American Association For Cancer Research Research Articles

Abstract 735: Sensitive detection of ctDNA in early stage non-small cell lung cancer patients with a personalized sequencing assay

Abstract Introduction Identification of minimal residual disease (MRD) following curative intervention of localized non-small cell lung cancer (NSCLC) holds promise for identifying patients who are at higher risk of relapse and who would benefit from adjuvant treatment. Current routine clinical practice involves serial radiographic imaging following surgery to detect macroscopic disease. Liquid biopsy can identify patients who have MRD without macroscopic disease. Currently available assays have only identified circulating tumor DNA (ctDNA) in a limited number of cases with early stage NSCLC. More sensitive methods are needed to accurately identify the majority of patients who will relapse. Here we evaluate the performance of InVision®MRD, a personalized sequencing assay for plasma cell-free DNA, for detection of ctDNA in patients with early-stage NSCLC undergoing treatment with curative intent. Methods InVision®MRD is a highly sensitive in vitro diagnostic assay, currently available for research use only (RUO), that can detect the presence of tumor DNA traces in cell-free DNA from plasma samples of cancer patients. InVision®MRD identifies tumor-specific variants from exome sequencing of tumor tissue and tracks them in plasma specimens by multiplex PCR and high-depth next-generation sequencing. We evaluated the detection of ctDNA in plasma samples collected from the LUng cancer - CIrculating tumor DNA (LUCID) study, which collected plasma samples from 100 patients with NSCLC stages I-III who underwent radical treatment with curative intent, either surgery or radiotherapy ± chemotherapy. Of patients in the LUCID study, 60% had stage I NSCLC and 40% patients had stage II/III disease, according to TNM 7th edition. Results To evaluate the InVision®MRD assay, a subset of samples from the LUCID study were analyzed. Samples were collected before and after surgery and chemo-radiotherapy from patients with early-stage NSCLC. Using multiplexed analysis of 48 patient-specific variants and high-depth sequencing, ctDNA was detected in 50% of pre-treatment samples analyzed from the first set of 18 patients, at ctDNA fractions ranging from 20 ppm (equivalent to 0.002%) to 19576 ppm (equivalent to 1.958%). Conclusions These findings highlight an opportunity to improve ctDNA detection for early stage NSCLC using a patient-specific plasma sequencing assay. Initial detection rates have reached 50% for patients with early-stage disease prior to treatment, including detection of ctDNA to levels as low as a few parts per million. Together with further data to be presented, this suggests a possible route to improving treatment for early stage NSCLC by detection of residual disease post treatment and for monitoring for early detection of relapse. Citation Format: Katrin Heider, Davina Gale, Andrea Ruiz-Valdepenas, Giovanni Marsico, Garima Sharma, Malcolm Perry, Robert Osborne, Karen Howarth, Tadd Lazarus, Viona Rundell, Jelena Belic, Jerome Wulff, Susan Harden, Doris M. Rassl, Robert C. Rintoul, Nitzan Rosenfeld. Sensitive detection of ctDNA in early stage non-small cell lung cancer patients with a personalized sequencing assay [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 735.

Abstract 1327: Genome-wide loss-of-heterozygosity calling from low-pass sequencing of ligation-mediated whole genome amplified DNA in single tumor cells

Abstract Background: Genome-wide analysis of Loss-of-heterozygosity (LoH) has been shown to be important in several contexts, including the assessment of BRCAness signature, a marker of homologous repair deficiency in several cancer types associated with efficacy of platinum therapy and PARP inhibitors. Methods to infer genome-wide LoH status require high-coverage whole-genome sequencing or low-pass sequencing of a large number of single cells. We present a method to infer genome-wide LoH status of as low as one single-cell from low-pass sequencing of libraries produced by Ampli1™ LowPass kit, at a fraction of the cost of current NGS-based methods. Methods: We used low-pass sequencing data generated from 294 single Circulating Tumor Cells (CTCs) from prostate, multiple myeloma and lung cancer patients, and from 24 single cells from 8 cell lines. DNA amplified with Ampli1™ WGA kit (based on ligation-mediated PCR) combined with Ampli1 LowPass kit for Illumina generates a reduced genome representation, whereby a significant fraction of fragments is expected to be covered by more than one read. We reasoned that, even at a mean coverage insufficient for variant calling (≤0.05x), a statistically significant decrease in loci showing 2 alleles should be expected in LoH regions. A pileup of mapped sequences at dbSNP loci was obtained (minor allele frequency ≥5%; read depth ≥2). Mono- and bi-allelic loci (pileup containing 1 or 2 different alleles, respectively), were counted in each uniform copy-number (CN) genome segment within a single chromosome arm. Regions with CN equal to cell ploidy were used as internal reference. A one-sided Fisher test on biallelic sites depletion (multiplicity corrected p-value threshold of 0.05) was used to call LoH on each region. Results: Compared with libraries obtained using random shearing, Ampli1 LowPass single-cell libraries showed an increase in mean coverage of covered genome from 1.1 to 1.9 starting from the same number of reads (1M) thus increasing the number of loci for LoH calling. Analysis of 10 patients' single cells with known LoH events, corresponding to regions at copy number 1, detected 1006 events out of 1142. Several (492) copy neutral LoH events or LoHs in copy gained regions were also detected. Representatives of such events were verified successfully by high pass sequencing of the same library. Performance was confirmed by analysis of 24 single cells from cell lines whose LoH status was inferred from CCLE database. Conclusions: This innovative method leverages reduced genome representation inherent in Ampli1 LowPass libraries, to compute genome-wide LoH along with copy number (CN) status, meeting the need for LoH/CN analysis at single-cell level on rare-cells such as CTCs. Citation Format: Alberto Ferrarini, Claudio Forcato, Marianna Garonzi, Genny Buson, Massimiliano Pellicano, Francesca Fontana, Nicolò Manaresi. Genome-wide loss-of-heterozygosity calling from low-pass sequencing of ligation-mediated whole genome amplified DNA in single tumor cells [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1327.

Abstract 4520: Preclinical characterization of ALG-031048, a novel STING agonist with potent anti-tumor activity in mice

