Mycoplasma pneumoniae P1 Genotype Indicates Severity of Lower Respiratory Tract Infections in Children

ABSTRACT

Mycoplasma pneumoniae strains can be classified into two major genetic groups, P1 type 1 (P1-1) and P1 type 2 (P1-2). It remains unknown if clinical manifestations of lower respiratory tract infections (LRTI) in children differ between the two genotypes. We aimed to determine if the M. pneumoniae P1 genotype is associated with severity of LRTI in children. Medical charts of 420 children (≤15 years old) with signs of acute LRTI who were PCR positive for M. pneumoniae from pharyngeal swabs in a recent M. pneumoniae epidemic were analyzed. We used a culture and pyrosequencing approach for genotyping PCR-positive samples. We compared epidemiological and clinical data of children with either P1-1 or P1-2 LRTI. P1-2-infected children presented with a significantly higher median baseline C-reactive protein level and were admitted to the hospital more often. The P1 genotype had a significant predictive value in a multiple linear regression model predicting C-reactive protein levels in our study sample. Moreover, the P1 genotype significantly affected the likelihood of hospital admission in a logistic regression model. Our modeling results were also confirmed on an additional independent sample of children with M. pneumoniae LRTI. Results from our large patient group indicate that the two M. pneumoniae P1 genotypes may have different pathogenic potential and that LRTI with P1-2 strains may have a more severe disease course than those with P1-1 strains in children. P1 genotyping is not routinely performed but could be used as a predictor of M. pneumoniae LRTI severity, enabling patient-tailored treatments.

INTRODUCTION

Mycoplasma pneumoniae is a common cause of community-acquired lower respiratory tract infections (LRTI) in children. It is most frequently detected among children from 5 to 15 years of age and is responsible for up to 40% of community-acquired pneumonias in this age group (1–3). Although M. pneumoniae infections are often mild and self-limiting, patients can develop severe and fulminant disease (4). The factors determining clinical severity of M. pneumoniae LRTI remain poorly understood (1).

M. pneumoniae strains can be classified into two major genetic groups called P1 type 1 (P1-1) and P1 type 2 (P1-2), according to nucleotide differences in two repetitive elements (RepMP2/3 and RepMP4) in the MPN141 gene, which code for the P1 adhesion protein (5). Protein P1 is the primary adhesin in the polarized attachment organelle, where it functions as a key ligand during adhesion, which is required for virulence of M. pneumoniae (6).

Previous studies have shown that the two genetic groups differ in vitro in their pathogenic potential (7, 8). However, it remains largely unknown if clinical manifestations of M. pneumoniae LRTI in children differ between the two genetic groups.

A large number of M. pneumoniae LRTI in children during a recent epidemic (9) provided us with a unique opportunity to examine the association between the M. pneumoniae P1 genotype and clinical characteristics of acute M. pneumoniae LRTI in children. We aimed to assess if the M. pneumoniae P1 genotype is associated with severity of LRTI in children.

MATERIALS AND METHODS

Study subjects.

Children referred to University Children's Hospital Ljubljana and Department of Infectious Diseases Ljubljana with clinical signs of acute mycoplasmal LRTI from January 2014 to December 2014 were tested for M. pneumoniae. These two hospitals are the largest children's referral hospitals in Slovenia (population of 2,062,731 during the designated period of time).

All patients younger than 15 years with signs of acute LRTI who were PCR positive for M. pneumoniae from pharyngeal swabs and had the P1 genotype successfully defined were identified from a laboratory database and recruited for the study. The diagnosis of LRTI was made based on clinical and/or radiographic appearance. We excluded cases where additional testing provided evidence that other infectious agents were the more likely cause of the disease. Immunocompromised children and children with chronic pulmonary disease were excluded from the study because the underlying conditions could affect the severity of the disease.

We verified our findings on an independent sample of children with M. pneumoniae LRTI in a 4-year period from January 2009 to December 2012. All previous inclusion and exclusion study criteria were applied to this independent sample for participation in the study.

The study was approved by National Medical Ethics Committee.

Study design.

We performed an observational retrospective study. The main objective was to assess if the M. pneumoniae P1 genotype is associated with severity of LRTI in children.

