Impact of somatic mutations on patterns of metastasis in colorectal cancer

Abstract

Somatic mutation status in metastatic colorectal cancer (mCRC) is becoming increasingly clinically relevant as it may be correlated not only with response to biologic therapies, but also with site-specific pattern of metastatic spread and outcome. In this review, we describe our current understanding of associations between mutational activation of the KRAS, BRAF, PIK3CA, and NRAS oncogenes and clinical outcomes and metastatic patterns of mCRC. The presence of a KRAS mutation is associated with a distinct pattern of metastatic spread with decreased liver metastases and increased lung, brain, and bone metastases. In patients who undergo resection of colorectal liver metastases (CLM) with curative intent, KRAS mutation is associated increased risk of recurrence, worse survival, and increased recurrence outside of the liver, particularly in the lung, but also in the brain and bone. BRAF mutation, a poor prognostic factor in mCRC, is associated with decreased liver-limited metastasis and increased peritoneal and distant lymph node metastases. PIK3CA mutation does not clearly affect outcomes in the metastatic setting, but is associated with concurrent KRAS mutations, and has been associated with an increased incidence of lung and brain metastases, metastatic sites preferentially involved in KRAS mutant mCRC. NRAS mutation may confer worse survival and early studies suggest NRAS mutation may promote tumorigenesis in the setting of colorectal inflammation. As metastasectomy with curative intent is increasingly considered in patients with mCRC, understanding patterns of metastasis associated with tumor mutations may help focus medical treatment, surgical management, and surveillance in patients with mCRC.

Introduction

It has long been noted that outcomes for patients with metastatic colorectal cancer (mCRC) vary widely. As sequencing technology has become less expensive and tumor genotyping has become standard practice for mCRC, clinicians now often have information on the mutational status of oncogenes, including the KRAS, BRAF, PIK3CA, and NRAS oncogenes. Many of these mutations represent early events occurring in the adenoma or adenoma-adenocarcinoma transition (1,2), but recent data suggest these mutations may affect the metastatic behavior of tumors and patterns of metastatic spread. In addition to the prediction of response to targeted agents, an understanding of the association between these somatic mutations and site-specific pattern of metastatic spread in mCRC is important as metastasectomy with curative intent is increasingly considered in patients with mCRC and as this may correlate with disease prognosis. Moreover, knowledge of clinical features associated with somatic mutation in mCRC may help focus treatment and surveillance for mCRC patients. In this review, we describe our current understanding of associations between mutational activation of the KRAS, BRAF, PIK3CA, and NRAS oncogenes and clinical outcomes and metastatic patterns of mCRC.

KRAS oncogene

KRAS is the most commonly activated oncogene in colorectal cancer (CRC). Approximately 30-50% of CRCs harbor somatic KRAS mutations, most frequently in exon 2 at codon 12 or 13 and less often at codons 61 or 146 (3). KRAS is a member of the RAS oncogene family. RAS proteins are small GTPases that are active when bound to GTP and regulate cell proliferation, survival, and differentiation. Point mutations in KRAS lead to constitutive activation by preventing hydrolysis of GTP. Currently, evaluation for mutations in KRAS is part of standard of care for mCRC to guide the use of the anti-epidermal growth factor receptor (EGFR) antibodies cetuximab and panitumumab as the presence of a KRAS mutation predicts for insensitivity to these agents (4-8).

KRAS mutant mCRC has been associated with a higher risk of recurrence following resection of hepatic metastases with curative intent, a procedure performed in about 25% of patients with colorectal liver metastases (CLM). Karagkounis et al. found that KRAS mutation status was an independent predictor of recurrence-free survival (RFS) and overall survival (OS) after surgical resection of CLM. Patients with KRAS mutant mCRC experienced worse RFS (HR 1.89) and OS (HR 2.13) after hepatectomy for CLM (9). Similarly, Vauthey et al. found that RAS (KRAS/NRAS) mutation was associated with worse OS (HR 2.3, P=0.002) and overall RFS (HR 1.9, P=0.005) after resection of CLM in multivariate analyses (10). In their dataset of 193 patients who underwent single-regimen modern chemotherapy before resection of CLM with curative intent, the 3-year RFS was 33.5% for RAS wild-type cases and 13.5% for RAS mutant cases (P=0.001). In a population of patients undergoing CLM resection plus adjuvant hepatic arterial infusion (HAI) and systemic therapy, Kemeny et al. identified a significant reduction in RFS at three years for KRAS mutant mCRC compared to KRAS wild-type mCRC (30% vs. 46%, P=0.005) and a trend towards decreased 3-year OS in the KRAS mutant mCRC cases (81% vs. 95%, P=0.07) (11).

The presence of a KRAS mutation may impact recurrence patterns. KRAS mutation has been associated with increased risk for lung recurrence (HR 2.1, P=0.007) in patients with resected stage II and III colorectal tumors (12). In patients undergoing resection of CLM with curative intent, RAS mutation has been associated with worse lung RFS on multivariate analysis (HR 2.0, P=0.01), and in patients undergoing resection of CLM plus adjuvant HAI therapy with curative intent, KRAS mutation was associated with an increased cumulative incidence of recurrence in the lung (58% vs. 33%, P<0.01), brain (14.5% vs. 2%, P=0.05), and bone (13.4% vs. 2%, P<0.01) in comparison to patients with KRAS wild-type tumors (11).

