Assessing the numerator and denominator of ‘weakly methylated’ MGMT, including tumour content correction

Aims MGMT methylation status is a key prognostic and predictive biomarker for patients with glioma. However, there is substantial heterogeneity across clinical laboratories regarding the classification of ‘weakly’ or ‘partial’ methylated cases, including the reporting of such a category and the thresholds used to define it. Despite this variability, significant clinical trial data support a benefit from temozolomide for such patients. In this study, we outline the challenges identified in consistent identification of cases with weakly methylated (WM) MGMT, focusing on the inter-replicate variability and result correction based on tumour content. Methods MGMT promoter methylation was interrogated using a commercial assay (EntroGen). Whole slide images scanned on the Grundium Ocus Scanner were analysed in QuPATH using an image analysis algorithm to calculate tumour % and compared with the pathologist’s assessment. Results The WM range of 0.025%–0.08% (corresponding to log ratios of −0.28 to 1.27) demonstrated several challenges: (1) overlaps with range of methylation observable in non-lesional brain tissue (0.011%–0.062%); (2) overlaps with the assay’s lower limit of detection (~0.05%); (3) associated with inter- and intra-run (inter-replicate) heterogeneity; and (4) vulnerable to report category changes, depending on algorithm used for normalising the output with the tumour %. When comparing pathologist-scored and image analysis-derived tumour % quantification workflows, changes to categorical results were observed in 2/76 (~2.6%) of cases. Conclusion WM cases represent true edge cases in MGMT methylation testing, associated with heterogeneous molecular features. Repeat testing and tumour content correction may provide additional data points, particularly in specimens with low tumour content.