SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001 29

DRUGS, COSMETICS, FORENSIC SCIENCES

Charm Safe-Level b-Lactam Test for , , Ceftiofur, Cephapirin, and G in Raw Commingled Milk

ROBERT S. SALTER,DAVID LEGG,NANCY OSSANNA,CHERYL BOYER,JOAN SCHEEMAKER,ROBERT MARKOVSKY, and STEVEN J. SAUL Charm Sciences, Inc., 659 Andover St, Lawrence, MA 01843-1032

The Charm Safe-Level b-Lactam Test was evalu- the NCIMS may be used for tanker testing (2, 3). The Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 ated by a U.S. Food and Drug Administration (FDA) AOAC–Research Institute (AOAC–RI) was chosen to admin- test protocol administered by the AOAC–Research ister the third party-verification program following the August Institute. The sensitivity and selectivity of the test 23, 1994, FDA Center for Veterinary Medicine (CVM) proto- were evaluated with >800 negative raw commin- col: Outline of the CVM Data Requirements for Milk gled and drug-fortified milk samples by the manu- Screening Tests Labeled for Testing Milk at the Bulk facturer and an independent laboratory. Probit Tank/Tanker Truck for Drug Residues: Drug Class: analysis by the independent laboratory determined β-lactams. The procedure calls for the test kit manufacturer the following 90% positive levels with 95% confi- (Charm Sciences, Inc.) to submit concentration-response data dence: amoxicillin, 5.6 ppb; ampicillin, 8.5 ppb; for review followed by verification by an independent labora- cephapirin, 13.7 ppb; ceftiofur, 46.2 ppb; and peni- tory. The Dairy Quality Control Institute (DQCI) was chosen cillin G, 3.6 ppb. These values were within a range as the independent laboratory. Because the Charm Safe-Level of ±20% of the manufacturer’s data. Selection of β-Lactam Test used a bacterial receptor with sensitivity ad- negative samples met confidence specifications. justed on the basis of SL/T rather than the observed biological Ruggedness parameters were studied and defined, activity (4, 5), the levels detected were closer than those de- and the stability of frozen milk was verified. There tected by other receptor methods to the regulated SL/T (6). were no interferences from somatic cells The SL design, ROSA (Rapid One-Step Assay), was developed (1 000 000 somatic cell count/mL) or bacteria in response to the NCIMS-recognized problem of nonviolative (300 000 colony-forming units/mL), or from positives that resulted from hypersensitive tests (7, 8). The 27 other non-b-lactam animal drugs. Test perfor- evaluation criteria and performance specifications for the mance with raw milk samples containing incurred Charm SL β-Lactam Test that led to its approval are presented penicillin, ampicillin, and amoxicillin was consis- in this paper. tent with the dose responses determined with forti- fied milk samples. Incurred in raw METHOD milk samples was detected at lower levels than was cephalosporin in fortified milk samples, presumably Materials because of the presence of metabolite, as verified by other test methods. Quality control data support (a) Standard Reference Materials.—Amoxicillin lot I, consistency in manufacture between batches and ampicillin (anhydrous) lot J, cephapirin sodium lot I, and po- the stability of refrigerated test reagents for up to 1 tassium penicillin G lot I reference standards were obtained year. Successful fulfillment of these criteria led to from the U.S. Pharmacopeial Convention, Inc. (Rockville, FDA certification of the test when used with a reader MD). Ceftiofur sodium lots 69BDJ and 67BTM were ob- in U.S. milk testing programs. tained from Pharmacia & Upjohn Co. (Kalamazoo, MI). Stock buffers listed in the United States Pharmacopeia (USP; 9) were used for the preparation of β-lactam standards as fol- -Lactam screening of milk tankers is required under lows: penicillin G and cephapirin, USP buffer No. 1 (B1); βU.S. milk regulations developed by the National Con- ampicillin, water for 1 mg/mL stock solution and USP buffer ference for Interstate Milk Shipments (NCIMS) and en- No. 3 (B3) for subsequent dilutions; amoxicillin, B3–0.025N forced by the U.S. Food and Drug Administration (FDA; 1). NaOH (2 + 1) for 1 mg/mL stock solution and B3 for subse- Only β-lactam screening tests that have been verified by a quent dilutions; and ceftiofur, water. Stock solutions at third party to detect at least 4 of 6 target β-lactams at U.S. safe 1 mg/mL were stable at 4°C for the following number of days: levels/tolerances (SL/T) and that have been recommended by penicillin G, 4 days; ceftiofur, 4 days; ampicillin, 7 days; and cephapirin, 3 days; amoxicillin required daily preparation (9). Subsequent dilutions were made daily by using class A volu- Received December 1, 1999. Accepted by JM June 23, 2000. metric glassware. 30 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001