Abstract Introduction: The STING (STimulator of INterferon Genes) protein has been identified as an attractive target for cancer immunotherapy due to its central role in inducing T-cell mediated tumor control. Here we describe the preclinical characteristics of the novel STING agonist, ALG-031048, that demonstrates potent anti-tumoral activity in in vivo mouse models. Methods: Binding to STING was determined using differential scanning fluorimetry (DSF). Cellular activation and cytokine release were assessed using HEK 293T reporter cell lines as well as THP-1 and Raw264.7 cells. In vitro stability was measured in the presence of snake venom phosphodiesterase (SVPD). In vivo anti-tumoral activity was tested in the syngeneic CT26 and B16F10 mouse model. Results: ALG-031048 demonstrated potent binding to the STING R232 protein in the DSF assay with a mean Kd of 3.12 ± 0.12 μM and a thermal shift of 12.8 °C, similar to the natural ligand 2'3'cGAMP (Kd of 2.75 ± 0.78 μM, shift of 16.7°C) or the clinical stage STING agonist, ADU-S100 (Kd of 4.61 ± 0.42 μM, shift of 10.7°C). In the HEK 293T R232 reporter assay, ALG-031048 was more potent than ADU-S100 with a mean EC50 of 55.9 ± 24.1 and 30.3 ± 17.1 nM for the IFNβ and IRF reporter, respectively, compared to ADU-S100 with EC50 values of 233 ± 179 and 88.7 ± 35.6 nM. Cellular activation was confirmed in THP-1 and Raw264.7 cells where ALG-031048 induced the secretion of IFNβ and CXCL-10. ALG-031048 was highly stable and exhibited no degradation for up to 24 hours in the presence of SVPD while ADU-S100 and 2'3'cGAMP were quickly degraded with less than 10% remaining after 24 hours. In vivo, 3 intra-tumoral (IT) doses of 100 μg ALG-031048 q3d caused complete tumor regression in 90% of animals in the CT26 mouse colon carcinoma model, while treatment with 25 μg ALG-031048 IT 3xq3d resulted in survival of 60 % of animals. In contrast, only 10 and 44% of mice showed complete tumor regression when treated IT 3xq3d with 25 or 100 μg ADU-S100, respectively. To study whether induction of a protective immune response was associated with STING agonist treatment, surviving animals were re-challenged with a second inoculation of CT26 cells on the contra-lateral side. All naïve animals formed tumors within 23-30 days. In contrast, 5 out of 5 and 8 out of 9 mice pre-treated with 25 or 100 μg ALG-031048, respectively, did not exhibit any significant tumor growth over 40 days, while 3 out of 4 animals pretreated with 100 μg ADU-S100 showed complete tumor suppression. Initial results using the B16F10 mouse melanoma model confirmed the potent anti-tumoral activity of ALG-031048 observed in the CT26 model. Summary: ALG-031048 is a novel STING agonist combining high stability and excellent in vitro potency. In preclinical models, ALG-031048 demonstrated potent and long-lasting tumor suppression. ALG-031048 is advancing in toxicology studies for further profiling as a clinical candidate. Citation Format: Andreas Jekle, Santosh Thatikonda, Sarah Stevens, Caroline Williams, April Kinkade, Suping Ren, Ruchika Jaisinghani, Qingling Zhang, Dinah Misner, Antitsa Stoycheva, Jerome Deval, Sucheta Mukherjee, Francois Gonzalvez, Sushmita Chanda, David B. Smith, Julian A. Symons, Lawrence M. Blatt, Leonid Beigelman. Preclinical characterization of ALG-031048, a novel STING agonist with potent anti-tumor activity in mice [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4520.

Abstract 2276: Preclinical development and mechanism of action studies of NM21-1480, a PD-L1/4-1BB/HSA trispecific MATCH3 therapeutic clinical candidate

Abstract Antagonistic molecules targeting the PD-1/PD-L1 axis have shown excellent activity in the clinic. However the majority of patients do not respond to the therapy due to multifaceted reasons implicating a non-effective activation of the immune system in those patients. The co-stimulatory molecule 4-1BB has been shown to be a key signalling component of T cells and the combination of 4-1BB activation and PD-1/PD-L1 antagonism has been shown to be highly active in preclinical models. Systemic application of first generation anti-4-1BB antibodies however have resulted in dose limiting hepatic toxicities. We have generated a novel 4-1BB/PD-L1/HSA trispecific MATCH3 immunomodulatory drug candidate (NM21-1480) that agonizes 4-1BB conditionally upon PD-L1 binding / blockade. Here we show the preclinical development data package for NM21-1480 demonstrating exquisite tumour-specific T cell activation associated with both 4-1BB agonism as well as PD-L1/PD-1 antagonism. We also demonstrate tumour specific localisation and accumulation of NM21-1480 in a mouse xenograft model. We report on the toxicity profile and pharmacokinetic properties of the molecule in non-human primates. We plan to initiate a first-in-human clinical study in the second half of 2020 to determine the safety, tolerability and first signs of clinical activity of the molecule. Citation Format: Daniel Snell, Tea Gunde, Stefan Warmuth, Peter Lichtlen, Julia Tietz, Matthias Brock, Alexandre Simonin, Christian Hess, Weinert Christopher, Robin Heiz, Naomi Flueckiger, Julia Zeberer, Dania Diem, Dana Mahler, Diego Morenzoni, Belinda Wickihalder, Simone Muntwiler, Antonia Poelderl, Benjamin Kuettner, Sandro Wagen, Sebastian Meyer, Timothy Egan, David Urech. Preclinical development and mechanism of action studies of NM21-1480, a PD-L1/4-1BB/HSA trispecific MATCH3 therapeutic clinical candidate [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2276.

Abstract 4188: The ERK1/2 inhibitor, JSI-1187, demonstrates preclinical efficacy in tumor models with MAPK pathway mutations

Abstract The MAPK pathway, via RAS-RAF-MEK-ERK, is one of the most important pathways for relaying signals from cell surface receptors for cell survival, proliferation, and differentiation. Many mutations were identified in this pathway and intensely studied, resulting in a number of therapeutics for various cancers (for example, EGFR mutant inhibitors, BRAF V600 mutant inhibitors and MEK inhibitors). A common mechanism of resistance to these therapeutics has emerged to be the reactivation of ERK1/2, the terminal kinase in this pathway. Inhibition of ERK1/2 activity could therefore be instrumental in various mutations in the MAPK pathway, either as single therapy or in combination with other compounds. Through internal effort, we discovered a novel ERK1/2 small molecule inhibitor, JSI-1187, which reversibly inhibited ERK2 with a Ki of less than 1 nM. In anti-proliferation assays, JSI-1187 had IC50 of 37 NM in the BRAF V600E melanonma cell line A375 and 48 NM in the BRAF V600E CRC cell line COLO 205. For cell lines with KRAS mutations, JSI-1187 had anti-proliferation IC50 of 130 NM in the CRC cell line HCT116 and 150 NM in the NSCLC cell line Calu-6. Furthermore, JSI-1187 inhibited the phosphorylation of the ERK substrate RSK1 in these cell lines. We also tested JSI-1187 using mouse CDX models. JSI-1187 showed tumor growth inhibition in a dose dependent manner in the xenograft models of A375, COLO 205, HCT116, and Calu-6. In the COLO 205 model, JSI-1187 caused tumor regression at the dose of 20 mg/kg QD. In the Calu-6 model, BID and QD dosing schedules were compared, with the result that 10 mg/kg BID and 20 mg/kg QD inhibit xenograft tumor growth similarly. Moreover, in the HCT116 model, JSI-1187 concentrations in plasma and tumor were measured, as well as phopspho-RSK1 levels in tumor. A PK/PD correlation was observed. In addition to being tested alone, JSI-1187 was also tested in combination with the BRAF inhibitor dabrafenib in the COLO 205 xenograft model. JSI-1187 and dabrafenib demonstrated synergistic efficacy in tumor growth inhibition. In conclusion, JSI-1187 can be potentially used for treatment of cancer with MAPK pathway mutations. Citation Format: Ao Li, Shanzhong Jian, Xia Yuan, Fengge Song, Shengwei Yang, Chengyi Du, Yue Tao, Liucheng Wang, Mingming Pan, Penghui Dong, jiajun Zhou, Zhengkang Ge, Qinghong Zhu, Wenshan Hao, Wen Xu, Jintao Zhang, Qun Li, Shaohui Wang. The ERK1/2 inhibitor, JSI-1187, demonstrates preclinical efficacy in tumor models with MAPK pathway mutations [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4188.

