Journal of Virological Methods 78 (1999) 163 – 169
Influenza virus serology—a comparative study Philip H. Rothbarth a,*, Jan Groen a, Arthur M. Bohnen b, Ronald de Groot c, Albert D.M.E. Osterhaus a,d a
Department of Virology, Uni6ersity Hospital Rotterdam, PO Box 2040, 3000 CA, Rotterdam, The Netherlands b Department of Family Practice, Erasmus Uni6ersity Rotterdam, Rotterdam, The Netherlands c Department of Paediatrics, Sophia Children Hospital Rotterdam/Erasmus Uni6ersity Rotterdam, Rotterdam, The Netherlands d Institute of Virology, Erasmus Uni6ersity Rotterdam, Rotterdam, The Netherlands Received 17 August 1998; received in revised form 30 November 1998; accepted 30 November 1998
Abstract Virus isolation or influenza virus antigen detection are the most rapid tests for diagnosis in the acute stage of influenza virus infection. As serology is easier to carry out, the synthesis of serum IgM, IgA and IgG was studied in two well-defined patient groups, infected with influenza B virus (cohort 1, n= 37) and influenza A virus (cohort 2, n= 40), diagnosed by antigen detection and/or virus isolation within 36 h after onset of symptoms. IgM was found in 13 influenza B patients (35%), IgA in 12 patients (32%), whereas a significant antibody rise was found in 33 patients (92%) by enzyme-linked immunosorbent assay (ELISA) and 74% by haemagglutination inhibition assay (HAI). For the influenza A cohort these numbers were respectively 18 (45%), 27 (68%) and 24 (62%) HAI (72%). In age-matched controls, who were bled on the first day of illness of the enrolled patient low prevalence was found for IgA and IgG, for influenza B respectively in 2 and 18%, and for influenza A in 4 and 39%. Studying the kinetics of the antibody response, we found that virus specific IgA and the bulk of IgG is synthesised within the first week of the infection. It is concluded that the finding of a specific serum IgA is highly indicative of an acute influenza infection. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Influenza virus; Enzyme immunoassay; IgA; IgG
1. Introduction
* Corresponding author. Tel.: +31-10-463-6747; fax: +3110-463-6560; e-mail:
[email protected].
Influenza virus infection is an annually recurring important cause of hospitalisation and excess mortality in elderly people and patients with chronic heart and respiratory diseases.
0166-0934/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 6 - 0 9 3 4 ( 9 8 ) 0 0 1 7 4 - 8
164
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
Various methods can be used for the rapid identification of viruses that cause flu-like illnesses. The diagnosis of influenza can be made by antigen detection, virus isolation, polymerase chain reaction (PCR) and serology. Influenza serology is traditionally performed using one of two classical assays: the haemagglutination inhibition assay (HAI) and the complement fixation test (CFT) (Couch and Kasel, 1995; Glezen and Couch, 1997). From a clinical point of view, however, these tests have several limitations: paired sera are needed for a conclusive result, both tests are labour-intensive, standardisation is difficult and no immunoglobulin classes can be distinguished. Furthermore, complications of influenza often arise more than 1 week after onset of the influenza-like symptoms; virus isolation is usually no longer possible and also classical serology will not be conclusive because of the lack of a significant rise in antibody titre. Several investigators have therefore sought alternative assays determining IgM, IgG and IgA separately, which could be helpful in the evaluation of patients with a recent history of influenza, preferably from a single serum (Julkunen et al., 1984; Do¨ller et al., 1986; Vikerfors et al., 1989). It may be concluded from these studies that IgA can be found frequently in sera from patients with recent influenza, whereas the presence of IgM proved to be less indicative. A prospective study is described in which specific serum IgM, IgA and IgG were determined in patients with a proven influenza A or B infection. The sensitivity and specificity of the subclass assays were compared with those of the HAI.
2. Patients and methods
2.1. Patients Human sera were obtained from adult patients (age 18–65 years) who were enrolled in clinical studies. Two cohorts of patients were studied during two consecutive influenza outbreaks. In the first cohort (March – April 1995) 37 patients were enrolled with a confirmed influenza B virus
infection. During the second influenza episode (December 1995) 40 patients were included, all with influenza A virus infection (8 H1N1, 31 H3N2, 1 not typable). Patients were enrolled if the onset of their complaints was shorter than 36 h and influenza was either diagnosed by the direct immunofluorescence assay (DIFA) or virus culture within that time. After enrolment blood was taken on days 1, 6, 28 (cohort 1) and 1, 6, 21 and 60 (cohort 2). Sera were stored at − 20°C until tests were performed. The first cohort was matched for age with otherwise healthy persons visiting the outpatient clinic for sexually transmitted diseases on day 1 of their matched counterpart, all being HIV-seronegative. In the observation period, also influenza A was circulating. Before enrolling the patients informed consent was given; the protocol was approved by the medical-ethical committee of the University Hospital Rotterdam as project number MEC 129.827/ 1993/116.
