Drug-Induced Liver Injury (DILI): From Liver Test Abnormalities to Causality and Regulatory Safety Assessment

Purpose and Scope

Drug-induced liver injury (DILI) is one of the most important forms of medicine-related organ toxicity because liver injury may range from a transient laboratory abnormality to acute liver failure, transplantation or death. It is also one of the more difficult safety problems to interpret. Abnormal liver tests are common, patients frequently have underlying liver disease or other causes of abnormal results, and the relationship between biochemical injury and clinically important hepatic outcomes is not linear.

A useful DILI assessment therefore cannot be reduced to a single laboratory threshold. The reviewer must establish what pattern of liver injury occurred, when it occurred, whether the pattern is biologically and temporally plausible, whether alternative explanations are credible, what happened after treatment interruption, and whether similar cases or findings exist elsewhere in the development or post-authorisation safety data.

This article develops that assessment progressively. It begins with recognition and classification of liver injury, then moves to clinical and causality assessment, severe outcomes and Hy's Law, and finally to pharmacovigilance signal assessment and regulatory decision-making. The objective is to connect clinical interpretation with the evidence and governance processes used in drug safety.

The article is deliberately broader than Hy's Law. Hy's Law is an important marker of potentially serious hepatocellular injury, but it is only one part of the wider DILI assessment. EMA material on liver safety likewise emphasises the need to distinguish true drug-induced injury from fluctuations or progression of underlying liver disease and identifies causality assessment and identification of potential Hy's Law cases as important components of liver-safety evaluation. [1]

Why DILI Requires Structured Assessment

Liver-test abnormalities are not synonymous with clinically significant hepatotoxicity. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) indicate hepatocellular injury but do not, by themselves, establish hepatic dysfunction. Alkaline phosphatase (ALP) can indicate cholestatic injury, while bilirubin elevation may reflect hepatocellular dysfunction, cholestasis, haemolysis or other processes. Albumin and coagulation parameters provide different information about hepatic synthetic function and are influenced by factors outside acute liver injury.

The distinction between injury, dysfunction and clinical outcome is therefore fundamental.

A patient may have a marked ALT elevation with preserved hepatic function and subsequently recover. Another patient may have a less dramatic initial laboratory abnormality but develop jaundice, coagulopathy and hepatic failure. The safety significance depends on the complete clinical course rather than the magnitude of one laboratory value.

DILI assessment is consequently a reconstruction exercise:

baseline → exposure → biochemical change → clinical phenotype → competing causes → treatment change → recovery or progression → recurrence or further cases

The quality of that reconstruction determines how confidently a potential safety signal can be interpreted.

Regulatory and Scientific Framework

DILI sits at the intersection of clinical medicine, toxicology, clinical development and pharmacovigilance. EMA maintains a scientific reflection paper on the non-clinical evaluation of DILI and describes the use of non-clinical findings to identify and characterise hepatotoxic potential. The paper highlights the importance of dose and systemic exposure, metabolism, species relevance, historical controls and consistency across studies when interpreting non-clinical liver findings. [2]

For pharmacovigilance, the general EU framework remains the GVP system. GVP provides the framework for collection, evaluation and management of suspected adverse reactions and for signal management, while product-specific clinical and scientific guidance can provide additional context for hepatic safety. Current EMA GVP materials include Module VI for individual case safety reports and Module IX for signal management. [3]

The regulatory framework should not be confused with a single mandatory DILI algorithm. Many numerical thresholds and analytical approaches used in DILI assessment are scientific or regulatory conventions rather than universal legal requirements. The article therefore distinguishes regulatory obligations from scientific interpretation and recommended operational practice.

Recognising a Potential Liver Injury

A potential DILI case may first appear through an individual adverse-event report, a laboratory abnormality identified during clinical development, an aggregate safety analysis, a signal-detection activity, a medical-information enquiry, literature surveillance or another source.

The initial task is not to label the event as DILI immediately. It is to determine whether there is a credible hepatic signal that warrants structured assessment.

Relevant information includes:

• ALT and AST;
• ALP;
• total and, where useful, direct bilirubin;
• gamma-glutamyl transferase (GGT);
• albumin;
• coagulation parameters such as INR where clinically relevant;
• symptoms and clinical examination;
• imaging where appropriate;
• hepatitis and other infectious investigations where indicated;
• alcohol and metabolic history;
• concomitant medicines and supplements;
• pre-existing liver disease;
• timing of treatment and dose changes.

