Surgical Tumor Resection Deregulates Hallmarks of Cancer in Resected Tissue and the Surrounding Microenvironment

Abstract

Surgery exposes tumor tissue to severe hypoxia and mechanical stress leading to rapid gene expression changes in the tumor and its microenvironment, which remain poorly characterized. We biopsied tumor and adjacent normal tissues from patients with breast (n = 81) and head/neck squamous cancers (HNSC; n = 10) at the beginning (A), during (B), and end of surgery (C). Tumor/normal RNA from 46/81 patients with breast cancer was subjected to mRNA-Seq using Illumina short-read technology, and from nine patients with HNSC to whole-transcriptome microarray with Illumina BeadArray. Pathways and genes involved in 7 of 10 known cancer hallmarks, namely, tumor-promoting inflammation (TNF-A, NFK-B, IL18 pathways), activation of invasion and migration (various extracellular matrix-related pathways, cell migration), sustained proliferative signaling (K-Ras Signaling), evasion of growth suppressors (P53 signaling, regulation of cell death), deregulating cellular energetics (response to lipid, secreted factors, and adipogenesis), inducing angiogenesis (hypoxia signaling, myogenesis), and avoiding immune destruction (CTLA4 and PDL1) were significantly deregulated during surgical resection (time points A vs. B vs. C). These findings were validated using NanoString assays in independent pre/intra/post-operative breast cancer samples from 48 patients. In a comparison of gene expression data from biopsy (analogous to time point A) with surgical resection samples (analogous to time point C) from The Cancer Genome Atlas study, the top deregulated genes were the same as identified in our analysis, in five of the seven studied cancer types. This study suggests that surgical extirpation deregulates the hallmarks of cancer in primary tumors and adjacent normal tissue across different cancers.

Implications: Surgery deregulates hallmarks of cancer in human tissue.

Figure 1.

Breast Cancer Study Design, significantly deregulated genes in individual patient-specific analysis. and heat map representation of MSigDB analysis using Gene Set Enrichment Analysis. A, Study sampling schema and experimental design. B, Patient recruitment and assay inclusion C. Histogram representation of genes deregulated in each comparison for a patient. D, Recurrence of differentially expressed genes across all patients in each of the three comparisons E. Venn diagram representation of 737 deregulated genes in at least three patients for all comparisons, across 32 patients. F, 141 genes commonly deregulated across three comparisons A versus B, B versus C, and A versus C, in 32 patients. G, −log of “k/K” values of Hallmarks significantly enriched in Hallmarks module analysis of significantly deregulated genes in the three comparisons A versus B, B versus C, and A versus C represented as a heat map. H, −log of “k/K” values of Canonical Pathways significantly enriched in the Canonical Pathways module analysis of significantly deregulated genes in all the three comparisons A versus B, B versus C, and A versus C represented as a heat map I, −log of “k/K” values of Gene Ontological Processes significantly enriched in the Gene Ontology module analysis of significantly deregulated genes in all the three comparisons A versus B, B versus C, and A versus C represented as a heat map.

Figure 2.

Microarray analysis of 9 HNSC patient intra-operative samples. A, Dendrogram representing sample relationship based on genome-wide expression profiles in microarray data from patients with HNSC. B, Heat map representation of 16 differentially expressed genes based on multiple class comparisons for all samples from all 9 patients (T1 v T2 v T3 v T4). Red denotes upregulation and green denotes downregulation. Genes were clustered on the basis of hierarchical clustering. Samples were manually arranged as class labels. C, Heat map representation of differentially expressed genes based on binary class comparisons for all samples at time point T1 v all samples at time point T2 for all 9 patients D. Heat map representation of differentially expressed genes based on binary class comparisons for all samples at time point T1 v all samples at time point T3 for all 9 patients. E, Heat map representation of differentially expressed genes based on binary class comparisons for all samples at time point T1 versus all samples at time point T4 for all 9 patients. Red denotes upregulation and green denotes downregulation. Genes and samples were clustered on the basis of hierarchical clustering. F, Venn diagram representation of differentially expressed genes in C–E. Six genes are commonly deregulated at all time points compared with tumor tissue at T1.

Figure 3.

Validation of top de-regulated genes. A, Cancer types and surgical and biopsy specimens in TCGA Expression cohort. B, Heat map representation of top genes recurrently de-regulated in TCGA RNA-Seq cohort for 5/7 tumor types. Class comparison between biopsy and tumor resected after surgery reveals differentially expressed genes. The top recurrently upregulated genes across tumor types in TCGA RNA-Seq cohort are AP-1 transcription factor-network family members. C, Gene expression validation results from qRT²⁻Profiller. D, Gene expression validation results from custom NanoString nCounter assay.