The value of seroprevalence data as surveillance tool for Lyme borreliosis in the general population: the experience of Belgium

Infectious disease surveillance aims to assess the size of a health problem (disease burden), to identify high risk groups and/or areas to target interventions, to monitor trends and to detect outbreaks, in order to guide public health practice. In addition to routine disease surveillance through physicians and laboratories, serological surveillance, based on the measurement of the presence of specific antibodies in a given population, is frequently used to monitor levels of immunity to or presence of particular diseases within different age groups [1]. This is particularly relevant for vaccine-preventable diseases, for example in the context of measles and rubella elimination targets, where population based seroprevalence studies provide important data on gaps in population immunity and the potential for future outbreaks [2]. For other communicable (non vaccine-preventable) diseases, seroprevalence studies allow the measurement of the occurrence of the disease and associated risk factors. Serosurveys can also be used to assess the intensity of transmission of mosquito borne diseases and to measure the magnitude of an outbreak, like has been shown for dengue [3‐5]. In an overview of surveillance strategies for Lyme borreliosis by van den Wijngaard et al., serosurveillance is proposed as a possible surveillance scenario [6].

Lyme borreliosis is a multisystem infectious disease caused by infection with spirochetes of the B. burgdorferi s.l. complex. These spirochetes are transmitted to humans through the bite of infected ticks. The disease is the most common tick-borne disease in Europe [7]. Although persons of all ages are at risk for infection, surveillance data suggest that most cases occur in children and elderly persons [7]. Clinical manifestations of infection may include dermatological, rheumatologic, cardiac and/or neurological symptoms, but infection is often asymptomatic. In prospective studies, antibody reactivity to B. burgdorferi s.l. (IgG seroconversion) after a tick bite in people without clinical symptoms was observed in 2.9 to 3.7% persons [8‐12].

In Belgium, two sources contribute to routine surveillance of Lyme borreliosis. A network of sentinel laboratories performs laboratory surveillance by weekly reporting the number of positive serological tests for B. burgdorferi s.l.. And the yearly number of persons hospitalized for Lyme borreliosis is monitored through the hospitals’ minimum clinical datasets. In addition, the incidence of erythema migrans (EM) is estimated based on repeated studies carried out by a sentinel network of general practitioners. Up to 2017, none of the mentioned surveillance sources identified a significant increase in the incidence of Lyme borreliosis [13, 14].

The aim of this study was to evaluate the utility of seroprevalence data as an additional source of information on the epidemiology of Lyme borreliosis in Belgium.

Out of the 3215 specimens tested for B. burgdorferi s.l. IgG with CLIA testing, 3093 (96.2%) were seronegative, 41 (1.3%) had an equivocal result and 81 (2.5%) were positive (Fig. 1). Further testing of the equivocal and positive samples with immunoblot resulted in 30 positive specimens (0.9%) and 7 equivocal (0.2%).

The highest prevalences were observed in men and in those younger than 15 years and those older than 59 years of age (Table 1). However, these differences were not statistically significant.

Likewise, there were no statistically significant differences observed in prevalence by region nor by province, except for the province of Limburg (in eastern Belgium), where a higher prevalence was registered compared to the province of West Flanders (in western Belgium), 2.65% (95%CI 1.25–5.53) versus 0.44% (95%CI 0.00–2.33) (Table 1). In this case the 95% confidence intervals overlap but performing Pearson X² -test showed a statistical difference (p = 0.03) in estimated prevalence.

Seroprevalence data provide information on the cumulative incidence of exposure to B. burgdorferi s.l. in a certain geographical area. In Belgium, an overall weighted seroprevalence for B. burgdorferi s.l. of 1.06% (95%CI 0.67–1.67) was observed after two-tier testing. The weighted seroprevalence after the screening test was 3.91% (95%CI 3.08–4.96).

Our results are in line with those of a smaller study in blood donors in both a rural (n = 209) and an urban area (n = 193) in Belgium in 2011 [22]. In that study, 4.3 and 3.1% of the donors respectively presented a positive or equivocal result for Borrelia IgG based on CLIA testing (Liaison XL) only, compared with our estimate of 3.9% using the same test. A smaller study on 50 blood samples collected from healthy blood donors in 2007–2009 reported an IgG seroprevalence of 2% [18]. And in an older study in the early nineties on 1916 young military blood donors, the seroprevalence of borderline and positive results was 3.2%, using an in-house Elisa kit [23]. Our results are also comparable to an estimated 4.7% prevalence based on C6-ELISA only as was reported in a study in a healthy population (n = 836) in our neighbouring country the Netherlands [24]. The slightly higher estimate in the Netherlands is in line with a higher incidence of erythema migrans diagnosed by general practitioners, with 140 diagnoses per 100,000 inhabitants in 2014 [25] compared to 97.6 per 100,000 in Belgium per year in 2015–2017 [14].

Studies in other European countries using two-tier testing in various study populations report a wide variation of prevalence results, with estimates ranging from 1.6% in blood donors in Slovakia up to 9.4% in healthy adults in Germany [26‐30]. However, seroprevalence estimates should be compared with caution, as the representativeness of the population covered and laboratory testing methods differ between studies.

Although the difference in prevalence observed by sex is not statistically significant in our study, a higher seroprevalence in men is also observed in studies in other countries [27, 28, 30]. This is likely due to a higher exposure to ticks during professional and leisure activities. Likewise, although not statistically significant in our study, an increased risk for infection is generally observed among older persons (> 59 years of age), reflecting the population’s cumulative exposure to B. burgdorferi s.l. [27, 30].

Based on the results of Lyme borreliosis surveillance and the geographical occurrence of tick bites reported through an online citizen-based platform (TiquesNet), the risk of getting a tick bite or developing Lyme borreliosis in Belgium is higher in the provinces of Luxemburg, Flemish and Walloon Brabant, Limburg, Antwerp and Namur [13, 31]. This is not reflected in our study, where the prevalence was significantly higher in the province of Limburg compared to the western part of the country only. Although trying to have geographical representativeness by allocating a fixed number of specimens to each laboratory and further correcting crude prevalence estimates for cluster sampling, specific hot spots for Lyme borreliosis might not have been reached by sampling. This is not surprising, since the distribution of ticks and the pathogens they can be infected with are heavily influenced by spatial factors (landscape composition and configuration, forest and wildlife management, host abundance), that can have important local variation [32, 33]. Although Belgium is a small country, the topography is very different from West to East, with a coastal flat region gradually changing over a central plane to a heavily wooded region in the east and southeast of the country. Even within provinces, the landscape can be diverse, with focal spots of suitable habitat for ticks. The sample collection method used for the current serum bank (based on residual samples) might thus not be suitable to study the seroprevalence in the general population of diseases with strong environmental determinants.

Using residual samples also has some general limitations. The information available for each sample is basic and data on possible risk factors cannot be collected. In addition, there might be a possible selection bias: serum specimens submitted to diagnostic laboratories may not be entirely representative of the general population, in particular for a disease like Lyme borreliosis, most often contracted during physical activities. Studies on vaccine preventable diseases have shown that a convenience sample of sera produces similar estimates of immunity to those obtained from a sample collected using a randomized cluster design [34, 35]. To our knowledge, a similar comparison has not been performed for other diseases. To limit the possible bias in our study and exclude people with chronic diseases and immune disorders, only samples from selected medical wards have been included. Moreover, our results are in line with estimates of the prevalence of Borrelia antibodies in blood donors (both from a rural or urban area) in Belgium. Generally, blood donors are expected to be less prone than average to chronic disorders and autoimmune diseases, given the permanent and temporary exclusion criteria for blood donors, as well as other confounding factors related to lifestyle resulting in a selection of healthier and more health-conscious persons [36]. Therefore, we believe that an oversampling of an ill low-risk population is unlikely to have occurred.

Using seroprevalence data for surveillance of Lyme borreliosis specifically presents other limitations. The results only provide information on historical exposure to B. burgdorferi s.l. but not on the incidence of disease, as asymptomatic infections do occur [27, 28]. Also, although the humoral immune response to an infection with B. burgdorferi s.l. is often long lasting, the persistence of antibodies can vary widely, going from several months to many years [37‐39]. A seroprevalence of 1% is therefore not an exhaustive estimation of exposure to the bacteria.

Finally, serological tests for Lyme borreliosis have their own limitations, widely discussed in the scientific literature [18, 40‐42]. They include the complexity of the antigen composition of B. burgdorferi s.l., cross-reactivity of B. burgdorferi antigens leading to false positive results and the lack of a clear European consensus on criteria of the specific bands required to be positive among the second-tier commercial immunoblotting kits. In our study, both equivocal and positive results after the confirmatory immunoblot were considered as positives, since the specificity of being both ELISA positive and at least equivocal in second tier is considered very high. This is possibly different in the clinical situation where the confirmation of patients with active disease, especially Lyme arthritis and acrodermatitis, would preferably require more antigens to be positive in order to distinguish the natural background seroreactivity [43]. A review by Leeflang et al. observed an important heterogeneity in sensitivity and specificity of different commercial and in house serological assays for Lyme borreliosis in Europe [44]. And a study by Ang et al. on the influence of assay choice on the results in a two-tier testing algorithm for the detection of anti-Borrelia antibodies concluded that the choice of ELISA-immunoblot combination severely influences the number of positive results, making the exchange of test results between laboratories with different methodologies hazardous [42]. In order to allow comparison of the results of our study with previous (smaller) studies in Belgium, the same serology kit was used. The results of Busson et al. suggested high performance of Borrelia IgG screening (Liaison® Borrelia IgG) and confirmatory (recomLine Borrelia IgG) kits in Belgium [18].

The main conclusion of our study is that the overall exposure to B. burgdorferi s.l. in the general population in Belgium is low; 99% of the general population has no antibodies. This implies that in the absence of other evidence supporting a Lyme borreliosis diagnosis, a person presenting with non-specific or subjective symptoms for more than six weeks with a negative serologic test result in Belgium would strongly support an alternate diagnosis [43].