Citations

75 Background: Despite progress of immune checkpoint therapies, many cases of non-small cell lung cancer (NSCLC) are refractory or acquire resistance to current therapies. Antibody blockade of semaphorin 4D (SEMA4D, CD100) can overcome resistance mechanisms of immune exclusion and myeloid suppression. Importantly, combinations of anti-SEMA4D with various immunotherapies enhanced T cell infiltration and activity, as well as durable tumor regression in preclinical models. Pepinemab (VX15/2503) is a first-in-class humanized monoclonal antibody targeting SEMA4D. Methods: The CLASSICAL-Lung clinical trial (NCT03268057) evaluates the combination of pepinemab with anti-PD-L1 antibody avelumab to couple beneficial modifications of the immune microenvironment via pepinemab with immune activation via checkpoint inhibition. This ongoing study evaluates the safety, tolerability and efficacy of the combination in patients with advanced (stage IIIB/IV) NSCLC, including immunotherapy-naïve (ION) patients and patients whose tumors progressed during or following immunotherapy (IOF). Results: The combination was well tolerated with no major safety signals identified. Among 29 evaluable IOF patients, two experienced confirmed partial response (PR) with 63% and 52% tumor reduction on study following acquired resistance to prior treatment with pembrolizumab, 15 additional patients experienced stable disease, and at least 5 patients with durable clinical benefit of ≥ 23 weeks. Among 21 evaluable ION patients, 5 experienced PR, clinical benefit ≥ 1 year was observed in 3 patients, and Disease Control Rate was 81%. Analysis of pre- and on-treatment biopsies demonstrated increased CD8+ T cell density correlating with response, reduction or elimination of tumor in 11/13 biopsies from subjects with PR or SD. Conclusions: Interim analysis suggests the combination of pepinemab plus avelumab is well tolerated and shows initial clinical signals of antitumor activity. Updated clinical response data (minimum of 6 mo. follow-up), as well as additional immunophenotyping of both inflammatory and suppressive myeloid cells will be presented. Clinical trial information: NCT03268057.

43 Background: Human leukocyte antigens (HLA) are crucial for cytotoxic T cell responses to cancer. Loss of HLA expression is a mechanism of tumor immune escape and may contribute to resistance to immunotherapy. In patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors we have observed discordant responses between brain metastases and extracranial disease and reported on differential PD-L1 expression and clonal T cell infiltration between paired primary lung lesions and brain metastases. In this project we sought to evaluate whether HLA expression was retained in metastatic NSCLC. Methods: Adult patients with paired primary NSCLC and brain metastases were identified from our institution’s tissue registry. HLA-A cell membrane expression on tumor cells was determined by immunohistochemistry with an anti-HLA-A antibody. Tumors with greater than 10% HLA expression were considered positive. Agreement statistics (κ) and Fisher’s exact test were used for analysis. Results: 51 patients with paired primary NSCLC and brain lesions were identified. The median HLA expression was 20% in the primary tumors (IQR 0-65%) and 10% in the brain metastases (IQR 5-40%). 27 primary tumors and 24 brain metastases were positive for HLA expression. There was disagreement in HLA positivity between paired lesions in 11 patients (22%, 95% CI 12-35%)(κ = 0.57, 95% CI 0.35-0.79)(p = 0.0001). There was no significant difference in the time between the primary tumor and brain metastasis resections in patients with HLA expression disagreement compared to those with HLA expression agreement. None of the patients received immune checkpoint inhibitors for treatment of these lesions. Conclusions: We found significant differences in HLA-A expression on tumor cells in nearly one quarter of paired primary lung cancers and brain metastases. Differences in HLA expression may help explain the discrepancies in response to immune checkpoint inhibitors at different sites of disease and warrants further study.

