Twelve years ago, Ronald Kavanagh—who at that time was a reviewer of psychiatric drugs for the FDA—turned into a whistleblower, telling the Office of the Inspector General that his superiors at the FDA were turning a blind eye to the risks of a new atypical antipsychotic, asenapine, and doing so in collusion with Schering-Plough, the company that was seeking to market the drug. Asenapine, he argued, was ineffective as a treatment for schizophrenia and for bipolar I patients with mild to moderate symptoms, and yet approval would expose those two groups of patients to its hazards, which included a heightened risk of death.
Kavanagh was soon fired for his efforts, and asenapine was approved as a treatment for both schizophrenia and bipolar I. Schering-Plough’s CEO, Fred Hassan, used the impending approval of the drug to negotiate a merger with Merck, a deal that brought him and other Schering-Plough executives more than $100 million.
Kavanagh has never given up on that whistleblower complaint. He filed a qui tam lawsuit in 2012, and in the years that followed, wrote letters to President Obama, President Trump and Senator Charles Grassley, none of which provided him with any relief. In May of this year, he renewed his complaint once more, writing members of Congress and the Office of the Inspector General.
It might seem that there is little reason to revisit his complaint, given that it has never gained any traction in Congress, the OIG’s office, or the courts. It is unlikely that it will fare any better this time around. However, a review of documents in this case, including the FDA’s reviews of the New Drug Application (NDA) for asenapine, reveals much about the mindset of the FDA at the time and the standards it applied for approving a psychiatric drug. The documents support Kavanagh’s complaints, including evidence that the FDA downplayed—or even obscured—risks with the drug.
Kavanagh’s complaint also alleged that asenapine’s potentially lethal side effects were common to other atypical antipsychotics that, like asenapine, had proven to be ineffective for bipolar patients with mild to moderate symptoms. He warned that the use of atypicals would lead to 5,000 or more deaths each year in such patients, and there is evidence that this concern has proven true. He also warned that asenapine and other atypicals could cause deaths in newborns whose mothers were exposed to these drugs during pregnancy, and FDA Medwatch records reveal that there have been a number of infants under two years old who died due to exposure to an atypical antipsychotic.
Asenapine was synthesized by the pharmaceutical company Organon in the 1980s. After Organon conducted early-phase testing of the drug, in 2003 it entered into a co-marketing agreement with Pfizer, which then took the lead in conducting phase III trials of asenapine as a treatment for both schizophrenia and bipolar I disorder. Pfizer had paid Organon $100 million when it negotiated the agreement, but after analyzing the phase III results, which didn’t provide “conclusive” evidence of efficacy, it pulled out of the deal in November 2006.
However, Organon continued to develop the drug, and in the spring of 2007, Schering-Plough’s CEO, Fred Hassan, negotiated a deal to buy Organon from its parent company, AkzoNobel, for $14.4 billion. Organon had several compounds in late-stage development, and Hassan perceived asenapine as of particular value since Organon was preparing to file a New Drug Application for it. Organon filed its NDA on August 31, 2007, and in November, Schering-Plough completed the deal.
At that point, Hassan needed the FDA to approve asenapine, and even as he was closing the Organon purchase, he began poking the FDA to do just that, telling Fortune magazine that “when bureaucrats come under pressure, they tend to choose the path of asking for more data, as opposed to approving the drug.” His political message was clear: An FDA that didn’t quickly approve new drug applications was throwing impediments in the way of American business.
Hassan’s Record as a CEO
Prior to his acquisition of Organon, Fred Hassan had been feted for turning around the fortunes of two pharmaceutical companies he’d led: Pharmacia and Schering-Plough. In both instances, he and his companies did so by hiding adverse effects of their best-selling drugs, which subsequently led to a Congressional investigation, lawsuits, and multimillion-dollar settlements and fines.
Hassan was hired as CEO of Pharmacia and Upjohn in 1997. Two years later, he negotiated a merger with Monsanto, which had a hot new drug that had just come to market, Celebrex. This was the drug that could remake Pharmacia.
Monsanto had been co-marketing Celebrex with Pfizer, and now Pharmacia, together with Pfizer, told the public that this Cox-2 inhibitor, an NSAID for treating pain, didn’t cause an increased risk of cardiovascular events, including heart attacks and strokes. The two companies made this claim even though results from a clinical trial, which they had known about since 1999, had shown otherwise. The hiding of this potentially deadly risk proved profitable for both, with Celebrex sales totaling $10.2 billion from 1999-2003.
The two companies brought a second Cox-2 inhibitor (Bextra) to market in 2002, and once again they told the public of a drug that didn’t elevate the risk of cardiovascular events. Hassan successfully sold Pharmacia to Pfizer for $62 billion at the end of 2002.
A little more than a year later, it became known that the two companies had hidden the cardiovascular risks from the public. Bextra was pulled from the market, and the FDA, in 2005, put a black-box warning on Celebrex. Pfizer then spent years fighting lawsuits, eventually paying $894 million to patients harmed by the two drugs; another $486 million to investors who bought Pharmacia and Pfizer stock from 2000 to 2003; and a $2.3 billion fine to the federal government to settle criminal charges for its fraudulent marketing of Bextra and other drugs.
After negotiating the sale of Pharmacia, Hassan—hailed by Financial Times as CEO of the Year in 1999 for his success at that company—was hired by Schering-Plough in April of 2003 to be its CEO. This was seen as a difficult period for Schering-Plough, as the patent on its best-selling drug, Claritin, had run out in December of 2002. The company’s future was now seen as tied to Zetia (ezetimibe), a cholesterol-lowering drug that had been approved recently by the FDA. “Mr. Hassan’s success may depend on his ability to hawk Zetia,” the Wall Street Journal reported.
Statins like Lipitor and Crestor, which were billion-dollar drugs, inhibited the body’s production of LDL cholesterol, which in turn had been shown to slow atherosclerosis, the buildup of plaque on arterial walls that leads to heart attacks and strokes. Ezetime worked in a slightly different manner. It inhibited the absorption of intestinal cholesterol. While this lowered LDL cholesterol levels, it wasn’t as effective as statins in this regard, and it also hadn’t been demonstrated that this method of lowering LDL cholesterol slowed atherosclerosis.
