The Danger Within Us
Page 25
Despite their PR effort, FDA advisers voted against approval of tPA. Instead they voted to approve streptokinase.
Then something strange happened.
Despite the vote by the expert medical panel to flunk tPA and approve streptokinase, FDA managers ruled that neither drug would be approved. The action, dubbed the Friday afternoon massacre, was a staggering setback for Genentech. Genentech stock values plummeted. Things looked even worse for Genentech when the FDA decided in early November of 1987 to reverse its position and approve streptokinase for the treatment of heart attacks.
But science would be no match for Genentech’s PR machine. Patients and cardiologists, influenced by Genentech’s PR and/or money, inundated the FDA with demands for the approval of tPA. Talk show host Larry King, who had been treated with tPA for a heart attack, railed against the FDA on his show. Most important, Eugene Braunwald jumped into action and composed a telegram to Senator Ted Kennedy to be signed by other prominent cardiologists. Braunwald’s name alone was so powerful that one cardiologist had just one question about the telegram: was Braunwald going to sign it? When told he was, the cardiologist said, “If Gene Braunwald is going to sign that, then I’ll sign it, too.”342
Braunwald’s intervention paid off. According to Linda Marsa’s book Prescription for Profits, Senator Kennedy reportedly spoke with FDA commissioner Frank E. Young, and on November 13, 1987, the FDA announced its approval of tPA for the treatment of heart attacks.
Once the drug was approved, Genentech still had to fight off its chief competitor, streptokinase, in the marketplace. It would be a hard sell: Genentech set the price of tPA at more than ten times the price of streptokinase—about $2,200 for a dose of tPA versus $200 for streptokinase. Genentech had to show that tPA was superior to streptokinase in order to justify the higher price. British cardiologist Peter Wilmshurst says that US cardiologists “made a virtue” of the greater cost of tPA, saying that the high price “was evidence that it was better than streptokinase” and that they told patients, “Greater price must mean better quality.”
Initially Braunwald and Genentech claimed that tPA was safer than streptokinase because it was “clot specific” and would break up only the clot causing the heart attack without affecting the normal mechanisms that stop bleeding. And, unlike streptokinase, which was routinely given with the blood thinner heparin, tPA could be given without heparin, which they claimed would make it less likely than streptokinase to cause troublesome side effects, especially bleeding and brain hemorrhage.
But Genentech’s safety claim failed to hold up. William R. Bell, the researcher who was testing tPA’s clot-specific nature at Genentech’s request, didn’t find a safety benefit. As he was preparing to submit his findings to the New England Journal of Medicine, he came under “big-time pressure” to withdraw the paper. Linda Marsa relates what happened:
One of his superiors came down to talk to him. “Is there any way that’s humanly possible that there might be a mistake?” he asked Bell.
“Well, I’m a human being,” Bell responded. “Sure, I can make one. But I don’t really think so. These are standard pedestrian-type assays that we’re doing—and it’s certainly not the first time we’ve done this.”
“Do you think there’s even a possibility that these things are wrong?” he pressed.
“Well”—Bell shrugged—“I’m not God.”
“Then I think you’d better withdraw this,” he was told.
Bell complied.
Other research also failed to support the clot-specific safety claims. In fact, tPA caused more strokes resulting from brain hemorrhage than streptokinase did. In a study of nearly 21,000 patients, 1.0 percent of patients treated with streptokinase developed a hemorrhagic stroke following treatment compared to 1.3 percent of patients given tPA.345 Following challenges to the safety claim, the company changed tack: tPA, it now said, might not be safer, but it was more effective. The claim was based on the open-vessel hypothesis, which maintained that tPA opened blocked arteries more quickly than streptokinase did, which they said would mean less heart muscle would be damaged.346, 347
Then, in 1985, Braunwald and his colleagues published the results of the landmark TIMI 1 trial in the New England Journal of Medicine.347 They reported that ninety minutes after treatment, twice as many patients treated with tPA compared to those treated with streptokinase had opening of a blocked coronary artery, which they demonstrated by angiography—shooting dye in the patients’ coronary arteries. They concluded that tPA “represents a major therapeutic advance in thrombolysis, having almost twice the thrombolytic effectiveness of intravenous streptokinase with the same or lower incidence of side effects.”347
Braunwald and his colleagues stopped the study early, after testing just 290 patients, claiming that tPA was so effective that it would have been “unethical” to continue the TIMI trial to completion because it would “deprive” patients of superior treatment.
