Aims: This study evaluates the use of local Borrelia burgdorferi sensu stricto and Borrelia afzelii strains in a single mixed antigen for in-house IgG western blots in the routine diagnostic setting by comparing it with the current in-house protocol.
Methods: Sera from 233 patients from areas of Scotland with low and high prevalence for Lyme borreliosis were tested by western blots prepared from reference strain antigen (B burgdorferi sensu stricto) and mixed antigen (B burgdorferi sensu stricto and B afzelii). Results were scored using original and revised criteria, and results were compared.
Results: The mixed antigen produced significantly more bands than the reference antigen. Using the original interpretation criteria the mixed antigen produced more positive results than the reference antigen (90 versus 85). When the revised criteria were applied to the mixed antigen there were 14 more patients with positive results than with the reference antigen (99 versus 85); this difference was significant. Although 22 patients were positive with the mixed antigen and revised criteria, but negative/equivocal with the reference antigen, eight patients who were positive with reference antigen remained negative with the mixed antigen. The positive predictive value of the two antigen preparations was the same (96%). The negative predictive value of the mixed antigen with revised criteria was higher than the reference antigen (96% versus 88%), but the specificity was similar (97% versus 98%).
Conclusions: The mixed antigen and revised interpretation criteria have been successfully incorporated into the routine diagnostic testing service, increasing the sensitivity of the in-house IgG western blot test for Scottish patients.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Local Borrelia isolates increase the sensitivity of western blots.
Diagnostic criteria for interpretation of western blots need to be devised for local populations.
Patients infected in different countries may require testing with reference antigens.
The serological diagnosis of Lyme borreliosis can be problematic.1 We previously published a method to mix two species of Borrelia (one strain each of B burgdorferi sensu stricto and Borrelia afzelii) as a mixed antigen in a single in-house IgG western blot preparation.2 Our preliminary results were encouraging. The mixed antigen appeared to be more sensitive than the single reference strain B burgdorferi sensu stricto antigen, which had been in routine use. However, the mixed antigen western blots were developed using western blot interpretation criteria based on the reference strain B burgdorferi sensu stricto antigen.3 Although interpretation criteria in Europe vary with strain and laboratory, a noticeable difference between B burgdorferi sensu stricto and B afzelii criteria is the presence of a B afzelii specific 43 kDa protein.4 5 As B afzelii was added into our western blot antigen, we believed that making the 43 kDa protein a specific band in our criteria would improve sensitivity. This study fully evaluates the use of the local B burgdorferi sensu stricto and B afzelii strains as a mixed antigen and the value of the revised criteria compared with the original criteria.
MATERIALS AND METHODS
Serum samples are referred to the Lyme Borreliosis Testing Laboratory, Raigmore Hospital, Inverness, UK, from throughout Scotland. All sera were tested following the internationally recognised two-step testing procedure.5 6 Samples were screened by enzyme immunoassay (EIA) for the detection of B burgdorferi total antibody (Zeus Scientific, Raritan, New Jersey, USA). All EIA-positive/equivocal samples, and EIA-negative samples with a high clinical suspicion of Lyme borreliosis were then confirmed by in-house IgG western blot as described previously.7 Western blots were prepared from reference strain B burgdorferi sensu stricto (Hb1) antigen or from locally isolated B burgdorferi sensu stricto and B afzelii 50:50 mixed antigen as described previously.2
Two groups of patient serum samples were tested. The first group consisted of sera from 133 patients that were tested prospectively. Prospective testing was performed on all samples received at the laboratory from January to May 2007. Each sample was tested with the two western blot preparations in parallel. Only the first sample from each patient tested during the study period was included in the study. The second group of patient samples were selected from stored sera previously tested outside the study period by only one western blot preparation. Fifty samples had been tested using the reference strain antigen (25 western blot positive, 25 western blot negative); and 50 samples had been tested using the mixed antigen (25 western blot positive, 25 western blot negative). The western-blot-positive and western-blot-negative groups for each antigen were age/sex matched as much as possible. The 50 samples tested with the reference antigen western blots were retested using the mixed antigen, and those previously tested with the mixed antigen western blots were retested with the reference antigen.
The reference antigen western blot results were scored according to the original laboratory western blot interpretation criteria.3 The mixed antigen western blot results were scored using the original laboratory western blot interpretation criteria (as above), and the western blot interpretation criteria revised to include the 43 kDa protein as a specific band. Demographics and clinical details for each patient were collated. The results from all patients were compared and statistical analysis performed where appropriate using a two-tailed t test, Fisher’s exact two-tailed test and McNemar test. The specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated using established formulae; it was assumed that patients who were negative for both antigens were true negatives.
In total, serum from 233 patients (prospective and stored samples) were tested by in-house IgG western blots prepared from reference and mixed antigens, and scored using the original and the revised interpretation criteria (mixed antigen only) (table 1). Of the 233 patients tested, 124 (53%) were from the Scottish Highlands, and the remaining 109 were from other areas of Scotland.
