Canonical amplifications and CDKN2A/B loss refine IDH1/2-mutant astrocytoma prognosis

Abstract

Background: Molecular features have been incorporated alongside histologic criteria to improve glioma diagnostics and prognostication. CDKN2A/B homozygous-loss associates with worse survival in IDH1/2-mutant astrocytomas (IDHmut-astrocytomas), the presence of which denotes a grade 4 tumor independent of histologic features. However, no molecular features distinguish survival amongst histologically defined grade 2 and 3 IDHmut-astrocytomas.

Methods: We assembled a cohort of patients ≥19 years old diagnosed with an IDHmut-astrocytoma between 1989 and 2020 from public datasets and several academic medical centers. Multivariate modeling and unbiased clustering were used to stratify risk.

Results: We identified 998 IDHmut-astrocytoma patients (41.5% female; 85.6% white). Tumor grade, CDKN2A/B loss, and/or ≥1 focal amplification were associated with reduced survival. Grade 2/3 patients with intact CDKN2A/B and no focal amplifications survived the longest (OS 205.7 months). Survival for grade 2/3 cases with either CDKN2A/B hemizygous-loss or focal amplifications (80.4, 88.7 months respectively) did not differ significantly from grade 4 cases with intact CDKN2A/B and no amplifications (91.5 months, P = .93). Grade 4 patients with either hemizygous or homozygous loss of CDKN2A/B had the shortest survival (OS 31.9, 32.5 months respectively), followed by grade 4 cases with intact CDKN2A/B and focal gene amplifications (OS 55.9 months). Integrating CDKN2A/B status and amplifications alongside histopathologic grade refined overall survival prediction. Unbiased clustering revealed 9 distinct molecular profiles, with differential survival. IDHmut-astrocytomas with any CDKN2A/B loss clustered together, regardless of grade, and exhibited the poorest outcomes.

Conclusions: Combining CDKN2A/B hemizygous-loss and focal gene amplifications reveals a group of IDHmut-astrocytoma patients with an intermediate prognosis, refining IDHmut-astrocytoma classification.

Keywords: CDKN2A/B loss; IDH1/2-mutant astrocytoma; focal amplification; glioma; molecular risk stratification.

Figure 1.

(A) Genomic signatures in IDH1/2-mutant astrocytomas, stratified by tumor grade. (B) Multivariate adjusted prognostic features in IDH1/2-mutant astrocytomas of all grades both negatively and positively influencing overall survival. (C) Overall survival of IDH1/2-mutant astrocytomas stratified by grade and CDKN2A/B status. Kaplan–Meier survival curves demonstrate that CDKN2A/B hemizygous loss is associated with worse survival in patients with IDH1/2-mutant astrocytomas of any grade.

Figure 2.

Kaplan–Meier survival curves of grade 2/3 (A) and grade 4 (B) IDH1/2-mutant astrocytomas demonstrate that canonical focal amplifications are associated with worse survival, independent of CDKN2A/B status. (C) Median overall survival of grade 2/3 and grade 4 IDH1/2-mutant astrocytomas, stratified by presence of CDKN2A/B loss or focal amplifications; P-values are of pairwise comparisons. (D) Kaplan–Meier survival curves of patients stratified across a molecular classification scheme incorporating amplifications, CDKN2A/B loss, and histopathologic grade, which partition patients into long-term, intermediate, and short-term survivors.

Figure 3.

(A) Unsupervised clustering using UMAP and DBSCAN of IDH1/2-mutant astrocytomas revealed that samples partition primarily based on CDKN2A/B-hemizygous or homozygous loss. (B) Distinct genomic signatures of IDH1/2-mutant astrocytoma clusters. (C) Kaplan–Meier curves demonstrate that loss of CDKN2A/B is associated with the lowest median overall survival compared to other molecular subgroups of IDH1/2-mutant astrocytomas (P < .001). Kaplan–Meier curves could not be generated for clusters 8–9 due to limited sample size.