Presentations, Causes and Outcomes of Drug-Induced Liver Injury in Egypt

Drug-induced liver injury (DILI) is a frequent cause of liver injury and acute liver failure. We aimed to review all hospitalized DILI cases in a tertiary Egyptian center from January 2015 through January 2016. Cases with elevated alanine aminotransferase more than 3-fold and/or alkaline phosphatase more than 2-fold the upper limit of normal value were prospectively recruited and followed for one year. Drug history, liver biopsy whenever feasible and application of Roussel Uclaf Causality Assessment Method (RUCAM) were the diagnostic prerequisites after exclusion of other etiologies of acute liver injury. In order of frequency, the incriminated drugs were: Diclofenac (31 cases, 41.3%), amoxicillin-clavulanate (14 cases, 18.7%), halothane toxicity (8 cases, 10.7%), ibuprofen (4 cases, 5.3%), Khat (3 cases, 4%), tramadol (3 cases, 4%), Sofosbuvir with ribavirin (2 cases, 2.7%), and acetylsalicylic acid (2 cases, 2.7%) with one offending drug in 93.3% of cases. Forty-four cases (58.7%) were males; while 56 cases (74.7%) had HCV related chronic liver disease. Thirty-two cases (42.7%) presented with pattern of hepatocellular injury, while 23 cases (30.7%) were with cholestasis, and 20 cases (20.7%) with a mixed hepatocellular/cholestatic injury. One case received a transplant (0.75%), 7 cases died (9.3%), 23 cases (30.6%) developed liver decompensation (hepatic encephalopathy and ascites), and 44 cases completely resolved (58.7%). In conclusion, Diclofenac is the commonest offender in DILI occurrence in an Egyptian cohort. Age and prothrombin concentration were the only predictors of unfavorable outcomes of DILI.

As defined by Aithal et al. 5 , DILI was defined by isolated increase of ALT ≥ 5 × upper limit of normal value (ULN) or increase of ALT values ≥ 3 × ULN and concomitant increase of total bilirubin (TB) values ≥ 2 × ULN or increase of alkaline phosphatase (ALP) values ≥ 2 × ULN and concomitant increase of gamma-glutamyl-transferase (γ-GT).
Acute on chronic liver failure (ACLF) was defined as acute hepatic injury manifested as jaundice (with bilirubin ≥5 mg/dl), coagulopathy (with international normalized ratio ≥1.5 or prothrombin activity <40%), and complicated within 4 weeks by ascites and/or hepatic encephalopathy with previously diagnosed or undiagnosed chronic liver disease (CLD) 6 . Evident drug history, along with exclusion of other causes of acute liver injury in patients with HCV related CLD were the criteria of inclusion in this study. No cases with hepatitis B Virus (HBV) monoinfection or co-infection with HCV were included in this study.
All cases were considered as suggestive of DILI according to meticulous drug history and causality assessed using Roussel Uclaf Causality Assessment Method (RUCAM) score 7 . Only cases that achieved values denoting highly probable or probable were included in the study.
The pattern of DILI was defined according to the R value which is defined as the ratio of ALT to ALP as multiples of their upper normal limits. Hepatocellular injury is defined when R value equals more than 5, mixed injury if it is between 2 and 5, while it is cholestatic injury when R value equals less than 2 5 .
Eosinophilia was defined as absolute eosinophil counts exceeding 450 to 550 cells/µL, depending on laboratory standards 8 .
Immune mediated DILI was defined by presence of features of type IV hypersensitivity reactions (fatigue, high-grade fever, arthralgia, skin rash, facial edema, and generalized itching) 9 .
Complete recovery was defined as complete restoration of the basal clinical and biochemical profile within 6 months of injury. Poor (unfavorable) outcome is defined as failure of restoration of the clinical and biochemical profile up to 6 months. Death, urgent liver transplantation, multiorgan failure, decompensation of compensated liver disease, and drug induced chronic liver injury in acute cases all were considered unfavorable DILI outcomes according to Lianos et al. 10 . Chronicity was defined as abnormal liver biochemistry, imaging test or histology one year after DILI recognition according to Medina-Caliz et al. 11 .
Statistical analysis. The data analyzed using SPSS (Statistical Package for Social Science) program for statistical analysis, (version 20; Inc., Chicago. IL). Two types of statistics were done; 1. Descriptive statistics: where quantitative data have been shown as mean, and SD, while qualitative data was expressed as frequency and percent, 2. Analytical statistics: where Chi-square test has been used to measure the association between qualitative variables, and Mann-Whitney test has been used to compare differences between two independent groups regarding variables with quantitative not normally distributed data. Fisher exact test was used for 2 × 2 qualitative variables when more than 25% of the cells have expected count less than 5. The p-value was considered statistically significant when it was <0.05.

