Does the monoclonal antibody AER002 improve Long COVID symptoms?

Testing viral persistence as a Long COVID driver

The trial treats a mechanistic hypothesis as its object of study. SARS-CoV-2 RNA or protein has been detected months after acute infection in some people, and persistence of virus has been proposed as one biological driver of Long COVID. If residual virus sustains symptoms, an antibody that binds SARS-CoV-2 spike protein might help clear it. AER002 is a fully human IgG1 monoclonal antibody against the receptor binding domain of spike, with Fc modifications intended to prolong its half-life; the paper describes it as a tool to probe whether clearing persistent antigen improves symptoms.

The design follows from that question. Thirty-six adults meeting the WHO Long COVID case definition were randomized 2:1 to a single intravenous infusion of AER002 1200 mg or matching placebo and followed for 360 days. The 1200 mg dose came from a prior phase 1 dose-escalation study in healthy adults and was chosen to maximize exposure, because no serum concentration threshold has been established for activity against presumed persistence. The trial is explicitly exploratory: it was designed primarily to assess safety and tolerability and to generate preliminary effect estimates for later, adequately powered trials.

A highly symptomatic cohort roughly two years in

Eligibility required at least two moderate or severe symptoms that were new or worsened since a test-confirmed SARS-CoV-2 infection, had lasted at least 60 days, and were not attributable to another cause. The infection also had to predate August 15, 2022—the period when AER002 was expected to retain neutralizing activity against circulating variants. People who had ever received a SARS-CoV-2 monoclonal antibody, or convalescent plasma or a vaccine in the prior 60 days, were excluded, as were those with active cardiovascular disease, uncontrolled HIV, moderate or severe immunocompromise, or ME/CFS or dysautonomia predating infection.

The enrolled group was sick and chronically ill. Median age was 42 years (IQR 32–53), 56% were female, and the infection to which Long COVID symptoms were attributed occurred a median of 740 days (IQR 576–1113) before enrollment. Participants reported a median of 14 Long COVID symptoms in the prior 30 days; 86% reported post-exertional malaise and 58% met 2015 Institute of Medicine criteria for ME/CFS. Median quality of life on a 100-point visual analog scale was 49 (IQR 35–63). That long illness duration matters for interpretation: the paper notes persistence has so far been linked to symptoms over roughly 3–14 months, considerably shorter than this cohort's exposure.

No group-level gain in physical health at day 90

The primary efficacy endpoint was change in the PROMIS-29 Physical Health Summary Score (PHSS) between baseline and day 90. PROMIS instruments are scaled to a T-distribution with a mean of 50 and a standard deviation of 10 points, and the physical and mental health summary scores are derived from seven domains including physical function, fatigue, pain interference, and sleep disturbance. Day 90 was chosen to give AER002 time to affect a putative viral reservoir while systemic concentrations remained substantial.

No significant difference in PHSS change emerged between arms. The same null pattern held across the secondary and exploratory measures the paper reports: quality of life on EQ-5D-5L and its visual analog scale, objective measures of physical function and cognition, and blood-based biomarkers. Safety was the other primary readout, and AER002 was safe and well tolerated, with the primary safety endpoint defined as grade 3 or greater adverse events judged definitely, probably, or possibly related to treatment. Because this was a small exploratory trial, the absence of a detected difference is a negative result for the group average, not a demonstration that the antibody has no biological effect in any subgroup.

A post-hoc responder signal tied to antibody levels

The one positive-looking result comes from an analysis the paper labels post-hoc. Using the Patient Global Impression of Change (PGIC)—a participant's own rating of change in activity limitations, symptoms, emotions, and overall quality of life attributed to the intervention—16 participants were classified as responders and 20 as non-responders. Responders had lower baseline SARS-CoV-2 antibody levels, and a multiple logistic regression using AER002 exposure (area under the concentration-time curve) plus baseline anti-SARS-CoV-2 levels predicted self-reported response, with p < 0.05 for anti-S, anti-S1, and anti-RBD.

This is a different question from the primary endpoint, and the two should not be merged. PGIC captures perceived change; PHSS captures self-reported physical health on a standardized scale; neither is an objective functional test. The authors themselves call these findings hypothesis-generating and state they do not establish efficacy. The subgroup counts are small—among AER002 recipients, 10 responders and 14 non-responders—so the pattern is best read as a candidate selection variable for future trials rather than as a treatment effect readers can apply.

Tissue immune activation without clinical translation

Exploratory work reached beyond blood. Optional [18F]F-AraG PET-CT imaging, available for 6 AER002 and 4 placebo recipients, compared baseline and post-infusion standardized uptake values in selected regions of interest, including submandibular glands; the paper reports potential biological effects on tissue-based immune activation from these exploratory analyses. Only one participant who underwent gut biopsy had detectable SARS-CoV-2 RNA in gut tissue at baseline, and that individual reported improvement.

These observations are too sparse to carry the efficacy question. The imaging subsets are single-digit samples, and the paper states that strict adjustments for multiple comparisons were not used given the limited sample size. Their value is directional: they suggest the antibody may engage tissue compartments, which is why the authors argue future trials should pair clinical endpoints with mechanistic ones rather than relying on blood biomarkers alone.

Why a negative trial can still inform design

The paper offers several explanations for the null primary result, and they point to design choices rather than to a refuted hypothesis. No validated, accessible biomarker of SARS-CoV-2 persistence existed when the study began, so participants could not be selected for actually harboring persistent virus; the authors call this a likely contributor to missing the primary outcome. Participants had also been ill for about two years, longer than the window in which persistence has been linked to symptoms, raising the possibility that downstream processes sustain disease independently of virus.

Two further constraints sit with the results. Thirty-three percent of participants had a SARS-CoV-2 reinfection after August 15, 2022, when AER002 was expected to have lost neutralizing capacity, and the antibody had reduced but retained in vitro activity against BQ.1 while later variants waned further. Separately, no outcome measure used here has been formally validated in Long COVID, so the instruments may have missed relevant dimensions of illness. The authors also note that immune dysfunction could promote persistence or sustain disease independently of virus, in which case immunomodulatory or combination approaches may be required.

The priority is confirming persistence before treatment

The paper's own conclusion is that AER002 showed no group-level benefit over placebo, and its forward-looking recommendations are specific: future trials targeting viral persistence should confirm persistence at enrollment, enroll larger samples to offset heterogeneity and reinfection, use agents with broad activity against historical and modern variants, and screen for favorable immune phenotypes such as low baseline antibody levels or use dosing strategies that sustain therapeutic levels.

For readers weighing this evidence, the decisive split is between the group average and the candidate subgroup. The randomized comparison answered the efficacy question negatively at day 90 on PHSS and across reported secondary measures; the responder analysis answered a different, post-hoc question about who perceived change, and it identified low baseline anti-S, anti-S1, and anti-RBD levels combined with higher drug exposure as the pattern worth testing. The trial's contribution is therefore a design lesson—select for persistence and exposure, not merely for the Long COVID diagnosis—rather than a treatment option.

Sources used in this answer