Gene expression based profiling of pleomorphic xanthoastrocytoma highlights two prognostic subgroups

Am J Transl Res. 2022 Feb 15;14(2):1010-1023. eCollection 2022.

Abstract

Background: Pleomorphic xanthoastrocytomas (PXAs) are rare, accounting for less than 1% of astrocytomas, and commonly occur in young patients. The majority are WHO grade II. A fraction of tumors that present or recur with malignant change are classified as anaplastic (APXA, grade III). Limited data are available on their molecular characteristics.

Methodology: Genome-wide expression profiling of 14 PXA and 6 APXAs was performed by microarray. Among differentially expressed genes (DEGs), Cyclin-Dependent Kinase 14 (CDK14) and Mitochondrial Fission Process 1 (MTFP1) were validated by qRT PCR.

Result: Unsupervised hierarchical clustering revealed two distinct molecular clusters (Cluster 1: 10 PXA, 3 APXA and Cluster 2: 4 PXA, 3 APXA) with grade II and III tumors distributed in both highlighting molecular heterogeneity within the same grade. There was an insignificant difference in age, sex, immunohistochemical profile, frequency of BRAF mutation, or CDKN2A deletion among the two clusters. Significantly, worse progression-free survival was observed in cluster 2 (P=0.003). mRNA profiling-based prediction of recurrence was superior to and independent of histological grade, BRAF mutation, or CDKN2A deletion status. A total of 10 upregulated and 418 downregulated genes were identified between the PXA clusters. qRT-PCR validation of CDK14 (upregulated in cluster 2) and MTFP1 (upregulated in cluster 1) showed strong concordance with expression array data.

Conclusion: This is the first comprehensive study highlighting distinct molecular subgroups of PXA. The differentially expressed genes between two clusters may potentially be used for developing histology independent classification schemes, prognostication and may serve as prospective therapeutic targets for PXA patients.

Keywords: BRAF; CDK14; CDKN2A; Expression profiling; MTFP1; PXA/APXA; molecular clusters; unsupervised hierarchical clustering.