Carlton Lee probably contracted hepatitis B when he was a Peace Corps forestry management volunteer in Sierra Leone in the 1980s. His young, vigorous immune system suppressed the initial assault, but the virus hung on tenaciously, a cranky, chronic nuisance living and breeding in his liver, a bomb that could go off at any time, that could cause fatal liver disease. In 1989, despite his own viral death threat--or perhaps because of it--Lee joined the National Commission on AIDS as its chief congressional liaison, the person responsible for representing the commission and its aims up on the Hill. He was so impressed by the activism of people with AIDS, by their aggressive pursuit of experimental treatments, that he decided to take a more vigorous role in the search for a cure for his own illness. Early in 1992 he was accepted into a clinical trial at the National Institutes of Health to test a drug against the hepatitis B virus. Last summer a toxic reaction to that drug killed him.
Four other volunteers in the same drug trial died in the space of less than three months; two more survived only after liver transplants. That casualty count makes this NIH trial the worst clinical disaster in recent memory. But it doesn't stop there: an investigation into earlier trials has revealed at least five other deaths and five hospitalizations that might have been caused by the same drug.
The trials and their fatal results have raised serious questions about how well scientists and government regulators minimize research risks for human volunteers who offer their bodies to advance medical science. Everyone wants to know the same thing: why this happened, and how to make sure it never happens again. To that end the NIH is reviewing the procedures used by its researchers during the trials; the Food and Drug Administration has assigned two task forces to study the matter; and the Institute of Medicine is embarking on an independent review of both institutions to ensure objectivity.
It was a medical nightmare to begin with, says Jay Hoofnagle, who as the leader of the NIH study has become the target of most of this scrutiny. Then it was an emotional and personal nightmare. Now it’s becoming an administrative, legal type of nightmare. It’s been very tough.
It may get even tougher. The drug in question--fialuridine, or FIAU--is a chemical cousin to several of the most prominent AIDS drugs today, including AZT and ddI, as well as several being tested in clinical trials right now. The FIAU fiasco has already irrevocably changed the way we think about clinical trials in this country; it may also irrevocably change the way we conduct them.
Hoofnagle is a gentle, young-looking 50-year-old doctor who, months after the nightmare began, still sounds somewhat stunned by it all. He has good reason. After more than 15 years of looking for a cure for hepatitis B, he got behind what he thought was the most promising drug of all, and it turned on him.
Hepatitis B strikes an estimated 5 percent of the world’s human population, some 300 million people concentrated mostly in Africa and Southeast Asia and especially in China. HBV--the hepatitis B virus--is transmitted sexually and through infected blood, just like HIV, the AIDS virus, though much more easily. In the United States it infects some 300,000 people a year; about a million Americans are living today with a chronic hepatitis B infection. Most of them never get seriously ill, although they may become jaundiced, their skin tinged yellow by the release of the pigment bilirubin from their damaged liver. But each year some 1,500 of those chronically infected Americans will develop a usually lethal primary liver cancer. Another 5,000 will suffer cirrhosis, a disease in which normal liver cells die and are replaced by scar tissue until the organ is no longer able to perform its vital functions: secreting bile, storing glycogen, breaking down proteins, and detoxifying drugs and food.
There has never been a really good treatment for hepatitis B. Indeed, there’s only one: alpha interferon, approved in 1992 by the FDA. Alpha interferon is an expensive drug--it costs about $300 per week and has to be injected into a patient three to seven times a week for four to six months. It often causes unpleasant side effects--fever, malaise, and other flulike symptoms and lowered white blood cell and platelet counts. It can also cause depression in some patients. For all that suffering, only 25 to 40 percent of patients wind up in long-term remission. But with even its limited success, alpha interferon renewed hope that HBV could be beaten, and it prompted American pharmaceutical companies to step up their efforts in looking for other potentially useful drugs.
FIAU and several of its chemical cousins were first synthesized at Memorial Sloan-Kettering Cancer Center in New York in the 1970s to treat herpesviruses. FIAU is a nucleoside analogue, a drug that structurally resembles a part of one of the building blocks of DNA--in this case thymidine--without mimicking it precisely. (The building blocks of DNA are called nucleotides; they consist of a chemical base with a sugar and a phosphate group attached. Nucleosides are nucleotides without the phosphate.)
Nucleoside analogues work by interfering with a virus’s ability to replicate itself. When a cell reproduces, it unwinds its double-stranded DNA and then duplicates each strand by picking up the complementary building blocks from those floating around in the cell; viruses, which can only replicate inside a cell by hijacking the cell’s reproductive enzymes, also pick up these same building blocks. Because the nucleoside analogues look so much like the nucleosides themselves, they will get picked up and incorporated into the growing DNA strands, both in the cell and in the virus. But once they become part of the DNA, these slightly abnormal nucleosides don’t act like nucleosides: they either prevent the addition of any more building blocks, halting the DNA replication in its tracks, or-- like FIAU--they prevent the DNA chain from replicating a second time by refusing to link up with any complementary building block. And if a virus can’t replicate--for whatever reason--it can’t take over the body and it can’t cause disease.
When FIAU was first tested, it actually showed little ability to block herpesvirus reproduction, and interest in the drug waned. Then, in the late 1980s, it and FIAC--a drug that converts to FIAU in the body--were revived by Oclassen Pharmaceuticals, a small company in San Rafael, California, best known for its medicated skin creams. At the time, Oclassen was thinking of using the drugs as a topical treatment for herpesvirus infection. But in reviewing the literature, Oclassen’s researchers became convinced that FIAC could be used to treat serious internal infections like those caused by cytomegaloviruses, viruses that are closely related to the herpesviruses and that commonly infect people with AIDS. Although the first clinical trials of FIAC, which began in November 1989, were unsuccessful-- indeed, only 3 out of 12 patients completed the study, and 4 died soon afterward--Oclassen still thought FIAC and FIAU worth another look.
In the summer of 1990 the company provided FIAU to the NIH for a 14-day-long study of six people infected with both HIV and hepatitis B. Jay Hoofnagle was one of the investigators in this study. We saw the most dramatic inhibition of HBV levels that we have seen with any drug, he remembers. Within a few days we would see 90 to 95 percent inhibition. Several of the patients became negative for hepatitis B virus DNA by the end of treatment. Usually with these antivirals, the levels go down, and when you stop they go right back up. But something was different here because the levels went down and they didn’t go right back up. It was quite dramatic.
Because of this success, in October 1990 Oclassen and the NIH-- but not Hoofnagle, who was short of lab help at the time--launched another trial of FIAU, which included 30 patients infected with both HIV and HBV. While there was no effect on HIV, there was again a dramatic decrease in the level of HBV in all the patients.
Hoofnagle was so impressed by the results of these studies that in April 1992 he began a 28-day, 24-patient trial of the drug in people with hepatitis B alone. These results were even more impressive. Six patients had permanent clearance of the virus, he notes. In addition, the side effects were minimal: the drug caused little more than fatigue and some upset stomachs. That happens all the time, says Hoofnagle. You get some gastrointestinal intolerance; it’s part and parcel of the drugs. Usually there is some kind of discomfort or nausea that goes away when you stop.
Essentially curing a quarter of his patients with a monthlong treatment causing few side effects was encouraging, but Hoofnagle believed that a longer treatment could give even better results. After all, it takes four to six months of interferon to help a maximum of 40 percent of hepatitis B patients. The same was probably true of FIAU. My feeling was, if you are going to treat this disease, you nail it, Hoofnagle says. You treat a person for three to six months. This is a chronic liver disease. People have it for life.
A longer trial, however, meant a lot more money, and Oclassen’s pockets just weren’t that deep. So in August 1992 Eli Lilly & Co. bought the overseas rights to market the drug and agreed to pay for its development and the needed safety tests. First they launched a six-month- long trial of the drug in dogs, to test the drug’s toxicity. There was none. The way was clear for long-term human trials.
The troubled trial began on March 24, 1993, with a great deal of optimism. All 10 of the patients enrolled in the first month of the trial had taken FIAU in the earlier trial and had no problems more serious than an upset stomach. They even liked taking the drug, a liquid Oclassen had reformulated to taste like Grand Marnier. After two months Hoofnagle added 5 more patients to the trial, for a total of 15. (The plan was to enroll 24 patients within six months.) By then the trouble was already brewing.
It’s difficult to say who got sick first. The early signs were unremarkable: fatigue, nausea, vomiting. Carlton Lee, however, was surely one of the early ones. He had participated in the monthlong 1992 trial and had felt nauseated and tired, but nothing too bad: indeed, 75 percent of the virus had cleared from his system. So he volunteered again for the longer trial in 1993. This time, however, after just a few weeks he began to experience terrible abdominal cramps and intense nausea.
Lee complained to Michael Fried, one of the NIH doctors in charge of his day-to-day care during the study. Fried performed extra tests but found nothing out of the ordinary. Despite the test results, the symptoms continued to worsen; on June 10, after taking FIAU for 11 weeks, Lee stopped taking it altogether.
