Prevention, Risk Reduction, and Hereditary Cancer

PREVENTION, RISK REDUCTION, AND HEREDITARY CANCER

10500 Oral Abstract Session

Breast cancer screening for carriers of ATM, CHEK2, and PALB2 pathogenic variants: A comparative modeling analysis.

Kathryn P. Lowry, H. Amarens Geuzinge, Natasha K. Stout, Oguzhan Alagoz, John M. Hampton, Karla Kerlikowske, Diana L. Miglioretti, Clyde Schecter, Brian L. Sprague, Amy Trentham-Dietz, Anna N.A. Tosteson, Nicolien Van Ravesteyn, Martin Yaffe, Jennifer Yeh, Fergus Couch, Peter Kraft, Eric Polley, Jeanne S. Mandelblatt, Allison W. Kurian, Mark E. Robson, Cancer Intervention and Surveillance Modeling Network (CISNET); Breast Cancer Surveillance Consortium (BCSC); Cancer Risk Estimates Related to Susceptibility (CARRIERS); University of Washington, Seattle Cancer Care Alliance, Seattle, WA; Erasmus Medical Center, Rotterdam, Netherlands; Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA; University of Wisconsin, Madison, WI; University of Wisconsin Car- bone Cancer Center, Madison, WI; UC San Francisco, San Francisco, CA; University of California, Davis, School of Medicine, Davis, CA; Department of Epidemiology and Population Health, Albert Einstein Col- lege of Medicine, Bronx, NY; University of Vermont, Burlington, VT; Geisel School of Medicine, Leba- non, NH; Odette Cancer Centre, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, Canada; Boston Children’s Hospital and Harvard Medical School, Boston, MA; Department of Labo- ratory Medicine and Pathology, Mayo Clinic, Rochester, MN; Harvard School of Public Health, Boston, MA; Mayo Clinic, Rochester, MN; Georgetown University Medical Center, Washington, DC; Stanford School of Medicine, Stanford, CA; Memorial Sloan Kettering Cancer Center, New York, NY

Background: Inherited pathogenic variants in ATM, CHEK2, and PALB2 confer moderate to high risks of breast cancer. The optimal approach to screening in these women has not been established. Methods: We used two simulation models from the Cancer Intervention and Surveillance Modeling Network (CIS- NET) and data from the Cancer Risk Estimates Related to Susceptibility consortium (CARRIERS) to project lifetime breast cancer incidence and mortality in ATM, CHEK2, and PALB2 carriers. We simulated screening with annual mammography from ages 40-74 alone and with annual magnetic resonance imaging (MRI) starting at ages 40, 35, 30, and 25. Joint and separate mammography and MRI screening performance was based on published literature. Lifetime outcomes per 1,000 women were reported as means and ranges across both models. Results: Estimated risk of breast cancer by age 80 was 22% (21- 23%) for ATM, 28% (26-30%) for CHEK2, and 40% (38-42%) for PALB2. Screening with MRI and mammography reduced breast cancer mortality by 52-60% across variants (Table). Compared to no screening, starting MRI at age 30 increased life years (LY)/1000 women by 501 (478-523) in ATM, 620 (587-652) in CHEK2, and 1,025 (998-1,051) in PALB2. Starting MRI at age 25 versus 30 gained 9-12 LY/1000 women with 517-518 additional false positive screens and 197-198 benign biopsies. Conclusions: For women with ATM, CHEK2, and PALB2 pathogenic variants, breast cancer screening with MRI and mammography halves breast cancer mortality. These mortality benefits are similar to those for MRI screening for BRCA1/2 mutation carriers and should inform practice guidelines. Research Sponsor: Breast Cancer Research Foundation, U.S. National Institutes of Health.

Life Years Gained False Positive Screens Mortality Reduction per 1000 Women per 1000 Women Mean (Range) Mean (Range) Mean (Range)

Strategy ATM CHEK2 PALB2 ATM CHEK2 PALB2 ATM CHEK2 PALB2

Mammography at 40 38% 38% 36% 291 370 621 2,224 2,174 2,092 (38-39%) (38-39%) (35-38%) (263-319) (330-409) (559-684) (2,222-2,227) (2,172-2,175) (2,085-2,099) +MRI at 40 54% 54% 52% 420 533 921 4,569 4,441 4,233 (53-54%) (53-54%) (51-53%) (388-452) (489-577) (876-967) (4,555-4,583) (4,438-4,443) (4,213-4,252) +MRI at 35 58% 57% 54% 473 591 992 5,001 4,871 4,661 (57-58%) (56-58%) (54-55%) (447-498) (555-627) (959-1,025) (4,979-5,023) (4,861-4,880) (4,635-4,688) +MRI at 30 59% 58% 55% 501 620 1,025 5,415 5,284 5,075 (58-60%) (57-60%) (55-55%) (478-523) (587-652) (998-1,051) (5,393-5,437) (5,249-5,319) (5,057-5,093) +MRI at 25 60% 59% 56% 510 630 1,037 5,932 5,802 5,592 (59-61%) (58-60%) (55-56%) (489-531) (599-661) (1,013-1,061) (5,907-5,957) (5,789-5,815) (5,563-5,621)

10501 Oral Abstract Session

Nearly half of TP53 variants are misattributed to Li-Fraumeni syndrome: A clinical evaluation of individuals with TP53 variants detected by hereditary cancer panel assays on blood or saliva.

Alison Schwartz, Sophie Hyman, Samantha Stokes, Danielle Castillo, Jeffrey N. Weitzel, Huma Q. Rana, Judy Ellen Garber; Cancer Genetics and Prevention, Dana-Farber Cancer Institute, Boston, MA; Dana-Farber Cancer Institute, Boston, MA; City of Hope Cancer Center, Duarte, CA; Oncogenomics for Precision Prevention, Sierra Madre, CA; Cancer Genetics and Prevention, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA

Background: Multigene panel testing (MGPT) has identified TP53 pathogenic or likely pathogenic (P/ LP) variants in patients with diverse phenotypes from no cancer to classic Li-Fraumeni syndrome (LFS). There is increasing recognition of variants at low allelic fraction (VAF) for TP53 in particular, which can be suggestive of post-zygotic mosaicism or aberrant clonal expansion (ACE), comprising clonal hematopoiesis of indeterminate potential (CHIP) or occult hematologic neoplasia. Distinguishing among these categories is essential because of widely different cancer risk and management implications for patients and their relatives. We report an evaluation of TP53 positive probands to determine germline versus somatic status from a cancer genetics clinic. Methods: We reviewed probands with TP53 P/LP variants by MGPT on blood (N = 83) or saliva (N = 1) samples from 2012-2019. Available VAFs were collected from commercial testing laboratories. Probands positive for a known familial variant, who met LFS testing criteria without indication of low VAF, or who carried the Brazil founder p.R337H variant were considered germline. For those with uncertain germline status, data was obtained from ancillary testing of family members, cultured skin fibroblasts, and other somatic benign or tumor tissues. TP53 variants were further categorized based on all available data. Results: Of the 84 probands, 28 (33%) had germline TP53 P/LP variants determined by above initial criteria; 18 (21%) were confirmed germline through ancillary testing. Seven (8%) individuals were classified as having constitutional mosaicism. In eleven (13%) individuals, the TP53 variants were consistent with ACE, in 9 (11%) with CHIP and in 2 (2%) with a hematologic malignancy (1 CLL, 1 NHL). Five (6%) cases could not be categorized despite ancillary testing. Fifteen (18%) probands declined any further workup. Conclusions: A TP53 P/LP variant found on peripheral blood or saliva MGPT does not always originate in the germline. In a clinical cancer genetics cohort, only 54% of patients had TP53 P/LP germline variants; these patients plus those with constitutional mosaicism (8%) require intensified surveillance. Assessment of VAF, family member testing, and analysis of TP53 in cultured fibroblasts or other tissue samples may distinguish germline and constitutional mosaicvariantsfrom the ACE spectrum. Expanding use of MGPT will increase this clinical challenge, which may motivate the modification of lab reports to include VAF and possible non- germline explanations. The findings of this study support a framework of multiple strategies to discern true constitutional status of a TP53 P/LP variant. Research Sponsor: None.

10502 Oral Abstract Session

Ancestrally unbiased polygenic breast cancer (BC) risk assessment.

Elisha Hughes, Placede Tiemeny, Shannon Gallagher, Stephanie Meek, Charis Eng, Monique Gary, Ora Gordon, Jennifer R. Klemp, Olufunmilayo I. Olopade, Holly Jane Pederson, Jeffrey N. Weitzel, Pat W. Whitworth, Lamis Yehia, Susanne Wagner, Thomas Paul Slavin, Alexander Gutin, Jerry Lanchbury; Myr- iad Genetics, Inc., Salt Lake City, UT; Genomic Medicine Institute, Cleveland Clinic Lerner Research Institute, Cleveland, OH; Grand View Health, Sellersville, PA; Providence Health & Services, Renton, WA; University of Kansas Cancer Center, Westwood, KS; Center for Clinical Cancer Genetics and Global Health, University of Chicago, Chicago, IL; Medical Breast Services, Cleveland Clinic, Cleveland, OH; Latin American School of Oncology, Sierra Madre, CA; Nashville Breast Center, Nashville, TN

Background: BC risk is influenced by single-nucleotide polymorphisms (SNPs) with small effects that can be aggregated into polygenic risk scores (PRSs). PRSs have primarily been developed and validated for populations of European descent. To make a PRS available for all women, we developed and validated a novel global PRS (gPRS) that utilizes individual ancestral genetic composition. Methods: Ancestry- specific PRSs corresponding to 3 continental ancestries were developed from 149 SNPs (93 BC and 56 ancestry-informative): an African PRS was developed using a cohort of 31,126 self-reported African American patients referred for hereditary cancer testing; an East Asian PRS was developed based on published data from the Asia Breast Cancer Consortium; and a European PRS was developed using data from the Breast Cancer Association Consortium and 24,259 European hereditary cancer testing patients. For each patient, ancestry-informative SNPs were used to calculate the fractional ancestry attributable to each of the 3 continents. The gPRS was the sum of ancestry specific PRSs weighted according to genetic ancestral composition. In an independent validation cohort (N = 62,707), we evaluated discrimination and calibration of gPRS, and compared performance against a previously described 86-SNP PRS for women of European ancestry. Associations of SNPs and PRSs with BC were analyzed using logistic regression adjusted for personal and family cancer history, age, and ancestry. Odds ratios (ORs) are reported per standard deviation within the corresponding patient population. P- values are reported as two-sided. Results: The gPRS was strongly associated with BC in the full validation cohort and in sub-cohorts defined by self-reported ancestry (Table). 95% (88/93) of BC SNPs had $1% frequency of risk alleles within each of the self-reported populations. Compared to the aforementioned 86-SNP PRS, the gPRS showed improved discrimination overall, and within each sub-cohort, with the exception of the Asian population where the sample size was too small to show superiority of either score. The 86-SNP PRS was calibrated for white non-Hispanic women but mis-calibrated for non- European ancestries. The gPRS was properly calibrated for all women. Conclusions: The 149-SNP gPRS is validated and calibrated for women of all ancestries. Combined with clinical and biological risk factors, this approach may offer improved risk stratification for all women, regardless of ancestry. Research Sponsor: Myriad Genetics, Inc.

Validation Cohort Total N N w/ BC OR (95% CI) P-value

All 62,707 15,137 1.41 (1.38 – 1.44) 2.5 x 10-212 Asian 1,325 396 1.25 (1.07 - 1.45) 3.7 x 10-03 Black/African 6,743 1,754 1.23 (1.16 - 1.31) 8.5 x 10-11 Hispanic 5,847 1,066 1.35 (1.24 - 1.46) 1.6 x 10-13 Mixed Ancestry 2,681 400 1.59 (1.39 -1.82) 2.4 x 10-12 Non-European 14,959 3,435 1.29 (1.23 –1.36) 2.5 x 10-25 White and/or Ashkenazi 42,897 10,288 1.44 (1.40 – 1.48) 6.3 x 10-172

10503 Oral Abstract Session

Adrenal-permissive HSD3B1 genetic inheritance and risk of estrogen-driven postmenopausal breast cancer.

Nima Sharifi, Megan Lynn Kruse, Jeffrey McManus, Yoon-Mi Chung, Wei Wei, Peter Bazeley, Fumihiko Nakamura, Aimalie Hardaway, Erinn Downs Kelley, Sarat Chandarlapaty, Mathew Thomas, Halle C. F. Moore, George Thomas Budd, Wai Hong Wilson Tang, Stanley L. Hazen, Aaron Bernstein, Serena Nik- Zainal, Jame Abraham; Cleveland Clinic, Cleveland, OH; Department of Hematology/Oncology, Cleve- land Clinic, Cleveland, OH; Cleveland Clinic Taussig Cancer Institute, Cleveland, OH; Cleveland Clini, Cleveland, OH; Cleveland Clinic Foundation, Cleveland, OH; Memorial Sloan Kettering Cancer Center, New York, NY; Department of Cardiovascular and Metabolic Sciences, Lerner Research Institute, Cleve- land Clinic, Cleveland, OH; Cambridge University, Cambridge, United Kingdom; University of Cam- bridge, Cambridge, United Kingdom

Background: Genetic factors that contribute to endogenous estrogen synthesis and postmenopausal breast cancer risk are unknown. We set out to test the hypothesis that homozygous inheritance of the common 1245A!C missense-encoding polymorphism in HSD3B1, which is common (8-10%) in White populations, functionally adrenal permissive and increases synthesis of the aromatase substrate, androstenedione, is associated with postmenopausal estrogen receptor-positive breast cancer. Methods: A prospective single institution study of postmenopausal estrogen receptor-driven breast cancer for determination of HSD3B1 genotype, circulating steroid concentrations, and adrenal-permissive genotype frequency compared with the genotype frequency in the general population and in estrogen receptor- negative breast cancer. Validation was performed in 2 breast cancer genomic studies with estrogen receptor documentation. The primary outcome is the adrenal-permissive genotype frequency in postmenopausal estrogen receptor-positive breast cancer and the general population. Genotype comparisons were also done with postmenopausal estrogen receptor-negative breast cancer and the association with circulating adrenal androgen concentrations determined. Results: The prospective and validation studies had 199 and 1628 women, respectively. The adrenal-permissive genotype frequency in postmenopausal White women with estrogen-driven breast cancer in the prospective cohort was 17.5% (21/120) compared with 9.6% (429/4451) in the general population [p = 0.0077]. The adrenal-permissive genotype frequency for estrogen-driven postmenopausal breast cancer was validated using the Cambridge and TCGA genomic datasets together: 14.4% (56/389) compared with 6.0% (9/149) for estrogen receptor-negative breast cancer (p = 0.007) and the general population (p = 0.005). Circulating androstenedione concentration was significantly higher for women with the adrenal-permissive genotype compared with the other genotypes (p = 0.03). Conclusions: The adrenal-permissive genotype is associated with estrogen-driven postmenopausal breast cancer. These findings link genetic inheritance of endogenous estrogen exposure to estrogen-driven breast cancer and have broad implications for risk stratification, prevention, potential biomarkers for hormonal therapy response and possibly other clinical outcomes related to estrogen physiology in postmenopausal women. Research Sponsor: U.S. Na- tional Institutes of Health.

10504 Oral Abstract Session

Underdiagnosis of germline genetic prostate cancer: Are genetic testing guidelines an aid or an impediment?

Edward D. Esplin, David J Cahn, Brian Mazzarella, Christopher Michael Pieczonka, Mukaram Gazi, Laurence H. Belkoff, Sean Heron, Rishi Modh, Aaron D. Berger, David Morris, Gautam Jayram, Charles Idom, Joseph Veys, Alexander Engelman, Richard Bevan-Thomas, Mary Kay Hardwick, Kathryn E. Hatchell, Sarah M. Nielsen, Robert Luke Nussbaum, Neal D. Shore; Invitae, San Francisco, CA; Colora- do Urology, Golden, CO; Urology Austin, Austin, TX; Associated Medical Professionals of NY PLLC, Syr- acuse, NY; University Urology Associates, Howell Township, NJ; Urologic Consultants of Southeastern Pennsylvania, Bala Cynwyd, PA; AUI Health—Advanced Urology Institute, St. Petersburg, FL; Associat- ed Urological Specialists, Chicago Ridge, IL; Urology Associates, P.C., Nashville, TN; Urology Associ- ates, Nashville, TN; North Georgia Urology, Dalton, GA; Cancer Center of South Tampa, Tampa, FL; Urology Partners, Arlington, TX; Targeted Medical Education, Cupertino, CA; Carolina Urologic Re- search Center, Myrtle Beach, SC

Background: Pathogenic/likely pathogenic (P/LP) germline genetic variants are estimated to occur in 10-15% of all prostate cancer (PCa) patients. However, genetic testing for PCa patients is underutilized, partially due to complicated and restrictive testing guidelines developed at a time when the cost of testing was high. We conducted a study based in community urology clinics to determine the incidence of P/LP variants in PCa patients who met and did not meet the NCCN 2019 PCa germline genetic testing criteria. Methods: An IRB-approved, multicenter, prospective registry was initiated with 15 community and academic urologists nationwide. Eligibility criteria included patients with a PCa diagnosis unselected for personal or family history, stage or histology who had not been previously tested. Consec- utive patients ages 18-90 were consented and underwent an 84-gene germline panel test. HIPAA-com- pliant electronic case report forms distributed to clinician collected information on patient diagnoses, NCCN testing criteria, and results-based recommendations. Results: To date, 640 enrolled patients have genetic testing results available. Overall, 69 (10.8%) patients had 72 P/LP variants detected, 15% of which were in BRCA1/2. Of the 532 patients for whom we have clinician-reported data, 293 (55%) met NCCN criteria and 239 (45%) did not. Median age was 70 (range 44-90). Overall, 11.1% (59/532) of patients with clinician-reported data had a P/LP variant. 36 (12.3%) of patients who met NCCN criteria and 23 (9.6%) of patients who did not meet criteria had a P/LP variant. The difference in P/LP rate between the two groups was not statistically significant (p=0.33). If only a conservative 12- gene PCa panel was considered, P/LP yield was 5.5% (29/532), with 8 (28%) of these patients missed by guidelines. Stratification by self-reported ethnicity was: 76% White/Caucasian (52 patients w/ P/ LP), 18% Black/African American (2 patients w/ P/LP), and <5% each of Hispanic or Asian. Conclu- sions: There was no statistically significant difference in the yield of P/LP variants between patients who met and those who did not meet NCCN PCa guidelines, reinforcing that a significant number of P/LP variants are missed if NCCN guidelines are required for genetic testing. Expanded panel testing yields more medically actionable P/LP variants than testing BRCA1/2 alone or PCa panels with 12 genes. While 18% of the cohort was Black/African American, there was a lower P/LP rate (2%) relative to other groups, indicating that more research is needed to understand genetic variation in underrepresented populations with PCa. Research Sponsor: Invitae.

10505 Oral Abstract Session

Germline mutations in DNA damage repair genes and HOXB13 among African American men diagnosed with early-onset prostate cancer.

Jennifer Lynn Beebe-Dimmer, Christopher Sample, Tara Baird, Azita Sadeghpour, Julie J. Ruterbusch, Kathleen A. Cooney; Barbara Ann Karmanos Cancer Institute/Wayne State University, Detroit, MI; Duke University Medical Center, Durham, NC; Barbara Ann Karmanos Cancer Institute, Detroit, MI; Wayne State University, Detroit, MI; Department Of Medicine, Duke University School of Medicine, Durham, NC

Background: Inherited defects in DNA damage repair (DDR) genes (e.g. ATM, BRCA1/2) and the tumor suppresser gene HOXB13 are rare in the general population, but have been observed at a higher rate among men with early-onset and advanced prostate cancer. However, most studies include few, if any, African American men, highlighting the need to understand the contribution of mutations in these genes in this high-risk population. Methods: A population-based cohort of 757 African American men diagnosed with prostate cancer at age 62 years or younger were identified and enrolled through the Metro- politan Detroit Cancer Surveillance System (MDCSS), one of NCI’s founding members of the SEER program. Participants completed a short survey to collect information on family history, medical history (including cancer-related treatment), surveillance, and health behaviors. Each participant submitted a saliva or blood sample for genetic analyses and consent for tumor tissue if available. All clinical data were collected through linkage with the MDCSS registry. Full exome sequencing was performed and herein, we report the mutation patterns observed in a panel of DNA repair genes and HOXB13. All variants were ranked according to frequency (MAF < 1%); REVEL, SIFT and PolyPhen scores for pathogenic potential; evidence from existing literature; and prevalence in the cohort. Results: Among the 744 African American prostate cancer cases with adequate DNA for sequencing and thus included in this analysis, the mean age at diagnosis was 55.6 years, 29% reported a family history of prostate cancer in a first degree family member, and 40% were initially diagnosed with intermediate- to high-risk disease (stage T3/T4 and/or Gleason 4+3 and higher at diagnosis). We identified 20 variants that were either known or predicted to be pathogenic in 11 candidate genes (ATM, ATR, BRCA1/2, BRIP1, CHEK2, FANCA, HOXB13, MSH2, PALB2, and PMS2). These particular variants were more common among men diagnosed before age 55 and in men with high grade cancer in this cohort. Conclusions: Our results suggest that mutations in DDR genes and HOXB13 may be important cancer risk factors for African American men diagnosed with early-onset and intermediate- to high-risk disease. Further study is necessary to describe the spectrum and prevalence of genetic mutations in this population including the characterization of variants of unknown significance. Research Sponsor: Department of Defense.

10506 Oral Abstract Session

Completion of lung cancer screening after a baseline order for LDCT at five diverse health systems.

Christine Neslund-Dudas, Amy Tang, Elizabeth Alleman, Jennifer Elston Lafata, Stacey A. Honda, Caryn Oshiro, Katharine A. Rendle, Anil Vachani, Oluwatosin Olaiya, Robert T. Greenlee, Michael J. Simoff, Debra P. Ritzwoller; Henry Ford Health System, Detroit, MI; Division of Pharmaceutical Out- comes and Policy, UNC Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, NC; Hawaii Permanente Medical Group and Center for Integrated Health Care Research, Kaiser Permanente, Honolulu, HI; Kaiser Permanente Center for Integrated Health Care Research, Ho- nolulu, HI; University of Pennsylvania, Philadelphia, PA; Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; Marshfield Clinic Research Institute, Marshfield, WI; Kaiser Perma- nente Colorado Institute for Health Research, Denver, CO

Background: In 2014 and 2015, the Affordable Care Act required coverage of, and CMS began reimbursing for lung cancer screening (LCS). Previous studies have shown that when new screening tests or treatments become available, disparities in disease outcomes often increase due to those with fewer resources having less access and greater barriers to care. African American men have historically had higher incidence of and death due to lung cancer than white males in the U.S., raising concerns regarding access to LCS and the potential for increases in disparities in lung cancer. We aimed to determine whether individual or neighborhood level factors were associated with completion of a baseline screening after an order for LCS low dose CT (LDCT) was placed. Methods: In a retrospective study conducted within the five health systems of the Lung Population-based Research to Optimize the Screening Pro- cess (PROSPR) Consortium, we determined adherence to baseline LDCT after a health care provider placed an order for LCS (January 2014 through June 2019). Follow-up was available through Septem- ber 2019. Patients of interest for this analysis were current or former smokers, age 55 to 80 with a 30+ pack-year smoking history. Smoking history and other individual level variables were determined through electronic medical records. Neighborhood factors were derived from the 2010 Census and multivariable logistic regression was used. Results: Of the 13,920 patients that had at least one order for a baseline LCS exam, 14.1% were non-Hispanic Black, 70.3% were non-Hispanic White, and 15.7% were of other or unknown race. Overall, 61.2% of patients completed a LDCT within 90 days and 71.9% completed a scan by the end of follow-up. Completion of a baseline scan differed by health system (LDCT at 90-days, range 51% - 84%, p<0.0001) and increased in general across scan year (range 49.1%-66.0%, p <0.001). In multivariate logistic regression models, males (aOR=1.15, 95% CI 1.07-1.23, p=<0.0001), former smokers (aOR=1.31, 95% CI 1.21-1.40, p <0.0001), and those with a prior history of any cancer (aOR=1.16, 95% CI 1.02-1.32, p=0.03) were more likely to complete LDCT. Blacks were marginally less likely to have completed a baseline LDCT (aOR=0.90, 95% CI 0.81- 1.00, p=0.06) within 90 days of an order. Sex modified the associations of race on completion of orders (p=0.08) (Black men aOR=0.81, 95% CI 0.70-0.94, p=0.006 ; Black women aOR=0.99, 95% CI 0.86-1.14, p=0.89). Conclusions: This multisite study indicates Black men in particular may have a lower likelihood of completing a baseline LCS after an order for screening is placed. As lung cancer screening programs move forward, attention should be given to factors associated with reduced uptake and adherence of screening to ensure disparities in lung cancer outcomes do not persist and increase. Provider and health system factors that may impact LCS uptake should be explored in future studies. Research Sponsor: U.S. National Institutes of Health.

10507 Oral Abstract Session

Second primary malignancies (SPM) in African American (AA) and white patients with multiple myeloma in the National Veterans Affairs (VA) healthcare system.

Sarah Premji, Cenk Yildirim, Nathanael Fillmore, Sarvari Yellapragada, Mary T. Brophy, Nhan Do, Quillan Huang, Daphne Ruth Friedman, Nikhil C. Munshi; Baylor College of Medicine, Houston, TX; Boston VA Healthcare system, Boston, MA; VA Boston Healthcare System and Dana-Farber Cancer In- stitute, Boston, MA; VA Boston Healthcare System and Boston University School of Medicine, Boston, MA; Duke University and Durham VA Medical Centers, Durham, NC; Dana-Farber Cancer Institute, VA Boston Healthcare System, and Harvard Medical School, Boston, MA

Background: In the recent decade, novel therapies have led to significant improvements in overall survival (OS) in symptomatic multiple myeloma (MM). With this increase in OS, there is an increase in the incidence of second primary malignancies (SPM) in patients with MM. This is related to multiple factors; we hypothesize that racial disparities also play a role. There is a paucity of studies with large, high quality datasets evaluating racial disparities in SPMs in MM. Our goal is to explore patterns of incidence of SPMs in US Veterans with MM, focusing on racial disparities. Methods: We conducted a retrospective cohort study based on electronic health record (EHR) and cancer registry data recorded in the VA’s Cor- porate Data Warehouse (CDW) between January 1, 1999 and January 1, 2018. We compared incidence of SPMs (defined as malignancies occurring after the diagnosis and treatment initiation of MM) using the standardized incidence rate ratio (IRR). Univariate and multivariable analyses were performed, using a Cox model adjusting for age, race, gender, treatment era, smoking status, and stage at MM diagnosis. Results: There were 8467 Veterans (97.4% male) with MM identified with self-reported race: White, 5029 (59.4%); AA, 2178 (25.7%); and other or unknown, 1260 (14.9%). At 7.5 years of follow up, 430 (5.1%) MM patients developed SPM while 8037 did not. 982 had received HSCT (Hematopoietic Stem cell transplant) of whom 65 (6.62%), a significantly higher proportion developed SPM versus those without HSCT (n = 7485) of whom 365 (4.87%) developed SPM (p = 0.024). Among those Veter- ans who had SPM 380 (88.3%) had solid tumors, 50 (11.6 %) had hematological malignancies. Of 88.3% with solid SPM the distribution was: prostate cancer, 113 (29.7%); digestive tract cancers, 63 (16.6%); lung cancer, 55 (14.4%); and GU (bladder/renal/testicular), 48(12.6%). The cumulative incidence of developing SPM increased steadily over time for the duration of this study period of 7.5 years from diagnosis. The median age of diagnosis for Whites was 68.2 years and for AA was 64.3 years, (p < 0.001) demonstrating that MM occurred earlier in AA. The hazard ratio for AA versus whites to develop SPMs was 1.21 (0.975-1.505) (p = 0.0823). However, the risk of prostate cancer was significantly higher in AA with an adjusted hazard ratio of 1.81 (1.196-2.739) (p = 0.005). No racial disparities were observed in the incidence of other types of SPMs. Conclusions: Our study suggests that the risk of SPMs does not plateau over a patient’s lifetime. HSCT was found to be an independent predictor of SPM, while smoking and agent orange were not. Our cohort is one of the largest groups of AA with MM in published literature. AA did not have a higher incidence of SPMs overall; but had a significantly higher incidence of prostate cancer than whites. We hope to develop an individualized predictive model for SPM in patients with MM. Research Sponsor: Veterans Affairs.

10508 Oral Abstract Session

Disparities in pan-cancer patients undergoing germline cancer risk assessment by self- reported race/ethnicity and ancestry.

Ying L Liu, Anna Maio, Yelena Kemel, Erin E. Salo-Mullen, Margaret Sheehan, Prince Rainier Tejada, Magan Trottier, Megan Harlan Fleischut, Alicia Latham, Maria Isabel Carlo, Michael Francis Walsh, Mark E. Robson, Diana Mandelker, Luis A. Diaz, Carol Aghajanian, Carol L. Brown, Kenneth Offit, Jada G. Hamilton, Zsofia Kinga Stadler; Memorial Sloan Kettering Cancer Center, New York, NY; Clini- cal Genetics Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY; Clinical Genetics Service, Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY

Background: Disparities in access to germline testing for cancer patients (pts) have been demonstrated; however, disparities in post-testing care are unknown. We sought to evaluate germline findings and subsequent genetic counseling/care in cancer pts undergoing tumor-germline sequencing to explore differences by self-reported ancestry. Methods: Pan-cancer pts were prospectively consented to tumor- normal sequencing via a custom NGS panel (MSK-IMPACT) from 1/2015-12/2019 inclusive of germline analysis up to 88 genes. Germline analysis was performed as a research non-billable test in 97.5% of cases. Referral to clinical genetics service (CGS) was recommended for all pts with a new positive (likely pathogenic/pathogenic) germline variant (PV). Ancestry was defined using self-reported Federal definitions of race/ethnicity and designations of Ashkenazi Jewish (AJ) ancestry. Pts were categorized into mutually exclusive groups: AJ, White, Non-White (Asian, Black/African American, Hispanic, Other), and unknown. All pts self-identifying as Hispanic were classified as such, regardless of race. Abstracted data on germline findings and downstream CGS follow-up were compared across groups using non-parametric statistical tests. Results: Among the 15,775 pts in this cohort (59.6% White, 15.7% AJ, 20.5% Non-White [6.9% Asian, 6.8% Black, 6.7% Hispanic, 0.1% Other], and 4.2% unknown), 2663 (17%) had a PV. AJ pts had the highest rates of PV (n = 683, 27.6%), and Non-White pts had a lower proportion of PV (n = 433, 13.6%) compared to Whites (n = 1451, 15.5%), p < 0.01, with differences mostly due to increased prevalence of moderate/low penetrance variants in White and AJ pts. These findings were consistent across multiple tumor types. Prior knowledge of the PV (424/2663, 16%) was more common in Non-White (19.9%) and AJ (19.2%) than White pts (13.4%), p < 0.01. Among 2239 pts with new PV, all were referred to CGS, and 1652 (73.8%) pts were seen. Non-White pts had lower rates of completing visits (67.7%) than White (73.7%) and AJ pts (78.8%), p < 0.01, with the lowest rates occurring in Black (63%) and Hispanic (68.1%) pts. All pts without a visit (n = 587) received a close out letter including 139 pts (6.2% of pts with new PV) who had no documentation of receipt of results in the medical record. Higher rates of non-disclosure were observed in Non-White (6.7%) compared to White (5.4%) and AJ (3.4%) pts with new PV, p = 0.032; non-disclosure did not vary by gene penetrance. There was a non-significant trend towards lower rates of cascade testing at CGS in Asian and Black pts with ongoing analysis. Conclusions: Even when traditional barriers to genetic testing were min- imized, Non-White pts were less likely to receive recommended cancer genetics follow-up for subsequent cancer risk counseling, with potential implications for oncological care, cancer risk reduction, and at-risk family members. Research Sponsor: U.S. National Institutes of Health, This work was supported by the Robert and Kate Niehaus Center for Inherited Cancer Genomics. MSKCC is supported by the NCI Core grant P30 CA008748.

10510 Poster Discussion Session

Clinically sufficient vitamin D levels at breast cancer diagnosis and survival outcomes in a prospective cohort of 3,995 patients after a median follow-up of 10 years.

Song Yao, Haiyang Sheng, Marilyn L. Kwan, Qianqian Zhu, Janise M. Roh, Lia D’addario, Isaac J. Ergas, Emily Schultz, Ting-Yuan David Cheng, Warren Davis, Catherine Thomsen, Scarlett Lin Gomez, Christine B. Ambrosone, Lawrence H. Kushi; Roswell Park Comprehensive Cancer Center, Buffalo, NY; Kaiser Permanente Northern California, Division of Research, Oakland, CA; University of Florida, Gainesville, FL; Zero Breast Cancer, Oakland, CA; University of California, San Francisco, San Francis- co, CA

Background: There have been suggestive findings for better cancer survival with vitamin D supplementation in the recent VITAL trial. The findings are consistent with meta-analyses based on earlier randomized trials testing daily supplement vitamin D intake. As there is no ongoing or planned randomized trial of vitamin D supplementation in sight for women after breast cancer diagnosis, we evaluated relationships between serum levels of vitamin D and breast cancer outcomes in a large prospective cohort of breast cancer survivors. Methods: We measured 25-hydroxyvitamin D (25OHD) levels in serum samples collected at the time of diagnosis from 3,995 women with incident breast cancer enrolled in the Path- ways Study, a large prospective cohort established in 2006 at Kaiser Permanente Northern California with active follow-up (FU). Potential determinants of 25OHD levels, including a polygenic score, were examined. Vitamin D levels were categorized based on clinical cutoffs as deficient ( < 20 ng/ml), insuf- ficient (20 to < 30 ng/ml), or sufficient ($30 ng/ml). These levels were then evaluated in relation to overall survival (OS), breast cancer-specific survival (BCSS), recurrence-free survival (RFS), and invasive disease-free survival (IDFS). Cox proportional hazards models were adjusted for non-clinical, clinical, and treatment factors and were further stratified by stage, estrogen receptor (ER) status, and body mass index (BMI). Results: Vitamin D supplement use, lower BMI, and self-reported white race were the strongest determinants of higher 25OHD levels. The polygenic score was significantly associated with 25OHD levels but explained only 0.3% of the variance. The median FU was 9.6 years (range: 0.3-13). Compared to those with deficient vitamin D levels, patients with sufficient levels had significantly better survival outcomes, which remained after controlling for various covariates (OS: HR [95% CI] = 0.73 [0.58-0.91]; BCSS: HR = 0.78 [0.56-1.09]; RFS: HR = 0.79 [0.65-0.97]; IDFS: HR = 0.82 [0.68- 0.99]). Associations were similar by ER status, but stronger among patients with more advanced stage disease and those with under-weight or normal BMI. Black patients had the lowest 25OHD levels, which contributed to their poorer survival compared to white patients. Adding 25OHD levels to the Cox model of OS lowered the HR associated with Black vs. white race from 2.03 (1.57-2.62) to 1.79 (1.37-2.32). Conclusions: Sufficient vitamin D levels at the time of diagnosis were associated with improved breast cancer prognosis. Consistent with results from randomized trials, our findings from a large observational cohort of breast cancer survivors with long FU provide the strongest evidence to date for maintaining sufficient vitamin D levels in breast cancer patients, including among Black women and those with more advanced stage disease. Research Sponsor: U.S. National Institutes of Health.

10511 Poster Discussion Session

Dietary intervention influence on physical activity in the Women’s Health Initiative randomized Dietary Modification trial.

Kathy Pan, Aaron K. Aragaki, Yvonne Michael, Cynthia A. Thomson, Aladdin H. Shadyab, Linda Snetselaar, Jean Wactawski-Wende, David O. Garcia, Nazmus Saquib, Rowan T. Chlebowski; Lundquist Institute for Biomedical Innovation at Harbor-UCLA Medical Center, Torrance, CA; Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA; Drexel University, Philadel- phia, PA; University of Arizona, Tucson, AZ; University of California San Diego, La Jolla, CA; University of Iowa, Iowa City, IA; University of Buffalo, Buffalo, NY; Sulaiman AlRajhi University, Al-Bukairiah, Saudi Arabia

Background: In the Women’s Health Initiative (WHI) Dietary Modification (DM) randomized trial, after 8.5 years dietary intervention and 19.5 years cumulative (median) follow-up, dietary intervention participation was associated with a statistically significantly 22% lower breast cancer mortality (P = 0.02). In observational studies, physical activity has been associated with lower breast cancer risk with emerg- ing results now indicating, compared to inactivity, any increase in physical activity has health benefits. Currently, longitudinal data on whether an intervention targeting dietary change influences other health-related behaviors as a gateway effect is limited. To evaluate whether randomization to a dietary intervention was associated with self-directed change in physical activity. Methods: In the WHI DM trial, 48,835 postmenopausal women, ages 50-79 years, with no prior breast cancer and baseline normal mammogram were randomized at 40 US clinical centers to a dietary intervention (19,541) or a comparison group. Dietary goals were to reduce fat intake to 20% of energy and increase intake of vegetable, fruit, and grains addressed in 18 group sessions in year 1 then quarterly. Neither randomization group received specific or ongoing instructions to increase physical activity, but physical activity was refer- enced in written materials given to the intervention groups in 7 of the 56 sessions. Episodes per week of moderate or vigorous recreational physical activity (MVPA) were collected at baseline and serially through 15.9 years follow-up by self-report questionnaire. Marginal longitudinal logistic regression models were used to assess physically inactive (MVPA = 0) or physically active (MVPA > 0) participants by randomization group. Marginal Poisson regression models estimated mean weekly MVPA by randomization group. Results: 45.6% of participants reported 0 MVPA at baseline which largely persisted throughout follow-up. During cumulative follow-up, relative to the comparison group, dietary intervention group participation was associated with 7% lower physical inactivity rate (odds ratio [OR] 0.93 95% confidence interval [CI] 0.91, 0.95, P < 0.001) and a 4% higher mean MVPA (ratio of means [RM] 1.04 95% CI 1.02, 1.06, P < 0.001). The association between dietary intervention participation with higher physical activity level was stronger with increasing BMI (P-interaction 0.01) and for women with waist circumference $ 88 cm (P-interaction 0.02). Conclusions: In conclusion, in a randomized trial setting, a low-fat dietary pattern intervention was associated with a significantly lower physical inactivity rate and significantly higher moderate and vigorous physical activity level which could be associated with health benefits. Clinical trial information: NCT00000611. Research Sponsor: U.S. Na- tional Institutes of Health.

10512 Poster Discussion Session

Analysis of real-world (RW) data for metastatic breast cancer (mBC) patients (pts) with somatic BRCA1/2 (sBRCA) or other homologous recombination (HR)-pathway gene mutations (muts) treated with PARP inhibitors (PARPi).

Felipe Batalini, Russell Madison, Dean C. Pavlick, Ethan Sokol, Tamara Snow, Arjun Sondhi, Garrett M. Frampton, Colby Jenkins, Judy Ellen Garber, Gerburg M. Wulf, Jeffrey Michael Venstrom, Nadine M. Tung, Emily Castellanos, Alexa Betzig Schrock, Kimberly McGregor; Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA; Foundation Medicine, Inc., Cambridge, MA; Cancer Ge- nomics Research, Foundation Medicine, Cambridge, MA; Flatiron Health, New York, NY; Dana-Farber Cancer Institute, Boston, MA; Cancer Genetics and Prevention, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA; Beth Israel Deaconess Medical Center, Boston, MA; Foundation Medi- cine, Cambridge, MA

Background: PARPi are approved for treatment of pts w/ HER2-negative mBC and germline BRCA1/2 (gBRCA) pathogenic or likely pathogenic variants (muts); however, clinical benefit has also been demonstrated in mBC pts w/ sBRCA or other HR-pathway gene muts. Using a RW Clinico-Genomic Database (CGDB), we assessed outcomes for pts w/ gBRCA muts compared to pts w/ either sBRCA or other HR- pathway muts treated w/ PARPi. Methods: 6,329 mBC pts from ~280 US cancer clinics were included in the Flatiron Health (FH) -Foundation Medicine (FM) CGDB, which includes comprehensive genomic profiling (CGP) linked to de-identified, electronic health record (EHR)-derived clinical data. Eligible pts had mBC, received care in the FH network from 1/1/2011-9/1/2020, and had tissue CGP by FM. Pts classified as gBRCA: positive germline result in EHR and BRCA mut predicted germline per FM’s somatic, germline, zygosity algorithm (SGZ) (Sun et al PMID 29415044). Non-gBRCA: negative germline results in EHR and a somatic BRCA (sBRCA) mut per SGZ or BRCA wild-type w/ another HR mut per CGP result. Pts w/o a documented gBRCA result in EHR, unknown FM BRCA SGZ result, or conflicting results were excluded. RW overall survival (rwOS) and RW progression-free survival (rwPFS) from start of PARPi for pts w/ gBRCA and non-gBRCA mBC were compared using Kaplan-Meier analysis and Cox regression adjusted for mBC line number, prior platinum, age at PARPi initiation, race, and receptor status. Results: Among pts who received PARPi in the mBC setting, 44 had gBRCA and 18 had non- gBRCA: 9 sBRCA (5 BRCA1, 4 BRCA2), 4 PALB2, 2 ATM, and 1 each of ATM+CDK12, BARD1+FANC- F+RAD54L, and CHEK2. Of HR muts 76% were confirmed biallelic: 33/44 gBRCA (11 unknown), 8/9 sBRCA, 3/4 PALB2, and 3/5 other (1 unknown). Neither median rwPFS nor rwOS from start of PARPi were significantly different between the non-gBRCA and gBRCA cohorts (rwPFS: 7.0 [4.6-11.3] vs 5.5 [4.3-7.2] months (mos), aHR: 1.19 [0.57 – 2.43]; rwOS: 15.0 [7.95-16.3] vs 11.5 [9.46-NA] mos, aHR: 0.85 [0.36-1.98]). For 9 pts w/ sBRCA mut, median rwPFS was 7.1 mos (range 1.4-12.4) and all pts had progressed by data cut off. Conclusions: Despite small pt numbers and limitations from RW data, our results suggest that pts w/ biallelic non-gBRCA mBC may derive similar benefit from PARPi when tumor CGP detects a sBRCA mut or germline or somatic mut in other HR-pathway genes. These findings are consistent w/ the results from TBCRC-048 (Tung et al PMID 33119476) and support further randomized trials exploring the efficacy of PARPi in this population. Research Sponsor: Foundation Medicine.

gBRCA (n = 44) Non-gBRCA (n = 18)

Median age at PARPi, yrs 55.5 64.5 Community practice, n (%) 41 (93) 18 (100) Line Number, n (%) 1 2 (5) 4 (22) 2 11 (25) 3 (17) 3 7 (16) 4 (22) 4+ 24 (55) 7 (39) Prior platinum, n (%) 18 (41) 5 (28) Caucasian white, n (%) 33 (75) 12 (67) Combo therapy, n (%) 7 (16) 5 (28)

10513 Poster Discussion Session

The marginal diagnostic benefit of pancreatic cancer molecular profiling after germline testing.

