7
HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds ² Ping He and Robert G Ackman* Canadian Institute of Fisheries Technology, DalTech, Dalhousie University, Halifax, Nova Scotia B3J 2X4, Canada Abstract: A new method was developed to determine 1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline (EQ, ethoxyquin), 2,6-dihydro-2,2,4-trimethyl-6-quinolone (QI) and 1,8-di(1,2-dihydro-6-ethoxy- 2,2,4-trimethylquinoline) (DM) in fish meals or fish feeds, QI and DM being the oxidation products of EQ. The sample was first extracted with hexane. After the removal of hexane the three analytes were extracted from the resulting oil with acetonitrile and determined by C 18 reverse phase high- performance liquid chromatography with a UV detector set at l = 280 nm. The mobile phase was acetonitrile–0.01 M ammonium acetate (80:20 v/v). The recoveries for EQ, QI and DM from the samples spiked at different levels varied in the ranges 90–100 per cent, 75–85 per cent and 90–100 per cent respectively. At room temperature, QI and DM were the major oxidation products of EQ in stored fish meals and fish feeds. Loss of EQ from fish meals is faster than that from feeds, resulting in relatively higher accumulations of QI and DM in the fish meals. Both QI and DM, especially the former, were not stable during storage of either and could be further oxidised to unidentified compounds. The residue levels of these two compounds were thus unpredictable during storage intervals. When the storage temperature was increased to 50 °C, EQ disappeared more rapidly, but neither QI nor DM accumulated. # 2000 Society of Chemical Industry Keywords: ethoxyquin; oxidation; fish meal; fish feed INTRODUCTION Aquaculture products now total a fourth of seafood consumed on a world basis, and most safety aspects have been considered in reviews by Howgate, 1 Dalsgaard 2 and Alderman and Hastings. 3 Metals and organochemicals, including pesticides and PCBs, are transferred from the environment into fish and are of special concern in fish from the Great Lakes of North America. 4,5 On the other hand, intensive marine fish farming now requires veterinary drugs, primarily anti- biotics such as oxytetracycline, in which case residues in fish muscle are severely limited by regulation. 6 This is a case where a small portion of an actual synthetic and biologically active chemical administered to the fish with their feeds is accumulated in the final food product without change. Public concern over chemicals in foods does not usually extend to synthetic antioxidants in animal feeds as distinct from human foods. Ethoxyquin (EQ, 1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline) is an effective antioxidant long used in fish meals and fish feeds. 7,8 It behaves in a sacrificial manner to provide protection for other components. Thus both EQ and its oxidation products may be present in fish meals and fish feeds. The original levels of EQ added to the feed ingredients may be known, but the accumulation of an unaltered residue of EQ or of the oxidation products when actually fed has not been recorded. Unlike the case of food additives, any carryover of these materials into farmed fish is not intentional and also apparently has not been examined. The matter was not even considered in an extensive review of health and safety. 1 As a prerequisite to an investigation of the accumu- lation of EQ and its products in animal food products we have focused on methods for the determination of EQ oxidation products in both the fish meal and compounded fish feeds used in salmon aquaculture. The oxidation products of EQ are numerous, and only some have been identified. 9–13 When EQ was added to oxidised fish meal or oxidised fish oil, a quinone imine compound (QI, 2,6-dihydro-2,2,4-trimethyl-6-quino- lone) and a dimer (DM, 1,8-di(1,2-dihydro-6-ethoxy- 2,2,4-trimethylquinoline)) were reported as the major oxidation products. 12 The structures are shown in Fig 1. Methods for determination of EQ itself in fruits, Journal of the Science of Food and Agriculture J Sci Food Agric 80:10–16 (2000) * Correspondence to: Robert G Ackman, Canadian Institute of Fisheries Technology, DalTech, Dalhousie University, Halifax, Nova Scotia B3J 2X4, Canada ² Presented in part at the 42nd Atlantic Fisheries Technological Conference, Newport, RI, 8–11 November 1997 Contract/grant sponsor: Natural Sciences and Engineering Research Council of Canada (Received 29 December 1998; revised version received 15 April 1999; accepted 15 July 1999) # 2000 Society of Chemical Industry. J Sci Food Agric 0022–5142/2000/$17.50 10

HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

Embed Size (px)

Citation preview

Page 1: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

HPLC determination of ethoxyquin and its majoroxidation products in fresh and stored fishmeals and fish feeds †

Ping He and Robert G Ackman*Canadian Institute of Fisheries Technology, DalTech, Dalhousie University, Halifax, Nova Scotia B3J 2X4, Canada

Abstract: A new method was developed to determine 1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline

(EQ, ethoxyquin), 2,6-dihydro-2,2,4-trimethyl-6-quinolone (QI) and 1,8'-di(1,2-dihydro-6-ethoxy-