He wasn’t the only one suffering. The only woman in the trial, Patient Six (to preserve their anonymity, Hoofnagle and his colleagues only talk about their patients by their trial numbers; Lee, Patient One, was enough of a public figure that his name made it into the newspapers) had such bad nausea and vomiting that the researchers took her off the drug on June 15. It didn’t help: she got worse and worse and became extremely jaundiced. Paradoxically, tests found no signs of HBV in her body. It seemed her hepatitis, at least, was in remission.
The real wake-up call came on Friday, June 25, when Patient Two was rushed to a hospital emergency room in Fredericksburg, Virginia. His liver had failed, sending his body into shock and causing his other organs to fall like dominoes. But what really caught the researchers’ eyes was his lactic acidosis. Lactic acid is thought to be the cause of the soreness felt in overworked muscles, a by-product of anaerobic respiration, the energy-production scheme a cell will use only when its need for energy exceeds its oxygen supply.
Normally, ions in the blood neutralize the excess lactic acid. But huge concentrations of lactate can overwhelm the blood’s buffering capacity. Patient Two’s insides were swimming in acid, and cells--whether liver, heart, or pancreas cells--simply can’t function in an acidic environment.
When word of Patient Two’s admission reached the NIH at 11:30 that Friday night, one of the doctors drove to Fredericksburg to see what was going on. He immediately had the patient transferred to the 500-bed NIH hospital. There the doctors realized that, despite normal blood tests just ten days earlier, the man was dying. On June 29 he was sent to the University of Virginia Health Sciences Center in Charlottesville for an emergency liver transplant. Five days later he got that new liver; two days after that he was dead. The acidosis had overwhelmed his body. The liver never got a chance to work, says Hoofnagle.
The day after Patient Two was first admitted to the hospital, Hoofnagle went to the NIH clinical center and began pulling the files of all 15 patients taking FIAU. There was no need to take any chances; he was taking them all off the drug. Since it was the middle of the summer, a few of the patients were on vacation. Hoofnagle sent the police to one man’s house to find out where he had gone. He reached another by tracking down one of his co-workers. When we reached them, he recalls, they would say, ‘I’m glad you called. I haven’t been feeling well for the last week.’ One person was on antibiotics--his doctor thought he had pneumonia or something. Another person said he’d caught a cold the week before and hadn’t been able to shake it and that he didn’t have an appetite and felt nauseated. This was a very common story. To us it was kind of scary. When the patients did show up at the NIH, it was clear that they had the same syndrome, liver failure and acidosis.
The trial was officially over. It was June 26, and the first ten patients had been taking the drug for as many as twelve and a half weeks. (The final five, who had taken it for just three weeks or less, never developed the severe problems the first ten did.) The initial ten were all suffering from some degree of neuropathy (tingling and pain in the extremities), pancreatic inflammation, or kidney dysfunction. But it was the lactic acidosis, a rare disorder, that caused the most problems. It resisted the usual treatment, an intravenous solution of sodium bicarbonate to soak up the excess acid and restore the body’s balance. Even after repeated treatments, the patients remained mysteriously acidotic.
They continued to worsen. On July 4, Patient Seven received a liver transplant; he died the next day. Patient Four--the man whose doctors were treating him for pneumonia--died on July 16 before he could undergo transplantation. Patient Ten had his transplant on July 19 and survived, though he’s suffering severe toxicity symptoms to this day. A liver transplant also saved Patient Three, who received one on August 3 and is now at home and doing well.
But even two liver transplants weren’t enough to save Patient Six--the female patient. She died at the University of Virginia on August 31, 53 days after she received her first transplant. That liver developed a blood clot and had to be removed; the second was working fine when she died from multiple organ failure.
Lee was admitted to the NIH hospital after he’d been off FIAU for nearly three weeks. He kept getting worse, says Hoofnagle, despite the drug’s having had plenty of time to clear from his body. The only thing that seemed to help was a daily infusion of three liters of highly concentrated glucose injected directly into one of the large central veins just outside the heart. It suppressed the lactic acidosis for reasons no one quite understands, but the suppression was only temporary. Lee’s liver continued to disintegrate and the acid continued to build up; ultimately the liver began to fail.
On July 28, Hoofnagle sent Lee to the University of Pittsburgh Medical Center for a liver transplant. When he arrived, the Pittsburgh doctors felt Lee was too healthy for an emergency transplant. But by the time his mother and sister arrived the next morning he was dying. His pancreas had begun hemorrhaging during the night, and only the most aggressive medical interventions were keeping him alive. Carlton Lee died the next afternoon, without a transplant.
Even in the midst of the clinical crisis, Hoofnagle, his colleagues at the NIH, and scientists from Lilly were struggling to figure out what was causing the poisonous reactions to FIAU. This is, I believe, a fairly unprecedented type of toxicity, says Hoofnagle. It was completely unexpected. It was sudden, and it was relentless.
The first real clues came from the failed liver of Patient Two. The liver was laced with tiny globs of fat. To Hoofnagle the organ looked like that of someone with Reye’s syndrome, an unusual illness that often follows childhood viral infections and has been associated with the use of aspirin to treat their symptoms. Reye’s syndrome is attributed to damaged mitochondria, the tiny structures in a cell that produce the energy needed to keep the cell alive. Mitochondria are semi-independent entities in a cell, and they come complete with their own set of genes, which are responsible for producing most of the enzymes that convert oxygen into energy. The researchers speculated that FIAU was somehow damaging the mitochondria’s genes, probably in much the same way it damaged the genes of the hepatitis B virus. If that were the case, then the tiny fuel factories would be useless; the cells, starved for energy, would switch to anaerobic energy production--which occurs in the cell’s cytoplasm, not in the mitochondria--and generate huge amounts of lactic acid.
The scenario seemed to make a lot of sense. Moreover, this wasn’t the first time a nucleoside analogue was associated with mitochondrial destruction. Yung-Chi Cheng, a Yale pharmacologist, had published a paper in 1989 describing how the nucleoside analogue dideoxycytodine, or ddC, could damage mitochondria and produce lactic acidosis in cell cultures.
Nucleoside analogues can put a halt to DNA replication only when added onto a cell’s or a virus’s genetic material by enzymes called DNA polymerases. There are several types of DNA polymerase in each cell: alpha and beta polymerase are typically active in a cell’s nucleus, while gamma polymerase tends to work best in the mitochondria and in a number of viruses. What Cheng found was that gamma polymerase was better able to introduce ddC into a DNA chain than were alpha or beta polymerase. Furthermore, because alpha and beta polymerase are active only in the nuclei of dividing cells, while gamma polymerase is active constantly in mitochondria, ddC could get incorporated more easily, and thus do far more damage, in a cell’s mitochondria than in its nucleus. While the DNA in quiescent cells was relatively safe from the nucleoside analogue, the DNA in mitochondria was always at risk. Two years after working out his mechanism for ddC poisoning, Cheng reported that five other nucleoside analogues used to treat AIDS--including AZT and ddI--can cause a similar problem.
At the time of his report Cheng had not yet tested FIAU, but he has since found that it does indeed behave like ddC. That would explain why FIAU’s toxicity is delayed, appearing only after several weeks of drug intake. There are between 300 and 1,000 mitochondria in a cell, each containing one circular piece of DNA. The suppression of the mitochondrial DNA does not affect the cell’s growth until the mitochondria drop to a certain level and can no longer support the cell, Cheng says. You have to knock out a sufficient number of the copies of the mitochondrial DNA. And it takes time to kill off that much mitochondrial DNA.
Cheng has a hypothesis that might also explain why the patients didn’t get any better after they stopped taking the drug. Most nucleoside analogues, once attached to the DNA strand, stop the addition of any more building blocks. Because they are thus always at the end of the strand and so in a somewhat vulnerable position, they can eventually be clipped off by the mitochondria’s primitive DNA repair mechanisms. FIAU, however, though it stops the replication of the strand, doesn’t stop its initial assembly. The analogue can thus end up in the middle of the DNA strand, and from there it can’t be removed. So once the mitochondria’s genes are damaged, they cannot be repaired; they cannot be used to direct the production of mitochondrial enzymes or to reproduce the mitochondria properly. The lactic acid comes pouring out, and the patient’s fate is nearly sealed.
It’s easy to have twenty-twenty vision in hindsight. Looking back, it’s clear that the first signs of trouble could be seen in Oclassen’s 1989 cytomegalovirus study. One of the four deaths in that study was attributed to fatal liver disease, which is not usually associated with cytomegalovirus. The second of the two studies involving patients infected with both HIV and hepatitis B--the one in which Hoofnagle was unable to participate--resulted in three deaths in which FIAU probably played a role, as well as the hospitalization of one patient.
Then came the first of Hoofnagle’s two studies on HBV-infected individuals alone, in April 1992. Paul Melstrom, a man in his fifties from Phoenix, was one of the 24 volunteers who took FIAU for a month in that trial. Some time afterward, in August, a neuropathy struck his toes and feet; the pain eventually became so great that he could hardly walk.
