“We Opened Pandora’s Box”

Between drugs, food, dietary supplements, tobacco, and medical devices, the FDA regulates $3.9 trillion worth of products each year, roughly one-eighth of the entire US economy. With a budget of approximately $6.9 billion in 2024, hundreds of offices in the United States and abroad, and more than 19,000 full-time employees (prior to recent firings), the FDA’s resources and scope make it one of the most powerful agencies in the country and highly influential around the world.

Its reputation was built on hard-won scientific standards put into place by Congress in 1938 and 1962 after a string of medical tragedies. These new laws, sometimes called “super-statutes” because they are so far-reaching, authorized the agency to require drug companies to provide evidence that their drugs are safe and effective before they could go on the market.

Then came AIDS. By 1988, the epidemic was in full swing: 46,000 people had died, and another 37,000 were living in the US with the poorly understood condition. AIDS activist groups wanted access to new drugs — and fast. With news cameras rolling, activists blockaded the entrance to FDA headquarters in Rockville, Maryland, holding placards saying “Federal Death Administration,” and lying on the street holding cardboard tombstones that read “Killed by the FDA.”

Activists found ready partners in drugmakers, who were eager to get their products to market. Together the activists and companies argued before the FDA that the compassionate, lifesaving strategy would be to loosen scientific standards for establishing the efficacy of drugs.

Among those existing standards was the strong recommendation that companies submit two or more randomized, controlled clinical trials that show a drug is effective in order to win approval. That’s because any single study, no matter how well it’s conducted, will often yield results that can happen by chance alone and do not stand up to later scrutiny.

But these trials can be expensive and time-consuming to conduct and analyze. Between that and a shortage of FDA reviewers, new treatments for AIDS were flowing through the chronically underfunded agency at the speed of chilled molasses. It took an average of seven to twelve years to take a drug through the development pipeline, three of which could be taken up while a company’s data sat on an FDA reviewer’s desk.

Under pressure from patient groups and drugmakers, the FDA created an “accelerated pathway” in 1992, which allowed companies to provide preliminary evidence that their AIDS drugs were effective. The new rules let these companies conduct trials tracking unreliable patient outcomes known as “surrogates,” with the expectation that drugmakers would provide more substantive evidence of real patient benefit after the drugs were on the market.

Surrogate outcomes are laboratory tests, or imaging studies like CT scans, that don’t themselves track quality or quantity of life but are hypothesized to be reasonably likely to predict a so-called “clinical benefit.” Clinical outcomes are the ones that matter to patients, like feeling better and living longer, but relying on surrogate outcomes means that drug companies could gain approval based on shorter and cheaper studies.

The problem is that improvements in a surrogate outcome often fail to correlate with whether a drug actually improves patients’ lives. In addition, a surrogate outcome can’t reflect the harm that a drug can cause, a limitation that has been repeatedly confirmed.

Some external experts therefore expressed concern about approving drugs based on surrogate outcomes — and those worries were borne out by the AIDS drug AZT.

When AZT was approved in 1987, many expected it to be a raging success based on research that counted T-cells, disease-fighting cells that the AIDS virus attacks, and other studies that tracked early survival data. But less than two years later, researchers at the Claude Bernard Hospital in Paris published their findings on 365 AIDS patients treated with AZT, concluding the results were “disappointing.” The drug was toxic to blood cells, and after six months, patients “returned to their pretreatment levels [of T-cells] and several opportunistic infections, malignancies, and deaths occurred.’’ (AZT would prove beneficial in later years when used in lower doses as part of a triple cocktail of AIDS drugs.)

Still, the pharmaceutical industry actively promoted the treatment. With a list price of $21,000 a year per patient in 2025 dollars, AZT delivered $2 billion in profits to its drugmaker Burroughs Wellcome in 1989. Seeing the rewards to be reaped, companies began lobbying Congress to force the FDA to apply the looser AIDS standards to all sorts of drugs.

Big Pharma’s campaign contributions jumped from $1.9 million in 1990 to $3.6 million in 1992, according to the Federal Election Committee and OpenSecrets, an independent research group. That year, Congress passed the Prescription Drug User Fee Act, which explicitly directed the FDA to allow approvals based on lower standards.

Rep. Henry Waxman (D-CA) slammed the legislation, saying that while it “sounds good . . . it is a hoax to imply that it will do more than the FDA is already doing to bring effective AIDS drugs to the market.”

