Altering Memories to Treat Addiction

This fall, a person addicted to cocaine arrived at the Medical University of South Carolina in Charleston. There, he or she watched a video of people using cocaine and was shown a small bag of what appeared to be either crack or powder cocaine, along with other drug-related paraphernalia. These cues were specifically designed to evoke memories of previous cocaine use and trigger intense craving. Shortly afterwards, the individual was given a powerful pharmaceutical called propranolol that interferes with the brain’s ability to form memories. All of this may sound like a bad idea, but it’s actually a new and promising experimental treatment for addiction.

Outside the controlled environment of a hospital, such a procedure would be risky, but this person — the last participant in a trial to test a new approach for preventing relapse — was carefully monitored on site and for six weeks after going home. Recently, much attention has focused on addiction as a chronic disease, but Mike Saladin, the research and clinical psychologist who led the study, explains that addiction is, in part, a disorder of learning and memory. The goal of his study is to reactivate memories of drug use — and then weaken them so they lose their power. The trial is one of 10 or more over the past decade inspired by lab experiments with rodents showing that memory can be manipulated to lessen its influence on addictive behavior. Researchers like Saladin aren’t trying to change the content of people’s memories. Instead, he says, “we’re turning down the volume” so that particular cues “don’t excite drug use the way they used to.”

Although the approach sounds like science fiction, it’s based on an understanding of learning and memory that goes back to an accidental discovery at the beginning of the last century, when the physiologist Ivan Pavlov, who was studying digestion in dogs, found that by repeatedly pairing a cue such as the tick of a metronome with the provision of food, the dogs would “learn” to salivate when they heard the noise — even before the meal arrived.

The same associative learning applies to addiction. When you learn that the pop of a wine cork predicts you will soon be enjoying a drink, you’ve created a memory. The more meaningful the experience and the more often you’ve paired those two events, the stronger the memory, even though it may be unconscious. Saladin points out that “people who have long substance use histories are responding to a myriad of cues, some of which they are not really aware of.”

Ravi Das, a psychologist at University College London, is also testing ways to modify memories in order to treat addiction. “Our brains are great at learning around rewards and what predicts them,” he says, noting that smells, sounds, and even people can serve as cues that promote relapse. This may help explain why 40 to 60 percent of people who have found a way to stop using drugs eventually relapse. Even if they’ve succeeded in staying off drugs for weeks or months or years, the paired association between cues and rewards can reactivate those drug memories, creating an overwhelming desire to use again.

Although we tend to think of longstanding memories as permanent, learning can be reversed, as Pavlov discovered with his dogs. When he repeatedly stopped providing food after starting the metronome, the dogs no longer salivated when they heard the cue.

Targeting memories in order to treat addiction, as well as anxiety, is not entirely new. The concept is the basis for extinction learning (also called exposure therapy), in which patients in the safe environment of a therapist’s office are exposed to environmental cues that normally trigger maladaptive memories. Because they’re exposed repeatedly to the cues without the response, those cues should gradually lose their power.

But it’s not that straightforward. It can be hard to transfer that new learning from the lab, to the therapist’s office, to the unpredictable chaos of the real world. “This is the million-dollar question,” says R. Kathryn McHugh, a research and clinical psychologist at McLean Hospital in Belmont, Massachusetts, who treats anxiety and substance use disorders. “People have been looking at this for decades now. In the lab, you absolutely can extinguish these cues. You can decrease craving.” Yet studies to assess the efficacy of incorporating cue exposure with cognitive behavioral therapy to treat alcohol addiction in earlier decades didn’t show much of a benefit, says McHugh. “If the basic research can make that jump more effectively, that would be a tremendous tool for us to have.”

David Epstein, who heads up the Real-world Assessment, Prediction, and Treatment Unit at the National Institute on Drug Abuse, is at the forefront of trying to make that jump. “We can’t treat addiction just by using extinction, in which someone is repeatedly exposed to drug-related cues until the cues cease to induce craving,” he said via email, “because extinction doesn’t generalize outside the environment where it occurs.”

Researchers in this nascent field aim to solve that problem by tinkering with how patients are exposed to cues. They were inspired in part by recent advances in treating post-traumatic stress disorder, or PTSD, in which maladaptive memories are triggered by stimuli associated with the original event, like a news report of an assault, or a sudden loud noise.

Nearly a decade ago, neuroscientists studying these so-called fear memories discovered what seemed to be a way around the weakness of extinction learning. Their method was based on the idea that when memories are reactivated, they are temporarily unstable. (It makes sense for memories to be fluid and dynamic, capable of incorporating new information, so they can help us make accurate predictions about the future.) “Normally the memory would subsequently re-stabilize,” says Das, who was not involved in these early studies. “But it would be slightly updated and that serves to keep your memories valid and up-to-date.” What if, researchers wondered, that instability provides a temporary window during which to intervene and add new information to the memory?

The trick was to make sure the behavioral therapy happened while the window of vulnerability was still open. Surprisingly, simply inserting a correctly timed pause between reactivating the memory and providing extinction learning was all that was needed to break the link between the cue and the fear memory, making the new approach more effective in both rodents and humans.

