Drug-induced immune-mediated liver injury is. Drug-Induced Autoimmune Hepatitis: Clinical Characteristics and Prognosis

Transcription

1 Drug-Induced Autoimmune Hepatitis: Clinical Characteristics and Prognosis Einar Björnsson, 1,2 Jayant Talwalkar, 2 Sombat Treeprasertsuk, 2 Patrick S. Kamath, 2 Naoki Takahashi, 3 Schuyler Sanderson, 4 Matthias Neuhauser, 2 and Keith Lindor 2 Drug-induced autoimmune hepatitis (DIAIH) has been reported to be caused by several drugs. There is a lack of data comparing these patients with other patients with autoimmune hepatitis (AIH). A search was performed using the Mayo Clinic diagnostic medical index for AIH patients and DIAIH patients identified over 10 years. Individuals with overlap syndromes and decompensated liver disease were excluded. Overall, 261 patients (204 females, median age 52) were identified, and 24 (9.2%) were DIAIH cases with a median age of 53 (interquartile range, 24-61). Two drugs, nitrofurantoin (n 5 11) and minocycline (n 5 11), were the main causes. A similar proportion of DIAIH patients had positive antinuclear antibodies (83% versus 70%) and smooth muscle antibodies (50% versus 45%) as compared with AIH patients. Histological grade and stage were similar in patients with DIAIH versus AIH; however, none of the DIAIH patients had cirrhosis at baseline; this was present in 20% of matched AIH cases. Liver imaging was normal in all minocycline cases. Eight of 11 (73%) nitrofurantoin patients had abnormalities on hepatic imaging (mainly liver atrophy), a finding seen in only 8 of 33 (24%) of a random sample of the rest of the AIH group (P ). Corticosteroid responsiveness was similar in DIAIH and the AIH patients. Discontinuation of immunosuppression was tried and successful in 14 DIAIH cases, with no relapses (0%), whereas 65% of the AIH patients had a relapse after discontinuation of immunosuppression (P < ). Conclusion: A significant proportion of patients with AIH have drug-induced AIH, mainly because of nitrofurantoin and minocycline. These two groups have similar clinical and histological patterns. However, DIAIH patients do not seem to require long-term immunosuppressive therapy. (HEPATOLOGY 2010;51: ) Drug-induced immune-mediated liver injury is an adverse immune response against proteins within the liver that can lead to a syndrome of autoimmune hepatitis (AIH). 1,2 Reactive metabolites created through hepatic metabolism of some drugs have been shown to bind to cellular proteins such as cytochrome P450. These can then be recognized by the immune system as neoantigens. 3,4 The underlying Abbreviations: AIH, autoimmune hepatitis; DIAIH, drug-induced autoimmune hepatitis; IgG, immunoglobulin G. From the 1 Department of Internal Medicine, Section of Gastroenterology and Hepatology, Landspitali University Hospital, Reykjavik, Iceland; 2 Division of Gastroenterology and Hepatology, Mayo Clinic Rochester, Rochester, MN; 3 Department of Radiology; and 4 Department of Laboratory Medicine and Pathology, Mayo Clinic Rochester, Rochester, MN. Received October 2, 2009; accepted January 11, Address reprint requests to: Einar Björnsson, Landspitali University Hospital, Hringbraut, 101 Reykjavik, Iceland. einarsb@landspitali.is; fax: Copyright VC 2010 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience ( DOI /hep Potential conflict of interest: Nothing to report. mechanisms have been elucidated for some drugs able to induce AIH but not currently in use, such as dihydralazine 5,6 and tienilic acid. 7 Among drugs still widely used, drug-induced AIH (DIAIH) has been well documented for nitrofurantoin, 8,9 which is widely prescribed for urinary tract infections, and minocycline, a treatment of acne. 10,11 However, available data on drug-induced AIH consist mainly of case reports and a few very small case series Autoimmune hepatitis induced by nitrofurantoin was reported in a series of five patients from the 1970s in the United States 12 and six patients from the Netherlands from the 1980s. 9 These small case series had very limited follow-up, and the long-term prognosis of patients with nitrofurantoin-induced AIH remains uncertain. Minocycline-induced hepatitis is associated with the appearance of antinuclear antibodies, and smooth-muscle antibodies, elevated gamma globulin levels, and histological features identical to those observed in AIH. 11,14-16 Information about need for long-term 2040

