Unconjugated hyperbilirubinemia is inversely associated with non-alcoholic steatohepatitis (NASH)


Correspondence to:

Dr S. A. Harrison, Division of Gastroenterology and Hepatology, Department of Medicine, Brooke Army Medical Center, Fort Sam Houston, TX 78234, USA.

E-mails: stephen.harrison@amedd.army.mil, stephen.a.harrison@us.army.mil



It has been recognised that unconjugated bilirubin contains hepatic anti-fibrogenic and anti-inflammatory properties and is a potent physiological antioxidant cytoprotectant. We believe that unconjugated hyperbilirubinemia may protect against development of non-alcoholic steatohepatitis (NASH).


This study was conducted to assess the association of serum unconjugated bilirubin levels and histological liver damage in non-alcoholic fatty liver disease (NAFLD).


This was a retrospective analysis involving adult patients from a tertiary medical centre undergoing liver biopsy to evaluate suspected NAFLD or NASH and a control group without NAFLD based on normal liver ultrasound, labs and history. Identification of unconjugated hyperbilirubinemia was based on the presence of predominantly unconjugated bilirubin ≥1.0 mg/dL (17.1 μmol/L) while fasting, in the absence of haemolytic disease or other hepatic function alteration.


Six-hundred and forty-one patients were included. Unconjugated hyperbilirubinemia was inversely associated with NASH (OR 16.1, 95% CI 3.7–70.8 P < 0.001). Of the patients without NAFLD (133 patients), 13 (9.8%) had unconjugated hyperbilirubinemia (range 1.0–1.8, mean 1.4). Of the patients with NAFLD without NASH (285 patients), 32 (11.2%) had unconjugated hyperbilirubinemia (range 1.0–3.0, mean 1.4). Of the patients with NASH (223 patients), three (1.3%) had unconjugated hyperbilirubinemia (1.0, 1.1, 1.4).


Unconjugated hyperbilirubinemia is inversely associated with the histopathological severity of liver damage in non-alcoholic fatty liver disease.


Non-alcoholic fatty liver disease (NAFLD) is the hepatic manifestation of the metabolic syndrome and has emerged in the last two decades as a burgeoning disease, rapidly increasing in prevalence as it is closely tied to the increasing prevalence of obesity, diabetes and the metabolic syndrome.[1] Patients who develop non-alcoholic steatohepatitis (NASH) have significantly worse clinical outcomes with the potential to develop cirrhosis, end stage liver disease and hepatocellular carcinoma.[2, 3]

The pathogenic mechanisms that result in the development of NASH remain incomplete. Dysregulation of fatty acid metabolism, development of hepatic insulin resistance and hyperinsulinemia lead to the development of NAFLD.[4] Fortunately, the majority of patients with NAFLD do not develop NASH.[1, 4] Among patients with NAFLD, up to 1/3 have histological lesions consistent with NASH[1] and approximately 11% of these patients progress to cirrhosis over 15 years.[5] Adaptive responses to sustained lipotoxicity are inadequate in these patients and this may be modulated through environmental and/or genetic processes that ultimately lead to hepatocyte necrosis and inflammation, activation of the fibrogenic cascade and subsequent fibrosis. Oxidative stress and impaired antioxidant defense mechanisms are widely believed to be factors involved in this process.[4, 6-8] Fatty acid accumulation in the liver provides a source of mitochondrial oxidative stress leading to cellular damage, inflammation and progressive fibrosis.[6]

Bilirubin, and particularly unconjugated bilirubin, is known to be a potent physiological antioxidant cytoprotectant.[9] Bilirubin protects against oxidative stress due to inhibitory effects on the activity of NAD(P)H oxidase which may be a basis for increased superoxide production.[10, 11] Furthermore, bilirubin can scavenge peroxyl radicals, hydroxyl radicals, singlet oxygen, reactive nitrogen species[12-14] and reduce the alpha-tocopheroxyl radical promoting recycling of vitamin E.[14] One study showed as little as 10 nmol/L bilirubin has the ability to protect neuronal cultures against 10 000-fold higher concentrations of H2O2.[9] In addition, bilirubin may have anti-inflammatory properties[15, 16] and can act as a major antifibrogenic agent via heme oxygenase-1 (HO-1).[17] There is also strong supporting clinical evidence for the beneficial cytoprotective effects of unconjugated bilirubin as observed in Gilbert's syndrome. It has been shown that unconjugated hyperbilirubinemia is associated with decreased risk of coronary and carotid stenosis,[13, 18, 19] peripheral atherosclerosis,[20, 21] ischaemic heart disease,[22] vascular complications in diabetics[23] and even cancer.[24] Patients with unconjugated hyperbilirubinemia have a significantly lower haemoglobin A1c, low-density lipoprotein cholesterol, total cholesterol, triglyceride and lower prevalence of hypertension.[23]

It is possible that unconjugated hyperbilirubinemia can reduce oxidative stress, decrease inflammation and prevent fibrosis ultimately impacting the development of NASH. We aimed to assess the association between unconjugated bilirubin levels and histological liver damage in patients with NAFLD and NASH.


Study population

Charts were reviewed in 641 patients,18 years and older who were eligible for care at Brooke Army Medical Center (BAMC), San Antonio, TX, comprising three groups: 133 patients without NAFLD (control group), 285 patients with NAFLD but without meeting strict histopathological criteria for NASH and 223 patients with NASH. Patients referred to the Hepatology Clinic to evaluate suspected NAFLD during the time period 1 January 2003 through 1 May 2010 and whose liver biopsy demonstrated NAFLD (not meeting strict criteria for NASH) or NASH were included in the chart review. The control group consisted of patients from a previous study at the same institution[1] who were recruited through study handouts or posters in the Primary Care Clinic waiting area and from patients presenting for colon cancer screening classes in the Gastroenterology Clinic. The control group patients had no prior history of liver disease, normal liver function tests, normal hepatic ultrasound examinations and demographic data similar to the general population. No patients were included if they had a diagnosis of liver disease from any other cause to include viral hepatitis, alcoholic hepatitis, haemochromatosis, alpha-1 antitrypsin deficiency, Wilson's disease; or were taking any medications associated with fatty liver disease (e.g. steroids, tamoxifen, amiodorone). The study protocol was approved by the Institutional review board of BAMC.

Data collection

The following blood test results that were reviewed included: serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin, direct bilirubin, alkaline phosphatase (ALP), total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL) and haemoglobin A1C. Tests to include total bilirubin, direct bilirubin, AST, ALT, ALP, HDL, LDL and total cholesterol were collected. Bilirubin values were gathered when they were collected at the same time as cholesterol lab tests thus it was presumed the values were drawn after at least an 8 h overnight fast. Identification of unconjugated hyperbilirubinemia was based on the presence of predominantly unconjugated bilirubin ≥1.0 mg/dL (17.1 μmol/L) in the absence of other hepatic function alteration.[18, 22, 25] Data points closest in time to the liver biopsy were used for statistical analysis.

Histology assessment

All liver biopsy specimens were prepared and stored at BAMC. Each biopsy was fixed in formalin and embedded in paraffin and Haematoxylin and Eosin (H&E) and Masson Trichrome stains were submitted for analysis. All liver biopsies were evaluated by a single expert hepatopathologist. A diagnosis of NASH was based on the following criteria: steatosis (typically in zone three/centrilobular), necroinflammatory hepatocellular injury, hepatocellular ballooning degeneration with Mallory-Denk bodies.[2, 26]

Hepatic ultrasound

The control group all had normal hepatic ultrasound examinations. One staff radiologist with 13 years of experience in sonography interpreted all the images for the presence or absence of steatosis. A hepatic ultrasound was considered normal if it had homogeneous echotexture with no acoustic attenuation, the diaphragm and portal veins were well visualised, and the echogenicity was either similar or slightly higher than that of renal parenchyma. The ultrasound was considered positive for fatty liver if there was increased echogenicity compared with renal parenchyma, there was attenuation of the ultrasound beam with the diaphragm indistinct and/or the portal vein walls were less visible.

Statistical analyses

Continuous variables were calculated as mean ± s.d. The student's t or Mann–Whitney test was used to evaluate any differences between two groups and the anova or Kruskal–Wallis was used to evaluate differences between groups of three or more. A Fisher exact test was used for assessment of potential differences between descriptive statistics calculated for all groups. spss 16.0 statistical software (SPSS Inc., Chicago, IL, USA) was used for all calculations.


A total of 641 patients were assessed. Demographic data for the three groups are shown in Table 1. The mean age was 52.2 ± 9.3 (range 19–81) and 49.5% were women. Not surprisingly, patients with NAFLD and NASH had a significantly higher body mass index and higher prevalence of diabetes, hypertension, AST, ALT, ALP, LDL cholesterol and HDL cholesterol than controls. Only 533 patients (83%) had haemoglobin A1c levels and higher levels were associated with NAFLD and NASH. Overall, NASH was associated with a lower mean unconjugated bilirubin then the control group and the NAFLD without NASH group (P < 0.001).

