What Is Known
- Neonates who receive prolonged parenteral nutrition are at risk for parenteral nutrition–associated liver disease.
- Fish oil–based lipid emulsions may improve parenteral nutrition–associated liver disease, but the long-term effects remain unknown.
- What Is New
- Fish oil–based lipid emulsion is well tolerated and safe in premature infants, resolves cholestasis, and decreases the need for intestinal and liver transplant.
- Growth during the first year is comparable in fish oil–based lipid emulsion–treated premature infants to controls.
- Cognitive and motor scores are decreased at 6 and 12 months in treated infants, likely due to prolonged hospital stay and severity of illness.
Parenteral nutrition–associated liver disease (PNALD) is now a well-established entity among infants who require prolonged parenteral nutrition and is a major cause of morbidity, liver transplantation, and death (1,2). As the name indicates, parenteral nutrition is the primary causative factor in PNALD, but the course and severity also appear to be affected by gestational age (GA), low birth weight, sepsis, duration of parenteral nutrition, and antecedent gastroschisis or jejunal atresia (1,2). The many components of parenteral nutrition make it difficult to pinpoint the cause of PNALD, but in recent years the lipid source commonly used in the parenteral nutritional regimen has emerged as the most likely culprit (3). Specifically, phytosterols and ω-6 fatty acids have been shown to have proinflammatory and cholestatic effects (4).
The only 2 parenteral lipid emulsions commercially available in the United States are both plant (soybean) based and rich in phytosterols and ω-6 fatty acids. Therefore, in recent years clinical investigators have turned to international sources to obtain lipid emulsions with increased ω-3 fatty acids and decreased ω-6 fatty acids. The fish oil–based lipid emulsion (FishLE) used in the present study was Omegaven (Fresenius Kabi, Bad Homburg, Germany), which consists primarily of ω-3 fatty acids and contains no phytosterols.
There are now multiple reports of the use of FishLE in infants requiring parenteral nutrition (4–6). In the United States where intravenous (IV) FishLE is not approved for marketing, these reports were based on compassionate use protocols using historical or case control cohorts. In all of these studies, most infants were born at term and/or FishLE was used at a corrected postmenstrual age (PMA) of term or thereafter. Some European studies have used an IV FishLE in preterm infants in conjunction with soybean and olive oil emulsion (not IV fish oil alone) and demonstrated a reduction in the incidence of retinopathy (7). Despite the paucity of data from randomized controlled trials, the encouraging uncontrolled results of the investigational use of FishLE and the high morbidity/mortality of PNALD leave investigators ethically unable to undertake new randomized controlled trials in the United States (8). Under these circumstances it is doubly important to probe the safety profile of FishLE, particularly in premature infants, and to evaluate short- and long-term outcomes related to growth and neurodevelopment due to the critically important role of proper lipid nutrition in development of the central nervous system. In the present study we report the short- and long-term outcomes of a group of preterm infants treated with FishLE under a compassionate use protocol compared with long-term outcomes of a large matched controlled cohort of preterm infants.
Methods are available online as Supplemental Digital Content 1 (https://links.lww.com/MPG/A771).
Cohort of Infants on Fish Oil
Thirteen patients were enrolled during the study period with birth weight of 1032 ± 783 g and GA of 28.3 ± 4.1 weeks. No consent was refused for patients who met the study criteria. The diagnoses that led to patients requiring parenteral nutrition are shown in Table 1.
Patients initiated FishLE treatment at a mean of 61 ± 9 days of life (range 15–131) and remained on the treatment for a mean of 73 ± 13 days (range 26–210). The nutritional status of each patient such as trophic feeds, enteral intake by discharge, parenteral nutrition duration, and number of days on FishLE are given in Table 1. The types of enteral feeds at the time of discharge were elemental formula in 30% of patients, 30% donor human milk, 16% mother's own milk, 8% preterm formula, 8% elemental/low lipid formula, and 8% custom made.
The trend of serum direct bilirubin values in study patients is shown in Figure 1. Resolution of cholestasis (direct bilirubin <2 mg/dL) was achieved in a mean of 68 ± 39 days after initiation of FishLE. Resolution of cholestasis was achieved in 54% of infants by the time FishLE was discontinued. Bilirubin continued to decrease after FishLE was discontinued and 100% achieved resolution of cholestasis by 4 to 8 weeks after discontinuation of FishLE. The highest direct bilirubin was 25.4 mg/dL at 28 days of life in an infant with a birth weight of 450 g. A full workup was performed in this patient by pediatric gastroenterologists to exclude other intrinsic causes of liver disease and it was determined to be PNALD.
Results for other indicators of liver function including gamma glutamyltransferase (GGT), aspartate aminotransferase (AST), and alanine aminotransferase are displayed in Figure 2. AST:alanine aminotransferase and AST:platelet ratio are displayed in Supplemental Digital Content 2, Figure, https://links.lww.com/MPG/A772. Two infants weighing <600 g at birth had GGT >900 mg/dL while on FishLE but eventually normalized.
No total free fatty acid deficiencies were detected by nonesterified fatty acid measurements in any patients. Mean nonesterified fatty acid serum concentrations were 0.22 ± 0.14 mmol/L at initiation of FishLE, 0.13 ± 0.05 mmol/L at 4 weeks, 0.15 ± 0.10 mmol/L at 8 weeks, and 0.18 ± 0.13 mmol/L at 12 weeks after FishLE initiation. One infant had a serum triglyceride of 1404 mg/dL after FishLE was inadvertently infused over 1 hour instead of 24 hours; this level normalized within 24 hours and FishLE was restarted and continued at 1 g · kg−1 · day−1 infused over 24 hours. A second infant developed hypertriglyceridemia to a max of 542 mg/dL and improved once FishLE was decreased to 0.75 g · kg−1 · day−1 and infused over 20 hours for 10 days and the remainder of 46 days FishLE was continued at 1 g · kg−1 · day−1 infused over 24 hours. The total cumulative days of FishLE infusion at a dose of 1 g · kg−1 · day−1 infused over 24 hours were 794/805 days that FishLE was infused in all patients. The mean peak triglyceride level during FishLE therapy was 216 ± 118 mg/dL and the mean triglyceride level at the end of FishLE therapy was 100 ± 53 mg/dL.
Central line−associated blood stream infections were lower than historically reported rates (2.3/1000 vs 8.4/1000 central line days historically in our neonatal intensive care unit [NICU]). No study patients developed a coagulopathy secondary to liver disease or FishLE administration, died, or required intestinal and/or liver transplant. Infants had a prolonged hospital stay at 218 ± 119 inpatient days.
Other major comorbidities in the FishLE group included: 62% of patients had a patent ductus arteriosus, 54% had bronchopulmonary dysplasia, and 31% had intraventricular hemorrhage.
Comparisons With Matched Controls
A total of 400 charts from the PREMIEre clinic visits during the study period were reviewed and 119 matched controls were identified. Baseline and demographic characteristics of the study patients and the matched control patients are shown in Table 2. The most noteworthy findings at baseline are the much larger proportion of small for gestational age (SGA) neonates (54% vs 19%) and the prolonged length of stay (218 ± 119 vs 88 ± 58) in the FishLE group compared with the matched control group, respectively. GA and birth weight were similar between the 2 groups. No other baseline demographic characteristics were significantly different.
Growth and neurodevelopmental outcomes at 6 and 12 months PMA for the study patients and matched GA control patients are also shown in Table 2. The difference in height noted at 6 months PMA was no longer observed at 12 months PMA. Of note, cognitive and motor scores were decreased in the FishLE group at 6 and 12 months PMA, whereas language scores were transiently decreased in the FishLE group at 6 months PMA but the difference disappeared by 12 months PMA.
Additional analyses were performed to investigate whether cognitive and motor score differences were due to treatment (FishLE vs control) or to other significant differences in the FishLE group such as prolonged length of stay, severity of cholestasis, or proportion of SGA patients. We compared the FishLE patients to matched control patients with prolonged lengths of stay (>100 hospital days, defined as the top quartile). The results of these comparisons are shown in Table 2. As expected, the peak serum direct bilirubin during the NICU stay in the matched controls was significantly lower than that in FishLE group (1.4 ± 2.3 vs 13.7 ± 6.6, P < 0.001); the peak direct bilirubin remained significantly lower when those matched controls with prolonged length of stay were compared (2.4 ± 3.6 vs 13.7 ± 6.6, P < 0.001). The differences in language and motor scores disappeared at the 6-month PMA comparison, and all differences disappeared by 12 months PMA in these analyses. No additional analyses were performed due to lack of appropriately matched controls for extremely prolonged length of stay and high proportion of SGA infants in the FishLE group.
Cognitive, motor, and language scores at 12 months PMA were classified as severe delay if <70 and mild delay if <85. After adjusting for GA, LOS, and treatment group, GA and LOS remained as significant predictors of severe cognitive delay (odds ratio [OR] 1.2, confidence interval [CI] 1.0–1.4, P = 0.04, OR 10.9, CI 1.0–112, P = 0.04, respectively), whereas LOS was the only significant predictor for severe motor delay (OR 19.4, CI 2.1–177, P = 0.009). We were unable to identify other predictors of language delay in this cohort by logistic regression analyses.
Once again, neonatologists in the United States find themselves faced with the conundrum of a serious clinical condition, PNALD, and a growing body of uncontrolled data that suggests benefits of an unapproved treatment, FishLEs such as Omegaven. The uncontrolled clinical experiences that have been reported are diverse in patient populations and the methods and circumstances of FishLE administration, but the preponderance of evidence in these reports indicates that the use of FishLE is associated with reduced serum direct bilirubin levels and decreased need for intestinal/liver transplant. An attempt to perform a randomized controlled comparison of a plant-based lipid emulsion to FishLE in preventing PNALD in infants at risk was terminated early after an interim analysis revealed much lower than expected incidence of PNALD in patients, making continuation unfeasible (9).
With the evidence of benefits of FishLEs increasing, and the possibility of performing randomized controlled trials ethically problematic, we aimed to gather additional experience with FishLE in a group of premature infants who represent a population at high risk for impaired growth and development owing to their prematurity and frequent comorbidities. In addition, questions remain about whether fish oil–based products could result in essential fatty acid deficiencies and have additional negative effect on short- and/or long-term growth and development. Because this population is the most vulnerable, we believed them to be in need of further study.
Our results demonstrate that the use of an IV FishLE in high-risk preterm infants with PNALD leads to resolution of cholestasis and decreases the risk of death and liver failure. This is the first study to report long-term growth and development data at 6 and 12 months corrected PMA in such premature infants. Nehra et al (9) reported results at 24 months of age with no significant differences from historical controls in Bayley scores, but their patients were on average 6 to 8 weeks more mature than ours. In the patients treated in our study there was no evidence that using a FishLE adversely affected somatic growth at 12 months PMA. This is in itself extraordinary because infants treated with FishLE are at the highest risk of poor growth due to severe intestinal disease and prolonged parenteral nutrition. A difference in length from matched controls at 6 months corrected PMA had disappeared by 12 months.
Development was assessed prospectively in all FishLE infants. The cognitive and motor scores were lower in treated patients at 6 and 12 months PMA. Noting that the LOS for treated patients was more than 100 days longer than GA-matched control patients, we wondered whether the decreased motor scores were related to prolonged hospitalization caused most likely by the complexity of multiple disease processes affecting these infants. After logistic regression analysis adjusting for GA, LOS, and FishLE treatment, length of stay was an independent risk factor for severe cognitive, language, and motor delays in premature infants at 12 months of age. We noted that the rate of some common comorbidities (intraventricular hemorrhage, bronchopulmonary dysplasia) were high in our treated patients compared with what is reported in preterm infants. It is possible that these other comorbidities also contributed to the decreased scores in treated patients, but the small number of patients did not allow us to state that with certainty. This is a limitation to the present study because we can only assume that adjusting for length of stay may take some of those factors into account. Another important factor that could play a role in the developing brain is a high direct bilirubin level at critical periods of development. In an attempt to tease out this factor, we compared the highest direct bilirubin during the hospital stay between groups. We found the direct bilirubin was 9.7-fold higher in the FishLE-treated group compared to the controls and 5.7-fold higher than controls with prolonged length of stay. The FishLE-treated group had similar neurodevelopmental outcomes compared to those controls with prolonged length of stay, which suggests that the main factor affecting neurodevelopment may be the multifactorial nature of an extremely long hospitalization and not the direct bilirubin level.
In a group of preterm infants who are severely ill enough to require prolonged parenteral nutrition, it is difficult to identify the many factors that affect their growth and development beyond hospitalization and infancy. Although the control groups used for comparison in our study were not randomly determined, the use of contemporaneous and concurrently treated matched control patients rather than historical controls should afford some confidence in the results. On the contrary, not having a historical control with the same disease process remains a limitation of the study. The differences we observed in cognitive and motor scores may also be explained by selection bias in the matched control group. Because these infants were selected from the patients who returned to follow-up clinic, it is possible they were less ill than those who did not return to clinic. This is an inherent problem when using data from follow-up patients.
Our study cannot completely put concerns about long-term development to rest, but on the other hand the results create no reason to stop further investigation. The results also suggest that pre-emptive, more intensive than usual inpatient developmental intervention tailored to these patients’ particular circumstances may be warranted to forestall possible developmental delays when patients are subject to these kinds of conditions.
Despite using FishLE in severely ill infants, previous reports of the safety profiles of patients during the period of infusion and immediately thereafter are reassuring. To date, no unexpected or untoward effects have been reported—no rise in infection rates, no free fatty acid deficiencies, no increased bleeding tendencies, or rebound in indicators of hepatic injury. Our results are consistent with those findings; indeed, we found a low rate of central line blood stream infections in FishLE patients compared to historical reports, a clinical result that may be related to the reported pre-clinical evidence of anti-inflammatory and immunostimulant properties of ω-3 fatty acids (4).
An unexpected finding in our study was the unusually high proportion of SGA infants among the treated group, approximately 54%. An explanation of that finding is not readily apparent, but it does obligate appropriate comparators in assessing long-term outcomes, as previously noted. There are reports of an increased risk of NEC and perforations in SGA infants (10), so one may speculate that SGA infants are more likely to develop intestinal dysmotilities and be over-represented in treatment groups. Other reports of FishLE treatment did not report the proportion of patients who were SGA. We also observed that a larger proportion of our patients required prolonged parenteral nutrition for functional gut dysmotility, rather than for anatomic short gut, which may reflect that our patients were more premature than those in other reports and more likely to have intestinal dysfunction on the basis of immaturity or NEC.
Although these results are nonrandomized and from a relatively small number of patients, our study prospectively followed patients for more than 2800 hospital days and captured a large amount of biochemical, nutritional, and outcome data. The results do not dissuade the growing interest in possible benefits of FishLEs, but cautionary notes must continue to be sounded. More information is needed in regards to long-term growth and development to avoid unexpected outcomes that are not immediately apparent in these early limited trials. Although the neurodevelopmental outcomes from the present study cannot completely assure long-term safety, it is important to disseminate these findings for comparisons with other centers that are tracking neurodevelopment in preterm infants with PNALD and perhaps for the design of future interventional trials. It continues to be a challenge to perform appropriate comparisons for these extremely fragile infants as premature infants enrolled in compassionate protocols for PNALD would likely succumb to the liver disease and not able to survive to liver/intestinal transplant. Also, although it is encouraging that declines occur in serum direct bilirubin and other biochemical indicators of liver dysfunction when FishLEs are used, the elevated GGT in 2 of our patients despite other biochemical improvements cannot be ignored. Reports of portal fibrosis despite biochemical improvements remain a concern and should not be forgotten as trials proceed (11).
In summary, the use of IV FishLEs in premature infants appears to be safe and improves PNALD despite significant liver disease and intestinal failure. We believe it is entirely warranted to continue compassionate use of FishLEs in premature infants and to pursue commercial access to a product in the United States to use for prospective trials of high-risk infants.
The authors thank the families and their children for participating and the amazing NICU nurses and Neonatologists at University Hospital for loving and taking care of the babies during many months. The authors also thank Laurie Weaver, RN for her diligence and accuracy on data collection. The PREMIEre clinic staff facilitated access and ensured inpatient testing and outpatient follow-up.
1. Kelly DA. Liver complications of pediatric parenteral nutrition—epidemiology. Nutrition
2. Christensen RD, Henry E, Wiedmeier SE, et al. Identifying patients, on the first day of life, at high-risk of developing parenteral nutrition-associated liver disease. J Perinatol
3. Premkumar MH, Carter BA, Hawthorne KM, et al. Fish oil-based lipid emulsions in the treatment of parenteral nutrition-associated liver disease: an ongoing positive experience. Adv Nutr
4. Fallon EM, Le HD, Puder M. Prevention of parenteral nutrition-associated liver disease: role of omega-3 fish oil. Curr Opin Organ Transplant
5. Gura KM, Lee S, Valim C, et al. Safety and efficacy of a fish-oil-based fat emulsion in the treatment of parenteral nutrition-associated liver disease. Pediatrics
6. De Meijer VE, Gura KM, Le HD, et al. Fish oil-based lipid emulsions prevent and reverse parenteral nutrition-associated liver disease: the Boston experience. J Parenter Enteral Nutr
7. Pawlik D, Lauterbach R, Turyk E. Fish-oil fat emulsion supplementation may reduce the risk of severe retinopathy in VLBW infants. Pediatrics
8. Abrams SA. Steps forward, backward, and sideways: intravenous lipid emulsions for critically ill neonates. J Parenter Enteral Nutr
9. Nehra D, Fallon EM, Potemkin AK, et al. A comparison of 2 intravenous lipid emulsions: interim analysis of a randomized controlled trial. JPEN J Parenter Enteral Nutr
10. Yee WH, Soraisham AS, Shah VS, et al. Incidence and timing of presentation of necrotizing enterocolitis in preterm infants. Pediatrics
11. Soden JS, Lovell MA, Brown K, et al. Failure of resolution of portal fibrosis during omega-3 fatty acid lipid emulsion therapy in two patients with irreversible intestinal failure. J Pediatr