Real-time detection of Mycoplasma pneumoniae in respiratory samples with an internal processing control

Mycoplasma pneumoniae is a common cause of community-acquired pneumonia transmitted by aerosol or close contact (Friedlaender et al., 1976). Major outbreaks occur periodically and the severity of clinical symptoms ranges from asymptomatic to pronounced pneumonia (Clyde, 1993). A wide variety of extrapulmonary manifestations have been reported such as haematological, gastrointestinal, musculoskeletal, dermatological and neurological complications (McMillan, 1998). Traditionally, serological methods for detecting specific IgM, IgG and IgA have been used to establish diagnosis of infection. However, detectable levels of antibodies may not be attained until 710 days after the onset of symptoms (Sillis, 1993; McMillan, 1998). Culture requires specialized media, is time-consuming (up to 21 days) and is rarely undertaken in a routine context.

Due to the vast reduction in time in comparison with culture, PCR has been used increasingly for M. pneumoniae detection. Several gene targets have been used for amplification, including the 16S rRNA gene, the elongation factor tuf, the P1 cytadhesin gene and repetitive elements located within the latter (Razin, 1994; Ursi et al., 2003; Michelow et al., 2004; Miyashita et al., 2004; Nour et al., 2005). The P1 cytadhesin gene encodes the 169 kDa P1 protein of M. pneumoniae. This protein is the major virulence and adhesin factor of M. pneumoniae and is located at the terminal attachment organelle, wherein it attaches to host cells (Svenstrup et al., 2002). Although similar genes are found in some other mycoplasma species, highly conserved regions of the sequence are unique to M. pneumoniae and this gene is therefore an attractive target for the design of species-specific PCR primers (Su et al., 1987). The use of the Roche LightCycler with fluorescence resonance energy transfer hybridization probes has been applied to the detection of several bacterial pathogens including Chlamydia pneumoniae (Mygind et al., 2001), Neisseria gonorrhoeae (Whiley et al., 2002), Bordetella spp. (Cloud et al., 2003) and M. pneumoniae (Ursi et al., 2003). Multiplex assays have also been described that enable the simultaneous detection of M. pneumoniae with other pathogens (Khanna et al., 2005; Raggam et al., 2005; Stralin et al., 2005). The purpose of this study was to establish a reliable, specific, sensitive and quantitative real-time PCR to detect M. pneumoniae in clinical specimens and to improve the current clinical diagnostic service. An internal processing control (IPC) was included to highlight amplification failure of the reaction due to inhibition.

Clinical samples. A panel of 167 clinical respiratory samples (sputa and bronchoscopy specimens, group A), urine and acute and convalescent serum samples were taken on admission to hospital in 19911992 as part of a prospective clinical trial. Patients were excluded from the study if they were undergoing antibiotic therapy immediately prior to presentation at hospital. Samples were taken from adult patients (mean age 46 years, SD±19·6, 95 % CI 41·451·4, range 1880) with radiologically confirmed acute community-acquired pneumonia and those of proven aetiology were made available for retrospective examination in this study. Clinical samples from the patients had previously been examined immediately after collection for evidence of a range of respiratory pathogens by standard culture or serological methods as shown in Table 1. An additional eight sputum specimens (group B) from patients with pneumonia that had been serologically confirmed as M. pneumoniae positive were included in the study; four of these samples were also confirmed as positive for M. pneumoniae by culture. Samples were then frozen at 80 °C and later tested for Legionella and other atypical pathogens, including serological testing for M. pneumoniae IgG and IgM antibodies by indirect immunofluorescence (Wreghitt & Sillis, 1985). After several years, respiratory samples were thawed, treated with Sputasol (Oxoid) for 30 min at room temperature, heated for 10 min at 65 °C and heated for 15 min at 100 °C to ensure inactivation of any tubercle bacilli or other pathogens. DNA was then extracted from 200 µl of each respiratory sample using the QIAamp DNA Mini kit (Qiagen). The eluted samples (200 µl) were tested immediately for M. pneumoniae DNA.

