Analytical comparison of 11 CE-IVD marked real-time PCR assays for Borrelia burgdorferi sensu lato detection

Abstract

Purpose

In some precise clinical manifestations of Lyme borreliosis, PCR can be helpful for diagnosis: suspicion of atypical erythema migrans, borrelial lymphocytoma, acrodermatitis chronica atrophicans, Lyme arthritis or in very early seronegative neuroborreliosis. This study aimed to evaluate: (1) the analytical sensitivity of commercial real-time PCR kits for Borrelia burgdorferi sensu lato (Bbsl), (2) their specificity, and (3) their practicability.

Methods

The performances of 11 CE-IVD marked commercial kits and the French NRC Borrelia’s in-house PCR were evaluated. PCRs were performed blindly on quantified DNA extracts, from 14 Bbsl strains corresponding to eight different genospecies, and on DNAs from relapsing fever (RF) Borrelia, Treponema, Leptospira, and 26 other micro-organisms.

Results

DNAs from Bbsl strains ranging from 10 to 10⁴ copies/5µL were detected by all the PCR kits evaluated, with the exception of B. spielmanii, which was not detected by one kit. Three kits had a higher limit of detection than the in-house PCR (P-value < 0.05). RF Borrelia generated positive PCR reactions for the majority of kits tested (9/11). The kits were generally easy to use, in terms of equipment, explanation of technical procedures, and realization in a laboratory competent in molecular microbiology whereas interpretation of results was moderately difficult to difficult to do for 9/11 kits.

Conclusion

This study shows that the analytical performance and practicability of the kits evaluated are broadly in line, but that advanced molecular biology skills are required to interpret the results. Simultaneous detection of Lyme Borreliosis and RF agents without precise distinction may result in a risk of confusion in the choice of biological matrices and in the wording of the result returned to the clinician. Diagnostic laboratories must pay particular attention to the amplification confirmation strategy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10096-025-05214-5.

Introduction

Lyme borreliosis is the most common human tick-borne disease occurring in the Northern hemisphere. In Europe, this zoonosis is caused by several spirochetes belonging to the Borrelia burgdorferi sensu lato (Bbsl) group which are transmitted by bites of Ixodes ricinus ticks. The Bbsl group is a complex of about 20 bacterial species including the four major pathogenic species: B. afzelii, B. garinii, B. bavariensis and B. burgdorferi sensu stricto (Bbss), some species less frequently encountered in clinical cases (B. bissettiae, B. spielmanii, B. lusitaniae, B. mayonii) and, other species not having yet proven pathogenicity for humans (e.g., B. valaisiana) [1, 2]. Lyme borreliosis is characterized by a localized skin manifestation, erythema migrans (EM), but in the absence of appropriate treatment after the initial infection, there is a risk of early (< 6 months) or late (> 6 months) disseminated infection, with some preferential tropisms: neurological (neuroborreliosis), articular (Lyme arthritis) and/or cutaneous (borrelial lymphocytoma, acrodermatitis chronica atrophicans-ACA) [3].

The laboratory diagnosis of Lyme borreliosis currently relies mainly on serological methods. Indeed, except in the very early clinical presentations, these indirect methods of detection are typically more sensitive than direct detection by PCR of Bbsl spirochetes in tissues and fluids due to low numbers of residing bacteria [4]. The diagnostic sensitivity of Bbsl PCR methods is depending on matrix type and clinical presentations: PCR sensitivity is medium to fairly good in synovial biopsies and joint fluids (36–100% [5, 6]), and in skin biopsies from erythema migrans (35–81% [5]) or acrodermatitis chronica atrophicans (16–92% [5]), and is low in CSF collected from early Lyme neuroborreliosis (5–30% [7–9]).

A Bbsl PCR positive will support a proven diagnosis in some atypical EM, borrelial lymphocytoma, ACA, very early neuroborreliosis and Lyme arthritis [5, 6, 10, 11]. Thus, the use of a sensitive as well as very specific PCR method, exhibiting no significant cross reaction is required to perform an accurate laboratory diagnosis. Several real-time (rt) PCR kits for Bbsl DNA detection in clinical samples are currently marketed in the EU. The assay performances displayed in the kit’s information inserts have not been recorded following the same methodology, which makes it difficult to compare the claimed data.

Our study aimed to evaluate (i) the analytical sensitivity of Bbsl rt PCR kits marketed in France for the different genomic species of Bbsl, (ii) the specificity of methods for other micro-organisms transmitted or not by tick bite, (iii) and the user practicability of kits. The analytical performances of 11 rt PCR kits have been compared with each other and to the in-house method used in the French National Reference Center for Borrelia. Importantly, the comparison was based on identical templates DNA thus avoiding differences that may occur through PCR upstream processes.