(b) Bacterial isolates from milk.—Gram-positive organ- rehydrated with milk that tested negative and then allowed to isms (G+; Streptococcus spp., Staphylococcus aureus, stand for 10 min before use. For storage, controls were refrig- Enterococcus faecalis), gram-negative organisms (G–; Pseu- erated for 48 h or frozen at –20°C for ≤2 months. Frozen con- domonas aeruginosa, Enterobacter aerogenes, Proteus trols were thawed in lukewarm water before use and centri- mirabilis, and Klebsiella spp.), and G+/G– were provided by fuged for 3 min at 1200–2000 ´ g before testing. Thawed FDA’s Laboratory Quality Assurance Branch (FDA–LQAB). controls were refrigerated and used within 24 h, and then dis- Freshly grown cultures (equal mixtures) were prepared and carded. diluted into milk at targeted concentrations of 150 000 and Apparatus 300 000 colony-forming units (cfu)/mL. (c) Matrix.—Fresh commingled raw bovine milk obtained (a) Test strip incubator.—Manufactured by Charm Sci- from a local dairy tank and/or silo and found to be ences, Inc.; operated at 56 ± 1°C (temperature verified by heat β-lactam-free by the Charm II β-Lactam Test–Quantitative strip or thermometer) and equipped with optional timing

Assay (10; unpublished pasteurized milk study, negative milk beeper and/or light. Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 verification, FDA, Lab Quality Assurance Branch, FDA Cen- (b) Pipet.—300 ± 15 µL with tips. ter for Food Safety, Summit Argo, IL) was used for testing. (c) Timer.—Necessary if not contained in incubator. The age of the milk from the time of cow milking could be as (d) ROSA reader.—Model SL Reader or Luminator great as 5 days. Imager, or equivalent; for date, time, operator, test identity, (d) Reagents.—SL β-Lactam Test strips were stored in line reading, and interpretation; required by FDA to conduct desiccated packages and refrigerated at 0–4.4°C. Raw milks NCIMS bulk milk tank/tanker screening programs. β determined as negative by the SL -Lactam Test were used as Charm SL b-Lactam Test Procedure (Figure 1) a negative control. Penicillin G (5 ppb) internal control stan- dards (Charm Sciences, Inc.) were prepared from freeze-dried (1) Check that incubator temperature is 56 ± 1°C (55°C preparations of USP stock material. Control standards were temperature strip indicator will be green) and that incubator is

Figure 1. ROSA (Rapid One-Step Assay) technology uses a lateral flow test strip and specific receptors to detect analytes in milk samples. (1) Strip is placed in a 56°C incubator, and the plastic tape is lifted. (2) Milk (300 mL) is slowly added to the strip, the plastic tape is resealed, and the incubator lid is closed. (3) After 8 min, the strip is removed. (4) Development of the lines is inspected to see the solid even development of the C-line, and then the strip is inserted in a reader to determine whether the test result is negative or positive. SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001 31 on level surface. Mix milk sample. Write sample ID on white Experimental Design label on test strip. Place test strip in incubator, flat side facing up. Reshape any dented sample compartments to allow test Sensitivity and selectivity.—The manufacturer’s initial data strip to fit into incubator. While holding strip flat, peel back submission included data for 2 or more raw milks spiked at 1/5, tape of SL test strip to edge of white label, exposing sample 2/5, 3/5, and 4/5 of the SL/T, and at the SL/T for each of the compartment well and pad. β-lactam drugs claimed. Thirty replicates of each drug and con- (2) Pipet 300 ± 15 µL milk into the side of the sample centration were blind-coded with 60 negative samples (total compartment well. Pipet slowly so that milk does not over- 810 samples = [5 β-lactams] × [5 concentrations per flow the well. Reseal tape over sample pad. Repeat steps 1 β-lactam] × [30 replicates] + 60 negatives). Samples were and 2 for up to 4 samples. Close cover on incubator and tested by 4 technicians for ≥2 days. Each sample was analyzed tighten latch. once. Reader and visual results were sent to reviewers. (3) Incubate for ≥8 min, but ≤10 min. Some incubators FDA–CVM determined test qualification according to the

