Experts in the field assert that the path forward requires a paradigm shift toward integrative analyses that encompass multiple classes of genomic aberrations and consensus classification of CRC based on genomic data to facilitate more effective management of this disease.
Although colorectal cancer (CRC) was among the first solid tumors to undergo molecular profiling, the clinical translation of this knowledge into effective therapies has been impeded by the startling level of complexity and heterogeneity revealed among these tumors. Despite the FDA’s approval of several new drugs in recent years, the success of these and other agents in development has been stifled by the complex nature of CRC. As a result, experts in the field assert that the path forward requires a paradigm shift toward integrative analyses that encompass multiple classes of genomic aberrations and consensus classification of CRC based on genomic data to facilitate more effective management of this disease.Every case of CRC is associated with genomic alterations that drive dysregulation of key signaling pathways involved in the hallmark processes that are essential to the transformation of a normal cell into a cancerous one.In the vast majority of cases, this results from the sporadic accumulation of somatic mutations over a patient’s lifetime. However, in approximately 5% of patients, CRC arises as part of an inherited syndrome that results from a germline mutation in a well-defined gene (hereditary). A further 10%-30% of CRC cases are associated with increased familial risk depending upon the analysis, but have no identified germline mutation (Figure A).
Several subtypes of hereditary mutations have been characterized, and protocols have been developed for screening patients with CRC at risk for these syndromes or individuals from families with a known high-risk syndrome associated with CRC.
The most common inherited subtype is Lynch syndrome, often called hereditary non-polyposis CRC, which results from germline mutations in one of several mismatch repair (MMR) genes and is believed to occur in approximately 2% to 5% of all CRC cases. Familial adenomatous polyposis (FAP), believed to occur in about 1% of CRC cases, stems from mutations in the adenomatous polyposis coli (APC) gene.Regardless of whether mutations present in CRC are inherited, the genomic instability that fosters cancer development can arise via several distinct pathways. Most common (about 70%-80% of cases) is the chromosomal instability (CIN) pathway, in which there is an accumulation of numerical or structural alterations to the chromosomes. These tumors also harbor somatic mutations in a number of oncogenes and tumor suppressors that activate signaling pathways with a critical role in carcinogenesis.
In about 15% of cases, colorectal carcinogenesis results from insertions or deletions in the microsatellites, which are short stretches of repetitive DNA scattered throughout the genome.
This microsatellite instability, which changes the length of the microsatellites, occurs predominantly because of defects in the MMR system. The MMR system is responsible for double checking the work of the DNA polymerase enzyme that synthesizes an identical copy of the DNA during cell replication. As a result of a malfunctioning MMR system, any errors introduced during replication cannot be corrected, impacting gene expression and function.Adding another layer of complexity, a third pathway recently has been described, reflecting increased appreciation of the importance of epigenetic alterations to the genome. Epigenetics describes the regulatory mechanisms that affect gene expression without altering the DNA sequence, most commonly through the addition of methyl groups (methylation).
CpG islands are rare, short DNA sequences that contain an unusually high frequency of a cytosine being followed by a guanine; hence, CpG, a cytosine and guanine separated by only one phosphate. As a result of this pattern of base pairs, CpG islands have a tendency to be methylated. When methylated, CpG islands that are located within the promoter region of genes silence the expression of that gene, thus tightly regulating the gene’s methylation status. Aberrant methylation of CpG islands in the promoter region of tumor suppressor genes can promote tumorigenesis via the CpG Island Methylation Pathway (CIMP).Several key genes involved in the carcinogenic process in CRC have been identified and include a number of well-known tumor suppressors and oncogenes (Figure B). The discovery of these driver genes has in some cases led to the development of drugs targeting the proteins encoded by these aberrant genes or components of the dysfunctional signaling pathways in which they play a role.
Components of the Wnt signaling pathway, most commonly the APC gene, are frequently mutated in patients with CRC. Indeed, the initial step in CRC development in some 80%-90% of CRCs is thought to be loss of the APC gene. APC acts as a negative regulator of beta-catenin, targeting it for destruction by the proteasome; therefore, the loss of the APC gene leads to stabilization of beta-catenin, which is then able to move into the nucleus where it acts as a transcriptional coactivator to upregulate expression of Wnt-target genes.
