# Defining the recommended gray zone in O6-methylguanine-DNA methyltransferase promoter methylation pyrosequencing reporting: A robust, translatable method to implement new EANO guidelines

## TL;DR

This paper introduces a method to define a gray zone in MGMT methylation reporting, improving survival prediction in glioblastoma patients.

## Contribution

A translatable method to identify the optimal MGMT methylation gray zone using pyrosequencing data.

## Key findings

- The optimal gray zone was identified as 5%–12% methylation, with distinct survival functions for three categories.
- The new categorization outperformed existing methods in predicting patient survival.
- Validation confirmed the method's effectiveness with sufficient sample size and follow-up time.

## Abstract

The DNA repair protein O6-methylguanine-DNA methyltransferase (MGMT) may cause resistance of tumor cells to alkylating agents and is a predictive biomarker in high-grade gliomas treated with temozolomide. Recent European Association of Neuro-Oncology (EANO) guidelines recommend internal validation of MGMT methylation cutoffs and reporting of gray zone values. This study aimed to develop a method to derive a gray zone from pyrosequencing MGMT methylation data.

We developed a method to find the optimal gray zone using pyrosequencing MGMT methylation values (CpG sites 72–83) from 308 glioblastoma cases with overall survival data. Each integer below the methylated threshold defined a new possible gray zone and categorization which was used as a variable in a multivariate Cox proportional hazards regression model. The optimal gray zone was selected as the option that had a statistically different survival function from the methylated and unmethylated groups, with the largest log-likelihood ratio test statistic. We applied the method to a validation cohort of 115 glioblastoma cases.

Our method successfully identified a gray zone in our development cohort. The following categorization gave 3 distinct survival functions: methylated ≥12% (n = 152 cases), gray zone 5%–12% (n = 43), and unmethylated <5% (n = 113). This categorization was better at predicting survival than the existing categorization (methylated ≥12%, unmethylated <12%). Validating our method showed a sufficient sample size and time to follow up is recommended to apply our method.

We have developed a translatable method to identify the optimal MGMT gray zone from pyrosequencing data in line with recent EANO guidelines, to enhance clinical decision-making.