Shortly before Hassan’s arrival, Schering-Plough had joined with Merck to launch a clinical trial called ENHANCE, which was designed to test whether a drug that combined Zetia with Merck’s Zocor (simvastin) would prove to be more effective than a statin alone in reducing the buildup of plaque. Simvastin was a statin that had gone off-patent, and their hope was that the trial would help make their combination drug, Vytorin, the No. 1 cholesterol-lowing medication. In 2004, before the ENHANCE trial had concluded, the companies obtained FDA approval to market Vytorin based on its cholesterol-lowering effects.
In late 2005, the two companies analyzed results from early enrollees in their trial, and found that there was no difference between the two arms of the study. The blind had not been lifted, but with this result it didn’t matter which arm was the combination drug and which was the statin alone—the fact that there was no difference between the two arms meant that the addition of ezetibime to simvastin had provided no benefit. The trial was completed in April of 2006, and not long after that, Schering-Plough and Merck definitively knew that Vytorin had failed the test.
ENHANCE was a high-profile study, and the cardiology community, as well as the investment community, expected that results would be announced at the fall 2006 meeting of the American Heart Association. That meeting passed without an announcement, and so too the association’s spring and fall meetings in 2007. During those 18 months, Hassan and Schering-Plough, along with Merck, busily promoted Vytorin’s effectiveness.
“The lower (cholesterol) is better story continues,” Hassan told investors in the fall of 2007. “Evolving medical science continues to find that reaching lower and lower goals for LDL is better for patients and Vytorin and Zetia provide very good options.”
The two companies filled the airways with ads in 2006 and 2007 that told of how Vytorin had been proven in clinical studies to lower cholesterol more than Pfizer’s Lipitor and AstraZeneca’s Crestor. The public pronouncements worked: Vytorin and Zetia generated $3.87 billion in global sales in 2006 and $5.2 billion in 2007, which produced nearly 70% of Schering-Plough’s profits that year.
Hassan was richly rewarded for this commercial success. In addition to his regular salary, his CEO contract provided for various cash and stock awards based on the company’s financial performance in 2007. Law firms that subsequently sued Schering-Plough and Merck for fraud calculated that Hassan received performance bonuses in 2007 worth $38.9 million. Other executives also received million-dollar bonuses, with one executive selling $28 million of Schering-Plough stock while its price was flying high.
However, Hassan and the other executives at Schering-Plough and Merck couldn’t keep this fraudulent story aloft forever. When the two companies didn’t announce results of the ENHANCE trial at the fall 2007 meeting of the American Heart Association, Congress initiated an investigation, asking for them to produce records by December. As a first response, the two companies—as the Wall Street Journal later reported—”created minutes of a crucial meeting about a major study on their cholesterol drug” to cover their tracks. Finally, on January 14, 2008, the company announced “preliminary results” from the ENHANCE trial, acknowledging that Vytorin had not shown any “statistically significant benefit” over placebo.
The attorneys general for New York and Connecticut announced they were launching an investigation, and then, on March 31, 2008, the two companies presented the results at a conference of the American College of Cardiology, with the New England Journal of Medicine publishing them the same day. Vytorin had produced “no result—zilch,” said principal investigator John Kastelein. “In no subgroup, in no segment, was there any added benefit.”
Five years later, Merck paid $688 million to settle lawsuits from investors who bought stock in one of the two companies while they hid the results from the ENHANCE study. But by that time, Hassan—with the many millions in his pocket from his performance awards—had moved on, his sights now set on getting asenapine approved. This drug, he told investors, was going to be a “blockbuster.”
Raising Red Flags at the FDA
Ron Kavanagh, who had come to work at the FDA in 1998, was an expert in clinical pharmacology, the study of how chemical agents are absorbed and metabolized, and their possible toxicities. He had earned his PharmD at the University of Texas and a PhD at the University of Washington, studying drug kinetics and pharmacodynamics. After finishing that training, he went to work for Merck in its division of internal regulatory affairs. “My job was to get drugs approved, and approved with regulatory agencies all over the world,” he said.
After he joined the FDA, he often gave talks and presentations to FDA reviewers and medical staff on how pharmaceutical companies developed drugs and the pharmacology tests a company needed to perform.
His first years at the FDA taught him to be wary of investigational new drugs. He started off in the agency’s endocrine division, where he worked with a colleague who had picked up on toxicities caused by a duo of obesity drugs, fenfluramine and dexfenfluramine, which led to their being banned in 1997.
Next he moved into the gastrointestinal division, where he reviewed Lotronex (alosetron hydrochloride), which was to be marketed for irritable bowel syndrome in females. Kavanagh thought there was evidence that the drug could cause ischemic colitis, a potentially fatal side effect. Although the drug was approved in 2000, it was recalled from the market a year later after its use led to five deaths.
The FDA’s recall of the drug, wrote Lancet editor Richard Horton, “reveals not only dangerous failings in a single drug’s approval and review process but also the extent to which the FDA—its Center for Drug Evaluation and Research in particular—has become a servant of the industry.”
Kavanagh then transferred into the division of neuro and psychiatric drugs, where he reviewed Eli Lilly’s NDA for Cymbalta (duloxetine). He warned of possible liver issues that could be expected with this drug, a concern that Eli Lilly “didn’t appreciate,” he said.
Next, he reviewed bifeprunox, an atypical antipsychotic that was similar to aripiprazole in its mechanism of action, as it had a mixed agonist/antagonist effect on both dopamine receptors and serotonergic receptors. However, bifeprunox caused a brain swelling and other toxicities that led to deaths in the clinical trials, and in 2007, thanks in part to Kavanagh’s review, the drug was not approved.
Even so, by that time, Kavanagh was becoming persona non grata at the FDA. Major companies had often complained about his critical review of their drugs, and he had also grown increasingly critical of the FDA. In 2005, he started taking his allegations to Congress, speaking to the Senate Finance Committee about “corruption in the psychiatry division and in the Office of Clinical Pharmacology and other FDA offices.” He also was upset with the off-label marketing of psychiatric drugs, particularly for their use in children, and in 2007, he was briefly suspended and told to stop his whistleblower efforts if he wanted to keep working there.