Clinical and demographic variables collected and recorded for each patient included age, gender, P1 type, macrolide susceptibility, interval between disease onset and antibiotic therapy initiation, laboratory markers of inflammation, respiratory virus codetections, hospital admission, duration of hospital stay, presence of extrapulmonary manifestations, and data related to treatment and complications.

We compared epidemiological and clinical data of patients with either P1-1 or P1-2 LRTI. The clinical impact was assessed by comparing several markers of disease severity, including inflammatory markers, hospital admission, duration of hospital admission, hypoxia, and admission to the critical care unit. Hypoxia was defined as oxygen saturation rate on admission of <92% using pulse oximetry on room air or use of supplemental oxygen at presentation. The patients were further divided into two age groups (≤5 years and 5 to 15 years), and their epidemiologic and clinical characteristics were compared.

Methods.

Pharyngeal swabs of children with signs of acute M. pneumoniae LRTI were subjected to DNA isolation using QIAamp DNA minikit (Qiagen, Germany) or MagNA Pure compact (Roche, Germany) and later tested by M. pneumoniae real-time PCR (Argene, France). The remainder of each PCR-positive sample was cultivated in order to obtain pure M. pneumoniae isolates. Culture was performed by using Mycoplasma selective broth and agar plates (Oxoid, United Kingdom) enriched with Mycoplasma supplement G or P (Oxoid, United Kingdom) according to standard methods (10). Isolates of M. pneumoniae were further classified into P1-1 and P1-2 types using pyrosequencing that targets the M. pneumoniae MPN141 and MPN528a genes, and macrolide resistance was recognized by pyrosequencing two parts of domain V in the 23S rRNA gene (9).

Multiplex PCR was performed on the nasopharyngeal aspirate specimens to assess viral codetection, including respiratory syncytial virus, influenza virus, parainfluenza virus, adenovirus, human bocavirus, metapneumovirus, rhinovirus, enterovirus, and coronavirus at the time of LRTI diagnosis.

Analysis.

Categorical variables were described with counts and percentages. Continuous variables were presented as mean (standard deviation [SD]) or median (interquartile range [IQR]) where appropriate.

Continuous variables were compared using independent samples t test or Mann-Whitney U test where appropriate, whereas categorical variables were compared by using the Pearson chi-square test. The differences were considered to be statistically significant when the P value was less than 0.05.

C-reactive protein (CRP) was, as an early sensitive marker of inflammation, used as a marker of disease severity similar to previous studies (11, 12). We used a multiple linear regression model to assess if the P1 genotype is associated with CRP. We also performed a multivariable logistic regression analysis to assess if the P1 genotype is associated with hospital admission.

RESULTS

In the epidemic period, 1,621 children were referred to our hospitals with signs of acute M. pneumoniae infection. Out of those, 37% (593/1,621) were PCR positive for M. pneumoniae from pharyngeal swabs. M. pneumoniae isolates were detected by culture from 70% (417/593) of the PCR-positive specimens.

After applying the inclusion and exclusion study criteria, we evaluated data from 420 patients (mean age, 6.7 years; SD, 3.2 years; 52% boys) with acute M. pneumoniae LRTI. Thirty-four percent of patients were younger than 5 years at presentation. The two main P1 genotypes, P1-1 and P1-2, accounted for 74% (310/420) and 26% (110/420) of total cultured isolates, respectively. Four hundred seventeen isolates were macrolide susceptible, and 3 were found to contain an A2063G mutation in the M. pneumoniae 23S rRNA domain V that infers macrolide resistance.

In the control independent sample, a total of 117 patients met the study criteria (mean age, 8.1 years; SD, 3.4 years; 52% boys).

The characteristics of patients with M. pneumoniae LRTI infected with either P1-1 or P1-2 strains are summarized in Table 1. Data of 310 patients infected with P1-1 were compared with 110 patients infected with P1-2. Demographic characteristics were similar between groups. Moreover, there was no difference in the median interval between disease onset and the initiation of antibiotic therapy in either group. The majority of patients in both groups received antibiotic treatment, 99% (307/310) versus 97% (107/110), with no difference in the choice of antibiotics. In accordance with Slovenian recommendations for prescribing antibiotics (13), midecamycin was the first-choice antibiotic in both groups, 65% (200/307) versus 60% (64/107), followed by azithromycin, 33% (101/307) versus 36% (39/107). Children did not improve on first-choice antibiotic therapy in 3% (9/307) of patients with P1-1 and 5% (5/107) of patients with P1-2 infection (P = 0.22). After alternative antibiotic therapy, all patients fully recovered. Despite no difference in X-ray observations, rate of viral codetection, and incidence of extrapulmonary manifestations, patients infected with P1-2 had a higher median baseline CRP level.

