FDA-rejected selective NRI with documented publication bias and a weak, contested evidence base
Look, reboxetine has a real mechanism. NET inhibition is measurable, the selectivity profile is genuine, and the pharmacokinetics are reasonably clean at a 13-hour half-life with the (S,S)-enantiomer carrying most of the noradrenergic activity. We’re not saying it’s a fake compound. We’re saying it’s a compound where the important questions have been answered, and the answers weren’t encouraging.
The FDA rejection in 2001 was the first signal, but the real reckoning came in 2010 when Eyding et al. published their BMJ meta-analysis after obtaining unpublished trial data from Pfizer. The conclusion was stark: the published literature had significantly overestimated efficacy, the unpublished trials showed the compound was no better than placebo on the full dataset, and the benefit-to-risk picture for depression essentially collapsed. Thats not a mixed evidence situation. That’s a case where systematic publication bias produced a sustained false positive that shaped European prescribing for over a decade. The visible published record looked credible. The full record didn’t.
For bench research, the situation is slightly better but not compelling. Reboxetine can function as a selective NRI probe to isolate noradrenergic contributions in a given assay or behavioral model, and its selectivity for NET over SERT and DAT is well-characterized. The problem is atomoxetine serves the same role with more extensive preclinical literature, FDA approval in ADHD providing a legitimate pharmacological anchor, and broader commercial availability. As a tool, reboxetine is functional but not first-choice for most research programs doing serious noradrenergic work.
The mechanism itself, single-target NET inhibition, doesn’t generate the kind of questions that attract serious investigation right now. The frontier in noradrenergic biology is adrenergic receptor subtype pharmacology, locus coeruleus circuit dynamics, and NE’s role in stress-resilience and threat-appraisal circuits. Reboxetine can contribute peripherally to some of that work, but it’s not driving any active research program we’re aware of. When you check recent literature, it appears mostly as a comparator or a cautionary example of publication bias, not as the primary agent of interest.
What we don’t like: the evidence base is actively misleading if you read only the published record. Anyone designing a study that touches noradrenergic pharmacology and leans on pre-2010 reboxetine efficacy claims is making a methodological error. The Eyding analysis should be considered required reading before this compound enters any design. It’s available as a research chemical in several jurisdictions with straightforward sourcing, and honestly that may be the most practically useful thing we can say about it. The scientific return on investing bench time here is low, and better NRI probes exist if selective noradrenergic inhibition is the angle you need.