Advances in CpG Island Methylator Phenotype Colorectal Cancer Therapies

With the aging of the population, the incidence of colorectal cancer in China is increasing. One of the epigenetic alterations: CpG island methylator phenotype (CIMP) plays an important role in the incidence of colorectal cancer. Recent studies have shown that CIMP is closely related to some specific clinicopathological phenotypes and multiple molecular phenotypes in colorectal cancer. In this paper, the newest progress of CIMP colorectal cancer in chemotherapeutic drugs, targeted agents and small molecular methylation inhibitors are going to be introduced. We hope to provide potential clinical treatment strategies for personalized and precise treatment of colorectal cancer patients.

histone modification, nucleosome localization and small noncoding RNAs etc. play a key role in the tumorigenesis of colorectal cancer (3).
In recent years, DNA methylation modification has been extensively studied. In humans, DNA methylation to form 5methylcytosine occurs mainly at the cytosine residues' fifth carbon position of CpG dinucleotides. Around 60-80% of CpG cytosines are methylated in human somatic cells, however CpG islands are regions with a high frequency of CpG sites mostly located near the transcription start site of promoter genes which are constitutively unmethylated. In 1999, Toyota et al. (4) first proposed a novel subset of CRCs positive for CpG island methylator phenotype (CIMP) that extensively displayed multiple cancer specific genes promoter DNA hypermethylation at some specific set of CpG islands in CRC tissues. CIMP is now considered as a distinct molecular subtype of sporadic CRC which is the initial event of the serrated neoplasia pathway in CRC's tumorigenesis (5). CpG island methylator phenotype (CIMP) is mediated by DNA methyltransferases (DNMTs) which promotes hypermethylation in promoter associated CpG-rich regions of tumor suppressor genes which are inactivated by transcription, leading to development and progression of CRC (6). Although there is no consensus on the definition of CIMP and no methodology has been proven to be superior to another (7)(8)(9), CIMP has been still recognized as a hotspot research direction of colorectal cancer in these past 20 years.

CLINICOPATHOLOGIC AND MOLECULAR CHARACTERISTICS IN CIMP CRC
CIMP in CRC was confirmed not only tightly associated with some specific clinicopathologic phenotypes, but also closely related to many molecular characteristics. In 2006, Weisenberger and colleagues (10) recommended a novel and sensitive panel of CIMP including five specific genes' promoter DNA hypermethylation markers to identify CIMP-positive CRCs. The study demonstrated that CIMP-positive was significantly related to female gender, proximal location, MSI-H status, MLH1 methylation, BRAF mutation, and KRAS mutation. In their further large population-based sample analysis, older age, family history of CRC and NSAIDs using before diagnosis related to CIMP-positive were additionally observed. Furthermore, smoke and overweight statistically associated with only female CIMP-positive CRCs were reported (11). Ogino and colleagues (12) selected different promoter loci as CIMP panel markers to identify CIMP-High, CIMP-Low and non-CIMP phenotypes. Follow-up experiments showed that CIMP-Low tumors different from CIMP-High and non-CIMP tumors were tightly correlated to male gender and KRAS mutation (13). In further CIMP subgroup analysis, CIMP-positive tumors independent of MSI status were significantly associated with mucinous or signet ring cell morphology, tumor infiltrating lymphocytes (TIL), peritumoral lymphocytes (PLS), presence of Crohn-like infiltrates, tumor necrosis, tumor cell sheeting, and poor differentiation (14). Interestingly, low level intake of folate was proven to be associated with a trend towards an increased risk of non-CIMP-High colon tumors (15). Recent years, CIMP status was also reported to be positively correlated with F. nucleatum, the gut microbiome component in CRC (16,17). Using more advanced methods or meta-analysis to evaluate the relationship between CIMP status and the clinicopathological and molecular characteristics in CRC, the conclusions obtained by Guinney et al. (18) and Advani et al. (19) were similar to those of Weisenberger and Ogino. In the past 20 years, CIMP-positive tumors were accepted as a consensus which were positively correlated with female, proximal location, MSI-H status, BRAF mutation and mucinous histology. These epidemiology associations help us explore the underlying cancer prevention and treatment strategies.