2.2. Methods 2.2.1. DIFA Nasopharyngeal aspirates were diluted with 5 ml Dulbecco MEM (BioWhittaker,Walkersville, MD) and thoroughly mixed on a vortex mixer for 1 min. The suspension was thus centrifuged for 10 min at 840× g. The sediment was spread on a multispot slide (Nutacon, Leimuiden, The Netherlands), the supernatant was used for virus isolation. After drying, the cells were fixed in acetone for 1 min at room temperature. After washing the slides were incubated for 15 min at 37°C with a FITC-labelled monoclonal antisera both for influenza A and B in duplicate (Imagen influenza A and B, Dako, Glostrup, Denmark). After three washings in PBS and one in tap water, the slides were included in a glycerol/PBS solution (Citifluor, UKC, Canterbury, UK) and covered. Results were interpreted positive if granular cytoplasmic fluorescence was observed in intact cells. 2.2.2. Virus isolation For virus isolation tertiary rhesus monkey kidney cells (RIVM, Bilthoven, The Netherlands)
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
were cultured in 24 well plates (Costar, Cambridge). The bottom of each well contained a glass slide, for immunofluorescence. Before inoculation the plates were washed with PBS and supplied with Eagle’s MEM with Hanks’ salt (ICN, Costa Mesa, CA) to which 0.02% trypsin (Gibco, Bethesda, MD) was added. After inoculation with 0.2 ml/well in triplicate, the plates were centrifuged at 840×g for 1 h. Plates were incubated at 37°C for a maximum of 14 days changing the medium twice a week. In DIFA-negative samples immunofluorescence was performed after 16 – 18 h incubation. After 48 h immunofluorescence was performed as described above. If negative, this procedure was repeated when cytopathic changes were seen. Isolates of influenza A were typed for either H1N1 or H3N2 with reference ferret sea in a haemagglutination inhibition assay as described by Kendal et al. (1982).
2.2.3. Serology 2.2.3.1. Influenza A and B 6irus specific IgM and IgA antibody detection in serum. Influenza A and B virus specific antibodies of the IgM and IgA classes were detected in sera by capture EIA essentially as described previously for RSV and hantavirus serology (Groen et al., 1994; Brandenburg et al., 1997). Briefly, for the measurement of specific IgM or IgA, either goat anti-human IgM or IgA (Cappel Cooper Biomedical, WestChester, PA) was used as capture antibody. Microtitre plates were coated over night with anti-human IgM or IgA in 0.1 M carbonate/bicarbonate buffer pH 9.6 with 100 ml volumes/well. After a wash with PBS/Tween 20 (PBST), 100 ml volumes of 1:100 diluted human serum in EIAbuffer (PBS supplemented with 0.1% BSA, 0.1% milk powder, 1% fetal calf serum and 2% goat serum) was added per well, incubated for 1 h at 37°C, washed three times with PBST and the optimal dilution of either influenza A or B in EIA-buffer plus 1% normal monkey serum was added to each well and incubated for 1 h at 37°C. After washing, the binding of the respective antigen was detected by adding 100 ml of the optimal working dilution of mouse monoclonal antibody (MCA) in EIA-buffer, MCA anti-influenza A NP
165
specific (Chemicon, Temecula, CA) and MCA anti-influenza B virus matrix specific (Chemicon) and incubated for 1 h at 37°C. Subsequently, the plates were washed three times in PBST. HRPO labelled goat anti-mouse (DAKO, Glostrup, Denmark) was used as the conjugate and incubated for 1 h at 37°C, washed three times and shaken dry. Then 100 ml tetramethylbenzidine/H2O2 (TMB) sustrate was added. The reaction was stopped after 10 min with 2 M H2SO4. The optical density (OD) was read spectrophotometrically at 450 nm. Values were calculated as S/N (signal/ negative control). The cut-off was defined as the negative value plus three times the standard deviation.