Table 1. Baseline characteristics of patients
 Normal liver (N = 133)NAFLD without NASH (N = 285)NASH (N = 223)P-value
  1. LDL, low-density lipoprotein; HDL, high-density lipoprotein; s.d., standard deviation.

Age (years)
Mean ± s.d.54.5 ± 7.251.4 ± 9.951.8 ± 9.30.004
Male gender, N (%)55 (41.4)164 (57.5)105 (47.1)0.004
Body mass index (kg/m2)
Mean ± s.d.28.2 ± 5.0032.3 ± 5.3033.1 ± 5.72 
<25, N (%)36 (27.1)14 (4.9)13 (5.8)<0.001
25–29.9, N (%)56 (42.1)79 (27.7)53 (23.8)0.001
≥30, N (%)41 (30.8)192 (67.4)157 (70.4)<0.001
Diabetes, N (%)14 (10.5)76 (26.7)101 (45.3)<0.001
Average A1C5.9 ± 0.76.1 ± 1.26.4 ± 1.20.003
A1C >6, N (%)19 (29.7)100 (39.1)99 (46.5)0.040
Hypertension, N (%)54 (40.6)183 (64.2)158 (70.9)<0.001
Race, N (%)
White80 (60.2)137 (48.1)108 (48.4)0.050
Hispanic25 (18.8)64 (22.5)50 (22.4)0.662
African American20 (15.0)23 (8.1)12 (5.4)0.007
Other8 (6.0)61 (21.4)53 (23.8)<0.001
Unconjugated bilirubin (mean ± s.d.)0.6 ± 0.30.6 ± 0.30.4 ± 0.2<0.001
Liver enzymes (mean ± s.d.)
Alanine aminotransferase (IU)24.5 ± 9.156.3 ± 44.762.1 ± 49.5<0.001
Aspartate aminotransferase (IU)25.4 ± 7.239.3 ± 21.746.4 ± 29.0<0.001
Alkaline phosphatase75.2 ± 26.389.4 ± 41.190.7 ± 36.7<0.001
Cholesterol (mean ± s.d.)
Total cholesterol186.8 ± 43.9197.6 ± 42.2190.4 ± 42.60.027
LDL cholesterol107.9 ± 35.3117.3 ± 36.0109.9 ± 33.40.009
HDL cholesterol57.2 ± 17.347.6 ± 14.046.9 ± 14.3<0.001
Histology, N (%)
1 (5–33%) 207 (72.6)96 (43.0)<0.001
2 (>33–66%) 60 (21.1)84 (37.7) 
3 (>66%) 18 (6.3)43 (19.3) 
Stage, N (%)
0–1  122 (54.7) 
2–4  101 (45.3) 
Grade, N (%)
1  110 (49.3) 
2  109 (48.9) 
3  4 (1.8) 

The univariate and multivariate comparison of patient variables with and without elevated unconjugated bilirubin are reported in Table 2. Unconjugated hyperbilirubinemia was an independent factor inversely associated with NASH among patients with fatty liver when controlled for other variables. Multivariate logistic regression analysis showed that there were no variables independently linked to unconjugated bilirubinemia which could explain the inverse association with NASH.

Table 2. Univariate and multivariate analysis of unconjugated hyperbilirubinemia
 Normal unconjugated bilirubin (N = 593)Elevated unconjugated bilirubin (N = 48)Univariate analysis P-valueMultivariate analysis OR (95% CI) P-value
  1. HDL, high-density lipoprotein; LDL, low-density lipoprotein; NASH, non-alcoholic steatohepatitis; s.d., standard deviation.

Body mass index (kg/m2)
Mean ± s.d.31.8 ± 5.830.4 ± 3.90.030  
<25, N (%)61 (10.3)2 (4.2)0.171  
25–29.9, N (%)168 (28.3)20 (41.7)0.0513.563 (0.741–17.138)0.113
≥30, N (%)364 (61.4)26 (54.2)0.3252.976 (0.595–14.872)0.184
Diabetes, N (%)179 (30.2)12 (25.0)0.4501.275 (0.568–2.864)0.556
Average A1C6.2 ± 1.26.0 ± 1.30.011  
A1C >6, N (%)207 (41.9)11 (28.2)0.094  
Hypertension, N (%)371 (62.6)24 (50.0)0.0850.795 (0.383–1.651)0.538
Race, N (%)
White299 (50.4)26 (54.2)0.618  
Hispanic126 (21.2)13 (27.1)0.3451.028 (0.473–2.234)0.944
African American53 (8.9)2 (4.2)0.2560.280 (0.057–1.365)0.115
Others115 (19.4)7 (14.6)0.4140.681 (0.262–1.774)0.432
Liver enzymes (mean ± s.d.)
Alanine aminotransferase50.7 ± 41.765.0 ± 67.70.2470.996 (0.985–1.008)0.529
Aspartate aminotransferase38.4 ± 23.145.6 ± 30.90.1471.019 (0.995–1.044)0.118
Alkaline phosphatase86.7 ± 37.689.2 ± 34.30.5241.006 (0.998–1.013)0.142
Cholesterol (mean ± s.d.)
Total cholesterol193.0 ± 43.5192.6 ± 33.80.9500.997 (0.982–1.012)0.658
LDL cholesterol112.6 ± 35.5116.7 ± 29.40.4321.006 (0.988–1.024)0.541
HDL cholesterol49.6 ± 15.646.6 ± 11.80.2011.002 (0.975–1.029)0.912

