Selected applications

3.1.6.1 The determination of PCBs in foodstuffs. In the European Community, 1999 will be remembered as the year of the Belgian dioxin crisis. The dioxins (mainly polychlorinated dibenzofurans or PCDFs) originated from the presence of used transformer oil in animal feed and many thousands of food samples had to be analysed for PCBs. During the crisis, the sample preparation technique applied at the Research Institute for Chromatography, Kortrijk, Belgium, was continuously updated and a high-throughput method based on UE was developed. The implementation of the method had to be accompanied with validation studies and the analysis of certified samples.

The analytical scheme of the official method Beltest I 014 for the analysis of PCBs in food consists of different steps: sample drying, extraction, clean-up and CGC analysis. Sample drying can be performed by freeze-drying or chemically by the addition of sodium sulphate. In the second step, the lipophilic contaminants are extracted from the matrix using an apolar solvent. The extract contains the lipids, PCBs, PCDDs, PCDFs and other apolar solutes such as organochloro-pesticides (OCPs), PAHs and mineral oil. Different extraction techniques may be applied namely Soxhlet extraction (or the automated versions Soxtec or Soxtherm), UE, ASE, MASE and SFE. All these techniques perform equally well for the extraction of fat and PCBs as will be illustrated further. It is obvious that especially in a crisis situation the selection should be based on sample throughput and cost. Next the PCBs are fractionated from the (co-extracted) fat matrix. For this fractionation, column chromatography on acidic silica gel and aluminium oxide is advised although other techniques such as gel permeation chromatography (GPC) and SPE may be applied if validated. Both sample extraction and clean-up require a concentration step. Finally, the cleaned extract is analysed by CGC-ECD and positive samples are confirmed by capillary gas chromatography with mass selective detection (CGC-MS).

UE and clean-up by a dispersive solid phase extraction (DSPE) technique nowadays also applied in the QuEChERS method, was performed as follows. Samples were homogenised using a blender. From fat samples (chicken or pork fat) 1 g sample was weighed in a 20-mL headspace vial. From eggs, 3 g egg yolk was taken. For animal feed samples or other meat products, a sample size corresponding to 200-500 mg fat was taken. To the sample, 2 g anhydrous sodium sulphate and 10 mL petroleum ether were added. Tetrachloronaphthalene or octachloronaphthalene may be added as internal standard to the sample at this stage although external standardization can also be applied. The headspace vial is closed and placed in an ultrasonic bath at 30°C for 30 min. In this step, the fat and PCBs are transferred from the matrix in the petroleum ether phase. The sodium sulphate adsorbs the water present in the sample. After extraction, the sample is allowed to settle. An aliquot (typically 5 mL) is transferred to a test tube and another aliquot (2 mL) is used to determine gravimetrically the fat content of the extract. To the test tube, 2 g of acidic silica gel (44% sulphuric acid) is added and the tube is placed in an ultrasonic bath for 30 min. This technique is nowadays called DSPE. In column chromatography and in SPE, the analytes are eluted through a bed and the fat is retained. In DSPE, the fat matrix is allowed to bind on the adsorbent that is mixed with the sample, while the solutes of interest stay in solution. For the fractionation of fat from PCBs, this method works very efficiently. After settlement of the adsorbent or centrifugation, an aliquot of the clear solution is transferred to an autosampler vial. The final volume is not important, since the concentrations are calculated to the initial 10 mL solvent and the sample or fat weight. The whole sample preparation takes approximately one hour and several samples can be prepared in parallel. One technician can handle more than 50 samples per day.

The extracts are analysed by CGC-ECD and/or CGC-MS. For MS detection, SIM mode is performed on two ions per congener group. The limit of detection (S/N>3) of the micro-ECD and the MS were 0.2 and 0.5 pg, respectively.

Figure 5 shows the RTL-CGC-ECD profiles of Aroclor 1260, egg, pork fat and mink fat recorded with a time interval of nearly one month. PCB congeners 153, 138 and 180 are the pre-dominant congeners.

