MS determination of florfenicol and florfenicol amine antimicrobial residues in tilapia muscle

Journal of Chromatography B, 1035 (2016) 8–15

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

UPLC–MS/MS determination of florfenicol and florfenicol amine antimicrobial residues in tilapia muscle Eduardo Adilson Orlando a , Aline Gabriela Costa Roque a , Marcos Eliseu Losekann (MSc) b , Ana Valéria Colnaghi Simionato c,d,∗ a

Institute of Food Technology, Campinas, SP, 13070-78, Brazil EMBRAPA Environment, Rodovia SP 340, Km 127,5, Tanquinho Velho, CEP 13820-000, Jaguariúna, SP, Brazil c Institute of Chemistry, University of Campinas − Unicamp, P.O. Box 6154, Campinas, SP, 13083-970, Brazil d National Institute of Science and Technology in Bioanalytics, Brazil b

a r t i c l e

i n f o

Article history: Received 9 June 2016 Received in revised form 8 September 2016 Accepted 10 September 2016 Available online 11 September 2016 Keywords: Florfenicol Florfenicol amine Residue Fish Tilapia SPE

a b s t r a c t Despite the benefits to fish farmers, the use of antimicrobials in aquaculture has concerned consumers and competent authorities. The indiscriminate use of such substances promotes the emergence of resistant microorganisms, decreases the effectiveness of treatments, and causes possible toxic effects in humans. In Brazil, florfenicol is the only antimicrobial registered for use in aquaculture and is often used in tilapia in cage creation. Thus, this study aimed to develop a method for determination of florfenicol residues and its metabolite florfenicol amine in tilapia fillet by UPLC–MS/MS. Analytes were extracted with ethyl acetate, followed by liquid-liquid partition clean-up with hexane and SPE. The sorbents C18, phenyl and HLB-Oasis were evaluated by SPE. Phenyl sorbent showed the best results, and the extraction conditions were optimized in the sample matrix with fractional factorial design 24−1 . The analytes were separated on a C18 chromatographic column (50 × 2.1 mm × 1.7 ␮m) using water (A) and acetonitrile (B) as mobile phase at a flow rate of 0.3 mL min−1 with a linear gradient (in% B): 0–2.0 min: 20%; 2.0–2.5 min: increase to 90%; 2.5–3.5 min: 90%; 3.0–3.5 min: decrease to 20%; 4.0–5.0 min: 20%. The analytes were monitored in a MS/MS triple quadrupole system by MRM mode with transitions at m/z 356.1 > 336.1 (florfenicol) and m/z 248.1 > 130.1 (florfenicol amine). The optimized method was validated obtaining LOQ values of 3 and 25 ng g−1 for florfenicol and florfenicol amine, respectively, precision between 20 and 36%, absolute extraction efficiency between 38 and 80%, and adequate linearity. The method was applied to samples intended for human consumption, and within the 15 evaluated samples, only one showed florfenicol residue at 30 ng g−1 , which is below the maximum residue limit established in Brazil. © 2016 Published by Elsevier B.V.

1. Introduction Tilapia (Oreochromis niloticus) is the most important species in Brazilian aquaculture and its production is practiced in net cages, an intensive cultivation characterized by high density of storage, high productivity and low cost of deployment. However, recent studies show that a higher incidence of disease in such production system is correlated with the increase in stocking density, which may result in delays in fish growth, high mortality and loss to the producer [1]. Thus, an effective microbial control of water envi-

∗ Corresponding author at: Institute of Chemistry, University of Campinas – Unicamp, P.O. Box 6154, Campinas, SP, 13083-970, Brazil. E-mail address: [email protected] (A.V. Colnaghi Simionato). http://dx.doi.org/10.1016/j.jchromb.2016.09.013 1570-0232/© 2016 Published by Elsevier B.V.

ronment is required, and the use of veterinary medicines, such as antimicrobials, is an important tool used in aquaculture [2,3]. Despite the benefits to fish farmers, the indiscriminate use of antimicrobials promotes the emergence of resistant microorganisms and causes possible toxic effects in humans due to the consumption of meat containing drug residues [4,5,6]. Moreover, as pointed out by Alexy et al. [7], antimicrobial agents used in aquaculture are released directly into surface water, where a high load of residues can accumulate in sediments. Florfenicol (FF) (Fig. 1) has a broadspectrum activity and is one of the most used antimicrobial agents for the treatment of streptococcosis. In Brazil, FF has been used to treat infected tilapia herds since 2007, and its maximum residue limits (MRLs) established in fish is 1000 ng g−1 [8,9]. Florfenicol amine (FFA) (Fig. 1) is the main metabolite of florfenicol, being regarded as a residue marker for surveillance regulatory purposes. [10,11].

