Progressing towards a modern animal-free and human relevant genotoxicity assessment: The role of GENOMARK: a transcriptomic biomarker in human HepaRG cells

The evaluation of a chemical’s (pharmaceutical, cosmetic ingredient, plant protection product, food additive,…) potential to induce damage to genetic material is a key element when assessing its safety for human health and the environment. A step-wise standardized approach is traditionally applied, starting with a battery of in vitro tests that usually consist of a bacterial reverse gene mutation test (also referred to as ‘Ames assay’), and one or two in vitro mammalian cell tests to detect gene mutations and chromosome damage (aneugenicity and clastogenicity). A positive result in any of these in vitro tests is followed up by an in vivo test for the same genotoxic endpoint. However, this genotoxicity testing strategy faces several limitations, including (1) low specificity, (2) limited or no information on the underlying mechanism, (3) relatively low throughput, and (4) limited quantitative assessment of the collected data. To modernize genotoxicity assessment and reduce reliance on experimental animals, high-throughput new approach methodologies (NAMs) to generate concentration-response data in human-relevant test systems are needed. An interesting group of NAMs are transcriptomic biomarkers, consisting of subsets of genes that robustly and consistently respond to chemicals from specific mechanistic classes. Previously, GENOMARK, a transcriptomic biomarker for genotoxicity was developed using microarray data collected in human metabolically competent HepaRG cells with a 72h exposure period. The goal of this doctoral thesis was to optimize the existing GENOMARK biomarker and explore its possible regulatory applications.
The first part ‘introduction and theoretical background’ outlines the context of the current work. More specifically, the current regulatory genotoxicity test batteries and their limitations are explained and the concept of NAMs with a focus on transcriptomics and strategies for their implementation are introduced. The second part summarizes the experimental work. Chapter 4 describes the development of two new improved prediction models for GENOMARK, which both demonstrated a predictive accuracy of 100% to de-risk misleading positives. In Chapter 5, a pilot study is presented which aimed to explore the quantitative assessment of the gene expression data, generated with qPCR, for two well-known genotoxicants using the benchmark dose (BMD) approach. The results showed that the gene expression data could be used for potency ranking of chemicals but that high-throughput approaches are needed to generate sufficient concentration-response data. Increasing the throughput of GENOMARK was therefore investigated in Chapter 6 by applying two high-throughput technologies including (1) RNA-Sequencing and (2) TempO-Seq®. GENOMARK showed a 90% predictive accuracy using a publicly available dataset of 10 chemicals generated with RNA-Sequencing in HepaRG cells, illustrating its applicability to this technology. Furthermore, new transcriptomic datasets were generated with TempO-Seq®. Analysis of this data revealed that GENOMARK could be used qualitatively (for hazard identification) and quantitatively (for potency ranking) on TempO-Seq® data. Moreover, genotoxic-specific transcriptomic point of departure (tPoDs) could be derived. In addition, in Chapter 6, hazard calls and potency rankings generated with GENOMARK were compared to those based on another biomarker for genotoxicity, TGx-DDI. Interestingly, although both biomarkers consisted of different genes, the hazard calls and potency rankings were highly similar for the 10 tested chemicals. Lastly, in Chapter 7, the correlation between general toxicological tPoDs based on a large set of genes (i.e. the customized human TempO-Seq® S1500+ panel) and biomarker genotoxic-specific tPoDs was investigated for 8 genotoxicants. This chapter also highlighted the use of high-throughput kinetic modeling to translate the in vitro tPoDs to human-relevant administered equivalent doses (AEDs). For the majority of the chemicals (6/8 in case of the large gene set, 5/7 in case of the biomarkers), the transcriptomics-based AEDs were more conservative compared to the in vivo PoDs. Finally, Chapter 8 summarizes and further discusses the regulatory application and implementation strategies of GENOMARK and other transcriptomic tools in the regulatory frameworks.
This chapter also highlights particular considerations and future challenges in the context of a modern genotoxicity risk assessment.