After Organon filed its NDA in August 2007 (Schering-Plough had not yet closed on its acquisition of the company), the FDA was required to assess whether the company had conducted the necessary pharmacological studies—a drug’s bioavailability, how it was metabolized, the possible toxicities of the metabolites and so forth—that would allow for a substantive review. In October, Kavanagh recommended that because of a “lack of validation of assays in pharmacokinetic studies,” the NDA was non-reviewable.
This turned out to be the first shot in what would become a war with his superiors at the FDA over asenapine. The director of the psychiatry division at the FDA’s Center for Drug Evaluation and Research, Thomas Laughren, overruled him, which meant that the review process—evaluating the efficacy and safety results from the trials—would now begin.
The Efficacy Data
Organon conducted four six-week trials of asenapine as a treatment for schizophrenia. The patients recruited into the trials were withdrawn from whatever antipsychotics they had been taking and randomized either to asenapine, placebo, or a comparator drug. The primary endpoint was reduction in symptoms as measured by the Positive and Negative Syndrome Scale. PANSS assesses 30 symptoms on a scale of 1 to 7, which means that total scores can range from 30 to 210. The patients could also be given concomitant medications during the study: zolpidem, zalepam, chloral hydrate, a benzodiazepine, and anticholinergic medications to treat extrapyramidal side effects.
In the four trials, there were a total of six tests of asenapine versus placebo: three tests of a 5 mg dose (administered sublingually twice a day); two tests of a twice-daily 10 mg dose, and one test of a flexible dosing schedule (from 5 mg to 10 mg). In four of the six tests, asenapine failed to beat placebo. In the fifth, a 5 mg dose did provide “statistically significant” superiority over placebo, but the comparator drug—risperidone—did not, and when the comparator drug in a trial fails to best placebo, the study is deemed to have failed. As a 2008 paper published in the journal Psychiatry explained, “No conclusion can be drawn about the investigational drug in a failed study.”
Thus, there was only one instance—out of six tries—where there was a “meaningful” positive finding for asenapine. And in that instance, the decrease in PANSS was 16.2 points for the 5 mg dose, 14.9 for the 10 mg dose, 15.4 for haloperidol, and 10.7 for placebo. Although there was virtually no difference in symptom reduction between the 5 mg and 10 mg doses of asenapine (1.3 points on a 210-point scale), the 5 mg dose just made it over the line for “statistical significance” over placebo, whereas the 10 mg dose did not. A total of 70 patients in the group of 114 randomized to the 5 mg dose completed the six-week trial.
Such was the efficacy data. The fact that concomitant medications were allowed—three different sedatives, an antianxiety medication, and a treatment for Parkinsonian symptoms—also meant that there had been no testing of asenapine as a monotherapy for psychotic symptoms.
Finally, researchers have determined that there needs to be at least a 15-point difference on the PANSS scale between the drug and placebo at the end of the trial for a drug to provide a “clinically meaningful” benefit. Yet, in the four clinical trials, there was not a single instance where any dose of asenapine came close to this standard.
Organon conducted two three-week studies of asenapine, both at a flexible dose from 5 mg to 10 mg (twice daily), as a treatment for manic and mixed episodes in bipolar 1 patients. The primary endpoint was reduction in symptoms on the Young Mania Rating Scale (YMRS). This is a 60-point scale, and in this study, those enrolled had to have baseline scores of greater than or equal to 20.
In both studies, asenapine was seen as providing a statistically significant benefit over placebo. The difference between the two groups in one study was 3.7 points on the YMRS scale and, in the other, 5.3 points. However, in both studies, the comparator drug, olanzapine, provided a greater reduction of symptoms than asenapine.
As for the real-world benefit of this drug for bipolar, researchers have concluded that a 6.6-point difference on the YMRS scale is needed for a drug to provide a “clinically meaningful” benefit over placebo. Asenapine did not reach this standard in either trial.
Summary Safety Data
In the six placebo-controlled studies (four for schizophrenia and two for bipolar), there were 2,251 patients who had been exposed to asenapine. Eleven died, and 14% suffered a serious adverse event (life-threatening or requiring hospitalization). Seventy-eight percent experienced an adverse event of some type; the most common side effects were sedation, dizziness, weight gain, extrapyramidal symptoms, and oral numbness.
Organon had also conducted a number of non-controlled studies, and thus the complete safety database consisted of 3,457 patients exposed to asenapine. Twenty-two had died, including eight by suicide. There were eight who died from cardiac and respiratory events, and one neonatal death. The baby died hours after being born prematurely at 32 weeks to a woman in the trial who had been treated with asenapine.
Although Kavanagh had earlier declared that Schering-Plough had not provided sufficient information to conduct an adequate pharmacology review, in the spring of 2008 he was assigned the responsibility for doing just that as part of the FDA’s overall assessment of asenapine’s efficacy and safety.
Early on, Kavanagh noted that the efficacy data for schizophrenia was unconvincing, particularly when the four studies—and six tests of asenapine—were viewed as a collective body of evidence.
At first glance, it did seem that the results from the bipolar studies were convincing, and easily passed the “two positive tests” standard for approval at the FDA. However, Kanavagh further analyzed the two bipolar studies, dividing the patients into “quintiles” based on the severity of their symptoms at baseline, and he found that in patients with mild to moderate bipolar symptoms (YMRS score less than 27), the drug had not provided any benefit. The rate of improvement in the placebo group and this subgroup of bipolar patients was virtually identical during the three weeks.
As such, the risk-benefit equation for schizophrenia patients and for around 50% of all bipolar I patients was evident. They would be exposed to the hazards of this drug without any benefit, and thus the treatment was certain to do more harm than good.
There were many evident worries in the safety database: cardiac risks, liver toxicity, kidney impairments, and so forth. And there was one other concern new to Kavanagh. He had been talking to a former FDA reviewer who had gone to work for industry and had experience with asenapine, and this “informant” had warned him that it was problematic because it acted as an “inverse agonist” of the 5HT2B receptor (a receptor for serotonin). Stimulation of this receptor by fen-phen had been identified as a possible cause of pulmonary arterial hypertension (PAH) and cardiac-valve stenosis. This led Kavanagh to worry that PAH could lead to miscarriages and an increase in neonatal deaths for babies that breast fed.
Kavanagh then focused his attention on the death records from the asenapine studies, reviewing the case report form for each one. These are the forms that the investigators at each site fill out for every patient, which are then summarized by the pharmaceutical company into a safety summary in its NDA. Kavanagh also revisited the animal studies and healthy volunteer studies that are expected to identify possible drug toxicities.
This investigation led him to identify a number of safety concerns. He also discovered specific instances where it appeared that Schering-Plough had obscured the possible risks with asenapine, or covered them up.
Although the case of neonatal death appeared in the safety summary, it hadn’t been flagged in the FDA’s preliminary medical reviews as a cause for real concern. The mother was said to have had previous miscarriages, and the thought was that this explained the infant’s death, even though the investigator who’d treated the woman had stated that the baby’s death was “possibly related” to asenapine.
The animal studies provided reason to think that might be the case. In both rats and rabbits, exposure to asenapine during pregnancy had led to an increase “in post-implantation loss at all doses,” meaning that there was an increase in deaths of the fetal pups prior to birth. In the rat studies, there was also a marked increase in “early pup deaths” through “day 21 postpartum.”
In addition, the animal studies warned of other congenital risks—abnormalities in bone formation and connective tissues, and—in one pup—a heart defect.
These findings had not been mentioned in the safety reviews Kavanagh had read up to that point. This was, Kavanagh thought, an instance where a possible toxic risk had been deliberately obscured when Schering-Plough filed its NDA.
There were eight patients exposed to asenapine who had died from cardiac and respiratory events. Although the FDA, in its preliminary reviews, didn’t identify these deaths as possibly related to asenapine, the drug had been found to cause QT prolongation in some patients, which meant that their hearts were taking longer than normal to recharge between beats. This was a warning signal of possible cardiac harm.
As Kavanagh dug through the healthy volunteer studies, he found that in one instance a 27-year-old man, described as fit and a non-smoker, went into cardiac arrest after being given asenapine. “There is no doubt that he became asystolic and as you pointed out you had to perform cardiac massage to sustain an output,” wrote the company’s cardiologist, Graham Jackson, in a December 1991 report. “This almost certainly has to be classified as a drug-induced effect with a serious adverse effect on the conducting system of the heart.”
Yet, Schering-Plough had described that incident differently in its summary of the healthy volunteer tests. The description there told of a 27-year-old man who had suffered an episode of “neurally mediated reflex bradycardia,” which, one FDA reviewer wrote, “is a benign, self-limiting event, and the most common cause of vasovagal syncope.”
These descriptions told of two very different events. Asystole is described in medical dictionaries as a “cardiac arrest rhythm with no discernible electrical activity on the EKG monitor. The heart is not functioning. It is a life-threatening condition that requires immediate action.” Vasovagal syncope is described in medical dictionaries as what “occurs when you faint because your body overreacts to certain triggers, such as the sight of blood or extreme emotional distress.”
To Kavanagh, this was a second instance of corruption: a cardiac arrest in a healthy volunteer had morphed into a fainting episode in the NDA.
As Kavanagh went through the case report forms for the 22 deaths, he found one that told of a 57-year-old woman with schizophrenia who, after being treated with asenapine for 470 days, suffered a bout of “Quincke’s edema.” This is an allergic reaction that causes a swelling in the back of the throat or uvula which can close the upper airway, and the case report told of the woman having been on a ventilator for four days after she suffered this “respiratory failure.”
However, the safety summary for this woman didn’t tell of a possible allergic reaction. It simply stated that she developed “sudden respiratory failure . . . the cause of death was pulmonary embolism.”
There were eight suicides in the 3,451 patients who were exposed to asenapine in the short-term and long-term studies of the drug (compared with no suicides in the 706 placebo patients in the short-term trials.) Five of the eight had occurred within 33 days of initial exposure to the drug: two by hanging, two jumped to their deaths, and the fifth by an unknown method.
This rash of suicides within a month of exposure to asenapine could be seen as a signal of suicide due to drug-induced akathisia. Akathisia is a known risk factor for suicide, and the safety summary in fact stated that 1 in 16 patients exposed to asenapine had experienced this adverse effect.
Moreover, Kavanagh reasoned, this drug acted on the serotonergic system, as did the SSRI antidepressants that had been found by the FDA to increase the risk of suicidal behavior in children and adolescents. This was another concern to be flagged and further investigated.
However, FDA reviewers, following the lead of Schering-Plough’s safety summary, were treating these suicides as a risk inherent with the “disease” (schizophrenia and bipolar). There had been four suicides in those treated with olanzapine, and once the suicide rate was calculated according to “patient-years” on each drug, it was the same for both: 1.3 suicides per 100 patient-years. Olanzapine had been on the market for more than a decade, and thus suicide didn’t appear to be a worry particular to this drug.
Is This a Class Effect?
After Kavanagh assessed the efficacy of asenapine for bipolar, finding that it provided no benefit for curbing mania in bipolar I patients with mild to moderate symptoms, he looked at the efficacy data for several other atypicals that had been tested for bipolar I: Zyprexa, Risperdal, Geodon, and Invega. He found that the pattern held true for those drugs as well: They were ineffective for those with mild to moderate symptoms.
As such, Kavanagh now saw a review process that had led to several atypicals being approved for a diagnosis even though 50% of bipolar I patients would receive no benefit and be exposed to the many hazards of these drugs. These drugs had a black-box warning that they increased the risk of dying in the elderly, and it was also well known that atypicals could induce metabolic dysfunction that increased the risk for early death. At the very least, Kavanagh reasoned, the FDA should put this information—that asenapine was ineffective for those with mild to moderate bipolar I symptoms—on the label, and add it to the labels for the other atypicals as well.
Kavanagh Blows the Whistle
Kavanagh’s clinical pharmacology review of asenapine was due on May 15, 2008, and at least a week before that he had sounded an alarm, stating in a large office meeting that he thought this drug was dangerous and largely ineffective, and that Schering-Plough had sought to hide the risks. While his 520-page report was filled with technical details regarding the drug’s bioavailability, how it was metabolized, and so forth, Kavanagh did have a personal history that was helping to fuel his opposition to this drug.
By his own admission, he suffered from “mental illness.” He had suffered more than a dozen episodes of depression, including one hospitalization, and he hadn’t found antidepressants to be particularly helpful. When he was about 40 years old, he told a psychiatrist of a time in his life when he had fallen in love, describing it as a period of great happiness. The psychiatrist viewed this as evidence of mania and newly diagnosed Kavanagh as bipolar. While Kavanagh didn’t agree with the diagnosis and didn’t take any bipolar medication, he spoke about this personal vulnerability when he raised his complaints about asenapine at the FDA.
