A Preliminary Study on the Presence of Salmonella in Lymph Nodes of Sows at Processing Plants in the United States

Salmonella-contaminated lymph nodes (LN), when included into edible meat products, are a potential source of Salmonella foodborne disease. In this survey, ventral superficial cervical and mandibular LN were tested for the presence of Salmonella from two sow processing plants in the midwestern United States. Results indicate that both LN can be contaminated with Salmonella; mandibular LN have higher prevalence (p < 0.05) of Salmonella than cervical LN (16% vs. 0.91%), and the majority (>90%) of Salmonella isolates are pan-susceptible or resistant to one antimicrobial, while 9.78% of isolates were multi-drug-resistant (MDR-resistant to three or more classes of antimicrobials). Intervention methods to prevent foodborne disease could include elimination of these LN from pork products or inclusion of LN only into products that are destined for cooking. Integrated multi-faceted intervention methods need to be developed to reduce Salmonella in the food chain.


Introduction
The Salmonella enterica (hereafter called Salmonella) is one of the top five foodborne pathogens globally and a leading cause of foodborne illness in humans [1]. Salmonella is reported to cause over 1.3 million infections annually in the United States (U.S.) with approximately 1 million traced back to consumption of contaminated food [2]. These health concerns have prompted the beef, pork, and poultry industries to continuously search for mechanisms to reduce the risk of Salmonella contamination of their products. It has been suggested that cattle lymph nodes (LN), when included into beef ground products, could contribute to Salmonella contamination [3]. Indeed, peripheral LN of cattle are in anatomic locations that their inclusion into beef products could be somewhat common. The prevalence of Salmonella-positive subiliac LN in U.S. cattle are reported to be 7.7% to 62% [4][5][6][7], whereas prevalence of Salmonella in inguinal LN of cattle in Mexico are 54-75% [6,8]. Of increasing concern is that 8% to 20% of Salmonella isolates in those studies were resistant to antimicrobials [5,7].
It is not unreasonable to believe that swine LN, when included in edible pork products, could serve as a source of Salmonella contamination for pork. Salmonella-positive inguinal LN of sows at slaughter are reported to range from 4.8% to 37% [9][10][11] in the U.S., whereas popliteal LN in market hogs is reported to be 12.6% [12]. A study in Mexican swine abattoirs reported that subiliac LN had a Salmonella prevalence of 10% to 20% [13]. There has been limited research on prevalence of Salmonella in swine LN. Salmonella-positive LN represent a risk of contamination to pork products when incorporated into pork products, such as sausage, chorizo, and others [13,14]. The objective of this preliminary survey was to determine the presence of Salmonella in ventral superficial cervical LN and mandibular LN of sows at processing.

Sample Collection
Two sow processing plants (Plant A and Plant B) in two separate states in the midwestern U.S. were chosen for sampling. Each plant was sampled on two consecutive days each month for twelve months (November 2016-October 2017). Samples consisted of 25 ventral superficial cervical LN on each of days 1 and 2, the first Monday and Tuesday of each month for both Plants A and B (50/month X 12 months = 472 samples for Plant A and 548 samples for Plant B). In addition, from February-October 2017 (9 months), the same day each month that cervical LN was sampled, we collected 150 mandibular LN from Plant A and 217 LN from Plant B (25/day/month/plant X 9 months = 367). Due to weather and other sampling difficulties, we were unable to collect the full number of samples (both cervical and mandibular) as projected from the original study design.

