- Email: [email protected]
Evaluation of a pan-reactive hantavirus enzyme immunoassay and of a hantavirus immunoblot for the diagnosis of nephropathia epidemica
Journal of Clinical Virology 21 (2001) 63 – 74 www.elsevier.com/locate/jcv
Evaluation of a pan-reactive hantavirus enzyme immunoassay and of a hantavirus immunoblot for the diagnosis of nephropathia epidemica Jo¨rg Schubert, Franz Tollmann, Benedikt Weissbrich * Institute of Virology and Immunology, Uni6ersity of Wu¨rzburg, Versbacher Str. 7, 97078 Wu¨rzburg, Germany Received 28 August 2000; received in revised form 16 October 2000; accepted 6 December 2000
Abstract Background: Nephropathia epidemica (NE) caused by the hantavirus serotype Puumala (PUUV) is endemic in large parts of Europe. The prognosis of this disease is usually good. However, a rapid serological diagnosis is important to differentiate NE from potentially more severe renal conditions. Objecti6e: To evaluate the diagnostic usefulness of a novel pan-reactive hantavirus enzyme immunoassay (EIA) and of a novel hantavirus immunoblot (IB). Study Design: Three groups of serum samples were tested with both assays: 79 samples from 43 patients with acute NE, 27 samples from healthy adults, and 29 tricky samples from patients with autoantibodies, with acute Epstein– Barr virus (EBV) or cytomegalovirus (CMV) infections, and from pregnant women. Results: With the EIA, all but two of the early samples of the NE patients and all of the follow-up samples were positive for hantavirus IgG. All control samples were negative. The IgM EIA was positive in 42 of the 43 primary NE samples. Weak IgM EIA reactions were observed for some of the serum samples from patients with acute EBV and CMV infections. Optimal sensitivity and specificity values for the EIA were achieved when both the IgG and the IgM results were considered for the diagnosis of acute NE. All of the early NE samples reacted with the hantavirus nucleocapsid proteins in the IgG IB and all but one of these samples in the IgM IB. Cross reactions between the PUUV and the Hantaan antigens were very common. Several of the control samples did show borderline or positive bands, but these were mostly bands against only one hantavirus antigen in either the IgG or the IgM IB. The presence of at least three hantavirus bands (PUUV or HTNV) in the IgG and IgM assays was highly predictive of acute NE. Conclusion: Both assays were highly sensitive for the diagnosis of acute NE. However, the specificity of the IB IgM was only 76%. The specificity of both the IB and the EIA can be increased by modifications of the result interpretation. © 2001 Elsevier Science B.V. All rights reserved.
Abbre6iations: CMV, Cytomegalovirus; CO, Cut-off; EBV, Epstein– Barr virus; EIA, enzyme immunoassay; HTNV, Hantaan virus; HFRS, Hemorrhagic fever with renal syndrome; IB, immunoblot; IFA, immunofluorescence assay; MWP, microwell plate; N, nucleocapsid; NE, Nephropathia epidemica; PBS, phosphate buffered saline; PUUV, Puumala virus; S/CO, Signal to cut-off; SEOV, Seoul virus; SFSV, Sandfly fever Sicilian virus; SNV, Sin Nombre virus.. * Corresponding author. Tel.: + 49-931-2013962; fax: + 49-931-2013934. E-mail address: [email protected] (B. Weissbrich). 1386-6532/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 6 - 6 5 3 2 ( 0 0 ) 0 0 1 8 7 - 6
64
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Keywords: Nephropathia epidemica; Hantavirus; Diagnosis; Enzyme immunoassay; Immunoblot
1. Introduction Hantaviruses belong to the family of the Bunyaviridae and are subdivided in several serotypes which differ in their clinical manifestation and epidemiological distribution (Schmaljohn and Hjelle, 1997). The medically most important serotypes are Hantaan (HTNV), Seoul (SEOV), Dobrava, Puumala (PUUV) and Sin Nombre (SNV). The first four are known to cause hemorrhagic fever with renal syndrome (HFRS) and SNV is the etiologic agent of the hantavirus pulmonary syndrome. The serotype PUUV is endemic in large parts of Europe and is the cause of nephropathia epidemica (NE), a less severe form of HFRS. The main clinical symptoms of NE include fever, abdominal and back pain, nausea, headache, renal dysfunction and thrombocytopenia. Other organ systems may be affected as well. Thus, hantavirus infection is part of the differential diagnosis in several clinical conditions (Settergren, 1991; Mustonen et al., 1994). The prognosis of NE is usually good. However, a rapid diagnosis is important to differentiate NE from more severe renal diseases. At the onset of symptoms, IgM and IgG antibodies against hantavirus are present in most of the patients. Therefore, the diagnosis of NE is traditionally based on serological tests. Immunofluorescence assays (IFA) and enzyme immunoassays (EIA) have been used for this purpose (Niklasson and LeDuc, 1987; Settergren et al., 1987; Niklasson and Kjelsson, 1988; Hedman et al., 1991). Since the predominant immune response is directed against the hantavirus nucleocapsid (N) protein, N proteins recombinantly expressed in E. coli or baculovirus are frequently employed as EIA antigens (Kallio-Kokko et al., 1993; Lundkvist et al., 1993; Wang et al., 1993; Zo¨ller et al., 1993; Elgh et al., 1995; Vapalahti et al., 1996; Brus Sjo¨lander et al., 1997; KallioKokko et al., 1998). In order to achieve a highly sensitive detection of all medically relevant hantavirus serotypes, the use of N proteins of both PUUV-like and HTNV-like serotypes is necessary
(Zo¨ller et al., 1993). The only commercial EIA kits available for many years have been the IgG and IgM assays for PUUV and HTNV from Progen (Heidelberg, Germany). Recently, two new serological tests have become commercially available, a pan-reactive hantavirus EIA and a recombinant bunyavirus immunoblot (IB). While the EIA uses a mixture of baculovirus recombinant N proteins of the hantavirus serotypes SEOV and SNV as antigen, the bunyavirus IB is based on the N proteins of the serotypes PUUV and HTNV expressed in E. coli. Additionally, the IB contains the N protein of the Sandfly fever Sicilian virus (SFSV), another member of the Bunyaviridae. This virus is endemic mainly in the Mediterranean region and causes a febrile illness. We investigated the usefulness of the hantavirus EIA and bunyavirus IB for the diagnosis of NE. The evaluation of the SFSV N protein was not part of this study.
2. Methods
2.1. Samples Three groups of samples were tested in this study: samples from NE patients (group 1), samples from healthy controls (group 2), and tricky samples (group 3). They were selected from samples that had been sent to the diagnostic laboratory of the Institute of Virology and Immunobiology of the University of Wu¨rzburg for routine serological analysis. All samples had been stored at − 20°C until testing. Group 1 consisted of 79 samples from 43 patients with acute NE from Underfranconia in Bavaria. The median age was 37 years (range 16–65 years), the male/female ratio 3.7/1. The patients presented with the classical clinical symptoms of NE and an elevation of creatinine values. Diagnosis of acute NE was confirmed by PUUV and HTNV IgM EIA (Progen, Heidelberg, Germany) in the first available serum sample from each patient. All samples were positive for PUUV
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
65
IgM, except for two patients. One was negative in both the PUUV and the HTNV IgM EIA (Table 3, patient 1), the other one was borderline in the HTNV IgM EIA and negative in the PUUV IgM EIA (Table 3, patient 4). Of all the patients with a positive Progen PUUV IgM EIA, all except one had higher absorbance values against PUUV than against HTNV. In addition, all serums samples were positive for PUUV IgG by an in-house IFA (see below). Precise information about the disease onset was available for 27 patients. The first serum sample of these patients that was tested for hantavirus antibodies was obtained between one and 18 days after disease onset (median 7 days). One to six follow-up samples were available from 18 of the acute NE patients with a median observation period of 5 1/2 weeks (range 10 days to 4 1/2 years). Group 2 consisted of 27 serum samples of 27 healthy adults from the institute staff. Group 3 contained one sample each of eight patients with acute CMV infections, ten patients with acute EBV infections, six patients with autoantibodies, and five pregnant women. All samples of group 2 and 3 were negative for PUUV IgG by IFA. The median age in these groups was 31.5 years (range 7 – 61 years) and the male/female ratio was 1/1.3. 2.2. PUUV IgG immunofluorescence assay and a6idity assay
2.3. Hanta6irus enzyme immunoassay
PUUV IgG antibodies were tested by an inhouse IFA according to standard procedures (Niklasson and LeDuc, 1987). Vero E6 cells infected with the PUUV strain Sotkamo were used for this purpose. The serum samples were tested in a twofold serial dilution starting at dilution 1:16. PUUV IgG avidity determination was performed essentially as described by Hedman et al. (1991). Twofold serial dilutions of a serum sample were spotted in parallel on two slides. Both slides were processed similarly, except for the washing step after the serum incubation. While one of the two slides was treated with the usual three PBS washes, the other slide was rinsed for 5 min in a 8 M urea solution and then in PBS. PUUV IgG avidity was calculated as the percent ratio of IgG titers with and without urea washing step.