have a beeper and/or solid red LED light, which switches to a protocol. Concentrations for verification of the submitted data Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 flashing yellow LED light when 8 min has elapsed. by the independent laboratory and for determination of the (4) Remove strips from incubator. Hold each strip verti- minimal 90/95% detection levels (90% positive levels with cally so sample compartment is down. Avoid squeezing sam- 95% confidence) were calculated according to the Probit and ple compartment. Visually verify that the C(control)-line is Gompertz Models used by FDA. Five concentrations, 4 defin- solid and complete. ing ROSA reader dose response (Robert Condon, RJC Asso- ciates, Inc., Myersville, MD, unpublished data, 1998) and one Determination of Results defining the SL/T (2), for each β-lactam drug were Visual determination.—Inspect strip to determine if C-line blind-coded with 120 negative samples and tested by the inde- is completely developed, (Figure 1, step 4). If the C-line is pendent laboratory. Samples were assayed and the results only partially developed, the test is invalid, and the strip were interpreted independently by the reader and visually. should be discarded; this sample will need to be retested. If the The experiment was divided into 2 separate experimental C-line is completely developed, the test strip may be read by blocks, each containing 60 negatives. The first block con- the ROSA reader. A visual result is obtained by comparing the tained 3 drugs and 60 negatives, 510 samples; the second darkness of the T(test)-line with that of the C-line. If the T-line block contained 2 drugs and 60 negatives, 360 samples. Data appears darker than or the same as the C-line, the sample is analyses were performed to determine the minimum 90/95% negative. If the T-line appears lighter than the C-line, the sam- detection levels (Robert Condon, FDA, unpublished data, ple is presumptive positive. If the T-line is only partially de- 1998) and to determine precision and accuracy (12). veloped, the sample is presumptive positive. Ruggedness, stability of frozen milk, chemical interfer- Reader determination.—Insert visually valid SL test strip ences, and QA manufacturing.—The manufacturer’s submis- into ROSA reader, or equivalent. Choose appropriate channel, sion included data on frozen and thawed milk samples; data on and then enter sample ID and operator ID if necessary. Press ruggedness perturbation obtained by following an approved “ENTER.” Reading and interpretation appear in 5 s. Negative multivariate experiment (13); data on interfamily animal drug readings and zero indicate that the T-line is darker than the C- cross-reaction obtained by using a cocktail of 27 animal drugs line, and the result is interpreted as negative. Positive readings at 100 ppb (sulfadiazine, sulfanilamide, sulfathiazole, indicate the T-line is lighter than the C-line, and the result is sulfamethazine, sulfapyridine, sulfadimethoxine, tetracycline, interpreted as positive. oxytetracycline, chlortetracycline, doxycycline, gentamicin, neomycin, streptomycin, ivermectin, erythromycin, pirlimycin, Quality Control (QC) tilmicosin, novobiocin, furosemide, trichlormethiazide, Positive and negative controls are run daily and when pre- thiabendazole, chlorothiazide, oxytocin, phenylbutazone, dexa- sumptive positive samples are retested. Negative controls are methasone, p-aminobenzoic acid, and dipyrone); and data on visually negative, with the T-line clearly darker than the interfamily cross-reaction. These studies were conducted with C-line, and typically give readings of <–600. Positive controls negative raw milk and with raw milk spiked with 4 ppb are visually positive, with the T-line clearly lighter than the penicillin G and 15 ppb cephapirin, and the projected 90/95% C-line, and typically give readings of >400. Calibration strips concentrations were calculated according to the Probit Analysis are supplied with the ROSA reader and must give daily read- Program (version 1.4 corrected for one tail) of the U.S. Envi- ings that are within 20% of their printed average. ronmental Protection Agency, Cincinnati, OH. Information about QA and QC manufacturing, data for 3 manufactured Quality Assurance (QA) batches, and shelf-life data were supplied to support consis- All data for the manufacturer’s method and the data from tency in manufacturing and shelf-life claims. the independent laboratory were obtained in compliance with Somatic cell interference study.—The independent labora- Good Laboratory Practices (GLP; 11). Data were made avail- tory identified farm raw milk with a high somatic cell count able to reviewers, and QA reviews of submitted data were per- per milliliter (SCC/mL), i.e., 1 100 000 SCC/mL and quali- formed by FDA’s Center for Veterinary Medicine fied it as negative for with the Charm II β-Lactam (FDA–CVM). Test (Quantitative Assay). Milk was spiked with 4 ppb peni- 32 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001

