Abstract
Background: Medulloblastoma is the most frequent malignant brain tumor in children. Currently, four distinct medulloblastoma molecular subgroups have been identified: SHH, WNT, Group 3, and Group 4. The nCounter® technology (NanoString) is a high-throughput platform, highly sensitive, robust and useful for analyses of highly degraded samples, such as formalin-fixed, paraffin-embedded (FFPE) samples. Recently, a gene panel of 22 genes employing the nCounter® technology was developed to distinguish meduloblastoma molecular subgroups.
Aim: To apply and to validate a gene panel for classification of Brazilian medulloblastoma molecular subgroups using the nCounter® technology.
Methods: We evaluated 47 tumor samples (FFPE) from patients diagnosed with medulloblastoma from Barretos Cancer Hospital, São Paulo, Brazil. Clinicopathologic features from all patients were collected and overall (OS) survival and 5-year OS were calculated. Gene expression profiling was carried out using the nCounter Elements™ custom panel (NanoString), and the custom CodeSet was composed of distinctive genes for each medulloblastoma molecular subgroup (WNT, SHH, Group 3, Group 4) plus housekeeping genes. Heatmap clustering was employed for initial classification of medulloblastoma molecular subgroups and the PAM method was applied for class prediction. hTERT hotspot mutations were screened by Sanger sequencing. Statistical analyses: Kappa index, Fisher's exact test, Kaplan-Meier method and log-rank test; level of significance: 95%.
Results: Forty-seven patients diagnosed with medulloblastoma were analyzed (average age = 14.5 years). The median OS was 66.5 months and the 5-year OS was 54%. After heatmap clustering, we distinguished 44 tumors: 21 SHH, 7 WNT, 5 Group 3, 11 Group 4, and 3 of them were not distinguished. After class prediction by PAM method, we classified 44 tumors: 22 SHH, 6 WNT, 5 Group 3, 11 Group 4, and 3 of them were not classified. Among the 3 cases which were not distinguished by heatmap clustering, 2 of them were predicted as SHH and 1 remained unclassified and 2 cases distinguished by heatmap clustering as WNT and SHH subgroups were unclassified by PAM method prediction. Classifications by heatmap clustering and PAM method were highly concordant according to kappa index. WNT group presented the lowest 5-year OS and the Group 3 presented the highest 5-year OS. hTERT hotspot mutation (C228A) was mostly found in the SHH subgroup.
Conclusion: We validated in a Brazilian cohort a modern approach for classification of medulloblastoma molecular subgroups. The nCounter® technology associated with PAM method is helpful to distinguish the medulloblastoma molecular subgroups and prospectively can be applied as a forthcoming diagnostic approach.
Citation Format: Leticia Ferro Leal, Adriane Feijo Evangelista, Flavia Escremim de Paula, Adriana Cruvinel Carloni, Gisele Caravina de Almeida, Bruna Mancano, Rui Manuel Reis. Molecular classification of Brazilian medulloblastoma by a NanoString gene panel [abstract]. In: Proceedings of the AACR International Conference held in cooperation with the Latin American Cooperative Oncology Group (LACOG) on Translational Cancer Medicine; May 4-6, 2017; São Paulo, Brazil. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(1_Suppl):Abstract nr A33.