Overall, 783 CAP patients were enrolled during March 2011–December 2012 and were tested for M. pneumoniae; 96 (12.3%) were positive. Patients who had M. pneumoniae infection were significantly younger and more frequently female, had fewer co-occurring conditions, and experienced significantly milder disease that did those who did not have M. pneumoniae infection (Table 1). M. pneumoniae–positive patients were more frequently pretreated with antimicrobial drugs; however, we found no significant differences in the classes of antimicrobial drugs administered, particularly not in the use of macrolides.

Using real-time PCR targeting the RepMP1 copies in the M. pneumoniae genome, we calculated a median of 7.8 × 10³ copies (range 4.1 × 10¹ to 1.5 × 10⁶) in the M. pneumoniae–positive samples. M. pneumoniae positivity showed a clear season-dependent trend; low positivity of ≈4% was found at the beginning and the end of the investigation period, but high positivity was found during October 2011–December 2011. Quarterly incidence ranged from 1.5% (quarter 3, 2012) to 27.3% (quarter 4, 2011) (Figure 1).

The percentage of M. pneumoniae–positive patients from each age group ranged from 28.1% for the 18- to 29-year age group to 13.5% for the >60-year age group (Figure 2). The prevalence of M. pneumoniae decreased by age group: 18–29 years, 38%; 30–39 years, 31%; 40–49 years, 17%; 50–59 years, 13%; >60 years, 3%. More than half (55%) of M. pneumoniae–positive patients were female; only the 18- to 29-year age group had more M. pneumoniae–positive men than women.

Regarding the P1 genotype, all strains in the 96 M. pneumoniae–positive samples could be typed culture independently. Subtype 1 strains dominated (60.4%; Figure 1, panel B), followed by variant 2a (19.8%), variant 2b (9.4%), variant 2c (8.3), and subtype 2 strains (2.1%). A high proportion of subtype 1 strains were found during the entire 22-month investigation period, and all P1 types detected during the period were found at nearly the same proportion during the high-incidence period of October–December 2011.

Highly discriminatory MLVA was carried out on all positive samples using nested PCR. For 87 of the 96 samples, the complete recommended panel of 5 tandem-repeat regions could be amplified and sequenced successfully. Overall, 23 MLVA types were identified. The most common MLVA types were 4/4/5/7/2 (n = 12, 13.8%) and 5/4/5/7/2 (n = 11, 12.6%). During the high-incidence period (quarter 4 of 2011), the 42 M. pneumoniae strains that showed valid MLVA results belonged to 16 MLVA types. No clear correlation was found between P1 and MLVA typing. The 51 classifiable subtype 1 strains can be assigned to 16 MLVA types (Table 2). Five of these MLVA types can also be found in subtype 2 and the related variant 2 strains.

On the basis of sequencing data, macrolide resistance can be assumed in 3 of the 96 M. pneumoniae–positive specimens (3.1%). All strains showed an A→G mutation at position 2063 of the 23S rRNA. The macrolide-resistant strains belonged to the predominant subtype 1 in P1 typing but differed in MLVA type (2/3/6/6/2, 2/4/5/7/2, and 5/4/5/7/2). The specimens containing resistant strains were sampled at different points of the study: December 2011, April 2012, and August 2012. All affected patients were female (ages 31, 42, and 42 years). For 2 of these patients, treatment with macrolides during the month before sampling was reported.

We further tested 794 patients for Chlamydia spp. (Table 3); 31 (3.9%) patients tested positive, 6 with dual infections (Table 4). In contrast to the M. pneumoniae findings, we found no significant differences in clinical characteristics between Chlamydia-positive and -negative patients (Table 3). Notably, no Chlamydia-positive patients received macrolides, whereas 8.3% of Chlamydia-negative patients did.

Test results identified C. psittaci (2.1%) as the most prevalent chlamydial species, followed by C. pneumoniae (1.4%). In addition, C. trachomatis, Simkania negevensis, and the animal pathogens C. suis, C. abortus and Waddlia chondrophila were identified in individual samples. The use of a DNA microarray assay combined with real-time PCR assays enabled us to detect multiple chlamydial infections; all 6 dual chlamydial infections (representing 19.3% of Chlamydia-positive patients) involved C. psittaci, 3 in conjunction with C. pneumoniae and 1 each with C. abortus, S. negevensis, and W. chondrophila. Co-infections with M. pneumoniae and Chlamydia spp. were detected in 3 samples (3.1% of M. pneumonia–positive and 8.1% of Chlamydia-positive specimens): 1 M. pneumoniae + C. psittaci, 1 M. pneumoniae + C. pneumoniae, and 1 M. pneumoniae + C. psittaci + C. pneumoniae.

