Development and validation of a real-time PCR test to detect Bartonella quintana in clinical samples

Abstract

This study reports on the validation of a real-time polymerase chain reaction test targeting the vomp region of Bartonella quintana. The assay displayed 100% sensitivity and specificity for the 52 bloods and 159 cultures tested. Molecular diagnosis of Bartonella quintana can aid clinical treatment during acute infection.

Human Bartonella quintana infection, transmitted via body lice, causes trench fever, endocarditis, bacillary angiomatosis, lymphadenitis, and peliosis hepatis [1]. Throughout World War I, B. quintana infected an estimated 800,000 Allied soldiers on the Western Front and accounted for at least one-fifth of illness in the British and Central Powers’ armies [2]. Infections today commonly occur in areas of high population density and poor sanitation, with persons experiencing homelessness at increased risk [3–6].

Humans are the main reservoir for B. quintana. The bacteria infect the bloodstream, causing fevers that last 2 to 4 days and relapse after 5 days intervals for several weeks [2,7]. Symptoms of B. quintana infection include headache, muscle and joint pain, chills, sweating, frequent urination, dizziness, nausea, and diarrhea. Although infection is typically not fatal, it can lead to significant morbidity, most commonly endocarditis, if untreated [1,7]. Diagnosis rests on clinical suspicion, as symptoms may be nonspecific [6,8]. Techniques for diagnosis include serology, culturing, and/or PCR. Serologic testing is not species-specific, may cross react with other pathogens, requires the host to develop a detectable antibody response to B. quintana, and is interpreted subjectively. Furthermore, persons infected with B. quintana can remain seroreactive for years after treatment [4,9,10]. Blood cultures are often negative, due to B. quintana’s fastidious nature and prolonged incubation time [10,11].

To help improve direct detection of B. quintana infection, this study focuses on the validation of a specific real-time PCR test for detection of B. quintana in blood or culture samples. We chose the vomp (variably expressed outer membrane protein) region as the DNA amplification target. The Vomp proteins (vompA-D) assist in escaping immune response through antigenic and phase variation [12]. Although the vomp region varies among B. quintana isolates, we designed oligonucleotides that bind to a conserved sequence that is present in at least 2 copies in all publicly available B. quintana genome sequences. They were confirmed to be specific to B. quintana by NCBI Primer BLAST [13].

Oligonucleotides consisted of a forward primer (5’CATCGCTCTGGTTATACTCTTATCGA3’), reverse primer (5’GATCCAAAATAACTTCCTGGGTCAT3’), and PrimeTime probe (5’/56-FAM/TGTATCGGCTGTTTTTGCCTCGACTTTACC/3BHQ_1/3’) (Integrated DNA Technologies; Coralville, Iowa). Each 20-μL PCR reaction included 750 nM concentrations of each primer and 250 nM of the probe, with PerfeCTa Multiplex Supermix (Quantabio; Beverly, MA). The run conditions included an initial denaturation at 95°C for 2.5 minutes, followed by 40 cycles of denaturation at 95°C for 12 seconds and annealing at 60°C for 45 seconds. We used the human endogenous retrovirus ERV3 as an endogenous control with previously described oligonucleotide sequences [14]. All DNA extractions were performed using the QIAamp DNA Mini Kit (Qiagen; Germantown, MD), unless otherwise specified.

We obtained all bacterial culture samples (Table 1) in house or from the American Type Culture Collection (ATCC). We performed PCR using 10 pg/reaction to assess analytic sensitivity and specificity. Among 18 different isolates of B. quintana, the average Ct was 26.80 (SD 1.26). 141/141 (100%) isolates for 40 non-B. quintana bacteria, including 15 other Bartonella species, were undetected.

Table 1.

Bacterial isolates used for analytical sensitivity and specificity testing.

Species	Number of isolates
Sensitivity
Bartonella quintana	18
Specificity
Acinetobacter radioresistens	1
Afipia felis	1
Bartonella alsatica	1
Bartonella bacilliformis	30
Bartonella clarridgeiae	5
Bartonella doshiae	3
Bartonella elizabethae	3
Bartonella grahamii	2
Bartonella henselae	55
Bartonella koehlerae	2
Bartonella rochalimae	1
Bartonella silvicola	1
Bartonella tamiae	1
Bartonella tribocorum	1
Bartonella vinsonii	1
Bartonella vinsonii subsp. arupensis	1
Bartonella vinsonii subsp. berkhoffii	1
Bartonella volans	2
Bartonella washoensis	1
Borrelia burgdorferi	2
Borrelia hermsii	1
Cedecea neteri	1
Chlamydophila pneumoniae	1
Enterobacter cloacae	1
Escherichia coli	2
Francisella tularensis	2
Klebsiella oxytoca	1
Klebsiella pneumoniae	1
Legionella pneumophila subsp. pneumophila	1
Leptospira interrogans	1
Moraxella catarrhalis	1
Mycoplasma pneumoniae	1
Ochrobactrum anthropi	1
Pasteurella multocida	1
Proteus mirabilis	1
Pseudomonas aeruginosa	1
Salmonella enterica subsp. enterica serovar Typhi	1
Staphylococcus aureus	2
Staphylococcus epidermidis	1
Streptococcus gallolyticus subsp. gallolyticus	1
Streptococcus pyogenes	1
Yersinia pestis	2