Table 1. Reader data obtained by the manufacturer Table 1. (continued) and the independent laboratory for spiked samples % Positive (no. positive/no. tested) % Positive (no. positive/no. tested) Drug added, ppb Manufacturera Independent labb Drug added, ppb Manufacturera Independent labb Penicillin G Nonec 1 0 (0/30) 0 0 (0/60) 2.5 (3/120) 2 13 (4/30) 7 (2/30)

Ampicillin 2.5 17 (5/30) 3 73 (22/30) 70 (21/30) 2 3 (1/30) 3.75 100 (30/30)

4 13 (4/30) 3 (1/30) 4 100 (30/30) Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 5 30 (9/30) 5d 100 (30/30) 100 (30/30) 6 83 (25/30) 33 (10/30) a Data resubmitted by Charm Sciences, Inc. and reviewed by 7 80 (24/30) FDA–CVM and AOAC–RI. b 8 100 (30/30) Data collected by the independent laboratory, Dairy Quality Control d Institute. 10 97 (29/30) 100 (30/30) c Negative raw milk. d U.S. safe level or tolerance. Amoxicillin

2 3 (1/30) cillin G and 15 ppb cephapirin. Negative samples were di- 2.5 7 (2/30) vided into 60 replicates and the 2 positive spiked milks were 3 10 (3/30) each divided into 30 replicates. The 120 blind-coded samples 4 70 (21/30) 63 (19/30) were sent to the manufacturer’s laboratory for testing. The in- 5.5 97 (29/30) dependent laboratory’s code and the manufacturer’s test re- 6 100 (30/30) sults were sent to AOAC–RI for decoding. 8 100 (30/30) Bacterial interference study.— The effects of G+, G–, and d G+/G– mixtures at 150 000 and 300 000 cfu/mL (Pasteurized 10 100 (30/30) 100 (30/30) Milk Ordinance raw milk limit) were assessed in a blind study. Ceftiofur The independent laboratory prepared blind-coded samples (negative, 3.6 ppb penicillin G; and 13.7 ppb cephapirin) by 10 0 (0/30) fortifying milk with freshly prepared bacterial G+, G–, and 20 0 (0/30) G+/G– cultures at 150 000 and 300 000 cfu/mL and then di- viding each of the samples into 6 replicates. The total number 30 13 (4/30) 27 (8/30) of samples was 108 (3 concentrations × 3 bacterial mix- 35 53 (16/30) tures × 2 bacterial levels × 6 replicates). Samples were sent to 40 70 (21/30) 73 (22/30) the manufacturer for testing, and the independent laboratory 48 100 (30/30) determined microbial levels on the same day. The independ- 50d 100 (30/30) 100 (30/30) ent laboratory’s code and the manufacturer’s test results were sent to AOAC–RI for decoding. Cephapirin Incurred residue study.—FDA–CVM prepared samples with incurred residues at levels of parent compound deter- 4 0 (0/30) mined by published liquid chromatographic (LC) methods for 6 0 (0/30) ceftiofur and cephapirin (14) and for 4 (15). The 7 0 (0/30) samples containing incurred residues were collected from a 8 50 (15/30) cow that had been dosed by intramammary injection. Follow- 10 13 (4/30) ing treatment, daily AM and PM milkings were collected and frozen. Samples that contained concentrations within a factor 12 97 (29/30) of 4 of the SL/T were chosen for additional dilution with nega- 12.5 87 (26/30) tive commingled bovine milk (milk that was collected before 16 100 (30/30) injection and tested negative by the Charm II β-Lactam 20d 100 (30/30) 100 (30/30) Test–Quantitative Assay). Samples were blind-coded and sent to the independent laboratory for analysis. The study con- sisted of samples containing incurred β-lactams with 10 replicates each at 1/10, 1/4, and 1/2 of the SL/T, at the SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001 33 Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021

