Fish Oil–Based Lipid Emulsions in the Treatment of Parenteral Nutrition-Associated Liver Disease: An Ongoing Positive Experience^,,

Abstract

We previously reported the beneficial effect of fish oil-based lipid emulsions (FOLEs) as monotherapy in the treatment of parenteral nutrition-associated liver disease (PNALD). In this report, we share our ongoing experience at Texas Children’s Hospital, Houston, Texas in the use of FOLE in treatment of PNALD as presented at the 2013 Experimental Biology meeting. We describe the findings of a single center, prospective, observational study of infants <6 mo of age with PNALD who received parenteral FOLE as monotherapy. A total of 97 infants received FOLE under the compassionate-use protocol for the treatment of PNALD. Eighty-three (86%) survived with resolution of cholestasis and 14 (14%) died. The median conjugated bilirubin (CB) concentration at the initiation of FOLE therapy was 4.8 mg/dL (range 2.1–26). The median time to resolution of cholestasis was 40 d (range 3–158). Compared with infants with mild cholestasis (CB of 2.1–5 mg/dL at the initiation of FOLE), nonsurvivors were significantly more premature and took longer to resolve their cholestasis. Gestational age at birth correlated inversely with CB at the beginning of FOLE and peak CB. Infants with an initial CB >10 mg/dL had a higher mortality rate than infants with an initial CB <5 mg/dL (35% vs. 6%; P < 0.05). Our experience with the use of FOLE in PNALD continues to be encouraging. Prematurity continues to be a major determinant in mortality and severity of cholestasis. This calls for further controlled studies designed to optimize dose and timing of intervention in the use of FOLE in neonates.

Introduction

Parenteral nutrition-associated liver disease (PNALD)⁶ is a life-threatening disease with high mortality and morbidity (1–3). Since the advent of total parenteral nutrition (TPN) nearly 5 decades ago, survival of extremely low-birth weight infants and infants with complex gastrointestinal surgical conditions is now possible (4). Though TPN has been lifesaving in these conditions, long-term TPN usage is associated with severe morbidities such as PNALD (5). In many cases, PNALD results in liver failure requiring liver transplant or even death (6).

The etiologies of the complications associated with the long-term use of TPN are unclear (7). However, evidence accumulated over the past few decades strongly suggests the lipid component of TPN as the most probable cause of PNALD. Both animal and human research during the past decade has implicated phytosterols and ω-6 (n–6) fatty acids as the most likely components in the current lipid formulations as primary reasons for PNALD (8–12). In contrast, fish oil-based lipid emulsions (FOLEs) have been shown to be associated with full resolution of PNALD (13–18). The beneficial effects of FOLE have been attributed to the abundance of ω-3 fatty acids, which, in contrast to ω-6 fatty acids, have been shown to possess considerable anti-inflammatory properties (11, 19–22).

In the United States, the plant-based lipid emulsions Intralipid (100% soybean oil, Baxter/Fresenius Kabi) and Liposyn II (50% soybean oil, 50% safflower oil, Hospira) are currently the only lipid formulations approved for clinical use, with Intralipid being more widely used. Since the emergence of evidence from animal studies and nonrandomized patient investigations incriminating the harmful role of plant-based lipid emulsions, the interest in the use of alternative lipid emulsions derived from fish oil rich in ω-3 fatty acids has increased. Though lipid emulsions that contain fish oil, such as SMOFlipid (30% soybean oil, 30% medium-chain TGs, 25% olive oil, and 15% fish oil, Fresenius Kabi) and Omegaven (100% fish oil, Fresenius Kabi), are in wide use in Europe and other parts of the world; in the United States, the use of FOLE is restricted exclusively to compassionate-use protocols in the treatment of PNALD. Several nonrandomized human investigations have accumulated substantial data to support the beneficial effects of ω-3-rich fatty acids in the treatment of PNALD (16–18). At Texas Children’s Hospital, we now have ~5.5 y of experience using Omegaven as an investigational new drug under a compassionate-use protocol in the treatment for PNALD.

We previously published our early experience with the use of FOLE in the treatment of PNALD at Texas Children’s Hospital. Since then, our overall positive experience in the management of PNALD with FOLE continues. We now publish the cumulative results of the same as presented at the 2013 Experimental Biology meeting held in Boston, Massachusetts.