Jerry Hoffman saw things differently. The New England Journal of Medicine paper, he said, “provided no information about clinical outcomes, so it seemed clear that this was simply another claim based on yet another surrogate marker.” What’s more, Hoffman pointed out, the paper “didn’t answer truly important questions: Does opening the vessels faster actually make a difference in patient outcomes? Will fewer patients die of a heart attack? Or develop congestive heart failure? Or return with chest pain or another heart attack? These are the things that matter to patients—clinically relevant so-called ‘patient-oriented’ end points—not laboratory measures or physiologic theories that might or might not mean anything.”
Hoffman also warned that stopping studies early for presumed benefit is known to increase the risk of bias that contributes to the illusion of benefit where there may be none.348
But Hoffman was waving his arms in the dark. The hypothesis that achieving vessel opening at ninety minutes would save lives and preserve heart function was so seductive and logically compelling that most doctors simply didn’t question the claim. The chorus supporting Genentech’s assertions was growing. The superiority of tPA seemed like a slam dunk. Surely opening a blocked coronary artery earlier had to provide a clinical benefit.
Hoffman says, “When the NIH concludes that the question is so clearly answered that the study doesn’t need to be finished, and the New England Journal of Medicine agrees that it’s important enough to publish the limited results, that attracts a great deal of interest.” But, he said, “The claim that tPA was more effective than streptokinase was purely theoretical and not yet verified by patient-oriented outcomes in this or any other study.”
Still, few doctors were paying close attention to the science behind tPA. Instead most relied on the recommendations of luminaries such as Braunwald. For them, tPA had proved superior, and its use became almost universal in the US.
Hoffman continued to urge caution. In his lectures, he hammered away at the unproved claims about the putative benefits of tPA. He sent a letter to the NIH pointing out that reliance on surrogate markers and a prematurely stopped study were not sufficient evidence of benefit.
Once again, his cautions were prescient. After the FDA approved tPA, the TIMI researchers gradually published the results of clinically important outcomes (though always in far less prestigious or widely read journals than the New England Journal of Medicine), and three years after their 1985 publication announcing tPA’s superiority to streptokinase, Braunwald and his colleagues acknowledged that “there was no significant difference” in either mortality or recurrent heart attacks between the two treatment groups.349
In light of this, the NIH was forced to reconsider its earlier decision that the superiority of tPA had been proved by the small number of test subjects in the TIMI 1 trial, and they called for a larger head-to-head trial of tPA versus streptokinase to examine clinical outcomes and overall death rates.
This sent Elliott Grossbard into a tizzy. He vehemently opposed such a trial. Genentech, basically
a one-trick-pony company that depended almost entirely on tPA for its profits, had already won control of the US market.350 During a private meeting in Washington, DC, with Braunwald and two of his colleagues, Grossbard argued against a new study: “Almost every cardiologist in America is convinced that tPA is so good that you had to stop the study,” he said. “We don’t know how another trial would turn out. And if we don’t come out ahead, we would have a tremendously self-inflicted wound. So why should we do it?…This could be a good thing for America, but it wasn’t going to be a good thing for us.”342 *
Grossbard was right. If tPA did prove superior to streptokinase, it couldn’t benefit the company because doctors in the US were already using the drug almost exclusively. But if tPA proved to be no better or even inferior to streptokinase, they could expect to lose, and lose big.