144 Background: Small cell lung cancer (SCLC) is highly responsive to cytotoxic therapy and can be cured in early stages of disease. In this setting, disparities in patient refusal despite provider recommendations are unknown. Methods: All incident limited stage (LS)-SCLC cases from the National Cancer Database were identified from 2004 to 2014. Logistic regression was used to determine factors associated with treatment refusal. Results: 65,664 patients (pts) were identified with LS-SCLC: 44% male, median age 68 years. 3.4% of pts refused radiation and 3.8% refused chemotherapy. The proportion of chemotherapy refusal increased over the study period: 3% in 2004-2006 compared to 5.4% in 2013-2014 [odds ratio (OR) 1.76, 95% confidence interval (CI) 1.55-2.01]; this was not observed for radiation. In multivariate analysis, women were more likely to refuse radiation (OR 1.18, 95% CI 1.06-1.32) and chemotherapy (OR 1.30, 95% CI 1.19-1.43) than men. Women who accepted treatment had higher overall survival compared to those who refused radiation (19.8 vs 5.2 months) or chemotherapy (17.4 vs 3.9 months) (both p < 0.001). Hispanic, Black, and Asian pts were not more likely to refuse treatment than White pts. Older pts were more likely to refuse radiation (OR 1.09 per year, 95% CI 1.08-1.09) and chemotherapy (OR 1.10 per year, 95% CI 1.09-1.11). Charlson comorbidity index (CCI) of 2 was associated with more frequent treatment refusal compared to CCI of 0 (radiation OR 1.96, 95% CI 1.68-2.30; chemotherapy OR 1.54, 95% CI 1.34-1.76). Medicaid as primary insurance predicted a higher risk of refusal compared to private insurance for radiation (OR 2.36, 95% CI 1.81-3.07) and chemotherapy (OR 2.23, 95% CI 1.78-2.80). Treatment at an academic facility predicted a lower risk of radiation refusal (OR 0.71, 95% CI 0.59-0.84) but not chemotherapy refusal. Conclusions: Female sex, comorbidities, and Medicaid insurance were predictors of treatment refusal in LS-SCLC, suggesting that socioeconomic and sex disparities may affect treatment decisions in this life-threatening disease. Further research to identify reasons for refusal via patient and provider interviews can improve care of vulnerable populations with potentially curable cancer.

Abstract Abstract 4259 Background: Patients with defective DNA repair mechanisms are at higher risk for developing cancer. BRCA1 and BRCA2 are proteins that coordinate DNA repair by homologous-recombination. Mutations in BRCA1 and BRCA2 are associated with an increased risk of cancers, namely breast, ovarian, and pancreatic, thus making these patients more likely to receive chemo- or radiation therapy. Additionally, secondary malignancies can be an unfortunate consequence of treatment with chemo- or radiation therapy. Although tumors with defective DNA repair mechanisms may be more susceptible to therapy, we hypothesized that patients with BRCA mutations that underwent chemotherapy or radiotherapy for treatment of breast or ovarian cancer would be at a higher risk of developing therapy- related myelodysplastic syndrome (t-MDS) or acute myeloid leukemia (t-AML). Methods: We accessed patient data from our institution maintained by the Mayo Clinic Life Sciences System using the Data Discovery and Query Builder. We searched for patients whose notes and diagnoses included the terms BRCA, and breast or ovarian cancer, and myelodysplastic syndrome or acute myeloid leukemia. We searched our records from 1/1/1995–7/15/2011. We cross-referenced our findings to a database of patients who were diagnosed with a BRCA mutation through our department of medical genetics. Secondly, we searched the National Cancer Institute's Surveillance Epidemiology and End Results (SEER) database for female patients who developed AML after the diagnosis or breast or ovarian cancer from the years 1973–2008 using ratio and multiple primaries standardized incidence ratios sessions. Results: We identified 220 patients with a BRCA mutation, of which 73 received chemotherapy. There were two patients with BRCA mutations that developed t-AML following treatment for breast cancer, and one patient with a BRCA mutation who developed t-AML following treatment for ovarian cancer (Table 1). Manual data extraction from the list of patients with BRCA mutations diagnosed at our institution only identified one of these patients, as two obtained their testing through another institution. We did not identify any patients with BRCA mutations who developed t-MDS following treatment for breast or ovarian cancer. The latency between diagnosis of the primary cancer and the diagnosis or t-AML was 24 months for two patients, and 22 years for the other. From our search of the SEER database we found 649 patients who developed AML after a diagnosis of breast cancer, and 98 patients who developed AML after a diagnosis of ovarian cancer. These patients represented 0.1% and 0.2% of evaluable breast and ovarian cases respectively. The latency from time of diagnosis of the primary cancer to that of AML was 58 months (29–113 interquartile range) for breast cancer and 47 months (29–91) for ovarian cancer. The observed number of patients with AML following the diagnosis of breast or ovarian cancer was 1.68 and 4.74 times higher than that expected of the general population, respectively (both p<0.05). We were not able to stratify patients based on the presence of a BRCA mutation due to the limitations of the database. Conclusions: We identified three patients with confirmed BRCA mutations and t-AML at our institution. Although this does not represent a strong association, patients at risk for BRCA mutations do not always undergo testing. Thus, our results may not fully represent whether there is an association with BRCA mutations and t-MDS or t-AML. Using the SEER database, we confirmed that patients who were diagnosed with breast or ovarian cancer subsequently had a higher incidence of AML than the general population, although we were not able to test if patients with BRCA mutations are at higher risk of t-MDS or t-AML. Disclosures: No relevant conflicts of interest to declare.

