To fully understand an experimental drug's potential for liver toxicity, clinical trials should be better designed to reflect real-world populations, including patients with liver disease, but neither FDA nor experts from industry and academia at a recent workshop on the topic were able to provide many answers on how to do so safely.

When FDA released its first draft guidance on drug-induced liver injury last October, it noted that "patients are sometimes excluded from clinical trials because of baseline liver test abnormalities or a history of liver disease, but there is not a well-established reason to do this." Instead, the guidance recommends that drug developers enroll patients with pre-existing liver disease in at least Phase III clinical trials, especially if the drug might eventually be used in such patients (¹ Also see "Liver Toxicity Guidance Urges Sponsors To Include Risky Patients In Trials" - Pink Sheet, 29 October, 2007., p. 29).

Industry appears to be open to the idea of enrolling patients with liver disease in clinical studies, but almost all of the drug companies that submitted comments on the draft guidance asked the agency to clarify how those patients should be included safely (² Also see "Pharma Seeks More Guidance On Drug-Induced Liver Injury" - Pink Sheet, 18 February, 2008., p. 21).

Pfizer, for example, pointed out that "the draft guidance does not provide clear guidance on the circumstances under which this can be done safely or appropriately, nor does it provide clear guidance on how DILI might be detected in the presence of non-drug induced liver impairment."

In part to address this issue, FDA partnered with the Pharmaceutical Research and Manufacturers of America and the American Association for the Study of Liver Disease to hold a two-day conference on "Detecting and Investigating Drug-Induced Liver Injury During Clinical Trials."

The March 26-27 meeting drew more than 250 people to hear 24 speakers tackle the challenges of DILI. But by the end of the conference, little headway had been made on how to include patients with underlying liver disease safely in clinical trials.

Still A Struggle For FDA, Woodcock Says

"Findings that can signal potential injury can disrupt and often stop the development programs of otherwise promising agents because we don't know what these signals mean," CDER Director Janet Woodcock said in opening the conference.

"We have been struggling with thinking about ways to better understand and predict drug-induced liver injury. It is a problem that has been a matter of academic inquiry for a long time, but we still don't have answers."

For the past three to four years, FDA has been urging members of academia, drug industry and government to come together to try to develop new ways of addressing liver toxicity as part of the Critical Path initiative, Woodcock said.

Answers will not be discovered by any single company, but instead "they are going to require population-based investigations that need more fire power put behind them than any single entity can muster," she said. "We need consortia, and we need to come together in new ways to address and solve these problems."

John Senior, associate director for science, Office of Surveillance and Epidemiology at CDER, agreed. "Such a study, a big study, is going to be very important, and it's going to cost a lot of money," he said. "We need to do something more. We need to find the Holy Grail. We need to chase it. We need to find a predictive biomarker. Not just a marker that's a more sensitive test of injury that has already occurred, but a predictive test of what will occur if you take a drug."

Quest For The Holy Biomarker

Diagnosis of DILI is challenging - and made all the more complicated by idiosyncratic patient characteristics, the difficulty of distinguishing drug-induced liver injury from that from most known liver diseases, and the difficulty of clearly establishing drug involvement in hepatocellular injury in clinical trials.

Finding a signal or biomarker for DILI would make enrolling patients with underlying liver disease safer and easier for pharmaceutical companies. But finding such a marker, experts said, will be extremely challenging.

"With rare exception, there is no biomarker that exist for the definitive diagnosis of drug-induced liver injury," Leonard Seef, NIH, said. "And because DILI can mimic virtually any form of liver disease, all of these causes of liver disease have to be excluded so that this is a diagnosis of exclusion."

However, establishing causality is still "very problematic," he said. "The possibility of re-challenging the person with the implicated drug may provide the best support for establishing a diagnosis for DILI." (see chart: "³ How To Better Understand DILI")

But re-challenging patients with a drug that previously caused elevated ALT or bilirubin levels comes with a whole host of safety and ethical problems. Proper ways to re-challenge a patient who has signs of DILI was a major topic of the meeting.

(Re)challenging Ethics

"Re-challenge is a setting in which we get into the ethical battle and tension between minimizing risk to the enrolled subjects and trying to ascertain as much information as possible about the drug in the study," FDA bioethicist Sarah Goldkind explained. "Re-challenge poses additional research-related risks to the enrolled subjects, but may provide important safety information of benefit to the public."

One possible way to do this, she said, is to wait for the biochemical levels to reset before re-challenging, and conduct frequent monitoring.

Seef termed this a "deliberate re-challenge." He also defined a "continuous re-challenge," when a patient with slight signs of DILI continues taking the drug, maybe at a lower dose, to see if the body adapts, and an "inadvertent re-challenge" - a possibly dangerous situation that occurs when a doctor unknowingly administers a drug to a patient for which he or she has already shown signs of DILI.