Evan Justin Walker, Amie Blanco, Julia C. Carnevale, Pelin Cinar, Eric Andrew Collisson, Margaret A. Tempero, Andrew H. Ko; University of California San Francisco, San Francisco, CA; University of Cali- fornia San Francisco, Helen Diller Family Comprehensive Cancer Center, San Francisco, CA; University of California San Francisco Helen Diller Family Comprehensive Cancer Center, San Francisco, CA

Background: Germline genetic testing is now universally recommended for patients (pts) with pancreatic ductal adenocarcinoma (PDAC) for purposes of both familial screening and therapeutic guidance. Treat- ment selection can be further informed by tumor molecular profiling (TMP) to identify targetable somatic alterations in pts with advanced disease, but this is inconsistently applied. Determination of the rate of actionable findings identified with TMP after germline testing, which we term marginal diagnostic benefit, may inform practice patterns and workflow in this patient population. Methods: This retrospective analysis included all pts with PDAC who underwent germline testing and TMP at UCSF over a 4-yr period. Medical records were reviewed for demographics, disease-specific data, and germline testing/ TMP clinical reports. Alterations classified as ‘pathogenic’ or ‘likely pathogenic’ were included, and were deemed ‘actionable’ if there was clinical or preclinical evidence of benefit from targeted therapy in any cancer, as previously described. Results: From 1/2016-1/2020, 144/738 (20%) UCSF pts with PDAC completed both germline testing and TMP. Germline testing identified actionable pathogenic alterations in 10 (7%). TMP confirmed 8/10 of these alterations and identified 3 additional therapeutic targets. Among the 134 pts without actionable germline findings, TMP identified 45 new therapeutic targets in 41 (31%) pts, increasing the overall rate of actionable findings from 7% to 35%. Most (35/ 58, 60%) actionable alterations involved genes associated with the Homologous Recombination DNA Damage Repair (HR-DDR) pathway (Table). 80% of pts with HR-DDR pathway alterations (9/10 germline, 19/25 somatic) received platinum-based chemotherapy. Four pts were treated with targeted therapy based on test results: PARP-inhibitor (n = 2, germline BRCA1 and PALB2 mutations), PARP- inhibitor + ATR inhibitor (n = 1, somatic ARID1A mutation) and mTOR inhibitor (n = 1, somatic STK11 deletion). Conclusions: In this analysis, PDAC TMP after germline testing increased the detection of actionable alterations (the marginal diagnostic benefit) by five-fold. As more somatic tumor alterations become actionable with the development of targeted therapeutics, TMP is a necessary complement to germline testing to fully inform personalized treatment decisions for all pts with PDAC. Research Spon- sor: None.

Detected by Germline Additional Alterations Detected Targetable Molecular Findings Testing (n = 10) by TMP (n = 48)

HR-DDR Pathway Alterations* 10 (100%) 25 (52%) PI3K/AKT/mTOR Pathway Alterations – 8 (17%) CDK4/6 Amplifications – 4 (8%) KRAS G12C Mutation – 4 (8%) HER2 Amplification – 3 (6%) BRAF Activating Mutation – 1 (2%) RET Activating Mutation – 1 (2%) EGFR Activating Mutation – 1 (2%) NOTCH2 Amplification – 1 (2%)

*ARID1A (15), ATM (6), BRCA1 (5), CHEK2 (3), BRCA2 (1), BRIP1 (1), FANCA (1), MRE11A (1), PALB2 (1), RAD51C (1).

10514 Poster Discussion Session

Clinical impact of medical policy supporting universal germline testing for patients with colorectal cancer.

Sarah M. Nielsen, Joline Dalton, Kathryn E. Hatchell, Stacey DaCosta Byfield, Chad Moretz, John Buresh, Taryn Hall, Robert Luke Nussbaum, Edward D. Esplin; Invitae, San Francisco, CA; OptumLabs, Minnetonka, MN

Background: Colorectal cancer (CRC) affects approximately 104,000 patients (pts) annually in the Unit- ed States, up to 45% of which are estimated to be genetic and/or familial. Aligned with clinical guidelines, in 2020, a large U.S. insurer established Medical Policy allowing for and reimbursing germline genetic testing (GGT) for all CRC pts. This study reports overall uptake of GGT in CRC pts under this inclusive policy, actionable findings and treatment implications for pts tested, stratified by self-reported ancestry/ethnicity. Methods: Two independent de-identified datasets were reviewed, including administrative claims data of commercially insured and Medicare Advantage enrollees, aged 18+ with CRC ($1 claim with ICD10 C18, C19 or C20 in the first position) who were continuously enrolled (CE) in the health plan from 1/2019-10/2020. Evidence of genetic testing based on CPT codes, was examined during 2020. A second de-identified dataset of CRC pts whose GGT was billed to the insurer under the Medical Policy, was also reviewed. Patient demographics, clinical information and GGT results were de- scriptively analyzed. Results: Of the >18,000,000 CE enrollees, 55,595 were identified as CRC pts, of whom 1,675 (3%) received GGT. From the GGT dataset, 788 pts had test results available for review. 143 (18%) pts had pathogenic/likely pathogenic (P/LP) variants in genes including MSH2,MLH1, PMS2, MSH6, CHEK2, APC, BRCA2, ATM, MUTYH (biallelic). Of pts with P/LP variants, 96 (67%) were potentially eligible for precision therapy and/or clinical treatment trials. Overall, 133 (93%) had P/ LP variants in genes with precision therapy, clinical trial and/or published management implications. In a subset of pts (n=674) with ethnicity data; Asian, Black/African-American and Hispanic pts showed lower relative uptake of germline testing than Caucasians (Table). Conclusions: Despite Medical Policy allowing for GGT for all pts with CRC, only 3% of eligible pts received testing. If all CRC pts had been tested, these data suggest up to 6,705 pts with P/LP variants conferring potential eligibility for precision therapy (PD-1/PD-L1 inhibitors) or clinical treatment trials (PARP inhibitors), and an additional 2,602 pts with mutations in genes with published management recommendations, could have been identified, but were missed. Additional research is needed to identify obstacles to systematic implementation of this Medical Policy, the best timing of GGT to prevent CRC and improve access to underrepresented populations. CRC patients with germline genetic testing. Research Sponsor: None.

Patients P/LP Negative VUS Self-reported ancestry/ethnicity N, % N, % N, % N, %

Asian 28, 4% 6, 21% 11, 39% 11, 39% African-American 65, 9% 15, 23% 26, 40% 24, 36% Hispanic 60, 8% 6, 10% 28, 46% 26, 43% Caucasian 521, 77% 99, 19% 241, 46% 181, 34% Total 674 126,18% 306, 45% 242, 35%

VUS - variant of uncertain significance.

10515 Poster Discussion Session

Limitations of direct-to-consumer (DTC) genetic testing for hereditary breast and ovarian cancer.

Neelam Vijay Desai, Elizabeth Dominic Barrows, Sarah M. Nielsen, Kathryn E. Hatchell, Edward D. Esplin, Nadine M. Tung, Claudine Isaacs; Beth Israel Deaconess Medical Center, Boston, MA; George- town Lombardi Comprehensive Cancer Center, Washington, DC; Invitae, San Francisco, CA; Stanford Univ Medcl Ctr, Stanford, CA; Georgetown Lombardi Comprehensive Cancer Center, Georgetown Uni- versity Medical Center, Washington, DC

Background: With the advent of DTC genetic testing, individuals have access to genetic testing without input from a healthcare professional. DTC testing now exists for the 3 Ashkenazi Jewish (AJ) BRCA1/2 founder variants. DTC testing may provide false reassurance to individuals that they do not carry a pathogenic or likely pathogenic variant (PLPV) in BRCA1/2 or other cancer-risk genes. Methods: Multi-panel genetic testing was performed in 348,692 individuals for a clinical indication of hereditary breast/ovarian cancer (Clinical cohort) and 7,636 self-referred ostensibly healthy individuals (Healthy cohort) by a clinical testing laboratory. The primary analysis evaluated PLPVs for Group 1 genes: BRCA1/2 AJ founder variants and Group 2: full sequence BRCA1/2. Secondary analyses assessed PLPVs in Group 3: high-risk breast cancer genes (BRCA1/2, CDH1, PALB2, PTEN, STK11, TP53), Group 4: all breast or ovarian cancer-risk genes (Group 3 genes plus ATM, BARD1, BRIP1, truncating CHEK2, EPCAM, MLH1, MSH2/6, NF1, PMS2, RAD51C/D) and Group 5: 41 cancer-risk genes; these analyses were limited to participants who tested for all 41 genes. Potentially mosaic variants were excluded. Results: Ta- ble illustrates PLPVs found in both cohorts. The BRCA1/2 AJ founder variants account for only ~11% (1513/13,987) and ~30% (19/64) of the BRCA PLPVs in the Clinical and Healthy cohorts, respectively. Even among AJ individuals, testing only for the 3 founder variants will miss ~10% (52/513) of all BRCA1/2 PLPVs. Evaluating only the BRCAAJ founder variants missed a higher percentage of PLPVs in other cancer-risk genes. Conclusions: The 3 BRCA1/2 AJ founder variants analyzed by DTC testing account for a small fraction of PLPVs in cancer-risk genes in the general population, and miss 10% of BRCA PLPVs even among AJ individuals. Greater public education is needed to dispel the misconcep- tion that DTC tests are equivalent to clinical assessment and comprehensive genetic testing. PLPVs identified in Clinical and Healthy Cohorts. Research Sponsor: None.

Primary Analysis Secondary Analyses (N=356,328) (N=83,101) Group 1 Group 2 Group 3 Group 4 AJ Group 2 % Founder / Group 1 Full Hi-risk Breast/ Group 5 Founder Full BRCA1/2 all BRCA1/2 AJ Founder BRCA1/2 breast ovarian 41 genes Cohort [N,%] [N,%] PLPV Cohort [N, (%)] [N, (%)] [N, (%)] [N, (%)] [N, (%)]

Healthy 19 (0.2%) 64 (0.8%) 29.7% Healthy 12 (0.2%) 46 (0.8%) 60 (1.0%) 137 (2.4%) 195 (3.4%) N=7636 N=5792 (5% AJ) Clinical 1,513 13,987 10.8% Clinical 367 2,721 3,489 6,069 6,749 N=348,692 (0.4%) (4.0%) 89.9% N=77,309 (0.5%) (3.5%) (4.5%) (7.8%) (8.7%) AJ (alone) 461 (4.4%) 513 (4.9%) 69.6% AJ (mixed) 218 (3.3%) 313 (4.7%) 1.0% Asian 10 (0.06%) 992 (6.3%) 0.5% Black 6 (0.02%) 1,118 (4.4%) 4.8% Hispanic 74 (0.3%) 1,542 (6.4%) 7.9% White 558 (0.3%) 7,028 (3.3%) 7.5% Other 186 (0.4%) 2,481 (4.9%)

10516 Poster Discussion Session

Early age of onset and broad cancer spectrum persist in MSH6 and PMS2-associated Lynch syndrome.

Ying L Liu, Karen Anne Cadoo, Anna Maio, Zalak Patel, Yelena Kemel, Erin E. Salo-Mullen, Amanda Catchings, Megha Ranganathan, Sarah R. Kane, Jinru Shia, Robert A. Soslow, Ozge Birsoy, Diana Mandelker, Jaclyn Frances Hechtman, Maria Isabel Carlo, Michael Francis Walsh, Arnold Markowitz, Kenneth Offit, Zsofia Kinga Stadler, Alicia Latham; Memorial Sloan Kettering Cancer Center, New York, NY; Clinical Genetics Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY; Memorial Sloan-Kettering Cancer Center, New York, NY

Background: Recently updated NCCN guidelines for management of germline MSH6 and PMS2 pathogenic/likely pathogenic (P/LP) variant carriers suggest a more modest phenotype with later onset colorectal cancer (CRC) and limited extra-colonic cancers compared to other Lynch Syndrome (LS) genes. However, data are limited, and a comprehensive understanding of the risk spectrum and age of cancer onset is critical for cancer screening and risk-reduction. We sought to characterize MSH6 and PMS2-associated cancers and age of diagnosis in those with mismatch repair deficient (MMRD)/ microsatellite instability-high (MSI-H) tumors, a hallmark of LS pan-cancer. Methods: Pts consented to IRB-approved protocols of tumor/germline sequencing or to a prospective registry of LS pts at a single institution from 2/2005-01/2021 were reviewed to identify those with germline heterozygous MSH6/PMS2 P/LP variants; pts with constitutional MMRD (CMMRD) were excluded. In cancer-affected pts, tumors were evaluated for MSI and/or MMR protein expression via immunohistochemistry. Tumor types were tabulated, and clinical variables were correlated with MMR/MSI status using non-parametric tests. Results: We identified 243 pts (159 female, 94 male) with P/LP germline MSH6/PMS2 variants, and 186 (77%) pts had a confirmed cancer [MSH6 111/148 (75%); PMS2 75/95 (79%)]. Overall, 51 (21%) pts had multiple primary cancers, 35 (24%) in MSH6 and 16 (17%) in PMS2 (p = 0.20), resulting in 261 total tumors, 160 in MSH6 and 101 in PMS2. Of the 191 tumors with molecular assessment, 118 (62%) were MMRD/MSI-H, including CRC (n = 54), endometrial (EC, n = 34), small bowel (SBA, n = 6), ovarian (OC, n = 5), urothelial (n = 5), pancreato-biliary (n = 4), gastroesophageal (n = 3), non-melanoma skin (n = 3), prostate (n = 2), breast (n = 1), and brain (n = 1). While CRC and EC were more likely to be MMRD/MSI-H (79% each) compared to other cancers (37%) (p < 0.001 overall, p = 0.001 for MSH6, and p < 0.001 for PMS2), 25% of all MMRD/MSI-H tumors in both genes were comprised of non-CRC/ EC cancers. Notably, there were 6 SBAs (5 in PMS2, 1 in MSH6), and all were MMRD/MSI-H. There were 17 OCs (12 in PMS2, 5 in MSH6), and of the 12 that underwent molecular assessment, 5 (41.7%) were MMRD/MSI-H (3 PMS2, 2 MSH6). Among MMRD/MSI-H CRC and EC, median age of diagnosis was 51.5 (range 27-80) and 55 (range 39-74) respectively, with 9/54 (17%) of CRC (4 in MSH6, 5 in PMS2) diagnosed < age 35, the suggested upper threshold for initiation of colonoscopy per NCCN. Conclusions: Despite being lower penetrance LS-associated genes, pts with MSH6/PMS2 P/LP variants remain at risk for a broad-spectrum of cancers and very early-onset CRC, with 17% of MMRD/ MSI-H CRC presenting prior to upper threshold of initiation of colonoscopic screening per NCCN. Re- search Sponsor: U.S. National Institutes of Health, Other Foundation.

10517 Poster Discussion Session

Cancer-specific mortality associated with germline genetic testing results among women with breast cancer or ovarian cancer treated with chemotherapy.

Allison W. Kurian, Paul Abrahamse, Ann S. Hamilton, Dennis Deapen, Scarlett Lin Gomez, Monica Morrow, Jonathan S. Berek, Steven J. Katz, Kevin C. Ward; Stanford Cancer Institute, Stanford Univer- sity School of Medicine, Stanford, CA; University of Michigan, Ann Arbor, MI; Department of Preventa- tive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA; University of Southern California, Los Angeles, CA; University of California, San Francisco, San Francisco, CA; Breast Service, Department of Surgery, Memorial Sloan Kettering Cancer Center, New York, NY; Stan- ford Women’s Cancer Center, Stanford Cancer Institute, Stanford, CA; Emory University, Atlanta, GA

Background: Breast and ovarian cancer patients increasingly undergo germline genetic testing. While studies suggest a greater chemotherapy benefit for carriers of BRCA1/2 pathogenic variants, little is known about whether pathogenic variants in other genes are associated with cancer mortality. Methods: Georgia and California Surveillance, Epidemiology and End Results (SEER) registry records of women diagnosed with breast cancer or ovarian cancer from 2013-2017 were linked to results of clinical germline genetic testing from four participating laboratories. Patients were included if they linked to a genetic result, had stages I-III breast cancer or I-IV epithelial ovarian cancer and received chemotherapy. Multivariable Cox proportional hazard models were used to examine the association of genetic results, demographic and clinical factors with cancer-specific mortality. Results: 21,348 breast and 4,320 ovarian cancer patients were analyzed with median follow-up of 41 months. Pathogenic variants were present in 12% of patients with estrogen and progesterone receptor-positive, HER2-negative breast cancer, 9% with HER2-positive breast cancer, 17% with triple-negative breast cancer and 18% with ovarian cancer. Pathogenic variants were most common in BRCA1/2, CHEK2, PALB2, ATM and BRIP1. Among triple-negative breast cancer patients, mortality was lower with pathogenic variants in BRCA1 (hazard ratio (HR) 0.27, 95% confidence interval (CI) 0.17-0.45) and genes other than BRCA1/2 (HR 0.33, CI 0.13-0.81) versus no pathogenic variant. Genetic results were not associated with mortality in other breast cancer subtypes. Among ovarian cancer patients, mortality was lower with pathogenic variants in BRCA2 (HR 0.36, CI 0.26-0.49) and in genes other than BRCA1/2 (HR 0.48, CI 0.33-0.70). Conclusions: Among breast and ovarian cancer patients treated with chemotherapy, those with germline pathogenic variants in several cancer-associated genes had equivalent or lower short- term mortality than those testing negative. These results may guide patient counseling and clinical trial design. Research Sponsor: U.S. National Institutes of Health.

10518 Poster Discussion Session

Chemoprotective effect of metformin against HR+/HER2- breast cancer among women with type-2 diabetes.

Soumya Chikermane, Susan M. Abughosh, Manvi Sharma, Meghna V. Trivedi, Rajender R. Aparasu, Michael L. Johnson; University of Houston, College of Pharmacy, Houston, TX; The University of Missis- sippi, School of Pharmacy, Oxford, MS

Background: Type-2 diabetes mellitus (T2DM) increases the risk of breast cancer among postmenopausal women. Metformin has demonstrated a chemoprotective effect in breast cancer, however its role in HR+/HER2- breast cancer (HR+/HER2- BC), the most common subtype, has not been studied among older women with T2DM in the United States. This study evaluated if increased exposure to metformin is associated with a reduced risk of HR+/HER2- BC among postmenopausal women with T2DM. Meth- ods: A case-control study was performed using the Surveillance, Epidemiology, and End Results (SEER)-Medicare data (2008-2015). Those diagnosed with HR+/HER2- BC as their first/only cancer after incident T2DM diagnosis were cases. The event date was the date of HR+/HER2- BC diagnosis in cases, and randomly assigned to non-cancer T2DM controls based on the distribution in cases. Cases were matched to up to 4 controls each using incidence density sampling with replacement. Metformin exposures were defined as cumulative dose, average intensity and adherence, measured during the 1- year lookback period prior to the event date. Dose (mg) was categorized as: 0, 0-30,000, 30,001- 136,000, 136,001-293,000, and > 293,000. Average intensity per day (mg/day) was categorized as: 0, 1-500, and > 500. To evaluate adherence, those without metformin claims during the lookback period were excluded. Adherence measures were: binary proportion of days covered (PDC) ($0.80, < 0.80) and adherence trajectories. Group based trajectory modeling was used to identify trajectories (adherent, slow decline, rapid decline, and early discontinuation). The Anderson Behavioral Model was used to guide selection of covariables: demographic and clinical variables (diabetes severity, metabolic syndrome, comedications, and health status). Conditional logistic regression was used to evaluate the association between exposure to metformin and the risk of HR+/HER2- BC. Results: The main cohort included 690 cases and 2747 controls. A decremental reduction in odds of HR+/HER2- BC in the highest cumulative dose (OR = 0.72, 95% CI: 0.55-0.95; OR = 0.60, 95% CI: 0.42-0.85) and intensity (OR = 0.61, 95% CI:0.46-0.82) categories of metformin was observed compared to the no-metformin group. Those non-adherent to metformin had 45% (OR = 1.45, 95% CI: 1.08-1.94) increased odds of HR+/HER2- BC compared to those adherent. The risk of HR+/HER2- BC in the adherent (OR = 0.67, 95% CI: 0.39-1.14), slow decline (OR = 0.75, 95% CI: 0.43-1.32) and rapid decline (OR = 0.73, 95% CI: 0.41-1.31) trajectories was not statistically significant compared to the early discontinuation trajectory. Conclusions: This retrospective study based on SEER-Medicare found an association between high dose and intensity of metformin use with reduced odds of incidence of HR+/HER2- BC among postmenopausal women with T2DM. Adherence to metformin also showed protective effect against HR+/HER2- BC. Research Sponsor: None.

10519 Poster Discussion Session

Development of AI-powered imaging biomarker for breast cancer risk assessment on the basis of mammography alone.

Ki Hwan Kim, Hyeonseob Nam, Eunji Lim, Chan-Young Ock; Lunit Inc., Seoul, South Korea

Background: There is increasing interest in early detection of breast cancer by utilizing MRI in high-risk populations. However, it is still challenging to define and enrich the high-risk population. In this study, we developed an artificial intelligence (AI)-powered Imaging Biomarker in Mammography (IBM) to dis- cover unique mammographic patterns, beyond simple density evaluations, that are related to breast cancer. Methods: A total of 49,577 mammography exams were collected to develop the AI-powered IBM, in which 6,218 were cancers. First, we evaluated the hypothesis that the unaffected breast of cancer patients would have a different pattern than that of non-cancer patients, by training AI (IBM-A) with unaffected breast in cancer patients and breasts of non-cancer patients. We then utilized further images of the cancer patients to train AI (IBM-B). This time we used both affected and unaffected breasts of cancer patients and breasts of non-cancer patients, allowing IBM-B to additionally learn patterns related to breast cancer. The IBMs were evaluated using the internal data (n = 2,058) that included 719 cancers. To demonstrate the feasibility of early detection by using IBM-B, it was tested with external data (n = 4,158) from an independent institution. This included pre-index exams (n = 292) taken prior to index exams acquired at the time of cancer diagnosis. Results: With the internal data, IBM-A showed AUC of 0.842, suggesting that AI could learn the difference between the normal breast of cancer patients and non-cancer patients. With IBM-B, which used additional cancer images to train, AUC was improved to 0.852. Based on the internal validation, IBM-B was chosen for the external validation, in which pre-index examinations were used only. IBM-B showed AUC of 0.777 in discriminating the pre- index exams of cancer patients and those of non-cancer patients. The radiologists excluded the apparent missed cancers (n = 87) by reviewing the pre-index exams retrospectively. After, the recalculated AUC of IBM-B was 0.770, suggesting that IBM-B can distinguish between mammograms of patients who will develop breast cancer in the future and those who will not. The mean IBM-B scores in pre-index exams of cancer group (0.580) were significantly higher than those in the normal (0.258, P < 0.001) and benign (0.258, P < 0.001) groups. Conclusions: AI-powered IBM could detect the unique parenchymal pattern associated with high breast cancer risks, and we show the potential of the AI-powered IBM to be used as an independent biomarker to select high-risk populations based on mammography alone. Research Sponsor: Lunit Inc.

Total (n = 292) Invasive cancer (n = 235) Mean score of IBM-B Index exams Pre-index exams Index exams Pre-index exams

Normal (n = 1693) 0.258 - - - Benign (n = 2173) 0.261 - - Cancer Visible in pre-index 0.687 0.587 0.721 0.580 ( = missed cancer, n = 87) Invisible in pre-index (n = 205) 0.627 0.572 0.638 0.580

10520 Poster Discussion Session

Gene methylation and cytological atypia in random fine needle aspirates for assessment of breast cancer risk: 10-year follow up in average risk women.

Erica Wrubel, Mary Jo Fackler, Peng Huang, Natalie Pulliam, Amanda Montanari, Carola Maria Zalles, Saraswati Sukumar, Vered Stearns, Seema Ahsan Khan; Northwestern University, Chicago, IL; Johns Hopkins Kimmel Cancer Center, Baltimore, MD; The Johns Hopkins University School of Medicine and The Sidney Kimmel Comprehensive Cancer Center, Baltimore, MD; Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins School of Medicine, Baltimore, MD; Mercy Hosp, Coral Gables, FL; The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins School of Medicine, Baltimore, MD; Northwestern Memorial Hospital, Chicago, IL

Background: Alterations in DNA methylation occur early in tumorigenesis, and are a potential breast cancer risk biomarker. We previously reported a study where healthy volunteers underwent random fine needle aspiration (rFNA) of the breasts; cumulative methylation index of eight preselected tumor suppressor genes (CMI) was associated with the presence of cytological atypia in rFNA samples. We now report 10-year follow-up of this population, to evaluate whether increased CMI is associated with subsequent breast cancer development. Methods: 380 women, unselected by breast cancer risk, were enrolled. Demographics, breast cancer risk factors, lifetime Gail model risk estimates (Gail-LR), and %breast density were obtained at baseline. rFNA samples were assessed for cytopathology (Masood Score, MS) and CMI. Patients were contacted annually for 10 years to ascertain development of invasive or non-invasive breast cancer. In univariate analysis, log-rank test was used to compare breast cancer incidence rates between individuals with high and low baseline measures (separated by median). Area under the ROC curve was used to evaluate the cancer prediction accuracy. In multivariate analysis, the effect of CMI (after log-transformation to reduce skewness) was further studied using Cox regression model adjusting for confounding baseline variables. Results: 362 women participated in follow up. At a median follow up time of 9.5 years after rFNA sampling, 16 women developed invasive or non-invasive breast cancer. There were no significant differences between women who developed cancer and those that did not in regard to demographic factors, %breast density, MS, or Gail-LR. On univariate analysis, Gail-LR was higher in women who developed cancer (13.0 vs. 16.5, p=0.08). The largest hazard ratios were observed from high breast density (2.30, 95% CI 0.8, 6.6) and high CMI (2.26, 95% CI 0.8, 6.6, p=0.07). In breast cancer prediction, the AUC for CMI was 0.64 (95% CI 0.51, 0.77). In separate bi- variable models that adjusted for age, Gail-LR, MS, and %breast density, the HR for log CMI was consistently above unity, with a p value consistently below 0.1, except for the model that included MS (see Table). Conclusions: Elevated CMI has potential as a robust predictor of future breast cancer occurrence in average risk women, even when adjusted for breast density or cytologic atypia. Our prior analysis established that CMI is not susceptible to variation with menstrual cycle phase and menopausal status. These features support its further evaluation in larger trials. Clinical trial information: NCT00896636. Research Sponsor: Avon Foundation #02-2011-109 awarded to S. A. Khan, V. Stearns, and S. Suku- mar, Other Foundation.

Models adjusting log CMI for covariates. Log CMI HR log CMI p value Covariate HR Covariate p value

Log CMI + age 1.56 0.0990 0.93 0.0624 Log CMI + % breast density 1.61 0.0739 1.01 0.6473 Log CMI + Gail-LR 1.62 0.0742 0.02 0.4898 Log CMI + MS 1.54 0.1470 1.54 0.8350

10521 Poster Session

Mismatch repair gene alteration subtypes impact age of onset of mismatch repair-deficient cancers.

Aifen Wang, Weidong Hu, Ninghan Feng, Yong Yin, Zhe Pei; Zhangjiagang TCM Hospital Affiliated to Nanjing University of Chinese Medicine, Suzhou, China; ZhongNan Hospital of Wuhan University, Wu- han, China; Wuxi Second Hospital Affiliated to Nanjing Medical University, Wuxi, China; Taizhou TCM Hospital Affiliated to Nanjing University of Chinese Medicine, Taizhou, China; CUNY Medical School, The City College of New York, New York, NY

Background: This study investigated whether mismatch repair (MMR) gene alterations and alteration subtypes impacted age at onset of MMR-deficient cancers. Methods: A retrospective study of association of MMR-deficient alterations and age of onset of MMR-deficient lung, breast, prostate, bladder, colorectal, endometrial, and/or ovarian cancers was conducted by using cBIPORTAL dataset. Results: A total of 815 participants with MMR-deficient alterations were enrolled, 346 males and 469 females (median age 63 years; age range, 20-90 years). Males were diagnosed as colorectal and bladder cancer at later ages than females. When stratified by alteration type, individuals with MSH6 or PMS2 missense alterations had later ages of onset of colorectal cancer than those with no missense alterations. Participants with MSH2 AMP alterations were older at the diagnosis of endometrial cancer than those without MSH2 AMP alterations. Females with MLH1 missense alterations had later ages of onset of ovarian cancer than those with no missense alterations. Individuals with PMS2 AMP alterations were confirmed with lung cancer at later ages than those with no PMS2 AMP alterations. Carriers with MSH6or PMS2 missense alterations were older at the time of diagnosis of prostate cancer than those with no missense alterations. Conclusions: Gene alterations and subtype of alterations could stratify carriers with MMR- deficiency in bespoke surveillance. The mechanism of the association of MMR-deficient alterations and alteration subtypes and age at the diagnosis of MMR-deficient cancers needs to be further study. Re- search Sponsor: ZhongNan Hospital of Wuhan University Science Technology and Innovation Cultivat- ing Fund.

Distribution of MMR deficient alterations, presenting age at onset of MMR-deficient cancers. MLH1 MSH2 MSH6 PMS2 Number Median Number Median MMR deficient Number Median Number Median (% of age (% of age cancers (% of total) age (range) (% of total) age (range) total) (range) total) (range)

Colorectal cancer 71 (30) 67 (20-86) 72 (30) 60 (24-90) 107 (45) 64 (27-90) 58 (24) 66 (24-86) Endometrial cancer 34 (24) 56 (43-87) 48 (31) 57 (33-83) 61 (38) 57 (33-87) 39 (36) 63 (33-87) Ovarian cancer 13 (24) 62 (48-77) 24 (45) 64 (41-83) 21 (40) 67 (41-83) 16 (30) 53 (37-73) Prostate cancer 28 (25) 59 (51-85) 44 (39) 62 (49-85) 45 (40) 62 (49-85) 39 (35) 62 (50-83) Bladder cancer 19 (24) 65 (49-90) 26 (31) 64 (31-87) 27 (38) 64 (31-87) 31 (39) 70 (36-85) Lung cancer 26 (20) 65 (35-81) 56 (44) 65 (33-87) 51 (40) 64.5 (33-76) 58 (46) 65 (38-84) Breast cancer 25 (25) 63 (31-83) 31 (31) 52 (32-90) 38 (38) 51 (29-90) 24 (24) 53 (29-78)

10522 Poster Session

The landscape of germline mutations and the relationship with tumor mutation burden in Chinese patients with lung cancer.

Jian Shi, Rongfeng Liu, Guanglei Huang, Lixing Wang, Baoen Shan, Ningning Luo, Qin Zhang, Didi Guo, Yaqing Wu, Yingxue Qi; Department of Medical Oncology, Fourth Hospital of Hebei Medical Uni- versity, Shijiazhuang, China; Department of Research Center, Fourth Hospital of Hebei Medical Univer- sity, Shjiazhuang, China; The Medical Department, Jiangsu Simcere Diagnostics Co., Ltd, Nanjing Simcere Medical Laboratory Science Co., Ltd, The State Key Lab of Translational Medicine and Innova- tive Drug Development, Jiangsu Simcere Diagnostics Co., Ltd, Nanjing, China

Background: Lung cancer is one of the most common types of cancer, ranking the first in the incidence and mortality of malignant tumors in the world and China. Although studies have been reported that genetic susceptibility to lung cancer is associated with certain germline mutations, the relationship between lung cancer risk and inherited genetic factors remains relatively elusive. However, the effect of germline mutation on TMB in lung cancer has not been explored. Herein, DNA genomic profiling was performed through NGS with a 539-gene panel to explore the germline mutations and the relationship with TMB in Chinese patients with lung cancer. Methods: We retrospectively analyzed the germline mutations through a comprehensive 539-gene profiling of 3541 Chinese patients with lung cancer. 539- gene panel contained germline mutations in 90 hereditary tumor-associated genes. We screened out the pathogenic and likely pathogenic germline mutations according to the standards and guidelines for the interpretation of sequence variants of The American College of Medical Genetics and Genomics (ACMG), and picked out there is no records in Clinvar database and no literature report. TMB of tissue or blood ctDNA in 3541 patients were further analyzed in with pathogenic mutations (P group), with likely pathogenic mutations (LP group), and no germline mutations group (Non-P group). The difference in TMB was analyzed via the Wilcoxon sign test. Results: In 3541 patients with lung cancer, 177 (4.999%) patients were identified harboring pathogenic or likely pathogenic germline mutations, in which 78 P group and 99 LP group, the rest 3364 were Non-P group. The highest prevalence of germline mutation was found in BRCA2 (0.565%), ATM (0.339%), MUTYH (0.282%), and BRCA1 (0.254%). In 177 patients with pathogenic or likely pathogenic germline mutations, 67 mutations were recorded as UNK (unknow) in Clinvar database and no literature report. The media TMB of tissue in P group, LP group and Non-P group were 5.149, 5.535 and 5.547 respectively. The media TMB of blood ctDNA in P group, LP group and Non-P group were 4.257, 3.945 and 4.483 respectively. There was no statistical difference in TMB between P and Non-P groups (tissue p = 0.98; ctDNA p = 0.5). Conclusions: In our study, we firstly identified 67 novel germline mutations and studied on the relationship between germline mutations and TMB in lung cancer, which expanded the understanding of germline mutations. Research Sponsor: None.

10523 Poster Session

Management impact of preoperative germline genetic testing in patients with breast cancer at the Lifespan Cancer Institute.

Kaitlyn P. Lew, Don S. Dizon, Chanika Phornphutkul, Cindy Benson, Doreen Leyden Wiggins, Mary Anne Anne Fenton, Charu Taneja, Lauren J. Massingham; Brown University, Providence, RI; Life- span Cancer Institute and Brown University, Providence, RI; Rhode Island Hospital/Alpert School of Medicine at Brown University, Providence, RI; Rhode Island Hospital, Providence, RI; University Surgi- cal/Alpert School of Medicine at Brown University, Providence, RI; Lifespan Cancer Institute, Provi- dence, RI; Rhode Island Hospital/ Alpert Medical School of Brown University, Providence, RI

Background: Breast cancer is a heterogenous disease and management is complex. Advances in next generation sequencing has allowed genetic testing to be more accessible. However, conveying results to patients and care team can be challenging due to various variant classifications. Diagnostic results have the potential to guide management. Nondiagnostic results can be misinterpreted. The extent to which preoperative genetic testing affects management of newly diagnosed breast cancer is unknown. Meth- ods: Newly diagnosed breast cancer patients were identified via review of breast tumor board between May 2019 and March 2019 followed by chart review to collect detailed information. Results: 408 newly diagnosed breast cancer cases were queried. Genetic evaluation was recommended and completed in 68%, not recommended (did not meet NCCN criteria) in 30% and declined in 2.7%. The genetic evaluation recommended cohort was associated with a higher mastectomy rate in comparison with when not recommended (31% vs. 9%, p=0.0001). Of those who completed genetic testing: 12% harbored a pathogenic/likely pathogenic variant (PV/LPV), 26% had a nondiagnostic variant of uncertain significance (VUS) and 61% had negative testing. Comparison between nondiagnostic test results (negative and VUS) and diagnostic test results revealed significantly increased number of women in the diagnostic group who chose mastectomy over breast conservation therapy (BCT, nondiagnostic 20% vs diagnostic 39%, p=0.018). When negative, VUS, and PV/LPV were each independently analyzed, diagnostic test results again revealed a significantly increased number of mastectomies over BCT (p<0.05). Com- parison of surgical choices in nondiagnostic VUS vs. negative results was not significantly different (Ta- ble). Comparing the surgical timelines, completing a genetic evaluation did not affect surgery timing (mean 2.3 vs. 2.2 months, p>0.5). Conclusions: Germline genetic testing in patients with newly diagnosed breast cancer impacts clinical management. Those harboring a diagnostic result were more likely to choose mastectomy over BCT. Not surprisingly, mastectomy rate was higher among those where genetic evaluation was recommended, possibly due to concerning personal or family history. The mastectomy rate was higher among those with a diagnostic result, indicating an understanding of the genetic testing significance by patients and the care team. More importantly, those who harbored nondiagnostic VUS did not make significantly different surgical choices compared with negative genetic testing, highlighting the critical role of proper genetic counseling and being part of the care team. We conclude consideration of genetic evaluation is clinically useful and feasible without affecting the surgical time- line. Research Sponsor: None.

BCT* MRM*

Normal 121 (77) 36 (23) PV/LPV 19 (61) 12 (39) VUS 49 (89) 6 (11)

*Number of patients (%).

10524 Poster Session

Uptake of genetic counseling and testing in a clinic based population of women with breast cancer.

Alexandra Wehbe, Mark A. Manning, Hadeel Assad, Kristen Purrington, Michael S. Simon; Wayne State University, Detroit, MI; Wayne State University/Karmanos Cancer Institute, Detroit, MI; Karmanos Can- cer Institute, Detroit, MI; Barbara Ann Karmanos Cancer Institute, Wayne State University, Detroit, MI; Barbara Ann Karmanos Cancer Institute/Wayne State University, Detroit, MI

Background: Carriers of pathogenic variants in cancer susceptibility genes have an elevated risk of developing breast, ovarian, and other cancers.We conducted a medical record review to determine the uptake of genetic counseling and testing in a clinic-based population of women with breast cancer. Methods: Medical records of 150 women with breast cancer seen at the Karmanos Cancer Institute between Janu- ary-December 2018 were reviewed to determine the proportion eligible for genetic testing according to National Comprehensive Cancer Network guidelines. We also assessed genetics referral rates, appointment completion and results of genetic testing. Using chi-square and ANOVA tests, we analyzed the association of demographic and clinical factors with eligibility and referral to genetic counseling. Results: The average age of diagnosis was 57.1 years old, with 68.7% of women diagnosed with stage I-III disease, and 31.3% diagnosed with stage IV disease. There were 91 (60.7%) women who met NCCN criteria for genetic testing, of which 46.2% ultimately underwent genetic testing. Eligible women were more likely to be younger (52.6 vs. 64.0 years old), White (75.0% vs. 54.5%), and have Medicaid (75.0%) or private insurance (72.9%) vs. Medicare (44.8%). Women who met NCCN criteria were 3.5 times more likely to be referred for genetic counseling than those that did not meet eligibility criteria. Women were also more likely to be referred if they had early-stage disease compared to stage IV (67.8% vs. 48.3%), and Medicaid or private insurance compared to Medicare (71.4%, 72.0% and 40.0%, respectively). Of eligible women, 59.3% had a genetic counseling appointment scheduled, and of those, 78.0% attended their appointment. There were no apparent differences in appointment completion based on race with similar percentages of Black and White women completing their appointments (74.0% and 77.0% respectively). Women with stage IV disease were more likely to complete their appointments (83.0%) compared to women with stages I-III (74.0%) and fewer women with Medicare completed their genetic counseling appointment (56.0%) compared to women with Medicaid (83.0%) and women with private insurance (83.0%). Among women who attended their appointment, 95.9% underwent genetic testing. Of women who had genetic testing, 8.5% had a pathogenic variant and 30.4% had a variant of unknown significance. Conclusions: The results of this study indicate that lack of genetic counseling referrals contribute to a gap between the need for and completion of genetic testing. By understanding barriers to genetic counseling and testing, future clinical initiatives could effec- tively improve accessibility to genetic counseling services. Research Sponsor: None.

10525 Poster Session

Closing the gap: Trends in inconclusive rates on hereditary cancer testing across racial/ ethnic groups.

Foluso Olabisi Ademuyiwa, Carrie Horton, Holly LaDuca, Timothy Komala, Jessica Profato, Tina Pesaran, Fergus Couch; Washington University School of Medicine in St. Louis, St. Louis, MO; Ambry Genetics, Aliso Viejo, CA; Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN

Background: Several groups have described disparities in genetic test results for inherited breast cancer predisposition, with a disproportionate number of variants of unknown significance (VUS) reported in non-Caucasian individuals. These disparities are due in part to the underrepresentation of non-Cauca- sians in reference databases and clinical genetic testing cohorts. Over the past few years, diversification efforts have been made however, little data exists on how ethnicity- and gene-specific VUS rates have changed over time and whether such disparities have improved or worsened. Methods: We retrospectively reviewed demographic information and test results for individuals who underwent hereditary cancer multigene panel testing between March 2012 and September 2020 at a single laboratory. Individuals who self-reported as African American, Asian, Caucasian, or Hispanic on the test requisition form and whose testing included ATM, BRCA1, BRCA2, CHEK2 and PALB2 (five commonly tested breast cancer predisposition genes with management guidelines) were included in the study (n = 284,130). The frequency of germline variants of unknown significance (VUS) in the five geneswas assessed in September 2015 and September 2020. Results: Amongst patients tested between March 2012 and September 2015, 82.8% of the study cohort self-reported as Caucasian and 17.2% were not Caucasian (6.5% Af- rican American, 6.0% Hispanic, and 4.6% Asian). The proportion of non-Caucasian individuals in the study cohort increased slightly by September 2020 to 22.8% (77.2% Caucasian, 9.2% African Ameri- can, 8.4% Hispanic, and 5.3% Asian). Consistent with previous reports, Caucasians had the lowest VUS rate overall in both 2015 and 2020. This was also true at the individual gene level, with the exception of CHEK2. Over time, we observed a relative decrease in VUS rates across all ethnicities. Between 2015 and 2020, the overall VUS rate for the five included genes in non-Caucasian individuals was reduced by 32.0% in non-Caucasians compared to 23.6% in Caucasians. The absolute difference in VUS rate between non-Caucasians and Caucasians decreased from 7.9% in 2015 to 4.5% in 2020. Conclu- sions: While VUS rates for commonly tested breast cancer predisposition genes remain higher in non- Caucasians relative to Caucasians, our results demonstrate that this gap has been reduced over a five- year time period. These findings may be indicative of efforts by clinicians and laboratories to reduce these disparities. Further studies are necessary to improve the clinical utility of genetic testing in under-represented populations. Research Sponsor: Ambry Genetics.

ATM BRCA1 BRCA2 CHEK2 PALB2 2015 2020 2015 2020 2015 2020 2015 2020 2015 2020

African American 10.2% 5.2% 1.5% 1.6% 4.2% 2.8% 1.1% 0.8% 2.0% 1.3% Asian 6.3% 5.1% 3.2% 2.3% 5.0% 4.2% 2.5% 2.5% 5.8% 2.2% Caucasian 4.1% 3.2% 0.9% 0.8% 2.0% 1.5% 1.9% 1.5% 1.5% 1.1% Hispanic 6.1% 4.2% 1.3% 1.0% 2.6% 2.5% 2.8% 2.1% 1.6% 1.2%

10526 Poster Session

Breast cancer ER, PR, and HER2 expression variance by germline cancer predisposition genes.

Grace Wei, Marilin Rosa, Maxine Chang, Brian J. Czerniecki, Xia Wang; USF Health Morsani College of Medicine, Tampa, FL; H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL; GeneHome, H. Lee Moffitt Cancer Center, Tampa, FL; Hosp of the Univ of Pennsylvania, Philadelphia, PA; Moffitt Can- cer Center, Tampa, FL

Background: The association between breast cancer characteristics and survival with estrogen receptor (ER) and progesterone receptor (PR) expression has been primarily studied via binomial categories, ER- positive and ER-negative. In order to better characterize germline genetic influences on these markers, we investigated their IHC expression semi-quantitatively in cancer predisposition germline pathogenic variant (PV) carriers of the following genes: BRCA1,BRCA2, PALB2, TP53, PTEN, CDH1, ATM, CHEK2, and Lynch syndrome genes. The HER2 expression was also analyzed. Methods: We conducted a retrospective chart review of patients with germline panel genetic testing for cancer predisposition genes at Moffitt Cancer Center’s GeneHome clinic. Inclusion criteria included 1) women $18 years old, 2) breast cancer diagnosis, 3) cancer predisposition germline panel genetic test results, 4) available ER and PR expression levels, and 5) available HER expression and/or amplification status. ER, PR, and HER2 status were compared between PV carriers and non-PV carriers via Mann-Whitney U at p>0.05. Results: A total of 847 cases were reviewed for the study. Among 658 patients with a breast cancer diagnosis and complete ER PR data, 365 cases (55.5%) were non-PV carriers and 293 cases (44.5%) carried a PV in at least one of the genes listed above. Among 635 cases with available HER2 expression/ amplification status, 355 (55.9%) cases were non-PV carriers and 288 (45.4%) cases were PV-carriers. When compared with non-PV carrier controls, BRCA1 PV carriers’ breast tumors had significantly lower ER and/or PR expression. Further, BRCA2 and TP53 PV tumors also displayed moderately lower ER expression. Contrarily, CHEK2 tumors displayed higher ER and PR expression compared to controls. Fur- ther, BRCA1 and BRCA2 PV carriers were more likely to have HER2- breast cancers. Conclusions: Differences in ER, PR, HER2 expression levels were observed in germline PV carrier breast cancers, sig- naling differential impacts by germline PVs on the tumor evolution process. It is likely that tumor differences in PV carriers influence responses to therapies, including hormone therapy, anti-HER2 therapy, and subsequent survival. Research Sponsor: None.

