The role of hepatitis E virus testing in drug-induced liver injury

Authors


Dr H. Dalton, Cornwall Gastrointestinal Unit, Royal Cornwall Hospital Trust, Truro TR1 3LJ, UK.
E-mail: harry.dalton@rcht.cornwall.nhs.uk

Summary

Background  Locally acquired hepatitis E is an emerging infection in developed countries and can be misdiagnosed as drug-induced liver injury.

Aim  To study the role of hepatitis E virus (HEV) testing in drug-induced liver injury.

Methods  Retrospective review of a cohort of patients with suspected drug-induced liver injury (n = 69) and hepatitis E (n = 45). The standard criteria for drug-induced liver injury were applied. Patients with suspected drug-induced liver injury who met these criteria were retrospectively tested for HEV on stored sera taken at the time of presentation. The two cohorts were compared to determine variables that predicted either of the diagnoses.

Results  Forty-seven out of 69 patients had criterion-referenced drug-induced liver injury. 22/47 were HEV negative and thus had confirmed drug-induced liver injury. 19/47 were not tested for HEV, as there was no sera available from the time of presentation. 6/47 were HEV positive and thus did not have drug-induced liver injury, but had hepatitis E infection. Compared to patients with confirmed drug-induced liver injury, patients with hepatitis E were significantly more likely to be male (OR 3.09, CI 1.05–9.08); less likely to present in November and December (0.03, CI 0.01–0.52); have lower serum bilirubin (P = 0.015); and higher serum alanine aminotransferase (P < 0.001) and alanine aminotransferase/alkaline phosphatase ratio (P < 0.001).

Conclusion  The diagnosis of drug-induced liver injury is not secure without testing for HEV.

Introduction

Drug-induced liver injury is common,1, 2 particularly in the elderly population.3, 4 It is estimated that up to 3% of all hospital admissions with jaundice are due to drug-induced liver injury.5, 6 An accurate diagnosis of drug-induced liver injury involves the application of the standard criteria for drug-induced liver injury (SCDLI).7 These criteria are based on a temporal relationship between the onset of drug therapy and biochemical evidence of liver injury; a temporal relationship between cessation of drug therapy and improvement in liver biochemistry; worsening of liver biochemistry on rechallenge; and the exclusion of alternative diagnoses (Table 1).

Table 1.   Standard criteria for drug-induced liver injury
  1. Reaction is ‘drug-related’ if all of first three criteria met, or if two of first three criteria met with positive re-challenge test.7

  2. LFTs, liver function tests.

(1) Onset of abnormal LFTs after intake of drug:
 ‘suggestive’: 5 to 90 days
 ‘compatible’: <5 days or >90 days
(2) Improvement (50% reduction) in liver function tests after cessation of drug:
 ‘very suggestive’: <8 days
 ‘suggestive’: <30 days for hepatocellular liver function tests
<180 days for cholestatic liver function tests
(3) Alternative causes excluded by detailed investigation, including liver biopsy
(4) Increase in liver function tests by 100% on re-challenge

Hepatitis E is endemic in parts of the developing world and occurs in large outbreaks in areas with poor water-related sanitation. It causes a brief icteric illness, similar to hepatitis A, and most often affects young adults.8 The prognosis of hepatitis E is generally good, except in the case of pregnant women and in patients with underlying chronic liver disease in whom the mortality rates are about 20% and up to 75% respectively.8–10

Hepatitis E in developed countries has previously been considered rare and usually seen in patients who have recently travelled to endemic areas.8 However, based on reports from developed countries throughout the world,11–21 autochthonous (locally acquired) hepatitis E is now considered an emerging infection in countries such as the UK.22 Unrecognised or subclinical infection with hepatitis E virus (HEV) is common, as we have previously shown that HEV IgG seroprevalence in blood donors in our community is 16%.23

During the course of our studies on autochthonous hepatitis E, it became clear that a significant minority of patients with hepatitis E were erroneously thought to have drug-induced liver injury, prior to the availability of HEV results.15, 20 The aim of the current paper is to study the role of HEV testing in the diagnosis of drug-induced liver injury in light of the emergence of autochthonous hepatitis E infection in developed countries. This was achieved by a retrospective study of a group of patients presenting with drug-induced liver injury and testing them for HEV. Patients with criterion-referenced drug-induced liver injury were then compared with a well-characterised cohort of patients with autochthonous hepatitis E to determine which demographic, clinical or laboratory variables predicted either of the diagnoses.

