5 Quantification of caffeine in horse urine and saliva by gas chromatography and ELISA Adriano B. Carregaroa*, Otávio A. B. Soaresb, Maria Isabel Mataqueirob, Antonio de Queiroz-Netob a Departamento de Clínica de Pequenos Animais, Centro de Ciências Rurais, Universidade Federal de Santa Maria, UFSM, Santa Maria, RS. b Departamento de Morfologia e Fisiologia Animal, Faculdade de Ciências Agrárias e Veterinárias, FCAV - UNESP, Jaboticabal, SP. *Autor Correspondente: Adriano B. Carregaro. Departamento de Clínica de Pequenos Animais, Centro de Ciências Rurais, Universidade Federal de Santa Maria, UFSM. Faixa nova de Camobi, km 9, Santa Maria, RS, Brasil. CEP.: 97105-900. Tel: (55) 3220-8460. e-mail: carregaro@smail.ufsm.br. This work was conducted at the Laboratório de Farmacologia e Toxicologia do DMFA da FCAV/UNESP, Câmpus de Jaboticabal. Resumo © 2006 Sociedade Brasileira de Toxicologia direitos reservados Revista Brasileira de Toxicologia 19, n.1 (2006) 5-9 SOCIEDADE BRASILEIRA DE TOXICOLOGIA REVISTA BRASILEIRA DE TOXICOLOGIA BRAZILIAN JOURNAL OF TOXICOLOGY recebido em 12/2005, aceito em 04/2007 Quantificação de cafeína em urina e saliva eqüina por cromatografia a gás e teste ELISA. Foram pesquisados resíduos organoclorados em sangue humano, colhido de 122 indivíduos. Foram compostos dois grupos: um urbano e outro residente em áreas de intensa atividade agrícola. Os resíduos organoclorados pesquisados foram: - HCH; - HCH; Aldrin; Dieldrin; Endrin; Heptacloro; Heptacloro epóxido; HCB; MIrex; o,p´DDD; o,p´DDT; Oxiclordane; p,p´Metoxicloro; p,p´DDD; p,p´DDE; p,p´DDT e Transnonacloro (limite de detecção = 0,010 g/L). Todos os indivíduos responderam a questionário com quesitos sobre idade, sexo, peso, atividade, local de residência, contato ou não praguicidas. Todas as amostras urbanas foram negativas para organoclorados analisados. 10,5% das amostras obtidas nas áreas de atividade agrícola foram positivas. Os resultados das análises do presente trabalho foram comparados com resultados pré-existentes no Rio Grande do Sul. Entre a pesquisa atual e a pesquisas anteriores usadas para comparação, verificou-se diminuição da incidência de organoclorados. Unitermos: Resíduos organoclorados, sangue. Abbreviations: ELISA: enzime-linked immunosorbent assay; GC: gas chromatography; NPD: nitrogen-phosphorus detector; IV: intravenous; N2: nitrogen. Introduction Caffeine became a powerful choice used with the purpose of increasing cardiopulmonar parameters, not only in humans but in animals, due its central stimulation activity (KUROSAWA et al., 1999). As it was already in the list of forbidden substances for the World Anti Doping Agency, studies were done to evaluate the caffeine concentration in vitamin and energy supplements in the attempt of discerning accidental from intentional doping (SALVADORI et al., 1994; KAMBER et al., 2001). In veterinary sports medicine, caffeine is forbidden by Federation Equestrian International, although studies with this substance continue being accomplished in equine (CARREGARO et al., 2004; SAVAGE et al., 2005). To evaluate caffeine and other doping agents, detection methods have become increasingly more sophisticate and accurate. Methods usually employed to quantify caffeine in biological specimens are gas chromatography (CG) and high-performance liquid chromatography (HPLC), both time consuming and expensive (SALVADORI et al., 1988). Therefore, it was of interest to apply methods capable of minimizing the assay time and cost without losing quality. Thin-layer chromatography (TLC) was the first technique employed as screening test for quantification of substances in biological matrices in horses (TOBIN, 1981). However, the only advantage was the cost, since the delay in the procedure and the high incidence of false negatives questioned its use. This was particularly the case when potent pharmacons came into use such as anesthetics, stimulants and narcotics, which produce effects even at very low concentrations. Later, immunofluoresence tests were utilized due to their relatively high sensitivity and rapid results (TOBIN, 6 1988). However, the difficulty in obtaining kits, which required their development, and the need for sample extraction and the use of a fluorimeter led to this technique being replaced by immunoenzymatic assays, which are now widely used in various lines of research and testing. Enzime-linked immunosorbent assay (ELISA) has been utilized routinely as a pre-selection step in drug assays (THURMAN et al., 1990; YEUNG & NEWSOME, 