ER, PR and HER2 expressions based on germline pathogenic variants. ER% mean P- PR% mean P- HER2- Case No. HER2+ Case No. P- (SD) Value (SD) Value % % Value

BRCA1 14.3 (32.7) 0 6.41 (20.2) 0 58 (98.3%) 1 (1.7%) 0 BRCA2 69.0 (38.0) 0.001 45.3 (37.7) 0.19 63 (95.5%) 3 (4.5%) 0 TP53 52.1 (46.7) 0.011 38.6 (44.7) 0.306 13 (61.9%) 8 (38.1%) 0.084 PALB2 68.2 (39.5) 0.114 39.5 (34.4) 0.163 20 (90.9%) 2 (9.1%) 0.112 ATM 80.1 (34.0) 0.609 51.0 (40.0) 0.631 21 (80.8%) 5 (19.2%) 0.471 CHEK2 93.3 (17.4) 0.003 72.0 (34.6) 0.001 32 (80.0%) 8 (20.0%) 0.462 MMRs 60.9 (47.8) 0.285 40.4 (42.5) 0.454 11 (61.1%) 7 (38.9%) 0.093 PTEN 96.8 (3.82) 0.305 83.8 (17.9) 0.066 4 (100%) 0 (0.0%) 0.368 Control/ 78.0 (34.5) - 48.8 (41.1) - 276 (77.7%) 79 (22.3%) - Reference

10527 Poster Session

Use of comprehensive next-generation sequencing to identify pathogenic germline variants with therapeutic relevance in metastatic breast cancer.

Nicole Margo Grogan, Yi-Mi Wu, Dan R. Robinson, James M. Rae, Norah Lynn Henry, Daniel F. Hayes, Michelle F. Jacobs, Kara J. Milliron, Bailey Hulswit, Lauren E. Hipp, Sofia Merajver, Elena Martinez Stoffel, Arul Chinnaiyan, Erin Frances Cobain; University of Michigan Rogel Cancer Center, Ann Arbor, MI; University of Michigan, Ann Arbor, MI; University of Michigain Health System, Ann Arbor, MI; Univ of Michigan, Ann Arbor, MI

Background: Among patients with early-stage breast cancer, approximately 6-10% have a pathogenic germline variant (PGV) conferring inherited cancer predisposition. In contrast, few studies have explored the frequency and types of PGVs identified in patients with metastatic breast cancer (MBC); therefore, additional data is needed. Methods: From 2011-2020, 278 patients with MBC underwent fresh tumor biopsy and blood sample collection for paired tumor/normal DNA (targeted exome capture with analysis of 1700 genes) and RNA (tumor transcriptome) sequencing through the Michigan Oncolo- gy Sequencing (Mi-ONCOSEQ) program. Somatic and germline alterations were annotated and classified according to degree of clinical actionability with results returned to treating oncologists. Retrospective chart review was performed to determine if: 1) a PGV was identified prior to Mi-ONCOSEQ testing, 2) patients met National Comprehensive Cancer Network (NCCN) guideline criteria for genetic testing on the basis of personal or family cancer history and 3) patients received subsequent therapy informed by a PGV. Results: Forty-eight of the 278 patients (17.3%) had at least one PGV identified, with a total of 50 PGVs identified in this cohort. Only twelve of these PGVs (24%) had been identified prior to Mi-ONCOSEQ testing. The most frequent PGVs identified were in CHEK 2 (n = 9, 18%), MUTYH (n = 6, 12%), BRCA 1 (n = 5, 10%), BRCA2 (n = 5, 10%), ATM (n = 4, 8%) and PALB2 (n = 4, 8%). Somatic loss of heterozygosity events (LOH) occurred in 30 of the 50 cases with PGVs identified (60%). LOH events were observed in 83.3% of BRCA1, BRCA2, ATM and PALB2 PGVs, but were less frequently observed with CHEK2 (33.3%) and MUTYH (66.7%). Two hundred sixteen out of 278 patients (77.7%) in this cohort met NCCN criteria for genetic testing, although six patients with a PGV identified (CHEK2: n = 5; MUTYH: n = 1) did not meet NCCN criteria. Twenty-nine PGVs identified (58%) had potential therapeutic relevance and 11 patients (22.9%) received targeted therapy based on the PGV. Conclusions: The frequency of PGVs identified in this cohort is nearly double the frequency reported for patients with early-stage disease, suggesting that certain PGVs may confer worse prognosis. CHEK2, the most frequently identified PGV, was less likely to have an identifiable LOH event. The direct role of CHEK2 PGVs in tumor pathogenesis is uncertain, but other mechanisms of silencing the wild type allele must be considered. Despite the majority of patients meeting NCCN criteria for genetic testing, those with PGVs in CHEK2 were less reliably identified by this mechanism. The majority of PGVs identified were of potential therapeutic relevance, supporting the recommendation for genetic testing in all patients with MBC. Research Sponsor: U.S. National Institutes of Health.

10528 Poster Session

Identification and management of pathogenic mutations in BRCA1, BRCA2, and PALB2 in a tumor-only genomic testing program.

Brittany L. Bychkovsky, Tianyu Li, Jilliane Sotelo, Nabihah Tayob, Joanna Mercado, Israel Gomy, Anu B. Chittenden, Sarah R. Kane, Samantha Stokes, Melissa E Hughes, Ji Seok Kim, Mark M. Awad, Panagiotis A. Konstantinopoulos, Brian M. Wolpin, Mary-Ellen Taplin, Bruce E. Johnson, Neal Ian Lindeman, Laura E MacConaill, Judy Ellen Garber, Nancy U. Lin; Cancer Genetics and Prevention, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA; Department of Data Science, Dana-Farber Cancer Institute, Boston, MA; Thermo Fisher Scientific, Waltham, MA; Genome Medical, South San Francisco, CA; Cancer Genetics and Prevention, Dana-Farber Cancer Institute, Boston, MA; Memorial Sloan Kettering Cancer Center, New York, NY; Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA; Lowe C, Dana-Farber Cancer Institute, Boston, MA; Dana-Farber Cancer Institute, Boston, MA; Lank Center for Genitourinary Oncology, Dana-Farber Cancer Institute, Boston, MA; Department of Pathology, Brigham and Women’s Hospital, Boston, MA; Center for Cancer Genome Discovery, Dana-Farber Cancer Institute, Boston, MA

Background: Tumor-genomic testing is increasingly used to guide treatment decisions in cancer patients. Although tumor-only testing cannot definitively distinguish between germline versus somatic alterations, the identification of pathogenic or likely pathogenic (P/LP) variants in certain genes should prompt consideration of germline testing. Germline P/LPs in BRCA1, BRCA2 and PALB2 (B1B2PAL) are associated with hereditary cancer syndromes. Methods: We reviewed tumor-only genomic data (Dana-Farber Oncopanel) between 10/2016 and 6/2018 to examine the prevalence of P/LPs in BRCA1, BRCA2,PALB2 among adult cancer patients at Dana-Farber Cancer Institute/Brigham and Women’s Hospital. We characterized the frequency of P/LPs by primary tumor type, confirmation by germline testing before or within 12 months after Oncopanel testing or not, and factors associated with germline testing. Results: Among 7,575 patients, the median age was 62 (range 18-99); 53.9% were female. A total of 272 (3.6%) had P/LPs in BRCA1 (n = 90), BRCA2 (n = 162) and/or PALB2 (n = 29). P/LPs in B1B2PAL were detected in 5.3% (38/712) of breast, 11.9% (34/285) of ovarian, 6.6% (18/ 272) of pancreatic, and 5.1% (12/234) of prostate cancers. P/LPs in B1B2PAL were also detected in other neoplasms (12.9% (8/62) of non-melanoma skin, 5.0% (43/855) of colorectal, 7.6% (20/264) of endometrial, and 4.6% (10/216) of head and neck cancers). Of 169 patients who had not had prior germline testing, 29/169 (17.2%) completed germline testing within 12 months after Oncopanel; 13 (7.7%) referred for testing declined or did not complete testing within 12 months, 14 (8.3%) died before or within 3 months of the Oncopanel results, and 113 (66.9%) had no documented germline testing within 12 months. Among 132 patients who had germline testing, 117 (88.6%) had a clinical indication based on personal or family history compared to 66/140 (47.1%) who did not undergo germline testing. Among 132/272 (48.5%) germline-tested patients, 70.5% were positive for a germline mutation in B1B2PAL; the remainder had somatic B1B2PAL mutations only. Germline testing was more often performed in patients with B1B2PAL-associated tumors (breast, ovarian, pancreatic and prostate cancers) or other clinical indications for germline testing. Conclusions: A low but clinically meaningful rate of P/LPs in BRCA1, BRCA2 and PALB2 was detected by tumor-only genomic testing in diverse malignancies. Given the implications of B1B2PAL alterations on treatment and familial cancer risk, our data support current NCCN guidelines recommending germline testing among patients with cancer and P/LPs in B1B2PAL detected on tumor-genomic testing and highlights the need for systems to ensure germline testing when indicated. Research Sponsor: None.

10529 Poster Session

Genetic counseling and testing rates among community cancer programs for patients with breast cancer following site-directed quality improvement.

Leigh Boehmer, Latha Shivakumar, Christine B. Weldon, Julia Rachel Trosman, Stephanie A. Cohen, Dawn Nixon, Zohra Ali-Khan Catts, Susan Miesfeldt, Stephanie Wiryaman, Lorraine Tafra, Lisa Muto, Natalie Fadrowski, Melissa C. Hulvat, Tiana Pallister, Catherine Gaines, Charles Hendrix Shelton, BRCA; Association of Community Cancer Centers, Rockville, MD; Northwestern University Feinberg School of Medicine, Chicago, IL; Center for Business Models in Healthcare, Chicago, IL; Ascension St. Vincent, Indianapolis, IN; Christiana Care Health System, Newark, DE; Maine Medical Ctr Cancer Inst, Scarborough, ME; Maine Medical Center, Portland, ME; Anne Arundel Medcl Ctr, Annapolis, MD; Ca- bell Huntington Hospital, Huntington, WV; University of Maryland Upper Chesapeake Health, Bel Air, MD; Bass Breast Ctr Kalispell Reg Medcl Ctr, Kalispell, MT; Kalispell Regional Medical Center, Kalis- pell, MT; Southeastern Health, Gibson Cancer Center, Lumberton, NC; Vidant Health, Nags Head, NC

Background: National Comprehensive Cancer Network (NCCN) guidelines recommend testing for highly penetrant breast/ovarian cancer genes in several scenarios, including women with early-onset (# 45 years) or metastatic HER-2 negative breast cancer regardless of family history. A 2018 Association of Community Cancer Centers (ACCC) survey (N = 95) showed that > 80% of respondents reported # 50% testing rate of patients with breast cancer who met guidelines. To improve rates of genetic counseling(GC)/testing, ACCC partnered with 15 community cancer programs to support site-directed quality improvement (QI) interventions. Methods: Pre- and post-intervention data from 9/15 partner programs for genetically at-risk women with early-onset or HER-2 negative metastatic breast cancer (MBC) were analyzed. Pre-intervention data were collected between 01/01/2017 and 06/30/2019 while post-intervention data were collected as early as 07/01/2019 and as late as 10/01/2020. QI project scope ranged from creation of testing eligibility education to implementation of a virtual GC clinic. De-identified data collected included: family history documentation; GC appointment; test results; and timing of results relative to surgical date. Results: The pre-intervention cohort included 2691 women and the post- co- hort included 3104 women who were eligible for GC. Early-onset patients in the post-intervention group attended a GC appointment 83% (331/401) of the time and 74% (296/401) had genetic test results, with 92% (271/296) receiving results before surgery. Sixty-one percent (1387/2267) of women with HER-2 negative MBC in the post-intervention group received GC, compared to 36% (658/1845) in the pre-intervention group. There was an overall increase in the number of MBC patients with documented test results following GC in the post-intervention cohort (55% (1243/2267) versus 15% (273/1845); p < 0.0001). Rates of GC appointments improved overall, regardless of family history documentation. Rates among those with a documented high-risk family history improved from 57% (729/1284) to 85% (1485/1741) following QI interventions (p < 0.0001). There was also a significantly higher rate of GC provided in the post-intervention group among women with negative family histories (40% (462/1155) versus 23% (181/778); p < 0.0001). GC also increased from 6% (35/629) to 45% (94/208) of women in the post-intervention cohort with no documentation of family history (p < 0.0001). Conclusions: Ge- netic testing is underutilized in women with breast cancer. Significant improvement was achieved with QI initiatives specifically designed to target easily identified populations meeting guidelines for GC/testing. This project demonstrates the importance of attention to practice-directed strategies aimed at improving identification of risk as well as follow through to GC/testing. Research Sponsor: Pfizer, Inc.

10530 Poster Session

Cancer surveillance in adults with germline TP53 pathogenic variants: A single-center observational study.

Thomas Meyskens, Vincent Vandecaveye, Steven Pans, Raphae€la Dresen, Chantal Van Ongeval, Ann Smeets, Ines Nevelsteen, Patrick Neven, Hans Wildiers, Patrick Schoffski, Ellen Denayer, Griet Van Buggenhout, Genevie�ve Michils, Hilde Brems, Eric Legius, Kevin Punie; Department of General Medi- cal Oncology, Leuven Cancer Institute, University Hospitals Leuven, Leuven, Belgium; Department of Radiology, University Hospitals Leuven, Leuven, Belgium; Department of Surgical Oncology and Multi- disciplinary Breast Centre, University Hospitals Leuven, Leuven, Belgium; Department of Gynaecology & Obstetrics and Multidisciplinary Breast Centre, University Hospitals Leuven, Leuven, Belgium; De- partment of General Medical Oncology and Multidisciplinary Breast Centre, Leuven Cancer Institute, University Hospitals Leuven, Leuven, Belgium; Center for Human Genetics, University Hospitals Leu- ven, Leuven, Belgium

Background: Germline pathogenic variants (PV) in the tumor suppressor gene TP53 are associated with a high risk of developing diverse malignancies, often at young age, and predispose to Li-Fraumeni syndrome (LFS). Surveillance programs for presymptomatic PV carriers have shown survival benefit in a non-randomized trial. Here we describe the surveillance findings and clinical outcomes of adults with TP53 PV undergoing a standardized screening protocol. Methods: We identified adults with germline PV in TP53 who underwent surveillance at the University Hospitals Leuven, Belgium, between 04/2013 and 08/2020. Patients with prior cancer were allowed, while patients with an active malignancy requir- ing treatment at diagnosis of the TP53 PV were excluded. Surveillance was performed per modified To- ronto protocol, including annual whole body diffusion-weighted MRI (WB-DWI/MRI), brain MRI, abdominal ultrasound (US), endoscopic surveillance, laboratory tests, dermatological examination and breast MRI/US in females. The primary aim was to evaluate the number and type of malignancies and premalignant lesions diagnosed during screening and to assess the proportion of malignancies detected by surveillance. Secondary outcomes were the cancer detection rate during the first year of screening, the proportion of carriers with false-positive findings, and overall survival. Results: We included 42 adults from 20 apparently unrelated families. Median age was 38y (range, 17-70y) and 23 had a history of prior cancer. After a median follow-up of 41.5mo, we diagnosed 18 cancers in 12/42 participants (29%). Overall survival was 95% in all participants, including 2 carriers who opted to discontinue surveillance. Surveillance detected 10/18 cancers (56%), the majority of whom through WB-DWI/MRI (6/ 10; 60%). No malignancies were identified with brain MRI. In 5/42 individuals (12%), surveillance detected a malignancy during the first year of screening. Only 2/10 cancers discovered with surveillance (1 soft tissue and 1 bone sarcoma) belong to the LFS core tumors. Cancers not detected with surveillance (8/18) were 6 non-melanoma skin cancers and 2 interval cancers (sarcoma post radiation, secondary acute leukemia). Additionally, we detected 27 premalignant lesions in 11/42 patients (26%), of whom 78% were diagnosed by colonoscopy. False-positive findings occurred in 7/42 patients (17%) and were mostly seen with WB-DWI/MRI. Conclusions: Adults with germline PV in TP53 that undergo surveillance have high cancer detection rates. The majority of malignancies were asymptomatic at diagnosis and detected with WB-DWI/MRI. Despite the high cancer incidence, few LFS core cancers were diagnosed and survival was encouraging. Increased genetic testing changes the clinical picture of germline TP53 carrier populations, justifying the transition from LFS to a wider concept of heritable TP53-related cancer syndrome. Research Sponsor: None.

10531 Poster Session

Twenty-one-gene recurrence score (RS) in germline (g)CHEK2 mutation-associated versus sporadic breast cancers (BC): A multi-site case-control study.

Anosheh Afghahi, Sydney Marsh, Alyse Winchester, Dexiang Gao, Hannah Parris, Lisen Axell, Leif W. Ellisen, Erin Wysong Hofstatter, Allison W. Kurian, Marie Wood, Dana Zakalik, Carol-Ann Mullin, Jennifer Lee Caswell-Jin, Virginia F. Borges, Nadine M. Tung; University of Colorado, Aurora, CO; Mas- sachusetts General Hospital Cancer Center, Boston, MA; Yale School of Medicine, New Haven, CT; Stanford School of Medicine, Stanford, CA; University of Vermont, Burlington, VT; Nancy and James Grosfeld Cancer Genetics Center, Beaumont Health, Royal Oak, MI; Stanford University, Stanford, CA; Beth Israel Deaconess Medical Center, Boston, MA

Background: Genomic assays, such as RS, are used to determine chemotherapy benefit in early-stage, estrogen receptor (ER)- and/or progesterone receptor (PR)-positive, HER2 negative BC patients (pts). Currently, guidelines to use pts’ germline genetic testing results to guide adjuvant therapy are lacking. Several reports have indicated worse outcomes for BC pts with gCHEK2 pathogenic variants (PV). We investigated whether PV in CHEK2 were associated with increased RS. Methods: Patient-level clinical data and RS were derived from electronic medical records of seven medical centers between years 2013-17. Confirmation of RS using the Genomic Health provider portal was performed. 38 pts with germline PV in CHEK2 (15 pts/39.5% with c.1100delC mutation) and RS score (cases) were matched with BC pts whose genetic testing did not identify PV (controls) using a 1:2 matching schema. Pts were matched based on age at diagnosis and lymph node (LN) status. LN negative pts were further matched based on T-stage. A multivariate random intercept linear mixed model of CHEK2 mutation status on RS was performed, adjusting for PR. A secondary ordinal univariate analysis was conducted that categorized RS into low, intermediate and high risk ( < 18, 18-30, and > 30, respectively). P-values were reported based on a null hypothesis of no effect against a two-sided alternative. Results: The median RS for cases was 19.5 (interquartile range [IQR]: 15 to 25) and the median RS for controls was 18 (IQR: 12 to 22). A greater proportion of cases were categorized as high risk (10.5%) compared to controls (5.6%), and a smaller proportion of cases were categorized as low risk (36.8%) compared to controls (49.3%). Cases had higher grade and increased proportion of PR-negative BC as compared with controls (grade 1: 12.1% of cases versus 32.4% of controls; PR-negative: 7.9% of cases versus 5.6% of controls). The variables used to match cases and controls (age, lymph node status, and T-stage) had similar summary statistics. The RS was 1.97-point higher in pts with gCHEK2 PV compared to controls, after adjusting for PR (95% confidence interval [CI]: 1.02-point lower to 4.96-point higher; p = 0.194). The secondary analysis of CHEK2 mutation status on an ordinal RS risk group yielded comparable results; on average, the odds of being high risk compared to the combined intermediate/low risk groups was 1.72 times higher in cases compared to controls (95% CI: 0.77 to 3.80; p = 0.181), but these differences were not significant. Conclusions: Our case-control study did not show a statistically higher RS for BC that develops in pts with gCHEK2 PV. Further studies are warranted to evaluate the association between type of CHEK2 PV (frameshift versus missense) and other modifying genetic variables and RS. Research Sponsor: None.

10532 Poster Session

Identifying individuals with primary central nervous system tumors at risk for hereditary cancer syndromes using the Utah Population Database.

Nicholas Shawn Whipple, Wendy Kohlmann, Samuel Cheshier, Zhe Yu, Karen Curtin, Joshua David Schiffman; Division of Pediatric Oncology, Department of Pediatrics, University of Utah, Salt Lake City, UT; Huntsman Cancer Institute, University of Utah, Salt Lake City, UT; Division of Pediatric Neurosur- gery, Department of Neurosurgery, University of Utah, Salt Lake City, UT; The University of Utah, Salt Lake City, UT

Background: CNS tumors are the most common solid tumors and the deadliest cancers in children. Ap- proximately 10% of children with a CNS tumor harbor a hereditary cancer syndrome (HCS), but many will not be tested for a HCS. The Utah Population Database (UPDB) contains comprehensive cancer registry data for Utah families and can determine multigenerational cancer pedigrees across an archive of 5.8 million individuals. Early identification of HCSs results in improved cancer surveillance and outcomes, reducing the impact of CNS tumors in children. We hypothesize that the UPDB can identify children and families with HCSs not previously identified. Methods: We queried the UPDB for individuals ages 0-39 diagnosed with a primary CNS tumor (malignant and benign) between 1966-2017 and generated cancer pedigrees of 3 generations or more for probands, extending to at least third-degree relatives. Specialized software calculated a familial standardized incidence ratio (FSIR) to determine families with excess clustering of CNS tumors. Clinical cancer genetics experts reviewed pedigrees to confirm patterns of HCS. Results: We identified 4,634 CNS tumors in 4,550 individuals, of whom 2,233 (49%) reside in high-quality pedigrees containing $2 grandparents, at least 1 from both mater- nal and paternal sides. To identify families with excess clustering of CNS tumors, we selected pedigrees with an FSIR P< 0.05 and $2 affected patients, resulting in 161 high-risk families with a mean of 170 (median 96) relatives per pedigree of 3-6 generations. Among these 161 families, there were 2,017 unique relatives (first-third degree) of CNS probands with 2,355 tumors (any site), for a per pedigree average of 14.7 tumors in 12.5 relatives. Review of the 10 highest risk pedigrees indicated that 4 meet HCS criteria, including Li-Fraumeni (n = 2), von Hippel-Lindau (n = 1), and rhabdoid tumor predisposition (n = 1). Conclusions: The UPDB can produce multigenerational cancer pedigrees that identify individuals and families at risk of harboring a HCS who warrant germline testing. These findings should encourage clinicians to perform thorough family history screening and to always consider workup for associated HCSs. Research Sponsor: Intermountain Foundation at Primary Children’s Hospital.

10533 Poster Session

Rare BAP1 variant of unknown significance (VUS) and analysis of BAP1 codon 146 genomics: Potential germline and therapeutic implications.

Jo-Ellen Murphy, Sara Sadan, Jessica Kim Lee, Jana Pruski-Clark, Rebecca Sutphen, Kimberly McGregor; Foundation Medicine, Cambridge, MA; Onc and Hem of White Plains PLLC, White Plains, NY; Foundation Medicine, Inc., Cambridge, MA; InformedDNA, St. Petersburg, FL

Background: The BAP1 gene (BRCA1-Associated Protein) encodes a protein ubiquitin carboxyl-terminal hydrolase (BAP1), which removes ubiquitin moieties and regulates various cellular functions including DNA repair. This association has driven interest in defining if BAP1 variants confer susceptibility to PARP inhibitors (PARPi). Germline and somatic BAP1 alterations are both rare, mostly unique, often classified as VUS’s, and associated with a broad range of overlapping tumor types. Based on the identification of a BAP1 R146K VUS variant in tumor, also previously identified as germline, an analysis of BAP1 codon 146 alterations was initiated to explore potential genetic and therapeutic implications. Methods: 394,756 tumor specimens were tested using hybrid capture-based genomic profiling (Foun- dation Medicine) for all variant types and Tumor Mutational Burden (TMB). BAP1 codon 146 cases were analyzed for clinicogenomic features and germline results when available. Results: BAP1 codon 146 variants were identified in 23 unique patients across the following tumor types: cholangiocarcino- ma (CCA) (4), mesothelioma (4), NSCLC (3), RCC (2), ocular melanoma (2), and carcinoma of unknown primary (CUP) (3); many of which overlap with known and suspected germline associated BAP1 syn- dromic tumor types. BAP1 R146 mutations were classified as VUS in 16 patients and 7 were likely or known pathogenic. In 20 of the 23 cases where zygosity could be determined 16 (80%) were homozygous and 4 (20%) were heterozygous. In 2 of the 3 NSCLC cases, the BAP1 variant appeared likely somatic and/or associated with a high TMB. A previously reported germline BAP1 variant in a RCC patient, R146K, occurred 5 times in our tumor database. One case which was homozygous in tumor and confirmed in germline occurred in a patient who had both breast and CCA. She also had a sibling with RCC who shared the germline BAP1 R146K variant along with multiple 1st and 2nd degree relatives with RCC, mesothelioma, melanoma, liver cancer, colon cancer, and a cancer of unknown primary. Con- clusions: Codon 146 of BAP1 localizes to the UCH (ubiquitin carboxyl hydrolase) domain, which includes the BARD1 interaction region. Loss of BAP1 activity as a consequence of germline or somatic mutation likely impacts ubiquitination status and activity of downstream proteins, such as those involved in DNA repair. For patients with suspected BAP1 inactivating alterations, often seen in non-homologous recombination deficiency related tumor types, PARPi therapy may be relevant. As demonstrated here, variants identified through tumor testing may also aid in re-classification of germline VUS’s. These results support the further investigation and validation of BAP1 alterations for germline risk stratification and therapeutic strategies with either PARPi and/or other therapies specific to tumors with impaired chromatin remodeling. Research Sponsor: None.

10534 Poster Session

Germline alterations among Hispanic men with prostate cancer.

Justin Shaya, Sarah M. Nielsen, Kathryn E. Hatchell, Edward D. Esplin, Robert Luke Nussbaum, Nicole Weise, Lisa Madlensky, James Don Murphy, Elena Martinez, Rana R. McKay; University of California San Diego, Moores Cancer Center, La Jolla, CA; Invitae, San Francisco, CA; UC San Diego Moores Can- cer Center, La Jolla, CA; UCSD Moores Cancer Center, La Jolla, CA; Department of Radiation Medicine and Applied Sciences, University of California San Diego, La Jolla, CA; UC San Diego Health Moores Cancer Center, San Diego, CA

Background: With the growing indications for germline testing in prostate cancer (PCa), there is accumu- lating evidence that African American and Hispanic men with PCa are significantly under-tested compared to non-Hispanic white (NHW) men. Given this, little is known about the pathogenic germline variant landscape in Hispanic men with PCa. Methods: This was a retrospective cohort analysis of 17,256 men with PCa who underwent diagnostic germline testing through a commercial laboratory (In- vitae) from 2015-2020. Self-identified Hispanic and NHW men were selected for comparative analysis. The primary endpoint was the rate of pathogenic/likely pathogenic (PLP) germline alterations in Hispan- ic men among 25 genes associated with PCa. Secondary endpoints included comparison of PLP rates in Hispanic vs NHW men, the rate of specific PLP alterations, and the rate of variants of uncertain significance (VUS). Fisher’s exact test was used to compare germline alteration rates for significance. Re- sults: We identified 508 Hispanic and 12,542 NHW men with PCa who underwent testing during the study period. Median age at the time of testing was 69 vs 67 years in Hispanic vs NHW cohorts. A family history of PCa was reported in 21.1% (N=108) vs 27.3% (N=3428) in the Hispanic vs NHW cohorts, respectively (p=0.002). The PLP alteration rate was 7.1% in the Hispanic cohort and this rate was numerically lower but not significantly different when compared to the NHW cohort (9.7%) (p=0.058). A significantly higher rate of VUS was seen in the Hispanic cohort (Table). The four most frequently detected genes with PLP alterations in both cohorts were ATM, BRCA1, BRCA2, and CHEK2. Only the rate of CHEK2 alterations was significantly different between cohorts among all 25 genes analyzed (Ta- ble). Conclusions: In this analysis, the PLP alteration rate among Hispanic men was 7.1%, a much higher rate than has been previously reported, and the germline genomic landscape was similar to that of NHW men. The VUS rate was significantly higher among Hispanic men, a known consequence of under- testing among minority populations.These data support germline testing in Hispanic men with prostate cancer and emphasize the importance of improving testing rates. Research Sponsor: None.

Germline characteristics. Germline Characteristic Hispanic Non-Hispanic White p-value

PLP alteration rate* 7.1% 9.7% 0.058 ATM 1.2% 1.6% 0.71 BRCA1 0.6% 0.6% 1.00 BRCA2 2.8% 2.5% 0.67 CHEK2 0.8% 2.6% 0.006** MSH2, MSH6, MLH1, or PMS2 1.2% 1.2% 0.82 VUS alteration rate 21.5% 16.6% 0.005** VUS detected without a co-occurring PLP alteration 19.3% 15.0% 0.010** VUS detected with a co-occurring PLP alteration 2.2% 1.6% 0.27 *Shown are the most commonly detected PLP alterations among the 25 genes analyzed. **Significance con- cluded if p<0.05.

10535 Poster Session

Pan-cancer evaluation of homologous repair deficiency somatic mutations and response to first-line anti-neoplastic therapy.

Jessica A Lavery, Samantha Brown, Gregory J. Riely, Philippe L. Bedard, Ben Ho Park, Jeremy Lyle Warner, Kenneth L. Kehl, Eva M Lepisto, Hira Rizvi, Michele LeNoue-Newton, Caroline G. McCarthy, Celeste Yu, Ritika Kundra, Brooke Mastrogiacomo, Nikolaus Schultz, Julia Elizabeth Rudolph, Shawn Sweeney, Deborah Schrag, Katherine Panageas, AACR GENIE Consortium; Memorial Sloan Kettering Cancer Center, New York, NY; Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY; Princess Margaret Cancer Centre, Toronto, ON, Canada; The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University, Baltimore, MD; Vanderbilt-Ingram Cancer Center, Nash- ville, TN; Dana-Farber Cancer Institute, Boston, MA; Memor Sloan-Kettering Cancer Ctr, New York, NY; AACR, Philadelphia, PA

Background: Homologous recombination is a major mechanism of defective DNA repair, but it remains uncertain whether homologous repair deficient (HRD) tumors have favorable prognosis or are more/less likely to respond to treatment than tumors lacking such mutations. Objective: To determine whether lung (NSCLC) and colorectal (CRC) HRD+ tumors have better survival or response to chemotherapy than HRD- tumors. Methods: Patients with de novo stage IV NSCLC or CRC who had next generation sequencing (NGS) between 2015-2018 from one of four cancer centers were identified. Records were cu- rated using the PRISSMM framework to ascertain treatment, overall survival (OS) and progression free survival based on imaging (PFS-I) and oncologists’ notes (PFS-M). Each NSCLC or CRC tumor was categorized as HRD+ if NGS revealed an oncogenic/likely oncogenic mutation in: ATM, BAP1, BARD1, BLM, BRCA1, BRCA2, BRIP1, CHEK2, FAM175A, FANCA, FANCC, NBN, PALB2, RAD50, RAD51, RAD51C, RTEL1, or MRE11A based on the OncoKB database. The tumor was categorized as HRD- if no oncogenic mutation in any of these genes was evident and HRD indeterminate (HRD?) if no mutation was identified but the panel did not include all genes. OS, PFS-I and PFS-M from start of first line therapy were reported by HRD status. The percentage with a good response to first line therapy ($2x the median) and exceptional response ($3x the median) was estimated for each endpoint. Results: For NSCLC 4% were HRD+, 59% HRD- and 37% HRD?. For CRC there were 5% HRD+, 60% HRD- and 35% HRD?. There were no significant differences for any survival endpoint between patients who were HRD+ vs HRD- in univariable analyses. The proportion of good and exceptional responders to first line systemic chemotherapy also did not vary by HRD status, though patients with HRD+ CRC were potentially more likely to be exceptional responders. Similarly, no differences between HRD+ and HRD- tumors were apparent for the subgroup receiving platinum containing therapy. Conclusions: NSCLC and CRC patients with somatic mutations in HRD oncogenic genes did not differ from patients lacking such a mutation with respect to OS or PFS. CRC patients with HRD+ tumors may be more likely to be exceptional responders, but sample sizes are limited. By May, the analysis will include breast and pancreatic cancer cases. Research Sponsor: American Association for Cancer Research GENIE BPC.

NSCLC CRC HRD- = 411; HRD+ = 29; HRD? = 257 HRD- = 406; HRD+ = 32; HRD? = 235 Endpoint Median Good Exceptional Median Good Exceptional HRD + - or ? (95% CI), y response Response (95% CI), y Response Response

OS - 1.6 (1.4, 2.0) 28% 9% 2.5 (2.0, 2.7) 18% 5% + 1.5 (.71, 2.8) 24% 7% 2.0 (1.2, —) 31% 22% ? 1.2 (1.1, 1.6) 31% 13% 1.9 (1.7, 2.2) 24% 5% PFS-I - .47 (.42,.56) 31% 16% 0.79 (.71, 0.84) 18% 10% + .43 (.21,.61) 14% 3% 0.85 (.69, 1.2) 19% 12% ? .42 (.34,.51) 27% 18% 0.73 (.63, 0.84) 21% 14% PFS-M - .59 (.54,.69) 28% 16% 1.0 (.90, 1.1) 28% 16% + .61 (.41, 1.0) 21% 10% 1.0 (.82, 2.1) 21% 10% ? .58 (.49,.67) 27% 17% 1.0 (.85, 1.1) 27% 17%

10536 Poster Session

Genetic ancestry and clinical outcomes to immune checkpoint inhibitors among seven common cancers.

Amin Nassar, Elio Adib, Sarah Abou Alaiwi, Elie Akl, Talal El Zarif, Pier Vitale Nuzzo, Tarek H. Mouhieddine, Habib El-Khoury, Stefan Groha, Guru P. Sonpavde, Robert I. Haddad, Kent William Mouw, Marios Giannakis, Guruprasad Ananda, Matthew L. Freedman, David J. Kwiatkowski, Laura E MacConaill, Toni K. Choueiri, Alexander Gusev; Brigham and Women’s Hospital, Boston, MA; The Lank Center for Genitourinary Oncology, Dana-Farber Cancer Institute and Brigham and Women’s Hospital, Boston, MA; Lank Center for Genitourinary Oncology, Dana-Farber Cancer Institute, Boston, MA; Dana Farber Cancer Institute, Boston, MA; Dana Farber Cancer Institute/Harvard Medical School, Boston, MA; Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA; Dana-Farber Cancer Institute, Brookline, MA; Dana-Farber Cancer Institute, Boston, MA; The Jackson Laboratory for Genomic Medi- cine, Farmington, CT; Center for Cancer Genome Discovery, Dana-Farber Cancer Institute, Boston, MA; Dana-Farber Cancer Institute, The Lank Center for Genitourinary Oncology, Boston, MA

Background: Prior studies and clinical trials report associations between self-reported race and clinical outcomes to Immune Checkpoint Inhibitors (ICIs). However, comprehensive studies of ancestry-associated differences in clinical outcomes have not been performed. We derived genetic ancestry scores and assessed clinical outcomes in 1341 patients with cancer treated with ICIs. Methods: Patients at the Dana-Farber Cancer Institute treated with ICIs only and with relevant cancer types and targeted exome sequencing data (Oncopanel) were included. Relevant cancer types included colorectal adenocarcinoma (CRC), esophagogastric adenocarcinoma (EGC), head and neck squamous cell carcinoma (HNSCC), melanoma, non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), and urothelial carcinoma (UC). We developed a bioinformatics pipeline to infer fine-scale genetic ancestry for each patient (n=1341) directly from tumor sequencing data by leveraging off and on-target sequenced reads and external ancestry reference panels. Three ancestry scores were determined (African, East Asian, Europe- an). Overall survival (OS) and time-to-treatment failure (TTF) were compared by Cox logistic regression between ancestral populations. Hazard ratio (HR) was derived using multivariable analysis, adjusted for single versus combination therapy, prior lines of therapy, and tumor mutational burden (TMB, as per- centiles). Results: Median follow-up was 37.8 months (m; interquartile range: 35.7-39.5m). Common cancer types included CRC (n=52), EGC (n=114), HNSCC (n=88), melanoma (n=274), NSCLC (n=571), RCC (n=99), and UC (n=143). A higher East Asian ancestry (EAS) was significantly associated with worse OS (p=0.03) and TTF (p=0.002) in patients with RCC, independent of the histologic subtype (Table). There was no significant association between any of the three ancestral populations and clinical outcomes in the other 6 cancer types. Conclusions: We described clinical outcomes to ICIs across three global populations in 7 cancers. As the medical field re-evaluates the use of self-reported race in clinical decision-making, we utilize a novel ancestry pipeline that can be readily applied to tumor-only sequencing panels and better characterize non-white populations. We find no ancestry differences in clinical outcomes except in patients with RCC treated with ICIs which will require future validation. We plan to analyze genomic correlates of response by ancestry in each of the cancer types to better understand these diverge clinical behaviors. Research Sponsor: None.

RCC cohort. HR (95%CI) Non-clear cell African Prior lines Single versus Percentile versus clear EAS American of therapy combination TMB cell histology

OS (58/99 events) 226 (2-32271) 2 (0.4-9.9) 1.3(1-1.7) 1.8 (0.9-3.6) 0.9 (0.3-2.7) 2.5 (1.2-5.2) TTF (83/99 events) 1000 (14-73792) 0.4 (0.1-1.7) 1.1 (0.8-1.3) 2.0 (1.2-3.5) 0.7 (0.3-1.7) 2.7 (1.4-5.1)

10537 Poster Session

The protean phenotype of MSH6 pathogenic variants (PV) in Lynch syndrome (LS) patients (pts).

Michelle J McSweeny, Susan Montgomery, Kristen Danielle Whitaker, Mary Beryl Daly, Michael J. Hall; Fox Chase Cancer Center, Philadelphia, PA

Background: LS is among the most common hereditary cancer (CA) syndromes. PVs in MSH6 are 2-4 fold more common in the population (1/758) than those in MLH1 (1/1946) or MSH2 (1/2841), and are increasingly regarded as lower penetrance for CRC due to published data supporting later mean age of CRC onset and lower CRC risk. Unlike for MLH1/MSH2, NCCN 2020 CA risk estimates recognize only endometrial CA (EC) and CRC risks in MSH6+ carriers as clearly above SEER population estimates. Fur- ther, risks of other LS manifestations such as skin disease/Muir-Torre, ovarian CA (OC), and possible rare tumors in LS like sarcoma, have been minimally characterized in MSH6+ carriers. Methods: Pedi- gree data for 44 MSH6+ index (first-evaluated family member by our program) pts consecutively ascertained since 2009 at Fox Chase (FCCC) were reviewed. 1 pt w/a rare MSH6 uncertain variant w/personal history (PHx) of MSH6-expression deficient EC (age 50) and MSH6-deficient sebaceous skin CA (age 50) and a strong family history (FHx) c/w LS is also included here. 34% (15/44) index pts were referred to FCCC for cascade testing due to a known MSH6 PV in the family. Of the remaining 29 index pts, as- certainment included: 14% w/positive universal LS tumor screening, 21% w/early-onset or synchronous LS CA, 14% w/multi-gene panel for PHx of OC, 10% w/incidental MSH6+ result (2 had testing for PHx breast CA, 1 tumor genomic profiling), and 28% w/PHx and/or FHx of LS CA warranting genetic testing. Age of CA onset and path data were verified in > 90% index pts. Results: Index pts had a mean age of 55.5 yrs, and 77% were female. Overall, 11% (5/44) of MSH6+ index pts were found to have LS at diagnosis of synchronous primary CAs (3 EC/OC, 1 CRC/CRC, 1 CRC/EC), and 4/5 of these occurred <50 yrs. An additional 20% (9/44) index pts reported PHx of >2 metachronous LS CAs. OC was the presenting CA in 14% (6/44) female index pts; 2 additional index pts had rarer OC variants (Mullerian duct @ 41, primary peritoneal CA @ 50). Skin manifestations of LS were documented in 9.1% (4/44) index pts (3 sebaceous, 1 SCC in-situ/Bowen’s disease); 1 other family had documented sebaceous CAs in an FDR (father) but the 2 daughters seen @FCCC (both 30s) had yet to develop skin lesions. 2 index pts were found to have LS after developing early-onset breast CA (age 39) and contralateral breast CA (ages 50 and 54). Finally, 7% (3/44) index pts had a PHx of sarcoma: 2 were liposarcomas (ages 57 and 67), and 1 was a dermatofibrosarcoma. 2 other index pts had siblings w/childhood sarcomas. Conclusions: Our data, encompassing 44 MSH6+ pts evaluated in our clinic and consecutively ascertained, suggest MSH6 PV carriers develop synchronous primaries (11%), common and rare OC histologic types (18%), sarcomas (7%) and skin disease/Muir-Torre (9%). While common in the population and lower penetrance for CRC, MSH6 PV can behave in uncommon ways and may have significant extra-colonic CA risks such as OC, sarcoma and skin manifestations. Research Sponsor: U.S. National Institutes of Health.

10538 Poster Session

Identification and characterization of de novo TP53 mutation carriers in Li-Fraumeni syndrome families: A single institution experience.

Carlos Christian Vera Recio, Jessica Corredor, Elissa Dodd-Eaton, Angelica M. Gutierrez-Barrera, Najat C. Daw, Banu Arun, Wenyi Wang; University of Texas MD Anderson Cancer Center, Houston, TX; The University of Texas MD Anderson Cancer Center, Houston, TX

Background: Li-Fraumeni syndrome (LFS) is an inherited cancer syndrome mainly caused by a deleterious mutation in TP53. An estimated 48% of LFS patients present due to a deleterious de novo mutation (DNM) in TP53. The knowledge of DNM status, DNM or familial mutation (FM), of an LFS patient requires genetic testing of both parents which is often inaccessible, making de novo LFS patients an understudied population. Famdenovo.TP53 is a Mendelian Risk prediction model used to predict DNM status of TP53 mutation carriers based on the cancer-family history and several input genetic parameters, including disease-gene penetrance. The good predictive performance of Famdenovo.TP53 was demonstrated using data collected from four historical US cohorts. We hypothesize that by incorporating penetrance estimates that are specific for different types of cancers diagnosed in family members, we can develop a model with further improved calibration, accuracy and prediction. Methods: We present Famdenovo.CS, which uses cancer-specific penetrance estimates that were derived previously using a Bayesian semi-parametric competing risk model, to calculate the DNM probability. We use our model to analyze 101 families recently collected from the Clinical Cancer Genetic program at MD An- derson Cancer Center (CCG-TP53) that includes 20 families with known DNM status and 81 families with unknown DNM status. We used the concordance index (AUC), observed:expected ratios (OE) and Brier score (BS) to measure our model’s discrimination, calibration and accuracy, respectively. We estimate the proportion of probands that present a DNM and compare DNM to FM carriers in several areas including: cancer types diagnosed, age at diagnosis, number of primary cancers diagnosed, sex, amino acid change caused by mutation in TP53. Results: Famdenovo.CS showed equally good discrimination and calibration performance to Famdenovo.TP53, while improving the overall accuracy, demonstrated by a decrease in the Brier score of -0.09 (95% CI: [-0.02, -0.19]). Of the 101 probands in the CCG- TP53 cohort, we predict 39 to be DNMs and 62 to be FMs. The cancer types and ages of diagnosis observed in FMs and DNMs are similarly distributed. Conclusions: Famdenovo.CS shows improved model accuracy in the CCG cohort. DNMs in TP53 are a prevalent cause of LFS and we did not find differences in the clinical characteristics of DNM and FM carriers. Our model allows for a systematic identification and characterization of TP53 DNM carriers. Research Sponsor: U.S. National Institutes of Health, Can- cer Prevention and Research Institute of Texas.

10539 Poster Session

Differential effects of the 2015 American Cancer Society guidelines on screening mammography exams based on socioeconomic status.

