Medicine should center on the patient, so I start this story with an anonymous person dropping dead.1 Perhaps this person first heard about the medication from a friend who raved about his new allergy pill. Then, the day of a routine doctor’s appointment, our patient saw an advertisement for Seldane™ in a Time Magazine in the waiting room. His doctor gave him a few free samples from the sample closet and wrote him a prescription. He was a good patient, so he had been dutifully taking the medication twice a day, exactly as prescribed. He was still taking the allergy medication when he came down with bronchitis at the beach. A friendly local physician started him on a brief course of erythromycin, which “probably isn’t necessary, but just in case.” Our patient was already starting to feel better a few days later, except for a few episodes of lightheadedness that got better on their own. Maybe he needed to drink a little more water?
Then, he died.
He didn’t clutch his chest and gasp. There was no time for him to ask “why me?” or contemplate mortality. No one, not his family, not the family doctor, not even the medical examiner found an explanation. Dysrhythmias leave no trace on autopsy. People die every day; sometimes for no reason.
A Breakthrough Medication
Another place to start the story is 1985, when terfenadine was released under the trade name, Seldane™. Before this time, seasonal allergy sufferers had two choices. They could take antihistamines that left them dazed and sleepy, or they could suffer a runny nose, itchy eyes, and sneezing associated with allergies. The FDA approval of terfenadine, marketed by Merrell Dow Pharmaceuticals, offered a third option – relief without sedation.2
Terfenadine, which is metabolized to its active form, was the first non-sedating antihistamine available in the United States. Like older antihistamines, the active metabolite of terfenadine interfered with the H1 histamine receptor.
In an allergic response, a foreign substance – an allergen – is bound by IgE, which, in turn, activates mast cells. The mast cells release histamine into the bloodstream, binding receptors in vasculature, skin, and airways. The H1 histamine receptor, which mediates allergic response, is coupled to a G protein that triggers a cascade of responses in vasculature, skin, and airways.
Antihistamines have been available for 50 years, but side effects limited their use. The first H1 antagonist, thymoxyethyldiethylamine, was identified in 1937.3 A much less toxic drug, diphenhydramine, aka beta-di-methylaminoethyl benzhydryl ether hydrochloride, aka Benadryl™, appeared in 1945. The H1 antihistamines are not structurally related to histamine and they do not bind the active site of the histamine receptor. Instead, they stabilize the “off” conformation of the receptor. In this way, the H1 antihistamines technically aren’t antagonists, they are “inverse agonists.”4
In addition to mediating allergic response, histamine is involved in wakefulness. Histamine receptors exist in throughout the cerebrum, cerebellum, posterior pituitary, spine and spine. Histamine receptors in the vestibular nucleus explain why antihistamines work for motion sickness. Histamine release by the tuberomammillary nucleus during the day causes arousal, while decreased nighttime histamine causes somnolence. First-generation antihistamines pass through the blood-brain barrier, triggering sedation. The prolonged duration of action of these drugs (hydroxyzine has a half-life of 20-25 hours) results in prolonged sedative effects and disruption of the sleep-wake cycle. Your histamine receptors literally don’t know whether it is day or night. First-generation antihistamines impair learning and work efficiency.5 Sleepiness is not just a nuisance. Children taking sedating antihistamines do worse on school examinations and adults are prone to their crash cars.6
Then, in 1985, came terfenadine. Unlike first-generation antihistamines, terfenadine is poorly lipid soluble and does not penetrate the blood-brain barrier. Therefore, it doesn’t cause sedation and cognitive impairment. The side effect profile in early trials was excellent. A 1982 randomized-controlled trial of nearly 400 patients that compared terfenadine vs 1st-generation antihistamine chlorpheniramine vs placebo found the active drugs had comparable effects on allergic symptoms (60% in both groups had moderate to complete relief).7 As expected, there was a huge difference in sedation. Nineteen percent in the chlorpheniramine group experienced sedation, compared to 7.6% in the terfenadine group, which was statistically no different from placebo.8
II. Partial Repolarization
Seldane was a blockbuster drug. Seldane™ was responsible for one of the very first television commercials marketing a pharmaceutical. To get around laws about mentioning side effects of the drug, the ad simply told viewers to ask their doctors about a non-drowsy allergy medication. Since Seldane™ was the only non-drowsy allergy medication available, the message was effective. In 1988, it was estimated that terfenadine accounted for 60% of US antihistamine prescriptions. By 1991, 17 million prescriptions had been written in the US while worldwide use reached 100,000,000.9 Sales of terfenadine reached $800 million per year. Allergy sufferers rejoiced. According to advertising directed at doctors, terfenadine turned the outdoors into psychedelic woodland ecstasy.