Table 1. M. pneumoniae LightCycler PCR results on respiratory samples from adults with pneumonia of proven aetiology Evidence for aetiology: Chlamydophila spp., Coxiella burnetii, RSV, influenza A and influenza B by complement-fixation test; Streptococcus pneumoniae cultured from blood, bronchoscopy samples, sputum or antigen in urine; Enterobacteriaceae, ß-haemolytic streptococci, H. influenzae, Moraxella catarrhalis and Staphylococcus aureus cultured from bronchoscopy samples or sputum; L. pneumophila antigen in urine or serology. Sputum isolates were considered causative if 106 c.f.u. ml1 (Streptococcus pneumoniae) or 107 c.f.u. ml1 (other organisms) were detected with 25 polymorphonuclear cells ml1. M. pneumoniae-positive samples were determined by immunofluorescence.

Reference strains. The following reference strains were included in this study. Mollicute species: Acholeplasma laidlawii NCTC 10116^T, Mycoplasma amphoriforme NCTC 11740^T, Mycoplasma buccale NCTC 10136^T, Mycoplasma faucium NCTC 10174^T, Mycoplasma fermentans NCTC 10117^T, Mycoplasma genitalium NCTC 10195^T, Mycoplasma hominis NCTC 10111^T, Mycoplasma lipophilum NCTC 10173^T, Mycoplasma orale NCTC 10112^T, Mycoplasma penetrans ATCC 55252^T, Mycoplasma pirum NCTC 11702^T, M. pneumoniae NCTC 10119^T, Mycoplasma salivarium NCTC 10113^T, Mycoplasma spermatophilum NCTC 11720^T, Ureaplasma parvum ATCC 33697, Ureaplasma urealyticum NCTC 10177^T. Respiratory species: Actinomyces odontolyticus NCTC 9935^T, Bordetella parapertussis NCTC 5952^T, Bordetella pertussis NCTC 10739^T, Burkholderia cepacia NCTC 10743, Chlamydophila pneumoniae IOL-207, Chlamydophila psittaci 6BC^T, Corynebacterium diphtheriae NCTC 10356, Enterococcus faecalis NCTC 775^T, Escherichia coli NCTC 9001^T, Haemophilus influenzae NCTC 8143^T, Klebsiella pneumoniae subsp. pneumoniae NCTC 9633^T, Legionella pneumophila NCTC 11192^T, Moraxella catarrhalis NCTC 11020^T, Pseudomonas aeruginosa NCTC 10332^T, Rothia dentocariosa NCTC 10917^T, Staphylococcus aureus NCTC 8532^T, group G streptococcus NCTC 9603, Streptococcus pneumoniae NCTC 7465^T, Streptococcus pyogenes NCTC 12067.

Extraction of DNA from bacterial cultures. Cultures of mollicute species, A. odontolyticus and R. dentocariosa (5 ml) were concentrated by centrifugation at 8000 g for 15 min, resuspended in 180 µl nuclease-free water and extracted with 20 µl Instagene matrix (Bio-Rad) according to the manufacturer's instructions. All other strains used in the specificity panel were extracted by using the Roche MagNAPure robot using the DNA isolation kit III (Roche Diagnostics) according to the manufacturer's instructions. DNA concentration was adjusted to approximately 20 pg µl¹ for use in the LightCycler assay.

Primers and probes. The sequences of all primers and probes are listed in Table 2. Primers specific for the RepMP2 repetitive element (GenBank accession no. X13087) of the P1 cytadhesin gene of M. pneumoniae were designed using the primer selection program OLIGO (Medprobe) to amplify a PCR product of 141 bp. Within this sequence, two probes were designed to a 43 bp region with a 2 bp gap located proximal to the 3' end of the sense sequence.