Materials and methods

Preparation of samples

Concentration ranges of genomic DNA of 14 cultivated Bbsl strains were prepared by the German National Reference Center for Borrelia. For each strain, the DNA concentration ranged from 10^− 1, 10⁰, 10¹, 10², 10³ and 10⁴ genome equivalents (GE)/ 5µL. The dilution buffer contained human thymus DNA at a concentration of 2 ng/µl to prevent DNA loss at low concentrations. Included isolates were: B. afzelii (PKo and PVPM), B. garinii (PBr (OspA type 3), PHei (OspA type 5), PLa (OspA type 8), PRef (OspA type 7) and PWudll (OspA type 6)), B. bavariensis PBi, Bbss (B31 and PBre), B. bissettiae PGeb, B. spielmanii PSigII, B. lusitaniae PotiB2 and B. valaisiana VS116. Culture of Bbsl strains, extraction and preparation of DNA samples were performed as previously described [12].

Other purified DNA extracts were also tested in the same conditions to assess the specificity of PCR tests (suppl. Table S1), i.e., six DNA samples from spirochetes outside the Bbsl group (including three RF borreliae: B. miyamotoi, B. hermsii and B. recurrentis), 28 DNA samples from other micro-organisms of bacterial, fungal and parasitic origins (including various agents of tick-borne diseases), and two of human origin.

All DNA samples were stored at -80 °C until use. They were thawed once to be aliquoted and once at the time of analysis.

Kits included

The analytical performances and user practicability of 11 rt PCR kits were compared with each other and to our in-house method used in the French National Reference Center for Borrelia. The list of included kits as well molecular targets and PCR thermocyclers employed are given in Table 1. The in-house rt Bbsl PCR was performed using two primers (forward-FlaBbsl 2a and reverse-FlaBbsl 4c) and two Taqman^® probes targeting a conserved region of the flagellin gene flaB, on the LightCycler 480 (Roche) analyzer (see more detailed information in [13]).

Table 1.

Overview of commercial and in-house PCR methods for Bbsl detection included in the study and PCR thermocyclers. The molecular targets were obtained from kit inserts at the time of the study

real-time PCR methods	Manufacturer (country)	Molecular target	PCR thermocycler used in this study
1-Fluorotype® Borrelia PCR	Hain Lifescience (Germany)	Not specified in the kit insert	FluoroCycler (Hain)
2-EliGene® Borrelia UNI (flagellin)	Elisabeth Pharmacon (Czech Republic)	Flagellin gene? (not specified in the kit insert but in the kit name)	LightCycler 480 (Roche)
3-Geneproof Borrelia burgdorferi PCR	Geneproof (Czech Republic)	16 S rRNA gene	LightCycler 480 (Roche)
4-Attomol Borrelia burgdorferi-Realtime LT*	Attomol (Germany)	Flagellin gene	LightCycler 2.0 (Roche)
5-RealLine Borrelia burgdorferi sl, Fla Format	Bioron Diagnostics (Germany)	Not specified in the kit insert	LightCycler 480 (Roche)
6-BactoReal® Borrelia burgdorferi sl	Ingenetix (Austria)	Lipoprotein OspA gene	ABI 7500 (Thermofisher)
7-Borrelia burgdorferi Real-TM	Sacace Biotechnologies (Italy)	16 S rRNA gene	ABI 7500 (Thermofisher)
8-Viasure Borrelia / Anaplasma / Coxiella Real Time PCR detection	Certest Biotec (Spain)	Not specified in the kit insert	ABI 7500 (Thermofisher)
9-Diarella Borrelia	Gerbion (Germany)	Not specified in the kit insert	ABI 7500 (Thermofisher)
10-Alphacube Borrelia	Gerbion (Germany)	Not specified in the kit insert	ABI 7500 (Thermofisher)
11-Mutaplex® Borrelia real time PCR*	Immundiagnostik (Germany)	Not specified in the kit insert	LightCycler 480 (Roche)
12-In-house PCR (French NRC for Borrelia)	-	Flagellin gene	LightCycler 480 (Roche)

*Since the implementation of the protocol, we have been informed that the kits Attomol and Mutaplex (Immundiagnostik) are no longer marketed

Assay implementation

Real-time PCR assays were performed according to the study protocol “reagents for biological diagnosis of Lyme borreliosis by PCR” (ANSM/CNR 22/05/2018). Four different thermocyclers have been used depending on the systems validated by the manufacturers: LightCycler 480 (Roche), ABI 7500 (Thermofisher), LightCycler 2.0 (Roche) or FluoroCycler (Hain) (see Table 1).