Kathleen M. Capaccione, Sophia Huang, Elizabeth West, Aileen Deng, Mary M. Salvatore, Elise Desperito; Department of Radiology, Columbia University Irving Medical Center, New York, NY; Colum- bia University Irving Medical Center, New York, NY; Novant Health Cancer Institute, Mooresville, NC

Background: Guidelines for screening mammography have changed several times since initiation of regular screening mammography in the 1970’s. Most recently, in 2015, the American Cancer Society (ACS) revised their screening guidelines, recommending that a patient discuss screening mammography with her primary care doctor (PCP) between the ages of 40-44 and should begin yearly screening at age 45; after age 54, ACS recommended screening every other year. Prior to this, from 2003-2015, ACS had recommended screening mammography every year beginning at the age of 40. We hypothesized that these guidelines were adopted to varying degrees in different patient populations and may have disproportionately reduced screening mammography utilization in socioeconomically disadvantaged populations. Methods: Here, we analyzed monthly screening mammography rates over time in two large New York City hospitals, one in a socioeconomically advantaged area and the other in a socioeconomically disadvantaged area. Using our radiology records query system, we searched for monthly screening mammography numbers for women by decade from 2012 to 2018. We performed statistical analysis to evaluate changes in number of exams over time. Student’s t-tests were used to evaluate for significant differences. Results: In both groups of 40-49 year old patients, monthly mammograms increased from 2012-2016. In the socioeconomically advantaged group, this increase continued until 2018 resulting in an overall 400% increase in screening mammograms over time. The change in ACS screening guidelines had no effect on the rate of screening mammography in this group. Conversely, after the revision of the ACS guidelines in 2015, there was a marked decline in screening mammography in the 40-49 year old group in the socioeconomically disadvantaged population. By 2018, there was a statistically significant difference in women screened in all age groups (40-49, p<.0001; 50-59, p<.0001; 60-69, p<.01; 70-79, p<.0001; 80+, p<.0001) between these two patient populations. Conclusions: These data suggest that implementation of the 2015 ACS screening guidelines had a disproportionate effect on patients from socioeconomically disadvantaged areas and that these effects have led to significant disparities in screening mammography trends over time. We postulate that lower levels of health literacy may have contributed to misunderstanding of the screening guidelines. More research is needed to elucidate the underlying etiology of these disparities and ensure that women from all socioeconomic backgrounds receive appropriate screening mammography. Over time, this may result in disproportionate breast cancer morbidity and mortality in populations not receiving appropriate screening. Research Sponsor: None.

10540 Poster Session

Community-based lung cancer screening adherence to Lung-RADS recommendations.

Roger Kim, Katharine A. Rendle, Christine Neslund-Dudas, Robert T. Greenlee, Andrea N. Burnett- Hartman, Stacey A. Honda, Michael J. Simoff, Jennifer M. Croswell, Debra P. Ritzwoller, Anil Vachani; University of Pennsylvania, Philadelphia, PA; Henry Ford Health System, Detroit, MI; Marshfield Clinic Research Institute, Marshfield, WI; Kaiser Permanente/Colorado Institute for Health Research, Denver, CO; Hawaii Permanente Medical Group and Center for Integrated Health Care Research, Kaiser Perma- nente, Honolulu, HI; Healthcare Delivery Research Program, National Cancer Institute, Bethesda, MD; Kaiser Permanente Colorado Institute for Health Research, Denver, CO; Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA

Background: In the NLST and NELSON trials, most low-dose CT (LDCT) screen-detected lung cancers were not diagnosed during the first round of screening, suggesting that longitudinal adherence to lung cancer screening (LCS) recommendations is key. Adherence was as high as 95% in clinical trials, but limited data exist regarding LCS adherence in clinical practice. We aimed to determine adherence to Lung-RADS recommendations among community-based patients undergoing LCS. Methods: We performed a multicenter retrospective cohort study of patients screened for lung cancer at healthcare systems within the Lung Population-based Research to Optimize the Screening Process (PROSPR) Consortium. We included 55-80 year-old current or former smokers who received a baseline (T0) LDCT with a Lung-RADS score between January 1, 2015 and September 30, 2017 and excluded patients who were diagnosed with lung cancer prior to the T0 scan. Over a 24-month period, we calculated the proportion of patients adherent to Lung-RADS recommendations and evaluated associations with patient-level (age, sex, race, ethnicity, smoking status, body mass index, Elixhauser comorbidities, year of T0 scan, and Lung-RADS score) and census tract (median family income, level of education) data, using multivariable logistic regression with mixed effects to account for site variability. Results: Of the 6,723 patients in our cohort (median age 65 years [IQR 60-69]; 45.1% female; 73.0% white; 59.0% current smokers), 5,583 (83.0%) had Lung-RADS 1 or 2 T0 scans, 733 (10.9%) Lung-RADS 3, 274 (4.1%) Lung-RADS 4A, and 133 (2.0%) Lung-RADS 4B or 4X. Overall, 55.2% (3,709/6,723) of patients were adherent (Table). In the final multivariable model, Black patients had reduced adherence compared to white patients (adjusted odds ratio [aOR] 0.79, 95% CI 0.66-0.94), while greater adherence was observed in former smokers compared to current smokers (aOR 1.33, 95% 1.19-1.49). Com- pared to individuals with a negative T0 scan (Lung-RADS 1 or 2), those with Lung-RADS 3 (aOR 1.56, 95% CI 1.31-1.86), 4A (aOR 1.63, 95% CI 1.24-2.15), or 4B/4X (aOR 3.59, 95% CI 2.30-5.60) T0 scans had greater odds of adherence. Conclusions: In the largest study of real-world patients receiving LCS to date, adherence to Lung-RADS recommendations is lower than previously observed in clinical trials. Our results highlight the need for further study of system-level mechanisms to improve longitudinal LCS adherence rates. Research Sponsor: U.S. National Institutes of Health.

Lung-RADS Score Lung-RADS Recommendation Definition of Adherence Adherence, %

1 12 month LDCT 9-15 month LDCT or CT chest 52.5% (2,931/5,583) 2 3 6 month LDCT 3-9 month LDCT or CT chest 67.1% (492/733) 4A 3 month LDCT or PET/CT 2-6 month LDCT, CT chest, or PET/CT 66.4% (182/274) 4B Chest CT, PET/CT, and/or <3 month LDCT, CT chest, PET/CT, or procedure (bronchoscopy, 78.2% (104/133) 4X tissue sampling mediastinoscopy, percutaneous procedure, thoracic surgery)

10541 Poster Session

Global tissue stiffness on breast MR elastography: High-risk dense breast patients have higher stiffness compared to average-risk dense breast patients.

Bhavika K. Patel, Kay Pepin, Kathy R Brandt, Gina L. Mazza, Barbara A. Pockaj, Jun Chen, Yuxiang Zhou, Donald W. Northfelt, Karen Anderson, Brenda Ernst, Patricia Cronin, Sarwat Ahmad, Juliana Kling, Denise Millstine, Heidi Apsey, Celine M Vachon, Ilyse Nelson, Richard Ehman; Mayo Clinic, Phoenix, AZ; Mayo Clinic, Rochester, MN

Background: Biomechanical tissue properties may vary in the breasts of patients at elevated risk for breast cancer. We aim to quantify in vivo biomechanical tissue properties in various breast densities and in both normal risk and high risk women using Magnetic Resonance Imaging (MRI)/MRE and examine the association of biomechanical properties of the breast with cancer risk. Methods: In this IRB–approved prospective single-institution study, we recruited two groups of women differing by breast cancer risk to undergo a 3.0 T dynamic contrast enhanced MRI/MRE of the breast. Low-average risk women were defined as having no personal or significant family history of breast cancer, no prior high risk breast biopsies and a negative mammography within 12 months. High-risk breast cancer patients were recruited from those patients who underwent standard of care breast MR. Within each breast density group (non-dense versus dense), two-sample t-tests were used to compare breast stiffness, elasticity, and viscosity across risk groups (low-average vs high). Results: There were 50 low-average risk and 86 high-risk patients recruited to the study. The risk groups were similar on age (mean age = 55.6 and 53.6 years), density (68% vs. 64% dense breasts) and menopausal status (66.0% vs. 69.8%). Among patients with dense breasts, mean stiffness, elasticity, and viscosity were significantly higher in high risk patients (N = 55) compared to low-average risk patients (N = 34; all p < 0.001). In the multivariate logistic regression model, breast stiffness remained a significant predictor of risk status (OR=4.26, 95% CI [1.96, 9.25]) even after controlling for breast density, MRI BPE, age, and menopausal status. Similar results were seen for breast elasticity (OR=4.88, 95% CI [2.08, 11.43]) and viscosity (OR=11.49, 95% CI [1.15, 114.89]). Conclusions: Structurally-based, quantitative biomarker of tissue stiffness obtained from global 3D breast MRE is associated with differences in breast cancer risk in dense breasts. As such, tissue stiffness could provide a novel prognostic marker to help identify the subset of high-risk women with dense breasts who would benefit from increased surveillance. Research Sponsor: Mayo Clinic MEGA grant.

MRE parameters MRI background parenchymal enhancement (BPE) by breast density and risk status. Non-Dense, Non-Dense, Non- Dense, Dense, Normal Risk High Risk Dense Normal Risk High Risk Dense Characteristic (N = 16) (N = 31) p-Value (N = 34) (N = 55) p-Value

Breast Stiffness Mean (SD) 1.07 (0.18) 1.69 (1.19) 0.008 1.16 (0.21) 2.14 (1.12) <0.001 Breast Stiffness Median 1.05 1.09 — 1.15 1.37 — (Range) (0.65–1.44) (0.54–4.36) (0.76–1.67) (0.92–4.09) BPE (#, % Minimal or Mild) 14 (87.5%) 28 (90.3%) >0.99 20 (58.8%) 35 (63.6%) 0.66

10542 Poster Session

Risk factors for early-onset colorectal cancer in China.

Zhe Pan, Junfeng Huang, Mingkai Huang, Zhiyuan Yao, Jiongqiang Huang, Jingsong Chen, Rongchang Wang; Department of Gastrointestinal Surgery, The First Affiliated Hospital of Guangzhou Medical Uni- versity, Guangzhou, China; State Key Laboratory of Respiratory Disease, National Clinical Research Center for Respiratory Disease, Guangzhou Institute of Respiratory Health, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China

Background: The incidence of colorectal cancer among persons aged < 50 years (early-onset colorectal cancer, EOCRC) has increased since the early 1990s. However, the risk factors contributing to this trend remain largely unknown. Methods: We conducted a retrospective study of participants who were aged < 50 years and without a previous cancer history, using the China Kadoorie Biobank cohort study. We analyzed data related to demographics, lifestyle habits, family history, and comorbidities of EOCRC cases with participants without colorectal cancer in this age group (controls). Univariate and multivariate-adjusted cox regression models were used to estimate the associations with risk factors. Results: We identified 225 EOCRC cases and 88842 controls that include the final analyses. Of the 225 EOCRC patients, 105 (46.7%) were colon cancers and 120 (53.3%) were rectum cancers. EOCRC cases were older, have more intake of fish and eggs, have higher BMIs, diabetes, and family history of cancer compared with controls (P < 0.05). After adjustment for potential confounding factors, increasing age (HR 2.18, 95%CI 2.05-2.31), BMI (HR 1.06, 95%CI 1.01-1.11), family history of cancer (HR 1.41, 95%CI 1.00-1.98), and more intake of fish (HR 1.54, 95%CI 1.09-2.19) were significantly associated with a higher risk of EOCRC. In sensitivity analyses stratified by cancer site (colon and rectum), the results remained consistent. Conclusions: Based on the large Chinese cohort study, we found increasing age, higher BMI or obesity, family history of cancer, and more intake of fish were independent risk factors for EOCRC. Further studies are needed to identify factors that cause the increasing incidence of EOCRC in China and other countries, and explore the potential mechanism behind. Research Sponsor: This work was supported by grants (2016YFC0900500, 2016YFC0900501, 2016YFC0900504,) from the National Key Research and Development Program of China, grants from the Kadoorie Charita- ble Foundation in Hong Kong and grants (088158/Z/09/Z, 104085/Z/14/Z, 1040, Grants from the Na- tional Natural Science Foundation of China (number 82072620) and Guangdong Basic and Applied Basic Research Foundation (number 2020A1515110056).

Unadjusted Models Adjusted Models Variables HR 95%CI P Value HR 95%CI P Value

Age 2.15 2.03-2.28 < 0.001 2.17 2.04-2.30 < 0.001 BMI 1.06 1.02-1.10 0.006 1.05 1.00-1.10 0.0305 Female 0.99 0.76-1.30 0.96 0.91 0.54-1.53 0.7144 Alcohol user 0.95 0.73-1.23 0.66 0.98 0.72-1.32 0.8724 Intake of Eggs 1.93 1.03-3.65 0.04 1.70 0.90-3.21 0.1052 Intake of Fish 1.66 1.23-2.24 < 0.001 1.39 1.02-1.90 0.0398 Family history of cancer 1.60 1.14-2.25 0.006 1.41 1.00-1.98 0.0477 Diabetes 2.66 1.31-5.38 0.007 1.21 0.59-2.47 0.6036

10543 Poster Session

Analysis of breast cancer screening results in urban areas of Henan Province.

Guo Lan-Wei; The Affiliated Cancer Hospital of Zhengzhou University, Zhengzhou, China

Background: Breast cancer is the most common female cancer in China. Reasonable and effective screening is an important means to reduce the mortality of breast cancer. This study was to evaluate the compliance and efficacy of breast cancer screening in urban areas of Henan province from 2013 to 2019. Methods: A cluster sampling method was used to select the residents of 40-74 years old in Henan province to investigate the risk factors and breast cancer risk assessment. For subjects with high risk of breast cancer, those aged 40-44 conduct breast ultrasound examination; those aged over 45 conduct breast ultrasound combined with mammography examination. BI-RADs classification was used as the evaluation standard in both examinations. BI-RADS 3 class was defined as suspicious and BI-RADS$4 class was defined as positive. Chi-square test was used to analyze the compliance rate and breast cancer detection rate among different groups. Results: A total of 29 111 residents at high risk for breast cancer were recruited in this study. They were 55.03 ± 7.91 years old, of which 13 760 took the following breast ultrasound or mammography examination, yielding a participation rate of 47.27% (13 760/ 29 111). The detection rates of suspected positive patients, positive patients and breast cancer patients were 23.40% (3 220/13 760), 2.55% (351/13 760) and 0.30% (41/13 760), respectively. Among the screening population, the highest detection rate of suspected positive patients was found in the 45-49 age group [27.79% (935/3 365)], the highest detection rate of positive patients was found in the 50-54 age group [2.98% (97/3 257)], and the highest detection rate of breast cancer was found in the 65-69 age group [0.49% (5/1 012)]. The detection rate of breast cancer positive patients by breast ultrasound combined with mammography examination was 2.95% (316/10 728), which was significantly higher than that of breast ultrasound alone [1.99% (213/10 728)] or mammography examination alone [1.25% (134/10 728)]. Conclusions: Breast ultrasound combined with mammography examination as a means of screening for breast cancer can significantly increase the detection rate of breast cancer positive patients. The next step should be to improve the compliance of the population, as well as the organization’s implementation and service capabilities of the screening provider, to improve screening effectiveness. Research Sponsor: Key Science and Technology Program of Henan Prov- ince, China (192102310353).

10544 Poster Session

An ultrasensitive method for noninvasive pan-cancer early detection based on targeted methylation sequencing of cell-free DNA.

Tiancheng Han, Yuanyuan Hong, Pei Zhihua, Song Xiaofeng, Jianing Yu, Geng Chen, Weizhi Chen, Ji He; Genecast Precision Medicine Technology Institute, Beijing, China; Genecast Biotechnology Co., Ltd., Beijing, China; Genecast, Beijing, China; Genecast Biotechnology Co.,Ltd., Beijing, China; Gene- cast Biotechnology Co., Ltd, Wuxi, China

Background: Screening the biomarkers from the cell-free DNA (cfDNA) of peripheral blood is a non-invasive and promising method for cancer diagnosis. Among diverse types of biomarkers, epigenetic biomarkers have been reported to be one of the most promising ones. Epigenetic modifications are widespread on the human genome and generally have strong signals due to the similar methylation patterns shared by adjacent CpG sites. Although some epigenetic diagnostic methods have been developed based on cfDNAs, few of them could be applied to pan-cancer and their sensitivities are barely satisfac- tory for early cancer detection. Methods: Targeted methylation sequencing was performed using our in- house-designed panel targeting regions with abundant cancer-specific methylation CpGs. The cfDNA samples from 80 healthy individuals and 549 cancer patients of 14 cancer types were separately sequenced. The dataset was randomly split into one discovery dataset and one validation dataset. More- over, cfDNA samples from four cancer patients were diluted with the healthy cfDNAs to generate 12 in vitro simulated samples with low circulating tumor DNA (ctDNA) fraction. Additionally, DNAs extracted from 130 unmatched tumor formalin fixation and paraffin embedding (FFPE) samples of 10 cancer types were sequenced to screen the diagnostic biomarkers. Adjacent CpG sites were first merged into methylation-correlated blocks (MCB) according to their correlations of methylation levels in tumor DNAs. The MCBs with higher methylation levels in tumor DNAs than that of healthy cfDNAs (from the discovery dataset) were defined as our hypermethylation biomarkers. For each cfDNA sample, a hypermethylation score (HM-score) was computed to measure the overall methylation level difference of selected biomarkers. The performance of our method was evaluated with the real-world dataset, while the limit of detection was estimated using the simulated low-ctDNA samples. Results: Our model based on 37 hypermethylation MCB biomarkers achieved an area under the curve (AUC) of 0.89 and 0.86 in the real-world pan-cancer discovery and validation cfDNA datasets, respectively. Furthermore, the overall specificity and sensitivity are 100% and 76.19% in the discovery dataset, and 96.67% and 72.86% in the validation dataset. In the validation dataset, 28/40 (70%) of early-stage colorectal cancer patients and 10/20 (50%) of non-small-cell lung cancer patients were successfully diagnosed. Additionally, all the simulated samples with theoretical ctDNA factions over 0.5% were predicted as diseased, demonstrating the ability of our method to detect tumor signals at early stages. Conclusions: Our cfDNA-based epigenetic method outperforms currently available methods in various cancer types, and is promising to be applied to early-stage cancer detection and samples with low ctDNA fractions. Research Sponsor: None.

10545 Poster Session

Impact of the sessile serrated polyp pathway on predicted colorectal cancer outcomes in the CRC-AIM model.

John B. Kisiel, Steven H. Itzkowitz, A. Burak Ozbay, Leila Saoud, Marcus Parton, David A. Lieberman, Paul J. Limburg; Mayo Clinic, Rochester, MN; Icahn School of Medicine at Mount Sinai, New York, NY; Exact Sciences Corporation, Madison, WI; Oregon Health & Science University, Portland, OR

Background: Approximately 20–30% of colorectal cancers (CRC) arise from the serrated polyp pathway. The multitarget stool DNA (mt-sDNA) test has greater sensitivity to detect sessile serrated polyps (SSPs) than a leading fecal immunochemical test (FIT). However, most modeling analyses do not account for the contribution of the SSP pathway to risk of CRC. We used the CRC-AIM model to assess the impact of the SSP pathway on predicted CRC outcomes with mt-sDNA or FIT screening. Methods: A simulated cohort of average-risk US patients underwent triennial mt-sDNA or annual FIT screening from ages 50–75. The percentage of CRCs arising from the SSP pathway were modeled at 0% (base case), 20%, and 30%, with stool screening adherence based on theoretical (100%) or previously reported (mt-sDNA 71%; FIT 43%) rates. Published SSP sensitivities for mt-sDNA and FIT were used. All other model inputs were identical to CISNET models. Sensitivity analyses used screening adherence rates of 40–70%. Key outcomes were life-years gained (LYG), CRC incidence and CRC mortality per 1000 patients. Results: Including SSPs in the model demonstrated a greater loss of LYG with FIT than mt-sDNA (Table). At 100% adherence, compared with base case, modeling 20% or 30% SSP pathway CRCs resulted in a decrease of 9–15 LYG with FIT and 2–4 LYG with mt-sDNA, a decrease in CRC incidence reduction of 3.9–6.1% with FIT and 0.7–1.1% with mt-sDNA, and a decrease in CRC mortality reduction of 2.6–4.0% with FIT and 0.4–0.8% with mt-sDNA. Using previously reported adherence, compared with base case, modeling 20% or 30% SSP pathway CRCs resulted in a decrease of 13–20 LYG with FIT and 2–5 LYG with mt-sDNA, a decrease in CRC incidence reduction of 4.4–6.9% with FIT and 0.6–1.1% with mt-sDNA, and a decrease in CRC mortality reduction of 3.5–5.4% with FIT and 0.4–0.9% with mt-sDNA. Assuming reported adherence and 30% SSP pathway CRCs, mt-sDNA had 48 more LYG, 14.6% greater CRC incidence reduction, and 12.4% greater CRC mortality reduction than FIT. Assuming 30% SSP pathway CRCs, outcomes favored mt-sDNA vs FIT even after modeling equivalent adherence rates ranging from 40–70%. Conclusions: After incorporating the SSP pathway into the model, outcomes with mt-sDNA neared those of FIT at 100% screening adherence rates and surpassed FIT at more realistic reported screening adherence rates. Research Sponsor: Exact Sciences Corporation.

CRC outcomes for triennial mt-sDNA and annual FIT after assuming 0%, 20%, or 30% SSP pathway CRCs and assuming either 100% or reported screening adherence rates. Triennial mt-sDNA Annual FIT % CRC CRC CRC CRC CRCs Incidence Mortality Incidence Mortality Screening Adherence from Reduction, Reduction, Reduction, Reduction, Rates SSPs LYG % % LYG % % 100% 0% 300 64.2 72.0 321 69.4 76.9 20% 298 63.5 71.6 312 65.5 74.3 30% 296 63.1 71.2 306 63.3 72.9 mt-sDNA, 71%; FIT, 0% 286 61.0 68.7 253 52.2 60.8 43% 20% 284 60.4 68.3 240 47.8 57.3 30% 281 59.9 67.8 233 45.3 55.4

Data are per 1000 individuals.

10546 Poster Session

Contribution of family history of melanoma to associated risk factors: Analysis of an Internet-based risk assessment tool.

Ryan M O’Keefe, Michael Joseph LaRiviere, Carolyn Vachani, Margaret K. Hampshire, Christina Bach, Karen Arnold-Korzeniowski, Marisa Healy, James M. Metz, Christine E. Hill-Kayser; University of Penn- sylvania Perelman School of Medicine, Philadelphia, PA; University of Pennsylvania, Philadelphia, PA; University of Pennsylvania Health System, Philadelphia, PA; University of Pennsylvania, Department of Radiation Oncology, Philadelphia, PA; Hospital of the University of Pennsylvania, Philadelphia, PA

Background: Risk factors for melanoma (ML) include UV exposure, sunburns, multiple nevi, and fair skin. Smoking and alcohol use may also play a role. Demographics associated with increased UV exposure include male gender, white race, age 18-29, and high-income. Those with family history (FH) of ML have increased risk of developing ML themselves yet are often unaware or do not engage in preventive behaviors. We sought to understand the association between FH of ML and other personal risk factors for cancer. Methods: Since 2009, voluntary participants could receive personalized information regarding their cancer risk via the OncoLink.org Reduce My Risk tool. Survey data was collected on demographics, FH, risk-factors, and risk-associated behaviors. Use of data was IRB approved. Differ- ences between respondents with v. without FH of ML were analyzed using chi-square test and adjusted via logistic regression. Results: 25255 responses were analyzed; 1928 (7.6%) had FH of ML. Of these, median age was 26 (range 18-75), 73.1% were female, 88.6% from North America, 91.4% White, 78.2% had at least some college, and 62.0% household income > $45,000. Comparing those with FH of ML to those without, no significant differences were observed in education, income, or home setting. Those with a FH of ML were less likely to be “light” smokers ( < 1 pack per day) but were more likely to be both “light” drinkers and “heavy” drinkers (Table). There were no significant differences in BMI or exercise habits. Those with FH of ML were more likely to sunbathe, to have had “blistering” sunburns, have 50+ moles, show signs of sun damage on their skin, and have dysplastic nevi (Table). There were no differences in use of indoor tanning beds (Table). Conclusions: Those with a FH of ML were more likely to sunbathe and to report skin damage and history of blistering sunburns; they were more likely to use tanning beds, though not statistically significant. These behaviors are modifiable and may suggest pa- rental influence on sun-protective behaviors. Those with FH also reported increased non-modifiable risks. Future work should continue to explore targeted intervention for those with a FH of ML to educate on risk and promote sun-protective behaviors. Research Sponsor: OncoLink - Hospital of the University of Pennsylvania.

Pos FH Neg FH OR 95% CI p-value

< 1 Pack / Day 129 (6.7%) 2043 (8.8%) 0.74 0.61–0.89 0.002 1+ Packs / Day 56 (2.9%) 644 (2.8%) 1.02 0.76–1.35 0.885 1-7 Drinks / Week 886 (46.0%) 9486 (40.7%) 1.22 1.10–1.36 < 0.001 8+ Drinks / Week 232 (12.0%) 2565 (11.0%) 1.29 1.09–1.51 0.002 "Love to Sunbathe" 286 (14.8%) 2510 (10.8%) 1.15 1.00–1.33 0.05 Tanning Bed Use 524 (27.2%) 4732 (20.3%) 0.94 0.84–1.06 0.316 Sunburns As Child or Teen 857 (44.5%) 5945 (25.5%) 1.71 1.54–1.90 < 0.001 50+ Moles on Skin 466 (24.2%) 3305 (14.2%) 1.46 1.29–1.63 < 0.001 Dysplastic Nevi 47 (2.4%) 203 (0.9%) 1.66 1.18–2.31 0.003

10547 Poster Session

Red blood cell folate, high-risk human papillomavirus risk and cervical intraepithelial neoplasia development: A large Chinese community-based cohort study of cervical screening.

Wei Wang, Aimin Yang, Yuanxing Li, Haixia Luo, Jing Yang, Weihong Zhao, Zhilian Wang, Zhe Wang, Chen Wang, Lili Zhang, Xiaoqiang Su, Wenhao Wang, Jintao Wang, Weiguo Lu, Li Li, Jinghui Song, Min Hao; The Second Hospital of Shanxi Medical University, Taiyuan, China; The Chinese University of Hong Kong, Hong Kong, China; Shanxi Medical University, Taiyuan, China; Women’s Hospital, Zhejiang University School of Medicine, Hangzhou, China; Guangxi Medical University Cancer Center, Nan Ning, China; Affiliated Hospital of Inner Mongolia Medical University, Huhhot, China; Second Hospital of Shanxi Medical University, Taiyuan, China

Background: Although low folate status has been implicated in cervical carcinogenesis, large-scale population-based prospective cohort studies controlling for high-risk types of human papillomavirus (hrHPV) infection are lacking. The aim of this study is to evaluate the associations of red blood cell (RBC) folate, hrHPV infection risk and cervical intraepithelial neoplasia (CIN) development. Methods: In this prospective, population-based cohort study, we analyzed the cross-sectional data of 2304 women from a large cervical cancer screening program of 40,000 women aged 19-65 years in the Chinese rural area from 2014-2015. We conducted a nested case-control study including 35 CIN1 progression cases and 105 CIN1 regression controls. A logistic regression model was used to evaluate the associations of RBC folate and hrHPV infection risk and CIN1 development. Results: The median RBC folate concentration decreased gradually with cervical lesion severity. The risks of CIN 1 and CIN2 or worse (CIN 2+) in the 1st quartile of RBC folate concentration were significantly higher than those in the 4th quartile (Odds Ratio [OR], 2.27; 95% confidence interval [CI], 1.71-3.01 and OR, 2.33; 95% CI,1.52-3.56; respectively). We did not observe a significant relationship between hrHPV infection and CIN1 risk in the unadjusted and adjusted models, however, a statistically significant association was observed for CIN2+. Interestingly, RBC folate concentration was not associated with hrHPV infection among women with CIN1 or CIN2+. After full adjustment for potential confounders, a highly significant inverse linear relation between RBC folate concentration and CIN2+ was observed (P-overall＜0.001, P-nonlinearity = 0.969). We further observed a positive additive interaction between RBC folate concentration and hrHPV infection on the risk of CIN2+ (P-interaction＜0.01). Moreover, during the 21-month follow-up, CIN1 progression risk was significantly higher in the lowest RBC folate quartile (1st quartile compared with 4th quartile: OR, 3.86; 95% CI,1.01-14.76). Conclusions: Our findings indicates that RBC folate is inversely associated with the risk of higher-grade CIN and CIN1 progression in the Chinese rural population, either with or without hrHPV infection. Therefore, improving folate status has the potential to prevent higher-grade CIN and cervical cancer among women in the areas without mandatory folic acid food fortification. Clinical trial information: ChiCTR-ROC-15006479. Research Sponsor: This project was supported by the Special Public Welfare Industry Research of National Health and Family Planning Commission of China (grant number: 201402010).

10548 Poster Session

Risk-reducing salpingo-oophorectomy and breast cancer incidence among BRCA-mutation carriers.

Tamar Perri, Shani Naor-Ravel, Perry Eliassi-Revivo, Dror Lifshitz, Eitan Friedman, Jacob Korach; Gyne- cologic Oncology Department, Sheba Medical Center, Ramat Gan, Israel; Sackler Faculty of Medicine, Tel Aviv University, Ramat Aviv, Israel; Sackler Faculty of Medicine, Tel Aviv University, Ramat Gan, Is- rael; Gynecologic Oncology Department, Sheba Medical Center, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel

Background: Uncertainty exists with regard to the role of bilateral salpingo-oophorectomy in altering the risk of breast cancer in BRCA-mutation carriers. Methods: Included were 1645 healthy Jewish Israeli BRCA1/2 -mutation carriers from a single center without prophylactic mastectomy. Carriers with and without risk-reducing bilateral salpingo-oophorectomy (RRBSO) were matched according to BRCA-mutation type (BRCA1 vs. BRCA2) and year of birth (±1 year). Hormonal and reproductive variables were compared and incidence of breast cancer recorded. Association between RRBSO and breast cancer was studied. Results: Seventy-seven and 50 matched-pairs had BRCA1 and BRCA2 mutation respectively. Fifty-two carriers had breast cancer, 21 in RRBSO-group and 31 in no- RRBSO group, with no statistically significant difference. When analysing each mutation group separately, stratified by age at surgery, no association between RRBSO and breast cancer incidence was found among BRCA1-mutation carriers. However, in BRCA2 mutation carriers, RRBSO was associated with a statistically significant decreased overall incidence of breast cancer, HR = 0.2 (confidence interval 0.44-0.913, p = 0.038). Breast cancer incidence was lower after 5, 10,15 and 20 years in BRCA2-mutation carriers with RRBSO compared to no-RRBSO. Age at menarche, age at surgery, parity and oral contraceptive use were not significant risk factors for breast cancer. Hormone replacement therapy was used by 62 mutation carriers, 52 in the RRBSO group and 10 in the no-RRBSO group, and its use did not alter breast cancer risk (p = 0.463). Conclusions: According to our findings, RRBSO is associated with a reduced risk of breast cancer only in BRCA2 mutation carriers, regardless of HRT use. Research Sponsor: None.

10549 Poster Session

Attention to diet, exercise, and weight in the oncology clinic: Results of a national patient survey.

Jennifer A. Ligibel, Lori J. Pierce, Catherine M. Bender, Tracy E Crane, Christina Marie Dieli-Conwright, Judith O. Hopkins, Larissa A. Korde, Gregory A. Masters, Caroline Schenkel, Liz Garrett-Mayer, Sweatha Katta, Janette K. Merrill, Jeannine M. Salamone, Abenaa M. Brewster; Dana-Farber Cancer In- stitute, Boston, MA; University of Michigan, Ann Arbor, MI; University of Pittsburgh, Pittsburgh, PA; University of Arizona Cancer Center, Tucson, AZ; University of Southern California, Los Angeles, CA; NSABP/NRG Oncology, and Novant Helath Forsyth Medical Center/Southeast Clinical Oncology Re- search Consortium, Winston Salem, NC; Clinical Investigations Branch, National Cancer Institute, Be- thesda, MD; Helen F Graham Cancer Center, Newark, DE; American Society of Clinical Oncology, Alexandria, VA; The University of Texas MD Anderson Cancer Center, Houston, TX

Background: Obesity and related factors are increasingly associated with increased risk of developing and dying from cancer. The American Society of Clinical Oncology (ASCO) conducted a survey of cancer patients to assess their experience in receiving recommendations and referrals related to weight, diet and exercise as a part of their cancer care. Methods: An online survey was distributed to potential participants between March and June 2020 via ASCO channels and patient advocacy organizations, with an estimated reach of over 25,000 individuals. Eligibility criteria included being 18 years, living in the US, and having been diagnosed with cancer. Logistic regression was used to determine factors associated with recommendation and referral patterns. Results: In total, 2419 individuals responded to the survey. Most respondents were female (75.5%), 61.8% had an early-stage malignancy, 38.2% had advanced disease, and 49.0% were currently receiving treatment. Breast cancer was the most common cancer type (36.0%). Average BMI was 25.8 kg/m2. The majority of respondents consumed £2 servings of fruits and vegetables per day (50.9%) and exercised £2 times per week (50.4%). Exercise was addressed at most or some oncology visits in 57.5% of respondents, diet in 50.7%, and weight in 28.4%. Referrals were less common: 14.9% of respondents were referred to an exercise program, 25.6% to a dietitian and 4.5% to a weight management program. In multiple regression analyses, racial and ethnicity minority respondents were more likely to receive advice about diet (Odds Ratio [OR] 1.92, 95% CI 1.56-2.38) and weight (OR 1.64, 95% CI 1.23-2.17) compared to non-Hispanic whites, individuals diagnosed with cancer in the past 5 yrs (vs > 5 yrs) were more likely to receive advice about exercise (OR 1.48, 95% CI 1.23-1.79), and breast cancer patients were more likely to receive advice about exercise (OR 1.37, 95% CI 1.11-1.68) and weight (OR 1.46, 95% CI 1.03-2.07) than other cancer patients. Overall, 74% of survey respondents had changed their diet or exercise after cancer diagnosis. Respond- ents reporting that their oncologist spoke to them about increasing exercise or eating healthier foods were more likely to report a change in behavior than those whose oncologists did not (exercise: 79.6% vs 69.0%, P < 0.001; diet 81.1% vs 71.4%, P < 0.001). Respondents whose oncologist had spoken to them about exercise were more likely to exercise > 2 times per week compared to respondents whose oncologists did not address exercise (53.5% vs 44.1%, P < 0.001). Conclusions: In a national survey of oncology patients, slightly more than half of respondents reported attention to diet and exercise during oncology visits. Provider recommendations for diet and exercise were associated with positive changes in these behaviors. Additional attention to diet and exercise as part of oncology visits is needed to help support healthy lifestyle change in cancer patients. Research Sponsor: None.

10550 Poster Session

Reductions in cancer screening: The consequence of changes in routine care during the COVID-19 pandemic.

Ashley Kim, Matthew Gitlin, Ela Fadli, November McGarvey, Ze Cong, Karen Chung; GRAIL, Inc., Menlo Park, CA; BluePath Solutions, Los Angeles, CA

Background: The COVID-19 pandemic imposes significant impact on daily activities with regard to public health orders and individual responses to the pandemic. Much of the direct or indirect impact is potentially in reductions in healthcare encounters for services such as preventive care. Here, we quantified changes in cancer screening rates to better understand the impact of the evolving COVID-19 implications and shifts in health-seeking behaviors. Methods: We conducted a retrospective analysis of cancer screening rates during March-June 2019 (pre-COVID-19) and March-June 2020 (post-COVID- 19 restrictions), using Optum’s de-identified Clinformatics Data Mart Database which includes Medi- care and commercially insured members. Members meeting age and/or sex criteria as detailed in the United States Preventive Services Task Force recommendations for breast, colorectal, lung, prostate, and cervical cancer screening represented the eligible membership for screening. Procedure and laboratory services were used to identify those who received cancer screening. Analyses were conducted cross-sectionally by cancer screening type. Results: Eligible cohorts were identified from insured members within March-June 2019 and 2020 (2019: 17,931,318; 2020: 17,521,411). The percent of eligible members screened in March-June 2019 was 19.3%, 9.4%, 16.7%, 0.4%, and 7.8% for breast, cervical, prostate, lung, and colorectal cancer, respectively. Changes in screening rates from 2019 to 2020 are summarized in Table, with the sharpest decline in April. The percent change from 2019 to 2020 during the combined March-June period for each cancer screening type was statistically significant (p<0.0001). Conclusions: Routine cancer screening rates from March-June 2020 showed meaningful reductions when compared to the same period in 2019, with substantial declines during the initial peak of the pandemic in April. These declines may be impacted by variations in regional restrictions with tighter restrictions leading to larger screening declines and loosening restrictions reflecting catch-up screening. Efforts to promote cancer screening in a safe and timely manner are crucial given individual risk factors, to reduce later stage cancer diagnoses and improve clinical outcomes. Research Sponsor: GRAIL, Inc.

Percent change in eligible members screened for cancer (March-June 2019/2020). Breast Cervical Prostate Lung Colorectal

Population Included Females aged 50- Females aged 21- Males aged 55-69 Males and Females Males and Females 74 years, no history 65 years, no history years, no history of aged 55-80 years, no aged 50-75 years, no of breast cancer of cervical cancer prostate cancer history of lung cancer history of colorectal cancer March-June -42.8% -42.8% -28.7% -31.5% -42.3% March -34.5% -35.3% -24.8% -14.6% -29.2% April -85.0% -78.9% -61.8% -84.7% -73.4% May -50.7% -49.8% -32.6% -40.6% -52.9% June -4.0% -2.2% 8.5% 19.0% -18.3%

10551 Poster Session

Trends in cancer screening volumes at an urban health center during the COVID-19 pandemic.

Mahir Khan, Margaret Wright, Karriem Watson, Shikha Jain; University of Illinois at Chicago, Chicago, IL; University of Illinois Cancer Center, Chicago, IL; University of Illinois College of Medicine, Chicago, IL

Background: The Coronavirus-19 (COVID-19) pandemic has disrupted cancer screening for reasons including healthcare resource preservation, infection control efforts, and patient factors. There is limited literature quantifying this interruption of care, particularly in vulnerable and racial/ethnic minorities. Methods: We compared the volume of cancer screening at the University of Illinois Hospital & Health Sciences System before and during the COVID-19 pandemic using data obtained from the electronic medical record. Modalities included mammogram, ultrasound, and MRI for breast; Pap test for cervical; colonoscopy, CT colonography, and flexible sigmoidoscopy for colorectal; low-dose CT for lung; and prostate-specific antigen test for prostate. Of note, screening and diagnostic tests could not be distin- guished for colorectal cancer. We examined percent changes in cancer screening counts for each month from February 2020-August 2020, using January 2020 as a reference. Results were stratified by gender, race, and ethnicity. Results: Screening volume declined rapidly after January 2020, with the nadir for each cancer site occurring in April 2020: breast (n = 0, -100%), cervical (n = 169, -84%), colorectal (n = 35, -89%), lung (n = 0, -100%), and prostate (n = 108, -72%). Values recovered by August 2020 for most cancer sites except cervical cancer, which remained decreased (-23%). There were no differences in screening trends by gender. With respect to race, breast screening volume in Black patients decreased earlier and exhibited slower recovery compared to White patients. White patients had poorer cervical screening recovery than Black patients by August 2020 (-60% vs. -23%). Hispanics had poorer recovery of breast screening compared to non-Hispanics by August 2020 (-23% vs. 6%). Conclusions: We observed widely decreased cancer screening attributable to COVID-19. Breast cancer screening data specifically showed persistent disparities affecting Black and Hispanic patients. Despite the reassuring recovery of multiple screening methods by August 2020, an increase above baseline is needed to compensate for initial declines. Further studies will likely reveal long-term consequences of this unprecedented situation. Research Sponsor: None.

Breast Cervical Colorectal Lung Prostate Count (% Count (% Count (% Count (% Count (% change) change) change) change) change)

Jan 2020 555 (Reference) 1041 333 (Reference) 14 (Reference) 390 (Reference) (Reference) Feb 2020 528 (-5%) 921 (-12%) 298 (-11%) 33 (136%) 428 (10%) March 2020 280 (-50%) 744 (-29%) 173 (-48%) 23 (64%) 303 (-22%) April 2020 0 (-100%) 169 (-84%) 35 (-89%) 0 (-100%) 108 (-72%) May 2020 0 (-100%) 224 (-78%) 67 (-80%) 3 (-79%) 209 (-46%) June 2020 222 (-60%) 580 (-44%) 168 (-50%) 6 (-57%) 266 (-32%) July 2020 523 (-6%) 690 (-34%) 266 (-20%) 9 (-36%) 373 (-4%) August 545 (-2%) 797 (-23%) 319 (-4%) 20 (43%) 353 (-9%) 2020

Cancer screening before and during the COVID-19 pandemic.

10552 Poster Session

Efficacy of HPV vaccination among seropositive, DNA negative cohorts.

Colm Mac Eochagain, Robert Power, Imelda Parker; St James’ Hospital, Dublin, Ireland; Trinity College Dublin, Dublin, Ireland; Cancer Trials Ireland, Dublin, Ireland

Background: Prophylactic HPV vaccination of naïve cohorts is known to be highly effective in the prevention of incident HPV infection and HPV-associated cervical premalignancy. Conversely, vaccination of women with active (DNA+) HPV infections has been demonstrated to be ineffective. Vaccine efficacy among previously exposed, but currently uninfected women, i.e. those who have serological evidence of a prior HPV infection without corresponding detectable HPV DNA, remains incompletely defined. This meta-analysis assessed the serotype-specific efficacy of prophylactic HPV vaccination against HPV16/ 18 persistent infection (PI) and cervical intraepithelial neoplasia (CIN) among seropositive, DNA negative (SPDN) women enrolled to RCTs of HPV L1-based vaccines. Methods: The study protocol was prospectively registered on PROSPERO (CRD42020206888). Searches were conducted on 08/16/20 on MEDLINE, EMBASE, SCOPUS and CENTRAL. RCTs of L1-based prophylactic bivalent or quadrivalent HPV vaccines, reporting serotype-specific clinical efficacy endpoints in the HPV16/18 seropositive, DNA-negative populations were included. Two authors independently screened studies, extracted data and assessed for bias. Data for SPDN women were extracted from subgroup analyses within primary and secondary publications, publication supplements, and manufacturers’ clinical study reports. Relative risks (RR) of 6-month persistent infection (6mPI), 12-month persistent infection (12mPI), CIN1+ and CIN2+ were pooled using a random-effects model. Results: A total of 1727 citations were reviewed. 8 studies, with a total of 9569 SPDN participants, met all eligibility criteria. The relative risk of 6mPI (RR: 0.22, 95% CI 0.08-0.61, p = 0.018), 12mPI (RR: 0.20, 95% CI 0.05-0.80, p = 0.035), CIN1+ (RR:0.13, 95% CI 0.05-0.30, p = 0.003) and CIN2+ (RR: 0.15, 95% CI 0.04-0.59, p = 0.022) was significantly reduced in the vaccinated compared to the unvaccinated group. The number needed to vaccinate (NNV) to prevent one case of CIN1+ and CIN2+ was 152 and 208 respectively. Conclusions: Our findings suggest high serotype-specific efficacy for HPV vaccination among cohorts of women with evidence of prior HPV 16/18 infections. Women without DNA evidence of active infection may be sero- negative, implying naïve status and > 99% efficacy; or seropositive, implying SPDN status and 87% efficacy against CIN1+ for HPV 16/18. Women without DNA evidence of HPV 16/18 infection should be offered prophylactic HPV vaccination regardless of prior exposure history. Research Sponsor: None.

10553 Poster Session

Sociodemographics, lifestyle determinants, and viral infections in patients with head and neck cancer undergoing radiotherapy in Taiwan: A prospective cohort study.