Abstract 1031: Liver metastasis mediated control of distant tumor-specific immunity and response to checkpoint immunotherapy

Abstract Background: PD-1 blockade achieves objective responses in 30-55% of melanoma patients, but there remain 45-70% of patients where it is ineffective, either as a single agent or in combination with anti-CTLA-4. We lack salvage strategies for this population and a mechanistic understanding of checkpoint inhibitor refractory melanoma. Previously, we reported that patients with liver metastasis have worse response rate and survival than those without liver metastasis despite treatment with checkpoint immunotherapy and cutaneous biopsy samples from patients with liver metastasis have significantly less activated tumor-infiltrating lymphocytes (TILs) than those with metastasis elsewhere. However, the mechanism of how liver metastasis influences the systemic antitumor immune response in this context is unclear. We hypothesize that tumor antigen-specific tolerance can be induced when metastatic cells invade the liver, resulting in reduced systemic antitumor immunity and immunotherapy resistance. Results: In this study, we developed a syngeneic two-site tumor animal model in which tumor cells are injected simultaneously into a subcutaneous (SQ) site and into the liver or other organ sites of wild-type C57BL/6 mice to investigate the unique effects of the intrahepatic tumor on antitumor immunity at a distant site. We found that the presence of tumor or tumor antigen within the liver, as opposed to other organs, reduced systemic antitumor immunity and effector T cell function that is evident at the distant SQ site. Tetramer-based flow cytometry and single cell-RNAseq analysis were done on the TILs within the SQ site and the results suggest that tumor-specific CD8 T cells were preferentially dysfunctional but not sequestered by the liver or clonally deleted, with reduced expression of activation markers and effector cytokines (ICOS, CTLA-4, IFNγ, and TNFα). Collectively, the data suggest a liver and antigen-specific immunoregulatory process mediated by the coordinated activation of Foxp3+ regulatory T cells (with increased expression of CTLA-4 and ICOS) and significant changes in the non-T cell immune composition at the distant SQ site in the liver-tumor bearing mice. Genetic studies using a mouse strain expressing the diphtheria toxin receptor under the control of Foxp3 (Foxp3-DTR) with or without the systemic depletion of Tregs revealed that in the absence of Tregs, there was no difference in the distal immunity against the SQ tumor between mice with and without experimental liver metastasis. Finally, we found that this form of systemic antitumor effector T cell dysfunction was not reversed by anti-PD-1 monotherapy but rather by combination therapy with the deletion (anti-CTLA-4) or destabilization (EZH2 inhibitor) of Tregs, suggesting a distinct regulatory mechanism contributing to anti-PD-1 resistance in the setting of liver metastasis that could be overcome to achieve complete tumor regression at both sites. Conclusion: Our results reveal a novel mechanism of checkpoint inhibitor resistance and suggest a strategy to improve the outcome of immunotherapy treatments for stage IV cancer patients with liver metastasis. Citation Format: James C. Lee, Sadaf Mehdizadeh, Arabella Young, Jennifer Bridge, Ilgiz A. Mufazalov, Adil Daud, Jeffrey A. Bluestone. Liver metastasis mediated control of distant tumor-specific immunity and response to checkpoint immunotherapy [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1031.

Abstract LB-005: Senescence reprogramming in primary tumors initiates prostate cancer metastases

Abstract Recent studies have raised fundamental questions about the possibility that senescence induced either by oncogene activation or chemotherapy, while initially restricting tumorigenesis, may promote cancer metastases at a late stage. This coexisting dichotomy remains unclear due to the lack of evidence on the mechanism by which senescent tumor cells initiate metastases, a fatal and therapeutically challenging hallmark of cancer. Here, we put forward the hypothesis that the genetic make-up of the tumors accounts for the distinct outcomes of senescence in cancer. Using different genetically engineered mouse models we demonstrate that inactivation of Timp1 is capable to switch senescence driven by the loss of PTEN and chemotherapy from a tumor-suppressive into a tumor-promoting response that drives metastases. Of note, depletion of senescent tumor cells by the senolytic compound ABT263 or p53 genetic inactivation significantly impeded metastases formation in Ptenpc-/-; Timp1-/- mice, indicating the involvement of senescence in the metastatic process. Moreover, enhancement of senescence by Docetaxel therapy decreased the proliferation of prostate tumors in Ptenpc-/- mice, whereas it accelerated metastases formation in Ptenpc-/-; Timp1-/- mice. Mechanistically, TIMP1-loss in PTEN-deficient cells reprogrammed the senescent-secretome significantly changing the levels of GDF15, FGF and IGFBP thereby promoting cellular invasion, migration and the up-regulation of a metastases-associated gene signature. In prostate cancer patients, we found that TIMP1 is commonly lost in combination with PTEN in advanced and metastatic prostate tumors and that TIMP1 inactivation is associated with TIMP1 promoter methylation. Altogether, these findings highlight the crucial role of TIMP1 as the gatekeeper that limits the pro-metastatic function of senescent tumor cells, implying that senescence inducing therapies should be administered with caution in patients affected by cancers with TIMP1 inactivation. Citation Format: Ilaria Guccini, Ajinkya Revandkar, Mariantonietta D'Ambrosio, Manuel Colucci, Emiliano Pasquini, Simone Mosole, Martina Troiani, Andrea Alimonti. Senescence reprogramming in primary tumors initiates prostate cancer metastases [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-005.

Abstract 6483: Performance characteristics of the first FDA-cleared droplet digital PCR (ddPCR) IVD assay for monitoring chronic myelogenous leukemia

Abstract Introduction: Chronic myelogenous leukemia (CML) is a cancer of myeloid cells that accounts for 15-20% of all cases of leukemia. Most CML cases are caused by a translocation between chromosomes 9 and 22 which creates an abnormal fusion gene, BCR-ABL1. The BCR-ABL1 gene expresses an abnormal tyrosine kinase protein that causes unregulated growth and survival of granulocytes. CML is treated with TKIs (tyrosine kinase inhibitors). The concentration of BCR-ABL1 transcript RNA is a marker of disease progression and the effectiveness of TKI treatment.Bio-Rad's QXDx BCR-ABL %IS Kit is an in vitro diagnostic test for the quantification of BCR-ABL transcripts in total RNA from whole blood from individuals diagnosed with Chronic Myeloid Leukemia expressing BCR-ABL1 Fusion transcript types e13a2 and/or e14a2. Materials and Methods: The QXDx BCR-ABL %IS Kit is a newly launched FDA-approved ddPCR assay from Bio-Rad Laboratories. Reproducibility of results across multiple sites, days, instruments, and users was evaluated using contrived samples consisting of BCR-ABL positive patient samples mixed with negative patient samples. Clinical performance of the kit was evaluated on patient samples, and compared to an already existing FDA-approved test. Clinical performance was further tested in the College of American Pathologists challenge. Results: The reproducibility study noted negligible contributions to variance from site, instrument, day, and user for samples spanning from MR 0.7-4.2. The mean between the QXDx™ BCR-ABL %IS Kit and the comparator test using a Bland-Altman analysis was 0.16 MR units. The limits of agreement between the two methods indicate that the difference between the two methods should be between 0.47 and -0.14 MR on 95% of tested samples. The QXDx™ BCR-ABL %IS assay demonstrated excellent correlation with the comparator test using a Weighted Deming regression with a Pearson R correlation coefficient of 0.99, slope 1.037 and intercept of 0.1084. College of American Pathologists(CAP) challenges (n=6) were conducted from 2016-2019, QXDx™ BCR-ABL %IS system results were regressed vs the mean values published by CAP yielded a slope of 1 and R2=0.99. Conclusions: The QXDx™ BCR-ABL %IS Kit on the QXDx™ AutoDG™ Droplet Digital PCR System can be used safely and effectively in the laboratory on t(9;22) positive CML patients for monitoring of treatment with Tyrosine Kinase Inhibitors (TKIs). Citation Format: Dawne N. Shelton, Prasanthi Bhagavatula, Nathan Sepulveda, Lan Beppu, Jerald Radich. Performance characteristics of the first FDA-cleared droplet digital PCR (ddPCR) IVD assay for monitoring chronic myelogenous leukemia [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6483.