CIMP AND COLORECTAL CANCER TREATMENT
A growing number of studies suggest that CIMP might be a potential epigenetic predictor or prognostic biomarker contributed to individualized and precise treatment of colorectal cancer patients (20,21).

CIMP and Chemotherapy
As the most widely used chemotherapeutic drug for colorectal cancer, 5-fluorouracil (5-FU) principally acts as a thymidylate synthase (TS) inhibitor by interrupting DNA replication. In 2007, Shen et al. (22) investigated CIMP status in 188 advanced CRCs who received 5-FU based chemotherapy and found that the median survival in the CIMP-positive subset was 6 months versus 17 months in CIMP-negative subset (P < 0.001) and twoyear survival rate was 8% in the CIMP-positive group versus 28% in the CIMP-negative group ( Table 1). In multivariate analysis, CIMP-positive cases had a significantly shorter survival (hazard ratio, HR=2.9; P < 0.0001). Jover et al. (23) studied 196 stage II-III CRCs and found that CIMP-positive CRCs did not benefit from 5-FU based adjuvant treatment. The disease-free survival (DFS; log-rank=0.02) of CIMP-positive patients receiving adjuvant 5-FU based chemotherapy was lower than that of CIMP-negative patients. In CIMP-negative CRCs, adjuvant 5-FU based chemotherapy significantly prolonged DFS (log-rank=0.00001). However, it failed to improve DFS (log-rank=0.7) in CIMPpositive CRCs. Multivariate analysis showed adjuvant 5-FU based treatment was not an independent predictor of prognosis in CIMP-positive CRCs (HR=0.8; 95% confidence interval, CI=0.3-2.0). Min and colleagues (24) reached the opposite conclusion by an independent Asian population clinical trials. They performed 124 stage II-III CRCs and reported that CIMPhigh CRCs (n=17; 3-year recurrence-free survival, RFS: 100%) who received 5-FU based regimen after surgery had significantly better RFS than those accepted surgery alone (n=7; 3-year RFS: 71.4%; P=0.022). Furthermore, Rijnsoever et al. (25) considered that CIMP-positive status was an independent significant predictor for the survival benefit treated with adjuvant 5-FU based chemotherapy in CRCs. They evaluated CIMP-positive status in 103 stage III CRCs treated with surgery alone and 103 The results were demonstrated that the OS were significantly different among the three CIMP groups with a median of 9.77, 22.2, and 35.7 months for the high, low and negative groups, respectively (P< 0.001). In 5-FU and oxaliplatin first-line chemotherapy (n=128), the median OS was 6.77, 23.8, and 37.9 months for the high, low and negative groups, respectively (P<0.001), while the median PFS was 1.83, 7.87 and 9.97 months, respectively (P=0.002). CIMPhigh cases were significantly associated with worst efficacy of therapy. In 5-FU and irinotecan second-line chemotherapy (n=86), only the median OS was shown a significant difference according to the CIMP status with values of 2.90, 13.4, and 20.4 months for the high, low and negative groups, respectively (P<0.001). The CIMP-high status was considered as a negative prognostic factor for metastatic CRCs received with chemotherapy. Bae and colleagues (33) inspected 1,370 stage I-IV CRCs treated with surgery and/or chemotherapy. Compared with CIMP-P1 (CIMP-L), CIMP-negative CRCs showed better 5-year cancerspecific survival (CSS; HR=0.47; 95% CI: 0.28-0.78) and better 5year RFS (HR=0.50; 95% CI: 0.29-0.88). The CIMP-H CRCs displayed best 5-year CSS from chemotherapy was observed, however no such trend was found in no chemotherapy analysis.
Multiple clinical studies seemed to reach an agreement on CIMPpositive CRCs associated with poor survival, but failed to display a prognostic value of CIMP-positive CRCs who were treated with oxaliplatin-based adjuvant chemotherapy regimen. Albumin-bound paclitaxel is an anti-microtubule drug, which interferes with the rearrangement of microtubules, leading to the cessation of mitosis, thus inhibiting the growth of cancer cells. Overman et al. (34) reported a phase II clinical trial which enrolled 21 CIMP-high metastatic CRCs and no efficacy of nab-paclitaxel was observed. Oxaliplatin-based or paclitaxel-based chemotherapy might be indirectly affected by CRCs' CIMP status.
The DNA repair gene O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation is a frequent and early event in colorectal tumorigenesis which was considered benefit from alkylating agents such as temozolomide (TMZ) (38). In a phase II study, TMZ showed a modest activity and achieved an average 10% RR in heavily pretreated metastatic CRC patients with MGMT hypermethylation (39). A recent study was shown that irinotecan and TMZ (TEMIRI) combination regimen was reached the primary end point in irinotecan-sensitive, MGMT methylated and MSS pretreated metastatic CRC patients (40). Six out of 25 patients achieved PR (ORR=24%; 95% CI, 11-43%). The mPFS and mOS were 4.4 and 13.8 months, respectively. All patients with MGMTpositive IHC were non-responders. Consistently, patients with MGMT-negative/low tumors had a significantly longer mPFS than others (6.9 vs 2.0 months; HR = 0.29, 95% CI 0.02-0.41; P = 0.003). The reason of the efficacy of TEMIRI regimen for metastatic CRC patients with MGMT methylation and absent/low might be the inhibition of topoisomerase II enhances the cytotoxicity of alkylating agents.