TABLE 1.

Characteristics of patients infected with either M. pneumoniae P1 type 1 or M. pneumoniae P1 type 2 strains in 2014^a

Characteristic	Patients with P1 type 1	Patients with P1 type 2	P value
No. of subjects	310	110
% boys/% girls	50/50	57/43	0.190
Age (IQR) (yrs)	6.5 (4.4–9.4)	6.1 (4.0–8.6)	0.244
≤5 yrs (% [no. of subjects/total no. of subjects])	33 (102/310)	37 (41/110)	0.406
5–15 yrs (% [no. of subjects/total no. of subjects])	67 (208/310)	63 (69/110)
Interval between disease onset and antibiotic therapy initiation (IQR) (days)	8 (6–11)	8 (5–11)	0.061
Macrolide therapy (% [no. of subjects/total no. of subjects])	99 (307/310)	97 (107/110)	0.292
CRP (IQR) (mg/liter)	17 (6–35)	26 (8–45)	0.008
WBC (IQR) (×109/liter)	9.5 (6.9–12.1)	9.6 (7.0–13.3)	0.320
X-ray effusion (% [no. of subjects/total no. of subjects])	29 (70/244)	28 (26/94)	0.851
Viral codetection (% [no. of subjects/total no. of subjects])	29 (46/157)	27 (17/64)	0.663
Extrapulmonary manifestations (% [no. of subjects/total no. of subjects])	13 (40/310)	11 (12/110)	0.651

^aData on X-ray observations and viral codetection were not collected in all subjects. Significant differences (P < 0.05) are higlighted in bold. Abbreviations: CRP, C-reactive protein; IQR, interquartile range; WBC, white blood cell count.

Out of 420 patients, 36% (150/420) required hospital treatment. Patients infected with P1-2 were admitted to the hospital more often (Table 2). However, there was no difference in the requirement for oxygen treatment or in the average duration of hospital stay between hospitalized patients. No intensive care treatment was required in either group.

TABLE 2.

Characteristics of hospitalized patients infected with either M. pneumoniae P1 type 1 or M. pneumoniae P1 type 2 strains in 2014^a

Characteristic	Patients with P1 type 1	Patients with P1 type 2	P value
No. of subjects	102	48
Age (IQR) (yrs)	4.9 (3.1–7.3)	5.2 (3.5–6.8)	0.897
≤5 yrs (% [no. of subjects/total no. of subjects])	46 (47/102)	50 (24/48)	0.654
5–15 yrs (% [no. of subjects/total no. of subjects])	54 (55/102)	50 (24/48)
Admission rate (% [no. of subjects/total no. of subjects])	33 (102/310)	44 (48/110)	0.044
Hospital stay (IQR) (days)	3 (2–4)	2 (1–4)	0.204
Oxygen therapy (% [no. of subjects/total no. of subjects])	46 (47/102)	44 (21/48)	0.789

^aSignificant differences (P < 0.05) are higlighted in bold. Abbreviation: IQR, interquartile range.

P1 type had a significant predictive value in a multiple linear regression model predicting CRP levels in our study sample (Table 3). Moreover, after adjusting for age and interval between disease onset and initiation of antibiotic treatment, the P1 type significantly affected the likelihood of hospital admission in the 5- to 15-year-olds in a logistic regression model (Table 4). Patients infected with P1-2 strains were almost twice more likely to be hospitalized than patients infected with P1-1 strains. No similar impact was observed in the age group of ≤5-year-olds.

TABLE 3.

Multiple linear regression model predicting C-reactive protein levels in children ≤15 years with M. pneumoniae lower respiratory tract infection^a

Independent variable	Regression coefficient (β)	SE	P value
Constant		6.45	0.024
P1 type	0.12	3.56	0.021
Gender	0.05	3.17	0.354
Age	0.10	0.49	0.047
Interval between disease onset and antibiotic therapy initiation	−0.11	0.25	0.034

^aC-reactive protein is the dependent variable. Significant differences (P < 0.05) are higlighted in bold. Abbreviation: SE, standard error.