Abstract 4655: Social jetlag and lung cancer incidence in the Alberta's Tomorrow Project: A prospective cohort study

Abstract Objectives: Increased lung cancer risk has been associated with markers of circadian disruption (e.g., overly short or long sleep duration, shift work, circadian gene mutation) in both human and animal studies. However, no study to date has reported the association of social jetlag, the misalignment between the internal clock and the socially required timing of activities, with lung cancer risk. We, therefore, aim to investigate this association in the Alberta's Tomorrow Project (ATP), a population-based prospective cohort study in Alberta, Canada, and the potential effect modification of smoking behavior. Methods: Data on sleep patterns were collected on 19,436 (37.6% male) Albertans aged 35-69 years recruited from 2001-2015. Social jetlag was defined as the absolute difference in waking time between free day and workday that accounted for sleep debt, and categorized into three levels, 0 hours (32.52%), >0-1 hours (32.44%), and >1-4 hours (35.04%). Incident lung cancer cases were determined through linkage with the Alberta Cancer Registry by June 2018. Smoking behavior was defined as never smokers, former smokers, and current smokers. Hazard ratios (HR) were estimated using a series of Cox proportional hazards regressions, adjusting for age, sex, recreational physical activity, sitting time, daily calorie consumption, body mass index, marital status, education, employment status, total household income, smoking and drinking behavior, family history of cancer and preexistence of medical condition. Results: A total of 117 incident lung cancer cases were identified during a medium of 9.6 years of follow-up, yielded 179,028 person-years. In the multivariable-adjusted model, a suggestive significant association between social jetlag and lung cancer risk was observed (>1-4 hours vs. 0: HR=1.67, 95% CI: 0.96-2.92, p for trend = 0.047). The increased lung cancer risk with social jetlag was retrained and heightened among daily smokers (>1-4 hours vs. 0: HR=2.72, 95% CI: 1.26-5.84, p for trend = 0.010), and attended to null among never smokers (>1-4 hours vs. 0: HR=0.98, 95% CI: 0.15-6.54, p for trend = 0.91) and former smokers (>1-4 hours vs. 0: HR=0.76, 95% CI: 0.27-2.12, p for trend = 0.63). Latency multivariate-adjusted model excluding those diagnosed with cancer within 2 years of baseline assessment returned similar results. Conclusion: Increased risk for lung cancer was observed among adults who reported social jetlag, particularly daily smokers. Our findings suggest a potential lung cancer risk associated with switching sleep patterns between work and free days among smokers. Further studies are necessary to confirm these findings and elucidate the biological mechanisms. Citation Format: Benny Rana, Andrew Harper, Grace Shen-Tu, Cheryl Peters, Darren Brenner, Don G. Morris, Lin Yang. Social jetlag and lung cancer incidence in the Alberta's Tomorrow Project: A prospective cohort study [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4655.

Abstract 4719: Novel germline variants in hereditary renal cell carcinoma genes elucidated by whole genome sequencing

Abstract Whole genome sequencing (WGS) may be utilized to reveal novel germline variants in renal cell carcinoma (RCC) syndrome genes that cannot be characterized by conventional methods. Patients who present with clinical manifestations of von Hippel-Lindau (VHL), Hereditary Leiomyomatosis Renal Cell Carcinoma (HLRCC), or Hereditary Papillary Renal Carcinoma (HPRC) are routinely evaluated for germline mutations in the VHL, FH (fumarate hydratase), or MET gene, respectively, by genetic testing. Rarely, either a negative result or an unusual finding is reported. For one patient who exhibited multiple clinical manifestations of VHL in the kidneys, pancreas, cerebellum and epididymis but no germline mutation, karyotype analysis revealed that the patient possessed a germline translocation between chromosomes 1 and 3. Two patients with family history of papillary type 1 RCC but no point mutation in MET were determined to have novel germline duplications in all or part of the MET gene. Finally, seven patients with clinical manifestations of HLRCC including cutaneous and uterine leiomyomas and/or family history of RCC, had negative genetic testing. Paired-end WGS was performed on the DNA derived from peripheral blood leucocytes from these patients. WGS for the patient with VHL manifestations revealed that the breakpoint between chromosomes 1 and 3 falls directly within intron 2 of the VHL gene, leading to a novel disruption of VHL in the germline. Sanger sequencing was performed to determine the exact breakpoints on both chromosomes. We believe this to be the first report of a germline chromosomal translocation resulting in VHL syndrome Likewise, both MET duplications were sequenced across the exact breakpoints, revealing a tandem duplication of 601,339 bp including the entire MET gene in one patient, and a tandem duplication of 101,464 bp including exons 5 through 21 in the other patient. As MET is an oncogene, we theorize that these duplications result in increased expression levels. Finally, 4 of 7 patients with clinical manifestations of HLRCC, including a family of three affected members, were found to possess a novel intronic A to G variant in intron 9 of FH that has not been reported in any human genome databases. This variant introduces a cryptic splice acceptor site and thus adds an extra exon of 57 bp to the transcript, resulting in a premature stop codon and thus a truncated protein of 474 amino acids. A cutaneous leiomyoma from one patient demonstrates loss of heterozygosity, substantiating loss of FH function. In conclusion, germline whole genome sequencing should be considered for patients who exhibit clinical manifestations of a hereditary RCC disease syndrome but for whom no sequence variants are detected by standard genetic testing, in order to improve the detection rate of novel germline variants. Citation Format: Cathy D. Vocke, Christopher J. Ricketts, Daniel R. Crooks, Martin Lang, Laura S. Schmidt, Mayank Tandon, Bao Tran, Christina Fitzsimmons, Pedro J. Batista, Mark W. Ball, W. Marston Linehan. Novel germline variants in hereditary renal cell carcinoma genes elucidated by whole genome sequencing [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4719.