Kevin Sheng-Kai Ma, Jasmine Tan, Angel Alfonso Velarde Lopez; Department of Life Science, National Taiwan University, Taipei City, Taiwan; School of Medicine, National Taiwan University, Taipei City, Tai- wan; Center of Global Health, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA

Background: Human papillomavirus (HPV) screening has been implemented to monitor both cervical cancer and head and neck cancer. In this prospective cohort study, we determined sociodemographic, behavioral, and infectious etiology for head and neck squamous cell carcinoma (HNSCC) in Taiwan using data collected from an anonymous sexually transmitted infections screening program. Methods: An anonymous sexually transmitted infections screening program was conducted at a medical center during 2016, in which sociodemographic characteristics including gender, age, marital status, education level, and occupation; medical history regarding underlying comorbidities and history of receiving HPV and other vaccines; lifestyle determinants including betel quid chewing, drug using, and sexual behaviors, were inquired. Blood, anal swab, and penile swab samples were collected to determine viral infections using polymerase chain reaction (PCR). With PCR, 37 HPV genotypes were detected. Regular follow-ups were made for patients enrolled in the screening program until end of 2020, during which all suspected malignancies were recorded upon referrals to oncologists. From this prospective cohort, odds ratios (ORs) of HNSCC for sociodemographic, lifestyle, and infectious variables were derived with logistic regression (R version 4.0.1). P < 0.05 was considered statistically significant. Results: A total of 376 patients were enrolled. Most patients were men (n = 372), with a median age of 27 years. There were 124 (32.98%) HPV-positive patients and 78 (20.74%) HIV-positive patients. Among HPV-positive patients, 20 (25.64%) were of high-risk genotypes. During the follow-up, 44 patients developed HNSCC and all received radiotherapy. Multivariate analysis revealed that patients who were single (OR = 1.43, 95% CI = 1.12-1.83, P = 0.01) or widowed (OR = 2.47, 95% CI = 1.88-3.25, P < 0.001) had higher risk of HNSCC than patients who were married. Patients aged 51-60 (OR = 2.93, 95% CI = 2.10-4.09, P < 0.001) and over 60 (OR = 1.89, 95% CI = 1.45-2.47, P < 0.001) presented higher risks of HNSCC, compared with those aged below 20. Patients addicted to betel quid chewing had high HNSCC risk (OR = 1.29, 95% CI = 1.11 – 1.50, P < 0.001). However, patients with HPV infections did not present with higher HNSCC risks (OR = 0.925, 95% CI = 0.852 – 1.003, P = 0.058). Conclu- sions: In this prospective cohort study, the elderly, unmarried patients, and patients addicted to betel quid chewing, presented with high HNSCC incidence. On the contrary, the association between HPV infection and HNSCC was insignificant. As both betel nut-chewing and HPV infection could be prevented, we advocate for comprehensive screening and patient education for HNSCC prevention. Research Sponsor: None.

10554 Poster Session

One-stop-shop for cancer screening: A model for the future.

Ezra Bernstein, Shiran Shapira, Shahar Lev-Ari, Ari Leshno, Udi A. Sommer, Lior Galazan, Humaid O. Al-Shamsi, Ido Wolf, Meital Shaked, Mori Hay Levy, Miri Sror, Menachem Moshkowitz, Ori A. Segal, Eyal Gur, Alfred I. Neugut, Nadir Arber; Integrated Cancer Prevention Center, Tel-Aviv Medical Center, Tel-Aviv, Israel; Health promotion and Integrated Cancer Prevention Center, Tel Aviv Medical Center, Tel Aviv, Israel; Integrated Cancer Prevention Center, Tel Aviv, Israel; Tel Aviv Sorasky Medical Center, Tel Aviv, Israel; Integrated Cancer Prevention Center, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel; Burjeel Cancer Institute, Sharjah, ON, United Arab Emirates; Tel Aviv Medical Center and Sackler School of Medicine, Tel Aviv, Israel; Integrated Cancer Prevention Center, Tel-Aviv Medical Center, Tel Aviv, Israel; Health Promotion and Integrated Cancer Prevention Center, Tel Aviv Medical Center, Tel Aviv, Israel; Tel Aviv University, Tel Aviv, Israel; Tel Aviv Sourasky Medical Center, Tel Aviv, Israel; Co- lumbia University Herbert Irving Comprehensive Caner Center, New York, NY; Health Promotion and In- tegrated Cancer Prevention Center, Tel Aviv, Israel

Background: Cancer is the second leading cause of death globally. Early detection will often greatly reduce mortality for many cancers, increase treatment effectiveness, and improve the quality of life for cancer patients, and, by implementing evidence-based prevention strategies, 30–50% of cancers can be prevented. Screening for different cancer types separately is inefficient. A solution is the Integrated Cancer Prevention Center (ICPC), a program with specialists in each discipline who test for multiple cancers during one visit. Methods: This is a prospective cohort study of 17,104 self-referred, asymptomatic patients who visited the Integrative Cancer Prevention Center (ICPC) between January 1, 2006, and December 31, 2019. Clinical, laboratory, and epidemiological data were recorded by multiple specialists. Patients were given follow-up recommendations and diagnoses when appropriate. The primary measure was the detection and staging of new malignant lesions. Secondary measures included cost- benefit and mortality benefit. Results: We screened 8618 men and 8486 women with an average age of 47.11 ± 11.71 years. Of 259 cancers detected through the ICPC, 49 (18.9%) were stage 0, 115 (44.4%) were stage I, 31 (12%) were stage II, 25 (9.7%) were stage III, and 32 (12.4%) were stage IV. Seventeen cancers were missed, only six of which were within the scope of the ICPC, and 189 cancers developed > one year after the last visit to the ICPC. Compared to the stage of detection for cancers in the US, all cancers except for colon were detected at an earlier stage at the ICPC. Lung was the most significant with 86.7% of cancers detected at stage 0, I, or II at the ICPC compared to only 49.3% caught at those stages in the US. Conclusions: This is a proof of concept for a one-stop-shop approach to asymptomatic cancer screening in a multidisciplinary outpatient clinic. It offers evidence that this screening framework can and should be replicated in other healthcare settings and on a national policy level as it saves lives and money. The encouraging results presented here should further the conversa- tion about the utility of screening and add momentum to the movement for increased screening. Re- search Sponsor: None.

Stages of detection for cancers detected by the ICPC. Stage 0 Stage I Stage II Stage III Stage IV

Bladder 1 0 0 0 0 Brain 0 0 0 0 0 Breast 10 12 7 11 0 Cervical 9 1 0 0 0 Colon 1 7 3 5 4 Hematological 0 0 0 1 12 Hepatobiliary 0 0 0 0 2 Laryngeal 0 1 1 0 0 Lung 0 11 2 1 1 Neuroendocrine 0 1 0 0 1 Oral 0 2 0 0 0 Other 1 0 0 0 2 Ovarian 0 2 0 1 2 Pancreatic 0 1 0 0 3 Prostate 0 23 10 1 2 Renal 0 8 4 0 1 Skin (BCC) 14 2 0 0 0 Skin (SCC) 4 3 0 0 0 Skin (Melanoma) 8 10 0 0 0 Soft Tissue 0 0 1 0 0 Testicular 0 5 0 0 0 Thyroid 0 22 0 5 0 Upper GI 0 0 1 0 1 Uterine 1 2 1 0 0 Total 49 (18.9%) 115 (44.4%) 31 (12%) 25 (9.7%) 32 (12.3%)

10555 Poster Session

Adverse consequences of the COVID-19 pandemic on breast cancer stage distribution and breast cancer disparities.

Genevieve A. Fasano, Yalei Chen, Solange Bayard, Melissa Davis, Vivian Bea, Michele B. Drotman, Elizabeth Kagan Arleo, Jennifer Marti, Manmeet Malik, Rache M. Simmons, Alexander J. Swistel, Rulla Tamimi, Lisa A. Newman; Weill Cornell Medicine/New York Presbyterian, New York, NY; Henry Ford Health System, Detroit, MI; Weill Cornell Medicine, New York, NY; Department of Surgery, Weill Cornell Medicine, NewYork-Presbyterian Hospital, New York, NY; New York Presbyterian Queens, Flushing, NY; New York-Presbyterian Hospital/Weill Cornell Medicine, New York, NY; NewYork-Presbyterian Hospital/ Weill Cornell Medical Center, New York, NY

Background: The COVID-19 surge in March 2020 resulted in a hiatus placed on screening mammography programs in support of shelter-in-place mandates and diversion of medical resources to pandemic management. The COVID-related economic recession and ongoing social distancing policies continued to influence screening practices after the hiatus was lifted. We evaluated the effect of the hiatus on breast cancer stage distribution on the diverse patient population of a health care system in New York City, the first pandemic epicenter in the United States. Methods: Breast cancer patients diagnosed Jan- uary 1, 2019 to December 31, 2020 were analyzed, with comparisons of stage distribution and mammography screen-detection for three intervals: Pre-Hiatus, During Hiatus (March 15, 2020 to June 15, 2020), and Post-Hiatus. Results were stratified by African American (AA), White American (WA), Asian (As) and Hispanic/Latina (Hisp) self-reported racial/ethnic identity. Results: A total of 894 patients were identified; of these, 549 WA, 100 AA, 104 As, and 93 Hisp comprised the final race/ethnicity-stratified study population. Overall, 588 patients were diagnosed Pre-Hiatus, 61 During-Hiatus, and 245 Post- Hiatus. Nearly two-thirds (65.5%) of the Pre-Hiatus cases were screen-detected versus 49.2% During- Hiatus and 54.7% Post-Hiatus (p = 0.002). Frequency of tumors diagnosed < 1 cm declined from 41.9% Pre-Hiatus to 31.7% Post-Hiatus (p = 0.035). WA patients were more likely to have screen-detected disease compared to AA in the Pre-Hiatus period (69.1% vs. 56.1%; p = 0.05) but non-significantly more likely to have screen-detected disease compared to As and Hisp patients (66.2% vs. 56.9%; p = 0.08). In the Post-Hiatus period, the frequency of screen-detected disease was highest among WA patients (63.0%) compared to all other racial/ethnic groups (AA; 48.1%, As-33.3%, and Hisp-40%; p = 0.007). Similar patterns were observed for frequency of tumors diagnosed #1cm Pre- Hiatus (WA-44.3% vs AA-26%, p = 0.02; and vs. As-41.3%, Hisp-48%; p = 0.09), and Post-Hiatus (WA-37.7% vs. AA-18.2%, As-30.8%, Hisp-23.5%; p = 0.25). Conclusions: The 3-month pandemic- related mammography screening hiatus resulted in a more advanced stage distribution for New York City breast cancer patients, and worsened pre-existing race/ethnicity-associated disparities, especially for AA pts. Research Sponsor: None.

10556 Poster Session

Impact of a sustainable breast and cervical cancer screening program in spite of COVID-19 pandemic: The AMPATH experience in Kenya.

Stephen Kiptoo, Naftali Wisindi Busakhala, Peter Itsura, Philip Tonui, Terry Vik, Patrick J. Loehrer, Omenge Elkanah Orang’o; AMPATH Oncology, Eldoret, Kenya; Teaching and Referral Hospital, Eldoret, Rift Valley, Kenya; Eldoret Teaching and Referral Hospital, Eldoret, Kenya; Indiana University, India- napolis, IN; Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, IN; Moi University College of Health Sciences, Eldoret, Kenya

Background: Cancer is the third leading cause of death with about 48,000 new yearly diagnoses in Ken- ya. Breast and cervical cancers are the major leading cancers in females, both of which are curable with access to timely and effective care. To meet population health goals, early abnormalities of the cervix and breast must be treated promptly to maximize the chance for cure. The AMPATH Breast and Cervical Cancer Control Program (ABCCCP) was initiated to improve access to screening and diagnostic services for breast and cervical cancer in Kenya by addressing the barriers of cancer care through a population health approach, working with communities and the Ministry of Health in Kenya with a potential for scaling these efforts to other parts of the region. Methods: We performed an interim analysis 3 years into a 5-year program, to assess the impact of COVID-19 on our screening program. Statistical descriptive summaries were used to show the trend of screening using visual inspection with acetic acid and breast clinical examination. The screening was conducted facility-based along with community screening upon requests across nine counties. Also, we conducted capacity building through mentoring of health care providers and initiating a telemedicine program to improve patient care and management plans. Results: From 2018-2021, we conducted training, connected 12 centers with telemedicine capacity and screened a total of 100,973 persons were for breast and cervical cancer. The yearly trends demonstrate that the facility routine screenings were maintained: 23,421 (2018); 27,997 (2019); and 28,045 (2020). The total women seen through organized mass screenings however declined (10,304 (2018); 10,107 (2019); and 1,099 (2020), respectively) as this type of screening was stopped after the onset of COVID-19 pandemic. Of all women screened, 3,019 (2.98%) had clinical abnormalities re- quiring follow-up per standard of care including 1,781(1.8%) who were eventually histologically confirmed to have cancer. During our first and second year of the program, 83 physicians were trained on cancer management and treatment, 341 nurses were trained on breast and cervical cancer screening procedures, and 247 community health workers (CHW) were trained on the importance of screening to enlighten the community on awareness. However, this training was suspended in our year three due to COVID-19. Conclusions: An integrated training program utilizing CHW, nurses and physicians are an effective means for breast and cervical cancer screening in LMC, such as Kenya. This capacity building allows flexibility and sustainability even in the midst of the global COVID-19 pandemic. We also demonstrated successful integration with the county government for program sustainability. The use of telemedicine has greatly enhanced our screening and patient care across several facilities in western Kenya. Research Sponsor: Lily Foundation.

10557 Poster Session

Cancer screening utilization in patients diagnosed with cancer types with and without recommended screening modalities.

Ariella Cohain, Christine Hathaway, Mudit Gupta, Braxton Lagerman, Yali Li, Jackie Blank, Sam Asgarian, John Keilty; Thrive, An Exact Sciences Company, Cambridge, MA; Geisinger Health System, Danville, PA; Geisinger, North Besthesda, MD; Third Rock Ventures, Boston, MA

Background: Several studies have shown screening methods can detect cancer at earlier stages and improve cancer prognosis; however, only four cancer types (breast, colorectal, cervical, and lung) currently have screening methods recommended by the United States Preventive Services Taskforce (USPSTF). In 2021, these four cancers are expected to make up roughly 40% of new cases and cancer deaths, meaning that the majority of cancer deaths will be associated with cancer types lacking recommended screening. We sought to characterize patients who were diagnosed with cancer types with and without recommended screening modalities to demonstrate the gaps in screening faced by the majority of cancers today. Methods: The Geisinger Health System (GHS) Phenomics Initiative Database (PIDB) provides deidentified data from electronic health, billing, and imaging records, and a tumor registry. PIDB was used to identify patients aged 50 to 76 who had cancers diagnosed between 2008 and 2020 and a record of USPSTF-recommended cancer screenings within GHS prior to diagnosis. Analysis focused on patients who received care at GHS during their screening-eligible intervals. Results: Between 2008 and 2020, 13,347 incident invasive cancers were identified in the GHS tumor registry. Of these, 40% (N = 5,331) were cancer types with a recommended screening modality. 57% of these cases (N = 3,039; 23% of all incident cancers) occurred in patients who underwent screening in the interval preceding diagnosis. Screening adherence was significantly associated with stage at diagnosis; patients who were not screened for their diagnosed cancer were more than twice as likely to have a late-stage diagnosis as compared with patients who received screening (multivariate ordinal logistic regression, OR = 2.16, p < 0.001). Patterns of screening adherence in this population are complex; however, 57% of these patients had received screening for a different cancer type. The majority of incident cancers were of those types with no recommended screening modality (N = 8,016; 60% of all incident cancers). Of these, most (N = 6,252; 78%) had been screened for at least one of breast, lung, colon, or cervical cancer and nearly half (N = 3,607; 45%) were current for all guideline-recommended screenings. Not surprisingly, stage at diagnosis was not associated with adherence to any or all screening modalities (multivariate ordinal logistic regression, p = 0.11 and p = 0.45). Conclusions: The majority (79%) of individuals diagnosed with cancer had a history of adherence to at least one screening recommendation. Out of all cancer patients, only 23% were screened specifically for the cancer with which they were subsequently diagnosed, a group that is associated with a lower odds of a late-stage diagnosis. This suggests that the majority of cancer patients who underwent any cancer screening did not benefit from earlier stage diagnosis. Research Sponsor: None.

10558 Poster Session

TNF-alpha immunosuppressive use and future malignancy risk.

Conor Driscoll, Kaitlyn O’Shea, Hui Zhang, Jordan Rich, Christopher Yang, Dylan Isaacson, Eric Li, Mohammad Siddiqui, Roopal Kundu, Edward M. Schaeffer, Shilajit Kundu, TNF Alpha ASCO 2021; Northwestern Medicine, Chicago, IL; Northwestern University Feinberg School of Medicine, Chicago, IL; Northwestern University, Chicago, IL

Background: Tumor Necrosis Factor Alpha (TNF-a) inhibitors suppress the immune system in patients with systemic inflammatory conditions. Long term data assessing future cancer risk for these patients is not known. We assessed long term risk of malignancy in patients with Rheumatoid Arthritis [RA], In- flammatory Bowel Disease [IBD], Psoriasis [PS], and Ankylosing Spondylitis [AS], who were or were not exposed to a TNF-a inhibitor. Methods: This was a retrospective, cohort study conducted using electronic medical record data for patients with complete demographic and treatment data at Northwestern Medicine from years 1998 to 2020 (RA: n = 10763; IBD: n = 12106; PS: n = 1920; AS: n = 5103). In- verse Probability of Treatment Weighting (IPTW) was used to balance the distributions of age, race, gender, smoking status, and follow-up time across exposure groups within each inflammatory condition type. Relative risk (RR) of malignancy based on TNF-a exposure was assessed using logistic regression. Results: 2583 (24.0%) of RA, 2185 (18.0%) of IBD, 1811 (94.3%) of PS, 572 (11.2%) of AS patients had TNF-a exposure. Median follow-up for patients was 43 months. The RR for any cancer was higher for patients exposed to a TNF-a agent with rheumatoid arthritis (RR 1.121 (95% CI 1.02-1.23, p = 0.015) and psoriasis (RR 1.763 (95% CI 1.32-2.37, p < 0.001). The relative risk of any cancer was lower in patients exposed to a TNF-a agent with IBD (RR 0.858 (95% CI 0.78-0.94, p = 0.001). No significant difference in relative risk associated with TNF-a exposure was detected with ankylosing spondylitis (RR 0.929 (95% CI 0.8-1.08, p = 0.344). Conclusions: Patients with RA or PS and TNF-a exposure had higher RR of overall malignancy. Patients with IBD and TNF-a exposure had lower risk of overall malignancy. TNF-a immunosuppression may alter cancer risk differently based on the disease states for which it is being used. This information is critical when counseling patients on long term risk and screening strategies when considering TNF-a inhibition. Research Sponsor: None.

10559 Poster Session

Understanding factors associated with uptake of lung cancer screening among individuals at higher risk.

Abdi Gudina, Anapaula Cupertino, Charles Stewart Kamen, David Adler, Elizabeth Belcher, Nikesha Gilmore, Amber Kleckner, Evelyn Arana, Lee Kehoe, Sara Hardy, Michelle Christine Janelsins, Gary R. Morrow; University of Rochester Medical Center, Rochester, NY; University of Rochester, Rochester, NY

Background: Lung cancer is the leading cause of cancer death in the U.S, accounting for about 25% of all cancer mortality. The U.S Preventive Services Task Force has recommended annual screening for lung cancer using low-dose computed tomography (LDCT) scanning for individuals at higher risk (aged 55-80 years with a >30 pack-year smoking history). Early detection using LDCT scanning reduces lung cancer specific mortality by 20%. Despite its efficacy, the uptake of annual lung cancer screening among high-risk individuals remains low ( < 18%). The purpose of this study was to identify factors associated with the uptake of lung cancer screening in high-risk individuals in the U.S population. Meth- ods: Data for this study were obtained from the 2017-2019 Behavioral Risk Factor Surveillance System (BRFSS), a population-based survey conducted annually by the Centers for Disease Control and Preven- tion (CDC) in collaboration with health departments in all 50 states, Washington, DC, and the U.S territories. We restricted our sample to high-risk individuals aged 55-80 years with a >30 pack-year smoking history. Only subjects with complete data on all predictor variables (age, gender, race/ethnicity, marital status, education, income, insurance, COPD, current smoking status, primary care provider) and the outcome variable (uptake of lung cancer screening) (n = 11, 714) were included in the final analysis. Chi-square tests were used to compare the uptake of lung cancer screening by demographic and socioeconomic factors. Multivariable logistic regression models were used to model the association between the predictors and the outcome variable. Results: Individuals with no health insurance (OR: 0.64; 95%CI: 0.46-0.90), no primary healthcare provider (OR: 0.40; 95%CI: 0.31-0.52), no chronic obstructive pulmonary disease (COPD) (OR: 0.35; 95%CI: 0.31-0.0.40) and who were females (OR: 0.86; 95%CI: 0.76-0.96) were less likely to participate in annual lung cancer screening. Individuals aged 65–69 years (OR: 1.65; 95%CI: 1.38-1.97), 70–74 years (OR: 1.77; 95%CI: 1.46-2.14) or 75–80 years (OR: 1.42; 95%CI: 1.16-1.76) were more likely to receive annual lung cancer screening compared with those aged 55-59 years. Race/ethnicity, level of education, level of income, marital status, and current smoking status had no significant association with the uptake of annual lung cancer screening. Conclusions: Our study identifies factors associated with lower uptake of annual lung cancer screening (no health insurance coverage, no primary healthcare provider, no COPD, and female gender). The findings from this study have important implications for the design of more effective interventions to target specific subgroups for the uptake of annual lung cancer screening. Research Sponsor: Univer- sity of Rochester Medical Center.

10560 Poster Session

Young patients with cancer related to modifiable behaviors in the United States: What can we learn from them?

John Chan, Michelle Ann P. Caesar, Chloe Chan, Michael Richardson, Daniel Stuart Kapp, Alex Andrea Francoeur, Cheng-I Liao; Palo Alto Medical Foundation, Palo Alto, CA; California Pacific Medical Center Research Institute, San Francisco, CA; Palo Alto Medical Foundation, San Francisco, CA; University of California, Los Angeles, Los Angeles, CA; Stanford University, School of Medicine, Stanford, CA; Kaoh- siung Veterans General Hospital, Kaohsiung, Taiwan

Background: We proposed to examine trends in modifiable behaviorally related cancers among younger men and women in the United States. Methods: Alcohol-associated cancers, HPV-associated, obesity- associated, physical inactivity-associated, and tobacco-associated were defined using ICD-O-3 site codes. From 2001 and 2017, registry data were obtained from the United States Cancer Statistics database. SEER*Stat 8.3.8 and Joinpoint regression program 4.8.0.1 were used to calculate the trends of associated cancers expressed per 100,000. Results: Of the young women (ages 20-49 years) the incidence of cancers in 2017 associated with alcohol, smoking and obesity were 89/100,000, 43/ 100,000 and 64/100,000 respectively. Based on analysis of trends of women from 2001 to 2017, obesity, physical inactivity and alcohol related cancers increased with an annual percent change (APC 2.31%, 1.67%, 0.46%, p < 0.001). Using a projection model, in Hispanic women, obesity related cancers will become the highest incidence cancers by 2035, surpassing alcohol and physical inactivity. Of the young men (ages 20-49 years) the incidence of cancers in 2017 associated with alcohol, obesity, and tobacco were 23/100,000, 36/100,000 and 44/100,000 respectively. On trends analysis, obesity, physical inactivity, and alcohol related cancers have increased (APC 2.0%, 1.65%, 0.17%, p < 0.001, p < 0.001, p = 0.044), whereas tobacco-related cancers are decreasing with an APC of -0.44% (p < 0.001). When examining different regions, the highest APC for obesity and physical inactivity related cancers was 2.43% in the Midwest (p < 0.001). Using a prediction model, obesity is predicted to sur- pass alcohol and physical inactivity related cancers for men 20-49 years old by 2035. Conclusions: In women, most modifiable factors associated with cancer are increasing except in HPV related cancers. In men, these rates of cancer are increasing except in tobacco related cancers. However, rates of obesity related cancers are on the rise in Hispanic women and younger men in southern U.S. regions. Obesity is projected to become the major modifiable factor for many associated cancers. Research Sponsor: Fisher Foundation and Denise Cobb Hale.

10561 Poster Session

Validation of a high performing blood test for multiple major cancer screenings.

Linhao Xu, Jun Wang, Weifeng Ma, Xin Liu, Sihui Li, Siyu Chen, Qian Hu, Ting Yang, Zhujia Ye, Zhiwei Chen, Jianbing Fan; AnchorDx Medical Co., Ltd, Guangzhou, China; Anchordx, Guangzhou, China; An- chor DX, Fremont, CA; AnchorDX LTD, Guangzhou, China; AnchorDx, Inc., Guangzhou, China

Background: Early detection at the localized stage is pivotal for the successful treatment of various cancer types. Although several cancers already have routine screening approaches, the comprehensive util- ities are impeded for various reasons, e.g., low accuracy, high cost, limited availability of required facilities, especially in the developing countries. Therefore, an accurate, cost-effective, and non-invasive test for multiple major cancer screening is in high demand. We previously reported a cfDNA methylation test, which can detect five major cancer types with high specificity and sensitivity, especially at the early stage (stage I). These five major cancers, including lung cancer (LC), breast cancer (BC), colorectal cancer (CRC), gastric cancer (GC), and esophageal cancer (EC), account for 56% of new cancer cases and 60% of cancer-related deaths yearly in China. Here, we report the result in an independent cohort as a further validation of this multi-cancer screening test. Methods: The high-throughput targeted methylation profiling platform, Aurora, was used to analyze the plasma samples from an independent retrospective cohort containing 505 healthy controls and ~200 cases for each cancer type. A locked model based on our previous pilot study (reported in AACR 2020 and 2021) was applied to this data set to assess the overall performance. Results: The Area Under Curves (AUC) of the classifier for LC, BC, CRC, GC and EC are 97.3%, 96.2%, 92.0%, 94.0% and 93.5%, respectively. At a fixed specificity of 99%, the sensitivities for LC, BC, CRC, GC and EC are 84%, 75%, 82%, 85% and 78%, respectively. Conclusions: A methylation blood test for five major cancer screening has been validated in a large retrospective cohort. Its high sensitivity for each cancer type, especially at the early stage (stage I), and easy to use suggests it can be implemented in real clinical world. A large prospective clinical trial is undergoing to further validate this test in asymptomatic populations. Research Sponsor: AnchorDx Medical Co., Ltd.

10562 Poster Session

Factors associated with breast self-examination in Mexican young women with breast cancer.

Andrea Becerril Gaitan, Bryan Vaca-Cartagena, Ana Sofia Ferrigno, Fernanda Mesa-Chavez, Alejandra Platas, Melina Miaja, Alan Fonseca, Marlid Cruz-Ramos, Dione Aguilar, Lucero Labra, Enrique Bargallo Rocha, Alejandro Mohar Betancourt, Cynthia Villarreal-Garza; Centro de Cancer de Mama, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, NL, Mexico; Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Gar- cia, Nuevo Leon, NL, Mexico; Joven & Fuerte: Programa para la Atencion e Investigacion de Mujeres Jo- venes con Cancer de Mama, Mexico City, EM, Mexico; Depto. Tumores Mamarios y Depto. de Investigacion, Instituto Nacional de Cancerologia, Ciudad De Mexico, EM, Mexico

Background: Breast cancer (BC) is the most common cause of cancer-related death and morbidity among young women in Latin America. This group has a higher prevalence of advanced disease stages at diagnosis compared to their older counterparts. Thus, strategies aimed at detecting BC at early stages are imperative. Breast self-examination (BSE) remains a useful strategy for BC detection, especially in women who do not routinely undergo screening with imaging studies. This study aims to evaluate factors related with BSE practice in Mexican young women with BC and assess its association with earlier disease stages. Methods: Women aged #40 newly diagnosed with BC from 2014 to 2020 at three cancer referral centers in Mexico accrued in the Joven & Fuerte cohort were included and asked to complete a socio-demographic survey. Fisher’s exact and Mann-Whitney U tests were used to evaluate associations between BSE and socio-demographic characteristics, as well as disease stages. Results: A total of 554 patients with a median age at diagnosis of 36 years (range: 19-40) were analyzed. Most patients (65%) were married or in a domestic partnership, and the majority were housewives (63%). Regarding educational background, 64% had completed at least high school, and up to 84% had a monthly income < 11,600 Mexican pesos (US$ 581). Overall, 85% of patients had public insurance, 6% had private insurance, and 9% were uninsured. The distribution of clinical stages at diagnosis was: 0 (2%), I (11%), II (45%), III (32%), and IV (10%). BC detection methods were: 85% by self/partner exam, 11% by an imaging study, and 4% by a healthcare professional. A total of 443 (80%) patients practiced BSE, of which 50% did it on a monthly basis, 18% every 2-3 months, and the remaining 12% every 4-12 months. Notably, a higher educational level ($ high school v # middle school) was positively associated with BSE practice (RR: 1.28; 95%CI 1.06-1.54; p= 0.005). No significant association was found between BSE and age (#35 v > 35), marital status (in a relationship v no), occupation (housewive v other), monthly income ( < 11,600 v $11,600) or medical coverage (public/uninsured v private). Patients that performed BSE were more likely to be diagnosed with early BC (stages 0-II) compared to those that did not (61% v 45%; p= 0.003). No association was found between BSE frequency and stage at diagnosis. Noteworthy, patients with private insurance were more likely to be diagnosed with stages 0-II compared to those with public or no insurance (80% v 56%; p= 0.007). Conclusions: The significant association between BSE and earlier stages at diagnosis found in this study highlights the need to raise awareness and promote this practice among young women with the objective of downstaging BC diagnoses. Public health interventions such as educational and social media campaigns that aim to improve the correct practice of BSE might be particularly useful in settings with inadequate screening programs. Research Sponsor: None.

10563 Poster Session

Young-onset colorectal cancer: A call for action.

Iosune Baraibar, Francesc Salva, Raquel Comas, Javier Ros, Ariadna Garcia, Mireia Sanchis, Jose Luis Cuadra, Nadia Saoudi, Augusto Valdivia, Nuria Mulet, Ana Virgili, Jorge Hernando, Jaume Capdevila, Marc Mart�ı, Stefania Landolfi, Helena Verdaguer, Eloy Espin, Paolo Nuciforo, Josep Tabernero, Elena Elez; Vall d’Hebron Institute of Oncology (VHIO), Medical Oncology, Vall d’Hebron University Hospital (HUVH), Barcelona, Spain; Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain; IOB-Quiro�n, Barcelona, Spain; Medical Oncology, Vall d’Hebron University Hospital (HUVH), Barcelona, Spain; In- stituto Catalan de Oncologia de Hospitalet, Barcelona, Spain; Hospital Sant Pau, Barcelona, Spain; De- partment of General and Digestive Surgery, Vall d’Hebron University Hospital, Barcelona, Spain; Pathology Department, Vall d’Hebron University Hospital, Barcelona, Spain; Department of General and Digestive Surgery, Vall d’Hebron University Hospital (HUVH), Barcelona, Spain; Molecular Oncolo- gy Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain; Vall d’Hebron University Hospi- tal and Vall d’Hebron Institute of Oncology (VHIO), UVic-UCC, Barcelona, Spain

Background: Young-onset colorectal cancer (YOCR) is defined as diagnosis below the age of 50. Over the past decades, the incidence of YOCRC has increased at an alarming rate, but causes and pathogenesis still remain unknown. Early detection of colorectal cancer (CRC) has demonstrated to improve survival. Despite these facts, adults < 50 years old are not yet included in screening programs and YOCRC is not well characterized. We aimed to characterize the clinical and molecular characteristics of YOCRC in patients (pts) diagnosed at our institution. Methods: Consecutive pts with a diagnosis of CRC below the age of 50 visited for the first time at Vall d’Hebron University Hospital in Spain between January 2017 and October 2020 were included in the analysis. Data of clinicopathologic features and treatment were collected retrospectively from medical records. Results: 205 pts met the inclusion criteria, 111 (54%) were females, 8 (4%) presented a personal history of cancer at diagnosis and 109 (53%) a family history of cancer. Age at diagnosis was: < 30: 10 (5%), {30 – 40): 52 (25%), {40-45): 51 (25%), {45-50): 92 (45%). Site of primary tumor was: right colon: 50 (24%), left colon: 107 (52%): rectum: 48 (24%). Stage at diagnosis was I: 3 (1%), II: 14 (7%), III: 60 (29%), IV: 128 (63%). 6 of 14 (43%) and 44 of 60 patients (73%) with stage II and III CRC presented disease progression after initial treatment, respectively. Molecular status was: KRAS mutation: 74 (36%), NRAS mutation: 7 (3%), BRAF mutation: 12 (6%), MSI-H: 12 (6%). 43 pts (21%) had documentation of genetic counseling. Median (range) number of lines of treatment for metastatic disease was 3 (1-7), 53 pts (30%) received at least 4 lines of treatment. Median (range) number of metastatic sites was: 2 (1-6). 114 patients (55.6%) had died at the cut-off timepoint. Conclusions: YOCR is usually diagnosed with a more advanced stage than standard-onset CRC, with a poorer course of the disease. Further studies in young adults with CRC should address this phenomenon to understand the underlying causes, and prioritize genetic counseling. Our results support the unmet need of initiating screening programs in adults younger than 50 years, the ur- gency for a global consensus and a call for action. Research Sponsor: None.

10564 Poster Session

Interactive online skin cancer training game “Whack-a-Mole” assesses training strategies and real-time feedback on melanoma identification among U.S. adults.

Margaret Sanchez, Nicholas Carcioppolo, Soyoon Kim, Di Lun, Kate Malova, Bingjing Mao, Ashley Renee Reynolds, Clay Ewing, Shasa Hu; Sylvester Comprehensice Cancer Center-University of Miami Miller School of Medicine, Miami, FL; University of Miami, Miami, FL

Background: Cutaneous melanoma is a deadly form of skin cancer. Several studies have shown that early melanoma detection is associated with decreased mortality through self-examination and dermatology full-body skin exams. ABCD rule and the ugly duckling sign (UDS) are used to identify melanomas, but little research has explored the comparative efficacy of these approaches. This investigation compares the effectiveness of different mole identification training strategies and explores the effect of real-time feedback on decision-making. Methods: We developed an online melanoma identification game that tests differences between training types and expert feedback on mole identification. This online RCT tests a 4 (training: ABCD, UDS, both, control) X 3 (feedback: standard, motivational, control) factorial design on melanoma identification, skin cancer beliefs (perceived susceptibility and self-efficacy), and skin cancer prevention intentions. Standard feedback included expert evaluations of moles, whereas motivational feedback added statements grounded in fear appeals theory to encourage skin self-examination. An online research panel service was used to recruit 1025 US adults. Participants were randomly assigned to condition, completed a pretest, participated in the game intervention, and completed a posttest. Gameplay incorporates the same mechanics as Tinder (swipe left on benign moles and swipe right on malignant moles). Results: In total, participants reviewed 48 moles, 12 of which were melanomas. We used two-way ANCOVA for the analysis. ABCD training resulted in significantly higher melanoma identification than the control (p =.011). Every training type resulted in significantly higher self- efficacy than the control (p =.007). Additionally, there was a significant main effect of feedback on self- efficacy (p =.001), where both standard and motivational feedback elicited significantly higher levels of self-efficacy than the control condition. Around 88% of participants intend to conduct skin self-exams and wear sunscreen. Conclusions: Our data suggests that “Whack a Mole” is an efficacious tool for melanoma training. ABCD and UDS training with interactive feedback are important to improve accuracy and ability for melanoma identification. Research Sponsor: None.

10565 Poster Session

The impact of socioeconomic status on stage at presentation, receipt of diagnostic imaging, receipt of treatment, and overall survival in colorectal cancer patients.

Rajan Shah, Kelvin K. Chan; University of Toronto, Faculty of Medicine, Mississauga, ON, Canada; Sun- nybrook Odette Cancer Centre, University of Toronto, Toronto, ON, Canada

Background: Socioeconomic factors have been identified to influence patterns of care in colorectal cancer yet current literature findings are sparse, conflicting, and often incomplete. As such, this study in- vestigates the impact of socioeconomic status (SES) on stage at presentation, receipt of diagnostic imaging, receipt of treatment, and overall survival (OS) in a universal healthcare system. Methods: The Ontario Cancer Registry was accessed to identify a cohort of patients diagnosed with colorectal adenocarcinoma from 2007-2016 in Ontario, Canada. Linkage to administrative datasets allowed study of the impact of SES, measured by mean neighbourhood household income divided into quintiles (Q1-Q5; Q1 = lowest income), on stage, imaging, treatments, and OS. Multivariable regression analyses of all endpoints were adjusted for age, sex, comorbidity, and rurality with OS models also adjusting for imaging and treatment. Results: 39,802 colon and 13,164 rectal patients were identified. Lower SES patients were more likely to present at a higher stage in both cohorts. Lower SES colon patients were less likely to receive magnetic resonance imaging (MRI) of the abdomen, liver resection, adjuvant oxaliplatin, and all palliative systemic therapies studied. In rectal patients, lower SES was associated with decreased receipt of MRI pelvis, rectal cancer resection in early stages, adjuvant oxaliplatin, and most palliative chemotherapies studied. All OS models found that lower SES was associated with poorer OS. Conclusions: These findings suggest disparities across the continuum of cancer care persist even within a universal healthcare system. Further efforts should be directed towards temporal research, identifying barriers, and subsequently applying this information to actionable policies. Research Sponsor: None.

10566 Poster Session

Impact of COVID-19 pandemic in 2020 on the diagnosis and management of breast cancer in Korea: A multi-institutional study.

Young-Joon Kang, Se Jeong Oh, Jong Min Baek, Yong-seok Kim, Ye Won Jeon, Tae-Kyung Yoo, Jiyoung Rhu, Chang-Hyun Shin, Shijin Cho, Hoon Choi; Incheon St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, South Korea; Yeouido St. Mary’s Hospital, The Catholic University of Korea, Seoul, South Korea; Uijeongbu St. Mary’s Hospital, Uijeongbu-si, South Korea; St. Vincent’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, South Korea; Department of Surgery, Seoul St. Mary’s Hospital, The Catholic University of Korea, Seoul, South Korea; Bucheon St. Mary’s Hospital, The Catholic University of Korea, College of Medicine, Seoul, South Korea

Background: Since the COVID-19 pandemic began in early 2020, there have been many reports that it has had a significant impact on screening, case identification and referral in cancer diagnosis. We investigated the diagnostic and therapeutic status of breast malignancy before and after the COVID-19 pandemic at the multi-institution level. Methods: We have reviewed the records of patients with breast cancer from February 2019 to July 2020 in six university hospitals in Korea. The patients were divided into two groups according to the initial date of cancer diagnosis: Period A, from February to April and Period B, from May to July in 2020. The two groups were compared for the same periods in 2019. The goals were to determine whether breast cancer screening and diagnosis have been delayed and thus resulted in stage migration. We also examined the difference in the number of surgeries in patients diagnosed with breast cancer during those periods. Results: The total of 1,669 breast malignancy diagnosis was made in the grouped periods of 2019, and 1,369 diagnoses in 2020. All patients were screened by PCR test for COVID-19 prior to hospitalization, and none of them tested positive. Overall, there was a 9.9% reduction in the number of diagnoses than in 2019 and the decrease was more significant in Peri- od A (11.1% vs. 8.7%). According to the age, there was no difference until the 30s but decreased from those in their 40s and above. The decline was more pronounced in the elderly. The COVID-19 pandemic has affected breast cancer screening (decreased by 27.4%) and more diminished in Period A (41.0% vs. 19.0%). Invasive breast cancer stage was not statistically different in Period A compare with 2019 (p = 0.170). But the stage in Period B was different (p = 0.032), and more patients were observed in advanced stages in 2020. The decrease in surgery was noticeably observed in Period A (4.6%, from 480 to 438 surgeries) and not in Period B. The analysis of reconstruction surgery was similar. Conclu- sions: Patients with COVID-19 increased exponentially from late February in Korea. However, the number of patients per day decreased to less than 100 on March 15 and then flattened. The health care system for cancer was not overloaded and restrictions on visiting hospital were minimal. Analysis in the pandemic period of the 6-month showed that the number of breast cancer screening, diagnosis and surgeries decreased compared with the previous year. Those decreases were prominent in Period A when the COVID-19 patient surged. The upstage migration of breast cancer was generally insignificant but slightly occurred in Period B. The outbreak of infectious disease makes patients reluctant to come to the hospital, especially in the elderly. We need to discuss the potential long-lasting deleterious effect of the COVID-19 pandemic on cancer diagnosis and management. And we should prepare for how to deal with the backlog caused by the COVID-19 pandemic. Research Sponsor: None.

10567 Poster Session

Outcomes for cancer patients on systemic anti-cancer therapies during the COVID-19 pandemic from the CAPITOL (COVID-19 Cancer PatIenT Outcomes in North London) cohort study.

Valerie Crolley, Daire Hanna, Nalinie Joharatnam-Hogan, Neha Chopra, Ekin Bamac, Meera Desai, Yuk- Chun Lam, Sabiq Dipro, Ruhi Kanani, Jack Benson, William Wilson, Thomas Andrew Fox, Kai-Keen Shiu, Martin Forster, John A. Bridgewater, Daniel Hochhauser, Khurum Hayat Khan; Barts Cancer Insti- tute, London, United Kingdom; Royal Free London NHS Foundation Trust, London, United Kingdom; University College Hospital, London, United Kingdom; University College London Hospital, Pinner, United Kingdom; University College Hospital, Londond, United Kingdom; North Middlesex University Hospital, London, United Kingdom; Cancer Research UK and UCL Cancer Trials Centre, University Col- lege London, London, United Kingdom; University College Hospital, NHS Foundation Trust, London, United Kingdom; University College London Hospitals NHS Foundation, London, United Kingdom; Uni- versity College London Cancer Institute, London, United Kingdom; The Royal Marsden NHS Foundation Trust, Sutton Surrey, United Kingdom

Background: One of the major challenges with COVID-19 has been the changes to cancer services, including changes to the type of systemic anti-cancer treatment being delivered to patients. There needs to be a better understanding of which cancer patients are at the greatest amount of risk to make informed decisions on how cancer treatment can be altered to protect patients from COVID-19 infection. The CAPITOL (COVID-19 CAncer PatIenT Outcomes in North London) study investigated the outcomes of patients receiving systemic anti-cancer therapies (SACT) with regards to COVID-19 infection, as patients with cancer are hypothesised to be at higher risk. Methods: CAPITOL collected data from all patients receiving SACT at two cancer centres. The effect of clinical characteristics on the incidence and severity of COVID-19 infection in patients on SACT was the primary outcome, and we used univariable and multivariable models in our analysis, adjusting for age, gender and comorbidities. Results: 2871 patients were analysed from 2nd March to 31st May 2020, all of whom received SACT; during this time period 68 (2.4%) were diagnosed with COVID-19. Receiving SACT increased the risk of death when contracting COVID-19 (adjusted (adj.) OR 9.84; 95% CI 5.73 – 16.9). The risk of contracting COVID- 19 was increased by receiving chemotherapy (adj. OR 2.99; 95% CI = 1.72 - 5.21), with the risk significantly increased by high dose chemotherapy (adj. OR 2.36, 95% CI 1.35 – 6.48). Patients with comorbidities (adjusted OR 2.29; 95% CI 1.19 - 4.38), or with a respiratory or intrathoracic neoplasm (adj. OR 2.12; 95% CI 1.04 - 4.36) were also at increased risk of contracting COVID-19. Cancer patients who received targeted treatment had a reduced risk of contracting COVID-19 (adj. OR 0.53; 95% CI 0.30 – 0.95), while there was no significant change in risk caused by treatment intent (curative versus palliative), hormonal- or immunotherapy and solid versus haematological cancers. Conclusions: To the best of our knowledge, this is one of the first investigations into the risk of contracting COVID-19 in a cohort of all cancer patients on SACT. We found that patients on SACT are more likely to die if they con- tract COVID-19. The type of SACT received by cancer patients can affect their likelihood of contacting COVID-19, with chemotherapy increasing risk, targeted therapy decreasing risk and a potential protective effect for hormonal and immunotherapy. Research Sponsor: None.

10568 Poster Session

Prior tonsillectomy and subsequent risk of breast cancer in females: Systematic review and meta-analysis.

