Financial toxicity of front-line platinum etoposide plus atezolizumab versus durvalumab in extensive-stage small-cell (ES-SCLC) lung cancer

23

Background: With the continued advent of new therapeutics, burgeoning costs and financial toxicity are more frequently a complication in comprehensive cancer care. At Levine Cancer Institute (LCI), a multidisciplinary Financial Toxicity Tumor Board has driven $55-60 million in cost savings for patients annually, and presented patient cases can serve as a continual impetus for iterative quality improvement and cost-savings research. One such example highlights the augmented risk of financial toxicity resultant from increasing utilization of front-line chemoimmunotherapy (chemo-IO) in ES-SCLC. Both atezolizumab and durvalumab in concert with platinum-etoposide chemotherapy separately portend an overall survival benefit. Retrospective analyses have failed to demonstrate a PFS or OS advantage between the two regimens. Amidst comparable efficacy, we sought to ascertain institutional variance in cost profiles and financial toxicity. Methods: Billing and demographic data for patients with ES-SCLC who received chemo-IO with durvalumab or atezolizumab in an LCI hospital-based outpatient infusion center were queried. Treatment dates encompassed October 1, 2019 through April 30, 2022. Patient costs were compared and exploratory analyses were performed related to patient demographics, patient geography (with financial characteristics per Census data) and bad debt proportions. Results: 211 patients received chemo-IO with atezolizumab, and 108 patients received chemo-IO with durvalumab. Medicare (+/- supplementary insurance), Medicare Advantage, and commercial insurance constituted 85% of payors. Institutional covenants restrict publication of commercial and supplementary insurance data; for compliance and generalizability, reported is the Medicare-only population. Medicare allowables of $6791 and $8429 were utilized per dose of atezolizumab 1200 mg and durvalumab 1500 mg, respectively. Amidst 20% patient responsibility with Medicare, durvalumab patient cost was $327.60 higher per cycle. Incorporating observed mean number of cycles of 5.20 for atezolizumab and 5.31 for durvalumab, an average course of durvalumab was $1836.56 more expensive. Assessing the entire population, 24% of patients in both groups met criteria for bad debt; however, mean quantity of bad debt was 62% higher with durvalumab ($8988.77) than with atezolizumab ($5555.11). Bad debt was disproportionately congregated in lower income zip codes (data to be displayed pictorally). Conclusions: Chemo-IO with durvalumab was $1834.56 more expensive per treatment course in ES-SCLC, with 62% increase in quantity of bad debt relative to atezolizumab. Future directions may focus on institutional interventions to mitigate financial toxicity related to therapy choice and geographic allocation of resources, as well as the assessment of cost variance in other solid tumors.

Associations between biomarker testing and characteristics of patients with metastatic non–small cell lung cancer (mNSCLC): An analysis of CancerLinQ Discovery (CLQD) data

9127

Background: Guidelines have evolved from 2011-2019; there are now 23 approved therapies targeting various predictive biomarkers in mNSCLC. 2021 NCCN Guidelines advocate for a minimum of ALK, EGFR, BRAF, METex14, NTRK1/2/3, RET, KRAS, and ROS1 testing before determining a treatment regimen. The study objective was to estimate the association between the presence of biomarker testing and smoking status, age, sex, race, ethnicity, histology, and diagnosis year in patients with mNSCLC. Methods: CLQD is a real-world data source that provides de-identified electronic health record (EHR) data from more than 60 U.S. oncology practices utilizing 10 different EHRs. This retrospective analysis included patients initially diagnosed with mNSCLC from January 1, 2011, to December 31, 2019. Standard logistic regression models were fit separately by practice to estimate practice-specific odds ratios to assess variability across practices in associations between covariates (smoking status, age, race, etc.) and the primary outcome of biomarker testing, defined as documented testing for EGFR, ALK, ROS1, BRAF, KRAS, MET, RET, ERBB2, and/or PD-L1 within -60 to +365 days of mNSCLC diagnosis. Random effects logistic regression was then used to estimate associations with random intercepts, accounting for clustering by practices. Results are reported as odds ratios (OR) with 95% confidence intervals (CI). Results: 8704 patients from 31 practices were eligible. Testing rates increased from 31.5% in 2011 to a peak of 62.3% in 2017. Patients with a smoking history were half as likely to receive testing than patients without a smoking history (OR = 0.50, 95% CI: 0.41, 0.60); patients with unknown smoking history were 0.66 times as likely (95% CI: 0.52, 0.84). Females were more likely to be tested than males (OR = 1.19, 95% CI: 1.07, 1.32). After adjusting for other covariates, Asian patients were 1.51 times more likely to be tested than patients of other races (95% CI: 1.05, 2.17); Hispanic patients were 1.33 times more likely to be tested than patients without Hispanic ethnicity (95% CI: 0.99, 1.78). The odds of receiving biomarker testing were 6x greater for patients with non-squamous mNSCLC versus squamous mNSCLC (95% CI: 5.45, 7.20). Patients > 70 years old were less likely to be tested (OR = 0.83, 95% CI: 0.75, 0.93) than younger patients. Conclusions: Our data demonstrate annual increases in testing rates, reflecting guideline changes. However, in this cohort of patients with mNSCLC, biomarker testing was more likely for non-squamous mNSCLC patients, females, Asians, Hispanics, or those who did not have a history of smoking. Patient characteristics should no longer factor into obtaining biomarker testing. Non-discriminant, broad panel-based reflex molecular testing in mNSCLC can reduce treatment choice ambiguity and enhance patient opportunities.

