Selection of study participants
The study was based on samples collected from 188 patients clinically diagnosed with EM in Norwegian general practice during the years 2012–2013. The EM-patients were originally enrolled in a clinical trial. Thus, the inclusion criteria and study patient characteristics are presented elsewhere . A skin biopsy was taken from 149/188 patients at the time of inclusion (day 0). DNA from Bbsl was detected in 104/149 (69.8%) skin biopsies using real-time PCR . Serum samples from all 188 patients were collected at day 0, day 14 and after 3 and 12 months. Although samples were collected as part of a previous study, all analyses and results presented in this paper are new.
Real-time PCR for tick-borne microorganisms other than Bbsl was performed on 139 of the 149 skin biopsies already analyzed for Bbsl DNA (Fig. 1a) . Ten of the 149 biopsies contained an insufficient amount of material for further PCR analysis. Real-time PCR (conventional PCR for Babesia spp.) was also performed on the inclusion (day 0) serum samples of patients reporting fever during the initial 14 days (8/188) and patients with an EM duration of ≥ 21 days (69/188). This was done as fever is relatively uncommon in patients with EM but is often reported in infections caused by other tick-borne microorganisms. Two patients reported both fever and an EM duration ≥ 21 days.
Most of the molecular analyses were performed at Linköping University, Linköping, SE. A summary of the molecular methods is presented in Table 1a. Total nucleic acids were extracted from the patient specimens and turned into cDNA, as described elsewhere . Different real-time PCR assays were used to detect N. mikurensis, Rickettsia spp., A. phagocytophilum and Babesia spp. Additional PCR analyses were performed on the inclusion (day 0) samples of patients displaying a fourfold or higher rise in antibody titers against SFG Rickettsia spp., A. phagocytophilum or Babesia microti. The molecular analyses of sera for Babesia spp. was done using conventional PCR at Statens Serum Institut, Copenhagen, DK, using the same primers and target gene as the real-time Babesia assay, as described elsewhere .
Real-time PCR assays
Detection of Rickettsia spp. was done using a TaqMan real-time PCR assay, as previously described . The primers CS-F and CS-R, and probe CS-P are designed to target the Rickettsia spp. citrate synthase gene (gltA) to amplify a 74-bp long amplicon (Table 1). As a positive control, a synthetic plasmid containing the target sequence of the TaqMan real-time PCR assay was used. The plasmid contained the target sequence, spanning the nucleotides 1102–1231 of the Rickettsia rickettsii gltA gene (GenBank: U59729), synthesized and cloned into pUC57 vector (GenScript). The assay has a limit of detection of 10 copies of the plasmid per reaction.
Detection of A. phagocytophilum was done using a TaqMan real-time PCR assay, as previously described . The primers ApF and ApR, and the probe ApM are designed to target the A. phagocytophilum citrate synthase gene (gltA) to amplify a 64-bp long amplicon (Table 1). As a positive control, a synthetic plasmid containing the target sequence of the TaqMan real-time PCR assay was used. The plasmid contained the target sequence, spanning the nucleotides 304–420 of the A. phagocytophilum gltA gene (GenBank: AF304137), synthesized and cloned into pUC57 vector (GenScript, Piscataway, NJ, USA). The assay has a limit of detection of 30 copies of the plasmid per reaction.
Detection of N. mikurensis was done using a TaqMan real-time PCR assay, as previously described . The primers and the probe are designed to target the N. mikurensis groEL gene to amplify a 169-bp long amplicon (Table 1). As a positive control, cDNA samples positive for N. mikurensis confirmed by sequencing in an earlier study  were used in each run.
Detection of Babesia spp. in the skin biopsies was done using a SYBR green real-time PCR assay, as previously described . Primers BJ1 and BN2 are designed to target the Babesia 18S rRNA gene to amplify a 411–452 bp long amplicon (Table 1). As a positive control, a synthetic plasmid containing the target sequence of the SYBR green real-time PCR assay was used. The plasmid contained the target sequence, spanning the nucleotides 467–955 of the B. divergens 18S rRNA gene (GenBank: AJ439713), synthesized and cloned into pUC57 vector (GensSript). The assay has a limit of detection of 10 copies of the plasmid per reaction.
Samples were analyzed using both indirect immunofluorescence antibody (IFA) assays and ELISA assays as shown in Fig. 1b. A summary of the serological assays used is presented in Table 1b. All samples were examined using current serological “gold standards” with high reported sensitivities and specificities [25,26,27]. Of the 135 patients who were serologically analyzed for other pathogens than Bbsl, 105 (78%) had their skin biopsies undergo molecular analysis using real-time PCR (Table 1a).
135 of the 188 patients had sufficient sera for further serological analyses using indirect IFA assays (Fig. 1b). Sufficient 3-month sera were available for 126 of these patients and were screened for the presence of IgG and IgM antibodies against SFG Rickettsia and A. phagocytophilum as well as IgG antibodies against B. microti. In patients missing 3-month sera, the day 14 sera (n = 8) and day 0 sera (n = 1) were screened instead (Fig. 1b).
The serological analyses for SFG Rickettsia, A. phagocytophilum and B. microti were done at SSI, Copenhagen DK, using commercially available IFA assays (Focus Diagnostics, Inc., Cypress, CA, USA). All samples were analyzed as part of the same experiment, exclusively for the present study. Samples were prepared according to the manufacturer’s instructions and all IFA samples underwent two-fold dilutions until end-point fluorescence, as described in a previous study . Samples were read in a dark room by two independent, experienced microscopists using a fluorescence microscope to establish the titer level. In cases of disagreement between the two microscopists, the sample was discarded and prepared again. Patients with detectable IgG antibodies in the screening sample had the remaining study samples further tested for antibodies against the corresponding microorganisms. The time of inclusion was chosen as baseline and we interpreted a fourfold rise in IgG antibodies at any of the follow-up visits (day 14, 3 months, 12 months) as serological evidence of recent exposure for all IFA assays. Patients with only weak antibody reactivity as indicated by a rise in antibody level from < 1:64 to 1:64 were excluded in the analyses. Further, a high IgG titer of ≥ 1:512 (clinical cut-off at SSI) or ≥ 1:128 (one titer above manufacturers recommendation) in at least one of the samples was considered evidence of prior exposure to SFG Rickettsia and A. phagocytophilum, respectively.
Day 14 and 3-month sera were tested for IgG antibodies against TBEV (187/188) and Bbsl (175/188), respectively, using ELISA.
For TBEV, the analysis was done using a commercially available IgG ELISA assay (Virion/Serion, Würzburg, Germany) at the Norwegian Institute of Public Health (NIPH), Oslo NO. We chose to analyze the day 14 samples for TBEV as the time from tick bite or EM was, at this point, expected to be sufficient for any seroconversion to have occurred. Equivocal results were retested using the same assay. Positive tests were reanalyzed with a different ELISA assay according to standard operating procedure (Euroimmun Anti-TBE Virus IgG, Lübeck, Germany).
The Bbsl analysis was performed at Sørlandet Hospital, Kristiansand NO, using a commercial ELISA kit (Enzygnost Borrelia, Lyme link VlsE/IgG), as described elsewhere . At the time of inclusion, 50.0% of the patients (94/188) had detectable IgG antibodies against Bbsl  and to assess any further seroconversion, the 3-month sera were analyzed. For both Bbsl and TBEV, the IgG cut-off levels were set in accordance with the manufacturers’ recommendations.
Data on concomitant symptoms was obtained through a patient diary kept for the first 14 days after inclusion . The following 16 symptoms were inquired upon: tiredness, headache, arthralgia, neck stiffness, fever, palpitations, myalgia, sore throat, tender skin, dizziness, nausea, chest pain, diarrhea, chills, hot flushes, and coughing. Further, clinical data on the size, duration, diagnostic accuracy and appearance of the EM lesions were collected through clinical examination of the patients at days 0 and 14.
For comparing proportions, we used a chi-squared test (Χ2) or a Fisher’s Exact Test when the expected number in one or more cells in the crosstabs was < 5. Means were compared with a t-test and for the continuous variables we used a Mann–Whitney U test. Missing data are considered to be randomly distributed. P-values ≤ 0.05 were considered statistically significant. Analyses were performed using IBM SPSS Statistics for Windows (v. 25; IBM Corp., Armonk, NY, USA).