METHODS

A retrospective case note review was performed on all patients found to have hepatic epithelioid granulomas on liver biopsy performed between January 1991 and December 2001 at Glasgow Royal Infirmary. A list of all biopsies reporting granulomas, regardless of coexistent histological hepatitis (that is, “granulomatous hepatitis”), was generated from the pathology department computer database. All biopsies revealing only lipogranulomas (clusters of lipid droplets surrounded by macrophages and lymphocytes) were excluded from our study because these are not related to the usual aetiologies of true epithelioid granulomas.¹¹ All the remaining biopsies were stained for mycobacteria and fungi and reviewed by our pathologist.

Aetiology was determined by results from autoantibodies, immunoglobulins, hepatitis serology, and full drug history. Chest radiography had been performed if sarcoid or TB were deemed likely. Serum angiotensin converting enzyme and calcium measurements were performed to aid with the diagnosis of sarcoidosis. The above results and clinical assessment were recorded in a standard proforma, as were liver function tests at the time of biopsy. Patient outcome was assessed on the basis of the most recent clinic review.

RESULTS

Incidence of hepatic granulomas

Granulomas were detected in 96 patients. Of these, 24 biopsies contained only lipogranulomas and were excluded. The remaining 72 patients with epithelioid granulomas were included in our study. Five case notes could not be traced and four patients did not have full appropriate investigations to determine an aetiology. Of the 63 remaining patients, 47 were female, with a mean age of 49 years (range, 17–81). Figure 1 shows the number of liver biopsies revealing hepatic granulomas and the total number of biopsies performed each year. Over the study period, the percentage of liver biopsies revealing granulomas has been fairly consistent, with a mean of 4.2% out of a total of 1662 liver biopsies performed over the study period.

Diagnoses

Table 1 reveals the various underlying aetiologies expressed both as the number of cases and as a percentage of the total number of liver biopsies revealing epithelioid granulomas.

Fifteen cases were attributed to primary biliary cirrhosis (PBC), and of these four patients had negative antimitochondrial antibodies (AMA) at the time of biopsy. However, each of these four patients had raised IgM, cholestatic liver function derangement, and compatible liver histology, and were therefore thought to have “AMA negative PBC”. Interestingly, none subsequently developed positive AMA during the study period.

Sarcoid accounted for seven of the 63 cases. Liver function tests in all these seven patients were deranged in a predominantly cholestatic pattern.

Seven patients fell into the idiopathic category, six were female and the overall mean age was 52.6 years. This diagnosis was made only after all the investigations described above were negative. In this idiopathic group, liver function tests were deranged in a cholestatic pattern.

Five patients in our series were deemed to have drug induced hepatic granulomas. The granulomas in all six of these biopsies were noted to have an associated eosinophilic infiltrate. The drugs in question were identified as glibenclamide, metronidazole, baclofen, nitrofurantoin, and allopurinol.

Autoimmune hepatitis (AIH)/PBC overlap syndrome was the diagnosis in four cases. The diagnosis in all four was made on the basis of mixed liver function test derangement, positive AMA and antismooth muscle antibodies, and a mixture of hepatitic/cholestatic histopathology.

There were three cases of AIH, a condition not normally associated with hepatic granulomas. In all three cases, other causes of granulomas were excluded and in addition typical autoantibodies on serology and interface hepatitis on histology were found. Of note, AMA antibodies were persistently negative in all cases.

Resolving biliary obstruction of duration more than six months as a result of gallstones accounted for two cases. The histological findings of poorly formed granulomas in close association with necrotic hepatocytes and bile pigment have been described previously. These were accompanied by features of extrahepatic biliary obstruction, including bile stasis, ductal proliferation, and a neutrophil infiltrate.

One case was ascribed to each of the following aetiologies; non-Hodgkin lymphoma, polymyalgia rheumatica, juvenile chronic arthritis, graft versus host disease, jejuno–ileal bypass surgery, and intravesical BCG. In one other case, it was felt impossible to determine the cause of a patient’s hepatic granulomas because both HCV and non-Hodgkin lymphoma were present at the time of biopsy, and both (either alone or in combination) could have been responsible.

Outcomes of patients with hepatic granulomas

Table 2 shows the outcome of patients diagnosed with hepatic granulomas over a follow up period ranging from 4.4 to seven years.

Of particular interest to us was the outcome of the idiopathic cohort. From this group, one patient died from a cerebrovascular accident, one was lost to follow up, and the remaining five are well and still attending our liver clinic, with a mean 6.2 years of follow up.

DISCUSSION

Our study confirms the broad range of underlying aetiologies of hepatic granulomas. In addition, HCV was seen to be an important cause of hepatic granulomas in our patient cohort. Despite this, a large cohort with presumed idiopathic hepatic granulomas persisted.