2,2,4-trimethylquinoline) (DM) in ®sh meals or ®sh feeds, QI and DM being the oxidation products of

EQ. The sample was ®rst extracted with hexane. After the removal of hexane the three analytes were

extracted from the resulting oil with acetonitrile and determined by C18 reverse phase high-

performance liquid chromatography with a UV detector set at l=280nm. The mobile phase was

acetonitrile±0.01M ammonium acetate (80:20v/v). The recoveries for EQ, QI and DM from the samples

spiked at different levels varied in the ranges 90±100 per cent, 75±85 per cent and 90±100 per cent

respectively. At room temperature, QI and DM were the major oxidation products of EQ in stored ®sh

meals and ®sh feeds. Loss of EQ from ®sh meals is faster than that from feeds, resulting in relatively

higher accumulations of QI and DM in the ®sh meals. Both QI and DM, especially the former, were not

stable during storage of either and could be further oxidised to unidenti®ed compounds. The residue

levels of these two compounds were thus unpredictable during storage intervals. When the storage

temperature was increased to 50°C, EQ disappeared more rapidly, but neither QI nor DM

accumulated.

# 2000 Society of Chemical Industry

Keywords: ethoxyquin; oxidation; ®sh meal; ®sh feed

INTRODUCTIONAquaculture products now total a fourth of seafood

consumed on a world basis, and most safety aspects

have been considered in reviews by Howgate,1

Dalsgaard2 and Alderman and Hastings.3 Metals and

organochemicals, including pesticides and PCBs, are

transferred from the environment into ®sh and are of

special concern in ®sh from the Great Lakes of North

America.4,5 On the other hand, intensive marine ®sh

farming now requires veterinary drugs, primarily anti-

biotics such as oxytetracycline, in which case residues

in ®sh muscle are severely limited by regulation.6 This

is a case where a small portion of an actual synthetic

and biologically active chemical administered to the

®sh with their feeds is accumulated in the ®nal food

product without change.

Public concern over chemicals in foods does not

usually extend to synthetic antioxidants in animal

feeds as distinct from human foods. Ethoxyquin (EQ,

1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline) is an

effective antioxidant long used in ®sh meals and ®sh

feeds.7,8 It behaves in a sacri®cial manner to provide

protection for other components. Thus both EQ and

its oxidation products may be present in ®sh meals and

®sh feeds. The original levels of EQ added to the feed

ingredients may be known, but the accumulation of an

unaltered residue of EQ or of the oxidation products

when actually fed has not been recorded. Unlike the

case of food additives, any carryover of these materials

into farmed ®sh is not intentional and also apparently

has not been examined. The matter was not even

considered in an extensive review of health and safety.1

As a prerequisite to an investigation of the accumu-

lation of EQ and its products in animal food products

we have focused on methods for the determination of

EQ oxidation products in both the ®sh meal and

compounded ®sh feeds used in salmon aquaculture.

The oxidation products of EQ are numerous, and only

some have been identi®ed.9±13 When EQ was added to

oxidised ®sh meal or oxidised ®sh oil, a quinone imine

compound (QI, 2,6-dihydro-2,2,4-trimethyl-6-quino-

lone) and a dimer (DM, 1,8'-di(1,2-dihydro-6-ethoxy-

2,2,4-trimethylquinoline)) were reported as the major

oxidation products.12 The structures are shown in Fig

1.

Methods for determination of EQ itself in fruits,

Journal of the Science of Food and Agriculture J Sci Food Agric 80:10±16 (2000)

* Correspondence to: Robert G Ackman, Canadian Institute of Fisheries Technology, DalTech, Dalhousie University, Halifax, Nova ScotiaB3J 2X4, Canada† Presented in part at the 42nd Atlantic Fisheries Technological Conference, Newport, RI, 8–11 November 1997Contract/grant sponsor: Natural Sciences and Engineering Research Council of Canada(Received 29 December 1998; revised version received 15 April 1999; accepted 15 July 1999)

# 2000 Society of Chemical Industry. J Sci Food Agric 0022±5142/2000/$17.50 10

Page 2: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

spices, animal feeds and foodstuffs have appeared in

many publications.14±18 Techniques include spectro-

metry, thin layer chromatography (TLC), high-per-

formance thin layer chromatography (HPTLC), gas

chromatography (GC) and high-performance liquid

chromatography (HPLC). GC and HPLC are the

most important methods for the determination of EQ

residues. GC with a ¯ame ionisation detector (FID)

has both the advantage of very high sensitivity and the

disadvantage of possible decomposition of EQ under

high temperature in the column.12,19 In contrast with

GC, HPLC methods avoid EQ decomposition, but

usually the sensitivity is less. In one study to compare

the determination of EQ in ®sh meal by GC (FID),

HPLC (UV or ¯uorescence detection) and HPTLC

(¯uorescence detection) methods, the mean values

showed no signi®cant differences.20

Much less work has been done on developing

methods for the determination of EQ oxidation

products. So far, only one study has dealt with the

quantitative analysis of these compounds. De Koning

and van der Merwe21 developed a GC method which

used methoxyquin (MQ) and methoxyquin dimer

(MQDM) as the internal standards to determine the

amount of EQ, QI and DM in ®sh meal. This GC

method is somewhat time-consuming, because sepa-

rate runs are required for the different compounds.

