JCM Accepted Manuscript Posted Online 9 December 2015 J. Clin. Microbiol. doi:10.1128/JCM.02343-15 Copyright © 2015, American Society for Microbiology. All Rights Reserved.
1
2
Diagnostic performance of five assays for anti-HEV IgG
3
and IgM in a large cohort study
4 5
Heléne Norder1*, Marie Karlsson1, Åsa Mellgren2, Jan Konar3, Elisabeth Sandberg1, Anders
6
Lasson4, Maria Castedal5, Lars Magnius6, Martin Lagging1
7 8
1
9
Gothenburg, Gothenburg, Sweden
Department of Infectious Medicine, Institute of Biomedicine at Sahlgrenska Academy, University of
10
2
11
3
12
4
13
5
14
Gothenburg, Gothenburg, Sweden
15
6
Clinic of Infectious Diseases, Södra Älvsborgs Hospital, Borås Sweden
Dept. of Transfusion Medicine, Sahlgrenska University Hospital, Gothenburg, Sweden Dept. of Internal Medicine, Södra Älvsborgs Hospital, Borås Sweden Transplant Institute, Sahlgrenska University Hospital and Sahlgrenska Academy, University of
Ulf Lundahl Foundation, Stockholm, Sweden
16 17
Corresponding author: Heléne Norder, Dep of Infectious Medicine, Clinical Microbiology/Virology,
18
Guldhedsg 10B, 413 46 Gothenburg, Sweden, e-mail:
[email protected] 19 20
Key words: Hepatitis E virus, viral infection, diagnostic kits, HEV RNA
21
Running title: Diagnostic performance of anti-HEV tests
22 23 24 25
1
26
ABSTRACT
27
Determination of anti-hepatitis E virus antibodies (anti-HEV) is still enigmatic. There is no
28
gold standard, and results obtained with different assays often diverge. Herein, five assays
29
were compared for detection of anti-HEV IgM and IgG. Sera from 500 Swedish blood donors
30
and 316 patients, of whom 136 had suspected HEV infection were analysed. Concordant
31
results for IgM and IgG with all assays were obtained only for 71% and 70% of patients with
32
suspected hepatitis E. The range of sensitivity for anti-HEV detection was broad, 42% to
33
96%, which was reflected in the detection limit varying up to 19-fold for IgM and 17-fold for
34
IgG between assays. HEV RNA was analysed in all patients and in those blood donors
35
reactive for anti-HEV in any assay and was found in 26 individuals. Amongst all assays, both
36
anti-HEV IgG and IgM were detected in ten individuals, whereas twelve had only IgG, and
37
seven of those twelve only with the two most sensitive assays. Three of the HEV RNA
38
positive samples were negative for both anti-HEV IgM and IgG in all assays. With the two
39
most sensitive assays, anti-HEV IgG was identified in 16% of the blood donor samples, and
40
in 66% of patients with suspected HEV infection. Because several HEV RNA positive
41
samples had only anti-HEV IgG without anti-HEV IgM, or lacked anti-HEV antibodies,
42
analysis for HEV RNA may be warranted as a complement in the laboratory diagnosis of
43
ongoing HEV infection.
44 45 46 47 48 49 50
2
51
INTRODUCTION
52
Hepatitis E virus (HEV) is a faecal-orally transmitted virus, that, globally, is a frequent cause
53
of acute hepatitis. HEV is classified into the Hepeviridae family within the genera
54
Orthohepevirus. Orthohepevirus comprises four species, A through D, infecting mammals
55
and chickens, and Pischihepevirus infecting trout(1). Four out of seven genotypes of
56
Orthohepevirus A are known to infect man, with genotypes 1 and 2 being endemic in Asia
57
and Africa causing recurrent large outbreaks. Genotype 3 is endemic in Europe, Japan and the
58
USA. Infection by direct or indirect contact with living animals or with food products
59
contaminated with HEV is probably the most common route of infection with this genotype,
60
but blood transmissions also occur(2).
61
Hepatitis E virus infection is usually mild or asymptomatic without sequelae and less than 5%
62
of exposed individuals develop hepatitis(3-5). However, fulminant infection sporadically
63
occurs and in some patients chronic infection may ensue, often with rapid fibrosis progression
64
leading to cirrhosis; this most commonly occurs in immunocompromised individuals, such as
65
solid organ recipients and patients on chemotherapy(6-11). Additionally, hepatitis E may be
66
associated with neurological manifestations such as Guillain-Barré syndrome(12), neurologic
67
amyotrophy(13), and meningitis(14, 15).
68
Acute, chronic, and past HEV infection can be diagnosed by immunoassays for detection of
69
anti-HEV IgM and IgG in serum as well as by assays for HEV RNA (16). Despite
70
improvements of the assays, their specificities have been difficult to determine and several
71
studies have evaluated up to seven different assays with anti-HEV serum panels from
72
immunocompetent and immunocompromised patients(17-23). However, many issues remain
73
unresolved regarding the sensitivity and specificity of these assays and the confirmatory
74
immunoblot is reported unreliable(24). Not surprisingly, discordant results are reported
75
regarding anti-HEV seroprevalence(25, 26).
