From the AGA Journals

Genomic study reveals five subtypes of colorectal cancer

View on the News

Stage now set for functional studies

Genomic, epigenomic, and transcriptomic information has revealed molecular subclasses of CRC, which has refined our understanding of the molecular and cellular biology of CRC and improved our treatment of patients with CRC. Several reliable and clinically useful molecular subtypes of colorectal cancer have been identified, including microsatellite unstable (MSI), chromosomal unstable (CIN), CpG island methylator phenotype (CIMP), and CMS 1-4 subtypes. Despite these substantial advances, it is also clear that we still only partially grasp the molecular and cellular biology driving CRC.

Dr. William M. Grady

The studies by Fennell et al. provide new insights into the CIMP subtype of CRC that address this knowledge gap. Using a large CRC cohort and more detailed molecular information than available in prior studies, they have identified previously unrecognized CRC CIMP subtypes that have unique methylomes and mutation patterns. These 5 CIMP subclasses vary with regard to location in the colon, frequency of mutations in KRAS, BRAF, and MSI, as well as alterations in epigenetic regulatory genes. The observations related to differences in frequencies of MSI, and mutations in KRAS and BRAF help demystify the heterogeneity in clinical and cellular behavior that has been seen in the broader class of CIMP cancers. Perhaps most importantly, their studies identify plausible driver molecular alterations unique to the CIMP subclasses, such as subclass specific mutations in epigenetic regulatory genes and activated oncogenes. These are promising novel targets for chemoprevention strategies and therapies. Fennell and colleagues have now set the stage for functional studies of these molecular alterations to determine their true role in the cellular and clinical behavior of CRC.

William M. Grady, MD, is the Rodger C. Haggitt Professor of Medicine, department of medicine, division of gastroenterology, University of Washington School of Medicine, and clinical research division, Fred Hutchinson Cancer Research Center, Seattle. He is an advisory board member for Freenome and SEngine; has consulted for DiaCarta, Boehringer Ingelheim, and Guardant Health; and has conducted industry-sponsored research for Jannsenn and Cambridge Epigenetic.



Colorectal cancer can be divided into five DNA methylation subtypes that predict molecular and clinical behavior and may offer future therapeutic targets, according to investigators.

In 216 unselected colorectal cancers, five subtypes of the CpG island methylator phenotype (CIMP) showed “striking” associations with sex, age, and tumor location, reported lead author Lochlan Fennell, MD, of the QIMR Berghofer Medical Research Institute in Queensland, Australia, and colleagues. CIMP level increased with age in a stepwise fashion, they noted.

Further associations with CIMP subtype and BRAF mutation status support the investigators’ recent report that sessile serrated adenomas are rare in young patients and pose little risk of malignancy. With additional research, these findings could “inform the development of patient-centric surveillance for young and older patients who present with sessile serrated adenomas,” the investigators wrote in Cellular and Molecular Gastroenterology and Hepatology.

“CIMP can be detected using a standardized marker panel to stratify tumors as CIMP-high, CIMP-low, or CIMP-negative.” In the present study, the investigators expanded these three existing subtypes into five subtypes, allowing for better prediction of clinical and molecular characteristics associated with disease progression.

Initial genomic testing showed that 13.4% of cases carried a BRAF V600E mutation, 34.7% were mutated at KRAS codon 12 or 13, and almost half of the patients (42.2%) had a TP53 mutation. Sorted into the three previously described subtypes, CIMP negative was most common (68.5%), followed by CIMP low (20.4%), and CIMP high (11.1%). About two-thirds (66%) of BRAF mutant cancers were CIMP high, compared with just 3% of BRAF wild-type cases (P less than .0001). KRAS mutated cases were more often CIMP-low than KRAS wild-type cancers (34.6% vs. 12.8%; P less than .001).

With use of Illumina HumanMethylation450 Bead Chip arrays and recursively partitioned mixed model clustering, five methylation clusters were identified; specifically, these were CIMP-H1 and CIMP-H2 (high methylation levels), CIMP-L1 and CIMP-L2 (intermediate methylation levels), and CIMP-negative (low methylation level). As described above, methylation level demonstrated a direct relationship with age, ranging from CIMP-negative (61.9 years) to CIMP-H1 (75.2 years). The investigators also reported unique characteristics of each new subtype. For instance, the CIMP-H1 cluster had many features in common with cases of serrated neoplasia, such as BRAF mutation positivity (73.9%; P less than .0001).

“BRAF mutations are a hallmark of the serrated neoplasia pathway, and indicate that these cancers probably arose in serrated precursor lesions,” the investigators wrote. “We previously showed that the colonoscopic incidence of sessile serrated adenomas does not differ between patients aged in their 30s and patients who are much older, whereas BRAF mutant cancers were restricted to older individuals, suggesting these BRAF mutant polyps may have limited malignant potential in young patients.”

In contrast with the CIMP-H1 cases, CIMP-H2 cancers were more often KRAS mutant (54.5% vs. 17.4%). Other findings revealed associations with subtype and location; for example, CIMP-L1 cases were located equally in the distal and proximal colon, whereas CIMP-L2 cases more often localized to the distal colon and rectum. Of note for CIMP-negative cancers, most (62.3%) occurred in the distal colon, and none had a BRAF mutation.

The five methylation subtypes also showed associations with consensus molecular subtypes (CMS) to varying degrees. The two strongest correlations were found in CIMP-H1 cancers and CIMP-H2 cancers, which were most frequently classified as CMS1 (69.6%) and CMS3 (54.5%), respectively.

Using CIBERSORT, the investigators detected a variety of associations between the five subtypes and stromal immune cell composition. For example, CIMP-H1 cases were enriched for macrophages, compared with the other subtypes, except CIMP-L2. Mast cells showed a stepwise relationship with subtype; they contributed the most to the immune microenvironment of CIMP-negative cancers and the least to cases classified as CIMP-H1. A converse trend was found with natural killer cells.

Of note, in CIMP-H1 and CIMP-H2 cancers, oncogenes were significantly more likely than tumor-suppressor genes to undergo gene body methylation, which is positively correlated with gene expression, and oncogenes in these subtypes had significantly greater gene body methylation than normal colonic mucosa.

“The five subtypes identified in this study are highly correlated with key clinical and molecular features, including patient age, tumor location, microsatellite instability, and oncogenic mitogen-activated protein kinase mutations,” they wrote. “We show that cancers with high DNA methylation show an increased preponderance for mutating genes involved in epigenetic regulation, and namely those that are implicated in the chromatin remodeling process.”

Concluding, the investigators explained the role of their research in future therapy development. “Our analyses have identified potentially druggable vulnerabilities in cancers of different methylation subtypes,” they wrote. “Inhibitors targeting synthetic lethalities, such as SWI/SNF component inhibitors for those with ARID mutations, should be evaluated because these agents may be clinically beneficial to certain patient subsets.”

The study was funded by the National Health and Medical Research Council, the US National Institutes of Health, Pathology Queensland, and others. The investigators disclosed no conflicts of interest.

SOURCE: Fennell L et al. CMGH. 2019 Apr 4. doi: 10.1016/j.jcmgh.2019.04.002.

Next Article: