The Food and Drug Administration is seldom accused of being too transparent. But in late September it looked like the agency had overshared. In an attempt to achieve the “greatest level of transparency,” the FDA caused the stock prices of four biotech companies to hemorrhage. Jittery traders, sifting through scraps of context-free information provided by the agency, dumped their drug stocks, triggering a brief but brutal plunge.

It shouldn't have happened that way. The FDA's attempt at transparency was far from revolutionary; it was the release of a Web interface to a drug side-effects database known as the FDA Adverse Event Reporting System (FAERS). Not only was FAERS already public, albeit in a slightly less user-friendly form, the database is also messy, context-free and subject to all sorts of biases and errors, making it nigh impossible to interpret properly. It's not something that the public should be very excited about; FAERS certainly isn't a precise enough tool to tell market speculators whether to dump a stock, any more than a sledgehammer is a precise enough tool to amputate a limb. Yet soon after the FDA announced the new interface, the stocks of four companies—Sarepta Therapeutics, Ionis Pharmaceuticals, Biogen and Acadia Pharmaceuticals—plunged.

The underlying cause of the bloodbath, ironically, is the FDA's opacity regarding certain important data about the performance of drugs. Data about adverse events patients had when taking the drug. Data that can give us insight into what elements of a clinical study researchers haven't made public. Even data hinting at research misconduct in key drug trials. The agency refuses to release these data, yet without them, the public is unable make an informed decision about whether or not to take a drug. The reason: doing so might hurt a pharmaceutical company. The FDA's refusal is a graphic demonstration of how the agency feels obliged to protect corporate secrets, even at the expense of consumer safety. That's the precise opposite of transparency, and in the FDA's case, it might be putting people's lives at risk.

The FDA has the unenviable position of making life-and-death decisions every day. No matter how good the agency is, some of the time it's going to get a decision wrong, and Americans will be put at risk and even die.

When things go wrong, though, it's often damnably hard to figure out precisely what happened. The FDA has long had a reputation for opacity; even congresspeople, who have the power of subpoena, have been frustrated by the agency's refusal to provide basic information that could help reveal the true story behind a bad decision. A decade ago Senator Charles Grassley of Iowa, when looking into a case where the FDA made a bad call in approving a dangerous drug, said the FDA put up “every excuse under the sun” to prevent the release of requested documents:

“The Department [of Health and Human Services] and FDA say ... they have been responsive to the Finance Committee's Ketek investigation because they made available millions of pages of documents to the Committee. But what they provided is quantity, not quality.

“They delivered hundreds of pages simply marked, for example, ‘57 pages removed,’ or ‘43 pages removed.’ ... Other documents have whole pages, paragraphs or sentences redacted with no explanation for what has been withheld or redacted and why. In fact, the FDA redacted some of the same documents differently and even redacted one of my own letters to them on a different matter ....”

Reporters like me have encountered similar roadblocks when covering the FDA. (For example, when I was trying to investigate how the agency was handling a massive case of fraud that undermined the data behind a number of approvals, the agency refused to release the names of the drugs that were affected.) On occasion, the agency is not just slow to release information about problems it is finding with drugs, but it has even been willing to reassure the public about products that later turn out to be dangerous. So the public can't rely on the FDA to release everything that is important to know, and objective data that the agency uses to make its decisions have become quite valuable for understanding not just the FDA's decision process but also whether it has been making good calls or bad ones.

It so happens that one of the most controversial calls of the past few years has to do with one of the companies involved in September's stock plunge, Sarepta. In September 2016 the FDA decided to approve Sarepta's first drug, eteplirsen. Eteplirsen is a cleverly designed compound that's supposed to help certain patients with Duchenne muscular dystrophy (DMD), a deadly disease that strikes young boys.

Initially the FDA's answer was that the drug should be rejected, but Janet Woodcock, the head of the agency's Center for Drug Evaluation and Research, overruled the FDA's own review team, a very unusual step, and declared that the drug should be allowed to come to market. Since then, a cloud has hung over eteplirsen as third parties, such as insurers, physicians and independent researchers, try to figure out whether or not the drug actually works.

The agency has released thousands of pages of information about eteplirsen, but I knew that there was a lot of critical information missing from those pages—information that might help determine whether the agency approved an entirely ineffective drug based on faulty clinical trials and undue industry influence or whether there is unjust suspicion about eteplirsen's safety and effectiveness.

Last May I sued the FDA under the Freedom of Information Act to understand the eteplirsen decision better; the lawsuit was designed to force the agency to release information about the drug and the FDA's decision. Late last year the agency released thousands of pages of previously undisclosed documents about eteplirsen and its approval. Despite the volumes of papers the FDA is disclosing, once again, the agency is far from transparent. What's so striking in those documents is not the information that the FDA is releasing but the information that it refuses to release.

For example, in several of the documents, frequently encountered adverse events—side effects and other negative consequences that occur during a treatment—are occasionally redacted. In some cases, other sources give us a hint of what these adverse events likely are. For example, one document states that “the most commonly reported [adverse events] included procedural pain, oropharyngeal pain, [REDACTED], cough, nasal congestion, and extremity pain.” Luckily, one table below a nearly identical redacted section is intact, and it implies that the censored portion is drawn from the following: hypokalemia (low levels of potassium), vomiting, “balance disorder,” headache, fever, back pain or a certain kind of blood clot known as a hematoma. (Vomiting and balance disorders are listed as possible side effects on eteplirsen's label.) In other cases, though, it's all but impossible to figure out what the FDA is attempting to block us from seeing: an updated listing of adverse events in a follow-up study says that “the most commonly experienced [adverse events] were procedural pain [REDACTED].” Also expunged from the document were possible indicators of kidney problems and issues related to blood clots.

The FDA has to make an active decision to prevent the public from seeing what's behind those black bars in the document. And it's not just side effects where the FDA seems determined to prevent the public from getting the full picture about the scientific case for eteplirsen.

A drug approval revolves around how patients perform with respect to so-called outcome measures in key clinical trials. Outcome measures are the yardsticks by which patient improvement is measured. For a muscle-wasting disease like DMD, there are many possible measures to choose from: how far a patient walks in six minutes, how long it takes to run 10 meters, how much time it takes to get up from the floor, and so on. But it's crucial to choose those yardsticks ahead of time and publish the results from all of them; otherwise it's possible to game the system by “outcome switching.” It's easy to make a worthless drug seem effective by hiding the outcome measures that don't show good results and publishing only outcomes that do. It's like going through a deck of cards and selecting only the ones you want; you're guaranteed a royal flush every time.

Eteplirsen researchers observed at least nine outcome measures designed to gauge patients' muscle strength and tone. The results from at least two of these measures have been buried: they're missing from the peer-reviewed literature. That's not unexpected; drug companies and researchers do this all the time. But it's surprising that the FDA would be complicit in hiding buried outcomes. The agency censors all reference to the results of those measurements and even to the names of the outcome measures that disappeared.

Pretty much every mention of those two measures is blanked out. Tables: censored. Sarepta's evaluation of these outcome measures: censored. Even tables of contents: censored.

From other sources, I have been able to piece together that the two missing outcome measures are a “nine-hole-peg test,” in which a patient is timed on putting pegs into holes, and “maximum voluntary isometric contraction testing,” or “MVICT,” which measures the force with which a patient pulls against a strap. The results for these tests are nowhere to be found, even though they've been in Sarepta's hands for years. All my requests for the results of these missing measurements—from the researchers and from Sarepta itself—have been refused. And this is even though eteplirsen researchers apparently “presented” the results in a poster session at a meeting last October—not long after the FDA started handing over documents in response to my lawsuit. (Another poster presented at the same meeting is on Sarepta's Web site, yet there's nothing on the missing outcome measures.) Smart money is that the results of the nine-hole-peg test and MVICT have been redacted because they hurt eteplirsen's cause rather than help it.

More evidence of outcome switching has to do with the number of certain types of white blood cells, known as CD3, CD4 and CD8 cells, found in each patient's muscle. It's not clear precisely what the researchers had in mind, because Sarepta's description of this “key secondary efficacy endpoint” on the national clinical trial registry Web site was vague. But we do know that sometime between July 2011, when the trial began, and July 2015, three years after it ended, the “key secondary efficacy endpoint” had mysteriously become a test of walking; the white blood cell outcome measure was nowhere to be found. (Neither the lead investigator of the eteplirsen trial nor Sarepta would answer questions about what the outcome measure was or what the results were.) The FDA knows the answers, but it ain't telling: the major references to the analysis of these white blood cells tend to have a big block of censored text where the results should be.

Outcome switching is a way for researchers and drug companies to distort the context around a clinical result, to make a drug look more effective or safe than it really is. The FDA's job is supposed to be exactly the opposite—to counter industry-distorted science and provide an objective measure of safety and effectiveness to help physicians make the best choices for their patients. Yet when it comes to eteplirsen, the FDA is siding squarely with the industry and against the public interest.

This is even true when it comes to allegations of outright fraud. One of the most alarming documents to come out of my lawsuit is a chain of e-mails in which an FDA reviewer suggests that Sarepta or eteplirsen researchers might be manipulating and misrepresenting scientific images. Of most concern are so-called Western blots.

Scientists use Western blots—which, when photographed, look like a bunch of messy stripes—to gauge the types and amounts of protein in a sample. Western blot images are ubiquitous in the medical and biological literature, but because they're such simple images, they're easy to fake, and blot fraud is surprisingly common.

The eteplirsen studies had Western blots—and those images raised the eyebrows of an important FDA reviewer: “There seems to be reason for concern of misrepresentation of the data,” he wrote. Apparently he was concerned that the images were misleading and perhaps even manipulated in an inappropriate way.

