Efficacy of a long-term home parenteral nutrition regimen containing fish oil-derived n-3 polyunsaturated fatty acids: a single-centre, randomized, double blind study

Background Data on the use of lipid emulsions containing fish-oil (FO) derived n-3 polyunsaturated fatty acids (n-3 PUFAs) in addition to medium- and long-chain triglycerides (MCT/LCT) for long-term home parenteral nutrition (HPN) are limited. This study aimed to compare HPN regimens containing either MCT/LCT/FO-derived n-3 PUFAs (test group) or MCT/LCT (control group) with respect to efficacy and safety during 8 weeks of HPN using a non-inferiority trial design with change of body mass index (BMI) as primary endpoint. Methods This prospective, randomized, double-blind study was conducted at the Charité, Berlin, Germany, from 02/2008 until 01/2014. Adult patients (n = 42; aged 18 to 80 years) requiring HPN for at least 8 weeks were randomly assigned to the test or control group. Assessments included weight, height, physical examination (cardiovascular system, abdomen, respiratory tract, liver, spleen, kidney, urine tract, skin, mucous membrane, neurology, psyche, musculoskeletal system, lymph nodes), bio impedance analysis, calorimetry, blood samplings (haematology, biochemistry, fatty acid analysis) and quality of life questionnaire. Results BMI increased in both groups with 8 weeks of HPN (ΔBMI(test group) = 1.3 ± 1.1 kg/m2; ΔBMI(control group) = 0.6 ± 0.9 kg/m2) demonstrating non-inferiority of the test regimen regarding nutritional efficacy. Assessment of secondary efficacy endpoints revealed that after 8 weeks of HPN with the test regimen, the proportion of n-3 PUFAs in serum, platelet and red blood cell phospholipids significantly increased, while the proportion of n-6 PUFAs decreased. The fatty acid pattern in the control group remained mostly stable. No statistically significant differences were detected between groups regarding inflammatory markers or quality of life. Laboratory parameters reflecting the safety endpoints liver function, bone metabolism, renal function, metabolic activity, lipid metabolism, coagulation and haematology were stable in both groups and no group differences were detected regarding (serious) adverse events. Conclusions The HPN regimen prepared with MCT/LCT/FO-derived n-3 PUFAs was at least as efficient in maintaining or even improving nutritional status during HPN as the control MCT/LCT regimen. Administration of FO-derived n-3 PUFAs for 8 weeks altered the fatty acid pattern of serum, platelet and red blood cell phospholipids. Both regimens were safe and well tolerated. Trial registration www.clinicaltrials.gov, registration number: NCT00530738. Electronic supplementary material The online version of this article (10.1186/s12937-018-0419-x) contains supplementary material, which is available to authorized users.


Introduction 84
Home parenteral nutrition (HPN) was first introduced in the early 1970s and is nowadays an 85 established therapy in the home care setting in western countries, as morbidity and mortality 86 associated with HPN are low [1]. HPN aims to provide adequate amounts of amino acids, 87 glucose, lipids, electrolytes and water in order to prevent malnutrition in patients requiring 88 long-term parenteral nutrition (PN) due to prolonged gastro-intestinal tract failure [1, 2]. As 89 prolonged malnutrition leads to weight loss, reduction of quality of life, increase in morbidity 90 and mortality, and is associated with poor clinical outcome due to slow wound healing or 91 Several lipid emulsions are available that differ in terms of lipid composition [11]. Lipid 105 emulsions derived from soya-bean oil deliver long-chain triglycerides (LCT) and are rich in 106 n-6 PUFAs (mainly linoleic acid, the precursor of arachidonic acid). Those based on soya-107 bean and coconut oil deliver medium-and long-chain triglycerides (MCT/LCT) and have a 6 reduced content of n-6 PUFAs compared with soya-bean oil. Lipid emulsions based on soya-109 bean oil, coconut oil and oil from cold-water fish (i.e. fish oil; FO) deliver a mixture of MCT,110 LCT and FO-derived n-3 PUFAs and have a reduced content of n-6 PUFAs while being rich 111 in n-3 PUFAs. 112 The safety of parenteral administration of FO containing lipid emulsions has been established 113 in several clinical trials and beneficial effects of FO supplementation including modulation of 114 inflammatory markers, reduced length of hospital stay as well as reduced infectious morbidity 115 have been shown for surgical patients [15][16][17][18][19][20][21][22], as reviewed in [23]. Concerns have been 116 raised regarding an increased risk of bleeding due to the administration of n-3 PUFAs, based 117 on early observations in the Greenland Inuit population which indicated a longer bleeding 118 time associated with high consumption of fish [12]. However, clinical trials found no 119 evidence for an increased risk of bleeding upon n-3 PUFA administration [13,14]. 120 Data on efficacy and safety of FO containing regimens during long-term PN in the home-care 121 setting are limited [1]. Indeed, a recent systematic review [24] identified only one randomised 122 controlled trial of FO-containing HPN in adult patients which indicated that long-term 123 administration of n-3 PUFAs in the setting of HPN was safe for a period of four weeks and 124 led to an increased n-3/n-6 PUFA ratio in plasma and red blood cells [25]. 125 The current clinical trial extended the study duration and for the first time a period of eight 126 weeks of HPN with n-3 PUFAs was assessed. The trial was designed to show non-inferiority 127 of an HPN regimen prepared with a MCT/LCT/FO-derived n-3 PUFA containing lipid 128 emulsion as compared to a conventional HPN regimen without FO-derived n-3 PUFAs with 129 respect to nutritional efficacy (primary endpoint: change of body mass index (BMI) after 8 130 weeks of HPN). Secondary endpoints of this clinical trial covered safety parameters and 131 assessed potential beneficial effects of such a regimen on quality of life and body composition 132 Furthermore, physical examination of cardiovascular system, abdomen, respiratory tract, liver 211 and spleen, kidney and urine tract, skin and mucous membrane, neurology and psyche, 212 musculoskeletal system and lymph nodes was performed, and vital signs were assessed. 213 All investigations except fatty acid pattern were part of the routine assessment at the trial site. 214 Fatty acid patterns were determined at the Faculty of Medicine, University of Southampton, 215 according to established methods of fatty acid extraction, fatty acid methyl ester (FAME) 216 formation and FAME separation using gas chromatography (for details see [27,28]). FAME 217 were detected by flame ionization detection and identified by comparison with run times of 218 authentic standards. Peak areas and the percentage contribution of each peak to the total were 219 calculated. 220

Statistics 221
This study was designed to show non-inferiority of the test lipid emulsion regarding the 222 primary endpoint 'difference of BMI between V2 and BL' (ΔBMI). The non-inferiority 223 margin for the treatment difference ΔBMI test -ΔBMI control was defined as -1.1 kg/m² based on 224 the following assumptions: 225  For patients considered for study participation, cachexia is one of the most life 226 threatening risks and maintenance of BMI (ΔBMI test = 0) is a clinical success. 227  Based on routinely generated data in everyday practice at the study site, BMI increase 228 over 8 weeks HPN support in adults was expected to be 1.1 kg/m 2 (i.e. ΔBMI control = 229 1.1 kg/m 2 ). 230 Sample size was initially calculated based on routinely generated data in everyday practice at 231 the study site indicating an average BMI increase over 8 weeks of 1.1 kg/m 2 with a standard 232 deviation σ of 1.57 kg/m 2 (significance level α = 0.025, power 1-β = 0.80). Considering a 233 drop out rate of 10 %, sample size was determined to be 74, i.e. 37 patients per group. In the 234 course of this study, the standard deviation of BMI increase was adjusted to σ = 1.07 kg/m² 235 based on the evaluation of data originating from the pilot phase [29], and sample size 236 calculation was amended accordingly resulting in 32 completely evaluable subjects. or Kruskal-Wallis test (ordered categorical counts, non-paired data), Wilcoxon or Friedman 252 test (ordered categorical counts, paired data), exact Fisher test (dichotomous variables), 253 Pearson chi-square test (categorical character with more than two categories), or McNemar 254 test (paired categorical data) were used. Data are presented as mean ± standard deviation 13 (SD). Means were compared via the 2-sample t-test or analysis of variance (ANOVA) as 256 appropriate and the significance level was defined as 5 %. 257

Study population 259
A total of 43 adult patients requiring HPN for at least 8 weeks, recruited from the ambulatory 260 nutritional service at the University Hospital of the Charité Berlin, were screened for study 261 participation. 42 patients were eligible and randomly assigned to receive either the test lipid 262 emulsion (MCT/LCT/FO-derived n-3 PUFAs as Lipidem ® , test group) or the reference lipid 263 emulsion (MCT/LCT as Lipofundin ® , control group). A total of nine patients prematurely 264 discontinued the study (n = 6 and n = 3 in test and control groups, respectively, see figure 1 265 for further details). ITT analyses comprised data of 42 patients (n = 21 in each treatment 266 group) while PP analyses were based on data from 33 patients (n = 15 and n = 18 in the test 267 and control group, respectively). FAS and PP were identical in this study. 268

Figure 1 269
Test and control group were homogenous for most demographic and anamnestic parameters; 270 only the proportion of patients that experienced diseases within three months prior to study 271 start was significantly higher in the control group (see Table 2). Most patients had 272 concomitant diseases and required concomitant medication. 273 Table 2 274

Extent of exposure, treatment compliance 275
During the course of this study, the mean amount of lipid emulsion taken was 36.4 ± 11.4 276 bottles in the test group and 41.0 ± 10.4 bottles in the control group. As mean study duration 277 was 48.0 ± 16.6 days and 59.1 ± 14.6 days in test and control group, respectively, this 278 corresponded to a daily lipid intake of approximately 76 g in the test and 70 g in the control 279 group. Patients in the test group therefore received about 7.6 g fractionated FO per day. BMI changes over the 8 weeks of HPN were based on a gain of body weight in both study 292 groups (+ 3.7 ± 3.1 kg and + 2.0 ± 2.9 kg in test and control groups, respectively) which was 293 also reflected by a comparable increase of body cell mass (BCM) in the test and control 294 groups as determined via BIA (ΔBCM (test group) = 3.4 ± 5.3 %, ΔBCM (control group) = 3.2 ± 7.7 295 %). Covariance analyses revealed that weight gain was not correlated to the lipid emulsion 296 assigned, the amount of lipid emulsion administered, weight loss within three month prior 297 study start, or chemotherapy during the year before study start. 298 BMI increase and weight gain were more pronounced during the first 4 weeks of HPN: while 299 body weight and BMI increased by 4.2 ± 3.9 % and 2.6 ± 2.9 % in test and control groups, 300 respectively, during the first 4 weeks of HPN, BMI and body weight increased by 1.5 ± 3.0 % 301 and 0.6 ± 4.2 % in test and control groups, respectively, during the subsequent 4 weeks of 302 HPN (see also Table 3). Weight gain during the first 4 weeks of HPN was correlated to the 303 conduct of chemotherapy during the year before study start. 304 Table 3 305

Influence of nutritional regimen on fatty acid pattern 306
Lipid composition of erythrocytes, platelets and serum phospholipids was significantly altered 307 after 8 weeks of administration of the test lipid emulsion. In the test group, the proportion of 308 n-3 PUFAs (i. e. Eicosapentaenoic acid (EPA), Docosahexaenoic acid (DHA) and 309 Docosapentaenoic acid (DPA)) increased while the proportion of n-6 PUFAs (i.e. Linoleic 310 Acid (LA), Arachidonic Acid (AA), Dihomo-ɣ-linolenic acid (DGLA) and ɣ-Linolenic acid 311 (GLA)) decreased in erythrocytes, platelets and serum phospholipids. In the control group, the 312 proportion of n-3 PUFAs and n-6 PUFAs remained mostly stable (see Table 4). Significant 313 treatment differences were detected for EPA, DHA and DPA in erythrocytes, platelets and 314 serum phospholipids. Significant treatment differences for n-6 PUFAs were found in 315 erythrocytes (AA, DGLA and GLA), platelets (LA, DGLA and GLA) and serum 316 phospholipids (LA, AA). 317

Table 4 318
Influence of nutritional regimen on inflammatory parameters 319 IL-10 and TNF-α values were within the reference range at baseline and stayed stable during 320 nutritional treatment. IL-6 and CRP levels exceeded the reference range in both study groups 321 at baseline, probably reflecting the high incidence of co-morbidities. Mean values of IL-6 and 322 CRP increased in the test group during 8 weeks of HPN while they decreased in the control 323 group. However, IL-6 and CRP values in the test group showed a broad distribution especially 324 after 8 weeks of HPN (IL-6: min-value: 3.30 ng/l, max-value: 32.5 ng/l; CRP: min-value: 0.12 325 ng/l, max-value: 6.90 mg/dl) which is reflected by the high standard deviation for mean IL-6 326 and CRP-values in the test group (see Table 5). This indicates single outliers with a high 327 impact on mean values due to the small number of patients included for this investigation 328 (N=11 in each group). No statistically significant differences could be detected between 329 groups regarding the profile of inflammatory markers after 8 weeks of HPN Mean values of 330 inflammatory parameters are displayed in Table 5. 331 24.73 (V2); score range: 0-100). Statistical analysis of score changes between BL and V2 337 revealed no significant treatment dependent differences. 338

Safety of nutritional treatment 339
No differences could be detected between groups regarding the profile of laboratory 340 parameters determined to monitor liver function, bone metabolism, renal function, metabolic 341 activity, lipid metabolism, coagulation and haematology. All parameters stayed stable 342 throughout the nutritional treatment (for mean values ± SD and reference ranges see 343 'Additional file 2'). Only two individual clinically relevant abnormalities were reported (low 344 platelet count, already present at baseline, and CRP elevation, reported as AE, both in the test 345 group). 346 The number and intensity of reported adverse events (AEs) were comparable for test and 347 control group. A total of 11 patients in the test group and 12 patients in the control group 348 experienced at least one treatment emergent AE. In total, 76 AEs were reported (34 and 42 349 AEs in test and control groups, respectively). 350 No differences were detected regarding the AE pattern between study groups. Most AEs were 351 classified as "Gastrointestinal disorders" (i.e. diarrhea, nausea and vomiting, constipation), 352 "Musculoskeletal and connective tissue disorders" (mainly muscle spasm), "Nervous system 353 disorders" (headache and somnolence), "General disorders and administration site conditions" 354 (i.e. fatigue, chills and medical device complication), "Infections and infestations" (i.e. device 355 related sepsis), and "Skin and subcutaneous tissue disorders". None of the AEs was 356 considered to be related to the nutritional regimen. 357 A total of four patients in each treatment group experienced at least one AE that was rated as 358 serious. In total, 10 serious treatment emergent AEs were recorded. Although none of these 359 serious adverse events (SAEs) was related to the investigational products, the treatment was 360 prematurely terminated due to inability to continue IP administration during hospitalisation in 361 six patients (n=4 and n=2 in test and control groups, respectively). No unexpected SAEs 362 occurred. The most frequent SAE (device related sepsis) was expected as it represents a 363 common complication of PN therapy. No patient died during the study. One limitation of this study is that several study participants were on HPN therapy already at 446 the study start. It was intended to include only patients with a new indication for HPN in this 447 study. However, recruitment of patients was very difficult, as patients had to be able and 448 willing to mix the HPN regimen at home. The inclusion criteria therefore had to be amended 449 in order to also allow study participation of patients already receiving HPN. BMI changes 450 detected during this study therefore most probably underestimate the beneficial effects of 451 HPN on BMI. Furthermore, it cannot be excluded that differences between treatment groups 452  consent was obtained from all participants prior to any study procedure. 472

Consent for publication 473
Not applicable 474

Availability of data and material 475
The pseudonymised datasets generated and/or analysed during the current study are not 476 publicly available due to data protection regulations, but are available only in anonymised 477 form (e.g. ordered in ascending order) from the corresponding author upon reasonable 478 request.  were excluded from PP analysis because of premature study termination due to severe 622 protocol deviation (N=1), withdrawal of informed consent (N=2) or serious adverse events 623 (SAEs; N=6). SAEs leading to premature study discontinuation were not investigational 624 product related but required discontinuation of study medication due to necessary 625 hospitalisation. 626 Table 1 Composition of test and reference  Titration acidity/-alkalinity (pH 7.4) < 0.5 mmol/l < 0.5 mmol/l pH-value 6.5 -8.5 6.5 -8.5