Cocaine Killed 30,000 Americans Last Year. Scientists Just Found the Brain Switch Behind Relapse — and It Isn’t Willpower.

Marcus doesn’t describe his relapse in March 2024 as a decision. He describes it as waking up in the middle of one. “I didn’t choose to use,” he told Recovery Review in a published interview. “I was sitting in my car in a parking lot and then I was using. The part where I decided wasn’t there.” He had been in outpatient treatment for cocaine use disorder for eleven months. He had a sponsor. He attended meetings three times a week. He wanted, with everything in him, to stay clean.

He also had no medication that could help him, because none exists — not one, after thirty years of clinical trials and billions of dollars in research spending.

That last fact is what most people miss when they think about cocaine addiction. Opioid use disorder has buprenorphine and methadone. Alcohol use disorder has naltrexone, acamprosate, and disulfiram. Nicotine dependence has varenicline and bupropion. Each of these medications shifts the odds meaningfully toward recovery. Cocaine has none of them. And as stimulant deaths approach 30,000 annually, that gap has become something harder to defend than a scientific limitation. It is a policy and investment failure built on a decades-old assumption that cocaine users can simply stop if they want to enough.

In March 2026, a team at Michigan State University published findings in Science Advances that may represent the most significant step toward closing that gap in a decade. They identified the specific protein — called DeltaFosB — that accumulates in neurons during repeated cocaine exposure and physically rewires the reward circuitry that governs motivation and compulsion. The work doesn’t produce a cure. But it gives pharmaceutical developers their first precise molecular target in the specific neural pathway that makes relapse biology rather than moral failure, and it reopens a treatment pipeline that the pharmaceutical industry largely abandoned in the 2000s.

When 30,000 Deaths a Year Don’t Move the Policy Needle

Before the DeltaFosB findings can be understood, it helps to understand why the baseline situation is this bad.

Roughly 29,000 to 30,000 Americans die annually from overdoses involving cocaine — a figure that hasn’t moved much in five years and represents about 28% of all overdose deaths. A 2026 Frontiers in Psychiatry paper frames the scope plainly: the country is in the fourth wave of the overdose epidemic, characterized by the contamination and deliberate mixing of stimulants — cocaine and methamphetamine — with synthetic opioids including fentanyl and its analogs. Between 5.9% and 15% of unregulated stimulant samples now test positive for fentanyl. Users who believe they are using cocaine are being exposed to a drug that requires entirely different emergency intervention. Naloxone reverses fentanyl overdose. It does nothing for cocaine’s cardiovascular mechanism — the cardiac arrhythmia, the stroke, the coronary spasm that drives cocaine deaths. The fourth wave demands a simultaneous response to two distinct pharmacological threats, and the treatment field has the tools for one of them.

The FDA recognized the structural gap in 2023, issuing draft guidance specifically calling for development of treatments for stimulant use disorder — the agency’s first formal acknowledgment that the absence of approved medications is itself a regulatory and public health problem. The guidance was a signal to pharmaceutical developers that the FDA would clear a path for approval. It did not produce funding, and it did not produce candidates.

The reason pharmaceutical companies walked away from cocaine treatment is traceable to a series of expensive Phase III failures in the late 1990s and 2000s. Cocaine disrupts at least four neurotransmitter systems simultaneously: dopamine, serotonin, norepinephrine, and glutamate. Single-target compounds — targeting only the dopamine transporter, or only serotonin — couldn’t adequately address the cascade of effects, and trial after trial failed to show the clear efficacy the FDA requires (at minimum two well-designed trials, with a consistent effect size). The returns weren’t there, the trials were failing, and the stigma attached to cocaine users made it politically unattractive to advocate loudly for treatment investment. The pipeline dried up. Cocaine treatment became, for two decades, a behavioral health problem — managed with contingency management, CBT, and willpower.

Cocaine treatment became, for two decades, a behavioral health problem — managed with contingency management, CBT, and willpower.

Marcus had all three.

What DeltaFosB Actually Does

The MSU team, in their Science Advances paper, traced the molecular mechanism that makes cocaine compulsion different from cocaine pleasure. This distinction matters clinically, because the loss-of-control element of addiction — the relapse despite genuine intention to stop — has always been the hardest feature to explain with existing neuroscience models.

DeltaFosB is a transcription factor: a protein that doesn’t directly trigger neurotransmitter release but instead binds to DNA and regulates which genes are expressed in a neuron. Under normal circumstances, DeltaFosB appears transiently in response to stress and reward. Under repeated cocaine exposure, it accumulates in the nucleus accumbens — the brain’s reward hub — at abnormally high levels and stabilizes there. Unlike most transcription factors, which are degraded quickly, DeltaFosB is unusually stable and long-lasting. A single period of heavy cocaine use can produce DeltaFosB levels that persist for months after the drug is stopped.

