The number of invasive GBS infections in the elderly population is continuously increasing [5], but the reason for this phenomenon is unclear. In Denmark, between 2005 and 2018 the incidence of invasive GBS in adults above the age of 65 years old increased more than twofold (from 3.23 to 8.34 per 100,000). Similar trends have been reported in other countries [8,9,10,11,12]. The prevalence of comorbidities that increase in parallel with age or an increase in GBS colonization have been discussed as potential contributing factors. Although the prevalence of GBS colonization in pregnant women has been investigated in numerous studies, with an average estimate around 20%, the prevalence in the elderly population – notably a group with increasing invasive GBS infections – is unknown. The GBS colonization rate is associated with sexual experience and activity [22, 23]. Considering that sexual activity in older people can change over time [24] and may have increased in recent decades, we aimed to determine the vaginal GBS colonization rate in elderly women. Our study showed a prevalence of GBS colonization of 17% in postmenopausal women (mean age, 68 years), similar to that reported by Moltó-Garcia et al. (17.8%) in Spain [25]. Edwards et al. reported a colonization rate of 21.7% in 254 healthy elderly participants (mean age, 73 years) in 2005 [15]. Kaplan et al. found a prevalence of GBS colonization of 12% among 167 elderly home residents (median age, 84 years) in 1983 [26]. These data indicate that the GBS colonization rate in pregnant women and healthy elderly adults is similar [16].
All GBS isolates preserved susceptibility to penicillin. However, compared with our previous study conducted on GBS isolates from pregnant women tested between 2009 and 2010 in the same geographical area, we observed a higher proportion of isolates that were non-susceptible to erythromycin (22% vs 14.6%), to clindamycin (14% vs 8.2%), and to both clindamycin and erythromycin (11% vs 7.7%) and that displayed inducible clindamycin resistance (8% vs 5.8%) [20]. However, a scientific comparison is not possible, because no longitudinal data were obtained. Moltó-Garcia et al. reported a similar resistance rate to erythromycin (23.4%) among their GBS samples collected between 2011 and 2012. Although they detected a higher prevalence of constitutive clindamycin resistance (20.6%), they observed the MLSB phenotype in only 0.9% of their strains. Increasing trends of resistance were also reported elsewhere [7,8,9,10,11, 25, 27]. It is possible that elderly individuals were exposed to antibiotics more frequently than were pregnant women, and hence, GBS isolates display a higher resistance rate. However, we did not make such a comparison in our study.
We found one GBS isolate with L phenotype (clindamycin resistant but erythromycin susceptible). This phenotype is rare and may occur via the inactivation of lincosamide-specific nucleotidyl-transferases encoded by lnu genes [28]. Alternatively, this unusual mechanism of resistance may also be mediated by the ABC transporter, encoded by lsaC genes, and be responsible for cross-resistance to streptogramin A (LSA phenotype) and pleuromutilins (LSAP phenotype) [29, 30]. Recently, there have been increasing reports of such a phenotype in the United States, Europe, China, Korea and other countries, with a prevalence ranging from 0.26% in Italy to 15.9% in Korea [31,32,33,34]. This phenotype was detected in one of the most common clones circulating worldwide, ST19. This observation is worrisome because clindamycin is a frequently used alternative in patients with documented allergy to penicillin.
The most prevalent capsular serotypes in our population were III, V and Ia [20]. ST1, ST19 and ST23 were the predominant clones, accounting for 28% of our isolates. All three STs have been consistently reported to be significantly associated with asymptomatic colonization because of their limited invasive ability [14]. However, when it belongs to capsular serotype V, ST1 has been related to invasive disease, and a possible origin from a bovine ancestor has been hypothesized, similar to the case for hypervirulent clone ST17 [35]. Likewise, ST23 was found in carriage and invasive isolates [36]. Clone ST17 was identified in only one strain.
We confirmed significantly higher rates of resistance to macrolides and clindamycin associated with the genetic background of ST1, belonging to clonal complex 1, as previously described [27]. However, in contrast to Lopes et al., who reported the association of ST1 and capsular serotype Ib, we observed a relation to capsular serotype V [27]. The association of ST1 and capsular serotype V has also been described elsewhere [37].
Our study has limitations. Because of slow recruitment, the study was terminated prior to reaching the calculated target sample size, ending in a relatively small study population. However, the number of participants was comparable to (or even larger) than those in previous studies, and the GBS prevalence was similar to that of a study that included 600 individuals [15, 25, 26]. We only obtained vaginal and not recto-vaginal swabs, and may have missed an unknown proportion of GBS colonized individuals. However, we are convinced that the potential difference between the two sampling methods did not influence significantly the overall GBS colonization rate in our study population. Eight GBS isolates were lost for phenotypic and genotypic analysis (Fig. 1). Given the lack of association between risk factors, resistance testing and serotype, it is unlikely that the results of these eight lost GBS isolates would have changed the overall findings.