The reviewer should establish whether the abnormality is new, treatment-emergent, worsening from baseline, fluctuating within a pre-existing abnormal range, or more plausibly explained by another condition.

The Three Principal Biochemical Patterns

DILI is commonly characterised as hepatocellular, cholestatic or mixed according to the relative magnitude of ALT and ALP abnormalities.

A commonly used approach is the R-ratio:

R = (ALT / ALT ULN) ÷ (ALP / ALP ULN)

Using conventional classification:

The R-ratio is a classification tool, not a causality score. It describes the biochemical pattern at a defined point in the clinical course. The pattern may change as the injury evolves, and the result can be affected by baseline abnormalities and the timing of measurement.

The reviewer should therefore record the underlying laboratory values rather than relying only on the category.

Hepatocellular Injury

Hepatocellular injury is characterised predominantly by aminotransferase elevation. The differential diagnosis is broad and may include:

• viral hepatitis;
• autoimmune hepatitis;
• ischaemic or hypoxic injury;
• alcohol-related injury;
• metabolic-associated steatotic liver disease and steatohepatitis;
• biliary disease;
• muscle injury contributing to AST elevation;
• other medicines or supplements;
• hepatic malignancy or infiltration.

A high ALT does not therefore establish drug causality.

The clinical context becomes particularly important when the patient has a pre-existing hepatic disorder or a competing acute illness. EMA scientific material specifically highlights the difficulty of distinguishing progression or fluctuation of underlying liver disease from genuine drug-induced liver injury. [1]

Cholestatic Injury

Cholestatic injury is characterised by a disproportionate increase in ALP relative to ALT. The assessment should determine whether the ALP is genuinely hepatic in origin and whether there is evidence of biliary obstruction or another cholestatic disease.

GGT, bilirubin and imaging can help clarify the pattern. ALP may also originate from bone and other tissues, so an isolated ALP increase should not automatically be classified as hepatic injury.

Cholestatic DILI can have a different clinical course from hepatocellular injury. Some cases resolve relatively slowly, and persistent cholestatic abnormalities can require prolonged follow-up.

Mixed Injury

Mixed injury has features of both hepatocellular and cholestatic injury. It can be particularly challenging when the biochemical pattern changes during follow-up.

The reviewer should therefore avoid treating the first laboratory result as the complete phenotype. Serial values can show whether the pattern is stable, evolving or resolving.

Baseline Abnormalities

Baseline liver-test abnormalities are common in clinical practice and clinical development. They create a fundamental interpretive problem: a subsequent abnormal result may represent a treatment-emergent injury, worsening of an existing disease, ordinary biological fluctuation or another event.

Baseline information should therefore be captured before attributing causality. Relevant considerations include:

• the magnitude and pattern of baseline abnormalities;
• whether abnormalities were stable;
• known liver disease;
• recent infections or other acute illnesses;
• alcohol use where relevant;
• metabolic risk factors;
• concomitant medicines;
• previous hepatic events.

A patient who begins treatment with ALT already elevated should not be evaluated in exactly the same way as a patient with normal baseline liver tests who develops a new marked elevation. The analytical approach must account for the starting point.

DILI Is a Clinical Diagnosis Supported by Laboratory Evidence

No single laboratory result proves DILI. The diagnosis depends on the totality of evidence.

A useful mental model is:

pattern + timing + exposure + alternative causes + clinical course + dechallenge/rechallenge + biological plausibility + external evidence

Each component changes the probability of a drug-related explanation. None should automatically override the others.

This principle is central to the remainder of the article: the objective is not merely to identify abnormal liver tests, but to determine whether the medicinal product provides the most coherent explanation for the observed clinical course.

Clinical Characterisation and Causality Assessment

Once a potential liver injury has been recognised and its biochemical pattern described, the next question is whether the medicinal product plausibly caused or contributed to it. This is where DILI assessment becomes substantially more difficult than laboratory classification.

Establishing the Time Relationship

The chronology should be reconstructed using actual dates whenever possible.

Important points include:

• first exposure;
• dose and dose changes;
• cumulative exposure;
• onset of symptoms;
• first abnormal liver test;
• peak laboratory abnormality;
• treatment interruption;
• subsequent recovery;
• re-exposure, if it occurred.

Latency is supportive evidence, not proof. Different mechanisms can produce different latency patterns, and an apparently unusual latency should prompt investigation rather than automatic rejection of causality.

The temporal assessment should also consider whether exposure continued while the abnormality progressed. A rising ALT during continued exposure may carry different interpretive information from an abnormality that began after treatment was stopped.

