Overview of the Antimicrobial Resistance of Salmonella Recovered from U.S. Poultry Processing Plants

The objectives of this study were to analyze the results of antimicrobial susceptibility testing on Salmonella isolated from poultry carcass and parts rinsates using a scoring system for antimicrobial resistance (AMR) and to deter-mine whether the resistance of Salmonella to selected antimicrobials critically or highly important to human medicine changed from 2017 to 2019. Samples were collected from 26 plants in the United States, analyzed for the presence of Salmonella, and tested for susceptibility to 12 antimicrobials (n=734 for 8 antimicrobials; n=597 for 4 antimicrobials). The multidrug resistance (MDR) scores and AMR scores remained the same over time (P>0.05); however, MDR and AMR differed (P<0.0001) by serogroup and serogroup-by-year interactions. Most notably, MDR—and AMR for 7 out of the 12 antimicrobials—was greater (P<0.05) in serogroup C1 than other serogroups and/or lower (P<0.05) in serogroup D1 than other serogroups. The effect year-by-serogroup was also significant for MDR (P<0.0001) and—for 8 out of the 12 antimicrobials—AMR (P<0.05); differences (P<0.05) across years were identified in serogroup C1, B, and C2 but were highly variable. Resistance to ciprofloxacin and ceftriaxone,“highest priority critically important antimicrobials”to human medicine, were not different (P>0.05) across years, but there were significant (P<0.05) serogroup and serogroup-by-year effects for ceftriaxone resistance. Interestingly, gentamicin resistance across years differed (P<0.05) in serogroup B and C. Overall, mean Salmonella MDR and AMR scores were stable from year to year, but shifts in AMR in Salmonella serogroups across years were identified, emphasizing the need to continue monitoring AMR in Salmonella isolated from poultry products in the interest of food safety and human health.


Introduction
One of the major pathogens causing foodborne illness in the United States is Salmonella (NARMS, 2019a). It is estimated that over 1.2 million illnesses are caused by nontyphoidal S. enterica yearly (Scallan et al., 2011), resulting in 23,000 hospitalizations and 450 deaths (NARMS, 2018). Cases of salmonellosis in humans are often attributed to the consumption of Salmonella-contaminated foodstuffs. While Salmonella has been isolated from a variety of foodstuffs-including, but not limited to, fruits, vegetables, and animal products-contaminated poultry products are one of the major vehicles of Salmonella transmission to humans (Parveen et al., 2007).
Furthermore, there is growing concern over the prevalence of not just antimicrobial-resistant (AMR) but also multidrug-resistant (MDR) strains of Salmonella isolated from poultry products and the potential transmission and impact of those strains on human health (Parveen et al., 2007;Berrang et al., 2009;Shah et al., 2017). Severe cases of salmonellosis are more frequently associated with MDR strains of Salmonella than susceptible strains (Parveen et al., 2007) and are often fatal for the young, elderly, and immunocompromised (NARMS, 2018). The Centers for Disease Control and Prevention estimates that drug-resistant nontyphoidal Salmonella causes 212,500 infections and 70 deaths annually (CDC, 2019). Two classes of antimicrobials-quinolones/fluoroquinolones (e.g., ciprofloxacin [CIP]) and cephalosporins (e.g., ceftriaxone [CRO])-are commonly used to treat severe Salmonella infections in adults and children, respectively (NARMS, 2018). Thus, Salmonella isolates resistant to antimicrobials used to treat severe Salmonella infections pose a risk to human health and reduce the number of therapeutic options available to treat Salmonella infections (Parveen et al., 2007).
Due to the human health risk associated with AMR pathogens, the antimicrobial susceptibility of foodborne pathogens, like Salmonella, are monitored in ill persons by the Centers for Disease Control and Prevention, in retail meats by the US Food and Drug Administration, and in food-producing animals by the US Department of Agriculture as part of the National Antimicrobial Resistance Monitoring System (NARMS). Additionally, to protect food safety and human health, Pilgrim's Pride Corporation (hereafter referred to as Pilgrim's) internally monitors the resistance of Salmonella isolated from poultry to selected medically important antimicrobials recognized as highest priority critically important, high priority critically important, or highly important to human medicine by the World Health Organization (Table 1). The objectives of this study were to analyze the results of antimicrobial susceptibility testing on Salmonella isolated from poultry carcass and parts rinsates using a scoring system for AMR and determine whether the resistance of Salmonella to selected antimicrobials critically or highly important to human medicine changed over time.