2.2.3.2. HAI. The HAI was carried out essentially as described previously (Masurel et al., 1981), using turkey erythrocytes for agglutination and four haemagglutinating units of the influenza viruses. For influenza A the prototype antigen strains were A/Texas/36/91(H1N1) and A/Nanhang/7118/95 (H3N2) respectively, whereas for influenza B virus the Harbin/94 antigen strain was used. 2.2.3.3. IgG EIA. The indirect IgG EIA for influenza A and B virus was performed in microtitre plates essentially as described previously (Rimmelzwaan et al., 1997). Briefly, concentrated influenza A and B viruses were solubilized by treatment with 1% Triton X-100 and coated for 16 h at room temperature into microtitre plates in PBS after determination of the optimal working dilution by checkerboard titration. Subsequently, 100 ml volumes of 1:100 diluted human serum samples in EIA buffer were added to the wells and incubated for 1 h at 37°C. Binding of human IgG was detected by adding a goat anti-human IgG peroxidase conjugate (DAKO, Glostrup, Denmark). Plates were developed by adding TMB as substrate and OD was measured at 450 nm. The results were expressed as the S(ignal)/N(egative) ratio of the OD. A serum was considered positive for IgG, if the S/N ratio was beyond the negative control plus three times the standard deviation. A significant increase of the IgG titre was considered, if there was a rise in S/N ] 2.
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
166
Table 1 Results of the serological assays in patients and controls
HAIa IgM IgA IgG
Influenza B cohort (n= 37)
Influenza A cohort (n = 40)
Controlse (n =50)
Anti-influenza A n (%)
Anti-influenza B n (%)
Anti-influenza A n (%)
Anti-influenza B n (%)
Anti-influenza A n (%)
Anti-influenza A n (%)
6b (18) 1 (3) 0 (0) 0 (0)
25b (74) 13 (35) 12 (32) 33 (92)
27c (68) 18 (45) 27 (68) 24 (62)
0c (0) 1 (3) 0 (0) 0 (0)
19 (38) ndd 2 (4) 19 (39)
31 (62) ndd 1 (2) 9 (18)
a
Haemagglutination inhibition assay. Thirty four patients tested. c Thirty nine patients tested. d Not determined. e HAI and IgG: detectable titres (in the patient cohorts: significant rises). b
The first and third serum were compared for serum rises in the S/N ratios.
influenza A was found. No significant titre rises were seen in the HAI against influenza B antigens.
3. Results
3.2.2. Virus-specific IgM, IgA and IgG serum antibodies
3.1. Direct immunofluorescence assay and 6irus isolation In the first cohort 34 out of 37 (92%) patients showed a positive DIFA at enrolment; from the remaining three virus culture was positive after one overnight incubation. In only one case DIFA was positive without a positive result in the virus culture. In this case, however, the HAI later showed a significant titre rise later on. In the second cohort, nasopharyngeal aspirates of 37 out of 41 (90%) showed a positive result in the DIFA; influenza A was isolated from all patients.
3.2. Serology 3.2.1. HAI In the first cohort (34 paired sera) 25 patients (74%) showed a significant titre rise for influenza B in the HAI (Table 1). In six patients a significant antibody rise could also be seen against one of the influenza A virus antigens (18%). In the second cohort 39 paired sera were available for the HAI. In 27 patients (69%) a significant titre rise against the infecting type of
3.2.2.1. Influenza B cohort. Specific IgM against influenza B virus was demonstrated in 13 patients (16%, Table 1); one patient also showed an IgM titre to influenza A virus. The maximum number of positive sera was seen on day 6. IgA against influenza B virus was found in 13 patients (35%), whereas no simultaneous IgA response to influenza A virus was seen. The maximum number was also seen on day 6. For IgG significant a titre rise (sera taken day 1 and 28) could be seen in 33 out of 36 patients (92%). In contrast to the influenza A cohort the patients without a significant rise had no higher titres in the first serum. No significant rises were seen for influenza A in the influenza B cohort. The development of the different classes of serum antibodies against influenza B is given in Fig. 1: the mean titres show a considerable rise, with the exception of IgM. 3.2.2.2. Influenza A cohort. IgM type antibodies to influenza A nucleoprotein could be detected in 18 patients (45%, Table 1); in one case also an IgM response to influenza B virus was seen. In this cohort the maximum number of IgM positive sera was found on day 21.
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
167
Fig. 1. Results of serological analyses by enzyme immunoassay. S/N: optical density (450 nm) specific serum/optical density (450 nm) negative control. — , influenza A nucleoprotein antigen; — , influenza B nucleoprotein antigen.
IgA type antibodies were found in 27 patients (68%), the highest number being observed on day 21. In only one patient specific IgA to influenza B virus was found. A significant rise in specific serum IgG (sampled day 1 and 28) was demonstrated in 24 patients (62%). No specific rise of the titre against
influenza B virus was seen. The titres in the first serum of the 14 of the 15 patients, in whom a titre rise was not seen, were relatively high with a S/N ratio\ 6. This was the case in only 50% of the patients with a significant titre rise. Fig. 1 shows the kinetics of the mean IgM, IgA and IgG titres. A rapid rise was observed between
168
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
day 1 and day 6. Between day, the titre rise was more gradual. IgA tended to decrease after 4 weeks. In comparison to IgA and IgG, the IgM titres were only slightly above the cut-off value. As specific IgA was found in the majority of the patients, the question rose, whether this could be elicited by a previous infection or resulted from the recent infection. To answer this question healthy matched controls were studied to obtain information about background titres in the population (Table 1). The prevalence of specific serum IgA in the controls was extremely low (4 and 2% for respectively influenza A and B viruses). Positive reactions were found in the HAI (cut-off 40) and the IgG ELISA. IgG for influenza A was more prevalent than for influenza B (41 vs 12%). Because of both the low prevalence and low titres of specific IgM this parameter in the control group was not determined.
4. Discussion It was shown that influenza A and B specific IgM-, IgG- and IgA-class antibodies can be demonstrated in most patients at the end of the first week after onset of the disease. It is be concluded from these findings that rapid diagnosis by serology may be made 5 – 7 days after the onset. In the first days of a flu-like illness, there usually is abundant shedding of virus and viral proteins (McIntosh, 1996). Therefore, the most rapid and reliable way to diagnose an influenza virus infection is virus isolation, DIFA or PCR. Nevertheless, these methods have their limitations. First specimens should be transported to and processed in a virological laboratory as soon as possible. This makes the utility of diagnostic virology for influenza together with costs not satisfactory. Second, if the specimens are taken too late in the course of the disease, virus isolation will be negative. The classical objection to clinical virologists that serological diagnosis is too time-consuming, caused by the need of two sera, has been overcome for the diagnosis of many virus infections by the possibility of IgM determination. This im-
munoglobulin class is, however, essentially a parameter for primary infection and is often absent in re-infection or reactivation. On the other hand IgA is a good parameter for infection both in a primary and a booster infection. This was, for example, found for measles virus (Van Binnendijk, 1992) and hantavirus (Groen et al., 1994), others described IgA antibodies for the diagnosis of e.g. CMV (Van Loon et al., 1987). In the case of influenza A it was clearly demonstrated in a study in nonhuman primates that during secondary infection no IgM could be demonstrated whilst high levels of IgA were present (Rimmelzwaan et al., 1997). We found IgM antibodies in 45 and 24% respecyively of the patients infected with influenza A and B virus, respectively. Therefore determination of IgM is not sensitive enough for routine diagnosis of influenza virus infection. In contrast, the rapid increase of the IgA titre enables clinicians to diagnose influenza virus infections in the first week of disease. In our control group, in which sera were collected on the same day as from the patients, the prevalence of IgA against influenza viruses was low. This means that IgA-class antibodies against influenza viruses are generally short-lasting. On the base of a positive IgA the diagnosis of influenza virus infection can be made, followed by clinical and epidemiological decisions, e.g. treatment (Hayden et al., 1997). As treatment is available, it will be necessary to have bedside tests. For serum such a rapid test is already available for dengue virus infection (Sang et al., 1998; Vaughn et al., 1998). On basis of our data it should be possible to develop such serological tests for influenza virus infections. Finally, the results obtained by the IgG enzyme immunoassay are comparable with or better than the HAI. In our opinion the HAI could be replaced by the enzyme immunoassay in a diagnostic laboratory; it should be reserved for epidemiological surveys, in which it is an easy and cheap test. The fact that in the influenza A cohort in 14 patients no significant rise found could be attributed to the very high serum titres against the influenza A nucleoprotein on day 1. This may be the result of a rapid booster after being infected,
P.H. Rothbarth et al. / Journal of Virological Methods 78 (1999) 163–169
which may prevent a significant rise of antibody titre in sera taken at days 1 and 21. In summary, a few days after onset of disease specific serum IgA can be detected in influenza virus infections (both A and B). Although the direct immunofluorescence assay, virus isolation or PCR are more sensitive during the first week of disease, serology can be used more easily. After 1 week it is the first method of choice, because influenza virus cannot be isolated any more.