The bunyavirus IgG/IgM IB test kits were obtained from Mikrogen (Martinsried, Germany). The nitro-cellulose strips were coated with electrophoretically separated E. coli-expressed fulllength N antigens of PUUV, HTNV, and SFSV. The IgG and IgM assays were performed according to the instructions of the manufacturer. Briefly, the nitro-cellulose strips were incubated at room temperature for 2 h with 20 ml of serum diluted in 2 ml of sample dilution buffer. The strips were washed four times for 5 min with the supplied washing buffer, followed by 1 h incubation with peroxidase conjugated anti-human IgG or IgM. After another four washings, tetramethylbenzidine substrate was added. Colour development was allowed to proceed for approximately 10 min. The reaction was stopped by removing the substrate and three washes with deionized
The hantavirus IgG and IgM test kits were produced by MRL Diagnostics (Cypress, CA, USA) and were obtained through Mikrogen (Martinsried, Germany). The assays were performed as described in the kit manual. Briefly, serum samples and appropriate controls diluted 1:101 were incubated for 60 min in microtiterplate wells coated with a mixture of full length SEOV and SNV N proteins expressed in the baculovirus system. Subsequently, the wells were washed with the supplied washing buffer and peroxidase-conjugated anti-human IgG was added for 30 min. After another washing step, tetramethylbenzidine substrate was added to allow color development. After 10 min, the color reaction was stopped by addition of sulfuric acid and the plates were measured photometrically at 450 nm. If the validation criteria of the assay were fulfilled, an index value was calculated for each sample as the ratio of the specimen optical density and the mean value of the cut-off calibrator. Index values of \ 1.1 were considered as positive, index values of B0.9 as negative, and index values E0.9 and 0 1.1 as equivocal. This index is also referred to as signal to cut-off (S/CO) ratio.
2.4. Bunya6irus immunoblot assay
66
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
water. As recommended by the manufacturer, a positive IgG intensity control and a negative control serum were included in each run. A positive IgM control was not provided. The results of the IgG and IgM blots were recorded independently by two investigators. The band intensity of the IgG control was used as cut-off. For the serum samples, bands similar to or stronger than the bands of the IgG control were scored as positive ( + or + + ). Visible but weaker bands than the IgG control were scored as borderline (9 ).
3. Results
3.1. Results of the enzyme immunoassay To evaluate the sensitivity and specificity of the pan-reactive hantavirus EIA, all samples of groups 1–3 were tested with the IgG and IgM EIA. The S/CO ratios obtained for the three sample groups with the IgG and IgM are presented in Fig. 1. All but two of the primary samples of the NE patients and all of the followup samples were positive in the hantavirus IgG assay. All control samples from group 2 and 3
were negative. The IgM assay was positive in 42 of the 43 primary samples of the NE patients. One patient was negative in the IgM EIA (see below). The time-course of the IgM S/CO ratios of the patients for whom the precise disease onset was known is shown in Fig. 2. All serum samples obtained during the first 6 weeks after disease onset were positive. While all samples of the healthy adult control group were hantavirus IgM negative, weak IgM reactions in or just above of the equivocal zone were observed for some of the serum samples from patients with acute CMV and EBV infections. The results obtained with the EIA are summarized in Table 1. The sensitivity and specificity values for the IgG and IgM assay are shown in Table 4. Though the specificity of the IgM assay using the evaluation criteria of the manufacturer was only 92%, it increased to 100% when the IgM cut-off was raised slightly. However, the number of equivocal samples was also increased. To circumvent the specificity problem of the IgM EIA, a combination of both positive or equivocal IgG antibodies and positive or equivocal IgM antibodies was used to diagnose acute NE. With this modification of the interpretation criteria, the sensitivity
Fig. 1. Results of the hantavirus EIA. (a) IgG S/CO ratios of the primary and the follow-up samples from NE patients and of the control samples. (b) IgM S/CO ratios of the primary samples from NE patients and of the control samples. The equivocal zones are illustrated by horizontal lines.
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
67
Fig. 2. Time-course of the IgM results (S/CO) of NE patients with known disease onset. The equivocal zone is illustrated by horizontal lines.
and specificity of the EIA were 95 and 100%, respectively, and there were no patients remaining with indeterminate NE status (Table 4).
3.2. Results of the immunoblot The samples from the NE patients and the control samples were also tested with the bunyavirus IB in order to determine its sensitivity and specificity values and to compare its performance with the EIA. All samples from the NE patients reacted with various degrees of intensity with the hantavirus N proteins in the IgG and/or IgM IB assay (Table 2). Representative examples of two patients with several follow-up samples are shown in Fig. 3. According to the instructions, an IB band was regarded as positive when the intensity was equal or stronger than the supplied cut-off control. When the presence of a positive hantavirus IgG or IgM band was used as criterion to diagnose hantavirus infections, the sensitivity of the IB was very high, but the specificity was poor because several of the healthy adult and tricky control samples did also show positive bands (Table 4). However, the bands observed in the control groups were mostly single IgG or IgM bands against only one hantavirus antigen (Table
2). They were more likely to occur in the IgM IB and with the HTNV N protein. The band intensity was often only weakly positive. In contrast, the bands of the samples of the NE group were usually stronger, there was a high degree of cross reaction between the PUUV N and the HTNV N protein, and both IgG and IgM bands were usually observed in samples from the acute phase of NE. When both positive and borderline bands were considered, at least three of the four possible hantavirus bands (two each in the IgG and IgM assay) were present in almost all primary NE samples. Less than three bands were observed in the first available serum sample after disease onset of only two NE patients. Follow-up samples of these two patients revealed three and four hantavirus bands (Table 3). Using the presence of three hantavirus bands as criterion to diagnose acute NE, the sensitivity and specificity values of the IB were 98 and 100%, respectively, with only three indeterminate samples (Table 4).
3.3. Discordant results Detailed results of the samples with false positive, false negative or discordant results are presented in Table 3. One patient (Table 3, patient 1;
68
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Fig. 3. Follow-up samples of three patients with acute NE. IgG and IgM results are shown in the upper and lower row, respectively. S/CO ratios of the EIA are shown below the IB strips. Patient C corresponds to patient 1 from Table 3. The time span from the approximate disease onset (defined as day 0) is shown above the strips. d, days; w, weeks; m, months.
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Fig. 3, patient C) with acute NE was peculiar because he displayed a strong IgG response in the first sample examined after disease onset but was negative in both the MRL and the Progen IgM EIA. In the IgM IB, weak PUUV N and HTNV N bands were present only in follow-up samples. The patient had the typical clinical findings of NE with acute renal failure, abdominal pain, elevated serum creatinine, and thrombocytopenia. Acute NE was confirmed serologically in this case by demonstrating the low avidity of the PUUV IgG antibodies. In the first serum samples of patients 2 and 3, a positive IgG response was lacking in both the EIA and the IB. This was probably due to the fact that these samples were obtained early after disease onset. The IB reaction pattern of patient 4 was unusual. While the IgM response was directed only against the HTNV antigen, the IgG response was stronger with the PUUV antigen. The routinely performed Progen IgM EIA was borderline and negative for HTNV IgM in the first and second serum sample of this patient and was negative for PUUV IgM in both samples. This patient may have been infected with a hantavirus serotype other than PUUV. The control patients 5 – 12 in Table 3 displayed weak unspecific reactions in the hantavirus IgM EIA. Most of them were also positive for HTNV
69
IgM in the IB. Patients 13 and 14 were the only control patients reacting with two bands. Several other control patients not shown in Table 3 had single positive bands in the IB, but one of the control patients reacted with three or four bands.