In the present work we have developed a new

method using high-performance liquid chromatogra-

phy (HPLC) to determine these compounds in

commercial ®sh feeds, and monitored their tendency

to change in ®sh meals and ®sh feeds during storage.

EXPERIMENTALMaterialsFreshly produced herring ®sh meal untreated with EQ

was provided by Connors Bros Ltd (Blacks Harbour,

NB, Canada). After receipt in our laboratory a portion

of this ®sh meal was spiked with EQ22 to prepare

sample FM1. An EQ-treated ®sh meal (FM2) and two

®sh feeds (FF1 and FF2) were provided by EWOS

(Canada) Ltd (Surrey, BC, Canada). Four commer-

cial salmon feeds were obtained from Moore-Clark (St

Andrews, NB, Canada) and the ®fth was obtained

from Corey Feed Mill Ltd (Fredericton, NB,

Canada). The commercial ®sh feed samples were

stored in their original packages atÿ15 toÿ20°C after

receipt.

Pure EQ, QI and DM were prepared as standards as

described elsewhere.23 a-Tocopherol, BHA and BHT

were purchased from Sigma Chemical Co (St Louis,

MO, USA). Ascorbic acid was purchased from Nutri-

tional Biochemical Corp (New York, NY, USA).

TBHQ and a-tocopherol acetate were purchased from

Fluka (Buchs, Switzerland).

The concentrations of standard solutions were 1, 2,

4, 6, 10, 15, 20, 30 and 40gmlÿ1 for QI, 10, 20, 40,

60, 100, 150, 200, 300 and 400mgmlÿ1 for EQ and 10,

20, 40, 60, 100, 150, 200, 250 and 300mgmlÿ1 for

DM. All solvents were either HPLC grade or analytical

grade and were redistilled before use in our laboratory.

ApparatusHPLC was carried out with a Waters 6000A solvent

delivery system, a Waters U6K injector and a Waters

450 variable-wavelength UV detector (Waters, Mil-

ford, MA, USA). The mobile phase was acetonitrile±

0.01M CH3COONH4 (80:20v/v) at a ¯ow rate of 1ml

minÿ1. The column was a CSC-S ODS-2 column

(30cm�0.39cm, 10mm; CSC Inc, Montreal, PQ,

Canada) with a guard column (3cm; Waters, Milford,

MA, USA).

Gas chromatography±mass spectrometry (GC±MS)

was carried out with a PE 990 gas chromatograph

(Perkin-Elmer, Norwalk, CT, USA) ®tted with a

DB-1 fused silica capillary column (59m�0.25mm

id; Chromatographic Specialities Inc, Brockville, ON,

Canada) passed directly into a Finnigan 700 ion trap

detector (Finnigan Corp, San Jose, CA, USA) as the

mass spectrometer. The column temperature was

150°C for EQ and QI determinations and 260°C for

DM determination. Data acquisition was executed by

a computer program provided by Finnigan Corp (San

Jose, CA, USA).

UV absorption curvesThe standards were separately dissolved in a mixture

of acetonitrile and 0.01M CH3COONH4 (80:20v/v).

The solutions were scanned on an HP 8453 UV/VIS

spectrophotometer from 250 to 450nm for UV

absorption.

Sample analysisA sample of 1.0g of ®sh meal or ground ®sh feed and

10ml of hexane were added to a 15ml test tube. After

careful ¯ushing with N2, the tube was tightly capped,

vortexed for 1min and sonicated in a 40°C water bath

for 10min. Then it was vortexed for another 0.5min.

After centrifuging for 5min at 600�g, the clear hexane

layer was transferred to another tube. The extraction

was repeated twice with vortexing, but without the

sonicating procedure, with 5ml of fresh hexane each

time. The hexane solutions were combined and the

hexane was removed with a stream of nitrogen. The

®sh meal lipid extract was then extracted with aceto-

nitrile three times using 1ml each time. The combined

acetonitrile solutions were concentrated to exactly 1ml

before HPLC and GC±MS analysis.

Figure 1. Structures of (I) EQ, (II) QI and (III) DM.