Abstract Introduction Cancer associated thrombosis is the second most common cause of mortality among patients with malignancy. While preventive low molecular weight heparin is effective in preventing venous thromboembolism (VTE), lack of agreement on risk stratification methods have precluded its widespread use. The most studied VTE prediction tool is the Khorana score, but it does not capitalize on the use of novel biomarkers. We aimed to evaluate the value of biomarkers in prediction of VTE among patients with active cancer receiving outpatient chemotherapy. Methods We prospectively collected pre-chemotherapy blood samples on adult patients with solid organ tumors. We excluded patients on anticoagulation and those unable or unwilling to consent. Along with demographics and cancer specific variables, potential novel predictors of VTE were selected a priori and included: coated platelets (CP), D dimer (DD), protein C (PC), fibrin monomers (FM), mean platelet volume (MPV), platelet microparticles (PMP). Type of cancer was grouped according to the Khorana score (Very high risk: Stomach, Pancreas; High risk: Lung, Gynecologic, Bladder, Testicular; Low/Moderate risk: all other sites). VTE events were adjudicated at 6 months of follow up. For all statistical analyses we used SAS 9.2 (SAS Institute Inc., Cary, NC, USA). A p value of <0.05 was considered as statistically significant. We only performed univariate analysis at this point of our study due to the stage of recruitment and period of follow-up. Results We have recruited a total of 101 patients (mean + standard deviation age 59.7+11 years, 68% female), 20% of the 75 patients with complete follow up have developed a VTE. We found no significant association between the incidence of VTE and age, gender, or baseline Khorana score. The variables associated with incident VTE were: BMI (odds ratio [OR] 8.5 for BMI<25 kg/m2, p<0.001), type of cancer (OR 14.5 for very high risk compared to high risk, p<0.01) and protein C (OR 3.3 for protein C<110ug/mL the lowest tertile, p=0.05). The rest of the tested potential biomarkers did not have a strong association with incident VTE in the current analysis (Table). Twenty percent of the patients were on aspirin and none of them developed VTE compared to 23% of those not taking aspirin who developed VTE, (p=0.2). Conclusion Pre Chemotherapy level of Protein C is a potential biomarker for the incidence of VTE. We confirm that cancer type and BMI are predictors of cancer associated VTE. In addition, lower BMI is not usually recognized as a potential predictor of VTE in other prediction scores. We were underpowered to detect differences in the VTE incidence according to Khorana score. Finally, aspirin may have a role in outpatient cancer associated VTE prevention and deserves further investigation. Disclosures: Tafur: Stago Diagnostica: Research Funding. Rathbun:Stago Diagnostica: Research Funding.