4. Discussion A correct and rapid serological confirmation of cases with suspected acute NE is desirable, because the differential diagnosis includes disorders such as rapid progressive glomerulonephritis which must be treated with immunosuppressive drugs as soon as possible. Furthermore, a timely diagnosis of hantavirus infections may help to avoid renal biopsies. We therefore evaluated the performance of a new EIA and a new IB to diagnose or exclude acute NE. The sensitivity of both assays was high. The specificity of the IgM EIA and of the IB was not optimal but could be increased by modifications of the result interpretation. Thus, both assay formats appear to be suitable for use in routine diagnosis. While the pan-reactive hantavirus IgG EIA was 100% specific for the detection of hantavirus IgG, weak unspecific reactions were observed in the IgM EIA when testing serum samples from patients with acute EBV or CMV infections. EBV infections are associated with a polyclonal B-cell
Table 1 Results of the hantavirus enzyme immunoassay Hantavirus IgG EIA Samples (n)
Negative
Acute NEb (43), group 1 1c Healthy adults (27), group 2 27 Tricky samples (29), group 3 29 Tricky samples, subgroups: Acute CMV (8) 8 Acute EBV (10) 10 Autoimmune (6) 6 Pregnant (5) 5 a
Hantavirus IgM EIA
Equivocal
Positive
Negative
Equivocal
Positive
1c – –
41 – –
1c 27 21
– – 4c
42 – 4c
41 – –
– – – –
– – – –
5 5 6 5
2 2 – –
1 3 – –
– – – –
Both IgG and IgM positive or equivocal. Only the results of the first sample after disease onset of each patient are included. c For details see Table 3. b
IgG and IgM Positive or equivocala
P−/9 H+/++
b
a
P+/++ H+/++
6 2 1 – – 1 –
– 25 26 8 10 4 4
– – – –
8 – – – – 1 1
– – 2 – – – –
29 – – 3 5 3 3
1 21 14 2 1 1 2
6 4 6
P or H 9
– – – –
5 – –
P+/++ H−/9
P− H−
P+/++ H−/9
P− H−a
P or H 9
Hantavirus IgM blot Number of samples reacting with
Hantavirus IgG blot Number of samples reacting with
4 P: Puumala, H: Hantaan, -/ 9/+/++: band intensities (see Section 2). Only the results of the first sample after disease onset of each patient are included. c For details see Table 3. d 9 , + and ++ bands are included.
4
Acute NEb (43), group 1 Healthy adults (27), group 2 Tricky samples (29), group 3 Tricky samples, subgroups: Acute CMV (8) Acute EBV (10) Autoimmune (6) Pregnant (5)
Samples (n)
Table 2 Results of the hantavirus immunoblot
3 4 2 –
6 2 9
P−/9 H+/++
– – – –
25 – –
P+/++ H+/++
3 5 3 2
– 19 13
0 band
5 5 1 3
1c 8 14
1 band
– – 2 –
1c – 2c
2 bands
Hantavirus IgG and IgM blot Number of samples reacting withd
– – – –
41d – –
E2 bands
70 J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
71
Table 3 Detailed results of patients with indeterminate or discordant resultsa EIA
IB
No.
Sex
Age
Classification
IgG S/CO
IgM S/CO
IgG PUUV
IgG HTNV
IgM PUUV
IgM HTNV
Total bands
PUUV IgG IFTb
1
M
24
2
M
27
3 4
F M
45 55
5 6 7 8 9 10 11 12 13 14
M M F F M M F F F F
38 19 7 29 46 35 17 19 42 50
Acute NE (ca. d7c) Follow-up (d10) Follow-up (d13) Follow-up (d20) Acute NE (d 2) Follow-up (d 11) Acute NE (d 5) Acute NE (d 17) Follow-up (d 35) Follow-up (d 57) Acute CMV infection Acute EBV infection Acute EBV infection Acute EBV infection Acute CMV infection Acute CMV infection Acute EBV infection Acute EBV infection Autoantibodies (ANAe) Autoantibodies (GBMe)
6.23 7.71 9.98 8.98 0.40 2.60 0.91 2.10 2.90 4.50 0.57 0.60 0.23 0.40 0.14 0.20 0.34 0.22 0.31 0.44
0.55 0.78 1.03 0.67 2.79 \8.75 1.97 6.32 3.86 3.01 1.50 1.27 1.38 1.50 1.02 0.98 1.03 1.10 0.49 0.25
++ ++ ++ ++ 9 ++ 9 + ++ ++ − − − − − − − − − 9
− − 9 9 − + − − + ++ − − − − − − − − + −
− 9 9 9 + ++ 9 − − − − − − − − − − − − −
− 9 9 9 + ++ 9 ++ ++ ++ + − + − 9 + + + + 9
1 3 4 4 3 4 3 2 3 3 1 0 1 0 1 1 1 1 2 2
2048 2048 1024 10,240d 64 128 256 64 n.d. \512 B16 B16 B16 B16 B16 B16 B16 B16 B16 B16
a S/CO of the MRL EIA and the band intensities of the Mikrogen IB (−/9 /+/++, see Section 2) are listed for the following patients: Acute NE patients with a negative IgM EIA (1) or a negative (2) or indeterminate (3) IgG EIA, or less than three IB bands (4). Control patients with a positive (5–8) or indeterminate (9–12) IgM EIA or more than one IB band (13, 14). The indeterminate or discordant results are shown in bold. b Reciprocal titers. c Time span after disease onset for the patients with acute NE. The exact disease onset was unknown for patient 1. It was estimated to be approximately 7 days before the first sample was obtained. d, days. d PUUV IgG avidity: 1%. e ANA, anti-nuclear antibodies; GBM, glomerular basement membrane antibodies.
