Hepatitis E virus coinfection in patients with HIV infection

Authors


Dr Harry R. Dalton, Royal Cornwall Hospital, Penventinnie Lane, Truro TR1 3LJ, UK. Tel: 0044 1872253858; fax: 0044 1872252794; email: harry.dalton@rcht.cornwall.nhs.uk

Abstract

Objectives

Hepatitis E virus (HEV) infection is an emerging infection in developed countries and is thought to be a porcine zoonosis. HEV can cause chronic infection and cirrhosis in the immunosuppressed, including patients with HIV infection. Little is known about HEV and HIV coinfection. The aim of the study was to document the incidence of chronic HEV coinfection in patients with HIV infection and to determine the anti-HEV seroprevalence and compare it with that of a control population.

Methods

A cohort/case–control study was carried out in two teaching hospitals in southwest England. A total of 138 patients with HIV infection were tested for HEV using an immunoassay for anti-HEV immunoglobulin M (IgM) and IgG and reverse transcriptase-polymerase chain reaction (RT-PCR), and 464 control subjects were tested for anti-HEV IgG. Demographic, lifestyle and laboratory data were prospectively collected on each patient with HIV infection. The anti-HEV IgG seroprevalence in patients with HIV infection was compared with that in controls and demographic risk factors for HEV exposure were explored using logistic regression models.

Results

There was no difference in anti-HEV IgG seroprevalence between the HIV-infected patients and controls. The only risk factor predictive of anti-HEV seropositivity was the consumption of raw/undercooked pork; sexual risk factors were unrelated. No patient with HIV infection had evidence of chronic coinfection with HEV

Conclusions

Anti-HEV seroprevalence is similar in controls and patients with HIV infection. Risk factor analysis suggests that HEV is unlikely to be transmitted sexually. Chronic coinfection with HEV was absent, indicating that chronic HEV/HIV coinfection is not a common problem in this cohort.

Introduction

Hepatitis E virus (HEV) is endemic in many parts of the developing world and globally it is the commonest cause of acute viral hepatitis. In developing countries, hepatitis E usually results in a self-limiting hepatitis, except in pregnant women in whom the mortality is approximately 20% [1].

Autochthonous (locally acquired) HEV infection is an emerging health issue in developed countries [1] and is thought in many cases to be a porcine zoonosis. In developed countries, acute HEV infection mainly affects the middle-aged and elderly and is more common in male individuals [2–7]. Recently, chronic HEV infection with rapidly progressive cirrhosis has been demonstrated in immunosuppressed transplant recipients [8], and in individuals with haematological malignancies [9].

In 2009, chronic HEV coinfection was documented in the UK [10] and France [11] in two HIV-infected patients, in association with established cirrhosis. However, little is currently known about the extent or outcomes of HEV and HIV coinfection. The aim of this study was to document the incidence of chronic HEV coinfection in an unselected group of patients with HIV infection and to determine the anti-HEV seroprevalence in patients with HIV infection and compare it with that of a control population.

Methods

Consecutive, unselected patients with documented HIV infection were approached to participate in the study between July 2009 and May 2010. The patients were attending the Departments of HIV Medicine at two teaching hospitals in southwest England (Royal Cornwall Hospital, Truro, and Southmead Hospital, Bristol, UK). After the patients had provided informed consent, a serum sample was taken and frozen at −70 °C, prior to being tested for HEV by reverse transcriptase-polymerase chain reaction (RT-PCR) and anti-HEV immunoglobulin G (IgG) and IgM immunoassays. Samples were also tested for hepatitis A virus (HAV) RNA by RT-PCR. Demographic, lifestyle and laboratory data were prospectively collected on each patient. These data included sexual orientation, place of birth, ethnicity, duration of HIV infection, current antiretroviral therapy (ART) and its duration, CD4 cell count, probable mode of acquisition of HIV infection and liver function tests.

After they had provided informed consent, 464 patients were used as controls. The controls were aged ≥18 years and were identified from in-patients and out-patients attending the Royal Cornwall Hospital, Truro between 2007 and 2009 and a General Practice in Cornwall, UK during 2009. These patients had a range of acute and chronic general medical conditions, but no history of liver disease. A blood sample was taken for anti-HEV IgG testing.

Samples were tested for anti-HEV IgG and IgM antibodies using the Wantai HEV IgG enzyme immunoassay (EIA) (Wantai Biological Pharmacy Enterprise, Beijing, China). This assay uses antigens encoded by a structural region of open reading frame 2 (ORF-2) from a Chinese isolate of genotype 1 HEV [12]. The assay was performed according to the manufacturer's instructions. Sera giving an absorbance greater than the cut-off value were considered to be positive for anti-HEV IgG. In addition, every HIV-infected patient's serum sample was tested for HEV RNA by real-time polymerase chain reaction with amplification within the ORF-2 region [10]. Additional HAV RNA detection was also undertaken through amplification across the VP1/2PA junction as previously described [13].

Differences in the risk of anti-HEV IgG seroprevalence between the HIV-infected patient population and the control group were assessed by fitting age/sex-adjusted logistic regression models with HEV IgG seroprevalence as the exposure variable. The shape of the relationship between age and anti-HEV IgG seroprevalence among the controls was explored by adding polynomial functions of age to logistic regression models in which HEV seroprevalence was specified as the binary dependent variable. Differential effects of age by sex were assessed through the inclusion of appropriate interaction terms in the models. Associations between anti-HEV IgG seroprevalence and risk factors in the HIV-infected population were assessed using basic age/sex-adjusted logistic regression models (with each risk factor considered in a separate model). Exposure effects were expressed as odds ratios with 95% confidence intervals. Demographic risk factors considered for inclusion in the models were whether the patient was born in an endemic area and whether the patient was of White ethnic origin (both coded as binary variables). Sexual orientation was included in the model as a dichotomous risk factor (categorized as ‘heterosexual’ and ‘homosexual or bisexual’). The following aspects of HIV history were entered as risk factors in the models: duration of HIV infection (categorized as ‘<5 years’, ‘≥5 and <10 years’ and ‘≥10 years’), current treatment with ART (binary coding), duration of therapy with ART (categorized as ‘<12 months’, ‘≥12 and <48 months’ and ‘≥48 months’), mode of acquisition of HIV {categorized as ‘men who have sex with men (MSM)’, ‘parenteral routes [injecting drug use (IDU) and blood transfusions]’, ‘high endemic area contact’ and ‘not known’}, HIV viral load (categorized as ‘≤40 HIV-1 RNA copies/mL’ and ‘>40 HIV-1 RNA copies/mL’) and CD4 count (categorized as ‘<250 cells/μL’ and ‘≥250 cells/μL’). Other risk factors assessed in the models included whether the patient had ever had contact with live pigs and whether the patient had ever eaten raw or undercooked pork (both categorized as binary variables).

All statistical analyses were conducted using the R software [14].

This study had ethical approval from the Plymouth and Cornwall Ethics Committee.

Results

A total of 138 patients with HIV infection were included in the study. Of these, 109 (79%) were male with a median age of 43 years (range 19–70 years). The demographic and laboratory variables for the study patients are shown in Table 1. It was found that 31 patients (22.5%) had abnormal liver function tests, but in most cases these were mild and only seven patients had an alanine aminotransferase (ALT) value greater than twice the upper limit of normal. The median CD4 count was 520 cells/μL and only 10 patients (7.2%) had a CD4 count of <250 cells/μL. No patients had an ALT value more than twice the upper limit of normal and a CD4 count of <250 cells/μL. Nineteen patients (13.8%) recalled contact with live pigs, and 15 (10.9%) recalled consuming undercooked or uncooked pork products in the past.

Table 1.  Demographic and laboratory data for the HIV-infected patients (n=138)
Characteristic 
  • These data are those obtained at the time at which the patient was tested for hepatitis E virus (HEV).

  • ALT, alanine aminotransferase; ART, antiretroviral therapy; HBV, hepatitis B virus; HCV, hepatitis C virus (chronic coinfection confirmed by molecular techniques); IDU, injecting drug user; MSM, men who have sex with men; ULN, upper limit of normal.

  • *

    Twenty-one patients were born in Central Africa and three in South-East Asia.

  • High endemic contact refers to patients with HIV infection who had had sex with a partner from an area considered highly endemic for HIV, such as Africa.

Age (years) [median (range)]43 (19–70)
Gender [n (%)]
 Male109 (79)
 Female29 (21)
Country of birth [n (%)]
 UK101 (73)
 Other developed country8 (6)
 Developing country endemic for HEV24 (17)*
 Developing country nonendemic for HEV5 (4)
Ethnicity [n (%)]
 White British104 (75)
 Black/British African19 (14)
 White other5 (4)
 All other ethnicities10 (7)
Sexual orientation [n (%)]
 Heterosexual61 (44)
 Bisexual3 (2)
 Homosexual68 (49)
 Not recorded6 (4)
Mode of acquisition of HIV [n (%)]
 MSM74 (54)
 IDU3 (2)
 Blood transfusion0 (0)
 High endemic contact17 (12)
 Commercial sex0 (0)
 Not known44 (32)
Duration of HIV infection (years) [median (range)]6 (0.3–20)
Taking ART [n (%)]
 Yes119 (86)
 No19 (14)
Duration of ART (months) [median (range)]40 (1–240)
CD4 count (cells/μL) [median (range)]520 (72–1465)
CD4 count<250 cells/μL (n)10
HIV viral load [n (%)]
 <40 copies/mL106 (77)
 40–400 copies/mL13 (9)
 401–10 000 copies/mL12 (9)
 >10 0000 copies/ml7 (5)
Coinfection [n (%)]
 HBV1 (1)
 HCV6 (4)
Liver function tests [n (%)]
 Abnormal31 (22)
 ALT>twice ULN7 (5)
 ALT>twice ULN and CD4 count <250 cells/μL0 (0)

None of the 138 HIV-positive patients tested had HEV or HAV RNA detected in their serum by RT-PCR. One hundred and thirty-seven of the 138 patients were anti-HEV IgM negative; the remaining sample gave an equivocal result. Thirteen of the 138 patients (9.4%) were anti-HEV IgG positive, compared with 64 of the 464 controls (13.8%). The seroprevalence of anti-HEV IgG in the control group increased with age (P<0.001) from a mean of 4% at age 20 years to 30% at age 80 years. After adjusting for age and sex, there was no difference in anti-HEV IgG seroprevalence between the HIV-infected patient population and the control group (P=0.8). Of the seven HIV-infected patients with ALT greater than twice the upper limit of normal, none was anti-HEV IgG positive.

Table 2 shows risk factor analysis for anti-HEV IgG seroprevalence in the HIV-infected population, with one risk factor tested at a time in age/sex-adjusted models. Eating raw or undercooked pork was associated with a significant increase in the risk of anti-HEV IgG seroprevalence in the HIV-infected population [odds ratio (OR) 5.45; 95% confidence interval (CI) 1.2–22.9; P=0.02], after adjustment for age and sex. The only other significant risk factor in basic adjusted models was ethnicity, with non-White patients more likely to test seropositive (OR 5.31; 95% CI 1.1–29.5; P=0.03). After fitting a multivariable model using a forward stepwise selection approach, the association between eating undercooked pork and anti-HEV IgG seroprevalence remained (P=0.02) but the effect of ethnicity was no longer significant (P=0.3).

Table 2.  Hepatitis E virus (HEV) risk factors in the HIV-infected population: a comparison of immunoglobulin G (IgG) seropositives and seronegatives
 nHEV IgG seropositiveOR95% CIP-value
Yes (%)No (%)
  • ART, antiretroviral therapy; CI, confidence interval; MSM, men who have sex with men; OR, odds ratio.

  • The following were excluded from this analysis because of missing data:

  • *

    One patient with missing ethnicity.

  • Six patients with missing sexual orientation.

  • Two patients with missing data on contact with live pigs.

  • §

    Four patients with no information on whether they had eaten undercooked pork.

  • Not estimable because of a lack of parenteral acquisition among HEV IgG-seropositive patients.

Born in HEV endemic area
 No11415.417.6   
 Yes2484.682.41.500.20–7.410.65
Ethnicity*
 White10969.280.6   
 Non-White2830.819.45.311.09–29.530.03
Sexual orientation
 Heterosexual6158.345.01.00  
 Homosexual/bisexual7141.755.00.360.09–1.340.12
Contact with live pigs
 No11784.686.21.00  
 Yes1915.413.81.060.15–4.550.95
Eaten raw/undercooked pork§
 No11966.791.01.00  
 Yes1533.39.05.451.21–22.940.02
Current treatment with ART
 No1923.112.81.00  
 Yes11976.987.20.380.09–1.920.20
Duration of therapy with ART
 <12 months5223.139.21.00  
 12–48 months3823.128.01.340.23–7.740.73
 >48 months4853.832.82.270.56–11.420.27
Duration of HIV infection
 <5 years5838.542.41.00  
 5–10 years4838.534.40.910.23–3.630.89
 >10 years3223.123.20.700.13–3.320.66
CD4 count
 ≥250 cells/μL12892.392.81.00  
 <250 cells/μL107.77.21.030.05–6.470.98
Viral load
 ≤40 copies/mL10776.977.61.00  
 >40 copies/mL3123.122.41.220.25–4.500.78
Mode of acquisition of HIV
 MSM7438.555.21.00  
 Parenteral40.03.2
 High endemic contact1715.412.02.610.30–16.60.33
 Not known4346.229.63.460.90–13.70.07

Discussion

Previous studies have shown relatively high anti-HEV IgG seroprevalence in patients with HIV infection [15,16]. The current study showed that, after accounting for age and sex, there was no difference in anti-HEV seroprevalence between patients with HIV infection and control subjects. Following multivariable analysis, the only risk factor that was identified as predictive of anti-HEV seropositivity in patients with HIV infection was consumption of raw or undercooked pork. There was no relationship between any of the sexual risk factors examined, IDU or probable mode of HIV acquisition and anti-HEV seropositivity. These data suggest that HEV is unlikely to be transmitted sexually and that the main route of infection is probably oro-faecal, possibly via consumption of infected pork meat. The consumption of raw/undercooked pork is of particular interest, as HEV genotype 3 has been isolated in retail pork products on sale to the public in a number of developed countries, including the UK [17–20].