Abstract Background Platinum chemotherapy (carboplatin or cisplatin) + etoposide is the current 1L standard of care treatment (tx) for ES-SCLC. Despite high response rates on this regimen, patient (pt) outcomes remain poor. IMpower133 is a Ph1/3, double-blind, randomized, placebo (PBO)-controlled trial that is evaluating the efficacy and safety of adding PD-L1-inhibitor atezo or PBO to 1L tx for ES-SCLC (NCT02763579). Methods IMpower133 enrolled pts without prior tx for ES-SCLC; PD-L1 testing was not required. Pts were randomized 1:1 to four 21-day cycles of carboplatin (AUC 5 mg/mL/min IV, Day 1) + etoposide (100 mg/m2 IV, Days 1-3) with either atezo (1200 mg IV, Day 1) or PBO, then maintenance atezo or PBO until PD, unacceptable toxicity or loss of clinical benefit. Maintenance prophylactic cranial irradiation (PCI) was allowed per protocol, while definitive thoracic radiation (TR) was not. Coprimary endpoints were OS and investigator-assessed PFS. A key exploratory endpoint was outcomes by blood tumor mutation burden (bTMB) analyzed at prespecified cutoffs (≥ 16 vs &lt; 16 and ≥ 10 vs &lt; 10 mutations/Mb). Results Adding atezo to carboplatin + etoposide demonstrated significant improvement in efficacy (OS and PFS) in 1L ES-SCLC (Table). Survival benefits were consistent across key subgroups and prespecified bTMB cutoffs. 44 pts (22 in each arm [per ITT]) received PCI, and 7 (3 in atezo arm, 4 in PBO arm) received TR. Gr 3-4 tx-related adverse events (AEs) occurred in 57% of atezo pts vs 56% of PBO pts; serious tx-related AEs occurred in 23% vs 19%. Occurrence of CNS-related AEs in pts who had PCI are shown in the Table. Conclusion Adding atezo to carboplatin + etoposide resulted in a significant improvement in survival in this 1L ES-SCLC all-comers population. There were no new safety signals, including in pts who received PCI or TR. Atezo + carboplatin + etoposide may represent a new standard regimen for untreated ES-SCLC. Atezo + carboplatin + etoposiden = 201PBO + carboplatin + etoposiden = 202Co-primary endpointsMedian OS, mo12.310.3HR 0.70 (95% CI: 0.54, 0.91); P = 0.0069Median PFS, mo5.24.3HR 0.77 (95% CI: 0.62, 0.96) P = 0.017Secondary efficacy endpointsInvestigator-assessed confirmed ORR, %60.264.4Median DoR, mo4.23.9CNS-related AEs in pts who had PCI, n (%)a,bn = 23n = 21Headache8 (34.8)4 (19.0)Asthenia5 (21.7)2 (9.5)Dizziness2 (8.7)0Insomnia2 (8.7)1 (4.8)Fall2 (8.7)1 (4.8)aSafety population; safety analyses conducted according to tx received (1 pt randomized to the control arm received a dose of atezo).bIncludes AEs occurring in ≥ 2 patients. Citation Format: Aaron S. Mansfield, Stephen V. Liu, Aleksandra Szczęsna, Libor Havel, Maciej Kzrakowski, Maximilian J. Hochmair, Florian Huemer, György Losonczy, Melissa L. Johnson, Makoto Nishio, Martin Reck, Tony S. Mok, Sivuonthanh Lam, David S. Shames, Juan Liu, Fairooz Kabbinavar, Alan Sandler, Leora Horn. IMpower133: Primary efficacy and safety + CNS-related adverse events in a Ph1/3 study of first-line (1L) atezolizumab (atezo) + carboplatin + etoposide in extensive-stage SCLC (ES-SCLC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr CT199.

Abstract Lung cancer is the deadliest and most frequently diagnosed type of tumor worldwide, with 1.6 million deaths reported annually. Non-small cell lung cancer (NSCLC) represents 85% of all lung cancer cases and carries a poor 5-year survival rate below 15%. Prognoses remain dismal due to the large number of patients diagnosed with advanced stage disease and the development of resistance to current therapies. A better understanding of acquired drug resistance will help to circumvent the progression of lung cancer and make significant strides in improving NSCLC patient treatment. Our recent work demonstrates that dopamine and cAMP-regulated phosphoprotein, Mr 32000 (DARPP-32) and its amino-terminally truncated splice variant (t-DARPP) promote lung tumor growth in orthotopic mouse models. IHC staining of 62 human lung adenocarcinoma tissues showed that t-DARPP expression is elevated with increasing tumor (T) staging score, which represents the size of the primary tumor and whether it has grown into nearby areas, as a metric of tumor progression and growth. Correspondingly, a computational biology analysis of 513 lung adenocarcinoma patients revealed upregulation of t-DARPP isoform expression correlates with advanced T stage and is associated with poor overall survival. We identified a novel physical interaction between DARPP-32 and inhibitory kappa B kinase-α (IKKα) that promotes NSCLC cell migration through activation of non-canonical NF-κB2 signaling. It has been shown that DARPP-32 overexpression inhibits gefitinib-induced apoptosis in gastric cancer and t-DARPP contributes to the trastuzumab resistance phenotype in breast cancer. Our recent findings suggest DARPP-32 and t-DARPP overexpression in NSCLC promotes resistance to specific molecular targeted inhibitors by enabling tumor cells to evade apoptosis via Akt- and Erk-dependent cell survival mechanisms. Current ongoing studies are focused on manipulating expression of DARPP-32 isoforms in lung tumor cells to prevent resistance to targeted therapies in NSCLC patients. Citation Format: SK Kayum Alam, Matteo Astone, Ping Liu, Li Wang, Abbygail M. Coyle, Erin N. Dankert, Dane K. Hoffman, Stephanie R. Hall, Wei Zhang, Rui Kuang, Anja C. Roden, Aaron S. Mansfield, Luke H. Hoeppner. Molecular mechanisms of non-small cell lung cancer growth and drug resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 873.

Abstract Introduction: While most small cell lung cancer (SCLC) patients die within a few months, a sub-group of patients survive for many years. Factors determining long-term survivorship remain largely unknown. We present the first comprehensive comparative genomic and tumor microenvironment analyses of small cell lung cancer (SCLC) between patients with long term (LTS) and expected (EXS) survival times. Methods: We compared surgically resected tumors of 23 LTS (survival &gt; 4 years) and 18 EXS (survival ≤ 2 years). There were no differences in clinical variables including TNM staging and curative versus non-curative intend surgery between the groups. Gene expression profiling was performed by microarrays and tumor microenvironment analyses were by IHC of prominent immune related markers. Results: Immune related genes and pathways represented the majority of the differentially overexpressed genes in LTS compared to the EXS. The differences in the immunological tumor-microenvironment were confirmed by quantitative immuno-staining. Increased numbers of tumor infiltrating and associated lymphocytes were present throughout tumors of LTS. Several differentiating patterns of enhanced anti-tumor immunity were identified. While some areas of LTS tumors also harbored higher numbers of suppressive immune cells (monocytes, regulatory lymphocytes, and macrophages), ratios of these suppressive cells to CD3+ lymphocytes were generally lower in LTS tumors indicating a more tumor suppressive microenvironment. Conclusions: Our data demonstrate that long-term survivorship of SCLC patients is strongly influenced by the presence of anti-tumor immune cells in the tumor microenvironment. Characterization of the anti-tumor immune responses may identify opportunities for individualized immunotherapies for SCLC. Citation Format: Farhad Kosari, Prasuna Muppa, Simone Terra, Anurag Sharma, Aaron Mansfield, Marie Christine Aubry, Nafiseh Janaki, Aqsa Nasir, Tobias Peikert. Anti-tumor immunity is a key determinant of small cell lung cancer survivorship [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4577.

Abstract Introduction: IL-15 has been used in clinical trials with PD-1 antibody to overcome cancer resistance to PD-1 monotherapy in cancer patients. However, it is not clear what T cell population would be expanded to mediate tumor control in IL-15 in combination with PD-1 antibody. We recently found CX3CR1+ cytotoxic CD8+ T cells, which increased in responders compared with non-responders after anti-PD-1 in melanoma patients, demonstrated an increased transcription of CD122 (IL2RB). Since CD122 is a subunit of IL-15 receptor, we hypothesized that IL-15 in combination with anti-PD-1 therapy may have the ability to expand CX3CR1+ cytotoxic CD8+ T cells for tumor control. Methods and findings: To examine whether IL-15 plus anti-PD-1 expand CX3CR1+ cytotoxic CD8+ T cells in tumor tissues, we treated tumor-bearing mice with IL-15/IL-15Rα complexes alone or with PD-1 antibody and measured the frequency of CX3CR1+Granzyme B+ CD8+ T cells. As only the combination therapy induced dramatic tumor regression, we found the highest increase of CX3CR1+Granzyme B+ CD8 T cells in tumor tissues in the group treated with IL-15 plus anti-PD-1 (28.87±3.19% vs. 7.91±1.32%, p&lt;0.01) compared with groups treated with either IL-15 alone or anti-PD-1 alone. In contrast to wild type mice, IL-15 plus anti-PD-1 did not suppress tumor growth in CX3CR1-KO mice. To determine whether IL-15 alone can also expand human CX3CR1+ cytotoxic CD8+ T cells, we cultured peripheral blood mononuclear cells (PBMCs) isolated from healthy donors (n = 5) with IL-15 for 24 hours in vitro. We found IL-15 (5 ng/mL) only modestly increased the frequency of CX3CR1+ Granzyme B+ CD8+ T cells than those cultured with no IL-15 (39.9±6.44% vs. 27.56±4.98%). To test whether this increase is due to enhanced proliferation of this subset of CD8+ T cells, we measured the expression of Ki-67, a cell proliferative marker, in CX3CR1+ CD8+ T cells by intranuclear staining. We found IL-15 modestly induced more proliferating CX3CR1+ CD8+ T cells in vitro culture compared with control group without IL-15 (6.25±0.54% vs. 0.91±0.25%). Summary: Taken together, we found IL-15 combined with anti-PD-1 therapy has the ability to significantly expand CX3CR1+ cytotoxic CD8+ T cells, however IL-15 alone may have limited ability to significantly expand this effector cell population. Our studies provide a new rational for combination therapy of IL-15 and PD-1 antibody to overcome cancer resistance for patients who did not respond to initial PD-1 therapy. Citation Format: Henan Zhang, Xin Liu, Siyu Cao, Lingling Chen, Susan Harrington, Ying Li, Aaron Mansfield, Sean Park, Yiyi Yan, Roxana Dronca, Haidong Dong. Expansion of CX3CR1+ cytotoxic CD8+ T cells provides a rationale for IL-15 in combination cancer therapy with PD-1 antibody [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4080.