CIMP and Targeted Therapy
Cetuximab, an epidermal growth factor receptor (EGFR) inhibitor, is an IgG1 monoclonal antibody specifically targeting EGFR overexpression and widely used in metastatic CRCs. In the 2 studies described above, Zhang et al. (27) demonstrated that the PFS of CIMP-positive and KRAS wild-type CRCs who treated with cetuximab was shorter than that of CIMP-negative CRCs with KRAS wild-type (mPFS, 2.1 vs. 5.1 months, P = 0.11) and objective response rate (ORR) was also decreased (20.0 vs. 24.4%, P = 0.90). Although this study did not show statistical significance, it seemed to suggest that CIMP-positive phenotype might be a biological negative predictor of efficacy of anti-EGFR antibody ( Table 2). Also Gallois and colleagues (31) concluded that the application of cetuximab in CIMP-positive stage III CRCs bought a non-significant trend of negative efficacy. Ouchi et al. (41) analyzed 97 KRAS wild-type metastatic CRCs received anti-EGFR antibody by advanced genome-wide DNA methylation technique and divided patients into highly methylated epigenotype (HME), intermediate methylated epigenotype (IME), and low methylated epigenotype (LME). The results were shown that ORR (35.7 vs 6.3%, P = 0.03), disease control rate (DCR; 75 vs 31.3%, P = 0.005), PFS (HR = 0.22; 95% CI, 0.13-0.57; P < 0.001) and OS (HR = 0.19; 95% Cl, 0.06-0.54; P < 0.001) were significantly better in LME subgroup CRCs compared with HME subgroup CRCs. Although merely one study displayed statistical significance, we still believe that CIMP-positive phenotype is a negative efficacy indicator for KRAS wild-type metastatic CRCs who received anti-EGFR antibody. First, CIMP was significantly associated with the right side CRC which was considered as an independent negative prognostic factor to cetuximab therapy (42,43). A clear mechanistic understanding to explain the worse outcome of CIMP-positive patients is currently lacking. EGFR promoter hypermethylation status may be responsible and has been reported to be more relevant than primary right colon in the prediction of negative response to anti-EGFR therapy in patients with metastatic CRCs (44). Secondly, we speculated that some hypermethylation genes increased the resistance of anti-EGFR antibody and lead to drug resistance. Demethylation agents were given to CIMP-positive CRCs which might reverse the resistance and improve patients' survival. This is a potential research direction to change the therapeutic strategy of metastatic CRC in the future.

