Sidney M. Wolfe, M.D.
Director, Public Citizen’s Health Research Group
FDA Endocrine and Metabolic Drugs Advisory Committee Meeting on Troglitazone
The last time I appeared before this advisory committee to present information about a diabetes drug was 22 years ago. The drug was phenformin, also said to have unique advantages and whose withdrawal from the market was opposed by many diabetes experts. Like troglitazone, it had a unique type of toxicity, in that case lactic acidosis, fatal in about 50% of those patients who developed this metabolic abnormality from the use of phenformin. After a lawsuit against the FDA by Public Citizen, it was banned as an imminent hazard to the public health in 1977.
Case Reports of Liver Failure or Liver Damage
John Doe was an otherwise healthy 45 year old Washington D.C. policeman who was started on Rezulin in the late summer of 1998, preceded by perfectly normal liver tests. Within a week after he started taking the drug, well before he was due for another liver test, he became seriously ill, was hospitalized at Johns Hopkins with acute liver failure (liver tests of ALT and AST around 3000 – 70 times higher than the upper limits of normal) and had a consultation for a possible liver transplant. After a week in the hospital, he turned out to be one of the few lucky patients with drug-induced liver failure who neither died nor required a liver transplant.
Because of the incompleteness of the spontaneously reported adverse reaction data currently made available to the public by the FDA, it is quite difficult to accurately count even the reported number of deaths which appear to be caused by Rezulin or any other drug. This is because individual adverse reaction reports do not have a unique numerical patient identifier which would allow several reports to be determined to be just follow-ups on the same patient. In our efforts to sort this out, it is possible to overmatch and undercount (assume that what are actually several different patients are the same one ) or under match and over count (assume that what are reports actually referring to the same patient represent different patients).The fault lies with the cumbersome and difficult-to-use format FDA is now using to supply data to the public. Compounding this uncertainty is the widely agreed upon under reporting which may result in as few as 10% or even 1% of cases which actually occur being reported to the FDA.
With these caveats in mind, our estimates of liver deaths from Rezulin up through the beginning of February, 1999 are 43 deaths, including American and Japanese cases. Although Warner Lambert has stated that there have been no new liver deaths since the July relabeling, this statement appears to be false. In addition to the deaths, there are several additional cases of American patients who had liver transplants and survived. This makes an estimated 45 to 50 cases of liver failure apparently caused by Rezulin. An estimated additional 60 or so patients who neither died nor had liver transplants were hospitalized in whom there was evidence of liver damage.
In a memo dated November 26, 1997 – 16 months ago – from FDA’s Mac Lumpkin (Deputy Director of FDA’s drug Center) to Director Dr. Janet Woodcock, Dr. Lumpkin cited 135 cases of “serious hepatotoxicity”, 110 in the U.S. and 25 in Japan which had been reported to the British Government. By June 5, 1998, according to a February 19, 1999 letter published in the Annals of Internal Medicine by FDA physician Dr. Robert Misbin, “FDA had received 560 reports of troglitazone-associated hepatotoxicity”.
Incidence of Liver Toxicity
The best source of information as to how many patients get liver damage from Rezulin comes from the 2510 patients given the drug in randomized placebo-controlled clinical trials before it was approved for marketing. Of these 2510 patients, 48 or 1.9% had abnormal liver tests (ALT) of three times or greater than the upper limit of normal which is 40 (only 0.6% of patients getting a placebo had this much elevation). Treatment was discontinued in 20 of the 48 Rezulin patients because of these abnormal liver tests but in none of the patients getting a placebo. Two patients were jaundiced and two had a liver biopsy. In 18 of the 20 cases, the patients were described as having “liver injury” of a hepatocellular type. Although information about 1.9% of patients with liver test abnormalities is mentioned in the FDA-approved label for the drug, the following sub-group analysis is not: 10 out of the 2510 patients or 1/250 had an ALT of more than 10 times the upper limit of normal (greater than 400) and five of these patients (1/500) had an ALT more than 20 times normal or greater than 800. I would estimate that if even 1/5 of these patients in the real world with the highest enzyme elevations, most of whom are not being adequately monitored instead continue on the drug because it is not known that they are developing a toxic reaction, then 1/2500 (one-fifth of 1/500) might develop hepatic failure, fatal or requiring liver transplants in most cases. The real world of medical practice is unlike the artificially vigilant circumstances of the clinical trial where the drug was immediately discontinued in those with the worst abnormalities (>20 times normal) because of the close monitoring and all recovered.
Although this clear evidence of Rezulin hepatotoxicity known to the company prior to marketing was eventually published (New Eng J Med on March 26, 1998 – vol 338, page 916), it was not discussed at the December, 1996 FDA advisory committee meeting at which there was a unanimous recommendation to approve the drug. In addition to almost no discussion of liver toxicity during that meeting, briefing materials handed out by Warner Lambert to members of the committee were dismissive of liver toxicity claiming, in those studies reviewed (not including all of the 2510 patients given Rezulin before marketing) , that the incidence of elevations of liver tests above the upper limit of normal “was lower for troglitazone-treated patients compared to placebo” (page 80 of handout) and the text concluded that “Elevations meeting threshold criteria for clinically important changes occurred at a similar incidence in both groups [troglitazone and placebo]”.
Given the findings described in the published article about the 1 out of 500 patients in the randomized trials who developed severely abnormal liver tests (greater than 20 times above normal) and liver damage, it demands explanation why no requirement to do liver tests was made in Rezulin’s initial labeling and why such recommendations were to wait until it had been on the market for seven months and many people were already starting to die of predictable liver failure.