1993; SAWAYA et al., 1998 CARREGARO et al., 2001). Moreover, it is starting to be used as the only test in drug testing because of its practicality and low cost (EDGAR et al., 1996; IL'YASOVA et al., 2004). In view of this, ABAD and colleagues (1993) compared the results obtained with ELISA and GC in the determination of nicotine and found a correlation coefficient (r) of 0.88. In preliminary studies, it was possible to compare GC and ELISA for the determination of caffeine in human plasma, whereby an r of 0.82 was obtained (CARREGARO et al., 2001). It was also shown that samples diluted close to the midpoint of the ELISA standard curve (I 50 ), that is, where there is an inhibition of 50% of the maximum absorbance of the colorimetric reaction, give the best correlation. The caffeine levels determined by ELISA were higher than those by GC, probably as a result of caffeine metabolites and of other methylxanthines from human nutrition feed which was not controlled in these experiments. The immunoenzymatic assay does not discriminate the above compounds and is thus susceptible to cross-reactions. The aim of the present study was to compare GC and ELISA methods in the determination of caffeine in urine, a classical matrix, and saliva, due the facility and noninvasive technique to obtain it, from horses fed with caffeine or any other methylxanthines free meals. M aterial and M ethods Seven thoroughbred mares were utilized in the study. They weighed between 380 and 550 kg, and were maintained on pasture grass (Cynodon dactylon) supplemented with commercial ration, mineral salt and coast-cross hay (Cynodon sp), all known to be devoid of methylxanthines. A dose of 2 mg/kg of caffeine (Sigma Chem. Co., USA) IV was administered diluted in 10 ml of saline solution. Urine and saliva samples were collected, in duplicate, before the administration of caffeine and at 15 min, 30 min, 1 h, 2 h, 4 h, 6 h, 8 h, 12 h, 24 h, 48 h and 72 h post-administration. A pharmacokinetic study of this drug was carried out in parallel. For the determination of caffeine by GC, samples were extracted according to the method recommended by GREENE and colleagues (1983). Standard solutions of caffeine were diluted in methanol to establish a standard curve using 5 known concentrations (0.2, 1, 5, 10 and 20 g/ ml) which served as a reference for the assay of samples. Samples (500 l) were placed in assay tubes to which 500 l of deionized water, 1ml of saturated sodium chloride solution and 2ml of dichloromethane were added. The tubes were agitated for 4 min and centrifuged for 1 h at 4 °C, at 626 g. After centrifugation, the organic phase was separated and dried with the help of a stream of N 2 , in a warm water bath at 35 °C. Subsequently, the residue was reconstituted with 50 l of methanol and injected into a chromatograph with a split less injector at 230 °C. The initial temperature of the capillary column was 150 °C and the final 280 °C (20 °C/min). The GC (CP-8400. Varian Ind. Com. Ltda., Brazil) used was equipped with a nitrogen-phosphorus detector (NPD). Hydrogen was used at a rate of 4.7 ml/min and air at a rate of 167 ml/min. The polymer capillary column (CP SIL 5CB) 30 m X 530 m X 1 m received a constant flow of helium at a rate of 7 ml/min. The injection volume was 1 l. The ELISA method was performed using kits (Neogen Corporation, Lexington, USA) specific for caffeine, according to EDGAR and colleagues (1996). At duplicate, 20 L of standard solution were added to each well to establish standard solutions which were prepared in buffer at concentrations of 1, 5, 10, 50 and 100 ng/ml. In the same way, 20 L of the samples were added to each well. Afterward, 180 l of drug-enzyme complex were added, diluted 1:180. The solution formed was homogenized lightly for 1min and incubated at room temperature for 45 min. After this period, the liquid was discarded and the plate completely dried. All the wells were washed three times with 300 l of washing buffer. Subsequently, 150 l of substrate were added and the plate was agitated for 30 min at room temperature. The reactions were determined with the use of a plate reader (Multiscan Ascent. Labsystem Rec. Tech., Finland) with a light filter for a wavelength of 650 nm. The results of the two methods were compared by Pearson's correlation coefficient (r), according to CALLEGARI-JACQUES (2003), utilizing the GraphPad Instat (GraphPad Software Inc.) program. Results The standard curves obtained with GC were accepted when at least four out of the five dilutions