4.1. Hoki Liver Oil Over-Oxidation
The present study reproduced the oxidative conditions employed by Albert et al. [28
] as well as possible, even though a precise description of the experimental setup, such as the wavelength range and intensity of the illumination used, the rate of oxygen gas delivery, and the geometry and material of the vessel used for the incubation were not reported. By inspecting the temporal profile of PV in the present study, it was nevertheless possible to assess at which degree of oxidation the chemical changes would be approximately similar to those that may have occurred in the original study. Albert et al. reported a PV of 48.8 meq O2
/kg for the hoki liver oil after one month of accelerated oxidation [28
]. The final PV attained in the current study was ≈125 meq O2
/kg, indicating that the conditions applied in the present study were more intense. In the present study, a similar level of hoki liver oil oxidation was attained at approximately 15 days.
Remarkably, only a relatively small increase in secondary oxidation was found in the study by Albert and colleagues; after 30 days, p-AV had increased from 0.6 to 4.5, which is a value still well within the quality requirements for omega-3 EPA/DHA oils. For comparison, the present study showed a ≈3-fold higher p-AV (13.3) at 15 days and a ≈10-fold higher p-AV (43.8) at 30 days. A range of other indicators, such as oligomers and volatiles, confirmed the marked degree of secondary oxidation that was obtained. One of two plausible reasons may explain the apparently low level of secondary oxidation in Albert et al. [28
]. Either fatty acid hydroperoxyl radicals were not effectively further transformed into secondary oxidation products, which is highly improbable, or a substantial portion of the volatile para
-anisidine-reactive aldehydes formed as secondary oxidation products were stripped out of the incubations more effectively than in the present study. The permanent oxygen gas stream is likely continuously removing a portion of the volatiles formed, suggesting a more intense contact (smaller bubbles) or longer contact time (a more vertically shaped container) of the gas bubbles with the oil in the study by Albert et al.
Both oxidation conditions employed in the present study represent good examples of “open” incubations, in which volatile compounds formed in the oil are easily lost to ambient air as the equilibrium is shifted toward net evaporation [13
]. Since volatile compounds produced during the sustained oxidation incubations can evaporate from the incubations, the pattern of detected volatiles needs to be interpreted as those present in the oils at the moment of sampling. The accumulation of secondary oxidation products with boiling points below ambient temperature (oxidation condition A) or 50 °C (oxidation condition B) will not occur; i.e., they will already have evaporated to a large extent from the oils at the time of sampling. In Condition A, the formation of volatile secondary oxidation products is likely to be underestimated even more, since the continuous oxygen gas stream will strip volatiles out of the incubations faster than the spontaneous evaporation of volatiles from the incubations. This may explain the lower concentrations of volatiles observed in both oil types under Condition A compared to B. This situation does not correspond to the situation of encapsulated or bottled fish oils, which are closed systems without any, or very limited, gas exchange. It is probable though that the extent of secondary oxidation was markedly under-estimated in the study by Albert et al., and that the levels of the secondary oxidation products with low volatility, such as the isoprostanoids and oxysterols, could be present at comparable levels as in the current study at day 15, or between day 15 and 30.
Other small differences in experimental conditions may cause relatively large differences over the long and sustained oxidation period, but their influence is not possible to estimate. Differences in oxygen tension in the oil might also be considered to be a factor, since the rate of lipid peroxidation depends on oxygen tension, as shown for rapeseed oil [62
]. Air-saturated water at 20°C contains an oxygen concentration of ≈0.29 mM [63
], whereas at full saturation, dissolved oxygen is ≈1.1 mM. The solubility of oxygen in marine triglyceride fish oils is in the range of 1.22–2.44 mmol/kg oil at 20 °C, and it is expected to be higher at 50 °C under Condition B [64
]. With continuous oxygen delivery to the oils, oxygen tension is likely maximal, suggesting that a small deviation in the precise rate of continuous delivery of oxygen gas to the oils should not be expected to affect the oxidation rate.
It is clear that a very high level of oxidation was achieved in the hoki liver oil by the photooxidative conditions at high oxygen tension after one month (condition A). The attained TOTOX number of 296 is ≈11 times larger than the maximum limit of 26 set voluntarily by industry and with which many fish oil producers and retailers comply [18
]. With the employed antioxidant-free hoki liver oil, this limit was exceeded already within 1 day of incubation. Furthermore, the marked rise in oligomers indicates that the oil was severely damaged, exceeding the pharmacopeial monograph limit of 1.5% that is considered acceptable for refined EPA/DHA omega-3 triglyceride oils [65
]. The present study also confirms that massive oxidative damage to an oil, exceeding regulatory quality limits many-fold, is needed to achieve a relatively modest 10% reduction in EPA and DHA content in an unprotected oil, as exemplified by the hoki liver oil in this study. It has previously been shown that the minor oxidative changes that can occur during normal shelf-life conditions of fish oil finished products do not lead to measurable decreases in EPA and DHA content [19
]. The content of saturated and monounsaturated fatty acids was not affected, which was expected, since these fatty acids are insensitive, or much less sensitive, to oxidation than PUFA. Other LCPUFA, such as AA, DPA n-3, DPA n-6, and AdA, did not display a decrease in content, but small amounts were nevertheless converted by lipid peroxidation to the corresponding isoprostanoids.
The sustained oxidation of oils and biological membranes rich in PUFA is understood to occur in three stages: initiation, propagation, and termination [13
]. Initiation involves the formation of fatty acid peroxyl radicals, propagation involves an unimpeded chain reaction of further free radical-mediated lipid peroxidation, and termination forms nonradical products. Once formed, a fatty acid peroxyl radical can abstract a bis-allylic hydrogen atom from another PUFA, thereby forming the corresponding fatty acid hydroperoxide, the major primary oxidation product, and a new PUFA radical. The hydroperoxides can further react with radicals to form alkoxyl radicals or fragment by chain scission, leading to the formation of an array of hydroxylated derivatives, saturated and unsaturated aldehydes, ketones, and short-chain hydrocarbons, which are collectively called secondary oxidation products.
The acid value is sometimes used as an indicator for oxidative quality in fish oils. However, the absence of detectable changes in acid value did not support the hydrolysis of esterified fatty acids to increased free unesterified fatty acids to occur under the experimental conditions. Possibly, the low water content in the oils precluded changes in the acid value. However, hoki liver oil had a three-fold higher baseline acid value (1.49 ± 0.01 mg KOH/g oil) than the anchovy oil (0.49 ± 0.00 mg KOH/g oil), which may be attributable to a higher level of free fatty acids. Carboxylic acids can catalyze the chain scission of fatty acid peroxides, thereby increasing the rate of secondary oxidation of an oil [67
]. Although we did not measure the level of free fatty acids directly, it is possible that the more marked development of secondary oxidation in hoki liver oil under thermal oxidation Condition B, compared to the anchovy oil, is promoted by a higher level of free fatty acids.
Another important aspect of reproducing the experimental conditions was that the hoki liver fish oil used in the study by Albert et al. did not contain added antioxidants, whereas the large majority of fish oils, as ingredients and finished products, contain added antioxidants (typically α-tocopherol or mixed tocopherols) to protect the oil from oxidation. This hoki liver oil was minimally refined and still contained some tocopherol (all α-tocopherol, which originates from the liver tissue from which the oil was extracted) but did not contain added antioxidants [68
]. The approximately 15-fold lower total tocopherol content than the standard refined and antioxidant-stabilized anchovy oil rendered it much more susceptible to oxidation.
The progress of oxidation in hoki liver oil is well illustrated by the measured changes in induction time, which is a parameter that reflects the remaining resistance of an oil to oxidation. This resistance reflects the level of antioxidant that remains at any time point, which still serves to protect the oil from entering in an uncontrolled radical propagation phase that will consume all available peroxidizable polyunsaturated fat. Tocopherols are chain-breaking antioxidants that donate hydrogen to reduce fatty acid peroxyl radicals and alkoxyl radicals to the corresponding alcohols, lowering the possibility for radical reactions to progress [69
]. When the induction time had decreased approximately 6-fold, which occurred between day 5 and 15, both primary and secondary oxidation ramped up substantially. Marked increases in PV, p-AV, TOTOX, and oligomers occurred at this time. By day 15, the total tocopherol level was reduced by half and was thereafter completely consumed in the second half of the incubation period. This antioxidant activity was reflected in the near-perfect correlation between induction time and tocopherol content (r = 0.997). In contrast, the formation of total volatiles lagged in terms of the change in induction time compared to the other indicators of secondary oxidation, supporting the idea that a major portion of volatiles is continuously stripped out of the oil by the delivery of oxygen gas to the oil.
4.2. Contrasting Distinct Oxidation Conditions
For fish oils and other EPA/DHA omega-3 finished products that consumers purchase, a more realistic way to experience oxidation is exposure to storage conditions that are too hot for the intended formulation but without incident light, since products are usually packaged and bottled in tainted plastic or glass containers that minimize incident light. Exposure to air and light may theoretically occur during fish oil production or encapsulation, but experienced manufacturers generally know how to control this. Nevertheless, it is important to study such different product environments to determine if qualitative and quantitative differences in specific oxidation products may occur. Since the present study looked at two different conditions, it was possible to appreciate that the continuous exposure of hoki liver oil to oxygen gas at room temperature under light conditions led to a much stronger formation of primary oxidation products (fatty acid peroxides) than at elevated temperature (50 °C) in combination with exposure to ambient air in the dark. Between days 2 and 15, secondary oxidation, as measured by p-AV, was stronger under thermal stress in the dark and ambient air exposure, while at day 30, p-AV was similar under both conditions in hoki liver oil. Interestingly, the distribution and absolute levels of specific volatiles was entirely different: at day 15 under Condition A, the volatiles with the highest abundance were 1-penten-3-ol (2.46 ± 0.06 mg/kg) and 1-penten-3-one (1.69 ± 0.02 mg/kg). Under Condition B, the most abundant volatiles were 1-penten-3-ol (124.20 ± 0.90) and (E,E)-2,4-heptadienal (14.84 ± 0.01 mg/kg). The difference in absolute levels again illustrates the probable accelerated evaporation of volatiles under Condition A, leading to at least a log-order lower level of volatiles under this condition compared to thermal stress in the dark. At day 30, although p-AV was the same under both conditions, the summed level of volatiles that were measured was about 2.5 times higher in Condition B. This indicates that non-volatile aldehydes represented a relatively greater proportion of para-anisidine-reactive substances in the hoki liver oil under Condition A than in B. These results show how the pattern of secondary oxidation products is very sensitive to the precise experimental and oxidative conditions.
The finding that a near-perfect negative correlation existed between tocopherol content and PV of hoki liver oil under Condition A (r = −0.993), but less strongly under Condition B, suggests that tocopherols are particularly effective in protecting fish oils from primary oxidation under conditions of photo-oxidation at high oxygen tension. For anchovy oil, this was observed under Condition B for the relationship between tocopherol levels and p-AV (r = −0.999) and volatiles (r = −0.9997), suggesting that tocopherols are excellent protectors of hoki liver oil from secondary oxidation under conditions of elevated temperature in the dark. Thermolysis did not take place under Condition B, since the acid value did not rise.
4.3. Comparison of Oxidation in Different Fish Oils
While hoki liver oil is a good source of EPA and DHA, on a global scale, it is used in relatively small volumes compared to other fish oils. Therefore, the susceptibility to oxidation and associated chemical changes of the hoki liver oil were contrasted with the far more commonly used refined anchovy oil. In Condition B, the induction time dropped more gradually over time in hoki liver oil compared to the anchovy oil, which showed drastic decreases in the second half of the incubation. Nevertheless, the induction times attained in the end of the incubation were nearly identical, as were PV, p-AV, and TOTOX, indicating that over the longer period of time, both oils had lost antioxidant protection to the sustained condition of thermal oxidation in the dark (Condition B). The final TOTOX number of 117 meq O2
/kg (Condition B) is four times larger than the voluntary industry limit of 26, again illustrating the marked rancidity achieved [18
]. This result indicates that while at earlier time points the well-stabilized anchovy oil resisted thermal oxidation better than the antioxidant-free hoki liver oil, at longer periods of time, the anchovy oil also quickly lost antioxidant protection.
In contrast, a marked difference in sensitivity to photo-oxidation at high oxygen tension (Condition A) was found between hoki liver and anchovy oil. The anchovy oil proved to be remarkably resistant throughout the 30-day period to this form of oxidative challenge, while the antioxidant-free hoki liver oil became rancid rapidly. While the total polyunsaturated fat content of both oils was comparable, the tocopherol contents were vastly different. The high level of mixed tocopherols present in anchovy oil likely support the delayed occurrence of free radical propagated lipid peroxidation as reflected in reduced primary and ensuing secondary oxidation, as seen in the profile of PV, p-AV, TOTOX, volatiles, and induction time, as compared to hoki liver oil where the antioxidant protection conferred by the naturally present α-tocopherol is considerably lower. The oligomer level in anchovy oil remained below the 1.5% limit set by the European Pharmacopoeia [65
], indicating that the oil remained relatively intact even after 30 days. When, after 30 days, the anchovy oil also started to succumb to the strong oxidative environment, tocopherol levels had decreased by only ≈20%. The correct use of mixed tocopherols is an example of how a refined fish oil, which in theory is highly susceptible to oxidation, can be stabilized appreciably through the appropriate use of antioxidants [17
]. The level of transition metals was not determined but would provide further information on the sensitivity of the tested oils toward facilitating lipid peroxidation.
4.4. Specific Groups of Secondary Oxidation Products and Biological Activity
In the study by Albert et al., about 30% of newborn rat pups, born from dams that received over-oxidized hoki liver oil throughout gestation, had died by day 2 postnatally [28
]. Although litter sizes were not affected, no further information on placentation, intra-uterine growth restriction, or other developmental or biochemical alterations during gestation was reported. Pups that did survive had no reported alterations from pups born from dams that had received unoxidized fish oil. Dams that continued to be fed with the rancid hoki liver oil postnatally developed increased insulin resistance at day 22 when offspring was weaned. No further information on dam health during gestation or at parturition was reported. New research is necessary to understand if specific oxidation products formed in oxidized oils that were identified in the present study may be involved in postnatal mortality and other potential perinatal alterations. Specific volatiles, isoprostanoids, and oxysterols have been reported to have potent effects on physiological processes, as discussed next.
Volatiles are frequently measured as an indicator for oxidative quality of marine oils, although no consensus currently exists about which volatiles should be monitored, since the profile can vary depending on oil type and oxidation mechanism. With the exception of 1-penten-3-ol, the volatiles tested in the present study are conjugated aldehydes that will react with the para
-anisidine reagent and contribute to p-AV value. The formation of unesterified secondary oxidation products originating from esterified LCPUFA, such as various oxygenated α
-unsaturated aldehydes, has been shown to occur in thermally treated aerated edible oils [71
]. Such conjugated aldehydes are bio-available upon oral intake, and their electrophilicity toward proteins and nucleic acids is linked to cytotoxic and genotoxic activity [72
]. The reactive aldehydes can react with sulfhydryl groups in proteins, thereby interfering with cellular metabolism and physiological processes. Under Condition A, 1-penten-3-one, (E
)-2-pentenal, 1-pentene-3-ol, (E
)-2,4-heptadienal, and (E
)-2,6-nonadienal were all detectable during the over-oxidation of hoki liver oil at 15 and 30 days. Hence, these volatiles will likely have been present in the oil administered to rats in the study by Albert et al. In a future study, it would therefore be interesting to assess which of these substances, at the concentrations uncovered in this study and administered in correctly allometrically adjusted doses, display perinatal toxicity in pups and dams. To draw meaningful conclusions about the biological risk resulting from exposure to reactive oxidation products, such in vivo studies should also evaluate the dose in relation to endogenous antioxidant defenses and phase 2 metabolism of reactive oxidation products [74
Furthermore, biological activity that is unrelated to electrophilic reactivity should be considered. For example, (E
)-2-pentenal is a strong irritant of sensory nerves, being a ligand for the TRPA1 channel [76
]. Even though these volatiles are highly reactive, they do not show an indiscriminate reactivity, because steric factors are important as well; for example, the α
-unsaturated oxo-compounds 1-penten-3-one and (E
)-2-pentenal inhibit glucose 6-phosphatase, but conjugated dienals do not [77
]. Limitations of the present study are recognized, since a wider array of LCPUFA-derived volatiles can be formed in fish oils [78
]. The major low molecular weight volatile acrolein, the dialdehyde malonaldehyde, as well as 4-hydroxy-α
-unsaturated aldehydes such 4-hydroxy-hexenal, for which cytotoxic effects are documented, were not covered by the present study [78
]. From the perspective of refined anchovy oil, further research could focus on volatiles formed particularly under conditions of thermal stress in the dark.
The present study reports for the first time that several isoprostanoids in unesterified free form are present in refined fish oils, and that specific isoprostanoids are formed during the development of rancidity. Isoprostanoids, and specifically F2
-isoprostanes, were originally identified as esterified products within phospholipids in peroxidized biological membranes [80
]. Isoprostanoids can also be formed from unesterified omega-3 LCPUFA under peroxidative conditions [81
]. Given the absence of active deacylating enzymes in refined oils, either the isoprostanoids are released by hydrolysis during oxidation or they are uniquely formed from free fatty acids in the oils (although we do not know what the levels are in this study). It is also possible that isoprostanoids formed during oxidation will be present in esterified form within the triglycerides that make up the bulk of the hoki liver oil and anchovy oil. This will need to be further investigated. Interestingly, all the tested isoprostanoids were also present at baseline, which suggests that these are also formed during the extraction and refining of the oil, and/or they originate from the fish from which the oil was originally extracted. Since isoprostanoids are chemically relatively stable at ambient conditions, it will be interesting to evaluate if specific members could be suitable markers for secondary oxidation in omega-3 EPA/DHA oils, including low-grade oxidation in ingredient oils and finished products. Recently, neuroprostanes derived from DHA and isoprostanes from EPA were shown to be present in two encapsulated marine dietary supplement oils: calanus oil and a fish oil composed of a mixture of sardine, anchovy, mackerel, and herring oil [83
Several phytoprostanes, isoprostanes, and neuroprostanes identified in the over-oxidized hoki liver oil at day 30 were present at higher levels than before the exposure to oxygen and light or at day 5 or day 15. Several of these substances were also present in considerably higher levels than in anchovy oil. However, strictly taken, at day 15, none of the measured isoprostanoids had increased compared to baseline, which in many cases had initially decreased from baseline to day 5. Between day 5 and day 30, a range of isoprostanoids were formed increasingly, ultimately reaching levels markedly surpassing baseline, typically about 3.5 times. The most abundant isoprostanoids formed from day 5 onwards were 5-epi-5-F3t-IsoP (0.32 ng/g oil) and 5-F3t-IsoP (0.1 ng/g oil), which were both derived from EPA, and the DHA-derived 4(RS)-4-F4t-NeuroP (0,2 ng/g oil), 20-epi-20-F4t-NeuroP (0.023 ng/g oil), and 14(RS)-14-F4t-NeuroP (0.014 ng/g oil). Thus, specific isoprostanoids could be considered of interest to further evaluate to contributing to, or modulating, the postnatal and maternal toxicity as observed by Albert et al.
The abundance of individual PUFA species in each oil was related to the concentrations of isoprostanoids formed. EPA and DHA are clearly a chemical substrate for the formation of specific isoprostanoids, but also other major LCPUFA species present in fish oils, such as AA and ALA, as well as LCPUFA found in lower abundance, such as adrenic acid and DPA n-3, were converted to the corresponding isoprostanes and phytoprostanes (and related furans). In vivo, tissue levels of EPA- and DHA-derived isoprostanes and neuroprostanes increase following dietary supplementation with EPA and DHA [37
]. While all the measured isoprostanoids were found in relatively low concentrations, some of these substances have potent biological activity. For example, 15-F2t
-IsoP has potent vasomotor activity, and it activates platelets [84
-IsoP regulates neurotransmitter release via activity on the prostanoid EP1 receptor [36
]. With regard to the perinatal period, F2
-isoprostanes are strongly related to the consequences of intra-uterine hypoxia and neonatal morbidity [85
]. Elevations in 15-epi
-IsoP and total F2
-isoprostanes in the cord blood of newborn infants are strongly correlated with the severity of hypoxic–ischemic encephalopathy [86
]. At birth, the transition to a higher oxygen tension involves increased oxidative stress, leading to the generation of specific isoprostanoids that are involved in the closure of the ductus arteriosus [87
]. The possibility that circulating isoprostanoids derived from a high intake of over-oxidized fish oil may be involved in these processes needs to be investigated.
Although isoprostanoids are useful biomarkers of oxidative stress in organisms, and their formation has been documented in several disorders, their biological activities are varied. The reported activities of EPA-derived isoprostanes and the DHA-derived neuroprostanes are not necessarily negative for health [37
]. For example, the 4(RS
-neuroP and F3
-IsoPs have anti-arrhythmic and cardioprotective activity [38
], and 4(RS
-NeuroP and 14(RS
-NeuroP have potent anti-inflammatory activity in macrophages but less than enone A4
]. So far, the DHA-derived 4(RS
-NeuroP is the most potent LCPUFA-derived isoprostanoid identified and possesses antiarrhythmic activities both in vivo and in vitro via the protection of the ryanodine receptor [89
] and similarly protects against ventilator-induced diaphragmatic dysfunction in vivo [90
]. Recently, Lee et al. reported that 4(RS
-NeuroP upregulates the transcriptional level of the antioxidant enzyme heme oxygenase-1 in SH-SY5Y cells and primary neuronal culture, confirming the bioactivity of this neuroprostane [91
We also addressed if oxysterols might be formed in hoki liver oil during the prolonged over-oxidation induced by exposure to oxygen and light. Three oxysterols, 5α,6α-epoxycholesterol, 5β,6β-epoxycholesterol, and 7-hydroxycholesterol, proved to be very sensitive markers of the oxidation process, showing increases in level already at day 5. That these correspond to two epoxides and a primary alcohol suggests that these oxysterols reflect primary oxidation, which is analogous to the formation of fatty acid hydroperoxides. Interestingly, the two 5,6-epoxy derivatives, which are highly sensitive to oxidative conditions, were also the most abundant oxysterols detected in the oil after the oxidation process.
We did not determine the content of cholesterol of the oils used in the present study. However, the level of cholesterol in refined hoki liver oil has been reported to be 1.41 g/kg [92
]. Refined anchovy oil that has been refined, bleached, deodorized, and further processed by molecular distillation contains cholesterol in the range of 3–6 g/kg [93
]. Thus, the levels of cholesterol in these two refined oils are likely to be present in a similar concentration range, and our initial hypothesis that cholesterol level would be substantially higher in hoki liver oil than in anchovy oil does not hold. Thus, the availability of cholesterol as an oxidizable substrate is not the explanation for the higher formation of oxysterols in hoki liver oil than in the anchovy oil, although understanding the relative contributions of esterified and free cholesterol to oxysterol formation merits further research. The main difference remains the large difference in susceptibility to oxidation, which is mainly defined by the difference in antioxidant protection. Antioxidants such as tocopherols are able to reduce the levels of oxysterols in a variety of foods [94
]. The finding that specific oxysterols can be formed during the over-oxidation of fish oil has not been reported previously. Nevertheless, the levels are low (nmol per kg oil range). Interestingly, both unoxidized hoki liver oil and anchovy oil already contained low levels of various oxysterols, some of which are unique to each oil type.
The formation of oxysterols has been associated with the onset and progression of a range of disorders [95
]. The pathophysiological activities of oxysterols are inter-related with the multiple biological functions of cholesterol, including being a precursor molecule for the synthesis of steroid hormones, neuroactive steroids, and bile acids. Many oxysterols are cytotoxic toward endothelial and smooth muscle cells at low micromolar concentrations, including 5,6-epoxycholesterol and 7-hydroxycholesterol, which increased markedly in the over-oxidized hoki liver oil [97
]. The addition of oxidized cholesterol to salmon oil has been shown to place a marked stress on antioxidant defenses in rat liver [100
]. Oxysterols may also mediate compromised insulin sensitivity, resulting from the consumption of cholesterol-rich diets, as well as toxic effects on the central nervous system [101
]. Several oxysterols, such as 25-hydroxycholesterol, 7-ketocholesterol, and 22-hydroxycholesterol, can impair the differentiation and fusion of term trophoblast cells and alter their secretion of chorionic gonadotrophin, via activating nuclear liver X receptors (LXR) and inhibition of Sterol Response Element Binding Protein-2 (SREBP-2), which may have important implications for placental development and function [103
]. Although specific research related to perinatal morbidity has not been carried out, there are multiple possibilities regarding how specific oxysterols might compromise the health of rat pups and dams following the exposure to excessive amounts of dietary oxysterols.