1. National Cancer Institute. Genetics of colorectal cancer for health professionals (PDQ). http://goo.gl/FF1W6e. Updated January 15, 2016. Accessed February 8, 2016.
2. My Cancer Genome. http://goo.gl/2gnF4M. Accessed January 29, 2016. 3. Guinney J et al. Nature Med. 2015;21(11):1350-1356.
Another commonly affected signaling cascade in CRC is the transforming growth factor beta receptor (TGFβR) pathway. The pathway is initiated by TGFβ binding to the TGFβR, which subsequently phosphorylates a family of intracellular proteins called SMADs. The receptor-regulated SMADs (SMAD1, 2, 3, 5 and 8) bind to SMAD4 to form a complex that enters the nucleus and regulates transcription.
Mutations in SMAD4 are found in between 10% and 35% of CRC samples, depending upon the analysis. Indeed, activation of the Wnt signaling pathway or inactivation of TGFβ signaling, resulting in increased activation of the transcription factor MYC, are near-ubiquitous events in CRC. Mutations have also been identified in KRAS and BRAF, both involved in the mitogen-activated protein kinase (MAPK) pathway, and in the tumor suppressors TP53 and PTEN.In an effort to further molecularly characterize CRC tumors, genomewide sequencing studies have been performed. The most comprehensive study to date was published by The Cancer Genome Atlas (TCGA) in 2012. Researchers performed whole-exome sequencing, as well as DNA copy number, promoter methylation, mRNA and microRNA expression analyses on 224 colon cancer samples and matched pairs from normal tissue.
The samples were divided into hypermutated (16%) and non-hypermutated (84%), based on the degree of MSI and the presence of mutations in the MMR pathway, with the hypermutated tumors more representative of MSI and the non-hypermutated tumors of CIN. A total of 15 and 17 genes in the hypermutated and non-hypermutated tumors, respectively, were found to be significantly mutated.
The vast majority of samples (93% of the non-hypermutated tumors and 97% of the hypermutated tumors) had a mutation in one or more members of the Wnt signaling pathway, most commonly in the APC gene. Mutations were found not only as expected in genes such as TP53, SMAD4, and KRAS, but also in novel genes such as FAM123B, a negative regulator of Wnt signaling; ARID1A, a tumor suppressor; and SOX9, which encodes a protein that plays a critical role in embryonic development. Mutations in the latter previously have not been identified in cancer.
Mutations in BRAF and TGFRB2 were found in around half of hypermutated samples, but in under 5% of the non-hypermutated samples, while TP53 and APC mutations were more frequently observed in non-hypermutated tumors. MAPK activation was common in the nonhypermutated samples, typically as a result of KRAS and PIK3CA mutations. These variations suggested that carcinogenesis occurred via different sequences of genomic alterations in hypermutated and non-hypermutated tumors. Interestingly, the TCGA study also suggested that, although patients with colon and rectal cancers are treated differently and display epidemiological differences, there were no differences between samples from the two anatomical locations at the molecular level when only non-hypermutated samples were considered.As the third most commonly diagnosed cancer worldwide, with more than 1 million new cases diagnosed annually, CRC represents a significant burden and has been the focus of extensive research efforts. When diagnosed early, a goal that has been substantially aided by the development of effective colonoscopy screening efforts, therapy can be curative. Yet still more than a quarter of patients are diagnosed with advanced or metastatic disease, which is significantly harder to treat.
During the past decade, advances in combination chemotherapy and the development of biological agents targeting key drivers of colorectal carcinogenesis have improved survival for patients with advanced disease. Drug development efforts have focused on targeting growth factor receptors that are commonly overexpressed on the surface of advanced-stage tumors.
Most of the FDA-approved agents are directed at the vascular endothelial growth factor (VEGF) pathway inhibitors; these include bevacizumab (Avastin), ramucirumab (Cyramza), regorafenib (Stivarga), and ziv-aflibercept (Zaltrap). Also approved are the epidermal growth factor receptor (EGFR) pathway inhibitors cetuximab (Erbitux) and panitumumab (Vectibix), with companion diagnostics that test for KRAS and EGFR mutations available for both agents. Cetuximab is specifically indicated for patients with KRAS wild-type, EGFR-expressing metastatic CRC. Targeting driver mutations has proved much less fruitful to date. Most advanced is the development of BRAF inhibitors, which have revolutionized the treatment of patients with BRAF-mutant melanoma. However, despite harboring the same BRAF V600E mutations as melanoma, CRC tumors are significantly less responsive to BRAF inhibition.
Recently, studies have suggested that BRAF inhibition activates the EGFR pathway and that combining BRAF inhibitors with drugs targeting the EGFR and phosphatidylinositol-3-kinase (PI3K) pathways has shown synergy in preclinical trials.
Phase II trials of combination therapies are ongoing. These include a trial testing the BRAF/MEK/EGFR triplet of dabrafenib, trametinib, and panitumumab (NCT01750918), the pairing of cetuximab and the BRAF inhibitor vemurafenib (NCT02164916), and cetuximab in combination with the novel RAF inhibitor LGX818 and the PI3K inhibitor BYL719 (NCT01719380).
Genome sequencing studies like the one carried out by the TCGA have revealed many other potentially druggable pathways, including the Wnt, PI3K, TGFβR, and p53 pathways. Drugs targeting components of these pathways are being evaluated in preclinical or early clinical trials in patients with advanced solid tumors, some including patients with colorectal cancer.A significant obstacle to development of targeted therapies for the treatment of patients with CRC is a lack of biomarkers that enable researchers to accurately predict which patients are going to benefit from a particular treatment regimen.
Predictive biomarkers for EGFR-directed therapy have met with the most success; activating mutations in KRAS exon 2 have been shown to predict a lack of response to these agents. Molecular testing for KRAS mutations and KRAS wildtype status is required for the use of these drugs in the clinic. However, this does not guarantee a response, and identifying predictive biomarkers beyond KRAS is a key area of research. Furthermore, several studies have shown that clinicians should be expanding testing beyond KRAS exon 2 mutations. Other RAS mutations may also predict a lack of response to anti-EGFR therapy, and not all exon 2 mutations may confer resistance. Finally, it has recently been suggested that a “quadruple negative” genotype—KRAS exon 2, BRAF V600E, PIK3CA exon 9, and PTEN mutations—could be considered as a biomarker of sensitivity to anti-EGFR therapy.In addition to identifying predictive and prognostic biomarkers, classification of patients according to molecular features can help to stratify individuals for therapeutic decision making and could render treatments more precise. There have been many efforts to categorize CRC according to genomic characteristics, using different algorithms and sample sizes, but clinical translation of a classification system has been hindered by the failure of these studies to reach a consensus regarding the heterogeneity of CRC.
In an attempt to achieve a consensus classification system, members of the research community formed an international group, the Colorectal Cancer Subtyping Consortium (CRCSC), whose aim is to share datasets and analyses across areas of expertise to assess and establish the molecular subtypes of CRC. The efforts of the CRCSC recently culminated in the publication of a consensus classification for CRC based on genomic data, in which more than 4000 patient samples were examined. The study, published in Nature Medicine, proposes four consensus molecular subtypes (CMSs) (Figure C). These subtypes, their prevalence, and their characteristics include:
The classification system must still be validated in clinical trials before it can be translated into practice, but it represents the most robust method currently available for organizing CRC mutations.A key take-home message from this study and others evaluating molecular classification systems is that none of the subgroups is identified by a single clear genomic or epigenomic marker or group of markers, as has been observed in other cancer types. This has important implications for anticancer strategies.
For example, although wild-type RAS tumors are currently considered as one group for therapeutic decision-making purposes, they were actually found across several subgroups that are profoundly different in terms of their biological makeup, which could impact drug response.
This could potentially help to explain why wildtype KRAS is not sufficient to predict response to anti-EGFR therapy. It also implies that molecular classification based on multiple different features has greater potential to refine disease stratification for treatment purposes compared with currently validated biomarkers.
Jane de Lartigue, PhD, is a freelance medical writer and editor based in New Haven, Connecticut