Causality assessment in hepatotoxicity by drugs and dietary supplements

Abstract

Structured causality assessment of hepatotoxicity by drugs and dietary supplements (DDS) is a major clinical challenge, since temporal associations as the sole criteria for a valid evaluation are not acceptable. Initially, a clear intuition for an ad hoc evaluation is necessary, but only provisional, and must be followed by a diagnostic algorithm using a pretest, main test and post test. The evaluation is based on a variety of items such as latency period, course of alanine aminotransferase and alkaline phosphatase after DDS discontinuation, risk factors, co-medication, previous information on hepatotoxicity of the DDS, response to rechallenge, and exclusion of other diseases. It is essential that practising and hospital physicians as well as other key health professionals, such as pharmacists, gather all information required for a sound causality assessment, obviating major discussions by expert panels, manufacturers and health agencies in face of scanty and fragmentary data. Because pharmacogenetic alterations may trigger metabolic hepatotoxicity by a few DDS, levels in plasma and urine should be measured and may be helpful for diagnosis. Concomitant genotyping of cytochrome P450 and other enzymes may also be useful in future to minimize the risk of unwanted side-effects, including toxic liver disease elicited by DDS.

Hepatotoxicity is a rare, unpredictable and inherent risk associated with the use of various drugs and dietary supplements (DDS) at recommended doses [1–9], and several excellent reviews have covered the broad spectrum of DDS hepatotoxicity [10–23]. Potentially hepatotoxic drugs include prescribed and over-the-counter medications, which are intensively studied before marketing both experimentally and in several thousands of healthy persons and patients regarding unwanted side-effects, including hepatotoxicity [24, 25]. Nevertheless, rare toxic liver injury cannot be excluded with certainty, as after marketing hepatotoxic reactions occur, for example, in 1:10 000 patients treated for specific diseases [11, 18]. Moreover, there is little if any legal regulation of dietary supplements, which comprise a basket of herbs and teas as botanical supplements and weight loss supplements containing herbs and various substances [26–32]. In face of > 1100 potentially hepatotoxic drugs [17] and the emerging consumption of dietary supplements [13, 26], DDS hepatotoxicity has a major impact not only on morbidity, including hospitalization, disability and liver transplantation, but also on mortality, liability and litigation.

Early recognition and comprehensive documentation of DDS hepatotoxicity is mandatory at all stages of the disease, but current procedures for efficient, transparent and impartial causality assessment in suspected DDS hepatotoxicity are inadequate. At present, practising and hospital physicians, expert groups, manufacturers and healthcare agencies each use their own individual causality approach, open to discussion and not free of conflicts of interest [24]. These difficulties are perpetuated and also evident in the databases of the World Health Organization (WHO) and various national health agencies, showing causality assessments being either hardly practicable or highly debatable [1, 6]. Consequently, there is an urgent need for a universally accepted stepwise causality assessment comprehensively established by the physicians and reviewed by the manufacturers and health agencies to achieve a grading of causality accepted by all parties. A case with verified causality should then undergo further pharmacovigilance evaluation and be submitted to the databases of the national health organizations and the WHO. The present review will focus on possible surrogate markers and structured causality assessment methods suitable for the diagnosis of DDS hepatotoxicity.

Surrogate markers

Various biochemical, functional or morphological criteria are commonly used as surrogate markers to facilitate diagnosis and/or to evaluate treatment efficacy. Presently, for DDS hepatotoxicity laboratory, immunological and genetic data have to be considered as surrogate markers, partially so under certain test conditions.

Suspicion of hepatotoxicity by DDS requires a detailed exploration of the patient for the use of prescribed or over-the-counter drugs as well as herbal and other dietary supplements. Constituents of DDS may interact with different cell populations of the liver involving various mediators and could thereby be harmful to this organ [10, 12], but specific diagnostic clues have not emerged. Similarly, since clinical symptoms such as fatigue, loss of appetite, right upper quadrant discomfort, scleral icterus, jaundice, dark urine and pale stool as well as liver histology commonly lack specificity and are not helpful to establish the diagnosis of DDS hepatotoxicity firmly, further approaches are therefore warranted using biochemical criteria as possible diagnostic surrogate markers.

