Sampling
A total of 672 NTS isolates were available for analysis, 359/672 (53%) originating from humans and 313/672 (47%) from animals. Among the NTS isolates from humans, 148/359 (41%) originated from blood and 211/359 (59%) from stool. In the animal isolates, 136/313 (43%) originated from chickens, 75/313 (24%) from ducks, 65/313 (21%) from pigs, and 37/313 (12%) from rodents.
Serovar and sequence type distribution
We identified 82 unique STs, comprised of 45 different NTS serovars (Tables S2-4). Segregation of the various STs into their origin (animal, human blood, and human stool) depicted a complex structure with some STs overlapping in origin and others being more restricted (Figure 1). The most common serovars isolated from human blood were Enteritidis (ST11) and Typhimurium (ST34, 19, and 1544) accounting for 62/148 (42%) and 44/148 (30%) of human blood isolates, respectively (Table S2). From the human stool samples, 46/211 (22%) of isolates were S. Typhimurium (ST19, 34, 36, 99, 313, and 1544) and 41/211 (20%) were S. Weltevreden (ST365 and 1500) (Table S3). Weltevreden (ST1500 and 365) and Typhimurium (ST1544, 19, 34, and 36) were also the most commonly isolated serovars in animals, accounting for 43/313 (14%) and 32/313 (10%) of animal isolates, respectively (Table S4).
We next compared the distribution of NTS STs between humans and animals. From 461 NTS isolates from human blood and animals, we identified 60 different STs; 9 STs were associated with human blood only, 51 STs were associated with animals only, and 16 STs were shared between the human blood and animal isolates. From 524 NTS isolates from human stools and animals, we identified 74 STs; 23 STs were unique to human stool, 18 STs were unique to animals, and 33 STs were shared by both (Figure S1).
Both serovar and ST were significantly associated with invasiveness of human NTS isolates (p<0.001; Chi-squared test). Invasiveness index for serovars with ≥10 isolates was highest for serovars Choleraesuis (100%), Enteritidis (91%), and Typhimurium (49%). Among S. Typhimurium isolates, the most invasive STs with ≥10 isolates per category were ST34 and 19 with invasiveness indices of 58% and 40%, respectively (Table 1).
We aimed to identify the potential origin of NTS in human blood by modelling the ST data. Our model found that 56% (95% CrI 41,3-67.6) of blood infections could be attributed to human non-invasive NTS isolates; chickens were the second most important contributor (39%; CrI 26.4-51.0) (Figure 2). The results of this model were comparable when an unsampled source was incorporated (Figure S2). We performed the same analysis on the NTS isolates from human stool. Humans with invasive disease, pigs, and rodents, were all significant contributors, with overlapping ST profiles ranging from 27% to 25% (Figure 2). When an unsampled source was included in the model, this was the main contributor, accounting for approximately 55% of NTS in human stool (Figure S2).
Antimicrobial susceptibility
There was a significant association between source of NTS isolates and antimicrobial susceptibility (p<0.05, Chi-squared test). Exceptions were ceftriaxone (p=0.12, Chi-squared test) and ceftazidime (p=0.23; Chi-squared test) where resistance was <4% in each of the sources. The highest proportion of antimicrobial resistant NTS isolates was in the organisms from human blood (p<0.001; Chi-squared test). Blood isolates were resistant to a mean of 2.5/8 antimicrobials (SD=2.1), human stool isolates to a mean of 1.2/8 antimicrobials (SD=1.7), and animal isolates to a mean of 1/8 antimicrobials (SD=1.5). Consequently, NTS isolates from human blood were significantly more resistant than human stool and animal isolates (p<0.001; t-test). Ciprofloxacin resistance was notably high in human blood isolates (76/148; 51%) in comparison to human stool (17/211; 8%) and animal isolates (61/313; 19%) (p<0.001; Chi-squared test). Of the 148 isolates from human blood, 61/148 (41%) were MDR. MDR phenotypes were less common in human stool isolates (50/211; 24%) and animal isolates (47/313; 15%) than human blood isolates (p<0.001; Chi-squared test) (Table 2) (Figure S3).
After combining data from all NTS isolates, we identified 43 different antimicrobial susceptibility profiles. Comparing NTS isolates from human blood and animals, 19 profiles were unique to humans, 16 were unique to animals, and 18 were shared. Comparing NTS isolates from human stool and animals, 19 antimicrobial susceptibility profiles were shared by both species, 11 were found in humans only, and 15 in animals only (Figure S4). (Tables S5-7). Modelling the potential source of AMR in NTS from human blood and stool found that chickens and pigs were the main contributors, accounting for 37% (95% CrI 23.9-48.5) and 45% (95% CrI 30.5-58.7) of NTS cases, respectively (Figure 3). When allowing for an unsampled source, the most likely origin of AMR in NTS from human blood remained pigs (Figure S5).
Sequence type-antimicrobial susceptibility profile distribution
To investigate the potential role of zoonotic sources in the AMR-NTS phenotypes in humans, we compared the combined ST-AMR profiles of human and animal isolates. We identified 242 unique ST-AMR profiles. When comparing ST-AMR profiles by the source of isolate, 47 profiles were unique to human blood, 114 to animals, and 19 were shared. Among human stool and animal isolates, 67 ST-AMR profiles were found only in human stool only, 90 in animal stool only, and 43 in both (Figure 4) (Tables S8-10). Modelling the potential source of NTS in human blood by ST-AMR profile found that 54% (95% CrI 39.9-66.1) could be attributed to chickens and 32% (95% CrI 18.6-46.4) to humans with non-invasive disease. For NTS in human stool, the majority could be attributed to pigs (40%; 95%CrI 31.7-49.4) (Figure 5, Table S11). When allowing for an unsampled source using ST-AMR profile, modelling again identified pigs as the principal contributor for NTS in human blood and human stools (Figure S6).