Simultaneous Detection and Differentiation of Clinically Relevant Relapsing Fever Borrelia with Semimultiplex Real-Time PCR

ABSTRACT

Bacterial vector-borne diseases, including Borrelia species, present a significant diagnostic, clinical, and public health challenge due to their overlapping symptoms and the breadth of causative agents and arthropod vectors. The relapsing fever (RF) borreliae encompass both established and emerging pathogens and are transmitted to humans by soft ticks, hard ticks, or lice. We developed a real-time semimultiplex PCR assay that detects multiple RF borreliae causing human illness and classifies them into one of three groups. The groups are based on genetic similarity and include agents of soft-tick relapsing fever (Borrelia hermsii and others), the emerging hard-tick-transmitted pathogen B. miyamotoi, and the agent of louse-borne relapsing fever (B. recurrentis). The real-time PCR assay uses a single primer pair designed to amplify all known pathogenic RF borreliae and multiple TaqMan probes to allow the detection of and differentiation among the three groups. The assay detects all RF borreliae tested, with an analytical limit of detection below 15 genome equivalents per reaction. Thirty isolates of RF borreliae encompassing six species were accurately identified. Thirty-nine of 41 residual specimens (EDTA whole blood, serum, or plasma) from patients with RF were detected and correctly classified. None of 42 clinical samples from patients with other infections and 46 culture specimens from non-RF bacteria were detected. The development of a single-assay real-time PCR approach will help to improve the diagnosis of RF by simplifying the selection of tests to aid in the clinical management of acutely ill RF patients.

INTRODUCTION

Relapsing fever (RF) Borrelia species include multiple recognized and emerging pathogens (1). The RF borreliae are spirochetes related to, but genetically and clinically distinct from, Borrelia burgdorferi sensu lato species (also known as Borreliella spp.), the causative agents of Lyme borreliosis (2–5). Infection with any of the RF borreliae is typically characterized by fever, headaches, chills, myalgia, arthralgia, and nausea/vomiting (1, 6). The name RF refers to the classical disease presentation of these infections, characterized by multiple episodes of high fever separated by asymptomatic periods (1, 6).

The RF borreliae that are pathogenic to humans can be divided into three groups on the basis of genetic similarity (7, 8). One group, referred to here as the B. hermsii group, contains a number of Borrelia species found primarily in the Americas and transmitted by argasid (soft) ticks (Ornithodoros spp.) (9). B. hermsii, the most common cause of tick-borne relapsing fever (TBRF) in the United States, is transmitted by Ornithodoros hermsi ticks, which are found in the Western states. B. parkeri has also been reported to cause TBRF in the Western United States, although direct evidence of human infection is lacking (10). B. turicatae infections are transmitted by Ornithodoros turicata ticks, which are found primarily in Florida and Texas, although their range overlaps that of O. hermsi (9). TBRF has also been reported in Central and South America, possibly associated with B. venezuelensis or other poorly characterized species (9, 11). Recently, “Candidatus Borrelia johnsonii” was detected in the blood of a Wisconsin patient with suspected tick-borne disease; it had previously been identified in bat ticks, Carios kelleyi, from Iowa (12).

Another group of RF borreliae consists solely of B. miyamotoi, which is the only pathogenic RF species transmitted by hard ticks (Ixodes spp.) (13). Although this species has only recently been described as a cause of human disease, its associated clinical syndrome (Borrelia miyamotoi disease) is likely the most common disease caused by RF borreliae in parts of North America and Eurasia (14–18). In North America, B. miyamotoi is found in the Northeast, in the Mid-Atlantic, and on the West Coast; the latter region overlaps the range of B. hermsii. B. miyamotoi disease is typically milder than clinical syndromes caused by other RF borreliae, with recurring febrile episodes being reported less frequently than in classical RF (13–15, 19, 20).

The final RF group, referred to here as the B. recurrentis group, comprises several species found primarily in Africa. B. recurrentis is transmitted by body lice and is the causative agent of louse-borne relapsing fever (LBRF), a disease with epidemic potential. The closely related species B. duttonii and B. crocidurae, along with other species that are less well characterized, cause TBRF and are transmitted by Ornithodoros species ticks endemic to Africa (1, 6, 21). B. recurrentis and B. duttonii cause the most severe human illnesses associated with RF borreliae, with reported mortality rates of >30% if left untreated (1, 6, 22).

