About 100,000 Americans die each year from overdoses, primarily caused by opioids like illicitly manufactured fentanyl. Fortunately, we have two treatments for opioid use disorder proven in multiple studies to reduce the death rate by 50 percent or more. These are methadone and buprenorphine. This data should make these drugs the gold standard for treatment. But a third medication—often promoted based on a pivotal trial that we now know missed some key overdose data—also vies for that position.
Vivitrol, a monthly injection of long-acting naltrexone, is the opioid treatment preferred by the criminal prosecution system, including jails, prisons, probation officers and drug courts. This is in part because it is not a controlled substance, unlike the other two medications, and in part because the drug’s maker, Alkermes, has heavily promoted it to those markets by claiming it is as safe as the other medications and easier to administer since it doesn’t have to be taken daily.
However, a new analysis of data from that pivotal trial in 2018, which compared Vivitrol and Suboxone (a combination of buprenorphine and naloxone), shows that the researchers conducting the trial miscoded several overdoses in people taking Vivitrol, leading them to conclude that both drugs were equally safe and effective. With these overdoses included, people on Vivitrol are actually more than twice as likely to overdose as those on Suboxone (Many opioid treatments couple buprenorphine with naloxone, which makes it harder to misuse buprenorphine.)
With additional research now also suggesting that naltrexone is not as safe as buprenorphine-based drugs or methadone, hundreds of addiction experts are demanding that the study be retracted. At a minimum, the carceral system must stop its preferential use of Vivitrol. Physicians and people using this drug must be made aware that the evidence heavily favors methadone and buprenorphine above and beyond any other treatment approach.
Understanding how each medication works is critical to understanding what’s at stake.
Methadone and buprenorphine are themselves opioids, typically taken daily. Methadone is a full agonist of the mu opioid receptor, meaning that when the drug meets the receptor in the brain, it activates it, causing a biochemical reaction that leads to euphoria and pain relief. Buprenorphine is a partial agonist, meaning it has those effects at some doses but also has a “ceiling effect”—higher doses don’t produce more euphoria and are less likely to cause respiratory depression, which is the main cause of opioid overdose death.
These medications work because opioids have an interesting pharmacological property: if a person takes the same amount at the same time every day, they become tolerant to the drug. Consequently, at the right dose, a person taking methadone or buprenorphine is not “high” or impaired. They have fewer cravings, don’t have to suffer withdrawal and can work and participate in social life as fully as anyone else.
However, naltrexone, in both short- and long-acting forms, has the opposite pharmacology: it antagonizes mu receptors, occupying them without activating them. This prevents any opioid—whether it’s one our bodies make, like endorphins, or something introduced, like heroin—from causing euphoria or relief. Therefore, if people relapse, the opioids they are misusing won’t work. Critically, people have to be completely free of opioids for around a week before they can even start taking Vivitrol, since it will otherwise cause withdrawal symptoms.
These mechanisms of action are important in understanding how these drugs affect overdose. Because of the tolerance that builds with opioid use, people take increasing doses to get high and to stave off withdrawal symptoms. There is no maximum. Many wind up taking amounts that could be fatal to those without tolerance. But mu agonists preserve tolerance. This means that if people use street opioids while taking methadone or buprenorphine, they are at much lower risk of dying.
However, when someone goes through complete withdrawal, which usually takes a week or two, they lose tolerance. A large proportion of overdose deaths occur after a period of abstinence—often after release from jail or rehab—when people return to use with little or no tolerance.
Vivitrol, however, stops tolerance. While it protects people from overdose by blocking the biochemistry that opioid receptors initiate, this protection may decline during the last week before their next monthly shot. Vivitrol may also sensitize opioid receptors because the system reacts to having less opioid activity when they are blocked. These sensitized receptors may make relapse even more dangerous than after abstinence or other meds.
Because methadone and buprenorphine are opioids, there has long been an enormous stigma against using them—especially in carceral settings where it might seem like people are being rewarded by being given a drug similar to the one they prefer. Vivitol’s manufacturer capitalized on this prejudice, marketing to drug courts, which mandate specific treatments instead of punishment and have often forced people to stop taking the other two medications. Alkermes engaged in intense political lobbying, emphasizing the dangers of the other drugs to Congress. In 2019, it was called out by the FDA for minimizing the risk of opioid overdose associated with Vivitrol as it wears off or following treatment. Yet Vivitrol’s use continues.