Rechallenge test

A positive rechallenge test is commonly considered as a surrogate marker and gold standard for the diagnosis of hepatotoxicity [10]. Historically, rechallenge tests have frequently been performed to establish drug-induced liver injury firmly, but this approach has now been abandoned due to high risks. Nevertheless, results are still rarely available from unintentional rechallenges reported for some DDS [2, 33].

Although a positive rechallenge test supports a DDS aetiology, the sensitivity of the test is not 100% [10]. For example, in isoniazid-associated liver injury, the drug can sometimes be introduced again after a convincing positive dechallenge without development of a new liver injury after the rechallenge due to some kind of adaptive process.

DDS levels

Levels of DDS constituents or their metabolites in plasma and urine are rarely measured [10, 19]. High levels reflect cumulative doses of slowly metabolized DDS or genetic deficiencies and may be useful surrogate markers in metabolic rather than immunological DDS hepatotoxicity [13, 34–36].

Cytochrome P450

Determination of genetic polymorphism of cytochrome P450 isoenzymes and various other enzymes is of interest in future as surrogate markers in DDS hepatotoxicity [34–36], since genetic deficiencies decrease metabolic rates of exogenous compounds and concomitantly enhance, for example, their levels, leading to various side-effects, including hepatotoxicity [11, 15, 16]. Other new technologies for predicting hepatotoxicity related to transcriptomics, proteomics and metabolomics may emerge [11, 22].

Lymphocyte transformation test

This test is cumbersome, debated regarding validation, and therefore not suitable as a common surrogate marker for DDS hepatotoxicity [10, 37, 38].

Immunological features

In a few patients with DDS hepatotoxicity signs of hypersensitivity such as peripheral and hepatic eosinophilia and autoimmune parameters such as antinuclear antibodies, antimitochondrial antibodies, smooth muscle antibodies, anti-liver kidney microsome (LKM)-1 and anti-LKM-2 are present, but their use as surrogate markers has not yet been sufficiently validated [10, 22].

Ad hoc evaluation

Patients with abnormal liver tests, jaundice or acute liver failure (ALF) in temporal association with the use of DDS require ad hoc causality assessment and possible discontinuation of its intake. This initial step in the diagnostic assessment is inevitable for the sake of the patients, but has some limitations, since other diagnoses may be missed as evaluated by structured causality assessment [39–53]. The list of missed diagnoses in actual cases includes Wilson's disease [2, 45], haemochromatosis [43], autoimmune hepatitis [2, 45, 47], virus hepatitis [2, 43, 45], Epstein–Barr virus (EBV) [48], cytomegalovirus (CMV) [48], ischaemic hepatitis [2, 43, 45, 47], cardiac hepatopathy [43], chronic liver disease [48], liver cirrhosis [48], fatty liver [48], non-alcoholic steatohepatitis [2, 45], alcoholic liver disease [2, 45, 47], Gilbert's syndrome [48], tumours [2, 45, 47], lymphoma [48], bile duct diseases [2, 43, 45], systemic sepsis [2, 45, 47], chlamydial infection [43], thyroid disease [2, 45] and postictal [48].

Structured causality assessment methods

Causality assessment in DDS hepatotoxicity may be achieved by a variety of scales [39–53]. With laboratory data of serum alanine aminotransferase (ALT) and alkaline phosphatase (ALP) as basis, two different categories of hepatocellular and cholestatic (±hepatocellular) injury are evident (Table 1) [42]. Based on this differentiation, numerous clinical studies have been performed in DDS hepatotoxicity with several evaluation systems [2, 3, 42–49].

Table 1.

