Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis

Clinical Nutrition xxx (2016) 1e7

Contents lists available at ScienceDirect

Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu

Original article

Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis J.M. Llop-Talaveron a, *, M.B. Badia-Tahull a, E. Leiva-Badosa a, J.M. Ramon-Torrel b a b

Pharmacy Department, Hospital Universitari Bellvitge, IDIBELL, L'Hospitalet de Llobregat, Barcelona, Spain Preventive Medicine Department, Hospital Universitari Bellvitge, IDIBELL, L'Hospitalet de Llobregat, Barcelona, Spain

a r t i c l e i n f o

s u m m a r y

Article history: Received 10 November 2015 Accepted 28 June 2016

Background & aims: Intravenous fat emulsions are associated with liver disease and there is some evidence that the administration of intravenous fish oil (FO) may be useful in reversing it. The aim of our study was to assess whether there are differences in the changes of liver function tests (LFTs) in hospitalized adult patients with parenteral nutrition (PN) with FO and vegetal lipids vs patients without FO. The secondary aim was to study the relationship between impaired LFT and FO. Methods: This was a 4-year, propensity score-matched analysis including patients aged
18 years treated with PN for
10 days. The exclusion criteria were previous liver disease, biliary disorders or pancreatic cancer, and altered initial LFT values. Patients were classified into 2 groups: FO cohort (patients who received FO e in addition to vegetal oil e after the first week of PN) and the vegetal oil cohort (patients who received only vegetal oil). A propensity score matched cohort design was developed. Univariate analyses were used to study the changes in LFTs. To evaluate whether LFT alterations vary with FO administration, four stepwise multiple linear regression models were conducted. Results: 52 patients were included, 52% men, median 66 (55e75) years and 69 kg (61.7e78.8), with 18.5 (14e31.8) days of PN treatment. Maximum FO supplementation was 23%. During the first week with PN (none of the groups receiving FO), gammaglutamyl transferase (GGT), alkaline phosphatase (AP) and total bilirubin (BIL) increased significantly. Comparing LFT values at seven days of PN with at the end of PN treatment, the univariate analysis showed a better response for the FO group. The group without FO showed a significant increase for GGT and AP. In multivariate models, the percentage of FO administered was associated with a decrease in GGT, B ¼ 0.33 [CI 95% ¼ 0.54/0.12], in AP, B ¼ 0.12 [CI 95% ¼ 0.20/0.03] and ALT, B ¼ 0.12 [CI 95% ¼ 0.21/0.024]. Conclusions: Lipid composition plays a significant role in LFT alteration associated with PN, and FO intravenous lipid emulsions (ILEs) minimize disturbance of LFTs in hospitalized adult patients. © 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Keywords: Fish oil Omega-3 Fatty acid Parenteral nutrition Liver function test

1. Introduction Intravenous lipid emulsions (ILEs) are traditionally administered through parenteral nutrition (PN), providing a dense source Abbreviations: ALT, alanine transaminase; AP, alkaline phosphatase; FO, fish oil; GGT, gamma glutamyltransferase; BIL, bilirubin; ILEs, intravenous lipid emulsions; LFT, liver function tests; MCT, medium chain triglycerides; OO, olive oil; PNALD, parenteral nutrition associated liver disease; PUFAs, polyunsaturated fatty acids; SO, soy oil. * Corresponding author. Pharmacy Department, Hospital Universitari Bellvitge, Feixa Llarga s/n, Bellvitge 22907, L'Hospitalet de Llobregat, Barcelona, Spain. Tel.: þ34 932607504; fax: þ34 932607507. E-mail address: [email protected] (J.M. Llop-Talaveron).

of non-protein calories, and being indispensable for supplying the requirements of essential fatty acids [1]. The soybean oil (SO)-based ILEs have been the reference. However, the emergence of different types of ILEs has been postulated for different clinical situations despite the criteria not being clearly established [2]. Conventional ILEs, such as SO-based emulsions, have caused concern because of the potential adverse effects involving oxidative stress, inflammation, and immune response probably caused by undesirable fatty acid composition. Apart from SO, new ILEs include other oils such as medium chain triglycerides (MCT) or olive oil (OO), which result in more favorable metabolic parameters because of a more desirable lower polyunsaturated fatty acid (PUFAs)-u6 content. Finally,

http://dx.doi.org/10.1016/j.clnu.2016.06.027 0261-5614/© 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

2

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7

the emulsions containing fish oil (FO), primarily composed of u-3 PUFAs, may provide anti-inflammatory effects and offer the potential for a targeted approach in disease specifically associated with inflammatory states. The development of significant hepatic dysfunction is a wellrecognized complication associated with PN administration [3,4] and PN associated liver disease (PNALD) is well established in patients, but the mechanisms leading to these remain largely unknown and are likely to be multifactorial [3,5,6]. A known established cause is the source of fat used and some evidence has suggested that the administration of FO-based ILEs may be useful in reversing PNALD [7]. The use of FO ILEs and their influence on liver disorders has become an important area of study in pediatric patients [8]. Studies in hospitalized adult patients that show improved liver function tests (LFTs) with FO administration are scarce [9,10]. As all vegetal oil-based emulsions contain phytosterols, it has been postulated that the long-term use of these emulsions leads to a progressive increase of phytosterols in plasma and cell membranes. Clayton et al. [11] published the first study in which the presence of phytosterols was associated with liver injury in pediatric patients. In adult patients with home PN (long-term treatment), even though there are fewer studies than in the preterm newborn population, there is some evidence regarding the advantages of FO use. In 2007, working in a multidisciplinary team, we were the first to determine the use of plasma phytosterols in adult patients included in our home PN program. We found a significant association between plasma phytosterol values and liver damage determined by LFT values, liver biopsy and the clinical situation [12]. In two later studies in hospitalized adult patients, after multivariate adjustment, our group found an association between the doses of FO and an improvement in some LFTs [13,14]. In both studies, the use of FO was based on high levels of C-reactive protein (CRP) and low levels of acute-phase protein prealbumin as surrogate variables of inflammation (SIRS) and low protein synthesis associated with hypercatabolic stress. In both studies there were differences in the number of patients in each group as well as in the initial parameters. Going further in this observational cohort study and with the aim of corroborating previous results, we selected a matched homogeneous group without previous hepatic or biliary disorders or pancreatic cancer, who had received PN for more than 10 days. Our hypothesis was that the use of FO in significant amounts in hospitalized adult patients with PN prevented LFT alterations. Hence, the aim of our study was to assess whether there are differences in LFT changes in hospitalized adult patients with PN with FO and vegetal lipids vs patients without FO. The secondary aim was to study the relationship between impaired LFTs and FO. 2. Material and methods This was a cohort study in a non-pediatric 600-bed, single tertiary-referral hospital between January 2005 and December 2008. The study included patients aged
18 years who received PN treatment for
10 days for maintenance and improvement of nutrition and with initial, final and weekly analytical registers. All patients received PN with the three macronutrients (glucose, lipids and aminoacids), electrolytes, and micronutrients (vitamins and trace elements). The bags were compounded as “all-in-one” formulas. The administration was performed through a central line during 24 h with a perfusion pump. The exclusion criteria were previous liver disease, biliary disorders or pancreatic cancer, and initial values of gamma glutamyltransferase (GGT), alkaline phosphatase (AP) and alanine

transaminase (ALT) two and a half times the upper limit of normal, and total plasma bilirubin (BIL) within normal values according to our clinical laboratory tests (normal reference ranges for the general population: GGT <1.11 mkat/L, AP <1.5 mkat/L, ALT <0.73 mkat/L and BIL <18 mmol/L). 2.1. Patients Eligible patients were classified into 2 groups upon registration according to the lipid administered in PN: the FO cohort consisted of patients who started to receive FO e in addition to vegetal oil e after the first week of PN; and the vegetal oil cohort consisted of patients who received only vegetal ILEs. FO administration was clinically based on the status of each patient. Total fluid intake, total calorie intake, and other nutritional recipes were also planned and administered based on clinical evaluation by clinicians as established in the protocol. Our hospital protocol established the use of FO after a week of treatment with PN in the case of stable or rising inflammatory response. If, by any exceptional cause, a patient had received FO during the first week of PN treatment, he was not considered for the study. 2.2. Data collection Measurements collected were: demographics (sex, age and weight); clinical (diagnostic, length of stay [LOS] in the intensive care unit [ICU], infection, sepsis and mortality); nutritional (length of treatment with PN and amounts of parenteral macronutrients administered); and analytical (CRP, glucose, creatinine, triglycerides, prealbumin, albumin, and LFTs: GGT, AP, ALT and BIL). We registered analytical data at the time of study initiation (first day with PN) and weekly until the day that PN ended. Sepsis was defined according to the American College of Chest Physician/Society of Critical Care Medicine (ACCP/SCCM) consensus, sepsis was established when the patient presented at least two of the following criteria: (a) body temperature
38  C or 36  C, (b) tachycardia with heart rate
90 beats/min, (c) tachypnea with paCO2  32 mmHg or mechanical ventilation, (d) leukocytes >12  109 cells/L [15]. 2.3. Nutritional data The six ILEs included in the study can be grouped into four types taking into account the lipid constituents: a) Emulsions with 80% of OO and 20% SO (Clinoleic® and Oliclinomel®) used in patients with moderate metabolic stress; b) SO emulsions with MCT (Structolipid® and Structokabiven®) used in patients with severe metabolic stress and hypertriglyceridemia; c) Emulsions containing fixed doses of FO in combination with other vegetal lipids (SMOF®) for patients with severe metabolic stress and high levels of inflammatory response; and d) Emulsion composed exclusively of FO (Omegaven®) that was used in combination with other ILEs in cases where severe inflammatory response required higher FO doses. Since it is not compulsory to declare the amount of a-tocopherol in ILEs, all patients included in the study received a daily multivitamin solution (Cernevit®) containing 10.2 mg (11.2 IU) of atocopherol. 2.4. Ethical considerations Written informed consent was considered not necessary for the study, as it was a non-interventional study. Parenteral nutrition and nutritional and safety variables were prescribed and collected according to clinical practice. Patient data were anonymized for

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7

the purposes of this analysis. Confidential patient information was protected according to national normatives. This manuscript has been revised for publication by the Clinical Research Ethics Committee of our Hospital. 2.5. Statistical methods We used a propensity score matched cohort design to minimize the effects of confounding variables when estimating treatment effects given the non-random nature of the study and the multiple factors that can influence them. Candidates for FO emulsion were selected according to our protocol, which states that patients with persistent inflammatory response associated with hypercatabolism, assessed by high CRP, and low values of prealbumin are the target group to receive FO. The main strategy here was to match the treated and untreated patients. The most common method is to pair a treated patient with an untreated one having a similar probability of receiving the treatment. This probability is defined through a range limited by the probability of the treated patient plus a given proportion of the SD of the probability. This calculated single probability usually does not present a normal distribution, so a logarithmic transformation of the probability of receiving treatment was performed and the range defined by 1/4 SD was used to pair. On the basis of our calculated propensity score, we matched people who received FO for at least five days during PN treatment with those who received PN with FO-free ILEs by using a greedy matching algorithm. The greedy matching algorithm first identifies matched pairs (a patient treated with FO and a patient without FO) and is made with a selected logistic regression model with a closeness range of 0.05. Descriptive statistics were calculated as frequencies, means (standard deviation), or medians and interquartile ranges (IQRs). To determine demographic and clinical differences between patients in both groups (those receiving FO or not), standard statistical analyses were performed including c2 test with Fisher exact tests or 2-sample t-tests. To study the LFT changes, we considered two periods of time. In the first period, we studied differences in GGT, AP, ALT and BIL between the beginning and the seventh day of PN treatment; in this case any patient receiving FO. In the second period, we studied differences in GGT, AP, ALT and BIL between the seventh day and the end of treatment with PN; in this case, there were two different groups: the FO group and the vegetal oil group without FO. In both periods the differences were studied by means of t-tests for paired samples. To evaluate whether LFT alterations varied with FO administration, four stepwise multiple linear regression models were conducted (GGT, AP, ALT and BIL). The difference for each LFT between the seventh day of PN treatment and the end of PN treatment was included in every model as a dependent variable. The main independent variable included in each of the four models was percentage of FO during PN treatment. Adjustment variables included as potential confounders were: prealbumin average, CRP average, glucose average, creatinine average, sepsis, days of ICU stay, days of PN treatment and proteins (g/kg/d) administered. They were included in the multivariate model for p < 0.1. All data were analyzed using SPSS 19.0 (SPSS INC, Chicago IL, USA). Statistical significance was reported with a 95% CI at the conventional p < 0.05 level (two-tailed). 3. Results For the studied four years, the number of patients hospitalized was 109.161. Digestive neoplasm (3.144 patients; 2.88%) represented

3

almost 22% out of all malignancies (14.835 patients; 13.59%). In this period 1.555 received parenteral nutrition (1.42%). In the study 52 patients were included. Median age was 66 (55e75) years, median weight was 69 kg (61.7e78.8) and 52% of participants were men. During PN treatment, 18% were admitted to ICU with a low stay ratio (0e3 days), 18 patients (34.6%) were septic and 3 (5.8%) died during the follow-up period. Table 1 details diagnoses. All patients were suffering from gastrointestinal diseases, 51.9% in the form of neoplasms. The median length of treatment with PN was 18.5 (14e31.8) days. FO supplementation was received by 50% of patients. Only 15.4% of patients had a FO supplementation of more than 10% out of the total administered lipids, and in no case was it higher than 23% (Table 2). The most frequent ILEs used were Structokaviben® 1407 (Table 3). Omegavenos® was always used as a supplement. Table 4 depicts baseline characteristics of patients treated and untreated with FO in the final propensity score matched sample and Table 5 depicts baseline CRP and prealbumin together with clinical progress data for the two groups of patients. Significant differences between both groups were only found in CRP and prealbumin. The FO group had more days of PN treatment, more days of ICU stay, higher values of LFTs; and a greater incidence of sepsis than the group without FO. However, those differences were not statistically significant. During the first week with PN (none of the groups receiving FO) GGT, AP and BIL increased significantly (Table 6). Comparing LFTs values at seven day of PN with values at the end of PN treatment, the univariate analysis showed a better response in the FO group. In the FO group no significant increases occurred for GGT and AP with a significant decrease in BIL. In contrast, the group without FO showed a significant increase for GGT and AP. In multivariate models (Table 7) in which the dependent variable was the difference between LFT values at the seventh day of PN treatment and the values at the end of PN administration, the percentage of FO administered was associated with a decrease in GGT, B ¼ 0.33 [CI 95%: 0.54/0.12], in AP, B ¼ 0.12 [CI 95%: 0.20/0.03] and ALT, B ¼ 0.12 [CI 95%: 0.21/0.024]. The percentage of FO administered had a tendency to a significant association with a decrease in the case of BIL, B ¼ 0.63 [CI 95%: 1.28/0.01]. Higher values of CRP plasma average were associated with a significant increase of GGT and ALT; and a tendency to a significant increase in AP. Higher values of average plasma triglycerides were associated with a significant increase of BIL. Table 1 Patient diagnoses. Diagnoses Gastrointestinal neoplasm Gastric cancer Rectal cancer Colonic cancer Esophageal cancer Small intestine cancer Retroperitoneal cancer Non-neoplastic gastrointestinal disease Pancreatitis Peptic ulcer Morbid obesity Intestinal adherence Evisceration Appendicitis Achalasia Volvulus Diverticulitis Crohn's disease Caustic ingestion Total

n

%

11 5 4 3 2 2

21.2 9.6 7.7 5.8 3.8 3.8

5 4 4 3 2 2 1 1 1 1 1 52

9.6 7.7 7.7 5.8 3.9 3.9 1.9 1.9 1.9 1.9 1.9 100

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

4

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7

Table 2 Characteristics of parenteral nutrition composition and the amount and type of lipids administered.

Table 5 C-Reactive protein and prealbumin at the beginning of parenteral nutrition and clinical progress data. p Valuea

Without fish oil With fish oil n ¼ 26 n ¼ 26

Parameter

n

Median (interquartile range)

Parameter

Days with parenteral nutrition Amino acids (g/kg/d) Lipids (g/kg/d) Fish oil (g/kg/d) Soy oil (g/kg/d) Olive oil (g/kg/d) Medium chain triglycerides (g/kg/d) Fish oil Without fish oil Fish oil < 5% Fish oil 5e10% Fish oil >10%

52 52 52 26 52 43 50 n 26 8 10 8

18.5 1.05 0.70 0.14 0.28 0.45 0.40 % 50 15.4 19.2 15.4

Baseline value of clinical criteria for fish oil administration C-reactive protein (mg/L), mean SD 66.6 (±73.1) 193.6 (±101.6) Serum prealbumin (mg/L), mean SD 139.2 (±50.2) 101.6 (±64.8) Nutrition data and clinical progress Lipids (Kg/day), mean SD 0.72 (±0.18) 0.74 (±0.20) Amino acids (Kg/day), mean SD 0.96 (±0.32) 1.06 (±0.17) Kcal/Kg/day, mean SD 24.3 (±2.96) 24.8 (±2.43) Days PN, mean SD 22.6 (±9.2) 29.2 (±22.1) Days ICU stay, mean SD 1.4 (±2.9) 4.5 (±9.4) Sepsis, n (%) 7 (25.9) 11 (44.0) Exitus, n (%) 2 (7.4) 1 (4.0)

(14e31.8) (1.04e1.17) (0.53e0.82) (0.12e0.17) (0.20e0.40) (0.33e0.59) (0.30e0.44)

0.00 0.02 0.85 0.16 0.51 0.18 0.13 0.17 0.53

SD: standard deviation. a Based on Fisher exact tests or 2-sample t tests. Table 3 Distribution of lipid emulsion trademarks used. Lipid emulsions trademarks

n

Days of ILE administration Median, IQR

Structokabiven 1407 (36% SO and 64% MCT) Oliclinomel (20% SO and 80% OO) Clinolenic (20% SO and 80% OO) Omegavenos (100% FO) Structolipid (36% SO and 64% MCT) Smoflipid (15% FO, 30% MCT, 30% SO and 25% OO) Structokabiven 1970 (36% SO and 64% MCT)

47 37 31 23 15 4 4

12 (6e21) 2 (1e4) 6 (3e13) 7 (4e16) 9 (5e13) 11.5 (3.25e31) 1.5 (1e5)

FO: Fish oil; IQR: interquartile range; MCT: medium chain triglycerides; OO: olive oil; SO: soy oil. Each patient might receive more than one trademark during the treatment.

4. Discussion This study confirms our previous results [13,14] in which an association between parenteral administration of FO and an improvement of LFTs in hospitalized adult patients was established. In this current work the association was stronger since the decrease in ALT was significant and a tendency to a significant decrease in BIL was also found, together with a loss of weight for the rest of the adjustment variables. The most striking effect of FO in this study was due to a paired matched distribution in both arms of the study (patients with or without FO). The main advantage or contribution of this study is the design. A propensity score method is an alternative to estimate the effect of receiving treatment when random assignment of treatments to subjects is not feasible. Propensity score matching refers to the pairing of treatment and control units with similar values on the propensity score, and possibly other

covariates, and the discarding of all unmatched units. Generally, if a treated subject and a control subject have the same propensity score, the observed covariates are automatically controlled for. Therefore, any differences between the treatment and control groups will be accounted for and will not be as a result of the observed covariates. The main difficulty was finding PN-treated patients with normal baseline LFTs since in the hospital setting candidates chosen for this type of therapy often have a significant degree of hypercatabolism and inflammation, as shown in the altered values of prealbumin and CRP. For patients selection, we followed the severity grading in drug induced liver injury proposed by LIVERTOX® [15], which establishes which ranges of LFT alteration should be used to classify the presence of moderate liver alteration, as well as its degree. ILEs with FO, known as third generation lipids, were developed in order to modulate the inflammatory response to damage resulting from catabolism. There is a sufficient theoretical base to assume that the addition of higher amounts of u-3 fatty acids to ILEs would positively modulate inflammation and immunosuppression after injury [16]. Its clinical use in different studies bears this out. In a first meta-analysis [17], the use of ILEs with FO in patients post major elective surgery demonstrated a reduction in infectious complications, reduced ICU stays and a trend to reduced hospital stays. Later, Clayton et al. [11] found an association between vegetal ILEs and liver injury in pediatric patients, and FO was tested in the treatment and prevention of PNALD. The first case reports of Gura et al. [18] in children were promising suggesting LFT improvement (and PNALD) when using FO ILE. Since then, there have been publications showing different strategies for dealing

Table 4 Baseline characteristics of all patients selected in the final propensity score matched groups: treated and untreated with fish oil. Parameter

Without fish oil n ¼ 26

With fish oil n ¼ 26

ORa

CI 95%a

Men, n (%) Age, mean (SD) Weight (Kg), mean SD Patients with gastrointestinal neoplasia, n (%) Triglycerides (mmol/L), mean (SD) Creatinine (mmol/L), mean (SD) Glucose (mmolL), mean (SD) Leucocytes (cellX109/I), mean (SD) Gamma-glutamyl transferase (mkat/L), mean (SD) Alkaline phosphatase (mkat/L), mean (SD) Alanine aminotransferase (mkat/L), mean (SD) Bilirubin (mmol/L), mean (SD)

16 (59.3) 64.1 (±14.2) 71.8 (±21.5) 13 (48.1) 1.81 (±0.9) 66.7 (±40.1) 7.3 (±2.0) 7.93 (±3.29) 1.21 (±0.97) 1.36 (±0.34) 0.54 (±0.85) 6.8 (±3.9)

11 (44.0) 64.2 (±14.4) 68.6 (±14.7) 14 (51.9) 2.0 (±0.7) 57.9 (±24.5) 7.9 (±3.5) 9.60 (±6.76) 1.56 (±0.93) 1.56 (±0.49) 1.1 (±0.65) 7.8 (±3.6)

2.01 0.98 0.99 4.33 1.15 0.98 1.15 1.88 0.14 13.52 4.91 1.32

0.41e10.62 0.93e1.03 0.95e1.03 0.63e29.86 0.66e5.33 0.95e1.01 0.85e1.55 0.88e1.15 0.02e1.25 0.88e206.5 0.76e31.82 0.98e1.76

SD: standard deviation; OR: Odds Ratio; CI: confidence interval. a Propensity score model with matched variables included.

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7

5

Table 6 Univariate study of liver function test evolution during parenteral nutrition treatment. First study period: Seventh day with PN versus initial PN treatment Patients group n ¼ 52 (all without fish oil) Liver function tests Gamma-glutamyl transferase (mkat/L), mean SD Alkaline phosphatase (mkat/L), mean SD Alanine aminotransferase (mkat/L), mean SD Bilirubin (mmol/L), mean SD

Initial PN 1.37 (±0.5) 1.46 (±0.43) 0.84 (±0.76) 7.3 (±3.7)

Seventh day 2.86 (±2.72) 1.97 (±1.11) 0.74 (0.91) 9.4 (±7.1)

a p Value 0.000 0.001 0.987 0.011

Seventh day 1.75 (±1.83) 1.52 (±0.56) 0.53 (±0.68) 8.3 (±8.2)

Final PN 3.62 (±3.65) 2.61 (±1.69) 0.48 (±1.02) 11.6 (±19.9)

a p Value 0.009 0.001 0.731 0.311

First week 4.06 (±3.04) 2.46 (±1.34) 0.89 (±1.01) 10.6 (±5.5)

Final PN 4.32 (±3.21) 2.78 (±1.13) 0.59 (±1.39) 7.9 (±5.1)

a p Value 0.772 0.310 0.111 0.027

Second study period: final PN treatment versus seventh day with PN Without fish oil group n ¼ 26 Liver function tests Gamma-glutamyl transferase, mean SD Alkaline phosphatase (mkat/L), mean SD Alanine aminotransferase (mkat/L), mean SD Bilirubin (mmol/L), mean SD Fish oil group n ¼ 26 Liver function tests Gamma-glutamyl transferase, mean SD Alkaline phosphatase (mkat/L), mean SD Alanine aminotransferase (mkat/L), mean SD Bilirubin (mmol/L), mean SD PN: parenteral nutrition; SD: standard deviation. a Based t-paired t tests.

Table 7 Multiple lineal regressions between liver parameter variations and FO percentage adjusted by different covariables. Liver parameter Variable

Percentage of fish oil C-Reactive protein average Triglyceride average Determination coefficient

Gamma-glutamyl transferase

Alkaline phosphatase

B [CI 95%]

p

B [CI 95%]

p

Alanine aminotransferase B [CI 95%]

p

Bilirubin B [CI 95%]

p

0.33 [0.54/0.12] 0.02 [0.001/0.031] e R2: 0.18

0.003 0.04

0.12 [0.20/0.03] 0.006 [0.00/0.012] e R2: 0.14

0.009 0.067

0.12 [0.21/0.024] 0.007 [0.00/0.014] e R2: 0.14

0.000 0.044

0.63 [1.28/0.01] e 5.89 [0.82/10.98] R2: 0.14

0.054 0.024

CI: confidence interval. Creatinine, glucose, Intensive care unit stay and sepsis do not feature in any model.

with this problem in paediatrics [19e22]; and also, though less frequently, in adult patients with long-term parenteral nutrition [9,10]. In short-term PN treatment in adults, different case reports have established the benefits of FO [23e26], but there are few studies with significant numbers of patients. In any case, those patients had established liver disease, which they had developed after a long period of PN treatment, when they started treatment with FO. Xu et al. [7] in a 15-patient study, found significant BIL and ALT improvement when the SO content of PN was partially replaced by FO, up to 10 g daily (0.15e0.2 g/kg/d); GGT values also decreased, though statistically significant differences were not found. With this current study we confirm that those patients who received a higher proportion of FO had better values for all LFTs. At the same time as we were recording these results, possible explanations about support mechanisms were published explaining the impact of lipid emulsions on LFT alteration and theoretically enabling the establishment of an alternative and more efficient treatment. The works of El Kassmi et al. [27] and Ng K et al. [28] are particularly relevant because they highlight the role that the presence of PUFAs u-6 and u-3, phytosterols and a-tocopherol could play. The authors [27] suggested that in vitro the combination of cytokine signaling and plant sterols could potentially exert potent synergistic inhibitory effects on hepatocyte expression of canalicular transport systems and thus promote intracellular bile salt retention or cholestasis. These results provide direct experimental evidence that phytosterols play a role in the pathogenesis of PNALD, and established that the absence of phytosterols in FO lipid emulsions and in lipid-free emulsions is

the likely mechanism of protection against LFT alteration. On the other hand, Ng K et al. [28] showed compelling and novel evidence that vitamin E plays an important hepatoprotective role in preventing LFT alteration. In premature piglets receiving a-tocopherol enriched lipid emulsions, both cholestasis and lipidemia were markedly reduced compared to those receiving SO-based emulsions (Intralipid®) alone without a-tocopherol. Thus, the absence of u-3 in SO and other vegetal ILEs versus u-3 enriched FO containing ILEs could alter the balance between hepatic inflammatory pathways and metabolic function during PN. In our series, parenteral FO was never administered as a single lipid source and never represented more than 23% of the total lipid intake. The mechanisms described in Kassmi et al. and Ng K et al. [27,28], detailed in the wide revision of Burrin et al. [29], would explain that hepatoprotective effects obtained by reducing phytosterol levels would be increased by the u-3 antiinflammatory effects and by reduced lipoperoxidation associated with a-tocopherol. In fact, a recent experimental study [30] provides evidence that novel, pure FO emulsion and a multicomponent lipid containing 15% of FO protect against LFT alterations in preterm TPN-fed pigs, suggesting that phytosterols are not the only factor associated with these alterations during ILE administration in healthy animals. Given the experimental nature of these contributions, the challenge would lie in knowing which hospitalized adult patients, under what conditions and at what dose should receive FO for the prevention and treatment of LFT alteration. In this field, our group is

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

6

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7

conducting a clinical trial to evaluate the effect of FO compared with OO in hospitalized patients treated with PN and with GGT alteration after one week of PN treatment. There are some limitations to our study despite the stronger selection criteria used in comparison with previous series, the propensity matched score and the multivariate adjustment; it has to be taken into account that the FO cohort showed greater altered initial values of CRP and prealbumin. In fact the patients had higher LFT values when FO was introduced (after seven days of PN) with a greater margin of improvement. However, this result could not be extended to all patients on PN treatment. For the multivariate adjustment we chose a stepwise model that only takes in consideration those variables that were statistically significant in the univariate model. We think that this approach gives a clearer idea of the importance of the main variables but R2 values are lower than if a multivariate enter model was performed. With an enter model R2 values vary between 40 and 53% but, since a lot of variables are taken into account, it is more difficult to know the exact weight of the studied variable FO dose. Another limitation is that we did not study the exact amount of a-tocopherol administered because the a-tocopherol present in ILEs cannot be calculated. This could be an interesting further avenue to explore since presently it is unclear whether the beneficial effects are caused by FO, atocopherol or by the exact mixture of the lipid components. Finally, as we use data from clinical routine registers, it has not been possible to introduce accurate variables relating to the initial clinical status and the clinical course and administration of potentially hepatotoxic drugs.

5. Conclusions These results indicate that FO ILEs minimizing the disturbance of LFTs should be considered in hospitalized adults patients. This study also further strengthens the idea that lipid composition plays a significant role in LFT alterations associated with PN and warrants further study into the metabolic effects of new-generation lipid emulsions in the prevention of this frequent alteration in adult hospitalized patients. Most randomized trials are performed in pediatric populations and the most relevant trials to establish preventive mechanisms of LFT alteration with FO are experimental, making it necessary to validate our hypothesis in randomized clinical trials in an adult hospitalized patient population incorporating the determination of plasma phytoesterol, a-tocopherol and more specific response markers.

Statement of authorship JLT designed the study, did the statistical analysis and wrote the manuscript, MBT carried the study out, analyzed the data and wrote and corrected the manuscript, ELB carried out the study and wrote and corrected the manuscript; and JRT did the statistical analysis, revised and corrected the manuscript.

Conflict of interest None.

Funding This research received no specific grant from any funding agency, commercial or not-for-profit sectors.

Acknowledgments We thank Maren White for editorial support in the preparation of this manuscript. References [1] Wretlind A. Development of fat emulsions. J Parenter Enter Nutr 1981;5(3): 230e5. [2] Adolph M, Heller AR, Koch T, Koletzko B, Kreymann KG, Krohn K, et al., Working group for developing the guidelines for parenteral nutrition of The German Association for Nutritional Medicine. Lipid emulsions-guidelines on parenteral nutrition [Chapter 6] Ger Med Sci 2009;7. Doc 22. € hm KD. Hepato[3] Piper SN, Schade I, Beschmann RB, Maleck WH, Boldt J, Ro cellular integrity after parenteral nutrition: comparison of a fish-oilcontaining lipid emulsion with an olive-soybean oil-based lipid emulsion. Eur J Anaesthesiol 2009;26:1076e82. [4] Nanji AA, Anderson FH. Sensitivity and specificity of liver function tests in the detection of parenteral nutrition-associated cholestasis. J Parenter Enter Nutr 1985;9:307e8. [5] Tillman EM. Review and clinical update on parenteral nutrition-associated liver disease. Nutr Clin Pract 2013;28:30e9. [6] Carter BA, Shulman RJ. Mechanisms of disease: update on the molecular etiology and fundamentals of parenteral nutrition associated cholestasis. Nat Clin Pract Gastroenterol Hepatol 2007;4:277e87. [7] Xu Z, Li Y, Wang J, Wu B, Li J. Effect of omega-3 polyunsaturated fatty acids to reverse biopsy-proven parenteral nutrition-associated liver disease in adults. Clin Nutr 2012;31:217e23. [8] Smithers LG, Gibson RA, McPhee A, Makrides M. Effect of long-chain polyunsaturated fatty acid supplementation of preterm infants on disease risk and neurodevelopment: a systematic review of randomized controlled trials. Am J Clin Nutr 2008;87(4):912e20. [9] Vanek VW, Seidner DL, Allen P, Bistrian B, Collier S, Gura K, et al. Position paper: clinical role for alternative intravenous fat emulsions. Nutr Clin Pract 2012;27(2):150e92 [REVISAR]. [10] Zaloga GP. Phytosterols, lipid administration, and liver disease during parenteral nutrition. J Parenter Enter Nutr 2015;39(1 Suppl):39Se60S. [11] Clayton PT, Bowron A, Mills KA, Massoud A, Casteels M, Milla PJ. Phytosterolemia in children with parenteral nutritioneassociated cholestatic liver disease. Metabolic Disease Unit, Institute of Child Health, London, England Gastroenterology 1993;105:1806e13. [12] Llop J, Virgili M, Moreno-Villares JM, Garcia-Peris P, Serrano T, Forga M, et al. Phytosterolemia in parenteral nutrition patients: implications for liver disease development. Nutrition 2008;24:1145e52. [13] Badia-Tahull MB, Leiva-Badosa E, Jodar-Masanes R, Ramon-Torrell JM, LlopTalaveron J. The relationship between the parenteral dose of fish oil supplementation and the variation of liver function tests in hospitalized adult patients. Nutr J 2015;14:65. n J, Leiva-Badosa E. Impact of intravenous lipid [14] Badia-Tahull MB, Llop-Talavero emulsions on liver funcion test: contribution of parenteral fish oil. Nutr Nutr 2015 Sep;31(9):1109e16. [15] Grading system developed by the acquired immune deficiency syndrome clinical trials group. February, 23 2015. http://livertox.nih.gov/Severity.html. [16] Grimble R. Fatty acid profile of modern lipid emulsions: scientific considerations for creating the ideal composition. Clin Nutr Suppl 2005;1(3):9e1561. [17] Wei C, Hua J, Bin C, Klassen K. Impact of lipid emulsion containing fish oil on outcomes of surgical patients: systematic review of randomized controlled trials from Europe and Asia. Nutrition 2010;26(5):474e81. [18] Gura KM, Parsons SK, Bechard LJ, Henderson T, Dorsey M, Phipatanakui W, et al. Use of a fish oil-based lipid emulsion to treat essential fatty acid deficiency in a soy allergic patient receiving parenteral nutrition. Clin Nutr 2005;24:839e47. [19] Angsten G, Finkel Y, Lucas S, Kassa AM, Paulsson M, Lilja HE. Improved outcome in neonatal short bowel syndrome using parenteral fish oil in combination with u-6/9 lipid emulsions. J Parenter Enter Nutr 2012;36: 587e95. [20] Soden JS, Lovell MA, Brown K, Partrick DA, Sokol RJ. Failure of resolution of portal fibrosis during omega-3 fatty acid lipid emulsion therapy in two patients with irreversible intestinal failure. J Pediatr 2010;156:327e31. [21] Yang CF, Lee M, Valim C, Hull MA, Zhou J, Jones BA, et al. Persistent alanine aminotransferase elevations in children with parenteral nutrition-associated liver disease. J Pediatr Surg 2009;44:1084e7. discussion 1087e88. [22] Park KT, Nespor C, Kerner Jr J. The use of Omegaven in treating parenteral nutrition-associated liver disease. J Perinatol 2011;31(Suppl. 1):S57e60. [23] Venecourt-Jackson E, Hill SJ, Walmsley RS. Successful treatment of parenteral nutrition-associated liver disease in an adult by use of a fish oil-based lipid source. Nutrition 2013;29:356e8. [24] Burns DL, Gill BM. Reversal of parenteral nutrition-associated liver disease with a fish oil-based lipid emulsion (Omegaven) in an adult dependent on home parenteral nutrition. J Parenter Enter Nutr 2013;37:274e80. [25] Jurewitsch B, Gardiner G, Naccarato M, Jeejeebhoy KN. Omega-3-enriched lipid emulsion for liver salvage in parenteral nutrition-induced cholestasis in the adult patient. J Parenter Enter Nutr 2011;35:386e90.

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027

J.M. Llop-Talaveron et al. / Clinical Nutrition xxx (2016) 1e7 [26] Zhu X, Wu Y, Qiu Y, Jiang C, Ding Y. Effect of parenteral fish oil lipid emulsion in parenteral nutrition supplementation combined with enteral nutrition support in patients undergoing pancreaticoduodenectomy. J Parenter Enter Nutr 2013;37:236e42. [27] El Kasmi KC, Anderson AL, Devereaux MW, Vue PM, Zhang W, Setchell KD, et al. Phytosterols promote liver injury and Kupffer cell activation in parenteral nutrition-associated liver disease. Sci Transl Med 2013 Oct 9;5(206). http://dx.doi.org/10.1126/scitranslmed. 206ra137. [28] Ng K, Stoll B, Chacao S, Saenz de Pipaon M, Lauridsen C, Gray M, et al. Vitamin E in new-generation lipid emulsions protects against parenteral nutrition-

7

associated liver disease in parenteral nutrition-fed preterm pigs. JPEN J Parenter Enteral Nutr 2016 Jul;40(5):656e71. pii: 0148607114567900. [Epub ahead of print]. n MS. Impact of new-generation lipid [29] Burrin DG, Ng K, Stoll B, De Pipao emulsions on cellular mechanisms of parenteral nutrition associated liver disease. Adv Nutr 2014;5(1):82e91. [30] Vlaardingerbroek H, Ng K, Stoll B, Benight N, Chacko S, Kluijtmans LAJ, et al. New generation lipid emulsions prevent PNALD in chronic parenterally fed preterm pigs. J Lipid Res 2014 Mar;55(3):466e77. http://dx.doi.org/10.1194/ jlr.M044545.

Please cite this article in press as: Llop-Talaveron JM, et al., Parenteral fish oil and liver function tests in hospitalized adult patients receiving parenteral nutrition: A propensity score-matched analysis, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.06.027