Antimicrobial susceptibility of Flavobacterium psychrophilum isolates 1 from the United Kingdom 2

16 Routine application of antimicrobials is the current treatment of choice for rainbow trout 17 fry syndrome (RTFS) or bacterial coldwater disease (BCWD) caused by Flavobacterium 18 psychrophilum. In this study, the antimicrobial susceptibilities of 133 F. psychrophilum 19 isolates, 118 of which were from the UK, were evaluated by broth microdilution and disc 20 diffusion methods following VET04-A2 and VET03-A guidelines of Clinical and 21 Laboratory Standards Institute (CLSI), respectively. Isolates were categorised as wild 22 type (fully susceptible, WT) or non-wild type (NWT) using normalised resistance 23 interpretation (NRI) determined cut-off values (COWT). Broth microdilution testing 24 showed that only 12% of UK isolates were WT to oxolinic acid (MIC COWT 0.25 mg L 25 ) and 42% were WT for oxytetracycline (MIC COWT 0.25 mg L). In contrast, all the 26 isolates tested were WT (MIC COWT 2 mg L) for florfenicol, the main antimicrobial 27 for RTFS control in the UK. Disc diffusion-based COWT values were ≥51 mm for 10 μg 28 amoxicillin, ≥44 mm for 30 μg florfenicol, ≥30 mm for 2 μg oxolinic acid and ≥51 mm 29 for 30 μg oxytetracycline. There was a high categorical agreement between the 30 classifications of the isolates by two testing methods for florfenicol (100%), 31 oxytetracycline (93%), and oxolinic acid (99%). 32 33


36
Flavobacterium psychrophilum, a Gram-negative, filamentous, psychrotrophic bacterium, 37 is the aetiological agent of rainbow trout fry syndrome (RTFS) and bacterial coldwater 38 disease (BCWD), which was first described in USA in 1946(Borg 1948. F. 39 psychrophilum infection has been found throughout North America, Europe and     (Table 1). However, within this set of 125 isolates, there were five 102 groups of two or three isolates that were recorded as having the same site, sampling time   The MIC assays were determined using the broth microdilution protocol 121 recommended for F. psychrophilum in the CLSI guideline VET04-A2 (CLSI 2014a).

122
Colony counts on inoculum suspensions were performed to ensure that the final inoculum 123 density was close to 5.0 x 10 5 colony-forming units (CFU) per millilitre. Mueller-Hinton medium (Sigma-Aldrich, UK; 3 g L -1 ) containing 1.5% agar (Agar No.   The NRI analyses for zone histograms were performed using a modification of the     Table 5. The distribution of disc diffusion zones of the isolates for six 196 antimicrobials is presented in Figure 1 and the zone data-based COWT values of 197 antimicrobial agents are shown in Table 6.  (Table 5). Applying this cut-off, fifty-six (42%) of the 133 isolates analysed 204 were categorised as WT.

205
The disc diffusion zone sizes for OTC30 showed considerable diversity at the high 206 zone end ( Figure 1A). However, NRI analysis of these data identified a high zone modal 207 group with a standard deviation of 7.44 mm. If this modal group was assumed to represent zones obtained from WT isolates, a provisional COWT value of ≥51 mm could 209 be calculated (Table 6). Applying this cut-off, sixty-five (49%) of the 133 isolates 210 analysed were categorised as WT.

211
The categorisation of isolates resulting from applying the cut-off of ≤0.25 mg L -1 to 212 the MIC data agreed with the categorisation resulting from applying the disc zone cut-off 213 of ≥51 mm to the zone data for 93% of the 133 isolates studied (Figure 2A).  (Table 3). On this basis, neither of these data sets was considered suitable for NRI  The disc diffusion zone sizes for AMOX10 were also bimodal ( Figure 1B). NRI 227 analysis of these data calculated a standard deviation of the normalised WT distribution 228 of 5.2 mm and a COWT value of ≥56 mm (Table 6) (Table 3). This modal group was 234 assumed to represent the WT isolates. NRI analysis calculated a standard deviation of the 235 log2 normalised WT distribution of 0.68 and a COWT value of ≤2 mg L -1 (Table 5).

236
The disc diffusion zone sizes for FFN30 were also unimodal ( Figure 1C). NRI 237 analysis of these data calculated a standard deviation of the normalised WT distribution 238 of 5.6 mm and a COWT value of ≥45 mm (Table 6).

239
Applying the cut-off of ≤2 mg L -1 to the MIC data and the disc zone cut-off of ≥41 240 mm to the zone data categorised 100% of the 133 isolates studied as WT ( Figure 2C). The MIC values of OXO, FLUQ and ENRO were bimodally distributed (Table 3). NRI  (Table 5). When these COWT values were applied, 21 (16%), 20 248 (15%) and 20 (15%) of the 133 isolates were categorised as WT with respect to OXO, 249 FLUQ and ENRO respectively. The disc diffusion zone sizes for OXO2 were bimodal ( Figure 1D). NRI analysis of 256 these data calculated a standard deviation of the normalised WT distribution of 8.5 mm.

257
This high standard deviation is probably a result of the fact that high zone modal group 258 was diverse and composed of only a few observations. This suggests that the disc COWT 259 value calculated by NRI analysis of ≥30 mm (Table 6) should only be treated as a 260 provisional value. Applying the cut-off of ≤0.25 mg L -1 to the MIC data for OXO and the 261 disc zone cut-off of ≥30 mm to the zone data resulted in 99% agreement in the 262 categorisation of the 133 isolates studied ( Figure 2D).

263
The disc diffusion zone sizes for FLUQ were not determined and those for ENRO 264 did not show any visually obvious high zone modal group and were not subject to NRI 265 analysis ( Figure 1E).  The distributions of the MIC values for these two potentiated sulfonamide agents were 275 diverse but appeared to be unimodal (Table 3). NRI analysis generated provisional COWT 276 values for PRI and SXT of ≤320 mg L -1 and ≤160 mg L -1 , respectively. However, the 277 standard deviations calculated for the normalized distribution of these putative WT 278 observations, 1.39 log2 mg L -1 and 1.67 log2 mg L -1 for PRI and SXT respectively, were 279 higher than those recorded for all other agents in this work (Table 5). Therefore, the 280 validity of these COWT values was questionable.

281
The disc diffusion zone sizes for SXT did not show any visually obvious high zone 282 modal group and were not subject to NRI analysis ( Figure 1F).  (Table 5) was 0.72 log2 mg L -1 . This suggests that the MIC data sets obtained in this work 295 for these agents were of an acceptable level of precision and were of sufficient quality 296 that they could be used to calculate COWT values.