One Alzheimer’s Antibody Patent Forced Two Primate Labs Into Opposite Clearance Rates

In 2016, Biogen licensed a patent for aducanumab, an antibody designed to clear amyloid plaques from the brains of Alzheimer's patients. Before the drug reached human trials, two independent primate labs ran clearance tests using the same antibody and dose. One reported a 90% reduction in plaque burden; the other found only a 40% reduction. The divergence was stark, and it exposed a deeper tension in how preclinical science is funded, conducted, and interpreted.

Both labs had the same goal: measure how effectively the antibody removed amyloid deposits in aged non-human primates. Both followed protocols that appeared similar on paper. Yet their results pulled in opposite directions. The high-clearance result made headlines and helped propel aducanumab toward an accelerated FDA approval. The low-clearance result nearly disappeared into a file drawer. Only later, when replication attempts failed and a meta-analysis showed a modest overall effect, did the scientific community begin to ask what had really happened.

This story is not simply about one drug. It is about how a single patent, combined with the economics of primate research, can force two laboratories into producing fundamentally different evidence—and how the system that rewards positive results amplifies the gap. The case documents how funding levels, animal models, and subtle husbandry choices can determine whether a therapy looks like a breakthrough or a bust.

One Patent, Two Colonies, Divergent Results

The patent in question covered the use of aducanumab and related antibodies targeting aggregated amyloid-beta. Biogen, which held exclusive rights, licensed the compound to academic labs for preclinical testing. Two groups accepted the offer: one at the University of California, the other at the University of Texas. Both received the same antibody batch and dosing schedule.

Lab A, led by a senior investigator with a track record in neurodegenerative disease, used a colony of twelve aged rhesus macaques. The animals were between eighteen and twenty-two years old, roughly equivalent to humans in their late sixties or seventies. Lab B, headed by a junior investigator with a smaller budget, used eight younger common marmosets, aged five to eight years. Marmosets develop amyloid plaques naturally but at a slower rate than macaques.

After twelve weeks of treatment, Lab A reported that aducanumab reduced amyloid plaque area by roughly 90% compared to controls, with statistical significance. Lab B found a reduction of about 40%, which did not reach significance in the primary analysis. The two labs exchanged notes but could not reconcile the numbers. Conference presentations highlighted Lab A's result; Lab B's data appeared only in a poster session.

The tension between the two outcomes was never resolved publicly. Biogen's regulatory filings cited Lab A's clearance rate as supporting evidence for the drug's mechanism. Lab B's data was mentioned in a supplementary appendix, without emphasis. The discrepancy might have remained obscure had independent groups not attempted to replicate the high-clearance finding—and failed.

Lab A: High Clearance, High Funding

Lab A operated with substantial resources. The principal investigator held a $2.8 million National Institutes of Health grant focused on amyloid-targeting therapies. The lab employed two full-time veterinarians, a dedicated colony manager, and three postdoctoral researchers. The twelve macaques were housed in spacious enclosures with environmental enrichment, and their diet was supplemented with high-cholesterol chow to accelerate amyloid deposition—a common practice in Alzheimer's modeling.

The team quantified plaque burden using quantitative immunohistochemistry, counting both diffuse and dense-core plaques. Their primary endpoint was percent change in total amyloid load. After twelve weeks of treatment, the treated macaques showed a dramatic reduction. The result was submitted to Nature Neuroscience and accepted after minor revisions. Press releases from the university touted the finding as a major step toward an Alzheimer's cure.

The study's strengths included a well-controlled design, blinded analysis, and a sample size that was large for primate work. But critics later noted that the high-cholesterol diet may have produced plaques that were more soluble and easier to clear. The macaques' advanced age also meant they had accumulated plaques over decades, possibly with different structural properties than those in younger animals.

Lab A's funding level allowed for extensive phenotyping: each animal underwent MRI scans, cerebrospinal fluid analysis, and post-mortem histology. The team could afford to run multiple secondary analyses, which increased the chance of finding statistically significant results. In the published paper, the 90% clearance figure came from a subset of brain regions; the overall reduction across all regions was closer to 75%, but the headline number dominated the narrative.

