Comparison of TaqMan PCR Assays for Detection of the Melioidosis Agent Burkholderia pseudomallei in Clinical Specimens

TEXT

Melioidosis is an emerging infectious disease caused by the Gram-negative soil bacterium Burkholderia pseudomallei (13). Infection is usually via cutaneous inoculation or inhalation, and disease presentations range from asymptomatic, to localized skin infection or pneumonia, to disseminated disease with abscesses in multiple organs, resulting in fulminant sepsis with mortality rates of >50% (9). Increasing numbers of cases are being observed globally, likely reflecting both improved diagnostics (12) and a true increase in cases in those living in or traveling from regions where melioidosis is endemic (11, 16, 17). Culture remains the “gold standard” for diagnosis of melioidosis, but is problematic due to sensitivity issues, lack of familiarity with B. pseudomallei in laboratories in areas where the disease is nonendemic (17), and poor specificity of biochemical tests (18). Subsequent delayed diagnosis can result in life-threatening delays in appropriate antimicrobial therapy (9).

Other diagnostic techniques for B. pseudomallei detection include antigen detection by immunofluorescence microscopy (34) or latex agglutination (3); however, these suffer from reduced sensitivity or dependence on an initial culture step, delaying time to diagnosis (1). Culture is also required for matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry (14). Serological diagnosis is unreliable due to background antibody levels in areas of endemicity and low sensitivity and specificity (10, 33).

While high-throughput sequencing technologies are not yet feasible for routine diagnostics (23), various other molecular platforms have been developed for rapid identification of B. pseudomallei. These include DNA microarrays (25), gene sequencing (15, 32), isothermal DNA amplification (7), and real-time PCR assays targeting specific regions of the B. pseudomallei genome (2, 5, 19–21, 26–30) (see Table S1 in the supplemental material).

Despite this abundance of published assays, the techniques used for validating criteria vary substantially between studies. Furthermore, few have been evaluated directly on clinical samples (8, 20, 27, 28). Thus, it is difficult to determine which of these assays would perform best in a diagnostic or forensic setting, in which high specificity and sensitivity with a low limit of detection (LoD) are paramount.

The aim of this study was to focus on real-time TaqMan PCR assays and assess the best available genomic target to date for B. pseudomallei detection in clinical samples. Seven real-time PCR assays were directly compared by assessing their analytical and diagnostic specificities and sensitivities (4, 6).

Based on superior reported specificity and LoD (see Table S1 in the supplemental material), four previously published real-time TaqMan PCR assays were included, namely, TTS1-orf2 (22) and TTS1-orf11 (29) targeting the type III secretion (TTS) system gene cluster, lpxO (19) and 8653 (27) (Table 1). The mprA target based on a previously published PCR assay (21) was validated with a TaqMan probe (Primer Express 3.0 software; Life Technologies). A multiplex TaqMan assay targeting the Yersinia-like fimbrial/Burkholderia thailandensis-like flagellum and chemotaxis (YLF/BTFC) region (31) was also assessed, together with a newly developed dual-probe assay, 266152, which targets the methylmalonate-semialdehyde dehydrogenase locus and differentiates between B. pseudomallei and B. thailandensis (Table 1).

For analytical sensitivity and specificity, real-time PCR was carried out as previously reported (5). In brief, PCR mixtures consisted of 10 μl of 900 nM primers, 200 nM probe, 1× Applied Biosystems genotyping master mix (Life Technologies), and 0.5 ng template DNA. Thermal cycling was performed on an AB 7900HT sequence detection system (Life Technologies) at 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 58°C for 1 min. The 266152 assay was performed with 1× Applied Biosystems universal master mix and 300 nM primers. The LoD was defined as the lowest possible template concentration detectable with 95% probability with at least 61/64 positive replicates (4, 6). Specificity was evaluated by screening 365 B. pseudomallei strains and 115 non-B. pseudomallei strains (with 71 Burkholderia spp. and 44 bacterial species of differential diagnostic importance [see full list in the supplemental material]). A threshold cycle (C_T) value of 40 was the declared cutoff for a positive result.

Clinical evaluation was performed by screening each assay across 50 clinical specimens (22 sputum, 20 blood, and 8 urine samples) from 22 patients with acute melioidosis, 59% of whom presented with pneumonia and 18% with genitourinary infection, with 55% overall being bacteremic. These samples were part of a study comparing DNA extraction techniques for molecular detection of B. pseudomallei in clinical specimens (24); while all samples were from culture-confirmed melioidosis cases, not every specimen from each patient was cultured for B. pseudomallei. Blood samples from 22 healthy volunteers were used as negative controls. DNA was extracted using the QIAamp DNA minikit (Qiagen, Australia) and PureGene blood core kit B (Qiagen). PCR conditions were as previously described (24), using the 1× Applied Biosystems environmental master mix. Samples were declared positive if 2/2 duplicates had C_T values of <40.

All assays showed high analytical specificity, with the TTS1-orf2, 8653, mprA and 266152 assays being 100% specific for both B. pseudomallei and nontarget strains (Table 2). The lpxO assay showed reduced specificity as it also amplified 14/23 Burkholderia mallei DNA targets. The TTS1-orf2 and YLF/BTFC assays had the lowest LoD of 5 genome equivalents (GE) per reaction (Table 2). Variations from previously reported LoD were apparent from our data and are likely due to the strict LoD definition we used to determine the lowest possible template concentration detectable with 95% probability (4, 6). The low LoD of TTS1-orf2 and YLF/BTFC assays was also reflected in their high diagnostic sensitivity. Of the 43 clinical samples that tested positive for B. pseudomallei by one or more assays, 42 were positive by several assays and one only by TTS1-orf2. The mprA assay performed least well in the clinical evaluation and had a significantly lower detection rate than the TTS1-orf2, YLF/BTFC, and lpxO assays (McNemar's test for paired samples, P < 0.001 for all, 2-tailed) (Table 2).

These data support the TTS1-orf2 assay as the best-performing assay to date for direct detection of B. pseudomallei in clinical specimens. The YLF/BTFC multiplex assay also performed well and in addition to B. pseudomallei detection provides information on the potential geographic origin of the tested isolate, with BTFC being common in Australia (88%) but rare in Thailand (2%) (31). As the YLF locus was also found in some close relatives of B. pseudomallei (A. Tuanyok, unpublished data), this assay should be used with caution on environmental samples.

As an additional informative assay, we included a dual-probe assay discriminating between B. pseudomallei and B. thailandensis. This assay was designed for screening culture isolates and showed high specificity. No differences in clinical detection rates were found when including both probes or only the B. pseudomallei-specific probe. Due to cross-hybridization, this assay should be used with caution on environmental samples where potentially both B. pseudomallei and B. thailandensis could be present.

Although the probe chemistry was tested as in the original publications, the PCR conditions were not adjusted to the original optimized conditions for each assay; we used an adaptation of the TaqMan universal PCR protocol (Life Technologies), with the same conditions for all assays. The inferior performance of the mprA assay in the clinical evaluation may reflect that the original assay was not probe based and had an annealing temperature of 68°C, which differs from the conditions we used.

In summary, we have shown that the TTS1-orf2 assay provides the best available molecular target to date for B. pseudomallei detection directly from clinical samples. Furthermore, the YLF/BTFC multiplex assay, which provides additional information on the possible geographic origin of a B. pseudomallei isolate, also showed high diagnostic sensitivity.