The colorectal liver metastasis growth pattern phenotype is not dependent on genotype

Abstract

Background: The histopathological growth patterns (HGPs) of colorectal cancer liver metastases broadly classify patients into two groups post-liver metastasectomy, with encapsulated HGP indicating a more favourable prognosis. The potential association between HGPs and specific mutations is poorly understood.

Methods: Using next-generation sequencing data of 461 resected patients (104 patients with encapsulated versus 357 patients with non-encapsulated HGP), 19 putative colorectal cancer driver genes, tumour mutational burden (TMB), and microsatellite instability (MSI) or POLE mediated hypermutation were compared.

Results: Most putative drivers, including KRAS (q = 0.89), NRAS (q = 0.98),) and BRAF (q = 0.97)), were not associated with HGP. However, mutations in B2M and PTEN were associated with a encapsulated phenotype (7% vs. 0%, q = 0.001, and 9% vs. 2%, q = 0.02, respectively). TMB was higher in encapsulated patients (median 5.8 vs. 5.1 mutations per megabase, p = 0.009). Multivariable overall survival analysis corrected for genetic and patient factors confirmed that the encapsulated phenotype was an independent prognostic factor (adjusted hazard ratio, 0.60; 95% confidence interval: 0.36-0.99). Upon stratified analysis, all identified genetic associations were equivocal between the cohorts.

Conclusions: While an association between genetic drivers of adaptive immune responses seems probable and could explain a minority of encapsulated patients, these results primarily demonstrate that HGP phenotype is independent of the tumour genotype.

Fig. 1. Comparison and graphical representation of the mutation rates of 19 colorectal cancer driver genes stratified by histopathological growth pattern and regarding microsatellite instability high and POLE mutant cases and genetic sample site (i.e., primay colorectal cancer or colorectal liver metastasis).

The percentages represent the mutation frequency for each gene in each group. The q-value represents the result of the χ2 test with correction for multiple testing according to Benjamini & Hochberg applied. *q < 0.05. CRC colorectal cancer, CRLM colorectal liver metastasis, INF inframe, MISS missense, MSI-H microsatellite instability high, TRUNC truncating.

Fig. 2. Tumour mutational burden and hypermutation frequency in encapsulated and non-encapsulated patients.

a The distribution of tumour mutational burden is plotted for encapsulated and non-encapsulated patients with the number of mutations per megabase on the Y-axis (logarithmic scale), and the relative proportion of patients within the cohort on the X-axis. Each point represents a single patient. The horizontal line represents the cut-off for hypermutated forms of colorectal cancer (tumour mutational burden >12 mutations per megabase), and the dashed vertical line represents the intersection with this cut-off on the X-axis. b Violin and boxplots displaying the tumour mutational burden and (c) number of driver gene mutations on a logarithmic scale in encapsulated and non-encapsulated patients. The box and corresponding horizontal line represent the interquartile range and median, respectively, and the whiskers represent the range excluding outliers (defined according to the 1.5 rule). The p-value represents the result of the Mann–Whitney U test. d Barplots displaying the frequency of hypermutated tumours (defined as a tumour mutational burden greater than 12 mutations per megabase) with and (e) without microsatellite instability high and POLE mutant cases in encapsulated and non-encapsulated patients. The error bars represent the binomial 95% confidence interval according to Clopper-Pearson. The p value represents the result of the χ2 test. MSI-H microsatellite instability high, mt mutant.

Fig. 3. Co-occurence and double mutation rates of 19 driver genes in encapsualted and non-encapsulated patients.

a Encapsulated and (b) non-encapsulated patient groups are shown, with genes grouped by genomic pathway. Within the inner circle, the frequency of co-occurring mutations between gene-pairs as a proportion of the total number of mutations is shown using ribbons, where the ruler indicates the proportion expressed as percentage; i.e., a percentage of 20 means that the co-occurring mutations in gene X and Y represented 20% of all mutations identified within the cohort. On the outer circle, the double mutation rates within the cohort for each gene pair is plotted using bar plots; i.e., a percentage of 20 means that in all patients, 20% had a mutation in gene X and a co-occurring mutation in gene Y.

Fig. 3. Co-occurence and double mutation rates of 19 driver genes in encapsualted and non-encapsulated patients.

a Encapsulated and (b) non-encapsulated patient groups are shown, with genes grouped by genomic pathway. Within the inner circle, the frequency of co-occurring mutations between gene-pairs as a proportion of the total number of mutations is shown using ribbons, where the ruler indicates the proportion expressed as percentage; i.e., a percentage of 20 means that the co-occurring mutations in gene X and Y represented 20% of all mutations identified within the cohort. On the outer circle, the double mutation rates within the cohort for each gene pair is plotted using bar plots; i.e., a percentage of 20 means that in all patients, 20% had a mutation in gene X and a co-occurring mutation in gene Y.

Fig. 4. Kaplan–Meier overall survival estimates for encapsulated versus non-encapsulated patients after resection of colorectal liver metastasis.

The p value represents the result of the overall log-rank test.