Neonatal sepsis contributes 25% to neonatal mortality in developing countries (1) and accounts for about 0.52 million deaths annually in the world (2). Gram-negative organisms are mainly responsible for sepsis in developing countries and are commonly associated with multiorgan dysfunction, resulting in high mortality and poor outcome. Hepatobiliary dysfunction in the form of cholestatic jaundice or elevated liver enzymes has been reported in more than two-thirds of preterm neonates experiencing Gram-negative septicaemia (3). Cholestatic jaundice in a neonate may also signify mechanical obstruction of the biliary tract because of biliary atresia or a metabolic disorder. To plan further investigations towards a primary liver disease and appropriate management, including referral to specialised centres, it is important to distinguish the newborns having cholestatic jaundice because of septicaemia from those having obstructive, hereditary, or metabolic disorders. The exact course, pattern of abnormalities, and outcome of sepsis associated hepatobiliary dysfunction in neonates have not been described. The present study was conducted to determine the prevalence, pattern, and course of hepatobiliary dysfunction in neonatal sepsis, and to evaluate its effect on the survival and growth.
The study was conducted amongst the hospital-born neonates during a period of 9 months from March 2006 to November 2006 in the University College of Medical Sciences and Guru Tegh Bahadur Hospital, Delhi, India. All of the neonates with clinical suspicion of sepsis or a positive sepsis screen (μ-erythrocyte sedimentation rate, C-reactive protein, total leukocyte count, band count [any 2 positive out of 4]) (4) were admitted to the neonatal intensive care unit of the hospital and venous blood culture was done. All of the symptomatic newborns having a positive blood culture for pathogenic bacteria were defined as having confirmed sepsis, and were eligible for inclusion in the study. Neonates with major congenital malformations were excluded from the study. Informed consent was obtained from the parents of eligible participants. The study was approved by the institutional research board, including the ethics clearance.
Data Collection and Case Management
At recruitment, the baseline information of mother and infant were recorded. Maternal history of fever, diarrhoea, vomiting, leaking or bleeding per vaginam, and foul-smelling liquor was recorded. Any history of consanguineous marriage or family history of jaundice/cholestasis/liver disease was noted. Details on the mode of delivery and resuscitation at birth were recorded. Birth weight was recorded to the nearest 5 g, and the gestational age was calculated from the first day of the last menstrual period. In case the last menstrual period was not known, gestational age was estimated by modified Ballard score (5). The symptoms that suggested infection such as poor feeding, prefeed gastric residues, respiratory distress, fever, hypothermia, apnoea, bleeding, and the like were also recorded. Clinical examination included anthropometry, clinical assessment of severity of jaundice, clinical assessment of liver and spleen sizes, and noting the colour of urine and stools.
A total of 3 mL of venous blood was collected in the bile broth bottle for bacterial culture. The blood culture bottle was kept in incubation for 24 hours and the report was available after 72 hours. Only those newborns in whom blood culture was positive for pathogenic bacteria were included in the study. Blood samples for estimation of liver function (total and direct bilirubin, serum alkaline phosphatase, alanine aminotransferase [ALT], serum proteins, prothrombin time) were taken after 72 hours of clinical suspicion of sepsis. Liver function tests (LFTs) were repeated on the 10th day of onset of symptoms in all of the cases. In babies with abnormal LFT (direct bilirubin >20% of total with a minimum level of 2 mg/dL or ALT >50 U/L), the tests were repeated weekly until 1 month and then fortnightly until they normalised. Ultrasonography of the abdomen was done in all of the babies to evaluate for any choledochal cyst or extrahepatic biliary atresia. Those with unresolving or worsening levels of bilirubin even after 2 weeks of life were investigated in detail, including an ophthalmic fundus examination, immunoglobulin M levels for toxoplasma, cytomegalovirus, and rubella, and urine for reducing substances.
The neonates were started on intravenous antibiotics and other appropriate supportive management as soon as the sepsis was suspected on the basis of clinical signs and symptoms or a positive sepsis screen. The first-line antibiotics were a combination of ciprofloxacin and gentamicin. The other antibiotics used were cefotaxime, ceftriaxone, and amikacin according to the sensitivity pattern. In cases wherein the organism was found resistant to all of the tested drugs, the drugs used were piperacillin plus tazobactam or meropenem. Vancomycin was used when the cultured organism was methicillin-resistant staphylococcus. Antibiotics were given for 10 to 14 days in cases of sepsis and 21 days in cases of meningitis. Breast-feeding or expressed breast milk feeding with cup was started whenever babies were considered stable. Parenteral nutrition was not received by any of the babies.
Children having cholestatic jaundice were started on multivitamin supplementation initially in intravenous form and orally once they improved clinically. Oral supplementation was composed of vitamin A, 25,000 IU given on alternate days, cholecalciferol 6000 U/day, vitamin E 100 IU/day. Vitamin K 2.5 mg (intravenously or intramuscularly) was given weekly. These were continued until the LFTs normalised.
Monitoring and Follow-up
Subjects were monitored for clinical improvement in cholestasis in the form of clearing of dark-coloured urine or pale stools, and change in liver and spleen size. Biochemical monitoring was done as per the schedule outlined above. The babies were discharged once they improved clinically, feeding was fully established, and antibiotic course was completed. They were followed up after 1 week (+2 days) in the neonatal high-risk clinic, where a detailed clinical and anthropometric examination was done. This was continued weekly until 1 month (+1 week) of age and fortnightly thereafter until at least 3 months (+1 week) of age.
Sample Size Estimation and Statistical Analysis
In a study by Shamir et al (3), liver enzyme abnormality was seen in 46% of neonates with Gram-negative septicaemia. Based on this estimated prevalence of 46%, sample size considered adequate for finding out the proportion of septicaemic neonates having hepatobiliary dysfunction was 95 with a confidence limit of 95% (α = 0.05) and specified absolute precision of 10% (6).
The data were entered in Microsoft Excel and analysed with the help of SPSS version 12.0 software (SPSS Inc, Chicago, IL). Descriptive statistics were used to calculate the frequencies and percentages. Proportions were compared with the χ2 test and intergroup means were compared using the unpaired Student t test.
During the study period, a total of 153 neonates were diagnosed as having confirmed (culture positive) sepsis. The mean (±SD) age at enrollment of these babies was 3.18 ± 1.38 days. The mean (±SD) gestational age of the enrolled subjects was 35.45 ± 3.42 weeks with a mean (±SD) birth weight of 2.02 ± 0.63 kg. The majority (71.2%) of enrolled subjects were low-birth-weight babies, and almost one-third (32.7%) were preterm. Parental consanguinity was present in 2 neonates, and there was a history of neonatal jaundice requiring phototherapy in siblings of 2 of the included subjects. Family history of cholestatic jaundice or any chronic liver disease was not present in any of the included subject.
The most common organism isolated was Klebsiella pneumoniae, seen in 146 of 153 (95.4%) subjects. Other organisms isolated were Staphylococcus aureus in 3 babies, and α-hemolytic streptococci, Staphylococcus epidermidis, and Escherichia coli and citrobacter in 1 case each. Klebsiella was sensitive to ciprofloxacin in 82% (120/153) babies followed by amikacin (70%), gentamicin (69%), chloromycetin (26%), and cefotaxime (14%).
Prevalence of Hepatobiliary Abnormalities
Any hepatobiliary dysfunction (direct bilirubin >20% of total with a minimum level of 2 mg/dL or ALT > 50 U/L) was found in 83 (54.2%) subjects. Cholestatic jaundice was seen in 65 (42.5%) subjects, whereas deranged ALT was seen in 57 (37.3%) cases. One-fourth (25.4%) of septicaemic babies had cholestatic jaundice as well as derangement of ALT (Table 1).
Onset, Pattern, and Course
The onset of cholestasis (n = 65) was seen at day 3 of onset of sepsis in 80% (52 of 65) babies. It was present by day 10 in another 15%. Only <5% of babies developed cholestasis after day 10. The maximum value of direct bilirubin was achieved by day 10 of onset of sepsis in 68% (44 of 65) babies, whereas in 22% the maximum value was achieved as early as day 3 of sepsis. In the remaining 10% cases, maximum value of direct bilirubin was seen at or after day 17 of sepsis. The direct bilirubin level normalised by day 10 of sepsis in 15% of babies, and in 48% it normalised by day 17. In 85% of babies having cholestatic jaundice, the levels normalised by 1 month. Cholestatic jaundice resolved in all survivors by the age of 2 months.