As he later wrote: “I also said that the drug, as well as others, could be used not only for what it was currently under development for, but in the future could also be approved for illnesses that I or my child suffered from, or that I had been misdiagnosed with (which is common for people with psychiatric illnesses), or that it could be promoted for use or used off-label for our illnesses and so we were personally affected by this.”
On May 16, he wrote his superiors at the FDA that he was changing his recommendation to “non-approval.” He was worried about the deaths in the trials, and stated that he believed that this toxicity was “mediated by agonism of the 5HT2B receptor.” Moreover, he wrote, there was evidence that the “sponsor knew of this toxicity and specifically tried to prevent our detecting it.” And finally, there was the harm that would come from off-label uses of asenapine.
“I simply do not believe that there is anything that we can do that will adequately educate physicians and patients to the risks and that with off-label use we will be looking at an epidemic of potentially lethal cardiac and pulmonary toxicities in children several years from now,” he wrote.
A week later, with his emotions ramping up into higher gear, he emailed Janet Woodcock, the director of the Center for Drug Evaluation and Research, to inform her of an “imminent public health concern” related to asenapine and the prescribing of already approved atypicals to bipolar patients with milder symptoms.
He warned her that he was ready to go outside the agency with his complaints. “I have notified the clinical division of apparent criminal activity by the sponsor (i.e. not reporting Serious Adverse Events and deaths as required), and have requested a criminal investigation,” he told Woodcock. “Yet the clinical division has apparently tried to cover this up.”
That last sentence was sure to land him in hot water with his colleagues, and on May 30, Kavanagh wrote Woodcock that his superiors were “bullying him” and that he no longer trusted the agency’s internal “process” for responding to his complaints. From this point forward, he told her, “I prefer to deal with Congress.”
The agency, he told her, was failing to do its job.
“I don’t like what has happened over the past few years. Based on my experience I believe the review process has gone (at least at the reviewer level) from collegial, collaborative, efficient and trying to make to balanced decisions to being bogged down in bureaucracy, reviewers just cutting and pasting what the sponsors say without critical evaluation, being grossly inefficient, cutting corners just to make deadlines and to avoid being retaliated against, and everyone being defensive.”
Kavanagh contacted the Office of the Inspector General and Iowa Senator Charles Grassley. In June, he filed two amendments to his May 15 review, further detailing and documenting his concerns. His colleagues at the FDA, he wrote on June 30, were “complicit” with Schering-Plough in criminal activity.
By then, he knew that his future with the FDA was over. “When I began raising these safety issues in the beginning of May, I knew this was the end of my career,” he said.
Thomas Laughren’s FDA
By going to Congress to blow the whistle, Kavanagh was presenting the agency’s impending approval of asenapine as a deviation from the norm. Yet, there was a 20-year record of psychiatric drug approvals that showed that this was not the case. Under the leadership of Thomas Laughren, the FDA had a history of approving psychiatric drugs that were, at best, of marginal efficacy, and a history of doing so knowing that the sponsor company had sought to hide, or at least obscure, the drug’s side effects.
Given this history, which has been well documented, Kavanagh’s accusations in this case were really a charge against the FDA’s standard operating procedures, at least in the psychiatry division. And the asenapine NDA wasn’t a particularly egregious case.
Laughren had come to the FDA in 1985, first serving as a team leader in the Division of Neuropharmacological Drug Products. As Laughren later wrote, in this position he “directly supervised the medical officers involved in the review of all drug development activities conducted under INDs and the review of all NDAs and supplements for new psychiatric drug claims.” In 2005, he became the director of the Division of Psychiatry Products, where he continued to oversee the “review of all NDAs.”
The first NDA that Laughren would have reviewed in his team leader position would have been for Prozac (fluoxetine). While this drug was promoted to the public as a breakthrough medication for depression following its approval in late 1987, the clinical trials told a different story altogether.
Early during the clinical trials of fluoxetine, it was found that it could induce an intense agitation in some patients. In response, Eli Lilly amended the trial protocols so that patients could be given a benzodiazine as a concomitant medication. As Eli Lilly’s Dorothy Dobb later admitted in a legal case, this was “scientifically bad,” since it would “confound the results” and “interfere with the analysis of both safety and efficacy.”
Eli Lilly conducted five clinical studies of its drug, and when the results from all five studies were pooled, the improvement on HAM-D scores—the scale used to assess depressive symptoms—was only one point greater for the fluoxetine patients than for the placebo group, a difference that is meaningless. In addition, Eli Lilly had engaged in various coding shenanigans, directing its investigators to record drug-related adverse events as “symptoms of depression,” and changing reports of “suicidal ideation” on case report forms to “depression.”
The FDA’s medical reviewers spotted much of this deception. Eli Lilly, wrote FDA reviewer David Graham, had engaged in “large-scale underreporting” of the harm that fluoxetine could cause.
Even so, the FDA gave fluoxetine its approval, and the SSRI explosion was on. Other companies developed similar SSRI antidepressants, and the FDA kept to the standard it had set with fluoxetine. Pfizer conducted six clinical studies of sertraline (Zoloft), and in four of the six, it failed to beat placebo. There was a fifth study that was “questionable,” and a sixth that was positive for sertraline. As a Pfizer employee confessed in an April 11, 1991 memorandum, sertraline had “received an unfavorable review in a number of countries. The common key issue is that regulators are not convinced of sertraline’s efficacy versus placebo.”
But the FDA gave Pfizer the green light, and soon Zoloft was on its way to becoming a best-selling drug. The efficacy data for the SSRIs that followed were much the same.
Next up were atypical antipsychotics, with risperidone (Risperdal) and olanzapine (Zyprexa) leading the charge. Johnson & Johnson, in its risperidone trials, compared multiple doses of risperidone to a high dose of haloperidol (the comparator drug), a design that allowed Janssen to cherry-pick the best results from one of the three risperidone doses and compare it with a dose of haloperidol certain to cause many adverse events. This was a biased design that could be expected to make risperidone look good, and FDA reviewers weren’t fooled. As they noted, these studies were “incapable of providing any meaningful comparison of the two drugs.”
