Specific Mutations in APC, but Not Alterations in DNA Damage Response, Associate With Outcomes of Patients With Metastatic Colorectal Cancer

Lay Summary

In an integrated genomic and clinical analysis, we evaluate the effects of Wnt and DNA damage response pathway alterations on metastatic colorectal cancer.

Introduction

Alterations in the Wnt and DNA damage response (DDR) pathways have been proposed as modifiers of outcomes and therapeutic targets in patients with metastatic colorectal cancer (mCRC)^1–3. The Wnt signaling pathway is a critical mediator of intestinal stem cells and is the most commonly dysregulated pathway in CRC; oncogenic alterations release β-catenin from cytosolic regulation, allowing it to translocate to the nucleus and act as a transcriptional promoter. The DDR pathway is crucial for maintaining genomic stability. Alterations in this pathway have been associated with response to platinum compounds and PARP inhibitors; the functional significance of DDR alterations in mCRC is not clear.

Methods

Consecutive patients with unresectable mCRC treated at Memorial Sloan Kettering with first-line chemotherapy whose tumor underwent genomic profiling with MSK-IMPACT⁴ between 2014 and 2017 were analyzed (Supplementary Figure 1A-B). FACETs, an allele-specific copy number algorithm, was used to assess tumor purity and ploidy and to detect regions of loss of heterozygosity⁵. Large state transitions (LST) was defined as a chromosomal break between adjacent regions ≥10 Mb and was estimated for each chromosome arm independently without accounting for centromeric breaks. Progression-free survival (PFS) and overall survival (OS) were calculated from first-line treatment start date using the Kaplan-Meier method.

Results

In microsatellite stable (MSS) unresectable mCRC (n=430), only APC and BRAF alteration frequencies significantly varied by duration of response (Figure 1A) and were associated with PFS and OS (Figure 1B). Multivariate analysis confirmed the association of APC mutation with longer PFS (HR 0.68, p<0.001) and OS (HR 0.56, p<0.001) versus no Wnt pathway alteration or alterations in other Wnt pathway genes (Figure 1C).

Figure 1. Associations of genomic alterations with survival in MSS metastatic colorectal cancer.

A, Frequency of genomic alterations across progression-free survival (PFS) interval groups. B, Kaplan-Meier analysis of overall survival for microsatellite stable patients with oncogenic APC mutations (left panel) or BRAF alterations (right panel). C, Multivariate models incorporating known prognostic genes, such as BRAF and KRAS, and the Wnt pathway genes (APC, AMER1, CTNNB1, RNF43). D, Distribution of mutations in the APC gene and their position relative to the split point for the N-terminal and C-terminal groups. E, Frequency of oncogenic pathway alterations by APC mutation side. F, Frequency of concurrent oncogenic alterations by APC mutation side. G, Distribution of primary tumor locations within N-terminal and C-terminal side APC mutant groups. Abbreviation: NA, not available. Asterisk indicates statistically significant difference.

We first assessed if APC alterations have different associations depending on mutation site. APC truncating mutations were split into N-terminal and C-terminal sides at amino acid (aa) 1400 (Supplementary Methods, Figure 1D, Supplementary Figure 1C-E) where the N- terminal group contains up to two 20R β-catenin-binding repeats. We performed a cutpoint analysis for OS by APC alteration site and inspected hazard ratios across 50 aa intervals. Survival differences were first apparent at aa 600, and survival became significantly shorter in the C-terminal group at aa 1400. Patients with N-terminal side APC mutations had significantly longer OS and PFS versus those with C-terminal side or no APC alteration. Survival outcomes were similar for dual N-terminal side mutations and single N-terminal side mutations; outcomes of patients with either dual C-terminal side APC mutations or N- and C-terminal side mutations were consistent with single C-terminal side mutations. Survival differences by APC mutation site persisted after adjusting for tumor sidedness.

To understand the underlying mechanism for these differences in outcomes, we investigated co-mutation patterns at the pathway and gene level (Figure 1E-F). RAS and phosphatidylinositol 3-kinase (PI3K) pathway alterations were significantly enriched in CRCs with C-terminal side APC mutations; TP53 mutations were more frequent in tumors with N-terminal side APC mutations. KRAS and BRAF alterations occurred in 46% and 2% of N-terminal side, respectively, versus 72% and 13% of C-terminal side APC mutant cases, respectively (p<0.001, q<0.01). PIK3CA and PTEN alterations occurred in 28% and 12% of C-terminal APC mutant mCRC, respectively, versus 11% and 3% of N-terminal side cases, respectively (p<0.01, q<0.1). Overall about three quarters of tumors with C-terminal side APC alterations had concurrent activating mitogenic alterations compared to about half of tumors with N-terminal side APC alterations. APC mutation sites varied by primary tumor location with N-terminal side APC mutations more common in left-sided colon/rectal primaries and C-terminal side APC mutations more common in right-sided colon tumors (Figure 1G).

