Prospective comparison of two enzyme-linked immunosorbent spot assays for the diagnosis of Lyme neuroborreliosis

Abstract

Commercial cellular tests are used to diagnose Lyme borreliosis (LB), but studies on their clinical validation are lacking. This study evaluated the utility of an in-house and a commercial enzyme-linked immunosorbent spot (ELISpot) assay for the diagnosis of Lyme neuroborreliosis (LNB). Prospectively, peripheral blood mononuclear cells (PBMCs) were isolated from patients and controls and analysed using an in-house Borrelia ELISpot assay and the commercial LymeSpot assay. B. burgdorferi B31 whole cell lysate and a mixture of outer surface proteins were used to stimulate the PBMCs and the numbers of interferon-gamma-secreting T cells were measured. Results were evaluated using receiver operating characteristic (ROC) curve analysis. Eighteen active and 12 treated LNB patients, 10 healthy individuals treated for an early (mostly cutaneous) manifestation of LB in the past and 47 untreated healthy individuals were included. Both assays showed a poor diagnostic performance with sensitivities, specificities, positive and negative predictive values ranging from 44.4-66.7%, 42.0-72.5%, 21.8-33.3% and 80.5-87.0%, respectively. The LymeSpot assay performed equally poorly when the calculation method of the manufacturer was used. Both the in-house and the LymeSpot assay are unable to diagnose active LNB or to monitor antibiotic treatment success.

Keywords: Borrelia; ELISpot; Lyme neuroborreliosis; T cells; interferon-gamma.

Figure 1

Results of the in‐house Borrelia enzym‐linked immunosorbent spot (ELISpot) assay (a–d) and the LymeSpot assay (e–f) expressed in the numbers of spot‐forming cells (SFCs). (a) (50 µl), (c) and (e) (both 100 µl) are the results after peripheral blood mononuclear cell (PBMC) stimulation with Borrelia burgdorferi B31, and (b) (50 µl), (d) and (f) (both 100 µl) are the results after PBMC stimulation with outer surface protein (Osp)‐mix among active Lyme neuroborreliosis patients (ANB), treated Lyme neuroborreliosis patients (TNB), treated healthy individuals (THI) and untreated healthy individuals (UHI). The displayed P‐values are corrected and interpreted using the Benjamini–Hochberg procedure with a false discovery rate of 2.5% for multiple comparisons (only false discovery rates < 0.025 are displayed).

Figure 2

The receiver operating characteristic (ROC) curves for both the in‐house Borrelia enzyme‐linked immunosorbent spot (ELISpot) assay (solid lines) and the LymeSpot assay (dashed lines) to discriminate active Lyme neuroborreliosis (LNB) patients from the other three groups. The dotted grey line represents the random predictor. (a) ROC curves based on the numbers of spot‐forming cells after stimulation with 100 µl of B. burgdorferi B31. (b) ROC curves based on the numbers of spot‐forming cells after stimulation with 100 µl of Osp‐mix. (c) ROC curves based on the outcomes of the two binary logistic regression models (M) for which the combined results of both Borrelia antigens, which were based on the numbers of spot‐forming cells, without (M1) and with (M2) their interaction term, were included as risk factors. P (M1) represents the adjusted P‐value for the comparison of both assays using the outcomes of model 1, P (M2) represents the adjusted P‐value for the comparison of both assays using the outcomes of model 2, P (M1 versus M2 in‐house Borrelia ELISpot) represents the adjusted P‐value for the comparison of the outcomes of models 1 and 2 for the in‐house Borrelia ELISpot assay, P (M1 versus M2 LymeSpot) represents the adjusted P‐value for the comparison of the outcomes of models 1 and 2 for the LymeSpot assay. (d) ROC curve of the LymeSpot assay based on the final LymeSpot result (a combination of the stimulation indices of both antigens following the protocol of the manufacturer (Supporting information, Fig. S2).