Equity for under-served populations in lung cancer screening and treatment: Does mobile low-dose CT scanning lead to stage shift and diagnosis with potential cures at 4 years of follow-up?

8551

Background: Randomized trials proved that screening high-risk patients with LDCT of chest reduces lung cancer mortality. Under-served patients have missed this benefit in most studies through access issues. We showed that mobile LDCT improves access and now assess if this translates to equity of survival. Methods: We used two coaches with BodyTom © portable 32 slice low-dose CT scanners (Samsung) to screen uninsured and under-served heavy smokers for lung cancer (Oncologist, 2019). All films were reviewed by central panel using LUNG RADS technique. Protocol was approved by Advarra IRB. Medicare pts were excluded as insurance covered them for LDCT (causing negative bias for diagnosis as the elderly are at high risk). Results: We initially screened 1200 uninsured/under-insured subjects, mean age 61 years (range 55-64), with average pack year history of 47.8 (30-150); 61% male; 18% Black, 3% Hispanic/Latino; 78% rural. We found 97 pts with LUNG RADS 4 (high risk) lesions, 30 lung cancers (2.5%), including 18 at stage I-III treated with curative intent (60%); 5 incidental non-lung cancers (renal CA 2, head & neck CA 1, pancreas CA 2); > 50% with intercurrent cardiovascular disease and COPD seen on LDCT. Of eligible first-screen subjects, 51% attended 12 month repeat LDCT and 27% attended third LDCT. One pt (6%) treated with curative intent has relapsed to date (median follow up 2.5 years, with 25% beyond 3 years). An additional 288 screened pts revealed 9 lung cancers (5 stage I-III), confirming shift to early stage disease at diagnosis. Conclusions: Mobile LDCT yields higher screening rate for under-served pts than prior hallmark trials, with shift to early-stage detection of lung cancer, with sustained treatment-induced remissions beyond 4 years. This approach could be applied to improve national lung cancer survival in the under-served.

LUNG cancer: Supportive oncology program development for patients and care partners: A qualitative study

e18652

Background: Lung cancer patients (LCP) have various unmet needs. Evidence supports tailored services to provide individuals information and interaction with healthcare professionals (HCP). We conducted a qualitative study of the ideal supportive care services and structure for LCP and care partners (CP) within Levine Cancer Institute, a regional, multi-site cancer system. Methods: Methods included Focus Groups (FG) and a paper-based survey (PBS) in the Section of Thoracic Oncology. FG included HCP, local Lung Cancer Support Group leadership, LCP, and their CP. FG identified important topics for LCP, CP, and HCP. FG recordings were transcribed, analyzed, and incorporated with the evidence base to inform PBS development. LCP and CP participated in the PBS and responses analyzed with REDCap (v11.1 11, ©2022 Vanderbilt). FG and PBS data were compared for overlap and discordance. Results: FG (total N = 18) identified four Themes: 1) unmet needs 2) program structure 3) barriers to care 4) program content; each analyzed further into three codes (Table). FG Themes are presented in order of frequency (FQ) of unique comments. Unmet needs (FQ=25) included: loneliness, care for the CP, onboarding burden. Structure (FQ=24) included: innovation, variability, virtual. Barriers to care (FQ=20) included: seeking/needing knowledge, transportation, psychosocial. Content (FQ=18) included: combined content and services, financial, educational. Among the responses (total n = 44), strong overlap existed in six PBS areas. Most agreed on seeking/needing knowledge and psychosocial barriers. The majority agreed with combined content and services and the importance of financial and educational content. Variability was specified by agreement on printed and patient portal digital materials with a moderate preference for virtual meetings. Discordance was found across loneliness and transportation while care for the CP, onboarding burden, and innovation were neutral or irrelevant. Conclusions: FG findings confirmed unmet needs, program structure, barriers to care, and program content as integral LCP and CP program concerns. Identified overlap between these Themes and PBS responses are addressed in the ideal LCP and CP support program. This methodology can be generalized to guide future program development to serve other populations and cancer institutes. [Table: see text]

High incidence of concurrent disease states detected during mobile lung cancer screening in an underserved population

6540

Background: Studies such as the National Lung Screening Trial (NLST; N Engl J Med 2011;365:395-409) have shown a survival benefit to low-dose Lung CT screening in high-risk smokers. Levine Cancer Institute (LCI) initiated the first mobile low dose computerized tomographic (LDCT) lung screening program for underserved populations in 2017. In addition to being able to intervene early in the natural history of lung cancer, the project has also shown a previously unreported high incidence of incidental diseases in this population. We characterize these findings in 1198 patients. Methods: From May 2017, subjects with criteria eligible for NLST screening were identified and underwent LDCT for lung cancer detection. Patients screened in the program were all uninsured or underinsured, mean age 60.8 years, 18% were African American, 3% Latin-x and 78% were rural with an overall 47.1 mean pack-year smoking history. These patients were screened using a novel mobile LDCT (J Clin Oncol 37, 2019 suppl; abstr 6567) created for this program. By December 2020, 1198 patients completed their first screening. All CT scans were reviewed by two separate radiologists and were reviewed for quality assurance by a separate expert multidisciplinary team. Results: Of the 1198 subjects, 84% (1006 subjects) were found by LDCT to have a least one incidental disease. More than half of the subjects (645, 53.8%) had coronary atherosclerosis. Of those, 25% (183) were described to have at least moderate disease with 8% (96) described as severe. Overall, 42% (504) were found to have emphysema and 25% (299) had vascular atherosclerotic disease; 1.8% (22) of those screened had a detected aortic aneurysm. In total, thirty separate disease findings were found (listed from fourth to tenth most common finding: degenerative spine changes [205], cholelithiasis [59], hiatal hernia [52], pericardial effusions [38], fatty liver [32], kidney stone [3]), and cardiomegaly [30]). 3.5% (42) were found to have an undiscovered breast, adrenal, liver or kidney mass that required further workup. Conclusions: The number of incidental findings in our mainly rural underserved subject group was very high (84%). 35.5% of patients in the National Lung Screening Trial died from heart disease or respiratory disease. These numbers have not been overtly discussed and our study confirms the number of concerning incidental diseases that can lead to morbidity or mortality. In this high-risk, underserved population of heavy smokers, the opportunity for positive impact on other disease states can be increased by a mobile lung cancer screening program by increasing access to care.

