Potential conflict of interest: Nothing to report.
Liver Failure/Cirrhosis/Portal Hypertension
Persistent abnormal liver fibrosis after weaning off parenteral nutrition in pediatric intestinal failure
Article first published online: 17 JUN 2013
Copyright © 2013 American Association for the Study of Liver Diseases
Volume 58, Issue 2, pages 729–738, August 2013
How to Cite
Mutanen, A., Lohi, J., Heikkilä, P., Koivusalo, A. I., Rintala, R. J. and Pakarinen, M. P. (2013), Persistent abnormal liver fibrosis after weaning off parenteral nutrition in pediatric intestinal failure. Hepatology, 58: 729–738. doi: 10.1002/hep.26360
This work was supported by grants from the Mary and Georg C. Ehnroth Foundation, the Finnish Medical Association, the Sigrid Juselius Foundation, and the Finnish Pediatric Research Foundation.
- Issue published online: 29 JUL 2013
- Article first published online: 17 JUN 2013
- Accepted manuscript online: 4 MAR 2013 12:24PM EST
- Manuscript Accepted: 20 FEB 2013
- Manuscript Revised: 3 FEB 2013
- Manuscript Received: 28 NOV 2012
The aim of this study was to evaluate the long-term effects of pediatric intestinal failure (IF) on liver histology. Altogether, 38 IF patients (median age: 7.2 years; range, 0.2-27) underwent liver biopsy, gastroscopy, abdominal ultrasound, and laboratory tests. Sixteen patients were on parenteral nutrition (PN) after 74 PN months (range, 2.5-204). Twenty-two had weaned off PN 8.8 years (range, 0.3-27) earlier, after 35 PN months (range, 0.7-250). Fifteen transplant donor livers served as controls. Abnormal liver histology was found in 94% of patients on PN and 77% of patients weaned off PN (P = 0.370). During PN, liver histology weighted with cholestasis (38% of patients on PN versus 0% of patients weaned off PN; P = 0.003) and portal inflammation (38% versus 9%; P = 0.050) were found. Fibrosis (88% versus 64%; P = 0.143; Metavir stage: 1.6 [range, 0-4] versus 1.1 [range, 0-2]; P = 0.089) and steatosis (50% versus 45%; P = 1.000) were equally common during and after weaning off PN. Plasma alanine aminotransferase (78 U/L [range, 19-204] versus 34 [range, 9-129]; P = 0.009) and conjugated bilirubin (43 μmol/L [range, 1-215] versus 4 [range, 1-23]; P = 0.037) were significantly higher during than after weaning off PN. Esophageal varices were encountered in 1 patient after weaning off PN. Metavir stage was associated with small bowel length (r = −0.486; P = 0.002) and number of septic episodes (r = 0.480; P = 0.002). In a multivariate analysis, age-adjusted small bowel length (ß = −0.533; P = 0.001), portal inflammation (ß = 0.291; P = 0.030), and absence of an ileocecal valve (ß = 0.267; P = 0.048) were predictive for fibrosis stage. Conclusion: Despite resolution of cholestasis and portal inflammation, significant liver fibrosis and steatosis persist after weaning off PN. Extensive small intestinal resection was the major predictor for liver fibrosis stage. (Hepatology 2013;58:729–738)
AST-to-platelet ratio index
body mass index
IF-associated liver disease
international normalized ratio
body mass index for age
activated partial tromboplastin time
plasma tromboplastin time
Intestinal failure (IF) results from reduction of functioning gut mass, most often resulting from either short bowel syndrome or severe intestinal dysmotility disorders. IF patients often require prolonged parenteral nutrition (PN) to maintain normal energy, fluid, electrolyte and/or micronutrient balance, and normal growth. IF-associated liver disease (IFALD) is a major complication and the leading cause of morbidity and mortality in pediatric and adult IF patients. Various risk factors have been linked to the development of IFALD, including lack of enteral nutrients, duration and composition of PN, different components of PN, such as plant sterols, septic episodes, prematurity, low birth weight, small bowel bacterial overgrowth, and massive intestinal resection.[4, 6] PN-associated liver disease, defined by serum liver enzymes, occurs in 15%-85% of neonates, children, and adults on long-term PN.[3, 5] Retrospective studies on selected children on long-term PN have reported liver fibrosis, cholestasisis, and steatosis in up to 94%, 84%, and 41% of patients, respectively. After weaning off PN, serum liver enzymes usually slowly normalize,[10, 14] but histological liver fibrosis may persist or even progress.