Exposure and Dose

Drug exposure should be considered in relation to:

• prescribed dose;
• actual dose received;
• dose escalation;
• adherence;
• treatment duration;
• renal or hepatic impairment;
• pharmacokinetic interactions;
• drug accumulation;
• active metabolites.

A dose relationship can strengthen biological plausibility, but idiosyncratic DILI may occur without a simple dose-response relationship. Conversely, apparent dose dependence may reflect exposure to a metabolite or an interaction rather than direct toxicity.

The reviewer should therefore distinguish dose from systemic exposure.

Competing Causes

Alternative explanations are often the most important part of a DILI assessment.

Depending on the clinical context, the differential diagnosis may include:

• acute viral hepatitis;
• chronic viral hepatitis with flare;
• autoimmune hepatitis;
• biliary obstruction;
• gallstone disease;
• hepatic ischaemia or hypoxia;
• sepsis;
• alcohol-related injury;
• metabolic-associated steatotic liver disease;
• hepatic metastases or infiltration;
• cardiac failure with hepatic congestion;
• other hepatotoxic medicines;
• herbal products and dietary supplements.

The appropriate investigation depends on the phenotype and clinical context. There is no universal laboratory panel that excludes every alternative cause.

The strength of an alternative explanation should be assessed explicitly. Merely listing another diagnosis is insufficient. The reviewer should ask whether it explains the timing, biochemical pattern and clinical course better than the medicinal product does.

Concomitant Medicines

Polypharmacy can make DILI assessment particularly difficult. Several medicines may have known or suspected hepatic effects, and some may have been started shortly before the event.

The assessment should reconstruct the complete medication timeline, including non-prescription products where information is available.

The key question is not simply whether another medicine is hepatotoxic. It is whether that medicine provides a more plausible explanation for this patient's event.

Dechallenge

Dechallenge describes the course after treatment is interrupted or discontinued.

Improvement after withdrawal can support causality, particularly when the timing and biological course are plausible. It does not, however, prove causality.

Liver injury may improve spontaneously, and improvement may coincide with treatment of another cause. Some injuries also continue to worsen briefly after the suspected drug is stopped because the biological process does not terminate immediately when exposure ends.

A positive dechallenge should therefore be interpreted alongside:

• the magnitude of the original injury;
• the half-life of the medicine and active metabolites;
• expected biological recovery;
• treatment of competing causes;
• the timing of improvement.

Rechallenge

Rechallenge refers to re-exposure after recovery or improvement.

Recurrence of a compatible liver injury after re-exposure can provide powerful evidence of causality. However, rechallenge is not ordinarily something that should be undertaken merely to establish pharmacovigilance causality when safer alternatives exist. Clinical necessity, ethics and patient safety take precedence.

An unintentional or clinically necessary re-exposure may nevertheless provide important evidence.

The reviewer should document:

• whether the re-exposure was intentional or accidental;
• dose and duration;
• interval between withdrawal and re-exposure;
• timing of recurrence;
• whether the recurrent phenotype resembled the initial event.

Biological Plausibility

Mechanistic evidence can strengthen a DILI assessment. Relevant evidence may include:

• known class effects;
• hepatic metabolism;
• reactive metabolites;
• mitochondrial effects;
• bile-acid transporter inhibition;
• immune-mediated mechanisms;
• dose/exposure relationships;
• non-clinical findings.

Mechanistic plausibility should not be treated as proof. A plausible mechanism that does not fit the clinical course remains weak evidence, while an apparently unusual mechanism may still be credible when the clinical and epidemiological evidence is strong.

Evidence From Other Patients

The assessment should not stop at the individual case.

Relevant external evidence includes:

• other cases in the clinical programme;
• spontaneous reports;
• aggregate safety analyses;
• literature;
• epidemiological studies;
• regulatory assessments;
• known class effects.

A single case can be highly informative, particularly when the phenotype is distinctive and the temporal relationship is strong. Conversely, a large number of reports may provide little causal evidence if they are poorly documented or heavily confounded.

Severity and Clinical Consequences

DILI severity should be assessed separately from causality.

A strongly suspected drug-related injury may be mild, while a severe hepatic event may ultimately have another cause. Conflating these two questions can distort the safety assessment.

Laboratory Abnormality Versus Hepatic Dysfunction

Aminotransferase elevations primarily indicate hepatocellular injury. They do not by themselves establish loss of hepatic function.