Mobile low-dose computerized tomography (LDCT): Three-year follow up of solution for early diagnosis of lung cancer in under-served populations

6507

Background: Randomized trials have proven that screening high-risk patients with LDCT of chest reduces lung cancer mortality compared to screening with chest x-ray. Under-served patients lack access to this test due to geographic and socio-economic factors. We hypothesized that a mobile screening unit would improve access and increase survival in this group, which is most at risk of lung cancer deaths. Methods: We installed a BodyTom portable 32 slice low-dose CT scanner (Samsung Inc) into a 35 foot coach (Frazier Inc), reinforced to avoid equipment damage during road travel. It includes waiting area, high speed wireless internet connection for rapid image transfer, and electronic tablets to deliver smoking cessation and health education programs and shared decision-making video aids. We used LUNG RADS approach to lesion classification, yielding high sensitivity and specificity in assessment. All films were reviewed by a central panel. This is certified as a lung cancer screening Center of Excellence by the Lung Cancer Alliance. Protocol was approved by Advarra IRB. Medicare pts excluded as insurance covered them for LDCT, although this reduced potential number of cases diagnosed as this is highest risk population. Results: We screened 1200 uninsured or under-insured subjects, mean age 61 years (range 55-64), with average pack year history of 47.8 (30-150); 61% male; 18% Black, 3% Hispanic/Latino; 78% rural. We found 97 pts with LUNG RADS 4 lesions, 30 lung cancers (2.5%), including 15 at stage I-II treated with curative intent; 5 incidental non-lung cancers (renal CA 2, head & neck CA 1, pancreas CA 2); more than 50% with cardiovascular disease or COPD seen on LDCT. Of eligible first-screen subjects (J. Clin. Oncol., 2019, 37, 383S), 440 attended Year 1 repeat LDCT and 161 attended Year 2 LDCT. Only one pt with surgically resected CA lung has relapsed to date. Conclusions: Mobile LDCT yields higher screening rate for under-served pts than prior international studies, with strong protocol adherence and paucity of early cancer deaths in high-risk population with traditionally poor compliance.

Upper respiratory tract SARS-CoV-2 viral shedding in cancer patients

e18776

Background: SARS-CoV-2 virus has been shown to persist in respiratory tract in immunocompromised patients. However, such data are lacking for both asymptomatic and symptomatic SARS-CoV-2 infection in cancer patients. We share our single center experience on duration of SARS-CoV-2 viral presence in the upper respiratory tract of cancer patients with SARS-CoV-2 infection (asymptomatic and symptomatic) detected by viral PCR. Methods: This is retrospective review of cancer patients with documented SARS-CoV-2 infection and measurement of viral shedding at Levine Cancer Institute. Testing indications were COVID-19 symptomatic illness, pre-procedural and pre-chemo testing. Prolonged shedding was defined as presence of viral RNA beyond 30 days after first positive test. To document viral clearance, patients required 2 negative SARS-CoV-2 PCR test separated by at least 24 hours and maximum 3 weeks apart either by nasopharyngeal or nasal PCR swab. Differences in distributions were identified between patients shedding virus more than 30 days and less than 30 days using uni- and multivariable logistic regression models. Statistical significance was set at p < 0.10 to enter the multivariable model, and p < 0.05 to remain. Results: Demographic data: median age 62 (range 20-93); 58.5% females; 70% White, 21% Black, and 7.4% Hispanics. Comorbidities included hypertension 43.2%, diabetes 16.7% and chronic lung disease 3.7%. Underlying malignancies were breast cancer 25%, hematologic cancer 22%, lung cancer 16% and genitourinary 11%. Chemotherapy was received by 26.5% patients within 4 weeks prior to testing. 162 patients were identified median duration of 18 days (range 4-90 days). Of these, 76% patients were tested for non-symptomatic indication with median duration of shedding 17 days (range 6-80) and 23% were tested for clinical symptoms with median duration of shedding 29 days (range 4-90) (p = < 0.001); 50% of patients never developed symptoms, whereas 35% patients with non-symptomatic testing indication, subsequently developed symptoms. Viral clearance by day 30, day 45, day 60 and day 90 was 78%, 93%, 97% and 100% respectively. Univariate analysis did not show difference between patients with prolonged shedding vs those shedding less than 30 days for age, gender, race, ethnicity, underlying malignancy, co-morbidities including body mass index, diabetes, chronic lung conditions, hypertension, or receipt of cytotoxic chemo. Multivariable analysis showed that presence of symptoms at any point during SARS-CoV-2 infection (OR 5.9, 95% CI 2.4-14.5, p < 0.001) was associated with prolonged shedding. Conclusions: Symptomatic SARS-CoV-2 infection is associated with prolonged viral shedding in cancer patients. Cancer patients can have asymptomatic SARS-CoV-2 infection. More studies are warranted to understand viral kinetics and its clinical implications in cancer patients.