stimulation which may cause or contribute to the unspecific IgM reactions (Alpers et al., 1994). There are two ways to deal with this problem. Firstly, it appears from our data, that IgM results above 1.5 S/CO are always specific, i. e. the stronger the IgM reaction the higher the likelihood of a specific result. Secondly, 93% of the patients with acute NE were positive for both IgM and IgG in their first serum sample. This is in accordance with the results reported in several other studies (Elgh et al., 1998; Kallio-Kokko et al., 1998). Therefore, it is recommended that hantavirus IgM and IgG are always tested in parallel. When the diagnosis of acute NE was based on a positive IgM EIA and a positive IgG EIA, the
specificity was 100% and there were no indeterminate samples while there was only little loss in sensitivity. Unless the analysed sample was drawn very early after disease onset, the diagnosis of acute NE is unlikely with the constellation, ‘‘IgM positive/IgG negative’’. A second serum sample should be analysed in doubtful cases. When NE is clinically suspected, a rapid diagnosis is important for patient management. On the other hand, though NE is endemic in large parts of Europe, the overall incidence and the number of samples to be tested is low in many of these areas. As a consequence, test runs often consist of only one patient sample. In this situation, the required number of three controls and
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
72
like infections based on band intensities would require the use of less cross reactive antigens, such as the truncated amino-terminal portions of the N proteins or N proteins with deleted aminotermini (Elgh et al., 1997; Morii et al., 1998). The expression systems used for the production of the hantavirus N proteins differed for the EIA and the IB. While the EIA was based on baculo recombinants, E. coli-expressed proteins where used for the IB. Because of the higher rate of unspecific reactions observed in the IB, baculo recombinants may be more suitable for the production of hantavirus N proteins for serological assays. Furthermore, it has been reported that the PUUV N protein expressed in E. coli lacks antigenic sites compared to the native and to baculoexpressed PUUV N proteins (Vapalahti et al., 1996). The suggested modifications of the interpretation criteria for both the EIA and the IB rely on the consideration of the IgG results in addition to the IgM results. Since the positive predictive value of hantavirus IgG tests depends on the overall prevalence of hantavirus infections, the strategy of basing the NE diagnosis on both IgM and IgG tests is less appropriate for areas where the prevalence is high. Further assessing the specificity of
three calibrators in each IgM and IgG EIA run leads to a considerable increase in the costs per reported EIA result. It is desirable that future versions of the EIA will require less controls and calibrators per run. Similar to the hantavirus EIA, the IB was highly sensitive for the diagnosis of acute NE, but there were some problems with unspecific results. The specificity was considerably increased when the interpretation criteria were changed to require the presence of at least three of the four possible hantavirus bands (PUUV IgG, HTNV IgG, PUUV IgM, HTNV IgM) for the diagnosis of acute NE. This recommendation is based on two observations. Firstly, as with the EIA, hantavirus IgG was detected in most NE patients by the time when a serological diagnosis was sought. Secondly, there was extensive cross reaction between the PUUV N and HTNV N proteins of the IB. Highly cross-reacting antibody responses in the acute phase of HFRS have been commonly observed with the use of full length N proteins as in the present assay format (Lee et al., 1979; Svedmyr et al., 1980; Zo¨ller et al., 1993; Elgh et al., 1998). While the cross reaction can be exploited to distinguish between specific and unspecific bands, a differentiation between PUUV-like and HTNV-
Table 4 Sensitivity and specificity values of the MRL EIA and the Mikrogen IB for the diagnosis of acute NE depending on the interpretation criteria Sensitivitya (%)
Interpretation criteria Enzyme immunoassay IgG, IgM, IgM, IgG/IgM, Immunoblot IgG, IgM, IgG/IgM
a
pos.: pos.: pos.: neg.: pos.: neg.:
\1.1 S/COc \1.1 S/COc \1.5 S/CO B0.9 S/CO both E0.9 S/CO IgG or IgM B0.9 S/CO
pos.: pos.: pos.: neg.:
E1 E1 E3 01
band +/++c band +/++c bands 9 /+/++ band 9 /+/++
Indeterminateb (n)
98 98 98
100 92 100
1 4 11
95
100
–
100 97 98
96 76 100
9 16 3
For the NE patients, only the results of the first samples were considered. These samples were excluded from the analysis. c Interpretation criteria of the manufacturer (see Section 2). b
Specificity (%)
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
the testing strategy in these areas will require the inclusion of old-immunity serum samples (i. e. hantavirus IgG positive and IgM negative) in the control groups. While an evaluation of the SFSV N protein of the IB was not part of this study, the usefulness of the concept of combining serological assays for hantavirus and sandfly fever virus infections is doubtful. Because the clinical presentation of both infections is different, a situation will be exceedingly rare when testing one serum sample for both infections is required. Instead, positive SFSV bands in IB assays performed for hantavirus diagnosis and vice versa may lead to interpretation difficulties. Overall, with some modifications of the interpretation criteria to overcome specificity problems, especially of the IB IgM, both the EIA and the IB are suitable for the diagnosis of acute NE. Choosing one of these assay formats for routine diagnostic will probably depend on factors such as assay costs, the involved labour, the number of samples, the potential for automation, and the available experience. If the IB were changed to include antigens with minimal cross-reaction between the different hantavirus serotypes, a sequential testing strategy of a pan-reactive EIA as first line assay and an IB for serotyping of positive EIA results would become feasible.
Acknowledgements We thank Bernd Kra¨mer, Mikrogen, for providing the test kits and for helpful discussions. We also thank Thomas Kerkau for providing the serum samples with autoantibodies.
References Alpers C, Scheid C, Bauer G. Acute EBV infection results in IgM responses against unrelated viruses. Med Microbiol Letters 1994;3:306– 14. Brus Sjo¨lander K, Elgh F, Kallio-Kokko H, Vapalahti O, Hagglund M, Palmcrantz V, Juto P, Vaheri A, Niklasson B, Lundkvist A, . Evaluation of serological methods for diagnosis of Puumala hantavirus infection (nephropathia epidemica). J Clin Microbiol 1997;35:3264–8.
73
Elgh F, Linderholm M, Wadell G, Tarnvik A, Juto P. Development of humoral cross-reactivity to the nucleocapsid protein of heterologous hantaviruses in nephropathia epidemica. FEMS Immunol Med Microbiol 1998;22:309– 15. Elgh F, Lundkvist A, , Alexeyev OA, Stenlund H, Avsic-Zupanc T, Hjelle B, Lee HW, Smith KJ, Vainionpaa R, Wiger D, Wadell G, Juto P. Serological diagnosis of hantavirus infections by an enzyme-linked immunosorbent assay based on detection of immunoglobulin G and M responses to recombinant nucleocapsid proteins of five viral serotypes. J Clin Microbiol 1997;35:1122– 30. Elgh F, Wadell G, Juto P. Comparison of the kinetics of Puumala virus specific IgM and IgG antibody responses in nephropathia epidemica as measured by a recombinant antigen-based enzyme-linked immunosorbent assay and an immunofluorescence test. J Med Virol 1995;45:146– 50. Hedman K, Vaheri A, Brummer-Korvenkontio M. Rapid diagnosis of hantavirus disease with an IgG-avidity assay. Lancet 1991;338:1353– 6. Kallio-Kokko H, Vapalahti O, Hedman K, Brummer-Korvenkontio M, Vaheri A. Puumala virus antibody and immunoglobulin G avidity assays based on a recombinant nucleocapsid antigen. J Clin Microbiol 1993;31:677– 80. Kallio-Kokko H, Vapalahti O, Lundkvist A, , Vaheri A. Evaluation of Puumala virus IgG and IgM enzyme immunoassays based on recombinant baculovirus-expressed nucleocapsid protein for early nephropathia epidemica diagnosis. Clin Diagn Virol 1998;10:83– 90. Lee HW, Lee PW, Lahdevirta J, Brummer-Korventkontio M. Aetiological relation between Korean haemorrhagic fever and nephropathia epidemica. Lancet 1979;1:186– 7. Lundkvist A, , Ho¨rling J, Niklasson B. The humoral response to Puumala virus infection (nephropathia epidemica) investigated by viral protein specific immunoassays. Arch Virol 1993;130:121– 30. Morii M, Yoshimatsu K, Arikawa J, Zhou G, Kariwa H, Takashima I. Antigenic characterization of Hantaan and Seoul virus nucleocapsid proteins expressed by recombinant baculovirus: application of a truncated protein, lacking an antigenic region common to the two viruses, as a serotyping antigen. J Clin Microbiol 1998;36:2514– 21. Mustonen J, Brummer-Korvenkontio M, Hedman K, Pasternack A, Pietila K, Vaheri A. Nephropathia epidemica in Finland: a retrospective study of 126 cases. Scand J Infect Dis 1994;26:7– 13. Niklasson B, Kjelsson T. Detection of nephropathia epidemica (Puumala virus)-specific immunoglobulin M by enzymelinked immunosorbent assay. J Clin Microbiol 1988;26:1519– 23. Niklasson B, LeDuc JW. Epidemiology of nephropathia epidemica in Sweden. J Infect Dis 1987;155:269– 76. Schmaljohn C, Hjelle B. Hantaviruses: a global disease problem. Emerg Infect Dis 1997;3:95– 104. Settergren B. Nephropathia epidemica (hemorrhagic fever with renal syndrome) in Scandinavia. Rev Infect Dis 1991;13:736– 44.
74
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Settergren B, Juto P, Wadell G. Detection of specific serum immunoglobulin M in nephropathia epidemica (Scandinavian epidemic nephropathy) by a biotin-avidin-amplified immunofluorescence method. J Clin Microbiol 1987;25:1134– 6. Svedmyr A, Lee PW, Gajdusek DC, Gibbs CJ, Jr, Nystrom K. Antigenic differentiation of the viruses causing Korean haemorrhagic fever and epidemic (endemic) nephropathy of Scandinavia. Lancet 1980;2:315–6. Vapalahti O, Lundkvist A, , Kallio-Kokko H, Paukku K, Julkunen I, Lankinen H, Vaheri A. Antigenic properties and
diagnostic potential of puumala virus nucleocapsid protein expressed in insect cells. J Clin Microbiol 1996;34:119– 25. Wang M, Rossi C, Schmaljohn CS. Expression of non-conserved regions of the S genome segments of three hantaviruses: evaluation of the expressed polypeptides for diagnosis of haemorrhagic fever with renal syndrome. J Gen Virol 1993;74:1115– 24. Zo¨ller L, Yang S, Go¨tt P, Bautz EK, Darai G. Use of recombinant nucleocapsid proteins of the Hantaan and nephropathia epidemica serotypes of Hantaviruses as immunodiagnostic antigens. J Med Virol 1993;39:200– 7.
