When to Conduct Hepatic Impairment Studies and When Modeling May be Enough

For drugs that undergo hepatic metabolism or biliary excretion, hepatic impairment (HI) studies are a critical component of the regulatory pathway. Yet with guidances that are now over twenty years old, a rapidly evolving modeling toolkit, and complex decisions around study design and timing, navigating HI requirements remains one of the more nuanced challenges in clinical pharmacology. Here we’ll discuss a practical framework for minimizing risk at every stage: from deciding whether a study is needed, to leveraging physiologically based pharmacokinetic (PBPK) modeling as a regulatory-accepted alternative. 

Why hepatic impairment studies matter

The core objective of any hepatic impairment evaluation is to inform prescribing information. Specifically, it helps physicians select an appropriate dose for patients with mild, moderate, or severe hepatic dysfunction. The consequences of getting this wrong are significant: an inadequate label may restrict patient access, trigger post-marketing commitments, or lead to unsafe dosing in a vulnerable population. 

Both the FDA (2003) and EMA (2005) have issued guidance on when HI studies are required, and while the two share common ground, there are meaningful differences that teams must account for in their strategy. 

When is a hepatic impairment study required?

The FDA guidance sets a specific threshold: if hepatic metabolism or biliary excretion accounts for more than 20% of the absorbed dose (parent drug or metabolite), a hepatic impairment study is generally expected. The EMA guidance takes a similar approach in principle but does not specify a numeric threshold, instead providing recommendations based on likely use in hepatically impaired patients and the probable impact on exposure. 

A study is typically not required when: 

• The drug is gaseous or volatile (eliminated via the lungs) 

• Less than 20% of the drug is hepatically metabolized 

• The drug has a wide therapeutic index and no significant hepatic involvement 

A practical decision tree for teams to work through, guiding sponsors whether a dedicated study in patients with hepatic impairment is required, is presented in Figure 1. Alternative strategies, including the use of modeling or a combination approach is discussed further below. 

Study design options: full vs. reduced

When a dedicated clinical HI study is required, two main designs are recommended based on current guidance: 

Full design study

Includes patients with mild, moderate, and severe hepatic impairment, along with age- and gender-matched healthy controls. This design is considered appropriate when specific dosing recommendations are needed across the full spectrum of hepatic dysfunction. Enrollment is typically guided by the Child-Pugh classification, with approximately six subjects per group. 

Reduced design study

Focuses on moderate hepatic impairment only, with matched controls. The FDA requires at least eight subjects per group for this design and specifies that any meaningful PK change seen in moderate impairment should also be reflected in labeling for mild patients. The EMA adopts a slightly different approach, recommending that sponsors may need to extend the study to severe or mild groups depending on the results in moderate hepatic impairment. 

Single-dose studies are typically sufficient when PK is dose-proportional and there is no time-dependent behavior. Multiple-dose designs should be considered when induction, inhibition, or time-dependent PK changes are present. 

Alternative strategies: when a dedicated study is not the only answer

A dedicated study is not the only path to adequate HI labeling. There are several alternative approaches: 

• Population PK (PopPK): Patients with hepatic impairment can be included in early efficacy studies, and hepatic function assessed as a covariate in a pre-specified PopPK analysis. Both the FDA and EMA support this approach, which aligns with the recently issued ICH M15 guidance. 

• PBPK modeling: Predict exposure in different levels of hepatic impairment, extrapolate from single to multiple dose, and supplement limited clinical data. While not explicitly mentioned in the FDA guidance, the EMA guidance references PBPK as a viable approach. 

• Combination approaches: In many cases, the most efficient strategy combines early PopPK data (e.g., from patients with mild impairment enrolled in phase 2/3) with PBPK modeling to address the full spectrum of impairment. 

• No study: If a sponsor decides not to conduct an HI study, labeling will reflect a lack of data. This will likely result in contraindication language for hepatically impaired patients and may also trigger a post-marketing commitment. 

PBPK modeling for hepatic impairment: a closer look

Simcyp® Simulator is a PBPK modeling software that separates drug data from systems data, allowing the software to model how physiological changes in a hepatically impaired population affect drug disposition. Key parameters built into the model include: 

• Demographic distributions (age, weight, height) 

• Organ weights and blood flows, including reduced liver volume and altered portal blood flow 

• Plasma protein concentrations (albumin and alpha-1-acid glycoprotein), which decrease with increasing disease severity 

• Enzyme and transporter expression levels across Child-Pugh A, B, and C categories 

• Glomerular filtration rate and gastric residence time

Three cirrhosis population models are available in Simcyp Simulator, corresponding to Child-Pugh A, B, and C classifications. 

Model performance

Across nine compounds and thirteen clinical studies, Simcyp Simulator’s hepatic impairment models predicted AUC ratios (impaired vs. healthy) within twofold across all Child-Pugh classifications, with average fold errors generally below 1.5.  

What increases the chance of regulatory acceptance?

Based on our PBPK modeling consultancy experience, the following factors improve the likelihood that PBPK modeling will be accepted in lieu of a clinical HI study: 

• A well-characterized ADME profile, including mass balance and enzyme phenotyping data 

• Good understanding of any PK nonlinearities and single-to-multiple dose transitions 

• Availability of drug-drug interaction (DDI) study data to refine fraction metabolized estimates 

• A predicted exposure shift that is modest or clinically manageable 

An evolving frontier: non-CYP enzymes and absorption

Most of our PBPK HI experience to date involves CYP-mediated clearance, particularly CYP3A4, where the modeling framework is well validated, but the field is evolving. Two key areas of active development include: 

• Non-CYP enzymes: Changes in UGT and other enzyme expression across Child-Pugh categories are increasingly well characterized in the literature, but validated probe substrates for these enzymes are limited. For example, only two clinical studies are currently available to validate Simcyp’s UGT1A1 predictions (using raltegravir as a probe substrate). 

• Absorption impairment: Work on avapritinib and other compounds is revealing that hepatic impairment affects not just elimination but also absorption through mechanisms including changes in gastric residence time, bile salt concentrations, and portal-systemic shunting. A robust mechanistic model for these absorption changes is still in development. 

Conclusion

Hepatic impairment studies sit at the intersection of regulatory obligation, patient safety, and drug development efficiency. With two regulatory guidances that predate many of today’s modeling capabilities, the field has evolved considerably beyond what those documents describe. The combination of well-designed clinical studies, pre-specified popPK analyses, and an EMA-qualified PBPK modeling platform like Simcyp Simulator now provides development teams a genuinely flexible toolkit for meeting HI labeling requirements while minimizing unnecessary burden. 

The key is to start early, understand your compound’s elimination pathways, and work with experienced clinical pharmacologists who can help you design the most efficient path to adequate labeling.