Manipulations were performed following the protocol of instruction for use provided in the kit’s information inserts and the material recommended (thermocycler, consumables). For all assays, the test sample was 5µL of DNA extract. For the four kits where this volume was lower than those recommended in the instructions (kits No. 1, 3, 5, 7 of the Table 1), we adjusted the volume with molecular grade water. For the three kits for which a test sample volume of 5µL was higher than that indicated in the instructions, we were specifically authorized by the manufacturers to adapt the protocol, with an increase of the test sample (5µL instead of 4µL for kits No. 9,10 and 11 of the Table 1). Positive and negative controls were included in each assay according to manufacturer’s recommendations. All assays have been carried out twice for each sample and for each kit, i.e., in two different series of manipulation according to the same testing procedure.

Limit of detection, reproducibility, linearity and efficiency

The detection of Bbsl DNA in purified extracts using a range of concentration from 10⁴ to 10^− 1 GE per reaction allowed to evaluate the analytical sensitivity of the kits. This is defined here as the lowest concentration of DNA for which the two duplicates of PCR (in two series) are positive; this is also called “limit of detection” (LOD). The lowest DNA concentration enabling positivity of one only of the two PCRs was also considered. The non-detection of the DNA from other micro-organisms or human will make it possible to check the analytical specificity of the kits.

Reproducibility was evaluated by delta cycle threshold (Ct) calculation corresponding to the absolute difference between two analytical replicates performed in independent series of analysis. These calculations were not possible for Hain Lifescience kit that did not display the PCR Ct value; for this kit, we only performed a qualitative evaluation of repeatability.

For each PCR method displaying Ct values and each Bbsl strain its semi-logarithmic, we studied regression line Y = aX + B; where (i) Y is the Ct value displayed by the analyzer, (ii) “a” is the slope, (iii) X is the DNA concentration expressed in log GE per 5µL (excluding the lowest concentration), and (iv) B is the intersection at the origin. From these representations, we extracted the slope and Pearson’s correlation coefficient (r²) values, and calculated the amplification efficiency (E) values of PCR assays using the formula E = 10^−(1/slope) – 1 [14].

Practicability

The practicability of rt PCR kits was evaluated according to the following criteria: (1) adequate equipment provided with the kit, (2) quality of description of material not provided, (3) precise description of procedures, (4) presence of positive and negative controls, (5) ease of realization and (6) ease of interpretation. Scores were then attributed for each criterion from 1 (not planned or very poorly adapted) to 4 (very well adapted).

Statistics

All quantitative variables were compared using Wilcoxon signed rank test for paired data, wherein a p-value ≤ 0.05 was considered significant. All these computations were performed using the R project software (http://www.r-project.org). Exact confidence intervals of a frequency, i.e., binomial confidence interval, were determined for the percentage by using the web server http://statpages.info/confint.html.

Results

PCR detection, reproducibility, linearity and efficiency

Borrelia DNA of the samples containing between 10¹ and 10⁴ GE/5µL was detected for all strains by all evaluated kits as well as by the in-house PCR, with the exception of the B. spielmanii strain which was not detected by the BactoReal kit, whatever the DNA concentration. At these concentrations ranging from 10¹ to 10⁴ GE /5µL, results were positive in both analytical replicates, except for B. lusitaniae 10² GE /5µL, B. garinii PHei 10¹ GE /5µL and B. bavariensis 10¹ GE /5µL, with Sacace, Viasure and Geneproof kits, respectively. The Ct values obtained and the delta Ct between replicates are presented for each PCR method and for each strain tested in Suppl. Table S2.

Regarding the reproducibility of positive PCRs, results were more frequently inconsistent between the two replicates for the least concentrated samples (10^− 1 to 10⁰ GE /5µL): 39% (17-64_CI95%) of the results for the BactoReal kit, 44% (22-69_CI95%) for the Hain kit, 47% (24-71_CI95%) for the Attomol kit, 50% for the in-house method (26-74_CI95%) and the Sacace kit (28-72_CI95%), 60% (36-81_CI95%) for the Eligene and Geneproof kits, 67% (22-96_CI95%) for the RealLine kit, 72% (47-91_CI95%) for the Viasure kit and 75% (19-99_CI95%) for the Mutaplex kit; with the exception of the Diarella and Alphacube kits which, respectively, had no or only one positive reaction (Suppl. Table S2).