In babies with deranged ALT, in almost two-thirds (68.4%), the onset was seen by day 3 of sepsis, and in another 26.3% babies by day 10 of sepsis. The maximum value was achieved by almost one-third (31.5%) of babies at day 10 of sepsis and in another one-third (31.5%), it was achieved at day 17 of sepsis. The ALT normalised by 1 month (day 31) in about half (49.2%) of the babies, whereas in the remaining half (50.8%) the raised levels persisted for variable periods up to 3 months.
Of the 83 babies who had hepatobiliary dysfunction, 35 (42.1%) babies had direct bilirubin in the range of 2 to 5 mg/dL (Table 2), 25 (30.1%) babies had direct bilirubin in the range of 5 to 10 mg/dL, and 10 (12%) had levels between 10 and 15 mg/dL. Only 1 baby had direct bilirubin beyond 15 mg/dL. The levels of ALT ranged between 51 and 150 IU/dL in 34 (41%) babies and 23 (28%) had ALT levels >150 IU/dL. Total protein <4 mg/dL was seen in 50 (60.2%) babies, whereas serum albumin <2.5 mg/dL was found in 66 (79.5%) babies.
Comparison of Parameters and Outcomes Between Children With or Without Hepatobiliary Dysfunction
The mean (±SD) birth weights and gestational ages of children with or without hepatobiliary dysfunction were comparable (birth weight 2.07 ± 0.64 kg vs 1.96 ± 0.61 kg [P = 0.62]; gestational age 35.75 ± 3.32 weeks vs 35.10 ± 3.54 weeks [P = 0.95]). Table 3 presents the mean and peak values of total bilirubin, direct bilirubin, and ALT in babies with hepatobiliary dysfunction (group I), and in babies without hepatobiliary dysfunction (group II).
Of the 153 babies with sepsis, 49.7% babies died during the course of the stay with mean (±SD) age of death being 6.1 ± 4.7 days. The prevalence of hepatobiliary dysfunction was higher in survivors as compared with deaths (56.7% vs 43.4%; P < 0.01). On comparing the biochemical parameters between the deaths and survivors, it was found that mean direct bilirubin levels in survivors were higher at day 3 and day 10 of onset of sepsis; however, this was not statistically significant. The maximum value of direct bilirubin reached was significantly higher (P < 0.01) in survivors as compared with deaths (Table 4). Similarly, the levels of ALT were higher at all time periods (days 3 and 10) in survivors as compared with the deaths. This was again not significant, except the maximum ALT, which was significantly higher in survivors (P < 0.01) (Table 4).
On follow-up up to 3 months after discharge, there was a significant serial increase in weight, length, and head circumference in both groups of survivors, but the gain in any of the parameters was not significantly different in children with or without hepatobiliary dysfunction.
This prospective observational study, conducted in 153 neonates with blood culture–confirmed sepsis, documented that hepatobiliary dysfunction is extremely commonly seen early in the course of neonatal septicaemia. Although cholestatic jaundice (42.5%) was more common than raised ALT (37.3%), almost one-fourth of the babies had both abnormalities. As an overwhelming majority (96.7%) of infections in our study were caused by Gram-negative organisms, especially Klebsiella, the results are reflective of the hepatobiliary abnormalities in Gram-negative sepsis only and may not be true for Gram-positive organisms. Earlier studies from our institute and from other regions in developing countries also report Klebsiella to be the most common organism responsible for neonatal sepsis (3,7–10). This may be explained by higher incidence of surface colonization with this bacteria and higher virulence of Klebsiella strains in our population (9,10).
Shamir et al (3) conducted a study in premature babies with Gram-negative septicaemia (n = 54) and Gram-positive septicaemia (n = 31). Liver enzyme abnormality was seen in 46.3% of neonates with Gram-negative sepsis and only 12.9% of Gram-positive sepsis. In addition to elevated liver enzymes, conjugated bilirubin was significantly elevated, but the exact prevalence was not studied. These infants had significantly elevated conjugated bilirubin without increased alkaline phosphatase. Our study documented that these abnormalities are not restricted to premature infants because two-thirds of our study infants were born after completing 37 weeks.
The pathogenesis of hepatic dysfunction secondary to sepsis is believed to be multifactorial. Enhanced haemolysis causing overload of immature liver function or hepatocellular damage has been implicated (11); however, histopathologically, a majority of cases reveal a pattern of intrahepatic cholestasis with little or no evidence of haemolysis or hepatic parenchymal damage (12,13). The increased frequency of cholestasis associated with Gram-negative sepsis suggests a pathogenetic role of circulating endotoxins in causing biliary stasis and hepatic parenchymal injury (14). The intrinsic toxicity of endotoxins has been linked to lipid A portion, which has a similar structure in all Enterobacteriaceae.
We documented that cholestatic jaundice associated with sepsis was early onset (within 3 days), and subsided by 1 month in most of the neonates. Hepatic enzyme abnormalities take a slightly longer time to appear, and the course is also more protracted. Tiker et al (15), in a retrospective analysis of 42 cases with early-onset conjugated hyperbilirubinemia (mean age of onset 10 days), documented culture-proven sepsis to be the cause in 35.7% of cases. Most other series of conjugated hyperbilirubinemia in infants have focused on the chronic cholestatic jaundice in older neonates and infants, and thus have documented a small contribution of bacterial infection in causation. The main causes found in these studies are congenital infections and extrahepatic biliary atresia (16). Our study suggested that the acute cholestasis associated with sepsis has an early onset (within 1 week in the majority) and is transient (peaking by the second week and subsiding in about 1 month). In settings wherein facilities for detailed metabolic workup are not available, and initial workup for metabolic disorders is negative, it may be reasonable to monitor infants with documented septicaemia carefully for a period of 1 to 2 months, especially when there is a trend towards improvement after the second week, before referring for invasive and expensive investigations towards a primary liver disease.
In our study, hepatobiliary dysfunction was seen more commonly in those babies who survived. This was statistically significant (P = 0.001). It may be argued that in some neonates who died early, there was not enough time for the development of any hepatobiliary dysfunction either in the form of cholestatic jaundice or elevated liver enzymes; however, the mean (±SD) age of death in our study subjects was 6.1 ± 4.7 days, whereas in majority of our cases, the onset of jaundice was extremely early (80% by day 3). Thus, it appears unlikely that this significant difference can be explained just on the basis of early death of some neonates before the development of cholestatic jaundice. It has been suggested earlier that accumulation of bilirubin could act as a protective mechanism for the survival of preterm newborns (17). Bilirubin is recognised as an antioxidant of possible physiological importance (18). It can scavenge the chain-carrying peroxyl radical by donating a hydrogen atom attached to the C-10 bridge of the tetrapyrrole molecule to form a carbon-centred radical bilirubin (19,20). Both conjugated and unconjugated bilirubin can serve as antioxidants protecting human low-density lipoprotein from peroxidation (21). It is possible that elevated bilirubin is an attempt by newborn babies to cope with increased exposure to reactive oxidation stress generated as a result of sepsis. The babies who are capable of generating this possibly protective response may have a better prognosis. To the best of our knowledge no direct evidence for this hypothesis, however, is available.
On serial follow-up, the increase in weight was seen at a lower rate in children with hepatobiliary dysfunction, but this was not statistically significant. The gain in height and head circumference was also comparable between the 2 groups. Thus, hepatobiliary abnormalities including cholestasis did not seem to affect the growth of the child during the first 3 months to a significant extent. Although it is expected that children having cholestasis would have impaired growth caused by malabsorption of nutrients, it was not observed in the present study. This could be attributed to the transient nature of abnormalities in our study because the nutritional and developmental effects of chronic cholestasis usually take several months to become clinically evident.