Mortality risk for patients undergoing cancer treatment who acquire SARS-CoV-2: ASCO registry

6509

Background: The ASCO Registry was created to analyze the impact of COVID-19 (COVID) on treatment (Tx) and outcomes of patients (pts) with cancer. Methods: The Registry includes pts with 1) a confirmed COVID diagnosis (Dx) and 2) clinically evident cancer receiving Tx/supportive care or resected cancer on adjuvant Tx <12 mos since surgery. Practices report data on cancer Dx and Tx at COVID Dx, COVID symptoms, comorbidities, cancer/COVID Tx, and survival. Kaplan-Meier estimation provided 30- and 90-day mortality rate estimates for pts with COVID Dx before or since 6/1/20 within pt subgroups with 95% confidence intervals (CI). Data submission cutoff for all practices was 10/24/20, except one that was 11/16/20. Results: This analysis reports on 453 pts with COVID Dx 3/5/20 to 10/22/20 who were on anticancer drug Tx for regional (9%) or metastatic (53%) solid tumors or hematologic cancers (38%) at COVID Dx. 38 practices entered data: health system-owned 51% of pts, privately-owned 25%, academic 24%. 53% of pts have ≥30 days follow-up or died ≤30 days from COVID Dx. Median age is 64 years; 53% of pts are female; 28% of pts are asymptomatic at COVID Dx. Multiple myeloma was most frequent cancer (17%). All-cause mortality rates (30 and 90 days) increased with pts’ age at COVID Dx [Table]. No mortality difference was seen based on sex, race, or comorbidities (hypertension, diabetes, pulmonary disease). Pts with COVID Dx before June 1 had worse survival than pts diagnosed on/after June 1. Pts with B-cell malignancies had higher mortality rates than pts with solid tumors. Conclusions: Severity of COVID illness and mortality were greater for patients with COVID Dx pre-June 1 than on/after June 1. Differences on/after June 1 may be attributed to improvements in COVID management, higher COVID testing rates, and more asymptomatic pts diagnosed. Variations in COVID-19 pt populations over time due to these changes should be considered when analyzing and interpreting pandemic data. Cancer pts with advanced age and B-cell cancers are at greatest risk of death but mortality rates for all groups (except those admitted to ICU) improved after 6/1/2020.[Table: see text]

Observed survival benefit with limited exposure of durvalumab in unresectable stage III non-small cell lung cancer at a large community-based institution

e20539

Background: The PACIFIC trial ushered in a paradigm shift in the management of unresectable, non-metastatic non-small cell lung cancer (NSCLC), demonstrating improvement in 12,24,36-month overall survival (OS) and leading to the 2018 FDA approval for durvalumab in unresectable or locally advanced stage III NSCLC. With almost 3 years of FDA approval, we performed a retrospective analysis of patient experiences and outcomes at Levine Cancer Institute analyzing patient data to assess survival and potential points of clinical significance. Methods: Patients over the age of 18, who met criteria similar to the PACIFIC trial (i.e. unresectable or locally advanced stage III NSCLC) from February 2018 through September 2020 were analyzed. Those who were receiving active treatment at the data cutoff were excluded. Patient characteristics, prior treatment, durvalumab administration, immune-related adverse events (irAEs), and efficacy data were summarized and evaluated. OS and progression free survival (PFS) were evaluated with Kaplan Meier methods. Results: A total of 159 patients were evaluated. 40.9% were female and 59.1% were male. The median age was 67 (range 38-83 years). Of note, 86.8% of patients were white, whereas 13.2% were nonwhite. 50.3% patients experienced an irAE. The most common reasons for discontinuation of durvalumab were completion (at least 24 doses), progressive disease, or toxicity (33.3%, 30.8%, 26.4%, respectively). The median number of doses of durvalumab received was 14 (range 1-26 doses). The median PFS was 15.3 months with 12-and 24-month PFS being 54% and 41.1 %, respectively. Median OS was 42 months with 12-and 24-month OS being 78.1% and 67.8%, respectively. Our analysis compared outcomes in those who completed adjuvant durvalumab versus those who did not complete adjuvant therapy (Table). Conclusions: Data shows the best survival in those who completed durvalumab (comparable to historic values) and novel data shows a perceived survival benefit in those completing 12 doses compared to those who did not. Thus, partial treatment may provide a survival advantage. Further multivariate analysis will look for possible correlations to increased immune events and inability to complete therapy. Further investigation will delve into this cohort’s small proportion of non-white patients, evaluating for possible barriers to care that may lead to more patients being diagnosed with stage IV NSCLC.[Table: see text]

Physician concern about delaying lung cancer treatment while awaiting biomarker testing: Results of a survey of U.S. oncologists