Both the B. hermsii and B. recurrentis groups are associated with high spirochetemia (>10⁵ spirochetes/ml blood) during symptomatic episodes in infected patients, with the diagnosis of acute infections often being performed by microscopic visualization of spirochetes in blood (23–25). However, spirochetes are typically not detectable by microscopy during asymptomatic periods or after antibiotic treatment (22). The limit of detection (LOD) of microscopy is dependent on the number of fields examined, but even examination of hundreds of fields yields an LOD no lower than 10³ to 10⁴ spirochetes/ml (25, 26). B. miyamotoi usually cannot be detected by microscopy as it results in lower spirochetemia (≤10⁴ spirochetes/ml blood) than other RF borreliae (15, 19, 27). Culture is not useful for timely clinical management as RF borreliae are fastidious and slow growing, requiring specialized media (23, 28). Serological techniques for RF detection are not standardized, and serology is not sensitive for the detection of early infections (29–32).

Molecular techniques can overcome the limitations of other diagnostic methods for RF. As the analytical sensitivity of PCR is higher than that of microscopy, it can be used when spirochetemia is low, such as during afebrile periods, after treatment, and in B. miyamotoi disease (15, 16, 33–36). Existing clinical diagnostic real-time PCR assays for RF borreliae are intended to target a single species or group, an approach that is reasonable when there is a strong index of suspicion for a single RF species (27, 34–39). However, species-targeted assays are limited in the ability to detect unexpected species resulting from travel-associated infections and to detect and distinguish multiple groups of RF borreliae in regions where they overlap, such as the U.S. West Coast (B. hermsii and B. miyamotoi) and the upper Midwest (B. miyamotoi and “Candidatus Borrelia johnsonii”).

Given the presence of multiple RF species in overlapping geographic regions, the overlapping clinical symptoms of RF borreliae, and the recent identification of several new pathogenic RF species, the aim of this study was to produce a molecular diagnostic assay capable of detecting all RF Borrelia species known to be pathogenic to humans in a single real-time PCR and accurately classifying each species into one of the three groups (B. hermsii, B. miyamotoi, or B. recurrentis). This unified, geographically unbiased approach will simplify assay selection for clinicians who suspect a patient’s illness to be relapsing fever.

MATERIALS AND METHODS

In silico analysis.

Genetic analyses were performed with Geneious (Biomatters, Inc., Auckland, New Zealand) unless otherwise stated. Chromosome sequences were downloaded from GenBank (see Table S1 in the supplemental material) and aligned using the progressiveMauve algorithm (40). The eight housekeeping genes (uvrA, rplB, recG, pyrG, pepX, clpX, nifS, and clpA) used for Borrelia multilocus sequence typing (MLST) were extracted and concatenated (41). The concatenated sequences were aligned and phylogenetic trees were produced using MEGA X (42).

Bacterial culture.

Non-Borrelia control organisms were obtained in-house or purchased from the American Type Culture Collection (ATCC). Borrelia species were grown in modified Barbour-Stoenner-Kelly (BSK-R) medium (28). For spiking experiments, borreliae were grown to mid-log phase and counted by microscopy at a ×400 magnification as previously described (28, 43). Nonmotile cells were included in calculations to account for DNA present in nonviable cells.

For DNA extraction, Borrelia culture material was centrifuged at 10,000 × g for 8 min. DNA was extracted from the resulting pellets and from nonborrelial bacteria using the QIAamp DNA minikit (Qiagen, Germantown, MD), quantified using the Qubit system (Thermo Fisher, Waltham, MA), and normalized to the DNA quantities described for each experiment.

Real-time PCR.

Oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). Probes were PrimeTime dual-quenched probes. Oligonucleotide sequences and probe modifications are shown in Table 1. Optimized PCR conditions included 500 nM concentrations of each primer and 100 nM each probe, with PerfeCTa multiplex supermix (Quantabio, Beverly, MA). Real-time PCR was performed on a QuantStudio Dx instrument (Thermo Fisher). Cycling conditions included an initial 20-s denaturation step at 95°C followed by 45 cycles of 3 s at 95°C and 30 s at 57°C. Data were collected from the Cy5, VIC (used to detect hexachlorofluorescein [HEX]), and 6-carboxyfluorescein (FAM) channels during the annealing/extension phase. Based on standard-curve analysis, a cutoff of 45 cycles was chosen. Human endogenous retrovirus 3 (ERV3) was used as an endogenous control in separate reactions using previously described oligonucleotide sequences with a FAM-labeled probe and the same amplification conditions as the ones used for the detection of RF borreliae (44).