.
Evaluation of a pan-reactive hantavirus enzyme immunoassay and of a hantavirus immunoblot for the diagnosis of nephropathia epidemica Jo¨rg Schubert, Franz Tollmann, Benedikt Weissbrich * Institute of Virology and Immunology, Uni6ersity of Wu¨rzburg, Versbacher Str. 7, 97078 Wu¨rzburg, Germany Received 28 August 2000; received in revised form 16 October 2000; accepted 6 December 2000
Abstract Background: Nephropathia epidemica (NE) caused by the hantavirus serotype Puumala (PUUV) is endemic in large parts of Europe. The prognosis of this disease is usually good. However, a rapid serological diagnosis is important to differentiate NE from potentially more severe renal conditions. Objecti6e: To evaluate the diagnostic usefulness of a novel pan-reactive hantavirus enzyme immunoassay (EIA) and of a novel hantavirus immunoblot (IB). Study Design: Three groups of serum samples were tested with both assays: 79 samples from 43 patients with acute NE, 27 samples from healthy adults, and 29 tricky samples from patients with autoantibodies, with acute Epstein– Barr virus (EBV) or cytomegalovirus (CMV) infections, and from pregnant women. Results: With the EIA, all but two of the early samples of the NE patients and all of the follow-up samples were positive for hantavirus IgG. All control samples were negative. The IgM EIA was positive in 42 of the 43 primary NE samples. Weak IgM EIA reactions were observed for some of the serum samples from patients with acute EBV and CMV infections. Optimal sensitivity and specificity values for the EIA were achieved when both the IgG and the IgM results were considered for the diagnosis of acute NE. All of the early NE samples reacted with the hantavirus nucleocapsid proteins in the IgG IB and all but one of these samples in the IgM IB. Cross reactions between the PUUV and the Hantaan antigens were very common. Several of the control samples did show borderline or positive bands, but these were mostly bands against only one hantavirus antigen in either the IgG or the IgM IB. The presence of at least three hantavirus bands (PUUV or HTNV) in the IgG and IgM assays was highly predictive of acute NE. Conclusion: Both assays were highly sensitive for the diagnosis of acute NE. However, the specificity of the IB IgM was only 76%. The specificity of both the IB and the EIA can be increased by modifications of the result interpretation. © 2001 Elsevier Science B.V. All rights reserved.
Abbre6iations: CMV, Cytomegalovirus; CO, Cut-off; EBV, Epstein– Barr virus; EIA, enzyme immunoassay; HTNV, Hantaan virus; HFRS, Hemorrhagic fever with renal syndrome; IB, immunoblot; IFA, immunofluorescence assay; MWP, microwell plate; N, nucleocapsid; NE, Nephropathia epidemica; PBS, phosphate buffered saline; PUUV, Puumala virus; S/CO, Signal to cut-off; SEOV, Seoul virus; SFSV, Sandfly fever Sicilian virus; SNV, Sin Nombre virus.. * Corresponding author. Tel.: + 49-931-2013962; fax: + 49-931-2013934. E-mail address: [email protected] (B. Weissbrich). 1386-6532/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 6 - 6 5 3 2 ( 0 0 ) 0 0 1 8 7 - 6
64
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Keywords: Nephropathia epidemica; Hantavirus; Diagnosis; Enzyme immunoassay; Immunoblot
1. Introduction Hantaviruses belong to the family of the Bunyaviridae and are subdivided in several serotypes which differ in their clinical manifestation and epidemiological distribution (Schmaljohn and Hjelle, 1997). The medically most important serotypes are Hantaan (HTNV), Seoul (SEOV), Dobrava, Puumala (PUUV) and Sin Nombre (SNV). The first four are known to cause hemorrhagic fever with renal syndrome (HFRS) and SNV is the etiologic agent of the hantavirus pulmonary syndrome. The serotype PUUV is endemic in large parts of Europe and is the cause of nephropathia epidemica (NE), a less severe form of HFRS. The main clinical symptoms of NE include fever, abdominal and back pain, nausea, headache, renal dysfunction and thrombocytopenia. Other organ systems may be affected as well. Thus, hantavirus infection is part of the differential diagnosis in several clinical conditions (Settergren, 1991; Mustonen et al., 1994). The prognosis of NE is usually good. However, a rapid diagnosis is important to differentiate NE from more severe renal diseases. At the onset of symptoms, IgM and IgG antibodies against hantavirus are present in most of the patients. Therefore, the diagnosis of NE is traditionally based on serological tests. Immunofluorescence assays (IFA) and enzyme immunoassays (EIA) have been used for this purpose (Niklasson and LeDuc, 1987; Settergren et al., 1987; Niklasson and Kjelsson, 1988; Hedman et al., 1991). Since the predominant immune response is directed against the hantavirus nucleocapsid (N) protein, N proteins recombinantly expressed in E. coli or baculovirus are frequently employed as EIA antigens (Kallio-Kokko et al., 1993; Lundkvist et al., 1993; Wang et al., 1993; Zo¨ller et al., 1993; Elgh et al., 1995; Vapalahti et al., 1996; Brus Sjo¨lander et al., 1997; KallioKokko et al., 1998). In order to achieve a highly sensitive detection of all medically relevant hantavirus serotypes, the use of N proteins of both PUUV-like and HTNV-like serotypes is necessary
(Zo¨ller et al., 1993). The only commercial EIA kits available for many years have been the IgG and IgM assays for PUUV and HTNV from Progen (Heidelberg, Germany). Recently, two new serological tests have become commercially available, a pan-reactive hantavirus EIA and a recombinant bunyavirus immunoblot (IB). While the EIA uses a mixture of baculovirus recombinant N proteins of the hantavirus serotypes SEOV and SNV as antigen, the bunyavirus IB is based on the N proteins of the serotypes PUUV and HTNV expressed in E. coli. Additionally, the IB contains the N protein of the Sandfly fever Sicilian virus (SFSV), another member of the Bunyaviridae. This virus is endemic mainly in the Mediterranean region and causes a febrile illness. We investigated the usefulness of the hantavirus EIA and bunyavirus IB for the diagnosis of NE. The evaluation of the SFSV N protein was not part of this study.
2. Methods
2.1. Samples Three groups of samples were tested in this study: samples from NE patients (group 1), samples from healthy controls (group 2), and tricky samples (group 3). They were selected from samples that had been sent to the diagnostic laboratory of the Institute of Virology and Immunobiology of the University of Wu¨rzburg for routine serological analysis. All samples had been stored at − 20°C until testing. Group 1 consisted of 79 samples from 43 patients with acute NE from Underfranconia in Bavaria. The median age was 37 years (range 16–65 years), the male/female ratio 3.7/1. The patients presented with the classical clinical symptoms of NE and an elevation of creatinine values. Diagnosis of acute NE was confirmed by PUUV and HTNV IgM EIA (Progen, Heidelberg, Germany) in the first available serum sample from each patient. All samples were positive for PUUV
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
65
IgM, except for two patients. One was negative in both the PUUV and the HTNV IgM EIA (Table 3, patient 1), the other one was borderline in the HTNV IgM EIA and negative in the PUUV IgM EIA (Table 3, patient 4). Of all the patients with a positive Progen PUUV IgM EIA, all except one had higher absorbance values against PUUV than against HTNV. In addition, all serums samples were positive for PUUV IgG by an in-house IFA (see below). Precise information about the disease onset was available for 27 patients. The first serum sample of these patients that was tested for hantavirus antibodies was obtained between one and 18 days after disease onset (median 7 days). One to six follow-up samples were available from 18 of the acute NE patients with a median observation period of 5 1/2 weeks (range 10 days to 4 1/2 years). Group 2 consisted of 27 serum samples of 27 healthy adults from the institute staff. Group 3 contained one sample each of eight patients with acute CMV infections, ten patients with acute EBV infections, six patients with autoantibodies, and five pregnant women. All samples of group 2 and 3 were negative for PUUV IgG by IFA. The median age in these groups was 31.5 years (range 7 – 61 years) and the male/female ratio was 1/1.3. 2.2. PUUV IgG immunofluorescence assay and a6idity assay
2.3. Hanta6irus enzyme immunoassay
PUUV IgG antibodies were tested by an inhouse IFA according to standard procedures (Niklasson and LeDuc, 1987). Vero E6 cells infected with the PUUV strain Sotkamo were used for this purpose. The serum samples were tested in a twofold serial dilution starting at dilution 1:16. PUUV IgG avidity determination was performed essentially as described by Hedman et al. (1991). Twofold serial dilutions of a serum sample were spotted in parallel on two slides. Both slides were processed similarly, except for the washing step after the serum incubation. While one of the two slides was treated with the usual three PBS washes, the other slide was rinsed for 5 min in a 8 M urea solution and then in PBS. PUUV IgG avidity was calculated as the percent ratio of IgG titers with and without urea washing step.
