Agreement between the categorization of isolates of Aeromonas salmonicida and Yersinia ruckeri by disc diffusion and MIC tests performed at 22 ℃

Standard disc diffusion and MIC test procedure were used to investigate the susceptibility of two hundred and fifty- one isolates collected from infected fish in France to florfenicol, oxolinic acid and tetracycline. The tests were performed at 22 ± 2 ℃ and for the 177 Yersinia rucker i they were read after 24– 28 hr incubation and for the 74 Aeromonas salmonicida isolates they were read after 44– 48 hr. Applying epidemio logical cut- off values to the susceptibility data generated in these tests, the isolates were categorized as wild- type or non- wild- type. The agent- specific categories into each isolate were placed on the basis of the data generated by the two methods were in agreement in 98% of the determinations made. It is argued that, with respect to categorising isolates, disc diffusion and MIC methods can be considered as equally valid at this temperature and after both periods of incubation.

categorized as WT. Isolates that show a susceptibility outside this range which possess some mechanism that confers reduced susceptibility are categorized as NWT (Silley, 2012).
With respect to their relative performance characteristics, the original paper of Ericsson and Sherris (1971) argued that the precision of disc diffusion test data was slightly superior to that of MIC data. In a study of the influence of incubation temperature on the precision of susceptibility tests, Smith et al. (2018) also suggested that, when the data were obtained at 35℃, the precision of disc diffusion data was equal to or superior to that of MIC tests. However, they noted that the precision of disc diffusion data, but not those of MIC data, decreased when the temperature of incubation was decreased, and the incubation time was increased. They argued that putative epidemiological cut-off values calculated from data with low precision should not be applied or, at the very least, should be applied with extreme caution. They further argued that the extent of the loss of precision at lower incubation temperatures was such that for organisms requiring incubation at temperatures <22℃ for >48 hr the disc diffusion method could not be recommended.
This work was, therefore, undertaken to investigate the relative performance of disc diffusion and MIC methods when they were performed at 22 ± 2℃ and incubated for either 24-28 or 44-48 hr. In this work, the relative performance of disc diffusion and MIC methods was investigated by calculating the extent to which the categorization of isolates as WT or NWT, made from the data they generated, were in agreement with each other. These categorizations were generated by the application of epidemiological cut-off values to disc diffusion and MIC data for the susceptibility to florfenicol (FLO), oxolinic acid (OXO) and tetracycline (TET) obtained for isolates of Yersinia ruckeri and Aeromonas salmonicida collected from French laboratories. These susceptibilities were determined at 22 ± 2℃ by the standard CLSI disc diffusion and MIC tests (CLSI, 2020a), and incubated for 24-28 and 44-48 hr. However, CLSI have not yet developed epidemiological cut-off values (ECVs) for any agents against Y. ruckeri or for TET against A. salmonicida. The guideline VET03Ed3 (CLSI, 2020b) gives ECVs for both inhibition zone and MIC data for FLO and OXO against A. salmonicida generated in tests at 22℃ with incubation for 44-48 hr. However, the guideline M23Ed5 (CLSI, 2018) recommends that data from at least three laboratories are needed to set such ECVs and those for A. salmonicida MIC data were estimated using data from only a single laboratory and those for inhibition data from only two laboratories. Therefore, in this work the epidemiological cut-off values (CO WT ) applied to the data sets for the French isolates studied in this work were calculated from those data themselves aggregated with additional published data sets that had been obtained using the same CLSI protocols.

| Bacterial isolates
The 74 Aeromonas salmonicida isolates and 177 Yersinia ruckeri isolates studied in this work were collected as part of the DIAMIC project, supported by the French EcoAntibio Plan. They were contributed from eleven veterinarians and veterinary laboratories working in France. The criterion for inclusion was that the isolates had been obtained during clinical investigations of diseased fish in France during the period 1985-2020.
The Mueller-Hinton agar (MHA) used in this work was obtained from Bio-Rad, the cation-adjusted Mueller-Hinton broth (CAMHB), and Brain Heart Infusion broth (BHI) were obtained from Thermo Fisher. Each isolate received in the laboratory was first growth on MHA at 22°C to check purity and stored in duplicate at −70°C in BHI broth supplemented with 20% of glycerol.
Confirmation of the identification of the presumptive A. salmonicida isolates was performed by Maldi-Tof and the PCR method of Byers et al. (2002). Only isolates, which were identified A. salmonicida by Maldi-Tof with score higher than 2.0 and gave positive result with PCR, were included in this study.
Confirmation of the classification of the presumptive Y. ruckeri was also performed by Maldi-Tof. Only isolates, which were identified as Y. ruckeri with a score higher than 2.3, were included in this study.