The type and reversibility of histological liver injury and its risk factors during and after weaning off PN are insufficiently characterized. Previous reports and our own clinical experience have pointed out that significant abnormalities in liver histology may persist and liver damage may, in some cases, continue to proceed, even after weaning off PN.[11, 15] Currently, liver biopsy remains as the gold standard for assessing pathological changes in liver histology in chronic and acute liver diseases and is generally considered to be safe also in children.
To this end, we performed a population-based, cross-sectional study on liver histology in relation to the presence of portal hypertension (PH) and liver function during and after weaning off PN in children and young adults with pediatric-onset IF. Furthermore, we evaluated the effects of previously identified potential risk factors of IFALD on liver histology.
Patients and Methods
This study has an ethical approval by the Helsinki University Hospital (Helsinki, Finland) ethics committee.
Patients and Study Design
Medical records of patients with pediatric-onset IF treated by our IF rehabilitation program from January 1984 to August 2010 were reviewed. A total of 56 patients were identified, and 52 of them were alive. IF was defined as over 50% resection of the small bowel or duration of PN over 30 days.[6, 12] Eligible patients were invited to participate in this cross-sectional study, including clinical examination, laboratory tests, abdominal ultrasound (US), gastrointestinal endoscopy, and liver biopsy. An informed written consent was received from all patients and/or their parents. Detailed management of pediatric IF by our institution, either resulting from short bowel syndrome or intestinal motility disorders, has been described previously.
US-guided percutaneous core needle liver biopsy and gastroscopy were performed during the same general anesthesia. An experienced pediatric radiologist performed liver biopsies, after which patients were followed overnight at the hospital. One complication of liver biopsy occurred: a small right-sided pneumothorax, which resolved spontaneously. All endoscopies were performed by an experienced endoscopist. Esophageal varices were graded as described previously. Blood samples were collected the day before the liver biopsy. An abdominal US was performed during the same admission to evaluate the overall appearance of liver, biliary tract pathology, portal venous flow, and spleen size.
Liver biopsies of liver transplant donors (n = 15) were used as age-matched controls (median age for controls: 14.9 years; range, 2.2-19.8; P = 0.069).
Clinical and Laboratory Data
Clinical data, including gestation age, birth weight, weight and height at liver biopsy, duration of PN, composition of PN during 3 months preceding liver biopsy, number of blood culture-positive septic episodes from birth to study date, and surgical procedures, were collected from patient records. Anatomy of the remaining bowel, including length of small bowel, ileum, and colon and presence of an ileocecal valve, was obtained from the original operative records. Age-adjusted bowel length was calculated based on published age-specific normal values, where, at 38 weeks of gestation, normal small bowel and colon length is approximately 140 and 40 cm, respectively. Type of intestinal circuit was recorded as end-enterostomy, jejunocolic anastomosis, or jejuno-ileocolic anastomosis (27). Body mass index (BMI; weight [kg]/height [m2]) was calculated for adults and Finnish reference value-based body mass index-for-age (ISO-BMI) for children over 2 years of age.
Blood samples were analyzed for platelets, plasma alanine aminotransferase (ALT), aspartate aminotransferase (AST), glutamyl transferase (GT), albumin (ALB), pre-ALB, bilirubin, conjugated bilirubin, platelets, and coagulation markers (e.g., plasma tromboplastin time [P-TT], international normalized ratio [INR], and activated partial tromboplastin time [P-APTT]) by routine hospital laboratory methods. AST-to-platelet ratio index (APRI) was calculated according to Wai et al.
Liver Biopsies and Histological Analyses
All control samples were surgical wedge biopsies, and all follow-up biopsies were core needle biopsies. Biopsies were fixed in formalin, embedded in paraffin, sliced, and stained with hematoxylin and eosin. Additional stainings included reticulin, Periodic acid-Schiff (PAS), copper, and iron. Immunostaining for cytokeratin 7 (CK7) was performed using SP52 monoclonal antibody and the Ultra View Universal Detector Kit (Vienna, Tucson, Arizona). Two experienced pediatric liver pathologists and the primary researcher, blinded to clinical data, reviewed the slides together until a consensus was reached.
Lobular (0 = absent, 1 = present, and 2 = prominent), portal (0 = absent, 1 = fibrous expansions of most portal areas, 2 = focal portal-to-portal bridging, 3 = marked bridging, and 4 = cirrhosis), and overall fibrosis (Metavir fibrosis stage) was assessed. Steatosis was evaluated as the proportion of hepatocytes affected (0 = absent, 1 = <25%, 2 = 25%-50%, and 3 = >50% of hepatocytes) and classified as macro- or microvesicular. Foamy degeneration in hepatocytes was recorded (0 = absent, 1 = <25%, 2 = 25%-50%, and 3 = >50% of hepatocytes). Cholestatic changes included intracellular, canalicular, and ductular cholestasis (0 = absent, 1 = minimal, 2 = marked, and 3 = prominent). For analytical purposes, cholestasis was defined as the highest of the three cholestasis grades. Ductular proliferation was graded from 0 to 2 (0 = absent, 1 = focal, and 2 = generalized). Chronic cholestasis was assessed by CK7 expression in periportal hepatocytes (0 = absent, 1 = rare, 2 = present, 3 = prominent, and 4 = extensive). In addition, CK7-positive ductular reaction was assessed (0 = absent, 1 = present, and 2 = prominent). Portal inflammatory cell infiltrate was graded from absent to extensive (grade 0-4). When present, distribution of inflammatory cells was recorded. Accumulation of copper and iron in hepatocytes was scaled as absent to extensive (grade 0-4).[11, 13, 30]
Descriptive statistics are presented as mean (range), unless otherwise stated. An independent samples t test and Fisher's exact test were used to compare differences between two groups. Correlations were tested by Spearman's rank-correlation test. To identify predictors of liver fibrosis, a multivariate stepwise regression and a multivariate logistic regression model was performed. Potential risk factors of fibrosis, including grade of portal inflammation, age-adjusted small bowel length, presence of an ileocecal valve, type of nutrition (PN or weaned off PN), duration of PN, and number of septic episodes, were entered in the regression models. Level of statistical significance was set at 0.05.
Altogether, 38 (73%) patients (median age: 7.2 years) participated (Table 1). Causes of IF included short bowel syndrome (necrotizing enterocolitis: n = 10, midgut volvulus: n = 7; and small bowel atresia: n = 7) and intestinal dysmotility disorders (extensive aganglionosis of Hirschsprung's disease: n = 8; chronic intestinal pseudo-obstruction: n = 6). Short bowel patients had an average of 39 cm of small bowel remaining. Demographic variables and disease characteristics were comparable between participants and nonparticipants, making significant selection bias unlikely (Table 1). Causes of death among nonparticipants included end-stage liver failure (n = 2) and septic complications (n = 2) during PN.
|Gestation age at birth, weeks||35 (24-42)||35 (26-40)||0.870|
|Gestation weight, g||2,610 (580-4,680)||2,400 (430-4,320)||0.589|
|Age, years, median (range)||7.2 (0.2-27)||9.3 (0.7-22)||0.912|
|Weight Z-score at biopsy||−0.7 (−3.5-1.8)|
|Height Z-score at biopsy||−1.1 (−4.0-1.0)|
|BMI, kg/m2||18.4 (12.8-23.5)|
|Age at PN start, months||6.9 (0-130)||3.0 (0-34)||0.538|
|Weaned off PN, n||22||13||0.008|
|PN duration, months||49 (0.7-253)||31 (1.9-220)||0.292|
|Time after weaning off PN, years||8.8 (0.3-27)||7.2 (0-16)||0.338|
|Short bowel syndrome, n||24||11||0.764|
|cm||39 (7-123)||57 (15-120)||0.101|
|%||28 (3-86)||33 (15-91)||0.552|
|Ileum||4 (0-49)||8 (0-40)||0.360|
|cm||34 (18-57)||42 (13-85)||0.180|
|%||85 (60-100)||81 (31-100)||0.614|
|Dysmotility disorders, n||14||5||0.764|
|cm||200 (50-450)||200 (62-450)||0.978|
|%||65 (16-100)||25 (28-100)||0.711|
|Ileum||21 (0-80)||37 (0-150)||0.479|
|cm||11 (0-56)||26 (0-90)||0.279|
|%||21 (0-100)||24 (0-60)||0.896|
|Ileocecal valve preserved, n||16||8||0.765|
|Blood culture-positive septic episodes, n|
|n||1.8 (0-10)||1.5 (0-8)||0.672|
|per 1,000 catheter days||3.2 (0-28)||3.2 (0-25)||0.994|
Sixteen patients were on PN after 74 PN months (range, 2.5-204), and 22 had weaned off PN 8.8 years (range, 0.3-27) earlier, after 35 PN months (range, 0.7-250) (Table 2). The sixteen patients on PN received six (range, 2-7) PN infusions per week and 48% (range, 6%-100%) of total energy parenterally. Of the parenteral energy, 74% (range, 53%-92%) was given as glucose and 17% (range, 0%-33%) as fat in 14 patients. PN fat was given as an olive-oil–based regimen (0.6 g/kg/day; range, 0-1.6) combined with fish oil (1.0 g/kg/day; range, 0.2-1.9) in 4 patients. The absolute number of septic episodes and per 1,000 catheter days was equal in patients weaned off PN and in patients on PN (Table 2).