Chronic HCV has only been recognised as a cause of hepatic granulomas since the availability of routine serological testing in 1991, and was therefore not described as a cause of hepatic granulomas in the previous large UK series.^4,7,8,10 Hepatic granulomas have been described in up to 10% of liver biopsies from HCV infected patients,^12–15 although the clinical relevance of this finding is unknown.^16,17

Thus, we expected that there would have been a smaller proportion of “idiopathic” cases in our series. Interestingly, this was not the case because we identified a subgroup of patients (11.1%) in our study in whom an underlying aetiology could not be found, despite extensive investigations. The most likely explanation for this is that those patients with HCV, most of whom were infected through intravenous drug abuse, were rarely investigated for abnormal liver function tests before 1990 (when widespread HCV serology testing became available), and certainly rarely underwent liver biopsy before the HCV era. Thus, the HCV granuloma group probably represents a new cohort of previously undiagnosed hepatic granulomas.

In our experience, those patients with so called idiopathic hepatic granulomas tend to be middle aged women with cholestatic liver function tests who have an excellent outcome. The benign prognosis of this patient group has been noted previously.^7,18 Earlier publications on this subset of patients^19,20 have highlighted the systemic upset, with fever, myalgia, weight loss, and weakness that may accompany the deranged liver function. In our “idiopathic” cohort, no patient exhibited such a clinical picture and, in addition, the fact that nine of the 10 patients in our study¹⁹ were male suggests this may represent a distinct disorder. Indeed, it is possible that these cases may represent underlying polymyalgia rheumatica, a recognised cause of hepatic granulomas and systemic upset. This is supported by the reported good clinical/laboratory response of such patients to corticosteroids.¹⁹

PBC remains the most common diagnosis in patients with hepatic granulomas and is consistent with previous granuloma series from the UK.⁸ However, it should be noted that PBC may be underestimated as a cause in our series because we do not routinely biopsy all patients with presumed PBC, particularly if they are middle aged women with positive AMA and cholestatic liver function test derangement.

“Those patients with so called idiopathic hepatic granulomas tend to be middle aged women with cholestatic liver function tests who have an excellent outcome”

In sarcoidosis, the histological findings are not diagnostic, and a wide range of pathological features is recognised.^21–23 Thus, a combination of multisystem involvement, chest radiograph abnormalities, raised serum calcium, and raised angiotensin converting enzyme values were used to make the diagnosis.

In our drug induced group, all of the patients had a favourable outcome and none developed acute liver failure. Many other medications have been reported to be associated with hepatic granulomas.²⁴

It has long been recognised that Hodgkin disease is associated with the development of hepatic granuloma, and a study in 1974²⁵ revealed that 8% of staging liver biopsies in 103 patients with Hodgkin disease contained granulomas. An earlier study had suggested a figure of 17%.²⁶ The presence of hepatic granulomas does not appear to be related to disease outcome or prognosis in Hodgkin lymphoma. Cases of hepatic granulomas associated with non-Hodgkin lymphoma,²⁷ polmyalgia rheumatica,^28,29 and jejuno–ileal bypass surgery^30,31 have previously been described.

In our series, similar to that reported by McGluggage and Sloan,⁸ infections accounted for only a small proportion of cases. This is in sharp contrast to other parts of the world. A series from Turkey³² revealed that infectious agents including TB, hydatid disease, brucella, and typhoid fever accounted for over half of their hepatic granuloma cases. In addition, in a series from Saudi Arabia,⁵ schistosomiasis accounted for over half of the cases, with TB the next most common cause. We noted one case that was caused by systemic BCG infection after an intravesicular BCG injection for a bladder tumour, which has been described previously.^33–36 We may have expected some opportunistic infections (for example, Mycobacterium avium intracellulare, toxoplasmosis) to account for some cases in human immunodeficiency virus infected patients, but these patients are exclusively cared for in a dedicated infectious diseases unit at another hospital in our city.

Perhaps surprisingly, in none of our cases of hepatic granuloma was inflammatory bowel disease the underlying aetiology, despite our hospital having a cohort of approximately 1000 patients with inflammatory bowel disease. For many years, it has been recognised that hepatic granulomas are associated with this disease, and are particularly prevalent in patients with Crohn’s disease.^37,38 The series of McGluggage and Sloan,⁸ carried out in Northern Ireland during the 1980s, revealed three (1.8%) cases of underlying Crohn’s disease. It is possible that endoscopic retrograde cholangiopancreatography or magnetic resonance cholangiopancreatography are being performed in preference to liver biopsy for the investigation of abnormal liver function tests in our patient cohort.

As previously mentioned, steroids have been advocated in the treatment of idiopathic granulomatous hepatitis; some believe their benefit depends on the fact that a proportion of these are cases of undiagnosed sarcoid. However, this course of action risks exacerbating underlying TB, and an earlier paper⁴ described a patient who died from miliary TB after empirical steroids in this setting. Therefore, caution must be exercised, particularly in parts of the world with a high incidence of TB, where empirical antituberculous treatment should be considered as a first line treatment. Of our three cases of TB, one showed central caseation and none had acid fast bacilli revealed on Ziehl-Neelsen staining. This is consistent with earlier studies,^1,4 highlighting the poor diagnostic yield for TB on hepatic histology, and underlines the need to consider this diagnosis.

In conclusion, our study confirms that hepatic granulomas have a wide aetiology. HCV is emerging as an important causal agent in the West. Finally, despite extensive investigations, 10–15% of our patients had no aetiology identified and were thus labelled “idiopathic”. These patients were typically middle aged women with cholestatic liver function tests and had an excellent prognosis.