Abstract 3201: Molecular subtype specific biomarkers improve prediction of temozolomide therapy in low grade glioma

Abstract

Abstract Introduction: Based on IDH mutation and 1p/19q codeletion, low grade glioma (LGG) could be classified into three groups with different prognosis : IDHmut-codel, IDHmut-non-codel and IDHwt,. Temozolomide (TMZ) is commonly used to reduce LGG recurrence and mortality. However, increasing studies have demonstrated MGMT promoter methylation status could not predict the therapeutic effect of TMZ in the different molecular subtypes of LGG. Therefore, promising molecular markers of TMZ chemotherapy in each subgroup is in urgent need. Methods: LGG data sets containing information of mutations and copy number aberrations (CNVs), as well as clinical records and treatment history were retrieved form The Cancer Genome Atlas. TMZ treated patients with intact PFS/OS information were selected(n=256). Kaplan-Meier and Cox's proportional hazard model were used to estimating survival and selecting predictive molecular makersin these TMZ-treated LGG.The Chinese Glioma Genome Atlas (CGGA) was used as independent validation data. Results: These patients were classified into IDHmut-codel (n=69), IDHmut-non-codel (n=127) and IDHwt (n=60) subgroups. MGMT promoter is methylated (MGMTpo-m) in 99% patients with IDHmut-codel .And the status of MGMT promoter methylation was not correlated with either PFS or OS in both IDHmut-non-codel and IDHwt cohorts. Based on the cox regression analysis, PIK3CA, mutated in 20% IDHmut-codel patients, significantly correlated with PFS (p=0.03) and OS (p=0.02) in this cohort. TTN obviously correlated with PFS (p=0.01) and OS (p < 0.01) in IDHmut-non-codel cohort. C9orf53, homozygous deletion in 40% IDHwt patients, significantly correlated with PFS (p=0.03) and OS (p < 0.01) in IDHwt cohort. Multivariate cox regression confirmed these three genes were independent predictors in each cohort. Notably, the combination of PIK3CA with TTN mutations (marked as PIK3CA/TTN) exhibited striking correlation with PFS (p < 0.01) and OS (p < 0.01) in the IDHmut cohort (IDHmut-codel and IDHmut-non-codel cohorts), and multivariate cox regression confirmed it was also an independent predictor of PFS (p < 0.01) and OS (p < 0.01). The predictive effect of PIK3CA/TTN was validated in an independent dataset CGGA. OS was significantly higher in PIK3CA/TTN patients (p = 0.04) and multivariate cox regression confirmed its independent predictive effect (p=0.04). Conclusion: Although MGMT methylation status is considered as the predictive markers for TMZ chemotherapy in LGG. In this work, we proved it could not predict TMZ efficacy in each molecular subtypes of LGG . However, PIK3CA, TTN, PIK3CA/TTN and C9orf53 were identified as promising molecular markers of TMZ in different subtypes of LGG: PIK3CA for IDHmut-codel, TTN for IDHmut-noncodel, C9orf53 for IDHwt subtype, and PIK3CA/TTN mutation for all IDH mutant LGG. Citation Format: Feng Jia, Feilong Zhao, Xiaoyan Zhang, Jiao Feng, Tonghui Ma, Tao Lv, Xiaohua Zhang. Molecular subtype specific biomarkers improve prediction of temozolomide therapy in low grade glioma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3201.