Study design and sample collection

Analytical standards and chemicals

Morphine, morphine-d₃, tramadol, ¹³C-tramadol-d₃, buprenorphine-d₄, chlorpromazine-d₃, and phenylbutazone were obtained as solutions from Sigma-Aldrich Chemie GmbH (Buchs, Switzerland). Butorphanol and buprenorphine were obtained as solutions from ReseaChem GmbH (Burgdorf, Switzerland) and Lipomed AG (Arlesheim, Switzerland), respectively. Ketoprofen-d₃ and phenylbutazone-(diphenyl¹³C₁₂) were purchased as powders Sigma-Aldrich Chemie GmbH (Buchs, Switzerland). Detomidine, acepromazine, meloxicam, ketoprofen, chlorpromazine, flunixin and fluphenazine were analytical standards available as powders in the authors’ laboratory. All solvents and other chemicals were purchased from Sigma-Aldrich Chemie GmbH (Buchs, Switzerland). All standards were dissolved in methanol, except for meloxicam that was dissolved in dimethylformamide, to yield stocks of 1 mg/ml. Firocoxib was obtained as an injectable for veterinary use. Chromatography-grade water used for LC-MS/MS analysis was processed by a PURELAB Option-Q system by ELGA LabWater (Labtec Services AG, Villmergen, Switzerland).

Preparation of working solutions

Spiking solutions for calibrators and quality control (QC) samples were prepared in methanol to yield concentrations comparable to those found in hair. An internal standard (IS) solution with the following concentrations was prepared in methanol: morphine-d₃ (15 ng/ml), ¹³C-tramadol-d₃ (8.0 ng/ml), buprenorphine-d₄ (20 ng/ml), ketoprofen-d₃ (10 ng/ml), chlorpromazine-d₃ (10 ng/ml) and phenylbutazone-(diphenyl¹³C₁₂) (30 ng/ml).

Hair sample preparation

Mane and tail hair samples were cut into 2- or 4-cm segments; one very long hair sample of 70 cm was analyzed in 6-cm segments. For a thorough decontamination according to the laboratory standard procedure for forensic hair analysis, 10 mg of hair were washed once with 5 mL deionized water and twice with 5 mL acetone for 3 min each. After drying at room temperature, hair segments were chopped in snippets with scissors. For extraction, 100 μl IS solution and 1900 μl methanol were added and samples were sonicated for 4 h at 50 °C. After centrifugation for 7.5 min at 10,000 g, 1500 μl of the clear solution was transferred into a vial for evaporation under a stream of nitrogen at 40 °C. For injection into the LC-MS/MS system, the residue was reconstituted in 60 μl 5 mM ammonium formate^a (pH 3) with 10 % (v/v) methanol.

LC-MS/MS parameters

Method validation

Method validation

An LC-MS/MS method was developed for the detection and quantification of 13 sedative, analgesic and anti-inflammatory drugs commonly used in horses. A representative chromatogram including all analytes is given in Fig. 1. The method, validated according to the Guidelines for Quality Control in Forensic-Toxicological Analyses [17, 18], was selective and very sensitive for all analytes. The detection of buprenorphine, butorphanol, detomidine, firocoxib, flunixin, ketoprofen, meloxicam, morphine and phenylbutazone fulfilled all validation criteria (Table 4). For chlorpromazine and tramadol, matrix effects exceeded the limits of ± 25 % standard deviation, but this was compensated by the inclusion of deuterated standards. The detection of acepromazine and fluphenazine displayed deviations in accuracy as well as precision, and variable matrix effects were observed. However, inclusion of labeled analogs of the two compounds allowed for comparisons between samples from the same horse. For all other drugs, the method was successfully validated and proven to be suitable for the comparison of hair samples from different horses.

