Evaluation of laser microdissected primary breast tumors for RNA-Seq over bulk processing

Introduction: RNA-Seq based gene expression profiling of breast tumor samples is widely used to subgroup patients and to identify gene signatures of prognostic value. However, tumor samples are highly heterogeneous, and so bulk processing of tumor tissue will consist of several different cell types. Here, we evaluated the advantage of using laser microdissected (LMD) breast tumors for RNA-Seq over bulk processing. Methods: Patients for the in-house dataset were duly consented under an IRB-approved protocol of the Clinical Breast Care Project. A total of 118 primary breast tumors embedded in OCT (Optimum Cutting Temperature) were selected and processed by LMD. Total RNA and protein were extracted using the Illustra triplePrep kit. Paired-end RNA sequencing of 118 cases was performed using the Illumina HiSeq platform and the reads were preprocessed using a PERL-based pipeline involving PRINSEQ, GSNAP and HTSeq. The Cancer Genome Atlas (TCGA) primary breast cancer RNA-Seq data for 1097 samples was downloaded. Differential expression of genes (DEG) was assessed using DESeq2. Significance was described for DEG with fold change >2 and p-adjusted value of 0.05. Results: A total of 24,518 genes with a mean expression of ≥ 10 raw counts across 118 tumor samples were identified in the in-house LMD dataset. In TCGA breast cancer RNA-Seq, 14,281 genes with a mean expression of ≥ 100 raw counts across 1097 tumor samples were identified. The conventional PAM50 classifier was used for intrinsic subtyping of in-house data, yielding 36 Basal-like, 14 HER2-enriched, 43 Luminal A, 22 Luminal B and 3 Normal-like calls. The provided PAM50 calls were used for TCGA which are 192 Basal-like, 82 HER2-enriched, 566 Luminal A, 217 Luminal B and 40 Normal-like calls. Within commonly expressed 13,165 genes, LMD and bulk processing exhibited approximately 40-78% non-overlap in significantly differentially expressed genes (SDEG) among the intrinsic subtypes. 21 unique stromal genes were present in SDEG unique to TCGA whereas there were only 5 SDEG unique to in-house dataset. Overall high positive correlation is observed among the stromal genes present in SDEG unique to TCGA suggesting strong stromal contribution in bulk processing. Pathway analysis of SDEG unique to LMD data suggested alterations in known cancer pathways (B-cell immune response, RNA metabolism and splicing, phagocytosis, and signaling components). Conclusion: Analysis of The Cancer Genome Atlas breast cancer RNA-Seq data set (based on bulk processing tissue) suggested contribution of stromal signature genes and important differences from LMD specimens. Thus, tumor selection via LMD can result in better expression profiling by RNA-Seq which has the potential to uncover many cancer genes and pathways. The views expressed in this abstract are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, Department of Defense, or U.S. Government. Citation Format: Praveen-Kumar Raj-Kumar, Lori A. Sturtz, Albert J. Kovatich, Brenda Deyarmin, Jeffrey A. Hooke, Leigh Fantacone-Campbell, Anupama Praveen-Kumar, Jianfang Liu, James Craig, Leonid Kvecher, Jennifer Kane, Jennifer Melley, Stella Somiari, Stephen C. Benz, Justin Golovato, Shahrooz Rabizadeh, Patrick Soon-Shiong, Richard J. Mural, Craig D. Shriver, Hai Hu. Evaluation of laser microdissected primary breast tumors for RNA-Seq over bulk processing [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 3402.