Comparison of Established Diagnostic Methodologies and a Novel Bacterial smpB Real-Time PCR Assay for Specific Detection of Haemophilus influenzae Isolates Associated with Respiratory Tract Infections

Abstract

Haemophilus influenzae is a significant causative agent of respiratory tract infections (RTI) worldwide. The development of a rapid H. influenzae diagnostic assay that would allow for the implementation of infection control measures and also improve antimicrobial stewardship for patients is required. A number of nucleic acid diagnostics approaches that detect H. influenzae in RTIs have been described in the literature; however, there are reported specificity and sensitivity limitations for these assays. In this study, a novel real-time PCR diagnostic assay targeting the smpB gene was designed to detect all serogroups of H. influenzae. The assay was validated using a panel of well-characterized Haemophilus spp. Subsequently, 44 Haemophilus clinical isolates were collected, and 36 isolates were identified as H. influenzae using a gold standard methodology that combined the results of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) and a fucK diagnostic assay. Using the novel smpB diagnostic assay, 100% concordance was observed with the gold standard, demonstrating a sensitivity of 100% (95% confidence interval [CI], 90.26% to 100.00%) and a specificity of 100% (95% CI, 63.06% to 100.00%) when used on clinical isolates. To demonstrate the clinical utility of the diagnostic assay presented, a panel of lower RTI samples (n = 98) were blindly tested with the gold standard and smpB diagnostic assays. The results generated were concordant for 94/98 samples tested, demonstrating a sensitivity of 90.91% (95% CI, 78.33% to 97.47%) and a specificity of 100% (95% CI, 93.40% to 100.00%) for the novel smpB assay when used directly on respiratory specimens.

INTRODUCTION

Haemophilus influenzae is a Gram-negative, coccobacillary, facultatively anaerobic bacterium, which is considered a normal human commensal that frequently colonizes the upper respiratory tract (URT) (1, 2). H. influenzae is also an important human pathogen that has been associated with invasive disease and a range of upper and lower respiratory tract infections (LRTIs). Other Haemophilus species that are occasionally isolated from the URT include Haemophilus haemolyticus, Haemophilus parainfluenzae, and Haemophilus parahaemolyticus; however, these species are rarely associated with human infections (3).

Typeable H. influenzae produces six distinct antigenic capsules (a to f), with type b being historically associated with a significant burden of invasive disease in children prior to the widespread use of the Haemophilus influenzae type b (Hib) vaccine (4, 5). Since introduction of the vaccine, nontypeable strains of H. influenzae have caused the majority of invasive disease with a lesser incidence of disease caused by other capsule types (6). Nontypeable strains are also commonly associated with noninvasive diseases, such as otitis media, sinusitis, pneumonia, and exacerbations of chronic obstructive pulmonary disease. As such, there is a need to develop rapid diagnostic assays that have the ability to detect all strains of H. influenzae, regardless of their antigenic status.

Traditional culture- and phenotypic-based methodologies for the identification of H. influenzae are slow and in many cases cannot differentiate H. influenzae from the closely related Haemophilus haemolyticus or other Haemophilus spp. (7–9). Recent studies have demonstrated that phenotypic methods, such as the API NH (bioMérieux, Nürtingen, Germany) or Vitek NH cards (bioMérieux), can fail in 1% to 10% of the analyses (3, 10, 11).

In recent years, a number of novel approaches have been described for the detection of H. influenzae, including matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) and nucleic acid-based diagnostic assays (12–15). As mass spectrometers are still relatively expensive and require specialist training, they are not available in all diagnostic and clinical laboratories. More importantly, the requirement for bacterial culture prior to analysis by MALDI-TOF MS takes a minimum of 1 day and often longer. Several real-time PCR diagnostic assays exist for specific detection of H. influenzae (14–17). However, many of these diagnostic assays lack specificity, which can pose challenges for the clinical laboratory (15). Accordingly, there is a need for the identification and validation of novel molecular targets that can be used for rapid and specific detection of H. influenzae, allowing appropriate targeted antimicrobial therapy to be administered.

