Theophylline improves steroid sensitivity in acute alcoholic hepatitis


  • Potential conflict of interest: Nothing to report.


Corticosteroid therapy has shown some benefit in severe acute alcoholic hepatitis (AAH); however, this is limited by uncertainty in patient selection and variable clinical response. Theophylline has been shown to ameliorate impaired steroid sensitivity in chronic obstructive pulmonary disease by facilitating corticosteroid-induced silencing of proinflammatory genes. We aimed to explore the mechanistic basis of the variable response to corticosteroid therapy seen in patients with AAH and to address the extent to which theophylline can improve this response. The ability of dexamethasone to inhibit phytohemagglutinin-induced lymphocyte proliferation was assessed by 3H-thymidine incorporation in 12 severe AAH patients and age-matched and sex-matched controls. Steroid sensitivity was measured in terms of Imax, the maximum inhibition of proliferation. The effect of 10−5 M theophylline and, in survivors, change in Imax during recovery were observed. Lymphocyte steroid sensitivity was found to be significantly reduced in AAH compared with controls (Imax 67[±4.5]% versus 95[±2.3]%, P = 0.0002) and correlated with clinical markers of steroid responsiveness. In survivors, Imax increased in recovery. Theophylline 10−5 M significantly increased lymphocyte steroid sensitivity (Imax 86[±6.6]% versus 67[±5.0]%, P = 0.027). Conclusion: Acute alcoholic hepatitis is associated with significant lymphocyte steroid insensitivity, which improves in recovery and can be ameliorated ex vivo by theophylline. This offers potential to rationalize corticosteroid prescribing in AAH and, furthermore, justifies investigation of this novel role for an existing pharmacological agent in this common and frequently fatal condition. (HEPATOLOGY 2010;)

A significant proportion of deaths from alcoholic liver disease occur in patients with rapidly progressive liver failure caused by acute alcoholic hepatitis (AAH). Patients who die of AAH frequently do so on their first presentation to health services, and so do not have the opportunity to reap the benefits of health education and abstinence. A dearth of applicable and effective therapeutic options means that mortality remains at approximately 30% in the most severely affected patients.1

It is well established that AAH is characterized by a significant inflammatory response with histological inflammatory cell infiltrate and high circulating levels of pro-inflammatory cytokines, including interleukin (IL)-6, IL-8, and tumor necrosis factor alpha, being seen in AAH patients.2-5 The presence of such an inflammatory element to AAH provides an obvious potential approach to therapy through anti-inflammatory drugs such as corticosteroids. Indeed, in meta-analysis corticosteroid therapy has been demonstrated to have a moderate effect on outcome in the group at highest risk of death, reducing 1-month mortality from 35% to 16%.1 However, one in six patients in this group die despite treatment, and the applicability of corticosteroid therapy is limited by concerns about heightened risks of sepsis and gastrointestinal hemorrhage. Steroid responsiveness in AAH is indicated by an early change in the serum bilirubin level (ECBL), with those patients whose bilirubin has not fallen by the seventh day of treatment having a particularly high mortality and gaining no benefit from continuation of therapy.6 Treatment outcomes could be improved by early identification of the 27% of patients who are unlikely to benefit from steroids or by strategies to improve steroid sensitivity in this group.

Impaired steroid responsiveness is not, however, unique to AAH. It has been well characterized in inflammatory skin diseases and ulcerative colitis, where it is evident as a reduced maximum inhibitory effect of dexamethasone on ex vivo phytohemagglutinin-stimulated lymphocyte proliferation.7 Its prevalence is higher in chronic obstructive pulmonary disease (COPD), in which it has been shown to result from smoking-induced abnormalities in transcriptional regulation of pro-inflammatory genes (through reduced capacity of the glucocorticoid receptor to recruit histone deacetylases (HDACs) to actively transcribed pro-inflammatory genes because of the inhibitory effect of smoking-induced oxidative stress8). The traditional bronchodilator theophylline was shown to counteract this effect, exerting an anti-inflammatory effect by improving HDAC recruitment to silence pro-inflammatory genes,9 even at 10−5 M, a concentration at which clinically useful bronchodilator effects do not occur.

In this study, we set out to explore corticosteroid insensitivity in AAH using an ex vivo approach and to explore the hypothesis that in vitro corticosteroid sensitivity can be improved by the actions of theophylline, thereby identifying theophylline as a potential novel therapeutic agent for the treatment of AAH.


AAH, acute alcoholic hepatitis; COPD, chronic obstructive pulmonary disease; ECBL, early change in bilirubin level; HDAC histone deacetylase; Imax, maximum inhibition; IC50, concentration for half maximal inhibition; IL, interleukin.

