Mutational change of CTX‐M‐15 to CTX‐M‐127 resulting in mecillinam resistant Escherichia coli during pivmecillinam treatment of a patient

Abstract Pivmecillinam (amdinocillin pivoxil) is the recommended first‐choice antibiotic used to treat urinary tract infections (UTIs) in Denmark. The frequency of mutation to mecillinam (MEC) resistance is described as high in vitro; however, treatment of UTI has a good clinical response and prevalence of mecillinam resistance in Escherichia coli remains low despite many years of use. We describe occurrence of in vivo mecillinam resistance in a clinical isolate of ESBL‐producing E. coli following pivmecillinam treatment. The identified phenotypic differences in the mecillinam resistant isolate compared with the original mecillinam susceptible isolate were a full‐length LPS with O‐antigen (O25), mecillinam resistance and a lower MIC for ceftazidime. Regarding genotype, the resistant isolate differed with a mutation in bla CTX‐M‐15 to bla CTX‐M‐127, loss of a part of a plasmid and a genomic island, respectively, and insertion of a transposase in wbbL, causing the rough phenotype. The observed mecillinam resistance is expected to be caused by the mutation in bla CTX‐M‐15 with additional contribute from the serotype shift. We continue to recommend the use of pivmecillinam as first‐line treatment for UTI.

Denmark, we have seen a low prevalence of mecillinam resistance in E. coli with 7% among urine isolates in Danish hospitals and 6% of primary care urine isolates, despite many years of use (DANMAP, 2016). The resistance mechanisms reported from in vitro studies include a large variety of mutations in genes associated with, for example energy metabolism and LPS synthesis (Thulin et al., 2015). Thus, mecillinam resistance is difficult to detect based on genomic data.
In this study, we describe the development of mecillinam resistance following pivmecillinam treatment of a patient with complicated urinary tract infection caused by an extended-spectrum β-lactamase (ESBL)-producing E. coli isolate sampled from a previous study by Jansåker, Frimodt-Møller, Sjögren, and Dahl Knudsen (2014). Two clinical isolates without and with mecillinam resistance from the same patient were isolated before and after pivmecillinam treatment, respectively (MEC-S and MEC-R, respectively). The isolates caused complicated lower UTI in the patient, and resistance to mecillinam was observed, after treatment with pivmecillinam 400 mg t.i.d during the primary infection. The patient had recurrent symptoms of UTI, and the second urine sample, two weeks after the initial sample, revealed the same ESBL-producing E. coli, however, now resistant to mecillinam.

| MATERIAL S AND ME THODS
A range of phenotypic and genotypic characterizations were performed to characterize the isolates. With respect to phenotype, the isolates were subject to MIC determination using E-test®, disk diffusion (EUCAST method) and serotyping. Additionally, the isolates were tested for synergy between mecillinam (10 µg) and ampicillin (10 µg) and amoxicillin/clavulanate (30 µg), respectively, by disk diffusion in order to evaluate the activity of mecillinam + ampicillin

| RE SULTS AND D ISCUSS I ON
The phenotypic analyses proved identical MIC values between the two isolates for all tested antimicrobials, apart from mecillinam (4 → >256 µg/ml) and ceftazidime (128 → 8 µg/ml) ( Genomic analyses revealed no differences between the two isolates with respect to MLST (ST131) and phylogroup (B2) ( Table 1) Regarding serotype, SeroTypeFinder identified O25:K-:H2 in both MEC-S and MEC-R. However, from the phenotypic serotype, it was evident that MEC-S was a rough isolate that did not produce the O25 antigen, but genetically, it belonged to the O25:K-:H4 group (Table 1). SeroTypeFinder does not identify rough phenotypes. wbbL has previously been correlated with rough phenotypes so we compared this gene and the O-antigen cluster of the two isolates by alignment in Geneious. The results showed that MEC-R had a complete wbbL, whereas MEC-S had a transposase inserted in wbbL disrupting the gene. We consider this the genetic cause of the rough serotype. Mutations in LPS have previously been described to affect mecillinam susceptibility in combination with other mutations (Antón, 1995). It is possible that this contributed to the increased MIC of mecillinam in MEC-R, but this should be investigated further.
We performed GenAPI followed by visual inspection of read mappings of the GenAPI identified areas of the genomes. These analyses revealed two regions with differences between the genomes: isolates, and therefore, we conclude that both isolates carried two copies of bla TEM-1B .
The MIC for mecillinam of the parental MEC-S strain was 4 mg/L, which was below the breakpoint of 8 mg/L, but higher than the ECOFF (1 mg/L). Whether this lower susceptibility to mecillinam increased the risk for development of resistance after exposure to mecillinam should be investigated, since it could imply for a change of the clinical breakpoint for mecillinam.

| CON CLUS IONS
In summary, here we describe clinical occurrence of mecillinam resistance in an ESBL-producing E. coli following pivmecillinam treatment. We expect that the mutation in bla  in combination with the serotype change caused the increased MIC for mecillinam and the lower MIC for ceftazidime.
Due to the low risk of resistance development, we continue to recommend the use of pivmecillinam as first-line treatment for UTI.

ACK N OWLED G M ENTS
The study was supported by Danish Council for Independent Research.

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C A L A PPROVA L
The study has been approved by the Danish Data Protection Agency (I -suitnr. 01755 and id.nr. HVH-2012-022) for a previously published study (Jansåker et al., 2014). Included patients were asked to participate and gave written consent prior to inclusion.  Antón, D. N. (1995). Resistance to mecillinam produced by the co-operative action of mutations affecting iipopolysaccharide, spoT, and cya or crp genes of Salmonella typhimurium. Molecular Microbiology, 16, 587-595. https ://doi.org/10.1111/j. 1365-2958.1995.tb024 21.x