Results
Seventy-five DILI cases were enrolled and followed for up to 6 months. Nineteen cases presented with acute DILI, while the remaining 56 cases presented with acute DILI on top of chronic HCV infection.

comparison between acute DiLi and acute DiLi on top of chronic HcV infection revealed.
There was statistically significant difference between the two groups regarding gender with male predominance in acute DILI on top of chronic HCV and female predominance in acute DILI (p = 0.025) ( Table 1). Regarding age: The group of acute on top of chronic HCV were significantly older (46.4 ± 12.3 years old) (p = 0.001) ( Table 1). Laboratory findings at presentation had statistically significant difference between the two groups as regards aspartate aminotransferase (AST), alanine aminotransferase (ALT), prothrombin concentration and platelet counts ( Table 1). Regarding R value; in acute DILI, the injury was mostly hepatocellular with an R value >5 in 89.5% of cases, with only 2 cases (10.5%) presented by mixed hepatocellular/cholestatic injury with an R ratio between 2 and 5. While in cases of acute on top of chronic HCV infection: the injury was cholestatic with an R ratio <2 in 41.1%, mixed hepatocellular/cholestatic type with an R ratio between 2 and 5 in 32.1% with the hepatocellular injury with an R ratio >5 in 26.8% as the least common significantly different (p < 0.001) ( Table 1). The eosinophilia presentation was evident in 31.6% of acute DILI cases while present in only 7.1% in cases of acute on top of chronic hepatitis C, with a statistical significance (p = 0.013) ( Table 1).

Drugs incriminated in DiLi occurrence.
Diclofenac was the most common offending agent in this study responsible in 31 (41.3%) patients, amoxicillin clavulanate was the second cause (18.6%), halothane was the third (10.7%), followed by ibuprofen which was the causative drugs in 4 cases (5.3%), khat and tramadol were the causative drugs in 3 cases (4.0%) for each and sofosbuvir plus ribavirin dual therapy was the causative drugs in 2 (2.7%) cases, the same prevalence for acetyl salicylic acid and chemotherapeutics (indoxan and vincristine) induced DILI (2.7%). Other drugs incriminated with single occurrence in our study (1.3%) were carbimazole, tenoxicam, methimazoles, progesterone therapy, carbamazepine and treatment with triple therapy by sofosbuvir plus interferon and ribavirin ( www.nature.com/scientificreports www.nature.com/scientificreports/ equally cholestatic or mixed in the remaining cases. Ibuprofen cases mainly presented with hepatocellular injury (75.0% of cases). All 3 cases of khat induced liver injury were hepatocellular (Table 2). Time interval between drug intake and symptoms was variable; it was a short interval with halothane, diclofenac and amoxicillin clavulanic acid induced liver injury, with mean ± SD 8.3 ± 3.3, 11.2 ± 5.3, 3.9 ± 1.5 days respectively. Longer intervals were shown with tramadol and khat induced liver injury with mean ± SD 29.6 ± 12.6 and 25.6 ± 4.1 days respectively; with more than 2 months interval in carbamazepine and carbimazole induced liver injury ( Table 2).
outcomes. Complete recovery occurred in 58.7% of cases. Those with unfavorable outcome included 9.3% of cases who died during follow up, 1.3% of them were transferred to urgent liver transplantation and 30.7% of patients had persistent liver injury for more than 6 months. The only patient who was transferred to transplantation due to DILI was 30 years old female who presented with acute hepatocellular injury due to halothane.
There were 13 cases that underwent liver biopsies, in 6 cases focal necrosis with lymphocytes and plasma cells caused by diclofenac in 5 cases and amoxicillin clavulanic acid in 1 case, all of which completely recovered. Auto immune with abundant eosinophilia was founded in 2 khat and 1 diclofenac cases. In two cases with amoxicillin clavulanic DILI: one presented with severe cholestasis, while steatohepatitis was prevalent in the other one. Cholestatic hepatitis was found in one diclofenac case, while centrilobular necrosis was found in one halothane induced case (Table 3 and Fig. 1).
There were seven deceased cases; the mean age was 58.6 ± 3.2 years old, 5 of them were males (71.4%), one (14.3%) presented with hepatocellular injury, 5 (71.4%) with cholestatic injury and one (14.3%) presented by mixed injury, all having chronic liver disease. Four cases were from diclofenac induced liver injury, two patients with chemotherapy (indoxan, vincristine) induced liver injury and one case due to sofosbuvir plus ribavirin induced liver injury (Table 4).
Univariate analysis of factors associated with the outcome revealed the following: there was statistically significant relation between younger age and good outcomes, (p < 0.001) and eosinophilia was present more often in patients who completely recovered (20.5%). Higher prothrombin concentration was associated with good     www.nature.com/scientificreports www.nature.com/scientificreports/ outcome. There was statistically significant relation between higher serum bilirubin at presentation and poor outcome (p = 0.027) ( Table 5).
Multivariate analysis for detection of predictors of complete recovery among studied patients revealed that age less than 37.4 ± 12.5 years (p-value < 0.001 with odds ratio 0.898) along with higher prothrombin concentrations more than 66.1 ± 19.2% (p-value = 0.001 with odds ratio 1.06) were the only predictors of good outcome (Table 6).

Discussion
As most DILI studies are currently available from well developed nations, like Iceland, Spain and the United States, it is an important to characterize the features of DILI in a developing country. In a French study from a northern city, they reported an incidence of approximately 14 per 100,000 DILI patients per year 13 . A Korean study found a rate of 12/100,000 persons/year 14 . Recent study from Iceland estimated an annual incidence rate of DILI of 19.1 cases per 100,000 inhabitants 15 . www.nature.com/scientificreports www.nature.com/scientificreports/ Many prospective data registry studies (the American and Spanish studies), failed to report the relative incidence of DILI 16,17 .
In our study, similarly the true DILI incidence estimates could not accurately be determined. Due to: 1) possibility of undiagnosed cases along with those managed in outpatient clinics 2) those taken care of private health sector, as usual as many other referral centers. Consequently, we believed that our results are most likely underestimation of the true DILI occurrence in Egypt.
DILI prevalence in NLI accounted for 1.38% of all admissions over one year (75 patients from 5452 cases admitted to the Hepatology department from November 2014 to November 2015).
Similar to our study, Devarbhavi et al. had reported DILI contributed to 1.4% of all gastrointestinal admissions and 2.5% of hepatobiliary admissions over one year 18 . In a Thailand study; 1.2% of admissions was due to DILI 19  The higher prevalence of hyperthyroidism and tuberculosis in China had alleged their therapies to be the second and third most reported hepatotoxic drugs 22 . Licata et al. reported that antibiotics were involved for 23.4%, NSAIDs for 35.5%, immunosuppressants for 10.9%, statins for 4.3%, anti-platelets and anti-psychiatric drugs for 7.6% 23 .
In the current study, Diclofenac was the drug leading cause of DILI in this Egyptian cohort (41.3%). Amoxicillin clavulanate was the second most common (18.6%), while Halothane was the third in the list (10.7%).
The higher Egyptian rates of Cesarean section (CS) labors compared to the international reviews might explain the rise in halothane DILI reported in Egyptian females 24 .
In the present study, age appeared to be a risk factor for DILI occurrence with the highest rates in fifties. This might reflect the cumulative risk of drug exposure over years, and or the susceptibility of consumption of drug metabolizing and excreting hepatic enzymes. These results were consistent with many studies identifying age as a risk factor for DILI occurrence 25 .
Contradictory were the results of the large Spanish study based on hepatotoxicity registries with more than 600 patients; where age risk was not substantiated 26 .
Regarding outcomes in this study, 68 patients (91.7%) of our cohorts were survivors, one of them survived following liver transplantation. Complete recovery with return to normal liver functions in acute DILI or to pre-injury status in acute on top of chronic HCV cases was reported in 58.7% of cases. Poor outcome was experienced in 41.3% of cases with persistently deranged liver function tests for more than 6 months. Consequently, one patient (1.3%) had developed acute liver failure requiring urgent liver transplantation and 7 patients (9.3%) died. In concordance, Ostapowicz et al. had mentioned 8% mortality with 2% requiring urgent liver transplantation 27 . Examining drug related outcomes in 64.3% (9 patients out of 14) with amoxicillin clavulanic acid DILI, 62.5% with halothane DILI (5 patients out of 8) and 64.5% (20 patients out of 31 patients) of those with diclofenac related DILI. Khat, tramadol and ibuprofen were also associated with complete recovery while acetylsalicylic acid and chemotherapy with cyclophosphamide and vincristine induced DILI were associated with poor outcome with incomplete recovery to normal. Those drug postulations lacked similarities in literature for the small number of DILI cases caused by these agents. Björnsson and Olsson in 2005 reported that mortality reached 40% with halothane-induced liver injury, whereas all patients with erythromycin-induced liver injury survived 28 . In another study by Björnsson et al. revealed that only 1 out of 35 (3%) cases of erythromycin-induced liver injury were associated with fatal outcome 29 .
In our study, patients with amoxicillin clavulanic acid induced DILI completely recovered from their injury. Mortality in diclofenac induced DILI cases was 12.9% with complete recovery in 64.5%. While the two cases who presented with cholestasis and received chemotherapy (cyclophosphamide and vincristine) had abruptly died. Halothane induced DILI represented 62.5% of cases and they had complete recovery. All khat induced DILI cases presented with hepatocellular injury (3 cases) with complete recovery in 66.7% of khat cases.
The prognosis of patients with cholestatic DILI was mostly presumed to be worse than both hepatocellular and mixed liver injury patterns 6 . According to our cohort, cholestatic injuries represented more than half of poor outcome cases.  www.nature.com/scientificreports www.nature.com/scientificreports/ Björnsson et al. reported a worse prognosis in patients with acute hepatocellular DILI than those with cholestatic or mixed liver injury pattern 29 . They had reported that in the first study in 2007 hepatocellular injury patterns are more common in younger patients and cholestatic patterns in older patients.   Table 5. Univariate analysis of factors affecting outcome. *p-value statistically significant. n = number, % = percentage. n = number, SD = standard deviation. AST = Aspartate Aminotransferase, ALT = Alanine Aminotransferase, n = number, SD = standard deviation, RUCAM = Roussel Uclaf Causality Assessment Method, % = percentage, *Two cases are missing in RUCAM score of bad outcome group of patients (n = 29).  Table 6. Multivariate analysis for detection of predictors of complete recovery. *OR (odds ratio), CI; confidence interval.