Fish oil supplements in New Zealand are highly oxidised and do not meet label content of n-3 PUFA

We evaluated the quality and content of fish oil supplements in New Zealand. All encapsulated fish oil supplements marketed in New Zealand were eligible for inclusion. Fatty acid content was measured by gas chromatography. Peroxide values (PV) and anisidine values (AV) were measured, and total oxidation values (Totox) calculated. Only 3 of 32 fish oil supplements contained quantities of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) that were equal or higher than labelled content, with most products tested (69%) containing <67%. The vast majority of supplements exceeded recommended levels of oxidation markers. 83% products exceeded the recommended PV levels, 25% exceeded AV thresholds, and 50% exceeded recommended Totox levels. Only 8% met the international recommendations, not exceeding any of these indices. Almost all fish oil supplements available in the New Zealand market contain concentrations of EPA and DHA considerably lower than claimed by labels. Importantly, the majority of supplements tested exceeded the recommended indices of oxidative markers. Surprisingly, best-before date, cost, country of origin, and exclusivity were all poor markers of supplement quality.


Methods
All encapsulated fish oil supplements sold at retail or online stores in New Zealand were eligible for inclusion in the study. Marine oils from other sources (such as krill, calamari, or algae, which have important differences in composition) fell outside the scope of this study and were not included. We selected encapsulated oils because this is the most common medium for consumption of fish oil supplements. If there were multiple fish oil supplements available from a particular company, all were included in the study, provided there was a claimed difference in oil composition. One vial of each supplement was purchased for sampling. Where products were found in multiple stores, the first identified vial with a best before date between 12 and 24 months from the time of purchase was selected, otherwise vials 9 to 30 months prior to best before date were eligible.
The recommended retail price and best before date were recorded. Immediately after purchase, oil capsules were removed from their packaging, stored in amber glass jars that were numbered and sealed, and placed inside a cardboard box. The box was then stored at 4uC. Thus, the oil capsules were stored in conditions expected to minimise oxidation 7 . All products were subsequently tested for the oil content, n-3 PUFA composition, PV, and AV. All tests and data analyses were performed by investigators who were blinded to the identity of individual supplements.
Mass of oil content. Capsule oil mass was measured by weighing the oil capsule (m 1 ), piercing and evacuating the oil contents, then washing the halved capsule in hexane. After drying the capsule shell, it was re-weighed (m 2 ). Mass of capsule content was calculated as: m 1 -m 2 . This procedure was performed in triplicate.
A 12 ml sample of oil was produced by combining the contents of 8-24 capsules (depending on capsule size). From this pooled oil sample, PV, AV, Totox, and fatty acid concentration were measured in triplicate.
PV, AV, and Totox. PV was measured according to the European pharmacopoeia 16 by visual titration of iodine. 2.5 g of oil was weighed in a volumetric flask. 50 ml of 352 (glacial acetic acid:trimethylpentane) was added with 500 ml of saturated potassium iodide solution. The flask was occluded and vigorously shaken for 60 seconds, and 30 ml of water added. The yellowed solution was titrated with 0.01 N sodium thiosulphate solution until almost colourless, when 500 ml of 1% starch solution was added. The resulting solution was titrated until colourless using 0.01 N sodium thiosulphate solution. The sodium thiosulphate volume was used to calculate the peroxide value in meq/l according to the formula ''PV 5 [10 3 (V 2 V control )]/ m'', where m is the oil mass previously obtained. Based on triplicate measures, the intra-and inter-assay coefficients of variation were 2.6 and 1.8%, respectively. AV was measured according to the European pharmacopoeia 16 using absorbance spectrophotometry after reaction with p-anisidine. Briefly, 0.2 g of oil was weighed into a small vial. 10 ml of trimethylpentane was added. The A 1 absorbance was measured at 350 nm against a reference solution of trimethylpentane. 1 ml of 2.5 g/l p-ansidine in acetic acid was added to 5 ml of the oil-trimethylpentane solution. After 10 minutes, the A 2 absorbance was measured against a reference solution of 5 ml trimethylpentane with 1 ml of p-anisidine in acetic acid. AV was calculated from these absorbances and the oil mass (m) previously obtained: ''10 3 [1.2 3 (A 2 2 A 1 )]/m''. Based on triplicate measures, the intra-and inter-assay coefficients of variation were 3.0 and 3.8%, respectively. Totox was subsequently calculated by the formula ''(2 3 PV) 1 AV'' 12 .