used fit a linear regression curve with r2 equal to or higher than 0.999. To determine the efficacy of the extraction method, a recovery test was performed with 10 samples with a known concentration of caffeine (1 g/ml). Recovery of the drug from urine was 94 % and from saliva 91 %. Concentrations obtained in urine and saliva were therefore multiplied by the correction factor 1.06 and 1.09, respectively. To ELISA, the validate method was similar to anterior, with r2 specified to minimum 0.99. Most of the ELISA results were higher than those determined by CG. To determine the correlation between the two methods, all the samples were assayed by ELISA and the results compared to those obtained with GC. After assaying 300 samples, including 160 of urine and 140 of saliva, an r was obtained for the two methods of 0.53, whereby r2 was 0.28 (y = 1.576 + 0.846x) (Figure 1). The results were considered valid when the I 50 of the standard Carregaro, A.B./Revista Brasileira de Toxicologia 19, n.1 (2006) 7 curve was up to 5 times the value of the sample I (I S ); otherwise, they were diluted until they were within the maximum limit. r = 0,53 r 2 = 0,28 y = 1,576 + 0,846x Cromatografia a gás 0 1 2 3 4 5 6 te s te E L IS A 0 2 4 6 8 10 12 ( g /m L ) ( g/mL)Gas chromatography E L IS A r = 0,53 r 2 = 0,28 y = 1,576 + 0,846x Cromatografia a gás 0 1 2 3 4 5 6 te s te E L IS A 0 2 4 6 8 10 12 ( g /m L ) ( g/mL)Gas chromatography E L IS A Figure 1. Regression curve between the gas chromatography and ELISA used to quantify caffeine in horses urine and salive samples, in g/mL. INTERVAL n r r2 Y = a + bx 0.5 0.6 26 0.75** 0.56 Y = 1.668 + 1.055x 0.6 0.7 23 0.85** 0.72 Y = 0.844 + 2.019x 0.7 0.8 25 0.83** 0.70 Y = 0.781 + 2.120x 0.8 0.9 15 0.71** 0.51 Y = 1.264 + 1.631x 0.9 1.0 20 0.59** 0.35 Y = 1.092 + 1.582x Table 1- Correlation coefficient ( r ), linear regression coefficient ( r2 ) and linear regression equation y= a + bx) of the samples of biological matrices of Thoroughbred horses for the determination of caffeine by gas chromatography and ELISA. The blocks were arranged based on the ratio between I 50 obtained in the standard curves of ELISA and I of the samples, corresponding to each curve. ** p < 0.05 In an attempt to obtain a higher correlation, the total samples (n) were divided into blocks, based on the ratio of the I 50 of the standard curve and the corresponding I S . The number of samples in each block was also kept about the same. Therefore, the samples were grouped into 5 blocks with about 60 samples each. The first block (n = 60), with a ratio between I 50 and I S (the inhibition caused by the sample) of up to 0.4, yielded an r of 0.70 with an r2 of 0.49 (y = 2.135 + 1.334x). The samples of 0.41 to 0.60 (n = 58) showed an r of 0.75 with an r2 of 0.56 (y = 1.749 + 1.3x). However, the block containing values with ratios of 0.61 to 0.90 (n = 64) showed an r of 0.79 with an r2 of 0.62 (y = 0.955 + 1.889x). For samples with ratios of 0.91 to 1.50 (n = 57), r was 0.70 with an r2 of 0.49 (y = 1.156 + 0.855x). The other results within the interval of 1.51 to 5.0 (n = 61), showed an r of 0.69 with an r2 of 0.47 (y = 0.645 + 0.557x). Despite that all the blocks showed a significant correlation coefficient, the samples with ratios between 0.6 and 0.9 had a better correlation. Therefore, it was decided to divide them, again, into smaller intervals to see if the r between the two methods would increase. The results that yielded a ratio of 0.5 to 1.0 were therefore divided into 5 blocks. The block with a ratio between 0.5 and 0.59 (n = 26) had samples with an r of 0.75 and r2 of 0.56 (y = 1.668 + 1.055x). The samples ratios of 0.60 to 0.69 (n = 23) showed an r of 0.85 with an r2 of 0.72 (y = 0.844 + 2.019x). However, the block containing values with ratios of 0.70 to 0.79 (n = 25) yielded an r of 0.83 with an r2 of 0.70 (y = 0.781 + 2.12x). For the interval of 0.80 to 0.89 (n = 15), r was 0.71 with an r2 of 0.51 (y = 1.264 + 1.631x). The results for the block with ratios of 0.90 to 0.99 (n = 20) showed an r of 0.59 with an r2 of 0.35 (y = 1.092 + 1.582x) (Table 1). Carregaro, A.B./Revista Brasileira de Toxicologia 19, n.1 (2006) 8 Discussion The correlation between the two methods studied showed to be positive, albeit discrete. According to CALLEGARI-JACQUES (2003), r varies between -1 and +1. Values between 0 and +1 indicate a positive correlation. However, to consider the two treatments as showing a good positive correlation, an r value above 0.70 is required. Preliminary studies