Specific Multilocus Variable-Number Tandem-Repeat Analysis Genotypes of Mycoplasma pneumoniae Are Associated with Diseases Severity and Macrolide Susceptibility

Clinical relevance of multilocus variable-number tandem-repeat (VNTR) analysis (MLVA) in patients with community-acquired pneumonia (CAP) by Mycoplasma pneumoniae (M. pneumoniae) is unknown. A multi-center, prospective study was conducted from November 2010 to April 2012. Nine hundred and fifty-four CAP patients were consecutively enrolled. M. pneumoniae clinical isolates were obtained from throat swabs. MLVA typing was applied to all isolates. Comparison of pneumonia severity index (PSI) and clinical features among patients infected with different MLVA types of M. pneumoniae were conducted. One hundred and thirty-six patients were positive with M. pneumoniae culture. The clinical isolates were clustered into 18 MLVA types. One hundred and fourteen (88.3%) isolates were resistant to macrolide, covering major MLVA types. The macrolide non-resistant rate of M. pneumoniae isolates with Mpn13-14-15-16 profile of 3-5-6-2 was significantly higher than that of other types (p≤0.001). Patients infected with types U (5-4-5-7-2) and J (3-4-5-7-2) had significantly higher PSI scores (p<0.001) and longer total duration of cough (p = 0.011). Therefore it seems that there is a correlation between certain MLVA types and clinical severity of disease and the presence of macrolide resistance.


Introduction
Mycoplasma pneumoniae (M. pneumoniae) is a common and important pathogen of community-acquired pneumonia (CAP), accounting for 10%-30% of cases [1][2][3][4]. Genotyping of clinical isolates is very important for clinical and epidemiological study of M. pneumoniae infections. However, isolates are not well differentiated using the most common typing methods based on the analysis of the gene encoding the P1 protein [5,6]. In 2009, Dégrange et al have established the multilocus variable-number tandem-repeat (VNTR) analysis (MLVA) method based on 5 VNTRs as Mpn1-13-14-15-16, and clustered 265 M. pneumoniae clinical isolates into 26 MLVA types [7]. In 2011, the same group has successfully confirmed a single-clone spread of type J (3-4-5-7-2) M. pneumoniae among children in a French primary school and their household contacts. Although the VNTR locus Mpn1 is reported to be unstable in both clinical isolates and in laboratory passages, MLVA genotyping of M. pneumoniae still exceeds the discriminatory power of previous typing methods based on P1 gene [8,9].
Macrolide-resistant M. pneumoniae have been spreading for 10 years worldwide, with prevalences ranging from below 10% in Europe, approximately 30% in Israel [10] and ranging from 69% to 92% in Asia [2,11,12]. One report found there was a possible association between the MLVA type Z (7-4-5-7-2) and macrolide resistance [13], whereas other reports found no association between macrolide resistance and MLVA types [9,12,14]. The evidence of clinical association between MLVA types and macrolide susceptibility is limited. More important, since almost all of the published reports focused on MLVA genotyping of M. pneumoniae isolates and method improvement [8,9,12,[15][16][17], the clinical relevance of MLVA types, such as the association with disease severity and clinical features, is still lacking.
In this multi-center and prospective study, we used MLVA typing method to analyze the M. pneumoniae clinical isolates, and tried to investigate the possible associations between MLVA genotypes and clinical features, including macrolide susceptibility and diseases severity.

Study setting and design
A prospective study was conducted in 12 general hospitals in Beijing, as described in former report [18]. Between November 2010 and April 2012, patients (aged 14 years or above) who came to the hospitals and met the inclusion criteria of CAP [19] were enrolled. Patients with HIV infection; neutropenia or chemotherapy; pregnant; known or suspected active tuberculosis, no informed consents or specimens were excluded. Adolescents were defined as aged from 14 to 17 years old, and adults were $18 years old. The study was approved by Institutional Review Board of Beijing Chao-Yang Hospital (project approval number 10-KE-49). Written informed consents were provided by all adults and the parents of patients aged less than 18 years.
The following data were recorded on enrollment: age, gender, smoking status, comorbidities and antimicrobial treatment prior to enrollment, clinical symptoms, etc. All surviving patients were followed-up by telephone within 7 to 28 days after discharge, symptoms and signs were recorded daily. Pneumonia severity index (PSI) scores were calculated [20].