At the lowest DNA concentrations (10^− 1 to 10¹ GE /5µL) and in case of congruence of positive results, delta Ct were quite often ≥ 1,00 between the two replicates (Suppl. Table S2): 18% (5-40_CI95%) of the observed results for the Eligene kit, 19% (5-42_CI95%) for the Geneproof kit, 21% (7-42_CI95%) for the Attomol kit, 26% (10-48_CI95%) for the in-house PCR, 27% (8-55_CI95%) for the Diarella kit, 44% (20-70_CI95%) for the Sacace kit, 46% (26-67_CI95%) for the BactoReal kit, and 50% for the Alphacube (23-77_CI95%) and Viasure (26-74_CI95%) kits; except for the Mutaplex kit with 0% of delta Ct ≥ 1.00 (0-22_CI95%).

Conversely, delta Ct > 1.00 were rare for the higher concentrations (10² to 10⁴ GE /5µL), with less than 10% of samples for the different PCR methods, with the exception of Bactoreal (24%) and Viasure (52%) kits whose reproducibility appears less good in our study.

The r² values were high for each method providing Ct value, with mean values ranging from 0.956 to 0.993 corresponding to the Diarella and Eligene kits respectively. These values close to 1 reflect a good linearity of the PCR methods studied (Suppl. Table S3). The efficiency of these different PCR methods was also satisfactory with average values ranging from 86% (RealLine kit) to 122% (BactoReal kit), with the exception of the Mutaplex kit with an average efficiency of 145% (Suppl. Table S3).

Analytical sensitivity assessment

LOD values, as well as the lowest DNA concentration enabling positivity for one PCR of two are presented in Fig. 1. In our study, LOD values generally ranged from 1 to 10 GE per reaction for the different PCR tests evaluated, regardless of the kit and the Bbsl strain studied (Fig. 1; Table 2). The orders of magnitude of the measured values corresponded to the values displayed in the kit’s information inserts, with the exception of the Ingenetix kit which appears slightly more sensitive than announced (Table 2).

Fig. 1.

(1 A-1 L) Individual values of limit of detection (LOD) and lowest positive value of the tested strains for each PCR method. Each radar chart corresponds to a PCR method evaluated (A to L). The rays correspond to the results obtained for each Borrelia burgdorferi sl. strain. The central axis represents in logarithmic format the LOD values (in orange) and the lowest positive value (in blue), in genome equivalent/5µL

Table 2.

Limit of detection (LOD) claimed and measured for the 12 real-time PCR methods for the detection of Borrelia burgdorferi Sl

real-time PCR method/molecular target	Claimed LOD (per reaction)	Measured LOD (per reaction)	Comparison with in-house PCR LOD (P-value of Wilcoxon signed rank test)	Lowest positive value (per reaction)
Fluorotype (Hain)/unknown target	3 copies	1–10 copies	0.26	0.1–10 copies
EliGene (Elisabeth Pharmacon)/flagellin gene	1–10 copies	1–10 copies	1	0.1-1 copies
Geneproof/16S rRNA gene	5.3 copies	0.1–100 copies	0.95	0.1-1 copies
Attomol /flagellin gene	< 10 copies	1–10 copies	0.82	0.1-1 copies
RealLine (Bioron) /unknown target	≤ 100 copies	1–10 copies	0.06	1–10 copies
BactoReal* (Ingenetix)/ospA gene	42 copies	1–10 copies*	0.09	0.1-1 copies*
Sacace/16S rRNA gene	5 copies	0.1–10 copies	0.43	0.1-1 copies
Viasure (Certest) /unknown target	10 copies	0.1–10 copies	0.34	1–10 copies
Diarella (Gerbion) /unknown target	10 copies	1–10 copies	0.026	1–10 copies
Alphacube (Gerbion) /unknown target	10 copies	10 copies	0.015	10 copies
Mutaplex (Immundiagnostik) /unknown target	10 copies	1–10 copies	0.042	1–10 copies
In-house PCR/flagellin gene	n/a	0.1–10 copies	n/a	0.1-1 copies

*Non-detected strain excluded from analysis for this kit

The sole higher LOD exception (100 GE per reaction) was obtained with the Geneproof kit for the B. bavariensis PBi strain, but the lowest DNA concentration enabling positivity of one PCR was lower: 1 GE per reaction (Fig. 1). The exceptions of very low LODs obtained during evaluation, i.e., 0.1 GE, concerned the Sacace kit for B. lusitaniae PotiB2 and B. garinii PLa strains, the Geneproof kit for B. afzelii PKo strain and the in-house method for B. bissettiae strain (Fig. 1).

