Evaluation of Selected Antibiotic Residue Screening

SAVE OFFLINE

Evaluation of Selected Antibiotic Residue Screening Tests for Milk from Individual Goats A. CONTRERAS,* M. J. PAAPE,† A. L. DI CARLO,† R. H. MILLER,† and P. RAINARD‡ *Enfermedades Infecciosas, Facultad de Veterinaria, Universidad de Murcia, 30071 Murcia, Spain †Livestock and Poultry Sciences Institute, ARS, USDA, Beltsville, MD 20705 ‡Laboratoire de Pathologie Infectieuse et Immunologie, Insitut National de la Recherches Agronomique, Centre de Recherches de Tours, 37380 Nouzilly, France

ABSTRACT

INTRODUCTION

Because somatic cell counts (SCC) of caprine milk are higher than SCC of bovine milk, the performance of antibiotic residue tests for screening bovine milk was investigated for caprine milk. Eighty-five does that were free of antibiotic usage for at least 30 d and that were free of clinical mastitis were sampled at three milkings during a 37-d period. At each sampling, foremilk was collected for bacteriological analysis, and composite bucket milk samples were collected for antibiotic testing and SCC. Day of lactation, parity, 305-d mature equivalent milk yield, and SCC averaged 221 d (57 to 577 d), 2.3 lactations (one to nine lactations), 1160 kg (623 to 1750 kg), and 2.2 × 106/ml (0.3 to 30.7 × 106/ml), respectively. The mean Dairy Herd Improvement Association test day milk yield for the month of sample collection was 3 kg (1.4 to 6.4 kg). Intramammary infections were present in 54% of the goats and in 36% of the udder halves. Assays included positive ( 5 and 10 ppb of penicillin-G and 50 ppb of ceftiofur) and negative controls that had been prepared in caprine milk and controls supplied by the manufacturers. One false-negative outcome and one false-positive outcome were recorded. For one sampling day, a positive linear relationship existed between SCC and the results of one test, and a quadratic relationship existed between SCC and the results of another test. The antibiotic residue screening tests for milk from individual goats adequately identified milk that was free of antibiotic. These tests are therefore recommended for use with caprine milk. ( Key words: dairy goats, antibiotic residue, screening tests, somatic cell count)

Screening tests for antibiotic residues are currently used to detect antibiotic contamination of bulk tank milk and milk from individual cows. Reports indicated a high degree of false-positive outcomes when these kits were used with milk from clinically normal cows (3, 4, 19). A test was considered to be a false positive when no antibiotics were present in the milk or when antibiotics were present but were below the tolerance level. A false-positive outcome results in discarded milk that would be legal to sell and represents a loss of revenue to the producer. Several factors contribute to false-positive outcomes, including milk SCC, BSA, lactoferrin, lysozyme, and other products of inflammation (2, 7, 19). Cullor ( 3 ) was among the first to focus national attention on the errors that were inherent in several of the antibiotic residue tests that were being used to screen bovine milk. Based on his investigations ( 3 ) , residue kits that had low selectivity rates ( a high number of falsepositive outcomes) were removed from the market, and other residue kits were modified by manufacturers to minimize false positives. As a result, recent studies (1, 17) have reported fewer false-positive outcomes. In a cooperative study (including the FDA and the USDA) on composite milk from 130 clinically normal lactating cows that were free of antibiotic usage for at least 30 d, no false positives were recorded (S. A. Andrews, R. A. Frobish, N. Alderson, and M. J. Paape; 1994; unpublished data). Somatic cell counts ranged from 0.03 to 6.0 × 106/ml. Kits evaluated were Charm II sequential (Charm Sciences, Inc., Malden, MA), Charm I/cowside II (Charm Sciences, Inc.), Charm Bacillus stearothermophilus disk assay ( BSDA; Charm Sciences, Inc.), Delvotest-P (Gist-Brocades, Menomonee Falls, WI), Delvotest-SP (Gist-Brocades), SNAP (IDEXX, Westbrook, ME), LacTek CEF (Idetek, Sunnyvale, CA), LacTek B-L (Idetek), and Penzyme (Smith Kline Beecham, West Chester, PA).