2 HEPATOLOGY, Vol. 51, No. 6, 2010 BJÖRNSSON ET AL immunosuppression in these patients is unclear, and none of these reports have described long-term consequences for liver damage. We aimed to assess the proportion of drug-induced AIH among consecutive well-characterized patients with AIH. Furthermore, we sought to compare the clinical, biochemical, and histological characteristics of these two groups of AIH patients. Lastly, we wanted to investigate the prognosis in terms of liver-related mortality and compare the results of drug withdrawal in these patients. Patients and Methods We performed a search in the diagnostic medical index at the Mayo Clinic for the diagnosis AIH in the period 1997 to The search is based on the presence of AIH in the text, and the search will retrieve all patients that have AIH mentioned in the medical charts, such as the primary diagnosis, differential diagnosis, family history of AIH, and liver transplantation for AIH. By this method we identified 1536 patients. Only patients with available clinical and biochemical components of the new simplified criteria at baseline were included. 17 Thus, information on the presence of autoantibodies, immunoglobulin G (IgG) or gammaglobulins, viral serologies, and a liver biopsy were obtained before initiation of immunosuppressive therapy. Patients seen at the Mayo Clinic only for a second opinion, but who were diagnosed earlier and who had already started treatment, were excluded. Also, patients undergoing transplants for AIH and patients with decompensated liver disease at presentation were excluded as well as pediatric cases (younger than 16 years of age). The reason for excluding patients with liver failure or those who required transplantation was because the diagnosis of AIH is more uncertain in these conditions. Some features of AIH such as autoantibodies can be present in patients with liver failure and decompensated liver disease, probably secondary to the chronic liver injury. Furthermore, most patients with decompensated liver disease had been started on treatment for AIH elsewhere, and therefore there was often a lack of important biochemical parameters such as gammaglobulins or IgG and autoantibodies, making a diagnostic score almost impossible. Furthermore, patients with a diagnosis of primary biliary cirrhosis and primary sclerosing cholangitis were excluded, as were patients with clinical suspicion of overlap syndromes (AIH/primary biliary cirrhosis and AIH/primary sclerosing cholangitis). A retrospective review was performed on patients fulfilling the inclusion criteria. The following variables were obtained by the chart review: age, sex, date of liver biopsy, titers of antinuclear antibodies, smooth muscle antibodies, liver kidney microsomal, antimitochondrial antibodies, antinuclear cytoplasmic antibodies, IgG, gammaglobulins, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, total bilirubin, albumin, and international normalized ratio at baseline. Furthermore, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, total bilirubin, albumin, international normalized ratio, IgG, gamma globulins at 1 to 2 weeks, 2 months, 6 months, 1 year, and at last follow-up after start of immunosuppressive treatment were recorded. The presence of a suspicion of drug etiology in triggering the AIH was recorded. The liver biopsy results were analyzed, and histology compared between the DIAH patients and age (65 years of age at the diagnosis of AIH) and sex-matched patients (n ¼ 24) randomly chosen from the rest of the AIH patients. Sex was balanced, and no significant age difference was found at diagnosis. Furthermore, patients with nitrofurantoininduced and minocycline-induced AIH were compared. All biopsy materials were reviewed by a single liver pathologist (S.O.S.), blinded to the clinical context of the biopsy as well as the patient s outcomes. The overall inflammatory activity (grade) of the whole specimen, portal, and interface inflammation as well as fibrosis (stage) were recorded from the available materials on a scale from 0 to 4 and analyzed according to Batts and Ludwig. 18 Interface hepatitis was graded as none, minimal, mild, moderate, and severe interface hepatitis and fibrosis stage as no fibrosis, to portal, periportal, bridging, and cirrhosis. Additionally, perivenular (zone 3) necrosis and confluent necrosis were evaluated in the biopsy materials review. The new simplified score was calculated. Histological features were considered typical, compatible, or atypical according to Hennes et al. 17 The biopsy was considered typical if the biopsy demonstrated interface hepatitis with a lymphoplasmacytic infiltrate extending from the portal areas into the lobular parenchyma with associated rosette formation. The biopsy was considered compatible if the biopsy revealed features of chronic hepatitis without all of the typical features listed. Additionally, the designation of atypical was applied if the biopsy demonstrated distinct features of different diagnosis. In medical records with lack of information about the IgG levels, we used the gammaglobulin level for calculation.

