Drug-Induced Acute Liver Failure in Children and Adults

Results of a Single-Centre Study of 128 Patients

Harshad Devarbhavi; Mallikarjun Patil; Vishnu V. Reddy; Rajvir Singh; Tarun Joseph; Deepak Ganga


Liver International. 2018;38(7):1322-1329. 

In This Article


Demography and Clinical Features

During the study period from 1997 till April 2017, 905 patients were adjudicated as DILI by RUCAM, of whom 128 cases met diagnostic criteria for DIALF. DIALF accounted for 14% of DILI cases. The mean age was 38 years and 67 (53%) were females. Skin rashes, eosinophilia with or without lymphadenopathy, a surrogate marker of immunoallergy was seen in 27/128 (21.3%) patients and was predominantly seen in the non–antituberculosis group (42% vs 13% antituberculosis group, P = .001). The mean duration of treatment was 70 days. The baseline demographical and laboratory characteristics are summarized in Table 1. Fourteen of the last 105 patients (13.3%) whose information was available were diabetic. Seventy–four patients had grade I and II encephalopathy and 50 patients had grade III and IV encephalopathy of whom 44 (59.4%) and 40 (80%) died respectively. Coma grade in 4 patients was not available.

Twenty–one of the 128 patients were children (<18 years) constituting 16.4% of DIALF. Here too, girls outnumbered boys (52.4: 47.6). The baseline characteristics in children are depicted in Table 2 and contrasted with patients >18 years.


Only 44 patients (34%) recovered spontaneously and 84 (66%) died. The characteristics between survivors and non–survivors are shown in Table 3. Among children, 13 (62%) died. Patients who died were more likely to be women, older and with clinically detectable ascites (ascites in non–survivors 58% vs survivors 29%, P = .002). No relationship was observed with AST, ALT, ALP and platelets and mortality. However, serum protein and albumin was lower; bilirubin (total and direct) and INR were significantly elevated in non–survivors. GGT was elevated in survivors than non–survivors (190 vs 59, P = .001).

Two patients each with acetaminophen and ferrous sulphate overdose had 100% mortality, while mortality in anti–TB DIALF was 79%.

Drugs Implicated in DIALF

Combination antituberculosis drugs (isoniazid and rifampicin with or without pyrazinamide) were the leading cause (n = 92, 72.4%), followed by anti–epileptic agents (n = 12, 17.3%) and dapsone (n = 7, 5.5%). Dose–dependent toxicity (predictable) was seen in 4 patients, 2 each with acetaminophen and ferrous sulphate ingestion. Overall, only 18 different classes of drugs caused DIALF, with only 2 from the complementary and alternative medicines (CAM) group. (Table 4) Since combination antituberculosis drugs constituted a substantial proportion of DIALF, we compared characteristics between ALF caused by antituberculosis drugs vs non–antituberculosis drugs which is depicted in Table 5. Ascites and hyperbilirubinemia were significantly associated with antituberculosis DIALF. Skin rash and increased GGT were more often seen in the non–ATT group which comprised mostly of anti–epileptic drugs and dapsone.

APAP hepatotoxicity leading to ALF was seen in 2 patients, both ending in fatality. One was a 16–year–old boy with history of previous drug abuse. APAP toxicity was associated with concomitant dextropropoxyphene intake. Post–mortem liver biopsy showed necrosis with marked sinusoidal congestion. Another case was a 16–week pregnant lady who ingested APAP together with aspirin. Post–mortem liver biopsy showed extensive massive hepatic necrosis. Details of these 2 patients are provided in Table 6.


The overall mean MELD score was 31 ± 11.7. The MELD score between antituberculosis DIALF group and non–antituberculosis DIALF group was 31 and 28 respectively (P = .18). Fifty–four patients fulfilled non–acetaminophen KCC of ≥3.

Multivariate Analysis for Mortality Risk Factors

Variables that were significant were entered into a multivariate regression model. The results are depicted in Table 7. Only total protein and INR were independent predictors of mortality. Receiver operating characteristic (ROC) curve analysis was carried out to compare MELD score, King's college criteria score and ALFSG index with regard to mortality. ROC curve for MELD and ALFSG index was 0.76, while for KCC it was 0.51 (Table 8). The sensitivity and specificity for MELD score was 72% and 74%, respectively, and for KCC, was 41% and 51% respectively. The best cut–off value for MELD score was ≥28.5.