Development and evaluation of antibody-capture immunoassays for detection of Lassa virus nucleoprotein-specific immunoglobulin M and G

Abstract

Background: The classical method for detection of Lassa virus-specific antibodies is the immunofluorescence assay (IFA) using virus-infected cells as antigen. However, IFA requires laboratories of biosafety level 4 for assay production and an experienced investigator to interpret the fluorescence signals. Therefore, we aimed to establish and evaluate enzyme-linked immunosorbent assays (ELISA) using recombinant Lassa virus nucleoprotein (NP) as antigen.

Methodology/principal findings: The IgM ELISA is based on capturing IgM antibodies using anti-IgM, and the IgG ELISA is based on capturing IgG antibody-antigen complexes using rheumatoid factor or Fc gamma receptor CD32a. Analytical and clinical evaluation was performed with 880 sera from Lassa fever endemic (Nigeria) and non-endemic (Ghana and Germany) areas. Using the IFA as reference method, we observed 91.5-94.3% analytical accuracy of the ELISAs in detecting Lassa virus-specific antibodies. Evaluation of the ELISAs for diagnosis of Lassa fever on admission to hospital in an endemic area revealed a clinical sensitivity for the stand-alone IgM ELISA of 31% (95% CI 25-37) and for combined IgM/IgG detection of 26% (95% CI 21-32) compared to RT-PCR. The specificity of IgM and IgG ELISA was estimated at 96% (95% CI 93-98) and 100% (95% CI 99-100), respectively, in non-Lassa fever patients from non-endemic areas. In patients who seroconverted during follow-up, Lassa virus-specific IgM and IgG developed simultaneously rather than sequentially. Consistent with this finding, isolated IgM reactivity, i.e. IgM in the absence of IgG, had no diagnostic value.

Conclusions/significance: The ELISAs are not equivalent to RT-PCR for early diagnosis of Lassa fever; however, they are of value in diagnosing patients at later stage. The IgG ELISA may be useful for epidemiological studies and clinical trials due its high specificity, and the higher throughput rate and easier operation compared to IFA.

Fig 1. Sample-to-cut-off (S/CO) values for 880 sera tested in IgM ELISA, RF-based IgG ELISA, and CD32-based IgG ELISA and comparison with IFA results.

The S/CO values obtained with the ELISA are shown as histograms according to sample origin, type of ELISA, and IFA results. The cut-offs for the ELISAs are indicated by vertical dotted lines. The number of samples with OD < CO and OD > CO is indicated in left and right corner, respectively, of each diagram.

Fig 2. Comparison of RF- and CD32-based IgG ELISA results obtained from 776 sera.

The S/CO values obtained with the RF ELISA were plotted against the S/CO values of the CD32 ELISA. The cut-offs for both assays are indicated by horizontal and vertical dotted lines. The number of samples per quadrant is also given. The regression curve is shown as straight line. Samples from Lassa fever non-endemic and endemic countries were plotted separately.

Fig 3. IgM and/or IgG seroconversion detected by ELISA and IFA during follow-up of 21 Lassa fever patients.

Each of the panels A–U depicts an individual patient. The cut-off for the ELISAs is indicated by a horizontal dotted line. The timeline for each patient is given in days with day 1 being the day the first sample was taken for Lassa fever diagnostics. IFA results are coded by symbols according to the categories "clearly negative", probable positive", and "clearly positive" on the S/CO curves of the corresponding ELISA.

Fig 4. Clinical performance characteristics of the IgM ELISA as stand-alone test and in combination with the IgG ELISA depending on prevalence of Lassa fever and pre-existing IgG among all patients tested.

The calculations are based on the data shown in Table 3. PPV and NPV of the stand-alone IgM ELISA depend only on the Lassa fever prevalence in diagnostics. However, PPV, NPV, sensitivity, and positive likelihood ratio for combined detection of IgM and IgG (i.e. a positive test result means that both IgM and IgG is positive) depend on the prevalence of Lassa fever as well as the prevalence of pre-existing IgG among all patients tested. To simplify calculation, we assumed a ratio of 1:3 between prevalence of Lassa fever and pre-existing IgG, which roughly corresponds to the setting in Nigeria where the study was performed.