In 2020, Elizabeth Ajazi, then a graduate student in biostatistics at the University of North Carolina, Chapel Hill, decided to reanalyze the trial that compared Vivitrol to Suboxone, known as X:BOT. She was seeking publicly available data sets to try out alternate statistical approaches. But she couldn’t replicate the original analysis. Then she found the unreported overdoses.
“I thought it was very strange,” she told me, explaining how she eventually asked her advisor, Nabarun Dasgupta, a senior scientist at the University of North Carolina, for help. At first, he thought she was making a basic mistake. “She was like, ‘I keep finding more overdoses’ and I was like, ‘Dude, how hard is it to count the word overdose?’” he said.
When they talked over Zoom, he realized she was right.
It turned out that in the X:BOT data analysis plan, the researchers would count an overdose only if it was coded into a column carrying that term. So, if a person in the trial had overdose symptoms, but the clinician treating them entered a symptom like respiratory depression as the main issue in their medical record, that overdose would go in different column, and not be counted as an overdose, even though slowed breathing is a key symptom of opioid overdose and overdose was listed elsewhere in the medical record.
Researchers are often criticized if they stray from their prespecified plan, because it can affect their data analysis, and possibly invalidate the trial: since the discovery of these overdoses, this has been an argument made by people who support leaving the study as is, rather than correct or retract it. But Ajazi’s discovery demands if not a new trial, then a retraction or correction to the published data. The Lancet has refused to do either.
After a long delay, the journal published a research letter outlining Ajazi’s work. They simultaneously published a response from the research team, led by Joshua Lee, a professor of medicine at New York University.
The X:BOT team acknowledges the missing overdoses, while maintaining that the initial analysis was sound. In their published reply, Lee and his colleagues wrote, “We do not agree with Dasgupta and colleagues’ call to correct or retract our initial report. We note that, in this context, The Lancet sought independent expert opinions from the original reviewers of the X:BOT paper, who were unanimously in agreement that there is no case for correcting or retracting the paper.”
Nonetheless, counting all of the overdoses suggests a real safety advantage for buprenorphine over extended-release naltrexone—one that is statistically significant under the methodology that Ajazi and Dasgupta’s group used but that Lee, et al., did not. Moreover, since X:BOT was originally published, other research suggests that the risk of overdose Ajazi found is real.
In a study published in JAMA Network Open in 2020, Sarah Wakeman, of Massachusetts General and Harvard, and colleagues examined data from medical records of nearly 41,000 people with opioid addiction in the U.S., treated between 2015 and 2017. Compared to untreated people, those who took buprenorphine or methadone had a 59 percent reduction in overdose risk in the year after starting treatment. But they found no significant risk reduction with Vivitrol.
A 2019 study of a different database, using nearly 47,000 records from 2010–2017 had similar findings: a 60 percent reduction in risk for people on buprenorphine, but no significant risk decline for those taking long-acting naltrexone.
Lee, the author of the Lancet study, says that the only two randomized controlled trials that compared long-acting naltrexone to buprenorphine—X:BOT and a Finnish study—did not enroll enough people to detect differences in overdose rates and that larger studies have not been done. Medical records data are not as conclusive as such randomized trials, because the records are subject to biases related to who selects which type of treatment. Lee acknowledged in an email interview, however, that the medical records data are “highly suggestive that retention and overall mortality is lower on methadone or buprenorphine vs. naltrexone.”
Given the premiere status of controlled trials as data, in the absence of further studies, X:BOT’s results remain important in the literature. However, physicians and patients need to know that this trial did not include some cases of overdose that did occur among patients—and that this renders the claim that both medications are equally safe as suspect.
To protect people, X:BOT should be corrected or retracted. And treatment providers need to inform patients that the only treatment for opioid use disorder that has been consistently associated with saving lives is ongoing use of buprenorphine or methadone—not naltrexone or any abstinence-based approach.
This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of Scientific American.