Abstract 452: Simultaneous identification of clinically relevant RB1 Mutations and somatic copy number alterations in the Aqueous Humor of retinoblastoma eyes

Abstract Purpose: Retinoblastoma (RB) tumorigenesis is associated with mutations in the RB1 tumour suppressor gene and somatic copy number alterations (SCNAs). Both alleles of the RB1 gene must be mutated for tumour development. The initial RB1 mutation may be germline or somatic. However, the somatic RB1 mutation was only detectable in enucleated RB eyes, when the tumour samples were available. The aqueous humor (AH) can serve as a surrogate to direct tumour biopsy and can be done in vivo. In this study, we demonstrate that both whole genome SCNAs and RB1 mutations can be simultaneously assayed in the AH liquid biopsy. Methods: AH from 7 eyes of 6 RB patients and 4 matched tumor samples were included. AH was extracted via clear corneal paracentesis. Cell-free DNA (cfDNA) was isolated and constructed into whole genome libraries. Low pass whole genome sequencing was performed for copy number alteration analysis. The same libraries were further used for capture-based targeted NGS for mutation detection. Custom hybridization capture panel manufactured from Agilent includes the whole gene region of RB1 gene, MCYN gene, and all exons of BCOR and CREBBP genes (Agilent SureSelect Tier1). Captured libraries were sequenced on Illumina platform with paired-end 150 cycles mode. Bioinformatics analysis was performed for single-nucleotide variant (SNV) and loss of heterozygosity (LOH) detection using an in-house pipeline (information available on request). Comparison was also made to clinically available peripheral blood RB1 mutation screening (CHLA, CPM). Results: Somatic whole genome SCNAs were detected in all 4 enucleated tumours. Among 7 evaluated AH samples, 5 contained SCNAs and the remaining 2 were flat without alterations under current analysis setting. Mutational analysis of tumour DNA from 4 enucleated eyes identified 6 RB1 SNVs. Mutation results from AH cfDNA were concordant with results obtained from corresponding tumour tissue and peripheral blood screening, when available. In the 3 AH samples without paired tumours, both RB1 hits were identified with high variant allele frequency (VAF), indicating high tumour fraction in AH of RB eyes. Conclusions: Use of a customized hybridization capture approach facilitated identification of RB1 mutation(s), together with low pass whole genome analysis for SCNA profiling in the same AH liquid biopsy provides the opportunity to identify the genomic and genetic alterations simultaneously for one diseased eye. This refined approach provides the potential to further investigate the spectrum of RB1 mutations and SCNA subtypes in both enucleated and salvaged RB eyes. It is crucial to distinguish between these alternative mechanisms, for the wider implications for management of patient and family members. Citation Format: Liya Xu, Lishuang Shen, Ashley Polski, Rishvanth K. Prabakar, Mark Reid, Peter Kuhn, David E. Cobrinik, James Hicks, Xiaowu Gai, Jesse Berry. Simultaneous identification of clinically relevant RB1 Mutations and somatic copy number alterations in the Aqueous Humor of retinoblastoma eyes [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 452.

Abstract 3416: Androgen deprivation following JAK1/2 and PD-L1 inhibition improves antitumor efficacy in mouse models of Pten-deficient prostate cancer

Abstract Recently, the tumor suppressor PTEN has been shown to function as an immune modulator and its inactivation is associated with mediating tumor immune evasion. We previously examined the effects of programmed death-ligand 1 (PD-L1) immune checkpoint blockade (ICB) in transgenic mouse models of Pten-null prostate cancer. However, anti-PD-L1 monotherapy was largely ineffective in immunologically cold castration-naïve tumors from intact mice but showed some activity when administered to castrated mice. Our previous immune profiling studies also revealed that while androgen withdrawal increased tumor inflammation, it also mediated the recruitment and accumulation of immunosuppressive cells such as regulatory T cells, M2-polarized macrophages and myeloid-derived suppressor cells (MDSCs). Interleukin-6 (IL-6)/signal transducer and activator of transcription-3 (STAT3) signaling is a key immune modulating pathway that is upregulated in both intact and androgen deprived tumors. Here, we evaluate the effectiveness of targeting PD-L1 and JAK/STAT signaling in preclinical models of Pten-null prostate cancer. Intact tumor bearing conditional Pten-knockout were treated for four weeks with an a-PD-L1 blocking antibody (clone D265A, mouse/IgG1 kappa) and/or the JAK1/2 inhibitor, AZD1480. Pharmacological treatments targeting PD-L1 and JAK1/2 (PJ) combinations were also evaluated as concurrent therapy with androgen deprivation (AD) by orchidectomy (PJ+AD), and as sequential therapy; two weeks AD followed by PJ (AD>PJ), and one week of PT followed by AD (PJ>AD). Combined PD-L1/JAK followed by AD was the only treatment combination that improved antitumor immune responses over monotherapy. Notably, flow cytometry studies showed that combined PD-L1/JAK potently abrogated PD-L1 expression in circulating dendritic cells in all settings. Evidence for enhanced antitumor immune response in PJ>AD was supported by increased numbers circulating effector memory CD8 T cells and CD355+CD8+ T cells (CD355 Class I MHC-restricted T cell-associated molecule (CRTAM) a marker for activated CD8 T cell), and increased CD8 T cell infiltration in tumors and a reduction of CD25+CD4+ regulatory T cells. The efficacy of the PJ>AD treatment combination was further tested and confirmed in a subcutaneous syngeneic allograft model using mice grafted with tumor blocks from Pten-deficient castration-resistant prostate cancer tumors. Together these results indicate that pretreatment with combined PD-L1/JAK blockade can decrease the immunosuppressive effects of androgen withdrawal and have the potential to restore antitumor immune activity in Pten-deficient prostate cancer. Citation Format: Marco A. De Velasco, Yurie Kura, Naomi Ando, Noriko Sako, Kazuko Sakai, Kazuhiro Yoshikawa, Kazuto Nishio, Hirotsugu Uemura. Androgen deprivation following JAK1/2 and PD-L1 inhibition improves antitumor efficacy in mouse models of Pten-deficient prostate cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3416.

Abstract 6466: Serum immunoglobulins are an independent prognostic marker of time to first therapy in newly diagnosed chronic lymphocytic leukemia (CLL) and monoclonal B-cell lymphocytosis (MBL)