Pre-test

	Patient			Patient
Hepatocellular injury	Yes	No	Cholestatic (±hepatocellular) injury	Yes	No
I.Causality unrelated			I.Causality unrelated
1.Reaction occurred before starting the drug			1.Reaction occurred before starting the drug
2.Reaction occurred >15 days after stopping the drug (except for slowly metabolized drugs)			2.Reaction occurred >30 days after stopping the drug (except for slowly metabolized drugs)
3.Nondrug cause highly probable			3.Nondrug cause highly probable
4.Decrease of ALT <50% after the day 30 of drug discontinuation
5.Recurrent increase of ALT during drug discontinuation
II.Causality not assessable			II.Causality not assessable
1.No information to calculate time to onset			1.No information to calculate time to onset
2.No information about the course of ALT			2.No information about the course of ALP
3.Continued drug therapy			3.Continued drug therapy
4.Decrease of ALT ≥50% after the day 30 of drug discontinuation			4.Persistence or increase of ALP after drug discontinuation

For differentiation of the hepatocellular, cholestatic or mixed form of hepatotoxicity caused by drugs and dietary supplements (DDS), serum activities of alanine aminotransferase (ALT) and alkaline phosphatase (ALP) are measured on the day the diagnosis of DDS hepatotoxicity is suspected. Each activity is expressed as a multiple of the upper limit of the normal range (N), and the ratio (R) of ALT:ALP is calculated. Liver injury is (i) hepatocellular, when ALT > 2 N alone or R ≥ 5, (ii) cholestatic when there is an increase of ALP > 2 N alone or when R ≤ 2, and (iii) of the mixed type when ALT > 2 N, ALP is increased and 2 < R < 5. When at least one question is answered with yes, causality is either unrelated or not assessable. The pretest and subsequent main test evaluate the hepatocellular injury separately from the cholestatic (±hepatocellular) one.

The first two published causality assessments have a qualitative approach: the Roussel Uclaf Causality Assessment Method (RUCAM) [39, 40] and qualitative Council for International Organizations of Medical Sciences (CIOMS) [41]. A primarily quantitative assessment represents the improved CIOMS [42, 43], which is considered as the CIOMS in general and has only a few qualitative items. The scale of Maria and Victorino (MV) is purely quantitative [44]. The causality assessment of Aithal and Day (AD) [47] includes three approaches: the qualitative CIOMS system [41], the MV scale [44], and liver histology. The evaluation by Aithal, Rawlins and Day (ARD) [49] performs without liver histology and uses as the first step some criteria of the qualitative CIOMS [41] and subsequently the MV scale [44]. Finally, the grading system of Takamori, Takikawa, Kumagi et al. (TTK) [52] is worth mentioning.

The diagnostic significance of the scores has been discussed in detail and differently evaluated [46, 51, 53]. Based on pragmatic grounds and these comments, a causality algorithm in DDS hepatotoxicity is presented using a pretest, main test and post test. The other structured causality assessment methods will be discussed subsequently.

Pretest

The first step of the structured causality algorithm in toxic liver disease by DDS represents the qualitative oriented pretest for the hepatocellular or the cholestatic (± hepatocellular) type of injury, which clarifies with a few questions whether causality is unrelated or not evaluable (Table 1). The items of the pretest are based on qualitative criteria and validated as described in the qualitative CIOMS [41] and the CIOMS [42, 43]. Each DDS has to be evaluated separately. When the observed liver disease is found to be unrelated to DDS as evaluated by the pretest, no further steps for causality assessment are required. The pretest may therefore be a preferential tool for regulatory agencies and drug manufacturers as an easy and less time-consuming assessment.

Main test

The main test (Table 2) should be used in the second step, provided the pretest has failed to give a clear negative signal for causality assessment. Each individual DDS has to be evaluated separately. First, levels of DDS constituents or their metabolites in plasma and urine should be determined, which may facilitate causality assessment. The main test corresponds otherwise to the well-validated assessment scores of CIOMS [42] with various modifications for reasons of precision and actualization. In the main test there are no more qualitative items (Table 2) as in the CIOMS scores [42], all of which have been transferred to the previous pretest (Table 1). In the main test, confirmation of the exclusion of hepatitis B and C is achieved by the inclusion of hepatitis B virus-DNA and hepatitis C virus-RNA determination (Table 2). Similarly, the use of colour Doppler sonography for the assessment of hepatic vessels has been added. For the exclusion of EBV, CMV, herpes simplex virus (HSV) and varicella zoster infection, the determination of polymerase chain reaction as well as IgM and IgG antibodies with subsequent titre changes is required on grounds of precision. All other items of the CIOMS scale [42] have been incorporated in the main test (Table 2), making new validation unnecessary.