To assess the analytical limit of detection (LOD), we tested DNA from B. quintana ATCC 51694 in quantities ranging from 20 fg to 0.625 fg per reaction in 2-fold serial dilutions, with 8 replicates per dilution. All replicates down to 5 fg were detected, whereas 7/8 replicates of both the 2.5 fg and 1.25 fg were detected, and 3/8 replicates of the 0.625 fg were detected. We analyzed LOD by probit regression within the MedCalc software (MedCalc Software Ltd; Ostend, Belgium). The resulting LOD was 2.750 ± 0.859 fg/reaction (1.6 genome equivalents based on B. quintana’s median genome size in GenBank).

To assess the LOD in blood, we tested EDTA blood spiked with B. quintana OK90–268 at 3 concentrations in 15 to 20 replicates. We grew isolates on sheep blood agar for 48 to 72 hours at 37°C with 5% CO_2, and prepared standardized cell suspensions using a turbidity meter. We then spiked the suspensions into EDTA whole blood from healthy human donors (Innovative Research; Novi, MI) to the final concentrations of 1.2 × 10², 1.2 × 10³, and 1.2 × 10⁴ colony forming units (CFU/mL). B. quintana DNA was detected in 20/20 replicates at both higher concentrations and in 6/15 replicates of the 1.2 × 10² CFU/mL concentration, for an estimated LOD of 1.2 × 10³ CFU/mL.We also tested EDTA whole blood from healthy human donors (Reprocell; Beltsville, MD) to ensure B. quintana was not detected in the blood of healthy individuals. All (10/10) samples were undetected.

As B. quintana-positive clinical blood samples were not available to assess diagnostic sensitivity, we spiked 5 isolates of B. quintana (CA15–0058, CA15–0053, CO20–0321, CO20–0297, and CO20–0256) into EDTA whole blood from healthy human donors to a final concentration approximately 10 times the LOD (1.3 × 10⁴ CFU/mL). We froze the samples at ≤−65°C, extracted DNA, and ran PCR in duplicate. All (10/10) samples were detected with an average Ct of 31.46 (SD 0.74).

To evaluate reproducibility, 2 of these spiked blood samples were tested in 6 runs by 2 operators over 5 days. The coefficient of variation of these results was 5.5%. Additional extraction methods were evaluated by spiking blood with B. quintana OK90–268 and extracting five replicates separately using either the QIAamp DNA Mini Kit protocol, the Roche MagNA Pure 96 instrument, or the Roche MagNA Pure 24 instrument (Roche Diagnostics; Indianapolis, IN). The coefficients of variation were 0.6% when comparing samples extracted with the QIAamp kit to the MagNA Pure 96 instrument, and 2.4% when comparing the MagNA Pure 96 and MagNA Pure 24 instruments (Table 2).

Table 2.

Summary of assay performance characteristics.

Analytical specificity	100% (0/141 non-B. quintana bacteria detected)
Analytical sensitivity (limit ofdetection)	2.750 ± 0.859 fg/reaction (1.6 genome equivalents/reaction)
Diagnostic specificity	100% (0/42 samples from patients with other pathogens detected; 0/10 samples from healthy donors detected)
Diagnostic sensitivity	100% (10/10 human blood samples spiked with B. quintana detected)
Reproducibility	5.5% coefficient of variation (2 separately spiked samples tested in 6 runs by 2 operators)
Extraction platforms: comparison of Qiagen QIAamp DNA Mini Kit and Roche MagNA Pure 96 instrument with DNA and Viral Nucleic Acid Small Volume Kit	0.6% coefficient ofvariation (5 spiked samples extracted in parallel and run in the same run)
Extraction platforms: comparison of Roche MagNA Pure 96 instrument with DNA and Viral Nucleic Acid Small Volume Kit and Roche MagNA Pure 24 instrument with Total NA Isolation Kit	2.4% coefficient ofvariation (5 spiked samples extracted in parallel and run in the same run)

The CDC Institutional Review Board (protocol #7102) approved the use of residual specimens for assay development and validation. We used residual EDTA blood previously identified positive for other bacterial pathogens to assess diagnostic specificity, including Borrelia burgdorferi, Anaplasma phagocytophilum, Ehrlichia chaffeensis, Ehrlichia muris subsp. eauclairensis, Leptospira kirschneri, Staphylococcus aureus, Streptococcus pneumoniae, Legionella pneumophila, and Rickettsia rickettsii [15,16]. All (42/42) bloods were undetected.

In conclusion, the vomp region of B. quintana targeted by real-time PCR in this study is highly specific and sensitive (Table 2), likely due to the presence of at least 2 gene copies per genome. This assay is specific to B. quintana and does not require additional testing to obtain a species-level diagnosis [6,10,17,18]. A limitation of this validation is the use of spiked samples to calculate diagnostic sensitivity. Real-time PCR targeting the vomp region provides a rapid and specific adjunct to blood culture for the diagnosis and clinical management of B. quintana bloodstream infections.