The act also required drug companies to pay “user fees,” which the agency used to hire nearly 600 new staff. But critics say such fees have led some senior agency officials to view the industry as “partners.” The agency would go on to launch multiple “public-private partnerships” with the companies it is tasked with regulating.

Subsequent laws directed the FDA to expedite drug approvals by further lowering standards of evidence.

“We opened Pandora’s Box, and pharma exploited it,” says Gregg Gonsalves, an AIDS activist and professor of epidemiology at the Yale School of Public Health. In 1993, he told an FDA advisory committee that with the best of intentions, and in the face of a terrifying health crisis, AIDS activists had helped “unleash drugs with well-documented toxicities onto the market, without obtaining rigorous data on their clinical efficacy.”

The result, said Gonsalves, is “we’ve arrived in hell.”

The Missing Evidence

Today, hundreds of drugs intended to treat a wide range of diseases have been approved by the FDA without critical evidence demonstrating they work.

According to the database created for this investigation, nearly three-quarters — 311 — of the 429 drugs approved from 2013 through 2022 were put on the market despite inadequate evidence of their efficacy.

This analysis was based on the four essential criteria cited in the FDA’s own standards and case law for assessing whether drugs work and are safe:

• Control group: Patients taking the drug were compared to a control group that was given a placebo or a comparator drug.
• Replication: At least two “well-controlled” trials showed the drug was effective.
• Blinding: Subjects in the studies and the doctors who cared for them don’t know which patients are on the drug and which are in the control group.
• Clinical end point: The studies measured the drug’s effect on patients’ survival or function rather than a surrogate measure.

These criteria are by no means a complete list of sound scientific evidence. Instead, they comprise the minimum criteria needed to determine whether drugmakers have provided “substantial evidence” to support claims of effectiveness for new drugs.

Yet only 118 of the 429 drugs approved over that period, a mere 28 percent, met the four criteria. One hundred and twenty-three drugs (29 percent) met three of the criteria, while thirty-nine drugs, more than 9 percent of the total, failed to meet a single criterion. That doesn’t mean these drugs don’t work, but it does mean the FDA approved them without knowing whether they are more likely to help patients than hurt them.

More than a quarter of the drugs approved over the study period, 123, were for various forms of cancer. This might seem like good news for the war on cancer, but only three of those drugs met all four of the minimum criteria, and twenty-nine cancer drugs failed to meet a single criterion. In other words, one in four cancer drugs were approved based on studies with no comparison group, no blinding, no replication, and no clinical outcome.

The most common way drug companies get such cancer drugs approved is by using a surrogate outcome in their clinical trials, rather than measuring actual clinical benefit. Drug companies routinely study laboratory measures like how much a tumor shrinks in response to a drug, instead of mortality or symptom relief.

But experts say the FDA should have no illusions regarding the unreliability of such surrogate outcomes, especially since these outcomes fail to take into account the harms a drug may cause, such as liver failure and anemia. These criticisms were hammered home after the agency approved Avastin to treat metastatic breast cancer.

Originally approved for two other types of cancers, Avastin hit the market for breast cancer in 2008. At that time, women with metastatic breast cancer faced a dire prognosis: most would be dead within five years, even after multiple rounds of chemotherapy drugs. Avastin was hailed as a breakthrough because it doesn’t kill cancer cells the way chemotherapy does but instead inhibits the growth of blood vessels, which tumors need to grow.

Avastin was approved based on a surrogate outcome called “progression-free survival,” which doesn’t actually look at patient survival but instead measures how long a cancer drug keeps a tumor in check. But just because a tumor isn’t growing or even shrinks doesn’t mean the patient will live longer or enjoy a better quality of life.

In 2010, worldwide Avastin sales had hit $6.8 billion. The approval for breast cancer had come with the condition that the manufacturer, Genentech, conduct another study to determine whether the drug actually worked. Two years later, the company produced two studies that suggested the opposite: the drug did not help people live longer.

In total, the company had conducted five clinical trials, none of which demonstrated that Avastin helped breast cancer patients live longer or with less disability. The new studies also documented the drug’s more serious side effects, which included blood clots, perforated intestines, stroke, heart problems, and kidney malfunction. Genentech did not respond to repeated requests for comment.

But when the FDA proposed withdrawing approval for Avastin for breast cancer, Genentech and several patient advocacy groups fought back, sometimes viciously. At a two-day public hearing in 2011, one of the few patient advocates testifying in favor of the FDA’s decision came with a bodyguard after being threatened verbally by others who supported Avastin. FDA staff received violent threats, and police were posted outside the building.