Methods

Identification of patients with drug-induced liver injury

Patients with drug-induced liver injury were identified from patients attending the Jaundice Hotline service at the Royal Cornwall Hospital, a single centre teaching hospital, between 1998 and 2006. The Jaundice Hotline24 provides a fast track referral system for patients with jaundice in Cornwall, an isolated, rural/maritime region in the southwest of England with a population of 400 000.

Any patient with jaundice in the community may be referred by their primary care physician for an out-patient assessment at the Jaundice Hotline. Such patients are seen and assessed within 72 h of referral. A detailed history, including risk factors for viral hepatitis, alcohol consumption and drug history is taken, together with a full clinical examination. All patients undergo abdominal ultrasound examination for evidence of parenchymal liver disease or biliary obstruction. Those patients with no biliary obstruction have serological tests done to screen for parenchymal liver disease with autoantibodies, immunoglobulins, iron and copper studies and serum alpha-1 antitrypsin concentration. Patients are also tested for acute and chronic viral hepatitis including hepatitis A, B, C, Epstein Barr virus (EBV) and cytomegalovirus (CMV).

All patients attending the Jaundice Hotline in whom drug-induced liver injury was the suspected diagnosis underwent retrospective case-note review. In each case the consensus criteria (SCDLI) were applied. All cases of criterion-referenced drug-induced liver injury were scored using the Roussel-Uclaf causality assessment method (RUCAM).25 Patients with criterion-referenced drug-induced liver injury were subsequently tested for HEV using sera taken at the time of presentation that had been stored at −70 °C.

Definition of drug-induced liver injury

Criterion-referenced drug-induced liver injury

The SCDLI, in the absence of histological data, was proposed by an International Consensus Meeting in 1990.7 Cases of drug-induced liver injury were identified using these criteria (Table 1).

Confirmed drug-induced liver injury

This was defined as criterion-referenced drug-induced liver injury with negative HEV serology at the time of presentation.

Identification of patients with hepatitis E

Patients with unexplained hepatitis presenting to the Royal Cornwall Hospital (Truro UK) between 1999 and 2006 were tested for HEV. Patients presenting with unexplained hepatitis to the three main hospitals in Devon UK (Plymouth, Torbay and Exeter) during 2005–6 were tested for HEV. Finally, from April 2005 to October 2005 patients presenting with unexplained hepatitis to the Auckland City Hospital, Auckland, New Zealand were also tested for hepatitis E. Each case identified was asked to complete a detailed travel questionnaire to establish whether they had travelled to an area endemic for HEV or not.

Hepatitis E testing and case definition

All sera were stored at −70 °C prior to testing. Serum anti-HEV IgM and IgG were measured using the Genelabs Diagnostics HEV ELISA kits (Genelabs Diagnostics, Singapore). All reactive sera (test:cut-off ratio >1) were tested for HEV RNA by RT-PCR as previously described.14, 26 In PCR positive cases the HEV RNA isolated from the patients was sequenced and characterised to determine genotype.14, 26

Cases were defined as follows:

  • • Biochemical evidence of hepatitis (elevated alanine aminotransferase) and
  • • Strong reactivity for anti-HEV IgM (test:cut-off ratio>5), or
  • • A rising titre of anti-HEV IgG, or
  • • Detectable viraemia by reverse transcriptase-polymerase chain reaction (RT-PCR)

Comparison of patients with drug-induced liver injury and hepatitis E

The cohort of patients with confirmed (HEV negative) drug-induced liver injury was compared with the cohort of patients with autochthonous hepatitis E to determine if any patient-related demographic factors or laboratory values predicted either of the diagnoses.

Consent/ethics

All patients described in this study consented to serological testing to elucidate the cause of their jaundice.