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

Fig 1. chemical structure of florfenicol and its metabolite florfenicol amine.

LC coupled to MS [12,13] or with UV absorption detection [14,15] has been the main used technique for simultaneous analysis of FF and FFA in meat, allowing reliable and accurate quantification of residues. Moreover, recent advances in commercial chromatographic systems allowed the migration to ultra performance liquid chromatography (UPLC or UHPLC), which uses stationary phases with particles of less than 2 ␮m diameter, enabling faster analyses and reduced solvent consumption, allied to increased detectability and analytical resolution. UPLC has been a widely used technique for the determination of veterinary medicine residues in meat [16]. An adequate sample preparation is also critical to the effective analysis of residues in food, since it eliminates matrix interfering compounds and pre-concentrate analytes. For this purpose, solid phase extraction (SPE) [12,14,15,17,18] and liquid-liquid extraction [13,18,19,20] have been the main sample preparation techniques used for analysis of FF and FFA residues in fish. Alternative techniques for sample preparation as: immunoaffinity chromatography (IAC) [21]; ELISA [22]; matrix solid-phase dispersion extraction (MSPD) [23]; and molecularly imprinted solid phase extraction (MISPE) [24] have also been studied. However, these emerging techniques are still difficult to be implemented in routine residue analyses in meat, due to required analyst skills as well as initial costly investment. Tilapia fish farming is economically relevant in Brazil and requires reliable methods to control the proper use of antimicrobials in fish production. Therefore, this study proposes the development, optimization and validation of an analytical method for quantification of FF and FFA residue in tilapia meat, using SPE as sample preparation technique, followed by UPLC–MS/MS analysis.

9

tion of detection conditions was performed in order to maximize the signal intensity, evaluating the following parameters: cone voltage, capillary voltage, desolvation temperature, and nebulizer gas flow. The transition showing the most intense peaks was used for quantification, and the one showing the second most abundant peaks was used to confirm the structure of the molecule (qualitative purposes). The mobile phase (MP) consisted of water (A) and acetonitrile (B) at a flow rate of 0.3 mL min−1 . Linear gradient elution was as follows (in% of B): 0–2.0 min: 20%; 2.0–2.5 min: increase to 90%; 2.5–3.5 min: 90%; 3.5–4.0 min: decrease to 20%; 4.0–5.0 min: 20%. The injection volume was 10 ␮L and the column temperature was maintained at 40 ◦ C. The analytes were monitored by Multiple Reaction Monitoring (MRM) in the MS/MS system, towards quantitative and qualitative transitions, simultaneously. The voltage applied to the capillary was 2.57 KV, the desolvation temperature was 500 ◦ C and the N2 desolvation flow was 800 L h−1 . 2.3. Evaluation of SPE sorbents and solvent volumes A comparison of C18, phenyl and OASIS polymeric sorbents for the extraction of FF and FFA was performed by evaluation of the extraction efficiency. For this purpose, 5 mL of 5% acetic acid solution spiked with FF and FFA simulating sample extracts in three concentration levels (125, 500 and 1250 ng g−1 ) were used. The spiked solutions were percolated by pre-conditioned SPE cartridges, followed by washing with 2 mL of water (OASIS cartridge) or 5% acetic acid (phenyl and C18 cartridges), elution with 5 mL methanol (OASIS cartridge) or methanol containing 0.1% acetic acid (phenyl and C18 cartridges), drying of the extract, resuspension and analysis by UPLC–MS/MS. With the best performance sorbent, variations on the following parameters were evaluated: sample resuspension volume prior to insertion into the cartridge; volume of solution for cartridge washing; volume of solvent for analytes elution. Solutions simulating sample extract containing 500 ng g−1 of analyte was utilized and analyte solutions not extracted by SPE were used as maximum reference of extraction (100%).

2. Material and methods

2.4. Optimization of analytes extraction from the matrix

2.1. Material and equipment

Extraction of analytes from fish matrix was based on a method previously published for the extraction of FF and FFA residues in poultry[12]. The procedure was optimized for fish matrix aiming at extraction efficiency improvement. A fractional factorial design 24 − 1 allowed studying the influence of variation of four factors on two levels concomitantly. The significance of the effects was statistically verified (significance of 95%) by the analysis of variance and application of the Student’s t-test distribution. Table 1 shows the evaluated factors and their levels. A sequence of random experiments in duplicate was developed. Blank samples spiked at 500 ng g−1 of FF and FFA were used and the best response for each condition was indicated by the highest chromatographic peak area of each analyte. Evaluation of the results was performed with the worksheet provided by Teófilo and Ferreira [25], wherein the values and statistical significance of each effect are calculated.

SPE cartridges with sorbents C18 and phenyl were obtained from Applied Separations (USA), and the Oasis-HLB® cartridges with polymeric reversed stationary phase were purchased from Waters (USA). A chromatographic column with stationary phase C18 and dimensions 50 × 2.1 mm × 1.7 ␮M, ACQUITY UPLC BEH model was obtained from Waters (USA). An ultra performance liquid chromatography system, Acquity model, coupled to mass spectrometer with electrospray ionization and triple quadrupole analyzer, Xevo TQD model, both from Waters (USA) were used. Acetonitrile HPLC grade was obtained from Tedia (USA), methanol HPLC grade from J.T. Baker (Mexico) and acetic acid P.A. from Merck (Germany). Analytical standards of florfenicol (FF), and florfenicol amine (FFA) were obtained from Sigma-Aldrich (USA). All aqueous solutions were prepared with deionized water obtained from a MilliQ System (Millipore, USA).

2.5. Analytical validation 2.2. Chromatographic and mass spectrometer conditions Initially, the analytes conditions of ionization were evaluated seeking for the selection of a more intense precursor ion, followed by the assessment of appropriate fragmentation energy and finally the selection of the two most stable product ions. Then, optimiza-

The best extraction efficiency method was validated using the guidelines provided by INMETRO (National Institute of Metrology, Quality and Technology) as reference [26]. For this purpose, selectivity, matrix effect, linearity, limit of detection, limit of quantification, precision and accuracy were evaluated.

10

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

Table 1 Identification of evaluated factors and levels for optimization of analyte extraction in sample matrix. Code - Factor

Identification

Level (−)

Level (+)

C1 C2 C3 C4

Volume of ethyl acetate Ethyl acetate:Ammonium hydroxide ratio (v/v) Concentration of acetic acid Volume of n-hexane (clean up)

6 mL 98:2 2% 2 mL

10 mL 95:5 5% 4 mL

2.5.1. Selectivity and matrix effect Selectivity was assessed with blank samples analyses of tilapia fillets (n = 6), obtained from farms with non-intensive fish cultivation and non-commercial use (from the region of Campinas/SP), assured for exempted antimicrobial use for fish cultivation. The obtained chromatograms were compared with those from blank samples spiked with known analytical standard concentrations. For acceptable method selectivity, interfering peak areas could not present a peak area higher than 20% of the analyte in a low level of concentration within the method linear range. The matrix effect was evaluated by comparing the analytical signal (peak area) of analytes in three different mediums. The first one, named analyte in solvent (AS), corresponded to the analysis of an analytical standard solution prepared by dissolution in the chromatographic mobile phase, following the same dilutions used in the extraction method. AS was considered the analytical signal reference with no matrix effect. The second one, called analyte in spiked extract (ASE), accounted for fortification of blank matrix final extract with analytical standard solution (after elution from the SPE cartridge). ASE allowed the verification of suppression or enhancement of ionization of the analytes in the presence of extracted components from the matrix. The last one, called analyte in spiked matrix (ASM), corresponded to the fortification of blank matrix before all extraction steps, which enabled assessment of analytes losses during the extraction process, as well as the calculation of extraction efficiency or recovery. Matrix effect was evaluated in three levels of concentration within the method linear range, namely: 320, 560, and 1030 ng g−1 . 2.5.2. Linearity Linearity was evaluated in the concentration range from 125 to 1250 ng g−1 by preparing an analytical curve with blank matrix fortified in five concentrations, besides zero (blank), measured in triplicate. By constructing a graph of peak area as a function of concentration of FF and FFA, the linearity of the linear regression was evaluated using the coefficient of determination (R2 ), sensitivity (angular coefficient) and the existence of discrepant values or trends through the standardized residual chart.

of analyte that generated a peak intensity equal to 10 times the signal/noise ratio of the chromatographic baseline.

2.5.5. Accuracy Method accuracy was assessed by extraction efficiency or recovery, which was calculated by comparing the peak area obtained from the chromatographic analysis of a fortified blank sample extract (after all extraction steps), with the peak area obtained from the analysis of a spiked sample before the extraction steps in the same concentration. Accuracy was evaluated at three fortification levels, namely: 320, 560 and 1030 ng g−1 .