Table 2.

Main-test

Hepatocellular injury	Score	Patient	Cholestatic (±hepatocellular) injury	Score	Patient
1.Time to onset from the beginning of the drug			1.Time to onset from the beginning of the drug
•5–90 days (rechallenge: 1–15 days)	+2		•5–90 days (rechallenge: 1–90 days)	+2
•<5 or >90 days (rechallenge: >15 days)	+1		•<5 or >90 days (rechallenge: >90 days)	+1
2.Time to onset from cessation of the drug			2.Time to onset from cessation of the drug
•≤15 days (except for slowly metabolized drugs: >15 days)	+1		•≤30 days (except for slowly metabolized drugs: >30 days)	+1
3.Course of ALT after cessation of the drug			3.Course of ALP after cessation of the drug
Difference between peak of ALT and upper limit of normal range			Difference between peak of ALP and upper limit of normal range
•Decrease ≥50% within 8 days	+3		•Decrease ≥50% within 180 days	+2
•Decrease ≥50% within 30 days	+2		•Decrease <50% within 180 days	+1
•No information	0		•Persistence, increase or no information	0
•Decrease ≥50% after day 30	0
•Decrease <50% after day 30 or recurrent increase	−2
4.Risk factor ethanol			4.Risk factor ethanol or pregnancy
•yes	+1		•yes	+1
•no	0		•no	0
5.Risk factor age			5.Risk factor age
•≥55 years	+1		•≥55 years	+1
•<55 years	0		•<55 years	0
6.Concomitant drug(s)			6.Concomitant drug(s)
•None or no information	0		•None or no information	0
•Concomitant drug with incompatible time to onset	0		•Concomitant drug with incompatible time to onset	0
•Concomitant drug with compatible or suggestive time to onset	−1		•Concomitant drug with compatible or suggestive time to onset	−1
•Concomitant drug known as hepatotoxin and with compatible or suggestive time to onset	−2		•Concomitant drug known as hepatotoxin and with compatible or suggestive time to onset	−2
•Concomitant drug with evidence for its role in this case (positive rechallenge or validated test)	−3		•Concomitant drug with evidence for its role in this case (positive rechallenge or validated test)	−3
7.Search for nondrug causes			7.Search for nondrug causes
Group I (6 causes)			Group I (6 causes)
•Anti-HAV-IgM			•Anti-HAV-IgM
•Anti-HBc-IgM/HBV-DNA			•Anti-HBc-IgM/HBV-DNA
•Anti-HCV-IgM/HCV-RNA			•Anti-HCV-IgM/HCV-RNA
•Hepatobiliary sonography/colour Doppler sonography of liver vessels			•Hepatobiliary sonography/colour Doppler sonography of liver vessels
•Alcoholism (AST/ALT ≥2)			•Alcoholism (AST/ALT ≥2)
•Acute recent hypotension history (particularly if underlying heart disease)			•Acute recent hypotension history (particularly if underlying heart disease)
Group II			Group II
•Complications of underlying disease(s)			•Complications of underlying disease(s)
•Infection suggested by PCR and titre change for CMV (anti-CMV-IgM/IgG), EBV (anti-EBV-IgM/IgG), HSV (anti-HSV-IgM/IgG), VZV (anti-VZV-IgM/IgG)			•Infection suggested by PCR and titre change for CMV (anti-CMV-IgM/IgG), EBV (anti-EBV-IgM/IgG), HSV (anti-HSV-IgM/IgG), VZV (anti-VZV-IgM/IgG)
Evaluation of group I and II			Evaluation of group I and II
•All causes-groups I and II – reasonably ruled out	+2		•All causes-groups I and II – reasonably ruled out	+2
•The 6 causes of group I ruled out	+1		•The 6 causes of group I ruled out	+1
•5 or 4 causes of group I ruled out	0		•5 or 4 causes of group I ruled out	0
•Fewer than 4 causes of group I ruled out	−2		•Fewer than 4 causes of group I ruled out	−2
•Nondrug cause highly probable	−3		•Nondrug cause highly probable	−3
8.Previous information on hepatotoxicity of the drug			8.Previous information on hepatotoxicity of the drug
•Reaction labelled in the product characteristics	+2		•Reaction labelled in the product characteristics	+2
•Reaction published but unlabelled	+1		•Reaction published but unlabelled	+1
•Reaction unknown	0		•Reaction unknown	0
9.Response to readministration			9.Response to readministration
•Doubling of ALT with the drug alone	+3		•Doubling of ALP with the drug alone	+3
•Doubling of ALT with the drugs already given at the time of 1st reaction	+2		•Doubling of ALP with the drugs already given at the time of 1st reaction	+2
•Increase of ALT but less than N in the same conditions as for the first administration	+1		•Increase of ALP but less than N in the same conditions as for the first administration	+1
•Other situation	0		•Other situation	0
Total points			Total points