Table 2. Primers and probes

Construction of a PCR IPC. A 278 bp sequence from λ phage DNA into which M. pneumoniae primer sites were incorporated was amplified using primer sequences for M. pneumoniae with λ phage amplifying sequences at their 3' ends (Table 2; full sequence available as supplementary material in JMM Online). λ DNA (1 µg; Sigma) was digested with EcoRI for 3 h. The restricted DNA (10 ng) was used as a template in a 50 µl PCR containing 150 mM NaCl, 1·5 mM MgCl₂, 10 mM Tris/HCl (pH 8·0), 1 U Taq DNA polymerase (Gibco-BRL), 10 pM each of primers M-IPC-1 and M-IPC-2, and 200 µM deoxyribonucleotides. PCR was performed on a Peltier Thermal Cycler (225) DNA Engine Tetrad and reaction conditions were 30 cycles of 95 °C for 30 s, 40 °C for 1 min and 72 °C for 1 min, with a final extension step at 72 °C for 10 min. The product size was confirmed by electrophoresis and the concentration of the product was estimated using a GeneQuant spectrophotometer (Pharmacia). The product was then cloned into the pCR2.1TOPO plasmid using the TOPO TA cloning kit (Invitrogen), transformed into Escherichia coli INVαF' and positive transformants were selected according to the manufacturer's instructions. Positive transformants were cultured overnight in LuriaBertani broth (5 ml) containing 50 µg ampicillin ml¹ and plasmid DNA was extracted using a Plasmid Mini kit (Qiagen), adjusted to a concentration of 1 ng µl¹ and then heated at 95 °C for 5 min. Plasmid preparations with inserts of the expected size (278 bp) were confirmed by PCR selection as follows. PCR reagents were as above with the inclusion of primers Mpn-3 and Mpn-4 (Table 2) and reaction conditions were 30 cycles of 95 °C for 30 s, 58 °C for 30 s and 72 °C for 1 min, with a final extension step at 72 °C for 10 min. The resulting IPC plasmid was linearized by digestion with XbaI, adjusted to 10¹⁰ copies µl¹ in TE buffer [10 mM Tris/HCl (pH 8·0), 1 mM EDTA] containing herring sperm DNA (10 ng µl¹; Gibco-BRL) and stored as 35 µl aliquots at 80 °C. For use in the assay, stocks of IPC were diluted to 10² copies µl¹ in nuclease-free water.

Construction of a M. pneumoniae PCR positive control. To alleviate problems associated with growing large volumes of M. pneumoniae to generate stocks of standard DNA, a cloned template DNA sequence was prepared for use as a positive control. A 410 bp PCR product of the P1 cytadhesin gene, external to the amplicon in the LightCycler assay, was amplified using primers Mpn-1 and Mpn-2 (Table 2) from M. pneumoniae strain FH (NCTC 10119^T). This product was cloned and prepared as above and a stock solution containing 10¹⁰ copies µl¹ was prepared in TE buffer containing herring sperm DNA (10 ng µl¹). The stock solution was serially diluted in nuclease-free water to give a concentration range that covered the expected dynamic range of the clinical specimens. Each LightCycler run included positive-control samples with 10⁴, 10³, 10² and 10¹ copies per PCR and a negative control (nuclease-free water). To assess reproducibility and stability, the crossing-point (Cp) values of these standards were monitored over several months.

LightCycler assay. Assays were carried out using a LightCycler (Roche) and the MgCl₂ concentration was optimized according to the manufacturer's instructions. Reaction volumes (20 µl in total) in glass capillary tubes (Roche) were used containing 2 µl of LightCycler FastStart DNA Master Hybridization Probes (which includes reaction buffer, nucleotides and Taq polymerase; Roche), 1·0 µM Mpn-3 and Mpn-4 primers (Table 2), 5 mM MgCl₂, 0·25 µM each of the probes Mpn-DN, Mpn-AC, IPC-DN and IPC-AC (Table 2), 10² copies of the IPC, 1 U uracil-DNA glycosylase (Roche) and 5 µl sample or positive/negative control. LightCycler FastStart DNA Master Hybridization Probes (Roche) reagent contains dUTP (not dTTP), and uracil-DNA glycosylase (Roche) was added to each reaction to eliminate carry-over contamination (Longo et al., 1990). Capillaries were sealed and placed in the LightCycler and the following cycling conditions applied: initial denaturation at 96 °C for 10 min (transition rate 20 °C s¹; acquisition mode none); 45 quantification cycles of 95 °C for 10 s (transition rate 20 °C s¹; acquisition mode none), 62 °C for 10 s (transition rate 20 °C s¹; acquisition mode single) and 74 °C for 15 s (transition rate 3 °C s¹; acquisition mode none); melting curve cycle of 95 °C for 0 s (transition rate 20 °C s¹; acquisition mode none), 45 °C for 2 s (transition rate 20 °C s¹; acquisition mode none) and 85 °C for 0 s (transition rate 0·1 °C s¹; acquisition mode cont); cooling at 40 °C for 30 s (transition rate 0 °C s¹; acquisition mode none).

Data were analysed using Roche LightCycler software version 3.5 using arithmetic baseline adjustment and second-derivative maximum analysis (Fig. 1). Copy number was estimated from the Cp threshold relative to positive standards; samples that did not give a positive result and in which the IPC did not amplify were recorded as inhibitory and were repeated undiluted and diluted 1/10 in nuclease-free water (Promega).

