Search

e15155 Background: Novel immunotherapies are becoming a viable option for advanced lung cancer treatment. High-throughput gene expression profiling (GEP) has enabled better understanding of patient responsiveness to targeted immunotherapy. However, non-invasive blood-based signatures are needed to monitor and predict response. Methods: To evaluate the predictive and prognostic role of blood-based GEP, we designed a longitudinal study by enrolling 15 patients with advanced lung cancers. A total of 65 samples were collected for RNA sequencing, ~4 blood specimens per patient, before and during anti-PD-1 treatment. We used multiple analyses, including time-course differential gene expression, principal component, lymphocyte compartment deconvolution, and genetic mutation, to search for and assess potential predictive and prognostic aspects. Results: Of 15 patients, 11 were classified as Responders (partial responders) and four were Non-Responders (one stable and three progressive diseases). Our analyses demonstrated: 1) By comparing baseline GEPs from Responders vs. Non-Responders before the first treatment, we identified potential markers (e.g., LY6E is significantly lower expressed in Responders, with Log2 Fold change = -3.44 and p = 1.83E-04) that can be used as predictors of responsiveness of the patients; 2) Immunoglobulin subunits- and T cell receptor complex-related genes were differentially expressed in Responders (DAVID analysis, p = 6.7E-3 and 2.1E-2, respectively), but not in Non-responders; 3) γδ T cells in the lymphocyte compartment were relatively increased in Responders; 4) Despite a different set of genes differentially expressed at different time points, the biggest GEP changes were at ~ week 6, after the second treatment. Additionally, we observed certain genes consistently up- or down-regulated through the whole course of treatment. Furthermore, after the first treatment, genes in the immune response pathway were regulated to different directions in Responders and Non-Responders. For example, interleukin receptor genes, such as IL18R1 and IL18RAP, and CD24 were down-regulated in Responders, but up-regulated in Non-Responders (p = 0.042, 0.023 and 0.044 respectively, t-test for the differential expression in these two groups). Conclusions: The utility of blood GEP to identify predictive and prognostic factors for precision immunotherapy is encouraging. Nevertheless, these results, predictive of the anti-PD-1 clinical response, are preliminary and need to be validated in a larger cohort.

e15776 Background: Pancreatic cancer (PDAC) is projected to become the second leading cause of cancer related mortality by 2030. Bulk whole genome sequencing studies of PDAC have characterized the landscape of clonal mutations and highlighted the prominence of copy number alterations (CNAs) in PDAC genomes. However, little is known with regards to the extent of sub-clonal heterogeneity of somatic CNAs and it is hypothesized that this heterogeneity is a contributing factor to the limited effectiveness of existing therapies. Methods: We retrieved absolute copy number information using bulk sparse whole genome sequencing of multi-region sampled PDAC samples as well as matching primary-metastasis tumor samples from over 100 patients. In addition, we analyzed copy number variations in a subset of these samples (n = 15) at single-cell resolution (~1000 cells in total). Results: We describe a detailed picture of sub-clonal CNAs genetic heterogeneity. Our results illustrate, among other findings, (1) extensive sub-clonal diversity of CNAs giving rise to many genetically unique sub-clonal cancer populations, (2) somatic mosaicism of chromosomal amplicons in single-cancer cells, (3) variation in the dosage of cancer genes, including the KRAS oncogene, in different tumor sub-clones and (4) somatic alterations, such as amplification of 8q11 containing the metastasis promoting gene IKBKB, associated with primary PDAC progression to liver metastasis. Conclusions: Our results offer an in-depth view of the sub-clonal heterogeneity of somatic CNAs in pancreatic cancer and illustrate ways in which such heterogeneity could lead to therapeutic resistance.

e14107 Background: Anti-PD-1 monoclonal antibodies have improved the outcome for patients with advanced non-small cell lung cancer (NSCLC). The goal of this study is to characterize the longitudinal gene expression profiles of peripheral blood mononuclear cells (PBMCs) in patients with advanced NSCLC treated with anti-PD-1 monoclonal antibodies and to evaluate the potential of this technology to predict response to therapy. Methods: PBMCs were collected using density centrifugation from patients with stage III/IV NSCLC before the first (T0) and second (T1) treatments, and, for responding patients, before the third (T2), fourth (T3), fifth (T4) and sixth (T5) treatments. Total RNA was extracted, and libraries were prepared using TruSeq Stranded mRNA LT Sample Prep Kit and sequenced by Illumina NextSeq. The paired-end reads were mapped to human reference hg38 using STAR and gene expressions were calculated using RSEM. Longitudinal differential gene expression with pathway analysis and principal component analysis (PCA) was performed across all time points. Genes with Bonferroni-adjusted p values < 0.05 were selected for pathway enrichment analysis with DAVID. Results: Nine of 15 patients have been enrolled. Six patients had a PR by RECIST 1.1, 2 patients have had stable/PD, 1 patient too early. Differential gene expression (DE) is available for the 8 patients with response assessment. There was differential expression of certain genes between T0 and T1 in the following pathways in responders but not in stable/PD: T cell receptor complex (p = 3.1E-6), regulation of immune response (p = 2.5E-5), T cell costimulation (p = 3.9E-3) were upregulated. CD24 is downregulated, while C4B is upregulated. Using PCA analysis we detected strong signal from the beta-hemoglobin HBB transcript. We observed HBG1 and HBM consistently upregulated at T1 (adjusted p = 7.9E-19 and p = 2.3E-10 respectively) through to T4 (adjusted p = 8.8E-27 and p = 3.8E-18 respectively). Conclusions: Anti-PD-1 monoclonal antibodies induce DE of genes in PBMCs from patients with advanced NSCLC. This technology may provide a predictive test for response to anti-PD-1 monoclonal antibodies.

