Rise of multidrug-resistant non-vaccine serotype 15A Streptococcus pneumoniae in the United Kingdom, 2001 to 2014

Conjugate vaccines have reduced pneumococcal disease in vaccinated children and unvaccinated adults, but non-vaccine serotypes are of concern, particularly if antibiotic resistant. We reviewed Streptococcus pneumoniae collected via: (i) the British Society for Antimicrobial Chemotherapy (BSAC) surveillances from 2001–2014; (ii) Public Health England’s (PHE) invasive isolate surveillance from 2005–2014 and (iii) referral to PHE for resistance investigation from 2005–2014. Serotype 15A increased in all series, with many representatives showing triple resistance to macrolides, tetracyclines and penicillin. 15A was consistently among the 10 most prevalent serotypes from 2011 in PHE and BSAC invasive isolate/bacteraemia surveillance but never previously; 26–33% of these invasive 15A isolates had triple resistance. BSAC respiratory isolates were only serotyped in 2013/14 and 2014/15 (October to September); 15A was most prevalent serotype in both periods, comprising 9–11% of isolates, 38–48% of them with triple resistance. Serotype 15A represented 0–4% of S. pneumoniae referred to PHE for reference investigation annually until 2008 but rose to 29% (2013) and 32% (2014). Almost all multidrug-resistant 15A isolates were sequence type (ST) 63 variants, whereas susceptible 15A isolates were clonally diverse. The rise of serotype 15A suggests that pneumococcal conjugate vaccines will need ongoing adaptation.


Introduction
Seven-valent pneumococcal conjugate vaccine (Prevenar 7, PCV7) first became available internationally in 2000, and protects against invasive Streptococcus pneumoniae infection by serotypes 4, 6B, 9V, 14, 18C, 19F and 23F. Numerous countries have reported that deployment reduced the incidence of invasive (i.e. blood and cerebrospinal fluid (CSF)) S. pneumoniae disease both in children, who are vaccinated, and in elderly adults, who benefit through reduced carriage and transmission of virulent serotypes by children [1][2][3][4]. Antibiotic resistance was historically concentrated in five PCV7-targeted serotypes (6B, 9V, 14, 19F and 23F) [5] and several countries have reported reductions in the prevalence of resistance as these were displaced [6]. United Kingdom (UK) experience conforms to these general patterns [7], with the caveat that penicillin-nonsusceptible S. pneumoniae were uncommon before the vaccine's introduction to the childhood schedule in 2006/07, meaning that little further fall occurred; macrolide resistance was reduced, reflecting displacement of a resistant serotype 14 lineage [8,9].
The success of PCV7 was partly offset by rises in other serotypes; notably 19A, where multidrug resistance to antibiotics became frequent [10,11]. This was countered by replacing PCV7 with a 13-valent conjugate vaccine (PCV13), additionally covering serotypes 1, 3, 5, 6A, 7F and 19A. PCV13 replaced PCV7 in the UK in April 2010 and this switch was followed by (i) reduced infant carriage of these additional serotypes [12], and (ii) a further 56% reduction in invasive disease incidence from a post-PCV7 baseline [13]. Again, however, rises are being seen in other, non-vaccine, serotypes, principally 8, 10A, 12F, 15A and 24F [13]. Serotype 15A is of particular interest since multidrug-resistant isolates belonging to this serotype have been reported as far apart as east Asia [14][15][16], North America [17,18], Norway [19], Italy [20] and Australia [21]. Here, we explore the rise of serotype 15A S. pneumoniae in the UK and Ireland at epidemiological and molecular levels, using data from both the British Society for Antimicrobial Chemotherapy (BSAC) and Public Health England (PHE) surveillances.

British Society for Antimicrobial Chemotherapy surveillance
The BSAC Bacteraemia and Respiratory Surveillance Programmes have been described previously [22][23][24].

Identification, serotyping and susceptibility testing
All surveillance and referred isolates were confirmed as forming alpha-haemolytic colonies on horse blood agar and being inhibited by a 5 µg optochin (ethylhydrocupreine hydrochloride) disc (Oxoid-Thermofisher, Basingstoke, UK). Isolates with atypical colonial morphology, or which could not be serotyped (below), were confirmed as being lysed within 30 min by 2% sodium deoxycholate, and being catalase-negative when tested with 3% hydrogen peroxide. For serotyping, isolates were grown overnight in Todd Hewitt broth at 35 °C with 5% CO2, harvested by centrifugation at 453 g for 30 min, then re-suspended in a small residual volume of broth and subjected to slide agglutination tests with standard antisera (Statens Serum Institut, Copenhagen, Denmark) [26]. Agar dilution susceptibility tests were performed in accordance with BSAC guidelines [27], using IsoSensitest agar (Oxoid-Thermofisher) supplemented with 5% defibrinated horse blood and incubated at 35-37 °C in a 5% CO2 atmosphere. 'Triple resistance' was defined as resistant to erythromycin (minimum inhibitory concentration (MIC) > 0.5 mg/L) and tetracycline (MIC > 2 mg/L), and non-susceptible to penicillin (MIC > 0.06 mg/L), based on EUCAST breakpoints [28])

