We investigated changes in O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation status in 21 patients with malignant gliomas who underwent two or more surgeries. Tumor specimens from multiple surgeries were analyzed using methylation-specific polymerase chain reaction (MS-PCR) and immunohistochemistry (IHC) for MGMT. Patients were divided into two groups: Group A (13 recurrent cases with therapeutic intervention) and Group B (8 cases of recurrence without therapeutic intervention or planned two-stage surgeries). MGMT methylation status was compared between initial and recurrent specimens, and in Group A, progression-free survival (PFS) and overall survival (OS) were analyzed according to MGMT methylation status at both initial and recurrent surgeries. By MS-PCR, changes in Group A were as follows: methylated to unmethylated in 23.1%, unchanged in 61.5%, and unmethylated to methylated in 15.4%. In Group B, changes were methylated to unmethylated in 12.5%, unchanged in 87.5%, and no cases of unmethylated to methylated. The difference in proportions between the groups was not significant (p = 0.13). By IHC, changes in Group A were methylated to unmethylated in 7.7%, unchanged in 53.8%, and unmethylated to methylated in 38.5%. In Group B, changes were methylated to unmethylated in 12.5%, unchanged in 87.5%, and no cases of unmethylated to methylated. The difference in proportions between the groups was significant (p = 0.034). Excluding one grade 3 case, in the 12 patients of Group A, median OS from treatment initiation differed by MGMT status assessed by immunohistochemistry in recurrent specimens (26.5 vs. 90.0 months, p = 0.043). These findings suggest that immunohistochemical assessment may capture spatial heterogeneity of MGMT status that is not readily detected by MS-PCR, and that therapeutic intervention may be associated with dynamic changes in MGMT methylation status. Furthermore, MGMT status assessed by IHC in recurrent specimens was associated with overall survival, indicating its potential prognostic relevance.
Ohkuma et al. (Wed,) studied this question.