The mixed antigen produced significantly more bands than the reference antigen (p = 0.00194, two-tailed t test). The B afzelii specific 43 kDa band was detected in significantly more mixed-antigen-positive western blots than negative/equivocal western blots (74/90 versus 47/143; p<0.0001, Fisher’s exact two-tailed test).
Using the original interpretation criteria, the mixed antigen produced more positives than the reference antigen (90 versus 85) (table 1). When the revised interpretation criteria were applied to the mixed antigen there were significantly more positives than with the reference antigen (99 versus 85; p = 0.0176, McNemar test) (table 1). Therefore, in total there were 14 more patients who were western blot positive with the mixed antigen and revised criteria than with the reference antigen. However, these 14 patients were made up of: 22 patients who were positive with the local mixed antigen and revised criteria, but negative/equivocal with the reference antigen; eight patients who were negative with the mixed antigen but reference antigen positive. The clinical details of these 30 patients were examined further. Eighteen of the 22 (81.2%) patients positive with the mixed antigen but negative/equivocal with the reference antigen had symptoms consistent with Lyme borreliosis (16 early, 2 late). There was insufficient clinical information provided to assess the other four patients. Of the eight patients who were reference antigen western-blot positive but not detected by the mixed antigen and revised criteria, five had symptoms consistent with a clinical diagnosis of early Lyme borreliosis, again indicating that these were true positives. Interestingly, two of these five patients received tick bites abroad (Norway and Canada). The other three patients had non-specific symptoms.
The PPV of using the reference antigen and mixed antigen with revised criteria was calculated to be the same (96%). The NPV of the mixed antigen with revised criteria was higher than the reference antigen (96% versus 88%), but the specificity was similar (97% versus 98%).
Interstrain and intrastrain differences in borrelia protein expression are widely recognised, with some individual borrelia proteins from different species reported to have sequence heterogeneity up to 40%.1 8 9 Therefore, different western blot interpretation criteria for each of the three pathogenic species of Borrelia present in Europe (B burgdorferi sensu stricto, B afzelii and B garinii) are widely used.1 4 10 As we included two species of Borrelia (B burgdorferi sensu stricto and B afzelii) in the same western blot antigen it was necessary to have interpretation criteria that would account for both. The 43 kDa band was detected significantly more in the mixed-antigen-positive blots than in the negative blots. This confirmed our hypothesis that the 43 kDa band should be included in our criteria as it indicated specificity for B afzelii, as found by other groups.4 5
The mixed-antigen western blots produced significantly more bands and detected more positives than the reference antigen (p = 0.00194). This increased sensitivity became significant when the revised criteria were applied. Increased sensitivity may be due to the use of an additional strain, and the fact that the strains used were local. Although there is an element of cross-reaction between strains, the detection of infection caused by a particular strain can only be enhanced by using the same strain in the diagnostic assay. Likewise, local strains have been found to be more reactive with study sera, presumably as a result of intrastrain differences in protein expression.8 10 11 The increased sensitivity with the mixed antigen and revised criteria is extremely important as it may have led to a change in clinical management in the 22 patients that were positive with the local mixed antigen but negative/equivocal with the reference antigen. Crucially, specificity did not appear to be compromised using either the mixed antigen or the revised criteria, as in the majority of these patients (18/22) the positive result was consistent with the clinical diagnosis of Lyme borreliosis. The high specificity of the two antigen preparations is alluded to by their high PPV, and the fact that there were so few patients with false-positive results (three with the reference antigen, and four with the mixed antigen and revised criteria). In addition, a large number of patients (n = 126) remained negative with both antigen preparations and interpretation criteria. These patients had a range of clinical features and were from different areas of Scotland (56/126 from the Scottish Highlands, and 70/126 from other areas of Scotland). Thus, the specificity of the two antigen preparations is comparable.
It is important to note that, although the overall sensitivity of the western blot was increased by using the mixed antigen and revised criteria, there were eight patients for whom the serological diagnosis of B burgdorferi infection would not have been made. Again, the majority of these patients (5/8) had a clinical diagnosis of Lyme borreliosis. As noted, two patients had been exposed abroad. As the strains used in the mixed antigen are local, it is highly possible that they may have reduced reactivity with some infections acquired outside Scotland, again due to intraspecies differences. The reference strain antigen had a better performance with these patients than the mixed antigen and may therefore be more suitable for patients exposed abroad. Currently, in patients who remain seronegative and in whom Lyme borreliosis is highly suspected, we recommend testing with mixed antigen and reference antigen preparations. This practice will be specifically useful for patients who may have been infected abroad.3
This study has confirmed the importance of using local mixed antigens and the need to revise interpretation criteria when introducing new antigens into the western blot. In addition, an important finding was that the reference antigen should be used in certain clinical situations. Overall, our results confirm the complexity of the serological diagnosis of Lyme borreliosis, and improve patient management.
We would like to thank all the users of the National Lyme Borreliosis Testing Laboratory for their help and support; their co-operation was essential for this study. We would also like to thank Ruth Seaton and Edwina Kehoe for their invaluable help with testing and data entry.
Competing interests: None.