Markers of oxidation were compared to international guidelines, with recommended values for PV, AV, and Totox of ,5, ,20, and ,26 meq/l, respectively 10-12 . n-3 PUFA composition. 1 ml of fish oil was frozen at 280uC for storage prior to analysis. 50 mg of oil samples were weighed out into solvent washed methylation tubes and dissolved in 10 ml of toluene. From this, 200 ul (1.0 mg) was taken for trans-methylation followed by gas chromatographic analysis. Briefly, 2 ml of methanol:toluene (451 v/v, containing C1950, 20 mg/ml as internal standard) was added to the sample. Acetyl chloride (200 ml) was added while vortexing and the contents were heated for 1 hour at 100uC. The tubes were cooled in water (5 minutes) and K 2 CO 3 6% (5 ml) added and centrifuged (3000 3 g, 5 min, 4uC). The upper toluene phase was collected and stored in a gas chromatograph (GC) vial at 220uC for analysis.
Methylated fatty acid samples were analysed by gas chromatography using a fixed carbon-silica column 30 m 3 0.25 mm (DB-225) (J&W Scientific, Folsom, CA, USA). The GC was equipped with a flame ionization detector, auto-sampler and autodetector. Injector and detector ports were set at 250uC. Oven temperature was programmed: 170uC for two minutes, increased 10uC/minute up to 190uC where it remained stationary for one minute. Temperature then increased 3uC/minute up to 220uC, which was maintained for a total run time of 30 minutes per sample. A split ratio of 1051 and an injection volume of 3 ml was used. A known fatty acid mixture was used to identify peaks according to retention time and their concentration was determined using a 6890 Series gas chromatograph (Hewlett Packard, Palo Alto, CA, USA) with Chemstations Version A. 04.02 (Chemstations Inc, Houston, TX, USA) 17 . Each oil sample was analysed in triplicate.
The mass of individual fatty acids in each capsule were calculated in mg/capsule by the formula ''specific fatty acid content 5 specific fatty acid concentration 3 capsule oil mass''.
Data analyses. Adjusted values for PV, AV, and Totox were also calculated to account for the higher concentration of n-3 PUFA in some oils. Oils with a higher concentration of n-3 PUFA have more substrate for oxidation, and are therefore expected to have higher levels of oxidation markers. Adjusted values standardised the oxidative indices, reflecting the exposure to oxidised compounds per 300 mg of combined EPA 1 DHA (the most frequent labelled dose per capsule). This adjustment may provide a better estimate of a consumer's exposure to oxidation products, as less capsules of a more concentrated fish oil would be ingested to achieve a targeted n-3 PUFA intake. The adjusted PV was calculated as: (PV 3 300)/(EPA 1 DHA), where the concentrations are the labelled content in mg/g of oil. Similarly, the adjusted anisidine value was calculated as: (AV 3 300)/(EPA 1 DHA). Adjusted Totox was calculated as: (2 3 AdjPV) 1 AdjAV. Associations between variables of interest were assessed using simple linear correlations and Spearman's rank correlations. Parameters measured were compared between countries of origin (New Zealand, Australia, and Other) using general linear regression models.

Results
We initially studied 32 individual fish oil supplements available in the New Zealand market (Table 1). An extra four products were later identified and also examined, but did not have their EPA 1 DHA content measured (Table 1). There was a considerable variation in price among different brands, with the recommended retail price (RRP) per gram of fish oil varying from $0.05 to $0.77 (USD) ( Table 1).
EPA 1 DHA content. The total oil content of individual capsules exceeded 97% of the labelled oil content for all supplements tested (Table 1). However, there was a marked disparity between the labelclaimed content of EPA 1 DHA and the actual capsule content of these fatty acids (Table 1; Figure 1), with supplements containing on average 68% (SD 5 23%) of the claimed content. Only 3 of the 32 oils tested contained a quantity of EPA 1 DHA that was equal to or higher than that claimed by the label, with more than two-thirds of supplements tested (22; 69%) containing less than 67% (Table 1; Figure 1). Two supplements contained approximately one third of the label concentrations of EPA 1 DHA (Table 1; Figure 1).
There was some indication that the more expensive brands had more accurate labelling of EPA 1 DHA content, as RRP was positively correlated with the ratio of measured to claimed EPA 1 DHA (r 5 0.55; p 5 0.001). In addition, despite the majority of fish oils having less n-3 PUFA than claimed, the label EPA 1 DHA content was well correlated with the actual n-3 PUFA content (r 5 0.71; p , 0.0001).
Oxidative markers. There were high levels of oxidation in the fish oils assessed (Table 1), with the vast majority of supplements tested failing to meet recommended levels of oxidation markers ( Figure 2). 30/36 (83%) products exceeded the recommended PV, 9/36 (25%) exceeded AV, and 18/36 (50%) exceeded recommended Totox thresholds ( Figure 2). Only 3 of 36 oils tested (8%) met all the international recommendations, not exceeding any of these indices ( Figure 2). After adjustment for concentration, only 19% (7/36) were within the recommended limits, indicating that the high frequency of excess oxidation was not simply an artefact caused by the concentrated oils.
There were no associations between the time to best-before-date and concentrations of oxidation markers (data not shown). There was a positive correlation between RRP and both PV (r 5 0.41; p 5 0.013) and Totox (r 5 0.39; p 5 0.018). However, this association likely reflected increased n-3 concentrations in more expensive products, as there was no association between RRP and content-adjusted PV (p 5 0.34) and Totox (p 5 0.46).
There was an observed association between the missing EPA 1 DHA (i.e. label claimed minus actual content) and both AV (r 5 0.45; p 5 0.011) and Totox (r 5 0.35; p 5 0.053) (Figure 3). Missing EPA 1 DHA content was not associated with PV (r 5 0.06; p 5 0.75). (USD) from New Zealand (p , 0.0001) and 25 cents from Australia (p , 0.0001). However, these international supplements were not associated with better supplement quality. The percentage EPA 1 DHA content (in proportion to claimed label content) was virtually identical in supplements from the three sources, and there were no differences in levels of any oxidative markers assessed (data not shown).

Discussion
This study shows that almost all fish oil supplements available in the New Zealand market contain much lower concentrations of long chain n-3 PUFAs than claimed by the product label. Importantly, the majority of fish oils exceeded the recommended indices of oxidative markers. The discrepancy between actual and labelled n-3 PUFA content is consistent with a study of fish oils sold at retail in South Africa, and an older international survey where products were purchased in Canada, USA and UK; both found that more than half of the marine n-3 PUFA products surveyed contained less than 90% of the n-3 PUFAs claimed 18,19 . A small study in Ohio showed even greater discrepancy 20 . In contrast, an Australian survey of retail products showed relatively close agreement between the labelled and measured n-3 PUFA content 21 . However, the method used to convert qualitative data to capsule content in that survey did not take into account the glycerol content of the oil, and would therefore overestimate the n-3 PUFA content.
As the labelled content of most fish oil products was inaccurate, it is important to consider how and when the n-3 PUFA content had been measured. The product labels did not indicate the method used to measure n-3 PUFA content, or at what stage of production this occurred. However, it is interesting that 17 products listed the same concentration of n-3 PUFAs (180 mg EPA and 120 mg DHA per gram of oil). Similar concentrations would be expected given that most oils are sourced from oily fish from the west coast of South America. However, such striking uniformity suggests that the companies selling these products have relied on data provided by extrac-   tors who supply fish oil to multiple brands. Thus, the labelled composition may not reflect the final n-3 PUFA content of product, which may have changed due to oxidation during transport, encapsulation, packing, and storage. Further, while there is evidence of seasonal variation in the n-3 PUFA content of oily fish 22 , the labelled n-3 PUFA content of individual brands does not vary with the best before date (which is related to the date of manufacture). This suggests that the fatty acid content is unlikely to be measured in individual batches of fish oil.
