**4. Discussion**

RPE is a key regulator of retinal function and is directly related to the transport, delivery, and metabolism of n-3 PUFA in the retina. For this reason, we aimed to evaluate the effect of different formulations of DHA and EPA on this important cell type. Both n-3 PUFAs produced favorable

effects on RPE cells by increasing cell viability and proliferation, reducing the production of ROS and decreasing oxidative damage induced by H2O2.

Recent research has demonstrated that the long chain n-3 PUFA, has antiangiogenic, anti-vasoproliferative, and neuroprotective actions on factors and processes implicated in the pathogenesis of degenerative/vascular retinal diseases of greatest public significance, including DR [18]. It has been demonstrated that diets with high levels of n-3 PUFA, have known anti-inflammatory properties [36] due to their ability to promote the gene expression of various inflammatory mediators via di fferent intracellular signaling pathways [37] leading to the inhibition of the expressions of pro-inflammatory cytokines, leukocyte chemotaxis, and adhesion molecules. It is also known that they regulate the production of eicosanoids such as prostaglandins and leukotrienes and increase the synthesis of anti-inflammatory mediators such as resolvins, protectins, and maresins [38]. All these anti-inflamatory mechanisms add to their antioxidant e ffects by increasing the bioavailability of nitric oxide (NO) and the expression of superoxide dismutase and glutathione peroxidase, known as endogenous antioxidant enzymes, while decreasing the level of biomarkers of oxidative stress, such as malondialdehyde, in type II diabetic patients [30].

Many of the n3 supplements used in the routine clinical practice have a combination of EPA and DHA. For this reason, we explored the e ffect of supplements that mixed EPA and DHA in di fferent proportions. Interestingly, EPA/DHA 40/20 TG was the only formulation that showed a significant increase in viability and proliferation of ARPE-19 cells and the most favorable antioxidant e ffect. This observation is in agreemen<sup>t</sup> with previous publications that have shown di fferent e ffects of omega-3 supplements, depending on the proportion of EPA/DHA. In a study conducted on Wistar rats, it was found that the dietary intervention with 1:1 and 2:1 EPA/DHA supplements were the most effective treatments to reduce inflammation and oxidative stress when compared with a 1:2 EPA/DHA formulation [39]. Similar results were reported in a study performed in spontaneously hypertensive obese rats, where EPA/DHA supplementation at the ratios of 1:1 and 2:1 were more e ffective than a 1:2 formulation, lowering plasma total cholesterol and LDL concentrations, decreasing inflammation, and increasing the activity of antioxidant enzymes [40]. The reasons that underlie these di fferences are not completely clear, but it has been suggested that the higher unsaturation level of DHA may increase the susceptibility of the molecule to be oxidized compared to EPA, rendering a higher level of free radicals [39]. Furthermore, di fferences in their influence on transduction pathways, the release of inflammatory cytokines, and the expression of genes involved in lipid metabolism have also been reported [40]. In the field of ophthalmology, one article compared the e ffect of two di fferent EPA/DHA formulations (1/4.5 and 1.5/1) as adjuvants to topical tacrolimus in a model of keratoconjunctivitis sicca (dry eye disease) in dogs [41]. Authors reported better clinical and biochemical outcomes after supplementation with the oral formulation containing a higher proportion of EPA.

However, the proportion of EPA/DHA is not the only factor that distinguishes one supplement from the other, and omega-3 formulations vary depending on whether they are present as TG, EE, or PL. In the body, long PUFA stores exist mainly as PL and TG and in the retina, the latter represents the predominant lipid class [18]. Although most of the large interventional studies in the field of omega-3 supplements have been conducted with EPA/DHA EE, several authors have commented on their lower bioavailability when compared to TG and PL formulations [42–44]. Possible explanations include di fferences in their digestion and absorption [42]. EPA and DHA may be obtained directly through the diet or can be biosynthesized from linoleic acid (an 18-carbon essential fatty acid) in the liver or the retina. However, the e fficiency of tissue accretion is highest when they are ingested in the preformed state [18]. n-3 PUFA are hydrolyzed by pancreatic enzymes and are re-esterified to triglycerides and phospholipids within the intestinal epithelium. These triglycerides and phospholipids are integrated to chylomicrons and very low density lipoproteins (VLDLs) which are transported to the choriocapillaris. Further transport from the choriocapillaris to the RPE and the inner segments of photoreceptors appears to be mediated by high a ffinity receptors [18]. Unlike TG and PL, which are hydrolyzed mainly by a colipase dependent pancreatic lipase, EE requires additional digestion with carboxyl

ester lipase, a step that can slow down their absorption by 10 to 50 times [44,45]. In our experiments, the supplement with EPA/DHA in the form of EE, did not improve cell viability or proliferation and showed a lower antioxidant e ffect, despite having the same proportion of EPA/DHA (40/20) than the TG supplement with the most favorable outcomes. These di fferences could be clinically relevant, as suggested by data from The Age-Related Eye Disease Study 2 (AREDS2), where supplementation with a 650 mg EPA/350 mg DHA EE formulation failed to show clinically significant benefits [46].

As recent evidence has suggested that dietary EPA-DHA PL are superior to TG and EE forms in exerting their functional mechanisms [43], we evaluated the e ffect of DHA formulations that combined TG and PL. The most relevant e ffect of these formulations was their antioxidant capacity. They decreased the production of ROS production, and increased viability and proliferation of cells challenged with H2O2. This e ffect is important as the oxidative stress in the human eye is also primarily due to H2O2, which is naturally generated in RPE cells by solar radiation and POS phagocytosis [47]. Regarding the origin of DHA PL, some authors have suggested that n3-PUFA from a marine origin might be more beneficial than those from a vegetable origin [39,43,48]. However, in our experiments we did not find an association between the origin of the DHA PL and their results. This is probably because the proportion of PL in our formulations was small (3% or 5%).

The above-mentioned beneficial e ffects of the n3-PUFA have been demonstrated in several clinical studies. In patients with DR and well-controlled diabetes, increasing the n3-PUFA intake was associated with a reduced likelihood of the presence and severity of DR [24]. A sub-study of the PREDIMED randomized clinical trial showed that patients with type 2 diabetes who reported an intake of at least 500 mg/d of long-chain n3-PUFA at baseline had a 46% decreased risk of sight-threatening DR compared to those not meeting this target [25]. In an early stage of DR, supplementation with a high-dose DHA plus xanthophyll carotenoid multivitamin during 90 days was associated with a progressive and significant improvement of macular function measured by microperimetry [34]. In normal ocular tissues, angiogenic homeostasis is controlled by the balance between angiogenic stimulators, mainly VEGF, and angiogenic inhibitors such as pigment epithelium derived factor (PEDF) [49]. Moreover, this balance is important in the regulation of vascular permeability. While VEGF is increased in the vitreous of patients with diabetic macular edema, the vitreous level of PEDF is significantly lower in these patients [50]. Therefore, therapeutic strategies that can lower the VEGF/PEDF ratio are clinically beneficial. In a randomized single-blind controlled trial, the addition of a DHA dietary supplement to intravitreal ranibizumab (a monoclonal antibody against VEGF) was e ffective to achieve better sustained improvement of central subfield macular thickness outcomes after three years of follow-up compared with intravitreal ranibizumab alone [26]. Interestingly, n3-PUFA formulations in our study produced a small decrease in the VEGF/PEDF ratio, but the results were not statistically significant. Further studies analyzing the expression of VEGF and PEDF mRNA could help clarify the effect of EPA and DHA supplementation on the VEGF/PEDF ratio and their apparent clinical benefit in the treatment of diabetic macular edema.

As with any other clinical intervention, the safety of omega-3 dietary supplements should be considered. A recent meta-analysis that specifically addressed the safety and tolerability of prescription omega-3 fatty acids did not find any definitive evidence of serious adverse events [51]. The most commonly reported treatment associated adverse reactions are digestive disturbances (mainly dysgeusia or fishy taste) and skin reactions (eruption, itching, exanthema, or eczema). Although both EPA and DHA reduce the TG levels, they can increase the concentration of low density lipoproteins (LDL) [51]. However, this mild, but negative e ffect in lipid profile has not been observed in patients treated with supplements that only contain EPA. Monitoring of the lipid profile may be advisable in patients undergoing omega-3 supplementation.

To our knowledge, there has not been an in vitro study or clinical trial comparing the e ffects of so many di fferent formulations of EPA and DHA on ARPE-19 cells, not only under normal conditions, but also following an oxidative challenge. It was very interesting to see that DHA and EPA had di fferent e ffects when applied separately than they did when used in combination. Another relevant finding of our study was that the EE formulation had worse results, which cannot be explained by a lower bioavailability due to di fferences in digestion and absorption. This suggests that the EE form might decrease the biological e ffect of EPA and DHA. However, we know that our in vitro study has certain limitations. As an in vitro study, cells were not exposed to the same physiological environment as in the functioning retina. Moreover, analyzing the e ffect of the supplements in other retinal cell types would have also been of interest.