TABLE 4.

Logistic regression analysis of hospital admission of M. pneumoniae-infected patients^b

Strain (age group)b	Crude OR (95% CI)c	Adjusteda OR (95% CI)	P value (crude/adjusted)
P1 type	1.58 (1.01–2.46)	1.51 (0.94–2.41)	0.044/0.087
P1 type (≤5 yrs)	1.21 (0.58–2.51)	1.09 (0.52–2.32)	0.616/0.818
P1 type (5–15 yrs)	1.83 (1.01–3.30)	1.83 (1.00–3.33)	0.046/0.050

^aAdjusted for age and time period between disease onset and antibiotic therapy initiation.

^bP1 type 1 versus P1 type 2 subtype. Significant differences (P < 0.05) are higlighted in bold.

^cAbbreviations: CI, confidence interval; OR, odds ratio.

To better assess the relationship between the P1 genotype and disease severity and point out any differences related to age, additional characteristics of 143 patients younger than 5 years and 277 patients 5 to 15 years old were compared (Table 5). Younger patients had a significantly higher rate of viral codetection. When comparing the CRP level and hospital admission rate between the two genotypes in these two age groups, patients infected with P1-2 strains had a higher median baseline CRP level and a higher admission rate in the group of 5- to 15-year-olds (Table 6). No such difference was observed in the group of younger than 5 years. Similar results were also obtained when solely analyzing data of patients with no viral codetection (Table 6).

TABLE 5.

Characteristics of patients with M. pneumoniae lower respiratory tract infection divided into two age groups^a

Characteristic	Data for patients of:		P value
	Age ≤5 yrs	Age 5–15 yrs
No. of subjects	143	277
% boys/% girls	57/43	49/51	0.163
Macrolide therapy (% [no. of subjects/total no. of subjects])	97 (139/143)	99 (274/277)	0.181
Interval between disease onset and antibiotic therapy initiation (IQR) (days)	8 (6–11)	8 (5–11)	0.878
Viral codetection (% [no. of subjects/total no. of subjects])	43 (43/100)	17 (21/121)	<0.005
Extrapulmonary manifestations (% [no. of subjects/total no. of subjects])	11 (16/143)	13 (36/277)	0.502

^aData on viral codetection were not collected in all subjects. Significant differences (P < 0.05) are higlighted in bold. Abbreviation: IQR, interquartile range.

TABLE 6.

Comparison of C-reactive protein level and admission rate of patients infected with either M. pneumoniae P1 type 1 or M. pneumoniae P1 type 2 strains divided into two age groups^a

Characteristic	Data for patients infected with:		P value	Data for patients infected with:		P value
	P1 type 1, all	P1 type 2, all		P1 type 1, no viral codetection	P1 type 2, no viral codetection
≤5-yr age group
No. of subjects	102	41		38	19
CRP (IQR) (mg/liter)	18 (4–37)	23 (7–36)	0.395	18 (5–39)	26 (6–36)	0.543
Admission rate (% [no. of subjects/total no. of subjects])	54 (55/102)	59 (24/41)	0.616	66 (25/38)	63 (12/19)	0.844
5- to 15-yr age group
No. of subjects	208	69		72	28
CRP (IQR) (mg/liter)	16 (7–33)	28 (10–47)	0.006	14 (5–32)	28 (14–39)	0.015
Admission rate (% [no. of subjects/total no. of subjects])	23 (47/208)	35 (24/69)	0.045	26 (19/72)	50 (14/28)	0.024

^aSignificant differences (P < 0.05) are higlighted in bold. Abbreviations: CRP, C-reactive protein; IQR, interquartile range.

We verified our 2014 results on an independent sample of children with M. pneumoniae LRTI encompassing a 4-year period from January 2009 to December 2012 (Table 7). The P1-2-infected patients had a higher median CRP level and higher rate of hospital admission in both age groups, with a significantly higher median CRP level in the 5- to 15-year-old age group.

TABLE 7.

Comparison of C-reactive protein level and admission rate of patients infected with either M. pneumoniae P1 type 1 or M. pneumoniae P1 type 2 strains from 2009 to 2012^a

Characteristic	Data for patients infected with:		P value
	P1 type 1	P1 type 2
All patients
No. of subjects	37	80
CRP (IQR) (mg/liter)	19 (9–31)	29 (8–49)	0.099
Admission rate (% [no. of subjects/total no. of subjects])	24% (9/37)	36% (29/80)	0.200
≤5-yr age group
No. of subjects	9	16
CRP (IQR) (mg/liter)	25 (10–33)	9 (4–48)	0.516
Admission rate (% [no. of subjects/total no. of subjects])	44 (4/9)	69 (11/16)	0.234
5- to 15-yr age group
No. of subjects	28	64
CRP (IQR) (mg/liter)	18 (8–33)	34 (15–50)	0.034
Admission rate (% [no. of subjects/total no. of subjects])	18 (5/28)	28 (18/64)	0.295

^aSignificant differences (P < 0.05) are higlighted in bold. Abbreviations: CRP, C-reactive protein; IQR, interquartile range.

DISCUSSION

The factors determining disease manifestation and severity of M. pneumoniae LRTI are only partly understood, with undiscovered roles for both host- and pathogen-related factors. The aim of our study was to assess if the M. pneumoniae P1 genotype is associated with severity of LRTI in children. P1 protein is the primary adhesin of M. pneumoniae, as evidenced by antibodies against P1 blocking adherence to host cells and mutants of P1 resulting in nonadherent and avirulent strains (14, 15). P1-1 and P1-2 strains in vitro form biofilms that differ qualitatively and quantitatively (7). P1-2 strains form more robust biofilms that could affect the ability of these strains to resist complement and other antimicrobials with potential impacts on virulence. P1-2 strains may also have higher expression of community-acquired respiratory distress syndrome (CARDS) toxin that is implicated as a significant virulence factor specific to M. pneumoniae (8, 16, 17). CARDS toxin production is strongly correlated with disease severity in an animal model, reinforcing its role as a disease determinant (18). Therefore, delineating the genotypes of distinct M. pneumoniae infections could provide more refined diagnoses and patient-specific treatments.

We compared epidemiological and clinical data of 310 patients with signs of acute LRTI infected with P1-1 and 110 patients infected with P1-2. Patients infected with P1-2 strains presented with a higher median baseline CRP level and a higher rate of hospital admission. CRP is one of the most frequently used markers of inflammation. Plasma levels of CRP can rapidly and dramatically increase in response to infection, inflammation, and tissue injury (19). Moreover, it was shown that the magnitude of CRP increase usually correlates with disease severity, including LRTI (11, 12). The P1 genotype had a significant predictive value in a multiple linear regression model predicting CRP levels in our study sample, indicating an association of the P1 genotype and M. pneumoniae LRTI severity. Moreover, after adjusting for age and interval between disease onset and initiation of antibiotic therapy, the P1 genotype significantly affected the likelihood of hospital admission for 5- to 15-year-olds in a logistic regression model. The 5- to 15-year-old P1-2-infected patients had a higher median baseline CRP level and a higher admission rate than P1-1-infected patients. The difference remained statistically significant when assessing patients with no viral codetection. No similar impact of the P1 genotype on CRP level or hospital admission was observed in the age group of ≤5-year-olds.

When assessing the differences between the two age groups, we observed that younger patients had a significantly higher rate of viral codetection. A high prevalence of codetected respiratory virus in young children suggests that viruses might play a role in making pneumonia clinically apparent in this age group (20). However, when solely analyzing data of patients with no viral codetection, we still did not observe any impact of the P1 genotype on clinical LRTI presentation in younger patients. Recent studies have shown that clinical manifestations of M. pneumoniae LRTI in young children are milder than those in older children (1, 20). This could partly explain why we found a relationship between the P1 genotype and severity of LRTI in older children but not in younger ones, where probably other factors influence LRTI severity.

Our findings were reproduced on an independent sample of children with M. pneumoniae LRTI, with the 5- to 15-year-old P1-2-infected patients having a significantly higher median CRP level. This, combined with results from our large cohort study, suggest that the two M. pneumoniae P1 genotypes may have different pathogenic potential and that LRTI caused by P1-2 strains may have a more severe disease course than those with P1-1 strains in children, as evidenced by higher inflammatory marker levels and higher hospital admission rate. However, we observed no difference in other markers of disease severity, such as requirement for oxygen therapy, duration of hospital stay, or need for intensive care treatment.