I’ve watched this video 5 times. I have a few thoughts and questions (mostly questions):
- Is the kids sub-plot really a coming-of-age friendship (love?) story about the two boys?
- Both females in the video are spinning
- That owl stare, though?
- What is falling on the woman with the wings? It’s too warm to snow, is that some sort of magic dust?
- Is the entire world actually the eyeball of the woman with wings?
- The owl still haunts me
However, despite the frolicking sylvan vision quests, all was not well. Troubling side effects were emerging. In 1989, a case of “cardiotoxic convulsions” – a dysrhythmia – was reported from an overdose of terfenadine.10 This adverse effect had already been described on Poisondex™, the principal resource for poison information centers. The poison information system had noticed the effect in overdose but hadn’t yet published a paper on it.
In 1990, JAMA published a case of torsades de pointes (TdP)11, which is French for “twisting of the peaks”. Everyone in the U.S. spells it “torsades de pointes” and most of us pronounce the S, which am I told by francophones is tres gauche. The correct spelling should be “torsades des pointes”.12 You can read more about this in our upcoming spinoff French Toxicology Blog.
This case was not an overdose, but a woman taking her terfenadine exactly as prescribed. A 39-year-old woman had syncopal episodes 10 days after starting a terfenadine and cefaclor for recurrent sinusitis. On day 8, she began taking ketoconazole for vaginal candidiasis, which was probably a result of the cefaclor. Her admission terfenadine concentration was determined to be 57 ng/mL. Based on clinical trials, her expected therapeutic level was 10 ng/mL or less.
This was her admission ECG:
To a layperson, this appears more or less normal, but most medical observers will instantly recognize that the QT interval, the distance that signifies the beginning of depolarization and the end of repolarization, is far too long.13 Her QTc was read as 655 ms. Most investigators agree that anything longer than 470 ms is a predictor for dysrhythmia in women.
While she was in the hospital, this happened:
This is a classic example of TdP.
In TdP heart beats much too fast, or the electric signals come so fast that it cannot beat at all. In any case, the heart cannot perfuse blood in any organized, sustainable way. With drug-induced QT prolongation, TdP can happen at any time, although it is often triggered by bradycardia. Sometimes the trigger is an electrolyte abnormality, and sometimes there is no identifiable trigger. Once the episode begins, it may spontaneously stop, or it may continue until terminated by a medical professional, or the death of the patient. Patients may experience TdP as a feeling of extremely rapid heart rate, or they may experience it as loss of consciousness. Sadly, some patients never truly “experience” it; they just die.
Thankfully, this woman’s episodes resolved without any intervention. Cardiac testing revealed no structural or functional problems. After 7 days off medications, her QTc corrected to a normal 429 ms. The authors of the paper correctly surmised that terfenadine was a substrate for CYP enzyme (we now know it is subtype CYP3A4) and that the introduction of antifungal ketoconazole, a CYP3A4 inhibitor, resulted in the accumulation of the compound. This adverse outcome also makes the argument for antibiotic stewardship; reducing the unnecessary use of antibiotics. This patient took an antibiotic, cefaclor, which triggered a yeast infection. She took the ketoconazole to treat that infection, not knowing it would interfere with the terfenadine. Medications can lead to complications, which are treated with other medications, which lead to complications.