Results

Patients with drug-induced liver injury

In all, 1318 patients presented to the Jaundice Hotline during the study period. Of these, 69 patients had a suspected diagnosis of drug-induced liver injury after initial assessment. Following case note review and application of the SCDLI, 47 patients had criterion- referenced drug-induced liver injury. HEV testing of stored sera from presentation showed that of these 47 patients, 22 patients were HEV negative and thus had confirmed drug-induced liver injury, (mean RUCAM score = 8.7, s.d. = 1.01). Six patients were HEV positive and thus did not have drug-induced liver injury (mean RUCAM score = 8.7, s.d. = 0.82), but had hepatitis E infection. 19/47 were not tested for HEV, as there was no sera available from the time of presentation. The mean RUCAM score in these 19 patients was 8.4, (s.d. = 1.15). There was no statistical difference in the RUCAM scores between these three groups.

The different culprit drugs in the cohort of 22 patients with confirmed drug-induced liver injury were co-amoxiclav (n = 10), flucloxacillin (n = 6), amoxicillin, chlorpromazine, indoramin, mercaptopurine, Chinese herbal remedy and metamphetamine (all n = 1). The pattern of liver injury in these patients was hepatitic (n = 6), cholestatic (n = 9), and mixed (n = 7). Details of demographic and laboratory values of the drug-induced liver injury patients are shown in Table 2. Most patients had severe drug-induced liver injury, with only 2/22 being anicteric. Three patients developed complications (persistent cholestasis n = 1, liver failure and sepsis n = 2), and two patients died.

Table 2.   A comparison of the demographics, laboratory values and outcome in autochthonous hepatitis E and hepatitis E virus-negative drug-induced liver injury
 Normal range Hepatitis E (n = 45)Confirmed drug-induced liver injury (n = 22)Statistical testP value/ odds ratio
  1. Laboratory values are those at presentation. ALT/ALKP ratio is defined as the number of times the ALT was above the upper limit of normal, divided the number of times the ALKP was above the upper limit of normal. P values <0.05 were taken as significant.

  2. ALT, alanine aminotransferase; ALKP, alkaline phosphatase; INR, International Normalised Ratio; OR, odds ratio; CI, confidence intervals; M-WU, Mann-Whitney ‘U’ test.

Mean (s.d.) age63.7 (13.6)65.45 (17.1)M-WUP = 0.307
Sex (M:F ratio)3.091.00OROR = 3.09
CI = 1.05–9.08
Mean (s.d.) serum bilirubin (μmol/L)3–17118.3 (109.3)198.9 (150.3)M-WUP = 0.015
Mean (s.d.) serum ALT (IU/L) 3–351410.6 (799.4)397.8 (442.7)M-WUP <0.001
Mean (s.d.) serum ALKP (IU/L)10–105264.0 (132.0)366.6 (212.8)M-WUP = 0.028
Mean (s.d.) ALT/ALKP ratio19.5 (14.1)4.6 (6.4)M-WU<0.001
Mean (s.d.) C-reactive protein (mg/L)<521.4 (19.0)18.5 (44.7)M-WUP = 0.015
Mean (s.d.) INR0.9–1.11.2 (0.2)1.0 (0.2)M-WUP = 0.001
Mean (s.d.) WCC (×109/L)4–117.9 (2.9)7.4 (3.2)M-WUP = 0.249
Mean (s.d.) neutrophils (×109/L)2–7.54.7 (2.7)4.9 (3.2)M-WUP = 0.868
Mean (s.d.) eosinophil count (×109/L)0.04–0.440.3 (0.7)0.3 (0.2)M-WUP = 0.115
No of patients with eosinophilia (>0.5 × 109/L)23OROR = 3.00
CI = 0.46–19.50
Mean (s.d.) lymphocyte count (×109/L)1.3–3.52.2 (1.6)1.7 (0.8)M-WUP = 0.128
No of patients with lymphocytosis (>3.5 × 109/L)31OROR = 0.60
CI = 0.06–6.17
Complications63OROR = 1.00
CI = 0.23–4.44
Death32OROR = 1.40
CI = 0.22–9.05

Autochthonous hepatitis E cohort

Forty-seven patients with hepatitis E were identified during the study period. Of these, 45 had autochthonous hepatitis E, as they had no history of travel to an endemic area in the 3 months prior to presentation. All patients fulfilled the case definition for hepatitis E, 31/45 were PCR positive and all were HEV genotype 3. 41 patients with autochthonous hepatitis E were identified in Devon and Cornwall UK, with 31 cases in 2005–6. Four cases of autochthonous hepatitis E were identified in Auckland New Zealand during 2005. We have previously reported the clinical features and natural history of some of these cases.13, 15, 20, 21 Demographics and laboratory values of these patients with hepatitis E are shown in Table 2.