More concerning clinical findings may include:

• jaundice;
• coagulopathy;
• hepatic encephalopathy;
• hypoglycaemia;
• ascites in an appropriate context;
• acute liver failure;
• transplantation;
• death.

Bilirubin elevation is particularly important when it accompanies hepatocellular injury because it can identify a population at higher risk of serious outcomes. This is the clinical reasoning underlying the importance of Hy's Law.

Symptoms

Symptoms can include:

• fatigue;
• nausea;
• anorexia;
• abdominal discomfort;
• pruritus;
• dark urine;
• jaundice.

Symptoms are non-specific. Their value comes from their relationship to the laboratory abnormalities and overall clinical course.

Follow-Up

A liver-safety case should be followed sufficiently to establish the trajectory where clinically and operationally feasible.

Important questions include:

• Did liver tests continue to rise?
• Did they stabilise?
• Did they return toward baseline?
• How rapidly did recovery occur?
• Did bilirubin or INR worsen?
• Was another treatment introduced?
• Was the suspected medicine restarted?

The final interpretation may change as follow-up information becomes available.

Special Situations

Pre-Existing Liver Disease

Underlying liver disease can both increase the clinical complexity of a case and make causality assessment more difficult.

Examples include:

• chronic viral hepatitis;
• cirrhosis;
• metabolic-associated steatotic liver disease;
• autoimmune liver disease;
• cholestatic disorders;
• hepatic malignancy.

The reviewer should distinguish a new treatment-emergent pattern from fluctuation or progression of the underlying condition.

Liver Metastases

Patients with cancer may have liver metastases or other hepatic involvement that can alter liver tests independently of treatment.

Progressive disease may therefore provide a strong alternative explanation for worsening liver tests. Conversely, the presence of metastases does not exclude DILI.

Serial imaging, tumour status, baseline laboratory values and the timing of changes should be considered together.

Acute Illness and Hypoperfusion

Sepsis, shock and cardiac or respiratory compromise can cause substantial liver-test abnormalities. These events may overlap temporally with treatment and can create an apparent drug-related signal.

The clinical record should therefore establish whether there was:

• hypotension;
• hypoxia;
• sepsis;
• cardiac failure;
• intensive-care illness;
• other evidence of hepatic hypoperfusion.

Combination Therapy

When several medicines are administered together, attribution may be uncertain. A combination may also create an interaction that changes exposure to one component.

The reviewer should assess each plausible medicine rather than automatically attributing the event to the most recently introduced product.

From Individual Case to Safety Signal

A DILI case becomes more informative when considered with the broader evidence base.

The transition can be expressed as:

individual case → case series → pattern recognition → signal assessment → causality and clinical relevance → regulatory decision

GVP Module IX describes signal management as a process for detecting, validating, confirming, analysing and prioritising safety signals. EMA explains that a safety signal does not itself establish that a medicine caused the reported event; signal assessment is intended to determine whether a causal relationship is plausible and whether further regulatory action is warranted. [3,4]

DILI therefore illustrates a general pharmacovigilance principle: the existence of reports and the existence of a causal safety signal are different evidentiary states.

Quality of the DILI Evidence Chain

For an individual case, the evidence chain should allow another medically qualified reviewer to reconstruct:

exposure → baseline → onset → laboratory trajectory → clinical assessment → alternative causes → treatment changes → outcome → causality conclusion

For aggregate assessment, the chain should additionally allow reconstruction of:

case identification → case validation → case classification → case review → pattern analysis → signal conclusion → action or rationale for no action

This traceability is particularly important when cases are used in regulatory submissions, signal evaluations, periodic safety reporting or risk-management decisions.

Hy's Law and Serious Hepatic Injury

Hy's Law is one of the most important concepts in drug-related liver safety, but it is also one of the most frequently over-simplified. It is not simply a rule stating that ALT above a particular threshold plus bilirubin above another threshold equals drug-induced liver injury. The concept combines a pattern of hepatocellular injury, bilirubin elevation and the absence of a more plausible alternative explanation.

The original clinical observation is associated with Hyman Zimmerman. Modern regulatory applications commonly use laboratory criteria as a means of identifying potential cases that require further clinical and causality assessment.

The Core Concept

In clinical-development safety assessment, a potential Hy's Law case generally involves:

1. evidence of hepatocellular injury, commonly represented by substantial ALT elevation;
2. bilirubin elevation in the absence of a better explanation such as obstruction or another cause; and
3. no more plausible alternative explanation for the overall liver injury.