9067

Background: With rapid advancements in biomarker testing informing lung cancer treatment decisions, clinicians are challenged to maintain knowledge of who, what and when to test and how to treat based on test results. An ASCO taskforce including representatives from the American Cancer Society National Lung Cancer Roundtable and patient advocates conducted a study to assess biomarker testing and treatment practices for patients with advanced non-small cell lung cancer (aNSCLC) among U.S. oncologists. Methods: A survey was sent to 2374 ASCO members – lung cancer specialists and general oncologists. Eligibility required treating ≥1 lung cancer patient/month. Proportions were estimated across groups and compared using chi-square tests. Results: 170 responses were analyzed. 59% of respondents work at an academic center (i.e., have a fellowship program), while 41% work at a community (non-academic hospital/health system/private practice). Nearly all (98%) believe biomarker results should be received within 1 or 2 weeks of ordering, yet 37% wait an average of 3 or 4 weeks for results. Of respondents who usually wait 3 or 4 weeks, 37% initiate a non-targeted systemic treatment while waiting. Respondents from community practices were more likely to initiate non-targeted systemic treatment if results were not available after 2 weeks (59% compared to 40% of academic respondents; p = 0.013). ). When asked about reasons for not testing, respondents <5 years since training were more likely to report that delaying treatment while waiting for results was always/often a concern compared to those >6 years from training (41% vs 19%). Respondents reported high testing rates in both non-squamous and squamous aNSCLC. Roughly equal representation of generalists/specialists and academic/community respondents helps mitigate potential concerns about external validity. Conclusions: Respondents indicated that treatment decisions are impacted by delays in biomarker test results. Clinicians should be informed about when it is safe and appropriate to defer treatment while biomarker testing is pending. Respondents suggest that diagnostic biomarker testing companies should strive to expedite results.[Table: see text]

Survival outcomes with the use of immunotherapy (IO) in patients (pts) with metastatic non-small cell lung cancer (mNSCLC) at a large hybrid cancer institute

e21192

Background: Landmark trials have shown increased survival in pts who receive IO for NSCLC as compared to chemotherapy (chemo). Median overall survival (mOS) ranged from 17-26 months (mo); however, mOS in several published “real-world” retrospective studies are lower, ranging from 8-12 mo, with about 4-5 months on IO treatment. We sought to define mOS of pts with mNSCLC who received IO as monotherapy or in combination with chemo as first line therapy at Levine Cancer Institute. Methods: We retrospectively reviewed 315 adult pts with mNSCLC without driver mutations diagnosed between 2016 and 2019. The Kaplan-Meier method was used to estimate and compare OS between IO and IO + chemo. Univariate and multivariate Cox models were used to evaluate risk factors (RF) for OS. RFs considered included age, sex, race, smoking status, histology, first-line treatment type, and metastatic (mets) sites. Results: Baseline pt characteristics were: 40% female, 77% white, 20% Black, 34% current smokers and 60% former smokers. Median age was 69 years (45-88) in pts receiving IO alone and 63 years (28-84) in pts receiving IO + chemo. Tumor characteristics were: 76% adenocarcinoma and 17% squamous cell carcinoma. PD-L1 TPS distribution was: 39% for 0%, 22% for 1-49%, and 39% for ≥50%. Distribution of mets was 10% adrenal, 40% bone, 30% brain, 14% liver, 31% lung. mOS for pts receiving IO and IO + chemo as first line therapy was 17 and 14.8 mos, respectively (P = .209). Median duration of IO received was 4.25 months (0 to 43.6). mOS as stratified by PD-L1 TPS was 14.5 mos for PD-L1 0%, 13.3 mos for PD-L1 1-49%, and 19.5 mos for PD-L1 ≥50% (P = .163). OS was significantly different between IO and IO + chemo after adjusting for age. No OS differences were seen between white and Black or between all pts vs pts with brain mets (brain-specific interventions not reviewed). The table summarizes significant findings only (P < 0.05). Conclusions: Pts with mNSCLC treated first-line with IO either alone or with chemo at Levine Cancer Institute lived longer than those in similar published “real-world” cohorts. Median OS was highest in patients with PD-L1 TPS ≥50%, although not statistically significant. While not unusual to identify worse outcomes in those with bone and liver mets, interestingly brain metastasis was not associated with worse survival. In this cohort, when adjusted for age, IO alone trends toward improved survival. Although there was no OS difference based on race, further investigation will seek to uncover any other disparities contributing to outcomes, such as insurance status and zip code mapping. To our knowledge, this provides the largest analysis of this patient population outside of a clinical trial.[Table: see text]

American Society of Clinical Oncology Road to Recovery Report: Learning From the COVID-19 Experience to Improve Clinical Research and Cancer Care

This report presents the American Society of Clinical Oncology's (ASCO's) evaluation of the adaptations in care delivery, research operations, and regulatory oversight made in response to the coronavirus pandemic and presents recommendations for moving forward as the pandemic recedes. ASCO organized its recommendations for clinical research around five goals to ensure lessons learned from the COVID-19 experience are used to craft a more equitable, accessible, and efficient clinical research system that protects patient safety, ensures scientific integrity, and maintains data quality. The specific goals are: (1) ensure that clinical research is accessible, affordable, and equitable; (2) design more pragmatic and efficient clinical trials; (3) minimize administrative and regulatory burdens on research sites; (4) recruit, retain, and support a well-trained clinical research workforce; and (5) promote appropriate oversight and review of clinical trial conduct and results. Similarly, ASCO also organized its recommendations regarding cancer care delivery around five goals: (1) promote and protect equitable access to high-quality cancer care; (2) support safe delivery of high-quality cancer care; (3) advance policies to ensure oncology providers have sufficient resources to provide high-quality patient care; (4) recognize and address threats to clinician, provider, and patient well-being; and (5) improve patient access to high-quality cancer care via telemedicine. ASCO will work at all levels to advance the recommendations made in this report.