Among the 14 strains of Bbsl tested, five PCR kits showed LODs mostly equal to 10 GE per reaction for a majority of strains (Certest 9/14; Bioron 12/14; Mutaplex 13/14; Diarella 13/14; Alphacube 14/14), while six other kits had LODs predominantly equal to or less than 1 GE per reaction for a majority of strains (Eligene 8/14; Geneproof 9/14; Attomol 9/14; Sacace 9/14; Hain 10/14; Ingenetix 11/14). The in-house PCR method showed seven LODs at 10 GE per reaction and seven LODs less than or equal to 1 GE per reaction, varying according to the strains tested.

Comparison of the LODs paired by strain of each PCR kit, with the in-house method taken as a reference, found a statistically higher median of LOD for the Diarella, Alphacube and Mutaplex kits (P-values < 0.05; Table 2).

Analytical specificity

Table 3 presents the results obtained for the three RF Borrelia species tested using the 12 PCR methods, as well as information from the PCR kit’s information insert on the reactivity or lack of reactivity of the test for the different Borrelia species concerned. DNA of B. miyamotoi, B. hermsii and B. recurrentis strains generated a positive PCR reaction for respectively 36%, 64% and 9% of the kits tested (Table 3). All kits evaluated showed at least one cross-reaction among the three RF Borrelia DNAs, except the BactoReal and RealLine kits. These two latter kits target ospA and an unknown gene, respectively. The in-house method, targeting the flagellin gene, showed no cross-reactions with the three RF Borrelia DNAs tested. The Hain and Viasure kits were the most cross-reactive kits, with two and three RF Borrelia detected respectively out of three. Out of 13 experimentally observed cross-reactions, only six (46%) were mentioned in the kit’s inserts (Table 3, see PCR+^•).

Table 3.

Results of tests performed with DNA of the three relapsing fever Borrelia strains for the 12 real-time PCR methods. In the first column, targets of PCRs are indicated after the kit name. Results are indicated as positive (PCR+) or negative (PCR-) for each condition experimented. White circles (°) correspond to reactivity (PCR+°) or no reactivity (PCR-°) not mentioned in the kit insert; black circles (•) correspond to the presence of reactivity (PCR+•) or absence of reactivity (PCR-•) in accordance with the kit insert

	B. miyamotoi	B. hermsii	B. recurrentis
Fluorotype (Hain)/unknown target	PCR+•	PCR+•	PCR+/-°
EliGene (Elisabeth Pharmacon)/flagellin gene	PCR-°	PCR+°	PCR-°
Geneproof/16S rRNA gene	PCR+•	PCR-°	PCR-°
Attomol /flagellin gene	PCR-°	PCR+•	PCR-°
RealLine (Bioron) /unknown target	PCR-°	PCR-°	PCR-°
BactoReal (Ingenetix)/ospA gene	PCR-°	PCR-°	PCR-°
Sacace/16S rRNA gene	PCR+°	PCR-°	PCR-°
Viasure (Certest) /unknown target	PCR+•	PCR+•	PCR+°
Diarella (Gerbion) /unknown target	PCR-•	PCR+°	PCR-°
Alphacube (Gerbion) /unknown target	PCR-•	PCR+°	PCR-°
Mutaplex (Immundiagnostik) /unknown target	PCR-•	PCR+°	PCR-°
In-house PCR/flagellin gene	PCR-	PCR-	PCR-

Neither the spirochetes other than RF-Borrelia nor the other micro-organisms nor the two human DNAs caused cross-reactions with all the PCR kits tested.

Practicability of commercial kits

Scores of practicability for real-time PCR kits were attributed as shown in Table 4.

Table 4.

Evaluation of the practicability of PCR kits (material, instructions for use, ease of execution and interpretation of results)

	PCR thermocycler employed in this study	Adequate equipment provided	Quality of description of material not provided	Precise description of procedures	Presence of positive and negative controls	Ease of realization	Ease of interpretation
Fluorotype (Hain)	FC	●●●●	●●●●	●●●●	●●●○	●●●●	●●●●
EliGene (Elisabeth Pharmacon)	LC 480	●●●●	●●●●	●●●●	●●●●	●●●○	●●●●
Geneproof	LC 480	●●●●	●●●●	●●●●	●●●●	●●●●	●●●○
Attomol	LC 2.0	●●○○	●●●●	●●●●	●●○○	●●●○	●●○○
RealLine (Bioron)	LC 480	●●●●	●●●●	●●●●	●●○○	●●○○	●○○○
BactoReal (Ingenetix)	ABI 7500	●●●●	●●●●	●●●●	●●●●	●●●○	●●○○
Sacace	ABI 7500	●●●●	●●●●	●●●●	●●●●	●●●●	●●●○
Viasure (Certest)	ABI 7500	●●●●	●●●●	●●●●	●●●●	●●●●	●●●○
Diarella (Gerbion)	ABI 7500	●●●●	●●●●	●●●●	●●●●	●●●●	●●●○
Alphacube (Gerbion)	ABI 7500	●●●●	●●●●	●●●●	●●●●	●●●●	●●●○
Mutaplex (Immundiagnostik)	LC 480	●●●●	●●●●	●●●●	●●●●	●●●●	●●○○