J Sci Food Agric 80:10±16 (2000) 11

HPLC of ethoxyquin and two oxidation products in ®sh meals and feeds

Page 3: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

Recovery experimentsRecovery studies for EQ, QI and DM from ®sh meal,

®sh feed and ®sh tissue were carried out at different

levels. All amounts of standards were added to samples

just before ¯ushing with nitrogen so that oxidation was

minimised.

Storage studiesAll samples of ®sh meal or of ®sh feed were stored

respectively either at ambient temperature or at 50°Cin disposable aluminium foil pans open to the air.

Samples were examined at intervals of 1 month.

StatisticsAll statistical analyses were performed using the

Sigmaplot 3.0 software packages (SPSS Inc, Chicago,

IL, USA).

RESULTS AND DISCUSSIONQI and DM have been identi®ed as autoxidation

products of EQ in oxidised ®sh meal and ®sh oil.12

Two weaknesses may limit the comprehensiveness of

that work. Firstly, the samples were extracted with a

method using a chloroform-based solvent, which

probably introduced EQ oxidation products during

sample preparation. It is known that EQ oxidation is

extremely fast in chloroform.19 The colour of the

upper layer of the EQ±chloroform solution in a very

long and thin tube was observed to change from

colourless to brown in 5min in one of our experiments

owing to the contact with air and the exposure to light.

Secondly, in the last few decades, ¯ame driers have

gradually been replaced in modern plants with steam-

heated driers, reducing damage to the product.24 In

current manufacture, EQ is then added to freshly

produced ®sh meal before much lipid oxidation

happens.

To recon®rm the major autoxidation products of

EQ in ®sh meal, EQ was added to the freshly produced

herring ®sh meal. Solvents used to extract the analytes

were hexane and acetonitrile, both of which have

proven to be reliable solvents for EQ examination in

other studies.18,19 The sample preparation was per-

formed under the protection of nitrogen. No distinct

formation of DM and QI was found after the ®rst or

second month of storage at room temperature.

However, on the 90th day a clear formation of DM

and QI in ®sh meal was observed on HPLC (Fig 2).

The extract of the same sample was also examined by

GC±MS, and two small peaks were identi®ed as DM

and QI, since their MS spectra matched those of the

standards.

All the published HPLC methods for EQ determi-

nation are based on reverse phase C18 columns, since

EQ tails heavily on normal silica gel columns. Such

behaviour can also be observed on reverse phase

columns owing to the interactions between the amine

functional group and any free silanol groups of the

column packing. Mobile phases such as a mixture of

methanol or acetonitrile with water or diluted aqueous

ammonium acetate buffer effectively suppressed the

tailing phenomenon. In order to determine the

suspected major oxidation products of EQ (ie DM

and QI) with the antioxidant itself in one run, the

proportion of acetonitrile to buffer (0.01M ammonium

acetate solution) was optimised as 80:20 (v/v). The

three, QI, EQ and DM then eluted in that sequence. In

this solvent proportion, DM elutes in a reasonable

time and there is a good separation between QI and

EQ.

The UV absorption curves of the three highly

puri®ed compounds in the mobile phase mixture

showed that the most stable and highest absorption

was located at 360nm for EQ, 280nm for QI and

375nm for DM. Since ®sh meal normally contains a

very low level of QI but a large amount of EQ,21 the

detection wavelength at 280nm was selected to

provide high sensitivity for QI. One can argue that a

larger error in measuring EQ and DM could be

expected with a small variation of the detection

wavelength. Nevertheless, the accuracy of the modern

instrument is generally reliable, and we experienced no

dif®culty in obtaining very reproducible determination

results in all experiments.

This HPLC system proved able to successfully

separate other commonly used antioxidants from EQ,

DM and QI (Fig 3). Ascorbic acid (Vc), TBHQ and

BHA eluted before QI. BHT eluted neatly between

EQ and DM. a-Tocopherol and a-tocopherol acetate

did not elute in 30min in this system. Thus, even if a

sample contains the above antioxidants, they will not

disturb the determination of EQ and the two oxidation

products.

Unlike the GC method21 which requires two

Figure 2. Confirmation of QI and DM as major oxidation products of EQ infish meal during storage on reverse phase HPLC: A, fish meal without EQ;B, fish meal with added EQ, day 0;C, fish meal with added EQ, day 90.

12 J Sci Food Agric 80:10±16 (2000)

P He, RG Ackman

Page 4: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

separate runs for EQ and its oxidation products, the

present HPLC method can analyse the three com-

pounds in one run. The analysis time required to ®nish

one run is short, only about 20min. In addition, unlike

GC, HPLC does not produce any possible decom-

position products during analysis, since HPLC is

conducted at room temperature.

