Investigation of colorectal cancer in accordance with consensus molecular subtype classification

Abstract The classification of colorectal cancer (CRC) plays a pivotal role in predicting a patient's prognosis and determining treatment strategies. The consensus molecular subtype (CMS) classification system was constructed by analyzing genetic information from 18 CRC data sets, containing 4151 CRC samples. CRC was classified into four subtypes with distinct molecular and biological characteristics: CMS1 (microsatellite instability immune), CMS2 (canonical), CMS3 (metabolic), and CMS4 (mesenchymal). Since their designation in 2015, these classifications have been applied to basic and translational research of CRC, with the hope that understanding these subsets will influence a clinician's approach to therapeutic treatment and improve clinical outcomes. We reviewed CRC investigations in accordance with CMSs published in the last 5 years to further explore the clinical significance of these subtypes and identify underlying trends that may direct relevant future research. We determined that CMSs linked common features of CRC cell lines and PDX models in various studies. Furthermore, associations between prognosis and clinicopathological findings, including pathological grade and the stage of carcinogenesis, tumor budding, and tumor location, were correlated with CMS classification. Novel prognostic factors were identified, and the relationship between chemotherapeutic drug resistance and CMS has been fortified by our compilation of research; thus, indicating that this review provides advanced insight into clinical questions and treatment strategies for CRC.


Immune acƟvity Strong Weak
DifferenƟaƟon poor well tumors decreased and were replaced by murine counterparts in subsequent passages. 8 20 Another cohort also demonstrated that very few cases classified as CMS3 represent metastatic lesions (<1%) compared to primary lesions, of which CMS3 comprises 11%. 21 These findings indicated that CMS3 may become uncommon as CRC progresses ( Figure 3).

| C ARCINOG ENE S IS , C AN CER PROG RE SS I ON , AND ME TA S TA S IS IN ACCORDAN CE WITH CMSS
Interestingly, intra-patient heterogeneity between primary lesions and peritoneal metastatic lesions was frequently observed in a cohort of 28 patients, three-quarters of which were diagnosed with CMS4 peritoneal carcinomatosis. Fifteen of the 16 patients with paired tumors, a primary lesion and one to four metastatic lesions, had at least one CMS4-positive tumor. 22

| PROG NOS IS IN ACCORDAN CE WITH CMSS
Many prognostic factors were not associated with the CMS classification. NUSAP1, CD44, and COL4A1 have been detected among all CMSs and play a key role in CRC progression. NUSAP1 regulates BRCA1 protein levels, CD44 presents as an EMT marker, and COL4A1 is a tumor angiogenesis indicator. 23 Though many markers have been identified and subsequently associated with CMSs, several prognostic genes were discovered in studies that focused on CMS classification. We reviewed prognostic genes in accordance with CMSs as follows (Table 1 and Figure 4).

| Prognostic factors in CMS1
In CMS1, microsatellite stable (MSS) CRC patients with BRAF mutation are associated with a shorter overall survival (OS) compared with BRAF wild type; however, prognostic difference between BRAF mutation or BRAF wild type were not found in MSI CRC patients. 24 In the BRAF mutated metastatic CRC patients, CDX2 loss and CK7 positivity indicated unfavorable prognosis. 25 Loss-of-function mutations of JAK1 are found in 20% of CRCs.
These tumors show elevated transcriptional signatures that are associated with resistance to anti-programmed death-1 treatment.
Among the MSI tumors, the total mutation load correlated with the number of predicted neoantigens, but not with immune cell infiltration, which was dependent on the CMS. CMS1 in particular had higher immunogenic features compared with CMS2-4. 28 Additionally, the expression of Annexin A2 (ANXA2), which is associated with endocytic and exocytic events and cytoskeleton regula-

| Prognostic factors in CMS4
The and these genes were associated with poor prognosis in CMS4. 43 The HCAR3 module was associated with favorable prognosis in CMS4. HCAR3 acts as a tumor suppressor and has been implicated in multiple interactions as well as in the development of anti-cancer drugs. 23