Black circles correspond to the score of practicability: one for ‘not adapted’, two for ‘poorly adapted’, three for ‘moderately adapted’, and four black circles for ‘very well adapted’

All kits tested were satisfactory concerning the equipment provided or not (good description required), with the exception of the Attomol kit in which the PCR mix was not included, needed to be ordered separately from the kit and therewith impacting the ease of realization.

The test procedures mentioned in the kit inserts were easy to follow for all PCRs kits. Negative controls were provided with the kits except for Fluorotype, Attomol and RealLine tests. For Attomol kit, the positive control was not included and needed to be ordered separately. For RealLine kit, a compliance issue was encountered with a positive control batch that was not experimentally positive by PCR, but replaced by the manufacturer during protocol.

The majority of the tests was easy to run. However, some kits were less ergonomic due to non-automatic computer configuration during installation (Eligene kit) or the presence of different reaction tubes that were difficult to differentiate (BactoReal kit). The RealLine kit was the less practical one to run, related to the absence of ready-to-use PCR mix.

A problem was frequently observed concerning the interpretation of the PCR results since it was fully optimal only for two of the 11 kits studied (Fluorotype and Eligene). Interpretation was rated as moderately easy to do for five kits (Viasure, Alphacube, Diarella, Geneproof and Sacace), difficult for 3 kits (Attomol, BactoReal and Mutaplex) and very poorly adapted for 1 kit (RealLine). We encountered difficulties in interpreting the amplification curves for all kits tested on the ABI7500 thermocycler, and in particular for the BactoReal kit. The amplification curves were also a bit difficult to interpret for the Geneproof kit on the LC480 thermocycler. Tests done with the Attomol, Mutaplex, and RealLine kits generated many Ct results displayed without any associated exponential curve. The RealLine kit also provided several results of exponential curves with Ct values > 40, that needed to be interpreted as negative results according to the kit insert.

Discussion

In our study of analytical comparison of 11 CE-IVD marked real-time PCR assays for Borrelia burgdorferi sensu lato (Bbsl), we found good performances of Bbsl DNA detection with the LODs of 14 strains that ranged from 1 (or below) to 10 GE /5µL, with some exceptions. Cross-reactivity with RF Borrelia DNA was observed for the majority of the PCR kits, while they have not occurred for the DNA extracts of other micro-organisms in our panel. These kits were generally easy to use, in terms of equipment, explanation of technical procedures, and realization in a laboratory competent in molecular microbiology whereas interpretation of results was moderately difficult to difficult for some of them. Although all the analysers employed in the study were given as compatible by the manufacturers, the ease with which PCR test results can be interpreted may have been influenced by the type of analyser used.

Overall good analytical performances but differences: importance of the targeted gene

Regarding the detection spectrum, all the PCR methods allowed the detection of Bbsl DNA from the different species and strains tested, except for the strain B. spielmanii which was undetected by the kit BactoReal. For optimal diagnostic sensitivity, the detection spectrum required for PCR used in direct diagnosis of Lyme borreliosis should at least cover the assured and possible human pathogen species including the species most frequently involved in human disease, i.e. B. afzelii, B. garinii, B. bavariensis, Bbss, but also B. bissettiae, B. spielmanii, B. lusitaniae and B. mayonii. The most sensitive methods should cover the whole spectrum of the Bbsl group species.

Although OspA is a target vaccine for Lyme borreliosis, this lipoprotein is not free of heterogeneity with at least 17 distinct serotypes, of which one genospecies-specific serotype B. spielmanii [15]. Thus, it is likely that one or the two PCR ospA primers used in the BactoReal kit are located in a variable region of the ospA gene, preventing primer binding at the given temperature and therefore impeding the continuation of the reaction and detection of B spielmanii, a proven pathogen in Lyme borreliosis [16]. In the study of Lager et al. [12], DNA of B. spielmanii also was not detected by the ospA-targeted PCR, but by five protocols targeting Bbsl 16S rRNA gene.

