4 Discussion
The AOCS pAV test (Cd 18-90), as a simple colorimetric method to measure non-volatile carbonyls, primarily aldehydes, has limitations due to its non-specificity. The imine chromophore that forms between an aldehyde and the p -anisidine reagent differs considerably among aldehydes (Szabó et al., 2010), with the method being more sensitive to unsaturated than saturated aldehydes (Gordon, 2004). So, pAV results provide only relative aldehyde concentrations. Although pAV is expected to reflect the amount of aldehydes arising from hydroperoxide decomposition during oxidation, it is known to respond to extraneous aldehydes as well, such as those from flavors. The interference by flavors is dose and type-dependent. Jackowski et al. (2015) found that marine oil products with citrus flavors, as well as products marketed towards children (e.g., bubblegum-flavored products) often had higher pAV than other flavored products. This was also confirmed in this study; the pAV increase caused by 14 different flavors varied from 1.1 to 46.5, with chocolate-vanilla, citrus, and bubblegum flavors having the largest impact (Fig. 1). Despite these limitations, pAV is still one of the most common methods to measure marine oil oxidation, due to its simple procedure and low cost compared to sensory evaluation or volatile analysis using GCMS.
GOED recognized these limitations and provided advice for those who want to use the AOCS pAV method for fish oil testing (GOED, 2018). In the guidance document, GOED recommended a protocol for flavored fish oils, based on an assumption that flavors have a constant contribution to the measured pAV throughout shelf life studies. However, in this study we found chocolate vanilla and lemon flavors had a diminishing effect on the pAV (Fig. 2), indicating that the current GOED recommendation may not apply to all the flavored oils. For both flavors, the significant difference in interfering components (A ) in FBO and FAO compared to UFO was expected since the flavors introduced a large amount of stable interfering aldehydes. Based on the observed differences in pAV of FBO and FAO, we had also expected to find a significant difference inB (changing aldehydes) and k (rate constant for aldehyde formation) for both flavors. However, the uncertainty introduced by the 4-day sampling frequency limited our ability to identify statistical differences. A more frequent sampling plan would have given us a greater confidence in estimating the model parameters. Despite this, we did find that chocolate-vanilla FBO samples had a significantly lower rate constant (k ) than UFO. This might be caused by flavor components with antioxidant properties or by flavor aldehydes that degraded over time.
With the current experimental design, it is difficult to determine if this diminishing effect was due to antioxidant activity, degradation of flavor aldehydes, or both. It is not concerning if the diminishing effect on measured pAV was solely due to any antioxidant activity because in this case, the lower pAV truly reflected the slower oxidation in the flavored oil. It is concerning if the decreasing contribution to the pAV was caused by flavor loss. In this case, more oxidation would have occurred before the maximum pAV allowed by the GOED recommendation is reached. In other words, following the GOED recommended protocol, we would underestimate the amount of oxidation that was occurring in the oil. This underestimation of the extent of oxidation may lead to overly optimistic inferences about the quality and shelf-life of the flavored oil in question.
In the chocolate-vanilla flavor used in this study, vanillin had the highest ARIS (90.7%). Antioxidant activity of vanillin has been demonstrated in antioxidant capacity assays (Tai et al., 2011) and in food systems containing polyunsaturated fatty acids (Burri et al., 1989). Vanillin may be oxidized to vanillic acid, which is 3.3 times more effective as an antioxidant than vanillin in bulk oils (Mourtzinos et al., 2009). Vanillic acid has also been shown to exert antioxidant effects in corn oil subjected to deep-frying conditions (Naz et al., 2005) and in fish oil at temperatures between 35 to 55°C (Farhoosh et al., 2016). Thus, vanillin is very likely to have a real influence on the rate constant for aldehyde formation (k ) in the FBO samples, through an antioxidant effect.
In the lemon flavored oils, although pAV of FBO and FAO samples were significantly different on days 12, 16, and 20, no significant differences were observed in the rate constant (k ) between FBO and other samples. In the lemon flavor, citral had the highest ARIS (total of α- and β-citral at 48.8 %). Citral does not exert strong antioxidant effects in edible fruit coatings (Guerreiro et al., 2015, 2016) and in essential lemon oils (Misharina et al., 2011). In fact, an antioxidant is often added to inhibit citral degradation. GC analysis of cold-pressed lemon oil stored in the dark for two months at 30°C showed significant losses of citral (Nguyen et al., 2009). This effect is exacerbated at increased temperatures (Djordjevic et al., 2008; Nguyen et al., 2009). Compounding this, citral is quite volatile. In the open vials in this study, citral might have been evaporating throughout the study. Thus, it is very likely that loss of citral caused the lower pAV in FBO compared to FAO samples, without significantly changing the rate constant (k ).
As a small study to estimate the influence of some common flavors on pAV testing in fish oil products, the current results do not allow us to calculate a pAV that solely reflects the aldehydes generated through lipid oxidation. Future studies should include a systematic experimental design that tracks multiple complementary oxidation products, such as peroxides and volatile aldehydes, and monitors matrix changes, such as tocopherol and flavor loss. Such a study would demonstrate the overall oxidative status, elucidate the correlation between various oxidation indicators, and clarify the source of the diminishing pAV inflation. This would also allow us to define the maximum pAV through its correlation with other oxidation markers and their respective limits. When pAV is the only applicable method for routine analysis for quality control purpose, this predefined pAV can be used as the control limit. For example, if we are confident about the correlation between 2,4-heptadienal and pAV after systematic studies on a flavored fish oil product under normal storage conditions, and if we have established a standard for 2,4-heptadienal levels, we can predefine our maximum allowable pAV for this product. For daily quality control, pAV can be measured and compared with this maximum pAV.