Introduction
It has been demonstrated that nutritional support by feeding patients specialized diets, as a result of its capacity to interfere with a variety of biological processes, can modulate the chain of inflammatory responses [
1‐
3]. Recent pharmaceutical interventions proposed for sepsis have sought to focus on regulating the chain of pro- and anti-inflammatory mediators [
4,
5], which are responsible for causing the systemic characteristics of the disease and consequently leading to multiple organ failure. The inflammatory reaction is capable of activating the synthesis of several lipid mediators which are involved in the complex regulation of the inflammatory process [
6].
Lipid mediators are synthesized by three main pathways, cyclooxygenase, 5-lipoxygenase and cytochrome P450, by using fatty acids such as arachidonic acid (AA), eicosapentaenoic acid (EPA) and γ-linolenic acid (GLA) as substrates [
7], but the biological anti-inflammatory activities of EPA and GLA are far beyond the simple regulation of eicosanoid production. For instance, EPAs can affect immune cell responses through the regulation of gene expression and subsequent downstream events by acting as ligands for nuclear receptors [
8] and as control transcription factors [
9]. EPA can also affect the activity of the proinflammatory transcription nuclear factor κB (NF-κB), which regulates the expression of many proinflammatory gene-encoding adhesion molecules, cytokines, chemokines and other effectors of the innate immune response system [
10]. Researchers have recently described that the two main active fish oil pharmaconutrients, EPA and docosahexaenoic acid, are substrates of two novel classes of mediators called resolvins and protectins [
11,
12], which are involved in the resolution of the inflammatory process [
13‐
15].
In 1999, Gadek and co-workers [
16] demonstrated that the use of a diet enriched with EPA, GLA and antioxidants can improve oxygenation status in patients with acute respiratory distress syndrome (ARDS). The same study demonstrated that patients nourished with this diet spend fewer days in the ICU as well as in the hospital. This diet was further evaluated in two additional studies published in 2006 [
17,
18]. Singer
et al. [
17] demonstrated the effectiveness of an EPA/GLA diet in improving oxygenation status and decreasing the ICU and hospital length of stay (LOS) of patients with acute lung injury (ALI). This diet was also associated with lower mortality rates on the basis of 28-day all-cause mortality. Similar results were observed by Pontes-Arruda
et al. [
18] in which such a diet was fed to patients with ARDS secondary to severe sepsis and/or septic shock.
In a recently published meta-analysis of outcomes, the three above-mentioned studies were combined [
19]. For the patients considered evaluable (
n = 296), the use of an EPA/GLA diet was associated with a 60% reduction in the risk of 28-day in-hospital all-cause mortality (odds ratio (OR) = 0.40, 95% confidence interval (95% CI) = 0.24 to 0.68;
P = 0.001). With regard to the effects of the use of an EPA/GLA diet upon mortality on the basis of intent-to-treat (ITT) analysis (
n = 411 patients), a 49% reduction in the risk of 28-day in-hospital all-cause mortality was evident (OR = 0.51, 95% CI 0.33 to 0.79;
P = 0.002). Available clinical evidence is also consistent in showing a reduction in time on mechanical ventilation as well as ICU and hospital LOS. In fact, a second meta-analysis including data from unpublished studies [
20] demonstrated similar results.
In analyzing the data together, it was clear that feeding patients an EPA/GLA diet was associated with a reduction in the development of new organ dysfunction. Although the Singer
et al. study [
17] did not assess this variable, both Gadek
et al. [
16] and Pontes-Arruda
et al. [
19] demonstrated reductions in this particular outcome. Combined, the feeding patients an EPA/GLA diet was associated with an 83% reduction in the development of new organ failure. If we accept that the development of new organ failure is the pathway that leads patients with sepsis to severe sepsis and septic shock, it appears logical to evaluate the possible benefits of treating patients with this nutritional intervention in the early stages of sepsis (defined as systemic inflammatory response syndrome associated with confirmed or presumed infection and without any organ failure) as a way to prevent the evolution or to slow progression of the disease. Moreover, if we consider that the development of multiple organ failure is associated with increased mortality rates, we can hypothesize that feeding patients a diet including an enteral formulation enriched with EPA/GLA might be a determining factor in reducing the mortality rate.
All previously published trials were performed in critically ill, mechanically ventilated patients with at least one organ failure. Since the effects of the EPA/GLA diet in patients without any organ failure remain uncertain, the aims of this clinical study were to evaluate the role of an enteral formulation enriched with EPA/GLA in patients diagnosed in the very early stages of sepsis, despite respiratory failure, and to compare the results with those obtained with the use of a standard ICU formulation, isocaloric and isonitrogenous to the study diet, and not enhanced with lipids but higher in carbohydrates.
Discussion
Currently available strategies for the treatment of patients with sepsis focus on the management of the more severe forms of the disease (severe sepsis and/or septic shock), when the patients already have multiple-system organ dysfunctions, which is always associated with elevated death risk [
30,
31]. The contradiction is that there is also a consensus that, when treating patients with sepsis, most, if not all, available strategies are time-dependent [
32].
Unfortunately, information about possible strategies to be applied in the early stages of sepsis is extremely limited, as is the number of trials evaluating possible therapies this early in the disease time line. The main purpose, and probably the most interesting characteristic of the present trial, was to apply a nutritional strategy that has proven to be of great value in the management of patients with late sepsis-associated respiratory failure [
1], and is associated with an important reduction in the development of new organ failures [
20], early in the evolutionary stage of sepsis when no organ dysfunction has been identified and to test whether this nutritional strategy can be used to help slow the progression of the disease.
In this study, we found an association of the use of an EPA/GLA diet with reduced incidence of severe sepsis and/or septic shock. Although the mechanisms underlying the results of the present trial are unclear, there are several possible roles associated with EPA, GLA and antioxidants which, acting alone or together, may help us to understand how this diet can lead to the observed effects, among them being the regulation of AA levels in the inflammatory cell membranes, downregulation of NF-κB, lower production of inflammatory eicosanoids, antioxidant-mediated reduction in overall reactive oxygen species levels and a quicker recovery from inflammation due to production of resolvins and protectins, among several others.
In this trial, we used a control diet different from those used in previously published works [
16‐
18]. Our control diet was not enhanced with lipids but was higher in carbohydrates and was considered a standard ICU diet. The reason for our use of that control diet was that one major focus of criticism of previous work is precisely the control diet that was used (Pulmocare; Abbott Nutrition). Maybe the main reason for such criticism is that a high-lipid enteral diet is not considered by many to be a standard diet for critically ill patients. It is a common misunderstanding that enteral formulas enriched with Ω-6 lipids such as linoleic acid (LA, or 18:2n6; for example, from corn oil) can produce upregulation of the inflammatory response and, for that reason, such formulas may not be used to feed patients with hyperinflammatory diseases. Production of AA from LA involves three key enzymatic steps: Δ-6-desaturation to form 18:3n6 (GLA), followed by elongation to 20:3n6 (dihomo-GLA (DGLA)) and a Δ-5-desaturation step to produce 20:4n6. Thus, LA-enriched formulas cannot worsen inflammation simply because the activity of the enzymes Δ-5-desaturase and Δ-6-desaturase are severely compromised during critical illness by the release of stress and catabolic hormones (for example, glucocorticoids and catecholamines) [
33,
34] limiting the ability to form AA despite the provision of LA [
35]. GLA is a metabolite of LA that can bypass the decreased expression of Δ-6-desaturase. Its elongation product, DGLA, is incorporated into the inflammatory cell membranes, and the formation of DGLA suppresses leukotriene biosynthesis and can be metabolized to form prostaglandin E
1, a potent pulmonary vasodilator [
36,
37]. DGLA is also metabolized by 5-lipoxygenase to form 15-hydroxyeicosatrienoic acid, which inhibits the formation of leukotriene B
4. In fact, an enteral diet enriched with GLA cannot increase AA levels in immune cell membranes [
38] and, on the other hand, can increase anti-inflammatory activity by incorporating DGLA into the immune cell membranes. Despite all of the above-mentioned evidence, the previous criticism of other studies highlights the importance of this trial as the first one to demonstrate a positive result in terms of primary and secondary outcomes comparing the EPA/GLA diet with a standard ICU formula.
In this study, we report a reduction in both ICU and hospital LOS for patients fed a diet with EPA/GLA compared to the control population. The differences represent a mean of 6.5 additional ICU-free days and 8.9 additional hospital-free days associated with the use of the study diet. These results are in accord with previously published studies [
16‐
18] in which EPA/GLA diets were used to treat critically ill patients requiring mechanical ventilation, all of which reported reductions in ICU and hospital LOS. In fact, a recent meta-analysis [
19] associated the use of EPA/GLA with a mean of 4.3 more ICU-free days compared to patients fed a control diet. Sepsis represents an important financial burden on the healthcare system [
39,
40], and any reduction in terms of LOS must be considered to have a potential economic impact regarding reductions in the overall cost of care.
It is important to note that the reported number of patients requiring mechanical ventilation (invasive or noninvasive) was slightly different in the study groups with regard to the number of patients considered to have developed respiratory failures since the end of the study. The reason for that finding was that, during the 28-day follow-up period, one patient in the study group and three patients in the control group required noninvasive mechanical ventilation but did not fulfill the PaO2/FiO2 ratio criterion for respiratory failure.
Previously published works [
17,
18] and the present study have many relevant characteristics in common. They all used a study diet containing carbohydrates and proteins together with high levels of EPA/GLA. They all started enteral nutrition as early as possible and delivered the diet continuously using an enteral feeding pump. In a recent study (the EDEN-Omega Study) conducted by the EDEN-Omega NHLBI ARDS Network, investigators reported different findings when treating critically ill patients [
41]. They found no benefit in using a module containing EPA/GLA to treat patients requiring mechanical ventilation, and the study was stopped because of futility. Comparison of the EDEN-Omega Study with the previously published works can lead to dangerous misinterpretation of the current evidence. First and foremost, the patients enrolled in the EDEN-Omega study received a bolus of EPA/GLA twice daily, not an enteral formula containing EPA/GLA as part of it. It is uncertain whether these pharmaconutrients act differently if not provided using continuous feeding and as part of an enteral nutrition formula. Absorption of individual macronutrients (such as lipids) at the intestinal level can be drastically affected by the presence or absence of other nutrients. Additionally, it has been demonstrated [
42] that the delivery of an EPA/GLA diet by continuous feeding can produce important changes in the production of inflammatory mediators associated with modulation of plasma phospholipid levels. It is not clear whether a bolus of EPA/GLA given twice daily can provide similar changes in plasma phospholipids. Unfortunately, to date, the EDEN-Omega NHLBI ARDS Network investigators have not made publicly available the plasma phospholipid measurements of the patients included in the EDEN-Omega study, data that are of pivotal importance to allow a fair trial comparison between their trial and those previously published. Moreover, the EDEN-Omega study did not control for several other variables, including the different levels of other macronutrients (such as proteins) in the underlying diets fed to their patients, producing a huge source of bias in their data. Finally, the EDEN-Omega study was not designed as an early intervention trial, but was in fact an early versus late intervention study, whereas early enteral feeding was described as a key factor in all previous trials. The currently available nutrition guidelines continue to unanimously recommend the use of an EPA/GLA diet in critically ill and mechanically ventilated patients with ALI/ARDS [
1,
20]; therefore, not to use this strategy in the indicated population of patients merely on the basis of the results of an unpublished single trial in which so many important questions remain to be answered does not appear to be justifiable.
Interaction between fish oil and sepsis is very complex and variable, depending on the dose administered and probably the route of administration as well. Most of the available evidence associated with the benefits of fish oil was produced when this active pharmaconutrient was used as part of enteral nutrition formulations. However, the number of trials testing parenterally administered fish oil in a variety of clinical situations, including sepsis, is growing [
43,
44]. For instance, in a recently published study using a fish oil-based lipid emulsion in patients with sepsis, Barbosa and co-workers [
45] demonstrated significant improvements in the PaO
2/FiO
2 ratio that were similar to the effects reported with the use of enteral nutrition.
Although in the present study we have demonstrated a reduction in the development of cardiovascular and respiratory failure in the patient population nourished with EPA/GLA as compared to the control population, this was not associated with a significant reduction in mortality. The probable reason for this finding is that this study included patients with early sepsis and no organ dysfunction, a clinical situation usually associated with lower mortality rates than are found in patients with severe sepsis and/or septic shock [
46]. The sample size of this trial was not calculated to demonstrate mortality differences, and this particular variable should be evaluated in future studies.
Some important limitations need to be considered when evaluating the results of the present study. One of the most important is that this study included only those patients in need of enteral nutrition. During the enrollment of the patients, the investigators considered this to be very relevant and to represent a limitation in terms of the number of potential patients who could be included in this trial, since most patients whom we found to be in the early stages of sepsis did not require enteral nutrition, being able to receive oral feeding. In addition, the study included only patients at the ICU, which reduced even more the number of potential patients to be included, since the majority of the patients with early sepsis are usually not at the ICU but in the general wards. As a result of these important limitations, we believe that the population of patients included in this study was somehow selected, being constituted by elderly patients, most of whom were already receiving treatment and enteral feeding at their homes because of previous limitations such as stroke and eventually developing diseases such as community-acquired pneumonia, therefore requiring hospital treatment. In addition, this study excluded patients with a BMI ≥ 29, and for this reason generalization of the results to the obese population of patients is not possible. Finally, it was not possible to report changes in the Δ or component SOFA scores, since this trial was not designed to collect daily SOFA scores. Therefore, it is uncertain whether the observed reduction in the development of severe sepsis and/or septic shock is associated with reductions in the SOFA score over time. This association represents an important variable to be evaluated in future trials in which an EPA/GLA diet is fed to septic patients.
It will be important to produce further evidence and reproduce the present work in a more broad population of patients in the early stages of sepsis using an EPA/GLA diet, maybe as an oral supplement, and not delivering these pharmaconutrients only when an enteral feeding tube is in place.
Acknowledgements
This study was partially supported by funding from Abbott Nutrition/Abbott Laboratories (Chicago, IL, USA). The INTERSEPT study was sponsored by the Centro de Estudos João Pompeu Lopes Randal (Fortaleza, Ceará, Brazil) with the coordination of the Latin American Sepsis Institute (São Paulo, Brazil). We thank Frederico Rafael Moreira (Bioestatística, Montreal, QC, Canada) for statistical analysis. We thank all participating institutions and investigators of the INTERSEPT Study Group for their perspective and guidance, especially the following institutions for having included patients in the study: Hospital Fernandes Távora (Fortaleza, CE, Brazil), Hospital do Servidor Público Estadual de São Paulo (São Paulo, SP, Brazil), Hospital Santa Luzia (Brasília, DF, Brazil), Hospital Pró-Cardíaco (Rio de Janeiro, RJ, Brazil) and Hospital Português (Salvador, BA, Brazil). The following hospitals approved the study and were registered as participating centers with the regulatory authorities but did not include any patients during their participation: Centro Hospitalar UNIMED (Joinville, SC, Brazil), Hospital Dona Helena (Joinville, SC, Brazil), Hospital de Base de São José do Rio Preto (São José do Rio Preto, SP, Brazil), Hospital São Paulo/UNIFESP (São Paulo, SP, Brazil), Clínica São Vicente (Rio de Janeiro, RJ, Brazil), Hospital Salvador (Salvador, BA, Brazil) and Hospital Universitário da Universidade Federal da Paraíba (João Pessoa, PB, Brazil). The results of this study have been published previously only in abstract form [
47,
48].
Competing interests
APA received speaker honoraria and a research grant from Abbott Nutrition International. MM received honoraria and participated in events sponsored by Pfizer and Novartis. LFM, SML, AMI, DT, ER and EBM have no conflicts of interest associated with the present work.
Authors' contributions
APA was responsible for the study design and trial registration, participated in the coordination of the study and wrote the manuscript. LFM, SML, AMI, DT, ER, MM and EBM collaborated in the collection of data for this study and in the revision of the manuscript. All authors read and approved the final manuscript.