Abbreviation key: BSDA = Bacillus stearothermophilus disk assay, CAEV = caprine arthritisencephalitis virus.

Received March 18, 1996. Accepted September 30, 1996. 1997 J Dairy Sci 80:1113–1118

1113

1114

CONTRERAS ET AL.

There are no published evaluations of antibiotic residue screening tests for caprine milk. Caprine milk has a higher SCC and a larger percentage of neutrophils than does normal bovine milk (6, 15). Although gross composition of caprine milk is similar to bovine milk, the composition of the casein and whey protein fractions in caprine milk is different, and caprine milk has a higher percentage of free fatty acids (8, 14). Also, milk fat globules in caprine milk are smaller than those in bovine milk (14). Because milk secretion is apocrine in goats but merocrine in cows, caprine milk contains many cytoplasmic particles that are not found in bovine milk ( 6 ) . These differences could affect the outcome of antibiotic residue kits that were designed for use with bovine milk. The objective of this study was to evaluate the occurrence of false-positive outcomes in caprine milk using the rapid screening tests that have previously been shown to have a low frequency of false-positive outcomes in bovine milk. MATERIALS AND METHODS

collected by individual buckets. Milk was transferred into a pail, and a 50-ml sample was obtained to test for antibiotics and to determine SCC. Milk samples were maintained at 4°C and transported to Beltsville, Maryland. On the same day, the 50-ml samples were divided into aliquots into individual tubes, refrigerated at 4°C, and tested for antibiotic residue and SCC the following morning. Milk SCC Milk SCC were determined by an electronic cell counter (Fossomatic model 90; Foss Electric, Hillerød, Denmark). Samples were heated to 60°C for 15 min because Miller et al. ( 9 ) previously determined that fresh samples needed to be heated to 60°C to damage somatic cells in milk, making the cell membranes more permeable to the fluorescent dye, ethidium bromide. Samples were then maintained at 40°C until counted. The cell counter was calibrated monthly with somatic cell standards of bovine milk (Dairy Quality Control Institute Services, Inc., Mounds View, MN).

Goats

Bacteriology

Experiments were conducted using 85 lactating does in a Grade A dairy goat herd (138 lactating does) located in southern Maryland. The herd was enrolled in the DHI program, and the dairy maintained herd reproduction and health records. Breeds included Alpine ( n = 64), Saanen ( n = 17), and Nubian ( n = 4 ) goats. Age of the does, days of lactation, parity, and milk yield are shown in Table 1. All does were clinically normal at sampling. Based on herd health records, does had not received any drugs for 30 d prior to milk sampling. Infection of the caprine arthritis-encephalitis virus ( CAEV) was prevalent in 94% of the does (Animal Health Diagnostic Laboratory of Maryland, College Park).

In addition to collection on the day of sampling, aseptic samples of foremilk were collected monthly for 2 consecutive mo prior to the month of sampling. Milk (20 ml ) was spread onto the surface of a trypticase soy agar plate containing 0.1% esculin and 5% washed sheep erythrocytes (Remel, Lenexa, KS). Plates were incubated aerobically at 37°C and examined after 24 and 48 h. Infection was defined as growth of two or more identical colonies. Bacterial groups were identified according to the recommendations of the National Mastitis Council (11). Specific identification of staphylococci was made using commercial micromethods (API  STAPH; BioMerieux Vitek, Inc., Hazelwood, MO).