Acknowledgements The dedicated technical support by Georgina Arron and Cedric Copra is gratefully acknowledged. We thank GlaxoWellcome for kindly providing us with serum samples.
References Brandenburg, A.H., Groen, J., van Steensel-Moll, H.A., Claas, E.J.C., Rothbarth, Ph.H., Neijens, H.J., Osterhaus, A.D.M.E., 1997. Respiratory syncytial virus specific serum antibodies in infants under six months of age: limited serological response upon infection. J. Med. Virol. 52, 97– 104. Couch, R.B., Kasel, J.A., 1995. Influenza. In: Lenette, E.H., Lenette, D.A., Lenette, E.T. (Eds.), Diagnostic Procedures for Viral, Rickettsial and Chlamydial Infections, 7th edition. American Public Health Association, Washington, pp. 431 – 446. Glezen, W.P., Couch, R.B., 1997. Influenza viruses. In: Evans, A.E., Kasklow, R.A. (Eds.), Viral Infections in Humans, Epidemiology and Control, 4th edition. Plenum, New York, pp. 473 – 506. Do¨ller, G., Do¨ller, P.C., Gerth, H.J., 1986. Diagnostic significance of influenza subtype-specific IgG, IgA and IgM antibodies. J. Biol. Standard. 14, 163–175. Groen, J., Gerding, M., Jordans, H.G.M., Clement, J.P., Osterhaus, A.D.M.E., 1994. Class and subclass distribu-
.
169
tion of hantavirus-specific serum antibodies at different times after onset of nephropathia epidemica. J. Med. Virol. 43, 39 – 43. Hayden, F.G., Osterhaus, A.D.M.E., Treanor, J.J., Fleming, D.M., Aoki, F.Y., Nicholson, K.G., Bohnen, A.M., Hirst, H.M., Keene, O., Wightman, K., 1997. Efficacy and safety of the neuraminidase inhibitor zanamivir in the treatment of influenzavirus infections. New Engl. J. Med. 337, 874 – 880. Julkunen, I., Kleemola, M., Hovi, T., 1984. Serological diagnosis in influenza A and B infection by enzyme immunoassay. Comparison with the complement fixation test. J. Virol. Methods 9, 7 – 14. Kendal, A.P., Pereira, M.S., Shehel, J.J., 1982. Concepts and Procedures for Laboratory-Based Influenza Surveillance. Viral Disease Unit, WHO, Geneva. Masurel, N., Ophof, P., De Jong, P., 1981. Antibody response to immunization with influenza A/USSR/77 (H1N1) virus in young individuals primed or unprimed for influenza A/New Jersey/76 (H1N1) virus. J. Hyg. (Lond) 87, 201 – 209. McIntosh, K., 1996. Diagnostic virology. In: Fields, B.N., Knipe, D.M., Howley, P.M., et al. (Eds.), Field Virology, 3rd edition. Lipincott-Raven, Philadelphia, pp. 401 – 430. Rimmelzwaan, G.F., Baars, M., Van Beek, R., Van Amerongen, G., Lo¨vgen-Bengtsson, K., Claas, E.J.C., Osterhaus, A.D.M.E., 1997. Induction of a protective immunity against influenza virus in a macaque model: comparison of conventional and iscom vaccines. J. Gen. Virol. 78, 757 – 765. Sang, C.T., Hoon, Ls., Cuzzubbo, A., Devine, P., 1998. Clinical evaluation of a rapid immunochromatographic test for the diagnosis of dengue virus infection. Clin. Diagn. Lab. Immunol. 5, 407 – 409. Van Binnendijk, R., 1992. T cell function in measles. Thesis, Utrecht. Van Loon, A.M., Heesen, F.W., Van der Logt, J.T., 1987. Antibody isotype response after cytomegalovirus infection. J. Virol. Methods 15, 101 – 107. Vaughn, D.W., Nisalak, A., Kalayanarooj, S., Solomon, T., Dung, N.M., Cuzzubbo, A., Devine, P.L., 1998. Evaluation of a rapid immunochromatographic test for diagnosis of dengue virus infection. J. Clin. Microbiol. 36, 234 – 238. Vikerfors, T., Lindegren, G., Grandien, M., Van der Logt, J., 1989. Diagnosis of influenza A virus infections by detection of specific immunoglobulins M, A and G in serum. J. Clin. Microbiol. 27, 453 – 458.