DNA extraction, sequencing and bioinformatic analysis
Isolates were grown on horse blood agar (PHE Media Services) and treated by the Qiagen-recommended method for lysis of Gram-negative bacteria (Qiagen, Manchester, UK), which is effective for S. pneumoniae and simpler than the Gram-positive protocol. DNA was extracted from the lysates using a QIAsymphony SP automated instrument (Qiagen) and a QIAsymphony [13].
Serotype 15A isolates were encountered in each year from 2010 and the serotype was in the top 10 from 2011 onwards, whereas previously the type was sporadic.
Other types that had long been encountered at moderate to low prevalence also became more prominent,   Table 3 and comprise all isolates from hospitals that participate in the EARS-net surveillance along with those bloodstream isolates where the referring laboratory specifically sought susceptibility testing. Inclusion of the latter group may over-represent resistant organisms, although there is no reason why it should do so disproportionately within particular serotypes.

Serotype trends, isolates referred to Public Health England for investigation of resistance
Between 2005 and 2014, 1,536 S. pneumoniae from respiratory, ear and eye infections were referred to PHE ( Table 5) for investigation of unusual resistance. These submissions constitute a heavily biased sample and lack a denominator, but do provide a rolling snapshot of S. pneumoniae isolates that sending laboratories  Table 3 are line-listed.
(mostly in England and Wales) consider to be concerning.

Genomic sequencing and phenotypes of serotype 15A isolates
Genomic sequencing was performed on 156 serotype 15A S. pneumoniae. These represented a diversity of resistance patterns, and including 50 with triple resistance; a limitation was that all 156 sequenced isolates dated from 2013 and 2014. MLST types were deduced from the sequence data, and 78 (50%) of the isolates were identified as belonging to ST63 (n = 61) or its single or double locus variants (n = 17). All of these 78 ST63-related isolates were resistant to erythromycin (also clindamycin, not shown) and 49 (62.8%) had the triple resistance profile ( Table 6). The macrolide and clindamycin resistance correlated with the consistent presence of erm(B) genes, as detected by WGS. All 78 ST63-related isolates were found also to carry the tetracycline-resistance determinant, tet(M); those (n = 65, 83.3%) that expressed tetracycline resistance had the intact gene, whereas those (n = 13, 16.7%) that were tetracycline-susceptible (all of them classical ST63 isolates) had a deletion of two nucleotides at codon 339, generating a premature stop codon and thereby inactivating the gene. Most of the 49 isolates with triple resistance were susceptible to alternative agents: 37 remained susceptible to ampicillin, 47 to moxifloxacin, 48 to cefotaxime and all 49 to vancomycin, all based on EUCAST breakpoints. Sequence types (STs) 3811 (n = 19), 58 and its single locus variants (SLVs) (n = 21), and 73 and its SLVs (n = 11) were all heavily represented among the 78 ST63-unrelated serotype 15A isolates and, among all these, just one isolate had triple resistance and three or fewer were non-susceptible to any one of erythromycin, tetracycline or penicillin.
WGS data were available for a further 141 non-15A S. pneumoniae, predominantly investigated owing to multidrug resistance. Six had ST63-related profiles and these all had triple resistance; three expressed serotype 19F, one serotype 21 and one 23F; the final isolate was typed using antisera as serotype 20 but was predicted to be serotype 11A based on WGS; review suggests that the original serotype determination was in error. The association with 19F (a PCV7 serotype) is notable (see Discussion), but members of this serotype were highly variable in terms of ST; among a total of 25 serotype 19F isolates sequenced, 22 with triple resistance, we recorded 12 different known STs, along with two new variants. No single ST had more than four representatives.