At first Genentech forestalled the head-to-head trial the NIH had demanded by simply refusing to supply tPA. But the company was forced to change its stance after two massive European trials involving approximately sixty thousand patients found no difference between tPA and streptokinase in clinical outcomes.351, 352 Worse, the studies confirmed that patients treated with tPA suffered more strokes from brain hemorrhage compared to patients treated with streptokinase.350 European doctors by then were mostly using streptokinase. Facing a growing threat to the US market as a result of the megatrials, Braunwald and a famous University of Michigan cardiologist, Eric Topol, dismissed the European findings, saying they were flawed because the researchers didn’t give the anticlotting drug heparin with tPA, and the lack of heparin, they said, reduced tPA’s efficacy (of course Braunwald and his colleagues had earlier claimed that tPA was safer because heparin wasn’t needed with tPA, as it was with streptokinase). Topol wasn’t a disinterested expert regarding tPA; at the time, he held options on six thousand shares of Genentech.353 (Topol eventually had a change of heart, became a critic of drug companies, and called streptokinase the “gold standard drug” for heart attacks.)354
In the aftermath of the European trials, advocates of tPA circled their wagons and unleashed scathing attacks on skeptics, leading one cardiologist at the fortieth annual scientific session of the American College of Cardiology to say it was as if any criticism of tPA were “an attack on the American way of life.”354 The vitriol led to some head scratching about why US doctors would cling to a drug that was clearly less safe and more expensive than its rival. Theories ranged from America’s love affair with technology to unbridled faith in any new drug made in the US.
Following publication of the European trials in the 1980s, sales of tPA threatened to tank, and Genentech changed its tune. It was now ready to conduct the head-to-head trial the NIH had asked for, but it wasn’t about to do it with the NIH. Instead the company would run the trial itself with cosponsors and colleagues around the globe. The study, known as GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries), was led by Topol, and unlike the European trials it did report a mortality benefit for tPA over streptokinase.
But GUSTO came under strong criticism. Critics pounced on the study, pointing out weaknesses in the study design and “inexplicable” contradictory numbers. Experts in the US and the UK published their doubts, which one critic summed up by saying that the validity of the study was “not clear” but that one thing was “incontrovertible”: tPA “is a more dangerous and much more expensive drug than streptokinase.”355, 356
Jerry Hoffman said, “I believe there are multiple problems with GUSTO, but even if you ignore these, the benefit they claim to have found was tiny. And even that difference goes away if you combine GUSTO’s results with those of the two European megatrials.”
The crack in Braunwald’s reputation appeared to be widening. No longer were doctors automatically buying his claims. According to a New York Times report, when Braunwald continued to insist on tPA’s superiority to streptokinase at the 1989 annual meeting of the American College of Cardiology, “many cardiologists at the meeting were quick to criticize the study. They said that Braunwald’s interpretations far exceeded his data.”357
* * *
Braunwald’s biographer and fellow Harvard physician, Thomas H. Lee, suggests that Braunwald’s insights and research, along with Genentech’s development of tPA, were in large part responsible for the dramatic decline in the heart attack death rate, writing, “Since the 1950s, the death rate from heart attacks has plunged from 35 percent to about 5 percent—and fatalistic attitudes toward this disease and many others have faded into history. Much of the improved survival and change in attitudes can be traced to the work of Eugene Braunwald, M.D.”17
But Jerry Hoffman says that attributing such impressive numbers simply to the effect of treatment relies on yet another illusion. He explains that the death rates cited apparently refer to case fatality rates. That is, for every X number of people suffering heart attacks, Y percent die. If one hundred people have heart attacks and twenty of them die, the case fatality, or death rate, is 20 percent. What those numbers don’t reveal is how many individuals had a heart attack overall—and therefore they don’t tell us the total number of deaths. In other words, if millions of people suffer heart attacks, and 20 percent of them die, that means a lot of deaths, whereas if thousands of people suffer heart attacks, and 20 percent of them die, that’s far fewer deaths. And the number of people diagnosed with a heart attack has increased dramatically in the early part of the twentieth century—not because more people are having serious heart attacks but because increasingly sensitive tests can detect mild heart attacks that would not have been previously diagnosed.