Milk Sampling

Antibiotic Residue Screening Test Kits

Milk samples were collected at the p.m. milking (1600 h ) during three samplings over a period of 37 d until all 85 does were sampled. No more than 30 does were sampled per day. Before sampling, teats were sprayed with Oxy Gard teat dip (Klenzade Inc., St. Paul, MN) and dried with individual paper towels. Teat ends were scrubbed with cotton pledgets that had been soaked in 70% ethyl alcohol. The first two or three streams of milk were discarded, and the next two streams were collected aseptically for diagnostic bacteriology. Does were machine-milked; milk was

Each refrigerated milk sample was analyzed for antibiotic residues by the screening tests shown in Table 2. These screening tests represented a variety of analytical principles for assaying antibiotic residues of b-lactam and ceftiofur (LacTek CEF) in milk. The tests were conducted according to the recommendations of the manufacturers. Four individuals conducted the tests, each of whom was trained by a company representative. For each test kit, the positive and negative control samples provided by the manufacturer were run immediately

Journal of Dairy Science Vol. 80, No. 6, 1997

1115

PERFORMANCE OF SCREENING TESTS IN CAPRINE MILK TABLE 1. Age, day of lactation, parity, milk yield, milk SCC ( ×106/ml), and bacteriology of milk from udder halves of does that were sampled to evaluate antibiotic residue screening tests. Variable

Goats1

X

SD

Minimum

Maximum

Age, yr Day of lactation Parity Total milk yield, kg 305-d Milk yield, kg SCC Infected halves Infected goats

(no.) 81 82 82 82 82 85 170 85

2.9 221 2.3 780 1160 2.2 62 46

1.8 107 1.8 408 277 5.1

1 57 1 109 623 0.3

9 577 9 1768 1750 30.7

1Number

of goats studied.

before assaying the individual caprine milk samples. Control samples were prepared using bulk tank caprine milk obtained at the time of sampling. In addition to these controls, positive controls using penicillin-G (Phoenix Pharmaceutical, Inc., St. Joseph, MO) were prepared to a final concentration of 5 and 10 ppb, and ceftiofur (Upjohn Co., Kalamazoo, MI) was prepared to a final concentration of 50 ppb. Penicillin-G and negative control samples were prepared fresh daily using the bulk tank caprine milk obtained on the day of sampling. Idetek prepared the ceftiofur controls in caprine milk and shipped the controls frozen to Beltsville, Maryland, where they were stored at –70°C. These positive and negative controls were included in the appropriate assays together with the individual caprine milk samples. The possible outcomes for the Penzyme farm test and the Delvotest-P and Delvotest-SP tests were negative, positive, and caution. A sample with a caution outcome was considered to be positive for the purposes of this study. There were two detection systems in the SNAP test: a visual color change and an instrument reader. The present study included both methods of detection. The analysis of the Charm BSDA was made on unheated and heated (82°C for 2 min) milk samples. Heat treatment inactivates en-

zymes and proteins, such as lactoferrin, and causes the liberation of chlorine and volatile natural inhibitors that may be present in the milk, which otherwise could contribute to false-positive outcomes. Statistical Analysis Because the goats used in this study had not received drugs for 30 d prior to milk sampling, all positive outcomes for tests on composite milk from individual goats were considered to be false-positive outcomes rather than false-violative outcomes (20). A false-violative outcome occurs when milk contains a residue below the FDA violative level. All negative outcomes for tests on the caprine milk samples containing 5 and 10 ppb of penicillin-G and 50 ppb of ceftiofur were considered to be false-negative outcomes. For each screening test, the selectivity rate (defined as the rate of truly negative samples that were found negative by the assay) was calculated. The selectivity rate was calculated as the number of truly negative assays divided by the total number of samples analyzed. The selectivity rate can be converted to the false-positive rate by subtracting the selectivity rate from 1.0. Analyses were conducted to determine whether milk SCC of a sample influenced

TABLE 2. Antibiotic residue screening tests and their analytical principles. Test kit

Manufacturer

Analytical principle

Charm II sequential and Charm I/ cowside II Charm BSDA1 Delvotest-P and Delvotest-SP SNAP LacTek CEF and LacTek B-L Penzyme

Charm Sciences, Inc. (Malden, MA)