The chromatogram of the mink fat extract is interesting. The minks were fed with contaminated eggs and although the same PCB profile (mainly Arochlor 1260) is present, some differences are noted in the relative abundances of the PCB congeners. This can be explained by a different metabolism between the animal species. In the mink sample, the PCB concentration measured as the sum of the seven congeners was as high as 25ppm (25mg/kg fat). This concentration was fatal for most minks.

Some 4,000 samples, including animal feed, eggs, chicken fat, pork fat, pork meat, meat products (ham, sausages), etc. were analysed using this methodology. From a practical point of view, the splitless liner was replaced after 100 analyses and the column after 1,000 injections. On the column selected and with the

2000

1000 0

6000 4000 2000 0

6000 4000 2000 0

15000 10000 5000 0

PCB 153

PCB 138

PCB 180

PCB 153

PCB 138

6000 4000 2000 0

15000 10000 5000 0

B

A

, ..j . »1 . J

^ n ..AAA, „ aA" - ■ i 1 . . .

C

A

A . iL «1 - j

i^JUA-jJ/Lou

«A . .. A

D

»

. A J

L^-AJIA_»AAA.

L_l^A___«_h_

Time (min)

Figure 5 Retention time locked CGC-ECD analyses of Aroclor 1260 (A), egg extract (B), pork fat extract (C), mink fat extract (D). Analyses were performed on a HP 6890 GC (Agilent) equipped with split/splitless inlet and micro-ECD detection. Separations were carried out on a 30-m L x 0.25-mm ID x 0.25 mm df HP-5MS column (Agilent). Injection of 1 mL was done in the splitless mode at 250°C. The carrier gas was hydrogen at an initial head pressure of 71 kPa hydrogen. The pressure was adjusted via the RTL software to obtain a retention time of 26.999 min for p,p-DDT. The oven was programmed from 70°C (2 min) to 150°C at 25°C/min, to 200°C at 3°C/min and to 300°C (2 min) at 8°C/min. Nitrogen at 40mL/min was used as detector make-up gas. The detector was set at 320°C. Reprinted from Ref [52] with permission from Elsevier.

Time (min)

Figure 5 Retention time locked CGC-ECD analyses of Aroclor 1260 (A), egg extract (B), pork fat extract (C), mink fat extract (D). Analyses were performed on a HP 6890 GC (Agilent) equipped with split/splitless inlet and micro-ECD detection. Separations were carried out on a 30-m L x 0.25-mm ID x 0.25 mm df HP-5MS column (Agilent). Injection of 1 mL was done in the splitless mode at 250°C. The carrier gas was hydrogen at an initial head pressure of 71 kPa hydrogen. The pressure was adjusted via the RTL software to obtain a retention time of 26.999 min for p,p-DDT. The oven was programmed from 70°C (2 min) to 150°C at 25°C/min, to 200°C at 3°C/min and to 300°C (2 min) at 8°C/min. Nitrogen at 40mL/min was used as detector make-up gas. The detector was set at 320°C. Reprinted from Ref [52] with permission from Elsevier.

chromatographic conditions applied the PCB congeners 28 and 31 are not separated but this was not critical as both congeners were not relevant for this PCB pollution.

The performance of UE was compared with ASE and MASE. The extraction efficiency of the three methods was evaluated with three egg samples contaminated at different levels (low, medium and high). For ASE, a Dionex ASE 200 system (Dionex Corp., Sunnyvale, CA, USA) was used. A 1-g sample was extracted at 100°C and 1,500 psi using petroleum ether as solvent. The extraction time was 5 min oven heat-up time, 5 min static extraction and 3 cycles with 60% of the extraction cell volume (22 mL). The extract was then concentrated to 10 mL. For MASE, an ETHOS SEL system (Milestone, Analis, Gent, Belgium) was applied. A 1-g sample was extracted after 20min at 95°C. The extraction solvent was n-hexane (10 mL in extraction thimble, 10 mL outside thimble) using a Weflon stir bar to absorb the microwave energy in combination with the non-microwave absorbing solvent. After extraction, the extract was filtered and concentrated to 10 mL to obtain the same final concentration factor as the UE and ASE. Clean-up and analysis was done in the same way for the three extracts of the three samples. The results based on duplicate analysis are summarized in Table 1.