During the first half of the twentieth century, doctors could only identify a heart attack when patients developed symptoms and had changes on their EKGs, or heart tracings. These patients tended to have very severe heart attacks, and case fatality rates were high: up to 35 percent of these patients died.
But in the 1970s, researchers began to refine diagnostic blood tests by identifying enzymes that leak out of damaged heart cells. With each new and increasingly sensitive test, tinier and tinier bits of cellular damage to the heart were identified as “heart attacks.” Now, in the twenty-first century, even a patient with a normal EKG tracing and so little heart damage that he or she has no symptoms at all can be diagnosed with a heart attack. These patients tend to be far less likely to die. Adding these patients to the total number of heart attack patients dilutes the pool of symptomatic patients, creating the illusion that fewer people are dying. Before the era of such testing, twenty deaths among one hundred people suffering severe heart attacks (identified by symptoms and EKG changes) meant a 20 percent case fatality rate. Since such testing became widespread, sensitive tests might identify two hundred or even four hundred people as having a heart attack (many with minor or no symptoms), yet still only twenty people die. This slashes the case fatality rate in half, to just 10 percent (or, in the case of four hundred heart attacks, to 5 percent)—even though the overall death rate, or mortality rate, remains the same—creating a fantastic illusion of benefit from clot buster drugs, which happened to come onto the market at the same time that sensitive enzyme tests for heart attacks were developed.
Although increasingly sensitive tests for heart attacks account for a substantial decrease in the case fatality rate, this doesn’t explain the whole picture. Total heart attack deaths have declined somewhat since approximately 1970, when highly sensitive tests were introduced. However, according to the Centers for Disease Control and Prevention, smoking began to decline in 1968, and that was estimated to account for more than half the decline in deaths, while other lifestyle modifications were responsible for further reductions.
Drugs like tPA can be helpful, but they turn out to account for a very tiny reduction in overall mortality. In fact, tPA offers about the same amount of benefit as a single aspirin: a 2 percent absolute reduction in mortality for the individuals who take the drugs. When aspirin and tPA (or aspirin and st
reptokinase) are used together, the result is additive, for a 4 percent reduction in death rates. But tPA still carries the pesky side effect of causing brain hemorrhage and can’t be used in many heart attack patients, whereas aspirin can be used in most.
But none of this stopped the forward march of Genentech’s main money maker or the celebrations that followed the FDA’s approval of tPA in 1987, when fireworks exploded and air traffic was stopped. Braunwald and Genentech were celebrating the approval of a drug, yes, but in a larger sense they were celebrating the dawn of a new era, one in which doctors could be multimillionaires and healthcare would come to consume 20 percent of the entire US budget.
In the years to come, those involved in bringing tPA to market would prosper. Herb Boyer, a Genentech cofounder and scientist who helped develop the recombinant-DNA technique used to produce tPA, owned more than two million shares of Genentech stock by 1992—stock valued at $30 million. For his part, Braunwald launched his research empire promoting tPA and ultimately gained substantial funding from Genentech for his research. Meanwhile, it turned out that at least thirteen of the TIMI study authors either owned Genentech stock or held options to buy it at a discount, and none had reported those interests. A congressional subcommittee report determined that “[t]he research literature on t-PA has repeated examples of more positive evaluation of t-PA by scientists with relationships with Genentech, compared to scientists without such relationships.” Yet the study carried the blush of independent scholarship, supported by the NIH and led by Eugene Braunwald at Harvard. Genentech stock prices soared from $18 to $28 following publication of the TIMI trial results.
By 2010, the TIMI Study Group trials were almost wholly funded by industry, with some NIH grant support for ancillary studies.358 Just how much grant money Braunwald’s TIMI group would receive from Genentech is secret. When I asked them for the information, Braunwald, the TIMI Study Group, Harvard University, Harvard Medical School, Massachusetts General Hospital, and Genentech all declined to reveal the total amount of funding provided by Genentech for the TIMI trials.