3 2042 BJÖRNSSON ET AL. HEPATOLOGY, June 2010 Because obvious imaging abnormalities were seen in most of the nitrofurantoin cases (none in the minocycline cases), we compared the appearance of the liver on imaging between these patients and other AIH patients. We matched three AIH patients for sex and age (65 years of age) with each nitrofurantoin patient (33 versus 11) for this purpose and analyzed results of imaging between the two groups. The type of immunosuppressive treatment and its duration was recorded. Information on whether immunosuppressive therapy was discontinued during followup and the results of the discontinuation were obtained. Discontinuation was considered successful if no relapse was observed biochemically or histologically in patients with at least 12 months of follow-up. Biochemical remission was defined as alanine aminotransferase and aspartate aminotransferase values that were less than 1.5 times the upper limit of normal. The date of last follow-up was recorded, and the duration of follow-up was calculated. Complications of liver disease, clinical cirrhosis, ascites, esophageal or gastric varices, need for liver transplantation, and death were obtained. Response to therapy at 1 to 2 weeks and 2, 6, and 12 months as well as at last follow-up was determined. Statistical Analysis. Continuous variables are presented as medians and interquartile range. Dichotomous variables were compared using the Fischer exact test, and the Mann-Whitney test was used for continuous variables. All tests were two-tailed and conducted at a 5% level of significance. Results General Characteristics of the Patients. A total of 261 well-characterized AIH cases were identified with the available clinical, laboratory, and histological data required for diagnosis according to the new simplified criteria. Overall, 560 cases were excluded because of an earlier diagnosis of AIH outside the Mayo Clinic, and many of these patients were on treatment when first seen at the Mayo Clinic. Decompensated liver disease at presentation was found in 125 patients, and 70 had undergone transplantation for AIH, 57 had not undergone a liver biopsy, 39 did not have available data on autoantibodies or gamma globulins, three had only family history of AIH, and 256 did not have AIH as their final diagnosis. Other cases excluded were for suspicion of overlap syndromes with primary biliary cirrhosis and primary sclerosing cholangitis (n ¼ 97) and pediatric cases (n ¼ 68). Drugs were suspected to have induced the AIH in 24 of 261 (9.2%) cases: minocycline (n ¼ 11), nitrofurantoin (n ¼ 11), and cephalexin and Prometrium, in one case each. These were suspected on clinical grounds because of current use of these drugs at the time of diagnosis. None of the patients continued with the drugs past the date of presentation, and none were rechallenged. The demographic and biochemical data in the study cohort of the 24 DIAIH patients and the 237 other AIH patients is demonstrated in Table 1. The median follow-up of the DIAIH patients was 36 months (13-77); and in the other AIH patients, 36 months (15-79) (P ¼ NS). The DIAIH patients were referral patients from states other than Minnesota in 19 of 24 (79%) and in 160 of 237 (68%) in the other AIH patients (P ¼ NS). A similar proportion of patients had antinuclear antibodies and smooth muscle antibody seropositivity among the two groups (Table 1). Trial of discontinuation of immunosuppression was only tried in 14 of 24 (58%) of the DIAIH patients. A significantly higher proportion of patients with DIAIH were able to discontinue their immunosuppressive therapy compared with other AIH patients (100% versus 35%) (P < ) (Table 1) and showed no relapse at 36 (12-58) months in the DIAIH patients. In general, liver tests at presentation were higher and jaundice more common in the DIAIH group, but the differences were not significantly different (Table 1). Patients with nitrofurantoin-induced and minocyclineinduced AIH are shown in Table 2. Histology. A very similar proportion of the DIAIH and the rest of the AIH group had typical and compatible histology according to the histological characterization presented by Hennes et al. 17 (Table 3). None of the DIAIH patients had atypical histology, and only one patient from the rest of the AIH group had atypical histology. This patient fulfilled diagnostic criteria for AIH, although histology was similar to histology observed in primary biliary cirrhosis. The severity of inflammation and fibrosis was not significantly different between the two groups (Table 3). There was a tendency toward a higher stage score in the AIH than in the DIAIH group, but the difference was not significant (Table 3). However, cirrhosis was not present in any of the DIAH patients, whereas cirrhosis was found among 5 of 24 (20.8%) in the rest of the AIH patients (P ¼ 0.049). When divided into none and mild versus moderate and severe activity in terms of grade, portal inflammatory

4 HEPATOLOGY, Vol. 51, No. 6, 2010 BJÖRNSSON ET AL Table 1. Comparison of the Demographics, Seropositivity, AIH Score, Histology, Treatment, and Liver Tests at Presentation in Patients with Autoimmune Hepatitis (AIH) and Those with Drug-Induced Liver Injury (DIAIH) AIH Patients (n 5 237) DIAIH (n 5 24) P Value Age 52 (37-62) 53 (24-61) NS Sex, females (%) 184 (78%) 20 (92%) NS ANA positive (%) 165/237 (70%) 20 (83%) NS SMA positive (%) 106/237 (45%) 12/24 (50%) NS Both ANA and SMA (%) 69/237 (29%) 9/24 (38%) NS Seronegative (%) 29/237 (12%) 1/24 (4%) NS Simplified AIH score: Probable or definite (%) 181/237 (76%) 19/21 (90.5%) NS Immunosuppressive therapy (%) 222/237 (94%) 21/24 (88%) NS Steroids and azathioprine (%) 191/222 (86%) 12/21 (57%) Steroids alone (%) 31/222 (14%) 9 (43%) Trial of discontinuation successful (%) 18/52 (35%) 14/14 (100%) < AST (<48 U/L) 392 ( ) 679 ( ) NS ALT (<55 U/L) 480 ( ) 728 ( ) NS ALP (115 U/L) 241 ( ) 376 ( ) TB (<1.0 mg/dl) 2.0 ( ) 4.0 ( ) NS Albumin (>3.5 g/dl) 3.4 ( ) 3.1 ( ) NS INR (<1.2) 1.1 ( ) 1.1 ( ) NS IgG (<1500 g/dl) 2020 ( ) 1905 ( ) NS Gamma globulins (<1.7 g/dl) 2.5 ( ) 2.55 ( ) NS Jaundice at presentation 110/237 (46%) 12/24 (50%) NS activity, interface, and portal hepatitis in DIAH versus AIH, no significant differences were found (data not shown). A total of six patients with DIAIH had a follow-up liver biopsy, and all had either very mild or no inflammatory activity in the follow-up biopsy. Comparison between the histological features in the Table 2. Comparison of the Demographics, Seropositivity, AIH Score, Histology, Treatment and Liver Tests at Presentation in Patients with Nitrofurantoin-Induced and Minocycline-Induced AIH Nitrofurantoin AIH (n 5 11) Minocycline AIH (n 5 11) P value Age 61 (54-66) 24 (18-38) Sex, females (%) 11 (100%) 10 (91%) NS Duration of therapy (months) 24 (8-36) 12 (9-36) NS ANA positive (%) 8/11 (73%) 10/11 (91%) NS SMA positive (%) 6/11 (55%) 5/11 (45%) NS Both ANA and SMA (%) 4/11 (36%) 4/11 (36%) NS Seronegative (%) 1/11 (9%) 0/11 (0%) NS Simplified AIH score: Probable or definite (%) 7/9 (78%) 10/10 (100%) NS Compatible histology (%) 6/11 (55%) 2/11 (18%) NS Typical histology (%) 5/11 (45%) 9/11 (82%) NS Immunosuppressive therapy (%) 11/11 (100%) 9/11 (82%) NS Steroids and azathioprine (%) 3/11 (27%) 8/9 (89%) Steroids alone (%) 8/11 (73%) 1/11 (11%) Trial of discontinuation successful (%) 9/9 (100%) 5/5 (100%) NS AST (<48 U/L) 827 ( ) 329 ( ) ALT (<55 U/L) 778 ( ) 380 ( ) NS ALP (115 U/L) 426 ( ) 188 ( ) NS TB (<1.0 mg/dl) 7 ( ) 1.5 ( ) NS Albumin (>3.5 g/dl) 2.9 ( ) 3.4 ( ) NS INR (<1.2) 1.2 ( ) 1.1 ( ) NS IgG (<1500 g/dl) 1835 ( ) 1905 ( ) NS Gamma globulins (<1.7 g/dl) 2.5 ( ) 2.8 ( ) NS Jaundice at presentation 8/11 (73%) 3/11 (27%) NS In two nitrofurantoin cases and one minocycline case, gamma globulin levels were not available, which made it possible to calculate the AIH score in 19 patients.

5 2044 BJÖRNSSON ET AL. HEPATOLOGY, June 2010 Table 3. Comparison Between DILI/AIH and AIH Alone DIAIH AIH P Value Grade (Batts and Ludwig) 3 (2-3) 3 (2-3) NS Portal inflammation 2 (2-3) 2 (2-3) NS Lymphoplasmacytic (absent/present) 19/23 (83%) 22/23 (96%) NS Interface hepatitis 2.5 ( ) 2.0 ( ) NS Lobular hepatitis 2.0 ( ) 2.0 ( ) NS Zone 3 necrosis 15/23 (65%) 12/22 (55%) NS Confluent necrosis 7/23 (30.4%) 2/22 (9%) NS Rosette formation 7/22 (31.8%) 5/22 (22.7%) NS Stage 0 (0-2) 1 (0-3) 0.06 Compatible 8/24 (33%) 8/24 (33%) NS Typical 16/24 (66%) 15/24 (63%) Atypical 0 1/24 (4%) Results are expressed as medians and interquartile range. Grade, portal inflammation, interface, and lobular hepatitis as well as stage are scored as none, mild, moderate, and severe. Lymphoplasmacytic occurrence, zone 3 necrosis, confluent necrosis, and rosette formation are classified as present or absent. nitrofurantoin-induced and minocycline-induced AIH patients showed similar grade and stage in these patients (Table 4). In general the necroinflammatory activity was found to be higher in the nitrofurantoininduced AIH than in the AIH induced by minocycline (Table 4). Imaging. Changes on radiological images were evident in 8 of 11 (73%) of the nitrofurantoin-induced AIH cases, whereas this was not observed in any of the minocycline patients (P ¼ ). Furthermore, other AIH patients had abnormalities on imaging in only 24% (8/33) of cases versus 73% of the nitrofurantoin patients (P ¼ ). In most of the AIH patients with abnormalities on imaging, the liver showed mild atrophy (n ¼ 4), clear signs of cirrhosis (atrophy and signs of portal hypertension such as ascites) (n ¼ 2), or coarsening of the liver architecture compatible with chronic liver disease (n ¼ 2). In several nitrofurantoin patients, the appearance of the liver was considered cirrhotic on imaging but cirrhosis was not shown to be present histologically in any of the nitrofurantoin (or the minocycline) patients. Two patients had left lobe liver atrophy, two patients had right lobe liver atrophy, and two patients had diffuse general liver atrophy. In five patients, computed tomography or magnetic resonance imaging showed confluent area of abnormality with distortion of surrounding liver parenchyma (Figs. 1-3). The confluent abnormal area showed retention of contrast on delayed-phase images, consistent with confluent fibrosis or massive fibrotic bands. In some cases when images were obtained at presentation, the confluent area showed fairly intense enhancement during the portal phase of enhancement. This early enhancement is somewhat unusual for typical confluent fibrosis, but may have been due to active or subacute phase of the disease. The appearance of confluent fibrosis or massive fibrotic bands (Figs. 1-3) was only seen in the nitrofurantoin patients and in none of the AIH patients. One of the patients with right liver atrophy also had a large mass in the right lobe and hypertrophy of the left liver lobe. Another patient had heterogeneous echotexture, with a subtle 3-cm mass in the right lobe that on the original imaging was similar to focal nodular hyperplasia but was clinically considered secondary to the nitrofurantoin-induced liver damage, and the patient was in clinical and biochemical remission at follow-up. Individual Patients. The clinical characteristics of the individual patients with nitrofurantoin-induced and minocycline-induced AIH are shown in Tables 5 and 6. A total of 8 of 11 (73%) nitrofurantoin cases had jaundice at presentation (Table 5), whereas only 3 of 11 (27%) of the minocycline cases had jaundice at presentation (Table 6). All patients with nitrofurantoin-induced AIH were treated with immunosuppression (Table 5), whereas two minocycline patients were not treated (Table 6). In one case attributable to rapidly improving jaundice, liver tests before therapy were instituted, and not in another case attributable to mild liver enzyme abnormalities (Table 6). All cases with all parameters available to make a score for the likelihood of AIH received at least a probable score; however, in three cases, information about gamma globulins was not available for the AIH score. Outcome. At the end of follow-up, among those who had at least 6 months follow-up at the Mayo Table 4. Comparison Between Nitrofurantoin and Minocyclin-Induced AIH Nitrofurantoin Minocycline P Value Grade (Batts and Ludwig) 3 (3-3.8) 2 (2-3) Portal inflammation 3 (2-3) 2 (2-3) 0.08 Lymphoplasmacytic (absent/present) 10/10 (100%) 9/11 (82%) NS Interface hepatitis 3.0 ( ) 2.0 ( ) Lobular hepatitis 3.0 ( ) 1.0 ( ) Zone 3 necrosis 9/10 (90%) 6/11 (55.5%) NS Confluent necrosis 6/10 (30.4%) 1/11 (9%) Rosette formation 6/9 (66%) 1/11 (9%) Stage 0 (0-2) 0 (0-1.8) NS Compatible 6/11 (55%) 2/11 (18%) NS Typical 5/11 (45%) 9/11 (82%) Atypical 0 0 Results are expressed as medians and interquartile range. Grade, portal inflammation, interface, and lobular hepatitis as well as stage are scored as none, mild, moderate, and severe. Lymphoplasmacytic occurrence, zone 3 necrosis, confluent necrosis, and rosette formation are classified as present or absent.

