Initially purported hepatotoxicity by Pelargonium sidoides: the dilemma of pharmacovigilance and proposals for improvement.

Background: Spontaneous reports of herb induced liver injury (HILI) represent a major regulatory issue, and it is in the interest of pharmacovigilance to identify and quantify previously unrecognized adverse reactions and to confirm or refute false positive signals of safety concerns. In a total of 13 spontaneous cases, liver disease has initially been attributed to the use of Pelargonium sidoides (PS), a plant from the South African region. Water/ethanol extracts derived from its roots are available as registered herbal drugs for the treatment of upper respiratory tract infections including acute bronchitis.

Objectives: The present study examines whether and to what extent treatment by PS was associated with the risk of liver injury in these spontaneous cases.

Study Design: Overall, 13 spontaneous cases with primarily suspected PS hepatotoxicity were included in the study. Their data were submitted to a thorough clinical evaluation that included the use of the original and updated scale of CIOMS (Council for International Organizations of Medical Sciences) to assess causality levels. These scales are liver specific, validated for liver toxicity, structured and quantitative.

Results: None of the 13 spontaneous cases of liver disease generated a positive signal of safety concern, since causality for PS could not be established on the basis of the applied CIOMS scales in any of the assessed patients. Confounding variables included comedication with synthetic drugs, major comorbidities, low data quality, lack of appropriate consideration of differential diagnoses, and multiple alternative diagnoses. Among these were liver injury due to comedication, acute pancreatitis and cholangitis, acute cholecystitis, hepatic involvement following lung contusion, hepatitis in the course of virus and bacterial infections, ANA positive autoimmune hepatitis, and other preexisting liver diseases. In the course of the case assessments and under pharmacovigilance aspects, data and interpretation deficits became evident. Possible improvements include appropriate data quality of cases in spontaneous reports, case assessment by skilled specialists, use of a validated liver specific causality assessment method, and inclusion only of confirmed cases into the final regulatory case database.

Conclusions: This study shows lack of hepatotoxicity by PS in all 13 spontaneous cases as opposed to initial judgment that suggested a toxic potential of PS. Major shortcomings emerged in the pharmacovigilance section that require urgent improvements.