Managing Liver Injury (DILI) Risk

Having a clinical drug-induced liver injury (DILI) signal is a surefire way to kill a drug candidate in non-serious indications like pain or chronic indications like diabetes, and will even give oncologists second thoughts as cancer patients are living longer. What can we do to prevent it?

Plan A: Small Pill and No Greasebombs

A widely accepted strategy for minimizing DILI risk is “the rule of two” – avoid lipophilic drugs (LogP > 3) which require high doses (> 100 mg) (Figure 1). (Chen, M. et al. “High lipophilicity and high daily dose of oral medications are associated with significant risk for drug-induced liver injury.” Hepatology, 2013, 58, 388-396. doi: 10.1002/hep.26208)

This makes sense as lipophilic compounds tend to be more promiscuous (For a great recent article worthy of a separate article, see: Brown, D. G. et al, “Promiscuity of in Vitro Secondary Pharmacology Assays and Implications for Lead Optimization Strategies.” J. Med. Chem. 2019, doi: 10.1021/acs.jmedchem.9b01625).

They’re also more susceptible to bioactivating P450 oxidations, and lower doses reduce the risk that the liver’s detoxifying pools of glutathione (5-10 mM) can be depleted. Compounds that can be given at a lower dose are also likely to be much higher quality as high lipophilic efficiencies are often necessary for a drug to provide full target engagement in whole blood with good PK, but this is a lengthly discussion I’ll save for another post!)

"Rule of Two" for drug design is to avoid high daily dose (>100 mg) drugs that are lipophilic (LogP >3)
Figure 1. Several retrospective analyses have highlighted the association between higher drug doses and DILI risk. (a) Oral medications associated with DILI (red circles) are enriched in the high dose, high LogP quadrant compared to drugs not associated with DILI (green triangles). The “rule of two” is to avoid the top right quadrant (black box). (Note to self: find out who put LogP 8 candidate forward and why) (b) A similar analysis shows that drugs with a Cmax >1 uM are more likely to be associated with DILI. (Reproduced from Ref. 1.)

The “one rule to rule them all” in minimizing drug attrition due to off-target toxicity might be: target efficacious doses as low as possible (<10 mg ideally, Cmax <1 uM). A well-known analysis by Antonia Stepan and colleagues showed that low dose drugs (<100 mg) were less likely to have idiosyncratic adverse drug reactions with any target organ (Figure 2). (Stepan, A. F. et al. “Structural Alert/Reactive Metabolite Concept as Applied in Medicinal Chemistry to Mitigate the Risk of Idiosyncratic Drug Toxicity: A Perspective Based on the Critical Examination of Trends in the Top 200 Drugs Marketed in the United States.” Chem. Res. Toxicol. 201124, 1345-1410. doi: 10.1021/tx200168d)

Figure 2. (a) A 2009 retrospective analysis of drugs associated with idiosyncratic adverse drug reactions found that drugs with lower clinical doses were less likely to be withdrawn from the market due to IADRs and (b) less likely to have boxed warnings for idiosyncratic adverse drug reactions. (Idiosyncratic liver injury is the biggest contributor to BBW’s for IADRs: Uetrecht, J. and Naisbitt, D. J. “Idiosyncratic Adverse Drug Reactions: Current Concepts.” Pharmacol. Rev. 2013, 65, 779-808. doi: 10.1124/pr.113.007450)

Unfortunately, binding pockets, biology, budgets, and timelines rarely make a 5 mg daily dose possible or practical. (With the multi-million-dollar-a-month $$ burn rate most project teams and startups deal with, perfect really is the enemy of good.) Plus, plenty of perfectly reasonable-looking, low-dose compounds have been withdrawn or given black box warnings due to observed hepatotoxicity (some examples in Figure 3, below). So what else can we do?