FDA guidance commonly describes laboratory criteria involving ALT or AST elevation of at least three times the upper limit of normal together with total bilirubin above twice the upper limit of normal, with the absence of substantial ALP elevation and alternative explanations being important elements of the assessment. These criteria are a regulatory/scientific screening convention and should not be presented as an EU legal definition. [5]

EMA clinical and regulatory material similarly uses identification and detailed assessment of potential Hy's Law cases as part of liver-safety evaluation. [1]

Why Bilirubin Changes the Significance

ALT elevation indicates hepatocellular injury. Bilirubin elevation can indicate that the injury is associated with impaired bilirubin handling or excretion and may therefore identify a more concerning phenotype.

The concern is not that every patient meeting laboratory thresholds will develop liver failure. Rather, the combination identifies a subgroup in which serious drug-induced liver injury is possible and therefore deserves careful assessment.

The correct operational sequence is therefore:

laboratory criteria → potential case → clinical review → alternative-cause assessment → causality determination → aggregate interpretation

Stopping after the laboratory criteria risks over-classification.

Why ALP Matters

A substantial ALP elevation may suggest a cholestatic process or biliary obstruction rather than the classic hepatocellular pattern associated with Hy's Law.

The reviewer should therefore establish whether the bilirubin elevation is associated with:

• hepatocellular injury;
• cholestasis;
• biliary obstruction;
• haemolysis;
• another non-hepatic cause.

This is why the complete biochemical profile matters more than isolated ALT and bilirubin values.

Alternative Explanations

Potential alternative explanations should be investigated explicitly.

Examples include:

• viral hepatitis;
• autoimmune hepatitis;
• biliary obstruction;
• sepsis;
• hepatic ischaemia;
• liver metastases;
• progression of underlying liver disease;
• other hepatotoxic medicines;
• alcohol-related injury.

A patient can meet laboratory criteria while not representing a drug-induced Hy's Law case. EMA public regulatory assessments illustrate this distinction: patients may have laboratory values fulfilling screening criteria while progressive disease, pre-existing hepatic abnormalities, sepsis or other factors provide alternative explanations. [6]

Hy's Law Is Not a Causality Algorithm

A common error is to treat Hy's Law as if it were a formal causality score.

It is not.

Hy's Law identifies a clinically concerning pattern. Causality still requires evaluation of:

• timing;
• exposure;
• alternative causes;
• dechallenge;
• rechallenge;
• pharmacology;
• other cases;
• disease course.

A case may meet laboratory criteria but be judged unlikely to be caused by the medicine. Conversely, a serious drug-related hepatic event may not fit a classic Hy's Law pattern.

This distinction is essential when interpreting clinical-trial data and post-marketing cases.

eDISH and Aggregate Liver Safety Assessment

In clinical development, graphical approaches such as evaluation of Drug-Induced Serious Hepatotoxicity (eDISH) can help identify participants with potentially concerning combinations of ALT and bilirubin abnormalities.

An eDISH plot can support identification and visualisation of potential Hy's Law cases, but it does not replace medical review.

The reviewer still needs to determine:

• whether the abnormalities were treatment-emergent;
• whether baseline abnormalities affect interpretation;
• whether the pattern is hepatocellular;
• whether bilirubin has an alternative explanation;
• whether competing causes exist;
• whether the clinical course supports drug causality.

The value of eDISH is therefore organisational as well as analytical: it helps bring potentially important cases into a structured review process.

DILI in Clinical Development

Liver safety should be considered prospectively during clinical development.

The development programme should have an approach for:

• baseline liver assessment;
• identification of treatment-emergent liver abnormalities;
• repeat testing;
• clinical evaluation;
• treatment interruption or discontinuation where appropriate;
• follow-up;
• identification of potential Hy's Law cases;
• adjudication or expert review where justified;
• aggregate analysis.

The exact operational approach depends on the medicinal product, population, mechanism, development stage and therapeutic context. It should not be represented as a universal legally prescribed algorithm.

Adaptation and Resolution

Some patients develop aminotransferase elevations that subsequently decline despite continued treatment. This phenomenon can complicate interpretation because spontaneous or treatment-associated adaptation does not automatically prove that the initial elevation was unrelated to the medicine.

The reviewer should distinguish:

• adaptation during continued exposure;
• improvement after dose reduction;
• improvement after treatment interruption;
• recovery after permanent discontinuation.

Each provides different information.

