4.1 Counts per minute for blank samples
The blank samples used in the study were those fish tissue matrices which were carried through the complete analytical procedure, and no antibiotic residues were detected in them using the respective Charm II assay kits [21]. The blank fish samples to which the binder and tracer had been added but without addition of an antimicrobial agent were extracted with the different kits and read on the respective channels. The results of the counts per minute (cpm) for the blank fish samples are summarized in Table 1.
From Table 1, the cpm for tilapia, trout, salmon, pangasius, seabass, dorate, catfish, and lingue, fish species were statistically evaluated using ANOVA and it was found that the overall F-calculated (0.22) was less than F-Critical (2.5), which implied that there was no significant difference between results for the blank fish samples of the aforementioned species when using antimicrobial test kits for β-lactams, tetracyclines, macrolides and streptomycins. However significant difference in cpm values was observed with the sulfonamides extraction kit while testing catfish, lingue and pangasius. The cpm for these species were almost double those of the other types of fish and their F-calculated (15.1) was greater than F-critical (2.4). The big variation in cpm for the catfish, lingue and pangasius fish species as compared to the rest could be attributed to the high fish fat content extracted by the sulfonamide kit protocol. In this respect, the three fish species (catfish, lingue and pangasius) cpm blank values were handled separately when calculating control points to minimize chances of getting false negative or false positive results. For the rest of the fish species, the blank cpm results were used to derive the respective control points for the different residues.
4.2 Evaluation of the Control Points for the different drug residues
The control point (CP) of a sample is the cut-off point between a negative or positive result. Any antimicrobial agent present in the sample extract competes for the binding sites with the tracer, thus, the greater the cpm measured, the lower the antimicrobial drug concentration in the sample and vice versa. Samples with high counts are considered negative (tracer antimicrobials are largely bound to the binder) while those with low counts are considered positive (tracer antimicrobials are largely free in solution). The CP for the different antimicrobials were determined independently; and with the exception of tetracyclines, the MRL value for each drug was spiked to the respective blank fish sample. In order to cater for the deviations in the different fish matrices, a percentage tolerance was added to or subtracted from the obtained average cpm value of the spiked blank fish sample. The CP evaluation was performed according to the Charm II protocol, and the percentages added to the mean value of spiked samples at detection capability or subtracted from the mean value of blanks serve to minimise occurrence of false positive or negative readings [21, 24, 25].
In this respect, the CP for the beta-lactams was evaluated from averaging the results of 6 negative samples spiked with benzyl penicillin at 25 µg/kg (0.5 MRL) and adding 20% of the obtained average cpm value. Whereas, for the sulfonamides, the CP was evaluated by averaging results of negative samples spiked at 50 µg/kg with sulfamethazine and adding 30% of the average obtained cpm value. In this respect, a control point of 1530 was calculated for the beta-lactam (Table 2). On the other hand, the CP for tetracyclines was calculated by averaging cpm results of negative control standards provided in the tetracycline test kit and subtracting 40% of the obtained average cpm value.
For macrolides, the CP was derived from averaging the results of 6 negative samples spiked with erythromycin A at 100 µg/kg (0.5 MRL) and adding 20% of the obtained average cpm value. Using a similar approach, the CP for streptomycins was derived from averaging results of negative samples spiked at 25 µg/kg with streptomycin and adding 30% of the average obtained cpm value. The results of the calculated CP for the different antimicrobials involved in the study are presented in Table 2. During the analysis of antimicrobial residues in different fish samples, results less than or equal to each respective CP were interpreted as positive while those greater than the CP as negative.
A brief comparison of the control points (CP) for the different antibiotics obtained using the Charm II assay kits with the corresponding cut-off points (Fm), calculated following Annex II of the EU guideline, for Community Reference Laboratories Residues 20/1/2010 [26], is presented in Table 3.
The Fm value refers to the response or signal from a screening test which indicates that a sample contains an analyte at or above the screening target concentration [26]. From the results presented in Table 3, the Fm values obtained using the EU guideline and the respective calculated CP according to the Charm II protocol are comparable. However, in all cases the CP value for a particular family of antibiotics is slightly higher the corresponding Fm readings, with the exception of sulfathiazole; this implies that there will be less incidences of false negative readings in the detection of the different antimicrobial compounds in fish matrix.
4.3 Detection capability for the different antimicrobials in selected fish species
The detection capability (CCβ) is the lowest concentration of the analyte that could be detected in the sample giving at least a 95% positive result. In these studies, blank negative fish tissue samples were spiked with different antimicrobials and a summary of the CCβ for the selected drugs is presented in Table 4. Results of detection capability show that the Charm II analytical technique can detect tetracycline and chlortetracycline at 25 µg/kg (0.25 MRL) and oxytetracycline at 100 µg/kg (MRL) for the different fish species (cat fish, trout, salmon, seabass, tilapia, lingue, dorade, and pangassius) with 100% detection. However, the batch of the multi antimicrobial standard, provided in the Charm II kit was not sensitive enough for chlortetracycline to be detected at 100 µg/kg (MRL) level. This could be attributed to the deterioration of the chlortetracycline in the standard due to poor handling, probably during transportation. In this respect, a Sigma Aldrich standard was used and chlortetracycline detected at a concentration as low as 0.25MRL. Interestingly, it was observed that the technique is also capable of detecting other antimicrobials belonging to the tetracycline family (tetracycline, oxytetracycline) and not limited to the chlortetracycline provided for in the Charm II test kit.
The sulfa drugs including, sulfadimethoxine, sulfadiazine, sulfathiazole were detected at 25 µg/kg (0.25 MRL) for the different fish species (cat fish, trout, salmon, seabass, tilapia, lingue, dorade, and pangassius) at 100% detection, and sulfamethazine, 25 µg/kg (0.25 MRL) at 96.6% detection (3.4% false negatives). It should be noted that the catfish, pangassius, and lingue fish species gave high counts for blank samples in respect to sulfonamide drug residues as compared to other fish species and their control points were established separately. The results also show that the technique can detect other antimicrobials belonging to the sulfonamides group (sulfamethazine, sulfadimethoxine, sulfamerazine, sulfadiazine, sulfathiazole) which are not in the MSU multi antimicrobial standard mix, provided in the Charm II test kit. For the macrolides; erythromycin A, tilmicosin, and Tylosin A were detected at 100 µg/kg, for the different fish species (cat fish, trout, salmon, seabass, tilapia, lingue, dorade, and pangassius) with 100% detection. Whereas, results for the beta-lactams show that benzyl penicillin, ampicillin, amoxicillin, oxacillin, dicloxacillin and cloxacillin were detected at 25 µg/kg, 50 µg/kg, 50 µg/kg, 300 µg/kg, 300 µg/kg and 300 µg/kg respectively for the different fish species (cat fish, trout, salmon, seabass, tilapia, lingue, dorade, and pangassius). Thus, benzylpenicillin is detected at 0.5MRL, whereas ampicillin, amoxicillin, oxacillin, dicloxacillin and cloxacillin are all detected at their respective MRL. However, 4.5 and 5% of the results for dicloxacillin and cloxacillin respectively, were false negatives (Table 4). Further on, the Charm II technique is capable of detecting streptomycin at 25 µg/kg (0.05MRL) for the different fish species (cat fish, trout, salmon, seabass, tilapia, lingue, dorade, and pangassius) at 100% detection.
A comparison of the detection capabilities (CCβ) and maximum residue limits (MRL) for the different antimicrobials is shown in Figure 1. The results show that, CCβ for the validated antimicrobials were below or equal to the MRL for all drug residues in this study, with the exception of tilmcosin which was detected at 2 MRL. Most of the drug residues exhibited detection capabilities in the range 0.05 MRL to 0.5 MRL, with 100% accuracy. The Charm II technique exhibited better detection capability for tetracyclines at 25 ppb (0.25 MRL) compared to other rapid screening techniques such as the ELISA kit of R-Biopharm for screening tetracycline antibiotic residues in the muscle of chicken, beef, and shrimp, which detected the same at 100 ppb (MRL) [27]. In another study, results of the revolutionary Biochip Array Technology detected tylosin and oxytetracycline at 0.10 and 0.5 of the respective MRL in samples [28].
The limits of detection (LOD) obtained using the Charm Test II assays, and the limits of quantitation (LOQ) for selected literature chemical methods are presented in Table 1b (Supporting information). The LOD results for fish matrix obtained in this validation using the Charm II kits, are comparable to the manufacturer’s claims for the tissue matrix. However, some antimicrobial compounds could be detected in fish tissue at levels lower than the manufacturer’s claim (Table 1b, Supporting Information). The LOD results were also compared with the LC-MS/MS analysis of sulfadimethoxine [29], HPLC-MS/MS analyses of tetracyclines, chlortetracycline, oxytetracycline, sulfadimethoxine, sulfamerasine and sulfadiazine [30]; and LC-ESI-MS/MS analyses of a range of tetracyclines, β-lactams, aminoglycosides and sulfonamides [31]. Generally, the rigorous chemical techniques, as expected, offer lower LOQ values compared to the respective LOD obtained with the Charm II tests. However, the Charm II test demonstrated ability to detect a wider range of antibiotics belonging to different classes including tetracyclines, macrolides, β-lactams, aminoglycosides and sulfonamides at MRL or lower levels.
