Here, we present a protocol for a study comparing the efficacy, safety, and delivery of olive-oil-based 3CB and soybean-oil-based CoB formulations in adults requiring parenteral nutrition. The results revealed that olive-oil-based 3CBs is non-inferior and well tolerated compared to soybean formulations.
Limited evidence exists to precisely estimate efficacy and safety differences between parenteral nutrition (PN) prepared using olive-oil-based three-chamber bags (3CBs) and soybean-oil-based compounded bags (CoBs) in hospitalized adult patients. We designed a multicenter, randomized, prospective, open-label, noninferiority protocol to compare the efficacy, safety, and distribution of olive-oil-based 3CBs and soybean-oil-based CoB formulations in adult Chinese patients requiring PN during surgical intervention. Subjects were randomized to receive either one of the study treatments using an interactive voice or web-based recognition system in accordance with the randomization code. Randomization was further stratified based on the study site and surgical category. Both treatment groups received similar amounts of calories and protein. In addition, the two study treatments contained a similar composition of the amino-acid component. The only difference between the two PN formulations was the lipid constitution. The duration of administration of study treatments was a minimum of 5 days up to a maximum of 14 days after the surgical procedure. The primary efficacy endpoint was serum prealbumin levels on day 5 of the study. Noninferiority was proved if the anti-log of the lower bound of the 95% confidence interval (CI) of the treatment difference was at least 0.80. Other efficacy measures included treatment preparation time; duration to achieve tolerability of oral nutrition; associated infectious complications; length of hospitalization; and laboratory assessment of markers of nutrition, inflammation, metabolism, and oxidative stress. A total of 458 patients were enrolled in the study. The results showed that olive-oil-based 3CBs were non-inferior to soybean-based CoBs, besides being well tolerated. The infection rate was found to be significantly lower in the olive-oil-based 3CB group. Thus, this study may be used as a reference for future research on lipid emulsion and 3CBs.
Parenteral nutrition is an essential component of overall therapy for a wide spectrum of indications, such as major gastrointestinal surgery, transient enteral intolerance, severe burns, coma; or for use in critically ill patients. Improvements in intravenous (IV) nutritional formulations and knowledge advancement regarding the implementation of therapy allow the safe and clinically efficacious administration of IV nutrition. These characteristics are particularly important in a metabolically stressed patient1.
Parenteral nutrition is commonly administered to patients by mixing nutrients that are compounded in the hospital pharmacy. Compounding total parenteral nutrition solutions from individual components is a multi-step, time-intensive process associated with a greater risk of human error. Recently, triple-chamber bag (3CB) systems have been developed in which individual components are separated by nonpermanent breakable seals. The contents of a 3CB include a glucose solution, an amino-acid solution, a lipid emulsion, with or without electrolytes. Prior to administration, the seal separating the various components of the 3CB is broken, enabling the components of the chambers to be admixed. The advantages offered by the 3CB includes increased physio-chemical shelf life of components, reduction the extent of contamination during preparation, and cutting down on the steps required in the preparation of a PN product2.
Lipid emulsion is an important ingredient in a PN formula; it can produce different clinical effects, depending upon the constituent fatty acids. Soybean-oil-based lipid emulsions primarily consist of long-chain linoleic acid (ω-6 polyunsaturated fatty acid [ω-6 PUFA]), which is mainly proinflammatory. Experimental data suggest that ω-6 PUFA-rich lipid emulsions may amplify the inflammatory response during stress and traumatic conditions, as well as increasing the infection rate3. On the other hand, olive-oil-based lipid emulsions, which consist of long-chain oleic acid (ω-9 monounsaturated fatty acids, [ω-9 MUFAs]), have a neutral response on the immune system3,4. Substituting soybean-oil-based ω-6 PUFAs with olive-oil-based ω-9 MUFAs can make the PN safe and further widen its clinical application5,6. However, there are limited clinical data in this connection.
Therefore, the present study aims to evaluate the rate of infections in two different lipid emulsions that varied in the content of linoleic acid, in addition to having the primary objective of assessing the safety and efficacy of 3CBs compared to CoBs for delivering PN. The assessment was carried out in adult hospitalized patients scheduled to undergo surgery for whom enteral nutrition was either not possible, inadequate, or inadvisable.
For this prospective, randomized, multicenter, active-controlled, parallel-group investigational trial, the Ethics Committees of Shanghai Sixth People’s Hospital approved the study protocol.
1. Patient recruitment and enrollment
2. Study population
3. Method and clinical parameters
4. Study treatments
5. Statistical methods
NOTE: The assumption for the sample size calculation in this noninferiority trial was made such that the true ratio was 1, the coefficient of variance was 0.5, and the noninferiority margin was 20 %. A sample size of 98 patients per study treatment came out of all the assumptions, providing 90% power to claim noninferiority between groups for the primary efficacy endpoint (i.e., prealbumin levels on day 5).
Patient Disposition
Out of the 480 patients who gave their consent, a total of 458 patients were enrolled and randomized in the study. The ITT population included all randomized patients, of whom 226 constituted the test group and 232 the control group. The safety population included a total of 453 patients, of whom 222 belonged to the test group and 231 to the control group. The modified intention-to-treat (mITT) population comprised a total of 443 patients, of whom 219 were in the test group and 224 in the control group (Figure 4).
A total of 373 patients comprised the per-protocol (PP) population, of whom 183 were in the test group and 190 in the control group. A comparable percentage of patients discontinued from the study in both the groups. Additionally, for both the groups, the two main reasons for discontinuation were AEs and the withdrawal of consent by the patient (Figure 4).
Demographic and Baseline Clinical Characteristics
The demographic and baseline clinical characteristics of patients in the two treatment groups (ITT population) were comparable. Sixty-one percent of the total patients were male. The majority of the patients were identified as Chinese Han (95%) and had a mean age of 56 years. A total of 62% of patients underwent high-complexity surgery of a mean duration of 3 hours (Table 2).
Endpoint’s Result
In relation to the primary endpoint of the study, olive-oil-based 3CBs were found to be non-inferior to soybean-oil-based CoBs in increasing or maintaining the levels of serum prealbumin at day 5 in both the mITT population (p=0.0002) and the PP population (p=0.0006).
A similar trend was also observed when the subgroup analyses for age, gender, no surgery, surgery of medium complexity, and surgery of high complexity was performed for the two groups (Figure 5).
Increased levels of prealbumin and albumin were observed for the test group, while the control group showed decreased levels of both proteins. Serum prealbumin and albumin levels on day 5 of the study were found to be significantly higher in the test group compared to the control group. No statistically significant difference was observed in serum IGF-I levels at day 5 between both the groups. However, at day 14, serum IGF-I levels were found to be significantly higher in the test group compared to the control group. No statistically significant differences were observed during the between-group analysis for 6 h urinary urea nitrogen and 6 h urinary excretion of 3-methylhistidine.
Lipid Endpoints
A significant increase in serum oleic acid levels were observed in both the groups; however, the observed increase in the olive-oil-based 3CB group was greater. There were no statistically significant differences observed at any time point between the treatment groups for serum levels of linoleic acid, arachidonic acid, and EPA.
Inflammation, Oxidation, and Infections
A small but statistically significant difference was observed on day 5 in serum levels of interleukin (IL)-6 between the two treatment groups. There was a decrease in the level of IL-6 in both the groups.
No significant differences were observed in serum levels of cortisol, procalcitonin, C-reactive protein, or ICAM-1 between the two treatment groups. Additionally, for the serum levels of malondialdehyde or F2-isoprostane, no significant differences were observed on day 5 or day 14 of the study between the two treatment groups.
The overall incidence rate of infections was found to be low in the study. Patients in the control group had a significantly higher infection rate compared to the test group (Table 3). The most common infections observed in the study were lung infections, followed by incision/wound infections. No bloodstream infections were reported in the study7.
Preparation Time
The preparation time for study treatment was found to be significantly lower for the test group compared to the control group on all assessment days (Figure 6).
a: Pre-surgery study treatment period refers to the time period (up to 3 days) when the subject received study treatment prior to surgery. If the subject did not undergo surgery, or if the subject underwent surgery without preoperative study treatment, this time period did not apply and the initiation of study treatment was hour 0.
b: If a subject underwent surgery and started study treatment preoperatively, the study procedures during the pre-surgery period were performed and the initiation of the first study treatment after surgery was hour 0.
c: Informed consent/authorization must be obtained prior to performing any protocol-specific assessments.
d: Body weight must be measured and recorded at screening and at EOT. Body weight, if measured for routine standard of care, should be recorded for each treatment day and, in particular, at baseline (hour 0) and on day 5.
e: Vital signs: Include body temperature, respiration rate, pulse rate, and systolic and diastolic blood pressure. Body temperature was measured as an axilla temperature and was obtained in the supine or sitting position; respiration rate, pulse rate, and systolic and diastolic blood pressure were obtained after the subject had been in the supine position for a minimum of 5 minutes. Clinically relevant changes in vital sign measurements consistent with worsening clinical status were recorded as adverse events (AEs).
f: Complete physical examination assessing the major body systems.
g: Focused physical examination.
h: The inclusion/exclusion criteria were confirmed prior to randomization. For a patient undergoing surgery without peroperative parenteral nutrition (PN), inclusion/exclusion criteria confirmation and randomization were performed after surgery.
i: Administered using an infusion pump set to the daily volume per prescribing orders. The prescribed and administered (actual) volumes and duration of each study treatment were recorded.
j: Calculation of the proportion of daily nutrition requirement that was administered by liquid oral or enteral nutrition was performed daily from day 6 through the remainder of the study period.
k: Clinical efficacy assessments include: Time for study treatment preparation (day 1 through day 5, documented in pharmacy record); surgical incision assessment (day 1, day 5, and end of study [EOT]); intravenous (IV) site assessment; length of hospitalization, mechanical ventilation (if applicable), and ICU (if applicable); date of first bowel movement after surgery.
l: Insulin: Hematology: Red blood cells (RBC), hemoglobin, hematocrit, white blood cells (WBC) with differential, platelet count, and prothrombin time. Serum chemistry: glucose, blood urea nitrogen (BUN), creatinine, bicarbonate (total carbon dioxide), sodium, potassium, chloride, phosphorus, triglycerides, total cholesterol, conjugated bilirubin, total bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase, and gamma-glutamyl transferase (GGT). These parameters were measured in the local clinical laboratory. The investigator was expected to review the laboratory data in real-time, by signing and dating laboratory printouts (or other media). Clinically relevant changes in safety laboratory values consistent with worsening clinical status were recorded as AEs.
m: Venous blood samples for efficacy assessment were collected during screening and at hour 0 (just prior to initiation of study treatment on day 1) and within 3 to 4 hours from initiation of the study treatment on day 5 and day 14 (if applicable).
n: Venous blood samples were collected within 3 to 4 hours from the initiation of study treatment on day 3, for measurement of hormones. These samples were processed and sent to the central laboratory.
p: Venous blood samples were collected at least 1 hour after the last study treatment. These samples were processed and sent to the central laboratory. Samples to assess serum prealbumin were drawn for all subjects at EOT. For all other efficacy laboratory measurements, venous blood samples were drawn only from subjects who received their final study treatment on day 9 and later. A urine sample collection was obtained prior to initiation of study treatment, on day 5 and on day 14 (if applicable), or EOT. On day 5 and day 14 (if applicable), the collection began at least 1 hour from initiation of study treatment and continued for 6 hours. For subjects who received their final study treatment on day 9 through day 13, the EOT urine collection began at least 1 hour from the end of last study treatment and continued for 6 hours. The volume of the urine collection was recorded, and then an aliquot from the urine collection was taken. These samples were processed and sent to the central laboratory.
q: Serious adverse events (SAEs) were collected from the time of signing the informed consent/authorization to EOT. AEs were collected from the initiation of study treatment to EOT. AEs occurring after the initiation of study treatment were considered treatment-emergent.
r: Any day from day 6 through day 13 was the final treatment day if the subject received at least 80% of their daily energy requirement from liquid oral or enteral nutrition. At the completion of the final treatment, EOT procedures were performed.
o: Day-14 procedures were performed for subjects who still required at least 20% of their daily energy requirements from the study treatment on day 14.
Table 1: Schedule of assessment—Pre-study treatment period through study treatment period (day 1 through day 14) and end of treatment Please click here to download this file.
Variable | OLIVE 3CBs (n=226) |
SOYBEAN CoBs (n=232) |
p-value |
Sex, n (%) | 0.482 | ||
Male | 134 (59.3) | 145 (62.5) | |
Female | 92 (40.7) | 87 (37.5) | |
Race, n (%) | 0.673 | ||
Chinese Han | 216 (95.6) | 220 (94.8) | |
Chinese (Other minority) | 8 (3.5) | 11 (4.7) | |
Altro | 2 (0.9) | 1 (0.4) | |
Underwent surgery, n (%) | 195 (86.3) | 202 (87.1) | 0.805 |
Complexity of surgery, n (%) | 0.859 | ||
Medium complexity | 49 (21.7) | 48 (20.7) | |
High complexity | 140 (61.9) | 143 (61.6) | |
Missing | 37 (16.4) | 41 (17.7) | |
Age, years, mean ± SD | 55.8 ± 13.1 | 56.3 ± 11.7 | 0.656 |
BMI, kg/m2, mean ± SD | 21.7 ± 3.9a | 21.8 ± 3.9 b | 0.667 |
Duration of surgery, (hours) mean ± SD | 2.9 ± 1.3 | 3.0 ± 1.4 | 0.645 |
BMI: Body mass index;kg/m2: kilogram per square meter;SD: Standard deviation;p value <0.05 was considered statistically significant | |||
a= (n = 217) ;b= (n = 226); * Information on the complexity of the surgery was missing |
Table 2: Patient demographics and baseline characteristics (Intention-to-treat [ITT] population)
Infection | OLIVE 3CBs (n=222) (%) |
SOYBEAN CoBs (n=231) (%) |
Total infections | 8 | 26* |
Total patient infected | 8 (3.60) | 24*(10.4) |
Lung infections | 2 (0.09) | 13 (5.62) |
Incision/wound site infections | 5 (2.25) | 3 (1.29) |
Urinary tract infections | 1 (0.04) | 2 (0.086) |
Abdominal/gastrointestinal infections | 0 (0) | 2 (0.086) |
Scrotal infections | 0 (0) | 1 (0.043) |
Non-specified infections (Systemic infection, site not identified) | 0 (0) | 5 (2.16) |
*P <0.01 |
Table 3: Treatment-emergent adverse event infections in safety population
Figure 1: Representative schematic for subjects undergoing surgery without preoperative parenteral nutrition (PN). (Treatment A (test treatment) is olive oil-based 3CBs; Treatment B (control treatment): Soybean-oil based CoBs) Please click here to view a larger version of this figure.
Figure 2: Representative schematic for subjects undergoing surgery with preoperative parenteral nutrition (PN) (Treatment A (test treatment) is olive oil-based 3CBs; Treatment B (control treatment): Soybean-oil based CoBs) Please click here to view a larger version of this figure.
Figure 3: Representative schematic for subjects not undergoing surgery. (Treatment A (test treatment) is olive oil-based 3CBs; Treatment B (control treatment): Soybean-oil based CoBs) Please click here to view a larger version of this figure.
Figure 4: Flowchart for patient disposition in study. (Treatment A (test treatment) is olive oil-based 3CBs; Treatment B (control treatment): Soybean-oil based CoBs) Please click here to view a larger version of this figure.
Figure 5: Representation of efficacy analyses of olive oil in the modified intention-to-treat (mITT) population and prespecified patient subgroups. (LSGM ratio is the antilog of (log (GM) ± 1.96SE); p<0.05 Abbreviations: CI confidence interval, LSGM least square geometric means, mITT modified intention-to-treat, PN parenteral nutrition, SD standard deviation, SE standard error) Please click here to view a larger version of this figure.
Figure 6: Preparation time of olive oil-based 3CBs (Clinomel N4) and Soybean-oil based CoBs (PN admixture) (Day 1 through day 5). (*p<0.05 by Kruskal-Wallis was considered statistically significant. Error bars indicate standard deviations.) Please click here to view a larger version of this figure.
The randomized clinical trial protocol is a multi-purpose document. It not only provides guidance for the conduct of trial to the investigators, but it also makes ethics committees and institutional review boards aware of appropriate measures adopted to protect participants’ safety and interests. A proper design is crucial for the success of a clinical trial. It is often noted that the design of a trial is connected with its successes/failures8.
Additionally, selection of the intervention and control groups is a key step in protocol design. Compounded PN is the current standard for the administration of PN to patients who cannot obtain adequate nutrition intake from diet or enteral nutrition. A placebo would not represent ethical treatment for patients requiring IV nutrition for the prevention or treatment of malnutrition. Therefore, active treatment (compounded PN admixture) was used as a control instead of a placebo in this study, to maintain the standard of care that subjects required. Another reason for which the compounded PN admixture was selected is the stated research objective. This made it possible to deliver both similar amounts of calories (primarily from dextrose and lipids) and protein to both groups. Furthermore, the two study treatments contained a similar amino-acid component. The only intentional difference in the PN formula was lipid source.
The double-blind design is preferable for a clinical trial. For this study, blinding was not impractical because, as per the basic clinical practice, the clinician or nurse made sure that the integrity of the combined components in the PN formulation was maintained throughout the length of the infusion. Although it was an open-label study, the bias in the reporting of treatment effects was minimized by implementing blinding for the data management personnel, biostatisticians, and researchers at the central laboratory.
Inherent in all clinical trials is the issue of con-founder of relationship. Randomization is necessary to equally distribute both known and unknown con-founders to the study and control groups; this can reduce bias9. In this study, an IVRS/IWRS system was used by the sponsor to enroll/randomize patients to different treatments arms. It generated a unique enrollment/randomization number for each patient. This configurable-and-customizable system was accessible via telephone or the Web from anywhere in the world. It empowered the sponsor to proactively manage key aspects of their clinical trials, including enrollment/randomization, dosing/drug dispensation, clinical supplies, and unbinding, etc. In addition, it has eliminated the risk of bias at sites by automated randomization, dispensation, and unbinding. Furthermore, the system also has the potential to decrease the work burden of research staff10. Stratified randomization includes construction of strata based on various variables, such as age groups, race, or practice and randomizing within these developed strata. In this study, randomization was stratified by study site to maintain approximately equal proportions of patients randomized to each study treatment at each study site. Randomization within each study site was further stratified by the surgical category. This approach reduced the impact of operational differences between study sites on statistical comparisons. Stratified randomization ensures that randomization is achieved in a balanced manner for the important predefined confounder variables. It further helps in allowing researchers to analyze various subgroups11.
Block randomization divides randomized participants within different subgroups, termed as 'blocks,' in order to ensure equal distribution of participants to each group. The limitation of the block randomization approach is the predictable distribution of participants, leading to a selection bias in unmasked study groups. Selection bias in the block randomization approach can be reduced by ensuring random block sizes and by blinding the investigator with respect to the size of the block12. The block size was specified in the randomization code algorithm in this study.
The eligibility criteria of a clinical trial protocol should ensure that the enrolled participants in the trial are alike to the maximum extent and that the results obtained from them are applicable to the general populations as well9. In this study, we selected surgical patients incapable of receiving the required nutrition via the enteral or oral route, representing a population that may benefit from PN therapy when enteral nutrition is not feasible, insufficient, or challenging. We also enrolled non-surgery patients, aiming to make the results applicable to the nonsurgical population as well. The purpose of the exclusion criteria was to reduce noise and ensure the safety of the trial. To fulfill these, we excluded severely ill and dying patients, patients with contraindication to PN or allergic to PN components, and patients with a medical history that would interfere with metabolism.
The primary study endpoint must correlate directly to the investigational study product, should be clinically accordant, and assessable conveniently in a clinical trial. Endpoints are usually a biomarker or a patient-specific structural or functional endpoint13. In this study, the prealbumin level was selected as the primary endpoint, following a discussion involving various experts. Serum prealbumin level, one of the commonly used nutritional endpoints, serves as a composite pointer for the amino acid supply, protein-synthesizing extent, inflammation, and catabolism. However, several confounding variables affect this endpoint, including synthesis and degradation status and inflammation level. Thus, for the secondary endpoints, albumin and IGF-1 levels were selected to assess anabolism; excretion of nitrogen and 3-methylhistidine helped in assessing catabolism. Oxidation, lipid profile, glucose levels, insulin, and electrolytes levels helped in assessing the metabolic status; clinical outcomes are assessed based on infections, hospital stay, morbidities, mortality, and preparation time. In addition to establishing the efficacy of 3CBs and lipids in supporting the claim, the trial should also prove the safety profile, which includes the assessment of lab parameters, vital signs, and injection-site reactions.
More and more clinical trials are being undertaken in order to evaluate whether the new treatment is as efficacious as standard treatment. The new treatment possesses various advantages such as a good safety profile, ease of administration, and is economical, making it fruitful to establish noninferiority with regard to the efficacy parameter. The noninferiority trial offers the advantage of statistical significance to be only 1 tailed, as there is no assumption that the analysis addresses whether the treatment is better. The predefining of a noninferiority margin for the primary outcome measure is of prime importance while designing a noninferiority trial14. In this study, the noninferiority margin was defined as -20%. The noninferiority in the trial was justified if the anti-log of the lower limit of the treatment difference 95% CI was at least 0.80.
In the majority of hospitalized patients, the ability to absorb nutrients via the gastrointestinal route will return to normal levels within 1 to 2 weeks of the medical/surgical event that caused a disruption in feeding. It is, therefore, practical to compare olive-oil-based 3CB to a CoB (soybean-oil-based) for a course of therapy that is long enough to evaluate a signal of the comparative efficacy and safety, yet does not delay the re-administration of oral or enteral feeding. It is always a critical decision to specify the duration of follow-up, to come to a meaningful result. The relatively short duration of follow-up, i.e. a maximum of 14 days, may be considered as the main limitation of this study. Despite the short duration of follow-up, it is possible that with a longer duration of PN, additional differences may have been noted between treatment groups.
To sum up, several aspects of the trial design are appropriately selected to ensure the value of the study is duly considered for this standard protocol involving lipid emulsion and 3CBs. This study may be used as a reference for future research.
The authors have nothing to disclose.
The study was sponsored and funded by Baxter Healthcare, manufacturer/Licensee of OliClinomel N4. ProScribe was used in compliance with global guidelines for good publication practice.