Many studies have described the strictly time-dependent epidemiology of infections caused by M. pneumoniae (3). Although we did not include serologic testing to provide further information for the differentiation of colonization of patient from infection, the results of our study confirm a strong epidemic peak of M. pneumoniae–positive respiratory samples among adult patients with suspected CAP in Germany during 2011–2012. The incidence of infections temporarily rose to ≈27% during the fourth quarter of 2011, a level that is in accordance with other reports (5–12). Despite limited comparability (e.g., target used), the number of RepMP1 copies measured with real-time PCR (median 7.8 × 10³) is in the range of results of other studies (32,33). Because of the short duration of the epidemic peak and known deficiencies in testing routines for symptomatic M. pneumoniae patients, an increase in incidence could easily escape the notice of public health authorities. Moreover, because patients with M. pneumoniae infection were significantly younger than those without infection, it is possible that the true incidence might even be higher, given the fact that younger persons visit physicians less frequently and are admitted to hospitals less often than older patients. In addition, β-lactams are often used as the first-line antimicrobial drugs for CAP but are known to be inefficient in treatment of Mycoplasma infections; this conflict might represent another reason for the spread of this pathogen. In light of these results, revision of recent guidelines for management of CAP with antimicrobial drugs should be considered.

Typing of strains can help clarify the dynamics of epidemic peaks. There is no evidence that the incidence peak we registered was related to the genotype of circulating strains. M. pneumoniae is a genetically conserved organism, which implies limits to potential typing targets; most frequently used is P1, the main adhesin and most immunogenic protein, where sequence variation occurs mainly in the 2 copies of repetitive elements RepMP2/3 and RepMP4 of the P1-encoding locus mpn141. The epidemiologic importance of P1 genotypes is based on their ability to generate a specific host immune response (13). Therefore, P1 genotyping of circulating mycoplasma strains is helpful for understanding host-pathogen interactions and the infections ensuing. The current dominance of subtype 1 strains in combination with a rare occurrence of subtype 2 was also described in Europe and Asia (11,34). Subtype 2 strains have been replaced with the phylogenetically related variant 2 strains, which occurred in our study in different types. The variant 2c of this group was described in 2011 in several isolates from China (34); our detection of variant 2c strains in Germany confirms the parallel circulation of most variant 2 strains described this far.

In this study, only particular regions of the P1 gene were selected for analysis, and the occurrence of additional P1 types showing further sequence variations cannot be ruled out. The recently characterized variant 2d (35) shares an identical 3′ part of the RepMP2/3 with variant 2a but differs in the 5′ region of the repetitive element; on the basis of those results, we retested all variant 2a strains, but we could not confirm the occurrence of variant 2d.

For epidemiologic reasons, it is important that the period with a high proportion of M. pneumoniae–positive samples (October–December 2011) was not associated with a change of the dominating genotype nor with the presence of a particular P1 type. Previous reports have hypothesized that the circulation of genotype-specific antibodies in the human population can influence the number of infections and the dominating P1 type (13), but our data do not support this hypothesis.

In recent years, MLVA was introduced for typing of M. pneumoniae isolates (14) and extended to culture-independent typing from clinical samples (31). The determination of the number of 5 VNTR markers (Mpn1, 13–16) enables characterization of strains with a higher discriminatory power in comparison with P1 typing. Whereas the genomic regions used for MLVA are located mainly between genes and within genes of unknown function, the P1 protein plays a critical role in host–pathogen interaction. The most common MLVA types (4/4/5/7/2 and 5/4/5/7/2) we detected were also found in high abundance in strains recently identified in France, China, and the United States (11,16,36), thus indicating a worldwide dissemination of particular MLVA types.