Discussion
Deployment of PCVs has had clear public health benefits. The incidence of invasive pneumococcal disease has been reduced not only in vaccinated children, but also in elderly adults, who benefit from herd immunity [31]. There is also evidence of impact on noninvasive disease: thus, PCV7 deployment in the UK in 2006 also was followed by a 19% reduction in hospital admissions for community-acquired pneumonia (CAP) among children aged < 2 years, reversing a rising trend that had persisted during the preceding decade [32]. A similar reduction was reported in Italy [33]. Moreover, a Cochrane review concluded that PCV7 reduced the incidence of acute otitis media in healthy vaccinated children, although with less impact for those with a history of the illness or deemed to be 'high risk' [34]. Lastly, active PCV13 vaccination was recently shown to achieve a 50% reduction in the incidence of bacteraemia and non-invasive pneumonia in elderly adults, again reflecting displacement of vaccine serotypes [35].
A limitation to this pattern of successes is, however, that the PCV vaccines cover only the most prevalent pneumococcal serotypes, leaving scope for expansion of other types. Deployment of PCV7 was followed by increased prevalence of serotype 19A isolates, many of them multidrug-resistant, and, although serotype 19A is now covered by PCV13, a niche may be created for yet further types. Internationally, several groups have remarked on the increased prevalence of multidrug-resistant serotype 15A and 35B isolates [14][15][16][17][18][19][20][21] and a recent PHE analysis of invasive pneumococcal infections, using the data series of Table 3, noted 15A to be among several serotypes now increasing in numbers and proportion in the UK [13]. The present analysis extends these findings, confirming that serotype 15A S. pneumoniae are of growing importance, as also shown (i) in the BSAC bacteraemia series (Table 1), which overlaps the PHE series but also includes Scotland and Ireland, (ii) the BSAC series LRTI (Table 2), which is the sole UK surveillance to test S. pneumoniae from their predominant disease setting, and (iii) among PHE reference submissions, which provide a rolling snapshot of resistance phenotypes causing concern to microbiologists at sending laboratories, which are predominantly in England, Wales and Northern Ireland, although with a few isolates received from elsewhere (Table 5). By 2013 and 2014, serotype 15A was consistently (i) among the top 10 serotypes in both the PHE and BSAC surveillances of invasive S. pneumoniae (Tables 1 and  3), (ii) was the top serotype among respiratory isolates ( Table 2) and (iii) accounted for almost one third of all the S. pneumoniae sent for reference investigation as 'unusually' resistant. Critically, and unlike other rising pneumococcal serotypes (8, 10A, 11A, 12F, and 24Fsee Tables 1, 3 and ref [13]) serotype 15A isolates were commonly resistant or non-susceptible to multiple antibiotics, including macrolides, clindamycin, tetracycline and penicillin. While none of the surveillances captures clinical outcomes, the fact that serotype 15A is rising in invasive infections implies that these organisms are virulent.
Around one third of serotype 15A isolates had 'triple resistance' (i.e. to macrolides and tetracycline together with intermediate penicillin resistance), a higher proportion than for other serotypes (Table 4). This proportion did not change substantially over time (although assessment is complicated by small total numbers of isolates in the earlier years), indicating that the serotype was gaining prominence both generally and as a resistant type, again implying that the surface polysaccharides of serotype 15A support virulence.
Triple resistance among serotype 15A isolates was strongly associated (p < 0.0001, Fisher's exact of chisquared tests) with ST63 and its variants and extremely rare among serotype 15A isolates belonging to other STs. This association between serotype 15A, ST63 and multidrug-resistance has been made by others too [18,36,37] and it was suggested by Frazao et al. [38] that multidrug-resistant ST63-15A organisms arose by type transformation of ST63 strains previously expressing the 19F capsular serotype. The present results provide very little support for this hypothesis. Although 3/22 multidrug-resistant serotype 19F S. pneumoniae examined were ST63 single-or double-locus variants, the remaining 19/22 belonged to diverse sequence types; moreover, ST63 alleles have been reported to be associated with other serotypes besides 19F and 15A, including serotype 8 in Spain [39], where it is suggested that they may have arisen via serotype switching of earlier Sweden 15A lineages [40].
In summary, the present findings imply that conjugate vaccines will face an ongoing game of 'catch-up', as new serotypes rise to prominence, and that expansion beyond a 13-valent formulation will be needed. They are pertinent also to the debate as to whether PCV13 should be adopted for prophylactic vaccination against pneumonia in the elderly, as is advocated based on recent positive trial results in the Netherlands [35]. Such positive findings must be set against the fact that PCV13 strains now account for less than 20% of community-onset pneumococcal pneumonias in the UK ( Table 2).
While the rise of any new multidrug-resistant type is of concern for patient management, the ST63-15A S. pneumoniae had high level resistances only to macrolides, clindamycin and tetracyclines; MICs of penicillin mostly remained in the range 0.12 to 0.5 mg/L, and this level of 'non-susceptibility' is unlikely to compromise outcomes, except in meningitis. Susceptibility to moxifloxacin and cefotaxime remained near-universal, and ampicillin MICs were twofold below those of penicillin, remaining in the susceptible range and reversing the usual pattern for penicillin-non-susceptible S. pneumoniae, where ampicillin MICs mostly exceed those of penicillin. Treatment of infections therefore is unlikely to present especial problems, unless macrolides or tetracyclines are used alone, for example in beta-lactam allergic patients.