The FDA ultimately withdrew its approval of Avastin for breast cancer, but the episode, still referred to as “Armageddon” by some staff, had a chilling effect. The agency would not demand another drug withdrawal in the face of company opposition for more than two decades.

Today progression-free survival, along with a half-dozen other surrogate outcomes, has become the norm for approving cancer drugs. This investigation found that 81 percent of the 123 cancer drugs approved from 2013 through 2022 were based on studies that did not track overall survival and instead reported on progression-free survival or another surrogate.

Such approvals have real consequences for patients. Take Copiktra, a cancer drug manufactured by Secura Bio Inc. that was approved by the FDA in 2018 based on improved progression-free survival and other surrogate outcomes. According to a Lever review of FDA records, two-thirds of leukemia and lymphoma patients taking the drug developed serious complications. Patients treated with Copiktra also died eleven months earlier than patients treated with the comparator drug.

In December 2021, the manufacturer announced it would stop marketing the drug for follicular lymphoma. Six months later, the FDA issued a warning about “possible increased risk of death and serious side effects.” But it wasn’t until July 2024 — six years after studies showed the drug cut months off the lives of patients — that the FDA announced that Copitra should not be used as first- or second-line treatment for certain types of leukemia and lymphoma “due to an increased risk of treatment-related mortality.”

A spokesperson for Copiktra’s manufacturer, Secura Bio, told the Lever that in May 2022, the company sent a “Dear Healthcare Professional” letter to physicians about the results of the postmarket studies “in accordance with FDA requirements.” The company said Copiktra can still be prescribed for certain conditions if two other drug treatments have failed.

Copiktra is hardly an exception. According to a 2015 study, only 14 percent of cancer drugs approved based on a surrogate outcome, such as progression-free survival, between 2008 and 2012 were later shown to improve real-world survival. In other words, 86 percent were not supported by evidence that they helped patients live longer. What the drugs did cause were numerous serious side effects, while costing patients, taxpayers, and insurers billions of dollars a year.

Another, more recent study found that between January 2006 and December 2023, 147 cancer drugs that had been approved on a surrogate outcome were later tested to determine if they improved overall survival. Approximately three-quarters failed to do so.

Even when cancer drugs do improve survival, the gains can be meager. A 2022 study by researchers at the National Cancer Institute found that the median survival time for 124 cancer medicines approved from 2003 through 2021 to treat solid tumors was just 2.8 months. The average cost of such drugs is more than $24,000 per month.

In other words, after patients and their families spend financially crippling sums and go through side effects ranging from nausea to death, their outcomes are hardly better — and could be worse — than no treatment at all.

That outcome is not what most cancer patients expect when their doctor writes a prescription. It’s also at odds with what doctors believe about the meaning of FDA drug approvals. In 2016, researchers published the results of a survey sent to physicians that included the following question: “For [the] FDA to approve a drug, must the studies show: a.) a clinically important result; b.) a statistically significant result; c.) both; d.) None of the above.” The correct answer was none of the above.

Only 6 percent of doctors got the answer right. Aaron Kesselheim, a researcher at Harvard Medical School and coauthor of the report, said the result was “disappointing, but not entirely surprising. Doctors aren’t taught about the approval process.”

When the Science Never Comes

Nothing about ProAmatine would suggest it would ever get on the market.

This investigation found that between 1990 and 1996, the FDA rejected the blood pressure drug four times. The agency told the company, Roberts Pharmaceutical, that its studies showing ProAmatine could treat low blood pressure when a patient stands up were “poorly conducted.” Nor were the “incremental” increases in blood pressure sufficient proof of benefit since they were surrogate measures, according to the agency, which wanted evidence that patients would feel or function better. The agency instructed the company to conduct randomized, double-blind, placebo-controlled studies.

Before the agency rejected the drug for the fourth time, in March 1996, FDA medical reviewer Maryann Gordon noted a striking incidence of severe high blood pressure among treated patients. She suggested that this side effect might account for at least some of the strokes, heart attacks, and congestive heart failure that had occurred among patients on ProAmatine. She concluded, “The mostly unexplored benefits of midodrine do not outweigh its real risks, thus making midodrine [ProAmatine] not approvable.”

But less than three months after the FDA’s fourth rejection of ProAmatine, the agency reversed course, saying that following a meeting with the company, it had “reconsidered” its prior rejection and would approve ProAmatine after all, under “21 Code of Federal Regulations Subpart H,” a provision in the regulations that the FDA interprets as giving the agency “flexibility” in its approval decisions.