For the sample with the lowest concentration (egg 1) the relative standard deviation (RSD) on the PCB sums obtained by the three techniques is 12%; for the two other samples the RSDs are less than 6%. For the individual values, some small differences are noted but in general these differences are within 10% of the average values. These results clearly demonstrate that there is no statistically significant difference between the three techniques and that equally good results are obtained. The ultrasonic method exhibits by far the highest throughput and is extremely cheap compared to ASE and MASE.

The repeatability of the method (n — 6) was evaluated by the analysis of a contaminated egg and fat sample (Table 2). These samples were distributed in a round robin test for Belgian laboratories during the crisis. The RSDs are all below 10% for the congeners PCB 118,153,138 and 180 and below 5% for the sum of the congeners.

The linearity and method sensitivity were determined by spiking a blank pork fat sample at 6 levels with the individual PCB congeners. The spike levels were 5, 10, 25, 50,100 and 200 ppb (ng/g fat) per congener. The recovery was determined versus an external standard and the linearity was measured by plotting the

Table 1 Comparison of UE, ASE and MASE for PCB enrichment

PCB

Egg 1 (ppb)

Egg 2 (ppb)

Egg 3 (ppb)

UE

MASE

ASE

UE

MASE

ASE

UE

MASE

ASE

118

170

68

147

556

410

550

945

1,023

922

153

263

309

210

1,142

1,051

1,184

1,811

2,032

2,042

138

240

320

234

1,019

1,120

1,105

2,031

2,323

2,128

180

111

166

93

696

552

686

1,015

1,166

1,259

Sum

783

863

682

3,412

3,133

3,525

5,803

6,544

6,349

Table 2 Repeatability for a contaminated egg and fat sample (1999 Belgian dioxin crisis)

PCB Egg sample Fat sample

Concentration (ppb) RSD (%) Concentration (ppb) RSD (%)

Table 2 Repeatability for a contaminated egg and fat sample (1999 Belgian dioxin crisis)

PCB Egg sample Fat sample

Concentration (ppb) RSD (%) Concentration (ppb) RSD (%)

118

141

7

20

4

153

333

2

274

6

138

342

4

318

3

180

165

5

165

4

Sum

982

2

776

4

Table 3

Figures of merit for the fast PCB method

CB

Mean % recovery

Linearity

S/N at 5 ppb

28

110

0.9999

9

52

88

0.9903

8

101

91

0.9991

12

118

105

0.9996

12

153

101

0.9994

13

138

104

0.9991

11

180

105

0.9998

18

absolute peak areas versus the spiked concentration. The signal-to-noise ratio was also measured at the lowest spiked concentration. The results are summarised in Table 3. All recoveries are within 88-110% for the individual congeners. The linearity is better than 0.995 except for PCB 52 for which an interfering peak was observed in the CGC-ECD trace. The signal-to-noise ratio was better than eight for all congeners at the 5 ppb level.

Lastly, the reproducibility and accuracy of the ultrasonic method was evaluated with the determination of the PCB content in two certified reference materials of the European Community namely the cod liver oil sample CRM 349 and the mackerel oil sample CRM 350 (IRMM, Geel, Belgium). The analyses of these reference materials were performed by four laboratories to which the method was transferred and having the same CGC-ECD and CGC-MS instrumentation. The results are summarised in Table 4 for cod liver oil and in Table 5 for mackerel oil. Most values are within 88% and 110% of the certified samples. The values outside these ranges are noted in italic. For all laboratories and for both techniques, the sum values were always between these limits.

The relatively long analysis times (see Figure 7) prompted us to evaluate fast high resolution capillary GC for PCB analysis [37]. This reduced the chromato-graphic analysis time to less than 8 min. Later, extracts were directly introduced in a mass spectrometer operated in the negative chemical ionization mode. This reduced the analysis time to less than 2 min [38].

Table 4

Accuracy and reproducibility test for the cod liver oil

sample

PCB

Certified Concentration (ppb)

Lab 1

Lab 2

Lab 3

Lab 4

ECD

MS

ECD

MS

ECD

MS

ECD

MS

28

68

61

65

73

64

104

68

75

70

52

149

126

165

141

164

144

199

159

148

101

370

333

394

385

404

296

437

356

373

118

454

390

467

421

448

479

508

397

458

153

938

975

989

886

1,010

790

1,030

810

1,016

180

280

252

295

283

312

270

326

288

273

Sum

2,259

2,137

2,375

2,189

2,402

2,083

2,568

2,085

2,338

Table 5 Accuracy and reproducibility test for the mackerel oil sample

Table 5 Accuracy and reproducibility test for the mackerel oil sample

PCB Certified Concentration (ppb) Lab 1 Lab 2 Lab 3 Lab 4

ECD

MS

ECD

MS

ECD

MS

ECD

MS

28

22.5

25

24

21

21

21

19

19

16

52

62

75

72

56

56

65

71

54

63

101

164

152

181

175

175

152

143

150

172

118

142

163

152

117

117

138

125

131

134

153

317

337

345

319

319

287

350

290

316

180

73

73

79

75

75

76

83

70

60

Sum

778.5

825

853

763

739

739

791

714

761

3.1.6.2 The determination of nitrofuran metabolites in scrimp and poultry.

Nitrofuran antibiotics have been widely used as food additives for treatment of Gram positive and Gram negative bacteria in poultry and fish. The parent compounds and metabolites are suspect carcinogens and they have been banned around the world. The Rapid Alert System for Food and Feed Annual report 2005 [39] shows that these compounds and metabolites continue to be detected in food samples thus remaining today a major concern in food safety. The nitrofuran antibacterial drugs furazolidone, furaltadone, nitrofurazone and nitrofurantoin have been found to metabolize rapidly and the metabolites bind to muscle tissue. The determination of metabolites and not the parent compounds is required in samples of animal origin. The structures of the parent compounds, the metabolites and the derivatives are shown in Figure 6.

The European Union set a minimum required performance level (MRPL) at 1 mg/kg for each metabolite [40]. Such levels can easily be determined using state-of-the-art MS systems (ion trap, triple quad and high resolution TOF) if an appropriate sample preparation technique is applied. Sample preparation should include a homogenization step, the release of the metabolites from the proteins, transformation into derivatives to facilitate chromatographic elution and UV

Furaltadone

AMOZ

NBA-AMOZ

TV ^nh2

Nitrofluranzone

O SEM

NBA-SEM

TV YNH

Nitrofurantoin

Furazolidone

2 rH

O AHD

NBA-AHD

NBA-AOZ

Figure 6 Structures of the nitrofuran antibiotics, metabolites and formed derivatives.

(the metabolites are not UV absorbing) or MS detection and transfer to an organic phase for analysis. Fortunately, the labelled standards NBA-d5AMOZ (used to quantify NBA-AMOZ) and NBA-d4AOZ (used to quantify the other metabolites) are commercially available simplifying drastically their determination.

A typical sample preparation protocol is as follows. One gram of homogenized tissue (e.g., poultry and shrimps) is mixed with 5 mL of a 0.2-M hydrochloric acid and 50 p.L of a 2-nitrobenzaldehyde (2-NBA) solution (100 mM in methanol) and incubated 16 h at 37°C. The mild acidic conditions release the metabolites via the imine bond from the tissue and the formed amino groups react with 2-NBA to form an aromatic imine bond. The acidic medium is neutralized with 500 p.L of a 0.3-M Na3PO4 solution in water and the pH is adjusted to 7 with a 2-M NaOH solution. The sample is twice extracted and centrifuged with 4 mL ethyl acetate and the combined extracts are evaporated to near dryness and reconstituted in 500 p.L of initial LC mobile phase. Reversed phase LC-MS (ITD, QqQ or HR-TOF) with electrospray ionization in the positive mode is applied. A typical LC-QqQ analysis is shown in Figure 7. A fish sample (genus Tilapia) was spiked with 0.5ng/g of the four metabolites and extracted as described [41].

0 0

Post a comment