Abstract Immune dysfunction is a hallmark of CLL/MBL; ~25% individuals with newly diagnosed CLL have hypogammaglobulinemia at diagnosis. In this study of newly diagnosed CLL/MBL, we sought to a) investigate the association between serum immunoglobulins and specific mutations identified by next generation sequencing, and b) determine if serum immunoglobulin levels predict TFT, independent of the CLL-International Prognostic Index (CLL-IPI) and tumor mutational load (TML). We used the Mayo Clinic CLL Resource to identify newly diagnosed CLL/MBL individuals consented within 2 years of diagnosis. Baseline clinical characteristics and the CLL-IPI score was ascertained on all individuals. DNA was extracted from peripheral blood CD5+/CD19+ clonal B-cells; and the entire coding regions of 60 somatically recurring mutated CLL driver genes were enriched using a customized Agilent SureSelect gene panel and sequenced in Illumina Hiseq4000. Serum immunoglobulins were quantitated by radial immunodiffusion using Immunoplates. The relationship between serum immunoglobulins was assessed: a) with individual gene mutations (mutated vs. not); and b) using the TML score (using a value of 0, 1, and 2 or greater [2+] mutated genes). The impact of serum immunoglobulin levels on TFT was analyzed for all individuals. Of 356 individuals identified, 80 (22%) had MBL, median age was 63 years (28-87), 261 (73%) were male, and the CLL-IPI distribution was: low risk (n=149, 42%), intermediate risk (n=112, 32%), and high/very high risk (n=94, 26%). We observed 43%, 39%, and 18% individuals with 0, 1, or 2+ mutated genes, respectively. The most commonly mutated genes were NOTCH1 (13%), SF3B1 (13%), TP53 (8%), ATM (8%), BIRC3 (7%), and NFKBIE (7%). Using standard cutoffs, 135 (38%) individuals had low serum IgG, 73 (21%) had low serum IgA, and 154 (44%) had low serum IgM at diagnosis. There were no differences in individual gene mutation frequencies between individuals with low and normal serum IgG. In contrast, individuals with low serum IgA were more likely to have mutations in NOTCH1 (25% vs 10%) and ATM (14% vs 7%) compared to those with normal IgA, and individuals with low serum IgM were more likely to have mutations in NOTCH1 (19% vs 9%) and BIRC3 (11% vs 4%) compared to those with normal IgM (all p-values <0.05). There were no significant differences in individuals with low serum IgG based on TML scores of 0, 1, and 2+ (31%, 35%, 40%, respectively, p=0.28). However, individuals with low serum IgA and low serum IgM were more likely to have a higher TML (8%, 20%, 36% for IgA, p<0.0001; and 35%, 45%, 54% for IgM, p<0.0001, for TML scores of 0, 1, and 2+, respectively). After a median follow-up of 6.9 years, 157 individuals required CLL therapy. After adjusting for CLL-IPI and TML, low serum IgG, low serum IgA, and low serum IgM (hazard ratio [HR]: 1.8, 2.3, and 2.4, respectively, all p-values <0.05) were independently associated with a shorter TFT. Our study shows that individuals with CLL/MBL who harbor mutations in NOTCH1, ATM and BIRC3 are more likely to have low serum IgA and IgM at diagnosis. Serum IgA and IgM (but not serum IgG) levels decrease with an increase in the TML score. Finally, low serum immunoglobulins can predict TFT, independent of the CLL-IPI and TML, which has important implications for risk stratification in newly diagnosed CLL/MBL. Citation Format: Sameer A. Parikh, Cristine Allmer, Geffen Kleinstern, Nicholas J. Boddicker, Shulan Tian, Cecilia B. De Campos, Laura A. Bruins, Daniel R. o"brien, Brian F. Kabat, Kari G. Rabe, Aaron D. Norman, Huihuang Yan, Xing Li, Timothy G. Call, Wei Ding, Jose F. Leis, Saad J. Kenderian, Celine M. Vachon, James R. Cerhan, Neil E. Kay, Susan L. Slager, Esteban Braggio. Serum immunoglobulins are an independent prognostic marker of time to first therapy in newly diagnosed chronic lymphocytic leukemia (CLL) and monoclonal B-cell lymphocytosis (MBL) [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6466.

Abstract LB-167: Post-marketing safety surveillance of Tumor Treating Fields (TTFields) in patients with high-grade glioma in clinical practice

Abstract Introduction: Tumor Treating Fields (TTFields) are an antineoplastic treatment delivering low-intensity, intermediate-frequency, alternating electric fields through 2 pairs of transducer arrays locoregionally applied to tumor bed. TTFields are FDA-approved for glioblastoma (GBM; 200 kHz) and mesothelioma (150 kHz). Safety and effectiveness were demonstrated in the Phase 3 EF-11 and EF-14 trials in recurrent GBM (rGBM) and in newly diagnosed GBM (ndGBM), respectively. The main TTFields-related adverse event (AE) was array-associated manageable scalp irritation. We report AEs from TTFields-treated patients in the real-world, clinical practice setting. Methods: Unsolicited, global, post-market surveillance data from TTFields-treated patients (October 2011-February 2019) were retrospectively analyzed using MedDRA v21.1, stratified by region (US, EMEA [Europe, Middle East, Africa], or Japan), diagnosis (ndGBM, rGBM, anaplastic astrocytoma, anaplastic oligodendroglioma, or other brain tumors [includes brain metastases from different cancer types]), and age (years; <18 [pediatric], 18 to 64 [adults], or ≥65 [elderly]). Results: Of 11,029 patients, 53% had ndGBM, 39% had rGBM (at any line of recurrence), 6% had anaplastic astrocytoma/oligodendroglioma, and 1% had other brain tumors. Most were adults (73%) and 26% were elderly (≥65 years of age). The majority of patients was males (n=7313; 66.3%) compared to females (n=3716; 33.7%). The most commonly reported TTFields-related AE was array-associated local skin reaction, with an incidence of 38% in ndGBM, 29% in rGBM, 38% in anaplastic astrocytoma/oligodendroglioma, 31% in other brain tumors, 37% in pediatric, 34% in adults, and 36% in elderly patients. Most skin AEs were grade 1 and 2 and they were reversible with treatment. Other TTFields-related AEs in patients with ndGBM and rGBM, respectively, included heat sensation (under-array warmth; 11%, 10%), electric sensation (under-array tingling; 11%, 9%), and headache (7%, 6%). Conclusions: This retrospective, global, TTFields safety surveillance analysis revealed no new safety signals, with favorable safety and tolerability comparable to published TTFields/GBM trials. The most common TTFields-related AE was array-associated local skin reaction. The safety profile remained consistent among subgroups (diagnosis, age, or region) and total cohort, indicating feasibility in multiple subpopulations, including elderly patients. Citation Format: Wenyin Shi, Deborah T. Blumenthal, Nancy Ann Oberheim Bush, Sied Kebir, Rimas V. Lukas, Yoshihiro Muragaki, Jay-Jiguang Zhu, Martin Glas. Post-marketing safety surveillance of Tumor Treating Fields (TTFields) in patients with high-grade glioma in clinical practice [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-167.

Abstract 1092: A systematic analysis of BRAF mutations and their sensitivity to different BRAF inhibitors

Abstract The MAPK-ERK signaling cascade is among the most frequently mutated pathways in human cancer, with the BRAF V600 mutation being the most common alteration. FDA-approved BRAF inhibitors as well as combination therapies of BRAF and MEK inhibitors are available and provide survival benefits to patients with a BRAF V600 mutation in several indications. Yet non-V600 BRAF mutations are found in many cancers and are even more prevalent than V600 mutations in certain tumor types. As the use of NGS profiling in precision oncology is becoming more common, novel alterations in BRAF are being uncovered. This has led to the classification of BRAF mutations, which is dependent on its biochemical properties and affects it sensitivity to inhibitors. Therefore, annotation of these novel variants is crucial for assigning correct treatment. Using a high throughput method for functional annotation of MAPK activity, we profiled 151 different BRAF mutations identified in the AACR Project GENIE dataset, and their response to 4 different BRAF inhibitors- vemurafenib and 3 different exploratory 2nd generation inhibitors. The system is based on rapid synthesis of the mutations and expression of the mutated protein together with fluorescently labeled reporters in a cell-based assay. Our results show that from the 151 different BRAF mutations, ~25% were found to activate the MAPK pathway. All of the class 1 and 2 mutations tested were found to be active, providing positive validation for the method. Additionally, many novel activating mutations were identified, some outside of the known domains. When testing the response of the active mutations to different classes of BRAF inhibitors, we show that while vemurafenib efficiently inhibited V600 mutations, other types of mutations and specifically BRAF fusions were not inhibited by this drug. Alternatively, the second-generation experimental inhibitors were effective against both V600 as well as non-V600 mutations. Using this large-scale approach to characterize BRAF mutations, we were able to functionally annotate the largest number of BRAF mutations to date. Our results show that the number of activating variants is large and that they possess differential sensitivity to different types of direct inhibitors. This data can serve as a basis for rational drug design as well as more accurate treatment options for patients. Citation Format: Zohar Barbash, Dikla Haham, Liat Hafzadi, Ron Zipor, Shaul Barth, Arie Aizenman, Lior Zimmerman, Gabi Tarcic. A systematic analysis of BRAF mutations and their sensitivity to different BRAF inhibitors [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1092.