Similarly, FDA reviewers concluded that Eli Lilly’s trials of olanzapine were “biased by design” against haloperidol, and thus its large phase III trial, which wasn’t placebo controlled, provided “little useful efficacy data.” Twenty patients treated with olanzapine died in the trials, and the FDA’s Paul Leber warned that that “no one should be surprised if, upon marketing, events of all kinds and severity not previously identified [in the trials] are reported in association with olanzapine’s use.”
This was how testing of psychiatric drugs and their approval took shape during the first decade that Laughren served as a team leader in the FDA’s Division of Neuropharmacological Products. Drugs that showed little efficacy and were tested in biased trials were approved, even as risks were minimized or obscured, and then they were promoted as breakthrough medications, which rang up billions in sales.
Many of the pharmaceutical companies then aggressively marketed their drugs for off-label use in children and an ever-expanding percentage of adults. The FDA mostly stood by as this happened, and what can be understood today is that the selling of psychiatric drugs during this period evolved into a fraudulent enterprise. Eli Lilly, Johnson & Johnson, GlaxoSmithKline, Astra Zeneca, Abbot, Bristol Myers Squibb, Park Davis, Forest Laboratories, Novartis, Warner-Lambert, and Otsuka all eventually paid fines to the federal government for improper off-label marketing of their psychiatric drugs, with several pleading guilty to criminal charges.
Yet, even while he wore his FDA hat, Laughren helped promote psychiatric drugs and their off-label use. In a 2007 complaint to FDA Commissioner Andrew von Eschenbach, the Alliance for Human Research Protection (AHRP), a watchdog organization, prepared a detailed list of Laughren’s “collaborative ties with pharmaceutical industry officials and industry-financed psychiatrists in academia and professional associations.”
AHRP provided links to sourced documents in its complaint. Laughren’s activities included:
- Participating in an industry-sponsored consensus panel convened by the American Academy of Child and Adolescent Psychiatry that recommended off-label uses of psychiatric drug in children.
- Co-authoring more than a dozen articles with industry-funded “thought leaders,” who were being paid to act as consultants, advisors, and speakers. In one instance, he co-authored a book chapter with the chief medical officer of Eli Lilly, Leigh Thompson.
- Serving on a “development panel” for a “Mood Disorders” conference underwritten by major pharmaceutical companies. The 2002 report endorsed depression assessments for a broad range of patients with physical ailments—cardiovascular disease, cancer, Parkinson’s, AIDS, and others—that was sure to expand the market for psychiatric drugs.
Laughren also earned the ire of many parents when it became known that he had dismissed concerns about SSRIs stirring suicidal impulses in children. In 1996, AHRP’s complaint noted, a medical review officer at the FDA had reported a seven-fold greater incidence of suicidality in children prescribed sertraline. In response, Laughren wrote that “I don’t consider these data to represent a signal for suicidality in either adults or children.”
Then, in 2004, Laughren prevented an FDA reviewer, Andrew Mosholder, from speaking at a public hearing on whether antidepressants increased the risk of suicide in children and adolescents. Mosholder had concluded that trial data showed that to be so, but Laughren and other FDA officials, the Wall Street Journal reported, feared that he would “condemn the drugs too strongly before the advisory committee.” Laughren presented Moshholder’s analysis, but “stressed the unreliability of the data instead of the possible risk from the drugs.”
AHRP’s complaint detailed other instances where Laughren had worked to get a drug approved that other FDA reviewers had raised red flags about. In 2007, the FDA’s medical officers recommended that Eli Lilly’s application for pediatric use of Zyprexa should be rejected, both because of safety concerns and because nearly half of the data had come from Russia, which the FDA officers suspected might be fraudulent. Yet on April 29, 2007, Laughren overruled them and deemed Eli Lilly’s application for pediatric use “approvable.”
Such were the prevailing standards for approving NDAs of psychiatric drugs under Laughren’s leadership. There was a history of the agency bending over backward to find evidence of “efficacy,” tolerating trials that were “biased by design,” and letting sponsors get by with safety reports that, in one manner or another, sought to obscure the possible harms their drugs could cause.
While Laughren came in for a fair amount of public criticism, he was, in his efforts to get new drugs approved, following a congressional mandate that emerged in the 1990s. During the Reagan era, conservative groups regularly complained that the FDA, with its lengthy reviews of NDAs, was thwarting the commercial interests of pharmaceutical companies and by doing so—or so the argument went—compromising public health. Newt Gingrich, who rose to Speaker of the House in 1995, complained that the FDA was the “leading job killer in America.” The message was clear: The FDA needed to become an agency that helped bring new drugs to market.
The pharmaceutical companies also had newly gained financial influence over this review process. In 1992, Congress passed the Prescription Drug User Fee Act (PDUFA), which required pharmaceutical companies to fund the FDA’s reviews of its New Drug Applications. This provided the industry with financial leverage over the FDA—if the agency didn’t more readily approve drugs, the industry would lobby hard against this act when it came time, every five years, for Congress to renew it.
All of this served to corrupt the FDA. In a 2006 survey of FDA scientists, one-fifth responded that they had “been asked, for non-scientific reasons, to inappropriately exclude or alter technical information or their conclusions in an FDA scientific document.” Forty percent said they feared retaliation for voicing safety concerns in public.
Meanwhile, David Graham, who had warned about Eli Lilly’s underreporting of Prozac’s harms, told Congress in 2004 that he had been told “by his superiors to not warn the public about dangers of drugs like Vioxx,” which eventually was recalled. After he did warn the public, he testified that he was “marginalized by FDA management and not asked to participate in the evaluation of any drug safety issues. It’s a type of ostracism.”
The following year, Graham—who was then associate director of the FDA’s Office of Drug Safety—stated that the “FDA is inherently biased in favor of the pharmaceutical industry. It views industry as its client, whose interest it must represent and advance. It views its primary mission as approving as many drugs as it can, regardless of whether the drugs are safe or needed.”
Given this history and agency politics, Kavanagh’s protests, both within the agency and to Congress and the Office of the Inspector General, were almost certain to fall on deaf ears. There was nothing unusual with the agency’s impending approval of asenapine. And he could expect to be ostracized—and perhaps worse—for making a legal fuss about it.