| Susceptibility testing
Using both disc diffusion and MIC methods, the antimicrobial susceptibilities of the French isolates of both species were tested in a single laboratory (Anses MBA) against three agents FLO, OXO and TET. These agents are used in aquaculture in France and susceptibility to them are routinely investigated in French diagnostic laboratories. As recommended by Smith (2019), the abbreviations for these agents were generated by applying the EUCAST rules (www.eucast. org/ast_of_bacte ria/guida ncedo cumen ts/).
For the diffusion tests, FLO (30 µg) discs were obtained from Mast. The OXO (2 µg) and the TET (30 µg) discs were obtained from Bio-Rad. The microdilution MIC tests were performed using 96-well plates made in the ANSES laboratory using a Biomek automated liquid handler (Beckman Coulter). The concentration ranges in these plates were 0.003-64 mg/L for FLO, 0.004-16 mg/L for OXO and 0.015-128 mg/L for TET.
The susceptibility testing protocols used in this work were those specified for non-fastidious organisms in the CLSI guideline VET03Ed2 (CLSI, 2020a). The MIC tests were performed using CAMHB and disc diffusion tests were performed using MHA with the results being recorded after either 24-28 or 44-48 hr incubation at 22° ± 2℃. Reading of inhibition zone size was performed using an automatic reader of discs diffusion susceptibility tests "Sirscan micro" (I2A Diagnostic) including a system of image capture. The reading of each agar plate controlled by the same technician who performed all analysis. Following the standard practice of French diagnostic laboratories, the Y. ruckeri tests were read after incubation for 24-28 hr and the A. salmonicida tests after 44-48 hr.
For both disc diffusion and MIC tests, the control quality (QC) reference strains Escherichia coli ATCC25922 and Aeromonas salmonicida ATCC33658 were tested using the same test protocols as were used for the tests of the Y. ruckeri and A. salmonicida isolates.
The acceptable ranges applied were those given for FLO and OXO in the CLSI supplement VET04Ed3 (CLSI, 2020b). This supplement does not however provide acceptable ranges for TET for these reference strains using either the disc or MIC tests at 22°C. Because of this absence of the necessary quality control requirements, the susceptibility tests with this agent, although were performed using the testing protocols specified in VET03Ed2 (CLSI, 2020a), cannot be considered as having been performed according to the full CLSI method.

| Data sets for calculating epidemiological cutoff values
The epidemiological cut-off values (CO WT ) applied to the data sets for the French isolates generated in this work were calculated from those data themselves aggregated, when they were available, with additional data sets that had also been obtained at 22℃. These data sets had been generated using the CLSI protocols provided in VET03-A (CLSI, 2006). However, with respect to the non-fastidious bacteria studied in this work the protocols pro- For Y. ruckeri, additional data sets accessed were those for the MIC values for 83 isolates with respect to FLO and TET (Huang et al., 2014).
For A. salmonicida, the additional data sets accessed were the MIC and disc diffusion data for 217 isolates with respect to FLO and OXO (Miller & Reimschuessel, 2006) and the data for disc diffusion 107 isolates with respect to FLO  and OXO . All these data sets had been used to establish the ECV published in VET04Ed3 (CLSI, 2020b).

| Calculation of epidemiological cut-off values
Epidemiological cut-off values (CO WT ) were calculated using the automatic Excel spread sheets for normalized resistance interpretation analysis (NRI) (Kronvall, 2010;Kronvall & Smith, 2016) (www.biosc and.se/nri/). Isolates were categorized as wild-type (WT) or non-wild-type (NWT) by application of these cut-off values to the disc zone and MIC values determined in this work.
The acronyms used to refer to epidemiological cut-off values followed the recommendations of Smith (2019). The acronyms ECV was reserved for cut-off values published by CLSI and for all other epidemiological cut-off values the acronym CO WT was used.

