ABSTRACT
To determine suitability for national serosurveys, we compared two commercial enzyme-linked immunosorbent assays (ELISAs) for mumps antibody, Enzygnost Anti-Parotitis-Virus/IgG (which uses a whole-virus antigen) and Microimmune Mumps IgG Screen ELISA (which uses a recombinant nucleoprotein antigen), by testing 1,915 opportunistically collected sera submitted to diagnostic laboratories across Australia in 1997 to 1998. The proportion of positive results increased with age in both ELISAs but was significantly higher with the Microimmune than with the Enzygnost ELISA overall (88% versus 63%; P < 0.01) and in all age groups. However, the proportion of equivocal results was significantly higher with the Enzygnost than with the Microimmune ELISA (9% versus 4%; P < 0.01). Of the 572 sera with discrepant or equivocal results, 508 had sufficient sample remaining to perform the neutralization test (NT). A proportion with concordant results in both ELISAs were also tested by the NT. For sera with discrepant results, there was significantly better agreement between the NT and Microimmune than between the NT and Enzygnost (310/444 [70%] versus 135/348 [39%]; P < 0.01). Of 64 sera with equivocal Microimmune results, 45 (70%) were positive in the NT compared with 140 of 160 (88%) equivocal Enzygnost results (P < 0.01). Compared with the NT, the Microimmune ELISA is more sensitive (96% versus 80%) but apparently less specific (36% versus 85%) than the Enzygnost ELISA. However, this is likely to be due to the generally lower sensitivity of the NT, since the Microimmune results reflect expected seroprevalence, based on vaccine uptake in the age groups studied. We conclude that the Microimmune ELISA is a more appropriate assay than the Enzygnost ELISA for estimation of mumps seroprevalence.
Assessing population-based mumps seroprevalence is difficult. There is no international reference serum, and correlation between different assays differs according to the virus strain used and the type of antibodies detected (14). The lack of standardization is a major obstacle to international comparisons of mumps serosurvey results and evaluation of the impact of different mumps vaccination schedules.
The neutralizing antibody test (NT) is considered the most specific indicator of protective mumps antibodies, but it is labor-intensive and difficult to perform (4, 10). Enzyme-linked immunosorbent assays (ELISAs) are simple, rapid, and suitable for automation and so ideally suited to large-scale mumps serosurveys (17). Generally they are reported to be more sensitive than the NT (2, 5, 9, 13, 15, 16). However, the NT can detect functional antibody of any class and has been shown to detect low levels of specific measles antibody below the level of detection of immunoglobulin G (IgG) binding antibody assays, such as ELISAs (3). It is plausible that the same could apply to mumps antibody.
In this study we compared the performance of two commercial ELISAs (Enzygnost Anti-Parotitis-Virus/IgG [Dade Behring, Marburg, Germany] and Mumps IgG Screen ELISA [Microimmune Ltd., Brentford, Middlesex, United Kingdom]) for the detection of serum IgG antibody to the mumps virus and for suitability for a national serosurvey.
MATERIALS AND METHODS
Serum samples.To compare the performance of the ELISAs, 1,915 serum samples, from individuals aged 1 to 49 years, were tested. They were collected between January 1997 and August 1998 from 45 major public and private diagnostic laboratories throughout Australia. The sera had been submitted for diagnostic testing and would otherwise have been discarded. Sera from individuals who were known to be immunocompromised, to have received multiple transfusions in the past 3 months, to be infected with human immunodeficiency virus, or to have had serum collected for the diagnosis of measles were excluded. Only one sample from each individual was tested. All sera were stored at −70°C prior to testing.
Antibody assays.Both ELISAs detect IgG (antigen binding antibody) to the mumps virus by the indirect ELISA technique. The assays were performed on all sera according to the manufacturers' specifications with reagents supplied with the kits. Sera giving equivocal results were retested by the same method and reclassified as positive or negative if appropriate.
All sera with sufficient remaining volume that gave persistently equivocal results and/or discrepant results between the two ELISAs were tested by the NT. A representative sample of 7.5% (90/1,197) of sera that were positive and 70% (102/146) of sera that were negative in both ELISAs were also tested by NT, and the results were used to predict the distribution of NT results for the remaining sera.
(i) Enzygnost Anti-Parotitis-Virus/IgG.The materials supplied with the test kit included Reference P/N, containing human IgG specific for mumps virus, and microtitration plates containing six paired strips of eight wells. The first well of each pair was coated with antigen derived from simian kidney cells infected with the Enders strain of the mumps virus, and the second (control) well was coated with uninfected simian kidney cells. Each specimen and the P/N were tested in the paired wells at a dilution of 1:231. The P/N was run at the beginning and end of each plate. On completion of the assay, the difference in absorbance between the antigen-positive and control antigen wells, ΔA, was calculated for each specimen and the Reference P/N. A plate run was validated when ΔA for each Reference P/N was ≥0.5. The specimens were reported as negative (ΔA < 0.100), positive (ΔA > 0.200), or retest/equivocal (0.100 ≤ ΔA ≤ 0.200).
(ii) Microimmune Mumps IgG Screen ELISA.The Microimmune assay detects IgG antibodies specifically directed against the mumps nucleoprotein. Reagents supplied with the kit included positive-control and negative-control (NC) sera and a microtitration plate with each well coated with a recombinant mumps nucleoprotein antigen. The recombinant nucleoprotein was made in yeast cells and derived from a wild-type United Kingdom mumps isolate designated the Gloucester strain (18). The controls were tested as supplied, without further dilution, in the first four wells of each plate; the NC was tested in triplicate. The specimens were tested at a dilution of 1:201, one well per specimen. A plate run was validated when the optical density (OD) of the positive control was >0.4 and the OD of each of the three NC wells was ≤0.2. The cutoff (CO) was determined by adding 0.4 to the mean of the three NCs. The OD of each individual NC had to be within 20% of the mean of the NCs; if one of the three OD values differed by more than 20%, it was omitted and the mean value recalculated. Specimens were reported as positive [OD ≥ (CO × 1.1)], negative [OD ≤ (CO × 0.9)], or retest/equivocal [(CO × 0.9) < OD < (CO × 1.1)].
Neutralization test.Sera inactivated at 56°C for 30 min were prediluted to 1:2.5 in dilution buffer (minimal essential medium containing 4% fetal calf serum). Twenty-five microliters of the 1:2.5 dilution was then transferred to each of the first three wells of a tissue culture plate (Falcon; Becton Dickinson, Franklin Lakes, NJ). Commencing with the third well, doubling dilutions were made in 25-μl volumes of dilution buffer to a final dilution of 1:80. The Jeryl Lynn strain of mumps virus (MUMPSVAX*; Merck & Co., Inc., Whitehouse Station, NJ), grown and harvested after two passages in Vero cells, was diluted to give 100 50% tissue culture infective doses, and 25 μl was added to all wells with the exception of the serum controls (25 μl diluent). Following incubation at 4°C for 90 min, 200 μl of a Vero cell suspension (5 × 105 cells/ml) was added to all wells. The plates were incubated at 37°C under 5% CO2 and read microscopically on day 6 for evidence of cytopathogenic effect. The end point was read at total inhibition of all cytopathogenic effect. A neutralizing antibody titer of <2.5 (i.e., no well showed total inhibition) was considered negative. A positive serum control was included in each plate, and the viability of the virus and cell culture was confirmed in the last plate of each run.
Statistical analysis.The percentages of sera with positive, negative, and equivocal results were determined for each age group and assay. The results for each assay, both overall and for each age group, were compared using the chi-square test. P values of <0.05 were considered statistically significant.
Because different types of assay measure different types of antibody and there is no accepted “gold standard” for both sensitivity and specificity, we used two different “gold standards.” First, sensitivity and specificity were estimated by comparing the results of each ELISA with those of the NT. The NT results were extrapolated for all sera (see above), on the assumption that the sera tested in each category (equivocal, discrepant, or consistent between the two ELISAs) were representative. The criteria for the second “gold standard” were as follows: a result was regarded as positive when either the NT or both ELISAs were positive and as negative when the NT and one or both ELISAs were negative (on the assumption that false-positive results were more likely with ELISA).
Ethics approval.The study was approved by the Human Research Ethics committee of the Western Sydney Area Health Service and the State-wide Health Confidentiality and Ethics Committee of the New South Wales Health Department.
RESULTS
The proportion of positive sera was significantly higher with the Microimmune than with the Enzygnost assay, both overall and for each age group (Table 1). With both assays, the proportion of positive sera increased significantly between the age of 1 year and the ages of 7 to 11 years. For older age groups (≥12 years), the Microimmune assay gave consistently higher seropositive rates (≥93%) than the Enzygnost assay, with its lower and more variable rates (54 to 82%).
Results of two mumps enzyme immunoassays by age group
In all age groups, there were higher proportions of equivocal results with the Enzygnost than with the Microimmune assay (Table 1). The difference was statistically significant overall and in most age groups. For both assays, the proportion of sera with equivocal results was lower in adults aged ≥20 years than in younger age groups, but this difference was statistically significant only for the Microimmune assay (P = 0.02).
The overall agreement between the Enzygnost and Microimmune ELISAs was 70.7% (1,353/1,915) (Table 2).
Concordance between the results of the Enzygnost and Microimmune ELISAsa
All of the sera giving equivocal results in both tests (10/10) and 89% (498/562) of those with discrepant ELISA results had sufficient volume remaining to perform the NT (Table 3). For sera with discrepant results, there was significantly better agreement between the NT and Microimmune than between the NT and Enzygnost (310/444 [70%] versus 135/348 [39%], respectively; P < 0.01). The NT was positive for 88% (141/160) of sera giving equivocal results in the Enzygnost assay and for 70% (45/64) (P < 0.01) of equivocal sera in the Microimmune assay. Of those sera giving negative results in both ELISAs, 31% (32/102) were positive in the NT. However, all of these sera had titers of ≤10, with 71% (23/32) giving a titer of 2.5. Most (91%) were from children, who would have received only a single dose of the measles-mumps-rubella (MMR) vaccine and would have had minimal exposure to mumps infection: 68% (22/32) were aged 1 to 6 years, and 23% (7/32) were aged 7 to 16 years. Only 3% (3/90) of sera positive by both ELISAs were negative by the NT. By extrapolation of the NT results from the sample of sera tested, it was estimated that approximately 85% of all sera would have been positive by the NT. Using either “gold standard,” sensitivity was higher, but specificity lower, for the Microimmune than for the Enzygnost assay (Table 4).
Neutralization test results compared with results of the Enzygnost and Microimmune ELISAsa
Estimated sensitivities and specificities of the Enzygnost and Microimmune ELISAsa
DISCUSSION
In this study the Microimmune assay gave significantly higher proportions of positive results and lower proportions of equivocal results than the Enzygnost assay. However, both assays showed decreasing proportions of seronegative and equivocal sera with increasing age of the individuals tested.
The pattern by age for both assays reflects the history of vaccination in Australia. A monovalent mumps vaccine was licensed for use on infants aged 12 to 15 months in 1980, and routine mumps vaccination was introduced in 1982 as a combined measles-mumps vaccine (7). In 1989, this was replaced by the MMR vaccine for 12-month-old infants, and in 1994, a second dose of MMR was introduced for children aged 10 to 16 years. Therefore, most individuals aged up to 20 years in 1997 to 1998, when sera were collected for this study, would have been eligible for at least one dose of mumps vaccine. Hence, the higher seroprevalence and low proportion of equivocal results in the over-20 age groups in both assays are probably due to stronger immune responses induced by exposure to the wild-type mumps virus, compared with vaccine-induced immunity in younger subjects.
Despite the similar trends by age group, the assays provide significantly different estimates of seroprevalence. Both assays are indirect ELISAs, but they use different antigens to coat the wells of the microtitration plates: purified Enders strain mumps virus in the Enzygnost test and recombinant mumps nucleoprotein in the Microimmune assay. Exposure to mumps vaccine elicits an immune response to both the S (soluble nucleoprotein) and V (viral envelope) antigens; therefore, detection of antibody to nucleoprotein is an indicator of a specific antibody response to exposure. The Microimmune assay specifically measures IgG antibody to nucleoprotein. However, core proteins, such as nucleoprotein, are also prominent antigens in the whole-virus lysate (14). It is therefore not clear why there is such an apparent difference between the assays.
Comparison of ELISA (IgG antigen binding assay) and NT (a whole-antibody functional assay) results also demonstrates differences between them, as described above. The NT apparently can detect functional antibody below the level of detection by ELISA, as shown by the relatively high proportion of sera that were negative by both ELISAs but positive (at low titers) by the NT. The significantly higher starting dilutions for both ELISAs than for the NT (∼1:200 compared with 1:2.5) could also contribute to these differences, which are similar to those recently reported in a comparison between two measles IgG antibody ELISAs and a plaque reduction NT (3). We used a second “gold standard” in part to balance these differences, although both are still weighted in favor of the NT, because of its high specificity. Given that there is a difference between the two ELISAs and that no “gold standard” is ideal, we believe that the Microimmune assay provides a more valid estimate of population seroprevalence than the Enzygnost assay for the following reasons.
First, by both of the “gold standards” used in this study, the Microimmune ELISA is more sensitive than the Enzygnost ELISA. The higher sensitivity of the Microimmune assay produces a seroprevalence closer to that expected on the basis of known MMR immunization rates. For instance, mumps seroprevalence measured using the Microimmune assay is much closer to that of measles (8) than that estimated by the Enzygnost assay. In addition, the expected proportion of seropositive 2- to 6-year-olds, based on vaccination coverage and effectiveness estimates, is closer to levels demonstrated by the Microimmune assay; MMR uptake for 2- to 6-year-olds in 1998 was estimated to be ∼90% (11). Given that vaccine effectiveness is between 80 and 100% (19), we would expect 72 to 90% of 2- to 6-year-olds to be seropositive. This range encompasses the result from the Microimmune but not the Enzygnost assay.
Second, we believe that the Microimmune assay is more specific than suggested by comparison with both “gold standards.” Theoretically, low specificity could be due to cross-reactivity with the parainfluenza virus (9). Sera containing parainfluenza virus antibody were not available to confirm this hypothesis, but Frankova et al. (6) found that cross-reactivity was eliminated when purified mumps nucleoprotein antigen was used for ELISA, as it is in the Microimmune assay. In addition, although it is (probably) highly specific, the NT has been shown to be generally less sensitive (2, 5, 9, 13, 15, 16) than ELISA, notwithstanding its ability to detect low levels of functional antibody not detectable by ELISAs. Moreover, the strain of mumps virus used in the NT is known to affect its performance (14). Thus, the apparently low specificity of the Microimmune assay appears to be due to a lack of sensitivity of the “gold standards” used in this study.
The higher sensitivity of the Microimmune assay may also explain why seroprevalence in the 17- to 19-year age group was lower than in surrounding age groups only in the Enzygnost assay. This probably reflects a combination of waning vaccine-induced immunity and low rates of natural immunity in this first cohort of children eligible for immunization in the 1980s. Presumably the more sensitive Microimmune assay can detect lower antibody levels in this age group.
Although there are no internationally agreed serological correlates of protective immunity, detection of mumps IgG antibody is an indication of past infection or immunization (1). Thus, seroprevalence of IgG antibodies is central to the determination of herd immunity in a population. The level of immunity required to achieve herd immunity is 85% to 90% (12). In this study seroprevalence, as determined by the Microimmune assay, was at an acceptable level in all age groups from the 7- to 11-year group up. However, as determined by the Enzygnost assay, seroprevalence would not be considered sufficient for herd immunity.
In conclusion, these results exemplify how assay selection can have a significant impact on age-specific and thus population-based estimates of immunity to mumps. In this study, the Microimmune assay produced a higher proportion of positive results and fewer equivocal results than the Enzygnost assay and, based on expected levels of immunity and comparison with the results of two “gold standards,” probably provides a more realistic estimate of seroprevalence in the community.
ACKNOWLEDGMENTS
NCIRS is supported by The Commonwealth Department of Health and Ageing, The NSW Department of Health, and The Children's Hospital at Westmead.
We thank the staff of the 45 laboratories (http://immunise.health.gov.au/metadata/measeval.htm, p. 8-9) that provided the sera, the nurses at the Children's Hospital at Westmead Centre for Immunization Research, and laboratory staff at the Centre for Infectious Diseases and Microbiology, in particular Beverley Bowcock, for help in processing and testing the sera.
FOOTNOTES
- Received 23 June 2005.
- Returned for modification 4 August 2005.
- Accepted 21 April 2006.
- American Society for Microbiology
REFERENCES