TABLE 1.

Oligonucleotides

Oligonucleotide	Target	Label	Sequencea
RF-0509-F (forward primer)	All	None	TTG GTG ATA TTG CTA TTG GTT C
RF-0509-R (reverse primer)	All	None	TCA TTC ATT GCA CCA TCA T
hermsii-0509-P	B. hermsii group	HEX	/HEX/TT TAT AAT A/ZEN/T TGT TGA GAG TCT TAG AGC CTT TTT /IABkFQ/
parkeri-0509-P	B. hermsii group	HEX	/HEX/TT TTT AGA C/ZEN/C CAG ATC TCT TGA AGC /IABkFQ/
coriaceae-0509-P	B. hermsii group	HEX	/HEX/TT TTT AGA C/ZEN/C CAG GTA TCT TGA GGC /IABkFQ/
recurrentis-0509-P	B. recurrentis group	Cy5	/Cy5/TG TTT CTA A/TAO/T ATG GTT GAT AGT CTG AAG TCA TT/IAbRQSp/
miyamotoi-0509-P	B. miyamotoi	FAM	/6-FAM/CA CAA CAC C/ZEN/T TAA AGT TCA TGA GAA TG/IABkFQ/

^aSequence modifications are in boldface type and are placed between slashes. HEX, hexachlorofluorescein; 6-FAM, 6-carboxyfluorescein; IAbFQ, Iowa Black FQ; IAbRQ, Iowa Black RQ. ZEN and TAO are proprietary internal quenchers from Integrated DNA Technologies.

LOD analysis.

LODs were calculated by probit analysis within MedCalc software (MedCalc Software Ltd., Ostend, Belgium). The analytical LOD was determined by testing DNA from B. hermsii, B. turicatae, B. miyamotoi, and B. recurrentis cultures in quantities ranging from 20 fg to 1.25 fg in 2-fold serial dilutions, with eight replicates per dilution.

The diagnostic LOD was determined in mock clinical samples spiked with B. hermsii 95-0544 or B. miyamotoi 13-2396 (45). Both cultures reached approximately 2 × 10⁷ spirochetes/ml, with >90% of cells observed to be motile. Cultures were diluted in phosphate-buffered saline, followed by a final 1:20 dilution to the desired concentration in pooled EDTA blood or serum from healthy human blood donors (Innovative Research, Novi, MI). In one experiment, blood spiked with Borrelia cultures was separated to test the LOD in plasma. The spiked blood, serum, or plasma was divided into 22 to 24 aliquots and frozen. Preliminary side-by-side comparisons indicated that freezing of spiked blood specimens at −80°C did not change PCR results.

Analysis of residual human samples.

EDTA whole-blood, plasma, or serum specimens were originally submitted for tick-borne diagnostic testing. The use of residual specimens for assay development and validation was approved by the CDC Institutional Review Board (protocol number 7102). EDTA blood from healthy human donors (BioIVT, Westbury, NY, and ZenBio, Research Triangle Park, NC) was also tested to ensure that no reactivity would be detected in a larger group of samples.

DNA was extracted from clinical and spiked specimens using a MagNA Pure 96 instrument (Roche, Indianapolis, IN) with the MagNA Pure 96 DNA and viral NA small-volume kit using the DNA blood small-volume protocol, an input volume of 200 μl, and an output volume of 100 μl. Five microliters of the resulting DNA was used in each PCR.

RESULTS

In silico analysis and assay design.

Initial analysis was performed on a subset of Borrelia species (Fig. 1) and subsequently confirmed on a broader set (see Table S1 in the supplemental material). All chromosomes were aligned and found to be colinear. For grouping of RF species, eight housekeeping genes used for Borrelia multilocus sequence typing (MLST) were extracted, concatenated, and used to produce a phylogenetic tree (Fig. 1A) (41). Based on this tree, the RF borreliae were divided into three groups typified by B. hermsii, B. miyamotoi, and B. recurrentis, respectively.

FIG 1.

