Qualitative and quantitative variations in liver function thresholds among clinical trials in cancer: a need for harmonization.

Abstract

The liver is critically involved in drug metabolism pathways and the potential for hepatic toxicity is significant with specific cancer therapeutics. Variations in the definition of liver function thresholds that may generate heterogeneity of toxicity and efficacy outcomes across therapeutics trials in cancer require assessment. A random sample of therapeutic trials in cancer (n = 500, general category), trials using hepatotoxic drugs (abiraterone, pazopanib: n = 181), trials using drugs metabolized by the liver (doxorubicin, vincristine: n = 606), and therapeutic trials in hepatic dysfunction (n = 49) were each identified on clinicaltrials.gov. Definitions of liver function thresholds and their distribution were collated and categorized in each group. A third of all trials listed on clinicaltrials.gov across the four categories failed to provide an explicit definition of liver function. Among trials with an explicit definition, a combination of bilirubin and transaminase levels was used in 33-64%, whereas a miscellaneous combination of definitions (in the general category consisting of 11 unique liver function parameters creating 17 unique combinations) was used 29-58% of the time across the four categories of studies. The Child-Pugh or National Cancer Institute Organ Dysfunction Working Group (NCI-ODWG) criteria were rarely employed (0-12% studies). Allowance for Gilbert's disease in bilirubin thresholds was identified in only 6-23% studies and for liver metastases in 2-15% of studies. There is a marked heterogeneity in the liver function definitions used across cancer clinical trials even when the potential for drug toxicity and altered drug metabolism is significant. Harmonization of criteria will streamline eligibility and mitigate variations in key outcomes across trials.