This prospective, double-blind controlled clinical trial randomized 100 medical and surgical ICU patients to receive standard PN formulations containing soybean oil–based (Intralipid, Deerfield, IL) or olive oil–based (ClinOleic, Baxter Healthcare, Deerfield, IL) lipid emulsions. The olive oil–based lipid emulsion is comprised of ~80% olive oil and ~20% soybean oil plus glycerol, purified egg phospholipids, sodium oleate, sodium hydroxide, and water for injection. The soybean oil–based lipid emulsion is comprised of 100% soybean oil, plus egg yolk phospholipids, glycerin, and water for injection. We enrolled male and female patients between the ages of 18 and 80 yrs who were expected to require PN for longer than 5 days by conventional criteria (11
). We excluded patients who had received PN within 48 hrs prior to study entry, those with septic shock (defined as unstable blood pressure despite pressor support and mean arterial pressure <60 mm Hg on more than two occasions within 24 hrs prior to study entry), patients with a history of organ transplantation, active malignancy (defined as requiring chemotherapy, radiation, and/or surgical intervention within 90 days prior to entry), cirrhosis or with a serum total bilirubin level ≥10.0 mg/dL, acquired immunodeficiency syndrome, chronic renal failure (defined as requirement for hemodialysis or peritoneal dialysis therapy, or with a serum creatinine >3.5 mg/dL), or a mental condition rendering the subject unable to understand the scope and possible consequences of the study or when the legally authorized representative were not available. Also excluded were patients who were pregnant or breast-feeding, those with terminal illness and a life expectancy of <7 days, and patients with a baseline serum triglyceride concentration of >400 mg/dL.
The study was conducted at Grady Memorial Hospital, a major urban teaching hospital affiliated with Emory University and at Emory University Hospital, a tertiary referral academic institution in Atlanta, GA. The study protocol and consent forms were approved by the Institutional Review Board at Emory University. Informed consent was obtained prior to randomization from study subjects and/or their legally authorized representative. A research pharmacist at each institution coordinated treatment assignment following a computer-generated randomization table.
The study PN was administered using uniform guidelines for PN support in ICU settings established at both study sites. Briefly, the nutritional goals aimed to provide total daily calorie (kcal) intake at 1.3 times basal energy expenditure (per the Harris–Benedict equation using actual body weight or adjusted body weight in obese subjects per standard methods) (11
). The total amino acid/protein intake goal was 1.5 g/kg day. Parenteral vitamin K, multivitamins, electrolytes, and trace elements were added as required per standard methods (11
Study subjects received a maximum of 28 days of the study PN formulations. If a subject continued to require PN after day 28, the type of PN and need for lipid emulsion was decided by the primary care physician and the local nutrition support team. PN was started and continued only if deemed indicated by both the primary physicians and investigators. If a subject’s PN was discontinued but was later restarted during the initial 28 days after entry, study PN was restarted based on the initial randomization and continued, as clinically indicated, until day 28. All patients were managed by their primary care ICU team, who determined when to initiate and discontinue PN and transition to enteral feeds, in consultation with the investigator, per usual practices in the two participating hospitals. In general, transition of PN to enteral feeds (oral diet or conventional tube feedings) occurred as soon as tolerated; when enteral intake was ≥50% of energy requirements for ≥48 hrs, PN was discontinued.
The primary outcome of this study was the rate of new nosocomial infections, defined as culture-proven infection including wound, drain, bloodstream
, respiratory tract, and urinary tract infections (25
). The presence of nosocomial infections was diagnosed based on standardized Centers for Disease Control criteria (29
). We followed these Centers for Disease Control guidelines for laboratory-confirmed bloodstream infection and did not distinguish catheter-related infections per se. New nosocomial infections were diagnosed after 48 hrs of PN initiation in order to minimize the chance that the infection were present (but undiagnosed) prior to study PN initiation. The following daily information was evaluated by the study team for nosocomial infection surveillance: temperature (fever) curve, white blood cell counts (to evaluate leukocytosis/leucopenia), review of daily progress notes in the medical record, daily clinical microbiology laboratory culture data, orders for antimicrobial agents (agent, daily dose, and start/stop times will be recorded), review of all relevant dictated radiographic reports (e.g., chest radiographs, abdominal computer tomography), communication, as needed, with primary physicians and site infectious disease consultants, and use of the Centers for Disease Control guidelines for diagnosis of specific nosocomial infections (25
Secondary outcomes included differences between treatment groups in ICU and hospital length of stay, glycemic control, specific neutrophil and monocyte functions, inflammatory and oxidative stress markers, respiratory failure and need for mechanical ventilation, cardiac complications (defined as myocardial infarction, cardiac arrhythmia, or cardiac arrest documented by electrocardiogram and/or cardiac enzyme evidence), acute renal failure (new development of serum creatinine concentration >2.2 mg/dL or an increment >0.5 mg/dL from baseline), and ICU and hospital mortality (during PN infusion or after PN treatment was completed).
Blood glucose (BG) monitoring and glycemic management followed similar protocols approved by both participating institutions. Capillary BG was measured with a glucose meter at bedside during PN infusion. Regular insulin was added to the PN solutions at a starting dose of 0.1 units per gram of dextrose in nondiabetic patients and at 0.15 units per gram of dextrose in patients with a history of diabetes when the BG concentration was >120 mg/dL. Patients with repeated BG values >140 mg/dL received continuous intravenous insulin infusion following a standard algorithm, adjusted to maintain a target BG level between 80 and 140 mg/dL in the ICU and <180 mg/dL in the less-controlled setting of the general medical–surgical wards. Nosocomial infections were diagnosed following standardized Centers for Disease Control criteria (25
). New nosocomial infections were diagnosed after 48 hrs of PN initiation in order to minimize the chance that the infection was actually present (but undiagnosed) prior to study PN initiation. The investigators reviewed each subject’s records regarding potential new infection diagnosis daily on each weekday from Monday to Friday. Data from the weekends were collected and entered in the database the following Monday. Data on respiratory failure and need for mechanical ventilation were collected daily. The day the subject was weaned from the ventilator was recorded as a ventilator day. The presence or absence of acute respiratory distress syndrome was monitored and recorded daily using criteria set by The Acute Respiratory Distress Syndrome Network (30
Plasma glucose was measured on the CX7 Chemistry Analyzer (Beckman Diagnostics, Fullerton, CA) using reagents and calibrators from Beckman Diagnostics. Levels of interleukin-6, tumor necrosis factor-α, C-reactive protein, insulin, and C-peptide were measured in plasma using a solid phase, two-site sequential chemiluminescent immunometric assays on the DPC Immulite analyzer (Diagnostic Products, Los Angeles, CA). Plasma glutathione, cysteine and related redox potential of these pools were measured as indicators of oxidative stress, as previously outlined (31
). Briefly, samples were collected in a preservation solution and stored at −80° under conditions known to result in negligible oxidation. Samples showing visual evidence of hemolysis were discarded. Samples were treated to form dansyl derivatives, analyzed by high performance liquid chromatography with fluorescence detection and quantified relative to γ-glutamyl-glutamate as an internal standard. Redox potential values were calculated using the Nernst equation with Eo
−264 mV for the glutathione/glutathione disulfide couple and −250 mV for the cysteine/cystine couple at pH 7.4 (31
). Coefficients of variation for concentration measurements were 5%–6% for all parameters except cysteine and glutathione disulfide, which were 9%–10%.
The phagocytic and oxidative burst activity of monocytes and granulocytes in heparinized whole blood was assessed according to manufacturer’s instructions using specific reagent kits for this purpose at baseline and again at day 7. Briefly, granulocyte and monocyte phagocytic activity was quantitated by incubation of whole blood with fluorescein isothiocynate-labeled, opsonized Escherichia coli bacteria at 37°C (Phagotest, Orpegen Pharma, Heidelberg, Germany) with detection of fluorescence of internalized particles as a percentage of positive cells by flow cytometry (FACSort Becton Dickinson Biosciences, Franklin Lakes, NJ), analyzed using FlowJo software (Tree Star, Ashland, OR). Oxidative burst activity of monocytes and granulocytes was quantitated in whole blood using a kit containing unlabeled opsonized bacteria (E. coli), phorbol-12-myristate-13 acetate and the chemotactic peptide N formyl-Met-Leu-Phe as stimulants, and dihydrorhodamine-123 as a fluorogenic substrate to determine the percentage of phagocytic cells that produce reactive oxidants (Phagoburst, Orpegen Pharma, Heidelberg, Germany). A sample without stimulants served as a negative background control for each experiment. The percentage of positive cells was determined by flow cytometry and analyzed using FlowJo software.
The primary outcome of this study was the rate of nosocomial infections diagnosed after entry including wound, drain, catheter, respiratory tract, and urinary tract infections while receiving PN. Chi-square test (or Fisher’s exact test) was used to compare rates of infections between the soybean oil–based PN and olive oil–based PN study groups. For continuous secondary outcomes, we adopted nonparametric Wilcoxon tests to assess the difference between treatment groups. Differences in categorical secondary outcomes between treatment groups were evaluated using chi-square tests (or Fisher’s exact tests when needed). The Cochran–Mantel–Haenszel test was used to examine the difference in categorical outcomes between the two treatment groups while controlling for study center and Acute Physiology and Chronic Health Evaluation group. The absolute change in monocyte and granulocyte endpoints from baseline to day 7 was calculated.
A power analysis was conducted prior to the study based on preliminary data from Grady Memorial Hospital that indicated an overall infection rate of 0.36 during PN therapy (32
). Using an approximate two-sample proportion test, two-sided and α = 0.05, we estimated >80% power with 50 patients per group in order to detect a difference >0.24 in the occurrence rates of nosocomial infection between the two PN lipid emulsion study groups.