What those elevated levels do is measurable: they alter the expression of genes governing synaptic plasticity in the reward pathway, increasing the density and sensitivity of reward-related synapses in a way that makes cocaine-associated stimuli — the smell of a particular place, the sight of a particular person, a Tuesday afternoon with nothing planned — disproportionately attention-grabbing and desire-triggering. Research on attentional bias in methamphetamine use disorder — a closely related stimulant — shows that what cue reactivity looks like from the outside (someone craving) is, from the inside, a pre-attentive process: the eyes move toward drug-related stimuli before any conscious awareness, before any decision. The person in Marcus’s car wasn’t deciding to use. The DeltaFosB-modified circuits in his reward system were already acting on the environment.

The MSU team is now partnering with researchers at the University of Texas Medical Branch to develop compounds that block DeltaFosB from binding to DNA in the relevant neurons. The goal is to interrupt the transcription cascade that maintains the rewired state — to, in the researchers’ framing, “reset the circuit” without dulling normal motivation and pleasure. That’s the critical distinction from earlier approaches: this isn’t about blunting reward broadly (a problem that produces anhedonia and creates its own compliance challenges). It’s about specifically reversing the cocaine-induced plasticity while leaving the baseline reward system intact.

The MSU work is preclinical. Human trials are years away.

The Fourth Wave Forces the Issue

The polysubstance reality described in the 2026 Frontiers paper doesn’t wait for DeltaFosB inhibitors. Stimulant users in the current drug supply are navigating a market in which fentanyl contamination is common enough that harm reduction organizations in Philadelphia, Chicago, and Phoenix are now distributing naloxone specifically to people who identify as cocaine users — not because they want to use opioids, but because they have no reliable way of knowing whether the cocaine they’re holding contains a lethal dose of fentanyl.

The treatment infrastructure hasn’t caught up to this reality. Treatment programs designed around opioid use disorder — which have clear medication-assisted treatment protocols and established withdrawal management guidelines — are being asked to serve people whose primary substance is a stimulant, whose withdrawal isn’t physically dangerous but whose craving is neurobiologically severe, and who may simultaneously be opioid-dependent via contaminated supply. The existing contingency management protocols for stimulant use were designed for a drug environment where stimulants were stimulants. The fourth wave has fundamentally changed the clinical picture.

The Frontiers authors call for adapting and scaling contingency management specifically for polysubstance populations, particularly underserved and minoritized communities — Black Americans are disproportionately represented in cocaine overdose deaths — and for treatment approaches that address co-occurring stimulant and opioid use rather than treating them as separate presentations.

Those are reasonable calls. They’re also not solutions to the absence of medication.

What the Breakthrough Actually Changes Today

Honest accounting: the DeltaFosB discovery does not help Marcus today. The MSU-UTMB collaboration is at the preclinical stage. Even in an optimistic scenario, DeltaFosB inhibitors enter Phase I trials no earlier than 2028. By the time any such compound reaches widespread clinical use, the current generation of people with cocaine use disorder will have navigated — or not navigated — the disease entirely through behavioral intervention.

Even in an optimistic scenario, DeltaFosB inhibitors enter Phase I trials no earlier than 2028.

What the discovery changes is the argument. For investors, the DeltaFosB pathway gives pharmaceutical developers a specific target with a mechanistic rationale, which is the prerequisite for venture investment in drug development. For the FDA, it represents the kind of basic science foundation that can justify prioritizing a new development program. For researchers, it shifts the framing: cocaine addiction is not intractable — it has a molecular engine, and that engine can potentially be disrupted. The pipeline that closed in the 2000s can be reopened on entirely different scientific footing.

In the interim, what works: contingency management, which uses incentives (vouchers, prize draws) to reinforce negative drug tests, is the most evidence-supported approach for stimulant use disorder and is criminally underused, partly because Medicaid reimbursement for CM varies wildly by state and is actively debated in Congress. The naltrexone-bupropion combination studied in the ADAPT-2 trial for methamphetamine shows modest but real benefits for co-occurring opioid-stimulant use. And harm reduction — fentanyl test strips, naloxone, non-judgmental engagement with people in active use — keeps people alive long enough to reach treatment when they’re ready.

None of these are as powerful as a medication. But they’re what exists.

The Question We Should Actually Be Asking

When Marcus described his relapse, he was expressing something neuroscience has been building toward for two decades: that cocaine compulsion, once established, operates through neural pathways that are not under voluntary control in any meaningful sense of the word. DeltaFosB is the molecular substrate of that. It accumulates, it persists, it reshapes the attention system and the reward system and the decision-making system in ways that make relapse not a failure of character but a predictable output of biology.

The question, then, is not why Marcus relapsed. He relapsed because he had a disease with no approved medical treatment, operating on rewired neural circuitry that behavioral intervention alone could not fully counter, in a drug supply that has made his substance increasingly lethal.

The question is why, in a country that spent roughly $300 million on cocaine-related enforcement in FY2025, the treatment development budget for cocaine use disorder is effectively zero. The DeltaFosB finding gives researchers a target. What it cannot provide — what only sustained public investment and pharmaceutical commitment can provide — is the decade of clinical development that turns a target into a drug.

Marcus is still in recovery. He started over in April 2024. He doesn’t have a medication. He has the same tools he had before — meetings, sponsorship, outpatient support, his own determination. The DeltaFosB study will not change his treatment plan this month, or next year, or possibly for another decade.

That timeline is a choice. It should be named as one.