The mechanism for terfenadine interference with the QT interval wasn’t described until 1996. In addition to mucking with histamine receptors, terfenadine blocks a potassium channel called IKr, which is key to maintaining the potassium rectifier current necessary to repolarize the cardiac cell, preparing it for the next heartbeat.14 Work on terfenadine and other drugs led to the important discovery that this same potassium channel is encoded by the human ether-a-go-go-related gene (hERG). The channel was given its name by scientists who noticed that fruit flies with a mutated version of the gene will shake when given ether. Apparently, this reminded the investigators of dancing at the famous Whiskey a Go Go club in Hollywood.15 As it turns out, the hERG gene, which encodes a potassium channel, is an important link in understanding the relationship between drug-induced QT prolongation and inherited QT prolongation.
Things, shall we say, escalated. In June 1990, an FDA advisory committee issued a report identifying risk factors for TdP, which included concomitant use of terfenadine and imidazoles (like ketoconazole), or macrolides such as erythromycin.16 In August 1990, FDA ordered the manufacturer to send a “Dear Doctor” letter to clinicians addressing the issue, followed by a formal boxed warning (“black box warning”) cautioning that the drug was associated with “rare cases of serious cardiovascular adverse events, including death.” By 1992, the drug manufacturer had received 177 complaints of sudden death in otherwise healthy individuals taking the terfenadine. Prescriptions of the drug slowly declined. From 1990 to the first half of 1994, same-day dispensing of terfenadine and contraindicated drugs dropped by 84%.16 However, in 1996, despite every warning, terfenadine was still the most widely prescribed antihistamine.
Cracks in Our Defenses
Thousands of patients were studied in clinical trials leading up to the approval of terfenadine. How did investigators miss this critical issue? First, even large clinical trials are underpowered for rare events. A clinical trial with 1500 patients will barely detect an event with a frequency of 1 in 1000. Torsades was estimated to have occurred following terfenadine in somewhere between 1 in 28,500 and 57,000 patients.17 To have an 80% chance of identifying an event of this frequency, we would need to enroll 1,425,000 patients in the trial. That’s a lot of parking vouchers! It is simply impractical to study this many people to turn up a rare dysrhythmia. Yet, in a population of 100 million people, this small proportion will result in about 3500 catastrophic events, about the size of the September 11 death toll. The risk is simultaneously so small that it cannot be detected and so large that it must be detected.
The risk is simultaneously so small that it cannot be detected and so large that it must be detected.
We didn’t (and still don’t) have a perfect method of predicting risk for dysrhythmia. The first drug that was withdrawn for QT-related dysrhythmia was the anginal drug prenylamine, in 1988.18 Prior to the 1980s, QT-prolongation was considered to be a good thing; an indicator of quality for antidysrhythmic drugs. Even after QT was determined to be a problem, there were many legacy drugs on the shelf that had never had formal QT studies performed. Perhaps scientists and regulators probably weren’t paying much attention to QT prolongation because terfenadine emerged from a class of drugs (antihistamines) that hadn’t been associated with dysrhythmias. A table of QT-prolonging drugs from a 1988 review article contained about a dozen cardiovascular drugs and 2 psychiatric drugs.19 Perhaps there would have been a more regulatory focus on terfenadine if it had been known that it was originally developed as an antipsychotic before its antihistamine properties were discovered. This isn’t unusual. The antipsychotic Thorazine™ (chlorpromazine) was intended to be an antihistamine before it was an antipsychotic, sildenafil (Viagra™) started out as a potential anti-anginal drug, and penicillin was famously discovered in mold.
Although QT interval was studied in terfenadine trials, small clues were overlooked. While the mean peak increase in QTc interval across the dosing interval was only 6 ms, the mean peak increase was 18 ms.20 The first figure is unremarkable, and the second wasn’t reported. The prolongation of the QTc interval is a good, but not foolproof, predictor of TdP. Most patients with QT prolongation will not have a dysrhythmia.
Finally, we don’t fully study combinations of drugs or effects of drugs in at-risk populations. Participants in clinical trials tend to be healthier and more homogenous than the general population. They are often under tighter clinician scrutiny regarding co-morbid conditions and non-study medications. It’s not surprising that the effects of combining terfenadine with CYP3A4 inhibitors were missed. It’s likely that patients with conditions predisposing them to dysrhythmias are also underrepresented in the clinical trials.
