Naltrexone for Depersonalization Disorder: When Does Thin Evidence Justify a Trial?
A 27-year-old woman with four years of persistent depersonalization has tried an SSRI and eighteen months of therapy without resolving the core symptom. The team disagrees about whether a thin but real evidence trail on naltrexone is enough to justify a trial now, rather than later.
Four years of continuous depersonalization, two adequate SSRI trials, and eighteen months of weekly therapy have brought R.A., a 27-year-old woman, to the point where the next reasonable option is a drug with no approval for her condition and a total published evidence base of two small uncontrolled studies. That is the decision in front of the team, worth stating plainly first.
It started during finals week of her senior year — sleep-deprived, overcaffeinated, in the campus library — with a panic attack that did not end the way panic attacks end. What followed was its opposite: a flattened, persistent sense of watching herself from a short distance, as though her hands were typing a few feet in front of where she actually sat. She was 23. She builds her UX and graphic design client list almost entirely through referrals now, because cold outreach and first meetings have gotten harder; the five-mile route she has run most mornings for three years is, she says, one of the only stretches of her day that still feels unambiguously hers. She has no other psychiatric history and no substance use, and a neurologic workup early on ruled out anything structural. Sertraline, then escitalopram, each given an adequate trial, took the edge off her residual background anxiety and left the depersonalization itself untouched. Weekly grounding- and CBT-informed therapy has produced real gains — she can recognize and ride out her worst episodes now rather than being flattened by them — but the rate of that improvement has visibly slowed over the last six months compared with the twelve before it, and the baseline distance from her own decisions has not closed.
Naltrexone is what remains, and the evidence behind it is thinner than the idea of an opioid-antagonist trial makes it sound. Nuller and colleagues' 2001 pilot study reported marked improvement in seven of fourteen patients and complete resolution in three more — but it was open-label, uncontrolled, and used intravenous naloxone, not oral naltrexone. The only study of naltrexone itself is Simeon and Knutelska's 2005 open trial, also uncontrolled, in which four of fourteen patients improved markedly and symptoms fell by an average of 30% across three dissociation scales. That trial's mean dose was 120 mg daily, with some patients taken to 250 mg — two to five times naltrexone's labeled 50 mg dose.
The mechanistic rationale is coherent: the endogenous opioid system mediates stress-induced numbing, the same pathway that blunts pain and affect under extreme threat, and chronic depersonalization is hypothesized to represent that system staying engaged long after the threat that switched it on. Coherent is not demonstrated, and neither study was designed to show the difference.
Considering naltrexone, four years in
Four years without meaningful movement on the core symptom is long enough that I want naltrexone on the table now, not held in reserve after another year of therapy adjustments. Nuller and colleagues' 2001 pilot reported marked improvement in seven of fourteen patients and complete resolution in three more, and Simeon and Knutelska's 2005 open trial of naltrexone itself found four of fourteen markedly improved, with symptoms down an average of 30% across three dissociation scales. I'll say plainly that neither was controlled, and that Nuller used intravenous naloxone rather than the oral drug we'd actually be starting. This isn't a drug reached for because nothing else has worked; the endogenous opioid system is independently implicated in stress-induced numbing, and chronic depersonalization looks a great deal like that system staying engaged long after the threat that switched it on. It's also a low-risk trial to run — naltrexone isn't sedating, isn't habit-forming, and carries nothing like the burden a mood stabilizer or antipsychotic would.
You're right that the safety margin is real — I'm not worried about naltrexone hurting her. But "a genuine signal" is doing a lot of work in that description, and you've conceded most of my objection already — different drug, different route, no control arm, fourteen patients. That is the entire literature, and there has been no randomized trial in the twenty years since. What worries me more than the size of it is the dose: Simeon and Knutelska averaged 120 mg a day and went as high as 250, against a labeled 50. If we start her at 50 and she doesn't improve, we will not know whether the drug failed her or whether we never gave her what was actually studied. Her therapy has produced real, ongoing gains over eighteen months. Before reaching for a drug with an evidence base this thin, I'd want to see whether increasing session frequency or adjusting technique moves the baseline further, the way it's already been moving.
"Whether adjusting technique moves the baseline further" assumes we're still in the part of therapy where technique adjustments reliably produce movement, and I'm not confident we are — her gains over the last six months have been noticeably slower than the twelve before it, which is at least as consistent with approaching a real plateau as it is with simply not having found the right adjustment yet.
I don't think this has to be sequential. She's twenty-seven, she's been managing this since she was twenty-three, and a low-risk medication trial run alongside therapy — not instead of it — costs her very little if it fails and gives meaningful time back if it works. I'd propose eight to twelve weeks, tracked against the Cambridge Depersonalization Scale rather than her own global sense of "better," with an explicit agreement up front about what continuing versus stopping actually looks like.
Agreed: an eight-to-twelve-week naltrexone trial, started at 50mg daily with planned titration upward toward the dose range the published open trial actually used, run alongside unchanged weekly therapy rather than in place of it, with the Cambridge Depersonalization Scale readministered at the end of the trial window as the primary outcome measure rather than her own global impression alone.
Not agreed, and carried forward rather than smoothed over:
A meaningful partial reduction in CDS score would justify continuing past twelve weeks, even without full resolution.
Given the drug's off-label, thin-evidence status, only a clearly substantial reduction should justify continuing it beyond the trial window.
The disagreement wasn't resolved; it was made explicit enough that the twelve-week reassessment will have to settle it rather than default to whichever position speaks first at that visit.