Abstract 1137: Deep targeted and whole exome sequencing identifies driver mutations in the airway epithelium of individuals at high risk for lung cancer

Abstract Insights into tumor mutational burden of the airway epithelium may play important role in stratifying individuals at different levels of risk for lung cancer. Comparison of the genetic profiles of normal airway epithelial cells to tumor mutations can yield valuable information regarding shared mutations as well as presence of neoantigens. We hypothesized that cytologically normal bronchial epithelium of patients with lung cancer and those at high risk for lung cancer harbor driver mutations that can be identified by deep targeted sequencing of selected genes implicated in cancer and whole exome sequencing. We performed next generation sequence-based CAPP-Seq (CAncer Personalized Profiling by deep Sequencing) and whole exome sequencing (WES) of DNA from the bronchial epithelial cells from the bronchial brushings of individuals at-risk (low and high) for lung cancer and patients with lung cancer. Bronchial brushings were collected from consented individuals at the time of bronchoscopy at Vanderbilt University Medical Center and the Nashville Veterans Affairs Medical Center. We used Tammemagi risk model to calculate the risk of developing lung cancer. Lung tumor tissues from matched individuals were also sequenced. Using CAPPSeq, a total of 129 mutations were identified in bronchial brushings. Among individuals without cancer, driver mutations included TP53, PIK3CA, NOTCH1, KRAS, MET, CUL3, NF1, KEAP1 and EGFR mutations in 5 out of 13 high risk and 2 out of 7 low risk individuals. Among individuals with cancer, 13 out of 22 brushings harbored driver mutations in TP53, PIK3CA, ARID1A, PTEN, NOTCH1, KRAS, MET, NFE2L2, CUL3, NF1, and KEAP1 of which mutations in TP53 and PTEN were identified in both tumor and bronchial brushings. At least one mutation was called in 18 brushings of individuals with lung cancer. Mutations detected in both individuals with and without cancer were TP53, PIK3CA, NOTCH1, KRAS, MET, CUL3, and NF1. Using WES on the DNA from the same bronchial brushings (low risk-9; high risk-8 and lung cancer-17), we identified 90 mutations that were also identified by CAPPSeq. Most of the mutations identified by CAPPSeq with >5% variant allelic frequencies (VAF) were also identified by WES. However, WES was not sensitive enough to identify mutations with less than 0.5% VAF. Out of 9 mutations in TP53 identified in the brushings by CAPPSeq, 4 were identified by WES. Mutation in KMTD2, a histone methylation enzyme and believed to be a tumor suppressor, was identified by WES in the brushings and the tumor tissues of patients with lung cancer as well as in one high risk brushings. We identified mutations in cancer driver genes in the bronchial brushings of individuals with and without lung cancer. This field of cancerization effect may provide a tool for risk assessment for lung cancer. This research was funded by U01CA196405 to PPM. Citation Format: SM Jamshedur Rahman, Shih-kai Chu, Shilin Zhao, Yong Zou, Angela Hui, Thomas Stricker, Maximilian Diehn, Pierre P. Massion. Deep targeted and whole exome sequencing identifies driver mutations in the airway epithelium of individuals at high risk for lung cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1137.

Abstract 6346: Optimizing precision combination therapies based on single-cell profiling of brain tumor biopsies

Abstract Glioblastomas (GBM) display dramatic levels of intra-tumoral heterogeneity. This heterogeneity contributes significantly to resistance to standard and emerging treatments. As a result, GBM remains the deadliest primary tumor of the adult brain. Moreover, most preclinical studies of GBM use models of primary GBM, most studies of clinical specimens use tissue derived from primary GBMs, despite the fact that most clinical trials enroll patients at recurrence. There is a paucity of information about recurrent GBM. To address these critical barriers, we devised a strategy to optimize precision drug combinations based on an understanding of a patient's intratumor heterogeneity, ascertained via single-cell profiling. We conducted single-nucleus RNA sequencing (snRNA-seq) and/or single-nucleus assay for transposon-accessible chromatin (snATAC-seq) sequencing for 16 recurrent GBMs. For 6 of these cases we also derived cell lines from patient biopsies. We identified lineage hierarchies of proliferating cells and their more differentiated progeny. Based on this, we performed an in vitro drug-combination screen using FDA-approved chemotherapies, targeting pathways learned from these single-cell signatures. We found that targeting distinct cellular states in the inferred lineage hierarchy of a patient's tumor provides a more complete treatment than targeting one state alone. Additionally, we will present the results of our ongoing in vivo validation activities and profiling efforts. This work presents a novel paradigm for identifying precision combinations of chemotherapies targeted to phenotypes learned from single-cell profiling of a patient's tumor biopsy. Citation Format: Husam Babikir, Lin Wang, Karin Shamardani, Francisca Catalan, Gary Kohanbash, Manish K. Aghi, Joanna J. Phillips, Aaron A. Diaz. Optimizing precision combination therapies based on single-cell profiling of brain tumor biopsies [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6346.

Abstract 6269: Effect of 3AP on neuroendocrine tumor cell proliferation, apoptosis and activation of DNA repair pathway

Abstract Background: Neuroendocrine tumors (NETs) are a heterogeneous group of neoplasms arising from multiple organs. Peptide Receptor Radionuclide Therapy (PRRT) is a novel therapy approved for treating NETs with a treatment response rate of 20-30%. To improve therapeutic response of PRRT, 3-AP (Triapine), which is a ribonucleotide reductase (RR) inhibitor with antineoplastic and radiation sensitizing properties, can be used in combination with PRRT. The purpose of our study is to examine effects of Triapine treatment on cellular proliferation, apoptosis and activation of DNA repair pathway in NET cells in vitro. Methods: For these studies, we utilized two human NET cell lines: BON, a pancreatic carcinoid tumor, and QGP-1, a somatostatinoma. (i) To determine effective dose range of Triapine (500 - 2500nM), proliferation was measured at 48, 72, and 96 h with sulforhodamine B (SRB) assay. (ii) To evaluate the effects of radiosensitivity, cultures were treated with increasing doses (250 - 5,000nM), received 2 Gy radiation (XRT) 1h later, exposed to Triapine for 24 hours, rinsed, and fed Triapine-free media; colonies were counted and quantified with SRB assays 14 days later and survival curves were generated. (iii) To further characterize cellular responses, cells were treated with Triapine (250 - 5,000 nM), irradiated 1h later, and harvested 24 h later to examine expression levels of Cyclin D1, cleaved PARP, p-DNA-PK (S2056), p-ATM (S1981) and p-ATR (S428) by Western blot. Results: (i) SRB proliferation assay demonstrated the effective dose range, determined by half maximal inhibitory concentration (IC-50), was 1,3 µM in BON cells and 1.73 µM in QGP-1 cells at 96 h. (ii) Exposure to Triapine before irradiation resulted in a significant increase in the radiosensitivity of each of the tumor cell lines with dose enhancement factors at a surviving fraction of 0.10 of 1.8 and 1.6 for BON and QGP-1, respectively. (iii) Triapine treatment alone activated DNA repair pathway in a dose-dependent fashion as determined by pDNA-PK (S2056) expression. Increasing p-ATM (S1981) expression was observed in a dose-dependent fashion in QGP-1 cells, but not in BON cells. In addition, p-ATR (S428) expression was not significantly different in either cell line. (iv) Triapine treatment alone or in combination with 2 and 4 Gy XRT resulted in dose-dependent decrease in cyclin D1 expression in both cell lines, however, no significant increase in cleaved PARP expression at 24 and 48h. Conclusion: Our results demonstrated rapid and dose dependent decrease in cyclin D1 expression after Triapine treatment, dose-dependent activation of DNA-PK, the key component of the non-homologous end joining pathway and significant increase in the radiosensitivity of each of the tumor cell lines. In conclusion, our findings provide a rationale for inclusion of Triapine as a radiosensitizer in combination with PRRT in treating NETs. Citation Format: Zeta Chow, Aman Chauhan, Lowell Anthony, Courtney Townsend, B. Mark Evers, Piotr Rychahou. Effect of 3AP on neuroendocrine tumor cell proliferation, apoptosis and activation of DNA repair pathway [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6269.