6 HEPATOLOGY, Vol. 51, No. 6, 2010 BJÖRNSSON ET AL Fig. 1. Confluent necrosis in both right and left liver lobe. clinic, 22 of 23 (96%) of the DIAIH patients were in biochemical remission, as were 159 of 177 (90%) of the other AIH patients (NS). None of the DIAIH patients developed cirrhosis during follow-up, and all were alive at last follow-up, whereas 18 of 257 (7%) AIH patients had developed cirrhosis clinically (P ¼ NS). Overall, five of these patients had decompensated liver cirrhosis with mild ascites in most, and 13 patients had nonbleeding esophageal varices. Among Fig. 3. Confluent fibrosis centrally with general liver atrophy. the seven patients who died during the study period, only two deaths were liver related. One patient died of fulminant hepatic failure on the liver transplantation list a few months after presentation, and one other patient died in chronic liver failure and with a disseminated breast cancer. Others died of stroke, heart failure, and colon cancer, and two had unknown causes of death. At the end of follow-up, one of the abovementioned patients with cirrhosis had undergone liver transplantation, two were listed, and one was evaluated for transplantation. Fig. 2. Confluent fibrosis in both right and left lobe in another patient but with another distribution. Discussion The results of the current study suggest that among consecutive patients with well-characterized AIH, drug-induced AIH makes up a significant proportion, approximately 9%, of AIH cases. More than 90% of these DIAIH cases were associated with two drugs, nitrofurantoin and minocycline. Overall, the histological features of DIAIH seem to be identical to those of AIH. Severe imaging abnormalities with liver lobe and general liver atrophy were also commonly observed in nitrofurantoin but not in minocycline cases, indicating postnecrotic scarring. None of the patients with DIAIH developed cirrhosis during follow-up, and these patients seem to have a generally favorable prognosis. All patients with DIAIH in whom discontinuation of immunosuppressive therapy was tried could withdraw drug therapy without a relapse, whereas only approximately one third of the

7 2046 BJÖRNSSON ET AL. HEPATOLOGY, June 2010 Table 5. Clinical Characteristic of the Individual Patients with Nitrofurantoin-Induced AIH Subject, Duration of Therapy (months) ANA SMA IgG Gamma-globulins Score AST ALT ALP TB Therapy Response to Therapy F55 (27) neg 1: Steroids Fair* F55 (36) neg 1: Steroids Good F53 (6) 1:320 1: Steroids Good F64 (7) neg 1: Steroids Good F55 (4) 5.6 neg Steroids Good F75 (60) 5.3 1: Steroids Good F52 (36) >12 1: Steroids Good F67 (36) 1.8 1: SterþAza Good F67 (12) >12 neg SterþAza Good F66 (36) 1:80 1: SterþAza Good F64 (24) 6.3 1: SterþAza Good F, female; M, male. *Information not available on discontinuation of immunosuppression. Discontinuation of immunosuppression successful. Discontinuation of immunosuppression not tried. other AIH patients did not experience a relapse after withdrawal of immunosuppression. Taken together, our results suggest that DIAIH is a distinct phenomenon, and a trial of withdrawal of immunosuppression should be undertaken in these patients. Previous literature on DIAIH consist only of case reports and small case series. Our series of 24 wellcharacterized patients is by far the largest study on the clinical aspects of drug-induced AIH. No previous study has been able to assess the proportion of DIAIH out of AIH cases in general. We found that 9.2% of AIH patients were by definition induced by drugs. It is conceivable that this reflects referral bias, but no significant difference was found between the proportion of referral patients among the DIAIH versus the other AIH patients. Nitrofurantoin can lead to a broad spectrum of liver injury with mild liver test abnormalities, acute liver failure with fatal outcome, or need for liver transplantation 9,12,19-21 and also liver cirrhosis. 21 Nitrofurantoin is still widely used in many countries and was recently found to be one of the most common single agents associated with drug-induced liver injury (DILI) in a recent prospective study in the United States. 21 Nitrofurantoin has been documented to induce AIH previously in a number of reports. 8,9,12,13,22-24 All cases in the current series with laboratory parameters available to score the AIH by the new simplified criteria 17 had a score of at least probable. Zone 3 necrosis was observed in our DIAIH cohort and has been described in nitrofurantoin-induced liver injury 12 but has also been reported in AIH without drug involvement. 25,26 It seems that DIAIH caused by other drugs than nitrofurantoin and minocycline is a rare cause of AIH. Interestingly, one of the two other drugs suspected to have caused AIH, cephalexin, was also taken by one patient with minocycline-induced AIH in one series. 10 Cirrhosis was not found to develop in any of the DIAIH cases in the current study, whereas this was found in 20% of the other AIH patients at baseline. Table 6. Clinical Characteristic of the Individual Patients with Minocycline Induced AIH Subject, Duration of Therapy (months) ANA SMA IgG Gamma-globulins Score AST ALT ALP TB Therapy Response to Therapy F24 (36) 1:40 neg SterþAza Good* F39 (-) 2.9 neg SterþAza Good F60 (8) neg 1: Steroids Good F25 (42) 1:320 1: SterþAza Good F17 (-) 1:80 1: SterþAza Good F55 (2) 1:80 1: None - F33 (4) >12 1: SterþAza Good F25 (-) 1.5 1: SterþAza Good* F17 (12) >12 neg SterþAza Good* F18 (36) >12 neg SterþAza Good* M18 (12) 3.6 neg None - F, female; M, male. *Discontinuation of immunosuppression not tried. Discontinuation of immunosuppression successful.

8 HEPATOLOGY, Vol. 51, No. 6, 2010 BJÖRNSSON ET AL Our results are in agreement with Stricker et al., 9 who found no cases of nitrofurantoin-induced cirrhosis among 52 reported cases of suspected liver injury reported to the Netherlands Centre for Monitoring of Adverse Reactions to Drugs. Most of the patients with DIAIH in our study had been treated for a long period with the drug, with a median duration of 24 and 12 months in the nitrofurantoin-induced and minocycline-induced AIH, respectively. This long duration of treatment has been the experience in other series. 9,12 Interestingly, severe abnormalities were seen on imaging in 8 of 11 (73%) of the nitrofurantoin cases, whereas this was not found to occur in the other DIAIH cases. This has not been reported previously, but a recent case report revealed low attenuation patches in the liver parenchyma. 27 This unusual pattern of fibrosis was seen in the nitrofurantoin cases with confluent fibrosis and massive fibrotic bands not seen in other AIH patients. The appearance of this type of fibrotic process on imaging might raise a suspicion of nitrofurantoin-induced liver injury. However, it is not clear whether these changes are specific for nitrofurantoin-induced liver damage. Inadvertent rechallenge of nitrofurantoin has been reported to induce chronic liver injury 28 and severe liver injury 17 years after an initial hepatitis-like illness, 29 which underlines the importance of careful history in assessing chronic test abnormalities and changes seen on imaging. One of the most important findings of the analysis of the DIAIH cases in our series is that corticosteroid therapy could be discontinued without relapse. The need for continuous immunosuppressive therapy has not been analyzed in previous series on DIAIH. 9,10,12,13,15,30 In all cases of the current studies when this was tried, no relapse was found to occur during a median follow-up of 36 months. This argues for induction of AIH by nitrofurantoin and minocycline and not simply unmasking otherwise sporadic cases of AIH. To our knowledge, relapse after discontinuation of immunosuppressive therapy has only been reported in a single case of minocycline-induced AIH previously. 31 Our results indicate that DIAIH patients have a generally favorable prognosis, although our follow-up was rather limited. None of the DIAIH patients had histological cirrhosis at presentation, and none developed cirrhosis clinically during follow-up, and all except one patient (who only had 6 months follow-up) were in biochemical remission and experienced lack of relapse after immunosuppression withdrawal. A recent study analyzing chronic liver injury after a previous episode of severe DILI revealed development of AIH in several patients associated with several different drugs. In these patients, immunosuppression could also be discontinued without a relapse, 32 which is in line with the results of the current series. Our results suggest that in patients with DIAIH a trial of discontinuation of immunosuppression should be undertaken in all patients. References 1. Manns MP, Obermayer-Straub P. Cytochromes P450 and uridine triphosphate-glucuronosyltransferases: model autoantigens to study druginduced, virus-induced, and autoimmune liver disease. HEPATOLOGY 1997;26: Liu ZX, Kaplowitz N. Immune-mediated drug-induced liver disease. Clin Liver Dis 2002;6: Robin MA, Le Roy M, Descatoire V, Berson A, Lebreton FP, Maratrat M, et al. Plasma membrane cytochromes P450 as neoantigens and autoimmune targets in drug-induced hepatitis. J Hepatol 1997;(26 Suppl 1): Beaune PH, Bourdi M. Autoantibodies against cytochromes P-450 in drug-induced autoimmune hepatitis. Ann N Y Acad Sci 1993;685: Siegmund W, Franke G, Biebler KE, Donner I, Kallwellis R, Kairies M, et al. The influence of the acetylator phenotype for the clinical use of dihydralazine. Int J Clin Pharmacol Ther Toxicol 1985;(23 Suppl 1):S74-S Bourdi M, Tinel M, Beaune PH, Pessayre D. Interactions of dihydralazine with cytochromes P4501A: a possible explanation for the appearance of anti-cytochrome P4501A2 autoantibodies. Mol