This was just more of the same, and even as the FDA was going through the review process, Hassan publicly expressed dismay in a front-page Wall Street Journal article that asenapine had not yet been approved: “What will it take to get new drugs approved? The point is, we don’t know.”
The article, titled “Drug Makers Say FDA Safety Focus Is Slowing New-Medicine Pipeline,” was published on June 30, 2008. That was less than three months since the Zetia fraud had made headlines, and yet, in this article, Hassan was given a pulpit—and the high moral ground—to complain about the FDA being slow to approve asenapine.
Kavanagh’s Complaints Are Dismissed
Kavanagh’s whistleblower complaint was quickly dismissed. A couple of FBI agents interviewed him, but concluded this was a case of different opinions within the FDA about asenapine’s efficacy and safety. At the same time, both the FDA’s reviewer of safety data, Robert Levin, and Laughren wrote up summary reports that, within the agency, put the matter to rest.
In a June 26 review, Levin provided brief case histories of the 22 deaths of patients treated with asenapine. He dismissed all but two as “probably unrelated” to the drug, and even in the other two, he downplayed their possible significance.
There was a suicide by a 67-year-old man that an investigator, in the case report form, had concluded was “possibly” related to the drug, but, Levin said, the investigator had “provided no clear rationale for why that would be.” The neonatal death, while possibly related to the drug, could also be explained by the mother’s history “of 3 previous premature deliveries,” he wrote.
That was that: an approvable bill of health for the drug. There was no mention of a death due to an allergic reaction, either; as the company’s safety summary had said, the 57-year-old woman had died from a pulmonary embolism.
On August 1, Laughren determined that asenapine was approvable for both schizophrenia and bipolar I. He wrote a few lines about each of Kavanagh’s main complaints.
Re asenapine’s lack of efficacy for schizophrenia: “From what I have seen, he [Kavanagh] has not made any credible arguments to support these broad statements.”
Re asenapine’s lack of efficacy for bipolar I patients with mild to moderate symptoms: “I consider [his analysis] a flawed approach to looking at these data . . . . these severity scores have no diagnostic significance and it would not be appropriate to suggest that baseline severity could be used to select patients for treatment. In my view, the correct interpretation of these data is that asenapine has been shown to be effective in the acute treatment of mania and mixed episodes, and I think it should be left to clinicians to decide how to select patients for treatment.”
Re the company’s hiding of safety risks: There was no reason to think that the agonism of the 5HT2B receptor would be a particular risk; the pregnant woman whose newborn died had a history of troubled pregnancies; the healthy volunteer had suffered a fainting episode, not a stopped heart; there was no evidence that the 57-year-old woman had died from an allergic reaction; and suicide was a risk of the two diseases.
“I do not share [Kavanagh’s] view that the sponsor failed to report critical safety information that they possessed, or that they misrepresented what they did submit in an attempt to mislead, at least based on what I have reviewed,” Laughren wrote.
There was one aspect of Kavanagh’s Office of Clinical Pharmacology Report that Laughren found worthy, which was that the company needed to provide more information about the drug’s metabolites. “If OCP is correct in its assertions, however, we have little assurance that the animal carcinogenity data or reproductive toxicity data are relevant to humans, since we would know so little about what is circulating in humans. Until this issue is resolved, I am inclined to agree with OCP that this is a serious deficiency.”
Fifteen days after Laughren wrote his “approvable” report, Kavanagh was escorted from the FDA’s office. The termination letter he subsequently received stated that, in addition to behavior that could be linked to his whistleblower efforts, he was being fired for having been discourteous and aggressive in meetings with pharmaceutical companies.
The deficiency in Schering-Plough’s NDA regarding asenapine’s metabolites did slow the issuance of a formal approval letter, but that came in early 2009. An advisory committee meeting was held in July of that year, with the committee voting 10-2 to approve asenapine for schizophrenia and 12-0 to approve it for bipolar. Kavanagh was not allowed to speak at the meeting; his concerns were not aired. The drug, marketed as Saphris, was launched into the market that fall.
The Payoff for Hassan
When Hassan led Schering-Plough to acquire Organon, he knew that if asenapine were approved for both schizophrenia and bipolar, it would increase Schering-Plough’s value. In March 2009, shortly after the FDA notified Schering-Plough that asenapine was “approvable,” he negotiated the sale of his company, in the form of a merger, to Merck for $41.1 billion.
Published accounts of the merger stated that Hassan and nine other Schering-Plough executives would get $132 million in the deal. Hassan’s share of that was estimated to be $51 million in cash and pension benefits, although there were reports that put Hassan’s take much higher.
The merger with Merck closed that fall, and Hassan was once again hailed for his successful stewardship of a pharmaceutical company. In August of 2010, CBS News published an interview with Hassan titled: “How I beat the odds on a $14 billion drug deal.”
The key, Hassan told CBS News, was that the compounds that Schering-Plough acquired when it bought Organon, “proved valuable, specifically the schizophrenia drug asenapine, trade name Saphris.”
“Many people were generally not excited about Organon. There was a lot of skepticism about Saphris, which Pfizer had walked away from. The prevailing view was that Pfizer gave up on asenapine because they saw major problems with it either scientifically or commercially. Companies, Wall Street analysts, and even the press wondered aloud how we could succeed where Pfizer could not, and if eventually we would come to the same conclusion that Pfizer did. But Saphris did get approved by the FDA in August  and subsequently launched . . . So this was something that we accomplished that was not previously clear to others. Very few deals turn out as well as this one.”
The merger, however, did not turn out particularly well for Merck. Saphris didn’t offer any advantage over the atypicals already on the market, and in 2013, with Saphris generating net sales of only $150 million that year, Merck sold its marketing rights for asenapine to Forest Laboratories for $240 million.
Laughren Psychopharm Consulting
In 2012, Laughren left the FDA to form Laughren Psychopharm Consulting with the goal of helping pharmaceutical companies bring their drugs to market. One of his first clients was AstraZeneca, a company he had helped out three years earlier when it was seeking to get Seroquel approved for a wider range of conditions, including for use in adolescents.