3
76
In the present study, the performance of five commercial assays for the detection of anti-HEV
77
IgM and IgG were compared in a clinical setting using samples from blood donors and
78
patients with liver disease.
79 80
Materials and Methods
81
Sera
82
Sera from 500 Swedish blood donors sampled at the Department of Transfusion Medicine at
83
Sahlgrenska University Hospital, Gothenburg in 2012 were evaluated. The samples were
84
anonymized with only gender and age known. Another 137 sera were derived from patients
85
with suspected hepatitis E based on negativity for markers indicative of ongoing hepatitis A,
86
B, or C, who were hospitalized or attending outpatient clinics throughout Sweden. Sera from
87
these patients were sent to the Department of Clinical Microbiology/Virology at Sahlgrenska
88
University Hospital, which is the Swedish referral laboratory for hepatitis E. Another 156
89
patients had liver disease with other aetiologies and were attending the Departments of
90
Infectious Diseases or Internal Medicine at Södra Älvsborgs Hospital in Borås. Sera from 23
91
patients, who had undergone liver transplantation at the Transplant Institute at Sahlgrenska
92
University Hospital, were also investigated. One to three additional sera from 27 patients
93
sampled one month to five years apart were also analysed. All sera were analysed at the
94
Department of Clinical Microbiology/Virology. All patients presented with liver disease, and
95
the liver transplant recipients had given their consent to participate in the study, which was
96
approved by the Regional Ethical Review Board in Gothenburg, registration number 737-12.
97 98
ELISA assays
99
Five commercial assays were used for anti-HEV IgM and IgG detection in the 851 samples.
100
The assays were selected based on routine use in the laboratory as well as availability and
4
101
prior publications(17-20). The selected assays were as follows: (1) recomWell HEV IgG/IgM,
102
(Mikrogen Diagnostik, Neuried, Germany), (2) DS-EIA-ANTI-HEV-G/M (DSI Srl, Milan,
103
Italy), (3) Anti-Hepatitis E Virus (HEV) ELISA (IgG/IgM) (Euroimmun, Lübeck, Germany),
104
(4) HEV Ab / HEV IgM EIA (Axiom Diagnostics, Worms, Germany), and (5) HEV
105
IgM/HEV IgG (DiaPro, Milan, Italy). DS-EIA corresponds to the assay produced by RPC
106
Diagnostic Systems, Nizhniy, Novogorod, Russia, and Axiom corresponds to the assay
107
produced by Wantai, Beijing, China. The formats of the assays are given in Table 1. All sera
108
were kept frozen at -20 °C until tested for anti-HEV IgM and IgG in parallel. The cutoff of
109
each assay was calculated according to the manufacturer’s instructions, and the optical density
110
(OD) value for each sample was divided by the cutoff value of the assay [sample OD/cutoff
111
OD of the assay]. All samples with OD values above or at the cutoff were reanalysed in
112
duplicate for that particular assay and the reactivity of the sample was determined by the
113
mean OD value of the reanalysed duplicates. Due to highly different numbers of anti-HEV
114
IgG reactive samples in different assays, the cutoff values were analysed by Receiver
115
Operating Characteristic (ROC) curve using GraphPad Prism version 6.04 (GraphPad
116
software Inc. CA, USA).
117
To determine the detection limit, serial two-step dilutions (from 1/2 to 1/64) of the WHO
118
reference reagent for hepatitis E virus (NIBSC, code:95/584 (27)) containing 100 IU/mL anti-
119
HEV IgG and IgM were performed in anti-HEV and HEV RNA negative serum. Two sera
120
positive for anti-HEV IgM and IgG in all assays were also diluted in negative serum. Each
121
dilution was divided into six aliquots and frozen at -20 °C, and thawed only once before being
122
analysed in duplicate. For determination of endpoint titres, linear regression was performed
123
on the Log[mean OD/cutoff OD] of each dilution vs. Log[dilution of the sample], also using
124
the GraphPad Prism program.
125
5
126
HEV RNA detection by real-time PCR
127
All blood donors reactive for anti-HEV IgG and/or IgM in any assay and all patients were
128
analysed for HEV RNA. The RNA was extracted from 250 µl serum sample mixed with 2 mL
129
lysis buffer (NucliSENS® easyMAG, bioMérieux SA, France). The mixture was incubated for
130
ten minutes at room temperature before the addition of 50 µl of NucliSENS® easyMAG
131
Magnetic Silica and incubated for an additional ten minutes. RNA was eluted in 110 µl
132
distilled water by using the NucliSENS® easyMAG instrument according to the
133
manufacturer’s instructions (bioMérieux SA, France).
134
Real-time PCR was performed on 20 µl extracted RNA in 30 µl master mix consisting of 25
135
µl 2X Reaction Mix with ROX, 1 µl SuperScript® III RT/Platinum® Taq Mix, 1 µl RNase
136
OUTTM (Invitrogen), 200 nM of each forward primer (JVHEVF), reverse primer (JVHEVR),
137
and probe (HEVP), and the cycling conditions were performed as previously described (28).
