Additional elaboration of the time course for the likely key events would be extremely helpful in clarifying the hypothesized mode of action for liver.

Studies should be conducted in the setting of conditions that predispose large populations to liver cancers, e.g., metabolic syndromes, obesity, hepatic steatosis, alcohol, and hepatitis. The spatiotemporal analysis could help reveal the regulatory networks and pathways that may bear upon intracellular events and cell interactions in tissues. The single-cell analyses could help reveal the contribution of specific cell populations in carcinogenesis.        

Improved TK-refined identification of DX saturating doses following single vs repeated oral drinking water and inhalation exposures relative to the existing downstream MoA dose-response datasets would further inform the criticality of the metabolic saturation hypothesis.  The preliminarily described mouse cyp2e1 KO data certainly suggests DX parent at supra-saturating conditions is largely responsible for DX hypothesized KEs.  The observation of a minimal contribution of a non-cyp2e1 mediated MoA cannot be ruled in the reported KO experiments given only an approximate 85% attenuation of DX effect on key MoA biomarkers and appearance of additional oxidative stress biomarkers.  However, in the absence of cyp2e1 metabolism in KO mice, the systemic blood/tissue levels of parent DX are likely to be extraordinarily high (much higher than in WT mice) due to loss of its primary metabolic detoxification pathway, and thus could initiate any number of very high dose specific but human irrelevant MoA(s). It cannot be excluded, however, that loss of cyp2e1-mediated DX metabolic clearance might be somewhat counterbalanced by increased exhalation.  Regardless, this reviewer could not find any evidence the cyp2e1 KO experiments have ever been published, so obtaining the data from the authors and/or repeating/improving KO studies to include DX comparative systemic dosimetry between KO and WT mice could be further informative relative contributions of direct high-dose specific parent DX mediated MoA compared to the primary cyp2e1 induction hypothesis.
Much discussion/debate has focused on whether transcriptomic analyses in a 28-day conducted over a range of sub-saturating, saturating and ultimately tumorigenic (or higher) doses study might further inform the plausibility of the overall metabolic saturation hypothesis.  Although intuitively attractive, this reviewer has puzzled over the MoA implications of the 7, 28 and 90 day mouse drinking water TK data in Lafranconi et al. (2021; Fig.4).  These data appear to show that 6000 ppm high-dose specific TK "saturation" response may not have fully set in at 7 days (HEAA is measurable while DX is not) while at 90 days DX is clearly saturated (even in a likely induced-state?) as evidenced by the large increase in plasma DX only at 6000 ppm.  Not sure what these data mean at present, but a further TK clarification would likely prove inform how metabolic saturation fits in with temporality and dose-response considerations of subsequent MoA events

It appears plausible that any liver injuries due to or accompanied by enhanced vulnerability to oxidative stress, increased necrosis and reparative hyperproliferation etc. may act as co-carcinogenic factors together with DX. Nevertheless, I do not expect that such interaction becomes relevant at a DX dose level which does not cause cytotoxicity/organ toxicity by itself.

Not sure about contribution of ongoing other liver injuries - I am biased toward transcriptomics as a "first cut" to get at mode of action. I would also suggest the MOA involves clonally expanded cells that "harbor" pre-existing mutation and high dose 1,4 DX  via a ROS/cell proliferation MOA "promotes" these to grow and clonally expand to form tumors. This second point - clonal expansion of pre-existing mutants can be tested in mice for 1,4-DX occurs or in rasH2 mice, and livers can be readily analyzed for the pre-existing mutated cells and their clonal expansions using error corrected DNA sequencing. A HESI group is working on both of these areas.
Spatiotemporal location effects are likely possible since Cyp2e1 in liver is localized to the pericentral region of the liver and inducers like alcohol cause strong immunostaining of the pericentral area of the liver.

Yes single cell analysis over time may provide avenues to dissect the landscape but this may take time, 4 week study, 90 days, 6 mths, of 1 yr to capture the evolution of cells as the acquire "hits" (mutation/epigenetic effects). This would also require acceptance of these type of data by regulatory agencies and an  assumption that they have folks that can interpret and understand the data. I am not sure if this would move the regulatory needle beyond whare it would be using current data sets combine with transcriptomics.

I would stick with the ROS and proliferation threshold based mode of action to calculate a BMD/POD for risk assessment.
Since EPA is "all in" on transcriptomic based MOA and risk assessments, if I were to suggest new studies they would follow the EPA ETAP process and use transcriptomic based RfD.