At the advisory panel meeting, Wayne Ray of Vanderbilt University told of his research that had linked sudden cardiac death to Seroquel when it was used with certain other medications. However, as Science magazine later reported, Laughren dismissed Ray’s research and told the panel that AstraZeneca’s own clinical findings, which had found no increased risk of death, should be considered the definitive research.
The panel voted overwhelmingly to approve quetiapine for new indications and it didn’t require any labeling about the risk of sudden cardiac death. Within two years, the FDA had received reports of 220 deaths due to cardiac events related to Seroquel use, and it then required AstraZeneca to warn of this risk on its label.
Science magazine’s article was titled: “FDA’s revolving door: Companies often hire agency staffers who managed their successful drug reviews.” Laughren, it seems, illustrated this career path best.
A Prophetic Warning
Kavanagh, for his part, has never put his “whistleblower” efforts to rest. Again and again, he has tried to bring public attention to the FDA’s approval of this drug and other atypicals for bipolar 1 and to the harms that have come from other FDA approvals. In addition, in his subsequent legal actions and letters to presidents Obama and Trump and to members of Congress, he has sought to make the argument that he was fired for his whistle-blowing efforts in 2008.
He has gathered some additional ammunition for his whistleblowing efforts along the way.
During the first 10 months that asenapine was on the market, the FDA received 52 reports of patients suffering a “serious allergic reaction” to the drug (type 1 reactions, which include anaphylaxis). Since it is estimated that only 1% to 10% of adverse reactions to a drug are reported to the FDA’s Medwatch program, this suggests that there were between 520 and 5,200 patients treated with asenapine during its first 10 months on the market who had a severe allergic response to the drug. On September 1, 2011, the FDA issued a warning related to this risk, requiring that it be added to the label.
A second worry of Kavanagh’s was that asenapine and other atypical drugs, when prescribed to pregnant women, would lead to neonatal deaths. In 2011, as Kavanagh was getting ready to file his qui tam lawsuit, he searched Medwatch reports for deaths due to atypical drugs, and found the following:
- The Seroquel file told of 1,844 deaths, with more than 5% of those deaths in infants less than two years old.
- From 1% to 3% of deaths from Risperdal, Geodon, Abilify, and Zyprexa were in this age group.
In his qui tam lawsuit, Kavanagh calculated that from 2000 to 2012, there had been about 3,500 infants under two years of age who had died due to exposure to atypicals, with this exposure—it would seem—coming during the pregnancy and through breastfeeding. Yet, even this number is undoubtedly an undercount: A 2015 review of neonatal outcomes with “exposure to antipsychotics” during pregnancy found that there was an increased risk for “major malformations, heart defects, preterm delivery, small-for-age gestational births and decreased birth weight.”
In his lawsuit, Kavanagh also sought to tally up the harm done to bipolar I patients with mild to moderate symptoms who are nevertheless prescribed an atypical antipsychotic. He estimated that 500,000 patients fall into this subgroup and, given a mortality rate of 1% to 1.2% for volunteers in antipsychotic drug trials, he calculated that this would lead to 5,000 or more deaths each year. “Not to mention,” he added, “the numerous other serious adverse effects that occur with antipsychotics including seizures, blood clots, heart attacks, diabetes, neurologic toxicities, etc.”
Since a qui tam lawsuit alleges that state and federal governments are being defrauded, in this case because Medicaid and Medicare were paying for drug treatments that Kavanagh alleged were ineffective and harmful, the possibility of success is usually dependent on whether the federal government “joins” the lawsuit. And while Kavanagh’s lawsuit alleged that the manufacturers of these drugs had sought to hide their harms, it also alleged that the FDA had been complicit in that fraud. The federal government would have had to throw its own agency under the bus to join the action and it chose not to do so, which led to the dismissal of his lawsuit in 2014.
After that, Kavanagh periodically renewed his petitions to elected officials and government agencies. He did so in 2016 and 2017, and this past May made one last push, sending letters and summaries of his complaints to President Trump, Congressman Jamie Raskin, Senator Charles Grassley, the House Committee on Oversight and Reform, and the Department of Justice’s Office of Inspector General, Investigative Division. This last effort of his is still pending.
Counting Up the Deaths
In his estimate of deaths, Kavanagh focused on bipolar I patients with mild to moderate symptoms. Yet, the FDA’s approval of atypicals for bipolar 1 disorder naturally led to these drugs being prescribed to bipolar II patients and those diagnosed with bipolar spectrum disorder, meaning they too are exposed to the hazards of the medications, even though there may be little evidence from randomized clinical trials that they provide a meaningful benefit for mania and mixed episodes.
Asenapine’s label told of a long list of health risks associated with its use, which were presented as typical of “atypical” antipsychotics. They include: akathisia, somnolence, neuroleptic malignant syndrome, extrapyramidal symptoms, dystonias, tardive dyskinesia, hyperglycemia, diabetes mellitus, weight gain, syncope, leukopenia, neutropenia, agranulocytosis, QT prolongation, hyperprolactinemia, seizures, potential for cognitive and motor impairment, body temperature dysregulation, and dysphagia.
Various studies have found that antipsychotics double the standard mortality rate (SMR) in non-psychiatric patient groups, and in general mental health care as well. Given this fact, it is not surprising that the “mortality gap” between bipolar patients and the general population has widened since the prescribing of atypical antipsychotics for bipolar disorders became commonplace.
In a large UK study, investigators found that the SMR for bipolar patients jumped from around 1.4 in 2006 to 2.5 in 2014. In Denmark, SMR rates for bipolar patients increased from around 2.4 in 1995 to 3.0 in 2014.
Translating that increase in mortality rates into a numbered death toll for bipolar patients is an uncertain task. Yet, even conservative calculations produce an estimated count of 15,000 additional deaths per year, or a total of 180,000 additional deaths since 2008, when Kavanagh first sought to warn of this risk.
The FDA Under Review
Revisiting Kavanagh’s whistleblower complaint provides a fresh way to see the FDA’s review process and “approvable” standards for psychiatric drugs in action. What can be seen is that the FDA will approve a drug that has shown marginal “statistical” efficacy in clinical trials, even though that “benefit” falls far short of being clinically meaningful and its use will expose patients to a long list of hazards. That is a drug-review standard that is certain to introduce drugs into the market that, in public-health terms, will do more harm than good.
MIA Reports are supported, in part, by a grant from the Open Society Foundations