The Impact of Methylation Array Analysis on the Diagnosis of Central Nervous System Tumours: DNA Quality, Concordance and Cost Efficiency

Abstract Aim Genome-wide DNA methylation arrays are commonly used in diagnostic practice to classify central nervous system tumours; they are still regarded as a research tool and are not clinically validated. Epigenetic profiles may not match the pathology or return identities with low calibration scores (low confidence). We conducted a multi-centre audit on the impact of methylation arrays on brain tumour diagnosis, including DNA quality and cost efficiency. Method Seventy-six tumours were analysed between January 2020 and June 2022 using the v11b.4 Heidelberg classifier. Tumour pathology, methylome and calibration scores (high = ≥0.84, low = 0.30-0.83) were extracted and concordance assessed (histological versus epigenetic classification). Microscopic features of all cases were reviewed. Each methylation analysis cost £470. Results DNA quality of 22.37% of samples was inadequate for testing due to damage or low DNA concentration. Inadequate DNA depended on diathermic artefacts or size of viable tissue. There was no concordance between pathology and arrays in 30.51% of cases; 66.67% of these had low calibration scores. High calibration scores, low calibration scores and “unidentified tumours” occurred in 47.46%, 32.20% and 20.34% of samples, respectively. Of the total cost of methylation arrays (£35,72), £9,400 was spent on non-contributory tests. Conclusions Methylation arrays are helpful but are not free from limitations. Specimen size, extent of necrosis and coagulation artefacts affect DNA concentration. Interpretation of profiles with low calibration is challenging when methylation profiles and pathology are discordant. The 20% of unidentified tumours required further analysis including whole genome sequencing and gene fusion analysis.