Venous thromboembolism (VTE) incidence and risk factors in patients (pts) with non-small cell lung cancer (NSCLC) receiving front-line therapy

e19293

Background: VTE incidence varies based on factors such as tumor type, stage, and treatment. There is limited data on VTE incidence and risk factors in NSCLC pts receiving first-line therapies, including immune checkpoint inhibitors (ICIs) and targeted therapies (TTs). Methods: This is a single institution retrospective cohort study of adult NSCLC pts who received first-line treatment between July 2003 and July 2019. Treatments included chemotherapy (chemo) (platinums, taxanes, pemetrexed, gemcitabine, etoposide, bevacizumab), ICI (pembrolizumab, nivolumab, atezolizumab, durvalumab), chemo + ICI, or TT (erlotinib, gefitinib, afatinib, osimertinib, crizotinib, alectinib, ceritinib). Diagnosis codes (ICD 9/10 codes) confirmed VTE (deep vein thrombosis and/or pulmonary embolism) and presence of risk factors which are summarized in Table. Landmark VTE incidence was estimated from cumulative incidence curves for time to VTE, death as a competing risk. Time to VTE distributions were compared between groups with Gray’s tests. Univariable and multivariable competing risk analyses identified risk factors for time to VTE. Results: In 1,618 evaluable pts, the median age was 66 years, 53% were male, 79% White, 18% Black, 58% had adenocarcinoma, 32% squamous cell carcinoma, and 47% metastatic disease. 1178 received chemo, 172 ICIs, 157 chemo + ICI, and 111 TTs. 6-month VTE rates per arm were 5.3%, 7.0%, 7.2%, and 12.0% and 12-month rates were 8.9%, 8.1%, 11.7%, and 13.3%, respectively. Cumulative incidence of VTE was not significantly different between treatment groups (p = .27). Univariable and multivariable analyses are summarized in the Table below. Conclusions: Treatment type was not associated with VTE risk in first-line NSCLC, but rates were numerically highest in pts receiving TTs. Khorana risk score was significantly associated with VTE risk and may identify those likely to benefit from thromboprophylaxis. [Table: see text]

Phase II prospective trial of primary lung tumor stereotactic body radiation therapy (SBRT) followed by concurrent mediastinal chemoradiation and adjuvant immunotherapy for locally-advanced non-small cell lung cancer (NSCLC): Planned toxicity analysis

e21047

Background: The care of pts with locally-advanced NSCLC has continued to evolve. Dose escalation with conventional radiation and concurrent chemotherapy has failed to improve outcomes and is associated with low primary tumor control rates and high toxicity rates. We initiated a phase II study to evaluate full dose SBRT to the primary tumor followed by conventional chemoradiation to the involved lymph nodes followed by adjuvant immunotherapy in patients (pts) with unresectable locally-advanced NSCLC. Methods: Eligible pts included peripheral primary tumors ≤ 7cm or centrally based tumors that had at least 2 cm separation from involved nodal disease. Pts received SBRT to the primary tumor (50-54 Gy in 3-5 fractions) followed by standard radiation to 60 Gy in 30 fractions with concurrent chemotherapy (physician choice of carboplatin/paclitaxel or cisplatin/etoposide). Pts without disease progression after chemoradiation then received adjuvant durvalumab (PACIFIC trial). Planned analysis of early toxicity was performed given this is the first prospective trial to replace conventional fractionation and deliver full dose SBRT to the primary tumor in the locally advanced setting. Results: From May 2017 to January 2020, 35pts were enrolled. Median follow up is 7.9 months (range 0-27 months). Toxicity related to either SBRT or mediastinal radiation was lower than previous published prospective trials with a grade 2 radiation pneumonitis incidence of only 17%. There was no ≥grade 3 pneumonitis. Grade 3 esophagitis was observed in 1 pt (3%) with grade 2 esophagitis in 37%. No grade ≥3 toxicity has been attributed to SBRT with 26% of pts experiencing a grade 2 toxicity from SBRT. Overall incidence of grade ≥3 pulmonary toxicity attributed to any therapy was 9%. One pt died of neutropenic sepsis due to legionella while on therapy attributed to chemotherapy. No major cardiac events were reported to date. Conclusions: Full dose SBRT to the primary tumor followed by standard chemoradiation to the involved lymph nodes appears to have a favorable toxicity profile when compared with conventional dose escalation and to previously published experiences with SBRT boost following chemoradiation in the treatment of locally advanced NSCLC. Our trial continues to enroll with a target of 56 pts. Clinical trial information: NCT03141359 .

Early Impact of COVID-19 on the Conduct of Oncology Clinical Trials and Long-Term Opportunities for Transformation: Findings From an American Society of Clinical Oncology Survey

The coronavirus disease 2019 (COVID-19) pandemic has disrupted all aspects of clinical care, including cancer clinical trials. In March 2020, ASCO launched a survey of clinical programs represented on its Cancer Research Committee and Research Community Forum Steering Group and taskforces to learn about the types of changes and challenges that clinical trial programs were experiencing early in the pandemic. There were 32 survey respondents; 14 represented academic programs, and 18 represented community-based programs. Respondents indicated that COVID-19 is leading programs to halt or prioritize screening and/or enrollment for certain clinical trials and cease research-only visits. Most reported conducting remote patient care where possible and remote visits and monitoring with sponsors and/or contract research organizations (CROs); respondents viewed this shift positively. Numerous challenges with conducting clinical trials were reported, including enrollment and protocol adherence difficulties with decreased patient visits, staffing constraints, and limited availability of ancillary services. Interactions with sponsors and CROs about modifying trial procedures were also challenging. The changes in clinical trial procedures identified by the survey could serve as strategies for other programs attempting to maintain their clinical trial portfolios during the COVID-19 pandemic. Additionally, many of the adaptations to trials made during the pandemic provide a long-term opportunity to improve and transform the clinical trial system. Specific improvements could be expanded use of more pragmatic or streamlined trial designs, fewer clinical trial-related patient visits, and minimized sponsor and CRO visits to trial programs.

