A trial that didn't hit its mark, but didn't miss entirely either

BioVie Inc. announced Monday that its experimental drug bezisterim failed to reach statistical significance across the full population of its Phase 2 ADDRESS-LC trial for Long COVID, a genuinely disappointing top-line result for a disease with an estimated 17 to 20 million affected adults in the U.S. and zero FDA-approved treatments. But buried inside that overall miss sits a more encouraging finding: in pre-specified subgroups of patients with the highest baseline symptom burden, high fatigue, cognitive impairment, and post-exertional malaise, bezisterim produced statistically significant improvements over placebo.

That's a genuinely nuanced result, and it's worth being precise about what it does and doesn't mean. A drug that fails its overall trial population but succeeds in a specific, pre-specified subgroup isn't the same as a drug that simply worked, nor is it the same as a drug that simply failed. It's a signal that requires careful interpretation, and one that both the company and outside experts are treating with appropriate caution.

What the trial actually tested

The Phase 2 ADDRESS-LC study, fully funded by a grant from the U.S. Department of War, formerly the Department of Defense, enrolled approximately 200 adults with Long COVID experiencing cognitive impairment and fatigue. Participants received either bezisterim, administered as a 20 mg oral capsule twice daily, or a matching placebo, in a randomized, placebo-controlled, signal-finding proof-of-concept design. Notably, the trial evaluated 22 separate clinical outcome measures without designating any single endpoint as the primary determinant of study success, a design choice that reflects how genuinely difficult it is to define and measure Long COVID's diffuse, multi-symptom presentation with any single metric.

Bezisterim itself, also known by its development code NE3107, is an oral drug that crosses the blood-brain barrier and is designed to reduce inflammation and improve insulin sensitivity without broadly suppressing the immune system, working through modulation of the ERK, NFκB, and TNF-alpha inflammatory pathways. The therapeutic rationale for testing it in Long COVID rests on a specific hypothesis: that lingering neurological symptoms like brain fog and fatigue may be driven by persistent circulation of spike protein fragments that trigger ongoing inflammation through NFκB activation, precisely the pathway bezisterim is designed to modulate.

The number that actually moved the needle

Across the full study population, the results were mixed in a way that BioVie itself characterized carefully: bezisterim showed a consistent numerical advantage over placebo on 21 of the trial's 22 individual endpoints, but that consistency in direction did not translate into statistical significance across the overall population. That's an important distinction. A drug trending in the right direction on nearly every measure, without reaching the statistical threshold needed to rule out chance, is a genuinely different result than a drug showing no discernible effect at all, even though neither result supports an immediate approval claim.

The subgroup findings tell a more specific story. Among patients entering the trial with the highest baseline levels of fatigue, cognitive impairment, and post-exertional malaise, arguably the patients experiencing the most severe and disabling form of Long COVID, bezisterim produced statistically significant improvements over placebo. Post-exertional malaise in particular, a hallmark symptom in which physical or mental exertion triggers a severe worsening of symptoms, has been one of the more disabling and treatment-resistant features of Long COVID, making any signal of improvement in that specific measure genuinely noteworthy regardless of the trial's overall result.

Why the safety data matters almost as much as efficacy

For a drug being considered in a patient population already managing complex, chronic symptoms, safety and tolerability carry real weight independent of efficacy. Bezisterim's safety profile compared favorably to placebo in this trial: 41.6% of bezisterim-treated patients experienced at least one treatment-emergent adverse event, compared with 55.9% in the placebo group. The most frequently reported drug-related adverse event was headache, affecting 4% of bezisterim patients versus 4.9% on placebo, and no serious adverse events occurred in the bezisterim arm, compared with one in the placebo group.

That safety profile carries particular significance given what BioVie described as extensive concomitant medication use among trial participants, meaning many Long COVID patients enrolled were already taking other medications for their symptoms. A drug that doesn't add meaningful additional adverse events or drug interactions on top of an already complex medication regimen is a genuinely relevant practical consideration for any Long COVID treatment candidate, since this patient population frequently manages multiple overlapping symptoms and existing prescriptions simultaneously.

What outside experts are saying, and what remains genuinely uncertain

BioVie CEO Cuong Do framed the subgroup results as reinforcing the underlying scientific rationale for the drug's mechanism, noting that "the growing body of research surrounding long COVID continues to reinforce the rationale for bezisterim's proposed mechanism, particularly its modulation of inflammatory pathways associated with symptoms relating to fatigue and 'brain fog.'" Michael Peluso, an associate professor of medicine at the University of California, San Francisco and an investigator on the trial, joined the company's Monday conference call alongside Long COVID patient advocate Ezra Spier, suggesting the research community around Long COVID treatment is engaging with these results directly rather than dismissing the trial outright.

That said, one independent financial analysis of the results was notably more cautious, flagging directly that "the Phase 2 trial did not achieve statistical significance on any endpoint" across the overall population and cautioning that "results observed in subgroup analyses may not be replicated in future trials." That's an important scientific caveat that applies broadly to subgroup findings in clinical research: when a trial doesn't pre-register a subgroup as its primary hypothesis with sufficient statistical correction for testing multiple subgroups, a positive result within that subgroup carries a meaningfully higher risk of being a statistical fluke than a result from the trial's primary, pre-specified overall analysis. BioVie has stated these particular subgroups were pre-specified before unblinding, which reduces but doesn't eliminate that risk.

What this means for a disease with no approved treatments

The stakes here extend well beyond one company's stock price or one drug's regulatory pathway. Long COVID remains a condition with an estimated 17 to 20 million affected American adults and no FDA-approved treatment specifically targeting its neurological symptoms, leaving patients and physicians largely managing symptoms individually rather than treating an underlying cause. In a therapeutic landscape this empty, even a mixed, subgroup-driven signal carries real weight, both for the patients directly affected and for the broader research effort trying to identify any effective intervention at all.

BioVie has separately been developing bezisterim for Alzheimer's disease and Parkinson's disease, having already reported topline results from a Phase 2 trial in early Parkinson's showing statistically significant improvements on motor and non-motor symptom measures. That broader pipeline gives the company multiple shots at demonstrating the drug's underlying anti-inflammatory mechanism translates into meaningful clinical benefit, even if the Long COVID indication specifically requires a larger, subgroup-focused follow-up trial to confirm today's results before moving toward any regulatory submission.

What comes next

Given the trial's proof-of-concept, signal-finding design and its mixed top-line result, the most likely near-term path forward is a larger, more targeted follow-up study specifically enrolling patients matching the high-symptom-burden profile that showed statistically significant benefit in this trial, rather than an immediate move toward regulatory filing based on today's data alone. Whether that follow-up trial confirms the subgroup signal or fails to replicate it will determine whether bezisterim becomes a genuine treatment option for the most severely affected Long COVID patients, or joins the long list of promising mechanistic candidates that ultimately couldn't translate laboratory rationale into confirmed clinical benefit. For now, Monday's results offer neither a clear victory nor a clear defeat, a genuinely uncertain outcome that reflects just how difficult Long COVID has proven to treat with any single therapeutic approach so far.