Analyte	QC sample	Accuracy, bias (%)	Intra-day precision, SD (%)	Inter-day precision, SD (%)	ME, mean ± SD (%)
Morphine	Low	0.6	6.2	8.8	56 ± 6.9
	Medium	3.2	3.5	3.8	N/A
	High	−0.1	3.4	3.4	53 ± 9.8
Tramadol	Low	2.4	4.3	5.2	107 ± 21
	Med	0.4	4.5	4.5	N/A
	High	4.9	1.8	2.3	183 ± 90
Detomidine	Low	3.9	7.6	7.7	88 ± 24
	Medium	5.9	7.7	9.1	N/A
	High	2.9	4.5	5.9	79 ± 12
Butorphanol	Low	−0.4	10.2	15.2	87 ± 12
	Medium	1.9	4.5	4.5	N/A
	High	−1.8	4.0	6.6	84 ± 10
Buprenorphine	Low	−9.3	4.5	8.0	87 ± 23
	Medium	−1.2	4.5	4.5	N/A
	High	−0.6	2.0	2.0	88 ± 7.7
Acepromazine	Low	87	28	53	97 ± 31
	Medium	8.3	8.2	11	N/A
	High	−0.7	11	11	117 ± 14
Firocoxib	Low	4.7	3.0	6.7	82 ± 11
	Medium	2.6	1.7	2.0	N/A
	High	2.8	4.0	4.0	83 ± 8.7
Meloxicam	Low	4.1	2.9	5.7	78 ± 12
	Medium	4.2	3.1	5.2	N/A
	High	−4.1	3.1	3.5	70 ± 12
Chlorpromazine	Low	−3.5	3.2	6.3	77 ± 42
	Medium	6.2	2.9	2.9	N/A
	High	0.9	2.9	3.2	90 ± 41
Ketoprofen	Low	6.1	5.6	7.0	78 ± 13
	Medium	−1.9	1.8	2.1	N/A
	High	5.3	5.5	5.5	86 ± 12
Flunixin	Low	5.8	5.8	7.3	89 ± 16
	Medium	0.4	2.2	2.8	N/A
	High	3.5	3.2	3.2	75 ± 7.0
Fluphenazine	Low	7.3	5.6	14.8	92 ± 125
	Medium	12.2	20	21	N/A
	High	18.4	18.7	19.3	81 ± 51
Phenylbutazone	Low	6.1	9.8	10.7	75 ± 14
	Medium	1.2	4.4	6.1	N/A
	High	2.3	4.6	6.2	60 ± 6.7

Detection of opioids

Butorphanol was detected in the hair of four out of five horses having received the drug. The concentrations ranged from below LLOQ to 4.0 pg/mg. In horse P02, repeated administrations of butorphanol were reported 22 months before sample collection and the highest concentration (4.0 pg/mg) was detected in a tail hair segment corresponding to a distance from the skin of 36 to 42 cm (Fig. 2a). In horse P08, a single application of butorphanol 6 months before sampling resulted in drug concentrations in tail hair of 0.9–2.3 pg/mg spread across a distance of 6–16 cm from the skin. In many cases, more drugs were found in hair than reported to have been administered. For example, butorphanol was detected in multiple 2-cm segments of the mane and tail hair of horse P07 (peak concentration of 3.6 pg/mg at a distance of 18–20 cm from the skin) and P10 (peak of 0.6 pg/mg at 22–24 cm from the skin), although administration of this drug was not reported. However, the presence of butorphanol was not demonstrated in the dark tail hair of horse P06 treated once 17 months before sample collection.

A single dose of tramadol was administered to horse P08 around 6 months before sample collection. This treatment resulted in a distinctive analyte peak (6 pg/mg) in a unique 2-cm segment of tail hair located 8–10 cm from the skin (Fig. 2b). A distinctive peak of the same drug (1.1 pg/mg) was also found in the tail hair of horse P02 but without any documentation of previous tramadol use. Morphine was present in the tail hair sample of horse P02 at concentrations of up to 1.4 pg/mg, although medication with this analgesic drug had not been recorded. There was no report of buprenorphine administration and, accordingly, all samples were negative for this substance.

Detection of sedatives

A single detomidine application was given to horse P02 on the day of euthanasia. This resulted in detection of traces of the drug around or below the LLOQ in all tested mane and tail hair segments, indicating an external contamination by sweat or other biological fluids. Detomidine was also detected in the mane hair of horse P07, with maximum levels (1.8 pg/mg) at a distance of 4–6 cm from the skin, although administration of this drug was not reported. On the other hand, a single dose of detomidine 6 months before sampling could not be detected in the dark mane hair of horse P08. Acepromazine, given to horse P06 around 15–16 months before sampling, was retrieved at concentrations of 0.5–1.2 pg/mg in tail hair segments corresponding to a distance of 8–40 cm from the skin (Fig. 2c). There was no report of chlorpromazine and fluphenazine administration and none of the samples contained these drugs.

Detection of non-steroidal anti-inflammatory agents

All hair samples of the four horses with reported flunixin administration were positive for this common anti-inflammatory agent, the highest detected concentration being 39 pg/mg. A sharp peak of flunixin deposition was observed in the tail hair of horse P03. In this case, drug administrations occurring 8 months before sampling yielded a distinctive accumulation at a distance of 16–18 cm from the skin (Fig. 2d). It should be noted that flunixin concentrations were higher in tail (up to 3.3 pg/mg) than in mane hair segments (below LLOQ). This trend was confirmed by comparing flunixin residues in the hair of horses P07 and P09 (data not shown). Firocoxib was detected in the tail hair of horse P10 at concentrations up to 5.0 pg/mg and at a distance of 12–16 cm from the (Fig. 2e). This residue is due to firocoxib treatments given 10 months before sampling.

Phenylbutazone administration was reported in six horses, but the substance was detected in the hair of only two of these animals. This is a consequence of the relatively high LOD (10 pg/mg) linked to poor ionization of phenylbutazone in the MS system. To our knowledge, incorporation of phenylbutazone into hair has not been described before. However, we were able to detect this drug along the entire tail hair of horse P06 with the highest concentration of 1300 pg/mg at a distance of 52–56 cm from the skin. This localization corresponds to a non-recorded treatment that may have been carried out more than 2 years before sampling. Phenylbutazone was also detected in the mane hair of horse P08 (Fig. 2f). In this case, the observed localization pattern with a peak concentration of 3600 pg/mg in the 16–18-cm segment is explained with known phenylbutazone administrations 10 months before sampling. Finally, meloxicam was detected in the tail hair of horse P02 at levels close to the LLOQ (data not shown). Ketoprofen had been administered in two cases but without leaving any detectable residues in hair.

Influence of pigmentation

The analyzed hair samples differed in color. The mane and tail hair of horse P01 as well as the mane hair of horse P09 were light and white, respectively, due to their low melanin content. In view of the melanin-dependent incorporation of basic drugs into hair (see Discussion), it is not surprising to find that the light hair of horse P01 was negative for all tested analytes although acepromazine and phenylbutazone administrations had been documented. Notably, the white mane hair of horse P09 (Fig. 3a) displayed only moderately lower concentrations of flunixin than the corresponding black tail hair of the same horse (Fig. 3b). However, all examined hair segments contained flunixin despite the fact that this drug was applied only 1 week before sample collection. Flunixin incorporated via the hair follicle could not yet have grown out as the follicle is located a few millimeters below the epidermis. Thus, the most likely explanation for the presence of flunixin in mane and tail hair of this horse is an incorporation via sweat and sebum [20].

Correlation between time of treatment and hair segment

Frequently, the analysis of mane or tail hair segments revealed a clear peak of drug incorporation (see for example Fig. 2b). In case of a broader distribution, as for example in Fig. 2c, the median distance from the skin was taken as the measure for drug localization. Subsequently, the assignments of drug localization in hair were compared to the reported times of treatment (Fig. 4). This relationship demonstrated that the time of drug administration can be calculated by assuming an average hair growth rate of 1.84 cm per month.