Salah Eddine Oussama Kacimi, Mounir Ould Setti, Amira Yasmine Benmelouka, Mohammad Aloulou, Hanane Zoubida Sekkal, Laila Salah Shamseldin, Jovana Vidovic, Selma Nihel Klouche-Djedid, Jaffer Shah, Farah Yasmin, Omar Riffi, Ahmed M Afifi, Sherief Ghozy; Faculty of Medicine, University of Tlemcen, Tlemcen, Algeria; Institute of Public Health and Clinical Nutrition, University of Eastern Fin- land, Kuopio, Finland; Faculty of Medicine, University of Algiers, Algiers, Algeria; Faculty of Medicine, University of Aleppo, Aleppo, Syrian Arab Republic; Faculty of Medicine, Tanta University, Tanta, Egypt; Semmelweis University, Budapest, Hungary; Drexel University College of Medicine, Philadel- phia, PA; Department of Medicine, Dow University of Health Sciences, Karachi, Pakistan; University of Kentucky College of Medicine, Lexington, KY; Department of Anesthesiology, University of Florida Col- lege of Medicine, Gainesville, FL

Background: Exposure to recurrent infections in childhood was linked to an increased risk of cancer in adulthood. There is also evidence that a history of tonsillectomy, a procedure often performed in children with recurrent infections, is linked to an increased risk of leukemia, and Hodgkin lymphoma. Ton- sillectomy could be directly associated with cancer risk or it could be a proxy for another risk factor such as recurrent infections and chronic inflammation. Nevertheless, the role of recurrent childhood infections and tonsillectomy on the one hand, and the risk of breast cancer (BC) in adulthood remain understudied. Our study aims to verify whether a history of tonsillectomy increases the risk of BC in women. Methods: A systematic review was conducted using PubMed, Google Scholar, Scopus, Embase and Web of Science databases from inception through November 2020 to identify the studies which explored the association between history of tonsillectomy and BC in females. The Newcastle Ottawa Scale was used to assess the quality of included studies. Odds ratio (OR) was used to measure effect size. The Random/Fixed effects model was applied to synthesize the associations between tonsillectomy and BC risk based on heterogeneity. Heterogeneity was assessed using the I-squared statistic. A forest plot was generated, and publication bias was assessed. The leave-one-out sensitivity analysis was performed to check if results were driven by a single study. Results: Seven studies with a total of 7259 patients were included in our analysis; out of them, 2200 patients were diagnosed with BC. Patients with a history of tonsillectomy (n = 2843) showed higher subsequent risk of developing BC (OR = 1.252; 95% CI = 1.115 - 1.406; P < 0.001; I2 = 9%) as compared to patients without a history of tonsillectomy (n = 4416). Using the leave-one-out sensitivity analysis to iteratively remove one study at a time, we confirmed that no single study had a substantial influence on the overall effect size. Conclusions: Our study supports and confirms the evidence that a history of tonsillectomy is associated with an increased risk of breast cancer. These findings are also an argument in support of the hypothesis that recurrent childhood infections are linked with adulthood breast cancer. Research Sponsor: None.

10569 Poster Session

Trends in incidence and mortality of squamous cell carcinoma of the skin: An observational analysis of the Global Burden of Disease database from 1990 to 2017.

Dorothy Yang, Chinmay Jani, Conor Crowley, Richard Goodall, Joseph Shalhoub, Justin D Salciccioli, Dominic C Marshall; St George’s University Hospitals NHS Foundation Trust, London, United Kingdom; Department of Medicine, Mount Auburn Hospital, Cambridge, MA; Division of Pulmonary and Critical Care, Lahey Hospital, Burlington, MA; Department of Surgery and Cancer, Imperial College London, London, United Kingdom; Division of Pulmonary and Critical Care, Brigham and Women’s Hospital, Boston, MA; National Heart and Lung Institute, Imperial College London, London, United Kingdom

Background: Epidemiological data relating to non-melanoma skin cancer (NMSC), including squamous cell carcinoma (SCC), is highly under-reported and under-studied due to its lower metastatic potential. In recent years, incidence and prevalence of SCC has increased in many countries due to earlier detection, increased ultraviolet light exposure, as well as increasing life expectancy. This investigation compared trends in SCC incidence, mortality and disability-adjusted life years (DALYs) in 33 countries. Methods: We utilized the Global Burden of Disease (GBD) database for 33 countries, including the Euro- pean Union nations as well as other selected high-income countries including the UK and USA. We extracted data including age-standardized incidence rates (ASIRs), age-standardized mortality rates (ASMRs) and DALYs for SCC of the skin from 1990 to 2017. Joinpoint regression analysis was used to describe the trends. Results: For both sexes, the highest ASIRs were seen in the USA and Australia: ASIRs were 362.8/100,000 and 283.7/100,000 respectively for males, and 171.2/100,000 and 152.4/100,000 respectively for females. Males had higher ASIRs than females at the end of the observation period in all countries. In contrast, the highest ASMRs for males were observed in Australia (2.77/100,000) and Latvia (2.44/100,000), while the highest ASMRs for females were observed in Ro- mania (0.95/100,000) and Croatia (0.90/100,000). The highest DALYs for both sexes were seen in Australia and Romania: DALYs were 58.4/100,000 and 43.8/100,000 respectively for males, and 16.9/100,000 and 14.9/100,000 respectively for females. Over the observation period, there were more countries demonstrating decreasing trends in mortality than in incidence. There was also a disparity between which countries had comparatively high mortality rates and which had high incidence rates – for instance, the USA, which had by far the highest SCC incidence rates, had among the lower mortality rates. Overall reductions in DALYs were observed in 24 of 33 countries for males, and 25 countries for females. Conclusions: Over the past 27 years, although trends in SCC incidence have risen in most countries, there is evidence that mortality rates have been decreasing, especially towards the end of the observation period. Overall, burden of disease as assessed using DALYs has decreased in the majority of countries. Future work will explore potential explanatory factors for the observed disparity in trends in SCC incidence and mortality. Research Sponsor: None.

10570 Poster Session

Demographical differentials of lung cancer survival in Bangladeshi patients.

Muhammad Rafiqul Islam, ATM Kamrul Hasan, Ferdous ara Begum, Nazrina Khatun, Md. Rafiqul Islam, Asaduz Zaman, Farida Arjuman, Ishrat Nur Ridi; National Institute of Cancer Research and Hos- pital, Dhaka, Bangladesh; National Institute of Caner Research and Hospital, Dhaka, Bangladesh; Na- tional Institute Of Cancer Research & Hospital, Dhaka, Bangladesh; National institute of Cancer Research and Hospital, Dhaka, Bangladesh

Background: Lung Cancer is the leading cause of cancer-related mortality and most common cancer in worldwide with more than a million deaths annually. 20.8% cancer related death caused by lung cancer and more than half of lung cancer occurred in Asia. Differences In the epidemiology of lung cancer among the developing country may shed light on possible genetic and demographical influences on lung cancer survival. Demographic stratification of lung cancer patients of Bangladesh is remain unclear due of lack of data We tried to figure out the demographic pattern and its impact on survival in Bangladeshi lung cancer patient. Methods: Previously diagnosed primary lung cancer patients attending Medical Oncology department of National Institute of cancer research and Hospital, a tertiary care center of Bangladesh, between 2018 and 2019 were included. Demographic and clinical data were collected retrospectively from the medical records. Results: A total of 1868 consecutive patient (1580 males, 288females) diagnosed to have lung cancer; Mean age was 60 years which quite early compare to other countries. Older than 70-year age groups had worse survival outcome (hazard ratio 1.04: 95% confidence interval: 1.17–1.68). Below 50-year group had better outcome with standard adjuvant or palliative chemotherapy whereas older groups had better survival with sequential radiotherapy and chemotherapy or concurrent chemo radiation (Hazard Ratio 0.45; 95% confidence interval: 0.30–0.67). Sex was not a predicting factor for overall survival (Hazard ratios 1.04 95% confidence interval 0.89- 1.22, P = 0.621). But, Male had better treatment response than the female (Hazard ratio and 95% confidence interval: 0.51 and 0.42-0.61, P = < 0.001). Education level had significant impact on survival outcome (Hazard ratio 0.58 and 95% confidence interval: 0.47-0.71, P = < 0.001). Underweight group had worse survival than the normal BMI group (Hazard ratio1.18 and 95% confidence interval 1.05-1.31, P = 0.005). Having the Comorbid condition at the time of diagnosis had shorter survival (Hazard ratio 1.16 and 95% confidence interval 1.04-1.30 P = 0.007). Histological variation had no survival benefit among the squamous, small cell or other histological types (p = 0.214, 0.494, 0.658 respectively). But adenocarcinoma or small cell carcinoma had better treatment response outcome. Eastern Cooperative Oncology Group performance status (ECOG-PS) 4 had worse outcome (Hazard ratio 1.95, 95% confidence interval 1.37–2.79; P = < 0.001). Conclusions: The sociodemographic related survival in lung cancer needs to be fully elucidated because of its implication in the design of experimental protocols for targeted chemoprevention, early disease screening, molecular marker based staging, and individualized treatment. Due to its extraordinary disease burden and the international variability in demographic variables, the lung cancer requires continual monitoring. Re- search Sponsor: None.

10571 Poster Session

Prognostic value of pre-infection routine laboratory parameters for COVID-19 mortality in tumor patients: Results of the ADHOK Coronavirus Tumor Registry.

Romina Roesch, Thomas Suedhoff, Clemens M. Wendtner, Frank Kullmann, Thomas Kubin, Markus Schaich, Clemens U. Mueller-Naendrup, Marinela Augustin, Frank Hartmann, Gerald Illerhaus, Holger Frithjof Hebart, Ruth Seggewiss-Bernhardt, Ullrich Graeven, Ralph Naumann, Alexander Kiani; Technische Universita€t Dresden, Dresden, Germany; Klinikum Passau, 2. Medizinische Klinik, Passau, Germany; Mu€nchen Klinik Schwabing, Klinik fu€r Ha€matologie, Onkologie, Immunologie, Palliativmedi- zin, Infektiologie und Tropenmedizin, Mu€nchen, Germany; Klinikum Weiden, Medizinische Klinik I, Weiden, Germany; Klinikum Traunstein, Klinik fu€r Ha€matologie & Onkologie, Traunstein, Germany; Rems-Murr-Klinikum Winnenden, Klinik fu€r Ha€matologie, Onkologie und Palliativmedizin, Winnenden, Germany; Onkologische und Ha€matologische Schwerpunktpraxis im Medizinischen Versorgungszen- trum II, Olpe, Germany; Klinikum Nu€rnberg, Klinik fu€r Innere Medizin 5, Nuremberg, Germany; Klini- kum Lippe, Klinik fu€r Ha€matologie und Onkologie, Lemgo, Germany; Klinikum Stuttgart, Klinik fu€r Ha€matologie, Onkologie und Palliativmedizin, Stuttgart, Germany; Stauferklinikum, Zentrum fu€r Innere Medizin, Mutlangen, Mutlangen, Germany; Sozialstiftung Bamberg, Medizinische Klinik V, Bamberg, Germany; Kliniken Maria Hilf GmbH, Klinik fu€r Ha€matologie, Onkologie und Gastroenterologie, Mo€nchengladbach, Germany; Marien Kliniken Siegen, Klinik fu€r Ha€matologie, Medizinische Onkologie und Palliativmedizin, Siegen, Germany; Klinikum Bayreuth GmbH, Medizinische Klinik IV, Bayreuth, Germany

Background: Tumor patients (pts.) are considered susceptible to severe COVID-19 after SARS-CoV-2 infection. However, they represent a heterogeneous group of individuals with variable risk. Identification of vulnerable subgroups is important for prioritization of vaccination strategies and possible early therapeutic intervention after infection. Methods: Tumor pts. with PCR-confirmed SARS-CoV-2 infection were included in the multicentric ADHOK registry by 22 institutions. Detailed information about tumor disease and treatment, as well as routine laboratory parameters determined at least 10 days prior to SARS-CoV-2 infection, was collected retrospectively. The primary endpoint was defined as the outcome of the SARS-CoV-2 infection, graded according to the WHO: asymptomatic, mild, moderate, severe, critical, and COVID-19-related death. Results: Until Feb. 5, 2021, 215 pts. (67% with solid tumors, 33% with hematological neoplasms) were included in the registry. 74% of the pts. had an active malignancy. The course of SARS-CoV-2 infection was rather variable: 66% of the pts. remained asymptomatic or showed a mild-to-moderate course, while the rest developed severe or critical disease. The COVID- 19-related mortality rate was 24%. Pre-infection routine laboratory values were available for 104 pts., obtained at a median of 21 days before SARS-CoV-2 infection. Compared to COVID-19 survivors, COV- ID-19 non-survivors showed significantly higher median levels of absolute neutrophil count (ANC: 3.6 vs. 6.4 /nL; p = 0.006, n = 91), neutrophil-to-lymphocyte ratio (NLR: 2.2 vs. 7.2; p = 0.005, n = 75), C-reactive protein (CRP: 9.9 vs. 42.0 mg/L; p = 0.001, n = 104), and lactate dehydrogenase (LDH: 213.0 vs. 267.0 U/L; p = 0.016, n = 78). When categorized by a median split, COVID-19 mortality was significantly higher in pts. with ANC > 4.4 /nL (4% vs. 55%, p < 0.001), NLR > 4.1 (5% vs. 58%, p < 0.001), CRP > 15.4 mg/L (18% vs. 46%, p = 0.003), LDH > 236 U/L (15% vs. 49%, p = 0.003) and lymphocytes < 1.3 /nL (41% vs. 11% p = 0.002). In multivariable analysis, ANC and CRP showed a strong and significant association with COVID-19-related death (OR 23.0 and 7.7, p = 0.007 and 0.029, respectively). To develop an easy-to-apply pre-infection score, we combined ANC and CRP and were able to separate three groups of pts. with significantly different COVID-19 outcomes (p < 0.001) (Table). Conclusions: Our results unveil subgroups of tumor pts. who may be at increased risk of severe COVID-19 and point to pre-infection routine laboratory parameters with potential prognostic power: ANC and CRP may help identify pts. at risk for severe COVID-19 before SARS-CoV-2 infection. Re- search Sponsor: Arbeitsgemeinschaft der Haematologen und Onkologen im Krankenhaus e.V, Research funds of the Klinikum Bayreuth GmbH, Germany.

Score COVID-19-related mortality (%) Pre-infection parameters = risk for severe COVID-19 (n = 83)

Absolute neutrophil count (ANC) < 4 /nL 0 = low 0% ANC > 4 /nL and CRP < 20mg/L 1 = increased 30% ANC > 4 /nL and CRP > 20mg/L 2 = high 68%

10572 Poster Session

Outcomes for hospitalized cancer patients with COVID-19 during the height of pandemic in New York City.

Amelia Sawyers, Margaret Chou, Paul Johannet, Nicholas Gulati, Yingzhi Qian, Hua Zhong, Iman Osman; NYU Grossman School of Medicine, New York, NY; The Ronald O. Perelman Department of Der- matology, New York University School of Medicine, New York, NY; Department of Population Health, New York University School of Medicine, New York, NY

Background: Several reports have suggested that cancer patients are at increased risk for contracting severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and suffering worse coronavirus disease 2019 (COVID-19) outcomes. However, little is known about the impact of cancer status on presentation and outcome. Here, we report on the association between cancer status and survival in hospitalized patients who tested positive for SARS-CoV-2 during the height of pandemic in New York City. Methods: Of the 6,724 patients who were hospitalized at NYU Langone Health (3/16/20 - 7/31/20) and tested positive for SARS-CoV-2, 580 had either active cancer (n = 221) or a history of cancer (n = 359). Patients were classified as having active malignancy if they either received treatment within six months of their COVID-19 diagnosis or they had measurable disease documented at the time of their hospitalization. Patients were categorized as having a history of cancer if there was no evidence of measurable disease or there were no treatments administered within six months of their COVID-19 diagnosis. We compared the baseline clinicodemographic characteristics and hospital courses of the two groups, and the relationship between cancer status and the rate of admission to the intensive care unit (ICU), use of invasive mechanical ventilation (IMV), and all-cause mortality. Results: There was no differences between the two groups in their baseline laboratory results associated with COVID-19 infection, incidence of venous thromboembolism, or incidence of severe COVID-19. Active cancer status was not associated with the rate of ICU admission (P = 0.307) or use of IMV (P = 0.236), but was significantly associated with worse all-cause mortality in both univariate and multivariate analysis with ORs of 1.48 (95% CI: 1.04-2.09; P = 0.028) and 1.71 (95% CI: 1.12-2.63; P = 0.014), respectively. Conclusions: Active cancer patients had worse survival outcomes compared to patients with a history of cancer despite similar COVID-19 disease characteristics in the two groups. Our data suggest that cancer care should continue with minimal interruptions during the pandemic to bring about response and remission as soon as possible. Addi- tionally, these findings support the growing body of evidence that malignancy portends worse COVID-19 prognosis, and demonstrate that the risk may even apply to those without active disease. Research Sponsor: U.S. National Institutes of Health.

10573 Poster Session

Trends in breast cancer incidence among young women aged 20 to 49 years in the United States.

Shuai Xu, Yunan Han, Fei Wan, Adetunji T. Toriola; Washington University School of Medicine, St. Lou- is, MO; Division of Public Health Sciences, Department of Surgery, Washington University School of Medicine, St. Louis, MO; Washington University School of Medicine in St. Louis, St. Louis, MO

Background: Breast cancer in young women is diagnosed at more advanced stages and has a less favorable prognosis. We investigated trends in breast cancer incidence by race/ethnicity, hormone receptor status, and tumor stage in women aged 20-49 years over the past 25 years, as well as the impact of period and cohort effects on these trends. Methods: We used data from Surveillance, Epidemiology, and End Results (SEER) 13 registries for 1993-2002 and SEER 18 registries for 2003-2017. We calculated age-standardized incidence rates and annual percent change (APC), and stratified by race/ethnicity, hormone receptor status (estrogen receptor [ER] and progesterone receptor [PR]), and tumor stage (I- IV) for 222,424 women aged 20-49 years with a primary invasive breast cancer. We performed age-period-cohort analysis (presented as incidence rate ratios [IRR]) to investigate the effects of age, period, and cohort on incidence trends using the 1948 cohort and 1993-1997 period as the reference groups, respectively. Results: Between 2010-2017, invasive breast cancer incidence increased (APC = 0.67%, 95%CI: 0.32 to 1.03) among women aged 20-49 years, after being stable from 1993-2010. There were differences by race over the 25-year period (1993-2017). We observed significant increases in incidence among non-Hispanic White (NHW) (APC = 0.25%, 95%CI: 0.16 to 0.34), non-Hispanic Asia/ Pacific Islander (NHAPI) (APC = 0.58%, 95%CI: 0.34 to 0.82), and Hispanic women (APC = 0.59%, 95%CI: 0.34 to 0.83), but not among non-Hispanic black (NHB) women (APC = 0.14%, 95%CI: -0.06 to 0.34). Incidence increased for ER+ tumors but decreased for ER- tumors: ER+/PR+ (APC = 2.39%, 95%CI: 2.20 to 2.58), ER+/PR- (APC = 1.46%, 95%CI: 1.05 to 1.87), ER-/PR+ (APC = -6.33%, 95%CI: -7.31 to -5.33), and ER-/PR- (APC = -0.70%, 95%CI: -1.09 to -0.32). The decrease in ER-/ PR- tumors appeared largely driven by decreases among HNW women. Incidence for stages I (APC = 0.31, 95%CI: 0.07 to 0.55), II (APC = 0.99, 95%CI: 0.82 to 1.16), and IV (APC = 2.88, 95%CI: 2.37 to 3.39) tumors increased while that for stage III tumors decreased (APC = 0.81%, 95%CI: -1.04 to -0.59). Both the cohort and period effects impacted incidence, with the cohort effect almost 10 times larger than the period effect. Age-specific relative risk by birth cohort initially decreased between 1948 and 1958 but steadily increased from 1958 to 1993. Breast cancer incidence was higher among women born in the 1988 (IRR = 1.17, 95%CI: 1.07 to 1.28) and 1993 (IRR = 1.22, 95%CI: 0.99 to 1.51) cohorts than for those born in 1948 cohort. Conclusions: Breast cancer incidence is increasing among young women, mainly driven by increases in ER+ tumors. Prevention efforts need to focus on how we can address factors driving the increase in ER+ tumors and also learn from what has worked for decreasing ER- tumors. Research Sponsor: U.S. National Institutes of Health.

10574 Poster Session

Socioeconomic background in relation to stage at diagnosis in women with breast cancer.

Juan Manuel Ariza, Anne Cowppli-Bony, Solenne Billon, Stephanie Ayrault-Piault, Audrey Blanc- Lapierre, Florence Florence Molinie; Cancer Registry (Loire-Atlantique, Vendee), Nantes, France; Insti- tut de Cance�rologie de l’Ouest, Nantes Saint-Herblain, France

Background: Breast cancer (BC) has been associated with socioeconomic deprivation and rural residence. However, it’s still unclear how these factors interplay to affect the frequency and incidence of early and advanced BC. Methods: Taking advantage of the Loire-Atlantique/Vende�e cancer registry (France), we investigated the association between early (TNM stage < 2) and advanced (TNM stage $2) BC and the socio-economic background (SB) of women diagnosed in the study region from 2008 to 2015. Socioeconomic status was studied using the residence information of every patient linked to the European Deprivation Index (EDI), an ecological index constructed to reflect individual deprivation experienced at the smallest geographical unit of France (IRIS = 2000 inhabitants). To investigate SB, we created a composite variable using the EDI and the urban/rural context information, to define 4 categories: affluent-urban, affluent-rural, deprived-urban, and deprived-rural. Two statistical approaches were implemented: i) mixed-effects logistic regression models to examine the likelihood (relative risk, RR) of being diagnosed with advanced BC, and ii) Poisson regression for modeling incidence rates of early and advanced stages. Analyses were stratified by age ( < 45, 45 to 74, > 75 years) with a random intercept at the IRIS level. For the strata 45 to 74, the models were adjusted for organized screening. Results: During the study period, 14,542 BC cases were recorded. Compared to the women diagnosed in the most affluent-urban areas, a higher proportion of cancer diagnosed at an advanced stage was observed in the women living in more deprived-rural areas (aged < 45 years, RR = 1.48 95% CI 1.17 - 1.73; aged 45 to 74 years, RR = 1.22, 95% CI 1.02-1.42; aged > 74 years, RR = 1.08, 95% CI 0.85- 1.27). Furthermore, while in the population under 74 years, incidence rates of early BC in deprived-rural women were reduced by 25-85 % in comparison to affluent-urban women (IRR: 0.33 to 0.97 95% CI 0.16 -0.98), the rates of advanced BC were unaffected by deprivation-rurality (Deprived rural vs Af- fluent urban, IRR: 1.06 to 1.13 95% CI 0.79 -1.72). For the population over 74 years, we were unable to detect any associations between SB and BC incidence by stage at diagnosis. Conclusions: Advanced stage at diagnosis was more frequent among deprived–rural women aged under 74 years and coincided with the low incidence rates of early stages in this population. No disparities in the incidence rates of advanced BC were detected according to their SB in any age classes, suggesting other factors may be stronger contributors to the advanced stage at diagnosis. Future research should investigate whether screening practices may influence the disparities in the early stage at diagnosis for women under 74. Research Sponsor: SIRIC-ILIAD Nantes, France.

10575 Poster Session

Incidence and trend of Epstein-Barr virus-related cancer: A surveillance, epidemiology, and end results program based study.

Jiayi Shen, Meijuan Luo, LiTing Liu, Qiu-Yan Chen, Lin-Quan Tang, Hai-Qiang Mai; Department of Na- sopharyngeal Carcinoma, Sun Yat-sen University Cancer Center, Guangzhou, China

Background: Epstein-Barr virus (EBV) infection was highly prevalent, as was found in more than 90% of the adults globally. EBV infection has been found to be related with several types of cancer and classified as group 1 carcinogen by the International Agency for Research on Cancer. The association between EBV infection and malignancy was observed in Burkitt lymphoma (BL), Hodgkin lymphoma (HL), extra- nodal natural killer (NK)/T-cell lymphoma (NNKTL), gastric cancer (GC) and nasopharyngeal cancer (NPC). In this study, we aimed to analyze the incidence and the trend of incidence of these virus-related cancer and to identify whether the trend was similar between them. Methods: This was a retrospective analysis based on the data from Surveillance, Epidemiology, and End Results (18 registries, 2000- 2017), which totally included 71,415 patients. EBV-related cancers were defined as BL, HL, NNKTL, GC and NPC. Age-adjusted incidence rates were displayed as per 100,000 persons. In terms of incidence trend, we calculated the average annual percent change (AAPC). AAPC was considered significantly different from 0 when the P-value was smaller than 0.05. The impact of the epidemiological and clinical characteristics on the incidence trend was estimated, with cancer type, histology, age, sex and race considered. Results: Incidence rates of EBV-related cancers were 6.68 per 100,000 persons in 2000 and 5.80 in 2017, of which the AAPC was -0.8 (95%CI, -1.1 - -0.5, P-value < 0.001). (Table) Similar with EBV-related cancer as a whole, the APCCs of BL, HL and GC were statistically significantly smaller than 0, except that the APCCs of NNKTL and NPC were statistically significantly larger than 0 and not statistically significantly different from 0 respectively. The incidence of EBV-related cancer also decreased in mixed cellularity classical HL, nodular sclerosis classical HL, adenocarcinoma of GC, signet ring cell carcinoma in GC, undifferentiated carcinoma of NPC, squamous cell carcinoma of NPC, patients diagnosed at the age of 20-39 years old and 60-79 years old, male patients and race as white, black or Asian, but increased in classical HL, NOS, nodular lymphocyte predominant HL and non-kera- tinizing carcinoma of NPC. Conclusions: Incidence of EBV-related cancer decreased during 2000 and 2017, which was consistent in BL, HL and GC. Research Sponsor: None.

Rate and APCC of EBV-related cancers. EBV-related cancers Rate in 2000 Rate in 2017 APCC P-value EBV-related cancers 6.68 (6.46 - 6.90) 5.80 (5.62 - 5.99) -0.80 (-1.10 - -0.50) < 0.001 BL 0.37 (0.32 - 0.43) 0.32 (0.28 - 0.37) -1.60 (-3.00 - -0.10) < 0.001 HL 2.93 (2.79 - 3.07) 2.53 (2.41 - 2.65) -0.80 (-1.30 - -0.40) < 0.001 NNKTL 0.02 (0.01 - 0.03) 0.09 (0.07 - 0.11) 1.90 (0.20 - 3.60) < 0.001 GC 2.77 (2.63 - 2.92) 2.35 (2.24 - 2.46) -0.70 (-1.10 - -0.40) < 0.001 NPC 0.59 (0.53 - 0.66) 0.52 (0.47 - 0.58) -0.60 (-1.30 - 0.10) 0.10

10576 Poster Session

Leukemia in hospitalized patients with inflammatory bowel disease: An analysis of the National Inpatient Sample (NIS) database.

Colin Wikholm, Shiva Shankar Vangimalla, Ehab Abaza, Akram Ahmad, Ioannis Pothoulakis, Urvish Patel, Asma Ali Dilawari, Won Kyoo Cho; MedStar Georgetown University Hospital, Washinton, DC; MedStar Washington Hospital Center, Washington, DC; MedStar Georgetown University Hospital, Washington, DC; Icahn School of Medicine at Mount Sinai, New York, NY; Lombardi Comprehensive Cancer Center, Washington, DC

Background: Inflammatory bowel disease (IBD) and use of immunosuppressive therapy in IBD is linked with increased risk of leukemia. We studied the NIS database from 2003-2017 to analyze trends in any type of leukemia in IBD hospitalizations over time and examined the role of age, sex, and race. Methods: We analyzed NIS data of all adult hospitalizations for ulcerative colitis (UC) or Crohn’s disease (CD) with any type of leukemia as a primary or secondary diagnosis using validated ICD 9/10 codes. Age, sex, and racial demographics were collected. Trend analysis of leukemia was performed with Cochran-Armit- age and Jonckheere-Terpstra tests. Results: Overall Trends: From 2003-2017, a total of 11,385 of 2,235,413 (0.51%) CD hospitalizations and 8,105 of 1,324,746 (0.61%) UC hospitalizations contained diagnosis of leukemia. An increase in leukemia was seen in both CD and UC group from 0.24% to 0.79% (pTrend < 0.0001) and 0.28% to 0.81% (pTrend < 0.0001) respectively. Sex: In both UC and CD patients, leukemia diagnoses were predominantly male in 2003 but approximated a near 1:1 ratio by 2017 (Table). In CD, the proportion of female (FEM) leukemia diagnoses grew from 31.33% to 45.05% from 2003 to 2017 (pTrend = 0.1898). In UC, the proportion of female leukemia diagnoses grew from 27.49% to 45.79% from 2003 to 2017 (pTrend = 0.0030). Age: Leukemia was more common with increasing age, with no significant changes in proportion of cases between age groups over time (pTrend >.05). Ethnicity: White patients composed 87.80% and 84.24% of leukemia diagnoses in CD and UC, respectively. In CD, an increasing proportion of leukemia diagnoses occurred in black (BK) patients, and a decreasing proportion occurred in white patients (pTrends <.0001; Table 1) during the study time. No trends in race were observed in the UC group (pTrend = 0.4229). Conclusions: Our study showed an increased prevalence of leukemia in CD and UC hospitalizations from 2003-2017 which may be related to increasing use of immunosuppressants such as anti-TNF medications. In both CD and UC, leukemia was male-predominant, but increasingly female by 2017. Rate of leukemia diagnosis increased with age. In the CD group but not the UC group, leukemia was increasingly prevalent in black patients. Research Sponsor: None.

Leukemia (%) Prevalence Sex and Leukemia Race and Leukemia Year CD* UC* CD (% FEM) UC (% FEM)* CD (% BK)* UC (% BK) 2003 0.24 0.28 31.33 27.49 0 0 2004 0.34 0.27 43.15 46.71 2.86 11.53 2005 0.46 0.41 47.24 46.39 2.08 15.4 2006 0.41 0.4 52.58 43.62 0.92 2.17 2007 0.47 0.47 40.9 43.54 6.86 7.47 2008 0.39 0.51 50.57 43.07 1.72 4.57 2009 0.43 0.63 54.18 44.87 2.24 5.4 2010 0.49 0.62 52.44 42.74 8.76 6.72 2011 0.42 0.69 51 49.05 9.57 5.46 2012 0.52 0.6 50.27 46.22 5.62 6.19 2013 0.52 0.62 48.11 42.64 8.89 5.65 2014 0.55 0.66 56.65 53.85 11 4.96 2015 0.57 0.73 47.42 46.34 9.62 4.43 2016 0.67 0.85 48.99 46.43 5.44 6.91 2017 0.79 0.81 45.05 45.79 8.39 9.34 pTrend <.05 is designated by ‘*’.

10577 Poster Session

The global burden of 29 cancer groups from 2010 to 2019: A systematic analysis for the Global Burden of Disease study 2019.

Jonathan M Kocarnik, Kelly Compton, Franny Dean, Weijia Fu, Brian Gaw, James Harvey, Hannah Henrikson, Dan Lu, Alyssa Pennini, Rixing Xu, Force Lisa; Institute for Health Metrics and Evaluation, Seattle, WA

Background: Cancer is a major cause of morbidity and mortality worldwide, and global efforts to reduce health loss from cancer require systematic estimates that can measure progress from national to global levels. As part of the Global Burden of Diseases, Injuries, and Risk Factors Study 2019 (GBD 2019), we examined global cancer burden in order to highlight areas where cancer burden is inequitably distributed and to inform cancer control efforts around the world. Methods: Using estimation methods from GBD 2019, we analyzed the incidence, mortality, years lived with disability, years of life lost (YLLs), and disability-adjusted life years (DALYs) for 29 cancer groups and 204 countries and territories from 2010 to 2019. Cancer burden was compared to health burden from other categories of diseases and injuries in the GBD. Results were assessed globally and by socio-demographic index (SDI), a summary measure of income per capita, average educational attainment, and total fertility rate. Point estimates and 95% Uncertainty Intervals (UIs) are reported. Results: There were 23.6 million (95% UI 22.2-24.9 million) incident cancer cases globally in 2019 (17.2 [15.9-18.5] million excluding non-melanoma skin cancer), and 10.0 (9.36-10.6) million cancer deaths. There were 250 (235-264) million DALYs globally due to cancer, 97% of which came from years of life lost. The leading five cancers by DALYs in 2019 were: tracheal, bronchus, and lung cancer (45.9 [42.3-49.3] million); colon and rectum cancer (24.3 [22.6-25.7] million); stomach cancer (22.2 [20.3-24.1] million); breast cancer (20.6 [19.0- 22.2] million; and liver cancer (12.5 [11.4-13.7] million). Compared to other diseases and injuries in the GBD, cancer was responsible for the second-highest number of deaths, YLLs, and DALYs globally in 2019. These rankings of cancer burden differed by SDI quintile: cancer was the leading cause of absolute DALYs in high SDI countries but was ranked 10th in low SDI countries. From 2010-2019, the number of global cancer cases increased by 26.3% (20.3-32.3%), deaths by 20.9% (14.2-27.6%), and DALYs by 16.0% (9.29-22.8%). The largest annualized rate of change in absolute cases and deaths over this period occurred in the low and low-middle SDI quintiles. Conclusions: Cancer cases and deaths are growing globally, with the largest relative growth over the last decade occurring in low to middle SDI countries. Improvements in cancer prevention efforts and ensuring access to timely diagnosis and care will be necessary to make equitable progress in reducing the global burden of cancer. Research Sponsor: Bill and Melinda Gates Foundation, Other Foundation.

10578 Poster Session

The lung cancer obesity paradox: An analysis of 432,924 patients.

Brittany Miles, James David Mackey; UTMB Health, League City, TX; Genesis Medical Group, Spring, TX

Background: The lung cancer obesity paradox is the unexpected inverse relationship between body mass index (BMI) and lung cancer mortality. While there is a growing body of evidence to support the existence of the obesity paradox in lung cancer, little is known about its magnitude and relationship to cancer incidence and its impact on outcomes from surgery, chemotherapy, immunotherapy, and radiation treatment. Methods: To evaluate the impact of obesity on lung cancer incidence, we used TriNetX, a global federated health research network providing access to electronic medical records (diagnoses, procedures, medications, laboratory values, genomic information) from approximately 69 Million patients in 49 large Healthcare Organizations. We evaluated 2 patient cohorts of 216,462 adult smokers aged 18 to 75 that were matched for age, race, gender, and ethnicity. One cohort of patients carried a diagnosis of overweight and obesity (ICD-10 code E66), while the other cohort required exclusion of those diagnoses. Results: We found a statistically significant decrease in lung cancer incidence for patients with obesity (1.407% vs 2.039%, p < 0.0001), in addition to superior overall survival (95.344% vs 92.039%, p < 0.0001). A subset analysis of patients who contracted lung cancer showed a statistically significant benefit in median survival in favor of patients with overweight and obesity (851 vs 602 days, p value 0.0009). Conclusions: These findings support the existence of the obesity paradox in lung cancer, and its positive impact on both lung cancer incidence and outcome. Research Sponsor: None.

10579 Poster Session

Hereditary cancer testing in an ethnically diverse U.S. population.

Michele Basiliere, Stephanie Woods, Heidi Owen, Lindsay Dohany, Lisa McDaniel, Nicholas Wisniewski; Progenity, Inc., Ann Arbor, MI; Progenity, Inc, Ann Arbor, MI

Background: The clinical utility of hereditary cancer multigene testing is well-established. Most testing has been performed in the White European population, with a relative shortage of data in other ethnic groups. Evaluation of hereditary cancer variants within diverse ethnic populations is important to drive accurate variant interpretation. This study aims to review the outcomes of hereditary cancer testing in an ethnically diverse U.S. population. Methods: We conducted a retrospective analysis of 8,888 test results from a 31-gene hereditary cancer test using next-generation sequencing with copy number variant analysis. Results were separated into seven categories based on the ethnicity selected on the test requisition: African American (AA), Asian (A), Ashkenazi Jewish (AJ), Hispanic (H), White (W), Other (O), and Unknown (U). O was used for any handwritten ethnicity or if multiple ethnicities were selected and U if no ethnicity was chosen. We quantified the number of negative, positive (pathogenic or likely pathogenic (P/LP) variant) and variant of uncertain significant (VUS) results for each ethnic group. We tallied the gene and variant for each result with a positive, negative or VUS. We used a Fisher’s Exact test and Bon- ferroni-adjusted p-values to account for multiple hypothesis testing. Results: Of the 8,888 orders reviewed, 45% were non-White ethnic groups. The P/LP variant rate was lower for AA (4%, p=2e-04) compared to W (9%). The VUS rate was higher for O, AA and A (48%, 56%, and 57%; p=0.0045, 3.1e-17 and 4.8e-06, respectively), compared to 38% in W. Lower negative rates were found in A and AA (37% and 40%; p=0.00024 and 9.4e-10, respectively), and trending lower in O (46%, p=0.0097), compared to 58% for W. The VUS rate for H is 41% (p value = 0.095). Conclusions: Our results suggest that non-Whites may be at a disadvantage when it comes to hereditary cancer multigene testing due to the higher rate of VUS results. The inverse correlation between the overall rates of VUS and negative results when comparing these populations suggests there may be limited variant information for the non- White population in medical literature and available databases. This underrepresentation may make it more difficult to accurately characterize a variant. Despite the small sample size, these findings are consistent with previous publications; however, gene specific outcomes could not be evaluated. This find- ing suggests that providers could give these patients uninformative results more often, which has potential to impact screening protocols for these families. More research is needed to understand the impact of variant classification across ethnicities to decrease health disparities. Research Sponsor: None.

10580 Poster Session

Point of care genetic testing in a breast cancer survivorship clinic.

Lori Ranallo, Lauren Elizabeth Nye, Mary Williams, Carol J. Fabian F.A.S.C.O, Anne O’Dea, Jennifer R. Klemp; University of Kansas Cancer Center, Westwood, KS; Northwestern University, Feinberg School of Medicine, Chicago, IL; University of Kansas Health System, Westwood, KS; University of Kansas Medical Center, Kansas City, KS; University of Kansas Medical Center, Westwood, KS

Background: Breast cancer survivorship care (BCSC) includes the ongoing assessment of personal and family cancer history and offering genetic education, counseling and testing to survivors who meet NCCN, ASBrS and Medicare guidelines for germline genetic testing. It is reported that approximately 8% of patients with breast cancer (BC) will have a clinically actionable germline mutation. However, lower than expected rates of testing are seen in both the acute and extended phases of BCSC. We sought to identify the number of patients seen in a long-term survivorship clinic who had previously undergone or currently qualified for germline testing, and the prevalence of germline variants in BC survivors. Meth- ods: In a Nurse Practitioner (NP) led clinic, 2,184 non-selected BC survivors were screened to determine if: germline testing was previously completed or if update germline testing or initial germline testing is needed (with a 3-generation review of family history). BC survivors eligible for initial or update germline testing (411 patients) were provided with genetic education, counseling, and offered multigene panel testing. Seven (7) BC survivors declined testing. Results: From May 2019 – January 2021, 2,184 BC survivors were seen in the clinic. The average age of survivors = 60.2 yrs; average time since diagnosis = 10.7 yrs; and average age at diagnosis = 50.1 yrs, gPV were identified in 10.4%. Out of pocket cost on average was $50.00 for 2.0% of those tested. Conclusions: Within a comprehensive Breast Cancer program where genetic testing is common practice, there is an ongoing need to screen breast cancer (BC) survivors for genetic testing eligibility. A significant number of BC survivors will test positive for a pathogenic mutation (10.4%) a decade after an initial diagnosis. Genetic testing is a necessary step to stratify a BC survivors’ risk of developing secondary cancers, appropriate screening and prevention strategies, cascade testing, and for some, treatment planning. This individualized approach to BCSC is often described, but difficult to put into action. Time/access and drop rates with a referral model are barriers. Incorporating a point of care genetic testing model requires additional support (genetic extender), professional development, education, and a commitment to provide patient centric care. Research Sponsor: None.

Germline Testing Total Tested (n = 404) Positive Family History 75% (n = 300) Results Most common gPV identified Pathogenic Variant 10.4% (n = 42) CHEK2 (9), MUTYH (7), ATM (6) and APC (3), BRCA2 (3) VUS 30% (n = 121) No Pathogenic Variant 59.6% (n = 241) Positive Family History 75% (n = 300) Results Pathogenic Variant 11.2% (n = 45) VUS 29% (n = 116) No Pathogenic Variant 59.8% (n = 240)

10581 Poster Session

Germline pathogenic variants in cancer predisposition genes among women with invasive lobular cancer of breast.

Siddhartha Yadav, Chunling Hu, Susan M. Domchek, Jeffrey N. Weitzel, David Goldgar, Peter Kraft, Katherine L. Nathanson, Rachid Karam, Elizabeth Chao, Amal Yussuf, Tina Pesaran, Jill S. Dolinsky, Steven Hart, Holly LaDuca, Eric Polley, Fergus Couch, The CARRIERS Consortium; Mayo Clinic, Roch- ester, MN; University of Pennsylvania, Philadelphia, PA; Oncogenetics for Precision Prevention, and Latin American School of Oncology, Sierra Madre, CA; University of Utah, Salt Lake City, UT; Harvard School of Public Health, Boston, MA; Ambry Genetics Corp, Aliso Viejo, CA; UC Irvine, Irvine, CA; Am- bry Genetics, Aliso Viejo, CA; Department of Laboratory Medicine and Pathology, Mayo Clinic, Roches- ter, MN

Background: The prevalence of germline pathogenic variants (PVs) in cancer predisposition genes among women with invasive lobular breast cancer (ILC) and the risk of ILC in PV carriers is not well-defined. Methods: The study included 2,999 women with ILC and 32,544 unaffected controls from a population-based cohort; 3,796 women with ILC and 20,323 women with invasive ductal carcinoma (IDC) undergoing clinical multigene panel testing (clinical cohort); and 125,748 exome sequences from unrelated women without a cancer diagnosis in the gnomAD 3.0 dataset. Frequencies of germline PVs in breast cancer predisposition genes (ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, PALB2, PTEN, RAD51C, RAD51D, and TP53) were compared between women with ILC and unaffected controls in both cohorts and between women with ILC and IDC in the clinical cohort. Results: The frequency of PVs in breast cancer predisposition genes among women with ILC was 6.5% in the clinical cohort and 5.2% in the population-based cohort. In case-control analyses, CDH1 and BRCA2 PVs were associated with high risks of ILC (Odds ratio (OR) > 4), and CHEK2, ATM and PALB2 PVs were associated with moderate (OR = 2-4) risks. BRCA1 PVs and CHEK2 p.Ile157Thr were not associated with clinically relevant risks (OR < 2) of ILC. PV frequencies in these genes in ILC and IDC were similar except for PV frequencies in BRCA1 and CDH1. Conclusions: The study establishes that PVs in ATM, BRCA2, CDH1, CHEK2 and PALB2 are associated with an increased risk of ILC, whereas BRCA1 PVs are not. The similar overall PV frequencies for ILC and IDC suggest that cancer histology should not influence the decision to proceed with genetic testing. While, multigene panel testing may be appropriate for women with ILC, CDH1 should be specifically discussed in the context of low prevalence and attendant gastric cancer risk. Research Sponsor: U.S. National Institutes of Health.

10582 Poster Session

Rate of incidental germline findings detected by tumor-normal matched sequencing in cancer types lacking hereditary cancer testing guidelines.

Timothy A. Yap, Arya Ashok, Jessica Stoll, Anna Ewa Schwarzbach, Kimberly L. Blackwell, Tianhong Li, Hyunseok Kang, Judy Ellen Garber, Funda Meric-Bernstam; The University of Texas MD Anderson Can- cer Center, Houston, TX; Tempus, Chicago, IL; Tempus Labs, Chicago, IL; University of California Davis Comprehensive Cancer Center, Sacramento, CA; University of California, San Francisco, San Francisco, CA; Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA

Background: Up to 10% of all cancers are associated with hereditary cancer syndromes; however, guidelines for germline testing are currently limited to patients and families with specific cancer types (ovarian, breast, prostate, pancreatic, etc.). Although germline alterations have been shown in genes associated with cancers such as bile-duct, head & neck, brain, bladder, esophageal, and lung cancers, genetic testing is not routinely offered (PMID: 28873162). In such cancers, a guidelines-based approach may fail to detect cancer risk variants found by tumor-normal (T/N) matched sequencing. Here, we report the prevalence of incidental germline findings in patients with the aforementioned 6 cancer types and highlight frequently mutated genes by cancer type. Methods: We retrospectively analyzed next-generation sequencing data from de-identified records of 19,630 patients tested using Tempus|xT T/N matched assay. Incidental germline findings (i.e., single nucleotide variants and small insertions/ deletions) detected in 50 hereditary cancer genes were determined for: bile duct (n = 466), head & neck (n = 673), esophageal (n = 395), brain (n = 1,391), bladder (n = 810), and lung (n = 5,544), where n = total patients. For comparison, we also included 4 cancer types that frequently undergo germline testing: ovarian (n = 2,042), breast (n = 3,542), prostate (n = 2,146), and pancreatic (n = 2,621). Results: We detected incidental pathogenic/likely pathogenic germline variants (P/LPV) in 6.5% (601/ 9,279) of patients diagnosed with the 6 selected cancer types lacking hereditary cancer testing guidelines. The highest prevalence of P/LPV was identified in patients with bladder (8%), brain (6.9%), and lung (6.5%) cancers. Frequently mutated genes (Table) include ATM (n = 62), BRCA2 (n = 60), BRCA1 (n = 33), APC (n = 27), and CHEK2 (n = 21). Of note, the Ashkenazi Jewish variant (p.I1307K) was the most frequent mutation in APC. For cancer types where patients frequently undergo germline testing, the rates of incidental germline findings in descending order were ovarian (15%), breast (12%), prostate (9.4%), and pancreatic (8.5%) cancers. Conclusions: In addition to enhanced variant calling, T/N matched sequencing may identify germline variants missed by a guidelines-based approach to testing. The identification of such germline findings may have clinical implications for the patient, as well as at-risk family members, thereby resulting in the opportunity for genetic counseling and risk- stratified intervention. Research Sponsor: None.

