MGMT Promoter Methylation as a Target In Metastatic Colorectal Cancer: Rapid Turnover and Use of Folates Alter its Study—Letter

We read with great interest the work by Pietrantonio and colleagues (1). O⁶ methyl-guanine methyltransferase (MGMT) is one of the most important biomarkers in glioblastomas showing predictive and prognostic power and its epigenetic silencing has a critical role in many other cancers (2). Similarly to brain cancer, the hypothesis that reduced activity of O⁶ methyl-guanine methyltransferase (MGMT) enzyme could be associated to increased DNA damage by alkylating agent in metastatic colorectal cancer (mCRC) is highly suggestive and clearly described by the authors who also emphasize the association with RAS status. MGMT methylation was assessed by real-time methylation-specific PCR assay and its expression through IHC. The authors did not find efficacy differences between temozolomide plus capecitabine versus fluorouracil plus irinotecan (FOLFIRI) in second-line treatment of RAS-mutated, MGMT-methylated mCRC. They conclude in the article abstract that temozolomide-based therapy should be further explored in molecularly hyperselected subgroups of patients with mCRC. However, we would discuss some concerns that could affect their results.

Decades of studies demonstrated that cancers are genomically heterogeneous and that this heterogeneity is displayed both in genetics of primary tumors versus matched metastatic lesions and in the same neoplastic masses (3–5). The genetic concordance rates (shared mutations/total number of mutations) between primary and metastatic lesions are extremely variable. It has been hypothesized that, when genetically similar, the cancer might evolve via epigenetic modifications (6). Promoter methylation is an epigenetic mechanism that concurs to change gene expression by the addition of a methyl (CH₃) group to DNA. Recently, we have characterized a series of 15 neuroendocrine tumor patients. A detailed description of methods and results is beyond the scope of this letter. However, MGMT promoter methylation was evaluated in neoplastic tissues from 15 primary tumors through real-time methylation-specific PCR and expressed as unmethylated (UNMET) versus methylated (MET). In four cases, metastatic tissues, at different sites and times in the same patient, were available. Interestingly, the methylation status was not concordant in 3 patients: PT1, rectal NET UNMET, lymph node metastasis MET at presentation; PT2, pancreatic NET MET, livermetastasis UNMET at progression; PT3, pancreatic NET MET, lymph node metastasis MET at progression, liver metastasis UNMET at progression; PT4, gastric NET MET, liver metastasis MET at presentation. These data might account for a significant variability in terms of methylation status of MGMT in space and time potentially affected by folate intake and blood concentrations. Furthermore, as for neuroendocrine tumors, also mCRC may present an about 30% of discordance between MGMT methylation status in primary carcinomas and matched metastases (7, 8). Along with well-known technical issues (i.e., MGMT evaluation in archival tissues), this may account for additional difficulty in assessing MGMT as a reliable marker in mCRC. Genome methylation has the following characteristics: (i) it is highly dynamic, as an example in physiologic condition, it works to direct and reprogram embryos' evolution (9,–11); (ii) it can be rapidly reversed by enzymatic mechanisms (12). In fact, both methylation and demethylation are simultaneously active in the cell nucleus with a rapid turnover. Thus, measurement of MGMT promoter methylation is necessarily affected by these dynamics, as well as by intrinsic tumor heterogeneity and cannot be taken into consideration, on a therapeutic point of view, as the expression of an oncogene (HER2), the presence of a mutation (KRAS, NRAS, BRAF), or a specific biologic status (microsatellite instability).

Furthermore, the study compares, through a phase II randomized design, capecitabine plus temozolomide versus FOLFIRI. The use of folates in the FOLFIRI arm, which implies the use of folinic acid 200 mg/sqm intravenously over 120 minutes on days 1 and 2, could induce a strong (a priori) and uncontrollable (a posteriori) internal bias for study results. In fact, serum folates' level affects the methylations status of DNA decreasing the expression of DNA methyltransferase 1 (DNMT1) protein thus affecting also the methylation status of MGMT (13). Moreover, the effect of diet folates on the methylation status of several genes including MGMT has been already reported and increased folate levels in the blood could induce, at least in our opinion, increased methylation of MGMT in the third tertile in specific CpG sites of MGMT strongly altering its methylation/expression (14, 15). In addition, study of promoter methylation patterns of many different genes has demonstrated a prognostic power in both early and advanced colorectal cancer. This is an indirect proof that folates-induced methylation may alter gene expression in colorectal cancer (16–19). Although randomization still remains controversial in phase II studies, in the randomized ones, the control arm acts as an internal comparator. In fact, the authors clearly stated that their intent, through a phase II randomized design, was to verify whether capecitabine–temozolomide was superior to FOLFIRI. In our opinion, albeit lacking direct data, their “control” arm (FOLFIRI) was reasonably biased over time towards an uncontrolled and broadly hypermethylated status whose implications were completely unpredictable. Notably, no control of the folate blood levels in both arms was made and this could hamper any eventual normalization of folate levels/methylation status.

Therefore, we think that more dynamic evaluations of MGMT promoter methylation (i.e., liquid biopsies) as well as more adequate study designs (use of folates in both arms) should be applied to explore the efficacy of temozolomide in mCRC.

See the Response, p. 3495