Troglitazone entered the U.S. market in 1997 as the first PPAR-γ agonist available to treat Type 2 diabetes. Within less than 1 year of product launch, the company began receiving reports of liver failure associated with troglitazone treatment. At first, it seemed possible that these events could be unrelated to troglitazone therapy and instead reflect the relatively high prevalence of liver disease in diabetics, particularly nonalcoholic fatty liver disease (NAFLD) and hepatitis C.1 Review of severe liver event reports, however, revealed a characteristic pattern, or “signature” not compatible with diabetes associated liver diseases.2 The troglitazone associated injury was generally acute, occurring after weeks or months of treatment. The characteristic injury was hepatocellular in nature with high peak serum aminotransferases. Elevation of serum bilirubin was delayed relative to the aminotranferase elevations, with jaundice typically appearing weeks after the onset. Elevations in serum alkaline phosphatase were generally mild and a late phenomenon. A striking feature of this “signature” was that after stopping treatment with troglitazone, the serum aminotransferases generally continued to rise for many days or weeks. Resolution of liver injury was prolonged, often taking weeks or months.
In a public meeting in March 1999, members of an FDA advisory committee were told that there had been more than 40 reports of acute liver failure in patients receiving troglitazone therapy.3 The committee suggested some restrictions on the use of troglitazone, but felt that troglitazone's benefits outweighed its risks. A year later, when the next in class compounds were demonstrated to be safer, troglitazone was withdrawn from the market.
In the 4 years since the withdrawal of troglitazone, there has been a major increase in interest in drug induced liver disease (DILI). The withdrawal of troglitazone followed on the heels of other FDA actions concerning DILI, including the withdrawal of bromfenac, prescribing restrictions placed on felbamate, pemoline, tolcapone, and trovafloxacin, and a string of physician warnings concerning other medications.2 Troglitazone drew attention to the fact that DILI had become the major single cause for regulatory actions concerning drugs, and to the inability to consistently recognize potential for severe DILI during drug development.4 Within 1 year of the troglitazone withdrawal, an FDA working group consisting of representatives from the Pharmaceutical Research and Manufacturers Association of America (PhaRMA), and the AASLD released three documents addressing hepatoxicity: “Nonclinical Assessment of Potential Hepatotoxicity,”5 “Clinical White Paper,”6 and “Post-marketing Considerations.”7 These documents provided “a framework for discussion” for a public workshop held in Chantilly, Virginia in February 2001.8 The FDA/PhaRMA/AASLD Hepatotoxicity Working Group has had regular meetings since.
The prolonged litigation that has followed the withdrawal of troglitazone has also intensified the interest in DILI research. In addition to amplifying the financial losses and negative publicity that followed the withdrawal, the litigation has created a cohort of plaintiff attorneys well versed on issues regarding DILI, and who have identified expert witnesses with opinions helpful to their clients. There has been concern within the pharmaceutical industry that DILI may become a preferred target for “Big Pharma” litigation in the near future. The troglitazone litigation has therefore contributed to the sense of urgency within industry to improve means of screening out liver liabilities at early phases of drug development. Many companies are hoping to solve the problem with “toxicogenomics.”9 One common theme has been to look at changes in the rodent liver transcriptome, proteome and/or metabolome in response to treatment with drugs known to cause hepatotoxicity in man (like troglitazone) versus drugs without known liver liabilities. From these studies, a predictive set of messenger RNAs, proteins, or metabolites are chosen to incorporate into compound screening. It appears that these genomic changes can significantly precede biochemical or histological evidence of hepatotoxicity and may therefore speed up preclinical safety screening. However, it is not yet clear whether this approach will improve prediction for human hepatotoxic potential.
The troglitazone litigation is also beginning to have an impact on the field of hepatology outside the drug development arena. Many of the leading hepatologists and scientists from around the world have been approached to serve as consultants for either the defense or for the plaintiffs. The attorneys generally select the consultants whose opinions most bolster their side's case. These experts are then asked to write detailed and fully referenced reports addressing specific questions. These reports are then exchanged between the sides. New issues are identified and new reports generated, either by the original experts, or by new experts selected by the attorneys. Key aspects of each side's legal strategy emerge from this iterative process.
Many of the questions the consultants were asked to address dealt with the mechanisms underlying troglitazone liver injury. As reviewed by Chojkier in this issue of HEPATOLOGY,10 there is not yet an accepted mechanism for troglitazone liver injury. This uncertainty is probably surprising to judges and juries, and causes confusion that can be exploited by each side. It would be beneficial to the plaintiff case if the mechanisms involved (e.g., covalent binding, glutathione depletion, mitochondrial toxicity) could have been detected in assays available in the early 1990s. It could then be argued that the company was negligent in not identifying the problem early in development. Likewise, mechanisms that supported intrinsic rather than idiosyncratic toxicity helped label troglitazone as a “defective product.” In addition to the legal implications of this designation, the status of “intrinsic toxin” could mean to a jury that all treated patients, not just those with the “signature” events, were potentially “poisoned” and therefore entitled to compensation. Along these lines, some plaintiff experts even argued that because troglitazone could cause cells to undergo apoptosis under certain culture conditions, troglitazone might have caused a “silent” liver injury.11 If a jury could be convinced that this was a possibility, the plaintiff pool could potentially include all of the estimated 2 million patients exposed to troglitazone, even those whose physicians dutifully monitored liver chemistries as recommended.
It is important that the perspectives of both defense and plaintiff experts are now appearing in this and other journals. This allows the larger scientific community to obtain the educational benefits of countless hours of literature research. It also invites opinions from a much broader audience.
In spite of the recent increased interest in DILI, very little progress has been made toward understanding mechanisms underlying idiosyncratic hepatotoxicity. The problem is not just that there are no good animal models of idiosyncratic hepatotoxicity, but that the vast majority of humans are not good models. A key to identifying the mechanisms underlying idiosyncratic hepatotoxicity will therefore be to study individuals who have actually experienced idiosyncratic DILI. This is one of the goals of the National Drug Induced Liver Injury Network (DILIN) recently sponsored as a cooperative agreement (UO1) by The National Institute of Diabetes and Digestive and Kidney Diseases.12, 13 Genomic DNA, serum, and lymphocytes for immortalization will be obtained from patients who have developed idiosyncratic liver injury. These patients will remain in a registry for up to 20 years so they can be offered enrollment in carefully controlled “phenotyping” studies. The creation of this tissue bank and registry should speed progress in DILI research and hopefully prevent another troglitazone.