Molecular signatures reflecting microenvironmental metabolism and chemotherapy-induced immunogenic cell death in colorectal liver metastases

Abstract

Background: Metastatic colorectal cancer (CRC) is associated with highly variable clinical outcome and response to therapy. The recently identified consensus molecular subtypes (CMS1-4) have prognostic and therapeutic implications in primary CRC, but whether these subtypes are valid for metastatic disease is unclear. We performed multi-level analyses of resectable CRC liver metastases (CLM) to identify molecular characteristics of metastatic disease and evaluate the clinical relevance.

Methods: In this ancillary study to the Oslo-CoMet trial, CLM and tumor-adjacent liver tissue from 46 patients were analyzed by profiling mutations (targeted sequencing), genome-wide copy number alteration (CNAs), and gene expression.

Results: Somatic mutations and CNAs detected in CLM were similar to reported primary CRC profiles, while CNA profiles of eight metastatic pairs suggested intra-patient divergence. A CMS classifier tool applied to gene expression data, revealed the cohort to be highly enriched for CMS2. Hierarchical clustering of genes with highly variable expression identified two subgroups separated by high or low expression of 55 genes with immune-related and metabolic functions. Importantly, induction of genes and pathways associated with immunogenic cell death (ICD) was identified in metastases exposed to neoadjuvant chemotherapy (NACT).

Conclusions: The uniform classification of CLM by CMS subtyping may indicate that novel class discovery approaches need to be explored to uncover clinically useful stratification of CLM. Detected gene expression signatures support the role of metabolism and chemotherapy in shaping the immune microenvironment of CLM. Furthermore, the results point to rational exploration of immune modulating strategies in CLM, particularly by exploiting NACT-induced ICD.

Keywords: colorectal liver metastases; genomic profiling; immunogenic cell death; neoadjuvant chemotherapy.

Conflict of interest statement

CONFLICTS OF INTEREST All authors declare no conflicts of interest.

Figures

Figure 1. Somatic mutations and copy number alterations (CNAs) in colorectal liver metastases

Main panel: Selected genes (y-axis) with mutations (blue), amplifications (red) and deletions (green); Oslo-CoMet trial sample ID (x-axis). Black outlines indicate metastatic pairs. Consensus molecular subtype (CMS) is indicated when available. Top panel: Total number of mutations, amplifications and deletions per sample. Right panel: Total number of mutations, amplifications and deletions for selected genes. *, CNA data not available.

Figure 2. Genome wide copy number alterations

(A) Frequency plot of genome wide copy number alterations. The histogram shows percentage of samples with specific alterations. The genomic position is indicated by chromosome 1 on the left and up to chromosome 22 on the right. Copy number gains for each region are depicted in red, and copy number losses are depicted in green. The plot shows high frequency of CNAs in chromosomes 7, 8, 13, 18 and 20. (B) Metastatic pairs displaying discrepancies in copy number profiles. Estimated ploidy and purity (tumor cell fraction) values are listed at the top of the plot. Metastatic pair 8M1 and 8M2 show discrepancies in segments of chromosomes 4, 13 and 14; metastatic pair 36M1 and 36M2 differ in total ploidy, with 36M1 having ploidy close to 4n and 36M2 having ploidy close to 2n.

Figure 3. Gene expression variability in tumor-adjacent liver samples and metastases

(A) Dot-plot showing distribution of variability of individual transcripts in tumor-adjacent liver samples (x-axis) and metastases (y-axis). Ten genes in tumor-adjacent liver samples and 111 genes in metastases were identified by a variance filter of the respective gene expression data sets (variance>5). (B) Principle component analysis comparing gene expression from tumor-adjacent liver samples (yellow dots) compared to metastases (blue dots). The two sample types formed distinct non-overlapping clusters. (C) Heat map representing the expression profiles of the 111 genes found to be highly variable in metastases and subjected to unsupervised clustering analysis. Red-blue scale: red represents high expression and blue low expression. Annotation of samples includes administration of neoadjuvant chemotherapy (NACT) and mutations in SMAD4 and NRAS.

Figure 4. Survival outcome after liver metastasis surgery

Kaplan-Meier estimates comparing overall survival using log-rank test in patients having neoadjuvant chemotherapy (NACT+) or not (NACT-).