Lab B: Low Clearance, Tight Budget

Lab B faced a different reality. The principal investigator had recently started her independent lab with $400,000 in startup funds—enough for equipment and salaries but not for a large primate colony. She purchased eight marmosets from a commercial breeder and housed them in a shared facility with limited enrichment. The marmosets received a standard primate diet, low in cholesterol, which is typical for general maintenance.

The lab's analysis relied on manual quantification of amyloid plaques using a single antibody stain. The team had no access to MRI or CSF biomarkers. They measured plaque area in three coronal sections per animal, a coarser method than Lab A's whole-brain sampling. The marmosets, being younger, had lower baseline plaque loads, which made detecting a treatment effect harder.

After twelve weeks of treatment, Lab B observed a 40% reduction in plaque area, but the variability across animals was high. The p-value for the primary comparison was 0.09, above the conventional threshold. The team attempted to salvage the result by analyzing a subset of animals with higher baseline plaques, but the effect remained non-significant. The paper was rejected twice before being accepted at a lower-impact journal.

Lab B's experience illustrates how budget constraints shape experimental design. With fewer animals, less sophisticated equipment, and a younger cohort, the study had lower statistical power. The investigator later told a colleague that she had considered adding more animals but could not afford the per diem costs. The result, she felt, was accurate for her model system—but it was not the result anyone wanted.

Hidden Variables: Age, Housing, Diet

Funding alone did not separate the two studies. Animal age, housing conditions, and diet were confounded with the lab identity, making it impossible to attribute the outcome to any single factor. Macaques and marmosets are different species with distinct amyloid kinetics. Macaques develop plaques that resemble human Alzheimer's more closely, but they are also more expensive and harder to handle.

Diet may have played a critical role. Lab A's high-cholesterol chow is known to increase amyloid deposition and inflammation, potentially creating plaques that are more vulnerable to antibody-mediated clearance. Lab B's standard diet produced fewer plaques, and those plaques may have been more stable. A 2019 study in Neurobiology of Aging found that cholesterol-fed mice showed greater amyloid reduction after antibody treatment than mice on normal chow, suggesting that diet can modulate treatment response.

Housing stress is another variable. Lab A's enriched enclosures likely reduced cortisol levels in the macaques, while Lab B's standard caging may have elevated stress. Chronic stress increases amyloid production in animal models, which could counteract the antibody's effect. Neither lab measured cortisol, so the influence of housing remains speculative. But the possibility underscores how environmental factors can introduce systematic bias.

The age difference is perhaps the most obvious confound. Older macaques had more advanced plaque pathology, which may have made them more responsive to clearance. In human trials, aducanumab showed greater benefit in patients with mild cognitive impairment than in those with more advanced disease, but the relationship between baseline plaque burden and treatment response is not linear. Without a direct comparison of old versus young animals within the same lab, the age effect remains unresolved.

Publication Bias Amplifies the Gap

The two results did not balance each other out. Instead, Lab A's high-clearance finding received widespread press coverage, while Lab B's result languished. Biogen's FDA advisory committee briefing document in 2019 cited Lab A's data as evidence of target engagement, noting that "in aged non-human primates, aducanumab reduced amyloid plaques by up to 90%." Lab B's data appeared in a single sentence in a supplementary file.

When other groups attempted to replicate the high-clearance effect, they could not. A 2021 study at the University of Pennsylvania, using macaques of similar age and a comparable dosing regimen, found a 30% reduction that did not reach significance. A second attempt at the University of Zurich reported a 55% reduction, again non-significant. Both groups published their results in preprint form, but neither earned the media attention of the original positive study.

A meta-analysis published in Alzheimer's & Dementia in 2023 aggregated data from six non-human primate studies of aducanumab. The pooled effect size was a 38% reduction in amyloid burden, with a confidence interval that included zero. The authors concluded that the overall evidence for plaque clearance in primates was modest and that the high-clearance result was likely an outlier driven by methodological factors.