CIMP and DNA Methylation Inhibitor
DNMT1, DNMT3A and DNMT3B are the canonical cytosine-5 DNMT enzymes. Their functions include not only the establishment and maintenance of DNA methylation patterns, but also the regulation of multiple gene functions, including transcriptional silencing, transcriptional activation and posttranscriptional regulation (45). Most widely studied DNA methylation inhibitors were 5-azacytidine (Azacitidine, 5-Aza-CR), 5-aza-2'-deoxycytidine (Decitabine, 5-Aza-CdR) and guadecitabine (SGI-110) formed irreversible complexes with DNMTs by substituting methylated cytosine targets during DNA replication, leading to the depletion of the enzyme and cytosine during cell division, passive DNA demethylation, tumor suppressor genes' re-expression, proliferation control and carcinogenesis process inactivation (46,47). It was suggested that CIMP-positive CRCs might potentially benefit from the treatment with DNA methylation inhibitors alone or combination (48,49). Garrido-Laguna et al. (50) confirmed 20 patients with KRAS wild-type metastatic CRC receiving sequential decitabin and panitumumab were well tolerated in phase I/II clinical trials ( Table 3). Two of patients previously received cetuximab had a partial response (PR). Ten patients had stable expression indicated decitabine combined with immunotherapy could be candidate treatment in the further study. From the above studies, it seemed that DNA methylation inhibitors combined with the traditional chemotherapy were shown no significant effective, while the efficacy of combination anti-EGFR antibody or immunotherapy seemed to be worth looking forward to. Although low-dose DNA methylation inhibitors show demethylation and promote apoptosis, inhibiting DNMT alone may not be sufficient to induce durable and robust transcriptional gene re-expression. DNA methylation inhibitors as immune modulators have been consequently considered inducing CTAs expression in CRC which stimulate cytotoxic T-cell responses and antitumor immunity (54). Furthermore, DNA hypermethylation of tumor-infiltrating immune cells or their ligands (e.g. PD-1, CTLA-4, TIM-3, TIGIT, PD-L1, and galectin-9) leading to tumor evasion from host immunosurveillance could be major contributors to the upregulation of immune checkpoints (55). Combining immunotherapy to evaluate the activity of DNA methylation inhibitors is the most promising research direction in the future. In addition, epigenetic therapies included not only DNA methylation inhibitors, but also HDAC inhibitor, BET inhibitor and EZH2 inhibitor. Different epigenetic alterations should be given appropriate interventions. Development of these therapies will provide exciting opportunities for novel and improved therapeutic interventions in CRC (56).

CONCLUSION
In summary, we reviewed the latest progress of CIMP CRCs' characteristics and treatment. We clearly realize that CRCs with CIMP phenotype are tightly related to the pathological features of female, older age and right side colon, as well as molecular characteristics of BRAF mutation and MSI-H status. Certainly, with the wide application of next generation sequencing technology, more accurate method to distinguish CRCs' CIMP status will be are constantly emerging. For chemotherapy, CIMPpositive CRCs were potentially benefit from irinotecan-based regimen rather than oxaliplatin-based regimen. For targeted therapy, negative efficacy from anti-EGFR antibodies seems to be associated with CIMP-positive CRCs. However, the mechanism of these phenomena needs to be further explored in the future. Clinicians are increasingly aware of the importance of CIMP phenotype in CRC. A various of DNA methylation inhibitors alone or especially combination with immunotherapy are undergoing clinical trials. These frontier studies provide potential individualized precise treatment strategies for patients with CRC.

DATA AVAILABILITY STATEMENT
All data collected, generated, or analyzed during this study are included in this published article.