Figure 3. Examples of drugs withdrawn due to hepatotoxicity or with hepatotoxicity black box warnings despite low doses. All look like perfectly reasonable compounds! (Update: I recently learned from one of the authors, Paul Morgan, that ambrisentan’s liver black box warning was removed in 2017 – likely at the time of first submission this was wrongly caught up in ‘class effect’ concerns with other endothelin antagonists such as bosentan and sitaxentan. Alpidem was never approved in the US due to concerns about lack of efficacy, and leflunomide requires a loading dose of 100 mg/day for 3 days for RA before decreasing to 30 mg. So these are not all the best examples but the point is that low dose is no guarantee of avoiding hepatotoxicity. Hope to update this figure soon w/ better examples when I have time!)
Plan B: De-risk Mechanistically

The mechanisms of DILI and some additional strategies scientists can use to minimize DILI risk in drug candidates were recently excellently summarized by an impressive consortium of scientists including from Servier, AbbVie, Merck, Lundbeck, GlaxoSmithKline, AstraZeneca, Janssen, Orion, and others including B. Kevin Park from the University of Liverpool (briefly summarized in Figure 4). (Weaver, R. J. et al. “Managing the challenge of drug-induced liver injury: a roadmap for the development and deployment of preclinical predictive models.” Nat. Rev. Drug Discov. 2019, https://www.nature.com/articles/s41573-019-0048-x) In this review they summarize the six main mechanisms by which drug-induced liver injury is thought to occur (Figure 4), along with a detailed description of assays and models available to assess whether any of these mechanisms are relevant to a molecule of interest.

Mechanisms of drug-induced liver injury (DILI)
Figure 4. A quick summary of the mechanisms of drug-induced liver injury and corresponding tools to assess these risks. Unfortunately, the impact of compounds on the immune compartment with respect to DILI is still poorly modeled, and assay development in this area is very much a work in progress.

The best human evidence for these mechanisms’ contribution to DILI is summarized in Table 1, below, along with a greatest-bang-for-your-buck assay for initially assessing whether any of these mechanisms are relevant to your molecule of interest. For example, polarized, functional hepatocyte assays where bile-canaliculi are recommended by the authors for assessing bile flow perturbation over assays measuring single transporters like BSEP. This is because not all drugs which cause cholestasis inhibit BSEP, and it was recently concluded that the inverted vesicle assay for BSEP inhibition were not predictive of the toxic potential of drugs. This is a rough guide for where to start an assessment, but more detailed risk assessment requires input and experimentation from a toxicologist. Getting a clean read in variants of these assays can definitely give a team peace of mind, but a signal in any of these assays is still just a starting point for a full risk evaluation discussion with the project team. Most of these key assays or variants of these are already incorporated in most companies’ late-stage compound evaluation cascades, but it’s worth checking some of these earlier if you have reason to suspect your lead series might be especially problematic (e.g. lipophilic amines w/ LogD ~ 4+).

Mitochondrial Perturbation
Bile Flow Perturbation
Reactive Metabolites
Lysosomal Perturbation
ER Stress
Immune Reaction
Table 1. Summary of the six major mechanisms of DILI injury, the best human evidence for their involvement in DILI, and key assays to initially assess whether each is a risk for compound development. (Summarized from Weaver, R. J. et al.)

For those newer to drug discovery, I hope this served as a helpful starting point for how to minimize DILI risk. For the more experienced, I hope this serves as a useful reminder for where to start your next conversation with your toxicologist.

Happy hunting! Explore dhdevsite0.wpengine.com for more.

RECENT ARTICLES

DRUG DISCOVERY RESOURCES

JOIN SUBSCRIBERS from

  • Johnson & Johnson
  • Roche
  • Pfizer
  • Novartis
  • Bayer
  • Merck
  • Sanofi
  • GlaxoSmithKline
  • AbbVie
  • Eli Lilly
  • Boehringer Ingelheim
  • Amgen
  • Bristol-Myers Squibb
  • Gilead
  • AstraZeneca
  • Teva
  • Novo Nordisk
  • Merck KGaA
  • Takeda
  • Celgene (BMS)
  • Biogen
  • Astellas
  • Daiichi Sankyo
  • Regeneron
  • UCB
  • Eisai
  • Servier
  • Kyowa Kirin
  • Shionogi
  • Vertex
  • Lundbeck
  • Taisho
  • Ono
  • Seagen
  • Incyte
  • LEO Pharma
  • Janssen
  • Galapagos
  • Idorsia
  • Astex
  • Arvinas
  • Kymera
  • Relay
  • Schrodinger
  • Mirati
  • Revolution
  • Nimbus
  • Plexxikon
  • Forma
  • Arcus
  • RA Capital
  • ARCH Venture Partners
  • MPM Capital
  • Orbimed
  • The Column Group
  • 5AM Ventures
  • Versant Ventures
  • Deerfield
  • Goldman Sachs
  • Citi
  • Piper Sandler
  • Softbank
  • …and hundreds of others