To evaluate the ef®ciency of extraction, the recovery

tests from the spiked samples (Table 1) were carried

out at three levels for each compound. The recoveries

for all three levels were 89±104%, 91±106% and 73±

89% for EQ, DM and QI respectively. The average

recoveries were 95% for EQ, 98% for DM and 83% for

QI, which were higher than or similar to the reported

recovery results from the application of the GC

method,21 in which the recoveries were 102%, 98%

and 55% for EQ, DM and QI respectively. The

method detection limits for HPLC analyses were

down to 5mgkgÿ1 for EQ, 5mgkgÿ1 for DM and

0.5mgkgÿ1 for QI, which were also similar to those of

the GC method. Thus the HPLC method and a

simpler extraction procedure can be used as an

alternative to the published GC method. Although

the detection limits restrict its use for samples contain-

ing very low levels of EQ and the oxidation products,

the method provides enough sensitivity for most ®sh

meal and ®sh feed samples.

Five commercial ®sh feed products were examined

by the method developed. Four of them contained EQ,

DM and QI, while the ®fth, from another manufac-

turer, contained traces of EQ but no DM and no QI

(Table 2). Health Canada speci®es 150mgkgÿ1 as the

maximum level of EQ that can be added to animal

feeds, and all the samples when analysed were far

below the legal limit. The legal limits of DM and QI

were not reported in the source consulted.

Among the four products from the same company

the EQ level varied from 17 to 58mgkgÿ1, the DM

level varied from 29 to 44mgkgÿ1 and the QI level

varied from 2.8 to 14.2mgkgÿ1. The levels of EQ and

its two oxidation products were probably dependent

on the formula proportion of, and the original residue

levels in, the two raw materials, ®sh meal and ®sh oil,

which likely contained these compounds. Fish meal in

North America usually contains 200mgkgÿ1 EQ or

less (SP Lall, personal communication), and the

normal ®sh feed formula contains up to 50% ®sh meal

and 10% ®sh oil. Fish oil may or may not be protected

by antioxidants, but EQ would be likely for this

purpose.

Since the oxidation level and the amount of EQ in

the raw materials were not known, it was impossible to

distinguish how much QI and DM were formed

through EQ oxidation after feed manufacture and

how much QI and DM already existed at the

beginning of production of these feeds. In our samples

it is reasonable to consider that most of the QI and DM

residues must have come from the raw materials, since

the samples were from recent production and EQ

oxidation would be very slow in our hands, where the

feeds were stored below ÿ15 to ÿ20°C. Finally, it was

also surprising to ®nd only a very small amount of EQ

Figure 3. Separation of other antioxidants with QI, EQ and DM on reversephase HPLC.

Table 1. Recoveries (%) of EQ, DM and QI from spiked fishmeala

Concentration EQ DM QI

Lowb 98.5�5.1 100.8�3.9 78.4�5.4

Moderatec 94.1�1.9 96.6�2.7 85.9�2.0

Highd 93.0�3.9 95.8�3.5 84.4�2.6

a Mean�sd (n =8).b Low concentration: 40mgkgÿ1 EQ/20mgkgÿ1 DM/3mgkgÿ1

QI.c Moderate concentration: 200mgkgÿ1 EQ/100mgkgÿ1 DM/

15mgkgÿ1 QI.d High concentration: 400mgkgÿ1 EQ/200mgkgÿ1 DM/

30mgkgÿ1 QI.

Table 2. EQ and its oxidation products (mgkgÿ1) in commer-cial fish feedsa,b

Sample ID EQ DM QI

Moore-Clark #1 58�2.7 44�1.5 14.2�0.5

Moore-Clark #2 52�1.4 30�1.2 8.0�0.5

Moore-Clark #3 28�0.8 29�0.3 3.4�0.1

Moore-Clark #4 17�0.4 36�0.3 2.8�0.1

Corey 6�0.3 BDL BDL

a Mean�sd (n =3).b BDL, below detection limit (0.5mgkgÿ1 for QI, 5mgkgÿ1 for

EQ and 5mgkgÿ1 for DM).

J Sci Food Agric 80:10±16 (2000) 13

HPLC of ethoxyquin and two oxidation products in ®sh meals and feeds

Page 5: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

occurring in some commercial products, such as the

Corey feed sample. However, a successful feed that

will not introduce EQ and its oxidation products into

cultivated ®sh may be eventually desirable.

The 4 month storage experiment was carried out to

study the change of EQ, DM and QI in ®sh meal and

®sh feed exposed to air under two different storage

temperatures, ie ambient temperature and 50°C. The

reason for storing the samples at 50°C was to

accelerate the oxidation rate of the lipids under

excessively harsh conditions. Samples were examined

for the contents of all three at 1 month intervals. The

results are listed in Tables 3±5 for EQ, DM and QI

respectively. Except for FM1, to which EQ was added

in our laboratory just before storage, all other samples

already contained DM and QI at the beginning of the

experiment. QI in FM1 at the ®rst analysis was due to

QI as an impurity in the EQ added to the sample.