Carcass and parts sampling for Salmonella analysis
Salmonella spp.-positive samples used for antimicrobial susceptibility testing were obtained from routine carcass and parts rinsate sampling conducted to meet the Modernization of Poultry Slaughter Inspection standards at all 26 of Pilgrim's plants in the US and Puerto Rico. Briefly, one prechill and postchill carcass rinse sampling pair was collected per 22,000 birds slaughtered at a plant. Of those, one sampling pair per shift per day was randomly selected and analyzed for Salmonella spp. The sampling pair selected for Salmonella spp. analysis consisted of a sample taken from one prechill zone and one postchill zone from the same flock. All plants followed the same zone categorization along the production line. One parts rinse sample was collected each processing day and analyzed for Salmonella. Rinsate samples were collected following the US Department of Agriculture carcass and parts rinsing procedures (FSIS, 2013). Briefly, samples were collected aseptically by rinsing a whole carcass or approximately 4 lb of parts with 400 mL of buffered peptone water for 1 min. Then, 100 mL of rinsate was poured into a sterile specimen cup. Samples were packaged in a cooler with ice packs and shipped to Pilgrim's laboratories for arrival and testing within 2 d of collection.

Salmonella analysis
Salmonella spp. analysis of carcass and parts rinsates was conducted by Pilgrim's laboratories in Athens, Georgia; Broadway, Virginia; and Pittsburg, Texas. Carcass and parts rinsates (30 mL) were tested

Salmonella antimicrobial susceptibility testing
The antimicrobial susceptibility of Salmonella spp. isolates was tested from July 12, 2017, through December 5, 2019. From July 2017 to mid-September 2017, the panel of antimicrobials used for antimicrobial susceptibility testing (Panel 1) included trimethoprim (TMP), gentamicin (GEN), amoxicillinclavulanic acid (AMC), streptomycin (STR), tetracycline (TET), CIP, ampicillin (AMP), chloramphenicol (CHL), erythromycin (ERY), penicillin (PEN), rifampin (RIF), and vancomycin (VAN). From late September 2017 onward, ERY, PEN, RIF, and VAN were replaced with cefoxitin (FOX), CRO, sulfisoxazole (SXZ), and meropenem (MEM) as recommended by the JBS Food Safety Advisory Board. The recommendation was made based on the fact that gramnegative bacteria are intrinsically resistant to those antimicrobials and the replacement of those antimicrobials allowed for the monitoring of more antimicrobials that are highest priority critically important, high priority critically important, and highly important to human medicine. Thus, the antimicrobials TMP, GEN, AMC, STR, TET, CIP, AMP, CHL, FOX, CRO, SXZ, and MEM were included on the latter panel (Panel 2). The susceptibility of Salmonella isolates (n = 137 analyzed with Panel 1; n = 597 analyzed with Panel 2) to selected antimicrobials was determined using the HardyDisk Antimicrobials Sensitivity Test (Hardy Diagnostics, Santa Maria, CA). Standardized methods for agar diffusion testing described by the manufacturer for Enterobacteriaceae were followed. The following antimicrobial disks were used: The results were categorized as susceptible, intermediate, or resistant using zone diameter (in millimeters) interpretive standards (Table 2) provided by the manufacturer for each antimicrobial, derived from the Clinical and Laboratory Standards Institute (CLSI, 2019).

Statistical methods
Statistical analyses were conducted using JMP version 14.3.0 (SAS Institute Inc., Cary, NC). A significance level of P < 0.05 was used for all statistical tests. Assumptions necessary for the results of the statistical analyses to be valid were assessed and found to be met.
Prior to statistical analysis, the antimicrobial sensitivity test results were transformed using a scoring system adapted and developed from Moore et al. (2013) and Ewing et al. (2017). The antimicrobial sensitivity test results of susceptible, intermediate, or resistant were assigned a numerical value of 0, 0.5, or 1, respectively. The values represent Salmonella isolate AMR scores for each antimicrobial on a panel (n = 12). An MDR score for each Salmonella isolate was calculated by adding the 12 AMR scores together for each sample, with a minimum MDR score of 0 and a potential maximum MDR score of 12 (maximum MDR in this Serogroup Ai-Vi was excluded from the analysis due to a lack of data in 2017. A two-way analysis of variance was conducted to assess the overall effects of year, serogroup, and year by serogroup on the means of Salmonella MDR scores (samples analyzed with Panel 2 antimicrobials only, n = 597) and AMR scores (n = 734, antimicrobials present on both panels; n = 597, only Panel 2). A Student t test was used to test pairwise comparisons of year, serogroup, and year-byserogroup combination means.

MDR scores
Least-square estimates of the means (with standard errors) for MDR scores are listed in Table 3. Mean MDR scores were not different across years (P = 0.8140); however, the effects serogroup and year by serogroup on mean MDR scores were significant (both P < 0.0001). Figure 1 illustrates differences in mean MDR scores by serogroup. Most notably, serogroup C1 had the greatest mean MDR score, whereas serogroup D1 had the lowest mean MDR score (P < 0.05; 3.27 ± 0.21 vs. 0.63 ± 0.13). Figure 2    2019 (P > 0.05; 1.59 ± 0.13 vs. 1.55 ± 0.15). Furthermore, there were no significant year-by-serogroup interactions in group D1 or group E. Contrary to these data, the most recent NARMS integrated summary reported a substantial increase (9.5% to 18%) in MDR Salmonella recovered from routinely sampled chickens during 2015 to 2017 (NARMS, 2019a). Additionally, the report stated that the increase in the percentage of MDR Salmonella over time was attributed to a rise in MDR Salmonella Infantis isolates. This is interesting because Salmonella Infantis belongs to serogroup C1 (Grimont and Weill, 2007), and in the data presented here, serogroup C1 had the greatest MDR score compared to other serogroups, and MDR nearly doubled from 2017 to 2018. However, MDR Salmonella did not change over time because MDR in serogroup B and C2 decreased over time, averaging out the rise in serogroup C1 MDR ( Figure 2). Table 4 lists least-square estimates of the means (with standards errors) for the AMR scores of 12 antimicrobials critical to human medicine (WHO, 2018; Table 1). AMR scores were not significantly different across years for all antimicrobials tested, with the exception of CHL (P = 0.0037). CHL mean AMR score increased from 2017 to 2018 (P < 0.05; 0.03 ± 0.02 vs. 0.12 ± 0.02) but remained the same from 2018 to 2019 (P > 0.05; 0.12 ± 0.02 vs. 0.08 ±.0.03). In agreement with these data, the "NARMS Now: Integrated Data" show an increase in the percentage of carcass rinsate Salmonella isolates resistant to CHL from 2016 to 2017 (2.9% vs. 8.2%) (NARMS, 2019b). In general, Salmonella resistance to CHL is still low but should continue to be monitored since CHL is a highly important antimicrobial in human medicine (WHO, 2018). It is also worth noting that CHL is not approved for use in food-producing animals in the US (eCFR, 2020).
Of    critically important antimicrobial used to treat severe Salmonella infections in humans (NARMS, 2018). Serogroup C1 isolates were also more resistant to the critically important antimicrobials GEN (an aminoglycoside) and AMP (a beta-lactam), as well as the highly important antimicrobials TMP (a sulfonamide) and CHL (an amphenicol). Several Salmonella isolates belonging to serogroup C1 are commonly isolated from poultry, some of which include Salmonella Infantis, Salmonella Thompson, Salmonella Montevideo, and Salmonella Mbandaka. In comparison, previous reports indicate that Salmonella Infantis is pansusceptible, Salmonella Thompson is pan-susceptible, Salmonella Montevideo has some resistance to sulfonamides, and Salmonella Mbandaka has some resistance to sulfonamides and aminoglycosides    Meat and Muscle Biology 2020, 4(2): 1, 1-11 Waldbusser et al. Salmonella antimicrobial resistance in broilers (Shah et al., 2017 Liljebjelke et al. (2017) for Salmonella Enteritidis, a commonly isolated D1 serogroup in poultry (Shah et al., 2017), in which STR resistance was 3.6% and TET resistance was 0% (Liljebjelke et al., 2017).
The majority of significant differences (P < 0.05) in serogroup C1 mean AMR scores across years followed 2 trends: (1) mean AMR scores in serogroup C1 increased from 2017 to 2018 and decreased from 2018 to 2019 (AMP, CRO, and CHL), and (2) mean AMR scores in serogroup C1 increased from 2017 to 2018 and remained the same from 2018 to 2019 (GEN, SXZ, TET, and TMP). STR showed a different trend as mean AMR scores in serogroup C1 increased across years and were significantly (P = 0.0001) different between 2017 and 2019.
Of the AMR scores in serogroup C1 that increased from 2017 to 2018 and decreased from 2018 to 2019, the changes in Salmonella CRO resistance are of interest because CRO is a highest priority critically important antimicrobial in human medicine, belonging to the cephalosporin class of antimicrobials used to treat severe Salmonella infections in humans (NARMS, 2018). While mean CRO resistance in serogroup C1 increased from 2017 to 2018 (P < 0.05; 0.14 ± 0.08 vs. 0.48 ± 0.04), mean CRO resistance decreased from 2018 to 2019 (P < 0.05; 0.48 ± 0.04 vs. 0.28 ± 0.04). The NARMS integrated summary reported an increase in the percentage of carcass rinsate Salmonella isolates resistant to CRO from 2015 to 2017 (6.5% vs. 9.3%), but the magnitude of CRO resistance reported by NARMS was less than that reported here (NARMS, 2019a).
For the AMR scores in serogroup C1 that increased from 2017 to 2018 and remained the same from 2018 to 2019, changes in the mean AMR score of GEN over time (years) are of particular interest since Pilgrim's discontinued the in ovo use of GEN in hatcheries on January 1, 2017. The serogroup C1 mean AMR score for GEN increased from 2017 to 2018 (P < 0.05; 0.15 ± 0.05 vs. 0.41 ± 0.05) and remained similar from 2018 to 2019 (P > 0.05; 0.41 ± 0.05 vs. 0.34 ± 0.05). Although not a direct year-to-year comparison, NARMS (2019b) reported the percent resistance of Salmonella Infantis (serogroup C1) to GEN increased from 2015 to 2016 (1.7% vs. 8.9%) and decreased in 2017 (5.3%), while the percent resistance of Salmonella Montevideo (serogroup C1) to GEN decreased from 2015 to 2017 (4.8% vs. 0%). Notably, the GEN resistance of serogroup C1 Salmonella in 2017, 2018, and 2019 (0.15, 0.41, and 0.34, respectively) reported here is greater than that reported by NARMS in 2015, 2016, and 2017 for Salmonella Infantis (1.7%, 8.9%, and 5.3%, respectively) and Salmonella Montevideo (4.8%, 0%, and 0%, respectively). Furthermore, despite discontinuing the in ovo use of GEN, it appears that resistance to GEN still persists in serotype C1 Salmonella isolated from carcass and parts rinsates at Pilgrim's.

Conclusions
The results presented in this paper demonstrate how a scoring system for AMR can be used to monitor AMR in Salmonella isolated from poultry products and assess how changes in production practices affect AMR. The MDR and AMR scores of Salmonella isolated from carcass and parts rinses at Pilgrim's processing plants remained the same from 2017 to 2019; however, MDR and AMR scores differed by serogroup and serogroup-by-year interactions. Most notably, MDR scores-and AMR scores for 7 out of the 12 antimicrobials tested-were greater in serogroup C1 than other serogroups and/or lower in serogroup D1 than other serogroups. The effect of year by serogroup was also significant for MDR scores, and for the AMR scores of 8 out of the 12 antimicrobials tested. Significant differences in both MDR and AMR scores across years were identified in serogroup C1, B, and C2 but were highly variable.
Of particular interest was the AMR of CIP and CRO because of the classification of these antimicrobials as highest priority critically important antimicrobials to human medicine and the use of these antimicrobials in the treatment of severe Salmonella infections in humans. The results indicated that Salmonella isolated from carcass and parts rinsates were susceptible to CIP across the years and serogroups evaluated. In contrast, CRO resistance did not differ by year but was greater in serogroup C1 Salmonella isolates, and it increased from 2017 to 2018 and decreased from 2018 to 2019. Salmonella resistance to GEN was also of interest since the in ovo use of GEN was discontinued on January 1, 2017. GEN resistance in serogroup B was not different from 2017 to 2018 but decreased from 2018 to 2019, while GEN resistance in serogroup C1 increased from 2017 to 2018 and remained similar from 2018 to 2019, highlighting the complexity of AMR. Overall, mean Salmonella MDR and AMR scores were stable from year to year, but shifts in AMR in Salmonella serogroups across years were identified and emphasize the need to continue monitoring AMR in Salmonella isolated from poultry products in the interest of food safety and human health. Furthermore, the scoring system used for AMR in Salmonella is a useful and effective method for monitoring changes in AMR.