have been conducted comparing GC and ELISA in the quantification of caffeine concentrations in human plasma samples (CARREGARO et al., 2001). The r value obtained (0.82) demonstrated a good correlation between the two methods. The samples diluted close to the I50 of the standard curve showed higher correlation. However, it was shown that measurements obtained with ELISA were consistently higher than those determined by GC, which was the same as the finding reported by LI and colleagues (1989). The plausible explanation for this finding could be susceptibility to cross-reaction. THURMAN and colleagues (1990) compared ELISA to GC-MS in the determination of thiazine herbicides in water samples and obtained an r value of 0.91. Nonetheless, the authors pointed out the fact that some thiazines are more detectable when samples are diluted to below the 50% level of absorbance for the ELISA standard curve. Conversely, others have shown a greater correlation when dilution of the samples is above the I 50 of the standard curve. This could be explained by the reaction of the immunoenzymatic assay, which contains haptens utilized for the production of antibodies with a molecular conformation resembling some thiazines. Therefore, depending on the concentration of each substance, determined by the dilution, antibodies may or may not be activated more, resulting in cross-reaction. Meanwhile, it is noted that ELISA should be considered as good quantification method when combined with GC, because not only is it economical, but it can also be performed on a small volume of sample which is easy to transport. In the present study, after determining the r between the two methods of quantification, the samples were grouped into blocks according to ratio between I 50 of the standard curve obtained with ELISA and the IS corresponding to the samples. An increase in correlation coefficient was thereby shown in samples where the ratio fluctuated between 0.6 and 0.9, where an r value of 0.79 was considered good according to CALLEGARI-JACQUES (2003). Based on this rationale, the samples with ratios of 0.5 to 1.0 were divided into sections. In this situation, the r values obtained between the groups differed more than in the previous case. In this manner, the correlation between samples with ratios of 0.9 to 1.0 (r = 0.59) can be rejected. However, the results with ratios between 0.6 and 0.7 and 0.7 and 0.8 showed higher indices of correlation of 0.85 and 0.83, respectively. In comparison with studies previously cited, the results were shown to be somewhat similar. It should be noted that in this study the biological samples were not submitted to any treatment that could alter the substances present, unlike in the study by YEUNG & NEWSOME (1993). EDGAR and colleagues (1996) cited the possibility of cross reactions in the kits formulated to determine caffeine, since some of the metabolites of this drug are isomers or differ only by the number of methyl groups. Due to simplicity and economy, ELISA can be used for caffeine quantification in the discussed matrices since it presented good correlation with GC. It is plausible that the higher correlation coefficients obtained here compared to that reported to date, could result from improving the dilution of the samples, attempting to approximate the ratio with the best correlation. In addition, all the methylxanthines could be converted to caffeine by the addition of new methyl groups, eliminating thus the possibility of cross reactions, which would also alter quantification by GC permitting a more reliable comparison. Acknowledgments The authors thank the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) for the support received. Dr. Albert Leyva is also acknowledged for his help with the preparation of the manuscript. Resumo O teste ELISA tem sido empregado como teste de pré- seleção na detecção de substâncias utilizadas como antidoping. Todavia, ultimamente tem sido utilizado como única técnica de quantificação em vários países, uma vez que proporciona resultados rápidos e com custo reduzido. O estudo objetivou avaliar a precisão do teste ELISA na quantificação de cafeína tanto em urina quanto em saliva eqüina, comparando-se os valores desse teste aos obtidos em cromatografia a gás (CG). Para o teste ELISA, as amostras foram diluídas para a obtenção de uma concentração próxima ao ponto médio da curva-padrão (I 50 ), correspondente a 50% de absorbância, e posteriormente comparadas aos valores da CG. Para determinar se a proximidade entre o I 