The bunyavirus IgG/IgM IB test kits were obtained from Mikrogen (Martinsried, Germany). The nitro-cellulose strips were coated with electrophoretically separated E. coli-expressed fulllength N antigens of PUUV, HTNV, and SFSV. The IgG and IgM assays were performed according to the instructions of the manufacturer. Briefly, the nitro-cellulose strips were incubated at room temperature for 2 h with 20 ml of serum diluted in 2 ml of sample dilution buffer. The strips were washed four times for 5 min with the supplied washing buffer, followed by 1 h incubation with peroxidase conjugated anti-human IgG or IgM. After another four washings, tetramethylbenzidine substrate was added. Colour development was allowed to proceed for approximately 10 min. The reaction was stopped by removing the substrate and three washes with deionized
The hantavirus IgG and IgM test kits were produced by MRL Diagnostics (Cypress, CA, USA) and were obtained through Mikrogen (Martinsried, Germany). The assays were performed as described in the kit manual. Briefly, serum samples and appropriate controls diluted 1:101 were incubated for 60 min in microtiterplate wells coated with a mixture of full length SEOV and SNV N proteins expressed in the baculovirus system. Subsequently, the wells were washed with the supplied washing buffer and peroxidase-conjugated anti-human IgG was added for 30 min. After another washing step, tetramethylbenzidine substrate was added to allow color development. After 10 min, the color reaction was stopped by addition of sulfuric acid and the plates were measured photometrically at 450 nm. If the validation criteria of the assay were fulfilled, an index value was calculated for each sample as the ratio of the specimen optical density and the mean value of the cut-off calibrator. Index values of \ 1.1 were considered as positive, index values of B0.9 as negative, and index values E0.9 and 0 1.1 as equivocal. This index is also referred to as signal to cut-off (S/CO) ratio.
2.4. Bunya6irus immunoblot assay
66
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
water. As recommended by the manufacturer, a positive IgG intensity control and a negative control serum were included in each run. A positive IgM control was not provided. The results of the IgG and IgM blots were recorded independently by two investigators. The band intensity of the IgG control was used as cut-off. For the serum samples, bands similar to or stronger than the bands of the IgG control were scored as positive ( + or + + ). Visible but weaker bands than the IgG control were scored as borderline (9 ).
3. Results
3.1. Results of the enzyme immunoassay To evaluate the sensitivity and specificity of the pan-reactive hantavirus EIA, all samples of groups 1–3 were tested with the IgG and IgM EIA. The S/CO ratios obtained for the three sample groups with the IgG and IgM are presented in Fig. 1. All but two of the primary samples of the NE patients and all of the followup samples were positive in the hantavirus IgG assay. All control samples from group 2 and 3
were negative. The IgM assay was positive in 42 of the 43 primary samples of the NE patients. One patient was negative in the IgM EIA (see below). The time-course of the IgM S/CO ratios of the patients for whom the precise disease onset was known is shown in Fig. 2. All serum samples obtained during the first 6 weeks after disease onset were positive. While all samples of the healthy adult control group were hantavirus IgM negative, weak IgM reactions in or just above of the equivocal zone were observed for some of the serum samples from patients with acute CMV and EBV infections. The results obtained with the EIA are summarized in Table 1. The sensitivity and specificity values for the IgG and IgM assay are shown in Table 4. Though the specificity of the IgM assay using the evaluation criteria of the manufacturer was only 92%, it increased to 100% when the IgM cut-off was raised slightly. However, the number of equivocal samples was also increased. To circumvent the specificity problem of the IgM EIA, a combination of both positive or equivocal IgG antibodies and positive or equivocal IgM antibodies was used to diagnose acute NE. With this modification of the interpretation criteria, the sensitivity
Fig. 1. Results of the hantavirus EIA. (a) IgG S/CO ratios of the primary and the follow-up samples from NE patients and of the control samples. (b) IgM S/CO ratios of the primary samples from NE patients and of the control samples. The equivocal zones are illustrated by horizontal lines.
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
67
Fig. 2. Time-course of the IgM results (S/CO) of NE patients with known disease onset. The equivocal zone is illustrated by horizontal lines.
and specificity of the EIA were 95 and 100%, respectively, and there were no patients remaining with indeterminate NE status (Table 4).
3.2. Results of the immunoblot The samples from the NE patients and the control samples were also tested with the bunyavirus IB in order to determine its sensitivity and specificity values and to compare its performance with the EIA. All samples from the NE patients reacted with various degrees of intensity with the hantavirus N proteins in the IgG and/or IgM IB assay (Table 2). Representative examples of two patients with several follow-up samples are shown in Fig. 3. According to the instructions, an IB band was regarded as positive when the intensity was equal or stronger than the supplied cut-off control. When the presence of a positive hantavirus IgG or IgM band was used as criterion to diagnose hantavirus infections, the sensitivity of the IB was very high, but the specificity was poor because several of the healthy adult and tricky control samples did also show positive bands (Table 4). However, the bands observed in the control groups were mostly single IgG or IgM bands against only one hantavirus antigen (Table
2). They were more likely to occur in the IgM IB and with the HTNV N protein. The band intensity was often only weakly positive. In contrast, the bands of the samples of the NE group were usually stronger, there was a high degree of cross reaction between the PUUV N and the HTNV N protein, and both IgG and IgM bands were usually observed in samples from the acute phase of NE. When both positive and borderline bands were considered, at least three of the four possible hantavirus bands (two each in the IgG and IgM assay) were present in almost all primary NE samples. Less than three bands were observed in the first available serum sample after disease onset of only two NE patients. Follow-up samples of these two patients revealed three and four hantavirus bands (Table 3). Using the presence of three hantavirus bands as criterion to diagnose acute NE, the sensitivity and specificity values of the IB were 98 and 100%, respectively, with only three indeterminate samples (Table 4).
3.3. Discordant results Detailed results of the samples with false positive, false negative or discordant results are presented in Table 3. One patient (Table 3, patient 1;
68
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Fig. 3. Follow-up samples of three patients with acute NE. IgG and IgM results are shown in the upper and lower row, respectively. S/CO ratios of the EIA are shown below the IB strips. Patient C corresponds to patient 1 from Table 3. The time span from the approximate disease onset (defined as day 0) is shown above the strips. d, days; w, weeks; m, months.
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Fig. 3, patient C) with acute NE was peculiar because he displayed a strong IgG response in the first sample examined after disease onset but was negative in both the MRL and the Progen IgM EIA. In the IgM IB, weak PUUV N and HTNV N bands were present only in follow-up samples. The patient had the typical clinical findings of NE with acute renal failure, abdominal pain, elevated serum creatinine, and thrombocytopenia. Acute NE was confirmed serologically in this case by demonstrating the low avidity of the PUUV IgG antibodies. In the first serum samples of patients 2 and 3, a positive IgG response was lacking in both the EIA and the IB. This was probably due to the fact that these samples were obtained early after disease onset. The IB reaction pattern of patient 4 was unusual. While the IgM response was directed only against the HTNV antigen, the IgG response was stronger with the PUUV antigen. The routinely performed Progen IgM EIA was borderline and negative for HTNV IgM in the first and second serum sample of this patient and was negative for PUUV IgM in both samples. This patient may have been infected with a hantavirus serotype other than PUUV. The control patients 5 – 12 in Table 3 displayed weak unspecific reactions in the hantavirus IgM EIA. Most of them were also positive for HTNV
69
IgM in the IB. Patients 13 and 14 were the only control patients reacting with two bands. Several other control patients not shown in Table 3 had single positive bands in the IB, but one of the control patients reacted with three or four bands.