2.6. Analysis of fish samples The validated method was applied to the analyses of tilapia samples collected from Furnas reservoir (state of Minas Gerais − Brazil). Furnas reservoir concentrates a region of fish farming, with emphasis on the tilapia production in cages. Thus, method applicability could be evaluated by the investigation of occurrence of FF and FFA residues, as well as whether the concentrations complied with the MRLs established in Brazil and other countries. Fifteen samples were obtained from three different fish farming with five different collection points each. Samples were collected by EMBRAPA Environmental center (Brazilian Agricultural Research Corporation) personnel of the Laboratory of Aquatic Ecosystems, located in Jaguariúna − São Paulo.

2.7. Statistical analysis Statistical evaluation of the SPE optimization and the matrix effect results were performed by variance analysis (ANOVA) and Tukey’s test (95% confidence) by SAS (version 9.1, SAS Inst. Inc., Cary, NC) statistical package program.

3. Results and discussion 2.5.3. Precision Repeatability or intraday precision was evaluated by the coefficient of variation (CV) of the analytes peak areas of the same sample analyses in quintuplicate, for two fortification levels − 250 and 1000 ng g−1 − on the same day and with the same reagents. The intermediate or interday precision was also evaluated by CV of analyte peak areas of the same sample on three different days, with different solutions and reagents in the same fortification levels described for repeatability. 2.5.4. Limit of detection (LOD) and limit of quantification (LOQ) LOD was calculated as the analyte concentration in matrix that generated a peak intensity equal to 3 times the signal/noise ratio of the chromatographic baseline. Therefore, analyses of spiked samples were performed with decreasing concentrations of FFA and FF, beginning from the point of the lowest analytical curve concentration (125 ng g−1 ). LOQ was calculated as the concentration

3.1. Optimization of MS conditions and detection Firstly FF and FFA precursor ions were selected, aiming maximum signal intensity. Positive electrospray ionization mode generated several FF adducts within the mass range 100 to 1000 Da, resulting in decreased signal strength. On the other hand, only ion m/z 356.0 showed high intensity by negative electrospray mode due to the presence of halogen atoms with high electronegativity in FF structure, resulting in more stable ions. For FFA, ion scanning in the positive electrospray ionization mode showed only one high intensity ion at m/z 248.1 ([M+H]+ ), which was selected as the precursor ion. The abundance of ion m/z 248.1 is influenced by the amino group present in FFA structure, which has the tendency to attract protons. Table 2 presents the detection conditions for FF and FFA. With the optimum MS detection conditions, a solution simulating 500 ng g−1 of FFA and FF was analyzed by UPLC–MS/MS, monitoring ions for quantitative analysis (Fig. 2).

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

11

Table 2 Detections conditions for FF and FFA by MS. Analyte

Precursor Ion (m/z)

FF

356.0

FFA

248.1

Fragment Ions (m/z)

Cone Voltage (V)

Collision Energy (eV)

Denomination

185.1 336.1 130.1 151.1

24 24 20 20

20 8 24 27

Confirmative Quantitative Confirmative Quantitative

Fig. 2. Negative (a) and positive (b) UPLC–MS/MS chromatograms of spiked solution with analytes simulating 500 ng g−1 FF (a), and FFA (b) under separation and detection optimized conditions. Mobile phase comprised of water (A) and acetonitrile (B) at a flow rate of 0.3 mL min−1 , with linear gradient (in% B): 0–2.0 min: 20%; 2.0–2.5 min: increase to 90%; 2.5–3.5 min: 90%; 3.0–3.5 min: decrease to 20%; 4.0–5.0 min: 20%. Injection volume of 10 ␮L, column temperature at 40 ◦ C. C18 column 50 × 2.1mm × 1.7 ␮m, Acquity UPLC BEH.

120 (a) (a)

(a)

(a)

(a)

(a)

(a)

(a) (a)

(a)

100 (b) (b)

Extraction Efficiency (%)

80

OASIS

60

Phenyl C18

40 (b) (b)

(c) 20

(c) (b)

(b)

0 125 - FFA

500 - FFA

1250 - FFA

125 - FF

500 - FF

1250 - FF

Concentration (ng g-1) Fig. 3. normalized results of extraction efficiency of SPE sorbents evaluated at low, medium and high concentration levels within the method linear range for FF and FFA. Different letters on each level of concentration indicate statistically significant differences by Tukey’s test.