Among all patients with mCRC, the presence of a KRAS mutation appears to influence the pattern of metastatic spread. Tie et al. analyzed the frequency of KRAS mutations in a series of liver, lung, and brain metastases (12). They found that KRAS mutations were less prevalent in liver metastases (32.3%), but more prevalent in lung (62.0%) and brain (56.5%) colorectal metastases (P=0.003). At the time of diagnosis of mCRC, RAS mutant mCRC is more likely to have spread to the lungs compared to RAS wild-type mCRC (22% vs. 13%, P<0.01) (13). RAS mutation is associated with a significantly higher cumulative incidence of lung, bone, and brain metastases after diagnosis of metastatic disease and is an independent predictor of metastasis to these sites (HR 1.5, 1.6, and 3.7 for lung, bone, and brain respectively) (13).

The mechanism by which RAS activation affects metastatic patterns is still unknown. RAS activation has been associated with vascular invasion and hematogenous metastases (14). A recent study focusing on the mechanism of sequential metastasis from the liver to the lung in KRAS mutant colon cancer cell line models found that downregulation of p38 MAPK signaling led to increased lung colonization through increased expression of the cytokine parathyroid hormone-like hormone (PTHLH), contributing to tumor cell extravasation to the lungs through PTHLH-inducted, caspase-independent apoptosis of endothelial cells in the lung microvasculature (15).

BRAF oncogene

BRAF mutations occur in 5-15% of mCRC, and over 95% of these mutations consist of a conversion of valine to glutamic acid at codon 600 (V600E) that leads to constitutive activation of BRAF (5,16). BRAF and RAS mutations in CRC are nearly always mutually exclusive. BRAF mutation is associated with tumor development through the serous serrated pathway, rather than the classic adenoma-carcinoma pathway (17,18).

BRAF mutation has been found to be a poor prognostic factor in multiple clinical trials. While the median survival for patients with mCRC overall has improved to over two years, reported median survival for BRAF mutant mCRC remains less than 1 year (6,19,20). BRAF mutant CRC more frequently demonstrates adverse histologic features such as lymphatic invasion, mean number of lymph node metastases, perineural invasion, and high tumor budding (21). BRAF mutant mCRC is also associated with poorly differentiated morphology, mucinous histology, and signet ring histology (21-23).

BRAF mutant mCRC exhibits a distinct pattern of metastatic spread. These tumors are less likely to present with liver-limited metastasis and are associated with increased incidence of peritoneal and distant lymph node involvement (19,20,22,23), a pattern of metastatic spread that may contribute to the poor outcomes of BRAF mutant mCRC. Patients with BRAF mutant mCRC less commonly undergo metastasectomy (10,19), and BRAF mutation confers a poor prognosis in patients who undergo resection of metastases with curative intent (23). BRAF mutation is also associated with atypical distant lymph node metastasis (24). The mechanism of increased lymph node metastasis in BRAF mutant mCRC is under investigation. Interestingly, a recent study in patients with CRC identifies overexpression of BRAF-activated long non-coding RNA (BANCR) in specimens from patients with BRAF mutant CRC and finds that this overexpression induces the epithelial-mesenchymal transition (EMT) and correlates with increased lymph node metastasis (25).

PIK3CA oncogene

PIK3CA encodes the p110α catalytic isoform of phosphatidylinsolitol 3-kinase (PI3K) and exhibits activating, hotspot mutations in 15-20% of CRCs (16,26). Mutation of PIK3CA is thought to occur during the adenoma-carcinoma transition in the classic adenoma-carcinoma sequence of development of CRC and commonly co-occurs with mutations in KRAS or BRAF (5). About 70% of PIK3CA mutant mCRC cases have concurrent mutations in the classic mitogen-activated protein kinase pathway, predominantly KRAS.

PIK3CA mutation alone has not been found to correlate with patient survival in the metastatic setting (5,27,28). The effect of PIK3CA mutation on metastatic pattern is unclear and is complicated by the high frequency of concurrent KRAS mutations. Our group did not find a correlation of PIK3CA mutation with the cumulative incidence of liver, lung, or bone metastasis, but did find an increased incidence of brain metastasis in mCRC with mutant PIK3CA (1.4% vs. 0.8%, P=0.0013) (13). The association of PIK3CA mutation with brain metastasis appeared to be driven by the high concurrent frequency of KRAS mutation and the increased incidence of brain metastasis in KRAS mutant mCRC as six of the seven cases of brain metastasis that developed in PIK3CA mutant mCRC had concurrent KRAS mutations. PIK3CA mutation has also been associated with lung metastases in mCRC (23% vs. 8.7%, P=0.004), another metastatic site associated with KRAS mutation (22).

NRAS oncogene

NRAS is mutated in approximately 3% of CRCs (5,13). Activating mutations in NRAS are mutually exclusive with BRAF and KRAS mutations in CRC and also predict for lack of activity of anti-EGFR antibodies (6).

NRAS mutation has been associated with decreased OS in small series (29). NRAS activation has been suggested to lead to colorectal tumor development in the context of inflammation (29). Activated NRAS may suppress stress-induced apoptosis in CRC cells. A recent study suggests that in CRC, NRAS signals through STAT3, a transcription factor that normally requires IL-6 to promote growth and survival in a cytokine-independent pathway (29).

Conclusions

Increasing data suggest that the presence of mutations in the RAS and BRAF oncogenes impacts patterns of metastasis. Increased lung, brain, and bone metastasis may be seen in RAS mutant mCRC in addition to increased recurrence after hepatectomy. This recurrence is associated with non-liver metastatic sites, such as the lung. BRAF mutation is associated with increased peritoneal involvement, distant lymph node metastasis, and decreased incidence of liver-limited metastasis, which may contribute to the overall poor prognosis seen in BRAF mutant mCRC. Our expanding understanding of the correlation of somatic mutations with metastatic patterns should contribute to the clinical care of mCRC patients by focusing treatment, defining the role of metastasectomy, and guiding surveillance strategies for patients.