Abstract 2081: Empowering rare disease cohort biomarker discovery via comparative assessments of gene expression analysis platforms for FFPE pediatric brain tumor specimens

The gene expression landscape for biomarker discovery is still limited for many pediatric brain tumors due to insufficient numbers of biorepository collected fresh frozen specimens. The majority of available tissue are formalin-fixed paraffin embedded (FFPE) pathology diagnostic specimens. These specimens, are often of limited quantity and contain compromised RNA material. A number of emerging commercial platforms are described as supporting quantitative expression analysis for low quantity and poor quality materials. Utilizing available platforms, we designed a study to evaluate RNA and miRNA levels in pediatric brain tumor FFPE specimens with limited/compromised access.Experiments were performed with specimens and/or data obtained from Brain Tumor Tissue Consortium (CBTTC) at Children's Hospital of Philadelphia (CHOP). The mRNA gene expression and miRNA target analysis were performed with FFPE material utilizing two commercial platforms: HTG EdgeSeq and Nanostring. The analysis included 5 specimens of low grade glioma or primitive neuroectodermal tumors for mRNA gene expression and 4 specimens of medulloblastoma for miRNA target analysis. FFPE specimens were processed according to manufacturer's protocols. The results were compared with the RNAseq or miRNA sequencing data derived from corresponding tumors' flash frozen specimens' RNA material.

We evaluated requirements of each platform for data generation and established between platform analysis correlations. In performed tests, the HTG platform required lowest amounts of specimen's material. For the majority of analyzed genes (>700 genes), the gene expression profile was relatively similar between all three approaches, however each of the platform presented distinctive distribution profile for normalized data. The RNAseq mean read counts values were correlated highly with NanoString (0.81) and of lesser value with HTG (0.7) platform. The RNAseq presented significantly higher variance distribution than other platforms. The miRNA target analysis (>600 genes) distribution of normalized data revealed significantly lower dynamics for NanoString when compared with miRNAseq or HTG panel data. The miRNA mean read counts were highly correlated with HTG (0.77) and little with NanoString (0.22), while miRNASeq presented the highest variance distribution. In summary, we found a significant level of agreement between all three platforms tested for gene expression data generation. As for miRNA target analysis, the HTG platform presented significantly higher agreement with miRNAseq data. We conclude that tested technologies can support data generation from archived FFPE specimens, however, the platform selection process should involve pre-selection data quality analysis, as well as sample size requirements, gene cohort selection, and pricing evaluations.

Citation Format: Mateusz P. Koptyra, Namrata Choudhari, Zhang Zhe, Mariarita Santi, Angela Waanders, Adam Resnick. Empowering rare disease cohort biomarker discovery via comparative assessments of gene expression analysis platforms for FFPE pediatric brain tumor specimens [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2081.