This study, which was designed in accordance with the minimum information for publication of quantitative real-time PCR experiments (MIQE) guidelines (18), reports on the design and performance of an internally controlled real-time PCR diagnostic assay using a novel target sequence to rapidly detect and specifically identify culture-positive isolates of H. influenzae. This novel smpB real-time PCR diagnostic assay has also been applied to directly detect H. influenzae from respiratory samples, including sputum, bronchoalveolar lavage, and endotracheal aspirates.

MATERIALS AND METHODS

Ethics statement.

We used anonymized routine specimens surplus to clinical requirements for assay validation, adhering to a governance framework agreed by and with a University College London (UCL) ethics agreement relating to the use of specimens surplus to clinical needs.

In silico diagnostics target identification.

We evaluated a number of previously described gene targets. Specifically, publicly available nucleotide sequences for potential diagnostics targets were retrieved from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov), the functional gene pipeline and repository (http://fungene.cme.msu.edu/), and the transfer-messenger RNA (tmRNA) website (http://bioinformatics.sandia.gov/tmrna/). In silico analysis of each molecular target was performed following alignments of nucleotide sequences using Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/).

Bacterial strains, culture media, and growth conditions.

We used a panel of 32 well-characterized Haemophilus and closely related species and strains (see Table S1 in the supplemental material) and 30 other bacteria (see Table S2 in the supplemental material). These species and strains were purchased from the German Collection of Microorganisms and Cell Cultures (Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH [DSMZ], Braunschweig, Germany), the National Collection of Type Cultures (NCTC, Public Health England, Salisbury, United Kingdom), the American Type Culture Collection (ATCC, provided by LGC standards, Middlesex, United Kingdom), and the Culture Collection, University of Göteborg (CCUG, Sweden). All Haemophilus species and strains were cultured on Columbia chocolate agar (Oxoid, Hampshire, United Kingdom) at 37°C with 5% CO₂ for 18 to 24 h.

Genomic DNA isolation and quantification.

Genomic DNA (gDNA) from Haemophilus isolates and clinical samples were isolated using a modified procedure combining mechanical lysis (IDI lysis kit; Becton Dickinson, Canada) and purification using a quick gDNA kit (Zymo Research, Irvine, CA, USA). Briefly, a loop of culture was resuspended in 250 μl IDI lysis buffer. The suspension was transferred to a GeneOhm lysis tube and bead beaten (Mini-BeadBeater-16; Stratech, United Kingdom) for 3 min. After bead beating, 200 μl of the supernatant was transferred to a Zymo-Spin column in a collection tube, and steps 2 to 5 of the procedure for purification of total DNA from cell suspensions were followed, according to the manufacturer's instructions. For all other bacterial species tested, DNA was provided from stocks held within this laboratory (Nucleic Acid Diagnostics Research Laboratory [NADRL], Microbiology, National University of Ireland, Galway).

Genomic DNA concentrations for all species and strains used in this study were determined using the Quant-iT dsDNA high-sensitivity assay kit and the Qubit fluorometer (Invitrogen Corporation, CA, USA), as per the manufacturer's instruction. Prior to use, genomic DNA samples were stored at −20°C.

Real-time PCR primers and hydrolysis probe design.

Following nucleotide sequence alignments of each of the putative target genes selected for in this study, we adhered to general guidelines and recommendations for diagnostic assay oligonucleotides for the design of primers and hydrolysis probes (19). All oligonucleotides used in this study were supplied from Eurofins genomics (Essenberg, Germany). All oligonucleotide primers used in this study were designed to have a melting temperature (T_m) of 58 to 61°C, and all oligonucleotide hydrolysis probes a T_m of 7 to 10°C higher.

For the H. influenzae specific diagnostic assay, PCR primers HaemF and HaemR (Table 1) were designed to amplify a 160-bp fragment of the smpB gene, the selected target gene for development. The H. influenzae probe was labeled with 6-carboxyfluorescein (FAM) and black hole quencher 1 (BHQ1). The internal amplification control (IAC) PCR primers, IAC_Fw and IAC_Rv (Table 1), were designed to amplify a 206-bp region of the Bacillus subtilis subsp. spizizenii strain W23 ssrA gene. The IAC probe was labeled with Cy5 and BHQ2.

TABLE 1.

Oligonucleotide primers and probes developed in this study

Name	Function	Sequence 5′ to 3′	Accession no. (nucleotide position [bp])
HaemF	Forward smpB real-time PCR assay primer	ATTAAATGTTGCATCAACGC	NC_000907.1 (213–232)
HaemR	Reverse smpB real-time PCR assay primer	GACTTTTGCCCACGCAC	NC_000907.1 (356–372)
H. influenzae smpB P1	smpB real-time PCR probe	FAM-ACGRTTTTACCATAGTTGCACTTTCTC-BHQ1	NC_000907.1 (317–343)
IAC_Fw	Forward IAC primer	AACGTAGCATTAGCTGC	HG519928.1 (111–127)
IAC_Rv	Reverse IAC primer	CTCATCTTCTTGCCTGC	HG519928.1 (260–276)
Bacillus subtilis P1	IAC probe	Cy5-CACATCCAAGTAGGCTACGCT-BHQ2	HG519928.1 (179–199)

Development of IAC for real-time PCR.

To avoid false-negative results due to PCR inhibition, thermocycler malfunction, and/or reagent problems, a noncompetitive IAC assay targeting the B. subtilis subsp. spizizenii strain W23 ssrA gene, was incorporated into the real-time PCR diagnostic assays (20). Titration experiments were performed to determine the optimum level of B. subtilis DNA to incorporate per reaction to ensure that the IAC was always detected yet have the least impact on diagnostic assay robustness. Five hundred cell equivalents of B. subtilis DNA per reaction was determined as the optimum concentration of IAC target DNA to include in the duplex real-time PCR assay.

Development of duplex real-time PCR H. influenzae diagnostic assay.

To demonstrate the specificity and sensitivity of the Duplex real-time PCR, reactions were carried out on the LightCycler 480 using the LightCycler 480 Probes Master kit (Roche Diagnostics, Basel, Switzerland). The optimized PCR mix contained 2× LightCycler 480 Probes Master (6.4 mM MgCl₂), H. influenzae forward and reverse primer (0.5 μM final concentration), 6-carboxyfluorescein (FAM)-labeled probe (0.4 μM final concentration), IAC forward and reverse primer (0.25 μM final concentration), Cy5-labeled probe (0.2 μM final concentration), and template DNA (target, 5 μl; IAC, 2 μl) adjusted to a final volume of 20 μl with the addition of nuclease-free distilled water (dH₂O). The B. subtilis internal control DNA was diluted to contain 500 genome equivalents per 2 μl and all other DNA used in this study was diluted to contain ∼10⁴ genome equivalents per 5 μl.

The cycling parameters consisted of 10 min incubation at 95°C to activate the Taq, 50 cycles of 95°C for 10 s and 63°C for 30 s, followed by a single cooling step at 40°C for 10 s. The temperature ramp rate on the LightCycler 480 was 4.4°C/s while heating and 2.2°C/s while cooling. A color compensation file was generated to avoid fluorescence leaking from channel to channel prior to experimental analysis on the LightCycler 480, as per the manufacturer's instruction.

Clinical isolate evaluation.

To evaluate the performance of the duplex real-time PCR diagnostic assays developed in this study, a panel of 44 recent clinical isolates of the genus Haemophilus were collected (Table 2). These isolates were cultured from clinical samples (sputum, n = 29; endotracheal aspirate, n = 2; nasal swabs, n = 8; bronchoalveolar lavage, n = 2; eye swabs, n = 2; and unknown, n = 1) using standard laboratory procedures and identified using MALDI-TOF MS (Bruker Daltonics, Bremen Germany) and MALDI-Biotyper 3.1 software. In accordance with previously published guidelines, only scores of 1.9 or greater were considered reliable for species identification (21, 22).

TABLE 2.

Clinical isolates used in this study

No. of clinical isolates	Real-time PCR fucK assay	Real-time PCR smpB assay	MALDI-TOF identification	Serotype	Antimicrobial resistance(s) detecteda	Beta-lactamase
6	H. influenzae	H. influenzae	H. influenzae	Nontypeable	SXT	−b
2	H. influenzae	H. influenzae	H. influenzae	Nontypeable	AMP, ERY	TEM
1	H. influenzae	H. influenzae	H. influenzae	Nontypeable	AMP, CRO, TET	−
1	H. influenzae	H. influenzae	H. influenzae	Nontypeable	AMP, AMC	−
1	H. influenzae	H. influenzae	H. influenzae	Nontypeable	AMP, AMC, ERY	−
1	−	H. influenzae	H. influenzae	Nontypeable	AMP, TET	TEM
5	H. influenzae	H. influenzae	H. influenzae	Nontypeable	AMP	TEM
17	H. influenzae	H. influenzae	H. influenzae	Nontypeable	Not detected	−
2	−	H. influenzae	H. influenzae	Nontypeable	Not detected	−
3	−	−	H. parainfluenzae		Not detected	−
1	−	−	H. parainfluenzae		CRO, MXF	−
1	−	−	H. parainfluenzae		ERY	−
1	−	−	Haemophilus parahaemolyticus		Not detected	−
1	−	−	H. parahaemolyticus		TET	−
1	−	−	H. parahaemolyticus		TET, ERY	−

^aSXT, trimethoprim-sulfamethoxazole; AMP, ampicillin; ERY, erythromycin; CRO, ceftriaxone; TET, tetracycline; AMC, amoxicillin-clavulanate; MXF, moxifloxacin.

^b−, negative.

The antimicrobial susceptibilities of the isolates to ampicillin, amoxicillin-clavulanate, ceftriaxone, erythromycin, imipenem, moxifloxacin, tetracycline, and trimethoprim-sulfamethoxazole were determined by Etest, and results were interpreted according to EUCAST guidelines (23). All isolates were tested for beta-lactamase activity using Cefinase paper disks (Becton Dickinson). The beta-lactamase variant in positive isolates was identified by PCR for bla_TEM and bla_ROB as described previously (24, 25).

Subsequently, genomic DNA was isolated, as outlined above, from pure isolates of confirmed Haemophilus species (Table 2). These DNA samples were then tested blindly in triplicate with an H. influenzae real-time PCR diagnostic assay previously described in the literature targeting the fucK gene (15). This fucK assay was chosen for evaluation of clinical isolates and samples, as it has previously been demonstrated to be highly specific and sensitive for the detection of H. influenzae from clinical samples (15). For epidemiological purposes, any isolate that was determined to contain H. influenzae using the smpB real-time PCR assay was also serotyped using a previously described real-time PCR approach (17).

Direct clinical RTI sample evaluation.

To demonstrate the suitability for using the assay developed in this study directly on clinical samples (sputum, n = 67; endotracheal aspirates, n = 19; bronchoalveolar lavage, n = 12), a panel of 98 anonymized surplus specimens (Table 3) from patients with LRTIs was collected. Using 200 μl of sample, 10-fold serial dilutions were carried out down to 10⁻⁵ in phosphate-buffered saline. A neat sample (50 μl) and each dilution were spread onto Colombia blood agar (CBA), Colombia agar with chocolated horse blood (CHOC), and Brilliance UTI Clarity agar (Oxoid). CBA and urinary tract infection (UTI) plates were incubated at 37°C for 18 h, and CHOC plates were incubated in a 5% CO₂ environment at 37°C for 18 h. Distinct colonies were identified using MALDI-TOF MS (Bruker Microflex LT) using MALDI Biotyper version 3.1 with default settings. From 300 μl of the sample, the total nucleic acid was isolated in accordance with the procedure outlined above (genomic DNA isolation and quantification). Nucleic acids isolated from clinical samples were then blindly tested using the previously described fucK real-time PCR diagnostic assay and also the novel H. influenzae smpB real-time PCR diagnostic assay.

TABLE 3.

Clinical respiratory tract samples

No. of clinical samples	fucK real-time assay identificationa	smpB real-time assay identification	MALDI-TOF identification
25	H. influenzae	H. influenzae	H. influenzae
7	H. influenzae	H. influenzae	Haemophilus species not detected
1	H. influenzae	−	H. influenzae
1	−	H. influenzae	H. influenzae
3	−	−	H. influenzae
7	H. influenzae	H. influenzae	H. parainfluenzae and/or H. parahaemolyticius
13	−	−	H. parainfluenzae and/or H. parahaemolyticius
41	−	−	Haemophilus species not detected

^a−, negative.

RESULTS

In silico diagnostics target identification.

There is currently no single nucleic acid diagnostics target described in the literature that can unambiguously identify H. influenzae. As such, in this study, a number of gene targets, including ssrA and lepA and genes that have previously been described in the literature as suitable bacterial species specific molecular diagnostics targets, were evaluated in silico (26–28). A putative novel diagnostics gene target, smpB, selected and identified by the NADRL, was also evaluated in silico. From this in silico analysis, the ssrA and lepA genes were deemed unsuitable for further use due to the nucleotide sequence similarity observed between H. influenzae and other closely related Haemophilus species. However, the smpB gene demonstrated sufficient intragenic nucleotide sequence variation between closely related Haemophilus species to allow for the design of H. influenzae specific probes.

Development of IAC.

A noncompetitive approach was utilized for the development of the IAC in this study (20). This means that in order for a result to be considered valid using this assay, a positive signal must be obtained in the Cy5 detection channel on the LightCycler 480. For the purposes of this study, B. subtilis DNA was spiked into the PCR master mix to act as an internal control target.

Specificity and sensitivity of the diagnostic assays.

The specificity of the real-time PCR diagnostic assay developed in this study was confirmed using the specificity panel listed in Tables S1 and S2 in the supplemental material. For inclusivity and exclusivity testing, each sample was tested in triplicate at a concentration of ∼1 × 10⁴ genome equivalents. The smpB-based assay specifically detected all 11 H. influenzae isolates; conversely, no other species of the Haemophilus genus or other bacteria were detected. The specificity of the IAC assay was also tested against the Haemophilus and other bacterial panels and was specific for B. subtilis DNA. A typical representation of the amplification curves generated in each of the analysis channels for this duplex assay is provided in Fig. S1A and B in the supplemental material.

The lower limit of detection (LOD) of the assay developed was established using probit regression analysis. With an estimated genome size of 1.8 million base pairs, each H. influenzae cell contains approximately 1.96 fg DNA (29, 30). Genomic DNA was quantified, and 12 replicates of each of 15, 10, 7.5, 5, 3.75, 2.5, 1.25, and 0.5 H. influenzae genome equivalents were tested. LODs of 6.38 were determined (95% probability). The IAC, at a concentration of 500 genome equivalents per reaction, was included in all samples during sensitivity testing and detected as expected.

Clinical isolate evaluation.

For the purpose of this study, it was decided to evaluate the performance of the smpB assay against a gold standard that combined results of MALDI-TOF MS and the fucK assay. Using this approach, an isolate was identified as H. influenzae if a positive result was observed for either the MALDI-TOF or the fucK assay. An isolate that was not identified as H. influenzae using MALDI-TOF and was also negative with the fucK assay was defined as negative for H. influenzae.

We collected a panel of 44 recent clinical isolates of Haemophilus spp. to evaluate the performance of the duplex real-time diagnostic assays developed in this study (Table 2). Using the gold standard criteria outlined above, 36/44 isolates were identified as H. influenzae and 8 isolates were identified as other Haemophilus species. Genomic DNA from these isolates was also tested with the novel smpB diagnostic assay, and 100% concordance was observed with the gold standard method, identifying the same 36 isolates as H. influenzae. This demonstrated a sensitivity of 100% (95% confidence interval [CI], 90.26% to 100.00%) and a specificity of 100% (95% CI, 63.06% to 100.00%) for the smpB assay.

For epidemiological purposes, we also determined the capsular serotypes of all H. influenzae isolates and determined their antimicrobial susceptibility profiles. Using a previously developed capsular serotyping method, it was determined that all the H. influenzae isolates were nontypeable. Furthermore, 21 of the 44 isolates were resistant to one or more antibiotics. The most common resistance phenotypes were ampicillin (25%), trimethoprim-sulfamethoxazole (13.6%), and erythromycin (11.4%). ROB beta-lactamase was not detected in any isolates, but 18.2% of isolates were positive for TEM beta-lactamase.

Direct clinical RTI sample evaluation.

We determined that our developed novel smpB real-time PCR diagnostic assay may be used directly to test clinical samples for H. influenzae by evaluating a panel of 98 respiratory specimens (Table 3). Using traditional culture methods, all of the specimens that contained Haemophilus species were determined to contain a minimum of 1 × 10⁴ CFU/ml. Using the same criteria as outlined above for a gold standard reference, 44/98 clinical specimens were determined to contain H. influenzae, and 54 were negative for H. influenzae. Using nucleic acid purified from the above respiratory samples, the novel smpB assay identified H. influenzae in 40/98 clinical samples, and the remaining 58 samples were negative for H. influenzae. This demonstrated a sensitivity of 90.91% (95% CI, 78.33% to 97.47%) and a specificity of 100% (95% CI, 93.40% to 100.00%) for the smpB assay when used directly on respiratory specimens.

DISCUSSION

Lower RTIs are among the top four major causes of morbidity and mortality worldwide, causing 3.1 million deaths in 2012 (31). In the context of community-acquired pneumonia, H. influenzae has been considered a significant cause of human infection (32, 33). H. influenzae is also an important pathogen associated with invasive infection and sepsis, making a rapid and reliable method for its detection highly desirable.

Traditional culture-based methodologies for the detection of H. influenzae are slow to perform and lack specificity. Furthermore, false-negative reporting of results can occur if antimicrobial therapy has been administered prior to collection of the clinical samples. In response to this need, several real-time PCR-based diagnostic assays have been described in the literature, including those that target the fucK, hpdA, bexA, ompP2, and P6 genes, respectively (9, 15, 16, 34). However, none of these targets is 100% specific for the detection of H. influenzae. For example, in the case of hpdA and ompP2, they have been found to cross-react with other Haemophilus species resulting in the reporting of false-positive results (9, 16). Also, the fucose kinase operon containing the fucK gene can be deleted in some H. influenzae strains, resulting in the reporting of false-negative results (35). This prompts the need for the identification of novel molecular targets for the specific detection of H. influenzae.

In this study, a number of potential diagnostics targets were evaluated in silico, including the ssrA, lepA, and smpB genes. These gene targets were chosen, as they are either present in all bacteria sequenced to date and/or have previously been demonstrated to be suitable as bacterial species-specific molecular-based diagnostics targets (26, 28, 36, 37). Of these, only the smpB gene demonstrated sufficient in silico nucleotide sequence heterogeneity to allow for the design of an H. influenzae-specific diagnostic assay. The smpB gene, which codes for the RNA binding protein small protein B (SmpB), has been identified in all bacterial species to date (38). It is considered an essential component of quality control in bacteria, as it facilitates the binding of tmRNA to stalled ribosomes, which in turn allows for removal of incomplete polypeptides from the cell (39, 40). As SmpB is considered essential for the correct functioning of tmRNA in a bacterial cell, the smpB gene is considered among a core set of genes necessary to sustain bacterial viability in Haemophilus species (41).

A rapid, specific, and sensitive internally controlled real-time PCR diagnostic assay targeting the H. influenzae smpB gene was then developed and validated against a panel of well-characterized culture collection species and strains. Subsequently, a panel of culture-positive clinical isolates of various antimicrobial susceptibilities was evaluated with this smpB assay. The results obtained were then compared to a gold standard reference method, which combined results from MALDI-TOF MS and a previously described fucK real-time PCR assay. A gold standard combining two methods was required due to the fact that currently, no single real-time PCR assay can be used to reliably detect all strains of H. influenzae. The results of the smpB real-time PCR diagnostic assay demonstrated 100% agreement compared to the gold standard method. Some discordance would have been observed if only the results of the real-time PCR assays were compared, as 3 isolates were determined to be negative with the fucK assay but were positive using the smpB assay. This may be due to a deletion of the fucose operon in these isolates, which further demonstrates the need for evaluation and validation of novel molecular targets for the specific identification of H. influenzae.

For epidemiological purposes, the capsular serotype and mechanisms of antimicrobial resistance for each of these culture-positive isolates that contained H. influenzae as determined by MALDI-TOF MS and the smpB diagnostic assay were also determined. All isolates were determined to be nontypeable H. influenzae, which is consistent with recent findings in Europe (6, 42). The prevalence of antimicrobial resistance was similar to previous reports from the United Kingdom (43, 44). As the sample size here is small and restricted to a single location, no firm conclusions can be drawn; however, it appears prevalence of antimicrobial resistance in this species has stabilized. These results further validate the robustness of the smpB assay developed for this study, as they demonstrate that this novel assay can reliably detect the predominant strains of H. influenzae that are commonly causing LRTI.

Finally, to further demonstrate the potential clinical utility of this method, a panel of clinical LRTI specimens was evaluated using the gold standard reference mentioned above. Using the gold standard criteria, 44/98 specimens were determined to contain H. influenzae whereas the remaining 54 samples were negative for the presence of H. influenzae. Using the novel smpB assay, 40/98 isolates were determined to contain H. influenzae whereas 58 samples were negative for H. influenzae.

If considering only the results of the real-time PCR assays, 96/98 samples were concordant. One additional sample was also positive for H. influenzae using the fucK assay and was not detected using the smpB assay. This discordant result is likely due to the sensitivity of the fucK assay compared to the smpB assay. From the literature, the LOD of the fucK assay is ∼2.5 cells whereas the LOD of the smpB assay presented here is 6.8 cells. One additional specimen was positive for H. influenzae using the smpB assay but was not detected by the fucK assay, which may be a result of a deletion of the fucose operon. As such, if only considering the results of the real-time PCR assays, the smpB assay may report false negatives if very low numbers of H. influenzae are present in a clinical sample. However, the smpB assay has superior specificity than the fucK assay and is also internally controlled, which is important when evaluating clinical samples.

When comparing the results of the real-time PCR assays and MALDI-TOF MS, H. influenzae was identified in an additional 14 clinical specimens using the fucK and smpB. The discrepancy in these results may have been due to the H. influenzae cells in these specimens being nonviable and therefore not detected by culture. In such instances, if a real-time PCR assay was used, there is a possibility of suboptimal treatment for patients. However, it is also possible that these specimens may contain viable but nonculturable cells or that they were not detected during culture due to being present at low CFU in these specimens. Alternatively, the culture and MALDI-TOF MS approach may have misidentified them, either outright or because these samples contained mixed populations of different Haemophilus species. As different Haemophilus species demonstrate highly similar colony morphology on agar plates, mixed populations may result in false-negative results when only one colony is analyzed using MALDI-TOF MS. There were 3 remaining samples identified as H. influenzae by MALDI-TOF MS that were not detected by the fucK or smpB real-time PCR diagnostic assays. After real-time PCR on these samples, PCR products were electrophoresed on a 1.5% agarose gel, and no visible PCR products were observed for either the fucK or smpB or real-time PCR diagnostic assays (data not shown). In the case of the smpB real-time PCR diagnostic assay, the discordance observed was not due to PCR inhibition as the IAC was reliably detected, as such a possible explanation for this discordance is that the purified DNA was degraded between extraction and PCR or that H. influenzae cells were unevenly spread throughout the sputum samples, resulting in H. influenzae DNA not being extracted.

While MALDI-TOF MS is simple to perform, a limiting factor of the technique is the requirement for culture of isolates prior to testing. Methods that depend on conventional bacterial culture are not always successful, particularly if antimicrobial agents have been administered prior to taking patient samples. Furthermore, as may have occurred in this study, false negatives may be reported if low cell counts or mixed cultures of indistinguishable colony morphology are present.

As such, we set out to develop a novel real-time PCR diagnostic assay for the specific detection of H. influenzae and to compare it to other platform technologies for performance and robustness. The diagnostic method developed in this study is the first described high-performance internally controlled duplex PCR assay capable of rapidly detecting all serotypes of H. influenzae with no cross-reaction observed with other culture collection and cultured clinical isolates of Haemophilus species. The method has been validated on a large panel of well-characterized culture collection isolates, culture-positive patient isolates, and directly on patient samples. We believe that this assay's diagnostic performance justifies a further clinical evaluation in a routine clinical microbiology laboratory to confirm the results of this study as well as to investigate its potential for widespread implementation with accompanying analysis of its value—in economic and in terms of clinical relevance.