The findings from our study can only be generalised to the neonates having Gram-negative sepsis, especially caused by Klebsiella, in developing countries with a high low birth weight rate. Also, the group of neonates enrolled in our study had severe sepsis as is apparent from the high mortality rate, and the findings may not be extrapolated to low-grade sepsis with negative blood cultures. Our study had the limitation that the detailed investigations for other causes of neonatal cholestasis were not performed in all of the infants, and the contribution of other metabolic and genetic causes could not be conclusively ruled out; however, a high prevalence of cholestasis and enzyme abnormalities, and their reversibility in all the survivors suggested that these dysfunctions seen in our study were most likely to be the result of bacterial sepsis.
In conclusion, hepatobiliary dysfunction in the form of cholestatic jaundice and hepatic enzyme abnormalities is common in neonatal sepsis. The onset of these abnormalities is early in most cases but ultimately resolve within 2 to 3 months after sepsis. The presence of conjugated hyperbilirubinemia in neonatal sepsis may carry a better prognosis in terms of survival and has no effect on growth during early infancy.
1. Bassani DG, Kumar R, Awasthi S, et al. Causes of neonatal and child mortality in India: a nationally representative mortality survey. Lancet
2. Black RE, Cousens S, Johnson HL, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet
3. Shamir R, Maayan-Metzger A, Bujanover Y, et al. Liver enzyme abnormalities in gram-negative bacteremia of premature infants. Pediatr Infect Dis J
4. Tollner U. Early diagnosis of septicemia in the newborn. Clinical studies and sepsis score. Eur J Pediatr
5. Ballard JL, Khoury JC, Wedig K, et al. New Ballard score, expanded to include extremely premature infants. J Pediatr
6. Lwanga SK, Lemeshaw S. Sample Size Determination in Health Studies: A Practical Manual
. Geneva: World Health Organization; 1991.
7. Kumhar GD, Ramachandran VG, Gupta P. Bacteriological analysis of blood culture isolates from neonates in a tertiary care hospital in India. J Health Popul Nut
8. National Neonatology Forum. Neonatal morbidity and mortality: report of the National Neonatal Perinatal Database. Indian Pediatr
9. Kerur BM, Bhat BV, Harish BN, et al. Maternal genital bacteria and surface colonization in early neonatal sepsis. Indian J Pediatr
10. Ghotaslou R, Ghorashi Z, Nahaei MR. Klebsiella pneumoniae
in neonatal sepsis: a 3-year-study in the pediatric hospital of Tabriz, Iran. Jpn J Infect Dis
11. Wong MH, Oelkers P, Craddock AL, et al. Expression cloning and characterization of the hamster ileal sodium-dependant bile acid transporter. J Biol Chem
12. Kenny JF, Mediains DN, Kleyn S, et al. An outbreak of UTI and septicemia due to Escherichia coli
in male infants. J Pediatr
13. Watkins JB, Sunaryo FP, Berezin SH. Hepatic manifestations of congenital and perinatal disease. Clin Perinatol
14. Utili R, Abernathy CO, Zimmerman HJ. Cholestatic effects of Escherichia coli
endotoxin on the isolated perfused rat liver. Gastroenterology
15. Tiker F, Tarcan A, Kilicdag H, et al. Early onset conjugated hyperbilirubinemia in newborn infants. Indian J Pediatr
16. Dick MC, Mowat AP. Hepatitis syndrome in infancy: an epidemiological survey with 10 year follow up. Arch Dis Child
17. Hammerman C, Goldstein R, Kaplan M, et al. Bilirubin in the premature: toxic waste or natural defense? Clin Chem
18. Stocker R, Yamamoto Y, McDonagh AF, et al. Bilirubin is an antioxidant of possible physiological importance. Science
19. Wu TW, Fung KP, Wu J, et al. Antioxidation of human low-density lipoprotein by unconjugated and conjugated bilirubins. Biochem Pharmacol
20. Dennery PA, McDonagh AF, Spitz DR, et al. Hyperbilirubinemia results in reduced oxidative injury in neonatal Gunn rats exposed to hyperoxia. Free Radic Biol Med
21. Tomaro ML, Batlle AM. Bilirubin: its role in cytoprotection against oxidative stress. Int J Biochem Cell Biol