In addition to P1 strain typing, multilocus variable-number tandem-repeat analysis (MLVA) is a frequently used method for analyzing the molecular characteristics of M. pneumoniae. A recent study showed that infections with M. pneumoniae type M3-5-6-2 have a higher risk of progressing to severe pneumonia (21). This aligns with our results as in this study 97.9% of type M3-5-6-2 specimens belonged to the P1-2 genotype.

All our patients were recruited from university hospitals, which may have led to a disproportionate number of cases with more severe M. pneumoniae LRTI. The retrospective design of our study limited data collection to the most commonly used clinical variables, most of which were recorded in all patients. A prospective design would allow for assessing other measures of inflammation and disease severity to better understand the pathogenic role of M. pneumoniae genotype. Regardless, few clinical studies to date have investigated the influence of the M. pneumoniae P1 genotype on severity of M. pneumoniae LRTI. Our sample size of 420 patients is one of the largest study samples addressing this question either in children or adults. In addition, in order to better assess the difference in disease severity between the two genotypes, we included outpatients as well as hospitalized patients in our study. A recent Chinese study showed that patients with P1-1 community-acquired pneumonia (CAP) were at greater risk of developing severe pneumonia than P1-2 (22). However, the majority of the patients in their study were younger than 6 years old, while our observed association of M. pneumoniae genotype with severity of LRTI occurred in patients older than 5 years. Furthermore, only hospitalized patients with CAP were included in their study. Importantly, their results might also be affected by the choice of antibiotic treatment as well as the rate of macrolide-resistant M. pneumoniae (MRMP), which is reported to be high in China (23), though these data were not included. In contrast, MRMP was present in less than 1% of specimens in our study. In another report from Sweden based on 45 patients, most of whom were older than the participants in our study, the admission rate was associated with bacterial load, with the M. pneumoniae P1 genotype not related to disease severity (24). Although we have not directly determined the bacterial loads in our study, we achieved a high success rate in M. pneumoniae cultivation. M. pneumoniae was cultured from 70% of PCR-positives samples, indicative of a consistently high bacterial load in the samples provided. This is expected since only patients with signs of acute M. pneumoniae infection were included in our study.

There is a growing body of evidence that prior RTI could alter immunity and pathology to subsequent infections (25). It is likely that older children encountered infection with M. pneumoniae prior to the 2014 epidemic and that previous infections could offer protection against previous circulating types and influence the severity of the resulting illness. The only major M. pneumoniae epidemic encountered by our group of patients was in 2010, when the dominant P1 type was P1-2 (9). There was no information about prior infection with M. pneumoniae in any of the patients included in our study. However, it is interesting that P1-2 still resulted in more severe infections, even though we speculate that the population might have been more naive to P1-1.

As M. pneumoniae is the most commonly detected bacteria in school-aged children hospitalized with CAP (2), M. pneumoniae LRTI represent a significant health care burden. To date, typing M. pneumoniae strains on the basis of P1 protein has been mainly used to define M. pneumoniae epidemics (16) and is not routinely performed when assessing an individual patient with M. pneumoniae LRTI. Our results suggest P1 typing could be utilized to quickly predict the severity of the disease course. This would be beneficial for early recognition, individual selection of treatment plans, and analyzing prognoses. More efficient treatment regimens would help relieve some of the economic strain induced by this ubiquitous respiratory pathogen. Therefore, it is essential that future studies of other patient populations or other epidemics examine the relationship of M. pneumoniae genotypes to severity of LRTI.

Conclusions.

The results from our large cohort suggest that the two M. pneumoniae P1 subtypes may have different pathogenic potential and that LRTI caused by P1-2 strains may be more severe than those caused by P1-1 strains in children. This was especially observed in children older than 5 years. P1 genotyping is not routinely performed but could be used to predict the presentation and severity of the disease course, which could enable early recognition and lay the groundwork for individualized treatment plans.

Previous studies on clinical presentation of M. pneumonia LRTI have generally not considered the role of the P1 genotype. Prospective collection of respiratory samples for M. pneumoniae culture and typing could further elucidate the impact of M. pneumoniae genotype on disease presentation and severity of LRTI.