Body Composition of Infants With Biliary Atresia

Biliary atresia (BA) is a rare disease of infancy in which obliteration and fibrosis of the intrahepatic and extrahepatic bile ducts leads to cholestasis and liver fibrosis (1.). Treatment of BA primarily consists of Kasai hepatoportoenterostomy (KPE), which results in a successful clearance of jaundice in approximately 50% of cases (2., 3.). Irrespective of the result of the KPE, however, 70% to 80% of children with BA in Western countries will sooner or later develop end-stage liver disease, ultimately necessitating liver transplantation (LT) before adulthood (4., 5.).

Infants with chronic cholestatic liver disease are prone to malnutrition because of several factors: intestinal malabsorption of fat and fat-soluble vitamins, abnormalities in the metabolism of liver macronutrients’ metabolic pathways, depression of hepatic plasma insulin-like growth factors, impeded nutritional intake because of discomfort because of ascites, cholestasis and/or organomegaly, and increased metabolic demands (6.-9.). This multifactorial process can lead to neurodevelopmental delay, decreased bone mineralization, loss of muscle mass, and even to paradoxal increase of fat storage, making these infants prone to sarcopenia and sarcopenic obesity (10.-15.).

In infants with BA, a detailed analysis of the nutritional status has not been easy. A variety of measurements have been developed, such as anthropometric parameters including weight and mid upper arm circumference (MUAC), assessment of daily intake requirements, and laboratory tests, such as serum albumin concentration but their precise interpretation and the correlation between these parameters are rather unclear (16.). Weight alone is unreliable partly because of possible overestimation caused by hepatosplenomegaly and/or ascites (17., 18.). MUAC measurement is a noninvasive, inexpensive, and readily available technique, and has been described previously to assess growth in infants with BA (17.-20.). Therefore, MUAC has frequently been considered the gold standard of nutritional status. MUAC as determinant of nutritional status is, however, imprecise in quantitative analysis of muscle and fat stores, as well as in determining their relative distribution within the body (21.). Sarcopenia at the time of LT screening, however, is associated with waiting list mortality and early posttransplant graft failure and mortality in adults, underlining the importance to obtain reliable information that is not provided by current measures, such as MUAC (18., 22.-24.). Even more at risk for adverse clinical outcome after LT are sarcopenic patients with obesity (25., 26.). Sarcopenia and sarcopenic obesity can be reliably assessed by determining abdominal muscle area at computed tomography (CT) scans in adults (27., 28.). Total psoas muscle surface area (tPMSA) or cross-sectional skeletal muscle area (CSMA) are preferably used (29., 30.). tPMSA has been shown to change in chronically ill adults and children with end-stage liver disease (13., 31.).

The aim of this study was to assess whether CT-based body metrics can be reproducibly determined in infants with BA. In addition, the correlation between CT-based body metrics and MUAC was evaluated.

MATERIALS AND METHODS

Radiological Measurements

All infants underwent CT imaging of the abdomen as part of LT screening, mainly to determine vascular anatomy and to exclude liver lesions. The quantities of intra-abdominal fat and skeletal muscle area were determined using standard abdominal contrast-enhanced CT scans. Scans were obtained using our standard paediatric liver protocol, with a 512×512 matrix; slice thickness varied from 1 to 3 mm. Cross-sectional areas (in square millimeters) of different tissue compartments were measured on an axial CT slice at the level of L3 vertebra. Muscle boundaries were manually drawn independently by an abdominal radiologist (J.P.P.) and a research fellow (A.L.G.) who was supervised by another abdominal radiologist (R.J.d.H.). Within these borders, muscle and fat were defined based on their specific differences in attenuation. The thresholds used were −29 to 150 Hounsfield Units (HU) for skeletal muscle density, and −190 to −30 HU for adipose tissue (35., 36.). tPMSA, CSMA, and total abdominal fat area within these contours were automatically segmented by in-house developed software (SarcoMeas) (28.). tPMSA was defined as the sum of the left and right PMSA. CSMA concerned the sum of the psoas, abdominal wall, and paraspinal muscle measurements. Total abdominal fat area was defined as the sum of abdominal and subcutaneous adipose tissue areas. Figure 1 shows an example of the determination of CT-based body metrics on an axial CT slice. tPMSA index (psoas muscle index; PMI [mm²/meters²]), CSMA (skeletal muscle index; SMI [mm²/meters²]), and total abdominal fat area index (abdominal fat index; AFI [mm²/meters²]) were calculated by correcting measurements for length in square meters.

