Multicenter Prospective Validation of a DNA Methylation-Based Predictor of Meningioma Recurrence Risk and Molecular Classification

Background: Meningiomas have been demonstrated to have significant heterogeneity between patients and even within each WHO grade, making prognostication challenging with current standard of care classifications. We previously developed a DNA methylation-based predictor (PMID 31158293) of meningioma recurrence risk following surgery and validated this in retrospective cohorts. For this study we utilize prospectively collected samples from multiple institutions enriched for biologically aggressive meningiomas to confirm the utility of our predictor in prognosticating meningioma patients and informing selection for adjuvant radiotherapy (RT).

Methods: DNA was extracted from meningioma tumor tissue and bisulfite converted. Genome-wide DNA methylation profiles were generated with the Illumina EPIC Beadchip Array and imported, processed, and normalized as previously described. Clinical data elements including progression-free survival (PFS) were extracted from the medical records and corroborated with radiographic neuroimaging. The performance of our previous methylation-based predictor was compared with a WHO grade predictor using generalized boosted regression modeling using previously published prognostic methylation probes compared to WHO grade as the sole feature by generating time-dependent receiver operating characteristic (ROC) curves and computing area under the ROC curves (AUCs) along with their 95% confidence interval using bootstrap resampling with 10,000 resamples.

Results: 173 meningiomas treated from 2018 to 2021 were included in preliminary analysis. Most patients were female (115/173, 66%) and median age was 59 (interquartile range 49-71). The cohort was enriched for WHO grade 2 (59/173, 34%) and 3 (26/173, 14%) meningiomas. Most cases received a gross total resection (GTR; 97/173, 56%). DKFZ classification confirmed the histologic diagnosis of meningioma in all instances ([Fig. 1A]). Meningiomas were also classified into their molecular groups as previously published by our group (PMID 34433969), UCSF (PMID 35534562), and Baylor (PMID 35108039) ([Fig. 1B]) based on methodologies of the original publications. Meningiomas in could be largely dichotomized into either low-risk (MG1 or MG2, Immune-enriched or Merlin-intact, MenG A or MenG B) or high-risk molecular groups (MG3 or MG4, Hypermitotic, MenG C). When DNA methylation risk scores were calculated for each tumor, there was a significant increase in risk scores with increasing WHO grade and MG ([Fig. 1C, D]) with PFS reflecting these differences ([Fig. 1E, F]). WHO grade 2 meningiomas had a bimodal distribution of risk scores, reflecting their heterogeneity. Using Cox proportional hazards modeling, increasing methylome recurrence risk was associated with a significantly increased hazard of tumor recurrence (HR 4.16, 95% CI 2.14–8.09, p < 0.001) ([Fig. 2A]). When cases were dichotomized into a low- and high-risk methylation group, the high-risk methylation cases had significantly worse PFS compared to the low-risk cases ([Fig. 2B]). The methylome-based predictor had substantially improved performance in predicting 5-year PFS compared to WHO grade alone (ΔAUC = 0.10, 95% CI: 0.09–0.11, [Fig. 2C]). Following results of methylation modeling, 59 cases (34%) were prescribed adjuvant RT following surgery.

Conclusions: DNA methylation modeling outperforms conventional WHO classification in outcome prediction in a novel, independent, prospective cohort of meningiomas enriched for clinically aggressive cases and may be effectively used for real time prognostication, patient counseling, and referral for adjuvant RT.

Fig. 1

Fig. 2

Publikationsverlauf

Artikel online veröffentlicht:
05. Februar 2024