Competitive binding; 14C penicillin displacement Bacterial growth inhibition Bacterial growth inhibition Antibiotic-antigen capture system Competitive enzyme system Enzymatic

1Bacillus

Charm Sciences, Inc. Gist-Brocades (Menomonee Falls, WI) IDEXX (Westbrook, ME) Idetek (Sunnyvale, CA) Smith Kline Beecham (West Chester, PA)

stearothermophilus disk assay. Journal of Dairy Science Vol. 80, No. 6, 1997

1116

CONTRERAS ET AL.

the results of the test kit assays. Only test kits that provided numerical results could be used (SNAP , Charm II sequential, Charm I/cowside II, LacTek CEF, and LacTek B-L). Data from each kit were used to fit linear and quadratic regression equations on the natural log of milk SCC. Equations were plotted to determine the nature of the relationship when regressions were significantly different from zero. RESULTS The goats used in this study represented a wide range in age, day of lactation, parity, and milk yield (Table 1). The mean DHIA test day milk yield was 3 kg (1.4 to 6.4 kg) for the month that the samples were collected. Of the 170 caprine udder halves cultured for diagnostic bacteriology, 62 had persistent IMI over the 3 mo of sampling (Table 1). Staphylococcus spp. were the most frequently isolated (97%) and consisted of Staphylococcus epidermidis (67.8%), Staphylococcus xylosus (9.6%), Staphylococcus simulans (6.5%), Staphylococcus hyicus (4.8%), Staphylococcus caprae (3.2%), Staphylococcus hominis (1.6%), Staphylococcus lugdunensis (1.6%), and Staphylococcus spp. (4.9%). Other isolated pathogens were Gramnegative bacilli, Corynebacterium spp., Streptococcus spp., and Bacillus spp. The geometric mean SCC for the bucket milk samples was 2.2 × 106/ml (Table 1). The SCC for uninfected and infected goats averaged 2.0 and 2.5 × 106/ml ( P > 0.05). A large percentage (82.4%) of the goats had SCC > 1.0 × 106/ml, the legal limit for bulk tank caprine milk in the US. All positive and negative control standards supplied by the manufacturers tested appropriately immediately before the individual caprine milk samples were tested. These results confirmed that the screening assays were working properly and could differentiate between truly negative and truly positive samples. For the penicillin-G and ceftiofur standards that

Figure 1. Regression of the results ( n = 85) of the SNAP antibiotic residue screening test (IDEXX, Westbrook, ME) on the natural log of SCC (LSCC). Y (predicted value) = 0.5917 + 0.03517x. For this test, a ratio of ≥1 was considered to be a positive outcome. X = Actual value; ÿ = predicted value.

were run along with the milk samples, LacTek B-L failed to detect one of the 5-ppb penicillin-G samples. Of the 935 samples tested, only one false-positive outcome was recorded for Penzyme, and that sample was in the caution area (Table 3). All test kits except Penzyme had selectivity rates of 1.0 (no false positives). Penzyme also had a high selectivity rate of 0.99. For one of the sampling days, a positive linear relationship ( P < 0.05) existed between SCC and the SNAP results (Figure 1), and a quadratic relationship ( P < 0.01) existed between SCC and the Charm I/cowside II test results ( P < 0.05) (Figure 2). No significant relationships with SCC existed for the SNAP or Charm I/cowside II test results for the other two sampling days or with results from any of the other test kits. The R2 values for SNAP and Charm I/cowside II were 17.1 and 17.6%. DISCUSSION

TABLE 3. Number of samples, percentages, and cumulative percentages of milk SCC ( ×106/ml) for the composite of caprine milk samples that were analyzed. SCC

n

Percentage

Cumulative percentage

3.0

5 10 13 17 8 9 23

5.9 11.8 15.3 20.0 9.4 10.6 27.1

5.9 17.6 32.9 52.9 62.4 72.9 100.0

Journal of Dairy Science Vol. 80, No. 6, 1997

The results from this study demonstrated high selectivity rates for the various residue kits that were designed for use with bovine milk but were tested on milk from individual goats. The composite milk samples varied widely in SCC, and the goats used varied widely in age, day of lactation, parity, and milk yield. These results are in contrast to the reported (17, 19, 21) association of false-positive outcomes with increased concentrations of SCC in bovine milk. The only association that we observed was for the SNAP