LN Processing, Salmonella Cultivation
LN samples were processed as previously described for meat samples [15]. Tissues were individually weighed and recorded, surface sterilized by submersion in boiling water for 3-5 s, placed into sterile filtered-stomacher bags (Nasco, Fort Atkinson, WI, USA), macerated by pounding with a rubber mallet until flat, enriched in 80 mL of tryptic soy broth (Becton Dickinson, Sparks, MD, USA), and homogenized for 30 s with a laboratory blender (BagMixer 400VW, Interscience Laboratories Inc., Weymouth, MA, USA) at a medium speed (seven paddle strokes per second). The homogenized samples were incubated at 25 • C for 2 h, 42 • C for 12 h, and then held at 4 • C, for no more that 4-6 h, until further processing. For prevalence analysis, 1 mL from each enrichment culture was subjected to anti-Salmonella immunomagnetic separation (IMS). Each 1 mL aliquot received 20 µL of anti-Salmonella beads (Invitrogen, Carlsbad, CA, USA) and was incubated with shaking at room temperature for 15 min. The beads were extracted from the enriched samples and washed twice in PBS-Tween 20 (Sigma, St. Louis, MO, USA). The beads were transferred to 3 mL of Rappaport Vassiliadis soya (RVS; Remel Products, Lenexa, KS, USA) broth and incubated at 42 • C overnight. Salmonella present in these samples were detected by swabbing the RVS enrichment culture onto brilliant green agar (Becton Dickinson, Sparks, MD, USA) containing sulfadiazine (80 mg/L; Sigma, St. Louis, MO, USA). All plates were incubated at 37 • C for 18 to 20 h. After incubation, up to three presumptive Salmonella colonies were selected and confirmed biochemically with lysine iron and triple sugar iron agars according to the manufacturer's recommendations (BD Difco, Dickinson and Company, Sparks, MD, USA).

Statistical Analysis
Descriptive statistics comparing Salmonella prevalence across LN, plants, month, and day were tabulated in Stata (version 16, Stata Corp, College Station, TX, USA). Logistic regression (Stata version 16) was used to explore associations between Salmonella prevalence within and across LN for plant, month, and day (p < 0.05).

Results and Discussion
Plant A had 28 positive cervical LN/472 tested (5.9%), whereas Plant B had 5 positive cervical LN/548 tested (0.9%) (Figure 1). The Plant A results are similar to that reported for inguinal LN prevalence rates of 37%, 8.9%, and 4.8% in sows and 6.4% and 13% in market hogs [9][10][11], whereas Salmonella prevalence for popliteal LN was 12.55% in market hogs [12]. Similar to the present study, those studies were in the U.S., whereas a study from abattoirs in Mexico reported 10.2% and 20% Salmonella prevalence in subiliac LN [13]. The decreased Salmonella isolation rates (0.73%) of Plant B in the present study mirror those of 0% reported for superficial cervical LN [17] and 0.06% for subiliac LN [18] of market hogs in U.S. plants. Although not peripheral LN, similar Salmonella prevalence results (6.1%) were reported for mesenteric LN from breeder sows in Spain [19].

Statistical Analysis
Descriptive statistics comparing Salmonella prevalence across LN, plants, month, and day were tabulated in Stata (version 16, Stata Corp, College Station, TX, USA). Logistic regression (Stata version 16) was used to explore associations between Salmonella prevalence within and across LN for plant, month, and day (P < 0.05).