4.4 Repeatability of the method
Repeatability analysis was performed using the same Charm II protocol for a specific antimicrobial on different fish species performed by the same researcher. This was evaluated by means of the intraday coefficient of variations and the results are presented in Table 5. Results of the repeatability study characterized by the relative standard deviation (%RSD) was satisfactory with a precision of less than 12% for the different antimicrobial drugs including tetracyclines, macrolides, beta-lactam, aminoglycosides, and sulfonamides; spiked in blank fish samples at MRL, 0.5MRL or concentration less than 0.5MRL and analysed under repeatability conditions (n ≥ 6). The coefficient of variation expressed as percentage relative standard deviation (RSDr) ranged from 7.8 to 9.8% for tetracyclines (chlortetracycline and oxytetracycline), 2.8 to 6.3% for macrolides (erythromycin), 6.9 to 9.7% for beta-lactams (penicillin G), 10.01 to 11.5% for aminoglycosides (streptomycin); and for sulfonamides (sulfathiazole) it was from 1.2 to 8.7%. These results, ably demonstrate the protocol’s repeatability when used for testing different antimicrobial residues in fish tissue matrix.
A closer look at results obtained under repeatability conditions in the analysis of different fish samples spiked with 25 µg/kg sulfathiazole is presented in Table 6. The results showed that there was no significant difference in cpm readings for the same fish species, and amongst different fish species including dorade, salmon and seabass, spiked with sulfathiazole at the same concentration level (ANOVA, overall F-critical 3.35 > F-calculated 1.99) with RSD < 10%. Similar observations were made for the other antimicrobial agents, whose summarized results are presented in Table 5.
4.5 Reproducibility of the method
The reproducibility studies were performed by two different researchers following the same Charm II protocol on selected fish species, spiked with different antimicrobial agents and evaluated by means of intra-day and inter-day coefficient of variations. The reproducibility study characterized by the relative standard deviation (%RSD) was satisfactory with a precision of less than 15.3% for the different antimicrobial drugs (tetracyclines, macrolides, beta-lactam, aminoglycosides, and sulfonamides) spiked in blank fish samples at MRL, 0.5MRL or concentration less than 0.5MRL and studied under reproducibility conditions (n ≥ 6). The coefficient of variation calculated as percentage relative standard deviation (%RSD) for tetracyclines (chlortetracycline and oxytetracycline) was 7.2 to 11.4%; macrolides (erythromycin) ranged from 5.8 to 8.9%; beta-lactam (penicillin G) from 10.4 to 11.2%; aminoglycosides from 8.9 to 15.1% and sulfonamides (sulfathiazole) from 2.8 to 8.3% as indicated in Table 7.
An elaborate presentation of some results of the reproducibility studies performed by two different researchers following the same Charm II protocol on selected fish species, spiked with oxytetracycline at a concentration level of 100 µg/kg, is presented in Table 8. A comparison of the results obtained by the two researchers for the same fish species showed no significant difference; and the overall analysis showed no significant difference in the cpm results for the different fish species including seabas, pangasius and salmon (ANOVA, F-critical 4.1 > F-calculated 0.64), with RSD < 10%, which further demonstrates the technique’s reproducibility with little matrices interference. Similar observations were made for the other antimicrobial compounds, whose summarized results are presented in Table 7.
4.6 Robustness of the method
Analysis of batches of many samples often require a couple of hours before completion; and there is likely to be a time interval between the first and last analysis of the processed samples. In the robustness testing of the Charm II assay, the effect of variation in reading time interval for processed samples was studied. Robustness testing was performed on samples spiked with 50 µg/kg amoxicillin and analysed on the beta-lactam channel immediately after mixing (0 hour) and after 14 hours. The control point for beta-lactam was set at 1530, and the robustness results are presented in Table 9.
4.7 Specificity and cross reactivity of the technique
The specificity and cross reactivity analysis was carried out in order to determine whether the presence of non-target drugs may lead to false identification of the target drug; or whether the identification of the target analyte may be hindered by the presence of one or more interferences. Representative blank fish samples were spiked with different antimicrobial drugs at known concentration levels higher than those likely to interfere with the identification of the analyte of interest, and then analysed using the respective Charm II protocol for the target drug. The aminoglycosides (spectinomycin, neomycin and paromomycin) were analysed using a macrolide channel meant for (erythromycin, tilmicosin and tylosin) and the results of this study are presented in Table 10. In these studies, standard mix containing the respective antimicrobials (aminoglycosides) at 150, 300 and 500 µg/kg concentration were used to spike the samples.
Results show that although the macrolides which were the targeted antimicrobials tested positive, the non-target aminoglycosides intentionally analyzed on the same channel, tested negative since in all cases the observed cpm were above the set control point of the macrolides of 2118. In similar studies, cross reactivity was investigated by spiking residue-free, blank fish samples with high level concentrations (10 MRL) of antimicrobial substances belonging to other groups (sulfonamides, beta-lactams, macrolides, and tetracyclines) were analysed on the aminoglycosides channel; and the results are presented in Table 11.
The results of these studies also showed that no residues of the non-target drugs (tetracycline, penicillin G, sulfamethazine and tilmicosin) could be detected using the aminoglycosides channel as shown in Table 11. All spiked samples tested negative and the non-target compounds could not be detected even at the high concentration of 10 MRLs. Similar observations were made utilizing the Biochip Array Technology technique, none of the tested six antibiotics could be detected under cross-reactivity studies [27].