Total points/causality: ≤0, excluded; 1–2, unlikely; 3–5, possible; 6–8, probable; 8, highly probable. ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CMV, cytomegalovirus; DDS, drugs and dietary supplements; EBV, Epstein–Barr virus; HAV, hepatitis A virus; HBc, hepatitis B core; HBV, hepatitis B virus; HCV, hepatitis C virus; HSV, herpes simplex virus; VZV, varicella zoster virus.

The respective answers may be scored quantitatively with points ranging from −3 to +3. The total score shows the grade of causality: ≤ 0, causality excluded; 1–2, unlikely; 3–5, possible; 6–8, probable; > 8, highly probable [42].

Post test

The post test (Table 3), the third diagnostic step, evaluates other hepatic and extrahepatic diseases that are rare and not necessarily considered previously in the main test (Table 2). This approach is essential in the circumstances of a so far unclear diagnosis. The post test is qualitative and requires an answer yes or no (Table 3). The list of differential diagnoses presented in the post test is certainly far from complete, but may be a checklist of exclusions and useful as a reminder of various diagnoses worthy of consideration in some patients depending on their clinical characteristics. The further diagnostic approach should be based on clinical judgement, and the number of investigations should be individualized depending on the clinical context.

Table 3.

Post test

Differential diagnosis	Diagnostic evaluation/symptoms	Exclusion
1.Autoimmune hepatitis (AIH) Type I	Gamma globulins, ANA, SMA, AAA, SLA/LP, anti-LSP, anti-ASGPR	Yes/no
2.Autoimmune hepatitis (AIH) Type II	Gamma globulins, anti-LKM-1 (CYP 2D6), anti-LKM-2 (CYP 2C9), anti-LKM-3	Yes/no
3.Primary biliary cirrhosis (PBC)	AMA, anti-PDH-E2	Yes/no
4.Primary sclerosing cholangitis (PSC)	p-ANCA,MRC, ERCP	Yes/no
5.Autoimmune cholangitis (AIC)	ANA, SMA	Yes/no
6.Overlap syndromes	see 1–5	Yes/no
7.Wilson's disease	Copper excretion (24 h urine), ceruloplasmin in serum, free copper in serum, Coombs-negative haemolytic anemia, hepatic copper content, Kayser–Fleischer–Ring, neurological-psychiatric diagnostic, genotyping	Yes/no
8.Hereditary haemochromatosis	Serum ferritin, total iron-binding capacity, genotyping for C2824 and H63D mutation, hepatic iron content	Yes/no
9.Porphyria	Porphobilinogen in urine, total porphyrines in urine	Yes/no
10.α1-Antitrypsin deficiency	α1-Antitrypsin in serum	Yes/no
11.Non-alcoholic steatohepatitis (NASH)	Adipositas, insulin resistance, hepatomegaly, echogenicity of the liver	Yes/no
12.Toxic liver diseases	Toxicological assessment by occupational and household toxins	Yes/no
13.Hepatitis E	Anti-HEV-IgM, anti-HEV-IgG HEV-RNA	Yes/no
14.Other viral infections	Specific serology	Yes/no
Adenovirus, Coxsackie-B virus, echovirus, measles virus, Rubellavirus, flavivirus, arenavirus, filovirus parvovirus, HIV
15.Additional infectious diseases	Specific assessment	Yes/no
Bacteria, parasites, worms, mycosis
16.Systemic diseases	Specific assessment	Yes/no
M. Boeck, amyloidosis, lymphoma and other malignant tumours, sepsis
17.Coeliac disease	TTG antibodies, endomysium antibodies, duodenal biopsy	Yes/no
18.Addison's disease	Plasma cortisol, ACTH	Yes/no
19.Thyroid diseases	TSH basal, T4, T3	Yes/no
20.Heat stroke	Shock, hyperthermia	Yes/no
21.Grand mal seizures	Status epilepticus (duration > 30 min)	Yes/no
22.Rare intoxications	Toxin screening	Yes/no
23.Cocaine, ecstasy and other amphetamines	Toxin screening	Yes/no
24.Anorexia nervosa	Clinical context	Yes/no
25.Cardiopulmonary diseases	Cardiopulmonary assessment	Yes/no
Congestive heart disease, myocardial infarction, cardiomyopathy, cardiac valvular dysfunction, pulmonary embolism, pericardial diseases, arrhythmia by haemorrhagic shock
26.Other diseases	Clinical context	Yes/no

AAA, anti-actin antibodies; AMA, antimitochondrial antibodies; ANA, antinuclear antibodies; ASGPR, asialo-glycoprotein-receptor; CYP, cytochrome P450; HEV, hepatitis E virus; HIV, human immunodeficiency virus; LKM, liver kidney microsomes; LP, liver-pancreas antigen; LSP, liver-specific protein; MRC, magnetic resonance cholangiography; p-ANCA, perinuclear antineutrophil cytoplasmic antibodies; PDH, pyruvate dehydrogenase; SLA, soluble liver antigen; SMA, smooth muscle antibodies; TSH, thyroid-stimulating hormone; TTG, tissue transglutaminase.

RUCAM

RUCAM represents historically the first essential test trying to assess causality in drug-induced liver disease [39, 40]. Chronological and clinical criteria were the basis of the evaluation, which was merely qualitative and therefore not suitable for widespread use.

Qualitative CIOMS

The qualitative CIOMS [41] corresponds to an improved version of RUCAM [39, 40], and still lacks quantitative assessment. For the evaluation of cholestasis, either alkaline phosphatase or total bilirubin has to be considered, the serological parameters for hepatitis do not have sufficient predictive accuracy, those for CMV, EBV and HSV are only optional, and the importance of co-medication has not been properly recognized. The qualitative CIOMS should therefore no longer be applied.

CIOMS

CIOMS [42, 43] represents an update of both RUCAM [39, 40] and the qualitative CIOMS [41] and has now primarily a quantitative basis in combination with a minor separate part of qualitative items [42, 43]. The quantitative part of CIOMS is well validated [43] and in general use [53, 54], but some flexibility in applying the criteria has been recommended [10, 46]. The scores of CIOMS have been established by international experts and are based on the results of rechallenge tests [42, 43], validated with a sensitivity of 86%, a specificity of 89%, a positive predictive value of 93%, and a negative predictive value of 78% [43].

MV system

Based on a quantitative evaluation, the MV system [44] corresponds to a shortened and modified version of CIOMS [42], with major differences 44. It evaluates the time up to the normalization of ALT or ALP, considers a shorter latency period, asks for less accurate exclusion criteria of drug-independent causes, ignores concomitant drugs, underscores drugs with market presence of > 5 years without published hepatotoxicity and gives considerable weight to extrahepatic manifestations, thereby favouring hypersensitivity-type reactions. The MV system uses real and fictitious cases of liver disease, simulating DDS hepatotoxicity but with a nonpharmacological aetiology, as well as the opinion of experts as gold standard [44]. Compared with CIOMS [42], the MV system is equivalent only in cases of hypersensitivity, otherwise the CIOMS scale is superior to the MV system [45, 46]. These limitations restrict general use of the MV system.

