Two thousand and nine bacterial strains, reported on the 10th of January 2018 and deemed clinically relevant by the local DCM in charge, were collected for genetic characterisation. Sixty-two (3.1%) were H. influenzae. Selection of strains was at the discretion of the responsible DCM, and criteria varied. Based on electronic reporting, the true prevalence on the 10th of January was 103 isolates of H. influenzae in 103 samples from 103 patients, corresponding to 1.78 per 100,000 person-days (all samples), or 2.47 per 1,000 hospital bed-days (hospital samples). One Danish region reported 15 specimens with H. influenzae and submitted 15 strains for characterisation; the nadir was 21 reported vs four collected. However, 62 genetically characterised isolates constitute a solid and credible fraction of 103 point prevalence strains.
All 62 sequenced isolates belonged to phylogenetic group I and were unencapsulated. Phylogenetic characterisation of the species usually focuses on diversity, as well as strains from invasive infections [27, 31, 32], and this aim may explain the overrepresentation of phylogenetic group II and the six capsular types, as well as outgroup strains, in published reports. The current depiction of the species underlines the importance of phylogenetic group I strains as commensals and in mucosal infections.
The genomic characterisation disclosed nationwide, circulating clones. H. influenzae has rarely been subject to epidemiologic investigations, and the current study revealed some obstacles when genomic distances are interpreted in order to guide infection control. Due to inclusion of intergenic sequences, read mapping invariably generated larger, pairwise SNP distances than alignment of core genes. However, the “molecular clock” of the two bioinformatics tools was not congruent, as demonstrated by the discordance of the two methods to identify the most closely related genomes in the large cluster of ST103 strains (Fig. 2). Roary-estimation of shared genes of close neighbours did, in some cases, seclude annotated genes to presence in only one of the two strains. Uptake of genes can also reflect a molecular clock, but true assessment of genomic content will require closing of genomes, or use of long-read sequencing technology. The most significant or decisive bioinformatics tool for epidemiologic investigations has yet to be clarified.
Susceptibility to ampicillin was assessed by phenotype (growth in the presence of antimicrobial agent) for 47 strains, and by genotype (WGS) for 62 strains. Categorisation of resistance differed markedly, with ampicillin resistance rates of 36% and 53%, respectively. A similar difference was reported from Japan, where 39% of upper respiratory tract isolates were non-susceptible to ampicillin (MIC ≥ 2 µg/ml), while 65% were genetically β-lactamase-negative ampicillin-resistant (gBLNAR) [15]. Personal identifiers and original specimen numbers were blinded to this study, and the two samplings retrieved from the same prevalence material of 103 strains were not identical. Ampicillin-resistant strains may be overrepresented in conjunctival strains rarely subjected to AST, but it should be emphasised that the two definitions of resistance are not interchangeable. Mutational resistance comes with a wide range of MIC distribution, and the accepted precision of ± 1 dilution for reference broth microdilution AST [33] complicates the interpretation when recorded MICs are close to the breakpoint. Partly to address the difficult identification of strains with mutational resistance, EUCAST has recently suggested that H. influenzae can only be categorised as susceptible to oral amoxicillin if an increased dosage of 750–1000 mg X 3 is used (EUCAST Clinical Breakpoint Tables v. 10.0). Finally, the unstable growth of fastidious organisms can affect the categorisation of phenotypic resistance. It is well known that methods and media have important consequences for detection of resistance in H. influenzae [34]. General use of the EUCAST disk-diffusion method for categorisation of susceptibility to aminopenicillins has reduced the inter-laboratory inconsistency, but essential and categorical agreement with minimal inhibitory concentration determined by broth microdilution is still modest [35].
PBP3-substitution N526K was detected in 13 of 62 strains (21%), which is twice the rate that was reported from one Danish region in 2007 [36]. The current strains showed a conspicuous accumulation of certain additional substitutions in the transpeptidase region of the protein. PBP3 substitutions identical to ftsI clusters A and B were also common in Norway in 2007, being disclosed in 48 (43%) and 17 (15%) of 111 strains with the N526K substitution [37]. Although genetic comparison in that study was restricted to MLST, a similar tendency to congregation of ftsI cluster B, and dispersion of ftsI cluster A, was observed. This “linkage disequilibrium” between ftsI cluster A and other core genes can arise by selection of particular mutations associated with survival in the presence of repeated exposure to antimicrobials, or by lateral transfer of an optimised fragment of the PBP3 gene. Inactivation of essential competence genes was preferentially observed in strains with wild-type PBP3, but the prevalence of truncated competence genes was too low to firmly disclose the origin of the gene fragment(s) encoding N526K. There is extensive variation in natural competence in H. influenzae [38], and other genes may markedly depress transformation without being essential for competence. Biologic assessment of competence in a random sample of clinical strains is required to elucidate the putative importance of lateral transfer for mutational resistance in H. influenzae.
The point prevalence study has emphasised the significance of unencapsulated H. influenzae, which may be an under-recognised pathogen [1]. Circulating clones are frequent, but infections may be related to host factors, rather than direct transmission of epidemic strains. Revelation of aminopenicillin resistance in half of point prevalence strains by WGS must be addressed by meticulous AST, and confirms the concerns that have been raised by WHO.