The high levels of oxidation identified in this study are broadly consistent with other surveys, which have shown that 11-62% of fish oil products are oxidised above international recommendations 15,[23][24][25][26] . As oil oxidises, the concentrations of EPA and DHA (the purported active compounds) decrease, suggesting reduced efficacy. In our study, oxidation may at least in part account for the low n-3 PUFA levels observed as there was an association between AV and Totox and the missing n-3 PUFA. Note that it is not surprising that PV was not associated with missing n-3 PUFA as lipid peroxides are initially formed during oxidation but subsequently broken down over time to secondary oxidation products, such that a low PV is consistent with both minimal and severe levels of oxidation. Nonetheless, the relatively weak correlations between these measures suggest that other unknown factors are also important, which could include poor quality control during manufacture.
The health implications of oxidised fish oil consumption remain unclear 14,27 . There is some evidence that specific n-3 PUFA oxidation products (i.e. resolvins, protectins and n-3 PUFA derived isoprostanes) have a role in mediating the anti-inflammatory effects of n-3 PUFA supplementation 28,29 . However, it is not clear whether these anti-inflammatory oxidation products are produced in sufficient quantities during capsule storage, to confer a net benefit to consuming oxidised fish oil. In fact, evidence from animal studies show that large doses of oxidised lipids may cause organ toxicity 30 , growth retardation 30 , and accelerated atherosclerosis 31 . Only one relatively short study has compared the effects of oxidised and unoxidised fish oil in humans 32 , observing no evidence of acute oxidative toxicity. However, the effects of long-term exposure to oxidised oils (particularly on markers associated with atherosclerosis) have not been studied. To our knowledge, the oxidative state of fish oils adopted in clinical trials has never been reported, so the possible effects of oxidation on trial outcomes are unknown 14 . It is possible that the conflicting and often disappointing results in clinical trials may have resulted from the use of highly oxidised supplements 14 . The high levels of oxidation shown in this study underline the importance of understanding the effects of oxidised fish oil taken in dietary supplements, particularly for pregnant women, children, and consumers with inflammatory, metabolic, or cardiovascular disease.
Our results also indicate that consumers would be unable to identify unoxidised fish oil supplements. The time until the best before date printed on the packaging had no relationship to the level of oxidation. Cost was also unhelpful, for while there was a correlation between cost and oxidation such that the more expensive supplements had greater oxidation, this relationship was an artefact of concentration (concentrated supplements were more expensive). In addition, supplements from outside Oceania were as oxidised as those manufactured in Australia and New Zealand. Even the two products that could only be purchased after naturopathic consultation had excess oxidation, and though one was labelled with the peroxide value, the true PV exceeded the label by more than four-fold.
It is important to emphasise that all products were well within their best-before dates, with the shortest being 270 days from analysis. The range of best-before dates observed in this study reflects the substantial variability presented on retail shelves. All retailers with a physical store were observed to keep their oil under the same conditions; at room temperature under artificial lighting, (the storage conditions of online only stores could not be observed). As fish oil products were selected based on predefined rules that were applied to all supplements tested, selection bias is unlikely. The major limitation of this study is that each product was only purchased from a single store, so that our study would have been unable to identify variations between batches or stores. Nevertheless, this study represents a comprehensive sample of the fish oil products available to consumers in New Zealand at the time of sampling. As most fish oil products sold globally (including New Zealand) appear to be sourced from deep sea fish from the west coast of South America, the results of our survey are likely to have international relevance.
In this study, we included all identified fish oil products (the most popular form of n-3 PUFA supplement), but not n-3 PUFA supplements produced from other marine sources such as krill. There are important differences in the composition of krill oils, particularly in the incorporation of n-3 PUFA as phospholipids 33 and the presence of naturally occurring antioxidants such as the pigment astaxanthin 34 . These differences may affect the propensity of these oils to oxidise 7 . Thus, the results of our study cannot be generalised to other marine oils that are not sourced from fish, and krill oils should be studied separately.
In summary, the majority of fish oil supplements available for purchase in New Zealand not only have n-3 PUFA contents well below those claimed by labels, but are also considerably oxidised (with PV, AV, and Totox values above recommended levels). The associated health implications are unclear. Future studies should investigate the effect of environmental conditions on oxidation of encapsulated fish oils, particularly regarding the oxidation process when supplements are stored in retail or home environments. Further, clinical trials investigating the health effects of fish oil products should measure and report their peroxide and anisidine values, so that the importance of oxidation to efficacy and harms can be better understood.