Objective:

To assess whether pre-analytic or technical factors explain the reported association between MGMT expression and outcomes in H3 G34-mutant diffuse hemispheric glioma (DHG), and the association between MGMT expression and DNA methylation.

Approach:

• RNA-seq cohort and processing: Reanalyzed RNA-seq data from 110 tumors, all centrally sequenced at St. Jude on NovaSeq 6000. Applied TMM normalization, log2 CPM transformation, and limma batch correction; assessed collection year, preservation method, data release year, and tumor content.
• Outcome analysis: In the 25 patients from the initial report, evaluated continuous MGMT expression and outcomes after adjustment for each of the four factors.
• Methylation analysis: In 34 tumors with expression and methylation data, assessed associations at selected CpG sites before and after accounting for the four factors.

Key Findings:

• Batch correction removed the association between principal components and data release year. Four low-complexity FFPE samples and one fresh/frozen sample remained outliers; excluding them did not change subsequent results.
• Raw-count correlations between MGMT and CD34 or ACTB were weaker than those reported in the reanalysis cited by the authors, and were eliminated after correction.
• After adjustment for each assessed factor, higher continuous MGMT expression remained associated with shorter progression-free survival (HR 1.64 per doubling; 95% CI 1.20–2.24; p = 0.002) and overall survival (HR 3.00; 95% CI 1.73–5.21; p < 0.001).
• None of the four factors was significantly associated with MGMT expression, progression-free survival, or overall survival in the reported analyses. Preservation method was not associated with expression (p = 0.71), though only five tumors were fresh/frozen.
• MGMT expression remained positively associated with methylation at three gene-body/intronic CpGs after adjustment (partial Spearman ρ = 0.58, 0.78, and 0.75; all p < 0.001); promoter CpGs showed negative correlations.

Interpretation:

The authors report that the MGMT–outcome association persisted after accounting for the assessed pre-analytic and technical factors. They note that correlation does not establish regulation of MGMT by gene-body methylation.

Limitations:

• The outcome analysis included 25 patients, and the methylation analysis included 34 tumors.
• Only five tumors were fresh/frozen, limiting assessment of preservation-method differences.
• The study is retrospective; the authors state that MGMT expression requires validation in larger, prospective cohorts before guiding therapy.
• Low-complexity libraries remained outliers after correction, according to the authors.

Conclusion:

The authors state that their findings uphold the initial conclusions and recommend documenting pre-analytic variables, normalizing and correcting batch effects, excluding low-complexity libraries, testing whether adjustment changes marker–outcome associations, and confirming key results with an independent assay.