Abstract 868: Prediction of clinical benefit to checkpoint blockade in advanced NSCLC patients using radiomics of PET/CT images

Abstract Immunotherapy has emerged as a one of the most successful treatment approach in advanced Non-small Cell Lung Cancer (NSCLC) patients. However, only 25-50% patients obtain durable clinical benefit (DCB), and 7-43% patient may suffer from immune-related severe adverse events. Therefore robust biomarkers that are predictive of response immune-checkpoint blockades at baseline are needed. Though PD-L1 expression by immunohistochemistry (IHC) is the only clinically-approved biomarker to trigger treatment decisions, sampling bias and the dynamic nature of PD-L1 expression can affect the final results. In this study, we investigated the value of PET/CT images in the predicting progression-free survival (PFS) and overall survival (OS) of immunotherapy in advanced NSCLC patients. The retrospective arm of this study included 400 patients with histologically-confirmed NSCLC, and these were split into training (N=284) and test (N=116) cohorts. A residual-convolutional-network model was developed to generate a deep learning score (PDL1-DLS) to predict the PD-L1 expression status with PET/CT images. Another 146 retrospective patients with histologically confirmed stage IIIB-IV NSCLC treated with immunotherapy (IO) were split into training (N=99) and test (N=47) cohorts, and a multiparametric radiomics signature (mpRS) generated from conventional PET and CT radiomics features was developed to predict DCB using an improved least absolute shrinkage and selection operator (LASSO) method. Subsequently, multivariable Cox regression models for PFS and OS prediction were trained with the mpRS, PDL1-DLS, and clinical characteristics, and these were further validated with 48 prospective IO-treated NSCLC patients. The PDL1-DLS significantly discriminated PD-L1 positive and negative patients with AUCs of 0.89 (95%CI:0.84-0.94) and 0.84 (95%CI:0.76-0.92) for training and test cohorts. The mpRS could predict DCB with AUCs of 0.86 (95%CI:0.79-0.94) and 0.83 (95%CI:0.71-0.94) in the training and test cohorts. In both retrospective and prospective IO-treated cohorts, the combination of binarized PDL1-DLS and binarized mpRS significantly improved the prediction of PFS and OS compared to single binarized PDL1-DLS (PFS: p<0.001, OS: p<0.001) or single binarized mpRS (PFS: p<0.040, OS: p<0.10). Subsequent multivariable Cox regression models identified binarized mpRS, binarized PDL1-DLS, ECOG, and histology as significant independent variables, and could achieve significant higher C-indices of 0.75 (95%CI:0.70-0.79) and 0.79 (95%CI:0.73-0.86) in prediction of PFS, 0.79 (95%CI:0.73-0.85) and 0.76 (95%CI:0.61-0.92) in prediction of OS in the retrospective and prospective IO-treated cohorts. We have therefore observed that, prior to therapy initiation, PET/CT fusion images can be used independently to predict PD-L1 status or DCB of immunotherapy in advanced NSCLC patients. Both of these models can be leveraged to improve more precise and individualized decision support in the treatment of patients with advanced NSCLC. Citation Format: Wei Mu, Ilke Tunali, Jhanelle E. Gray, Jin Qi, Matthew B. Schabath, Robert J. Gillies. Prediction of clinical benefit to checkpoint blockade in advanced NSCLC patients using radiomics of PET/CT images [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 868.

Abstract 852: Improved tumor-only somatic variant calling using a gradient boosted machine learning algorithm

Abstract Background: Accurate identification of somatic variants in a tumor sample is often enabled by utilizing a paired normal tissue sample from the same patient that enables the separation of private germline mutations from somatic variant calls. However, a paired normal sample is not always available from patients, making accurate somatic variant analysis more challenging. Composite proxy normals and other filtering approaches can be used in lieu of a paired normal sample, but the resulting somatic call set may suffer from incomplete germline filtering and reduced sensitivity compared to paired tumor-normal analysis. To address these limitations, we developed a novel, machine learning based tumor-only somatic small variant classifier, which leverages gradient boosted decision trees to substantially increase somatic variant specificity from the tumor-only analysis without reducing overall sensitivity. Methods: We produced a ground truth set of somatic SNVs and indels from 350 whole exome-sequenced tumor-normal pairs using a validated cancer bioinformatics pipeline. We then generated a feature set from each tumor sample by aggregating pileup attributes including: allelic frequency and read depth, tumor cellularity estimations, germline variant calls from HaplotypeCaller, somatic variant calls from Mutect and Mutect2 using a proxy-normal, copy-number alterations, annotations from databases such as GnomAD and COSMIC, and problematic-region annotations including homopolymers. Using these features and the ground truth set, we trained a gradient-boosted decision tree to predict the somatic likelihood of each variant. Model hyperparameters were optimized using a random search during stratified cross-validation, and model performance was evaluated on a hold-out test set. Results: Using a classification threshold that optimized F1 score on the validation set, we observed a significant increase in model precision on the test set, with comparable sensitivity to somatic calling using a conventional proxy-normal filtering approach. Because our model outputs somatic probability, the classification threshold can be tuned to favor sensitivity or specificity of the call set, depending on the desired use case. To improve interpretability of our model, we employed shapely additive explanations (SHAP) to obtain feature importance values. Our analysis revealed that annotations such as population frequency and base quality scores were among the most important features. Conclusions: Our machine learning approach can greatly enhance germline filtering when making somatic variant calls when a paired normal sample is not available without decreasing sensitivity for true somatic variants. Depending on the use-case, classification thresholds can be tuned to improve sensitivity over conventional variant callers for more modest improvements in precision. Finally, model interpretation has revealed a subset of highly discriminative features, which may prove useful for variant interpretation, future feature set expansion, or model tuning. Citation Format: Nicholas Phillips, Patrick Jongeneel, John West, Richard Chen, Jason Harris. Improved tumor-only somatic variant calling using a gradient boosted machine learning algorithm [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 852.