(A) Phylogenetic tree of RF Borrelia species. Whole genomes were downloaded from GenBank (see Table S1 in the supplemental material for accession numbers) and aligned using progressiveMauve (40). Alignments of the eight housekeeping genes used in Borrelia multilocus sequence typing (uvrA, rplB, recG, pyrG, pepX, clpX, nifS, and clpA) (41) were extracted and concatenated, and a tree was created using maximum likelihood methodology and 500 bootstrap replicates in MEGA X (42). (B) Sequence alignment of the amplified region of Bh_0509 and homologs in other RF borreliae. The strains included are the same as those in panel A, except that B. hermsii YOR and B. miyamotoi FR64b are not included. Nucleotides that differ from the consensus are highlighted (red, A; yellow, G; blue, C; green, T). Arrows indicate primer and probe binding regions (5′ to 3′) and are annotated on the sequence to which they match exactly.

A chromosomal gene of unknown function, designated Bh_0509 in B. hermsii, was identified as a candidate for assay development based on sequence homology. It is present in all available Borrelia genomes, including both RF and B. burgdorferi sensu lato, and contains a predicted signal sequence but no identifiable conserved domains according to a search of the InterPro database (46). It exhibits approximately 86% pairwise nucleotide identity between RF borreliae from different groups and 73% pairwise nucleotide identity between RF borreliae and B. burgdorferi sensu lato species. Some other spirochetes, including Treponema spp. but not Leptospira spp., contain homologs with <40% nucleotide identity, and no homologs are detectable by a BLAST search in nonspirochete genomes.

PCR primers and TaqMan probes were designed to amplify a 187-bp fragment of the gene in RF borreliae but not B. burgdorferi (Fig. 1B and Table 1). For the detection of the B. miyamotoi and B. recurrentis groups, TaqMan probes were labeled with FAM and Cy5 fluorophores, respectively. For the detection of the B. hermsii group, three TaqMan probes labeled with the same fluorophore (HEX) were used, as the sequence diversity was too great to detect all members of this group with a single probe. All tested RF Borrelia species in all three groups were detected in the expected channels using this two-primer/five-probe assay, and there was no cross-reactivity between groups (Table 2). Additional in silico analysis found 100% sequence identity within the primer and probe regions of B. miyamotoi from the United States, Russia, and Japan and a single nucleotide polymorphism that is not predicted to affect amplification within the probe region of European strains of B. miyamotoi. Although fewer sequences from the B. recurrentis group are publicly available, those that are available display 100% nucleotide identity within the primer region and a single nucleotide difference in the probe region, which is not predicted to affect detection. The assay is not predicted to detect either B. anserina, an avian pathogen in the RF group, or the newly discovered “Candidatus Borrelia mahuryensis,” which is most closely related to the reptile-associated borreliae B. turcica and “Candidatus Borrelia tachyglossi” (47).

TABLE 2.

Cross-reactivity of probes

Template	Cross-reactivity with probe
	hermsii-0509-P	parkeri-0509-P	coriaceae-0509-P2	miyamotoi-0509-P	recurrentis-0509-P
B. hermsii	+	−	−	−	−
B. parkeri	−	+	−	−	−
B. turicatae	+	+	−	−	−
B. coriaceae	−	−	+	−	−
B. miyamotoi	−	−	−	+	−
B. recurrentis	−	−	−	−	+

Analytical validation.

For the determination of analytical specificity, high-concentration DNA (≥1 ng/reaction mixture) from 46 non-RF bacterial species was tested, including 10 B. burgdorferi sensu lato species, 1 reptile-associated Borrelia species (B. turcica), 3 other spirochetes (Leptospira interrogans, Leptospira kirschneri, and Treponema denticola), and 32 other bacterial species (Table S2). No amplification was detected.

For analytical sensitivity, DNA from cultures of 30 RF borreliae (14 B. hermsii, 7 B. miyamotoi, 5 B. turicatae, 2 B. recurrentis, and 1 each of B. parkeri and B. coriaceae) was used (Table S3). All were detected only in the expected fluorescent channel (B. hermsii, B. turicatae, B. parkeri, and B. coriaceae in HEX; B. miyamotoi in FAM; and B. recurrentis in Cy5). Average threshold cycle (C_T) values with 10 pg of the DNA template were 24.0 for B. hermsii, 25.6 for B. turicatae, 26.5 for B. miyamotoi, and 25.1 for B. recurrentis (Table S3).

