Meningioma is the most common primary brain tumor and carries a substantial risk of local recurrence. While WHO grade correlates with recurrence to a degree, there is substantial within-grade variation of recurrence risk and current risk stratification does not accurately predict which patients are likely to benefit from adjuvant radiation therapy. We hypothesized that recurrent tumors have unique gene expression profiles (GEP) that could be used to better stratify patients for radiation therapy. We optimized a recurrence predictor using a training/validation approach and a support vector machine classification method with radial-basis smoothing kernel. Three publicly available Affymetrix gene expression datasets ({type:entrez-geo,attrs:{text:GSE9438,term_id:9438}}GSE9438, {type:entrez-geo,attrs:{text:GSE16581,term_id:16581}}GSE16581, {type:entrez-geo,attrs:{text:GSE43290,term_id:43290}}GSE43290) combining 127 newly diagnosed meningioma samples served as the training set. Unsupervised variable selection was used to identify an 18-gene GEP model (18-GEP) that separated recurrences with a negligible root mean square error of 0.17. The characteristics of the training dataset were as follows: WHO grade [I-92(73%), II-32(25%), III-2(2%)]; median follow-up = 5.53 years (range:0.05-25.42); recurrences = 18. This model was tested on 62 cases from our institution [validation dataset (VD)] with similar demographics, but enriched for cases with either long clinical follow-up or known recurrence. When applied to the VD, the 18-GEP separated recurrences with a misclassification error rate of 0.25 (log-rank p = 0.0003). 18-GEP was significantly predictive of tumor recurrence, independently [p = 0.0007, HR = 7.91, 95%CI = 2.35-35.78)] and was predictive after adjustment for WHO grade, mitotic index, and Simpson grade [p = 0.047, HR = 4.73, 95%CI = (1.02-26.55)]. The expression signature included genes encoding proteins involved in normal embryonic development, cell proliferation, tumor growth and invasion (FGF9, SEMA3C, EDNRA), angiogenesis (angiopoietin-2), cell cycle regulation (CDKN1A), membrane signaling (tetraspanin-7, caveolin-2), WNT-pathway inhibitors (DKK3), complement system (C1QA) and neurotransmitter regulation (SLC1A3, Secretogranin-II). In conclusion, our gene expression classifier accurately stratifies patients with meningioma by recurrence risk and has the potential to guide therapy.
Read full abstract