Methods

The research protocol was approved by the Institutional Review Board of Baylor College of Medicine and Affiliated Hospitals. An informed consent was obtained from the parents or legal guardians of each enrolled infant after noting that this was a compassionate-use protocol, not a controlled study, and that no fully controlled studies existed. Between September 2007 and April 2013, 97 infants with PNALD younger than 6 mo were enrolled in a compassionate-use protocol to receive 1 g/(kg · d) i.v. infusion of Omegaven at Texas Children’s Hospital, Houston, Texas. Infants older than 2 wk and younger than 6 mo of age were eligible for enrollment if the plasma conjugated bilirubin (CB) was ≥4.0 mg/dL in the absence of prior gastrointestinal surgical procedure or ≥2.0 mg/dL with history of gastrointestinal surgical intervention or severe feeding intolerance. These infants were expected to receive TPN for at least 28 d.

Upon enrollment, the soy-based lipid emulsion Intralipid was discontinued and substituted with FOLE Omegaven at 1 g/(kg · d) infused for 24 h. Participants having a congenital diagnosis with lethal prognosis, clinically severe coagulopathy unresponsive to standard therapy, or cholestasis secondary to primary hepatic disease were excluded from the study. Weekly measurements of CB and TGs were performed until resolution of cholestasis (CB <2 mg/dL) or death. Other liver function tests were monitored as considered clinically appropriate. Evidence for coagulopathy, essential fatty acid deficiency, and sepsis were clinically monitored and appropriate confirmatory tests were performed when these were suspected.

Descriptive statistics were performed using SigmaPlot version 11.0 (Systat Software). The comparison of variables between the survivors and nonsurvivors was assessed using a Mann-Whitney test. Statistical comparison of baseline characteristics and time to resolution were assessed using t test when reporting means, Mann-Whitney test when reporting medians, chi-square test when reporting proportions, and 1-factor ANOVA when >2 groups were being compared. The associations between variables were assessed using simple linear regression and Spearman rank order correlation tests. P < 0.05 was considered significant; all analyses were 2-sided.

Results

During a period of 5 y and 7 mo, 97 infants received FOLE for a diagnosis of PNALD. The male:female ratio of patients was 1.8:1. The median gestational age at birth of the infants who received FOLE was 28 wk (range 22.7–40 wk). The median postmenstrual age (PMA) at the initiation of FOLE therapy was 36.5 wk (range 22.7–64.5 wk) and the median CB concentration was 4.8 mg/dL (range 2.1–26 mg/dL). Eighty-three (86%) infants survived with resolution of conjugated hyperbilirubinemia during a median period of 40 d (range 3–158 d). Fourteen infants died (14%); 13 of them had persistent cholestasis.

The baseline characteristics of the 83 survivors and the 14 nonsurvivors are shown in Table 1. As we analyzed in our previous report, to further understand the relation of the concentration of CB with the course of the disease following the initiation of FOLE, we categorized the survivors according the initial CB at the initiation of FOLE. Those with CB between 2.1 and 5.0 mg/dL were categorized as group A and those with concentrations 5.1–10 and >10 mg/dL were categorized as groups B and C, respectively (Table 2). The median gestational age at birth in group A (31.8 wk, range 23.1–40 wk) was higher compared with the nonsurvivors (25.7 wk, range 22.7–39.5 wk; P < 0.05) (Fig. 1). Similarly, the median PMA at the initiation of FOLE was significantly higher in group A (39.5 wk, range 27.7–64.6 wk) compared with that of the nonsurvivors (33.6 wk; P < 0.05). Gestational age at birth did not correlate with the time to resolution of cholestasis (r² < 0.001; P = 0.99); however, it inversely correlated with severity of cholestasis at the initiation of FOLE therapy (r² = 0.1; P = 0.001).

TABLE 1.

Baseline characteristics of infants who received FOLE monotherapy for conjugated hyperbilirubinemia¹

	Survivors (n = 83)	Nonsurvivors (n = 14)	P value
Gender, n (%)			0.23
Male	56 (66)	7 (50)
Female	27 (33)	7 (50)
Gestational age at birth, wk	28.4 (23.1–40.0)	25.6 (22.7–39.6)	0.05
PMA at start of FOLE, wk	36.8 (27.3–64.6)	33.5 (27–47.3)	0.06
CB at initiation of FOLE, mg/dL	4.4 (2.1–26)	10 (3.6–14.3)	0.004
Peak bilirubin at initiation of FOLE, mg/dL	6.5 (2.2–26)	14.25 (6.4–28.6)	<0.001

¹Values are median (range). CB, conjugated bilirubin; FOLE, fish oil-based lipid emulsion; NA, not applicable; PMA, postmenstrual age.

TABLE 2.

Characteristics of the survivors categorized on the basis of CB concentration at initiation of FOLE treatment¹

	Group A (n = 47)	Group B (n = 23)	Group C (n = 13)
CB at the initiation of FOLE, mg/dL	2.1–5.0	5.1–10.0	>10
Gender, n (%)
Male	29 (61)	18 (78)	9 (69)
Female	18 (39)	5 (22)	4 (31)
Gestational age at birth, wk	31.8 (23.1–40.0)	26.9 (24–37.9)	27.0 (23.7–37.0)
PMA at start of FOLE, wk	39.6 (27.7–64.6)	36.3 (27.3–53.7)	34.1 (29.1–43.6)
CB at initiation of FOLE, mg/dL	3.0 (2.1–4.8)	6.0 (5.1–10.0)	13.2 (10.4–26)
Time to resolution, d	35 (3–135)	45 (18–158)	53 (21–84)
Peak bilirubin at initiation of FOLE, mg/dL	4.6 (2.2–22.8)	8.3 (5.3–17.1)	13.8 (10.5–26)

¹Values are median (range). CB, conjugated bilirubin; FOLE, fish oil-based lipid emulsion PMA, postmenstrual age.

FIGURE 1.

Comparison of gestational age at birth among categories within the group of survivors with the group of nonsurvivors. The categorization of survivors was based on the concentration of CB at the start of FOLE: group A = 2.1–5.0 mg/dL (n = 47); group B = 5.1–10 mg/dL (n = 23); and group C = >10 mg/dL (n = 13). The lower and upper borders of the box indicate the 25th and 75th percentiles, respectively. The line within the box depicts the median. The error bars indicate the 10th and 90th percentiles. CB, conjugated bilirubin; FOLE, fish oil-based lipid emulsion.

The patterns of resolution of conjugated hyperbilirubinemia following the initiation of FOLE are shown in Figure 2. The median time for resolution of cholestasis (CB <2.0 mg/dL) in group A was 34 d (range 3–135 d), shorter than that in group C, which was 53 d (range 21–84 d; P < 0.05) (Fig. 3).

FIGURE 2.

The trends of resolution of CB (median values in mg/dL) as observed over weeks following initiation of FOLE in the group of survivors. The comparison was made among the categories of survivors based on the concentration of CB (median values) at the initiation of FOLE. The categorization of survivors was based on concentration of CB at the start of FOLE: group A = 2.1–5.0 mg/dL (n = 47); group B = 5.1–10 mg/dL (n = 23); and group C = >10 mg/dL (n = 13). CB, conjugated bilirubin; FOLE, fish oil-based lipid emulsion.

FIGURE 3.

The comparison of time (days) taken to complete resolution of conjugated hyperbilirubinemia (CB <2.0 mg/dL) following initiation of FOLE in the group of survivors. The comparison made among the categories of survivors based on the concentration of CB (median values) at the initiation of FOLE. The categorization of survivors was based on concentration of CB at the start of FOLE: group A = 2.1–5.0 mg/dL (n = 47); group B = 5.1–10 mg/dL (n = 23); and group C = >10 mg/dL (n = 13). CB, conjugated bilirubin; FOLE, fish oil-based lipid emulsion; ns, not significant.

As noted in our previous experience, the CB increased for ∼1–2 wk after the initiation of FOLE before showing a gradual decrease. The resolution of cholestasis coincided with an improvement but not complete normalization of the hepatocellular indices as assessed by aspartate aminotransferase, alanine aminotransferase, and γ-glutamyl transferase (Fig. 4).

FIGURE 4.

The trends of concentrations of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and γ-glutamyl transferase (GT; median values in mg/dL) in the group of survivors as observed over weeks following initiation of FOLE.

All survivors demonstrated resolution of cholestasis, whereas only 1 of the 14 nonsurvivors demonstrated resolution. None of the terminal events were identified as complications secondary to FOLE or direct complications of liver failure. The mortality in infants with an initial CB >10 mg/dL (7 of 20 infants, 35%) was higher (P < 0.05) than that in infants with CB <5 mg/dL (3 of 50 infants, 6%) and 5–10 mg/dL (4 of 27 infants, 15%). Details of the infants who received FOLE but did not survive are described in Table 3. Of the 84 babies who responded to the therapy, only 20 of them progressed to receive 100% of calories via enteral feedings during the first 6 wk after initiation of FOLE; the majority continued to receive partial TPN support while their cholestasis resolved.

TABLE 3.

Details of infants who did not survive after the initiation of FOLE¹

Infant	Gestation at birth	Pre-existing diagnoses	Age at initiation of FOLE	Cause of death
	wk		d
1	26	NEC, short-gut, transferred from a referring hospital	54	Died on day of life 165 due to multi-organ failure following an abdominal surgery.
2	25	Spontaneous intestinal perforation, short-gut; transferred from a referring hospital	33	Died on day of life 138 due to multi-organ failure following surgery for re-anastomosis.
3	29	Omphalocele, NEC	37	Died on day of life 107 following redirection of care in view of multi-organ failure.
4	29	NEC, short-gut, Kelbsiella sepsis, transferred on day of life 126 from a referring hospital	128	Died on day of life 148 due to sepsis and multi-organ failure.
5	23	Ileal perforation, short-gut, grade 4 IVH, transferred from a referring hospital	34	Died on day of life 50 following redirection of care in view of multi-organ failure.
6	25	Spontaneous intestinal perforation, NEC	64	Died on day of life 102 due to fungal sepsis and multi-organ failure.
7	28	NEC, short-gut, grade 4 IVH, seizures, transferred from a referring hospital	41	Died on day of life 99 following redirection of care in view of multi-organ failure and IVH.
8	22	NEC, short gut, transferred from a referring hospital	30	Died on day of life 124 following gram-negative bacteremia and multi-organ failure.
9	28	NEC, short-gut, transferred from a referring hospital	120	Died on day of life 158 due to multi-organ failure.
10	24	Duodenal atresia, patent ductus arteriosus	66	Died on day of life 77 due to gram-negative sepsis and multi-organ failure.
11	35	Cloacal exstrophy, short-gut. Initial admission resulted in resolution of cholestasis and full enteral feeds. Readmitted at 1 y of age following intestinal obstruction.	365	Died on day of life 379 following intestinal obstruction and multi-organ failure.
12	25	NEC, short-gut, transferred from a referring hospital	28	Died on day of life 44 following gram-negative sepsis and multi-organ failure.
13	25	Grade 4 IVH, bronchopulmonary dysplasia, transferred from a referring hospital	25	Died on day of life 35 following E. coli sepsis and multi-organ failure.
14	25	Intestinal perforation, short-gut, grade 4 IVH, bronchopulmonary dysplasia	24	Died on day of life 243 following multi-organ failure

¹IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis.

Discussion

Although TPN is lifesaving in providing nutrition to extremely low-birth weight infants and infants with serious gastrointestinal surgical conditions, it has serious side effects, with PNALD being the foremost among them (3, 8). Until recently, infants with PNALD lacked effective modes of treatment, but the recent emergence of FOLE has shown a significant reduction in the mortality and morbidity associated with established PNALD (17, 18).

Since the publication of our earlier report, we have treated 37 more infants with PNALD, of whom 4 infants failed to respond, developed persistent sepsis with associated cholestasis, and eventually died of overwhelming sepsis (17). The total number of infants with PNALD who have been treated with FOLE on a compassionate protocol at Texas Children’s Hospital since September 2007 is currently 97, of whom 14 infants have died. The cumulative rate of resolution with survival is 85.5% compared with 82.5% at the end of previous reporting.

Compared with the previous report, the salient demographic features of the cohort remain unchanged, with the predominance of the male sex in the patient population. Whereas the median gestational age at birth of the infants remained similar to the previous report at 28 wk, the median CB at the initiation of therapy with FOLE was lower (4.8 vs. 7.5 mg/dL). The unchanged lower gestational age at birth reiterates the association of prematurity as a risk factor with PNALD. Further highlighting this important association is the significant inverse correlation between these 2 variables. In our report, we found that the gestational age at birth and PMA at initiation were inversely correlated with both the concentrations of CB at the initiation of FOLE and the peak CB. Because therapies aimed at prevention of prematurity are still largely unsuccessful, this association stresses the need for interventions that can minimize the risks associated with cholestasis. Such measures include strategies devised to control line-associated sepsis, early feeding techniques, and use of FOLE to support nutrition (17, 23).

In this report, the PMA in infants with PNALD at the time of initiation of FOLE was younger than the earlier cohort (36.5 vs. 39.3 wk). The younger PMA at the initiation of FOLE suggests that these infants with PNALD were perhaps recognized at an earlier age, thereby limiting the prolonged exposure to states of cholestasis and exposing them to the beneficial effects of FOLE sooner. Because the majority of the FOLE recipients were born at referring hospitals, earlier FOLE initiation is perhaps due to the increased awareness of the referring hospitals regarding the risks of cholestasis and the benefits of FOLE. This increased awareness among the referring hospitals facilitates earlier recognition of infants with risk factors, closer monitoring practices for PNALD, and consequent referral of these infants for treatment with FOLE. This also meant that these infants received FOLE at an earlier age when the cholestasis and the effect on the liver were less severe.

Within our hospital during the past 2 y a structured multi-disciplinary team approach to the management of infants with intestinal failure has been adopted. Previous reports have shown the benefits of a multi-disciplinary team approach in improving the outcomes in intestinal failure and PNALD (24–26). The multi-disciplinary approach in our hospital includes a core group of neonatologists, pediatric gastroenterologists, surgeons, dietitians, nurses, social workers, and case managers. Importantly, the team responsible for the management and follow-up of infants post-discharge is involved right from the point of identification of these infants, well before the initiation of FOLE. Weekly review of the infants with PNALD facilitates detailed discussions of medical and surgical management of these infants among all the members of the team and devising a comprehensive plan for the management of these infants.

The median CB at the time of initiation of FOLE was 4.8 mg/dL compared with 7.5 mg/dL from our previous report. This suggested that the median concentrations of CB were lower than the previous report, probably a result of identifying these infants at an earlier gestational age and treating with FOLE when the effect of PNALD was less severe. However, recognition of these infants at an earlier PMA and a lower level of cholestasis did not reflect in a shorter time to resolution. The median time to resolution in survivors was 40 d in this report compared with 35 d in the previous report.

As in our previous report, the biochemical improvement in terms of liver enzymes did not match the course of improvement of the CB. The decline in the elevated amounts of various liver enzymes was more gradual and at the time of resolution of cholestasis, the liver enzymes had failed to completely normalize. This once again denotes that the improvement in cholestasis is an early marker in a prolonged course of recovery. Notably, none of these infants required liver biopsies or liver transplants.

An important finding from this study is the association of mortality with high concentrations of CB at the initiation of FOLE. The mortality rate in infants with an initial CB >10 mg/dL was 35% compared with 6% with an initial CB <5 mg/dL. Similar to previous reports, this re-emphasizes the risk of increased mortality with higher concentrations of CB (3). More importantly, this underlines the vastly improved survival when PNALD is recognized at early stages and treated promptly with FOLE.

Vital information that is lacking in the current data set is the long-term safety data of FOLE in these infants. However, we note that none of the surviving infants from Texas Children’s Hospital with resolution of their cholestasis redeveloped cholestasis. An interim analysis of a recent randomized controlled trial (RCT) demonstrated the safety and efficacy of FOLE in terms of biochemical and nutritional variables and neurodevelopmental outcomes up to 2 y of age (27). Although we agree that an RCT would be optimal, we think that the current widespread use of this product, albeit under compassionate use, makes such a trial extremely difficult. As we mentioned earlier, in light of the beneficial effects with FOLE, several centers find it unethical to randomize such infants to plant oil-based lipid emulsions (17). However, authorization by the FDA for unrestricted use in neonates and infants will pave the way for RCTs comparing the optimum doses of FOLE and combinations of FOLE with plant oil-based emulsions both in the prevention and treatment of PNALD.