This study set out to evaluate the effects of a lipid emulsion containing a mixture of MCT, soybean oil and fish oil on plasma phospholipid fatty acid profile, inflammatory mediators in plasma and produced by LPS-stimulated whole blood, routine biochemical and physiological markers, gas exchange and clinical outcomes in septic patients in the ICU. The control group received a 50:50 mix of MCT and soybean oil. This is the first study of this fish oil containing lipid emulsion (that is, Lipoplus) in septic patients in the ICU, although it has been used previously in post-surgery patients [17–19, 34, 35]. In these latter patients, Lipoplus was found to decrease production or concentration of inflammatory eicosanoids [17, 34] and cytokines  and to reduce length of hospital stay . A different fish oil containing lipid emulsion (Omegaven) has also been used in post-surgery patients where it decreased production or concentration of inflammatory eicosanoids  and cytokines , improved immune function [15, 16] and improved clinical outcomes [15, 16, 37]. Omegaven has also been used in septic patients [21, 22], in critically ill ICU patients  and in patients with severe acute pancreatitis . In some of these studies, use of Omegaven was associated with decreased inflammatory markers [21, 22, 25] and improved respiratory function . Heller et al.  used Omegaven in a heterogeneous group of patients including post-surgical, septic and trauma patients and identified a dose-dependent reduction in mortality predicted from SAPS II score at entry. However, a recent study reported no effect of parenteral nutrition including Omegaven on inflammatory markers, or on clinical outcomes including infections, ventilation requirement, or ICU or hospital stay in critically ill ICU patients .
The current study found that five-day infusion of a MCT, soybean oil, fish oil mixture providing 6.4 g fish oil/day (equivalent to 2.3 g EPA plus DHA/d), increased EPA in the plasma phospholipid PC by an average of 3.8-fold, with no significant effect on DHA content and that this was associated with improved gas exchange and a tendency towards a shorter length of hospital stay. These are important findings since they indicate that the use of such an emulsion in this group of patients will improve clinical outcomes in comparison with what is seen with the more standard mix of MCT and soybean oil.
The increase in EPA content of PC is consistent with the recent report of a 2.4-fold increase in EPA in plasma phospholipids in healthy subjects receiving this same emulsion over a period of five days . Likewise the lack of a significant change in either in DHA or arachidonic acid in plasma PC seen in the current study is consistent with what is reported by Simoens et al. . These observations would suggest that any clinical benefit seen from the emulsion is due to EPA rather than DHA.
The tendency towards a reduction in length of hospital stay seen here (Table 6) was not a result of shorter ICU stay, and is consistent with findings in post-surgery patients receiving parenteral fish oil [15, 16, 18]. A previous study using a different lipid emulsion in ICU patients reports reduced ICU stay with higher fish oil administration  but this study was not controlled and relied upon historical data for comparison. Thus, this is the first randomised, controlled study reporting reduced length of hospital stay in septic ICU patients as a result of use of a fish oil containing lipid emulsion. The average dose of fish oil administered in the current study (6.4 g/day or 0.09 g/kg/d) is consistent with the dose that Heller et al.  found to be clinically favourable (>0.1 g/kg/d).
The current study identified a benefit of parenteral fish oil on gas exchange (Table 5). This is consistent with the recent report by Wang et al.  using parenteral fish oil in severe acute pancreatitis patients and with findings in acute respiratory distress syndrome patients receiving enteral fish oil . The mechanism by which n-3 fatty acids improve respiratory function is not entirely clear, but recent work in the fat-1 mouse, which endogenously synthesizes n-3 fatty acids from dietary n-6 fatty acids, provides new information on this . When exposed to LPS intratracheally, fat-1 mice showed reduced leukocyte invasion, protein leakage and inflammatory mediator (thromboxane B2, macrophage inflammatory protein-2) levels in lavage fluid compared with wild type mice. Furthermore ventilator compliance was improved in the fat-1 mice. This study shows a close link between anti-inflammatory effects of n-3 fatty acids, in this case seen at the level of the lung, and improved respiratory function.
Patients receiving parenteral fish oil showed more of a marked reduction in plasma IL-6 concentration than those in the MCT/LCT group and they also showed a smaller reduction in the anti-inflammatory cytokine IL-10. These findings concur with observations in post-surgery patients where plasma IL-6 concentrations were lower with parenteral fish oil [16, 17]. These changes in plasma inflammatory markers may be part of the mechanism that explains the clinical benefits seen in this study. Differences in plasma TNF-α and IL-1β concentrations between the two groups were small, although significant.
In contrast to the effects on some plasma cytokines, parenteral fish oil did not affect LPS-stimulated production of inflammatory mediators from whole blood cultures. This contrasts with the observation of Mayer et al.  in septic ICU patients that LPS-stimulated production of inflammatory cytokines (TNF-α, IL-1β, IL-6, IL-8) by purified monocytes was lower in the fish oil group. However, the amount of fish oil and n-3 fatty acids used by Mayer et al. was much greater than the amount used in the current study (35 vs. 6.4 g fish oil/d; approximately 10 vs. 2.3 g EPA plus DHA/d). This may explain the difference in findings between the two studies.
In the current study the whole blood cultures responded well to LPS stimulation: the response to LPS increased with time in both groups. This is consistent with the recent observations of Kirchhoff et al.  who showed increased numbers of cytokine-positive monocytes following LPS stimulation of whole blood taken from patients with severe multiple injuries over the period 24 to 72 hours post-admission. Likewise, Heidecke et al.  demonstrated that in sepsis survivors there is an increase in LPS-stimulated production of IL-1β and IL-10 by monocytes over time. This recovery in cellular response appears to be associated with improved clinical outcome [39, 40]. The observation that a poor inflammatory response of cultured cells taken early in sepsis is associated with poor patient outcome [39, 40] seems to conflict with the many observations that a poor outcome is associated with higher concentrations of inflammatory cytokines in the circulation [41–43]. Thus there seems to be a miss-match between circulating pro- and anti-inflammatory cytokine concentrations which are elevated early in sepsis and the ability of leukocytes to produce pro- and anti-inflammatory cytokines which is impaired early in sepsis. Indeed, the current study indicates that, as plasma cytokine concentrations decline over time, the ability of leukocytes to produce those same cytokines when stimulated with LPS ex vivo increases. This seemingly paradoxical observation may be explained by considering the regulatory processes that occur to control inflammatory cytokine release. A strong inflammatory stimulus in vivo will lead to inflammatory cytokine production with an elevation in plasma cytokine concentrations. However, this will lead to negative feedback, for example inhibition of monocyte nuclear factor κB activation [44, 45]. Therefore, upon restimulation ex vivo, the monocytes are less responsive . Hence monocytes isolated from blood at a time when there is a high concentration of cytokines may show a low cytokine response when stimulated and vice versa.
The anti-inflammatory properties of n-3 fatty acids have been described and discussed in detail elsewhere [11, 47, 48]. The mechanisms involved include effects at the membrane level, on signal transduction pathways leading to transcription factor activation and altered patterns of gene expression, and on the pattern of lipid mediator generation. The discovery of resolvins generated from both EPA and DHA  has focussed attention on resolution of inflammation as a mechanism of action of n-3 fatty acids and on the differential effects of EPA and DHA on inflammatory processes. In the current study status of EPA, but not DHA, was increased in plasma PC, suggesting that the effects seen are due to EPA. EPA has been shown to decrease production of inflammatory eicosanoids and cytokines [see ] and is the precursor of inflammation resolving resolvin E1 . Thus EPA may exert effects on both the generation of inflammatory mediators and on the resolution of inflammatory processes.
Whatever the mechanism(s) involved, this study demonstrates that a parenteral nutrition regimen including fish oil at the level used here does not impair the recovery of the ex vivo response of monocytes, but enhances the reduction in plasma IL-6 and diminishes the reduction in plasma IL-10 concentrations seen in the control group. Given that poor outcome is associated both with high plasma concentrations of inflammatory cytokines, including IL-6 [41–43] and with impaired ex vivo monocyte responses to LPS [39, 40], the overall effects of fish oil seen in the current study appear to be of benefit.
Limitations of this study are its relatively small sample size, the difference in age between the two treatment groups (the average age of patients in the fish oil group was higher than in the MCT/LCT group), and the higher glucose supply in the fish oil group. However, despite the small sample size, significant effects on plasma phospholipid EPA content, plasma cytokines, and gas exchange parameters were observed. In order to account for the differences in age and glucose supply between the two groups, these were controlled for in statistical analysis of cytokines, gas exchange parameters and clinical outcomes. Even after accounting for the differences in age and glucose supply between the groups, effects of lipid emulsion on plasma cytokines and on gas exchange parameters remained significant and the trend for an effect on length of hospital stay was not altered.