Patients and Methods

Patients with a primary diagnosis of acute alcoholic hepatitis admitted to Newcastle upon Tyne Hospitals NHS Foundation Trust between October 2006 and March 2008 were invited to participate in the study if they fulfilled the accepted criteria for corticosteroid therapy (Maddrey discriminant function > 32, equating to a 35% 28-day mortality without treatment10). Patients were excluded if they had gastrointestinal bleeding, sepsis, recent surgery, or inflammatory disease, or were already receiving immunomodulatory therapy. Alternative causes of liver dysfunction were excluded by duplex ultrasonography and blood testing for hepatitis A, B, C, and E, ferritin, and HFE genotype, ceruloplasmin, alpha-1-antitrypsin, immunoglobulins, thyroid-stimulating hormone, antinuclear antibody, anti-smooth muscle antibody, antimitochondrial antibody, anti-liver/kidney microsomal antibody, anti-soluble liver antigen antibody, and anti-neutrophil cytoplasmic antibody. Liver biopsy was performed only when there was thought to be a significant probability of it altering clinical management or outcome. In patients who did not undergo biopsy, the presence of coexisting cirrhosis was determined either from diagnostic imaging or from the presence of varices at endoscopy. Age-matched and sex-matched controls were recruited locally. Control exclusion criteria were alcohol intake greater than 40 units per week for men and 30 units per week for women, active infection or inflammatory disease, surgery, trauma or immunomodulatory therapy. Patient and control characteristics are summarized in Table 1. None of the study subjects were receiving corticosteroid therapy at the time of their participation in the study. Patients were subsequently treated with oral prednisolone 40 mg daily for 28 days. Treatment was discontinued early in eight patients, four because of evidence of poor clinical response and four because of concern about sepsis or bleeding. Lymphocyte steroid sensitivity was assessed at recruitment and, for surviving patients who attended follow-up, at 4-weekly intervals for 6 months. The study protocol was prospectively approved by the Local Research Ethics Committee, and informed consent was obtained for each participant.

Table 1. Patient and Control Characteristics
CharacteristicAAH Patients (n = 12)Normal Controls (n = 12)
  1. Values are mean (range).

  2. DF, discriminant function.

Age (years)47 (34-60)51 (34-65)
Weekly ethanol (units)117 (20-560)19 (7-40)
Maddrey DF61 (32-141) 
Biopsy evidence of cirrhosis2 
Circumstantial evidence of cirrhosis only4 

Peripheral blood mononuclear cells were isolated from 25 mL blood by density centrifugation (Lymphoprep, Axis-Shield, Norway, Leucosep tubes, Greiner Bio-One, Germany) and washed three times in phosphate-buffered saline. 4 × 105 cells were added to each of 24 wells of a 96-well tissue culture plate in a final volume of 200 μL Roswell Park Memorial Institute medium with 2 mM L-glutamine and 10% fetal bovine serum (Lonza, UK). Seven triplicates were stimulated with the mitogen phytohemagglutinin (Sigma, St. Louis, MO) at a final concentration of 5 μg/mL. Six of these triplicates were treated with dexamethasone (Sigma) in serial 1:10 dilutions to give final concentrations from 10−6 to 10−11 M. For the initial patient samples, the reaction was duplicated in the presence of 10−5 M theophylline (Sigma). Plates were incubated at 37°C in 5% CO2 for 72 hours. Proliferation was measured by incorporation of tritiated thymidine (Amersham Biosciences, UK). To each well was added 5.29 nmol (37 kBq of activity), and it was incubated for a further 16 hours. Well contents were aspirated onto filter mats (Wallac, Finland) by a cell harvester (Harvester 96 Mach II, Tomtec, Hamden, CT), and radioactivity of the aspirate was determined by beta-counter (MicroBeta TriLux; PerkinElmer, Waltham, MA).

Incorporated 3H-thymidine activity counts were corrected for background and normalized to uninhibited proliferation without dexamethasone. Proliferative response was plotted against dexamethasone concentration, a curve was fitted, and the maximum inhibition of proliferation (Imax) and the concentration of dexamethasone required to achieve 50% of maximal inhibition (IC50) were calculated (Prism 4; Graph Pad, La Jolla CA).7 Mean (±standard error) Imax was compared between groups using the Mann-Witney U test (unpaired observations) and Wilcoxon signed-rank test (paired observations).


Representative control and patient lymphocyte steroid sensitivity curves are shown in Fig. 1. Mean ex vivo steroid sensitivity as determined by Imax was significantly lower in AAH patients than in controls (67% ± 4.5% versus 95% ± 2.3%, P < 0.0005, Fig. 2), suggesting a significantly higher level of corticosteroid insensitivity in patients with AAH than in normal controls. Evidence of established cirrhosis (n = 6) was not associated with significant difference in Imax (63% ± 4.9% versus 72% ± 7.4%, P = 0.31, Supporting Fig. 1). Groups did not differ significantly in mean IC50 (18.0 ± 10.1 versus 9.1 ± 1.7 nmol/L, P = 0.98, data not shown). Within the AAH group, mean Imax was greater in those patients who showed an early change in bilirubin level, an established clinical marker of glucocorticoid responsiveness in AAH, than in those whose bilirubin had not fallen by 7 days (74% ± 5.8% versus 58% ± 4.7%, P < 0.05, Fig. 3), suggesting that corticosteroid insensitivity in the in vitro assay is a predictor of subsequent clinical response to corticosteroid therapy. However, this study did not detect a difference in Imax between those who died during treatment and those who survived (65.7% ± 8.2% versus 68.1% ± 5.6%, P = 0.93, data not shown). Within the AAH group, eight patients survived and seven attended for at least one follow-up appointment. Of these, six reported abstinence from alcohol since admission. Mean Imax in this group was significantly higher during recovery than at presentation (92% ± 4.9% versus 70% ± 7.4%, P = 0.01, Fig. 4). The addition of Theophylline 10−5 M (a concentration previously demonstrated to modulate inflammation in COPD9) to the lymphocyte steroid sensitivity reaction resulted in an increase in Imax measured at presentation (86% ± 6.6% versus 67% ± 5.0%, P < 0.05, Fig. 5).

Figure 1.

Representative patient and control lymphocyte steroid sensitivity curves in peripheral blood mononuclear cells from control subjects (A) and patients with acute alcoholic hepatitis. (B) T lymphocyte proliferation was stimulated with phytohemagglutinin and suppressed with increasing concentrations of dexamethasone. At each concentration, proliferation was measured by incorporation of a radiolabeled nucleotide into new cells and expressed as a percentage of uninhibited proliferation in the absence of dexamethasone. For each subject, the maximum inhibition of proliferation (Imax) and the dexamethasone concentration required to achieve 50% of maximal inhibition (IC50) were calculated. Values are mean ± SEM.

Figure 2.

Lymphocyte steroid sensitivity in acute alcoholic hepatitis patients and normal controls. Steroid sensitivity is measured in terms of Imax, the maximal percentage inhibition of proliferation by dexamethasone. Bars represent mean Imax in each group. AAH, acute alcoholic hepatitis.

Figure 3.

Clinical correlation of ex vivo steroid sensitivity. Relationship between ex vivo lymphocyte steroid sensitivity expressed as Imax and early change in bilirubin level (ECBL), a clinical marker of steroid responsiveness in acute alcoholic hepatitis defined as any fall in bilirubin at day 7. Boxes represent the interquartile range divided at the median, with whiskers covering the whole range of the data.

Figure 4.

Change in lymphocyte steroid sensitivity after recovery from acute alcoholic hepatitis. Representative steroid sensitivity curves for the same patient at presentation and after recovery are illustrated in (A). Steroid sensitivity in terms of Imax is plotted at presentation and after recovery for surviving patients attending follow-up in (B).

Figure 5.

Change in ex vivo lymphocyte steroid sensitivity with addition of theophylline. In acute alcoholic hepatitis patients, lymphocyte steroid sensitivity was determined at presentation in the presence and absence of theophylline 10−5 M. Representative steroid sensitivity curves for a patient's lymphocytes with and without theophylline are illustrated in (A). Steroid sensitivity in terms of Imax is plotted in the absence and presence of 10 M theophylline in (B).

In this study, no relationships were seen between Imax and smoking status (P = 0.32), reported weekly alcohol intake (r2 = 0.007), or disease severity as measured by serum bilirubin, prothrombin time, creatinine, albumin, or discriminant function, Child-Pugh,11 MELD,12 or Glasgow Alcoholic Hepatitis13 scores (r2 < 0.15 for each).


The sensitivity of mitogen-stimulated lymphocyte proliferation to glucocorticoids is a well-established correlate of clinical steroid responsiveness in asthma, rheumatoid arthritis, renal transplantation, and ulcerative colitis.7, 14-16 Lymphocyte proliferative responses to liver-derived antigens are a feature of human alcoholic liver disease,17 and steroid therapy is effective in a proportion of patients with AAH.1. This study investigated ex vivo lymphocyte steroid sensitivity in patients with clinically severe AAH and found it to be significantly lower than in age-matched and sex-matched controls.

As is the case in steroid-resistant ulcerative colitis, the reduced steroid sensitivity in AAH was apparent in differences in the maximum inhibition of proliferation Imax, a measure of steroid efficacy, rather than in the IC50, which measures steroid potency.7 This implies that steroid resistance is unlikely to be overcome simply by increasing the steroid dose. The clinical marker of any fall in bilirubin by day 7 on treatment (ECBL) has been shown to identify patients who are clinically steroid responsive, with a 95% 6-month survival after treatment, as opposed to those with no fall in bilirubin whose 6-month survival is 25% and who do not benefit from continuation of glucocorticoid therapy.6 In our study, mean steroid sensitivity as measured by Imax was significantly lower in those patients who showed no ECBL, suggesting that, as in other diseases, Imax is an ex vivo correlate of clinical steroid responsiveness in AAH. The relatively small number of patients in this study limited its power to detect a relationship between survival and Imax in corticosteroid-treated patients, and it would be valuable to confirm the relationship between in vitro findings and clinical response in a larger sample. Further refinement of this technique might allow ineffective steroid therapy to be stopped at day 4 or earlier, helping to minimize steroid-related morbidity.

The high prevalence of reduced Imax in patients relative to controls strongly suggests that steroid resistance is a feature of AAH. However, there is known to be wide interindividual variation in steroid sensitivity, with studies suggesting that up to 30% of the healthy population would fail to respond to steroid therapy for severe inflammatory conditions.18 This raises the question of whether the patients with AAH are those who are intrinsically “steroid resistant” (suggesting that this intrinsic steroid insensitivity might predispose to severe AAH in heavy drinkers) or whether steroid resistance occurs as a consequence of earlier events in the pathogenesis of AAH. Our observation that in survivors of AAH Imax tends to increase (though not always normalize) during recovery suggests that steroid resistance is at least in part a consequence rather than a cause of the development of AAH. There was no relationship between weekly alcohol intake and Imax in our sample, but further study could investigate whether steroid insensitivity in other inflammatory conditions correlates with alcohol intake, and whether heavy drinkers without liver disease demonstrate relative steroid insensitivity by the same assay.

Glucocorticoid modulation of inflammatory responses results from the ability to control transcription of pro-inflammatory genes via a number of mechanisms, including the recruitment of HDACs to silence actively transcribed pro-inflammatory genes. Failure of HDAC recruitment leads to steroid insensitivity and ongoing inflammation.8 The oxidative stress associated with cigarette smoking has been shown to have this effect in COPD. No association between Imax and smoking history was observed in our patient sample. However, ethanol metabolism is itself a potent source of reactive oxygen species, which may impair the ability of the glucocorticoid receptor to recruit HDACs to “switch off” transcription of proinflammatory genes by the same mechanism. Alternatively, we have recently demonstrated that the increased concentrations of free acetate generated by ethanol metabolism favor increased histone acetylation and enhanced proinflammatory gene expression in a cell-line model of AAH,19 and this mechanism could antagonize the normal glucocorticoid effect.

In our ex vivo assay, the addition of theophylline improved (but not always normalized) steroid sensitivity as measured by Imax. Theophylline has a similar effect on steroid sensitivity in COPD, where it has been shown to improve HDAC recruitment at the same concentration (10−5 M),9 which is below the plasma concentration associated with a clinically significant bronchodilator action. The same mechanism could be responsible for the increased Imax observed in the AAH patient samples. Alternatively, theophylline may be exerting its effect through phosphodiesterase inhibition; a role has recently been demonstrated for PDE4B in the enhanced inflammatory responses observed after chronic ethanol exposure.20 However, significant phosphodiesterase effects are unlikely at the relatively low concentration of theophylline used in this study.9 Of interest is the fact that theophylline is a hepatic metabolite of caffeine and coffee drinking has been shown to protect against advanced alcoholic liver disease in large epidemiological studies.21 Furthermore, pentoxifylline, another methyl xanthine compound similar in structure to both caffeine and theophylline with documented anti-tumor necrosis factor alpha properties, has been shown in trials to improve survival in AAH.22 Pentoxifylline has been found to be ineffective as salvage therapy in AAH patients who have failed to respond to corticosteroids,23 but true combination therapy has not been tested. The mechanism of action of both pentoxifylline and theophylline in this setting remains to be fully investigated, and it is possible that theophylline provides a different or a more potent pharmacological effect.

This study highlights the possibility for existing, inexpensive treatments to be used more effectively in a common and frequently fatal condition. Early and more accurate determination of steroid sensitivity has the potential to rationalize corticosteroid treatment to minimize unwanted effects and maximize clinical benefit. The potential of low-dose theophylline to increase steroid sensitivity leads us to suggest that a clinical trial of its use in acute alcoholic hepatitis could inform and improve future clinical practice.


The authors thank the study participants for their time and commitment and acknowledge the support of their medical and nursing staff.