To Melstrom, a computer aficionado who describes himself as an angry advocate, the cause of the neuropathy was obvious: FIAU. But to the NIH doctors, it was not so clear-cut. At the time the pain began, Melstrom hadn’t taken the drug for four months. He had also suffered from neuropathy in the past, apparently caused by bouts of drinking. But he claimed he was no longer drinking, and so the doctors didn’t know why the neuropathy had returned. Melstrom got into quite a row with his NIH doctors over the cause of his pain. I am angry that all my complaints were dismissed as being unwarranted and lacking credibility, he told the FDA investigators.
Four other patients suffered some sort of liver ailment during that trial; the reactions were attributed to their hepatitis, but the FDA task force now thinks they may have been caused by FIAU. There was also one death--a perplexing case that, in retrospect, could have been an early warning missed by the NIH doctors. Patient 4D developed gastrointestinal problems two months after he stopped taking the drug. His local physician recommended removing his gallbladder even though he had no gallstones. The NIH doctors cautioned against the procedure because the man had mild cirrhosis, which can be exacerbated by surgery. The man decided to proceed anyway. After the operation he developed a somewhat uncommon complication of both gallbladder surgery and cirrhosis: fluid in the abdomen. He later developed lactic acidosis. He continued to deteriorate and doctors recommended a liver transplant, but he died in January 1993 before the transplant could be performed.
Should Hoofnagle and his colleagues have seen a disaster in the making? The FDA task force thinks the signs were all there. In all instances where observations and events encountered in the . . . studies might have suggested a drug toxicity, none was attributed by the sponsors to a toxic effect of FIAC or FIAU, they wrote in their report. But Hoofnagle, while declining to comment on the task force’s report, has noted that the delayed toxicity obscured the connection to the drug. The liver troubles in the earlier trials were easily attributed to the hepatitis itself; and with many drugs, patients often get worse before getting better, as their immune system revs up to beat the invaders.
There was another major obscuring factor: the system by which clinical trials are run and monitored. Scientists conducting clinical trials are not required to report serious complications or even deaths in patients unless they believe they were directly related to the drug in question. If they had--if Hoofnagle had known about all the prior problems up front, if he had been required to report all the complications in his own trials--perhaps FIAU’s deadly effects would have been detected sooner. To that end the FDA task force has made a number of recommendations, foremost among them being new criteria for reporting complications and deaths. The task force recommends that sponsors be required to report all expected and unexpected deaths, serious adverse experiences, and discontinuations that occur within six months of the last dose of the study drug or within the prescribed follow-up period, whichever is longer, they wrote.
The intent, obviously, is to ensure that researchers proceed with caution and conservatism when the lives of trial subjects are at stake. But the recommendation is at odds with other equally good intentions, like the attempt to get lifesaving drugs out to the people who need them as quickly as possible. Indeed, last November the government announced a new initiative to speed the development of anti-AIDS drugs. Will history repeat itself? I often find myself in the position of saying that a complaint is of an ‘unknown etiology’ because I really don’t know what causes it, says Donald Abrams, assistant director of the AIDS Activities Division at San Francisco General Hospital. If I can’t measure something, I say, ‘Well, it’s probably not important.’ This will sensitize me. I just hope I am not going to be paralyzed by it, and that we can come up with ways to ascertain what good we are doing and what potential damage we are doing to people when we enroll them in clinical trials.
The commissioner of the FDA, David Kessler, thinks the lessons learned from FIAU will prevent future disasters. But he warns that there’s always a danger of overregulating research to the point of stifling progress. No one wants to expose patients to needless risk, he says. And we have to do whatever we can to minimize that. But we are always balancing risks versus benefits. When we are dealing with life-threatening diseases, we have to be willing, because the benefits are potentially great, to take those risks.
Ironically, in the case of FIAU, those risks almost paid off. By four weeks, Hoofnagle says of the final, fatal study, there was a 90 percent inhibition of hepatitis B virus DNA. During the nine weeks of therapy, six of the ten patients became negative. Tests conducted since that time show that two more patients may have become negative before their deaths. Hoofnagle notes that fiau was one of the most powerful drugs he’s ever seen against hepatitis B and that it was effective at doses that were a fraction of those used with AZT and other nucleoside analogues. If researchers could find or develop a drug with FIAU’s potency minus its toxicity, its potential would be tremendous.
All these strengths, however, were overwhelmed by the drug’s fatal weakness. FIAU’s side effects were just too terrible for it to be used ever again in patients. The drug, says Hoofnagle, is dead.