3.2. Evaluation of SPE sorbents and solvent volumes Solutions simulating sample extracts in three concentration levels (125, 500 and 1250 ng g−1 ) were extracted by SPE with C18, phenyl and Oasis HLB sorbents under the conditions indicated at

session 2.3. Fig. 3 presents the extraction efficiency results (normalized to the most intense response at each concentration level) for each SPE sorbent valued. Sorbent C18 showed the lowest extraction efficiency for FF and FFA analytes. This result can be explained by the high polarity of

12

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

Table 3 Volumes assessed during SPE optimization with sorbent phenyl and extraction efficiency results. Experimental condition

Sample volume (mL)

Washing volume (mL)

Elution volume (mL)

FFA extraction efficiency (%)

FF extraction efficiency (%)

1 2 3 4 5 6

4 2 1 2 2 2

0 0 0 2 1 0

5 5 5 5 5 8

9.7 16.2 5.2 36.7 83.0 31.1

87.7 87.5 63.0 95.1 96.3 100.4

Table 4 Significance effects assessed with the matrix extraction method. Factors

FFA FF

C1*

C2*

C3*

C4*

NS SG (+)

SG (−) NS

SG (−) NS

NS NS

NS: No significant effect; SG (−): negatively significant; SG (+): positively significant. * Conditions shown in Table 1.

FF (log Kow = −0.04) and FFA molecules (log Kow = −1.18) [27], leading to high interaction with the flushing water, and decreasing the retention by the hydrophobic sorbent. Moreover, according to analytes pKa (8.3 and 7.4 for FF and FFA, respectively), in acidic medium analytes are protonated [27], reinforcing the low retention on the reversed-phase sorbent. The sorbents phenyl and OASIS presented suitable extraction efficiencies with values near or higher than 80%. Although sorbent OASIS has presented higher absolute values of extraction efficiency than phenyl, both are statistically equivalent (except for the lower level of concentration). Thus, for the next optimization steps, method validation was performed with extraction employing the sorbent phenyl due to better cost/benefit. Table 3 shows volumes of sample, washing and elution of analytes, as well as extraction efficiency results. The experimental condition 5 showed the best compromise of extraction efficiency for both analytes within all evaluated conditions. However, FFA elution with 8 mL of solution (condition 6) resulted in further extraction efficiency, when compared to 5 mL (condition 2). For this reason, condition 5 was selected for further experiments with an alteration on elution volume. Thus, resuspension of analytes was performed in 2 mL of 5% acetic acid, washing with 1 mL and elution with 8 mL were the conditions with the best extraction efficiency to both analytes. Under these conditions, FFA extraction efficiency improvement was achieved (above 80%) along with FF extraction efficiency better than 95%. The cartridge washing volume was the most relevant factor in FFA extraction, since extraction efficiency increased about 50% with 1 mL washing volume, regarding the use of 2 and 0 mL. This fact indicates that FFA was dragged from the sorbent when larger solution volume was used for washing the cartridge, drastically reducing their extraction efficiency. On the other hand, the absence of cartridge washing generated no change in the sorbent phenyl bed caused by the presence of an acid solution. 3.3. Optimization of the extraction method Blank samples spiked at 500 ng g−1 were extracted with different conditions (Table 1). Table 4 shows the results for the optimization of analytes extraction from the matrix, using fractional factorial design 24 − 1 . The increment on concentration of ammonium hydroxide from 2% to 5% in the extraction solution of ethyl acetate (C2) reduced by 20% the FFA extraction efficiency. Likewise, increasing the concentration of acetic acid from 2% to 5% (C3) reduced by 22% the FFA extraction efficiency. The volume factor change of ethyl acetate from 6 to 10 mL (C1) increased the FF extraction efficiency by 10%.