Epstein helped analyze the results of a clinical trial published in 2012 that sought to reduce heroin craving by inserting these properly timed cues prior to therapy. The research team, based at Peking University in Beijing, worked with hospitalized patients who had detoxified. If the researchers activated the drug-reward memory by showing a video of heroin use, but waited 10 minutes before beginning an hour-long extinction session, they could reduce their patients’ craving. However, if they waited six hours or more, it was too late. The memory had already reconsolidated without the new learning, and the craving for heroin was as strong as before.

“This is so remarkable on its face that I wouldn’t believe it if it hadn’t been replicated,” says Epstein, who points to follow-up studies in rats and humans.

Six years later, Epstein is now enrolling patients in the first clinical trial in this country testing a similar approach to treat alcohol misuse, but he’s added a new maneuver to address another weakness of extinction learning. Normally exposure cues that are just images, sounds, smells, or words associated with the drug are only capable of making a subset of drug-reward memories unstable. In Epstein’s trial, the cue for some participants will consist of drinking the alcohol itself. The hope is that this technique will, as Epstein puts it, “get you to the hub of where all learned associations are stored, instead of having to weaken them one by one.” For safety reasons, any test subjects who drink alcohol will stay in the hospital until their blood alcohol levels are very low and will then be taken home. Over the course of the following month, everyone in the trial will use a smartphone to keep track of their drinking and craving.

At University College London, Das is also testing memory manipulation to prevent relapse, but instead of using extinction learning to modify the memory, he uses drugs to block its reconsolidation. Ninety college graduates who drank heavily at least four times a week participated in a recent trial. They were at risk of transitioning to alcohol addiction and wanted to cut back.

As with the heroin users in the 2012 study, these so-called hazardous drinkers were exposed to visual cues such as pictures of beer and were asked to rate how appealing the images were. This procedure alone should reactivate drinking memories, making them temporarily unstable, but Das added a twist. “It’s harder with these reward memories because they’re really robust,” says Das. “So, you have to get the parameters right during the retrieval in order to destabilize.”

On the first day of the experiment, participants were allowed to drink a glass of beer at the end of the rating session. When they returned the following day, they began the session assuming that the procedure would be the same. But after rating the pictures and being instructed to pick up the glass and prepare to drink, the beer was withheld at the very last moment. Das’ hope is that this extra element of surprise, which neuroscientists call a prediction error, will destabilize the memory more than just a simple retrieval would. The theory is that if the function of memory is to make accurate predictions, any time your predictions are wrong memories require more updating than usual.

The coup de grace to the memory occurs five to 10 minutes later with the administration of a drug called ketamine, a safe and widely used anesthetic which, at the low dose used in this trial, causes a psychedelic dissociative experience. Ketamine blocks certain receptors on the surfaces of neurons called NMDA receptors, which Das says have been shown to be key to restabilizing memories once they’ve been destabilized. This blockade makes it difficult for the now-destabilized and weaker alcohol reward memory to reform.

In the trial, which concluded in September, one third of the hazardous drinkers went through this procedure, another third got saline solution instead of ketamine, and the final third received ketamine but only rated pictures of orange juice. All the subjects kept diaries and returned one week, one month, three months, six months, and nine months later to report on their drinking. Das called the preliminary results encouraging and presented a poster session on the trial at a recent conference in Barcelona.

Das acknowledges that there is some stigma against using a psychedelic drug to treat drug addiction. But he argues that if it proves effective, that concern should be put aside. “If it works, then it works, and you have to weigh up the cost-benefit of giving people a single dose of ketamine versus repeatedly drinking heavily on an almost nightly basis for many years.”

In the cocaine cue exposure trial in Charleston, South Carolina, Saladin uses a beta-blocking drug called propranolol — commonly used to treat high blood pressure — which accumulating evidence suggests may be effective in treating anxiety disorders and modulating memory. The five-year study he is now concluding is the follow-up to an earlier trial to treat cravings and reduce relapse in cocaine users. In that first trial, session participants who received propranolol reported less craving the next day, but when they returned for evaluation one week later, there was no significant difference between them and participants who did not receive propranolol.

In the new trial, Saladin is using two retrieval sessions instead of just one, and testing double the original dose of propranolol to see if drug reward memories will be weakened enough to prevent craving — and cocaine use — for longer than a day. Participants will be evaluated 18 times over the course of six weeks so researchers can document their craving in response to cues both in and out of the hospital, and to find out whether they have used cocaine.

Theories about precisely what’s going on are just that — theories. It’s possible that memories are not reconsolidating. “There’s no consensus on whether these reconsolidation-targeted treatments really weaken or erase memories at the neural level. It’s possible that the memories are still physically present, but difficult to access. In practical terms, that might be fine — unless the memories return,” says Epstein.

And memory manipulation does not address the underlying and complex reasons that lead to substance use disorders in the first place. So it’s not yet clear how effective this approach will be for individuals who struggle with depression, anxiety, or a history of trauma. Finally, it’s often the case that people use multiple drugs and are vulnerable to relapsing on multiple drugs. That could mean having to use separate memory manipulation treatments for each substance.

For these and other reasons, researchers are treading cautiously while also recognizing the urgent need for better treatment of addiction. “These people are in desperate circumstances and help is necessary,” says Saladin, adding that “reducing suffering is critical.”

Lauren Aguirre is a freelance science journalist based in Boston and a former staff editor, writer, and producer for the PBS series NOVA. She is currently writing a book about memory.