Results and Discussion
Plant A had 28 positive cervical LN/472 tested (5.9%), whereas Plant B had 5 positive cervical LN/548 tested (0.9%) (Figure 1). The Plant A results are similar to that reported for inguinal LN prevalence rates of 37%, 8.9%, and 4.8% in sows and 6.4% and 13% in market hogs [9][10][11], whereas Salmonella prevalence for popliteal LN was 12.55% in market hogs [12]. Similar to the present study, those studies were in the U.S., whereas a study from abattoirs in Mexico reported 10.2% and 20% Salmonella prevalence in subiliac LN [13]. The decreased Salmonella isolation rates (0.73%) of Plant B in the present study mirror those of 0% reported for superficial cervical LN [17] and 0.06% for subiliac LN [18] of market hogs in U.S. plants. Although not peripheral LN, similar Salmonella prevalence results (6.1%) were reported for mesenteric LN from breeder sows in Spain [19].  (Figure 1). During the collection periods, Plant A had significantly increased Salmonella-positive cervical and mandibular LN samples when compared to Plant B (p < 0.05). Additionally, there were seasonal differences in Salmonella-positive samples with August having the highest number of Salmonella isolates when compared to the other months; however, this difference was not statistically significant when analyzing both sample types together (p = 0.22 (data not shown)) or individually for cervical and mandibular lymph nodes (p = 0.18 and p = 0.77, respectively (data not shown)). For whatever reasons, the second day of samples in Plant A and Plant B had numerically increased Salmonella isolates from cervical lymph nodes when compared to the results of the first day; however, this difference was not significant (p = 0.08 (data not shown)). In both plants, the percentage of Salmonella-positive samples were significantly increased in  (Figure 1). During the collection periods, Plant A had significantly increased Salmonella-positive cervical and mandibular LN samples when compared to Plant B (p < 0.05). Additionally, there were seasonal differences in Salmonella-positive samples with August having the highest number of Salmonella isolates when compared to the other months; however, this difference was not statistically significant when analyzing both sample types together (p = 0.22 (data not shown)) or individually for cervical and mandibular lymph nodes (p = 0.18 and p = 0.77, respectively (data not shown)). For whatever reasons, the second day of samples in Plant A and Plant B had numerically increased Salmonella isolates from cervical lymph nodes when compared to the results of the first day; however, this difference was not significant (p = 0.08 (data not shown)). In both plants, the percentage of Salmonella-positive samples were significantly increased in mandibular LN when compared to cervical LN (p < 0.05). This is not a great surprise in that lymphatic tissues that are close to the oral cavity (such as tonsils and parotid LN) tend to have greater Salmonella numbers when compared to peripheral LN sources [13,20,21]. Similarly, head trim and cheek meat tissues, anatomically close to the oral cavity, can have a high percentage (65%) of Salmonella-positive samples [22,23].
The majority (52.5%) of Salmonella isolates tested were pan-susceptible (no or intermediate resistance) to 14 antimicrobials; 37.8% were resistant to one class of antimicrobials, and six isolates (9.7%) were MDR. Two of the six MDR were serotype Brandenburg, one each from serotypes Infantis and Orion_var_15+34, and two isolates from Derby (Table 1). There were four MDR isolates from cervical LN and two from mandibular LN. These results are similar to Salmonella prevalence of 6.1% in mediastinal LN from sows at slaughter in Spain [19]. In that study, serotypes of isolates were: Typhimurium (43.7%), Derby (18.7%), Enteritidis (12.5%), and Montevideo (12.5%) with 9/16 of isolates showing multi-resistance to three or more antimicrobial classes (SSuT, ACSSut, and ASSut-Nx-Cfx).
Although not LN tissue, the low levels of MDR in the present study contrast with a similarly designed study in which 63% of Salmonella isolates of cheek meat (anatomically close to mandibular LN) of market hogs were MDR [23].

Conclusions
On the basis of the results of this survey, it is evident that cervical and mandibular LN can be colonized by Salmonella and inclusion of these LN into pork products may increase the risk those products could be contaminated with Salmonella. Intervention methods to prevent foodborne disease could include elimination of these LN from pork products or inclusion of LN only into products that are destined for cooking. Additional research is needed to further understand the distribution and prevalence of Salmonella in the lymph nodes of commercial sows at slaughter. There is a high priority for the meat industry to utilize a multi-faceted approach to develop effective intervention methods to reduce Salmonella in the food chain.
Author Contributions: R.B.H. provided conceptualization, collection of samples, supervision of laboratory isolation and analysis, coordination with industry partners, and writing, editing, submission, funding, revisions, and re-submission of manuscript; K.N.N. provided conceptualization, methodology, analysis, writing-review and editing; R.C.A. provided visualization, data curation, and writing-original draft preparation; D.J.N. provided funding, supervision, and project administration. "A preliminary study on the presence of Salmonella in lymph nodes of sows at processing plants in the United States". All authors have read and agreed to the published version of the manuscript.