"As a clinician, I find this the most difficult part of the entire issue," Seef said. "If we're called to see a patient, the patient has been put on a drug, and they have an abnormal ALT, but bilirubin is normal, how do we know for certain that this is not something that will eventually progress when, maybe even months later, the bilirubin goes up? Or whether the patient will adapt and this is the adaptation phenomenon? I don't know if there is in fact a definitive means of distinguishing between these two possibilities."

But re-challenging could help distinguish between whether DILI is being caused by the drug, or the patient taking the drug. "Is it the drug, or is it the person getting the drug [that causes DILI]," Senior asked. "I think we need a shift in how we look at things; a new paradigm in toxicology. It's not the drug, the dose, or the regiment. It's the patient that gets it."

To test this, Senior proposed a large prospective trial of isoniazid, which is used around the world to treat tuberculosis, but carries with it the risk of DILI. Isoniazid is an interesting case, he said, because while it comes with a danger of liver toxicity, people are able to manage it.

Senior said isoniazid would not be approved today if it had to go through a re-approval process, but that FDA, industry and academia could learn a lot from studying how patients react to it.

"I think we need a large prospective study of isoniazid," Senior said. "We have learned a lot from isoniazid, but I think we could learn more. We know it can kill people, but it doesn't have to. This is a very valuable drug, and just because there's one or two or three cases of Hy's Law, it doesn't mean we should throw the baby out with the bath water."

Hy's law is a current indicator of potential liver toxicity, and is determined by aminotransferase and bilirubin levels.

Setting Up For Failure?

The current clinical trial paradigm excluding patients with liver disease or abnormalities raises dueling problems: it either pushes the problems to the post-market setting, or if problems are detected pre-market, it derails clinical development.

William Lee, a professor specializing in liver disease at the University of Texas, Southwestern Medical Center, raised the point that clinical trials are too homogenous. "Typically in the pre-approval setting we try not to get anybody with co-morbidities, knowing full well that once the drug is approved, all these people are in play," he said. "You're not going to see uncommon side effects unless you have a range of patients."

"The etiquette of clinical trials and the way we're designing them now, is to keep it as clean as you can, don't bring up anything controversial, do your AST monitoring, but don't look for funny side effects, keep it simple, but certainly don't use a controversial population," he said.

"So if we set up a trial to not include these kinds of individuals, then we're setting it up for success to get the drug approved, and questions raised post-approval."

As an example of a drug where pivotal clinical trials were designed with real-world patients and liver toxicity was detected, Lee cited the failure of AstraZeneca's Exanta (ximelagatran). "One of the reasons this drug didn't make it is because it did include a real-world population, including 80-year-olds with atrial fibrillation," he said. AstraZeneca ultimately withdrew the application for the antithrombotic because of liver safety concerns (⁴ Also see "AstraZeneca Rethinking Blockbuster Model After Exanta “Not Approvable”" - Pink Sheet, 18 October, 2004., p. 27).

"But let's put it in perspective," Lee said. "It would be hugely successful as a replacement for [Bristol-Myers Squibb's] Coumadin [(warfarin)], which is one of the worst drugs we use" because of side effects and difficult management. Exanta "might have had its place if we knew how to handle it better, and we have to recognize that while it's high risk, it's also high benefit."

There are also many examples of drugs where liver toxicity problems have been detected after approval. "So my point is that you may have short-term relief in this clean pre-approval process to get your drug approved, and then you pay the price later on by having all these complications show up after the drug has been marketed," Lee said.

"Drug trials are a crap shoot," he admitted. "If we hand-select cases too much, we're avoiding problems later on. I think we need to consider the real-world population that's going to be taking this drug." But, he added, "perhaps what we need to do is be a little bit more liberal in approval, but tighter in giving something like a provisional approval that then leads to a final approval later on when a larger subset of patients has been enrolled."

Temple's Take On The Situation

"There's a principle that we try to encourage and don't always succeed - that drugs are supposed to be tested in the population that will eventually receive them," CDER Office of Medical Policy Director Robert Temple acknowledged.

"The answer to the basic question of should people with underlying liver disease be included in studies has got to be yes," he said. "The idea that everyone with pre-existing liver disease has to be left out of a drug that increases a few transamanases doesn't seem to be ethically required."

While early phase trials should not necessarily include people with variable disease, if there is no early evidence of DILI in the first few trials of a drug, "it's hard to think why there's any reason not to include people with underlying liver disease," he said.

Even if the early trials do present signals of DILI that aren't "really scary," Temple said, "then you might still want to know what happens in people with underlying liver disease, but you'll have to be very careful." He recommended thorough monitoring.

The Last Word From FDA

In the absence of clear answers to the issues of DILI, FDA is postponing making final decisions and still looking for input.

The agency has reopened the comment period until June 30, 2008, due to the discussion generated from the conference. The agency will review the comments over the summer and expects the guidance to be finalized sometime in 2009.

- Lauren Smith (lauren.smith@elsevier.com)

Drug-Induced Liver Injury Workshop Offers More Questions Than Answers

More from Archive

More from Pink Sheet