RESULTS

Patient Characteristics

A total of 82 BA infants under the age of 2 years, underwent an abdominal CT scan as part of the LT screening process. Data of 2 infants (2%) were not usable, because of incomplete display of the abdominal muscle area at the level of L3 vertebra. Data from the remaining 80 infants (52 girls, 65%) were analyzed in the study. The median age at the time of KPE was 59 [IQR: 49--75] days. The median age at the time of screening for LT was 4.6 [IQR: 3.6--5.8] months. At LT screening, mean weight z-score was −0.46 (±1.12), mean MUAC z-score was −1.35 (±0.88), and mean length z-score was -0.39 (±1.04), respectively. Other baseline characteristics can be found in Table 1.

TABLE 1. Demographic and clinical characteristics

Computed Tomography-based Body Metrics

The ICC for determining CT-based body metrics varied between 0.92 and 0.99 (Table 2). The mean values of CT-based body metrics and indices of BA infants at the time of LT screening can be found in Table 2. Intrasubject correlation coefficient of CSMA with tPMSA was moderately strong (r = 0.71, Figure, Supplemental Digital Content 1, http://links.lww.com/MPG/B892). In the total study population, CT-based body metrics were converted to indices and correlated with MUAC z-score. The correlation coefficient of PMI with MUAC z-score was poor (r = 0.02, Fig. 2A). Also, the correlation coefficient of SMI with MUAC z-score was poor (r = 0.06, Fig. 2B). The correlation coefficient of AFI with MUAC z-score was fair (r = 0.43, Fig. 2C). The lowest 25th percentile for the CSMA-to-total abdominal fat area ratio was 0.93, the correlation coefficient of CSMA with total abdominal fat area was fair (r = 0.36). All infants in the lowest quartile group had a normal nutritional status according to MUAC z-score.

TABLE 2. Overview of determined computed tomography-based body metrics and interobserver agreement of each determined variable

DISCUSSION

The present study showed that CT-based body metrics can be reproducibly determined in infants younger than 2 years of age with end-stage liver disease because of BA. tPMSA, CSMA, and total abdominal fat area can be quantified with very strong interobserver agreement with an ICC varying between 0.92 and 0.99. Furthermore, our study demonstrates that CT-based body metrics correlated only poor to fair with MUAC.

The absence of a close relationship between MUAC and inner body composition may be explained by the fact that infants with chronic liver disease are susceptible to changes in body composition (6., 7.). Previous studies in children reported that the accuracy of MUAC in determining fat and muscle mass in the body was limited compared with CT or MRI (39., 40.). Validation of MUAC with Dual-energy X-ray Absorptiometry, however, showed that MUAC is a good predictor of total body fat, but not of total body muscle mass in healthy and chronically ill children (21.). In our cohort, MUAC z-score correlated better with AFI compared with SMI. As SMI is most important in determining nutritional status, CT-based body metrics provide relevant additional data on inner muscle mass.

For years, MUAC has been considered the gold standard for measuring clinical nutritional status in infants with BA (17.-20.). The value of nutritional assessment through MUAC is, however, limited as MUAC is imprecise in quantitative analysis of muscle and fat stores (21.). Hurtado-López et al debate the potential value of arm anthropometrics in evaluating nutritional status of infants and toddlers with chronic liver disease by comparing total, muscle, and arm fat areas calculated with MUAC and total subcutaneous fat based formulas (7.). The arm indicators that separate muscle from fat identified a lower proportion of cases with abnormal nutritional status compared with MUAC (7.). Therefore, MUAC z-scores should be interpreted with caution.