Regarding the strains included in our study, the assignment of P1 types to MLVA types confirmed that no clear relationship between the typing methods exists (Table 2). In agreement with the results of other reports (11,14,16), a low number of P1 subtype 1 strains belong to MLVA types that are typical for subtype 2 or variant strains (e.g., 5/3/5/6/2) and vice versa. In our opinion, both typing approaches are of practical importance and complement each other; the circulating P1 types reflect a more host-dependent pattern, whereas MLVA can differentiate strains with higher discriminatory power, enabling a better understanding of epidemiologic relationships.

Instability of VNTR marker 1 has been reported (36), and the removal of Mpn1 from the current MLVA scheme has been suggested (16). For our data, the removal of Mpn1 would reduce the number of MLVA genotypes to 9, with >80% of strains belonging to 2 types, 3/5/6/2 and 4/5/7/2. This change would result in a substantial decrease of the discriminatory power of the MLVA typing method and would require efforts to include further VNTR markers showing a stable number of repeats within a given strain.

The data from our study confirm that a nationwide peak of infections caused by M. pneumoniae is polyclonal, which is in agreement with results of other reports (11). In contrast, the clonal spread of M. pneumoniae can only be expected for small-scale endemic outbreaks with person-to-person transmission in close communities, such as families (37).

Since 2000, an increase of the worldwide occurrence of macrolide resistance in M. pneumoniae strains to 90% and higher has reported. Compared with data from Asia (16), the proportion of 3.1% of macrolide-resistant M. pneumoniae strains we detected is low. Results of studies from Germany in recent years have showed results in the same range (15), thus indicating a nearly constant prevalence of resistance. Nevertheless, resistant strains are circulating in the population we investigated, which requires further monitoring of strains to provide an updated overview of drug resistance. Several reports have confirmed that resistant strains were selected during antimicrobial drug treatment (38). However, in our study, increased prescription of macrolide antimicrobial drugs, which can be expected during a period of high CAP incidence, did not result in a measurable increase of resistant strains. In addition, because most current resistance data originate from pediatric patients, the results of this study are of general interest because we report on adult patients.

Although further sequence differences of the 23S rRNA locus of M. pneumoniae have been described, the A→G mutation at position 2063 is most common (15,16). Comparison of sequencing results with antimicrobial susceptibility testing confirmed that the mutation at positions 2063/2064 led to a high level of macrolide resistance (15). Thus, with the mutation detected, resistance of these strains can be assumed. We did not find macrolide resistance to be associated with a certain MLVA type, which confirms the findings of other reports (16), but the low number of resistant strains in our study is insufficient for us to draw a final conclusion.

The results of Chlamydia testing are remarkable for the comprehensive methodologic approach. In the past, the choice of diagnostic tests was usually limited: that is, either only C. pneumoniae was analyzed or all Chlamydia spp. identified were thought to be C. pneumoniae. Although the positivity of 1.4% for C. pneumoniae we found is in line with previous data from Germany (22), the proportions of positive samples of 2.1% for C. psittaci and 4.7% for all Chlamydiae spp. considered here deserve specific attention (Table 4). Chlamydial co-infection involving >2 Chlamydia spp. has been shown to be frequent in trachoma patients (39), but data from pneumonia patients remain scarce. Our finding that 19.3% of Chlamydia-positive patients harbored >2 Chlamydia spp. is in the same range as the 24% found in a recent trachoma study (40).

One unexpected finding in our study was that C. psittaci positivity could not be correlated with the patients having regular contact with birds at home or at work. Moreover, no seasonal patterns were identified. Nevertheless, our findings indicate that chlamydial species other than C. pneumoniae should be included in testing of CAP patients; in particular, C. psittaci should be included but also C. trachomatis. The relatively high prevalence of C. psittaci we found raises questions about its epidemiologic and etiologic importance, which should be addressed in future studies.

In summary, we confirmed a strong epidemic peak of M. pneumoniae–positive respiratory samples among adult patients with confirmed CAP in Germany during 2011–2012. During the fourth quarter of 2011, the incidence of infections temporarily rose to ≈27% of all patients investigated. Chlamydia spp. were found in 3.9% of samples, without epidemic peaks. Throughout the study period, the dominant M. pneumoniae P1 type was subtype 1; only 3.1% of M. pneumoniae strains were macrolide resistant. Both bacteria represent relevant pathogens in CAP, and awareness of their epidemiology, particularly among clinicians, is clearly warranted.