Impact of Chromosome Copy Number Thresholds on Meningioma Molecular Classification

Introduction: Molecular classification systems for meningioma have been shown to better predict tumor behavior and patient prognosis compared to traditional World Health Organization (WHO) histopathologic designation. Chromosome arm copy number variants (CNV) contribute to several molecular classification schemes for meningioma. However, the optimal threshold of chromosome arm gain/loss which results in a CNV call in meningioma remains unknown, with significant heterogeneity in thresholds used across institutions. We sought to characterize the landscape of chromosomal CNVs in meningioma, assessing typical CNV sizes, chromosome arm involvement, CNV hotspots, and their relationship to meningioma molecular classification performance.

Methods: We identified 1,042 patients who received surgery for meningioma with whole-genome microarray sequencing of their tumor. Demographic/clinical information and tumor histopathologic grade were extracted from the electronic medical record and pathology reports respectively. Probe locations were used to determine the start and end chromosome arm coordinates for CNVs. CNV thresholds were iteratively altered between 0 and 100% to determine chromosome arm call changes. Chromosome arm calls, CDKN2A/B loss, and mitotic count were used to calculate a molecularly Integrated Grade (IG) for meningioma across CNV thresholds. Cox regression models and Kaplan–Meier curves were used to assess the impact of differential CNV thresholds on predicting meningioma recurrence in a subset of patients with primary, nonradiated meningiomas with gross-total resection.

Results: We observed significant variability in the fraction of a chromosome lost or gained in meningiomas, with median losses on chromosome 1, 18, 22 and gains on chromosome 13, 18, and 20 exceeding 25% of the total chromosome. While CNVs often simultaneously involved p and q arms, losses were more frequently restricted to either p or q arms compared to gains. Most CNVs occurred as a single fragment, but 35.1% of 8q loss, 44.0% of 12q loss, 54.4% of 3q gain, and 74.1% of 5q gain occurred as two or more fragments. Across chromosome arms associated with recurrence risk, the frequency of CNV arm loss calls varied the greatest when CNV loss thresholds were set between 0–5% and 95–100%. As the CNV loss threshold was increased, 1p, 6q, 19p, and 19q showed steady declines in arm loss calls. Varying call thresholds shifted meningioma across IG grades, with 12.6% of IG-2 and 8.6% of IG-3 tumors shifting to lower grades when the CNV loss threshold was increased from 25% to 50%. Changing CNV thresholds impacted the power of certain chromosome arms to predict meningioma recurrence, with arm-dependent optimal CNV thresholds ranging from 12% to 86%. While CNV thresholds altered tumor distribution across IG grade, Kaplan–Meier curves showed no significant difference in predicting meningioma recurrence between IG grades calculated at a 25%, 50%, or arm-dependent optimized CNV loss threshold.

Conclusion: CNV gains and losses vary in size, distribution, and composition across chromosome arms. Altering CNV thresholds shifts the prevalence of chromosome arm gain and loss calls, impacting molecular classification of meningiomas. Characterizing the influence of optimal CNV call thresholds may therefore be an important step to ensure molecular classification paradigms are robust and reproducible.

Fig. 1

Fig. 2

Fig. 3

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
05 February 2024