Evidence of Serious Non-Compliance With Liver Tests
We have obtained data on the prescribing patterns and liver function (AST) testing patterns from an academic teaching hospital and medical center. Actual testing was compared to the amount of testing required by the label at the time Rezulin was first prescribed for the patient.
There were a total of 160 patients prescribed Rezulin through the beginning of March, 1999. Of these, 69 (43%) were first prescribed the drug prior to the first relabeling (October 28, 1997) and so were not subject to any testing requirement when they began to use the drug. Of the remaining 91, only 36 (40%) had a baseline AST liver test. Among the 40 patients prescribed the drug after the third and most recent relabeling (July 1998), 16 (40%) had a baseline test, indicating that compliance with that aspect of the testing requirement has not improved.
At the time of the first relabeling (October 1997), liver function testing was recommended 3 times in the first 6 months. This was increased to 7 in the first 6 months at the second relabeling (December 1997) and retained at this level in the third relabeling (July 1998). For example, we examined liver function testing patterns for patients prescribed the drug for the first time any time after the second relabeling (December 1997). None of the 21 patients who had been on the drug for at least 6 months had undergone the recommended 7 liver function tests in the first 6 months after being prescribed Rezulin. Six patients (29%) had no test in the first six months and the mean number of tests was only 2.0.
In sum, the overall compliance with the baseline testing requirement was dangerously inadequate with only 40% of the 91 patients who were started on the drug after the liver testing was added to the label in October, 1997 getting a baseline liver test. Only 1 of 26 patients (4%) who were taking the drug for six months or more after the liver testing requirements went into effect were in full compliance with the requirements for testing after the baseline. A necessary precondition for a labeling requirement to reduce toxicity is that the labeling requirement must be adhered to. This has clearly not been the case. The same lack of compliance with labeling was seen with Duract, also a hepatotoxic drug and Posicor, leading to their withdrawal from the market.
Cardiac Toxicity: Fluid Accumulation and Heart Failure
In a September 30, 1997 memo from FDA Diabetes Group Leader Dr. Alexander Fleming, there is a section entitled “Troglitazone continues to carry concerns about fluid distribution among body tissues and cardiotoxicity.” It states that:
Troglitazone is a member of a class of compounds that have been associated with cardiac toxicity… In study 032, 15 patients (4.6%) on troglitazone vs. 0 in the placebo group developed peripheral edema. One patient in study 055 developed pulmonary edema [evidence of heart failure]… It is hard to believe that patients with cardiac, liver or renal disease would not be adversely affected by the drug.
In reviewing the spontaneous adverse reports filed with the FDA, we found a total of about 50 patients with heart failure including 6 patients who died who were taking the drug.
Evidence that Others have Decided Rezulin is not Necessary
Dr. Joseph Loewenstein is an endocrinologist, for years at Case Western Reserve University School of Medicine and Meridia Hospital and now a Professor of Medicine at the School of Medicine at Texas Tech Health Sciences Center. He was a member of your Endocrine and Metabolic advisory committee from 1980 to 1984 and Chairman from 1982-1984. In response to my question to him about troglitazone, he replied, on March 11th (copy attached): ” You asked if the removal of troglitazone from the market would impair my ability to provide good care for diabetic patients. My answer is unequivocally, NO, it would not. I see a large number of diabetic patients in our faculty and resident clinics, and I have not seen a single patient in whom I thought troglitazone was essential. While troglitazone appears to increase tissue sensitivity to insulin, the published data fail to convince me that its benefits outweigh its risk of fatal liver toxicity. For this reason, I do not prescribe the drug, and when I see a new patient who is already taking it, I discontinue it.”
Dr. Loewenstein is hardly alone. Group Health Cooperative of Puget Sound in Seattle, one of the largest HMOs in the country, with about 450,000 patients and 575 physicians, also decided that Rezulin is not necessary for the practice of good medicine and have never had it on their formulary. To their credit, they also never put the three recently-banned drugs, Redux (dexfenfluramine), Posicor (mibefradil) and Duract (bromfenac), on their formulary, thus sparing the patients in their consumer-run cooperative a large number of serious adverse reactions.
In other countries, such as all of the European Union, diabetes patients are all cared for without the use of Rezulin as it has never been on the market in the rest of Europe and was taken off the market in the UK in December, 1997 by Glaxo Wellcome, its manufacturer there, because of an opinion, strongly shared by the British government, that its risks outweighed its benefits. Although Glaxo has now changed its mind, the British government has not and this week rejected Glaxo’s efforts to reintroduce the drug in the U.K.
In our petition to the FDA in July, 1998, to take troglitazone off the market because of hepatotoxicity, we stated ” How many more Americans will have to die or require liver transplants before Parke-Davis and the FDA take action to protect people in this country by banning the drug?” It is clear, one and one-half years after the label first recommended liver test monitoring, that most patients are not getting the recommended number of tests or even a baseline test. It is time for the United States to join the overwhelming proportion of countries where the drug is not on the market and protect American patients from a drug with no evidence of a long-term benefit in mortality and clear evidence of relatively short-term increased risks of death from liver toxicity. It is clear that the “label remedy” is not effective for those drugs which are inherently too dangerous to remain on the market such as Duract, Posicor and Rezulin. Every additional month on the market for Rezulin means longer duration of therapy for hundreds of thousands of patients with concomitant increased risk of liver damage and possible death or the need for transplantation because of liver failure.