AD system

AD causality assessment requires a liver biopsy to rule out an alternative cause of the liver disease [47]. It also uses the qualitative CIOMS [41] as well as the MV system [44]. Only when evaluation by the former scale shows drug-induced liver injury is a further assessment by the MV system performed and the grade of causality defined, provided the MV score is ≥11 points [47]. The original MV score defines 10–13 points as only a possible causality [44]. The AD system is complex and hampers evaluations.

ARD system

The ARD causality system [48] is based on the qualitative CIOMS [41], using only some of its items in modified form [48]. Reactions are considered to be drug induced when a clear temporal association to the intake of the substance is evident and probable alternative causes have been excluded. Causality is classified as not probable when the course of the reaction in its initial phase does not argue for a drug reaction and an alternative cause of the reaction can be verified. Finally, reactions are assessed as undetermined when a temporal relationship exists between intake of the substance and the reaction in the presence of a probable alternative cause or when a temporal association and an alternative cause are lacking. The causality assessment by the ARD system is therefore not without problems, since the analytical basis is only qualitative and in the evaluated patients the group with undetermined causality reachs almost one-half [48]. In a further analysis [49] incorporating also the MV scale [44], the problem of reactions with undefined classification of its causality could not be resolved [49, 51], and the additional value of the MV system [44] compared with the used qualitative CIOMS [41] has been questioned [51]. For all these reasons, the ARD system is less recommendable for general use.

TTK system

The TTK score [52] corresponds to CIOMS [42, 43] with major modifications, excluding the item of co-medication as a risk factor, implementing the lymphocyte proliferation test and including eosinophilia into the assessment system. The report requires further evaluation.

Pros, cons and caveats

Pros

There is general agreement that DDS hepatotoxicity represents a major diagnostic challenge, since surrogate markers for its diagnosis are not commonly available [10] and misdiagnosis is frequently associated with ad hoc causality assessments [2, 43–48]. To overcome this problem a diagnostic algorithm is presented which uses the pretest, main test and post test (Tables 1–3). This may add consistency to the diagnostic process by a stepwise approach, establishing the degree of probability and excluding almost all DDS-independent causes of liver disease. The questions asked are clearly defined and leave little room for discussion or uncertainty.

Whereas only a preliminary assessment may be obtained by the pretest (Table 1), the main test (Table 2) represents the crucial step in the diagnostic algorithm and provides a quantitative approach for systematic causality evaluation. It requires thorough data collection of the case by the physicians, beginning when DDS hepatotoxicity is first suspected. The main test is then being scored for its individual items and sent together with the post test (Table 3) to the manufacturers and regulatory agencies for review. When final agreement on a possible, probable or very probable causality is reached between all parties, the case will undergo further evaluation for pharmacovigilance by the health agencies and for strategic aspects by the manufacturers. The questions of major concern are signals that may emerge from this particular case alone and in connection with similar cases presented earlier. The tests should then be included in the databases of individual national health agencies and the WHO and be accessible to the health community.

When all these modalities are taken into consideration, the many general problems and shortcomings of the present procedure can be overcome. These include misdiagnosis through inappropriate causality assessment by the practising and hospital physicians [2, 48] and the regulatory agencies [6] with insufficient documented cases in databases [1, 6], differences in causality assessment of identical cases between various national drug regulatory agencies [6], the installation of expert groups that may differ in their causality assessments [22, 55], problems of interest between manufacturers and regulatory agencies [25], and insufficient causality assessment by case reports published in scientific journals [6, 56]. The main test and post test as frameworks for causality assessment in DDS hepatotoxicity are therefore of benefit for patients, the health communities, manufacturers and drug regulatory agencies, recognizing early and minimizing consecutively the risks of hepatotoxicity and ALF by DDS. Clearly, the post test (Table 3), as the third and last step in the diagnostic algorithm for causality assessment, rules out some but certainly not all other causes of liver disease and is essential in all cases of DDS hepatotoxicity, including ALF, setting out to avoid misdiagnosis. In 10–15% of all cases with ALF the causes were considered as undetermined and in 55–80% as DDS related, but only some data on causality assessments were presented [7, 18]. This raises the question, to what extent ALF was attributed to DDS merely on the basis of a temporal association and exclusion of few other diseases. Moreover, whereas in the main test (Table 2) as well as in other structured causality assessment methods [42, 44] the most frequent DDS-independent causes are more or less defined to be used for exclusion, this is not the case for other rare conditions described in the post test (Table 3). This pertains also to cardiac hepatopathies [56] that may not commonly be recognized in the setting of hepatic illness.

The importance of co-medication for DDS hepatotoxicity is taken into account by both the main test (Table 2) and the CIOMS scale [42], but not by the methods of MV [44], AD [47], ARD [49] or TTK [52]. DDS co-medication is not uncommon [6, 57] and increases substantially the risk of hepatotoxicity [4], partly due to common metabolic pathways and interactions at the cytochrome P450 isoenzyme level [34, 58].

Cons

There may be some concern that the results of the various tests (Tables 1–3) are not available in their entirety at the time when the diagnosis of DDS hepatotoxicity is first suspected. The point is well taken, but this disadvantage applies basically also to any other scoring system with items partially evaluated only in the further course of the disease. Nevertheless, preliminary causality assessment with the tests is possible using those items available at the time when DDS hepatotoxicity is first suspected.

Caveats

Caveats are well understood regarding causality assessment in some cases of ALF due to DDS using the main test (Table 2). The validity of the main test in ALF by DDS is not in question provided a concomitant fall of both serum ALT and bilirubin is observed upon discontinuation of DDS use as a sign of a resolution of hepatic injury. However, under conditions of falling ALT activities and a concomitant increase of bilirubin levels, deterioration rather than amelioration of toxic liver disease is evident [20], and causality assessment may thereby lack accuracy. In this situation the falling ALT activities in the serum reflect exhausting hepatic ALT availability, and increased serum bilirubin levels are the consequences of a diminution of liver function associated with a reduced hepatic microsomal bilirubin conjugation [20, 21]. This laboratory constellation is due to a self-perpetuating liver disease either independent of or causally related to DDS. Causality assessment with the main test (Table 2) may thus be difficult, requiring urgently the post test (Table 3) to rule out various DDS-independent causes of the observed liver disease.

Causality assessment may also be crucial when ALF is associated with liver transplantation or lethality and the temporal course of ALT or ALP is too short for adequate evaluation by the main test (Table 2), thus limiting the accuracy of the causality assessment due to a low score. A similar situation arises when steroid treatment is installed with the intention of accelerating improvement of liver function tests including ALT and ALP [10]. This may modulate the natural course of the enzyme following DDS discontinuation and excludes the enzymes from causality assessment, since no information is available of the natural course. For final assessment the post test (Table 3) is essential to consider various differential diagnoses.

Caveats are worth mentioning when causality assessment covers chronic DDS liver disease. In general, the main test (Table 2) and all other causality approaches are applicable primarily to acute rather than chronic toxic liver diseases [39, 53]. Little if anything is known about the natural course of chronic DDS hepatotoxicity, and chronic DDS liver disease may not resolve quickly and completely after DDS discontinuation [10]. Causality assessment may therefore not accurately reflect the degree of probability.

Concluding remarks

Structured causality assessment of hepatotoxicity by DDS is a major clinical challenge, since temporal associations as sole criteria for a valid evaluation are not acceptable. Initially, a clear intuition for an ad hoc evaluation is necessary, but provisional, followed by a diagnostic algorithm using the pretest, main test and post test. The evaluation is based on a variety of typical items such latency period, course of ALT and ALP after DDS discontinuation, risk factors, co-medication, previous information on hepatotoxicity of the DDS, response to rechallenge, and exclusion of other diseases. It is essential that practising and hospital physicians gather all information required for a sound causality assessment, obviating fruitless discussion by expert panels, manufacturers and health agencies in face of scanty and fragmentary data. Since pharmacogenetic alterations may be the clue to metabolic hepatotoxicity by some DDS, levels in plasma and urine should be measured to verify the diagnosis. Concomitant genotyping of cytochrome P450 and other enzymes may also be useful in future to minimize the risk of unwanted side-effects, including toxic liver disease elicited by DDS.

Competing interests

None declared.