50 da curva-padrão e o I da amostra (I a ) fornece valores mais precisos, as mesmas foram divididas em cinco grupos, dispostos de acordo com a razão entre I 50 /I a . As amostras com razão entre 0,6 a 0,7 demonstraram forte correlação (r = 0,85). A possibilidade de reação cruzada no teste ELISA com outras metilxantinas não pode ser descartada, uma vez que as concentrações foram superiores às obtidas por CG. Unitermos: cafeína, ELISA, cromatografia a gás, eqüino. Financial support by Fundação de Amparo à Pesquisa do Estado de São Paulo - FAPESP. Article from Master thesis entitled Estudo farmacocinético da cafeína em eqüinos da raça puro sangue Carregaro, A.B./Revista Brasileira de Toxicologia 19, n.1 (2006) 9 inglês, submetidos ou não à acidificação urinária. Comparação entre as técnicas de quantificação por cromatografia a gás e teste Elisa, in 2002, FCAV - UNESP - Campus de Jaboticabal. References Abad A, Manclús JJ, March C, Montoya A. Comparison of a monoclonal antibody-based enzyme-linked immunosorbent assay and gas chromatography for the determination of nicotine in cigarette smoke condensates. Anal Chem, 1993, 65: 3227-3231. Callegari-Jacques SM. Bioestatística - Princípios e aplicações. Ed. Artmed, Porto Alegre, 255p., 2003. Carregaro AB, Woods WE, Tobin T, Queiroz-Neto A. Comparison of the quantification of caffeine in human plasma by gas chromatography and ELISA. Braz J Med Biol Res, 2001, 34: 821-824. Carregaro AB, Mataqueiro MI, Soares OAB, Queiroz-Neto A. Study of caffeine in urine and saliva of horses subjected to urinary acidification. J Appl Toxicol., 2004, 24(6): 513-8. Edgar T, Tobin T, Tai HH, Schroedter D, Ice JT, Bass S. ELISA assay for caffeine. In: Proceedings of 11th International Conference of Racing Analysts and Veterinarians, Brisbane, Australia, (R & W Publications, Newmarket, UK; Occasional Publication), 1996, 478-480. Greene EW, Woods WE, Tobin T. Pharmacology, pharmacokinetics and behavioral effects of caffeine in horses. AJVR, 1983, 44:57-63. Il'yasova D, Morrow JD, Ivanova A, Wagenknecht LE. Epidemiological marker for oxidant status: comparison of the ELISA and the gas chromatography/mass spectrometry assay for urine 2,3-dinor-5,6-dihydro-15- F2t-isoprostane. Ann Epidemiol, 2004, 14(10):793-797. Kamber M. Fight against doping--national and international developments after Tour de France 1998, Ther Umsch, 2001, 58(4):220-5 Kurosawa M, Nagata S, Takeda F, Mima K, Hiraga A, Kai M, Taya K. Effects of caffeine and promazine hydrochloride on plasma catecholamines in thoroughbreds at rest and during treadmill exercise. Equine Vet J Suppl., 1999, 30:596-600. Li QX, Gee SJ, McChesney MM, Hammock BD, Seiber JN. Comparison of an enzyme-linked immunosorbent assay and gas chromatography procedure for the determination of molinate residues. Anal Chem, 1989, 61:819-823. Salvadori MC, Velletri ME, Camargo MMA, Araújo ACP. Identification of doping agents by chromatographic techniques and UV spectrophotometry. Analyst, 1988, 113(8): 1189-1195. Salvadori MC, Rieser EM, Ribeiro Neto LM, Nascimento ES. Determination of xanthines by high-performance liquid chromatography and thin-layer chromatography in horse urine after ingestion of Guarana powder. Analyst. 1994,119(12):2701-3. Savage KA, Colahan PT, Tebbett DR, Rice BL, Freshwater LL, Jackson CA. Effects of caffeine on exercise performance of physically fit thoroughbreds. Am J Vet Res, 2005, 66(4):569-73. Sawaya WN, Lone KP, Husain A, Dashti B, Al-Zenki S. Screening for estrogenic steroids in sheep and chicken by the application of enzyme-linked immunosorbert assay and a comparison with analysis by gas chromatography-mass spectrometry. Food Chem, 1998, 63:563-569. Thurman EM, Meyer M, Pomes M, Perry CA, Schwab AP. Enzyme-linked immunosorbent assay compared with gas chromatography/mass spectrometry for the determination of thiazine herbicides in water. Anal Chem, 1990, 62:2043-2048. Tobin T, Drugs and the Performance Horse, Thomas Books Pub, Springfield, 480p., 1981. Tobin T, Watt DS, Kwiatkowski S, Tai H, Blake JW. Non- isotopic immunoassay drug test in racing horses: A review of their application to pre- and post-race testing, drug quantification, and human drug testing. Res Commun Chem Pathol Pharmacol, 1988, 62:371-394. Yeung JM, Newsome WH. Survey of total tetrahydrophthalimide in baby foods using both enzyme- linked immunosorbent assay and gas chromatography/ mass spectrometry: A comparative study. J AOAC Int, 1993, 76:1225-1229. Carregaro, A.B./Revista Brasileira de Toxicologia 19, n.1 (2006) 10