NIMG-09. NONINVASIVE PERFUSION IMAGING BIOMARKER OF MALIGNANT GENOTYPE IN ISOCITRATE DEHYDROGENASE MUTANT GLIOMAS Free

A small subset of IDHmut astrocytomas behave aggressively, similarly to IDHwt glioblastoma. Genomic correlates of poor prognosis IDHmut astrocytomas include global relative DNA hypomethylation, MYCN amplification copy number variation (CNV) abundance, and CDKN2A/B homozygous deletion. We sought to identify a non-invasive imaging correlate of poor prognosis IDHmut astrocytomas.

30 IDHmut astrocytomas (NYU=18, TCGA=12) were included. Relative cerebral blood volume was obtained from 4 regions of interest within the highest perfusion areas. Genomic CNV analysis and CDKN2A/B copy number status was obtained using Illumina Infinium 450k or Illumina EPIC methylation array data. Associations between rCBV, survival, and genomic alterations were evaluated. An additional 22 IDHmut astrocytomas were obtained from another institution for validation.

CNV stable (CNV-S, n=14) astrocytomas demonstrated significantly lower mean rCBV than and CNV unstable (CNV-U, n=16) (P=0.0.0155) ones. IDHmut astrocytomas with CDKN2A/B homozygous deletion (n=7) had higher rCBV (3.9 ± 2.3, mean ± SD) compared to those without this alteration (n=22; 1.8 ± 1.9, mean ± SD; P=0.0489). In the validation set, there was no significant evidence in rCBV between CNV-S and CNV-U tumors (P=0.3976). These results were sensitive to a single outlier in the CNV-S group. Excluding this case with rCBV=7.15, the CNV-S tumors demonstrated significantly lower mean rCBV (P=0.0279). IDHmut astrocytomas with CDKN2A/B homozygous deletion (n=9) tended to have higher rCBV (4.2 ± 1.8, mean ± SD) compared to those without this alteration (n=16; 2.5 ± 1.5, mean ± SD; P=0.0757, t-test). This led to significant discrimination (P=0.0328, Wilcoxon).

Our study is the first to identify a non-invasive imaging biomarker for prognostic genomic alterations in IDHmut astrocytomas. Greater CNV abundance and/or CDKN2A/B homozygous deletion have higher rCBV and worse prognosis. Our methods can be applied to standard clinical practice and may enhance informed treatment decisions in the preoperative and immediate postoperative settings.