Pharmacol 1994;45: Lecoeur S, Andre C, Beaune PH. Tienilic acid-induced autoimmune hepatitis: anti-liver and-kidney microsomal type 2 autoantibodies recognize a three-site conformational epitope on cytochrome P4502C9. Mol Pharmacol 1996;50: Fagrell B, Strandberg I, Wengle B. A nitrofurantoin-induced disorder simulating chronic active hepatitis: a case report. Acta Med Scand 1976;199: Stricker BH, Blok AP, Claas FH, Van Parys GE, Desmet VJ. Hepatic injury associated with the use of nitrofurans: a clinicopathological study of 52 reported cases. HEPATOLOGY 1988;8: Gough A, Chapman S, Wagstaff K, Emery P, Elias E. Minocycline induced autoimmune hepatitis and systemic lupus erythematosus-like syndrome. BMJ 1996;312: Bhat G, Jordan J Jr, Sokalski S, Bajaj V, Marshall R, Berkelhammer C. Minocycline-induced hepatitis with autoimmune features and neutropenia. J Clin Gastroenterol 1998;27: Sharp JR, Ishak KG, Zimmerman HJ. Chronic active hepatitis and severe hepatic necrosis associated with nitrofurantoin. Ann Intern Med 1980;92: Black M, Rabin L, Schatz N. Nitrofurantoin-induced chronic active hepatitis. Ann Intern Med 1980;92: Herzog D, Rasquin-Weber AM, Debray D, Alvarez F. Subfulminant hepatic failure in autoimmune hepatitis type 1: an unusual form of presentation. J Hepatol 1997;27: Goldstein NS, Bayati N, Silverman AL, Gordon SC. Minocycline as a cause of drug-induced autoimmune hepatitis: report of four cases and comparison with autoimmune hepatitis. Am J Clin Pathol 2000;114: Lawrenson RA, Seaman HE, Sundstrom A, Williams TJ, Farmer RD. Liver damage associated with minocycline use in acne: a systematic review of the published literature and pharmacovigilance data. Drug Saf 2000;23: Hennes EM, Zeniya M, Czaja AJ, Pares A, Dalekos GN, Krawitt EL, et al. Simplified criteria for the diagnosis of autoimmune hepatitis. HEPATOLOGY 2008;48:

9 2048 BJÖRNSSON ET AL. HEPATOLOGY, June Batts KP, Ludwig J. Chronic hepatitis: an update on terminology and reporting. Am J Surg Pathol 1995;19: Bhagwat AG, Warren RE. Hepatic reaction to nitrofurantoin. Lancet 1969;2: Iwarson S, Lindberg J, Lundin P. Nitrofurantoin-induced chronic liver disease: clinical course and outcome of five cases. Scand J Gastroenterol 1979;14: ChalasaniN,FontanaRJ,BonkovskyHL,WatkinsPB,DavernT,SerranoJ, et al. Causes, clinical features, and outcomes from a prospective study of druginduced liver injury in the United States. Gastroenterology 2008;135: , 1934 e1-e Lindberg J, Lindholm A, Lundin P, Iwarson S. Trigger factors and HL- A antigens in chronic active hepatitis. Br Med J 1975;4: Stromberg A, Wengle B. Letter: chronic active hepatitis induced by nitrofurantoin. Br Med J 1976;2: Hatoff DE, Cohen M, Schweigert BF, Talbert WM. Nitrofurantoin: another cause of drug-induced chronic active hepatitis? A report of a patient with HLA-B8 antigen. Am J Med 1979;67: Pratt DS, Fawaz KA, Rabson A, Dellelis R, Kaplan MM. A novel histological lesion in glucocorticoid-responsive chronic hepatitis. Gastroenterology 1997;113: Kessler WR, Cummings OW, Eckert G, Chalasani N, Lumeng L, Kwo PY. Fulminant hepatic failure as the initial presentation of acute autoimmune hepatitis. Clin Gastroenterol Hepatol 2004;2: Koulaouzidis A, Bhat S, Moschos J, Tan C, De Ramon A. Nitrofurantoin-induced lung- and hepatotoxicity. Ann Hepatol 2007;6: Bjornsson E, Kalaitzakis E, Av Klinteberg V, Alem N, Olsson R. Longterm follow-up of patients with mild to moderate drug-induced liver injury. Aliment Pharmacol Ther 2007;26: Paiva LA, Wright PJ, Koff RS. Long-term hepatic memory for hypersensitivity to nitrofurantoin. Am J Gastroenterol 1992;87: Teitelbaum JE, Perez-Atayde AR, Cohen M, Bousvaros A, Jonas MM. Minocycline-related autoimmune hepatitis: case series and literature review. Arch Pediatr Adolesc Med 1998;152: Ramakrishna J, Johnson AR, Banner BF. Long-term minocycline use for acne in healthy adolescents can cause severe autoimmune hepatitis. J Clin Gastroenterol 2009;43: Bjornsson E, Davidsdottir L. The long-term follow-up after idiosyncratic drug-induced liver injury with jaundice. J Hepatol 2009;50: