Naltrexone for Self-Harm and Dissociative Symptoms in BPD: What the Evidence Actually Shows
A nursing student's dissociative flashback-like episodes have not responded to SSRIs or eighteen months of therapy. The disagreement is whether an open-label positive trial or a more rigorous, null-trending blinded one should carry more weight in the decision.
T.N., a 22-year-old nursing student, moved into her first apartment alone three weeks ago, after two years living with her older sister following a psychiatric hospitalization at twenty. The move was something she had wanted for over a year — a real marker of independence she had been working toward in therapy — but her flashback-like episodes, which had become less frequent while she lived with her sister, have returned twice in the past two weeks, both times without an identifiable external trigger she can name.
She was diagnosed with borderline personality disorder at nineteen, following years of self-harm that began in adolescence and a hospitalization after a significant overdose at twenty. Her episodes follow a specific, consistent pattern: a sudden sense of watching herself from outside her body, reduced pain sensation, and, on her worst days, a compulsion to cut that she describes as trying to “feel like I'm actually inside myself again.” Two SSRI trials over the past three years — sertraline, then escitalopram — have modestly helped her baseline anxiety but done nothing for these specific episodes, which is consistent with how dissociative symptoms generally respond to serotonergic antidepressants: they typically don't, since the symptom isn't primarily a serotonergic one.
She has been in weekly DBT-informed therapy for eighteen months, which has genuinely reduced her overall self-harm frequency, but the dissociative episodes themselves have not tracked with her therapy progress the way her mood and interpersonal reactivity have — they seem to run on a separate track, which is why she is now asking specifically about naltrexone, a drug she read about that targets the opioid system rather than serotonin. Her care team is being asked to write that prescription on the strength of a result that pointed the right direction and still came up short of significance — and has to decide, honestly, how much that's actually worth.
Follow-up visit, two weeks after moving out on her own
There's a real physiologic rationale here, and it's not just theoretical. Bohus and colleagues' 1999 open-label trial gave naltrexone to women with BPD experiencing exactly this kind of dissociation — flashbacks, depersonalization, analgesia — and found a highly significant reduction in both the duration and intensity of those episodes, along with a real drop in reported flashback frequency. Given how few pharmacologic options exist for dissociation specifically, and given that her episodes are recurring right as she's living alone for the first time with less immediate support around her, I think there's a reasonable case for trying it now rather than waiting.
The rationale is real and the open-label result is genuinely striking — I'm not dismissing that.
But the strongest evidence we actually have points the other way. Schmahl and Kleindienst's 2012 report — two small double-blind, placebo-controlled crossover trials, twenty-nine patients in total, built specifically to correct for the lack of blinding in the earlier open-label work — found that both the duration and intensity of dissociative symptoms were numerically lower under naltrexone than under placebo, but the difference was too small to reach statistical significance. That's a meaningfully different finding than “it doesn't work,” but it's also a meaningfully different finding than the open-label result implies on its own. Prescribing off the open-label data while treating the more rigorous null result as a footnote risks giving her more confidence in this drug than the actual evidence supports.
I don't think those two findings are as contradictory as they sound — the blinded trial found the same direction of effect, just underpowered to call it significant, which is a real possibility when the pooled total across both crossover trials is twenty-nine. That's different from a trial that found naltrexone doing nothing at all. Given the safety profile is favorable — non-habit-forming, generally well tolerated — I'd frame this as an individual trial rather than a population-level bet: a defined eight-week course, with her logging her own episode frequency and intensity the way she already tracks other symptoms for therapy, and a real decision point at the end based on what actually happens to her, not on what the average patient in a twenty-nine-person trial did.
Agreed: an eight-week naltrexone trial, with T.N. logging dissociative-episode frequency and intensity using the same tracking method she already uses for therapy. Escitalopram continues unchanged. The plan was explained to her directly with the actual evidence behind it, including the underpowered blinded-trial result, rather than presented as settled.
Naltrexone continues, understood explicitly as an individual response rather than proof the drug works generally — the distinction the Clinical Pharmacologist asked to keep visible in her chart.
Naltrexone is stopped, and the team returns to DBT-informed therapy alone as the primary approach to the dissociative episodes, without having lost meaningful time to find out.
Not resolved, and named directly to T.N. rather than smoothed over: an eight-week individual trial, even a positive one, cannot by itself distinguish a real drug effect from expectation or natural fluctuation — a limitation the Clinical Pharmacologist wanted on record regardless of how the trial turns out.