Optimization of Ultrasonic-Assisted Extraction of Antioxidants in Apple Pomace (var. Belorusskoje malinovoje) Using Response Surface Methodology: Scope and Opportunity to Develop as a Potential Feed Supplement or Feed Ingredient

Apple pomace showed a relatively low content of crude protein and a moderately low content of crude fiber compared to conventional feed material. The proximate content of this variety of apple pomace is presented in Table 3. Previous studies have reported similar results of proximate analyses of apple pomace, which include low crude protein and crude fat in apple pomace [57,83,84,85,86,87,88,89,90,91]. However, this variety of apple showed scanty crude fiber content, only up to 14.6%, which is significantly lower than previously reported crude fiber content of dehydrated apple pomace [92,93,94]. While apple pomace may have low protein and fat content, enhancing its nutritional profile by ensiling it with urea or ammonia or fermenting transforms it into a considerable alternative feed ingredient for ruminants. This process elevates its feeding value to a level comparable to that of grass silage for beef cattle [83]. Based on the literature review that has been carried out for this article, the content of protein and fat of apple pomace from variety B. malinovje is comparable to that of citrus pulp [95,96,97]. The crude fat and crude fiber content of apple pomace has shown to be more similar to pumpkin and citrus pulp; however, pumpkin peel has much greater protein content [95,96,97,98,99,100]. The results of metabolizable energy and metabolizable protein align with previously reported results and are presented in Table 4 [83,84,89,90,91].

Regarding fatty acid content, linoleic acid and oleic acid are proven to be two major unsaturated fatty acids of apple pomace. These results align with previous studies on the fatty acid composition of apple pomace [7,101,102]. While oleic and linoleic fatty acids are dominant fatty acids in apple pomace, apple pomace’s low-fat content means that this by-product is not considered a rich source of these fatty acids compared to other waste materials, which may yield higher amounts [7].

For optimizing the yield of TPC, DPPH^•, and ABTS^•+, the chosen extraction method was ultrasound-assisted extraction as this technique is quite efficient, energy and time-saving, and it is suitable for extraction of heat-sensitive compounds. For this study, two independent variables were selected for the optimization of TPC, DPPH^•, and ABTS^•+ yield: extraction time and amplitude. RSM was employed to ascertain the optimal condition of independent variables, to create a prediction model, and to evaluate the impact of these two factors on the TPC and antioxidant yield. The independent variables are presented in Table 3. In Table 6, actual and predicted values of TPC, DPPH^•, and ABTS^•+ are displayed, with visual representation of the yields shown in Figure S1. Based on the data acquired and summary statistics involving p-value, F-value, and R², the quadratic model was the best fit for maximizing all three yields (Tables S9–S11).

For the optimization of TPC extraction yield, the model p-value implies that the model is significant, as presented in Table 7 and Table S3. In this case, A, B, and A² are significant model terms (factors) for the optimization of TPC yield, with A being time and B being the amplitude, as shown in Table 7. p-values indicate that both factors (time and amplitude) impact the yield of TPC. The predicted R² of 0.7614 is in reasonable agreement with the adjusted R² of 0.9401; i.e., the difference is less than 0.2 (Table S2). The high R² value of 0.9775 implies that approximately 97.75% of the variability in TPC results can be explained by independent variables in the model (Table S2). The adjusted R² is also very high, which suggests that even with multiple factors included, the model can explain about 94% of the variability of response (Table S2).

Regarding the optimization of DPPH^• yields, the model’s F-value of 45.05 and p-value (p = 0.0051) indicate that the model is significant, as can be seen in Table 7 and Table S4. A, B, A², and B² are significant model terms, which tells us that both time and amplitude influence DPPH^• yield (Table 7). Just like in the results for TPC, the predicted R² is in reasonable agreement with the adjusted R², with the difference being less than 0.2 (Table S2). In addition, high R² and adjusted R² values suggest that the model is a good fit and that it can explain a large percentage of variability of DPPH^• extraction yield (Table S2).

For maximizing ABTS^•+ yield, the p-value (p = 0.03) suggests the model created is significant, as presented in Table 7 and Table S5. In this quadratic model, B (amplitude) and A² are significant model terms. In addition, R² of the ABTS^•+ optimization model is high: 0.9552, suggesting that 95% variability in ABTS^•+ extraction yield can be explained by independent variables used in the model (Table S2). However, the predicted R² of 0.56 is not as close to the adjusted R² of 0.88 as one might normally expect (Table S2). For this model, reducing the number of terms could be helpful.

Final equations describing the extraction yield of TPC (mg GAE/g DW) (1), DPPH^• (2), and ABTS^•+ (3) were the following:

where Y1 represents the yield of TPC (mg GAE/g DW), Y2 represents the yield of DPPH.^• (µM TE/g DW), Y3– ABTS^•+ (µM TE/g DW), A—time, and B—amplitude.

In Figure 2, a three-dimensional (3D) response surface plot is represented to show the visual effect of factors on each response, TPC (a), DPPH^• (b), and ABTS^•+ (c) extraction yield, as well as the relationship between the two factors. From the plot, it can be concluded that increasing the amplitude negatively affected the yield of TPC (Figure 2a). This can also be concluded by Equation (1). At the same time, with the increase in the time of extraction, TPC extraction increased as well. The lowest amplitude (20%) and middle set time (17.5 min) provided the highest yield of TPC, while the lowest amplitude and highest set time were a close second.

Based on the given Equation (2), which helps us understand the impact of the independent variable on the response, in this case, DPPH^•, we can conclude that increasing amplitude has a negative effect on DPPH^• extraction yield. This has been further proved by a 3D response surface plot, which provides the visual of the factor’s individual and combined influence on the response (DPPH^•), which is shown in Figure 2b. Much like with the results of TPC, the lowest set amplitude (20%) accompanied by the middle set time 17.5 provided the highest DPPH^• results. This also implies there is a positive correlation between TPC and DPPH^• results.

Looking at Equation (3), we can see that increasing both time and amplitude has a negative effect on ABTS^•+ yield. This is visually shown in the 3D response surface plot, which is presented in Figure 2c. Based on this, we can also include that the correlation between ABTS^•+ extraction yield and TPC and DPPH^• is slightly less. The highest ABTS^•+yield was obtained with the lowest amplitude (20%) and lowest extraction time (5 min). The ABTS^•+ assay yielded the highest results with the shortest extraction time, unlike TPC and DPPH^•, where the highest results were achieved with medium extraction time. The decline in ABTS^•+ results associated with prolonged extraction time could be attributed to the decomposition of antioxidative compounds within the sample [111,112]. Furthermore, the highest ABTS^•+ results obtained with the shortest extraction time could be attributed to the rapid response of certain antioxidants, leading to increased activity during shorter extraction time, while other antioxidants may require a longer period of time to reach their optimal antioxidant activity [113]. It is important to emphasize that while both DPPH^• and ABTS^•+ provide valuable information about the antioxidant capacity of the material, they might not extract the same antioxidants [113]. In addition, the antioxidants they extract could exhibit different response times influenced by their distinct chemical properties and interactions with other components in the matrix [113]. Moreover, some antioxidants could be more sensitive to certain extraction conditions such as ultrasound intensity, temperature, extraction duration, etc. [114,115,116]. Differences in assay conditions, especially the concentration of radicals and reaction kinetics, might contribute to the highest antioxidant yield being achieved with different extraction times [113].

The results for TPC, DPPH^•, and ABTS^•+ from apple pomace correspond with results reported in previous studies carried out on different varieties of apple pomace [47,50,53,56]. In this study, TPC values for the variety B. malinovoje were lower than the results previously presented for this variety; however, DPPH^• results are in agreement [52]. While this study reports lower total phenolic content (TPC) results compared to previous findings, several factors might explain this variance. Differences in extraction techniques, methodologies, variations in the treatment of apple pomace, and seasonal variations could contribute to the observed differences in antioxidant content [117,118,119,120]. Understanding and recognizing the factors that influence is very important for the optimization of antioxidants of apple pomace.

In addition, the positive values of linear correlation coefficients indicate that analyses are positively correlated one with the other (TPC × DPPH^• = 0.78, p value: 0.01388; DPPH^• × ABTS^•+ = 0.52, p value: 0.1497 and TPC × ABTS^•+= 0.57 p value: 0.109); however, the correlation is not as high as expected. Previously reported correlation between results of TPC, DPPH^•, and ABTS^•+ of apple pomace has been higher [47]. The correlation between TPC and DPPH^• results is statistically significant. In addition, ABTS^•+ results are higher than DPPH^• results. This can be attributed to the fact that ABTS^•+ assay applies to hydrophilic and hydrophobic antioxidant systems, while the DPPH^• applies to hydrophobic systems only [121,122,123]. This can result in ABTS^•+ capturing more antioxidant capacity, leading to higher readings [121,122,123].

In addition, the difference in results between ABTS^•+ and DPPH^• could be explained by higher radical reactivity and reaction kinetics in ABTS^•+ assay than in DPPH^• where the reduction in radicals occurs more slowly [124,125,126,127]. Due to higher radical reaction, absorbance decreases faster too, which may lead to higher antioxidant activity values observed in the ABTS^•+ assay compared to the DPPH^• assay [111,112,113,122]. Antioxidant compound solubility can also lead to differences in results [128,129]. In addition, ABTS^•+ is more sensitive than DPPH^• assay, and it can detect antioxidants with low concentrations, which can further explain the higher results obtained [122]. The choice of solvent can play a significant role and influence the results as well. ABTS^•+ radicals are soluble in both organic and aqueous solutions, while DPPH^• radicals are soluble in organic mediums only [122,124,127]. Therefore, the choice of solvent can affect the solubility and reactivity of antioxidants and affect the results. Also, because the chemical structure of ABTS^•+ and DPPH^• radicals is different, it impacts their reactivity with antioxidants, so some antioxidants may be more efficient at scavenging ABTS^•+ radicals compared to DPPH^• radicals, which can result in higher readings in ABTS^•+ assay [113].