Intensive Care Med (1998) 24: 1029±1033 Ó Springer-Verlag 1998 ORIGINAL

V.Smirniotis Long chain versus medium chain G.Kostopanagiotou J.Vassiliou lipids in patients with ARDS: N.Arkadopoulos P.Vassiliou effects on pulmonary haemodynamics A.Datsis E.Kourias and gas exchange

Abstract Objective: To compare tion was only associated with an ele- Received: 23 July 1997 Accepted: 15 May 1998 pulmonary haemodynamic and gas vation of oxygen consumption exchange alterations in septic pa- (VO2) from 329 ± 14 to 396 ± 12 ml/ tients with ARDS receiving long- min. During lipid infusion group 1 chain triglycerides (LCT) versus patients presented higher Qva/Qt medium-chain triglycerides (MCT). (37% ± 6% vs 25 % ± 4%), MPAP Design: Prospective, randomised, (33 ± 4vs27±3 mmHg) and VO2 clinical study. (359 ± 11 vs 396 ± 12 ml/min) and Setting: Surgical ICU patients in a lower PaO2/FIO2 (180 ± 35 vs University Hospital. 235 ± 30) values compared to Patients: Twenty-one septic patients group 2. with ARDS were randomly assigned Conclusion: In conclusion, we have to receive 50 % of their non-protein shown that, in septic patients with caloric requirements as either 20 % respiratory failure, LCT administra- LCT (group 1, n = 10) or 20% 1 : 1 tion was associated with more sig- mixture of LCT/MCT (group 2, nificant changes of Qva/Qt, MPAP n = 11). and PaO2/FIO2 compared to infu- Intervention: Intravenous infusion of sion of an LCT/MCT 1 : 1 emulsion. V.Smirniotis ()) × G.Kostopanagiotou × LCT and LCT/MCT combinations at Clinically, these transient alterations J.Vassiliou × N.Arkadopoulos × a rate of 12 g × hŸ1. might cause serious problems in pa- P.Vassiliou × A.Datsis × E.Kourias 2nd Department of Surgery, Measurements and results: The LCT tients with marginal arterial oxyge- University of Athens Medical School, infusion was associated with an in- nation and cardio-respiratory im- ªAreteionº Hospital, crease of pulmonary venous admix- pairment. Athens, Greece ture (Qva/Qt) from 24% ± 5%to 37% ± 6%, an increase of mean Key words Fat emulsion × Medium Mailing address: pulmonary artery pressure (MPAP) chain triglycerides Long chain 22 Hanioti St. GR-15452, Athens, Greece × Tel.: + 30-1-728-6000 from 25 ± 5to33±4 mmHg and de- triglycerides × Pulmonary Fax: +30-1-721-1007 crease of PaO2/FIO2 from 240 ± 30 haemodynamics × Gas exchange × email: [email protected] to 180 ± 35. LCT/MCT administra- ARDS

Introduction cleared from the serum and do not contribute to the adi- pose stores [13]. They are independent of carnitine acyl- Many studies have been carried out to examine the me- transferase and are easily transferred into mitochondria tabolic properties of long chain triglyceride (LCT) for beta-oxidation [15]. MCTs may also have a greater emulsion [1±6]. Recently, medium chain triglyceride nitrogen-sparing effect than LCTs [12]. In addition, (MCT) emulsion have emerged as an alternative energy LCTs have been shown to decrease arterial oxygen ten- source which may have significant advantages over stan- sion (PaO2) and increase pulmonary venous admixture dard LCT formulas [3±4, 6±15]. MCTs are rapidly (Qva/Qt) and mean pulmonary artery pressure (MPAP) 1030

Table 1 Clinical characteristics of the patients Group 1 (LCT) Group 2 (LCT/MCT) Sex/ Weight (kg)/ Cause Sex/ Weight (kg)/ Cause Age (year) Height (cm) Age (year) Weight (cm) M/65 70/175 Oesophagectomy M/75 70/180 Oesophagectomy M/45 54/170 Necrotising pancreatitis M/40 68/170 Necrotising pancreatitis F/68 65/170 Small bowel necrosis M/56 72/175 Colon resection M/42 70/175 Colon resection F/60 60/160 Colon resection F/62 60/155 Colon resection F/70 71/165 Rupture of aortic aneurysm M/72 58/165 Necrotising pancreatitis M/40 63/170 Colon resection F/38 68/170 Pancreatic resection M/48 73/165 Pancreatic resection F/65 71/165 Necrotising pancreatitis M/62 70/180 Multiple trauma M/45 52/180 Pancreatic resection F/72 56/180 Splenectomy M/48 68/181 Necrotising pancreatitis M/68 55/160 Necrotising pancreatitis F/42 60/170 Necrotising pancreatitis

[1, 5, 7, 13, 16]. These side effects appear to be aggravat- 2. Mean pulmonary artery pressure (MPAP), pulmonary artery oc- ed by the co-existence of sepsis [1, 8, 15]. LCT/MCT clusion pressure (PAOP), and mean systemic arterial pressure emulsions contain less linoleic, linolenic and arachido- (MAP) were measured using a Swan-Ganz catheter (Opticath P7110, Abbott Critical Care Systems, North Chicago, Ill.) and a nic acid compared to pure LCTs [17±18]. These differen- radial artery catheter (20G, Arrow, Reading, Pa.). ces may be important because the aforementioned acids 3. Thermodilution cardiac output was calculated from three elec- are precursors of prostaglandins and other eicosanoids, tronically integrated temperature decay curves, generated after which have been shown to affect pulmonary haemody- indicator injection (monitor: Horizon 2000, Mennen Medical namics, gas exchange and intrapulmonary shunt quo- Ltd, Rehovot, Israel). tient [17, 19±21]. 4. Qva/Qt and oxygen consumption (VO2) were calculated via standard formulas. We hypothesised that, in patients suffering from in- tra-abdominal sepsis complicated by ARDS, LCT/ MCT mixtures may be associated with less prominent Study design changes of pulmonary haemodynamic and gas exchange parameters compared to pure LCT emulsion. Twenty-one consecutive post-operative ICU admissions with sep- sis-related ARDS were randomly assigned to one of two groups: group 1 (n = 10) or group 2 (n = 11) (Table 1). In group 1, 50% of non-protein caloric requirements was infused as a 20% LCT emul- Material and methods sion (Intralipid, Pharmacia). In group 2, the corresponding energy needs were administered as a 20% mixture of LCTs/MCTs (Lipo- All procedures were conducted in full compliance with the stan- fundin MCT/LCT, B.Braun Melsungen AG, Melsungen, Ger- dards of the Institutional Committee for the Protection of Human many). Fat emulsion was intravenously infused over an 8-h period Subjects, in accordance with the Helsinki Declaration of 1975. using a pump (Life Care Pump, model 4, Abbott Laboratories, Our survey included patients with respiratory failure due to in- North Chicago, Ill.) at a rate of 12 g × hŸ1. The remaining 50% of tra-abdominal sepsis (Table 1). All patients met the standard cri- the non-protein energy requirements were covered with glucose, teria for diagnosis of ARDS, presenting lung injury scores of infused continuously i.v. over a 24-h period. more than 2.5 [22]. Sepsis was confirmed by clinical and laboratory Measurements were obtained on the third post-operative day findings as well as positive blood cultures [23]. Patients with pre- at three time points: existing cardiovascular, respiratory, renal or metabolic diseases were excluded from the study. Mechanical ventilation was insti- 1. 30 min before the 8-h lipid infusion was started (before), tuted via an endotracheal tube using a volume-cycled ventilator 2. 30 min before the infusion was completed (during) and (Puritan 7200 Series, Bennett Corporation, Carlsbad, Calif.) and 3. 4 h following suspension of lipid infusion (after). was titrated to maintain arterial PCO2 within the normal range. Pulmonary end-expiratory pressure (PEEP) levels were kept at Each sample was the average of three measurements. No further 8±15 cm H2O (12 ± 2; mean ± SD). All patients received low mole- therapeutic interventions were undertaken during the study peri- cular weight heparin for the prevention of deep vein thrombosis od. and low-dose (2 mg × kgŸ1×minŸ1 i.v.) dopamine. All patients re- ceived total parenteral nutrition and antibiotics. Sedation was maintained with fentanyl (100±200 mg × hŸ1 i.v.) and midazolam Statistical analysis (1±2 mg × hŸ1i.v.). The following measurements were obtained: The comparison of measurements in each cohort was performed 1. Arterial PO2, PCO2, haemoglobin concentration and haemoglo- using a non-parametric Wilcoxon signed rank test for paired data. bin oxygen saturation were measured using a blood gas analyser Mann-Whitney-U test was used to compare group 1 versus (ABL 300, Radiometer Medical A/S, Copenhagen, Denmark). group 2. Statistical significance was accepted if p was less than 0.05. 1031

Table 2 Pulmonary haemodynamic parameters during lipid infu- occlusion pressure, CO cardiac output, Before 30 min before the sion (HR heart rate, MPAP mean pulmonary artery pressure, infusion was started, During 30 min before the infusion was com- MAP mean systemic arterial pressure, PAOP pulmonary artery pleted, After 4 h following suspension of infusion) Group 1 (LCT) Group 2 (LCT/MCT) Before During After Before During After HR (/min) 125 ± 12 110 ± 10 115 ± 8 115 ± 10 126 ± 10 118 ± 8 MPAP (mmHg) 25 ± 533±4a,b 26 ± 626±627±325±5 MAP (mmHg) 90 ± 20 85 ± 15 90 ± 15 90 ± 20 95 ± 15 80 ± 15 PAOP (mmHg) 11 ± 213±210±211±29±310±3 CO (L/min) 8.6 ± 2.5 9 ± 2 9.5 ± 2.5 9.6 ± 2.1 10.5 ± 1.5 8.9 ± 2 Values are presented as mean ± SD a p K 0.05 vs before and after lipid infusion for the same cohort (Wilcoxon signed rank test) b p K 0.05 vs group 2 during lipid infusion (Mann-Whitney U-test)

Table 3 Gas exchange parameters during lipid infusion (PaO2 arterial PO2, Qva/Qt pulmonary venous admixture, FIO2 fractional in- spired oxygen, VO2 oxygen consumption) Group 1 (LCT) Group 2 (LCT/MCT) Before During After Before During After pH 7.36 ± 0.7 7.35 ± 0.9 7.36 ± 0.8 7.38 ± 0.2 7.35 ± 0.8 7.35 ± 0.9 a,b Qva/Qt (%) 24 ± 537±625 ± 423±525±424±4 a,b PaO2/FIO2 240 ± 30 180 ± 35 235 ± 35 240 ± 30 235 ± 30 235 ± 25 b a VO2 (ml/min) 333 ± 12 359 ± 11 338 ± 10 329 ± 14 396 ± 12 336 ± 13 Values are presented as mean ± SD a p < 0.05 vs before and after infusion in the same cohort b p < 0.05 vs the other cohort during infusion time

Results During lipid infusion, group 1 patients presented higher Qva/Qt and MPAP as well as lower PaO2/FIO2 The data are expressed as means ± SD. No statistical dif- and VO2 compared to group 2 (p < 0.05). In both ference was detected between the two groups regarding groups, no statistically significant changes of mixed ve- sex distribution (group 1: six male, four female; group 2: nous oxygen saturation were recorded between the dif- seven male, four female), age (group 1: 55 ± 4 years; ferent time-points. group 2: 58 ± 4 years), weight (group 1: 64 ± 2 kg; group 2: 65 ± 2 kg), height (group 1: 171 ± 2 cm; group 2: 170 ± 2 cm) and APACHE II score on the day of lipid in- Discussion fusion (group 1: 16 ± 3; group 2: 18 ± 4). Caloric require- ments were estimated at 36 ± 3 kcal × kgŸ1 ×dayŸ1 in Long chain triglyceride administration has been linked group 1 and 35 ± 3 kcal × kgŸ1×dayŸ1 in group 2 accord- to a number of adverse effects on pulmonary haemody- ing to Long's modification of the Harris-Benedict for- namics and gas exchange functions. In healthy volun- mula [24]. Nitrogen intake was 0.32 ± 0.03 g × kgŸ1 × teers, LCT infusion has been shown to correlate with a Ÿ1 Ÿ1 Ÿ1 day in group 1 and 0.31 ± 0.04 g × kg ×day in group rise in serum triglycerides, reduction of PaO2 and de- 2. Nitrogen balance was calculated by measuring 24 h ur- cline in pulmonary diffusing capacity for either carbon inary urea nitrogen excretion, multiplied by 1.25. Un- monoxide or carbon dioxide [16, 25, 26]. In critically ill measured nitrogen losses were estimated to be 2 g × patients, LCT-induced serum lipaemia has been shown dayŸ1. Differences in caloric requirements and nitrogen to correlate with significant changes of Qva/Qt, MPAP, intake between the two groups were not statistically sig- pulmonary vascular resistance and PaO2/FIO2 [1, 5]. nificant. Long chain triglyceride-induced pulmonary hyper- The results are summarised in Tables 2 and 3. The tension and gas exchange impairment is probably due administration of LCTs in group 1 was associated with to the production of arachidonic acid metabolites [7, statistically significant increases of Qva/Qt and MPAP 17±20]. LCT emulsion metabolism is diverted favour- as well as a reduction of PaO2/FIO2 (p < 0.05). Infusion ably towards vasoconstrictive end-products [5, 7, 14, 15, of the MCT/LCT mixture in group 2 was associated 19]. McKeen et al. have demonstrated that 10 % Intrali- Ÿ1 Ÿ1 with a statistically significant increase of VO2 (p < 0.05). pid infusion at a rate of 4.2 mg × kg ×min induced in- 1032 creases of both MPAP and transvascular fluid filtration control. LCT infusion appeared to have selective detri- resulting from an increase in microvascular pressure; mental effects on pulmonary function ± demonstrated these changes were attributed to vasoconstrictive ara- by the observed elevation of MPAP and Qva/Qt and de- chidonic acid by-products [27±28]. In some species and cline of PaO2/FIO2 ± without affecting systemic haemo- under particular metabolic circumstances, LCTs may dynamics. Although, these changes were completely re- also produce vasodilating products like PGE2 and PGI2 versed following the cessation of lipid infusion, they that may have adverse effects on regional ventilation to could have potentially destabilising effects on the vul- perfusion matching, resulting in an increase of Qva/Qt nerable septic patient. By contrast, LCT/MCT adminis- and decline of PaO2/FIO2 [21, 27, 28]. The diversion of tration did not appear to have any major effects on pul- lipid metabolism towards either vasoconstrictive or va- monary physiology. The observed elevation of 22 % of sodilating products seems to be dependent upon lipid VO2 during LCT/MCT infusion underlines the percep- concentration, infusion rate and fatty acid chemical tion that MCTs are easily diverted to b-oxidation and structure [8]. In individuals with compromised lung energy production. Similar findings have been docu- function, LCTs may abolish compensatory mechanisms mented by Chassard and his associates [6]. Our results that improve arterial oxygenation, such as the hypoxic confirm previous reports [14] on the advantages of pulmonary vasoconstriction [1, 6, 17, 18, 28]. LCT/MCT over pure LCT administration in septic pa- Contrary to pure LCT infusion, LCT/MCT emulsions tients. Furthermore, our findings demonstrate that the have been shown to provide less prostaglandin precur- beneficial effects of LCT/MCT infusion are preserved sors and produce considerably fewer effects on pulmo- in the setting of ARDS. nary physiology. Radermacher et al. measured urinary In conclusion, we have shown that, in septic patients thromboxane-B2 (TxB2) and 6-keto-prostaglandin-F2a with respiratory failure, LCT administration was associ- (6-keto-PDF2a) following the administration of LCT/ ated with more significant changes of Qva/Qt, MPAP Ÿ1 Ÿ1 MCT 1 : 1 emulsions, at a dose of 0.150 g × kg ×h , and PaO2/FIO2 compared to infusion of an LCT/MCT and concluded that LCT/MCT mixtures do not increase 1 : 1 emulsion in a comparable group of patients. Clini- the formation of thromboxane-A2 (TxA2) and prostacy- cally, these transient alterations might cause serious pro- clin (PGI2) [14, 20]. blems in patients with marginal arterial oxygenation In the present study, we focused on the pulmonary and cardio-respiratory impairment. Further prospec- and gas exchange effects of lipid infusion in patients tive, randomised, well controlled studies are needed in who developed significant lung injury in addition to order to determine if substitution of LCT emulsion their severe post-operative septic complications. The with a mixture of LCT/MCT can improve clinical out- design of our study and the statistical methods em- comes and, consequently, this should be recommended ployed allowed us to evaluate the effects of LCT and for septic patients with ARDS. LCT/MCT regimens by using each patient as his own

References 1. Venus B, Smith AR, Patel CB, Sando- 5. Venus B, Prager R, Patel CB, Sandoval 9. Eckart J, Adolph M, Van der Muhlen val E (1989) Haemodynamic and gas E, Sloan P, Smith RA (1988) Cardiopul- U, Naab V (1980) Fat emulsions con- exchange alterations during Intralipid monary effects of Intralipid infusion in taining medium chain triglycerides in infusion in patient with adult respira- critically ill patients. Crit Care Med 16: parenteral nutrition of intensive care tory distress syndrome. Chest 95: 587±590 patients. J Parenter Enter Nutr 4: 1278±1281 6. Chassard D, Guiraud M, Gauthier J, 360±366 2. Askanazi J, Carpentier YA, Elwin DH Gelas P, Berrada KR, Bouletreau P 10. Ball MJ, White K (1989) Comparison of (1980) Influence of total parenteral nu- (1994) Effects of intravenous medium- medium and long chain triglyceride me- trition on fuel utilisation in injury and chain triglycerides on pulmonary gas tabolism in intensive care patients on sepsis. Ann Surg 191: 40±46 exchanges in mechanically ventilated parenteral nutrition. Intensive Care 3. Carpentier YA, Richelle M, Haumont patients. Crit Care Med 22: 248±251 Med 15: 250±254 D, Deckelbaum RJ (1990) New devel- 7. Hwang TL, Huang SL, Chen MF (1990) 11. Bach AC, Babayan VK (1982) Me- opments in fat emulsions. Proc Nutr Effects of intravenous fat emulsion on dium-chain triglycerides: An update. Soc 49: 375±380 respiratory failure. Chest 97: 934±938 Am J Clin Nutr 36: 950±962 4. Crowe PJ, Dennison AR, Royle GT 8. Skeie B, Askanazi J, Rothkopf MM, 12. Jiang ZM, Zhang SY, Wang XR, Yang (1985) A new intravenous emulsion Rosenbaum SH, Kvetan V, Thomashow NF, Zhu Y, Wilmore D (1993) A com- containing medium chain triglyceride: B (1988) Intravenous fat emulsions and parison of medium chain and long chain Studies of its metabolic effects in the lung function: a review. Crit Care Med triglycerides in surgical patients. Ann perioperative period compared with a 16: 183±194 Surg 217: 175±184 conventional long chain triglyceride emulsion. J Parenter Enter Nutr 9: 720±724 1033

13. Dowes RFH, Royle GT, Dennison AR, 18. Hageman JR, Hunt CE (1986) Fat 24. Long CL (1980) Energy and protein Crowe PJ, Ball M (1986) Metabolic stu- emulsions and lung function. Clin Chest needs in the critically ill patient. Con- dies at a lipid emulsion containing me- Med 7: 69±77 temp Surg 16: 29±42 dium chain triglyceride in perioperative 19. Hunt CE, Cora P, Inwood RJ (1981) 25. Sundstrom G, Zauner CW, Arborelius and total parenteral nutrition infusions. Pulmonary effects of intralipid: the role M (1973) Decrease in pulmonary diffus- World J Surg 10: 38±46 of intralipid as a prostaglandin precur- ing capacity during lipid infusion in 14. Radermacher P, Santak B, Strobach H, sor. Prog Lipid Res 20: 199±204 healthy men. J Appl Physiol 34: Schror K, Tarnow J (1992) Fat emul- 20. Radermacher P, Santak B, Wust HJ, 816±820 sions containing medium chain trigly- Tarnow J, Falke KJ (1990) Prostacyclin 26. Greene HL, Hazlett D, Demaree R cerides in patients with sepsis syn- for the treatment of pulmonary hyper- (1976) Relationship between Intra- drome: effects on pulmonary haemody- tension in adult respiratory distress syn- lipid-induced hyperlipemia and pulmo- namics and gas exchange. Intensive drome: effect on pulmonary capillary nary function. Am J Clin Nutr 29: Care Med 18: 231±234 pressure and ventilation-perfusion dis- 127±135 15. Bach AC, Cuiraud M, Gibault FP,Schir- tributions. Anaesthesiology 72: 238±244 27. McKeen CR, Brigham KL, Bowers RE, ardin H, Frey A, Bouletreau P (1988) 21. Marthru M, Dries DJ, Zecca A, Fareed Harris TR (1978) Pulmonary vascular Medium chain triglycerides in septic pa- J, Rooney MW, Rao TLK (1995) Pul- effects of fat emulsion in unanesthe- tients on total parenteral nutrition. Clin monary haemodynamics and prosta- tised sheep. Am Soc Clin Invest 57: Nutr 7: 157±163 glandin metabolism. Chest 99: 426±429 1291±1297 16. Talbott GD, Frayser R (1963) Hyper- 22. Murray JF,Matthay MA, Luce JM, Flick 28. Inwood RJ, Gora P, Hunt CE (1981) In- lipidaemia: a cause of decreased oxygen M (1988) An expanded definition of the domethacin inhibition of Intralipid-in- saturation. Nature 200: 684±686 adult respiratory distress syndrome. duced lung dysfunction. Prostaglandins 17. Hageman JR, McCulloch K, Gora P, Am Rev Respir Dis 138: 720±723 Med 6: 503±514 Olsen EK, Pachman L, Hunt CE 23. Sirio CA, Knaus WA (1992) Triage and (1983) Intralipid alterations in pulmo- assessment of severity of illness. In: nary prostaglandin metabolism and gas Hall JB, Schmidt GA, Wood LDH exchange. Crit Care Med 11: 794±798 (eds) Principles of Critical Care. McGraw-Hill, New York, pp 551±557