The End of Terfenadine
In 1997, FDA finally recommended that terfenadine be removed from the market. However, several other daggers aimed at the heart of terfenadine probably doomed it already. Terfenadine’s patent exclusivity was expiring and a manufacturer was already planning to release a generic version. Other second-generation antihistamines, loratadine (Claritin™) and cetirizine (Zyrtec™) were either approved or on the horizon. Perhaps most damning was FDA approval of fexofenadine (Allegra™) in 1996. Fexofenadine, the active metabolite of terfenadine, was also made by the same manufacturer, but didn’t have the dysrhythmogenic effects, and wouldn’t have a generic competitor. Terfenadine simply wasn’t going to be lucrative moving forward. By agreeing with FDA without challenge, the manufacturer simultaneously eliminated a generic competitor and positioned itself as the sole producer of next best thing to terfenadine. (How generous of them!) Over the life of Seldane™, Merrell Dow had been rolled and kneaded into Marion Merrell Dow and then Hoechst Marion Roussel. Like its predecessor, fexofenadine became a huge hit, although it competed in a more crowded landscape with the other drugs. Today, Hoechst Marion Roussel exists as Sanofi, and Allegra marketing is entrusted to Sanofi subsidiary Chattem, Inc. For what it’s worth, I don’t think the pharmaceutical company is the villain of this story. They thought they had found a way to relieve a lot of suffering while simultaneously making a ton of money. Seemingly, everyone was winning.
Terfenadine was the first blockbuster drug to be taken off the market for dysrhythmia related to QT prolongation. Many drugs would follow. Terfenadine research led to a better understanding of the hERG gene and the relationship between drug-induced and hereditary QT prolongation. This undoubtedly saved lives.
Today, regulators are moving away from simply using QT interval from clinical trials as the sole prediction tool for adverse cardiac events. The Comprehensive in vitro Proarrhythmia Assay initiative by FDA hopes to combine in silica and in vitro models with phase I clinical trial data to assess dysrhythmia potentials of drugs. In this way, software and cell models, supplemented with ECGs performed on just a small number of trial participants, may be able to identify dysrhythmia risk without requiring large numbers of patients.21
In drug overdose patients, a growing body of evidence indicates cardiovascular events (shock, dysrhythmia, myocardial injury) are a key contributor to bad outcomes, and that TdP only accounts for a fraction of those outcomes. Clinicians who care for overdose patients should be on the lookout for predictors of cardiovascular events as a whole, rather than just TdP. From this standpoint, QT interval is one predictor, but not the only, of adverse cardiovascular events. Age, lower serum bicarbonate, cardiac disease, and altered mental status increase the risk of adverse cardiovascular events.22
Terfenadine taught us to heed early signals. In the mid-1980s, we were just beginning to understand the relationship between potassium channels, drugs, and dysrhythmias. A short increase in QTc in a clinical trial might portend trouble in the general population. The first clinical events we noticed were the effects of intentional terfenadine overdoses. A small number of overdoses gives a window of insight into the effects of drug accumulation. Overdoses serve as canaries that identify issues therapeutic dosing. These overdoses were reported in the medical literature and shared on Poisondex™, a database that Poison Control Centers still rely upon. Medical toxicologists serve the public by recognizing poisoning epidemics early. When a drug accumulates from metabolic interference, it is really no different to the body than if it accumulated from an overdose.
Recall our patient who died suddenly. Perhaps a family member called the local poison center asking for help, and mentioned that he was taking Seldane™. Perhaps his doctor had a second patient on Seldane™ who had palpitations, and this doctor reported it to the prescriber, or called his poison center, or published a case report. Reporting cases matters. Often, we belittle the meager case report. We ask ourselves, “Why should a case report of a single death matter?” To that, I respond, “It mattered to the patient.” Case reports and poison center calls add up.
“Why should a case report of a single death matter?” To that, I respond, “It mattered to the patient.”
Because clinicians took the time to share and report on these terfenadine overdoses, the first case of toxicity from a drug interaction was diagnosed. From this diagnosis, scientists were able to study the effects of terfenadine on a potassium channel, which led them back to the cell. In this way, science proceeds. The mystery of a sudden death unlocks the mystery of a tiny gene coding a tiny channel. Death leads to discovery and life.
tom fiero says
excellent Andrew. thank you very much. more than a little frightening though.