4. Discussion A correct and rapid serological confirmation of cases with suspected acute NE is desirable, because the differential diagnosis includes disorders such as rapid progressive glomerulonephritis which must be treated with immunosuppressive drugs as soon as possible. Furthermore, a timely diagnosis of hantavirus infections may help to avoid renal biopsies. We therefore evaluated the performance of a new EIA and a new IB to diagnose or exclude acute NE. The sensitivity of both assays was high. The specificity of the IgM EIA and of the IB was not optimal but could be increased by modifications of the result interpretation. Thus, both assay formats appear to be suitable for use in routine diagnosis. While the pan-reactive hantavirus IgG EIA was 100% specific for the detection of hantavirus IgG, weak unspecific reactions were observed in the IgM EIA when testing serum samples from patients with acute EBV or CMV infections. EBV infections are associated with a polyclonal B-cell
Table 1 Results of the hantavirus enzyme immunoassay Hantavirus IgG EIA Samples (n)
Negative
Acute NEb (43), group 1 1c Healthy adults (27), group 2 27 Tricky samples (29), group 3 29 Tricky samples, subgroups: Acute CMV (8) 8 Acute EBV (10) 10 Autoimmune (6) 6 Pregnant (5) 5 a
Hantavirus IgM EIA
Equivocal
Positive
Negative
Equivocal
Positive
1c – –
41 – –
1c 27 21
– – 4c
42 – 4c
41 – –
– – – –
– – – –
5 5 6 5
2 2 – –
1 3 – –
– – – –
Both IgG and IgM positive or equivocal. Only the results of the first sample after disease onset of each patient are included. c For details see Table 3. b
IgG and IgM Positive or equivocala
P−/9 H+/++
b
a
P+/++ H+/++
6 2 1 – – 1 –
– 25 26 8 10 4 4
– – – –
8 – – – – 1 1
– – 2 – – – –
29 – – 3 5 3 3
1 21 14 2 1 1 2
6 4 6
P or H 9
– – – –
5 – –
P+/++ H−/9
P− H−
P+/++ H−/9
P− H−a
P or H 9
Hantavirus IgM blot Number of samples reacting with
Hantavirus IgG blot Number of samples reacting with
4 P: Puumala, H: Hantaan, -/ 9/+/++: band intensities (see Section 2). Only the results of the first sample after disease onset of each patient are included. c For details see Table 3. d 9 , + and ++ bands are included.
4
Acute NEb (43), group 1 Healthy adults (27), group 2 Tricky samples (29), group 3 Tricky samples, subgroups: Acute CMV (8) Acute EBV (10) Autoimmune (6) Pregnant (5)
Samples (n)
Table 2 Results of the hantavirus immunoblot
3 4 2 –
6 2 9
P−/9 H+/++
– – – –
25 – –
P+/++ H+/++
3 5 3 2
– 19 13
0 band
5 5 1 3
1c 8 14
1 band
– – 2 –
1c – 2c
2 bands
Hantavirus IgG and IgM blot Number of samples reacting withd
– – – –
41d – –
E2 bands
70 J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
71
Table 3 Detailed results of patients with indeterminate or discordant resultsa EIA
IB
No.
Sex
Age
Classification
IgG S/CO
IgM S/CO
IgG PUUV
IgG HTNV
IgM PUUV
IgM HTNV
Total bands
PUUV IgG IFTb
1
M
24
2
M
27
3 4
F M
45 55
5 6 7 8 9 10 11 12 13 14
M M F F M M F F F F
38 19 7 29 46 35 17 19 42 50
Acute NE (ca. d7c) Follow-up (d10) Follow-up (d13) Follow-up (d20) Acute NE (d 2) Follow-up (d 11) Acute NE (d 5) Acute NE (d 17) Follow-up (d 35) Follow-up (d 57) Acute CMV infection Acute EBV infection Acute EBV infection Acute EBV infection Acute CMV infection Acute CMV infection Acute EBV infection Acute EBV infection Autoantibodies (ANAe) Autoantibodies (GBMe)
6.23 7.71 9.98 8.98 0.40 2.60 0.91 2.10 2.90 4.50 0.57 0.60 0.23 0.40 0.14 0.20 0.34 0.22 0.31 0.44
0.55 0.78 1.03 0.67 2.79 \8.75 1.97 6.32 3.86 3.01 1.50 1.27 1.38 1.50 1.02 0.98 1.03 1.10 0.49 0.25
++ ++ ++ ++ 9 ++ 9 + ++ ++ − − − − − − − − − 9
− − 9 9 − + − − + ++ − − − − − − − − + −
− 9 9 9 + ++ 9 − − − − − − − − − − − − −
− 9 9 9 + ++ 9 ++ ++ ++ + − + − 9 + + + + 9
1 3 4 4 3 4 3 2 3 3 1 0 1 0 1 1 1 1 2 2
2048 2048 1024 10,240d 64 128 256 64 n.d. \512 B16 B16 B16 B16 B16 B16 B16 B16 B16 B16
a S/CO of the MRL EIA and the band intensities of the Mikrogen IB (−/9 /+/++, see Section 2) are listed for the following patients: Acute NE patients with a negative IgM EIA (1) or a negative (2) or indeterminate (3) IgG EIA, or less than three IB bands (4). Control patients with a positive (5–8) or indeterminate (9–12) IgM EIA or more than one IB band (13, 14). The indeterminate or discordant results are shown in bold. b Reciprocal titers. c Time span after disease onset for the patients with acute NE. The exact disease onset was unknown for patient 1. It was estimated to be approximately 7 days before the first sample was obtained. d, days. d PUUV IgG avidity: 1%. e ANA, anti-nuclear antibodies; GBM, glomerular basement membrane antibodies.
stimulation which may cause or contribute to the unspecific IgM reactions (Alpers et al., 1994). There are two ways to deal with this problem. Firstly, it appears from our data, that IgM results above 1.5 S/CO are always specific, i. e. the stronger the IgM reaction the higher the likelihood of a specific result. Secondly, 93% of the patients with acute NE were positive for both IgM and IgG in their first serum sample. This is in accordance with the results reported in several other studies (Elgh et al., 1998; Kallio-Kokko et al., 1998). Therefore, it is recommended that hantavirus IgM and IgG are always tested in parallel. When the diagnosis of acute NE was based on a positive IgM EIA and a positive IgG EIA, the
specificity was 100% and there were no indeterminate samples while there was only little loss in sensitivity. Unless the analysed sample was drawn very early after disease onset, the diagnosis of acute NE is unlikely with the constellation, ‘‘IgM positive/IgG negative’’. A second serum sample should be analysed in doubtful cases. When NE is clinically suspected, a rapid diagnosis is important for patient management. On the other hand, though NE is endemic in large parts of Europe, the overall incidence and the number of samples to be tested is low in many of these areas. As a consequence, test runs often consist of only one patient sample. In this situation, the required number of three controls and
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
72
like infections based on band intensities would require the use of less cross reactive antigens, such as the truncated amino-terminal portions of the N proteins or N proteins with deleted aminotermini (Elgh et al., 1997; Morii et al., 1998). The expression systems used for the production of the hantavirus N proteins differed for the EIA and the IB. While the EIA was based on baculo recombinants, E. coli-expressed proteins where used for the IB. Because of the higher rate of unspecific reactions observed in the IB, baculo recombinants may be more suitable for the production of hantavirus N proteins for serological assays. Furthermore, it has been reported that the PUUV N protein expressed in E. coli lacks antigenic sites compared to the native and to baculoexpressed PUUV N proteins (Vapalahti et al., 1996). The suggested modifications of the interpretation criteria for both the EIA and the IB rely on the consideration of the IgG results in addition to the IgM results. Since the positive predictive value of hantavirus IgG tests depends on the overall prevalence of hantavirus infections, the strategy of basing the NE diagnosis on both IgM and IgG tests is less appropriate for areas where the prevalence is high. Further assessing the specificity of
