n-3 fatty acids, γ-linolenic acid, and antioxidants in sepsis

The usefulness of n-3 fatty acids, γ-linolenic acid and antioxidants in the critically ill is controversial. I propose that adverse outcome in the critically ill is due to excess production of proinflammatory cytokines and eicosanoids from polyunsaturated fatty acids (PUFAs), while generation of anti-inflammatory products of PUFAs may lead to a favorable outcome. Hence, I suggest that measurement of plasma levels of various cytokines, free radicals, and proinflammatory and anti-inflammatory products of PUFAs and correlating them to the clinical picture may pave the way to identify prognostic markers and develop newer therapeutic strategies to prevent and manage critical illness.


Introduction
An original study published in JAMA [1] and comments on the same in Critical Care [2] called into question the role or the involvement of ω-3 and ω-6 polyunsaturated fatty acids (PUFAs) and antioxidants as a therapeutic strategy in sepsis, systemic inflammatory response, and acute lung injury (ALI). Th e results of the OMEGA study -a randomized, double-blind, placebo-controlled, multi center trial in which all participants had complete follow-up and the intervention group received twicedaily enteral supplementation of n-3 fatty acids, γlinolenic acid (GLA), and antioxidants -did not show any improvement in the number of ventilator-free days or other clinical outcomes in patients with ALI, and in fact showed a trend towards increased mortality [1].
Th e results of the OMEGA study are in contrast to those of Gadek and colleagues, who reported an increase in 28-day ventilator-free days, a reduction in ICU length of stay and new organ failure when enteral diets were supplemented with omega-3 fatty acid, GLA, and anti oxidants [3]. Pontes-Arruda and colleagues reported a similar benefi cial eff ect of the same enteral preparation in the form of an increase in ventilator-free days, an increase in ICU-free days, and a 19% absolute reduction in mortality [4], and they performed a meta-analysis and suggested that eicosapentaenoic acid (EPA), GLA, and antioxidants have a benefi cial eff ect in patients with ALI or acute respiratory distress syndrome [5]. Singer and colleagues per formed a single-center, prospective, randomized, con trolled, unblinded study of an enteral diet enriched with EPA, GLA, and antioxidants on the respiratory profi le and outcome of patients with ALI, and reported that such a human diet in signifi cant amounts may be benefi cial in patients with ALI for gas exchange, respiratory dynamics, and requirements for mechanical ventilation [6].
Based on these studies [3][4][5][6], the Society of Critical Care Medicine and the American Society for Parenteral and Enteral Nutrition endorsed the use of these enteral supplements in patients with ALI [7]. Th e exact reason for the contrasting results obtained in the above-mentioned studies [1][2][3][4][5][6] is not clear, but could be attributed to the diff erences in the stage(s) at which the enteral preparation was started, the heterogeneity of the patients included in the studies, the fat load, diff erences in the level of oxidative stress between diff erent patient populations, the tissue content of various unsaturated fatty acids prior to the administration of an enteral diet enriched with EPA, GLA, and antioxidants, and the receipt of fi sh oil in acute respiratory distress syndrome with or without full nutrition provision -because all of the positive results for fi sh oil in acute respiratory distress syndrome occurred in the setting of full nutrition provision and all of the negative trials occurred when the fatty acid supplements were administered independent of complete nutrition provision.
I propose that the diff erences in the results reported in various studies [1][2][3][4][5][6] could be attributed to the complex nature of the metabolism of PUFAs, and their interaction(s) with cytokines, free radicals and nitric oxide. Since PUFAs form precursors to both pro infl ammatory and anti-infl ammatory products that have a modulatory infl uence on the generation and action of cytokines, free radicals and nitric oxide [8][9][10][11][12][13][14][15][16][17], enhanced production of proinfl ammatory products may probably lead to an adverse outcome, whereas generation of antiinfl ammatory products of PUFAs could have a benefi cial action in sepsis and other critical conditions. In this context, it is important to know the metabolism of GLA, an n-6 fatty acid, and EPA and docosahexaenoic acid (DHA), the main n-3 fatty acids present in the supplement used in these studies.

Metabolism of essential fatty acids
GLA is not present in the human diet and has to be formed from the dietary essential fatty acid (EFA), cis-linoleic acid, unless and otherwise one is consuming oils that are rich in GLA, such as evening primrose, black currant and/or borage oil. On the contrary, n-3 fatty acids EPA and DHA could be obtained from the diet since they are present in signifi cant amounts in marine fi sh. Vege tarians do not get signifi cant amounts of EPA and DHA in their diet but can form them from the dietary EFA; namely, α-linolenic acid. Both cis-linoleic acid and α-linolenic acid are desaturated and elongated by the same set of enzymes (see Figure 1) to form their respective long-chain PUFAs such as GLA, dihomo-γ-linolenic acid (DGLA) and arachidonic acid (AA), and EPA and DHA, respectively.
DGLA is the precursor of 1-series prostaglandins (PGs), whereas AA is the precursor of 2-series PGs and thromboxanes (TXs) and 4-series leukotrienes (LTs). EPA is the precursor of 3-series PGs and TXs and 5-series of LTs. Most of the PGs, TXs, and LTs are proinfl ammatory in nature. One should, however, note that prostacyclin (PGI 2 ), a benefi cial PG that has potent platelet antiaggregatory and vasodilator actions, is formed from AA. Although PGs, TXs, and LTs are also formed from EPA, they are less proinfl ammatory in nature compared with the PGs, TXs, and LTs formed from AA. One should, however, understand that PGs, TXs, and LTs formed from EPA still are proinfl ammatory in nature [8,9].

Polyunsaturated fatty acids, cytokines, and free radicals
Studies showed that GLA, DGLA, EPA, and DHA suppress the production of proinfl ammatory cytokines IL-6 and TNFα both in vitro and in vivo [10][11][12], which led to the suggestion that these fatty acids could be used to suppress excess infl ammation and restore normalcyforming the basis of testing GLA, EPA, and DHA supplementation in ALI, sepsis, and other systemic infl ammatory conditions. Worthy of note, however, is that metabo lism of GLA, DGLA, AA, EPA, and DHA is not as simple as outlined above. For instance, in the presence of free radicals, these fatty acids (especially AA) could give rise to the formation of F2-isoprostanes, a group of proinfl ammatory substances that are formed in a nonenzymatic fashion (see Figure 1). In a similar fashion, EPA is also vulnerable to the action of free radicals and gives rise to six classes of F3-isoprostanes, α-linolenic acid and GLA to two classes of E1-isoprostanes and F1isoprostanes, and DHA to eight classes of D4-isoprostanes and eight classes of E4-isoprostanes. Each of these classes comprise up to eight racemic isomers, leading to an astounding number of isoprostane molecules [13]. Nitric oxide can react with unsaturated fatty acids to form nitrolipids that have anti-infl ammatory action [14], while reactive oxygen species and nitrogen free radicals can also attack unsaturated fatty acids to form trans-fatty acids (especially AA to form trans-AAs) that have proinfl ammatory action [15].

Formation of anti-infl ammatory bioactive lipids from polyunsaturated fatty acids
In addition, AA is the precursor of lipoxin A 4 (LXA 4 ), a potent anti-infl ammatory molecule. LXs are formed via a transcellular biosynthetic mechanism by platelets that depend on leukocytes for LTA 4 , which is converted to LXA 4 and LXB 4 by the action of platelet 12-lipoxygenase enzyme. An analogous class of anti-infl ammatory bioactive lipids, designated resolvins, is derived from EPA and DHA. Furthermore, DHA also forms the precursor to another class of anti-infl ammatory compounds called protectins (see Figures 2 to 5 for the formation of LXs, resolvins and protectins and their structures) and maresins.

Polyunsaturated fatty acids may have a complex role in sepsis and other critical illnesses
Based on the preceding discussion, it is clear that PUFAs are converted into a diverse range of molecules that have very many actions. Hence, when enteral supplementation of n-3 fatty acids, GLA, and antioxidants is given, it is virtually impossible to know what products have been generated from the administered fatty acids. Furthermore, it is known that several factors and cofactors modulate the activity of desaturases and elongases infl uencing the formation of AA from the GLA administered. Similarly, the activity of cyclooxygenase and lipoxygenase enzymes is also infl uenced by several factors that, in turn, infl uence the formation of both proinfl ammatory and anti-infl ammatory molecules. For instance, it is known that TNFα induces an EFA-defi ciency-like state [20]. Hence, in sepsis and ALI in which plasma TNFα levels are high, an alteration could probably occur in EFA metabolism and the products derived from various PUFAs. During the formation of various PGs, TXs, LTs, LXs, resolvins, and protectins there is a critical role for free radicals, and therefore use of antioxidants along with PUFAs may complicate the issue further. Th is eff ect is evident from the observation that treatment with n-3 PUFAs (EPA and DHA), but not vitamin E, signifi cantly lowered the risk of death, nonfatal myocardial infarction, and stroke, the primary endpoint in the GISSI-Prevenzione trial. Th us, addition of vitamin E do not seem to have any benefi t [21], calling into question the benefi ts of antioxidants.

Lipoxins, resolvins, protectins and maresins have cytoprotective and anti-bacterial actions
Worthy of note is the fact that resolvins and protectins protect kidneys exposed to ischemia/reperfusion injury in mice by reducing the number of infi ltrating leukocytes and blocking toll-like receptor-mediated activa tion of macrophages [22]. Th is observation implies that acute renal injury that is likely to be seen in sepsis and other critical illnesses may be prevented provided adequate amounts of anti-infl ammatory LXs, resolvins, protectins, and maresins are generated either by the target tissue/ organ or by their administration in a timely manner.
Furthermore, Chiang and colleagues showed that resolvins and protectins were present predominantly in self-resolving Escherichia coli infection mice [23]. Adminis tration of resolvins reduced bacterial titers in blood and exudates, abrogated E. coli-induced hypo thermia, and improved survival of mice. Resolvins enhanced phagocytosis of E. coli, enhanced the antibacterial actions of ciprofl oxacin and vancomycin, and augmented the clearance of bacteria. Th ese results suggest that synthesis and release of optimal amounts of anti-infl ammatory LXs, resolvins, and protectins from PUFAs may contain bacterial infections and lower antibiotic requirements for bacterial clearance. Th is suggestion implies that optimal produc tion of LXs, resolvins, protectins, and maresins from various PUFAs is of paramount importance to contain infections, clear the invading organisms, and resolve the detrimental infl ammatory process and prevent tissue damage.

Balance between proinfl ammatory and anti-infl ammatory products of PUFAs infl uences outcome of illness
Under normal conditions there is a balance maintained between proinfl ammatory and anti-infl ammatory molecules, and when this balance is upset more in favor of the former it would lead to persistence of infl ammation and progressive cell/tissue and organ damage. In sepsis, ALI, and other systemic infl ammatory conditions, therefore, not only could an increase in proinfl ammatory molecules such as IL-6, TNFα, high-mobility group box 1, PGs, TXs, LTs, isoprostanes, trans-fatty acids and free radicals probably occur, but also a defi ciency in the production and action of anti-infl ammatory molecules such as IL-4, IL-10, LXs, resolvins, protectins, and maresins.  Th is hypothesis is supported by the observation that lipopolysaccharide or platelet-activating factor-induced intravascular volume loss (microvascular fl uid leak) was attenuated by LXA 4 , which may be partly mediated by the c-Jun N-terminal kinase signaling pathway [24]. In the cecal ligation and puncture model of sepsis, when LXA 4 (40 μg/kg, intra peritoneally) was given 5 hours after surgery, the 8-day survival increased and tissue injury was attenuated after 8 days. Plasma IL-6, monocyte chemo tactic protein-1, and IL-10 levels were reduced in LXA 4 -treated rats compared with control. LXA 4 reduced phosphorylation of the p65 subunit of NF-κB at serines 536 and 468 in peritoneal macrophages, reduced the blood bacterial load, and increased the peritoneal macrophage number without aff ecting phagocytic ability, suggesting that LXA 4 reduced the blood bacterial load by enhancing macrophage function and that LXA 4 increased survival in sepsis by simultaneously reducing systemic infl am mation as well as bacterial load [25]. Th e lipid mediators LXs, resolvins, protectins, and maresins have thus emerged as novel potent and stereoselective players that counter-regulate excessive acute infl ammation and stimulate molecular and cellular events that induce resolution of infl ammation.
Th is implies that in sepsis, ALI, and other systemic infl ammatory diseases, provision of PUFAs -which are precursors of both proinfl ammatory and anti-infl am matory bioactive lipids alone -may not be adequate to suppress the infl ammatory process that is in progress since one is not certain which products are being generated from these precursor lipids. In such a scenario, it is worthwhile to administer the anti-infl ammatory bioactive lipids LXs, resolvins, protectins, and maresins themselves or their stable synthetic analogs to ward off the infl ammatory process [16,26] (see Figure 1).

Clinical implications
In light of the preceding discussion, it evidently would have been informative if the investigators of various clinical trials [1][2][3][4][5][6] measured plasma levels of both proinfl ammatory and anti-infl ammatory cytokines, various PGs, TXs, LTs, isoprostanes, trans-fatty acids, LXs, resolvins, protectins, and maresins, nitrolipids, and isoprostanes at various stages of the disease(s) and correlated them with the clinical picture and outcome. Such an attempt was made by Stapleton and colleagues in their study while performing a phase II randomized placebocontrolled trial of n-3 fatty acids for the treatment of ALI in which broncho alveolar lavage fl uid IL-8 levels were measured, and they found no signifi cant diff erence between the placebo group and the treatment group [27]. In their study, however, Stapleton and colleagues did not measure other cytokines and various eicosanoids including LXs, resol vins, and protectins.
Measurement of various proinfl ammatory and antiinfl ammatory bioactive molecules would have given an indication of the products generated during the diff erent stages of the illness and how supplementation of n-3 fatty acids, GLA, and antioxidants altered their concentrations and ultimately infl uenced the outcome. For instance, it Based on the changes in concentrations of proinfl amma tory and anti-infl ammatory cytokines, various PGs, TXs, LTs, isoprostanes, trans-fatty acids, LXs, resolvins, protectins, maresins, and nitrolipids in the critically ill, one would certainly be in a better position to predict the outcome and institute appropriate intervention(s) (such as administration of AA, EPA, DHA, LXs, resolvins, protectins, maresins, IL-4, IL-10, antibodies to IL-6 or TNF, low-dose aspirin, and statins) to facilitate recovery and pave the way for resolution of the infl ammatory process. In view of this, I propose that it is essential to measure concentrations of proinfl ammatory and antiinfl am matory cytokines, various PGs, TXs, LTs, iso prostanes, trans-fatty acids, LXs, resolvins and protectins, maresins, nitrolipids in the plasma and other tissue fl uids to understand the role(s) of these bioactive molecules in the infl ammatory process and the dynamic alterations in their levels at various stages of the illness (see Table 1). Such a study would not only lead to a better understanding of the infl ammatory process itself, but would also help to elicit the exact role of cytokines and bioactive lipids in various stages of illness and how their alterations are associated with various clinical indices and, fi nally, the outcome in sepsis, ALI, and other systemic infl amma tory diseases. We should consider administration of LXs, resolvins, protectins, and maresins or their stable syn thetic analogs in these conditions to suppress inappro priate infl ammatory process and facilitate or institute resolution of the progressive infl ammation. If proinfl ammatory cytokines , infl ammation is dominant; if anti-infl ammatory cytokines , recovery process is on the anvil; correlation needs to be made among cytokine profi le, bioactive lipids and clinical picture Shown are bioactive lipids and cytokines that could be measured at various time points in patients with sepsis and other critical illnesses and correlated with prognosis. These bioactive lipids/cytokines and other compounds could also be measured in urine at diff erent time points and correlated with their plasma levels and prognosis of the patient. Although not mentioned in the table, other measurements that could be performed include lipid peroxides, nitric oxide, antioxidants (such as superoxide dismutase, glutathione, catalase) and adipose-fatty acid binding protein. AA, arachidonic acid; ALA, α-linolenic acid; DGLA, dihomo-γ-linolenic acid; DHA, docosahexaenoic acid; ELISA, enzyme-linked immunosorbent assay; EPA, eicosapentaenoic acid; GC, gas chromatography; GLA, γ-linolenic acid; HPLC, highperformance liquid chromatography; LA, linoleic acid; LC-MS, liquid chromatography-mass spectrometry; LT, leukotriene; MS-MS, tandem mass spectrometry; PG, prostaglandin; PUFA, polyunsaturated fatty acid; TX, thromboxane.
It is feasible to clinically administer anti-infl ammatory bioactive lipids LXs, resolvins, protectins, maresins, and nitrolipids by themselves or as their synthetic analogs by coupling them to human albumin, so that the albumin could render the bioactive lipids stable and their actions would not be too drastic because albumin is expected to release these bioactive lipids slowly. Also pertinent is to consider giving fi sh oil and/or GLA-EPA/DHA supplementation enterally. We recommend that studies need to be performed to deter mine the most suitable and clinically more useful method of administration of fi sh oil by comparing intravenous versus enteral feeding of fi sh oil and/or GLA-EPA/DHA supplementation.