It is likely that the seroprevalence results reported in the current study are more accurate than those previously reported as, in contrast to the previous studies, a highly sensitive assay was used. This assay has recently been validated against a large bank of acute and convalescent sera taken from patients with PCR-proven HEV genotype 3 infection [21], the predominant genotype in developed countries. However, a degree of caution needs to be used when interpreting results of any immunoglobulin assay in the context of HIV infection. HIV produces well-documented defects in T-cell immunity, but also more subtle defects in humoral immunity. Although the assay used in the current study has been validated in the immunocompetent, its accuracy in the context of HIV infection remains to be established.

To date, chronic HEV coinfection (defined as HEV viraemia persisting for more than 6 months) has been reported in two patients with HIV infection [10,11], one of whom had associated cirrhosis. The prevalence of HIV/HEV chronic coinfection is unknown. Globally this issue could be of considerable importance, as developing countries where HEV is endemic also have a high burden of HIV infection. The current study shows that, in southwest England, in an unselected group of HIV-infected patients, none of the patients tested showed evidence of HEV viraemia, thus excluding the possibility of chronic HEV coinfection. These data concur with those of a similar study from Spain [22] that found no evidence of HEV viraemia in 93 HIV-infected patients. A study from Germany also found no evidence of chronic HEV coinfection in 123 HIV-infected patients, but this study was somewhat limited, as only six patients (those who were IgG positive) were tested for HEV using molecular techniques [23]. Thus, the evidence to date indicates that chronic HEV/HIV coinfection is not a common problem, at least in Western Europe.

There are two factors that may have influenced our failure to demonstrate HEV/HIV coinfection in our cohort of patients. Firstly, compared with some regions in developing countries where HEV is endemic, southwest England has a modest anti-HEV seroprevalence. This reflects a lower incidence of circulating HEV in our community than that found in endemic areas, possibly resulting in a reduced risk of chronic coinfection with HIV. Secondly, most of our patients were receiving ART and had low HIV viral loads and most had CD4 counts >250 cells/μL. This indicates that, although they were infected with HIV, the immunosuppressive consequences in our cohort of patients were, on the whole, mitigated by effective therapy.

Chronic HEV infection occurs in the immunosuppressed, and it appears that the degree of immunosuppression is one of the key factors that determine failure of HEV clearance [8]. The two previously documented cases of chronic HIV/HEV coinfection have two important similarities [10,11]. Both patients had a low CD4 count (<200 cells/μL) and both had abnormal liver function tests (ALT more than twice the upper limit of normal). It is noteworthy that in the current study no patients had both of these characteristics. Although 50 patients in the Spanish series had a CD4 count <200 cells/μL, and 43 patients had ‘cryptogenic hepatitis’ [22], it is not clear if any patients had both. A further study is currently in progress to determine the prevalence of HIV/HEV coinfection in patients with both a low CD4 cell count and abnormal liver function tests.

In summary, anti-HEV seroprevalence was similar in controls and patients with HIV infection. Risk factor analysis suggests that HEV is unlikely to be transmitted sexually, and consumption of raw/undercooked pork was the only factor associated with HEV seropositivity. Evidence of chronic HEV coinfection was absent in 138 unselected patients with HIV infection, but none of these patients had both a CD4 count <250 cells/μL and abnormal liver function tests.

Acknowledgements

Author contributions: FK co-designed the study, collected data and reviewed the drafts; MG co-designed the study, collected data and reviewed the drafts; RB helped design the study and interpret the data and co-wrote the paper; RG and LJ entered patients into the study and reviewed the drafts; JB and GB collected the control data, collated the patient data and reviewed the drafts; NXL and WH helped design the study, performed the statistical analysis and reviewed the drafts; SLN and SI performed the virological studies and reviewed the drafts; HRD instigated the study, co-wrote the paper and is the guarantor.

Financial support: WEH was supported by funding from the National Institute for Health Research (NIHR). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

Conflicts of interest: HRD has had travel/accommodation costs and consultancy fees reimbursed by Wantai Pharmaceuticals and GlaxoSmithKline. There are no other conflicts of interest.

Ancillary