Introduction of IgM testing for the diagnosis of acute Lyme borreliosis: a study of the benefits, limitations and costs

Serological testing is recommended for all suspected cases of Lyme borreliosis (LB) other than those with clinically identified erythema migrans (EM) rash. Until recently, testing for LB in Scotland utilised a two-tier testing protocol: a screening enzyme-linked immunosorbent assay (ELISA) for the detection of IgG antibodies to Borrelia burgdorferi sensu lato (hereafter B. burgdorferi) followed by IgG immunoblot [1]. To aid earlier detection of LB and to comply with the National Institute for Clinical Excellence (NICE) guidelines for Lyme disease published in 2018 [2], immunoblot for the detection of IgM antibodies to B. burgdorferi was introduced in 2018, followed by the introduction of IgM (and IgG) chemiluminescent immunoassay (CLIA) in 2020.

Methods to detect IgG antibodies to B. burgdorferi lack sensitivity during early disease, and their persistence in serum can complicate interpretation [3‐8]. Although IgM antibodies are produced earlier than IgG, studies found that IgM tests have suboptimal specificity with high false positive rates due to cross-reactions with other infections and autoantibodies [9‐11]. IgM may also persist [12, 13].

Data for serum samples sent from throughout Scotland and tested at the Scottish Lyme Disease and Tick-Borne Infections Reference Laboratory (SLDTRL) for B. burgdorferi antibodies from 1 June 2018 to 17 October 2020 were analysed:

Clinical information from specimen request forms, additional information from questionnaires returned from the referring clinician and any data from subsequent samples were used to allocate individual patients with isolated IgM results into groups based on likelihood of acute LB (Table 1).

Of the 15,294 sera tested for LB for the period 1 June 2018 to 14 April 2020, 1304 (8.5%) were reactive by IgG ELISA and thus tested by IgG and IgM immunoblot. Of these, 188/1304 (14.4%) were IgM immunoblot positive and IgG negative or equivocal. These 188 sera came from 152 individual patients: 76 (50.0%) were classed as “Probable” acute LB, 34 (22.4%) as “Possible” and 30 (19.7%) as “Not consistent” with acute LB. Twelve (7.9%) patients had insufficient clinical details to assign a presumptive diagnosis (Fig. 1).

Of the 2895 sera tested for LB for the period 15 April 2020 to 17 October 2020, 661 (22.8%) were reactive by IgG and/or IgM CLIA and subsequently tested by IgG and IgM immunoblot. Of these, 346 (52.3%) were reactive for IgM CLIA only: the majority of which (73.1%) did not confirm by immunoblot, the remaining 93 (26.9%) sera were positive for IgM immunoblot only. These were from 73 individual patients: 23 (31.5%) patients were classed as “Probable” acute LB, 24 (32.9%) as “Possible” and 20 (27.4%) as “Not consistent” with acute LB. Nine patients (12.3%) had insufficient clinical details to assign a presumptive diagnosis (Fig. 2).

During the first study period (22.5 months), following the introduction of IgM immunoblot, a total of 1304 IgM immunoblots were performed, representing an increase in laboratory consumables costs of £34,556 (£18,430 per year). During the second study period (6 months), following the introduction of IgM CLIA, 2895 specimens were tested by IgM CLIA at an additional cost of £10,595. An additional 346 sera were tested by immunoblot as they had a reactive IgM CLIA, resulting in additional immunoblot costs of £18,338. Thus, the increase in laboratory costs after the introduction of IgM CLIA could be extrapolated to £57,866 per year.

This study has shown that IgM immunoblot is a valuable tool in the laboratory diagnosis of LB, allowing us to detect 110 patients with “Probable” or “Possible” acute LB over the initial 22.5-month study period that may otherwise have been missed with IgG immunoblot alone. Although false positive results were obtained with the IgM immunoblot, the estimated false positive rate of 48/188 (25.5%) was slightly lower than other recent studies [14, 15]. Introduction of IgM immunoblotting led to an increase in test costs of approximately £34,556. However, if more cases of acute LB are detected and treated early, significant cost savings to the health service could result. The increased risk of developing disseminated and late LB, along with the associated manifestations, in untreated patients has been well described [16‐18]. Whilst repeat samples are routinely requested in patients with negative serology and recent onset, there is a clear potential for cases to be missed from follow-up. A 2010 study in the Netherlands found that the mean cost of disseminated LB and Lyme-related persisting symptoms was around 5700 Euros per case [19]. Although unlikely, if all 110 of the above patients were missed and progressed to disseminated LB/persisting symptoms, this could equate to 627,000 Euros as well as a huge personal cost.

Introduction of the IgM CLIA produced a much higher false positive rate. Of the 346 sera with an isolated IgM CLIA result, 277 (80.1%) were potentially false positive as they either did not confirm by immunoblot or were assessed as “Not consistent” with acute LB. This highlights that IgM CLIA testing in sera should only be used as part of a robust two-tiered testing protocol with confirmatory testing. The high false positive rate obtained for the IgM CLIA meant that a much higher proportion of samples required immunoblot testing: 22.9% of sera tested by CLIA (April to October 2020) were reactive with IgG and/or IgM CLIA, in contrast with the 8.5% of sera that were reactive with the IgG ELISA in the previous study period. This put extra pressure on the laboratory staff and greatly increased test costs (£28,933). It could be argued that the benefits of IgM CLIA are only marginal and perhaps not cost-efficient; however, 47 patients with probable or possible acute LB were detected, which may otherwise have been missed. Again, although unlikely, if all 47 were missed and progressed to disseminated LB/persisting symptoms, this could equate to 267,900 Euros.

Our results show that there is a significant risk of reporting of inaccurate and misleading results if patients are diagnosed on the basis of an isolated IgM positive result without consideration of clinical details, disease duration and pre-test probability. Prior studies have found that the majority of patients tested for Lyme serology did not meet European or UK clinical case definitions, and recommend that pre-test probability of infection is considered [5, 6, 20]. As a degree of seroprevalence for B. burgdorferi–specific antibodies exists in the population, over testing can lead to high false positive rates. The cost of misdiagnosing someone with acute LB based on false positive results, leading to their inappropriate, ineffective or even harmful treatment with antibiotics, and potentially delaying further investigations into the cause of their symptoms, should not be ignored.

The authors concede that the categorisation of patients into “Probable” or “Possible” LB in this study was flawed as there was limited clinical information available. For this reason, assay sensitivity, specificity and positive predictive values could also not be assessed. Many of the samples tested were not accompanied with adequate clinical details such as symptoms and date of onset. The authors accept this may have influenced the study false positive rate and whilst this may indicate that our testing protocols are suboptimal, this is the real-world scenario for a large number of laboratories where demand for Lyme testing is high and illustrates the challenges faced when interpreting results. The authors also recognise that the study results may have been influenced by the actual assays utilised. It is widely recognised that B. burgdorferi assays lack inter-assay consensus, particularly IgM assays [21, 22]. Interestingly, one of these studies calculated that a small loss of specificity led to an additional 192,716 immunoblot tests required (4,625,183 Euros), with an additional 6191 IgM false positive results.