Clinical Trial Comparator Context

The frequency and magnitude of liver-test abnormalities should be interpreted in relation to:

• placebo or active comparator;
• background disease;
• baseline abnormalities;
• concomitant treatments;
• exposure duration;
• population characteristics.

A small number of liver-test abnormalities can be clinically important if the pattern is distinctive, while a higher number of mild abnormalities may have limited causal significance if they occur similarly in comparator groups.

DILI in Post-Marketing Pharmacovigilance

Post-marketing DILI assessment differs from clinical-development assessment because the evidence is more heterogeneous.

Spontaneous reports may contain:

• incomplete laboratory data;
• uncertain onset dates;
• missing baseline results;
• incomplete concomitant-medication histories;
• uncertain diagnoses;
• variable follow-up.

The pharmacovigilance reviewer must therefore work with imperfect evidence while preserving a consistent assessment framework.

Case Review

A medically meaningful DILI case review should establish, where available:

• exact exposure dates;
• dose and indication;
• baseline liver tests;
• serial liver tests;
• symptoms;
• imaging;
• relevant investigations;
• competing diagnoses;
• concomitant medicines;
• dechallenge;
• rechallenge;
• outcome.

The absence of information should be documented as uncertainty rather than converted into a negative finding.

For example, "no alternative cause identified" is not necessarily equivalent to "all alternative causes were excluded".

Case Series

When several cases accumulate, the assessment should look for common structure:

• similar latency;
• similar biochemical phenotype;
• similar dose or exposure relationship;
• recurrence on rechallenge;
• common risk factors;
• class effects;
• geographical or formulation patterns.

A case series can strengthen a signal even when individual cases are imperfect. Conversely, a large heterogeneous series can remain difficult to interpret if the phenotype is inconsistent and reporting is biased.

Signal Detection and Validation

DILI signals may arise from spontaneous reports, clinical development, literature, epidemiological studies or combinations of evidence.

Signal detection is only the beginning. EMA describes signal management as a process that establishes whether a safety signal represents a potential causal relationship and whether further assessment or action is required. [4]

For DILI, validation should ask whether the apparent signal represents:

• a genuine drug-related liver injury;
• a known background event;
• reporting stimulated by publicity;
• confounding by indication;
• confounding by concomitant treatment;
• a coding artefact;
• duplicated cases;
• a mixture of different hepatic phenotypes.

Regulatory Decision-Making

A confirmed or strongly suspected DILI signal does not automatically dictate one regulatory outcome.

Possible responses depend on:

• seriousness;
• frequency;
• strength of causality evidence;
• dose/exposure relationship;
• reversibility;
• affected population;
• therapeutic alternatives;
• disease severity;
• benefit-risk balance;
• preventability;
• existing risk-minimisation measures.

Possible actions may include:

• continued monitoring;
• targeted additional analysis;
• product-information changes;
• laboratory-monitoring recommendations;
• contraindications or warnings;
• risk-minimisation measures;
• post-authorisation studies;
• further clinical investigation;
• regulatory referral or other action where warranted.

The regulatory decision is therefore the endpoint of an evidence chain rather than a mechanical consequence of one laboratory threshold.

Relationship With Risk Management

When DILI becomes an identified or important potential risk, the safety concern may affect the risk-management system.

The assessment may influence:

• the safety specification;
• routine risk-minimisation measures;
• additional risk-minimisation measures;
• pharmacovigilance activities;
• additional monitoring;
• targeted follow-up of hepatic events.

The precise RMP consequences depend on the medicine and the regulatory assessment. GVP Module V provides the general EU framework for risk-management systems, while the underlying DILI evidence determines how a hepatic risk is characterised and managed. [3]

Inspection and Governance Perspective

Although DILI is primarily a medical and scientific subject, the process by which DILI cases are identified and assessed can be inspected.

An effective pharmacovigilance system should be able to demonstrate:

• how potential hepatic cases are identified;
• how medically important cases are escalated;
• how laboratory data are obtained and reconciled;
• how causality is assessed;
• how follow-up is requested;
• how aggregate DILI signals are reviewed;
• how decisions are documented;
• how conclusions feed into signal management, PSURs and risk management where appropriate.

A useful inspection question is not simply "Did the company identify DILI?" but:

Can the organisation demonstrate a controlled and scientifically defensible process from the initial hepatic observation to the final safety conclusion and any resulting action?

This connects DILI assessment directly to the broader pharmacovigilance quality-system principles.

Practical Assessment Framework

A structured DILI review can be organised into a sequence that is reproducible without turning the process into a rigid scoring exercise.

Step 1: Establish the Exposure

Confirm:

• medicinal product;
• indication;
• start and stop dates;
• dose;
• dose changes;
• treatment duration;
• concomitant medicines;
• relevant exposure modifiers.

Step 2: Establish the Baseline

Determine:

• baseline ALT and AST;
• ALP;
• bilirubin;
• relevant hepatic history;
• previous liver abnormalities;
• underlying liver disease.

Step 3: Characterise the Injury

Determine the biochemical pattern using the actual laboratory values and, where appropriate, the R-ratio.

Ask whether the pattern is:

• hepatocellular;
• cholestatic;
• mixed;
• or not adequately characterised.

Step 4: Reconstruct the Timeline

Map exposure, laboratory changes, symptoms, dose changes, treatment interruption and recovery.

Step 5: Investigate Alternatives

Assess the clinically relevant competing explanations rather than merely listing them.

Step 6: Assess Dechallenge and Rechallenge

Determine whether improvement followed withdrawal and whether recurrence followed re-exposure, while recognising the limitations of both observations.

Step 7: Consider External Evidence

Review other cases, clinical-trial findings, literature, class effects and mechanistic evidence.

Step 8: Assess Severity Separately

Determine whether the event involved:

• isolated laboratory injury;
• symptomatic hepatitis;
• jaundice;
• coagulopathy;
• hepatic failure;
• transplantation;
• death.

Step 9: Reach a Causality Conclusion

The conclusion should state the strength of the evidence and the principal uncertainties. Avoid false precision when important information is missing.

Step 10: Determine the Pharmacovigilance Consequence

Consider whether the case or case series requires:

• follow-up;
• signal validation or analysis;
• aggregate review;
• PSUR assessment;
• RMP consideration;
• product-information review;
• additional risk-minimisation or pharmacovigilance activity.

Illustrative Case: Hepatocellular Injury With a Strong Alternative Cause

Illustrative scenario — not a reported regulatory inspection finding.

A patient begins a medicinal product and develops ALT elevation several weeks later. Bilirubin also increases. The laboratory pattern appears compatible with a potential Hy's Law case.

During clinical review, the patient is found to have progressive hepatic metastases with imaging evidence of rapid disease progression. Baseline liver tests were already abnormal and had been worsening before the suspected medicine was introduced.

The correct conclusion is not to dismiss the laboratory abnormalities. They require review. However, the presence of a credible alternative explanation substantially changes the causality assessment. The laboratory criteria should trigger further investigation rather than automatically establish a drug-induced Hy's Law case.

This illustrates the central principle of DILI assessment:

screening criteria identify cases for evaluation; they do not replace evaluation.

Illustrative Case: Positive Dechallenge

Illustrative scenario — not a reported regulatory inspection finding.

A patient develops a new hepatocellular pattern after starting treatment. Alternative causes are investigated without identifying a convincing explanation. Treatment is discontinued, and ALT subsequently falls toward baseline over an appropriate period.

The positive dechallenge strengthens the drug-related hypothesis, but the conclusion should still consider the drug's pharmacokinetics, the expected natural history of the injury and whether another intervention occurred at the same time.

The case may therefore be assessed as strongly suggestive of causality without claiming that dechallenge alone proves it.

Illustrative Case: Rechallenge

Illustrative scenario — not a recommendation to rechallenge.

A patient develops a compatible hepatic injury, recovers after treatment withdrawal and is later inadvertently re-exposed. A similar liver-test pattern recurs shortly after re-exposure.

A reproducible recurrence can provide strong causal evidence. The case should document the circumstances of re-exposure and the similarity between the initial and recurrent events.

The existence of such evidence does not mean that deliberate rechallenge is appropriate. Patient safety and clinical necessity remain the governing considerations.

Common Analytical Errors

Several recurring errors can weaken DILI assessment.

Treating ALT Elevation as Causality

ALT elevation establishes neither DILI nor drug causality. The differential diagnosis must be considered.

Treating Hy's Law Laboratory Criteria as the Final Diagnosis

Laboratory criteria identify potential cases requiring clinical and causality assessment. They do not eliminate alternative causes.

Ignoring Baseline Values

Without baseline information, treatment-emergent change may be overestimated or underestimated.

Ignoring Serial Results

A single laboratory value can conceal the direction and duration of the injury.

Treating Missing Information as Negative Evidence

If viral investigations were not performed, the correct conclusion is usually that the alternative cause was not fully evaluated, not that viral hepatitis was excluded.