Currently, it is still not known which method (ie, assessment of tPMSA or CSMA) should be used to determine sarcopenia (41.). Measurement of tPMSA has been reported as simple and convenient, and has been stated to be predictive of morbidities in certain conditions (41.). Others have, however, argued that psoas muscle measurements are not representative of CSMA, and thus overall sarcopenia (42., 43.). In the present study, tPMSA and CSMA were only moderately correlated in infants with BA. Therefore, we have decided to include CSMA rather than tPMSA in the definition of sarcopenic obesity. Whether CSMA should be preferred over tPMSA to determine sarcopenia and sarcopenic obesity in infants, however, should be assessed in further studies.

The cut-off for sarcopenic obesity was a CSMA-to-abdominal fat area ratio below 0.93. We are the first to report CSMA-to-total abdominal fat area ratios in infants. Mangus et al observed an overall decreased muscle and increased fat mass on CT in children (mean age of 7.6 years) with chronic liver disease compared with controls (14.). In the literature on adult patients, a low CSMA-to-total abdominal fat area ratio before LT, has been found to be prognostically unfavourable (25.). A remarkable finding is that all infants in the lowest quartile group had a normal nutritional status according to MUAC z-score, suggesting that these sarcopenic obese infants nevertheless maintain normal MUAC values. This can be explained by the fact that our hospital has an intensive feeding protocol and dietary follow-up program for these fragile patients. Infants with BA awaiting LT are hypermetabolic in general and catabolic during fasting (44.). Therefore, adequate nutrition is essential to prevent that caloric intake is stored as fat rather than utilized for reversing the effects of catabolism. CSMA-to-total abdominal fat area ratio in infants with BA seems to provide essential additional information on inner body composition that is not recognized when using MUAC. A decreased CSMA-to-total abdominal fat area ratio may even prompt further investigation into: the progression of the liver disease of the infant and a potential acceleration of candidacy for transplantation.

The strong interobserver agreements are in line with the scarce literature on this topic. Lurz et al (31.) reported an ICC of 0.99 for the determination of tPMSA in children aged 0 to 18 years listed for LT. Our findings regarding the poor to fair correlation between CT-based body metrics and other traditional parameters of nutritional status are in agreement with previous literature that used parameters as weight and albumin level in infants (14., 31.). Therefore, it can be hypothesized that CT-based body metrics provide novel objective nutritional biomarkers for the comprehensive nutritional assessment of children with end-stage liver disease. Determining accurate markers of inner body composition is important as at the time of LT screening, sarcopenia, and even more important sarcopenic obesity, are associated with waiting list mortality and early posttransplant graft failure and mortality in adults (18., 22.-26.). In future studies, the potential correlation between CT-based body metrics and peri- and posttransplant outcomes in children should be investigated.

Our study has several limitations. First, because of its retrospective nature, 2 infants had to be excluded because of inadequate display of the complete abdominal muscle area at CT. Second, radiation protection has to be kept in mind, and thus, CT-based body metrics seem only applicable to infants who have a primary diagnostic reason to perform a CT scan, such as determining the vascular anatomy in preparation for an LT. Thus, it is not recommended to perform CT scans solely to assess CT-based body metrics. Third, reference values for tPMSA of healthy children only exist for ages between 1 and 16 years (45.). As our population consisted of infants with a median age of 4.6 months, we could not compare tPMSA values with existing validated reference values. Fourth, in our cohort, 12% of infants was born prematurely. As it is still unknown whether prematurity leads to changed body composition, especially increased fat mass, the relatively large percentage of prematurity could have influenced our results (46.-48.). Finally, we do not have reliable data on the day-to-day muscle activity of these infants. We speculate that, in the light of the feeding attention that these infants receive, the optimization of the child's condition could be found in “training” of muscle tissue as the only viable option for intervention.

In conclusion, in infants with BA under the age of 2 years, it is possible to determine CT-based body metrics during LT screening with very strong interobserver agreement. The present study shows that CT-based body metrics provide additional, relevant information on inner body composition in infants with biliary atresia that is not available when using MUAC alone. These findings could be the basis for future studies investigating nutritional status by using CT-based body metrics, as well as evaluating their potential impact on transplant outcomes.