Effectiveness of mobile computerized tomographic (CT) lung scanning unit for early diagnosis of lung cancer in under-served populations

6567

Background: The National Lung Screening Trial (NLST) demonstrated that screening high-risk patients with low-dose CT (LDCT) of the chest reduces lung cancer mortality compared to screening with chest x-ray. Uninsured and Medicaid patients lack access to this hospital-based screening test due to geographic isolation/socio-economic factors. We hypothesized that a mobile screening unit would improve access and confer benefits demonstrated by the NLST to this under-served group, which is most at risk of lung cancer deaths. Methods: In collaboration with Samsung Inc, we inserted a BodyTom portable 32 slide low-dose CT scanner into a 35-foot coach, reinforced to avoid equipment damage, to function as a mobile lung scanning unit. The unit includes a waiting area, high speed wireless internet connection for rapid image transfer, and electronic tablets to deliver smoking cessation and health education programs and shared decision-making video aids. It has been certified as a lung cancer screening Center of Excellence by Lung Cancer Alliance. We employed the LUNG RADS approach to lesion classification, yielding high sensitivity and specificity in assessment. All films were reviewed by a central panel of oncologists, pulmonologists and radiologists. The protocol was approved by Chesapeake IRB, which oversees all LCI cancer trials. Interim analysis at this time was approved by the Oversight Committee. Results: We screened 470 under-served smokers between 4/2017-1/2019; M:F 1.1:1, mean age 61 years (range 55-64), with average pack year history of 45.7 (30-150) (25% African-American; 3% Hispanic; 65% rural; 100% uninsured, under-insured or Medicaid - NC Medicaid does not cover lung cancer screening). Patients over the age of 64 years were excluded as they are covered by Medicare for lung cancer screening. We found at initial screen 35 subjects with LUNG RADS 4 lesions, 49 subjects with LUNG RADS 3 lesions, 10 lung cancers (2.1%), including 4 at stage I-II. 4 non-lung cancers were identified and treated. Other incidental non-oncologic findings are the subject of another presentation. Conclusions: In this small sample using the first mobile low dose CT lung screening unit in the United States, the initial cancer detection rate is comparable to that reported in the NLST but with marked improvement of screening rates in underserved groups and with better anticipated outcomes at lower cost than if they had first presented with metastatic disease.

Impact of Electronically Accessible Pathways (EAPathways) on clinical trial enrollment at a large multisite cancer center

6517

Background: Clinical pathways streamline evidence-based treatment decisions and provide consistent, high-quality, value-based care. A high-quality clinical pathway should enhance screening and access to clinical trials. Our healthcare system utilizes EAPathways to allow providers to select treatment regimens vetted by section experts, inquire about clinical trials, and refer to relevant programs (e.g. palliative medicine) or testing (e.g. genomics) at a main cancer center and 22 regional sites. With over 400 clinical trials, our goal is to provide access regardless of where a patient lives or receives treatment. We aim to explore the impact of EAPathways on clinical trial enrollment at our healthcare system. Methods: This study is a retrospective review to compare clinical trial inquiries through EAPathways and clinical trial enrollment using Oncore between 1/1/2017 and 7/31/2018. The primary outcome is the success rate reported as the total number of inquiries that resulted in clinical trial enrollment. Other outcomes include a comparison of inquiries and enrollments for hematology and solid tumor oncology, cancer treatment and non-treatment (e.g. specimen collection), and our main cancer center and regional sites. The number of and reason for opting out of treatments or trials was also analyzed. Results: A total of 29.1% (740/2539) of clinical trial inquiries through EAPathways resulted in clinical trial enrollment. Success rates for the following settings were reported: 39.5% (223/564) in hematology, 26.2% (517/1975) in solid tumor oncology, 27.0% (594/2203) in treatment trials, and 43.5% (146/336) in non-treatment clinical trials. Sixty-three percent of enrollments were at our main cancer center compared to regional sites. A total of 39.7% (3356/8453) of patients were enrolled into an opt-out pathway due to reasons such as performance status, organ dysfunction, or hospice. Conclusions: Clinical pathways can provide access to clinical trial enrollment in multiple settings. These baseline metrics will help assess process improvement needs to increase clinical trial enrollment success rates and address reasons for opt-out.

Effectiveness of ASCO’s adverse event reporting decision aid: Results from an interventional study

3065

Background: Investigators often send adverse event (AE) reports to sponsors that are incorrectly categorized as serious or attributed to the investigational drug. Such errors contribute to a high volume of uninformative IND safety reports that sponsors submit to FDA and all participating investigators, straining stakeholder resources and impeding the detection of valid safety signals. To improve the quality of AE reporting, ASCO developed and tested a Decision Aid Tool (DAT). Methods: An interventional study with a cross-over design was conducted. Physician investigators and research staff were randomized to receive case studies. Cases were assessed for seriousness and attribution, first unassisted and then with the DAT. Participants completed a feedback survey. Effectiveness of reporting and attribution were assessed using logistic regression. Results are reported as odds ratios (OR) with 95% confidence intervals (CI). Results: Most of the 29 participants reported that the DAT was helpful (93%), improved their decision-making time (69%) and confidence in reporting (83%), and that they would use it in practice (83%). The DAT did not significantly affect accuracy of determining seriousness (OR, 0.87; 95% CI: 0.31, 2.46) but it did significantly increase accuracy of attributing a serious AE to a drug (OR, 3.60; 95% CI: 1.15, 11.4). Conclusions: The DAT shows promise as a method to reduce errors in attribution of AEs, which may help to ensure the detection of valid safety signals. Many participants were experienced clinical trialists, and the DAT may show greater utility as an educational tool for novice investigators, research staff, and students.