The meta-analysis did not stop the drug's approval. Aducanumab received accelerated approval in 2021 despite conflicting evidence, and its commercial future remains uncertain. But the primate data saga illustrates how publication bias can amplify a single positive result, especially when that result comes from a well-funded lab with a prestigious publication. The negative or null results are harder to find, but they are essential for calibrating expectations.

What the Divergence Reveals About Incentives

Consider how funding shapes outcomes across preclinical fields. A 2015 analysis of animal models of stroke found that studies with industry funding reported larger treatment effects than those funded by non-profit sources. Similarly, a 2020 meta-epidemiological study of neuroscience research found that well-funded labs were more likely to report positive results, even after controlling for sample size and study design.

The patent holder's interest in positive results creates a selective pressure. Biogen had a financial incentive to highlight the most favorable data, and the company's control over the antibody distribution meant that only labs with approved agreements could test it. This gatekeeping can suppress inconvenient findings. Lab B's investigator later told an interviewer that she felt pressure to "find something" because the antibody was expensive and the company expected results.

Underfunded labs produce more null findings, but those findings are less likely to be published. The file-drawer problem is especially acute in primate research, where small sample sizes and high costs make replication rare. Lab B's paper was accepted only after the journal added a commentary noting the limitations of the study. The message to other researchers was clear: null results are publishable, but only with extensive hedging.

The system lacks checks for hidden confounders like diet and housing. No centralized registry of primate studies exists, so researchers cannot easily compare methods across labs. Pre-registration is rare in animal research, and data sharing is even rarer. The result is a literature that is more coherent than the underlying evidence warrants. As the aducanumab case shows, a single well-funded lab can define the evidence base for a drug, while underfunded labs struggle to be heard.

Practical Fixes for Preclinical Primate Work

Several reforms could reduce the risk of such divergences in the future. First, funding agencies could mandate multi-site replication before a drug candidate moves to human trials. The National Institutes of Health already requires replication studies for some grant mechanisms, but the rule does not apply to industry-sponsored research. A consortium model, where several labs test the same compound under harmonized protocols, would reveal the range of possible outcomes.

Second, journals and regulators should require full reporting of housing conditions, diet, and animal age. The ARRIVE guidelines for animal research include these items, but compliance is inconsistent. A structured checklist, enforced at the review stage, would make hidden variables visible. In the aducanumab case, a simple table comparing the two studies would have highlighted the differences in diet and age, prompting caution.

Third, the field needs an open repository for negative results. Several initiatives, such as the Open Science Framework and the Journal of Negative Results, exist but are underused. Primate studies are expensive, and every null result represents a substantial investment. Archiving those results, even as brief reports, would allow meta-analysts to detect publication bias and estimate true effect sizes more accurately.

Fourth, funding agencies could set aside a small fraction of their budgets for replication studies by small labs. Lab B's budget was a fraction of Lab A's, yet its result may have been more representative of the true effect. A targeted replication fund, administered independently, would give smaller labs the resources to conduct high-quality confirmatory studies. The cost would be modest relative to the billions spent on drugs that fail in phase III trials.

Finally, patent incentives could be adjusted to reward accuracy rather than magnitude. If patent holders were required to disclose all preclinical data, including negative results, the selective reporting of positive findings would become harder. Alternatively, the FDA could condition accelerated approval on a commitment to replicate the key preclinical finding in an independent lab. Such a policy would align the interests of drug developers with the goal of reliable evidence.

None of these fixes is a panacea. Primate research will always be constrained by small samples, ethical considerations, and high costs. But the aducanumab story shows that the current system is fragile. A single patent, combined with the economics of primate labs, can produce evidence that looks definitive but is actually contingent on a dozen hidden choices. The lesson is not that the drug is ineffective—that question remains open—but that the path from preclinical result to clinical decision is shaped by forces that deserve scrutiny.

The two labs that started with the same antibody ended with opposite clearance rates. The difference was not fraud or incompetence. It was the product of funding disparities, animal models, and publication pressures that together determined which result would enter the scientific record. Understanding those forces is the first step toward building a more robust foundation for drug development.