One of our earlier hypotheses was that EQ oxidisa-

tion would be very slow when the feeds were kept

under ÿ15 to ÿ20°C. Almost no decrease of EQ in

FF1 and FF2 during the 4 month open storage at

room temperature was, however, unexpected. Only a

small decrease (about 30mgkgÿ1, less than 10% of the

starting level) in ®sh meals was observed under the

same conditions. The slight increase (not signi®cantly

Table 3. EQ contents (mgkgÿ1) in fish meal (FM)and feed (FF) during storagea,b

Month

Sample and condition 0 1 2 3 4

FM1, room temp 360�16a 342�10ab 337�22ab 334�4ab 327�3b

FM2, room temp 262�5a 253�7ab 242�1b 233�1c 237�0d

FM1, 50°C 360�16a 269�2b 220�5c 191�7d 189�4b

FM2, 50°C 262�5a 192�9b 171�0c 111�3d 74�1c

FF1, room temp 197�7ab 202�1a 197�6ab 190�2b 188�1b

FF2, room temp 174�6abc 183�2a 183�1a 174�1b 168�3c

FF1, 50°C 197�7a 177�6b 151�1c 144�0d 114�2e

FF2, 50°C 174�6a 167�9ab 154�2b 132�5c 104�4d

a Mean�sd (n =3).b Numbers in one row with the same superscript letter are not signi®cantly different (P<0.01).

Table 4. DM contents (mgkgÿ1) in fish meal (FM)and feed (FF) during storagea,b,c

Month

Sample and condition 0 1 2 3 4

FM1, room temp BDLd BDLd 5�0.2a 4�0.2a 9�0.4b

FM2, room temp 6�0.2a 8�0.3b 10�1.0c 9�0.3c 12�1.0d

FM1, 50°C BDLd BDLd BDLd BDLd BDLd

FM2, 50°C 6�0.2a 6�0.8ab 8�0.1b 7�0.3b 6�0.4ab

FF1, room temp 14�0.4a 16�0.6b 15�1.0ab 13�1.2a 17�0.9b

FF2, room temp 8�0.2a 9�0.5b 9�0.4b 9�0.3b 10�1.0b

FF1, 50°C 14�0.4a 14�0.6a 14�1.0ab 13�0.4bc 12�0.3c

FF2, 50°C 8�0.2a 9�0.6b 8�0.5ab 7�0.7a 8�0.6ab

a Mean�sd (n =3).b Numbers in one row with the same superscript letter are not signi®cantly different (P<0.01).c BDL, below detection limit (<5mgkgÿ1).

Table 5. QI contents (mgkgÿ1) in fish meal (FM)and feed (FF) during storagea,b,c

Month

Sample and condition 0 1 2 3 4

FM1, room temp 2.1�0.1a 1.5�0.1b 3.6�0.4cd 3.2�0.1c 3.7�0.1d

FM2, room temp 2.7�0.2a 1.3�0.1b 3.2�0.2c 1.3�0.1bd 1.5�0.1d

FM1, 50°C 2.1�0.1a 1.6�0.1b 1.6�0.1b 1.3�0.1c 1.1�0.1c

FM2, 50°C 2.7�0.2a 1.4�0.1b 1.5�0c 1.1�0d BDLe

FF1, room temp 1.9�0.1a 1.1�0.1b 0.9�0.1bc 0.8�0c 0.8�0c

FF2, room temp 2.4�0.2a 1.0�0b 0.8�0.2bc 0.8�0c 0.7�0.2bc

FF1, 50°C 1.9�0.1a 1.5�0b BDLe BDLe BDLe

FF2, 50°C 2.4�0.2a 1.5�0.1b BDLe BDLe BDLe

a Mean�sd (n =3).b Numbers in one row with the same superscript letter are not signi®cantly different (P<0.01).c BDL, below detection limit (<0.5mgkgÿ1).

14 J Sci Food Agric 80:10±16 (2000)

P He, RG Ackman

Page 6: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

different, P>0.01) of EQ in ®sh feeds at the second

and third determinations was probably due to experi-

mental error and continued sample moisture loss. The

larger variation of the FM1 triplicates (in comparison

with FM2, FF1 and FF2) almost certainly resulted

from the uneven distribution of EQ in the sample.

It is considered good practice to let the freshly

prepared meal cool to below 60°C before adding

antioxidant.24 The decreases of EQ in both ®sh meals

and ®sh feeds were in fact remarkable when the storage

temperature was elevated to 50°C. Over 40% of the

EQ starting level was lost (even 71% for one ®sh meal

sample) during storage. The rate of decrease for EQ in

®sh meal was faster than that in ®sh feed, with a loss of

over 170mgkgÿ1 EQ in each ®sh meal sample and 70±

83mgkgÿ1 EQ in each ®sh feed sample over 4 months.