Tissue sampling for genomic testing in a community-based center

e18000

Background: Genomic testing requires greater quality than routine clinical diagnostic tests. As genomic testing proliferates and becomes more widespread, adjustments in processes may be needed by community-based pathology, radiology, and surgery groups. Both tissue quality and quantity parameters are essential for successful completion of next-generation sequencing (NGS) panels. Methods: A NGS genomic platform ( > 600 genes) was implemented in a hybrid academic community-based cancer institute, with success/failure rates recorded.. Data were collected from the electronic medical records and biorepository data system. Descriptive statistics assessed success or failure rates by tissue quality or quantity with subgroup analyses by disease sites (primary/metastases), collection procedure, and specimen age. Results: From 6/2015-2/2017, 809 NGS tests yielded 143 (17%) failures [specimen quality (34%) and specimen quantity (66%)]. Of the failed tests, specimens were collected from 26 primary disease sites [lung (21%)] and 18 metastatic sites [liver (29%), lymph node (19%), and lung, omentum, brain, and peritoneal specimens (all < 10%)]. Of the failed tests, all bronchoscopy, EBUS, and peritoneal fluid specimens and the majority of FNAs (78%) were due to insufficient tissue quantity, whereas surgical resection or excision had more issues with quality of the specimen (60%). Of note, surgical specimens ranged from 2005-2016; NGS quality initiatives were implemented system-wide in 2015, leading to substantially fewer failed tests. Cores biopsies had fewer quantity (24%) or quality (18%) issues. The number of tests failing due to insufficient quantity also declined due to education and communication processes implemented. Conclusions: The majority of NGS tests provided valuable information. Implementation of systematic tissue collection (core biopsies) and processing, for primary lung and lung and liver metastases using core biopsies consistently provided the fewest failures and implementing NGS-specific processes improved NGS success rates.

Utilization of consultative molecular tumor board in community setting

6508

Background: Physicians in the community have a broad range of experience using genomics data to inform treatment decisions. They typically have a heavier patient load than found in academic centers and treat a variety of tumor types. Genomic data has been reportedly used less than anticipated, even when results were actionable. Monthly didactic molecular tumor boards have been implemented in a number of cancer centers to try to fill gaps in knowledge. Methods: A weekly virtual consultative molecular tumor board (MTB) was implemented (Mar 2016) at an academic hybrid, multi-site community-based cancer institute to provide rapid molecularly-driven treatment guidance to physicians, augment genomics education, provide supporting documents for off-label use and clinical trials. A baseline survey was performed prior to first MTB. MTB assessments were summarized and provided to treating physician. Data was abstracted from the electronic medical records and clinical trials management system. Descriptive statistics were utilized to summarize utilization of MTB and treatment recommendations. Results: Genomics testing with a large panel (~600 genes) was requested for 809 patients (Jun 2015-Feb 2017). The MTB received 81 requests for review from 32 physicians from 14 locations. Most commonly reviewed disease sites were lung, ovary, pancreatic, colon, breast and head and neck cancers; 37% of reviews requested were for rare tumors. Median time to review request was 15 days from receipt of results. MTB recommendations were followed in 70% of cases, 16% continued current/other therapy, 11% declined rapidly (hospice/died), and 3% of patients decided against recommendations. Forty-four (44) percent were screened for recommended clinical trials; 26% went on study. Conclusions: Implementation of a weekly virtual consultative MTB facilitates molecularly-driven treatment decisions in community setting, especially in rare tumor types and enhances clinical trial accruals.

Collection and utilization of oncology big data

e18134

Background: Big Data includes data sets whose size is beyond the ability of typical database software tools to capture, store, manage, and analyze. We sought to gain insight on ASCO members’ perspectives towards Big Data in Oncology. Methods: The survey was conducted as part of the ASCO Leadership Development Program and was electronically distributed to US-based ASCO members. Survey questions explored demographics, practice type, EHR utilization, perceptions about Big Data, and awareness and understanding of ASCO Big Data initiative, CancerLinQ. Statistical analysis was conducted and the data were reviewed. Results: Seventy percent of survey respondents were physicians (n = 1261); 50% had some knowledge of Big Data but only a minority (38%) planned to champion participation in Big Data initiatives. The most frequently cited barriers to participating in Big Data collection were perceived added burden of data retrieval and submission (68%), privacy breach (49%) and potential misuse of patient data (49%). The most valuable aspect of Big Data for practices was population level information for patient management (75%); the most important factor in choosing a Big Data organization was emphasis on privacy and confidentiality of data (53.5%). Participants from NCI-Designated Cancer Centers were more likely to share oncology patient data. Years in practice was inversely correlated with a willingness to share data. Only 25% of respondents were very likely to share oncology data with Big Data organizations, with participants in oncology practice for > 30 years least likely to share data as compared to those with less than 5 years (p = 0.010) and those 11-20 years in practice (p = 0.004). Physicians in practice for > 20 years were least likely to trust information from Big Data (p = 0.004). A majority of patients had heard about Cancer LinQ mostly from ASCO sources, however only 25% were familiar with the details. Conclusions: Big Data has the great potential to provide physicians with novel insights and enhance care for cancer patients although some concerns and barriers exist. Education, outreach and physician champions are keys in further realizing the promise of Big Data for cancer patients.