M. pneumoniae Culture and Antimicrobial Susceptibility Test
Culture of M. pneumoniae was performed as described previously by Waites [21]. One hundred and thirty-six M. pneumoniae isolates were obtained from throat swabs (each specimen collected from one patient). All the isolates were identified by colony morphology and PCR assay.
Minimum inhibitory concentrations (MICs) were determined by broth microdilution methods with SP4 broth (Remel) [21]. Susceptibility tests were performed in triplicate. M. pneumoniae reference strain MPFH (ATCC 15531) were also included. According to CLSI guideline (M43-A, 31 (19), 2011) for antimicrobial susceptibility for human mycoplasmas including M. pneumoniae, susceptible strains have a MIC of erythromycin and azithromycin #0.5 mg/ml and resistant strains have a MIC $1 mg/ml. The total length of the 23S ribosomal RNA (rRNA) gene of each M. pneumoniae isolate was amplified and sequenced by the method described previously [22]. The mutations of 23S rRNA have been determined by alignment with sequence of MPFH (ATCC 15531).

MLVA and P1 Gene Typing
MLVA typing on M. pneumoniae isolates was performed using PCRs with fluorescently labeled primers targeting five VNTR loci, followed by capillary electrophoresis [7]. One reference strain of MPFH (ATCC 15531), was included as the positive control with MLVA type T (5-3-6-6-2). There were two peaks of molecular weight signal in Mpn1 of four samples, Mpn13 of one sample and Mpn15 of two samples. The numbers of VNTR were finally determined by sequencing of PCR products. P1 gene typing on M. pneumoniae isolates was performed using a multiplex PCR as described by Kenri [23].

Statistical Analysis
Continuous variables were presented as means (6 SD), or medians (with IQR) where appropriate. For categorical variables, the percentages were calculated. The PSI scores of patients infected with different MLVA types were compared using One-Way ANOVA and Tukey's pairwise comparison test. Comparisons of macrolide resistant rates with different VNTRs profiles were conducted using a Chi-square test, Bonferroni method was used for multiple comparisons with adjusted significance level as a = 0.05/3 = 0.0167. Comparisons of clinical features were conducted between MLVA type groups, using an Independent-Sample t-test or the Mann-Whitney test (SPSS for Windows 13.0).

Macrolide Susceptibility and MLVA Typing of Clinical Isolates
In vitro activities of nine antimicrobials are listed in Table S1. Among 136 isolates, 114 (83.8%) were resistant to macrolide with MIC $1 mg/mL to erythromycin. These erythromycin-resistant isolates also showed cross-resistance to clarithromycin and azithromycin. Results of 23S ribosomal RNA (rRNA) gene sequencing indicated that all macrolide-resistant isolates harbored an A2063G mutation, and no such mutation had been found in non-resistant isolates.

Discussion
M. pneumoniae is an important pathogen in adolescents and adults CAP patients, with a positive rate as 14.3% (136/954) by  culture in our study. Compared with those without M. pneumoniae infection, patients with M. pneumoniae pneumonia have several specific clinical characteristics, including younger age, less comorbidities or smoker, more proportion of cough, and lower PSI scores, similar with the findings from CAPNETZ [24]. Based on clinical data of the defined patients group and MLVA genotyping method (with a good discriminatory ability of M. pneumoniae), we investigate the possible clinical relevance of different MLVA genotypes.
There are two limitations in this study. First, since we have used M. pneumoniae cultures for both MLVA typing and determination of antibiotic resistance, and since cultures have much lower sensitivity in identifying M. pneumoniae in clinical samples we might have a bias towards patients with higher bacterial burden. In most recent papers, the clinical MLVA typing was performed directly on throat swabs without culturing [16]. Another limitation is that no children have been enrolled in our study. We could not answer whether there is any specific MLVA type responsible for the higher prevalence of M. pneumoniae in children than adolescents and adults [1]. But the distribution pattern of MLVA types we found here was similar with the finding from a study on pediatrics [9].