| S EN S ITIVIT Y TO C Y TOTOXI C DRUG S IN ACCORDAN CE WITH CMSS
The effects of anti-cancer drugs differ according to CMS classification (Table 2 and Figure 5). The sensitivity of CRC cell lines to the cytotoxic drugs, 5-fluorouracil (5-FU) and L-OHP, was reported in accordance with CMSs. The inhibitory concentration 50 (IC50) of 5-FU was lower to treat the CMS1-3 cell lines compared with CMS4.
The ratio of apoptotic cells in CMS2 was high after treating with 5-FU combined with L-OHP in comparison with CMS4. The CMS2 PDXs with 5-FU and L-OHP treatment resulted in longer survival than the placebo treatment; however, the survival of CMS4 PDXs showed no benefit with the combined treatment. 44 In PDXs, CMS1/4 had a poor response to 5-FU compared with CMS2/3. 45,46 In stage II and III, CRC patients classified as CMS1/4, no survival benefit was conferred by ACT. The prognosis of the stage II CRC CMS2 patients was improved by ACT, and the survival of the stage III CRC patients in CMS2/3 had a favorable prognosis in those receiving ACT compared to those who only received surgery. The stage II and III CRC CMS2 patients with CD8 positivity had a favorable prognosis without ACT. 35 In a different cohort, the stage III CMS2 patients with 5-FU and L-OHP treatment had a more favorable prognosis than those receiving 5-FU monotherapy. More specifically, only the enterocyte subtypes of CRC Assigner (CRCA) classification in CMS2 patients had a significant benefit from the 5-FU plus L-OHP treatment. The benefit of adding L-OHP to  was not shown in the other subtypes. 47 ZEB2 positivity was found in CMS4 and was elevated after L-OHP treatment. High ZEB2 expression correlated with reduced proliferation, however, ZEB2 positivity was also associated with resistance to chemotherapy and poor prognosis. 48 Regarding first-line chemotherapy, Irinotecan-based regimens were significantly superior to L-OHP-based regimens for PFS in CMS4. TOP1 and CES2 expression are predictive biomarkers for F I G U R E 5 Therapeutic effects of cytotoxic drugs and molecular target agents in accordance with CMS. Therapeutic effects of cytotoxic drugs were shown. CMS4 had a poor response to 5-FU compared with CMS2. CMS2 was a longer relapse-free survival (RFS) compared with CMS4 (A). Therapeutic effects of molecular target agents were shown. CMS2 responded most significantly to anti-epidermal growth factor receptor antibody, cetuximab, treatment compared with the other CMSs. CMS1 had the least significant response. CMS1/3 had a poor response to anti-vascular endothelial growth factor, bevacizumab, compared with CMS2/4 (B

| S PECIFI C MOLECUL AR TARG E TED AG ENTS AND CMSS
In an investigation with PDXs, CMS2 responded most significantly to the cetuximab treatment compared with the other CMSs, and CMS1 had the least significant response. 45 53 Both anti-EGFR antibody and anti-VEGF antibody treatments offered fewer benefits for CMS1/4 compared with CMS2/3. In the FIRE3 trial, prognostic difference between anti-EGFR or anti-VEGF agents were not found in accordance with CMSs. CMSs did not impact selection of these anti-molecular target agents in first-line chemotherapy 54,55 (Table 3 and Figure 5). Some CRC revealed intratumor heterogeneity. Multiple conflicting subtype assignments were sampled based on the tumoral region during tissue collection, using stromal-based classifiers like CMS specifically when using biopsy samples. 16 While clinical trial data was reanalyzed in accordance with CMS classification, further research will be necessary to translate the genetic data to clinical practice. Further investigation and analysis of CRC in accordance with CMSs will enable physicians to provide optimal, personalized treatment options and ultimately improve clinical outcomes for CRC patients.

CO N FLI C T O F I NTE R E S T
Authors declare no conflict of interests for this article.