Of the patients without NAFLD (133 patients), 13 (9.8%) had unconjugated hyperbilirubinemia (range 1.0–1.8 mg/dL, mean 1.4). Of the patients with NAFLD without NASH (285 patients), 32 (11.2%) had unconjugated hyperbilirubinemia (range 1.0–3.0 mg/dL, mean 1.4). Of the patients with NASH (223 patients), 3 (1.3%) had unconjugated hyperbilirubinemia (1.0, 1.1, 1.4 mg/dL) (OR 16.1, 95% CI 3.7–70.8, P < 0.001) (Figure 1). The histological findings of the patients with NAFLD without NASH and NASH with and without unconjugated hyperbilirubinemia are displayed in Table 3.

Figure 1.

The percentage of patients with unconjugated hyperbilirubinemia ≥1.0 mg/dL (17.1 μmol/L) and the associated NAFLD category.

Table 3. Histology of NAFLD
 Normal unconjugated bilirubin (N = 473)Elevated unconjugated bilirubin (N = 35)P-value
  1. NASH, non-alcoholic steatohepatitis.

Histology, N (%)
Steatosis grade
1 (5–33%)281 (59.4)22 (62.9)0.731
2 (>33–66%)136 (28.8)8 (22.9) 
3 (>66%)56 (11.8)5 (14.3) 
 0–1121 (25.6)1 (2.8)0.591
 2–499 (20.9)2 (5.7) 
 1108 (22.8)2 (5.7)0.823
 2108 (22.8)1 (2.8) 
 34 (0.8)0 


To our knowledge, this is the first study to show an inverse relationship between unconjugated hyperbilirubinemia and NASH. Unconjugated hyperbilirubinemia is known to be associated with lower haemoglobin A1c, low-density lipoprotein cholesterol, total cholesterol, triglyceride, lower prevalence of hypertension, lower risk of cardiac disease and atherosclerotic vascular disease.[9, 13, 18-23] It is possible that unconjugated hyperbilirubinemia may contain a protective effect against NASH as well, although the mechanisms in which this occurs remain undefined.

The cause of unconjugated hyperbilirubinemia in our patients is most likely Gilbert's syndrome; however, genetic studies identifying the defect in uridine-5′-diphosphoglucuronosyltransferase 1A1 (UGT1A1) were not performed. We found a prevalence of unconjugated hyperbilirubinemia in the control group at 9.8% and in the isolated steatosis group at 11.2%. This is comparable to the prevalence of Gilbert's syndrome in the general western population at 2–19%.[25] Interestingly, Lin YC, et al.[27] examined UGT1A1 genotypes in 234 obese children in Taiwan and found a lower risk of NAFLD associated with the UGT1A1*6 variant, however, histopathological data to grade and stage extent of NAFLD was not obtained. Consistent with our study, Kumar et al.[28] documented less advanced liver disease on histopathology and/or fibroScan in 204 patients with unconjugated hyperbilirubinemia and NAFLD.

This study was not designed to identify a pathogenic link between unconjugated bilirubin and severity of liver damage. We speculate that the relationship between unconjugated bilirubin and NASH is not an epiphenomenon of bilirubin alterations but involved in the inhibition of pathogenesis of NASH through the potent antioxidant, anti-inflammatory and anti-fibrogenic effect of unconjugated bilirubin. Bilirubin is abundant in blood plasma and is the final product of heme catabolism as heme oxygenase (HO-1) cleaves the heme ring to form water-soluble biliverdin. Biliverdin reductase then reduces biliverdin to bilirubin. Bilirubin oxidised to biliverdin and rapidly reduced back to bilirubin is a redox cycle which possibly amplifies 10,000-fold the physiological oxidative cytoprotection of bilirubin.[9] The anti-inflammatory effects of bilirubin were studied by Keshavan et al.[15] who found that in murine lung parenchyma bilirubin blocks lymphocyte migration decreasing the total leucocyte count and inhibits eosinophil and lymphocyte infiltration. Li et al.[17] discovered major hepatic antifibrogenic properties ascribed to bilirubin and its mediation of HO-1. The HO-1 inhibits proliferation of hepatic myofibroblasts and procollagen I mRNA expression and controls 15-d-PGJ2 (a prostaglandin which elicits apoptotic effects in hepatic myofibroblasts, inhibits collagen and proinflammatory chemokine synthesis and displays growth inhibitory effects).[17]

A strategy has been proposed for treating or preventing atherosclerotic vascular disease by inducing an ‘iatrogenic Gilbert's syndrome’ with therapies that decrease hepatic glucuronidation activity.[13, 29] If a protective association is identified between unconjugated bilirubin and NASH, inducing an ‘iatrogenic Gilbert's syndrome’ may be a consideration for individuals at high risk for NASH. Probenecid, a uricosuric medication, inhibits UDP-glucuronic acid transport from the cytoplasm into the endoplasmic reticulum and is listed as a possible cause of unconjugated hyperbilirubinemia.[30] Interestingly, studies have shown a link between higher uric acid serum levels and NAFLD independent of other metabolic risk factors.[31-33] A recent study found hyperuricemia to be associated with the severity of liver damage in NAFLD independent of other metabolic syndrome factors.[34] This may implicate uric acid as a factor in pathogenesis of metabolic disorders[35] and progressive liver damage in NAFLD[34] and could act as a potential therapeutic target.[29, 34] Other medications which raise serum bilirubin include rifampin[36] which inhibits a membrane transport protein OAP1B1 and valproate[37] which inhibits binding of bilirubin to albumin and increases free unconjugated bilirubin in tissue. Side-effect profiles of rifampin and valproate would likely exclude these medications for use in a large population and valproate has been linked to NAFLD due to increased body weight and insulin resistance.[38] Other proposed therapeutic avenues for utilising the oxidative properties of bilirubin include phycobilins and algal biliverdin.[13, 29, 39, 40] However, the association between unconjugated bilirubin and NASH would need to be validated and a causative effect would need verification with prospective randomised controlled studies prior to use of these proposed therapies. Furthermore, therapies that decrease hepatic glucuronidation would need safety evaluations due to the possible adverse impact on conjugation and clearance of drugs and toxins.

Bilirubin may have a physiological role in modulating tumorigenesis.[41-43] In fact, some epidemiological analyses demonstrate an inverse correlation between serum bilirubin levels and risk of cancer.[23, 44] Elevated serum bilirubin and HO-1 has been shown to be associated with reduced risk for breast cancer[45] and squamous oral cancer.[46] An analysis using data from the Third National Health and Nutrition Examination Survey found that each increase in serum bilirubin of 1 mg/dL was associated with a substantial decrease in prevalence of colorectal cancer (OR = 0.257, 95% CI 0.245–0.260).[24] However, an earlier study found no such association between bilirubin and colon cancer risk[47] and a subsequent study found no association between baseline serum bilirubin and incidence of colorectal cancer.[48]

The main limitation of this study is the retrospective design and further prospective studies are needed. Furthermore, genetic analysis defining elevated unconjugated bilirubin as Gilbert's disease was not done. For future studies this may be important to help define if the association is due to UGT1A1 expression and its contribution to NASH. Also, the differences in metabolic characteristics between the three groups likely represent characteristic mediating factors associated with NAFLD but could raise the possibility of selection bias. The control group consisted of 30% obese patients for which hepatic ultrasound determination of NAFLD is less sensitive and it is possible that some of the patient's with normal ultrasounds may have NASH. Another limitation is the possible narrow external validity of our results applied to different populations and settings. Our study cohort consisted of patients enrolled at a tertiary care centre with government and military health insurance and may not be applicable to the general population.

In conclusion, we found a significantly lower prevalence of unconjugated hyperbilirubinemia in patients with histopathological evidence of steatohepatitis (NASH) compared to NAFLD without NASH and absence of NAFLD. Additional study is needed to confirm this association as therapy that results in higher unconjugated bilirubin may be considered in the future to confer a histopathological benefit in patients at risk for NASH.


The opinion or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the view of the US Department of the Army or the US Department of Defense. Declaration of personal interests: Stephen A. Harrison has served as an ad hoc advisory board member for Amylin Pharmaceuticals and has received research funding from Rottapharm and Mochida Pharmaceuticals. Declaration of funding interests: None.