Changing the n-hexane (C4) volume, used in the clean up step, showed no significant effect for both analytes. Thus, the method initially proposed by Zhang et al. [12] for amphenicol residue analysis in poultry muscle was modified for the determination of FF and FFA in tilapia matrix as follows: i) weighing 2.00 ± 0.10 g of homogenized tilapia fillet sample and adding 10 mL ethyl acetate: ammonium hydroxide (98:2), followed by vortexing for 1 min and centrifugation at 6000 rpm for 7 min; ii) removal of the supernatant and re-extraction of the solid with 10 mL ethyl acetate; iii) combining of supernatants; iv) addition of 1 mL acetic acid to the extract and evaporation under N2 flow at 45 ◦ C until reducing the volume to c.a. 1 mL; v) transferring the extract to 15 mL centrifuge tube; vi) addition of 2 mL of n-hexane and vortexing to clean up co-extracted lipids with ethyl acetate; vii) centrifugation at 6000 rpm for 5 min, followed by removal of the organic phase; viii) sample extraction by SPE with sorbent phenyl, previously conditioned with 2 mL methanol and 2 mL water, followed by washing with 1 mL of 2% acetic acid; ix) elution of the analytes with 2 aliquots of 4 mL of methanol containing 0.1% acetic acid, x) evaporation under N2 flow at 45 ◦ C, resuspension in 2 mL mobile phase (water:acetonitrile 80:20), filtration with polyvinylidene fluoride membrane and UPLC–MS/MS analysis. 3.4. Analytical validation In order to evaluate the method selectivity, coelution of possible interfering species was inspected. There were no significant interfering peaks at the same retention time for any of the analytes, indicating that the method was selective to the analytes even in the matrix presence. Another selectivity criterion adopted for all quantitative analyses was the confirmation of the relation between the signal (peak area) of the quantitative and qualitative mass transition for each analyte, adopting a limit of ±20% of the value obtained in the analysis of analytical standards in solvent. Matrix effect investigation was conducted with the analyses of analytical standards dissolved in different mediums (Figs. 4 and 5). FFA showed higher matrix effect than FF. Comparing analyte addition to the sample extract (ASE) with analyte in solvent (AS), a signal decrease of about 15% was observed − compared to 4% for FF. This result indicates FFA ion suppression in the presence of matrix components, since the same concentration of analytes were added to the sample extract. Analytical signal loss for FFA spiked in matrix was even more pronounced with an average drop of 65% relative to AS, while FF presented a value of 25%. Since an average loss of 17% occurred in the SPE step, the main contribution to FFA decrease of analytical signal must be caused by the non-extracted analyte, possibly due to strong interaction with polar matrix components, such as proteins. Table 5 presents a summary of the obtained figures of merit of the validated method. The CV values ranged from 20 to 27% for intraday precision and from 24 to 36% for interday precision for both analytes. However, precision deviations have also been observed by other authors [17,28,29,30], which have analyzed veterinary drug residues in fish matrix. The main source of variation in intraday precision is the multiple sample preparation steps, although necessary to ensure the UPLC–MS/MS system is not damaged with sample residues

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

13

120 ASE (a) AS (a) ASE (a)

AS (a)

AS (a)

Relative intensity Signal

100

ASE (b) ASM (b) ASM (b)

80

ASM (c)

60

40

20

0

320

560

1032

Spiked concentration (ng g-1) Fig. 4. Matrix effect observed for FF. AS: analyte in solvent; ASE: analyte in spiked extract; ASM: analyte in spiked matrix. AS corresponded to 100% of the analytical signal. ASE and ASM corresponded to signal intensity in% of AS. Different letters on each level of concentration indicate statistically significant differences by Tukey’s test.

120

AS (a) AS (a)

AS (a)

Relative Intensity Signal

100

ASE (a) ASE (b) ASE (b)

80

60 ASM (b) 40

ASM (c)

ASM (c)

20

0

320

560

1032

Spiked Concentration (ng g-1) Fig. 5. Matrix effect observed for FFA. AS: analyte in solvent; ASE: analyte in spiked extract; ASM: analyte in spiked matrix. AS corresponded to 100% of the analytical signal. ASE and ASM corresponded to signal intensity in% of AS. Different letters on each level of concentration indicate statistically significant differences by Tukey’s test.

Table 5 Figures of merit of the developed method. Intraday Precision (%)

Interday Precision (%)

Recovery (%)

Concentration (ng g−1 )

250

1000

250

1000

320

560

1030

FF FFA

20 22

22 27

36 24

33 28

79.0 38.1

74.6 41.8

79.9 52.4

a b

LOD (ng g−1 )

LOQ (ng g−1 )

R2

6 1

25 3

0.9939a 0.9951b

y = 18.43x + 2752.80 - analytical curve linear range: 125–1250 ng g−1 . y = 333.72 x − 31149.28 - analytical curve linear range: 125–1250 ng g−1 , where y = ax + b; with a = slope and b = intercept.

accumulation. The observed interday variation may be the result of analytes ionization variations by the electrospray in the presence of a complex matrix, such as fish, which can lead to ion suppression. To overcome such variation in the analysis of unknown samples, analytical curves constructed with spiked blank matrix were analyzed simultaneously with the samples.

Method linearity presented significant regressions for analytes quantification in the investigated matrix (R2 values above 0.99). Moreover, the distribution graphs of the standardized residues, obtained from the linear regression, showed random priority orders, confirming that the linear regression model is appropriate (data not shown).

14

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15

Fig. 6. Comparison of chromatograms of FF analysis: blank sample (1), blank sample spiked at 25 ng g−1 (2), the unknown sample chromatogram (3) − monitoring the quantitative ion. Fig. 2 describes the chromatographic conditions.

LOD and LOQ values were below the MRLs values established in Brazil (1000 ng g−1 ), confirming the possibility of using the method for determination of florfenicol in residue level in tilapia muscle. Accuracy was expressed as actual extraction efficiency (with values above 70%), and was suitable for FF determination. However, accuracy for FFA determination varied from 38 to 52%, requiring analytical curve elaboration with blank matrix fortification followed by extraction in order to compensate analyte losses. Van der Riet et al. [13] also reported lower extraction efficiency for FFA in fish tissues, showing values below 60%. Nevertheless, even with such loss of analytes in the extraction step, the FFA detectability was suitable for the method purposes and comparable to the literature [12,13], qualifying the method as appropriate for quantification of these residues in tilapia meat. 3.5. Method application to the analyses of fish samples Within the 15 tested samples, only one showed FF residue with concentration above the LOD, but still below the initially assessed linear range. For quantification of FF residue in this sample, the analytical curve linear range was extended to lower concentrations, showing a concentration of 30 ng g−1 . Fig. 6 shows the chromatograms overlapping of: a blank sample (1), a blank sample spiked at 25 ng g−1 (2), and the sample chromatogram (3) − monitoring the quantitative transition of florfenicol. The residue presence was confirmed by the signal ratio between the quantitative and qualitative transition of FF, which was within ± 20% of the value found for the analytical standards.

4. Conclusions The analytical method developed in this study proved to be suitable for analysis of florfenicol antimicrobial residue and its metabolite florfenicol amine in tilapia muscle, and can be used as an evaluation tool of the correct use of this medicine in fish farming. The method showed high detectability, allowing analysis at concentrations below the MRLs established in Brazil. The selectivity of the MS/MS system enabled unequivocal quantification and confirmation of analytes. Precision and accuracy of the method

were suitable for florfenicol and comparable to values found in the literature for florfenicol amine. Due to a strong matrix effect observed in analyte extraction, especially for FFA, the analytical curve was prepared in spiked matrix. The SPE sorbents phenyl and polymeric OASIS presented satisfactory performance for analyte concentration and sample clean up in terms of extraction efficiency. However, sorbent phenyl was selected herein for offering best cost/benefit and presenting extraction efficiency statistically equivalent to OASIS in medium and high levels of analyte concentration within the method linear range. Within the analyzed real samples, only one showed FF residue of 30 ng g−1 , which is far below the MRLs established in Brazil. FFA residue was not found in the evaluated samples, indicating the correct use of antimicrobials within the sampled region. Acknowledgments The authors would like to thank São Paulo Research Foundation (FAPESP), Funding of Studies and Projects (FINEP) and National Council for Scientific and Technological Development (CNPq) for financial support. Authors are also grateful to Institute of Food Technology (ITAL) for facility availability. The project was co-financed by external source − Macroprogram 2 of Embrapa, subsidized to the Ministry of Fisheries and Aquaculture (MPA), Brazil. Title: Development of Monitoring System for Aquaculture Environmental Management at Furnas Reservoir (MG) − Support for the consolidation of indicators for monitoring plan and environmental management of aquaculture. The authors would also like to thank the Writing Department/General Coordination of Unicamp, for the text translation. References [1] F. Garcia, D.M. Romera, K.S. Gozi, E.M. Onaka, F.S. Fonseca, S.H.C. Schalch, P.G. Candeira, L.O.M. Guerra, F.J. Carmo, D.J. Carneiro, M.I.E.G. Martins, M.C. Portella, Aquaculture 410 (2013) 51–56. [2] R. Hirsch, T. Ternes, K. Haberer, K.L. Kratz, Sci. Total Environ. 225 (1999) 109–118. [3] S.H. Monteiro, J.G. Francisco, T.F. Campion, R.F. Pimpinato, G.C.R. Moura Andrade, F. Garcia, V.L. Tornisielo, Aquaculture 1 (447) (2015) 37–43. [4] F. Canada, M. Pena, A.A.E. Mansilla, Anal. Bioanal. Chem. 395 (2009) 987–1008.

E.A. Orlando et al. / J. Chromatogr. B 1035 (2016) 8–15 [5] P. Hernández Serrano, Responsible use of antibiotics in aquaculture, in: FAO Fisheries Technical Paper. No. 469, FAO, Rome, 2005, 97p. [6] V.F. Samanidou, E.N. Evaggelopoulou, J. Sep. Sci. 30 (2007) 2549–2569. [7] -R. Alexy, A. Schöll, T. Kümpel, K. Kümmerer, What do we know about antibiotics in the environment? in: K. Kümmerer (Ed.), Pharmaceuticals in The Environment, Springer-Verlag, Berlin, 2004, pp. 209–221. [8] T.F. Oliveira, Therapeutic Efficacy of Florfenicol Against Streptoccoccus Agalactiae Infection in Nile Tilapia (Oreochromis Niloticus), Universidade Federal de Vic¸osa, Belo Horizonte, MG, 2016, pp. 40p. [9] Brasil, Ministério da Agricultura, Pecuária e Abastecimento, INSTRUC¸ÃO NORMATIVA No. 17, DE 29 MAIO DE 2013, Anexo I. 2013. [10] C. Wrzesinski, L. Crouch, P. Gaunt, D. Holifield, N. Bertrand, R. Endris, Aquaculture 253 (2006) 309–316. [11] Florfenicol (Extension to Fish) Summary Report 5, European Medicines Agency (EMEA), Committee for Veterinary Medicinal Products, 2000, http:// www.ema.europa.eu/docs/en GB/document library/Maximum Residue Limits - Report/2009/11/WC500014280.pdf (accessed 11.04.2015). [12] S. Zhang, Z. Liu, X. Guo, L. Cheng, Z. Wang, J. Shen, J. Chromatogr. B 875 (2008) 399–404. [13] J. Van de Riet, R. Potter, M. Christie-Fougere, B. Burns, J. AOAC Int. 86 (3) (2003) 510–514. [14] C. Wrzesinski, L. Crouch, R. Endris, J. AOAC Int. 86 (3) (2003) 515–520. [15] C. Vue, L. Schmdit, G. Stehly, W. Gingerich, J. Chromatogr. B 780 (2002) 111–117. [16] J. Wang, B. Sherri Turnipseed, Chemical analysis: quantitative and confirmatory methods, in: J. Wang, J.D. MacNeil, J.F. Kay (Eds.), Chemical Analysis of Antibiotic Residues in Food, John Wiley & Sons, Inc, New Jersey, 2012, pp. 187–225.

15

[17] R.W. Fedeniuk, M. Mizuno, C. Neiser, C. O’Byrne, J. Chromatogr. B 991 (2015) 68–78. [18] J. Shen, X. Xia, H. Jiang, C. Li, J. Li, X. Li, S. Ding, J. Chromatogr. B 877 (2009) 1523–1529. [19] K. Xie, L. Jia, Y. Yao, D. Xu, S. Chen, X. Xie, Y. Pei, W. Bao, G. Dai, J. Wang, Z. Liu, J. Chromatogr. B 879 (2011) 2351–2354. [20] M.R. Rezk, S.M. Riad, F.I. Khattab, H.M. Marzouk, J. Chromatogr. B 978–979 (2015) 103–110. [21] P. Luo, X. Chen, C. Lian, H. Kuang, L. Lu, Z. Jiang, Z. Wang, C. Li, S. Zhang, J. Shen, J. Chromatog. B 878 (2010) 207–212. [22] T.L. Fodey, S.E. George, I.M. Traynor, P. Delahaut, D.G. Kennedy, C.T. Elliott, S.R.H. Crooks, J. Immunol. Methods 393 (2013) 30–37. [23] X. Pan, P. Wu, W. Jiang, B. Ma, Food Control 52 (2015) 34–38. [24] S. Sadeghi, M. Jahani, Food Chem. 141 (2013) 1242–1251. [25] R.F. Teófilo, M.C. Ferreira, Quím. Nova 29 (2) (2006) 338–350. [26] Instituto Nacional de Metrologia, Qualidade e Tecnologia (INMETRO). DOQ-CGCRE-008: Orientac¸ão sobre Validac¸ão de Métodos Analíticos, Rev. 04, 2011. [27] E. Alechaga, E. Moyano, M.T. Galceran, Analyst 137 (2012) 2486–2494. [28] R.P. Lopes, R.C. Reyes, R. Romero-González, J.L.M. Vidal, A.G. Frenich, J. Chromatogr. B 895–896 (2012) 39–47. [29] E.A. Orlando, A.V.C. Simionato, J. Chromatogr. A 1307 (2013) 111–118. [30] A. Freitas, S. Leston, J. Rosa, M.C. Castilho, J. Barbosa, P. Rema, M. Â. Pardal, F. Ramos, Food Addit. Contam. Part A 31 (5) (2014) 817–826.