparities

<h3>Objectives:</h3> Disparities in adjuvant treatment between Black and White women with endometrial cancer (EC) are well documented and contribute to worse outcomes among Black women. However, factors leading to disparate treatment patterns are understudied. We examined whether patient refusal of adjuvant treatment (chemotherapy or radiation) differed between Black and White women and whether refusal differences mediate racial disparities in EC survival. <h3>Methods:</h3> We used the National Cancer Database (NCDB), a hospital-based registry covering ~70% of U.S. cancer diagnoses, to identify non-Hispanic Black (NHB) and non-Hispanic White (NHW) women diagnosed with EC from 2004 to 2015 who either received or refused chemotherapy or radiation. NCDB data are abstracted from medical records by tumor registrars. We used logistic regression to estimate adjusted odds ratios (aOR) of patient refusal of treatment associated with race, and accelerated failure time models to estimate differences in overall survival (OS) by race. We used causal mediation analysis to estimate how much of the difference in OS by race could be attributed to racial differences in treatment refusal. We considered the overall study populations as well as strata defined by histology, and adjusted for factors assessed at diagnosis, including age, comorbidity burden, insurance, hospital characteristics, stage, histology, and zip code-level income and education. <h3>Results:</h3> We identified 49,518 women (83% NHW, 17% NHB) who received adjuvant chemotherapy and 12,710 women with documented refusal (90% NHW, 10% NHB). After adjusting for covariates, NHB women were more likely to refuse chemotherapy than NHW women (aOR=1.12, p=0.005), particularly among those with low-grade endometrioid (aOR = 1.22, p=0.02) or serous (aOR = 1.38, p<0.001) EC. Mean OS for NHB women was 43.5 months shorter than for NHW women. Of this disparity, 0.9 months of may be attributable to differences in chemotherapy refusal (p=0.01). Among women with serous EC, NHB race was associated with 17.6 months shorter OS, with 1 month potentially attributable to refusal differences (p=0.002). We identified 66,839 women (88% NHW, 12% NHB) receiving adjuvant radiation and 4,471 (88% NHW, 12% NHB) who refused. Among women with serous EC, NHB women were more likely to refuse radiation (OR = 1.39, p=0.01), and NHB women had mean OS 13.2 months shorter than NHW women, 0.4 months of which may be attributable to differences in treatment refusal (p=0.03). <h3>Conclusions:</h3> We observed differences in treatment refusal by race, and those differences may be responsible for up to 1 month of the OS disparity between NHW and NHB women. This represents a small fraction of the dramatic racial disparity in EC survival. While a better understanding of the reasons for patient treatment refusal and subsequent intervention may help improve outcomes for some women, other causes of disparate outcomes warrant further investigation.

Abstract 2618: Staged analysis of standard of care tumor molecular testing among patients with metastatic colorectal cancer in the community health system setting

Abstract Background: National Comprehensive Cancer Network guidelines have recommended metastatic colorectal cancer patients (mCRC pts) undergo BRAF, KRAS, NRAS, and microsatellite instability/mismatch repair deficiency (MSI/MMR) testing since 2015. Previous studies have reported molecular testing rates from academic and community cancer centers, but there is a lack of information on testing rates in community health systems (HS) which account for approximately half of US cancer care. Methods: Tumor molecular testing rates were assessed in a first stage analysis for pts with mCRC in HS, in order to inform a second stage analysis focused on treatment decisions and pt outcomes. For the first stage analysis, we randomly selected 200 pts (50 pts from each of 4 HS sites) diagnosed with mCRC between 1/1/2015 and 9/1/2020. The 200 pts were identified through technology-enabled curation, followed by retrospective review performed by certified tumor registrars. An anticipated cohort of 1000 evaluable pts is planned for the second stage analysis. Results: 53% of pts were tested for all four biomarkers at some point during their care (Table). Pts more likely to have been tested include those: from HS Site 1 (P = 4.2x10-5); age &amp;lt;65 years old at mCRC diagnosis (P = 0.014); and with commercial insurance (P = 0.015). Among pts tested, the median times from metastatic diagnosis to BRAF, KRAS, and NRAS testing results were 35, 33, and 34 days, respectively. Conclusion: This first stage analysis of molecular testing in mCRC pts revealed variation in testing rates by site, biomarker, and patient characteristics, which may in part be explained by the implementation of a precision medicine testing program at HS Site 1. A second stage analysis assessing the impact of biomarker testing on treatment decisions and pt outcomes will be guided by this initial characterization of testing rates and possible confounders. Biomarker testing rates among metastatic colorectal patients across four health system sitesSite 1 (n=50)Site 2 (n=50)Site 3 (n=50)Site 4 (n=50)BRAF tested43 (86%)25 (50%)24 (48%)26 (52%)Count (%)KRAS tested44 (88%)35 (70%)32 (64%)35 (70%)Count (%)NRAS tested42 (84%)23 (46%)26 (52%)30 (60%)Count (%)MSI/MMR tested49 (98%)37 (74%)44 (88%)44 (88%)Count (%)All biomarkers tested40 (80%)19 (38%)20 (40%)26 (52%)Count (%) Citation Format: Chenan Zhang, Mahder Teka, Francesca F. Coutinho, Joseph R. Burkhart, Louise E. Widmer, Ronda G. Broome, Mary T. Tran, Michael A. Thompson, Antony M. Ruggeri, Jennifer J. Godden, James L. Weese, Douglas J. Reding, Anna B. Berry, Yanina Natanzon, Thomas D. Brown. Staged analysis of standard of care tumor molecular testing among patients with metastatic colorectal cancer in the community health system setting [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2618.