One FDA insider who asked to remain anonymous told the Lever, “Subpart H was never meant to rescue failed drug development programs after trials had already failed to show direct benefit to patients.” Instead the provision was designed to allow the agency leeway in approving drugs that looked highly promising for life-threatening conditions for which there were few or no other treatments.

While the condition treated by ProAmatine is not life threatening, Roberts Pharmaceutical was able to use the regulation to get the drug on the market in 1996, agreeing to submit a follow-up study no later than 2000 to determine if the drug worked.

The year 2000 came and went. No study. In 2003, the drug patent expired. Still no study. The FDA tried to induce the company to conduct the promised postmarket study by granting a three-year patent exclusivity extension, allowing the company to maintain a monopoly on the drug and keep its prices high. In 2005, the company submitted two studies, but the FDA found them insufficient.

In 2013, researchers at the Mayo Clinic conducted a comprehensive review of all studies of the drug and found ProAmatine provided “no significant benefit” in blood pressure change on standing up. The drug also caused a higher incidence of adverse events, and the researchers concluded “there is insufficient and low-quality evidence” to support its use.

Yet the drug was allowed to remain on the market with no caution issued to patients or doctors until 2017. As of 2009, it had generated $257 million in sales; in 2023, the midodrine market was estimated to be $745 million. Twenty-nine years after the drug was approved, there is still no evidence that it works.

“I think rightly or wrongly, FDA managers have decided that completely bending their standards could still lead to good drugs,” says Matthew Herder, a legal scholar at the University of Dalhousie in Halifax, Canada. “They put a lot of faith in postmarket evidence.” But the history of ProAmatine demonstrates that once a drug has been approved on an expedited basis, a requirement for postmarket evidence can have little meaning.

According to a 2022 analysis by the US Department of Health and Human Services Office of Inspector General, more than one-third of drugs approved on an accelerated pathway have never seen a confirmatory trial. When they did conduct the studies, regulators found companies took anywhere from a few months to twelve years to do so. This investigation found that confirmatory trials can take even longer — up to thirty years — and may not be performed at all.

ProAmatine, no longer sold under its brand name, is currently manufactured by multiple companies under the generic name midodrine. Three generic manufacturers of midodrine failed to respond to requests for comment.

Even when postmarket studies are submitted, they often provide no new information. Bishal Gyawali, an oncologist and associate professor at Queen’s University in Kingston, Canada, and two Harvard colleagues looked at cancer drugs approved based on surrogate outcomes between December 1992 and May 2017. Twenty percent of the time, the company submitted a follow-up study that looked at the same surrogate it used to get the drug approved — even though the point of the additional study was to determine whether the drug offered an actual clinical benefit. In another 21 percent, the follow-up trial used a different surrogate end point, rather than clinical outcomes.

Such scenarios are so common that the agency has created a term, “dangling approvals,” for the status of drugs whose postmarket trials failed or weren’t conducted before their deadline. As of 2021, the FDA had allowed ten of thirty-five cancer treatments to remain on the market even after their follow-up studies failed. The manufacturers eventually withdrew eight of the ten treatments, but only after patients received ineffective treatments for years and companies reaped healthy financial rewards.

Among the drugs with dangling approvals was one that had generated controversy before: Avastin, the failed breast cancer drug. Avastin has been shown to improve survival for some cancers, and in 2009, the FDA authorized the drug for recurrent glioblastoma, a fatal brain cancer, based on progression-free survival. In 2017, after the requisite “confirmatory” trials, the agency granted it full approval — even though the follow-up studies failed to show it helped patients live any longer.

Avastin remains on the market for recurrent glioblastoma — at a cost of $153,000 per year. European regulators refused to approve the drug for that use, citing concerns about safety and efficacy. A spokesperson for Genentech said some patients could benefit, noting that 11 percent of patients were able to discontinue treatment with steroids.

Then there’s Keytruda, which is widely advertised on TV for a number of cancers. Granted expedited approval in 2017 to treat a certain type of gastric cancer, the drug subsequently failed its postmarket study. Keytruda was sold for this use for nearly four years before regulators withdrew its approval in 2021. It remains on the market for other cancers. Keytruda maker Merck did not respond to requests for comment.

While drug companies profit from the sales of unproven drugs, everyone else — patients, insurers, and the government — pays a heavy price. In just four years, from 2018 through 2021, the taxpayer-funded health insurance programs Medicare and Medicaid shelled out $18 billion for drugs approved on the condition that their manufacturers produce confirmatory trials that had yet to be delivered.

“Everybody from the [FDA] commissioner on down keeps telling us that approving these marginal drugs will lead to an accumulation of benefits to patients over time, but if a drug doesn’t work, it doesn’t work, and putting them all together doesn’t add up to an effective treatment,” says Fran Visco, a breast cancer survivor and president and cofounder of the National Breast Cancer Coalition. “The entire medical oncology treatment world is built on a shaky foundation of limited evidence and uncertain patient benefit, but rather than admit that, we keep tweaking it and building on top of it. And it’s not just that you are giving people things that don’t work; these drugs will harm them, and they are financially toxic.”

A Treatment World Built on Nothing

In defending dangling approvals, Richard Pazdur, a physician and the head of the cancer drug division at the FDA, wrote in 2021 that when a drug fails in a clinical trial, it “does not necessarily mean that the drug is ineffective.” In one sense, Pazdur is right. Not having convincing evidence that a drug works is not the same as knowing it does not work.

What the scenario does mean is no one knows if it works or not. This investigation’s database shows that hundreds of drugs are now being put on the market before anybody knows if they’re effective.

In the past, having a promising surrogate end point was used by drug companies to decide whether proceeding to a clinical trial was worth the investment. Now drugs are put on the market, sold to thousands if not millions of patients, and the evidence to determine if they actually work is allowed to come later, if it comes at all.

The arrangement can have devastating costs for people like Laura MacMillan.

Starting in 2001, Laura MacMillan was diagnosed with a series of puzzling medical conditions. First came interstitial cystitis. Two years later, when she was only forty-five, she developed severe abdominal pain and diarrhea that was diagnosed as colitis. “I could barely leave my house,” she says. “I had frequent accidents and could only go to stores where I knew there was a bathroom.”

Twelve years later, she began to see wavy lines as her vision progressively dimmed. One evening, she drove over a cement island in a parking lot, leaving her with a flat tire and $4,700 of damage to her car. Her ophthalmologist could find no known diagnosis for her eye condition and could offer no treatment. Between the colitis and progressing blindness, MacMillan’s world was closing in.

Finally, in 2021, MacMillan came across Nieraj Jain’s troubling findings of blindness linked to the interstitial cystitis drug Elmiron and instantly put two and two together. She had been on Elmiron for twenty years and had experienced the same retinal changes Jain uncovered. She stopped the drug, and within weeks, her colitis symptoms went away. A subsequent colonoscopy showed her gut had completely healed, but MacMillan’s vision loss is permanent.

“I can’t drive anymore,” she says. “I had to stop working thirteen years ago. When I get up in the morning, it can take about two hours for the blurriness to subside enough for me to see words on my tablet. Last summer, my husband and I drove [1,000 miles] to Toronto to see the Blue Jays play the Yankees. Even though we had good seats, I couldn’t see a damn thing. So we won’t ever be able to do that again.”

MacMillan is relatively lucky compared to the hundreds of others uncovered by this investigation who suffered serious injuries or died while taking the drug.

The FDA continues to approve medicines based on evidence as flimsy, contradictory, and inadequate as the data for Elmiron. Citing the need for “flexibility” and the importance of encouraging drug development when there are few treatment options, the agency has all but abandoned its hard-won standards for sound science, according to numerous experts.

Jerome Hoffman, professor emeritus of medicine at the University of California, Los Angeles, and the lead analyst of the database created for this investigation, says:

Most of us imagine that the primary goal of the FDA is to make sure that the drugs it approves are more likely to help people than to harm them. If so, the FDA would require drugmakers to submit rigorous studies. Instead, the agency seems to have forgotten about that goal and is more interested in promoting the interests of industry than protecting the public health.

In 2014, the Elmiron Study Group published their results of the long-awaited clinical trial that showed Elmiron was no better than a placebo. “I thought that the next day it would be over for Elmiron,” says Curtis Nickel, a urologist at Kingston Hospital in Ontario, Canada, who led the study group. “Doctors would stop prescribing it. The FDA would order it off the market.” But none of that happened.

Instead, after twenty-four years, Elmiron is still being sold, and doctors continue to prescribe it. Hundreds of thousands of patients have been exposed to the drug, and the American Urological Association lists it as the only FDA-approved medication for interstitial cystitis. The agency finally added a warning to the drug label in 2020, alerting doctors and patients to the risk of vision loss.

Nowhere is there any mention that studies have never shown Elmiron is effective.