Briefly, patients with autochthonous hepatitis E were predominantly elderly males. 33/45 presented with jaundice, but 12 patients remained anicteric throughout their illness. By definition, none of the patients had travelled to any area endemic for HEV in the period 3 months prior to their illness. One patient worked as a butcher and had regular exposure to uncooked pig meat, five patients lived adjacent to farms and may have had exposure to farm animals (pigs n = 3, sheep and cattle n = 2). 15/45 kept domestic animals. Three patients had no symptoms at all, and were diagnosed by an incidental finding of a transaminitis on their liver blood tests taken for screening purposes. Autochthonous hepatitis E shows a seasonal variation with no cases observed in the months of November and December. 10/45 (22%) of patients with hepatitis E were initially (erroneously) thought to have drug-induced liver injury prior to the availability of results from HEV testing (Table 3).

Table 3.   Cases of autochthonous hepatitis E infection, where the diagnosis was erroneously thought to be drug-induced liver injury, prior to HEV results being available
CaseAge and sexSymptomsDrugDuration of therapy (weeks)Drug stopped at presentationBilALTAlkPTime to normal LFTs (weeks) RUCAM score*
  1. Bil, serum bilirubin (μmol/L, normal range 3–17 μmol/L); ALT, serum alanine aminotransferase (IU/L, normal range 3–35 IU/L); AlkP, serum alkaline phosphatase (IU/L normal range 10–105 IU/L); LFTs, liver function tests. All laboratory values are those at the time of presentation. Cases 1–9 were identified in Cornwall and Devon, UK. Case 10 was identified from Auckland, New Zealand.

  2. * RUCAM scores are a numerical reflection of the causal relationship between drug ingestion and liver injury: >8 = highly probable, 6–8 = probable; 3–5 = possible, 1–2 = unlikely; 0 or less = unlikely. None of the patients were re-challenged with the drug in question.

 176MJaundice, nausea, anorexiaCandersartan 3Yes2111106243 8 9
 279MNilArtorvastatin52Yes 122542145 3 8
 353MJaundice, abdominal painDiclofenac 2Yes2201478376 6 8
 464FJaundice, abdominal painCefalexin 1Yes 77 752197 310
 559MJaundice, anorexia, malaise, weight lossArtorvastatin 6Yes1601381188Died 9
 675MAnorexia, myalgia, vomitingArtorvastatin70Yes 39260233210 8
 749MItchingIbuprofen 1Yes 12 476264 2 9
 867MJaundice, anorexia, malaise, weight lossSimvastatin 8Yes124 972393 8 8
 972MJaundice, anorexia, nauseaAmoxycillin 3Yes100 421736 4 9
1085FNilSimvastatin 4Yes 14 552256 4 9

Comparison of patients with confirmed drug-induced liver injury and hepatitis E

A comparison of the demographic data and laboratory values of patients with confirmed (HEV-negative) drug-induced liver injury (n = 22) and autochthonous hepatitis E (n = 45) shows that patients with HEV were significantly more likely to be male (Table 2); less likely to present in November and December (Figure 1); have lower serum bilirubin; and higher serum alanine aminotransferase (ALT), ALT/ALKP ratio, C-reactive protein (CRP) and International Normalised Ratio (INR). There were no differences in other laboratory variables including serum eosinophil and lymphocyte count (Table 2).

Figure 1.

 Seasonal variation in cases of autochthonous hepatitis E and confirmed drug-induced liver injury. Notes: DILI = confirmed (HEV negative) drug-induced liver injury. HEV = autochthonous hepatitis E infection. The months of the year have been abbreviated to their initial letter for the sake of clarity. Compared to patients with confirmed drug-induced liver injury, patients with autochthonous hepatitis E were significantly less likely to present in the months of November or December (OR 0.03, CI = 0.01–0.52). These data are based on 41 cases of autochthonous hepatitis E from Cornwall and Devon UK. For the sake of clarity, the cases from New Zealand have been excluded, due to the differences in Seasons between the Northern and Southern hemispheres.

