Materials and Methods

Animals—Sixty-one steers were used in a prospective cohort study. Steers were allocated to 6 pens (5 pens contained 10 steers each, and 1 pen contained 11 steers). All procedures used were reviewed and approved by the Amarillo-Area Cooperative Research, Education, and Extension Triangle Animal Care and Use Committee (protocol No. 2006-02).

Four cohorts were included in the study. Cohort 1 comprised 10 steers (5 steers in each of 2 pens); these steers were administered CCFA^a as a single-dose regimen (6.6 mg/kg, SC, on day 0). Cohort 2 comprised 10 steers (5 steers in each of 2 pens) administered CCFA as a two-thirds–dose regimen (4.4 mg/kg, SC, on day 0). Cohort 3 comprised 10 steers (5 steers in each of 2 pens) administered CCFA as a 3-dose regimen (6.6 mg/kg, SC, on days 0, 6, and 13). Cohort 4 comprised 31 steers (5 steers in each of 5 pens and 6 steers in the remaining pen) that served as untreated control animals.

Procedures—Commercially prepared CCFA was procured and administered in accordance with label instructions, except that a reduced dosage (4.4 mg/kg) was administered to steers of cohort 2. The CCFA for each treatment was administered SC in the middle third of the caudal aspect of the ear. No products (CCFA, other injectable antimicrobial, or control solution) were administered to control steers during the study.

Steers were moved through an animal handling facility on days 0, 2, 6, 9, 13, 16, 20, and 28. Fecal samples were collected directly from the rectum of each steer, placed in labeled cups, and stored on wet ice. All samples were transported to the laboratory for additional processing on the day of sample collection.

Microbiologic analysis—A direct-plating method was used to determine the number of CFUs of non–typespecific E coli per gram of feces. In brief, a 25-g portion of fecal matter was aseptically transferred into 225 mL of buffer peptone water. The mixture was shaken for 1 minute. To enumerate E coli, serial dilutions were spirally plated^b onto MacConkey agar plates. Plates were incubated at 37°C for 48 hours, and typical colonies were then counted by use of an automated colony counter.^c

Three separate colonies were selected from each plate with distinct colonies for further characterization. Because biochemical confirmation offered little additional advantage over simply selecting typical colonies from MacConkey agar plates, only a proportion of the isolates were subjected to automated biochemical confirmation^d to verify that they were E coli.

Susceptibility testing was performed on all isolates by use of microbroth dilution with a panel of antimicrobials^e (Appendix).^16,17 Three distinct colonies/sample were transferred onto tryptic soy agar plates and incubated at 37.3°C for 24 hours. Three to five clearly isolated colonies were subsequently selected from each tryptic soy agar plate and placed into 4 mL of sterile deionized water; these solutions were then adjusted to a 0.5 McFarland standard. A portion (10 μL) of the suspension was transferred into Mueller-Hinton broth and mixed on a vortex. An automated inoculator^f was used to inoculate 96-well microbroth dilution plates. Plates were incubated at 37.3°C for 18 to 24 hours; results were then manually determined.

The MIC was reported as the lowest concentration of antimicrobial that inhibited visible growth, as determined in accordance with the manufacturer's instructions (Appendix). When an isolate grew at the greatest concentration of an antimicrobial, the MIC was arbitrarily set at double the greatest concentration. Isolates were classified as susceptible or resistant on the basis of guidelines established by the Clinical Laboratory Standards Institute.¹⁶ No breakpoint has been established for ceftiofur by the Clinical Laboratory Standards Institute. As described by the National Antimicrobial Resistance Monitoring System for Enteric Bacteria,¹⁷ ≥ 8 μg of ceftiofur/mL was used to determine phenotypic resistance. Additionally, isolates that grew at concentrations ≥ 2 μg of ceftriaxone/mL were characterized as having reduced susceptibility.

Statistical analysis—Data were entered into an electronic database. The MICs were logarithmically transformed by use of log base 2, whereas estimates of the number of E coli per gram of feces were logarithmically transformed by use of log base 10. Descriptive statistics were generated. Estimates of the concentration of ceftiofur-resistant bacteria in fecal samples were calculated by multiplying the proportion of resistant isolates within a sample by the concentration (untransformed value) of E coli within the sample. Estimates were then logarithmically transformed by use of log base 10. Data were imported into a commercially available software package^g for subsequent analysis.

Outcome measures (ie, log₂ MIC, log₁₀ CFUs of E coli/g of feces, and log₁₀ CFUs of resistant E coli/g of feces), proportion of resistant isolates, and proportion of susceptible isolates were analyzed by use of linear or logistic regression techniques, when appropriate. An interaction between administration status and time was evaluated. When significant (P ≤ 0.05) values were detected, only pairwise evaluations of within-day and within-administration means were performed. Dependency within an animal over time was modeled by use of first-order autoregressive covariance matrices.¹⁸

Results

We recovered 1,441 isolates. Confirmatory biochemical assays were performed on 953 (66.1%) isolates, and 952 (99.9%) isolates were confirmed as E coli. Resistance to 1 or more antimicrobials was detected in 986 (68.4%) isolates. The most common antimicrobials to which E coli isolates were resistant were sulfisoxazole (901 [62.5%] isolates), tetracycline (586 [40.7%]), streptomycin (475 [33.0%]), chloramphenicol (335 [23.3%]), and ampicillin (302 [21.0%]; Table 1). Four hundred fifty-five (31.6%) isolates were susceptible to all drugs. The distribution of MICs was determined (Table 2).

On the basis of the number of antimicrobials to which isolates were resistant, a bimodal distribution in the proportion of isolates was evident (Figure 1). The 2 modes were attributable to bacteria typically being resistant to no antimicrobials or to 8 antimicrobials. The bimodal distribution was most pronounced for those cattle administered 3 doses of CCFA at 6.6 mg/kg and least pronounced for the control steers.

View Full Size

Proportion of Escherichia coli isolates, on the basis of the number of antimicrobials to which they were resistant, recovered from fecal samples collected from control steers (A); steers administered 6.6 mg of CCFA/kg, SC, once on day 0 (B); steers administered 4.4 mg of CCFA/kg, SC, once on day 0 (C); or steers administered 6.6 mg of CCFA/kg, SC, on days 0, 6, and 13 (D).

Citation: American Journal of Veterinary Research 68, 5; 10.2460/ajvr.68.5.501

• Download Figure
• Download figure as PowerPoint slide

Phenotypic expression of ceftiofur resistance was observed in 289 (20.1%) isolates. The mean number of antimicrobials to which ceftiofur-resistant isolates were resistant was 8.2, whereas the typical number of antimicrobial drugs to which ceftiofur-susceptible isolates were resistant was 1.1. Agreement was almost perfect between ceftiofur resistance and coresistance to ACSSuT (K, 0.97). Similarly, nearly perfect agreement (K, 0.97) was detected between ceftiofur resistance and an MIC of ≥ 2 μg of ceftriaxone/mL.

A significant (P = 0.01) interaction was detected between exposure status and day for the likelihood of recovering a ceftiofur-resistant isolate (Figure 2). The likelihood of recovering a ceftiofur-resistant isolate did not differ significantly (P = 0.20) between cohorts on day 0 or 28. Significant (P = 0.01) differences among cohorts were detected on remaining days. In addition, a significant (P = 0.01) effect of day was detected for each of those cohorts administered ceftiofur; however, a similar effect of day was not detected (P = 0.68) for the control cohort. Recovery of ceftiofur-resistant isolates was significantly (P = 0.01 for all within-day pairwise comparisons) more likely on days 2 through 20 from steers administered 3 doses of CCFA at 6.6 mg/kg, compared with recovery for control steers. A ceftiofur-resistant isolate was significantly (P = 0.03 for all within-day pairwise comparisons) more likely to be recovered on days 2, 6, 9, and 16 from steers administered 1 dose of CCFA at 6.6 or 4.4 mg/kg, compared with recovery for control steers. On day 13, only isolates recovered from steers administered either 3 doses of CCFA at 6.6 mg/kg (P = 0.01) or 1 dose of CCFA at 4.4 mg/kg (P = 0.02) were more likely to be resistant to ceftiofur than isolates recovered from control animals. With few exceptions, similar patterns were observed for coresistance to ACSSuT, an MIC of ≥ 2 μg of ceftriaxone/mL, and log₂ MIC of β-lactam antimicrobials (data not shown).

View Full Size

Proportion of isolates resistant to ceftiofur that were recovered from fecal samples collected on various days from control steers (black circles), steers administered 6.6 mg of CCFA/kg, SC, once on day 0 (white circles); steers administered 4.4 mg of CCFA/kg, SC, once on day 0 (inverted black triangles); or steers administered 6.6 mg of CCFA/kg, SC, on days 0, 6, and 13 (inverted white triangles). Error bars represent the SE of the proportion.

Citation: American Journal of Veterinary Research 68, 5; 10.2460/ajvr.68.5.501

• Download Figure
• Download figure as PowerPoint slide

Exposure was significantly (P = 0.01) associated with a time-dependent impact on E coli concentration in feces. The E coli concentration in all cohorts administered CCFA was significantly lower on days 2 and 6 (P = 0.02 for all within-day pairwise comparisons), compared with the concentration in control steers (Figure 3). On day 9, only cohorts administered 3 doses of CCFA at 6.6 mg/kg (P = 0.01) or 1 dose of CCFA at 4.4 mg/kg (P = 0.01) had lower E coli concentrations than control steers. We did not detect significant (P = 0.15) within-day variation on days 0, 13, 16, 20, or 28.

View Full Size

Total E coli concentration (A) and estimated concentration of ceftiofur-resistant E coli (B) in fecal samples collected from various cohorts of feedlot steers on various days. Notice that the scale on the y-axis differs between the 2 portions of this figure. Error bars represent the SE of the geometric mean. See Figure 2 for remainder of key.

Citation: American Journal of Veterinary Research 68, 5; 10.2460/ajvr.68.5.501

• Download Figure
• Download figure as PowerPoint slide

Administration of CCFA was associated with a significant (P = 0.01) time-dependent increase in the estimated concentration of ceftiofur-resistant E coli (Figure 3). Estimated concentration of ceftiofur-resistant E coli varied significantly (P = 0.02) on days 6, 9, 13, 16, and 20. Cohorts administered CCFA had greater numbers of ceftiofur-resistant E coli per gram of feces, compared with results for the control steers. The most pronounced (extent and duration) effect was evident in the cohort administered 3 consecutive doses of CCFA at 6.6 mg/kg; on days 13, 16, and 20, only the ceftiofur-resistant population of the 3-dose cohort was significantly (P = 0.02) greater than that of the control cohort.

Discussion

Subcutaneous administration of CCFA exerted a detectable selection pressure within the gastrointestinal tract that favored transient expansion of ceftiofur resistance among non–type-specific E coli. The decrease in population susceptibility to ceftiofur after administration of CCFA generally returned to preadministration values approximately 13 to 15 days after CCFA administration. Because prevalence and within-animal genetic diversity of Salmonella serovars is considerably lower than that of E coli, we used E coli as a model for the effect of ceftiofur administration on Salmonella serovars because the study described here had insufficient power to evaluate an effect of CCFA administration on Salmonella serovars. Despite the expected similarity in response of Salmonella serovars and E coli, a degree of caution should be used when generalizing our results beyond E coli.

Administration of CCFA also transiently facilitated expansion of isolates that were resistant to multiple antimicrobials. This phenomenon resulted because those isolates that phenotypically had ceftiofur resistance also had coresistance (or had reduced susceptibility) to other sometimes unrelated antimicrobials. Furthermore, ceftiofur-susceptible isolates were typically only resistant to approximately 1 antimicrobial, compared with ceftiofur-resistant isolates, which were typically resistant to approximately 7 other antimicrobials. This clustering of resistance patterns as broadly susceptible or broadly resistant was responsible for the bimodal distribution in number of antimicrobials to which the isolates were resistant; hence, the distribution was most pronounced in isolates recovered from steers administered 3 consecutive doses of CCFA (Figure 2). Because ceftiofur resistance was so intimately associated with resistance to other antimicrobials and administration of CCFA facilitated expansion of ceftiofur resistance, administration of CCFA led to the more pronounced bimodal distribution observed in isolates recovered from steers of the CCFA-administered cohorts.

Agreement was nearly perfect between ceftiofur resistance and coresistance to ACSSuT, and reduced susceptibility to ceftriaxone was evident in the data. Although not explicitly stated, a similar phenotype cluster was observed in ceftiofur-resistant isolates reported elsewhere.¹⁵ The ACSSuT coresistance phenotype has been described for Salmonella Typhimurium DT104¹⁹ and multiply drug-resistant Salmonella Newport that produces an AmpC β-lactamase (designated S Newport MDR-AmpC).^20,21 In the former, the integron complex that houses the genetic determinants responsible for coresistance to ACSSuT is chromosomally integrated,²² whereas in the latter, it is a plasmid-mediated event.²³ The consequence of this is that genetic elements are more likely to be disseminated horizontally when housed on a plasmid than when housed on a bacterial chromosome.²⁴ Additional studies are needed to determine whether the detected time-dependent increase in resistance was attributable to clonal expansion of specific isolates, such as for Salmonella Typhimurium DT104, or to nonspecific dissemination of resistance determinants housed on mobile genetic elements, such as for a conjugative plasmid. If the latter is true, then similar (or identical) plasmids harbored within multiple unrelated E coli genotypes would be expected.

The most likely determinant conferring ceftiofur resistance as well as reduced susceptibility to ceftriax-one is the bla_CMY gene.^8,25 It is not certain in the study reported here how, or whether at all, this gene may be linked to those conferring the coresistance ACSSuT pattern. It is possible that they are all housed on a single mobilizable or conjugative plasmid. Additional molecular characterization of the isolates from our study is required to better describe genetic determinants, transferability of resistance, and similarity of isolates.

Ostensibly, it would appear that more isolates recovered from those animals administered 4.4 mg/kg were resistant than were recovered from animals administered a single 6.6 mg/kg regimen (Figure 3). However, when the proportion of ceftiofur-resistant isolates recovered on days 0, 2, and 6 was compared between the 4.4 mg/kg cohort and cohorts 1 and 3, no difference was observed. Consequently, it is unlikely that a causal CCFA dose effect was evident in our study. In other words, detected variation between cohorts 1 and 2 was more likely attributable to within-pen factors than to varying selection pressures.

A herd-level association between ceftiofur use and ceftiofur resistance has been reported in another study.¹⁵ In that study, investigators did not detect an animal-level association between treatment and recovery of ceftiofur-resistant E coli. It is not completely clear as to the reasons we detected an animal-level response and those investigators did not. However, in that study,¹⁵ the investigators included ceftiofur treatments administered within the preceding 6 months in their analyses. Because the within-animal effect is relatively transitory, it may be difficult to detect an animal-level effect beyond a few days in many cases. The detection of a herd-level effect is interesting and may represent a treatment-dependent shift in baseline over time.

Even though CCFA provided selective pressure favoring expansion of multiply resistant E coli, the return to baseline values within approximately 2 weeks after completion of administration of CCFA indicated reduced bacterial fitness associated with maintenance of resistance determinants. Moreover, we did not detect an effect of day within the control cohort, despite the fact that they were commingled in pens with steers administered CCFA. Thus, the ceftiofursusceptible variants (those broadly susceptible) appeared to have the greatest fitness for the feedlot conditions in the study reported here when there was no ceftiofur selection pressure. Consequently, ongoing CCFA administration may be a prerequisite for dissemination of ceftiofur resistance in non–type-specific E coli. Therefore, prudent use of CCFA in animal health programs may prevent E coli from becoming an important reservoir of genetic determinants that encode ceftiofur resistance. Additional studies are required to determine whether an apparent fitness disadvantage is also evident among Salmonella serovars and whether it is possible to exploit differing metabolic fitness to mitigate resistance in situations where the burden of ceftiofur resistance has increased to become problematic within defined cattle populations.