Standard curves were run in duplicate for B. hermsii, B. turicatae, B. miyamotoi, and B. recurrentis with DNA quantities in 10-fold dilutions ranging from 10 ng to 1 fg. All four species displayed linear amplification over this 8-log₁₀ range (Fig. 2). Calculated amplification efficiencies were 100.0% for B. hermsii, 97.3% for B. turicatae, 102.5% for B. miyamotoi, and 107.9% for B. recurrentis. The analytical limit of detection of the assay against these four species ranged from 3.78 fg to 12.1 fg per PCR (Table 3). The genome sizes for these species range from 1.17 Mb to 1.39 Mb, leading to calculated LODs ranging from 2.8 to 8.8 genome equivalents (Table 3).

FIG 2.

(A to C) Standard curves for B. hermsii 95-0544 (HEX channel) (A), B. miyamotoi HT31 (FAM channel) (B), and B. recurrentis 99-0708 (Cy5 channel) (C). Serial 10-fold dilutions of bacterial DNA from 1 ng to 1 fg were tested in duplicate. ΔRn, fluorescent reporter signal normalized to baseline. (D) Linearity of standard curves. Data from panels A to C are plotted, along with data for B. turicatae 90-901 (amplification curves not shown). RF borreliae are indicated by different symbols.

TABLE 3.

Genome sizes and limits of detection

Species	Genome size (Mb)a	Analytical LOD (fg/PCR) (95% confidence interval)	Analytical LOD (genome equivalents/PCR)
B. hermsii	1.390	5.42 (3.68, 61.2)	3.61
B. turicatae	1.165	6.47 (3.98, 8.95)	5.14
B. miyamotoi	1.278	12.11 (7.63, 16.59)	8.78
B. recurrentis	1.242	3.78 (2.65, 4.90)	2.82

^aGenome sizes are based on genome sequences deposited in the NCBI database for B. hermsii HS1, B. turicatae 91E135, B. miyamotoi CT13-2396, and B. recurrentis A1.

Diagnostic validation.

To determine the limit of detection (LOD) in clinical specimens, cultures of B. hermsii and B. miyamotoi were diluted in serum or EDTA blood from healthy human donors. Plasma was separated from a subset of spiked blood samples to determine the effect of separation on the LOD. The LODs in spiked blood and serum were nearly identical for B. hermsii (22 spirochetes/ml blood and 17 spirochetes/ml serum), while a 4-fold difference was found for B. miyamotoi (20 spirochetes/ml blood and 88 spirochetes/ml serum). The LODs in plasma separated from spiked blood were 5-fold and 80-fold higher than those for whole blood for B. hermsii (125 spirochetes/ml blood before separation) and B. miyamotoi (1,629 spirochetes/ml), respectively.

Human EDTA blood specimens previously identified as being positive for B. hermsii (n = 4), B. turicatae (n = 3), “Candidatus Borrelia johnsonii” (n = 1), and B. miyamotoi (n = 27) were tested for diagnostic validation (Table 4). Amplification of DNA from all B. hermsii-, B. turicatae-, and “Candidatus Borrelia johnsonii”-positive blood specimens was detected solely in the HEX channel, indicating the B. hermsii group, with C_T values ranging from 16.4 to 38.0 (Table 4). Of 27 B. miyamotoi-positive EDTA whole-blood specimens tested, 26 were detected in the FAM channel, with C_T values ranging from 28.0 to 39.9 (average, 34.2). No amplification was detected in the remaining B. miyamotoi-positive blood specimen. The human endogenous control gene ERV3 was detected in this specimen, indicating that the lack of RF amplification was not caused by poor DNA quality or inhibition. The C_T of this specimen on a pan-Borrelia PCR assay was 39.1, indicating that the quantity of borrelial DNA was small (12).

TABLE 4.

