Reliability and usefulness of the proposed scale in causality assessment of hepatotoxicity

M Miljkovic1, S Dobric2, V Dragojevic-Simic3. 1ADOC d.o.o. Belgrade, Development and Regulatory Affairs, Serbia, 2Military Medical Academy, Scientific Informations, Serbia, 3Military Medical Academy, Centre for Clinical Pharmacology, Serbia

The general approach in causality assessment of hepatotoxicity is the use of standard CIOMS/RUCAM scale. However, certain deficiencies of this scale have been identified and its reliability and usefulness has never been fully confirmed. The objective was to compare the reliability and usefulness of the standard CIOMS/RUCAM scale and proposed modified CIOMS/RUCAM scale. Compared to the standard scale, the modified scale excludes the risk factors (age ≥55 years, alcohol use and pregnancy), includes laboratory and clinical data, gives lower points for the positive dechallenge and rechallenge and has different scoring system. The two scales (standard and modified) were compared in 80 cases of hepatotoxicity reported during the period 2004-2009 in evaluating their reliability and usefulness, and in 19 cases of unexpected hepatotoxicity in evaluating their reproducibility. Data of the cases were collected by a network of a medical specialist by using a structured reporting form. All cases were ascertained at the time of reporting by the initial clinical assessments of reporting physicians, which served as a control for comparison with the subsequent causality assessments. The performance of the scales was analysed by Kappa weighted (Kw) statistical test. In the 80 cases, the modified scale showed significantly better agreement with the initial clinical assessment (reliability) compared to the standard scale (Kw: 0.92 vs. 0.50). The 2 x 2 contigency tables with cut-off score of 6 points on the standard scale, and of 8 points on the modified scale showed that the modified scale had significantly higher sensitivity and negative predictive value than the standard scale (Se: 97% vs. 62%; NPV: 98% vs. 78%), but similar positive predictive value and specificity (PPV: 96% vs. 95%; Sp: 94% vs. 98%). In the 19 cases of unexpected hepatotoxicity, better agreement between the initial clinical assessments and the standard scale outcomes was achieved for the observer B than for the observer A (Kw: 0.56 vs. 0.72) as well as between the initial clinical assessments and the modified scale outcomes (Kw: 0.76 vs. 0.85). The agreement between the observers was achieved in 17 cases (89.47%, Kw: 0.84) using the standard scale, and in 18 cases (94.74%, Kw: 0.91) using the modified scale. The modified scale showed better agreement between the observers (reproducibility) than the standard scale (Kw: 0.91 vs. 0.84), and also large increase in sensitivity (from 62% to 97%) without significant loss in specificity (from 98% to 94%), accompanied with a very few false-negative and false-positive results. Our results shows that the fine-tuning of the standard CIOMS/RUCAM method enabled more accurate estimates of the likelihood that the drug caused hepatotoxicity, which is particularly important in the evaluation of the patient when decision on the withdrawal of the responsible drug should be without delay, and also in the evaluation of potentially hepatotoxic drugs involved in the process of the signal detection and consequential regulatory measures.

1. Miljkovic MM, Dobric S, Dragojevic-Simic V. (2011) Consistency between causality assessments obtained with two scales and their agreement with clinical judgments in hepatotoxicity. Pharmacoepidemiology and Drug Safety, 20, 272–285.

2. Miljkovic MM, Dobric S, Dragojevic-Simic V. (2011) Accuracy and reproducibility of two scales in causality assessment of unexpected hepatotoxicity. Journal of Clinical Pharmacy and Therapeutics, 36, 1-10. (in press)

3. Lucena MI, Garcia-Cortes M, Cueto R, Lopez-Duran JL, Andrade RJ. (2008) Assessment of drug-induced liver injury in clinical practice. Fundamental and Clinical Pharmacology, 22, 141-158.

4. García-Cortes M, Stephens C, Lucena MI, Fernandez-Castaner A, Andrade RJ; Spanish Group for the Study of Drug-Induced Liver Disease. (2011) Causality assessment methods in drug induced liver injury: Strengths and weaknesses. J Hepatol, 55, 683-691.

5. Rochon J, Protiva P, Seeff LB, Fontana RJ, Liangpunsakul S, Watkins PB, Davern T and McHutchinson JG. (2008) Reliability of the Roussel Uclaf Causality Assessment Method for assessing causality in drug-induced liver