(37K): Fig. 1. Replicate standards of M. pneumoniae cloned DNA from 101 to 104 copies per sample. Channel F2, Red 640 fluorescence.

The use of real-time PCR was employed to detect M. pneumoniae in clinical samples using a Roche LightCycler. In total, 30/175 (17·1 %) samples were positive by PCR, culture or serology. This figure could not be linked to prevalence or carriage in the population, as all patients had clinical signs of pneumonia and control samples from healthy patients were not available for study.

Sensitivity
The assay detected 10 copies or more of the target sequence in the 5 µl samples used in the assay [corresponding to 2·0x10³ organisms (ml extracted concentrated DNA)¹ and 1x10³ organisms (ml actual clinical sample)¹]. The estimated copy number in actual clinical samples from proven M. pneumoniae patients ranged from 5·0x10³ to 6·5x10⁷ organisms ml¹. Samples with a Cp equal to or less than the lowest standard in the assay were deemed negative. Due to the absence of control specimens from healthy individuals, this study could not ascertain the level of M. pneumoniae in subclinical colonization and the use of the method in population screening is therefore limited. The LightCycler is capable of measuring the genome load in clinical specimens accurately, yet in practice, it is of little use in applying quantitative analyses to bacterial numbers in the respiratory tract, as specimens vary considerably in volume, consistency and composition and contain substances inhibitory to PCR. The measurement of an estimated bacterial load in this case can only be considered as semi-quantitative, and without a casecontrol study examining M. pneumoniae load in clinical specimens from both healthy and pneumonic specimens over the infective period little clinical relevance can be drawn other than the presence or absence of detectable M. pneumoniae DNA. None the less, of the 175 pneumonic patient samples tested, 20 (11·4 %) were positive in the real-time assay (Table 1). In 25 of 175 patients, a diagnosis of M. pneumoniae infection had been previously established by serology, of which 15/25 (60·0 %) were positive by this real-time method (groups A and B): four had also previously been confirmed by culture (group B). Four samples were serologically positive, yet negative by culture and PCR (group B). This may be due to cross-reactive antibody tests, increased sensitivity of antibody detection methods in comparison with PCR and culture, clearance of the micro-organism from the body or early treatment with antibiotics. The latter is unlikely as patients receiving antibiotic therapy were excluded from the study. It is possible that a period occurs during the infection process that is optimal for the detection of M. pneumoniae by PCR. Specimens were taken on admission to hospital, which corresponded to between 0 and 17 days after the onset of clinical signs. Those found to be positive by PCR in group A were taken on average 7 days after onset (217 days, SD±4·48) of symptoms, in comparison with PCR-negative samples taken 3 days post-onset (012 days, SD±2·36).

The sensitivity was estimated at 60·0 % in comparison with serological positives obtained from paired acute and convalescent serum samples. Paired serum samples are not often received by clinical laboratories and comparison of the PCR assay with serology on single serum samples may have reflected use in practice more accurately. This level of sensitivity may be considered low; however, reports from other publications and the basis for calculating sensitivity have varied considerably. Michelow et al. (2004) reported a respiratory-sample PCR with a sensitivity of 57 % and specificity of 98 % when compared with M. pneumoniae ELISA. Hardegger et al. (2000) reported a P1 gene TaqMan real-time assay to detect M. pneumoniae that could detect a 1 : 100 dilution of a DNA sample. However, no indication of DNA concentration or target copy number in positive-control samples was given and it is not possible to make a direct comparison of sensitivity between this and other methods. Miyashita et al. (2004) noted that their multiplex assay could detect 100 copies ml¹, and Dorigo-Zetsma et al. (1999) reported an estimated sensitivity of 78 % in a block-based PCR. Using a similar assay to that described in this study, Ursi et al. (2003) noted a sensitivity of 5x10³5x10⁴ organisms ml¹, indicating that the assay described here is 550 times more sensitive. Our assay could be improved further by the addition of a second M. pneumoniae target, allowing confirmation of positive results simultaneously with detection. Sensitivity can be dependent on several factors, such as the method used for comparison and selection criteria for evidence of infection (PCR, culture or sero-conversion as the gold standard), the presence of competitive DNA and inhibitors, sample type (swab, aspirate, sputum), method, age, transport and storage, operator variance, sample dilution and the method of extraction. Ievens et al. (1996) compared the PCR sensitivities of a simple boiling/freezing extraction and a guanidium thiocyanate/phenol extraction in two separate laboratories and found that results varied considerably between techniques and laboratories. This retrospective study utilized archived samples that had been stored at 80 °C for several years and had been heat-treated; therefore deterioration in the quality of the DNA in the samples could have occurred. Furthermore, immunofluorescence for M. pneumoniae antibody detection is highly subjective and, due to the lack of standardization across the techniques, caution should be employed when comparing techniques for M. pneumoniae detection (Loens et al., 2003). These factors make inter-assay comparisons complex. Furthermore, no formal external quality assessment schemes for M. pneumoniae serology or molecular detection exist. Such a scheme could provide highly beneficial data regarding the detection of infection with this pathogen.

Specificity
The assay was found to be highly specific in that DNA extracted from reference cultures (15 human mollicute species and 19 common respiratory bacteria) did not give a positive signal in the assay, even when tested at concentrations significantly higher than the limit of detection. A search of GenBank using the BLAST algorithm (Altschul et al., 1990) revealed no significant homology of the target sequence with any other known genes (E value=5·8). Of the 150 patients where the primary diagnosis was not infection with M. pneumoniae infection, five (3·3 %) were PCR positive for M. pneumoniae. One had been serologically confirmed as positive for Chlamydophila spp. by complement fixation tests, one as H. influenzae positive by isolation from sputum and three as S. pneumoniae positive by isolation from sputum or antigen detection. This gives an overall percentage specificity of 96·7 % (5/150). These results could represent either false-positive reactions with other bacterial DNA, co-infection with M. pneumoniae prior to the formation of a detectable serum-antibody response or detection of residual M. pneumoniae DNA from a past infection. The former is unlikely, as none of the DNA from these three pathogens resulted in amplification in the assay. In addition, all five samples had low concentrations of detectable M. pneumoniae DNA (<260 copies per reaction). Low levels of colonization may be detected by PCR prior to a detectable serological response in patients with or without symptoms. Indeed, several specimens from patients that were negative serologically exhibited Cp values close to the limits of detection (<10 copies per sample). This may represent a technical artefact or could be detection of low levels of infection. Asymptomatic carriage has been reported to occur within the community and up to 13 % of healthy adults may act as carriers of M. pneumoniae, particularly during epidemic periods (Gnarpe et al., 1992; Foy, 1993). Interestingly, Principi et al. (2001) detected M. pneumoniae by PCR in 3·8 % (16/419) of children without serological evidence of acute infection. Patients positive for M. pneumoniae had a mean age of 37·8 (SD±18·3, 95 % CI 29·246·3, range 1878), significantly younger than patients in which M. pneumoniae was not detected (mean age of 50·6, SD±19·1, 95 % CI 44·656·7, range 1880) (Fisher's exact test, P=0·014).

Control of inhibition
It is essential that respiratory samples inhibitory to real-time PCR are identified to ensure that they are not falsely reported as negative. Single-tube methods that simultaneously detect IPC and target DNA amplicons of differing lengths with the same primers have been used in isothermal amplification and real-time methods for M. pneumoniae (Loens et al., 2002; Ursi et al., 2003). We developed a similar IPC consisting of a cloned PCR fragment of the λ phage genome flanked by primer sites for the M. pneumoniae PCR. When used with target and λ-specific probes that fluoresce at different wavelengths, the same set of primers amplifies both M. pneumoniae and IPC, detecting both products simultaneously. Furthermore, the size of the IPC amplicon (278 bp) is larger than that of the target (141 bp) amplicon and the latter reaction is driven preferentially at the expense of the former (Fig. 2). Only when target DNA is low in concentration or absent will any IPC sequence be amplified, and inhibition is detected in reactions with an absence of both target DNA and IPC amplicons. The IPC was stable when Cp values were monitored over several months with a SD of approximately 1·0 for 10 sets of standards (10¹ copies per reaction: Cp 32·135·4, mean 33·5, SD±1·03; 10² copies per reaction: Cp 28·831·3, mean 30·0, SD±0·87; 10³ copies per reaction: Cp 24·527·2, mean 25·9, SD±0·99; 10⁴ copies per reaction: Cp 21·524·4, mean 22·4, SD±0·98; 10⁵ copies per reaction: Cp 16·119·8, mean 17·9, SD±0·97).

(27K): Fig. 2. Internal processing control (IPC) at a concentration of 102 copies per PCR in the presence of no target DNA (A), 101 copies target DNA (B), or 102 copies target DNA (C). Channel F3, Red 705 fluorescence.

In the current study, 40/167 (24·0 %) of undiluted samples were inhibitory to the assay (note that eight samples were only tested at a 1/10 dilution due to limited sample volume). By retesting a 10-fold dilution, this was reduced to 7/175 (4·0 %). These results support those of Dorigo-Zetsma et al. (1999), in which 20 % of throat-swab extracts inhibited amplification for M. pneumoniae and this inhibition was removed by sample dilution. Abele-Horn et al. (1998) reported inhibition in 10 % of samples, which was reduced to 2 % by the use of a 1 day culture enhancement. Raggam et al. (2005) and Stralin et al. (2005) found no inhibition with respiratory samples. Ursi et al. (2003) reported that 2/115 (1·7 %) respiratory samples were inhibitory. Both sample viscosity (Loens et al., 2002) and the pre-treatment of samples with mucolytic agents (dithiothreitol) used to reduce viscosity (Deneer & Knight, 1994) have been reported to cause false-negative results. In this study, differing extraction methods and the resulting effect on assay performance were not examined. A single method of extraction was employed that is well established in the laboratory for extracting DNA from respiratory specimens (QiAamp DNA Mini kit; Qiagen). Total DNA content may also affect PCR whereby increasing amounts of non-target DNA inhibit amplification of the target DNA. In respiratory specimens, human DNA, as well as DNA from other micro-organisms, may cause inhibition (Ludwig & Schleifer, 2000).

This study is similar to that by Ursi et al. (2003) in that a real-time assay for the P1 gene of M. pneumoniae with the same primer IPC was employed to compare detection of M. pneumoniae in respiratory specimens. Both studies used the same platform, similar methods of DNA extraction and a similar methodology and detected comparable concentrations of M. pneumoniae in respiratory samples (Ursi et al., 2003: 5x10³5x10¹⁰ organisms ml¹; this study: 5x10³6x10⁷organisms ml¹). However, our study examined a larger number of patients (175 compared with 82), improved the limit of detection by at least fivefold (1x10³ compared with 5x10³5x10⁴ organisms ml¹) and was tested against a comprehensive panel of human Mycoplasma species and respiratory pathogens. Another important difference to note is that this study compared real-time PCR with serologically positive patients (and some that were culture-positive), whereas Ursi et al. (2003) compared the real-time assay with isothermal PCR using exactly the same target. Ursi et al. (2003) recommended testing lower respiratory tract specimens rather than upper respiratory tract specimens for M. pneumoniae due to the higher concentrations of detectable M. pneumoniae DNA. Here, lower respiratory tract samples from patients with clinically defined pneumonia that had been screened previously for a large panel of respiratory pathogens were tested. This may account for the difference noted in the number of inhibitory samples: Ursi et al. (2003) tested throat swabs, throat washings and sputa, with 2/115 (1·7 %) showing inhibition that was eliminated by sample dilution. This study comprised lower respiratory tract samples only, of which 24 % were inhibitory; this was reduced to 4 % by dilution. We therefore recommend testing of all lower respiratory tract DNA samples undiluted and at a 1/10 dilution for M. pneumoniae detection.

Conclusions
This assay for detecting M. pneumoniae in the respiratory tract is reproducible, with a sensitivity of 60·0 % in comparison with serology and a specificity of 96·7 %. The method is rapid and up to 27 extracted samples can be assayed in less than an hour, ideal for same-day screening of patients with respiratory symptoms. The use of a co-amplified IPC in the same tube and employing the same primers as the target sample increases the throughput of specimens for diagnosis and reduces cost. The use of undiluted and 1/10 dilutions of specimens decreased the observed inhibition from 24 to 4 %. Validation of extraction methods to resolve sample inhibition would be beneficial, but would require large volumes of clinical specimens. Dual infections were found in five (3·3 %) of the samples tested, and overall 20/175 (11·4 %) samples were PCR positive for M. pneumoniae. This study included validation of the assay on respiratory specimens that are highly defined for other respiratory infections, but was limited by the absence of control specimens from healthy individuals and the age of the specimens under test. It is difficult to gain access to respiratory specimens that have been tested for a full range of respiratory specimens with relevant clinical and matching serological data. Such a collection could be invaluable for assay comparison to ensure the best patient service as new techniques for M. pneumoniae and other fastidious respiratory pathogens are described.

The authors would like to thank Dr Margaret Sillis for advice and the provision of some specimens and Andrew Grant and Tom Nichols for statistical assistance.