| Categorical agreement
For each agent/species combination, the categorization of each isolate as WT or NWT on the basis of disc diffusion data was compared with the categorization of the same isolate based on the MIC data.
The percentage categorical agreement (CA) was calculated as the number of isolates which were placed in the same category by both disc and MIC methods as a percentage of the total number of isolates tested.

| RE SULTS AND D ISCUSS I ON
The zones of inhibition generated by FLO (30 µg) and OXO  (Table S1 and Table S2). For all twelve data sets analysed, the standard deviations (SD) of the distributions of normalized WT observations, calculated by NRI analysis, were within the limits suggested by Smith et al. (2018).

| NRI analysis of the Y. ruckeri isolates
Scatter plots of the paired disc zones and MIC values for the Y. ruckeri isolates are shown for each of the three agents in Figure 1. The frequency of isolates categorized as NWT phenotype and the agreement between the categorization based on disc zone data and MIC values are shown in Table 2.
Analysis of the FLO disc data obtained in this work generated a Analysis of the TET data obtained in this work generated a CO WT of ≥26 mm and application of this cut-off value categorized one French isolate as NWT. The MIC data obtained in this work aggregated with those of Huang et al. (2014) generated a CO WT of ≤1 mg/L which categorized all the French isolates as WT. With respect to TET, the CA between the disc and MIC methods was 99%.
Combining the Y. ruckeri data for all three agents, the total number of paired comparisons of the categorizations of isolates generated by the disc and MIC methods was 531. The categorizations were in agreement 528 (99%) of these pairs.

| NRI analysis of the A. salmonicida isolates
Scatter plots of the paired disc zones and MIC values for the A. salmonicida isolates are shown for each of the three agents in Figure 2.
The frequency of isolates categorized as NWT phenotype and the agreement between the categorization based on disc zone data and MIC values are shown in Table 2.
Analysis of the FLO disc data obtained in this work aggregated with those published by Miller and Reimschuessel (2006)

| Frequency of reduced susceptibility
In the isolates examined in this work, the frequencies of NWT phenotypes with respect to OXO were high in both species. Using the TA B L E 2 Summary of the categorization 177 Yersinia ruckeri and 74 Aeromonas salmonicida isolates as WT or NWT by applying the CO WT generated in this work (Table 1)  It should be noted, however, that the isolates studied in this work were collected from those held by eleven diagnostic laboratories operating in France. Smith et al., (2013) suggested that it is reasonable to assume that outbreaks which are not effectively controlled by chemotherapy with the first agent chosen will be investigated in more detail than those which were effectively controlled. Moreover, the diagnostic laboratories might have been more prone to keeping isolates with abnormal susceptibility profiles in their historical collection of strains. Thus, there is potential for isolates with reduced susceptibility among being over-represented in those collected by frontline diagnostic laboratories. As a result of the potential for this bias in the isolates studied in this work, the frequencies of NWT phenotypes reported here cannot be taken as providing an accurate picture of the frequency of reduced susceptibility in France as a whole.

| CON CLUS IONS
In the study of bacteria isolated from aquatic animals, the primary aim of either zone size or MIC data is to categorize isolates as either WT, fully susceptible members of their species, or as NWT isolates, that possess some mechanism that confers reduced susceptibility (Silley, 2012). In this work, the agreement between the categorization of isolates using disc diffusion data and that using MIC data both generated at 22°C according to the testing protocols specified by CLSI (2020a) was investigated for 735 sets of paired determinations. The overall categorical agreement between the two methods was 98%. The work reported here on susceptibility tests performed at 22°C suggests that disc diffusion or MIC methods were equally valid and found no reasons why one should be preferred over the other.

ACK N OWLED G EM ENTS
The assistance of the veterinarians involved in this project and who

CO N FLI C T O F I NTE R E S T
There were no conflicts of interest.

E TH I C S A PPROVA L , PATI E NT CO N S E NT A N D CLI N I C A L TR I A L R EG I S TR ATI O N S TATE M E NT
The work reported did not require ethical approval or patient consent and did not involve any clinical trials.

PE R M I SS I O N TO R E PRO D U CE M ATE R I A L FRO M OTH E R S O U RCE S
All material from other sources was available and accessed from published work that was in the public domain.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing not applicable to this article as no data sets were generated or analysed during the current study.