e12635Background: There is a dearth of reliable analytic solutions that enable integration of tumor and immune profiles to determine mechanisms of response to targeted therapies. We developed a sys...

e23162 Background: Cancer is characterized by a variety of heterogeneous genomic and transcriptomic patterns involving highly complex signaling biological pathways. The problem of identification of the factors driving tumor progression becomes even more challenging due to intricate interaction mechanisms between these pathways. Using novel approaches in machine learning, we demonstrate the ability to quantitatively describe characteristic signaling patterns in cancer based on transcriptomic data Methods: We used RNASeq data from 20531 genes in 174 samples of GBM from The Cancer Genome Atlas including 5 major histological subtypes – Classical, G-CIMP, Mesenchymal, Neural, and Proneural, anddeveloped predictive computational framework for molecular subtype differentiation from normal tissue relying on variance based gene selection and random forest algorithm. Results: We obtained a few key findings – (1) genes from cell signaling pathways alone differentiate each subtype from normal tissue with 100% accuracy; (2) predictive genes are specific to each subtype; (3) inferred pathway interactions are also specific to each subtype; (4) typically most of the predictive genes involved in signaling are down-regulated in tumor compared to normal tissue (MAPT, PRKCG, PDE2A, RYR2, ATP1B1, GRN1, GNAO1), however, in each subtype we observed a smaller subset of predictive genes which are highly up-regulated in tumor (ID3, FN1, JAG1, F2R, COL4A1, EDAR, CDK2, CDK4, MFNG, BIRC5, CCNB2). We detected and quantitatively evaluated characteristic signaling pathway involvement across the GBM subtypes for MAPK, RAP1, RAS, Notch, PI3K-Akt, mTOR, FoxO, Jak-STAT, Wnt, cAMP, and Calcium Signaling, providing a unique approximation for each subtype signaling profile. Conclusions: In this study, we identified gene expression profiles and associated signaling pathways for distinguishing GBM Multiforme subtypes from normal tissue. We observed and described a dense complex picture of interacting signaling pathways. The detected interactions may provide clinical insights and could be used to identify potential therapeutic targets, however, more research is needed to confirm this.

e13019 Background: Several mitochondrial DNA (mtDNA) mutations have been identified in various types of human cancer. While gene fusions encoded by the nuclear genome have an important role in tumorigenesis, the contribution of the mitochondrial genome to this process is unclear. mtDNA has limited protective mechanisms compared to that for nuclear DNA (nDNA), and hence more susceptible to genomic aberrations. Methods: We sequenced RNA from pre and post treatment breast cancer samples obtained from neoadjuvant clinical trials BrUOG 211A/211B. Seventy-three of these samples were considered baseline and 61 were from post-treatment samples, 54 samples at both time points. Patients were given a run-in dose of bevacizumab (B), nab- paclitaxel (N) or trastuzumab (T), followed by combination biologic/chemotherapy (HER2- with B/carboplatin/N; HER2+ with T/carboplatin/N). RNA was derived from biopsy pairs obtained pre/post 10-day exposure to run-in monotherapy. We used TopHat-Fusion and deFuse for fusion detection. Results: On average, in each patient 4.6 mtDNA gene fusions were detected at the baseline and 4 in the post treatment samples. In this cohort we detected in total 385 mtDNA fusions across the baseline and 344 across the post-treatment samples. A number of fusions had recurrently shown drastic increase from baseline to post-treatment, such as MT-ATP6:MT-ND2 ( N = 9), MT-ND6:MT-CO2 ( N = 6), MT-ND6:MT-CO1 ( N = 10), and MT-ND4:MT-ATP6 ( N = 9). We observe that some mitochondrial genes such as MT-ND6 have as many as 7 gene partners. MT-ND6 appeared in fusions that were most abundant among both baseline and post-treatment samples with MT:ND4 ( N = 39), MT:ND2 ( N = 37) and MD:ATP6 ( N = 36). Interestingly, gene fusions MT-ND6:J01415.25, MT-CYB:MT-ND2 and MT-ND6:MT-ND1 were observed respectively in 9, 5 and 4 baseline samples but not in their post treatment equivalent. Conclusions: While several studies have examined the presence of mtDNA mutations in breast cancer, the role of mitochondrial gene fusions is unknown in cancer. In our breast cancer cohort we observed quite a high number of mitochondrial fusions, and some of them seem to be recurrent in multiple samples. This could potentially have clinical implications as knowledge of fusions in solid tumours gains.

e13006 Background: Current analyses on somatic mutations mainly involve counting samples as either carriers or non-carriers. Such a binary approach misses to take into account the fraction of tumor cells that carry a mutation, which is a significant measure of the influence of a mutation on the overall phenotype, such as gene expressions, subtype and pathway activity. Methods: In this study, we re-analyzed the 72,084 non-synonymous somatic mutation in 16,164 genes of 817 TCGA breast cancer samples (HER2+: 65, Luminal A: 415, Luminal B: 176, Basal: 136 and Normal-like: 25) using variant allele frequency (VAF) adjusted by sample purity derived from multiple methods. We obtained a gene-based VAF by choosing the maximum VAF among all mutations found in each gene for each sample. For each breast cancer subtype, we filtered for top 100 genes with highest number of mutation carriers and then ranked them by decreasing average VAF across all samples. We assume genes with higher average VAF are more likely to harbor clonal mutations in early tumor progression. We further evaluated the Pearson correlation between VAF of genes having at least ten mutation carriers and the expression levels of ESR1, PGR and ERBB2 in Luminal A subtype. Results: The top two genes with highest VAF for each of the four subtypes are as follows: (a) HER2+: ERBB2 (VAF = 0.39 ± 0.21, n = 4), KAT6A (0.31 ± 0.09, 4), (b) Luminal A: CTCF (0.41 ± 0.19, 13), MAP2K4 (0.37 ± 0.14, 24), (c) Luminal B: MAP2K4 (0.46 ± 0.11, 5), TP53 (0.46 ± 0.17, 64), and (d) Basal: SCN10A (0.46 ± 0.29, 5), MYH9 (0.43 ± 0.16, 5). The strongest correlations for Luminal A are: (i) SPEN (corr. = 0.85, n= 14), KMT2C (0.73, 24) and DMD (0.73, 11) with ESR1, (ii) DMD (0.69, 11) and NBL1 (0.67, 10) with PGR, and (iii) TBX3 (-0.51, 12) and MUC12 (0.47, 14) with ERBB2. Conclusions: While carrier count is effective for identifying genes prone to mutations, average VAF opens another perspective for uncovering genes that tend to harbor clonal mutations. Our work shows the potential of VAF analysis for identifying driver genes, understanding tumor progression and evaluating the impact of a mutation on a patient. As future work, we may improve the VAF estimates by adjusting for copy number variations and weight each variant by pathogenicity.

1046 Background: Gene expression profiles from core biopsies taken during PT for EBC may identify markers associated with benefit. B was recently shown to improve pathologic complete response (pCR) to CT in two randomized trials, but is associated with increased toxicity. Predictive signatures are critical to optimizing benefit. We have shown that changes in TGF-β pathway activity after one dose of B are associated with pCR. To confirm these results and understand the mechanism of alteration of the TGFb pathway, we evaluated the 61-gene TGFb signature by RT-PCR, measured markers of TGFb pathway activation and assessed TGFb gene mutations. Methods: Real-time PCR was performed on paired (pre/post) core biopsy RNA samples from 21 patients (pts) who received B, trastuzumab (T) or nab-paclitaxel prior to combined CT and targeted therapy using Custom RT2 Profiler PCR Arrays (384-well plate, 64x6 format) containing 60 genes from our TGF-beta signature and 3 housekeeping genes, IPO8, ALAS1, and SDHA. RT- PCR reactions containing Qiagen RT2SYBR Green qPCR Mastermix and 5 ng of amplified cDNA were run on a Roche LightCycler480 machine. Cp values were determined using the Roche LightCycler480 software. Antibodies to p-Smad2, TGFBR1 and TGFBR2 were analyzed in FFPE IHC pre and post therapy. TGFb pathway genes with known mutations in TCGA were assessed using an Illumina Custom Amplicon and miSeq next generation sequencing. Results: The log 2 FPKM value of the 60 genes correlated well with the ΔCP values calculated normalized to ALAS1 housekeeping gene. The ΔΔCP values was determined by subtracting the 60 gene ΔCP values of baseline samples from post-exposure samples. Patients were clustered using the ΔΔCP. Pts exposed to B who achieved a pCR showed down-regulation of signatures genes, in contrast to a strong non-pCR cluster. Signature evaluation showed no association with response in the T arm. TGFb pathway protein expression and mutation frequency is under analysis. Conclusions: TGFb signature predicts response to B containing chemotherapy after one dose of B and is validated by RT- PCR as a potential predictive tool. TGFb pathway activation and gene mutation analysis will be presented. Clinical trial information: NCT00617942.