PERFORMANCE OF SCREENING TESTS IN CAPRINE MILK

Figure 2. Regression of the results ( n = 85) of the Charm I/ cowside II antibiotic residue screening test (Charm Sciences, Inc., Malden, MA) on the natural log of SCC (LSCC). Y (predicted value) = 680.7234 – 128.4915x + 7.7158x2. Based on the set points established for this test, a result was considered positive if the total counts of 14C accumulated for 2 min were ≤105. X = Actual value; ÿ = predicted value.

test. With that test kit, one of the three sampling dates yielded a positive linear relationship between SCC and test score. For the Charm I/cowside II test kit, the relationship was quadratic. Test scores increased as SCC increased; however, no false positives were obtained with either the SNAP or the Charm I/ cowside II test kits. The significant R2 obtained with results of either test kit indicated that the R2 was associated with a significant regression on SCC. Given the high SCC for this goat herd, finding only one false-positive outcome (Penzyme) in 935 tests (85 goats × 11 assays) was surprising. One would have anticipated that many of the products of inflammation that had been reported (2, 7, 19) to cause false-positive outcomes in bovine milk would have been present in a majority of the caprine milk samples used in this study. For the Charm BSDA test, because no false-positive outcomes were obtained on unheated milk, the presence of any volatile natural inhibitors in milk did not contribute to false-positive outcomes. However, because of the heavy use of sanitizing agents in the milking parlor, there is always the possibility of chlorine contamination of milk. Thus, in order to avoid false-positive outcomes, milk should be heated to liberate chlorine. Our results for the Delvotest-P, Charm I/cowside II, Charm II sequential, Charm BSDA, LacTek B-L, and Penzyme test kits agreed with the findings of Andrews et al. ( 1 ) , who observed selectivity rates of 0.97, 1.0,

1117

0.97, 1.0, 0.98, and 0.91, respectively, for these tests in bovine milk. The results are also in agreement with a recently completed study by the Center for Veterinary Medicine, FDA, and the USDA (S. A. Andrews, R. A. Frobish, N. Alderson, and M. J. Paape, 1994, unpublished data). That study evaluated the same antibiotic residue test kits that were used in the present study with composite milk from individual cows. No false positives were observed. The results of the present study and those unpublished findings indicate that false-positive outcomes are virtually nonexistent when these residue tests are performed on milk from either goats or cows according to the recommendations of the manufacturer. The high SCC in this herd of goats was not surprising in view of the recent report by Droke et al. ( 5 ) . In that study ( 5 ) , 65.5% of the bulk tank SCC from 71 commercial dairy goat herds located throughout the US had SCC >1.0 × 106. A number of factors have been reported to contribute to the high SCC for caprine milk, including day of lactation, parity, infection, and CAEV infection (6, 15, 16). In the present herd, the high incidence of infection from CAEV (94% of the goats), high IMI rate (36% of the udder halves), and extended days of lactation could have contributed to the high SCC. The SCC in milk from cows that are free from IMI are not affected by day of lactation and parity, which is in contrast to results with goats (12). Furthermore, the composition of the somatic cells in milk differs between goats and cows. For animals that are free of IMI, neutrophils constitute 5 to 20% of the somatic cells in bovine milk and 45 to 74% of the somatic cells in caprine milk (6, 10), which suggests that neutrophil migration into caprine milk is faster than that in bovine milk and may contribute to a naturally higher SCC. Interestingly, for this herd of 138 lactating does, the yearly incidence of clinical cases of mastitis over the last 10 yr averaged only 0.5% or