three calibrators in each IgM and IgG EIA run leads to a considerable increase in the costs per reported EIA result. It is desirable that future versions of the EIA will require less controls and calibrators per run. Similar to the hantavirus EIA, the IB was highly sensitive for the diagnosis of acute NE, but there were some problems with unspecific results. The specificity was considerably increased when the interpretation criteria were changed to require the presence of at least three of the four possible hantavirus bands (PUUV IgG, HTNV IgG, PUUV IgM, HTNV IgM) for the diagnosis of acute NE. This recommendation is based on two observations. Firstly, as with the EIA, hantavirus IgG was detected in most NE patients by the time when a serological diagnosis was sought. Secondly, there was extensive cross reaction between the PUUV N and HTNV N proteins of the IB. Highly cross-reacting antibody responses in the acute phase of HFRS have been commonly observed with the use of full length N proteins as in the present assay format (Lee et al., 1979; Svedmyr et al., 1980; Zo¨ller et al., 1993; Elgh et al., 1998). While the cross reaction can be exploited to distinguish between specific and unspecific bands, a differentiation between PUUV-like and HTNV-
Table 4 Sensitivity and specificity values of the MRL EIA and the Mikrogen IB for the diagnosis of acute NE depending on the interpretation criteria Sensitivitya (%)
Interpretation criteria Enzyme immunoassay IgG, IgM, IgM, IgG/IgM, Immunoblot IgG, IgM, IgG/IgM
a
pos.: pos.: pos.: neg.: pos.: neg.:
\1.1 S/COc \1.1 S/COc \1.5 S/CO B0.9 S/CO both E0.9 S/CO IgG or IgM B0.9 S/CO
pos.: pos.: pos.: neg.:
E1 E1 E3 01
band +/++c band +/++c bands 9 /+/++ band 9 /+/++
Indeterminateb (n)
98 98 98
100 92 100
1 4 11
95
100
–
100 97 98
96 76 100
9 16 3
For the NE patients, only the results of the first samples were considered. These samples were excluded from the analysis. c Interpretation criteria of the manufacturer (see Section 2). b
Specificity (%)
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
the testing strategy in these areas will require the inclusion of old-immunity serum samples (i. e. hantavirus IgG positive and IgM negative) in the control groups. While an evaluation of the SFSV N protein of the IB was not part of this study, the usefulness of the concept of combining serological assays for hantavirus and sandfly fever virus infections is doubtful. Because the clinical presentation of both infections is different, a situation will be exceedingly rare when testing one serum sample for both infections is required. Instead, positive SFSV bands in IB assays performed for hantavirus diagnosis and vice versa may lead to interpretation difficulties. Overall, with some modifications of the interpretation criteria to overcome specificity problems, especially of the IB IgM, both the EIA and the IB are suitable for the diagnosis of acute NE. Choosing one of these assay formats for routine diagnostic will probably depend on factors such as assay costs, the involved labour, the number of samples, the potential for automation, and the available experience. If the IB were changed to include antigens with minimal cross-reaction between the different hantavirus serotypes, a sequential testing strategy of a pan-reactive EIA as first line assay and an IB for serotyping of positive EIA results would become feasible.
Acknowledgements We thank Bernd Kra¨mer, Mikrogen, for providing the test kits and for helpful discussions. We also thank Thomas Kerkau for providing the serum samples with autoantibodies.
References Alpers C, Scheid C, Bauer G. Acute EBV infection results in IgM responses against unrelated viruses. Med Microbiol Letters 1994;3:306– 14. Brus Sjo¨lander K, Elgh F, Kallio-Kokko H, Vapalahti O, Hagglund M, Palmcrantz V, Juto P, Vaheri A, Niklasson B, Lundkvist A, . Evaluation of serological methods for diagnosis of Puumala hantavirus infection (nephropathia epidemica). J Clin Microbiol 1997;35:3264–8.
73
Elgh F, Linderholm M, Wadell G, Tarnvik A, Juto P. Development of humoral cross-reactivity to the nucleocapsid protein of heterologous hantaviruses in nephropathia epidemica. FEMS Immunol Med Microbiol 1998;22:309– 15. Elgh F, Lundkvist A, , Alexeyev OA, Stenlund H, Avsic-Zupanc T, Hjelle B, Lee HW, Smith KJ, Vainionpaa R, Wiger D, Wadell G, Juto P. Serological diagnosis of hantavirus infections by an enzyme-linked immunosorbent assay based on detection of immunoglobulin G and M responses to recombinant nucleocapsid proteins of five viral serotypes. J Clin Microbiol 1997;35:1122– 30. Elgh F, Wadell G, Juto P. Comparison of the kinetics of Puumala virus specific IgM and IgG antibody responses in nephropathia epidemica as measured by a recombinant antigen-based enzyme-linked immunosorbent assay and an immunofluorescence test. J Med Virol 1995;45:146– 50. Hedman K, Vaheri A, Brummer-Korvenkontio M. Rapid diagnosis of hantavirus disease with an IgG-avidity assay. Lancet 1991;338:1353– 6. Kallio-Kokko H, Vapalahti O, Hedman K, Brummer-Korvenkontio M, Vaheri A. Puumala virus antibody and immunoglobulin G avidity assays based on a recombinant nucleocapsid antigen. J Clin Microbiol 1993;31:677– 80. Kallio-Kokko H, Vapalahti O, Lundkvist A, , Vaheri A. Evaluation of Puumala virus IgG and IgM enzyme immunoassays based on recombinant baculovirus-expressed nucleocapsid protein for early nephropathia epidemica diagnosis. Clin Diagn Virol 1998;10:83– 90. Lee HW, Lee PW, Lahdevirta J, Brummer-Korventkontio M. Aetiological relation between Korean haemorrhagic fever and nephropathia epidemica. Lancet 1979;1:186– 7. Lundkvist A, , Ho¨rling J, Niklasson B. The humoral response to Puumala virus infection (nephropathia epidemica) investigated by viral protein specific immunoassays. Arch Virol 1993;130:121– 30. Morii M, Yoshimatsu K, Arikawa J, Zhou G, Kariwa H, Takashima I. Antigenic characterization of Hantaan and Seoul virus nucleocapsid proteins expressed by recombinant baculovirus: application of a truncated protein, lacking an antigenic region common to the two viruses, as a serotyping antigen. J Clin Microbiol 1998;36:2514– 21. Mustonen J, Brummer-Korvenkontio M, Hedman K, Pasternack A, Pietila K, Vaheri A. Nephropathia epidemica in Finland: a retrospective study of 126 cases. Scand J Infect Dis 1994;26:7– 13. Niklasson B, Kjelsson T. Detection of nephropathia epidemica (Puumala virus)-specific immunoglobulin M by enzymelinked immunosorbent assay. J Clin Microbiol 1988;26:1519– 23. Niklasson B, LeDuc JW. Epidemiology of nephropathia epidemica in Sweden. J Infect Dis 1987;155:269– 76. Schmaljohn C, Hjelle B. Hantaviruses: a global disease problem. Emerg Infect Dis 1997;3:95– 104. Settergren B. Nephropathia epidemica (hemorrhagic fever with renal syndrome) in Scandinavia. Rev Infect Dis 1991;13:736– 44.
74
J. Schubert et al. / Journal of Clinical Virology 21 (2001) 63–74
Settergren B, Juto P, Wadell G. Detection of specific serum immunoglobulin M in nephropathia epidemica (Scandinavian epidemic nephropathy) by a biotin-avidin-amplified immunofluorescence method. J Clin Microbiol 1987;25:1134– 6. Svedmyr A, Lee PW, Gajdusek DC, Gibbs CJ, Jr, Nystrom K. Antigenic differentiation of the viruses causing Korean haemorrhagic fever and epidemic (endemic) nephropathy of Scandinavia. Lancet 1980;2:315–6. Vapalahti O, Lundkvist A, , Kallio-Kokko H, Paukku K, Julkunen I, Lankinen H, Vaheri A. Antigenic properties and
diagnostic potential of puumala virus nucleocapsid protein expressed in insect cells. J Clin Microbiol 1996;34:119– 25. Wang M, Rossi C, Schmaljohn CS. Expression of non-conserved regions of the S genome segments of three hantaviruses: evaluation of the expressed polypeptides for diagnosis of haemorrhagic fever with renal syndrome. J Gen Virol 1993;74:1115– 24. Zo¨ller L, Yang S, Go¨tt P, Bautz EK, Darai G. Use of recombinant nucleocapsid proteins of the Hantaan and nephropathia epidemica serotypes of Hantaviruses as immunodiagnostic antigens. J Med Virol 1993;39:200– 7.
.