|Histology||Scale||On PN (n = 16)||Weaned Off PN (n = 22)||P Value|
|Age at biopsy, years, median (range)||4.2 (0.2-17)||9.9 (1.1-27)||0.001|
|Duration of PN, months||74 (2.5-204)||35 (0.7-250)||0.083|
|Time after weaning off PN, years||8.8 (0.3-27)|
|Blood culture-positive septic episodes|
|N||2.7 (0-10)||1.2 (0-7)||0.186|
|per 1,000 cathether days||3.6 (0-20)||2.9 (0-28)||0.750|
|Fibrosis, n (%)||14 (88)||14 (64)||0.143|
|Lobular||0-2||0.6 (0-1)||0.4 (0-1)||0.235|
|Portal||0-4||1.5 (0-4)||1.1 (0-2)||0.192|
|Metavir fibrosis stage||0-4||1.6 (0-4)||1.1 (0-2)||0.089|
|Steatosis, n (%)||8 (50)||10 (45)||1.000|
|Proportion of affected hepatocytes||0-3||0.8 (0-3)||0.7 (0-3)||0.943|
|Foamy degeneration||0-3||0.7 (0-3)||0 (0)||0.044|
|Cholestasis, n (%)||6 (38)||0||0.003|
|Intracellular||0-3||0.7 (0-3)||0 (0)||0.029|
|Canalicular||0-3||0.6 (0-3)||0 (0)||0.034|
|Ductular||0-3||0.1 (0-1)||0 (0)||0.333|
|Ductular proliferation||0-2||0.3 (0-2)||0.1 (0-1)||0.250|
|CK7 expression in periportal hepatocytes, n (%)a||7 (50)||2 (10)||0.015|
|0-4||0.9 (0-3)||0.1 (0-1)||0.024|
|CK7-positive ductular reaction, n (%)a||4 (29)||2 (10)||0.191|
|0-2||0.4 (0-2)||0.1 (0-1)||0.136|
|Portal inflammation, n (%)||6 (38)||2 (9)||0.050|
|0-4||0.5 (0-1)||0.1 (0-1)||0.093|
Abdominal US and Gastroscopy
Overall US appearance of liver (n = 34) was abnormal in 4 patients, including nodularity and increased hepatic echogenity. All of them had fibrosis (Metavir stage: 1.5; range, 1-2) and 2 had steatosis (grade 1 and 3) in liver biopsy. Excluding gallstones in 1 patient, no other biliary tract changes were observed. Two patients had undergone cholecystectomy for gallstones previously. Splenomegaly was found in 1 patient weaned off PN with Metavir stage 2 in liver biopsy and grade 2 esophageal varices in gastroscopy. Esophageal varices were not encountered in any other patient and all had normal liver vasculature.
Approximately half of patients on PN and up to 18% of patients weaned off PN showed increased plasma ALT, AST, GT, or conjugated bilirubin, or low plasma ALB, pre-ALB, and platelets (Table 3). INR was off normal range in 13% and 23% of patients on PN and weaned off PN, respectively (P = 1.000). APRI was comparable between groups.
|Biochemistry||On PN (n = 16)||Weaned off PN (n = 22)||P Value|
|Plasma ALTa (%)||9 (56)||4 (18)||0.015|
|U/L||78 (19-204)||34 (9-129)||0.009|
|Plasma ASTa (%)||8 (50)||4 (18)||0.037|
|U/L||104 (33-315)||38 (18-78)||0.013|
|Plasma GTa (%)||10 (63)||0||0.001|
|U/L||103 (16-297)||18 (10-51)||0.002|
|Plasma bilirubina (%)||5 (31)||2 (9)||0.082|
|μmol/L||61 (1-331)||13 (3-73)||0.071|
|Plasma conjugated bilirubina (%)||7 (44)||3 (14)||0.037|
|μmol/L||43 (1-215)||4 (1-23)||0.037|
|Plasma ALBa (%)||7 (44)||4 (18)||0.010|
|g/L||34.5 (23.7-41.8)||40.6 (31.5-49.9)||0.002|
|Plasma pre-ALBa (%)||9 (56)||3 (14)||0.005|
|mg/L||151 (40-282)||194 (89-252)||0.020|
|Blood plateletsa (%)||4 (25)||4 (18)||0.453|
|E9/L||317 (57-567)||264 (67-465)||0.265|
|INR||1.2 (1.1-1.3)||1.2 (1.0-1.5)||0.531|
|APRI||1.4 (0.1-8.9)||0.4 (0.03-1.9)||0.141|
Liver biopsies were considered representative because over 10 (ranging from 5 to over 20) portal tracts were found in 84% of samples. Overall, 84% of patients had abnormal liver histology. Control liver samples showed neither fibrosis, cholestasis, nor portal inflammation, whereas mild steatosis was found in 2 (13%). Frequency of liver fibrosis (in 74%; P = 0.001), portal inflammation (21%; P = 0.088), and steatosis (47%; P = 0.028) was increased among patients. Six patients (all on PN) displayed histological cholestasis (16%; P = 0.102), with increased intracellular (grade 0.3 [range, 0-3]: P = 0.032) and canalicular cholestasis (grade 0.2 [range, 0-3]; P = 0.037), compared to controls. Entirely normal liver histology was found in only 6 patients (16%) who had experienced less-septic episodes (0.3 [range, 0-2] versus 2.1 [range, 0-10]; P = 0.009) and had longer remaining age-adjusted small bowel length (79% [range, 42%-100%] versus 35% [range, 3%-100%]; P = 0.001), compared to patients with abnormal liver histology.
Overall, 94% (15 of 16) of patients on PN and 77% (range, 17%-22%) of patients weaned off PN displayed abnormal liver histology (P = 0.370). Although cholestasis had completely resolved and portal inflammation was significantly milder after weaning off PN, liver fibrosis and steatosis were comparable between groups (Table 2; Fig. 1).
Fibrosis similarly governed liver histology in patients on PN (88%) and patients weaned off PN (64%), concentrating to portal areas in both patient groups (Table 2; Fig. 1). Age at PN start, duration of PN, time after weaning off PN, absolute and percentage of age-adjusted small bowel length, ileum length, and number of blood culture-positive septic episodes correlated with Metavir fibrosis stage and portal fibrosis (Table 4; Fig. 2). Patients without an ileocecal valve had more frequently (20 of 22) and more advanced fibrosis, compared to those with a preserved Ileocecal valve (8 of 16; P = 0.008) (Fig. 3). Lobular fibrosis correlated with ileum length, duration of PN, and absolute (r = −0.334; P = 0.035) and age-adjusted colon length (r = −0.391; P = 0.015) (Table 4). In a multivariate stepwise linear regression model (adjusted R2 = 0.425), age-adjusted small bowel length (ß = −0.533; P = 0.001), grade of portal inflammation (ß = 0.291; P = 0.030), and absence of an ileocecal valve (ß = 0.267; P = 0.044) were significant predictors for Metavir fibrosis stage. In a multiple logistic regression model (for the full model: χ2 = 18.71; df, 4; P < 0.001), the strongest independent predictor of fibrosis was absence of an ileocecal valve (odds ratio = 8.9; 95% confidence interval: 1.0-79; P = 0.05). APRI correlated positively with Metavir stage (r = 0.404; P = 0.013).
|Clinical Variables||Fibrosis||Steatosis||Cholestasis||Portal Inflammation|
|Age at PN start,||r||−0.209||−0.332||−0.353||−0.122||−0.167||−0.199|
|Duration of PN,||r||0.466||0.324||0.387||0.390||−0.068||−0.157|
|Time after weaning off PN (n = 22),||r||−0.208||−0.325||−0.384||−0.139||−0.429||−0.272|
|Small bowel length,||r||−0.140||−0.510||−0.486||−0.362||0.021||−0.074|
Steatosis was equally common during (50%) and after weaning off PN (45%), including equal amounts of microvesicular (50%) and macrovesicular (50%) steatosis in both groups (Table 2; Fig. 1). No Mallory bodies were observed. Steatosis was associated with duration of PN and absolute and age-adjusted small bowel length (Table 4). Patients on PN had more foamy degeneration, compared to patients weaned off PN (Table 2). Neither steatosis nor fibrosis was related to BMI or weight for length (r = −0.016-0.027; P = 0.888-0.934).
Portal inflammation was more common during than after weaning off PN (Table 2; Fig. 1) and consisted mainly of neutrophils and lymphocytes similarly in both groups. Degree of portal inflammation associated with degree of cholestasis (r = 0.333; P = 0.041) and portal fibrosis (r = 0.333; P = 0.041).
Cholestasis was found in 6 patients on PN and in none after weaning off PN (Table 2; Fig. 1). Time after weaning off PN was inversely associated with cholestasis grade (Table 4). Expression of CK7 in periportal hepatocytes was increased in patients on PN (Table 3) and correlated with ileum length (r = −0.347; P = 0.041) and the number of blood culture-positive septic episodes (r = 0.421; P = 0.013). In patients on PN, canalicular cholestasis was associated with daily PN glucose dose (g/kg/day; r = 0.631; P = 0.009), but not with daily PN fat dose (r = 0.022; P = 0.934). The number of weekly PN infusions tended to associate with intracellular and intracanalicular cholestasis (r = 0.496 and P = 0.051 for both). Intracellular and intracanalicular cholestasis grade correlated with plasma ALT (r = 0.471-0.476; P = 0.003), AST (r = 0.491-0.520; P < 0.003), GT (r = 0.542-0.519; P = 0.001), total bilirubin (r = 0.516-0.527; P = 0.001), and conjugated bilirubin (r = 0.538-0.549; P = 0.001).
In this population-based, cross-sectional study on liver histology in pediatric IF, we found, first, that over half of the patients on long-term PN had significant or severe (Metavir stage ≥ 2) histological liver fibrosis accompanied with deranged liver biochemistry. Second, despite diminishing portal inflammation and resolution of cholestasis, significant liver fibrosis and steatosis persists after weaning off PN. Third, in addition to duration of PN, extensive small intestinal resection and loss of ileocecal valve as well as septic episodes are major risk factors of histological liver fibrosis, which was occasionally associated with signs of PH, such as esophageal varices or splenomegaly. Although laboratory markers of liver function usually normalize after weaning off PN, liver histology remains abnormal up to 9 years after weaning off PN in the majority of IF patients.
Since the first reports of IFALD, the liver injury is described as initially cholestatic with a variable degree of fibrosis and steatosis. During PN, elevated serum biomarkers of liver function, such as bilirubin, ALT, and AST, are the earliest signs for liver dysfunction. Biochemical alterations have been previously reported in up to 57% of children on long-term PN. With progression of IFALD, a fall in ALB and prolonged coagulation occurs, whereas thrombocytopenia suggests hypersplenism associated with advanced hepatic fibrosis or cirrhosis. Our results of abnormal liver histology in the majority of IF patients on long-term PN, characterized by cholestasis, portal inflammation, fibrosis, and steatosis with elevated biomarkers of liver function, are in accord with previous findings. An especially alarming observation was that nearly 60% of the patients on long-term PN had at least Metavir stage 2 liver fibrosis accompanied with deranged liver biochemistry. During PN, histological cholestasis was associated with portal inflammation, and fibrosis-binding cholestasis and portal inflammation close together, in the pathogenesis of liver fibrosis in IFALD. The fact that intracellular cholestasis correlated with parenteral glucose, rather than fat dose, may be explained by our clinical practice of avoidance of parenteral lipids among patients, who develop signs of IFALD.
Although we and others have demonstrated resolution of biochemical cholestasis after weaning off PN,[10, 14] some studies suggest that liver histology may still remain abnormal. Previous retrospective studies and case reports have reported persistent hepatocellular cholestasis, lobular disarray, hepatocyte ballooning, marked fibrosis, and even progression of the liver injury after weaning off PN.[11, 15] Transition from olive-oil–based PN lipid to fish-oil–based PN lipid emulsion may result in reversal of biochemical indicators of cholestasis. However, 2 patients with persisting liver fibrosis 3 and 11 months after transition to fish-oil–based PN have been reported on. In this study, we found abnormal liver histology, including mainly fibrosis (Metavir stage 2 in 50%) and steatosis with occasional portal inflammation, in 77% of patients after weaning off PN an average of 8.8 years before. Interestingly, degree of liver fibrosis was comparable during and after weaning off PN, although the weak inverse correlation between Metavir stage and time after weaning off PN suggest that some resolution of fibrosis may occur over time.
Bacterial overgrowth, epithelial changes, and impaired local immunity of the small intestine may provide potential mechanisms causing and maintaining liver injury in IFALD, both during and after weaning off PN. In a mouse model of IF, PN-induced increase in intestinal permeability promotes Toll-like receptor 4–dependent Kupffer cell activation and liver injury, presumably caused by bacterial translocation. In short bowel syndrome, loss of barrier function of the ileocecal valve, adaptation-induced bowel dilatation, and impaired motility after massive intestinal resection are known risk factors for bacterial overgrowth. Together with increased intestinal permeability, these alterations may promote bacterial translocation and subsequent liver injury also in IF patients. Here, the number of blood culture-positive septic episodes, reflecting both central venous catheter- and bacterial translocation-related septic episodes, correlated positively with liver fibrosis and chronic cholestasis (periportal CK7 staining). Moreover, the patients with the shortest remaining small bowel and those without an ileocecal valve had the most advanced liver fibrosis stage.
The exact mechanism of liver protection exerted by the small intestine is most likely multifactorial, but may involve enterohepatic circulation of bile acids. Short length of the remaining small intestine also reflects decreased enteral absorption with increased and prolonged PN requirements. Accordingly, duration of PN and extensive small intestinal resection positively correlated with both fibrosis and steatosis. The fact that liver fibrosis stage was inversely related to young starting age of PN emphasizes the vulnerability of newborn liver function. In a multivariate analysis, age-adjusted small bowel length, portal inflammation, and absence of an ileocecal valve were the most significant predictors of Metavir fibrosis stage.
Although APRI correlated with histological liver fibrosis, any of the conventional liver function tests were off normal limits only in 63% of patients on PN and in 18% of patients weaned off PN, whereas liver US was abnormal only in 4 patients. Thus, conventional liver biochemistry and abdominal US alone are insufficient for reliable evaluation of liver damage in IF patients. In this respect, noninvasive evaluation of the liver with FibroScan is a promising option.
We were able to reach a high participation rate. Altogether, 73% of the patients participated and the participants and nonparticipants were comparable in terms of demographic variables and disease characteristics, making significant selection bias unlikely (Table 1). Specifically, starting age of PN, duration of PN, length of the remaining small intestine, and number of septic episodes were comparable. A challenge in our study design is the wide age range of the patients, whereas treatment of IF patients has significantly developed over time. Composition of parenteral lipids has changed from soy-based to olive-oil– and fish-oil–based lipids, amount of PN fat is limited, and early initiation of enteral nutrition and cyclic PN infusions are persuaded.[5, 35, 47] Although the changes in clinical practice may have modulated our results and may hamper their applicability for newly treated children with IF, this study provides reliable population-based information regarding the current long-term outcomes.
The authors thank pediatric radiologists K. Lauerma, R. Kivisaari, and R. Seuri from the Medical Imaging Center, Helsinki University Central Hospital (Helsinki, Finland), for carrying out the abdominal US exams and US-guided liver needle biopsies.
- 30Assessment and differential diagnosis of pathological features. In: Scheuer PJ, Lefkowitch PJ, eds. Liver Biopsy Interpretation. Volume 1, 7th ed. Philadelphia, PA: Elsevier Saunders; 2006:35-51., .