PCR kits evaluated were linear and showed good overall efficiency (> 85%), although we were not able to evaluate precisely the method that did not show Ct values (Hain). It is noteworthy that the interpretation of PCR efficiency in our study is in part limited due to the use of only two replicates instead of three. Despite this, our results obtained for efficiency were in agreement with the study of Maes et al. which evaluated three kits (Geneproof, Sacace and Diarella) using three Bbsl strains [17]. Ranging from 1 to 10 GE per reaction, our LOD results were also congruent with those of this latter study for B. afzelii and B. garinii strains that ranged from 2 to 20 GE per reaction (10 to 100 when expressed per mL). However in the study of Maes et al. [17], the results of LODs were unfavorable when using thermocyclers not tested herein for Sacace (LC480) and Diarella kits (LC480 and CFX96), both tested on ABI7500 in our study. The range of LODs obtained in our study is also consistent with that previously observed in an analytical comparison study of six in-house PCR protocols performed in different Scandinavian laboratories using the same DNA panel of Bbsl [12]. Finally, our experimental LODs were generally in accordance with the values displayed in the kit’s insert.

Very low LODs observed in this study for several PCR methods (≤ 1 GE) may be related to targeting multicopy genes of Bbsl spirochetes, such as two copies of 16S rRNA gene harbored by some strains such as B. afzelii [18]. For example, the Geneproof and Sacace kits were characterized by very low LOD values, especially for the B. afzelii species, both using 16S rRNA gene as a PCR target. It can be noted, however, that although present in the Borrelia genome as a single copy, the flagellin gene is a PCR target with a low LOD PCR, i.e., from 0.1 to 10 copies per reaction for the in-house method used in our study. The plasmid origin of the ospA gene does not support herein a better analytical sensitivity as observed with the BactoReal PCR kit. Similar results for the ospA target compared to chromosomal targets were also observed in the study by Lager et al. [12]. However, the supposed impact of the nature of the target gene on the differences in analytical sensitivity observed cannot be further discussed because information is lacking concerning the PCR target displayed by several manufacturers.

Borrelia bacteria are unusual in that cell numbers are not easily determined using usual microbiology tools (e.g., counting chambers, OD measurements). The GE determined for the DNA panel are estimates and are likely not exact. The amplification of the Bbsl genomes within the spirochete without cell division during the exponential phase could also have a favorable impact on low LOD values of cultured spirochete DNA extracts [19]. However, since identical DNA dilutions and identical GE units were used for all tested PCR kits, LOD values, sensitivity and specificity can be compared between tested kits.

Undesirable cross-reactivity with relapsing fever spirochetes in the Lyme borreliosis PCR

The detection of DNA from one or more RF agents was commonly observed in this study, since we observed it for nine of the 12 real-time PCR methods studied. This cross-reactivity was even claimed in the kit insert of four out of the nine methods concerned. Given the close phylogenetic relationship between Bbsl and RF Borrelia spirochetes, genetic similarity is frequent, and it depends directly on the nature of the target genes (e.g., ospA specific for Bbsl spirochetes), and on the variability of the primer hybridization regions (e.g., flagellin gene harboring conserved as well as variable regions among Bbsl and RF Borrelia species). Despite phylogenetic relatedness of these spirochetes, indications of using these molecular methods for diagnosis differed, given the following facts:

• (i) the clinical presentations of the two entities are usually quite different. On one hand, the most frequent manifestations in case of Lyme borreliosis are erythema migrans, Lyme arthritis, meningoradiculitis, facial paralysis, acrodermatitis chronica atrophicans. On the other hand, relapsing fevers that are characterized by recurrent febrile episodes can be complicated by neurological manifestations [4, 8, 20];

• (ii) the epidemiological contexts are also often distinct, regarding the geographical area of distribution of pathogens and of vectors that are required for their transmission [21]. The modalities of tick exposure (exophilic or endophilic ticks) may also be very different, with the exception of B. miyamotoi, a RF agent transmitted by the same vectors as Bbsl spirochetes and that also show in part overlapping symptoms with Lyme borreliosis [22],

• and, (iii) the useful biological matrices for direct biological diagnosis by PCR are also mostly distinct. For Lyme borreliosis, PCR will be done on synovial samples, or on skin biopsies, and in very early neuroborreliosis on CSF [4], while for RF, PCR will be done mainly on whole blood samples and also on CSF samples in case of neurological manifestations [23, 24].

Considering these significant differences in clinical and epidemiological contexts, and diagnostic strategy, the dual detection of the agents of Lyme borreliosis and Relapsing fever is not justified and may even pose confusion in differential diagnosis. There is a risk of misprescription of biological tests for clinicians and a risk of misunderstanding test results, particularly in the absence of genospecies identification in presence of a positive result.

However, other diagnostic methods targeting multiple pathogens seem to be of particular interest in cases of post-tick bite febrile syndrome. Several infectious agents are concerned such as bacteria (e.g., B. miyamotoi, Rickettsia spp., Anaplasma phagocytophilum, Ehrlichia spp.), viruses (e.g., TBE virus, Powassan virus) or parasites (e.g., Babesia spp.). Their distribution in the tick vectors as well as clinical febrile contexts can be superimposed that may benefit from the use of different PCR detection panel strategies, such for instance A. phagocytophilum/Neoehrlichia mikurensis in an endemic French region [25], or A. phagocytophilum/Candidatus Anaplasma capra/B. miyamotoi/B. mayonii/Erhlichia spp./ Rickettsia spp. in Upper Midwest and Northeast US [26]. The Viasure (Certest) kit included in our study is an example for such a strategy and allows the realization of two multiplex PCRs targeting, on the one hand, the DNA of Bbsl, Coxiella burnetii and A. phagocytophilum and, on the other hand, the RNA of TBE virus and the DNA of Rickettsia spp., Babesia spp. and Ehrlichia spp.

PCR in the direct diagnosis of Lyme borreliosis: practice in laboratories

The good analytical sensitivity of most kits tested in this comparative study, as found in other studies [12, 17], may appear to contrast with the poor to moderate diagnostic sensitivity of Lyme borreliosis PCR in clinical practice. This dissociation is mainly explained by the scarcity of Lyme-spirochetes in tissues and fluids [27].

In Europe, Lyme borreliosis PCR is recommended as confirmatory testing on skin biopsies in case of suspicion of dermatological manifestations out of typical EM and in joint fluid and/or synovial biopsies in case of suspicion of seropositive Lyme arthritis, and optional on CSF in case of suspicions of very early neuroborreliosis [4, 5]. More rare presentations of Lyme borreliosis may also justify performing Bbsl PCR on other matrices such as heart valves [28]. Even if sampling is less restrictive, PCR remains usually not recommended on urine nor blood samples for Lyme borreliosis [4, 5]. This framework of precise indications and biological matrix must be followed by diagnostic laboratories using PCR kits or in-house PCRs. In addition to these prerequisites, the laboratory using these kits must have specific skills in molecular biology. Advanced skills are even required for some techniques of poor practicability in our study and for in-house PCRs. Regardless of the PCR kits used, laboratories performing direct diagnosis of Lyme borreliosis should also regularly participate in External Quality Assessment (EQA) programs.

In case of a positive Bbsl PCR on a clinical sample, molecular typing at least at the genospecies level appears of great importance for confirmatory diagnostic of the specificity of amplification and is useful for epidemiological monitoring. None of the kits in this study provided direct identification to the genospecies level, this can be done either by sequencing the produced amplicons if their sampling is accessible or by using another molecular typing method on positive samples; the most widely used are 16S rRNA or flaB sequencing, and MultiLoci Sequence Typing (eight genes) (e.g., [29]). In addition, if a positive Borrelia PCR result has been obtained by a method that is not specific to the Bbsl group and without distinction between the two groups of pathogenic Borrelia, depending on the clinical context, a confirmatory PCR test should be carried out for RF Borrelia DNA. The target gene must be specific to this phylogroup, such as a specific portion of the 16S rRNA gene [30] or the glycerophosphodiester phosphodiesterase gene, glpq [31].

The European In Vitro Diagnostic Regulation «IVDR» 2017/746 introduced new requirements, particularly in terms of traceability and performance evaluation for diagnostic tests marketed in the EU. Lyme Borreliosis PCR tests appear in class C/MDCG 2020-16 rev3, with mandatory certification in 2028. In this context, important elements such as the nature of PCR target genes, detection spectrum and specificity should be displayed by manufacturers.

This study showed that the analytical performance and practicability of the kits evaluated were broadly in line, but that advanced molecular biology skills were required to interpret the results. Simultaneous detection of Lyme Borreliosis and RF agents without precise distinction does not necessarily appear to be justified and may result in a risk of confusion in the choice of biological matrices and in the wording of the result returned to the clinician. Diagnosis laboratories using a PCR method detecting both group Bbsl and group RF Borrelia must pay particular attention to the amplification confirmation strategy by group or species typing. More standardized analytical performance testing, as performed in this comparative study, may also be desirable for Bbsl PCR in the future.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

BJ realized the conceptualization and supervision of the study. VF, ETR and LZ helped to conceptualization. MF, BS, GM, VF performed the preparation and qualification of samples. LZ, LB and ME performed the PCR experiments. ETR, LZ, LB and BJ analyzed the results. ETR wrote the article and prepared figures. VF, GM, BJ and LZ helped to write the article. All authors reviewed the manuscript.

Funding

This work was supported by Santé Publique France and Université de Strasbourg.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.