ABSTRACT
      The control of residues of antimicrobial substances in products of animal origin such as milk, is highly relevant because it can cause problems to the consumer (allergies, digestive disorders, resistance to medicines, etc.). Moreover, it can result in interferences in the manufacture of dairy products causing serious economic losses in the dairy industry.
      Therefore, it is necessary to establish an adequate system to control the presence of the most common residues of antimicrobials used in dairy cows, thus preventing them from reaching the consumer. This control should be based on efficient and practical methods which can be used in animal houses and dairy centers. 
      Due to the diversity of available analytical methods on the market to detect antibiotics, a specificity and sensitivity study of the most commonly used microbiological and specific methods in the screening stage was conducted. 
      Fourteen methods (7 microbiological and 7 specific) were used in the assessment. One hundred milk samples from non-treated animals were analysed in triplicate. The specificity obtained for the microbiological methods was high: BRT AiM (99%), Blue Yellow (100%), CMT Copan (100%), Delvotest MCS Accelerator (98%), Delvotest SP-NT (98%), Eclipse 50 (99%), and Eclipse 100 (99%), showing no significant differences among methods. The specificity calculated for the specific methods also reached high values: Beta Star (97%), Delvo XP (99%), Rosa MRL BL (100%), Rosa TET (99%), Snap BL and Snap TET (100 %), Twinsensor BL (100%) and Twinsensor (98%), showing no significant differences among them (p <0.05), whereas the Penzym method showed the lowest specificity (82%) with a high number of doubtful results and was statistically different from the other methods. 
      Method sensitivity was calculated using milk samples from a pool of 30 individual animals. Twenty antimicrobial substances were tested at three concentrations (0.5 MRL, MRL and 2 MRL), though in some cases, these concentrations differed (MRL, 2 MRL and 4 MRL or 0.25 MRL, 0.5 MRL, and MRL), depending on the proximity of the limits of detection of the methods with the Maximum Residues Limits (MRLs) of each antibiotic. Thirty repetitions were tested per concentration for the microbiological methods, and 10 repetitions for the specific methods. 
      In general, microbiological methods presented a high sensitivity to detect betalactam antibiotics at MRL equivalent concentration. Results from the specific methods to betalactams varied depending on the method and the tested molecule. Nevertheless, the sensitivity of these methods was high at MRL for the molecules of penicillin G, cephalexin, cefalonium, cefoperazone, cefquinome, and ceftiofur. 
      For the rest of antimicrobial groups (aminoglycosides, macrolides, quinoline, and tetracyclines), microbiological methods did not detect any of the substances studied at the MRL, except for neomycin, where BRT AiM, Blue Yellow, and Delvotest SP-NT methods presented a sensitivity of 100 %.
      As regards the specific methods to analyse tetracyclines, all the tested methods (Rosa MRL TET, Snap TET, and Twinsensor) showed a sensitivity of 100% for the detection of oxitetracycline at MLR (100 µg/Kg). Rosa TET and SNAP TET, however, presented this sensitivity at 0.5 MLR (50 µg/Kg). Gentamicin and enrofloxacin were analysed using the corresponding specific methods (Snap Gentamicin, Equinox, and Rosa Enrofloxacin) showing a sensitivity of 100 % in all the cases at a concentration equivalent to the Maximum Residues Limit concentration. 
      Multivariant cluster analysis and Principal Component Analysis (PCA) was applied to the sensitivity results. The obtained results showed that the microbiological methods BRT AiM, Blue Yellow, CMT Corner, Delvotest MCS Accelerator, Delvotest SP-NT, Eclipse 50, and Eclipse 100 presented similar sensitivity results, the same as for the specific methods Thread, Twinsensor, Delvo XP, Rosa MRL BL, and Snap BL.
      Clustering of the results of sensitivity and the frequencies of use of the antimicrobial substances most commonly used in dairy cattle was applied to recommend a combination of methods that can guarantee the major spectrum in the detection of antimicrobials in milk. These results indicate that there is no single method which can detect the whole range of antibiotics, i.e. using microbiological methods only, it is possible to detect between 51.3 and 70.4 % of antibiotics. On the other hand, protein receptor models can only detect between 29.7 and 54.6 % of the antibiotics used. In addition, the use of two simultaneous screening methods “betalactam microbiological-specific” did not improve the detection frequency of molecules (between 65.8 and 71.5 %) compared with microbiological methods. Nevertheless, the use of simultaneous screening methods resulted in an efficient control, with two controls per milk sample. 
      The incorporation of a specific method for the detection of residues of tetracyclines in milk to the combinations of “betalactam microbiological-specific” can increase the percentage of coverage of antibiotics in milk between 68.1 and 73.8 %. Furthermore, the periodic incorporation of gentamicin controls (5.7 %) with the Snap Gentamicin method, and enrofloxacin (2.3 %) with Equinox or Enroflox methods would detect up to 81.8 % of the antibiotics used in Spanish dairy cattle.
      In addition, due to the lack of research studies on the effect of methodological factors related with milk sampling, the evaluation of the influence of the time of refrigeration and the presence of acidiol on the response of some of the microbiological screening methods most commonly used in control laboratories for the detection of the residues of antibiotics in milk, was conducted.
      In this study, twelve concentrations of amoxicillin, ampicillin, penicillin G, and oxitetracycline were used, which were analysed after 0, 24, 48 and 72 hours of refrigeration at 4şC. The objective was to calculate the relative losses of antimicrobial activity for each antibiotic, based on the detection limits of each method, using milk samples without preservative and with acidiol. Results indicated that antibiotic loss of antimicrobial activity (AAL) increased with time of refrigeration of milk samples and those losses were the lowest in milk samples containing acidiol. Therefore, when milk is refrigerated, the use of acidiol is recommended as well as its analysis within the first 48 hours of its arrival at the laboratory.
      Given the implications of the presence of residues of medicines in milk for Food Safety, further studies which deal with detection methods of antibiotics are necessary. Moreover, analytical methodologies with wide spectra of detection for the analysis of other antimicrobial groups such as aminoglycosides, macrolides and quinolones should be developed, given their high frequency of use in dairy cattle and the lack of sensitivity of the current screening methods. To establish a more convenient analytical strategy than that presently used, other specific methods or methods with a wide spectrum of detection should be established in the current control system.