Figure 2. The sensitivity evaluation called for each detected drug to be tested by the manufacturer’s laboratory in 5 equal increments leading to the U.S. SL/T. In this example, the dose response of penicillin G was determined at 1, 2, 3, 4, and 5 ppb (U.S. safe level), ¾n¾. After probit analysis to determine 95% confidence intervals of dose response, n, FDA determined concentrations to be run by the independent laboratory to verify response. For penicillin G, 2, 2.5, 3, 3.75, and 5 ppb were selected for analysis by the independent laboratory, ¾s¾. To determine duplication, the independent laboratory’s 95% confidence curve, ],90% positive level, ¾¾, had to be within 20% of the 90% positive level at 95% confidence,-----,asdetermined from the submitted data set.

90/95% detection level (Robert Condon, FDA, unpublished mined 90/95% levels were 5.6 ppb amoxicillin, 8.5 ppb data, 1998), and at the SL/T (2) for each β-lactam as well as ampicillin, 13.7 ppb cephapirin, 46 ppb ceftiofur, and 3.6 ppb 120 negative samples. penicillin G (Table 2). For a β-lactam screening test to be used in the United States, ≥4 of 6 target β-lactams must be detected Results and Discussion at or below the SL/T values, and the differences between the 90/95% levels of the manufacturer’s laboratory and the inde- Table 1 shows the reader data from the manufacturer’s lab- pendent laboratory must be <20%. These specifications were oratory and the independent laboratory that illustrate the sen- met (Figure 2 and Table 2). Five of the 6 target β-lactam drugs sitivity and selectivity of the SL β-Lactam Test. FDA–CVM were detected. The sixth drug, , is detected at evaluated the data and selected 4 concentrations and the SL/T 50 ppb as determined by cross-reactivity experiments. In addi- in the positive response range for the independent laboratory tion to the detection levels of the β-lactams listed in Table 2, to verify the manufacturer’s data. Negative and SL/T concen- those of other cross-reactive β-lactams were , trations had to meet 90/95% detection level specifications, 50 ppb; cloxacillin, 50 ppb; , 50 ppb; , ≥29 of 30 SL/T samples had to test positive, and ≥58 of 50 ppb; and , 100 ppb. 60 negative samples had to test negative. In general, the test The concentrations in common between the 2 laboratories was less sensitive when visual data (not shown) were used may be used to estimate within-laboratory and be- than when reader data were used to indicate a positive re- tween-laboratories precision and repeatability (12). The aver- sponse at the concentrations studied. Figure 2 demonstrates age standard deviation (SD) for each data set of readings is the calculation of the 90% detection level with 95% confi- about 330. The overall replication SD is 98 and is consistent dence (90/95%) for penicillin G. The independently deter- among the individual concentrations studied within both labo-

Table 2. Minimum 90/95% detection levels calculated from reader data

90/95% Detection level, ppb

Drug Safe level or tolerance, ppb Manufacturer’s lab Independent lab Difference between labs, %