Diagnostic sensitivity and specificity testing using residual EDTA whole-blood specimens

Species	Original case classification method	No. of specimens	RF PCR result	CT ranged
B. hermsii	Microscopy/culture	4	4/4 positive (B. hermsii group)	16.4–24.5
B. turicatae	Microscopy/culture or serologya	3	3/3 positive (B. hermsii group)	27.1–38.2
B. miyamotoi	PCR or MLST/16S metagenomic analysisb	27c	26/27 positive (B. miyamotoi)	28.0–39.9 (avg, 34.2)
“Candidatus Borrelia johnsonii”	MLST/16S metagenomic analysis	1	Positive (B. hermsii group)	32.4
B. burgdorferi	MLST/16S metagenomic analysis	5	Negative	NA
Anaplasma phagocytophilum	16S metagenomic analysis	10	Negative	NA
Ehrlichia chaffeensis	16S metagenomic analysis	10	Negative	NA
Ehrlichia muris subsp. eauclairensis	16S metagenomic analysis	2	Negative	NA
Leptospira kirschneri	16S metagenomic analysis	11	Negative	NA
Rickettsia rickettsii	16S metagenomic analysis	1	Negative	NA
Staphylococcus aureus	16S metagenomic analysis	1	Negative	NA
Streptococcus pneumoniae	16S metagenomic analysis	1	Negative	NA
Legionella pneumophila	16S metagenomic analysis	1	Negative	NA

^aSerology was performed as previously described (32).

^bPCR was performed at the Mayo Clinic. MLST and 16S metagenomic analyses were performed as previously described (12, 41, 48).

^cBased on available data on deidentified specimens (age, sex, state, and collection date), two of these specimens (both positive) may have been from the same patient. All the other specimens are from unique patients.

^dNA, not applicable.

Diagnostic specificity was assessed using DNA extracted from whole human EDTA blood specimens identified as being positive for non-RF bacterial infections by 16S V1-V2 metagenomics (48). These specimens included 5 B. burgdorferi, 10 each Anaplasma phagocytophilum and Ehrlichia chaffeensis, 2 Ehrlichia muris subsp. eauclairensis, 11 Leptospira kirschneri, and 1 each Staphylococcus aureus, Streptococcus pneumoniae, Legionella pneumophila, and Rickettsia rickettsii (Table 4). No RF PCR amplification was detected for any of these specimens. DNA extracted from 96 EDTA whole-blood specimens from healthy human donors was also tested, with no RF PCR amplification detected. The human endogenous control gene ERV3 was amplified in all human specimens.

Serum and plasma specimens from patients diagnosed with TBRF or B. miyamotoi disease, who were not known to have received treatment before sample collection, were also tested (Table 5). These specimens comprised three sera from B. hermsii-infected patients, two sera from B. turicatae-infected patients, and one plasma specimen from a B. miyamotoi-infected patient. DNA extracted from five of the six specimens tested positive by RF PCR. The negative specimen was taken 23 days after onset from a patient with B. hermsii infection that was originally diagnosed by serology.

TABLE 5.

Diagnostic sensitivity testing using residual serum and plasma

Species	Matrix	Original case classification method(s)	No. of days after onset	RF PCR result	CT
B. hermsii	Serum	Microscopy and serologya	3	Positive (B. hermsii group)	26.4
B. hermsii	Serum	Microscopy and serology	7	Positive (B. hermsii group)	25.0
B. hermsii	Serum	Serology	23	Negative	NA
B. turicatae	Serum	Serology	Unknown; acute	Positive (B. hermsii group)	24.9
B. turicatae	Serum	Serology	42	Positive (B. hermsii group)	38.0
B. miyamotoi	Plasma	PCR	1	Positive (B. miyamotoi)	35.2

^aSerology was performed as previously described (32).

DISCUSSION

The development of a sensitive and specific molecular assay to detect RF Borrelia species and differentiate them into groups will help to simplify the diagnosis of RF, providing a unified, geographically unbiased approach for a broad and growing group of related pathogens that cause overlapping clinical symptoms. In devising the assay, we used a semimultiplex approach in which the primer sequence is designed to broadly amplify all RF Borrelia species known to be pathogenic to humans, while TaqMan probes differentiate among the three groups. Similar assays have been described for other pathogens, including the differentiation of European B. burgdorferi sensu lato species (49–51). The use of the same primer set streamlines assay design and helps to limit issues with competition and decreased sensitivity encountered in standard multiplex PCRs (52).

The specificity of the assay was excellent, with no false-positive amplification detected either in DNA from pure cultures of other bacteria or in clinical specimens from individuals found to have other tick-borne or spirochetal diseases. This includes B. burgdorferi sensu lato, which is closely related to the RF borreliae, as well as the more distantly related spirochete T. denticola, which encodes a gene with weak homology to Bh_0509.