# Top altered genes/cancer type (n) Patients P/LPV Cancer Total # with P/ incidence type Patients LPV (%) ATM APC BRCA1BRCA2 CHEK2 MSH2 MSH6 NF1 PALB2RAD51D TP53

Bladder 810 65 8 6 3 3 8 4 3 Brain 1391 96 6.9 7 5 6 7 11 Lung 5,544 362 6.5 47 22 17 44 16 Esophagus 395 22 5.6 4 3 2 2 Bile duct 466 26 5.6 3 3 3 2 2 Head & 673 30 4.5 2 2 3 2 3 Neck

10583 Poster Session

Variant reclassification and its impact on clinical care in an Asian cancer center.

Jianbang Chiang, Tze Hao Chia, Sock Hoai Chan, Joanne YY Ngeow; National Cancer Centre Singapore, Singapore, Singapore; Lee Kong Chian School of Medicine, Singapore, Singapore; Cleveland Clinic, Cleveland, OH

Background: Genetic testing has demonstrated clinical utility in the identification and subsequent surveillance of patients with cancer predisposition syndromes. However, the increased likelihood of en- countering a variant of uncertain significance (VUS) in individuals of non-European descent such as Asians may be challenging to both clinicians and patients in interpretation and management. VUS can be reclassified as more data becomes available. VUS reclassification is important, as it may have implications for surveillance and treatment. This study aims to evaluate the prevalence and patterns of variant reclassification in an Asian country and its impact on patient management. Methods: A prospective cohort of patients seen at the Cancer Genetics Service at the National Cancer Center Singapore between February 2014 to March 2020 was evaluated. The frequency, direction and time to variant reclassification was assessed by comparing the reclassified report against the original report. Results: A total of 1412 VUS were reported in 49.9% (845/1695) of patients. Over six-years, 6.7% (94/1412) of variants were reclassified. Most VUS (94.1%; 80/85) were downgraded to benign/likely benign variant, with a smaller proportion of VUS (5.9%; 5/85) upgraded to pathogenic/likely pathogenic variant. Actionable VUS upgrades and pathogenic/likely pathogenic variant downgrades, that resulted in management changes, happened in 31.0% (39/126) of patients. The median and mean time taken for reclassification were 1 and 1.62 year(s) respectively. Conclusions: Clinicians need to put in place a system for review of variants, as variant reclassification can lead to changes in management in nearly 1/3 of patients. Management should be based on the patient’s personal history, family history and variant interpretation. We propose a clinical guideline to standardize management of patients with VUS. For clinically relevant or suspicious VUS, follow-up is recommended every two years, as actionable reclassifications may happen during this period. Research Sponsor: Singapore Ministry of Health National Medical Research Council.

10584 Poster Session

An evaluation of gender discrepancies in genetic referrals for BRCA testing for indicated malignancies.

Wesley Smith, Kayla Smith, William Sessions, Connor Evins, Michael Baker, Mary Blumer; Prisma Health-Upstate, Greenville, SC; Prisma Health Upstate, Greenville, SC; University of South Carolina School of Medicine Greenville, Greenville, SC

Background: Tumor suppressing genes BRCA1 and BRCA2 were discovered in 1990 and 1994, respectively, with mutations linked to hereditary breast-ovarian cancer syndromes (HBOCs). The discovery of these mutations has led to screening of at-risk patient populations and their family members. Women with BRCA1 or BRCA2 mutations are generally recommended to have prophylactic bilateral mastectomies and oophorectomies to decrease their future risk of cancer. While the initial discovery mostly focused on cancers in women, research has shown that BRCA mutations increase the risk of other cancers such as prostate cancer and pancreatic cancer, that also affect men. Previous research suggests that men are three times less likely to receive genetic testing in cancer driven by a 10:1 disparity in HBOC genetic testing. This was thought to be due to the lack of information on the importance of HBOC testing along with social roles in health. We wanted to evaluate the magnitude of the potential gender gap in BRCA testing in men compared to women. Methods: This was an IRB-approved, single center retrospective study to evaluate the rate of referrals to genetics for BRCA testing. Eligible patients had a personal history of cancer meeting criteria for BRCA testing per NCCN recommendations. Chart review was performed for patients with ovarian cancer, female breast cancer 45 years and younger, female triple negative breast cancer 60 years and younger, metastatic prostate cancer, all male breast cancer that have made an office visit since 2017, and pancreatic cancer since 2019. Rates of referral for genetic testing was the primary outcome and the groups were compared via the Chi-Square test. Results: 1,320 patients were included in the study, of which 664 were men and 656 were women. 128/664 (19.3%) of men were referred to genetics for screening compared to 527/656 (80.3%) for women (p <.001). Addi- tionally, 42/128 (32.8%) men who were referred for screening did not complete genetic screening compared to 72/527 (13.7%) women (p <.001). A total of 62/541 (11.5%) patients who completed screening had either a BRCA1 or BRCA2 mutation. Conclusions: In our study, men were referred for BRCA testing significantly less than women for primary cancers, despite recommendation from the NCCN. In addition, men were also more than twice as likely not to complete genetic screening even if referred. The integration of genetics and oncology will continue to grow as personalized medicine contin- ues to drive more treatment options. Closing this gender gap is important not only for familial screening purposes but also for treatment implications as patients with germline BRCA mutations are eligible for poly ADP ribose polymerase (PARP) inhibitors (e.g. olaparib) in both metastatic prostate cancer and pancreatic cancer. Further quality improvement initiatives are needed in order to close this gap by increasing education of the importance of BRCA testing in men. Research Sponsor: None.

10585 Poster Session

Germline testing and somatic tumor testing for BRCA1/2 pathogenic variants in ovarian cancer: What is the optimal sequence of testing?

Janice S. Kwon, Anna Tinker, Jennifer Santos, Katie Compton, Sophie Sun, Kasmintan A. Schrader, Aly Karsan; University of British Columbia, Vancouver, BC, Canada; British Columbia Cancer Agency, Van- couver, BC, Canada; BC Cancer, Vancouver, BC, Canada; BC Cancer Agency, Vancouver, BC, Canada; British Columbia Cancer Research Centre, Vancouver, BC, Canada

Background: In 2020 ASCO recommended that all women with epithelial ovarian cancer have germline testing (GT) for BRCA1/2 mutations, and those without a germline pathogenic variant (PV) should have somatic tumor testing (TT), to determine eligibility for PARP inhibitor (PARPi) therapy (GT-TT strategy). An alternate strategy is to start with tumor testing first, and to conduct germline testing only in those with a PV in the tumor, or a significant family history (TT-GT strategy). The objective was to conduct a cost-effectiveness analysis comparing the 2 testing strategies. Methods: A Markov Monte Carlo simulation model compared the costs (USD) and benefits of the 2 testing strategies. According to local empiric data, a sufficient tissue sample for TT was available in 99% of cases, otherwise the patient would only have GT. Sensitivity of TT was 99% for detecting germline PV. Only those with BRCA1/2 PV were eligible for PARPi. Primary outcomes included the number of women eligible for PARPi, with progression- free years of life (PFLY) gained based on SOLO1 data, and the incremental cost-effectiveness ratio (ICER). Monte Carlo simulation estimated the number of women who would have GT and TT, and the total with germline or somatic BRCA1/2 PV eligible for PARPi. Sensitivity analyses accounted for uncertainty around various parameters. Results: The GT-TT strategy was more effective but more costly than TT-GT in identifying patients eligible for PARPi. Table summarizes the average lifetime costs, benefits, and Monte Carlo simulation estimates for 10,000 women diagnosed with advanced epithelial ovarian cancer annually in the USA. The incremental benefit from the GT-TT strategy would be achieved at substantial cost to the health care system, with an ICER of $119,340 per PFLY gained relative to the TT- GT strategy. The results were highly sensitive to the sensitivity of TT to detect germline PV, and the costs of GT and TT. Assuming that GT was less than 50% of the cost of TT, the sensitivity of TT had to exceed 98% for the TT-GT strategy to be cost-effective. Conclusions: Although the ASCO recommended strategy of BRCA germline testing followed by tumor testing for those without a pathogenic variant may be more effective in identifying ovarian cancer patients for PARP inhibitor therapy, it is more costly. The ASCO strategy is justified if the sensitivity of tumor testing is not sufficiently high. However, assuming high tumor testing performance rates, tumor testing first followed by germline testing if there is a PV in the tumor and/or family history is a cost-effective strategy. Research Sponsor: Michael Smith Foundation for Health Research, Pharmaceutical/Biotech Company.

TT-GT GT-TT

Average lifetime outcomes Cost $109,730 $111,115 Incremental cost - $1,384 Effectiveness (PFLY) 2.684 2.6956 Incremental benefit - 0.0116 ICER - $119,340 Monte Carlo simulation (n=10,000) Tumor testing 9902 8021 Germline testing 2748 10,000 Eligible for PARPi 2178 2197 Germline BRCA PV identified 1884 1898

10586 Poster Session

Clonal hematopoiesis association with cardiac function and mortality in patients with solid tumors.

Jessica A. Regan, Michelle Green, Chester Kao, Eric Powers, Jadee Neff, John H. Strickler, Matthew Stuart McKinney, Svati H. Shah, Tian Zhang; Duke University Medical Center, Department of Medicine, Durham, NC; Duke University Medical Center, Durham, NC; Department of Medicine, Duke University, Durham, NC; Duke University Department of Medicine, Durham, NC; Duke Cancer Center, Durham, NC; Duke University School of Medicine, Durham, NC; Duke Molecular Physiology Institute, Durham, NC; Duke Cancer Institute Center for Prostate and Urologic Cancers, Duke University, Durham, NC

Background: Clonal hematopoiesis (CH) is the presence of expanded somatic clones in hematopoietic cells and is associated with higher overall mortality (OM). Studies suggest atherosclerotic cardiovascular disease may drive mortality, but the detailed mechanisms remain unclear. CH mutations can be detected in solid tumor sequencing, often confounding genomic tumor analysis. We evaluated the association of CH in solid tumor next-generation sequencing (NGS) with echocardiographic findings and OM. Methods: Sequential adult patients treated at the Duke Cancer Institute with solid tumor NGS analysis by FoundationOne were captured retrospectively. CH mutations present at a variant allele fraction $2% across 57 genes previously associated with hematologic malignancies were included. Pa- tients with echocardiograms between 2 years before NGS testing and up to 5 years afterward were analyzed. Association between CH mutations with cardiomyopathy (CM, left ventricular ejection fraction < 45%) and global longitudinal strain (GLS) was determined using logistic and linear regression, respectively. In a subset of patients with detailed cancer diagnosis date and clinical follow-up, Cox proportional hazard models were used to associate CH mutations with OM, with or without TP53/KRAS (included in most CH analyses but highly prevalent in solid tumors). Analyses were adjusted for age, gender and race. Results: Of 3029 patients with NGS testing, 2212 (73.0%) carried at least one CH related mutation, the majority of which were in TP53/KRAS. When excluding TP53/KRAS, CH mutations were observed in 806 of 3029 (26.6%) patients. CH mutations were associated with age (est 2.1, 95% CI 1.1-3.2, p < 0.001). Excluding TP53/KRAS strengthened the association between CH and age (est 2.8, 95% CI 1.8-3.9, p < 0.001). Echocardiogram data were available in 828 patients, of whom 48 (5.8%) had CM. CH mutations were not associated with CM (OR 1.3, 95% CI 0.6-2.6, p = 0.5), however when excluding TP53 and KRAS, CH mutations were associated with lower odds of (OR 0.4, 95% CI 0.1-0.9, p = 0.03). GLS was available in 423 patients and was not associated with CH mutations (p = 0.8 with TP53/KRAS; p = 0.4 without TP53/KRAS as CH). In 222 patients with clinical information, OM did not differ between the CH vs no CH cohorts (HR 0.8, 95% CI 0.6 = 1.2, p = 0.3 inclusive of TP53/KRAS). When excluding TP53/KRAS mutations, in this population of patients with cancer, non- TP53/KRAS CH was associated with less OM (HR 0.6, 95% CI 0.4-0.9, p = 0.01). Conclusions: In this patient population with cancer, CH mutations did not associate with higher CM. In contrast to prior studies, CH detected in solid tumor does not associate with OM in this population. CH mutations con- found tumor sequencing and these findings support the value of paired tumor-blood sequencing to determine true CH. Consensus around CH variants should be undertaken in future studies. Research Sponsor: Duke University.

10587 Poster Session

Genetic assessment of hereditary breast and ovarian cancer in the Harris Health System: A five-year, single-center experience.

Nicole Higashiyama, Shaun Bulsara, Susan G. Hilsenbeck, Tiffaney Tran, Ria Brown, Mary Fang, Cathy Sullivan, Georgiann Garza, C. Kent Osborne, Mothaffar F. Rimawi, Julie R. Nangia; Lester and Sue Smith Breast Center, Dan L. Duncan Comprehensive Cancer Center, Baylor College of Medicine, Hous- ton, TX; Baylor College of Medicine, Houston, TX

Background: Identifying patients with hereditary breast cancer is critical since lifetime breast cancer risk is as high as 85% for those with germline BRCA1/2 mutations and preventive interventions can reduce that risk. However, genetic assessments and counseling are often underutilized among racial/ethnic minority populations. Reducing this genetic testing gap is important since hereditary breast/ovarian cancer syndromes occur among racial/ethnic minorities at least as frequently as non-Ashkenazi Jewish, non-Hispanic White populations. More information on variants in these populations is also needed to better define their genetic susceptibility. Methods: We conducted a retrospective study of adult patients evaluated for genetic testing for hereditary breast/ovarian cancer by a genetic counselor between Octo- ber 1, 2009 and September 30, 2014 in Harris Health System which is a large, county health system composed mostly of underserved and minority patients. Data from 2015-2019 is currently being extracted and we are reporting the first 5 years of data. Descriptive statistics were used to summarize patient data. Results: 659 patients underwent genetic counseling (10.5% non-Hispanic White, 24.4% Black, 56.9% Hispanic, 5.9% Asian, and 2.3% other). Five patients had Ashkenazi Jewish ancestry. The majority of patients completed testing (87.4%) with 72.7% receiving financial assistance. Among those who did not complete testing, only 12.0% declined, while 66.3% did not meet guideline-based criteria or were recommended to have an affected relative tested. Multigene panel testing was not available until April 2014, so most underwent BRCA sequencing (75.0%) and/or a BRCA large rearrange- ment test (61.0%). 36.1% received multigene panel testing, 4.6% single site analysis, and 4.4% p53 sequencing. Deleterious mutations occurred in 98 (14.9%) patients: BRCA1 (n = 60), BRCA2 (n = 25), PALB2 (n = 7), ATM (n = 3), and other (n = 3). The distribution of races/ethnicities among those with deleterious mutations was similar to the overall population (7.1% non-Hispanic White, 18.4% Black, 69.4% Hispanic, 3.1% Asian, and 2.0% other). 80.6% of those with deleterious mutations had breast cancer. High rates of bilateral mastectomies were performed in patients with deleterious mutations: BRCA1 60%, BRCA2 55%, PALB2 57.1%, and ATM 33%. Risk-reducing salpingectomy or salpingo-oophorectomy was performed in 56.7% BRCA1, 60% BRCA2, 28.5% PALB2, and 33.3% other mutation carriers. Conclusions: We demonstrate that with the support of financial assistance programs, most patients who receive genetic counseling will accept genetic testing in a socioeconomically underserved, racially/ethnically diverse population. Identification of high-risk patients in these groups is critical since pathogenic variants in this population were common and more than half underwent risk- reducing procedures. Research Sponsor: None.

10588 Poster Session

Testing for mutations in BRCA1 and BRCA2 among ovarian cancer patients at a diverse academic medical center.

Caitlin Taylor, Ryan Mooney, Yuan Liu, Yichun Cao, Elizabeth Sakach, Christine R. Tallo, Jane Lowe Meisel; Emory University Hematology Medical Oncology-Fellowship Program, Atlanta, GA; Emory Uni- versity, Atlanta, GA; Departments of Biostatistics and Bioinformatics, Emory University, Atlanta, GA; Winship Cancer Center, Atlanta, GA; Emory University Hospital, Atlanta, GA; Department of Hematolo- gy and Medical Oncology, Winship Cancer Institute of Emory University, Atlanta, GA

Background: Testing for mutations in BRCA1 and BRCA2 is recommended for all women with ovarian cancer (OC), given important implications for treatment and prognosis. Despite this recommendation, studies show that only a small percentage of OC patients (pts) undergo genetic testing (GT). In this study, we evaluated rates of genetics referral, counseling and testing among OC pts at an academic medical center. Our goal was to identify factors associated with lower rates of GT. Given the large Black population at our center, we specifically wanted to evaluate the association between race and GT given limited existing data on this issue. Methods: Retrospective chart review was performed evaluating rates of referral and uptake for GT, and percentages of BRCA mutation carriers among pts with OC diagnosed and treated at Emory’s Winship Cancer Institute between 2008 and 2018. Associations between age, race, histology, family history (FH), performance status, provider characteristics and genetics referral and testing were evaluated using logistic regression models. Results: Of the 171 pts who met inclusion criteria, the majority were age 55 or older (62%) with high grade serous carcinoma (60.8%). Pts were predominantly Caucasian (59.4%), followed by Black (29.1%), Asian (10.3%) and Hispanic (1.2%). Overall, GT rates were low with 44.7% of pts referred for genetic counseling and 39.8% receiving testing. Among pts who did receive GT, the percentage of deleterious BRCA1 and BRCA2 mutations identified was 11% and 8.8% respectively. Variables correlating with higher likelihood of genetics discussion, referral and testing included serous histology (50% vs 23.9% non-serous, p < 0.001), Cau- casian or Asian race (87.5% Asian, 58.8% Caucasian vs 42.2% Black, p = 0.003) and seeing a medical oncologist (67.5% vs 44.7% seeing gynecologic oncologist alone, p = 0.004). Notably, while fewer Black women were referred for GT (25.9% vs 74.1% Caucasian), those that did undergo GT were found to have higher rates of BRCA1 and BRCA2 mutations when compared to Caucasian pts (22.2% vs 8.2% BRCA1; 11.1% vs 6.0% BRCA2). Pts with a FH of OC were more likely to undergo GT (69.2% vs 37.9%, p = 0.027), and pts with a FH of breast cancer were more likely be referred for testing (57.1% vs 39.6%, p = 0.042), suggesting that FH impacted referral patterns. Conclusions: The rates of GT among OC pts at our institution were lower than expected despite the broad recommendation for GT in this population. It is imperative to improve access to GT for all OC pts regardless of FH, and in particular among Black pts given the higher rates of BRCA mutations in this population. Pts and providers must work together to overcome barriers to genetics referral and testing in order to improve GT rates and clinical outcomes. Further research is needed to design interventions that may help improve adherence to this important recommendation in the future. Research Sponsor: None.

10589 Poster Session

A comprehensive literature review and meta-analysis on prevalence of BRCAm, HRRm and HRD+ across tumor types.

Changxia Shao, Jun Wan, Huilin Tang, Andrew Marley, Chelsey Miller, Madeline Brown, Yiqing Song, Jiali Han, Gboyega Adeboyeje; Merck & Co., Inc., Kenilworth, NJ; Indiana University School of Medi- cine, Indianapolis, IN; Integrative Precision Health LLC, Carmel, IN; Indiana University Richard M. Fairbanks School of Public Health, Indianapolis, IN; Indiana University Richard M Fairbanks School of Public Health, Indianapolis, IN

Background: Poly (ADP-ribose) polymerase inhibitor (PARPi) may have broad application in the treatment of cancer patients with mutations in BRCA (BRCAm) or other homologous recombination repair genes (HRRm) or with homologous recombination deficiency positive (HRD+). A literature review and meta-analysis were conducted to evaluate the prevalence of BRCAm, HRRm, and HRD+ across tumor types. Methods: Comprehensive searches for eligible studies in Ovid MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, and Cochrane reviews were performed in May 2020 to capture studies published in English, within 10 years for manuscripts and 3 years for abstracts across geograph- ic regions. A weighted summary estimate was calculated for BRCAm. A summary estimate with corresponding 95% CI was calculated using random-effects models for HRD+ and HRRm. Results: A total of 342 eligible studies with at least 100 samples were included in the review of BRCA1/2m prevalence, containing a total of over 469,000 samples across 24 tumor types. The most frequent indications examined in the included studies were breast (study number n = 144), ovarian (53), prostate (17) and pancreatic (11) cancers. The prevalence of germline BRCA1m (gBRCA1m) and gBRCA2m was 5% and 4% for breast, 12% and 5% for ovarian, 1% and 3% for prostate, and 1% and 4% for pancreatic cancer, respectively. The prevalence of somatic BRCA1m (sBRCA1m) and somatic BRCA2m (sBRCA2m) was 3% and 3% for breast, 7% and 5% for ovarian, 3% and 5% for prostate, and 1% and 3% for pancreatic cancer, respectively. Few studies evaluated endometrial, lung and colon cancers, the prevalence of gBRCA1m or gBRCA2m was generally less than 1%, and the prevalence of sBRCA1m and sBRCA2 ranged 1 to 3%. Seven publications were identified where HRD+ was defined by either BRCAm or genomic instability score (GIS) $ 42 across breast, ovarian and pancreatic cancers. The overall HRD+ prevalence was 56% (95%CI: 48, 64), with similar prevalence observed across the 3 tumor types and was 50% (34, 66) for the 3 studies only counting GIS$ 42. 194 studies across 26 tumor types were identified that examined HRRm as mutations in one or more HRR genes other than BRCA1/2m. The definitions of HRRm varied substantially across the studies, and ATM (2.8%), CHEK2 (1.6%), and PALB2 (1.6%) accounted for most of the observed mutations among HRR genes. Conclusions: Prevalence of BRCAm, HRRm and HRD+ varied by cancer type. This comprehensive meta-analysis enriches the knowledge in this field and demonstrates the need to standardize the measurement of HRRm and HRD.Understanding the prevalence of these biomarkers could have important clinical implications. Re- search Sponsor: Merck & Co.

10590 Poster Session

Distribution of BRCA1/2 germline and somatic alterations across cancer type.

Jingde Chen, Ming Quan, Zhengqing Yan, Shiqing Chen, Mei Huang, Jiujie Cui, Zhiqin Chen, Yong Gao; Shanghai East Hospital, Tongji University School of Medicine, Shanghai, China; The Medical Depart- ment, 3D Medicines Inc. Shanghai, P.R. China, Shanghai, China; The Medical Department, 3D Medi- cines Inc., Shanghai, China; Department of Oncology,Yancheng Third People’s Hospital/the Affiliated Yancheng Hospital of Southeast University Medical College,Jiangsu, Yancheng, China; State Key Labo- ratory of Oncogene and Related Genes, Shanghai Cancer Institute, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China; East Hospital, School of Medicine, Tongji University, Shanghai, China

Background: PARP inhibitors (e.g. Olaparib or niraparib) have been approved by FDA as a targeted therapy for many tumors harboring germline or somatic BRCA1/2 (g or sBRCA1/2), including ovarian cancer, prostate cancer, breast cancer and pancreases cancer. It is imperative to study the distribution of BRCA1/2 across cancer type. In this study, we aim to assess the landscape of BRCA1/2 alterations in solid tumors and evaluate the feasibility of circulating tumor DNA (ctDNA) tested by next-generation sequence (NGS) as a tool to detect BRCA1/2 alterations. Methods: For tissue specimen, genomic DNA from formalin fixed paraffin-embedded (FFPE) tumor specimens or fresh tumor tissues was used for sequence analysis. Genomic DNA (gDNA) from white blood cells was extracted using the QIAamp DNA Mini Kit (Qiagen). For ctDNA, cell-free DNA libraries were prepared using the KAPA Hyper Prep Kit following the manufacturer’s protocol. The captured libraries were loaded onto a NovaSeq 6000 platform (Illumina) for 100bp paired-end sequencing. The testing was performed in the College of American Pathologists (CAP) and Clinical Laboratory Improvement Amendments (CLIA) -certified 3D Medicines Library. Results: A total of 27, 835 patients were tested using tumor tissue during Jan. 1th 2017 to June 1th 2020, including 43% (N = 12089) of non-small cell lung cancer, 19% (N = 5357) of colorectal cancer, 8% (N = 2181) of liver cancer, 6% (N = 1621) of gastric cancer, 5% (N = 1479) of biliary tract cancer, 4% (N = 1084) of kidney cancer, 4% (N = 1045) of pancreas cancer, 3% (N = 689) of breast cancer and 2% (N = 599) of ovarian cancer. Across all tumor types, the known or likely deleterious BRCA1/2 alterations were identified in 2147 (7.7%) patients. Ovarian cancer had the highest frequency of BRCA1/2 alteration (23.4%), followed by endometrial cancer (12.7%) and breast cancer (10.6%). BRCA1/2 alteration was found in 8.8% prostate cancer and 4.2% pancreas cancer respectively. Across all tumor types, the known or likely deleterious gBRCA1/2 alterations were identified in 369 (1.3%) patients. Notably, ovarian cancer had the highest frequency of gBRCA1/2 alteration (13.9%), followed by breast cancer (7%), prostate cancer (4.4%) and endometrial cancer (4.1%). No clear hotspot mutations and mutated codons were spread throughout g or sBRCA1/2 mutations. Addi- tionally, among 15699 patients who suffered ctDNA sequencing, any known or likely deleterious sBRCA1/2 alterations were identified in 358 (2%) patients. Similar to the results of tissue sequencing, ovarian cancer had the highest frequency of sBRCA1/2 alteration (16.67%), followed by endometrial cancer (9.68%), prostate cancer (7.18%) and breast cancer (5.58%) in the blood cohort. Conclusions: BRCA1/2 alterations existed across tumor types and the landscape of g or sBRCA1/2 alterations varied according to cancer type. Furthermore, ctDNA can be used as a potential tool to detect BRCA1/2 alterations. Research Sponsor: None.

10591 Poster Session

Racial disparities in genetic testing of breast cancer patients.

Solange Bayard, Yalei Chen, Genevieve A. Fasano, Melissa Davis, Eleanor M. Walker, Jessica Miley Bensenhaver, Lindsay Petersen, Jennifer Marti, Rache M. Simmons, Alexander J. Swistel, Lisa A. Newman; Weill Cornell Medicine, New York, NY; Henry Ford Health System, Detroit, MI; Weill Cornell Medicine/New York Presbyterian, New York, NY; Henry Ford Hospital, Troy, MI; University of Michigan Health Syst Comp Cancer Ctr, Ann Arbor, MI; Department of Surgery, Weill Cornell Medicine, NewYork- Presbyterian Hospital, New York, NY; New York-Presbyterian Hospital/Weill Cornell Medicine, New York, NY; NewYork-Presbyterian Hospital/Weill Cornell Medical Center, New York, NY

Background: TNBC is disproportionately prevalent in African American (AA) populations and in women with BRCA-1 germline mutations. BRCA mutation carriers are candidates for targeted therapy with PARP-inhibitors, and testing results may influence risk-reducing surgery choice. Methods: We evaluated genetic testing patterns and outcomes for TNBC patients treated in the prospectively maintained databases of academic cancer programs in two metropolitan cities in the Northeast (New York City, NYC) and Midwest (Detroit, Det), 1998-2018. Median follow up was 3.73 years. Testing patterns were also analyzed by time, comparing pts diagnosed before versus after the mid-2013 Supreme Court ruling that expanded testing availability by banning gene patenting. Results: Of 810 pts, 600 were from NYC and 200 from Det; 202 were AA and 488 WA. Pts undergoing genetic testing were younger (median age 50 vs 62; p < 0.0001). Compared to WA, AA pts were less likely to undergo genetic testing overall (23.8% vs 42.0%; p < 0.0001) and within site (NYC: 25.6% vs 42.8%, p = 0.008; Det: 22.3% vs 38.6%, p = 0.025). No significant differences were seen in frequency of pathogenic BRCA mutations (AA-14.6%; WA-29.3%) or VUSs (AA-6.3%; WA- 4.9%); p = 0.20. Genetic testing disparities were reduced among pts diagnosed after mid-2013 (AA-31.4% vs WA-44.0%; p = 0.01) compared to pre-mid-2013 (AA- 18.3% vs WA-40.7%; p < 0.0001). No differences were seen in local or distant recurrence free survival between patients with BRCA, BRCA variants of uncertain significance, non-BRCA mutations, and patients without genetic mutations (local recurrence p = 0.827; distant recurrence p = 0.574). This outcome equivalence was consistent when stratified by WA vs AA identity. Conclusions: Genetic testing has increased for TNBC pts following the mid-2013 Supreme Court ban on gene patenting, but race-associated disparities persist. Pts undergoing genetic testing are more likely to undergo risk-reducing mastectomy, but testing results do not affect survival outcomes, regardless of race. Addressing genetic testing disparities will become increasingly important as mutation-associated targeted therapies are identified through advances in precision medicine. Research Sponsor: None.

10592 Poster Session

Oncologists’ (ONCs) perceptions of tumor genomic profiling (TGP) and barriers to communicating secondary hereditary risk to African American (AA) patients.

Michael J. Hall, Sarah Bauerle Bass, Yana Chertock, Jesse A Brajuha, Paul D’Avanzo; Fox Chase Cancer Center, Philadelphia, PA; Temple University, Philadelphia, PA; Temple University College of Public Health, Philadelphia, PA

Background: TGP identifies targets for precision cancer treatments. TGP may also identify secondary hereditary cancer risks, necessitating complex decision support during informed consent. ONCs are poorly trained in the communication of genetic information, particularly for patients with low health literacy, poor knowledge of genetics, and high medical mistrust. AA patients are especially vulnerable in this setting. Methods: We conducted semi-structured interviews with 10 ONCs to assess perceived barriers related to communication of secondary hereditary risks of TGP, probing barriers unique to AA patients. Informed by results, an Internet-based survey was developed/distributed to a convenience sample of 50 ONCs nationwide to assess TGP knowledge, genomics confidence, and perceptions related to communication of secondary hereditary risk. Results: Six themes emerged from interviews: risk/benefits of TGP, knowledge of genetics, discussing hereditary risk, value/harm of TGP, unique risks in AA, and training needs. Most ONCs felt uncomfortable discussing hereditary risks of TGP w/patients. Seven out of 10 identified socio-economic status, medical mistrust, discrimination, genetic counseling non-compliance, low health literacy and family relationships as factors important to consider with AA patients. On- line survey participants were 52% White, 66% male, with median age of 42 years. Education in the interpretation/communication of TGP was largely informal (56% reported only informal training) and 46% reported perceived gaps in their education. Genomic confidence was associated w/higher use of TGP (p = 0.05), but was not associated w/knowledge or years in practice; however, low knowledge was associated w/more perceived barriers to TGP and w/negative attitudes toward the value of TGP and the challenge of communication of possible hereditary risks (p = 0.05). Early-career ONCs were more likely to endorse perceived barriers to communication of genetic risk information from TGP to AA patients. Overall 86% ONCs felt additional online training in communication of secondary hereditary risks of TGP would be useful. Conclusions: ONCs recognize unique needs and barriers for AAs related to communication of secondary hereditary genetic information from TGP. Many feel uncertain about how/whether to address barriers and recognize the need to improve their skillset to do so. Training is critical to ensure informed decision making in vulnerable populations. Research Sponsor: American Cancer Society.

10593 Poster Session

Evaluation of proband adherence and satisfaction with a prospective cascade testing protocol.

Maria Smith, Kristina Hwang, Julia Anne Smith, Bhavana Pothuri; NYU Grossman School of Medicine, Brooklyn, NY; NYU Perlmutter Cancer Center, New York, NY; New York University Langone Medical Center, New York, NY; New York University School of Medicine, New York, NY

Background: We sought to evaluate the feasibility of a Cascade Testing (CT) protocol for family members of probands with actionable germline mutations associated with endometrial or ovarian cancer. Here, we characterize proband compliance with contacting family members for CT and proband satisfaction/ regret. Methods: In this prospective study, consenting patients with pathogenic germline mutations associated ovarian or endometrial cancer completed a demographic survey and were asked to contact first- and second-degree relatives with genetic testing results. After a 1–3-month period, probands completed a survey indicating how many relatives had been contacted. At 3 months following consent, probands were asked to complete the validated Impact of Event Scale (IES) and Decision Regret Scales (DRS). Characteristics of probands who contacted relatives and those who did not were compared. Results: The study has accrued 57 probands since opening in March 2019. Germline mutations identified in the 57 probands include 27 BRCA1 (47.4%); 21 BRCA2 (36.8%); 3 BRIP1 (5.3%); 2 MLH1 (3.5%); 2 MSH2 (3.5%); 3 MSH6 (5.3%); 3 PMS2 (5.3%); 1 EPCAM (1.8%); 1 RAD50 (1.8%). Twenty-four (42.1%) probands had a history of cancer (breast 12; ovarian 8; uterine 2; other 5). Of the probands, 32 (56.1%) completed follow-up questionnaires and 29 (50.9%) had contacted relatives about participating in CT. In total, 67 relatives were contacted. Probands contacted an average of 1 relative, ranging from 1-20. Of the 29 probands who contacted relatives, 13 (44.8%) completed IES and DRS questionnaires. The median IES score was 0 out of 75 (IQR 0.0-4.5) and the median DRS score was 0 out of 100 (IQR 0.0-11.3). When comparing characteristics of probands who contacted relatives with those who did not, those with annual household incomes <$75,000 were more likely to contact relatives vs those with incomes $$75,000 (77.8% vs 39.5%; p=0.01). There was no association between contacting relatives and personal cancer history, race/ethnicity, education status, or age (Table). Conclusions: Half of probands enrolled in this study contacted relatives about CT, and those with household incomes <$75,000 were more likely to contact relatives than those with higher incomes. Overall, probands reported little/no regret or distress after contacting relatives about genetic testing results. Research Spon- sor: None.

Contacted (N=29) Did not contact (N=28) P-Value Personal Cancer History—N (%) 14 (58.3) 10 (41.7) 0.42 Race/Ethnicity—N (%) 1.00 Non-Hispanic White 20 (50.0) 20 (50.0) All Other 9 (52.9) 8 (47.1) Education—N (%) 1.00 College degree or higher 26 (50.0) 26 (50.0) No college degree 3 (60.0) 2 (40.0) Household income—N (%) 0.01 <$75,000/year 14 (77.8) 4 (22.2) $$75,000/year 15 (39.5) 23 (60.5) Age —median (IQR) 44.0 (34.0-52.0) 40.0 (27.5-49.0) 0.65

10594 Poster Session

Multicancer hereditary syndrome testing: Genetic counselors’ perspectives.

Christine B. Weldon, Su-Ying Liang, Kathryn A Phillips, Michael P Douglas, Maren Theresa Scheuner, Allison W. Kurian, Kendra Schaa, Breanna Roscow, Deanna Erwin, Julia Rachel Trosman; Northwestern University Feinberg School of Medicine, Chicago, IL; Sutter Health-Palo Alto Medical Foundation Re- search Institute, Palo Alto, CA; Center for Translational and Policy Research on Personalized Medicine (TRANSPERS), University of California, San Francisco, San Francisco, CA; UC San Francisco, San Francisco, CA; University of California, San Francisco, San Francisco, CA; Stanford School of Medicine, Stanford, CA; University of Iowa, Iowa City, IA; Myriad Genetics Inc, Salt Lake City, UT; City of Hope, Los Angeles, CA; Center for Business Models in Healthcare, Chicago, IL

Background: The accessibility of cancer hereditary syndrome testing has increased, and the cost has declined significantly in the past few years. We conducted a national, quantitative survey of genetic counselors (GCs) to assess their perspectives on what influences hereditary cancer genetic testing decisions and practices, with a focus on cost. This survey was funded by NIH, conducted by UCSF TRANSPERS, and supported by the National Society of Genetic Counselors (NSGC). Methods: The survey was developed through literature review, expert interviews, and a pilot. Sent to the NSGC Cancer Special Interest Group via email. Chi-square tests were used to examine variability. Results: The survey response rate was 56% (202/363). Multiple hereditary cancer syndrome tests are discussed often/always by 86% of genetic counselors (GCs). The existence of an institutional protocol on multiple hereditary cancer syndrome testing was reported by 35.4% of GCs. When asked about GC counseling encounters, GCs report insurance rarely/never pays for: 25.2% pre-test in-person, 39.7% for pre-test tele-genetics, 35.4% post-test in-person, and 52.9% post-test tele-genetics. GCs rated clinical factors higher than cost as influencing decision for multiple hereditary syndrome cancer testing (table); the total cost of the test was least important. These patterns were similar across the GCs institution types and years in practice. Conclusions: We found consistent use of multiple hereditary cancer syndrome tests, with less focus on cost, out-of-pocket, and insurance coverage and more of a focus on clinical indicators. GCs reported challenges with reimbursement for GC counseling encounters. The shift toward more genetic counseling encounters via tele-genetics necessitates evaluation of insurance reimbursement. Research Spon- sor: U.S. National Institutes of Health.

Moderately / Very How important are the following factors in your decision to discuss and possibly Important n = 202 order multiple hereditary cancer syndrome testing? %

Limited family structure, unknown family history, and/or patient adopted without 81 biological relative history knowledge Patient’s request and or question about multi-gene syndrome testing 81 Personal and family history reveal only one major cancer type 56 Single syndrome test is more clinically appropriate for patient 53 Concern about increased likelihood of variants of uncertain/unknown significance 51 Concern about unexpected pathogenic/likely pathogenic variant in a gene not associated 49 with phenotype Expectation that patient may be willing to pay for test if insurance doesn’t cover it 47 Standards or protocol of your clinic, organization, institution 44 Concern that patient share of payment / out of pocket is too much 43 Concern that patient’s insurance may not cover multiple-syndrome testing 41 Patient with no phenotype based on personal or known family history 38 Concern that total cost of test (cost to insurance/payer plus co-pay) is high 32

10595 Poster Session

Prevalence and spectrum of pathogenic variants among patients with multiple primary cancers.

Brittany L. Bychkovsky, Min-Tzu Lo, Yuan Tian, Amal Yussuf, Carrie Horton, Marcy Richardson, Holly LaDuca, Judy Ellen Garber, Huma Q. Rana; Cancer Genetics and Prevention, Dana-Farber Cancer Insti- tute, Harvard Medical School, Boston, MA; Ambry Genetics, Aliso Viejo, CA; Ambry Genetics, Alisa Vie- jo, CA

Background: Multiple primary cancers (MPCs) are a hallmark of cancer predisposition syndromes. We aim to characterize the frequency of germline pathogenic/likely pathogenic variants (PVs) among patients with MPCs. Herein we report the frequency of PVs by sex, number of cancers and age at diagnosis among a laboratory-based cohort of patients with MPCs. Methods: Patients with MPCs who underwent germline genetic testing with Ambry Genetics from 3/2012 to 12/2016 were included in our cohort. Eli- gible individuals had multigene panel testing, which included 21 genes, at minimum: ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, EPCAM, MLH1, MSH2, MSH6, MUTYH, NBN, NF1, PALB2, PMS2,PTEN, RAD51C, RAD51D, STK11, and TP53. Clinical factors including age at diagnosis, age at testing and cancer type were obtained from test requisition forms and clinical notes. Patients with > 1 PVs were excluded from the analysis. Using Rv.3.3.3., the frequencies of PVs by sex, number of cancers and age at diagnosis were compared using two-sided v2 tests or Fisher’s exact test when the number was < 10. Results: Of the 9820 patients with MPCs tested for the 21 genes above, 104 (1.1%) had multiple PVs and were excluded. Among the remaining 9716 patients in the analytic cohort, most were female (91.1%) and white (71.0%). The median age at testing was 63 years (IQR: 16) and the median ages of first and second cancer diagnosis were 49 (IQR: 18) and 58 (IQR: 17) years, respectively. Over- all, 1406 (14.5%) were found to have PVs: 14.3% of females and 16.2% of males. The prevalence of PVs increased with the number of primary cancers (PCs) as follows, 2 PCs: 13.1% (95% CI:12.4- 13.8%), 3 PCs: 15.9% (95% CI:14.0-18.0%), >4 PCs: 18.0% (95% CI:13.7-23.3%), (p < 0.01). Among patients with 2 PCs (n = 8145), differences in the prevalence of PVs by age at diagnosis were significant: 2 PCs diagnosed at an age < 50 (13.5%, 95% CI:12.0-15.1%), 1 PC diagnosed at an age < 50 (14.8%, 95% CI:13.4-16.5%), 2 PCs diagnosed at age >50 years (12.1%, 95% CI: 11.1- 13.2%), (p = 0.01). PVs were most frequently identified in: BRCA2 (2.2%) BRCA1 (2.0%), CHEK2 (1.9%) and ATM (1.5%). There were also significant differences in the frequencies of PVs in BRCA1, BRCA2 and MLH1 by sex (p < 0.05). Conclusions: These data demonstrate a high frequency of germline PVs among both males and females with MPC. The frequency of PVs was higher among patients with a higher number of PCs. Differences in the prevalence of PVs by age at cancer diagnosis while significant, were not meaningful as 12.1% of individuals with 2 PCs diagnosed at age >50 years had germline PVs. Limitations include the homogenous testing population (predominately female and white) and small numbers in some patient categories. These data may aid in counseling patients with MPCs and their families as well as encourage less restrictive genetic testing of this population. Further analysis of PV frequencies by specific cancer combinations was conducted and will follow. Research Sponsor: None.

10596 Poster Session

Identification of potential germline (GL) variants by routine clinical comprehensive genomic profiling (CGP) and confirmatory GL testing in 24 tumor types.

Kristen Hanson, Michael Mullane, Erica Schnettler, Dean C. Pavlick, Garrett M. Frampton, Maureen Cooper, Deborah Wham, Michael A. Thompson, Jennifer Godden, Mary Walters, Jeffrey Michael Venstrom, Kimberly McGregor; Advocate Aurora Health, Milwaukee, WI; Foundation Medicine, Inc, Cambridge, MA; Foundation Medicine, Inc., Cambridge, MA; Foundation Medicine, Cambridge, MA; Aurora Cancer Care, Advocate Aurora Health, Milwaukee, WI

Background: Tumor CGP may identify both somatic and GL variants, though confirmatory testing is required to verify which variants originate from the GL. Studies have shown CGP can identify patients who both do and do not meet criteria for genetic counseling (GC). Ideally, improved annotation from tumor CGP could more appropriately direct GC referrals. We explore how a computational algorithm might be used to influence GC and confirmatory GL testing for variants in inherited cancer predisposition genes. Methods: 849 patients from the Aurora Oncology Precision Medicine Program who had routine hybrid- capture based CGP by Foundation Medicine from 8/2018-8/2020 were eligible. A previously published algorithm, SGZ (Sun et al PMID 29415044) which incorporates allele frequency, aneuploidy, and ad- mixed copy number modeling was used to predict whether each single nucleotide variant (SNV) was GL or somatic. SGZ predictions for SNVs in 24 actionable inherited cancer predisposition genes were available to Aurora for review as part of standard screening to identify appropriate GC referrals. For patients who had GL testing, variants in genes on both assays were compared. Results: 76 pathogenic (P) or likely pathogenic (LP) variants predicted to be GL by SGZ were detected in 73/849 (9%) patients: ATM (7), BAP1 (2), BRCA1 (13), BRCA2 (8), BRIP1 (1), CHEK2 (18), FH (0), FLCN (2), MLH1 (1), MSH2 (0), MSH6 (3), MUTYH (12), PALB2 (3), PMS2 (1), POLE (0), RAD51C (1), RAD51D (0), RET (1), SDHA/ B/C/D (0,0,0,0), TSC2 (0), and VHL (3). 27/73 (37%) patients had GL testing. 25/26 (96%) variants were confirmed to be GL in origin and 1 additional variant was detected by CGP in a region not interrogated by the GL assay: ATM (2/2), BRCA1 (6/6), BRCA2 (2/2), BRIP1 (1/1), CHEK2 (9/9), FLCN (0*/ 1), MSH6 (1/1), MUTYH (2/2), PALB2 (1/1), RAD51C (1/1), and VHL (0/1). Variants were confirmed in bladder, breast, CRC, glioma, NSCLC, ovary, pancreas, prostate, sarcoma, and gastric cancer. The VHL variant was discordant in a leiomyosarcoma. Conclusions: We identified the potential real-world clinical impact of computationally screening solid tumor patients undergoing routine CGP for potential P/LP GL variants. Predicting GL results with SGZ for 24 inherited cancer predisposition genes was highly concordant with confirmatory GL testing independent of tumor type. CGP annotations can facilitate GC referral and GL testing for at-risk patients, particularly in tumor types which may not typically meet guidelines for GL testing. Research Sponsor: None.