Abstract Purpose: Delta-like protein 3 (DLL3) is being developed as a predictive biomarker to identify small cell lung cancer (SCLC) patients for treatment with the DLL3-targeting antibody-drug conjugate rovalpituzumab tesirine (Rova-T). Merkel cell carcinoma (MCC) is an aggressive neuroendocrine carcinoma of the skin, associated with Merkel cell polyomavirus. Given the neuroendocrine features of SCLC and MCC, we sought to evaluate DLL3 expression, its association with common clinicopathological factors, and its prognostic role in MCC. Methods: A total of 65 formalin-fixed and paraffin-embedded MCC cases were consecutively identified from an institutional biospecimen repository and were cut at 5 µm. Each case was stained for DLL3 protein with SC16.65 mouse monoclonal antibody at 3 µg/mL with the FLEX+ system (DAKO). Slides were then scanned at 20X with AT2 (Leica/Aperio) and read out for percentage of positive tumor cells with 1+, 2+, and 3+ intensity side-by-side with isotype control (IgG2a). Any subcellular staining pattern (membranous, cytoplasmic or punctate) was counted as overall positive. Polyomavirus status was determined by immunohistochemistry (IHC) using an antibody that binds to MCPyV large T-antigen (sc-136172, Santa Cruz Biotechnology, Inc.). A generalized additive model was used to visualize the trend between common clinical factors and overall survival (OS). A multivariable Cox proportional hazards model was applied to assess the association between DLL3 expression and OS. The association between DLL3 expression and other clinicopathological factors was evaluated with median and t tests. Results: A total of 67 patients were included with a median age of 74 years. Thirty-six (55%) had stage I/II disease; 29 (45%) had stage III/IV disease. Thirty-five (52%) patients died from MCC. Among the 65 cases with successful DLL3 stain, the median H-score of DLL3 expression was 60 (interquartile range: 30-100). Fifty-eight cases (89%) had ≥1% tumor cells positive for DLL3 expression with any intensity, of which the median DLL3 expression was 50% (interquartile range: 25-70%). Thirty-four cases (52%) had ≥50% tumor cells positive for DLL3 expression with any intensity. Older age significantly predicted shorter overall survival (p = 0.03). However, higher AJCC stage did not predict shorter OS (p = 0.3). H-score of DLL3 expression was not associated with AJCC stages (p = 0.4). However, higher H-score of DLL3 expression was associated with higher polyomavirus nuclear expression (p = 0.02) when it was dichotomized to 0/1+ and 2+/3+. In a multivariable Cox regression analysis, H-score of DLL3 expression did not predict OS of patients with MCC (p = 0.8). Conclusions: DLL3 overexpression is very common in MCC by IHC. DLL3 expression was significantly associated with Merkel cell polyomavirus expression but was not prognostic for OS in this cohort. The high levels of DLL3 expression in a subset of MCC may potentially be used to select patients to receive Rova-T. Citation Format: Hao Xie, Kumiko Isse, Yan Sun, Johanna Ramoth, Dorothy M. French, Laura R. Saunders, Aaron S. Mansfield. Delta-like protein 3 expression in Merkel cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3171.