Analytical LODs for the four primary target species (B. hermsii, B. turicatae, B. miyamotoi, and B. recurrentis) were all below 10 genome equivalents. The theoretical LOD for real-time PCR assays is 3 genome equivalents, indicating that the analytical sensitivity of the assay is nearly optimal (53). However, the true LOD may be slightly higher because the number of genome equivalents was calculated based on the genome size available in the NCBI database, which may be underestimated because borrelial genomes are highly fragmented (45).

The diagnostic LOD for B. hermsii and B. miyamotoi was calculated in spiking experiments to be approximately 20 spirochetes per ml blood. This level is well below expected spirochetemia levels in acutely ill patients, including those with B. miyamotoi disease, which tends to result in lower spirochetemia than other RF diseases (15, 27). This is supported by the fact that 26/27 B. miyamotoi specimens tested were positive, although the C_T values in these specimens were higher on average than those in specimens from TBRF patients. This LOD is likely sufficient to detect infection even in samples collected during afebrile periods or soon after treatment, when detection by microscopy is unlikely.

Although blood is the preferred specimen for the diagnosis of RF due to the high spirochetemia in symptomatic individuals, clinicians may not draw an acute-phase blood specimen if they suspect a different tick-borne disease such as Lyme disease, which is primarily diagnosed by serology. Molecular detection of RF borreliae in serum is therefore of potential utility but has rarely been examined. In one previous study, investigators were unable to amplify B. miyamotoi DNA from any of 27 acute-phase serum samples by conventional PCR (17). In contrast, here, we show that DNA from RF borreliae, including B. miyamotoi, can be detected by semimultiplex real-time PCR in acute-phase serum and plasma. Although the calculated LODs are 5- and 80-fold higher in plasma than in blood for B. hermsii and B. miyamotoi, respectively, these sensitivities are still higher than that of microscopy and may allow the timely detection of infection if a blood sample is not available. Although few serum or plasma specimens from untreated symptomatic patients were available for evaluation, our preliminary findings suggest that the semimultiplex PCR assay can detect RF borreliae in these sample types.

Assay development and validation were subject to several limitations. The assay can determine the group to which an RF specimen belongs but cannot differentiate it to the species level. Although species-level differentiation is desirable for epidemiological and surveillance purposes, it is not required for diagnostic testing and appropriate treatment and is not practical given the number of known RF species and the likelihood that new species will be discovered in the future. Group-level information provides enough information for clinical management, and additional sequence characterization can be performed if required for surveillance.

An additional limitation is that no clinical specimens from the B. recurrentis group, and only two cultures of B. recurrentis, were available. However, genome data suggest that other species in the B. recurrentis group also have sufficient sequence identity in their Bh_0509 homologs to be detected by this assay (Fig. 1) (21).

The approach described here allows the broad detection of RF borreliae and leaves open the possibility of detecting novel species. The breadth of RF species pathogenic to humans has not yet been fully defined, as the recent discoveries of “Candidatus Borrelia johnsonii” and of pathogenic African members of the B. hermsii group make clear (7, 12). Similarly, RF borreliae are suspected to cause disease in Central and South America, but these species are poorly understood (9). Many other RF borreliae have not yet been fully characterized, and their potential pathogenicity is unknown (11, 54). Therefore, we chose to optimize the assay for the detection of as many species as possible, including B. coriaceae and B. parkeri, which have not been demonstrated to cause human disease, in hopes of detecting novel species that may be discovered in the future. The detection of “Candidatus Borrelia johnsonii,” despite the fact that sequence information is not available for the PCR target gene, indicates that this approach was at least partially successful. If new pathogenic species are characterized in the future, the oligonucleotide composition of the assay can be modified as needed to ensure optimal performance.

The implementation of this semimultiplex real-time PCR assay will help simplify the diagnosis of acute RF in human clinical specimens. The ability to differentiate among the three major groups of RF borreliae allows an unbiased approach to diagnosis, which is important for a large group of related bacterial pathogens (both known and emerging) that are difficult to distinguish symptomatically, cause illness in overlapping geographic regions, and can cause travel-associated cases. Accurate and timely diagnosis of RF will improve the clinical management of patients and the understanding of these understudied and emerging pathogens.