Conclusions

The data presented in this paper show that 6/28 (21.4%) patients with criterion-referenced drug-induced liver injury did not have drug-induced liver injury, but had autochthonous hepatitis E. Moreover, 10/45 (22.2%) patients with autochthonous hepatitis E were erroneously thought to have criterion-referenced drug-induced liver injury prior to HEV testing. Thus, at least in our population, the diagnosis of drug-induced liver injury is not secure in the absence of HEV testing.

Whilst it is not possible to reliably distinguish between drug-induced liver injury and hepatitis E on clinical grounds without testing for HEV, the data presented in the current paper illustrate a number of points to bear in mind when assessing an elderly patient with a possible drug-induced hepatic injury. Patients with hepatitis E are more likely to be male; have a lower serum bilirubin; and a higher ALT, ALT/ALKP ratio, CRP and INR. When faced with a patient with a putative diagnosis of drug-induced liver injury, these differences may prompt a clinician to consider the diagnosis of hepatitis E. Clinicians should be particularly wary of making the diagnosis of drug-induced liver injury in elderly males with a predominant transaminitis without first testing the patient for HEV.

In the UK, hepatitis E shows a seasonal variation, with peaks in March and July, and no cases in November and December. In contrast, drug-induced liver injury shows no apparent seasonal variation, nor would one be expected. Our data shows that it is significantly less likely for a patient to present with hepatitis E in November and December compared to drug-induced liver injury. The reason for this observation is unclear, but it seems likely to be related to the mode of transmission of hepatitis E.

How applicable are our observations to other communities? There is considerable evidence that genotype 3 hepatitis E is an emerging infection both throughout England and Wales and in other developed countries too. There are reports of hepatitis E from other areas of the UK.12, 14 Over the last few years, the number of documented cases of locally acquired hepatitis E in England and Wales has increased rapidly, and these cases are from a wide geographical area14 (S Ijaz, unpublished observations). Autochthonous genotype 3 hepatitis E has been described in a number of other developed countries including Japan, Holland, France, USA and New Zealand.16–20 HEV IgG seroprevalence is 4–6% in most developed countries,20, 27, 28 but in some areas of the world the rates are much higher. For example, the HEV IgG seroprevalence is up to 14% in certain areas of Japan,29 and 17–21% in blood donors in the USA.30, 31 In our community, HEV IgG seroprevalence in blood donors is 16%.23 These data suggest that autochthonous hepatitis E is frequently subclinical and/or unrecognised. Finally, in the current paper we included four cases of autochthonous hepatitis E from New Zealand. One of these cases was erroneously thought to have drug-induced hepatitis due to simvastatin prior to the availability of HEV results. Thus, it is likely that our observations are applicable to the rest of the UK and other developed countries, in addition.

The main implication of our findings to clinical practice is that the diagnosis of drug-induced liver injury is not secure without testing for HEV. This is important for two reasons. At the individual patient level, it is important to make an accurate diagnosis of drug-induced liver injury and thus avoid erroneous discontinuation of a therapy that may lead to an adverse outcome. Our findings also have implications for research in to drug-induced liver injury. It is possible that a significant minority of patients classified as having drug-induced injury in papers published in the past may not have had drug-induced liver injury at all, but in fact had hepatitis E. Thus, such research papers should be interpreted with this possibility in mind.

In summary, we have shown that 21% of patients with criterion-referenced drug-induced liver injury did not have drug-induced liver injury at all, but had autochthonous hepatitis E. 22% of patients with autochthonous hepatitis E were erroneously thought to have criterion-referenced drug-induced liver injury prior to HEV results being available. We believe that these findings are likely to be applicable to the rest of the developed world. The diagnosis of drug-induced liver injury is not secure without testing for HEV, particularly in an elderly male patient with a predominant transaminitis.

Acknowledgements

Declaration of personal interests: None. Declaration of funding interests: This work was supported by a grant from the Duchy Healthcare Charity, Truro, Cornwall, UK.

Ancillary