Nationwide Children's Hospital pediatrician and lead eteplirsen researcher Jerry Mendell denies allegations of manipulation of images. “The studies were FDA reviewed/audited [and the drug was approved] and the articles were peer reviewed,” he wrote in an e-mail. Sarepta refused to discuss any allegations of misconduct.

Who's right? It's impossible to tell. We can't say whether or not there's scientific misconduct without looking at the raw, unprocessed Western blot images and comparing them with the ones that are published. Mendell did not respond to requests for those unprocessed images. Nor did Sarepta.

But those originals are in other hands, too. The FDA has them, and it was a big fight to get them. After extensive negotiations through my lawyers, the FDA turned over the documents—the very morning that this article was going to press. It has yet to be seen whether, in fact, the allegations have merit, but it is clear that the FDA certainly hasn't brought this issue to the public's attention; on the contrary, when pressed, agency officials denied any suspicions of misconduct. (A different reviewer, shortly after eteplirsen was approved, said that he viewed the issue as “sloppy science” rather than misconduct.) And the FDA appears to be actively withholding similar data: another place where such raw images reside is redacted, as are major portions of the analysis that might cast light on how the data were processed. The FDA won't release them.

Why would the FDA, an agency trying to be more transparent, block evidence of outcome switching and even hide references to a medication's side effects? The reasoning takes a bit of unpacking, but it boils down to a simple principle: the FDA is refusing to release this information because it might hurt Sarepta, the maker of eteplirsen.

The FDA has stated that the redacted sections represent “trade secrets and commercial or financial information obtained from a person, and privileged or confidential.” In this particular case, this tends to mean that the release of the information will cause “substantial competitive injury” to the company that turned it over to the FDA.

Before releasing the documents, the agency allowed Sarepta (which is intervening in my lawsuit against the FDA) to suggest redactions that it felt would cause such harm or are exempt from release for other reasons. And sure enough, Sarepta thinks that releasing certain adverse events and endpoints will hurt Sarepta and help its competitors. (For example, Sarepta's present position is that releasing which endpoints were used, much less the results of the tests, would give “invaluable information to competitors.”) If the agency didn't agree—if it didn't think that Sarepta was correct—it would still be required by law to release this information or, at the very least, to come up with a different reason for the redactions. So to all appearances, the FDA believes that in these cases releasing this information will hurt Sarepta and refuses to turn it over.

That's it in a nutshell. The FDA is blocking access to very basic information about eteplirsen—censoring side effects and hiding evidence of missing outcome measures—because releasing that information would hurt Sarepta. (The FDA refused to answer any of my questions about its conduct, citing the lawsuit as a reason.) The public's interest in knowing the truth about a drug is secondary to the interest in protecting a company from harm.

This is toxic for our confidence in the FDA and in the drugs that it allows to come to market. It may well be that there's no real case for scientific misconduct in the eteplirsen clinical trial. It may well be that we already know about all the drug's important side effects. Heck, it's even possible that the censored and missing outcome measures strengthen the case for the drug's effectiveness rather than weaken it. But the FDA's willingness to consider such basic information about a drug's performance as a “trade secret” or “confidential commercial information” and block it from public view means that we won't—and can't—know. There's a haze of uncertainty around every single one of the FDA's decisions.

And this, ultimately, was the trigger of the stock plunge at the end of September. When the FDA made its adverse-events database easier to search, investors immediately started searching through it and turned up scary-seeming reports of deaths and injuries, which caused a loss in confidence about certain drugs—in Sarepta's case, about eteplirsen. (Sarepta's stock has since recovered.) Even though these adverse-events reports weren't terribly useful for evaluating the drugs' safety, every little scrap of new data can send shudders through a market starved for information.

Had the FDA been more scrupulous about serving the public's interest—sharing all information about adverse events, endpoint switching and even intimations of fraud—the market wouldn't have been so reliant on the noisy and hard-to-interpret data in the adverse-events database. The public would have much more confidence in an FDA that's truly transparent than one that is willing to call such information a “trade secret” or “confidential commercial information” and hide it from view.

In other words, it's impossible to trust an agency that worries more about a drug's side effect on a company than on a patient.

Since this article was published, the Food and Drug Administration released additional portions of documents to the author's legal team. With these releases, the FDA and Sarepta are no longer redacting indications that patients taking eteplirsen experienced hypokalemia (abnormally low levels of potassium), or that researchers studying the drug measured patients' scores on the so-called 9-hole-peg test and MVICT muscle test. However, the agency is still redacting what those measurements showed, as well as other elements in the documents describing outcome measures and adverse events. 

The agency released images which accompanied a FDA reviewer's e-mail to a colleague that "There seems to be reason for concern of misrepresentation of the data" regarding eteplirsen. The images seem to support the reviewer's statement that the images in the raw gels do not always seem to correspond to their supposed equivalent in a public presentation given by Sarepta's then-CEO Ed Kaye in 2011.


Experienced hypokalemia:

9-hole-peg test and MVICT:

What those measurements showed:

Describing outcome measures:

And adverse events:

The images:

Public presentation: