You are here:
Transcriptomic dose response modeling in >21 year-old archival mouse liver tissue for risk assessment
Citation:
Wehmas, L., C. Wood, AND S. Hester. Transcriptomic dose response modeling in >21 year-old archival mouse liver tissue for risk assessment. Society of Toxicology, Anaheim, CA, March 15 - 19, 2020. https://doi.org/10.23645/epacomptox.19103639
Impact/Purpose:
Poster presented to the Society of Toxicology annual meeting March 2020. Gene expression data from archival studies shows promise in measuring dose effects that inform and precede toxicity outcomes of regulatory concern. This data can be related to later adverse pathological effects without the need for new animal studies. Advances in technology can help overcome the challenges of isolating and sequencing damaged nucleic acid from old FFPE tissue samples. This study performed transcriptomic benchmark dose modeling (BMDT, BMDExpress 2.0) on whole genome TempO-sequencing in >21 yr-old paired frozen (FR) and FFPE liver tissue harvested from male B6CF7 mice exposed to 0, 1, 2, and 3.5 g/L dichloroacetic acid via drinking water for 6 d. Previous total RNA sequencing on >21 yr. old FFPE tissue samples was problematic due to a 97% loss in gene count signal; however, BMDT analysis of whole transcriptome TempO-sequencing on the same FFPE tissue only suffered a 40% loss relative to paired FR. The median BMDT value for FFPE TempO-sequencing results was 335 mg/kg-d, 38% higher than paired FR and 16% higher than the FR total RNA sequencing BMDT results. While median TempO-sequencing BMDT values were still higher than the IRIS BMD, they were within an order of magnitude and calculated from a short in vivo exposure (6 d.) unlike cancer endpoints. Our results show very old FFPE tissue samples can be used to obtain quantitative gene expression data and with additional refinement, used to enhance traditional risk assessments.
Description:
Gene expression data from short-term in vivo studies shows efficacy in quantifying dose-dependent effects that inform and precede toxicity outcomes of regulatory concern. With archival formalin-fixed paraffin-embedded (FFPE) tissue samples, this data can be obtained and related to pathological effects without the need for new animal studies. Advances in technology can help overcome the challenges of isolating and sequencing damaged nucleic acid from old FFPE tissue samples. This study investigated the efficacy of performing transcriptomic benchmark dose modeling (BMDT, BMDExpress 2.0) on whole genome TempO-sequencing in >21 yr-old paired frozen (FR) and FFPE liver tissue harvested from male B6CF7 mice exposed to 0, 1, 2, and 3.5 g/L dichloroacetic acid via drinking water for 6 d. for which total RNA-sequencing data existed. BMDTs calculated from preselected Cyp4a targets in FR total RNA sequencing data were 277-284 mg/kg-d, which was ~6.5-fold higher than the IRIS BMD for liver adenoma and carcinoma. BMDT performed poorly on FFPE total RNA sequencing data due to a 97% loss in gene count signal; however, BMDT analysis of whole transcriptome TempO-sequencing on the same FFPE tissue only suffered a 40% loss relative to paired FR. Furthermore, DESeq2 analysis of TempO-sequencing data identified 8865 and 16434 differentially expressed genes (DEGs) (FDR<0.05, ±2-fold change) across all doses for FR and FFPE datasets, respectively. Of those DEGs, 5355 were common and 1022 FR and 726 FFPE passed the Williams Trend Test (p-value <0.05) and had BMDT values ≤ to the highest dose. The median BMDT value for FFPE TempO-sequencing results was 335 mg/kg-d, 38% higher than paired FR and 16% higher than the FR total RNA sequencing BMDT results. TempO-sequencing BMDT for Cyp4a target genes in FFPE samples was quite consistent (166-325 mg/kg-d) but not as consistent as the FR (202-225 mg/kg-d). While median TempO-sequencing BMDT values were still higher than the IRIS BMD, they were within an order of magnitude and calculated from a short in vivo exposure (6 d.) unlike cancer endpoints. These BMDTs may also be improved with molecular pathway level data, saving time and resources. Our results show very old FFPE tissue samples can be used to obtain quantitative gene expression data and with additional refinement, used to enhance traditional risk assessments. This abstract does not represent EPA policy.
URLs/Downloads:
DOI: Transcriptomic dose response modeling in >21 year-old archival mouse liver tissue for risk assessment
WEHMAS_SOT2020_POSTER_V7.PDF (PDF, NA pp, 645.354 KB, about PDF)