Ampicillin 10 7.1 8.5 19.7 Amoxicillin 10 5.5 5.6 1.8 Cephapirin 20 12.5 13.7 9.6 Ceftiofur 50 48.2 46.2 –4.1 Penicillin G 5 3.75 3.6 –4.0 34 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001 ratories. The overall reproducibility SD is 489 and displays gests a reproducibility of 20% for readings between laborato- more variance among the different concentrations, reflecting ries. The design of the experiment, which includes sample dif- possible sample differences between the laboratories. The ferences between laboratories, adds inherent variation to the reproducibility SD taken as a percent of the total measurement reproducibility calculation. scale (–1500 to +1500, or approximately 3000 points) sug- Detection in the SL β-Lactam Test is based on the use of bacterial receptors in a lateral flow design. The receptors have affinity for all β-lactam drugs, but not necessarily at the SL/T. Table 3. Results for the incurred residue study The receptors are hypersensitive to some β-lactams at the Concn of drug, ppb No. positive No. tested % Positive SL/T. The receptor response is also not predictable by the sen- sitivity of the microbe from which it is isolated (4, 5). To Negative avoid the hypersensitive features of the receptor in the test, β-lactams to which the receptors are hypersensitive are 0 0 120 0 β down-regulated by a proprietary technique. -Lactams near Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 the SL/T cause flowing receptors bound to reddish particles to Ampicillin flow past a β-lactam affinity T-line on the strip and collect on a C-line. The effect causes negative tests to have a T-line that 0.85 0 10 0 is darker than the C-line, and positive tests to have a T-line that 2.13 0 10 0 is lighter than the C-line (Figure 1). 4.25 4 10 40 Results for the negative samples examined by the inde- 8.5 10 10 100 pendent laboratory were 1 positive in 60 samples and 10a 10 10 100 2 positives in 60 samples. These results met FDA 90/95% lev- els, but they were the reason for the increase from 60 to 120 in Amoxicillin the number of negative samples in the experimental design of the incurred residue study. False positives (negative samples 0.56 0 10 0 testing positive) may result from exceeding ruggedness pa- 1.4 0 10 0 rameters that were determined by experiments. Allowable 2.8 4 10 40 perturbation levels were pipeting volume, 300 µL ± 5%; incu- 5.6 10 10 100 bation temperature, 56 ±1°C; incubation time, 8–10 min; am- a bient temperature, 10–30°C; milk temperature, 0–10°C; and 10 10 10 100 reading time (time from completing test to reading in ana- Cephapirin lyzer) of 10 min when strip stored at room temperature (1 week for frozen completed test strips). Parameters that af- 1.37 7 10 70 fect the constituency of milk, such as sampling and mixing, 3.43 10 10 100 are also considered critical and can cause invalid or erroneous results. Generally, perturbations greater than those indicated 6.85 10 10 100 led to more positive readings and invalid results. Because in- 13.7 10 10 100 valid test strips can lead to erroneous results, they must not be b 18 10 10 100 put in the reader.

Ceftiofur No false positive results were observed in the incurred resi- due study (Table 3). The positive response of samples contain- 4.7 10 10 100 ing incurred penicillin G, ampicillin, and amoxicillin was very similar to that of spiked samples. The sensitivity of the SL 14.3 10 10 100 β-Lactam Test was greater for cephalosporin-incurred sam- 28.5 10 10 100 ples (cephapirin and ceftiofur) than for spiked samples. This 47 10 10 100 greater sensitivity to samples containing incurred residues 50a 10 10 100 was also observed when other inhibition and receptor screen- ing methods were used and was presumably caused by the Penicillin G presence of metabolite. Metabolite was observed by liquid chromatography (16). The important conclusion of the in- 0.36 0 10 0 curred residue study is that the SL β-Lactam Test performed 0.9 0 10 0 acceptably with no false positives or false negatives. The sen- 1.8 0 10 0 sitivity of the SL β-Lactam Test is greater for incurred 3.6 10 10 100 than for the parent compound in spiked sam- 5a 10 10 100 ples; therefore, the test errs on the side of public health and safety. a U.S. SL/T. There were no effects on the SL β-Lactam Test from so- b Value closest to the SL/T (20 ppb). matic cells, bacteria, or 27 non-β-lactam animal drugs. SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001 35

Table 4. Results for the interference studies with bacteria, somatic cells, chemicals, and frozen milk

Negative milk, Penicillin G, Cephapirin, Interference challenge Spiking material No. positive/No. tested No. positive/No. tested No. positive/No. tested

Bacteria G+, 180K cfu/mL 0/6 6/6 6/6 G–, 160K cfu/mL 0/6 6/6 6/6 G–/+, 140K cfu/mL 0/6 6/6 6/6 G+, 270K cfu/mL 0/6 6/6 6/6 G–, 280K cfu/mL 0/6 6/6 6/6 G–/+, 390K cfu/mL 0/6 6/6 6/6 Bacteria summary 0/36 36/36 36/36

t = 1.3447 t = 0.9787 t = 0.6922 Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021 Somatic cells 1 100 000 SCC/mL 0/60 30/30 30/30 Chemicals Mixture of 27 animal drugs, 0/6 6/6 6/6 each at 100 ppb t = 0.674 t = 2.397 t = 0.030 Frozen milka (–15°C) Thawed weekly and tested 0/30 30/30 30/30 at 1, 2, 3, 4, 6, and 8 weeks t = 1.034 t = 1.5838 t = 1.0288 a Repeatedly frozen and thawed, centrifuged, and tested over 8 weeks.

Freezing the milk for up to 2 months, the length of the study, References had no effect on results, but frozen milk occasionally (5% of the time) needed centrifugation before testing to remove pre- (1) FDA, Public Health Service (1995) Publication 229, Pasteur- cipitated components that would have caused invalid results. ized Milk Ordinance, Appendix N Table 4 summarizes the results for the 4 studies. There were (2) FDA Coded Memorandum to NCIMS M-I-91-4, July 21, no false results from the experiments, and t-test analysis indi- 1991, Safe Levels/Tolerances for Drugs, IMS List of Sanitation Compliance and Enforcement Ratings of Interstate cated no significant difference in the readings when compared Milk Shippers, Vol. 91-3, U.S. Food and Drug Administration, with test results for samples without the somatic cells, bacte- Washington, DC, http://vm.cfsan.fda.gov/~ear/prime.html ria, or animal drugs. (3) FDA Coded Memorandum to NCIMS M-I-78, March 6, Submitted dose response data for 3 manufactured lots in- 1992, Screening Tests Allowable for Use in Compliance with dicate that the SL β-Lactam Test 90/95% levels of the manu- Appendix N, IMS List, Vol. 92-3 facturer were duplicated by the independent laboratory, with (4) Charm, S.E., & Chi, R. (1988) J. Assoc. Off. Anal. Chem. 71, differences of <20%. Shelf-life experiments indicate that 304–316 90/95% levels were duplicated with differences of <10% at (5) Charm, S.E. (1992) Current Problems Associated with the end of shelf life. These results met protocol acceptance Detection of Antibiotic Residues in Milk and Other Foods, criteria. Analysis of Antibiotic Drug Residues in Food Products of Animal Origin, Plenum Press, pp 31–56 (6) FDA Coded Memorandum to NCIMS M-A-85, Rev. 7 (May Summary 21, 1999) Approved β-Lactam Tests and Their Detection Levels, IMS List, HFS 626, Vol. 99-3 The Charm SL β-Lactam Test when used with a reader met (7) FDA (May 1997) Milk Safety Resolution No. 8 of the the conditions required for approval under the FDA–CVM re- NCIMS Conference, San Francisco, CA view process administered by AOAC–RI. The reader data (8) Charm, S.E. (1994) Dairy, Food and Environmental submitted by the manufacturer were satisfactorily duplicated Sanitation 14, 151–154 by DQCI, the independent laboratory, which verified the de- (9) U.S. Pharmacopeia (1995) 21st Ed., U.S. Pharmacopeial tection of β-lactam drugs at the following 90/95% detection Convention, Inc., Rockville, MD, pp 1161–1162 levels: amoxicillin, 5.6 ppb; ampicillin, 8.5 ppb; cephapirin, (10) FDA Coded Memorandum to NCIMS M-A-85, Rev. 5 (July β 13.7 ppb; ceftiofur, 46.2 ppb; and penicillin G, 3.6 ppb. The 31, 1996) Charm II -Lactam Quantitative Assay Screening and Detection Levels, IMS List, HFS 626, Vol. 97-2 SL β-Lactam Test met the other approval criteria for rugged- (11) 1994 Code of Federal Regulations (1994) U.S. Government ness, lack of interferences, stability of frozen milk, selectivity, Printing Office, Washington, DC, Title 21, Part 58, Good incurred residues, and QA. The use of the reader was accepted Laboratory Practices for Nonclinical Laboratory Studies under the protocol of the NCIMS milk-screening program. (12) ISO 5725-2 (1994) Accuracy (Trueness and Precision) of This acceptance resulted in NCIMS approval, in May 1999, of Measurement Methods and Results, International Organiza- the SL β-Lactam Test for screening U.S. bulk tank milk (6). tion for Standardization, Geneva, Switzerland, pp 6–9 36 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 1, 2001

(13) Youden, W.J., & Steiner, E.H. (1975) Statistical Manual of the AOAC, AOAC, Arlington, VA, Ch IV, D, pp 33–36 (14) Schermerhorn, P.G., Chu, P., Ngoh, M.A. (1998) J. AOAC Int. 81, 973–977 (15) Sorensen, L.K., Rasmussen, B.M., Boison, J.O., & Keng, L. (1997) J. Chromatogr. B 694, 383–391 (16) Zomer, E., Quintana, J., Saul, S.J., & Charm, S.E. (1995) J. AOAC Int. 78, 1165–1172 Downloaded from https://academic.oup.com/jaoac/article/84/1/29/5656549 by guest on 29 September 2021