Ignoring Concomitant Medicines

Attribution to the study or marketed medicine without reconstructing the complete exposure history can produce false signals.

Overinterpreting Dechallenge

Improvement after discontinuation supports but does not prove causality.

Overinterpreting Rechallenge

Recurrence can be highly informative, but the circumstances and clinical phenotype must be assessed carefully.

Confusing Severity With Causality

A severe hepatic event may have another cause. A mild event may nevertheless be drug-related.

Treating Every DILI Signal as a Regulatory Crisis

The appropriate response depends on evidence, seriousness, frequency, preventability and benefit-risk context. Signal identification is not synonymous with confirmation.

Key Takeaways

1. DILI is a clinical diagnosis supported by laboratory, temporal and contextual evidence rather than a single laboratory threshold.
2. The first task is to characterise the biochemical pattern and reconstruct the clinical timeline.
3. The R-ratio is a pattern-classification tool, not a causality score.
4. Baseline abnormalities and alternative causes are central to interpretation.
5. Dechallenge and rechallenge can provide important evidence but have limitations and must be interpreted clinically.
6. Hy's Law identifies a concerning hepatocellular injury pattern associated with bilirubin elevation; it is not itself a complete causality algorithm.
7. Potential Hy's Law cases require assessment of alternative explanations before being treated as drug-related cases.
8. Clinical-development and post-marketing DILI assessment use different evidence environments but share the same underlying reasoning principles.
9. A safety signal is a hypothesis requiring evaluation, not proof that the medicine caused the event.
10. The pharmacovigilance value of DILI assessment depends on a traceable evidence chain from exposure and laboratory findings to causality, aggregate interpretation and regulatory action.

References

1. European Medicines Agency (EMA). Reflection paper on regulatory requirements for development of medicinal products for primary biliary cholangitis and primary sclerosing cholangitis — section on liver safety, including distinction between underlying liver disease and DILI and identification of potential Hy's Law cases. EMA. 2023. https://www.ema.europa.eu/

2. European Medicines Agency (EMA). Reflection paper on non-clinical evaluation of drug-induced liver injury (DILI). EMEA/CHMP/SWP/150115/2006. https://www.ema.europa.eu/en/non-clinical-evaluation-drug-induced-liver-injury-dili-scientific-guideline

3. European Medicines Agency (EMA). Good pharmacovigilance practices (GVP), including Module V — Risk management systems and Module VI — Collection, management and submission of reports of suspected adverse reactions. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/good-pharmacovigilance-practices-gvp

4. European Medicines Agency (EMA). GVP Module IX — Signal management and EMA signal-management information. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/signal-management

5. U.S. Food and Drug Administration (FDA). Drug-Induced Liver Injury: Premarketing Clinical Evaluation — Guidance for Industry. FDA. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/drug-induced-liver-injury-premarketing-clinical-evaluation

6. European Medicines Agency (EMA). EU public assessment reports and risk-management documents illustrating clinical assessment of potential Hy's Law cases and consideration of alternative causes. EMA. https://www.ema.europa.eu/

7. European Medicines Agency (EMA). Risk minimisation measures and GVP Module XVI. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/risk-minimisation-measures-rmm

Regulatory Note

This article is a pharmacovigilance and drug-safety educational reference, not a statement that every scientific threshold or assessment convention described here is legally mandatory in the European Union.

EU pharmacovigilance obligations arise principally from applicable EU legislation and associated regulatory requirements. GVP provides detailed guidance on pharmacovigilance processes, including individual case management, signal management and risk-management systems. Scientific concepts such as the R-ratio, eDISH and Hy's Law laboratory criteria are used in clinical and regulatory safety assessment, but their use should not be represented as a single universally applicable legal algorithm unless the applicable regulatory document expressly establishes such a requirement.

The current EMA GVP framework should be checked before applying this article operationally because GVP modules and related EU requirements can be revised. The EMA identifies the current GVP modules and their effective dates on its GVP page. [3]

The FDA DILI guidance cited above is a U.S. regulatory document and is included for scientific and regulatory context. It must not be presented as an EU legal requirement. Similarly, illustrative cases in this article are hypothetical teaching examples and are not claimed to be actual regulatory inspection findings or published individual cases unless expressly identified as such.

For an actual medicinal product, the final safety assessment should take account of the product's indication, population, pharmacology, clinical-development evidence, post-authorisation evidence, applicable product-specific guidance, current product information, RMP and the current regulatory position.