N=73 patients harboring P/LP variants predicted to be GL by SGZ.

GL SGZ prediction GL testing Confirmed GL

ATM 7 2 2 BAP1 2 0 0 BRCA1 13 6 6 BRCA2 8 2 2 BRIP1 1 1 1 CHEK2 18 9 9 FH 0 0 0 FLCN 2 1 0* MLH1 1 0 0 MSH2 0 0 0 MSH6 3 1 1 MUTYH 12 2 2 PALB2 3 1 1 PMS2 1 0 0 POLE 0 0 0 RAD51C/D 1/0 1/0 1/0 RET 1 0 0 SDHA/B/C/D 0/0/0/0 0/0/0/0 0/0/0/0 TSC2 0 0 0 VHL 3 1 0

*Gene was not interrogated by GL assay.

10597 Poster Session

Prevalence of germline testing criteria in breast cancer patients in the Brazilian public health system: A retrospective study.

Robson dos Santos dos Santos Borges, Laisa Gabrielle Silva, Mariana do Nascimento Vilaca, Alexandre Ribas De Carvalho, Joa~o Paulo Solar Vasconcelos, Luiz Flavio Penna Coutinho, Ricardo Cembranelli Teixeira, Helena Flavia Cuba de Almada Lima, Octavio de Castro Menezes Candido, Arilto Eleuterio da Silva Junior, Carolina da Silva Cardoso, Isabella Rocha Franc¸a Staino, Anita Chavez Azevedo, Maria Helena Cruz Rangel Silva, Bruno Lemos Ferrari, Otto Metzger-Filho; Grupo Oncoclini- cas, Hospital Felicio Rocho, Hospital Alberto Cavalcanti, Belo Horizonte, Brazil; Hospital Alberto Caval- canti, Belo Horizonte, Brazil; Hospital Felicio Rocho, Belo Horizonte, Brazil; Hospital Felicio Rocho, Grupo Oncoclinicas, Hospital Alberto Cavalcanti, Belo Horizonte, Brazil; Grupo Oncoclinicas, Belo Hori- zonte, Brazil; Hospital Madre Teresa, Grupo Oncoclinicas, Hospital Felicio Rocho, Belo Horizonte, Bra- zil; Grupo Oncoclinicas, Brasilia, Brazil; Hospital Fel�ıcio Rocho, Belo Horizonte, Brazil; Grupo Oncoclinicas, Hospital Felicio Rocho, Belo Horizonte, Brazil; Dana-Farber Cancer Institute, Boston, MA

Background: Identification of a germline mutation in a breast cancer predisposition gene has implications for the patients and their families. The National Comprehensive Cancer Network (NCCN) has published guidelines for genetic testing. In Brazil, this assessment is covered by health insurance in accordance with criteria defined by the National Supplementary Health Agency (ANS). For the majority of the population, served by the public health system (SUS), the assessment is not routinely available. Methods: In order to determine the prevalence rates of NCCN and ANS criteria for germline testing in breast cancer (primary outcome) we retrospectively analyzed data from patients treated at two SUS oncology centers in Belo Horizonte, Minas Gerais, Brazil, between 01/01/18 and 12/31/19. The secondary outcomes were comparisons between the groups with and without germline testing criteria (NCCN and ANS) regarding overall survival, clinical and epidemiological characteristics. The association between qualitative variables was calculated using the Chi-square and Fisher tests. The Kaplan-Meier method was used to analyse the survival data and the differences between the groups were tested using the log-rank test. The level of significance was 5%. Results: A total of 357 patients were included in the final analysis. The presence of germline testing criteria were found in 126 patients (35%) according to NCCN guidelines and in 82 patients (23%) according to ANS guidelines. None of them were tested for germline mutations. The most common criteria were women up to 60 years old with triple negative tumors (n = 43, 12% of all patients) and diagnosis of cancer up to 45 years old (n = 75, 21% of all patients) according to ANS and NCCN criteria, respectively. When the group of patients who met at least one criterion for germline testing were compared with the group who did not met any criteria, we found in the first group: more ductal carcinomas and less lobular tumors (p = 0.009), more grade 3 tumors (p = 0.002), more triple negative tumors (p < 0.001), more neoadjuvant treatments (p = 0.008) and less hormonal therapies (p = 0.011). After a median follow up of 13.5 months there were 22 deaths in the cohort, 7 in the group with testing criteria (5.7%) and 15 in the group without testing criteria (6.4%). There was no statistical significant difference between the groups in terms of overall survival (p = 0.77). Conclusions: To our knowledge this is the first study to evaluate the prevalence of NCCN and ANS criteria for germline testing in patients with breast cancer treated in the Brazilian public health system. Our results show that more than a third of those patients are candidates for germline testing. Moreover, the data highlight a serious shortcoming in the management of breast cancer and must be considered in the development of public health policies for routine germline testing in that population. Research Sponsor: None.

10598 Poster Session

Impact of race on biomarker testing among HER2- advanced breast cancer (ABC) patients (pts) in the United States: Results from a real-world study.

Reshma L. Mahtani, Alexander Niyazov, Katie Lewis, Lucy Massey, Alex Rider, Bhakti Arondekar, Michael P Lux; Sylvester Cancer Center, University of Miami, Deerfield Beach, FL; Pfizer Inc., New York, NY; Adelphi Real World, Bollington, United Kingdom; Pfizer Inc., Collegeville, PA; Kooperatives Brustzentrum Paderborn, Frauenklinik St. Louise, Paderborn, St. Josefs-Krankenhaus, Salzkotten, Frauen- und Kinderklinik St. Louise, Paderborn, Germany

Background: African Americans (AA) have the highest breast cancer (BC) mortality rate. Access to treatment is a known contributing factor. In the past 4 years, several targeted therapies for HER2- BC have become available which require testing for specific biomarkers. This study assessed the impact of race on biomarker testing rates in HER2- ABC pts receiving treatment in the US. Methods: Oncologists were recruited to abstract data from medical charts for the next 8-10 pts receiving treatment with HER2- ABC during Sept 2019-Apr 2020. Pts records were stratified by race and categorized into 3 mutually exclusive cohorts [White/Caucasian (White), AA, Other]. The other race cohort was excluded from this analysis due to small sample size. Differences in pt demographics/clinical characteristics were analyzed via Fisher’s exact tests. Testing rates for actionable biomarkers (i.e. BRCA1/2, PIK3CA, PD-L1) were compared between White and AA pts utilizing logistic regressions controlling for age, known family history of a BRCA-related cancer, hormone receptor (HR) status and practice setting (academic vs. community). Further analyses by age will be presented. Results: This analysis included 378 pts records, provided by 40 oncologists. Mean age was 64 years; 77% had HR+/HER2- ABC; 20% had advanced triple negative breast cancer (TNBC), 3% had ABC with an unknown HR status. Compared to White pts, AA pts were significantly more likely to have advanced TNBC (27% vs. 18%, p<0.05). Compared to White pts, AA pts had significantly lower BRCA1/2 mutation (mut) testing rates (Table). Numerically lower rates of PIK3CAmut and PD-L1 testing were observed among AA pts (Table). BRCA1/2mut positivity rate (germline [g] and/or somatic [s]) was higher among AA vs. White pts (30% vs. 22%). Positivity rate for PIK3CAmut was lower for AA vs. White pts (8% vs. 11%). Conclusions: A higher than expected BRCA1/2mut positivity rate was observed than previously reported in the literature. This is likely because this analysis included sBRCA1/2mut and represented a high risk pt population. Across all biomarkers assessed, AA pts had lower testing rates than White pts. This suggests racial disparities in testing rates of actionable biomarkers. Consistent with guidelines, and with the increased availability of targeted therapies, focused efforts should be developed to increase biomarker testing in AA pts. Fund- ing: Pfizer Biomarker Testing Rates by Race. Research Sponsor: Pfizer.

Whitea n=231 AA n=88 ORb (95% CI) P-value All BRCA1/2mut tested, % 79 66 0.44 (0.24-0.81) <0.01 sBRCA1/2mut tested only 21 23 1.10 (0.60-2.01) 0.77 g +/- s BRCA1/2mut tested 51 37 0.56 (0.33-0.94) 0.03 Unknown BRCA1/2mut tested 8 6 0.71 (0.25-2.03) 0.52 PIK3CAmut tested, % 50 44 0.76 (0.44-1.29) 0.31 PD-L1 tested, % 59 54 0.80 (0.46-1.39) 0.43 aReference value (ref) bCompared to ref, <1 is lower testing rates, 1 = the same testing rates, >1 is higher testing rates.

10599 Poster Session

Compliance with germline testing in pancreatic cancer in a rural tertiary care hospital.

Catherine Travaline, Leighton Andrew Elliott, Nadia N. Ramdin, Joseph Vadakara; Geisinger Medical Center, Danville, PA

Background: Pancreatic cancer is the 7th most common cause of cancer death worldwide and is projected to be the second leading cause of cancer death in the next decade. Personalized care is becom- ing more of a reality with pharmacological regimens targeting specific genetic mutations. In March 2019, the National Comprehensive Cancer Network (NCCN) guidelines were updated to recommend germline testing (GT) in all patients with pancreatic adenocarcinoma (PDAC) considering 1 in 10 may have a germline mutation (GM). The goal of this study was to quantify compliance with these recently updated guidelines. Methods: The electronic medical records and survivorship data of all patients diagnosed with PDAC between January 1, 2017 and October 1, 2020 were reviewed. April 1, 2019 was used as the transition point (TP) for guideline updates. Descriptive statistics for all variables were determined. The rate of ordered referrals to genetic counseling (GC), as well as completion rate, was calculated. Results: A total of 304 patients were diagnosed with PDAC during the study period (223 prior to the TP). A total of 54 patients were referred for GC and 41 had GT ordered. The rate of GC referrals ordered after the TP was significantly higher than before the TP (22/81, 26.6% vs. 32/223, 14.4%; p-value 0.010). Almost 60% of patients who had genetic evaluation had private insurance. The patients who completed GT were significantly more likely to have a documented family history of cancer (61.0% vs 4.2%; p-value <.0001Patients who completed GT had more problems on their problem list (median 10 vs 7, p = 0.001). The median overall survival (OS) for all patients in the study was 7.8 months (95% CI: 6.3-9.8). Conclusions: Overall compliance with the updated NCCN guidelines significantly improved; however, it was below 25%. This study showed that there may be some lingering bias toward GT in PDAC solely for those who have a family history of cancer. Although patients with stage IV PDAC have poor outcomes, GT may still improve surveillance for family members. The approval of olaparib in patients with BRCA1/2 mutations based on the POLO trial is likely to increase provider compliance as it provides a viable maintenance strategy in these patients. Patient complexity was unlikely to affect GT rate. Assessment of provider awareness was outside the scope of this study. There is need for continued advocacy for awareness and implementation of guidelines that highlight the importance of germline evaluation on prevention, surveillance, and treatment in pancreatic adenocarcinoma. Research Spon- sor: None.

10600 Poster Session

Evaluation of cohort diversity in development and validation studies of hereditary cancer genetic risk assessment models.

Amanda Gammon, Ambreen Khan, Joanne M. Jeter; Huntsman Cancer Institute, Salt Lake City, UT; 2000 Circle of Hope Drive, Salt Lake City, UT; The Ohio State University, Columbus, OH

Background: Multiple models estimate a person’s chance of harboring a pathogenic variant increasing cancer risk. Some pathogenic variants are more common in individuals from specific ancestries, such as the BRCA1 and BRCA2 founder variants in Ashkenazi Jews. Yet data remains limited on the larger variant spectrum seen among people of different ancestral backgrounds and whether or not the pathogenic variant frequency differs in many populations. Due to this, it is important that genetic risk assessment models be validated in a diverse cohort including Black, Indigenous, People of Color (BIPOC). Methods: A literature search was conducted to identify published development and validation studies for the following genetic risk assessment models: BRCAPRO, MMRPRO, CanRisk/BOADICEA, Tyrer-Cu- zick, and PREMM. Validation studies that only evaluated the cancer risk prediction capabilities of the models (and not the genetic variant risk prediction) were excluded. The following participant information was abstracted from each study: total number of participants, gender, race, and ethnicity. Authors were contacted to obtain missing information (if available). Results: 12 development and 12 validation studies of the genetic risk assessment models BRCAPRO, MMRPRO, CanRisk/BOADICEA, Tyrer-Cu- zick, and PREMM were abstracted. Of the validation studies, five were internal validation studies conducted by the model developers, and seven were external validation studies. Four external validation studies compared multiple models. 75% (18/24) of papers did not include reporting of participant race or ethnicity information in their published reports. External validation studies (4/7, 57%) more often reported participant race/ethnicity than development (0/12, 0%) or internal validation (2/5, 40%) studies. The external validation studies for BRCAPRO reporting race/ethnicity information involved cohorts that ranged from 50-51% non-Ashkenazi Jewish white, 28% African American, 1% Asian, 2-49% His- panic, and 19-42% Ashkenazi Jewish. The external validation studies for MMRPRO and PREMM reporting race/ethnicity information involved cohort that ranged from 0-82% white, 4-100% Asian, 7% Black, and 7% Hispanic. Conclusions: Increased reporting of participant ancestry and ethnicity is needed in the development and validation studies of genetic risk assessment models. BRCAPRO’s validation cohorts have included a higher percentage of Hispanic and Black/African American participants, while MMRPRO and PREMM have been validated in a higher percentage of Asian participants. As debate con- tinues about the utility of currently used racial categories in genetics research, it will be important to determine how best to report on participant diversity. These findings highlight the continued need for genetics researchers to engage BIPOC and identify ways to diversify their participant cohorts. Research Sponsor: None.

10601 Poster Session

Assessing somatic and germline variants in cancer patients.

Charite� Ricker, Erika Amundson, Sandra Algaze, Marcia Ciccone, Stephen Dong, Anishka D’souza, Kimberly Felicetti, Daisy Hernandez, Priya Jayachandran, Irene Kang, Kavitha Narayanan, Jacob Stephen Thomas, Varsha Tulpule, Rebecca Umayam, Bing Xia, Gino Kim In; Division of Oncolo- gy, USC Keck School of Medicine, Norris Comprehensive Cancer Center, Los Angeles, CA; Keck School of Medicine, University of Southern California, Los Angeles, CA; USC Keck School of Medicine, Los An- geles, CA; Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Southern California, Los Angeles, CA; LAC+USC Medical Center, Los Angeles, CA; USC Norris Compre- hensive Cancer Center, Los Angeles, CA

Background: The increasing integration of somatic and germline testing into oncology practice allows physicians to target oncologic therapy and identify those with cancer predisposition. We explored the impact of a somatic assay (liquid biopsy, LB) on the identification of patients appropriate for germline genetic testing. Methods: We identified a cohort of diverse cancer patients with LB to assess for targetable somatic gene variants at LAC+USC Medical Center between 2016 and 2020 (n= 467). To enrich the cohort for variants that may reflect germline findings, we focused on the 46 patients (9.9%) who had at least one variant identified with a cell-free DNA (cfDNA) fraction of 25.00% or greater. Retro- spective chart review extracted demographics and medical history with variables related to cancer history and treatment. LB variants were classified based on whether germline confirmation was indicated and the results of germline tests, when done, were reviewed. Results: Table summarizes the characteristics of the 46 patients identified to have at least one variant on LB in # 25% of the cfDNA. The most frequently mutated genes on LB were TP53 (n=18, 39%), KRAS (n=11, 24%), APC (n=8, 17%), BRCA2 (n=7, 15%), PIK3CA (n=6, 13.0%), and BRCA1 (n=4, 9%). Seventeen patients (40%) were referred for genetic counseling and 13 (30%) underwent germline testing of whom 10 (77%) carried pathogenic variants (PV). All germline PV were concordant with LB variants identified. Four patients with PV BRCA2 on LB and confirmed to be germline, had lung or biliary tract diagnoses, which are not part of the typical BRCA-tumor spectrum. Thirty-three patients were not referred for genetic counseling, though 24 (72%) had LB-identified variants in cancer predisposition genes and 18 (54%) merited a genetics referral. Among patients with germline mutations, three (23%) had targeted therapy and two (15%) had preventive surgery to address second primary cancer risk. Among the 467 patients with LB results, there were an additional 13 patients (not included in the enriched group) known to have a cancer predisposition gene PV. Only two (15%) had findings on LB reports that suggested germline testing would be indicated. Conclusions: While the purpose of somatic testing is to identify targeted therapy, it can provide germline insight, especially for patients not typically referred for genetic assessment. Fur- ther education and guidance is needed to assure that this aspect of somatic testing is appreciated by oncology providers and acted upon. Research Sponsor: None.

Age Median (range) 55 (21-80) Sex - no. (%) Male 19 (41.3%) Female 27 (58.7%) Race/Ethnicity - no. (%) Hispanic/Latino 29 (63.0%) Asian 9 (19.6%) White 6 (13.0%) Black 2 (4.3%) Cancer Diagnoses: Lung/Thoracic Gastrointestinal 14 (30.4%) Gynecological 14 (30.4%) Breast 6 (13.0%) Skin/Soft Tissue 4 (8.7%) Other 4 (8.7%) Highest Cancer Stage: Limited 12 (26.1%) Advanced 34 (73.9%)

10602 Poster Session

Circulating tumor DNA (ctDNA) as a tool to help guide germline testing in patients with solid malignancies.

Nikita Pankaj Patel, Ami N. Shah, Andrew A. Davis, Lorenzo Gerattana, Saya Jacob, Neelima Katam, Firas Hazem Wehbe, Amir Behdad, Theresa Sciaraffa, Massimo Cristofanilli; Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL; Northwestern University, Chicago, IL; Department of Medicine, Division of Oncology, Washington University School of Medicine, St. Louis, MO; Division of Hematology and Oncology, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL; Northwestern University Feinberg School of Medicine, Chicago, IL; Department of Pathology, Northwestern University Feinberg School of Medicine, Chicago, IL; North- western, Chicago, IL; RH Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL

Background: As the use of circulating tumor DNA (ctDNA) is more widely implemented, incidental identification of pathogenic variants reflecting germline alterations in cancer predisposition genes will occur more frequently. Such mutations are expected to have a high mutant allele frequency (MAF) or occur in genes typically associated with inherited syndromes. When a similar analysis was conducted by our group, we found that MAF of about 30% or greater in BRCA1/2 was associated with confirmed putative germline mutations in patients with breast cancer (Jacob et al & Davis et al, SABCS 2020). In this study, we extended this analysis to non-breast malignancies. Methods: Patients with non-breast solid malignancies and ctDNA testing between 2015-2020 were retrospectively identified from Northwest- ern Medicine. All ctDNA was analyzed using Guardant 360 (Guardant Health, Inc. Redwood City, Ca). Patients with ctDNA samples with mutations at high MAF (>30%) and those with BRCA1/2 mutations at any MAF were identified. We reviewed these charts for referral to genetic counselors and/or CLIA-approved germline testing. Descriptive analysis was reported for these findings. Genetic alterations were classified as pathogenic or of unknown significance based on OncoKB (Chakravarty et al, JCO PO 2017). Results: We identified ctDNA samples of 548 patients with non-breast solid malignancies, of whom 56 had gene mutations occurring at high MAF (>30%). Predominant cancer subtypes were lung (48%), colorectal (21%), pancreatic (7%), ovarian (3.5%), prostate (3.5%), and gastroesophageal (3.5%). The most common gene mutations identified were TP53 (46%), BRCA1/2 (18%), EGFR (18%), APC (13%), and KRAS (9%). 87.5% were pathogenic and 12.5% were of unknown significance. 11 patients (19.6%) had germline testing of whom 6 tested positive. These germline mutations were in BRCA2 (n = 3), EGFR, APC, and TP53. In addition to the 10 patients with BRCA1/2 mutations at high MAF (>30%), we identified 70 patients with BRCA1/2 mutations at low MAF ( < 30%). 54% were pathogenic and 46% were of unknown significance. 11 patients (14%) had germline testing of whom 3 tested positive for BRCA2, all at high MAF. 1 patient with a BRCA2 mutation at low MAF of 1.4% tested positive for a different germline BRCA2 variant. Conclusions: In patients with advanced cancers, ctDNA analysis can reveal variants with MAF >30% that are reflective of a germline mutation. Unfortunately the rate of genetic testing in these patients was low (20%). Future studies with germline testing in patients with high MAF variants would help understand the prevalence of germline variants. This can facilitate developing a more standardized approach for genetic counselor referral to identify families that may benefit from interventions for early detection or prevention of future cancers. Re- search Sponsor: None.

10603 Poster Session

Prospective genomic testing of unselected cancer patients yields insights about cancer susceptibility and noncancer disease with therapeutic implications.

Stacy W. Gray, Kevin McDonnell, MD, PhD, Gregory Idos, Christine Hong, Ilana Solomon, Duveen Sturgeon, Joseph D. Bonner, Sidney A. Smith, Jennifer Morales Pichardo, Melissa Woodhouse, Elise Sobotka, Elyssa Zukin, Samir Courdy, Elisabeth King, Xiaoyu Xia, Szabolcs Szelinger, Janine LoBello, Stanley R. Hamilton, Jeffrey M. Trent, Stephen B. Gruber; City of Hope, Duarte, CA; City of Hope Na- tional Medical Center, Duarte, CA; University of Southern California Norris Comprehensive Cancer Cen- ter, Los Angeles, CA; Brogent International, LLC, New York, NY; City of Hope Comprehensive Cancer Center, Duarte, CA; City of Hope National Cancer Center, Duarte, CA; Ashion Analytics, Phoenix, AZ; The University of Texas MD Anderson Cancer Center, Houston, TX; Translational Genomics Research In- stitute, Phoenix, AZ

Background: Clinicians have used strict criteria to determine eligibility for cancer susceptibility (CS) testing and have limited genetic assessment to cancer-related genes. However, half of all CS mutation carriers are missed by criteria-based testing and there may be an unrecognized opportunity to modify care for patients who have rare but actionable genetic disorders as defined by the American College of Medical Genetics (ACMG). With the aim of improving patient outcomes through precision genomics, we initiated an enterprise-wide program to offer somatic and germline sequencing to all patients. Methods: We offer consented patients clinical grade paired somatic & germline WES/ RNA seq and panel germline testing for cancer (156 genes) and ACMG disorders (59 genes). Results are reviewed by a Precision Oncology Tumor Board. Somatic results are returned by the treating team. Germline results are returned by phone (genetic counselor, GC) followed by a clinic visit (GC+MD) for those with pathogenic/likely pathogenic (P/LP) mutations and selected variants of uncertain significant. We evaluated the propor- tions of patients with somatic findings suggestive of germline conditions and those carrying P/LP mutations in CS and ACMG genes. Results: 1,804 patients enrolled and received somatic sequencing: 52% female; 51% non-Hispanic White/ 20% Hispanic White/ 18% Asian/ 4% Black/ 7% other; median age 64. Review of somatic data suggest that 14% have findings suggestive of germline conditions based on factors such as TMB, MSI, and young age. Of the patients offered germline testing, >95% opted to receive CS/ACMG results. To date, we have sequenced 684 patients for CS and 647 for ACMG. 18% of patients had P/LP mutations in CS genes and 4% had P/LP mutations in ACMG non-cancer genes (Ta- ble). Conclusions: Prospective somatic/germline sequencing of unselected cancer patients reveals tumor findings suggestive of germline disorders and identifies patients with CS and non-cancer genetic conditions. These findings highlight the promise of a comprehensive sequencing approach to help guide cancer treatment, management of unrecognized cancer risk and the need for concomitant management of rare disorders such as arrhythmogenic cardiomyopathy and susceptibility to adverse reactions with anesthesia. Research Sponsor: Internal COH.

Result n (%) Cancer: all P/LP 182 (26.6) Cancer: all P/LP minus CFTR 125 (18.3) Highly penetrant cancer genes BRCA2 11 (1.6) APC, ATM, BLM, FH, MSH6 4 (0.6) each gene BRCA1, HOXB13, NF1, TP53, 3 (0.4) each gene MSH2, PALB2 2 (0.3) each gene CDH1, MEN1, PMS2, RB1, SDHA, SDHB, SDHD 1 (0.15) each gene Moderately-penetrant/autosomal recessive genes CFTR 75 (11.6) MUTYH 18 (2.6) CHEK2 13 (1.9) >23 genes with <1% each gene ACMG: all P/LP 72 (11.1) ACMG: P/LP non-cancer 26 (4) ACMG non-cancer gene frequency ATP7B 13 (2.0) RYR1 4 (0.6) PKP2 3 (0.5) LDLR, MYBPC3 2 (0.3) APOB, KCNQ1 1 (0.2)

10604 Poster Session

Subsequent primary cancers among survivors of adolescent and young adult onset cancers in the United States.

Hyuna Sung, Rebecca Siegel, Kimberly Miller, Ahmedin Jemal; American Cancer Society, Atlanta, GA

Background: Adolescent and young adult (AYA) cancer survivors are at increased risk of subsequent primary cancer (SPC); however, a comprehensive examination of risk patterns across cancer types is lacking in the U.S. Methods: SPC incidence and mortality was calculated among >1-year cancer survivors aged 15 to 39 years at first primary cancer (FPC) diagnosis during 1992-2016 in 12 Surveillance, Epi- demiology, and End Results registries. Rates were expressed as number of cases/deaths per 10,000 person-years and compared with those expected in the general population using standardized incidence (SIR) and standardized mortality ratios (SMR). Results: Among 202,440 survivors of AYA-onset cancers (mean age at FPC diagnosis, 31.8 years; 60.7% women), 6,675 SPC cases (34.3 per 10,000) and 3,786 SPC deaths (19.4 per 10,000) occurred during 1,955,119 person-years of follow-up (mean, 9.7 years), corresponding to an SIR of 1.58 (95%CI = 1.54-1.62) and SMR of 4.19 (95%CI = 4.06- 4.33. In men, overall incidence and mortality SPC rates were statistically significantly higher for each of 21 FPC types compared with risks in the general population, except for thyroid cancer mortality. In women, risk was statistically significantly higher for 14/23 FPC types for incidence and 19/23 FPC types for mortality. SIRs were highest in survivors of pancreatic cancer (SIR = 5.68, 95% CI = 2.94- 9.93; 84 per 10,000), Kaposi sarcoma (SIR = 5.15, 95%CI = 4.62-5.73; 116 per 10,000) and liver cancer (SIR = 4.97, 95%CI = 2.57-8.68; 68.4 per 10,000) in men, and acute lymphoid leukemia (SIR, 3.27, 95% CI = 2.22-4.64; 49.5 per 10,000), Hodgkin lymphoma (SIR = 2.47, 95% CI = 2.22- 2.73; 51.6 per 10,000), and bone sarcoma (SIR = 2.41, 95%CI = 1.80-3.16; 47.6 per 10,000) in women. SMRs were highest in survivors of pancreatic cancer, acute lymphoid leukemia, and stomach cancer in men, and liver cancer, acute lymphoid leukemia, and soft tissue sarcoma in women. Conclu- sions: Overall and type-specific risk patterns of SPCs among AYA cancer survivors differ considerably across FPC type, highlighting the need for targeted approaches for cancer prevention and surveillance in survivorship care planning. Research Sponsor: None.

10605 Poster Session

Mutant PPM1D and TP53 populate the hematopoietic compartment after peptide receptor radionuclide therapy (PRRT) exposure.

Abhay Singh, Nuria Mencia-Trinchant, Elizabeth A. Griffiths, Medhavi Gupta, Matthew Gravina, Rutaba Tajammal, Mark G Faber, LunBiao Yan, Eti Sinha, Duane C. Hassane, Monica L. Guzman, Renuka V. Iyer, Eunice S. Wang, Swapna Thota; Roswell Park Comprehensive Cancer Center, Buffalo, NY; Weill Cornell Medical College, New York, NY; Roswell Park Cancer Institute, Buffalo, NY; Roswell Park Can- cer Institute, Dept. of Medicine, Buffalo, NY; Weill Medical College of Cornell University, New York, NY; Roswell Park Comprehensive Cancer Institute, Buffalo, NY; University of Tennessee Health Science Center, Memphis, TN

Background: Mutations in TP53 and PPM1D are putative drivers associated with therapy related-myeloid neoplasm (T-MN) and have been identified in pre-treatment blood samples obtained at the time of primary malignancy, predating clinically evident T-MN. Genomic analysis of patients(pts) who undergo leukemogenic therapies will help understand T-MN biology and devise risk mitigation strategies. PRRT (Lu 177) for neuroendocrine tumors is associated with enhanced risk of T-MNs. The mechanism for T- MN induced by PRRT is largely elusive due to the novelty of this drug. Methods: We analyzed initial (n=13) and serial blood samples (n=4) prior to and following PRRT for clonal mutations in order to elucidate the role of PRRT in exerting selective pressures on HSCs. Genomic DNA was analyzed using a targeted myeloid 100-gene panel and a variant allele frequency (VAF) cutoff 1% was used to call clonal hematopoiesis (CH). Results: Fifty-four percent pts had CH, despite relatively young age of cohort (median age 58 years, range 41-75) and minimal chemo-radiotherapy exposure; baseline characteristics and molecular profile of cohort is published [Singh et al. Blood 2020; 136 (Supplement 1): 35–36]. Serial sample analysis in 4 pts (Table) demonstrates that PRRT exposure is associated with clonal evolution and accompanying cytopenias in 75% (3/4) pts. Pt-1 (age 67) with normal baseline hemogram developed persistent cytopenias after PRRT, accompanied by emergence and expansion of mutant-PPM1D (mPPM1D; VAF 20%). These data suggest that cytopenias result from repopulation of the HSC compartment by mPPM1D cells. In Pts 2 and 3 (age 74 and 75), we note expansion of mTP53 and mPPM1D clones respectively, also associated with the development of cytopenias. Pt-4 was younger (age 59) and developed no cytopenias. Exposure to PRRT was associated with loss of mTET2 and mDDX41, possibly due to lack of clonal fitness of mTET2/DDX41 clones and the relatively young HSC microenvironment. Conclusions: We conclude that mutations in PPM1D and TP53 are clinically relevant, contribute to clonal cytopenias and may increase risk of future T-MN. The temporal association of mTP53 and mPPM1D expansion with PRRT exposure in our analysis suggests selection of these clones in response to PRRT-induced stress, outcompeting wild type and less therapy-resistant HSCs. Our study along with others will inform future efforts to strategize methods of surveillance and early detection for clonality assessment and chemoprevention, to reduce adverse effects of leukemogenic therapies. Re- search Sponsor: KL2/BTC award from University of Buffalo’s CTSI (S.T.).

ID Gene Mutation Type VAF (%) Nucleotide Change Post PRRT (VAF%)

Pt-1 No mutation 20% (PPM1D) Pt-2 TP53 SNV 1.67 c.586C>G 5% ASXL1 SNV 1.04 c.1954G>A 0% Pt-3 TP53 SNV 1.78 c.842A>T 5% PPM1D Truncating 1.65 c.1508C>A 3% Pt-4 PPM1D Truncating 1.24 c.1709C>G 2.7% TET2 Frameshift 2.60 c.3732_3733del 0% DDX41 SNV 2.65 c.878G>A 0% *SNV = Single nucleotide variants.

TPS10606 Poster Session

A phase II open label, single arm study to evaluate the efficacy of pembrolizumab for leukoplakia.

Ashleigh Porter, Audrey J. Zeh, Thu Ly, Alyssa Serna, Arturo Villanueva, Wanxing Chai-Ho, Diana Messadi, St. John Maie, Marilene Wang, Eliot Abemayor, Ezra E.W. Cohen, Deborah J.L. Wong; UCLA, Los Angeles, CA; Beverly Hills Cancer Ctr, Los Angeles, CA; University of California, San Diego, San Die- go, CA; University of California Los Angeles, Los Angeles, CA

Background: The presence of pre-cancerous oral lesions such as leukoplakia or erythroleukoplakia are known risk factors for the development of squamous cell carcinoma of the head and neck (SCCHN), however preventative agents have not yet shown clinical benefit. The risk of malignant transformation varies but has been quoted as high as 36% in some studies. While the primary mode of treatment of these lesions is largely surgical, recurrence rates are high. Pembrolizumab is a potent and selective hu- manized monoclonal antibody that is designed to directly block the interaction between PD-1 and PD- L1 (as well as PD-L2) that is currently FDA-approved for treatment of SCCHN. We have hypothesized that the treatment of oral premalignant lesions with pembrolizumab would be an effective and well-tol- erated strategy to prevent transformation to invasive cancer. Methods: This study is an open-label, phase II study that will accrue 26 patients with leukoplakia, erythroleukoplakia, or proliferative verrucous leukoplakia with documented moderate to severe dysplasia or carcinoma in situ to be treated with pembrolizumab 200mg every 3 weeks for a total of 6 months. Patients must have visible and measurable lesions that will be both photographed and measured in two dimensions at each visit from the start of treatment until 12 months post-enrollment. Biopsies will be required at diagnosis and following the final treatment, with an optional biopsy following cycle 2 and at progression of disease. Major exclusion criteria include patients with mild dysplasia or hyperplasia, prior chemotherapy, targeted small mole- cule therapy, or radiation therapy within 2 weeks of Day 1 of study, or patients with a known additional malignancy that is active. Patients will also be excluded if they have received anti-PD-1, anti-PD-L1 or anti-PD-L2 treatments in the past. The primary objective is clinical response rate at 6 months, and will be quantified as the percentage of patients with a complete response (CR) and partial response (PR) at 6 months. A CR is defined as complete resolution by visual inspection for 4 weeks of more and a PR is defined as 50% or greater reduction of the product of the 2 dimensions of a single lesions or the sum of all lesions. Progressive disease (PD) is defined as unequivocal increase (greater than or equal to 5mm in one dimension and greater than 20% increase) or the development of new lesions. Secondary objectives will include histologic response rate at 6 months, change in clinical impression based on photographs, clinical response rate at 9 and 12 months, and toxicity. Additional exploratory objectives will include PD-L1 expression in leukoplakia lesions as well as p16 expression, presence of tumor infiltrating lymphocytes, and immunohistochemical as well as RNA sequencing gene expression profiling which may allow for the identification of novel biomarkers. Enrollment began in June 2019 and is ongoing. Clinical trial information: NCT03603223. Research Sponsor: Merck.

TPS10607 Poster Session

A PAncreatic cancer screening study in individuals with New-onset or DeteriOrating diabetes MEllitus (PANDOME study).

Richard C. Frank, Tammy Lo, Dugho Jin, Cornelius Ferreira, Ramanathan Seshadri, Xiang Da (Eric) Dong, James Bauman, Naveen Anand, Ronald Lee; Nuvance Health, Norwalk, CT

Background: Pancreatic adenocarcinoma (PC) has a persistently high mortality as it presents in the advanced stages and has largely not benefited from the genomic and immunotherapeutic revolutions in oncology. Improvements in screening to detect early stage cancers are therefore urgently needed. Screening studies such as those from the International CAPS Consortium have demonstrated improved survivals in hereditary high risk individuals. In the sporadic population, individuals with new-onset diabetes (NOD) or long-standing deteriorating diabetes (DD) are at substantially increased risk of PC in the 12 months following these diagnoses and have been proposed as target populations for screening efforts. This trial will study the benefits of PC screening in the latter populations in a community setting. Methods: Individuals $ 50 years of age with either NOD or DD will be eligible. Criteria for NOD (within the past 12 months) include: fasting blood glucose $ 126 mg/dL, random blood glucose $ 200 mg/dL, or HbA1c $ 6.5%, with confirmed prior normal values. For those without prior glycemic values, a HbA1c $ 7.0% is required. Transition from pre-diabetes requires an increase in HgA1c of $ 0.5%. DD is defined by an increase in HbA1c of $ 2% within the past six months that is not associated with medi- cation non-compliance or weight gain. Study participants will undergo every 6 months: evaluation by an APRN, testing for anxiety and depression and blood donation for biobanking purposes. High resolution MRI/MRCP with gadolinium will be performed at study entry and annually for 2 years. Images will be reviewed at a multi-disciplinary tumor board consisting of body-image certified radiologists, interventional gastroenterologists, hepatobiliary surgeons and medical oncologists. MRI results will be classified according to a novel PANC-RADS system. High-risk pancreatic findings will be further interrogated by endoscopic ultrasound (EUS). Study endpoints include detection rate of high-risk lesions, referrals for EUS and surgery and detection of incidental findings leading to unnecessary procedures. Adverse psy- chological impacts will be assessed through HADS testing. Target accrual: 500 patients within 3 years. Clinical trial information: NCT03937453. Research Sponsor: The Naughton Family Fund.

TPS10608 Poster Session

Comparing the clinical impact of pancreatic cyst surveillance programs: A trial of the ECOG-ACRIN cancer research group (EA2185).

David Weinberg, Jon Steingrimsson, Herbert Zeh, Ruth Carlos, Peter J. O’Dwyer; Fox Chase Cancer Cen- ter, Philadelphia, PA; Brown University, Providence, RI; Department of Surgery, University of Texas Southwestern Medical Center, Dallas, TX; University of Michigan Comprehensive Cancer Center, Ann Arbor, MI; University of Pennsylvania, Pennsylvania Hospital, Philadelphia, PA

Background: The optimal surveillance strategy for pancreatic cysts, which occur in up to 20% of the adult population, is ill defined. Although risk of malignant degeneration of these cysts is low, pancreatic cancer mortality remains high. Two cyst surveillance guidelines, one proposed by an international consensus group (Fukuoka) and the other by the American Gastroenterological Association (AGA), are ac- cepted standards. Both rely on radiographic and endoscopic ultrasonographic imaging. They differ in indications for, and intervals between, imaging tests, with the Fukuoka guideline advocating more intensive imaging. Clinical guidelines that provide discordant recommendations may undercut the quality and/or value of care, and have implications for societal health care costs. The primary objective of this prospective trial is to compare the clinical effectiveness and associated resource utilization of the Fuku- oka and AGA guidelines for pancreatic cyst surveillance. Secondary objectives include a comparison of resource utilization and patient reported outcomes. We are also collecting and banking radiomics data and biospecimens to identify novel methods that might improve cancer risk stratification. Methods: 4606 asymptomatic patients with newly identified pancreatic cysts $1cm in diameter are being randomized 1:1 to high intensity (Fukuoka) or low intensity (AGA) surveillance. The primary endpoint is a composite of any pancreatic cancer without surgery, unresectable pancreatic cancer or cancer > T1a, N0 at surgery, and benign disease at surgery. This sample size will provide 90% power to identify a 30% relative difference in the primary outcome at 5 years between the two study arms. Study duration is 8 years in total, allowing for 2 years of cohort enrollment, 5 years of prospective follow-up, and six months reserved for study initiation and close out. Study participants must be $ 50 years and # 75 years with an ECOG Performance Status 0-1 at baseline. Participants must have received a CT or MRI within 6 months of registration that identifies a new $1 cm pancreatic cyst. Patients with a prior diagnosis of a pancreatic cyst, pancreatic malignancy or a history of pancreatic resection are not eligible. Additional exclusion criteria include a history of acute or chronic pancreatitis, a family history of pancreatic adenocarcinoma in 1 or more first degree relatives, imaging findings or clinical signs that would prompt immediate surgical consideration (enhancing mural nodule, solid component in cyst, pancreatic duct > 10mm, cyst causing obstructive jaundice), a comorbid illness that precludes pancreatic cyst resection, pregnancy or current participation in an established surveillance program. As of February 4, 2021, thirty three (33) participants have been enrolled from two hundred (200) potential sites. Clinical trial information: NCT04239573. Research Sponsor: U.S. National Institutes of Health.

TPS10609 Poster Session

Engaging the radiology community in the National Clinical Trials Network: The ECOG- ACRIN TMIST experience.

Etta Pisano, Constantine Gatsonis, Mitchell D. Schnall, Martin Yaffe, Melissa A. Troester, Ilana F. Gareen, Laura C. Collins, Amarinthia Curtis, Elodia Cole, Jean Cormack, Ruth Carlos, Kathy Miller, Christopher Comstock; Beth Israel Deaconess Medical Center, Boston, MA; Brown University, Provi- dence, RI; Hosp of the Univ of Penn, Philadelphia, PA; Odette Cancer Centre, Sunnybrook Health Sci- ences Centre, University of Toronto, Toronto, ON, Canada; University of North Carolina-Chapel Hill, Chapel Hill, NC; Brown University–ECOG-ACRIN Biostatistics Center, Providence, RI; Beth Israel Dea- coness Medical Center and Harvard Medical School, Boston, MA; Spartanburg Medical Center, Gibbs Cancer Center, Spartanburg, SC; American College of Radiology, Philadelphia, PA; Brown University Center for Statistical Science, Providence, RI; University of Michigan Comprehensive Cancer Center, Ann Arbor, MI; Indiana University Simon Cancer Center Indianapolis, Indianapolis, IN; Memorial Sloan Kettering Cancer Center, New York, NY

Background: ECOG-ACRIN launched the Tomosynthesis Mammographic Imaging Screening Trial (TMIST) through the National Cancer Institute’s National Clinical Trials Network (NCTN)— a network of academic medical centers, community hospitals, and private clinical practices that are committed to participating in NCI-funded clinical trials. The NCI NCTN was developed to support rapid trial start-up of NCI-funded cancer control/prevention, cancer treatment, and non-therapeutic clinical trials that occur within the institution through centralized institutional administration and shared clinical resource allocation (personnel, lab services). TMIST is a randomized clinical trial assessing two breast cancer screening imaging modalities, tomosynthesis and digital mammography, in the population of women presenting for screening mammography and therefore requires active involvement of radiology. Methods: TMIST seeks to enroll 164,946 women, ages 45 to 74 years who present for screening mammography. Because the population under evaluation are women already scheduled for screening mammography, the mammography clinic is critical to successful recruitment as well as adherence to imaging randomization assignments over a 5-year period and therefore must be actively engaged in this trial with a breast imaging radiologist championing the trial within this service. To get active engagement of breast imaging radiologists, we needed to first make them aware of TMIST. Breast imaging radiologists that were already actively involved in the NCTN received notification of the trial through the NCTN email lists. So our goal was to come up with a strategy to reach out to breast imaging radiologists that were not active members in the NCTN. This was achieved through in-person informational sessions to introduce the trial at national and international breast imaging meetings, introduction of the trial and the workings of the NCTN network to the radiology community through articles placed in American College of Radiology (ACR) newsletters, ads promoting TMIST on ACR social media platforms, and direct email by the TMIST study chair to key radiology stakeholders. As of February 15, 2021, there are 115 sites open: 106 in the U.S. and 9 internationally with an additional 54 sites planning to open. A total of 39,366 women are enrolled in the trial with two-thirds also consenting to optional blood and/or buccal cell collection. Minority populations’ participation in the trial is over 20%. A significant drop in enrollment occurred in Spring 2020 coinciding with the suspension of mammography services globally due to COVID-19 beginning mid-March 2020. Enrollment and follow-up screening visits for TMIST restarted in May 2020 and gradually ramped back up to pre-COVID totals in September 2020. Our highest monthly accrual so far occurred in November 2020 with 2,148 subjects enrolled. Clinical trial information: NCT03233191. Research Sponsor: U.S. National Institutes of Health.