The loose structure of the meal, versus the ®rm feed

pellet, is the probable explanation.

The loss of EQ should result from oxidation,

thermal decomposition and evaporation, among which

oxidation is the major pathway. Under certain condi-

tions, such as high storage temperature, evaporation

should not be neglected for EQ disappearance, since

we did not observe the accumulation of the oxidation

products or other decomposition products concurrent

with the loss of EQ. Another hypothesis is that EQ

interacts easily and more tightly with proteins under

high temperature so that it could not be extracted with

hexane.

The overall tendency of DM level to change in ®sh

meals and ®sh feeds at room temperature resulted in

an increase with the storage time, but this increase may

not be continuous. For example, DM in FM2 was

lower at the end of the third month than at the end of

the second month. In the FF1 sample, DM dropped

from 16mgkgÿ1 (end of the ®rst month) to 13mgkgÿ1

(end of the third month) and then rose to 17mgkgÿ1

(end of the fourth month). The total increase of DM in

®sh meal, more than 8mgkgÿ1 for FM1 and 6mgkgÿ1

for FM2, was two times larger than the 2±3mgkgÿ1

increase in both ®sh feed samples. The lower DM

formation rate in the feed samples suggested that EQ

oxidation mainly happened at the surface of the

extruded feed pellet.

The DM level was, however, basically unchanged

during storage at 50°C. It was even slightly reduced in

one feed sample (FF1) at the end of storage. Although

sometimes a small increment could be observed during

the experiment, the overall pattern was totally different

from that of the sample held at room temperature.

There are two possible reasons for the above situation.

One is that high temperature accelerates the formation

of the secondary oxidation products of lipids, most of

which are aldehydes from decomposition of the

peroxides. Thus the existing lives of the peroxides

and the correspondingly peroxy radicals are shorter,

resulting in less DM formed and accumulated. Never-

theless, the excessively high storage temperature

clearly changed the oxidation process of EQ in ®sh

meal and ®sh feeds.

QI amounts in ®sh meal stored at ambient tem-

perature were very unpredictable. The level of QI rose

and then dropped by as much as 2mgkgÿ1 in two

consecutive months, with an initial value of

1.3mgkgÿ1. Such a large variation was not observed

in ®sh feeds, which showed a continuous pattern of

decrease during the whole storage period at room

temperature. When the samples were stored at 50°C,

QI had a tendency to decrease in both ®sh meal and

®sh feed. The latter was quicker, resulting in QI in

feeds dropping from 2.0±2.5mgkgÿ1 to almost zero in

2 months. The reason may be the accelerated rate of

further oxidation of QI at high temperature. The

experimental results also showed that QI could only

accumulate in very small amounts, the highest level

being 3.7mgkgÿ1 in ®sh meal.

It is interesting to compare our short-term (4

months, room temperature) storage results with a

long-term (over 1 year, 25°C) storage study which

used anchovy ®sh meal samples.21 The average loss

rate of EQ was not consistent among the four samples

in the long-term study, but all losses were much larger

than that observed in the present study (30mgkgÿ1 in

4 months). The tendencies of the changes of DM and

QI were both completely unpredictable in that

experiment. The maximum accumulations of DM

and QI even reached as high as 124 and 29mgkgÿ1

respectively. It seems likely that the species of the ®sh

and natural antioxidants, especially in the oil, in¯u-

ence EQ oxidation during storage. Furthermore, we

cannot exclude the potential effect of the storage

methods, since the storage conditions for the ®sh meal

samples were not described in the South African study.

In summary, EQ in ®sh meals and ®sh feeds does

not disappear quickly under normal storage condi-

tions, especially in the feed. This is evidence that the

®sh feed quality can be retained during feeding, as

most ®sh feeds are used within 6 months of manu-

facture. It is recommended that ®sh feeds should be

stored at lower temperatures and kept in the dark if

conditions permit. The major oxidation products of

EQ in ®sh meal and ®sh feed are DM and QI under

normal conditions, but the ®nal oxidation products

under high storage temperature are still not under-

stood.

ACKNOWLEDGEMENTSFunding for the study, provided by the Natural

Sciences and Engineering Research Council of

Canada through a research grant awarded to Dr

Ackman, is gratefully acknowledged. The authors

would like to thank EWOS (Canada) Ltd and Connors

Bros Ltd for providing the ®sh meal and ®sh feed

samples.

REFERENCES1 Howgate P, Review of the public health safety of products from

aquaculture. Int J Food Sci Technol 33:99±125 (1998).

J Sci Food Agric 80:10±16 (2000) 15

HPLC of ethoxyquin and two oxidation products in ®sh meals and feeds

Page 7: HPLC determination of ethoxyquin and its major oxidation products in fresh and stored fish meals and fish feeds

2 Dalsgaard A, The occurrence of human pathogenic Vibrio spp. and

Salmonella in aquaculture. Int J Food Sci Technol 33:127±138

(1998).

3 Alderman DJ and Hastings TS, Antibiotic use in aquaculture:

development of antibiotic resistanceÐpotential for consumer

health risks. Int J Food Sci Technol 33:139±155 (1998).

4 Zabik ME, Zabik MJ, Booren AM, Nettles M, Song J-H, Welch R

and Humphrey H, Pesticides and total polychlorinated biphe-

nyls in chinook salmon and carp harvested from the Great

Lakes: effects of skin-on and skin-off processing and selected

cooking methods. J Agric Food Chem 43:993±1001 (1995).

5 Zabik ME, Booren A, Zabik MJ, Welch R and Humphrey H,

Pesticide residues, PCBs and PAHs in baked, charbroiled, salt

boiled and smoked Great Lakes lake trout. Food Chem 55:231±

239 (1996).

6 Huang TS, Du WX, Marshall MR and Wei CI, Determination of

oxytetracycline in raw and cooked channel cat®sh by capillary

electrophoresis. J Agric Food Chem 45:2602±2605 (1997).

7 De Koning AJ and Mol TH, Lipid determination in ®sh meal: an

investigation of three standard methods applied to stabilised and

non-stabilised anchovy meals at increasing stages of maturity. J

Sci Food Agric 46:259±266 (1989).

8 De Koning AJ, A new method for measuring ef®cacies of

antioxidants in ®sh meal. Int J Food Properties 1:255±261 (1998).

9 Lin JS and Olcott HS, Ethoxyquin nitroxide. J Agric Food Chem

23:798±800 (1975).

10 Bharucha KR, Cross CK and Rubin LJ, Nitroxides derived from

ethoxyquin and dihydroethoxyquin as potent anti-nitrosamine

agents for bacon. J Agric Food Chem 35:915±917 (1987).

11 Taimr L and PrusõÂkova M, Chromatographic behaviour of the

antioxidant ethoxyquin and its transformation products. J

Chromatogr 587:355±358 (1991).

12 Thorisson S, Gunstone FD and Hardy R, Some oxidation

products of ethoxyquin including those found in autoxidising

system. Chem Phys Lipids 60:263±271 (1992).

13 Kriz J and Taimr L, A new dimeric derivative of ethoxyquin from

its reaction with alkylperoxy radicals. Collect Czech Chem

Commun 58:1914±1918 (1993).

14 Winell B, Quantitative determination of ethoxyquin in apples by

gas chromatography. Analyst 101:883±886 (1976).

15 Van Peteghem CH and Dekeyser DA, Systematic identi®cation

of antioxidants in lard, shortenings, and vegetable oils by thin

layer chromatography. J Assoc Off Anal Chem 64:1331±1335

(1981).

16 Perfetti GA, Joe FL and Fazio T, Reverse phase high pressure

liquid chromatography and ¯uorescence detection of ethoxy-

quin in milk. J Assoc Off Anal Chem 66:1143±1147 (1983).

17 Uchiyama S and Uchiyama M, Thin-layer chromatography

¯uorometry of ethoxyquin using triton X-100. J Chromatogr

262:340±345 (1983).

18 Schreier C and Greene RJ, Determination of ethoxyquin in feeds

by liquid chromatography: collaborative study. J AOAC Int

80:725±731 (1997).

19 Skaare JU and Dahle HK, Gas chromatographic±mass spectro-

metric studies of ethoxyquin in some organic solvents. I. J

Agric Food Chem 23:1091±1093 (1975).

20 Vega M, Saelzer R and Mendoza N, Analysis of ethoxyquin

(6-ethoxy-1,2-dihydro-2,2,4-trimethylquinoline) in ®sh meal.

Comparison between HPLC, GC, and HPTLC. J High Resol

Chromatogr 13:854±855 (1990).

21 De Koning AJ and van der Merwe G, Determination of

ethoxyquin and two of its oxidation products in ®sh meal by

gas chromatography. Analyst 117:1571±1576 (1992).

22 Spark AA, Ethoxyquin in ®sh meal. J Am Oil Chem Soc 59:185±

188 (1982).

23 He P and Ackman RG, Puri®cation of ethoxyquin and its two

oxidation products. Food Chemistry submitted (1999).

24 Bimbo AP, Technology of production and industrial utilization

of marine oils. In Marine Biogenic Lipids, Fats, and Oils, Vol II,

Ed by Ackman RG, CRC Press, Boca Raton, FL, pp 401±434

(1989).

16 J Sci Food Agric 80:10±16 (2000)

P He, RG Ackman