Our survival analysis suggests that both highly activating CTNNB1 mutations and less activating Wnt ligand alterations identify mCRC with a poor prognosis¹. These alterations commonly co-occurred with additional RAS pathway alterations (63% with CTNNB1, 71% with RNF43) (Supplementary Figure 1E).

DDR pathway alterations were not associated with survival across the cohort or with PFS in patients receiving oxaliplatin-containing chemotherapy (HR 1.1, 95% CI: 0.7–1.6, p=0.74) (Supplementary Figure 2A-D). Alterations in the DDR pathway were enriched in MSI-H mCRC. Median LST scores did not vary significantly between CRCs with somatic or germline DDR pathway alterations and those without alterations in this pathway, or in an analysis limited to somatic DDR alterations. No patient had an LST score greater than 15, which has been considered a cut-off for homologous repair deficiencies.

Discussion

Our data suggest that tumors with “just right” Wnt activation may be less aggressive than those with low or high Wnt activity, where there may be selection for a higher frequency of mitogenic alterations⁶. Weak activators of Wnt signaling, particularly distal APC mutations, appear to experience selective pressure for additional alterations that activate other mitogenic pathways, such as ERK or PI3K signaling. Concurrent activation of these pathways will further promote transcription of shared targets important for proliferation, such as cyclin D. This provides a mechanistic explanation for the shorter survival seen in patients with mCRC with C-terminal APC mutations. In contrast, mCRC with N-terminal side APC mutants may not need these concurrent activating alterations to become full-fledged cancers, leading to relatively less aggressive tumors.

Our analysis of DDR pathway alterations suggests that the presence of these alterations in mCRC do not indicate inactivation of this pathway and a new therapeutic vulnerability. Our findings are consistent with negative studies for PARP inhibitors in mCRC⁷. A recent study of BRCA alterations determined that these alterations often do not co-occur with loss of the second BRCA allele or with homologous recombination deficiency in non-BRCA associated cancer types⁸. Acknowledging the limitations of a small sample size and possible subclonal mutations, our data suggest that DDR pathway alterations, including germline changes, do not identify a distinct subtype of CRC, predict platinum sensitivity, or lead to homologous recombination deficiency.

Our study provides insights into the biology of the most aggressive CRCs. It suggests that we should shift resources from targeting the DDR pathway in this disease, identifies N-terminal side APC alterations as a clinically relevant prognostic marker, and exposes the role of different Wnt alterations in affecting concurrent mitogenic alterations and tumor behavior.

Supplementary Material

s1. Outcomes and concurrent alterations by APC mutation location.

A, Flow diagram of patients included in the study. Abbreviations: CRC, colorectal cancer; mCRC, metastatic colorectal cancer; CRS, cytoreductive surgery; EPIC, early postoperative intraperitoneal chemotherapy; HIPEC, hyperthermic intraperitoneal chemotherapy; BSO, bilateral salpingo-oophorectomy; HAI, hepatic arterial infusion; BSC, best supportive care. B, Clinical characteristics of study population. C, Survival data analysis of different cutpoints in APC gene with divisions at 50 amino acid intervals. Hazard ratios generated from comparing post-cutpoint samples to pre-cutpoint samples. D, Univariate analysis of single and dual C- and N-terminal APC groups. E, Kaplan-Meier analysis of overall survival and progression-free survival by N-terminal and C-terminal APC groups.

s2. Association of DNA damage response (DDR) pathway alterations with outcomes and phenotypic changes.

A, Kaplan-Meier analysis of progression-free survival (PFS) in patients receiving first-line oxaliplatin-containing regimens. Groups were stratified by whether they had an oncogenic alteration in a DDR pathway gene. B, Kaplan-Meier analysis of progression-free survival of mCRC patients, stratified into quartiles based on their large-scale state transitions (LST) scores. Asterisk denotes statistically significant difference at p<0.05 level. C, LST scores across cancer types, stratified by DDR pathway alteration (both germline and somatic variants) status. D, Frequency of DNA damage response (DDR) pathway alterations within microsatellite stable (MSS) and microsatellite instable (MSI) tumors.