138
Statistical analysis
139
All statistical analyses were performed by using the GraphPad Prism version 6.04 (GraphPad
140
software Inc. CA, USA). Specificity and sensitivity of the assays were calculated according to
141
Baratloo et al(29).
142 143
RESULTS
144
There were considerable differences in reactivity when using recommended cutoffs at the
145
initial testing of the material. The highest reactivity was obtained using DiaPro with 49 sera
146
reactive for anti-HEV IgM and 255 for anti-HEV IgG, and the lowest being noted with
147
Euroimmun, with 16 and 79 sera reactive for anti-HEV IgM and IgG, respectively (Figure S1
148
a-j). The values log [sample OD/cutoff OD of the assay] obtained for each sample and patient
149
category from two assays was plotted on a scatter plot, with the values obtained with one
150
assay on the x-axis and the values obtained with another assay on the y-axis (Figure S1).
6
151
Because the plot of Mikrogen and Euroimmun formed a straight line below the origin, these
152
two assays appeared to have similar reactivity for each sample. However, the cutoff seemed to
153
be set too low for Euroimmun (Figure S1). Most samples with low level reactivity slightly
154
above the cutoff were obtained with DiaPro. Therefore, all assays were subject to ROC curve
155
analyses, and based on these results, an optimal cutoff value of 0.8 for Euroimmun and 1.3 for
156
DiaPro for anti-HEV IgG were found and used in the further analyses (Table S1). Otherwise,
157
the recommended cutoffs were used.
158
After adjustment of the cutoff values, reactivity in two or more assays for the same marker
159
(i.e., anti-HEV IgG or IgM) was considered a specific outcome. Hepatitis E virus infection
160
was also considered confirmed for patients with sera reactive for both IgG and IgM with
161
different assays or positive in HEV PCR.
162 163
Anti-HEV IgM
164
Analysis of the dilution series of the WHO Reference Reagent and two patient sera revealed a
165
detection limit from 5.1 to 23.8 IU/mL, with Axiom and DiaPro being the most sensitive
166
(Table 2; Figure 2). There was also an up to 19-fold difference in the detection limit for the
167
diluted patient sera, with Axiom being the most sensitive, followed by DiaPro and DSI (Table
168
2).
169
For 749 (87%) of the 816 samples, there were concordant results for anti-HEV IgM with all
170
five assays (Table S2). IgM reactivity was confirmed for 68 of the 80 IgM reactive samples
171
(85%) by anti-HEV IgM reactivity with more than one assay and/or with anti-HEV IgG or
172
HEV RNA; (Table 3). Most, 46 of these 68 confirmed anti-HEV IgM samples, were from
173
patients with suspected hepatitis E (Table 3). The sensitivity for anti-HEV IgM detection was
174
low and varied from 24% for Euroimmun to 72% for DiaPro.
175
7
176
Anti-HEV IgG
177
The detection limit for anti-HEV IgG was determined as for anti-HEV IgM and revealed up to
178
11-fold differences between the assays (from 0.2 to 2.2 IU/mL, Table 3) and revealed up to
179
17-fold differences in endpoint titres of the diluted sera with Axiom and DiaPro being most
180
sensitive (Table 2).
181
More samples were reactive for anti-HEV IgG than for IgM. Concordant results were
182
obtained for 644 (80%) of the 816 samples (Table S3). The highest concordance in anti-HEV
183
IgG reactivity was found with Axiom and DiaPro being 99% for sera from patients with
184
suspected hepatitis E and 96% for all sera.
185
There were 215 samples with confirmed anti-HEV IgG reactivity. All 22 HEV RNA positive
186
samples with anti-HEV IgG were reactive with Axiom and DiaPro (Table 2). The highest
187
number of anti-HEV IgG singly reactive samples was obtained with DiaPro, with 22 out 235
188
sera being positive in just one assay. The lowest reactivity was obtained with Euroimmun,
189
with 94 reactive samples; 91 of those were confirmed, 15 of which by the presence of HEV
190
RNA. The sensitivity for anti-HEV IgG detection was higher than that for IgM and varied
191
from 42% with Euroimmun to 98% with DiaPro (Table 3).
192 193
HEV RNA
194
Five blood donor samples and 21 patient samples had detectable HEV RNA (Table 4). Ten of
195
these 26 samples were reactive for both anti-HEV IgM and IgG in all five assays. For the
196
other 16 HEV RNA containing samples, 12 were reactive for anti-HEV IgG. Four of these
197
were reactive for anti-HEV IgG with all five assays, one with all assays except with
198
Microgen, and seven only with Axiom and DiaPro.
199 200
8
201
Blood donor sera
202
The mean age of the 500 blood donors was 42 years, 44 years for males (range 18-75) and 41
203
years for females (range 19-74). The seroprevalence of anti-HEV IgG varied from 5% to 19%
204
depending on the assay used and was 16% when confirmed by reactivity in two assays: 17%
205
for males and 13% for females. There was a significantly higher seroprevalence in donors
206
older than 50 years for both sexes, 57/170 (78%) vs. 16/330 (4%) (p