Creator: Help Dementia Editorial Team
Published: 2026-06-13T22:02:53

Reviewed by the Help Dementia Editorial Team — our editors review every article for accuracy against guidance from the National Institute on Aging, the Alzheimer’s Association, and peer-reviewed sources.

Smaller trials sits at the center of this dementia and brain health question.

Yes, smaller clinical trials can work more efficiently with better biomarker screening—but “work” doesn’t mean they’re automatically faster or cheaper. When researchers can identify which patients actually have the biological disease being studied, they reduce noise and dropout rates. A smaller group of truly affected people can show drug effects that a larger, mixed population might miss. This is why the Lecanemab trial (Clarity AD) for Alzheimer’s disease enrolled only 1,795 participants across two trials instead of the 3,000+ typical for late-stage dementia trials—the inclusion of amyloid-PET biomarker positivity meant nearly every participant had confirmed amyloid pathology.

The catch is that biomarker screening adds its own cost, time, and attrition. Requiring a PET scan or lumbar puncture before enrollment stops many willing participants from even entering the trial. Sites must invest in imaging infrastructure. The statistical power still depends on effect size and disease progression rate, not just on sample size. Smaller trials with rigorous biomarker selection can work, but they’re not a shortcut around the fundamental biology.

Why Do Biomarkers Matter in Smaller Trial Designs?
The Cost and Complexity of Biomarker-Selected Enrollment
Real-World Trial Examples and Disease Heterogeneity
Statistical Power and the Smaller-Trial Math
Dropout, Placebo Response, and the Hidden Costs of Small Samples
Presymptomatic and Prevention Trials—Where Biomarkers Are Essential
Blood Biomarkers and the Future of Streamlined Trials

Why Do Biomarkers Matter in Smaller Trial Designs?

Traditional dementia trials were built on clinical symptoms alone—if someone scored low enough on cognitive tests, they qualified. This meant enrolling people at vastly different disease stages and with different underlying pathologies. A patient with primarily tau tangles, another with vascular disease, and a third with Lewy bodies all lumped together as “cognitive impairment.” The drug that works for amyloid accumulation looks ineffective in a population where only 40% have amyloid as the primary driver. Biomarker screening acts as a filter. Instead of testing a drug on 500 cognitively impaired people (half of whom may not have the target pathology), you test it on 200 people confirmed to have amyloid positivity or tau burden or neuroinflammation—whatever the drug targets.

The smaller group has higher signal-to-noise ratio. Studies of anti-tau antibodies or amyloid vaccines have begun requiring phosphorylated tau (p-tau) blood biomarkers or tau-PET positivity precisely because lumping in people without tau pathology dilutes the drug effect and requires more participants to reach statistical significance. The tradeoff is accessibility. A trial requiring amyloid-PET or CSF biomarkers automatically excludes anyone without imaging access or unwilling to have a spinal tap. Rural participants, those with contraindications to MRI, and people uncomfortable with invasive procedures drop out. You gain statistical efficiency but lose demographic diversity and generalizability.

The Cost and Complexity of Biomarker-Selected Enrollment

Biomarker screening sounds straightforward until you price it. A single amyloid-PET scan costs $3,000–$5,000. A tau-PET scan costs $4,000–$6,000. Add MRI for exclusion of structural abnormalities, blood draws for p-tau phosphorylation states, and you’re looking at $8,000–$12,000 in pre-enrollment screening per participant. If your trial enrolls 300 people but screens 600 to find 300 biomarker-positive candidates, you’ve spent $4.8 million before a single participant receives the investigational drug. This screening cost often gets absorbed by the pharmaceutical sponsor or research institution, not the participant. But it still determines trial timeline.

Imaging sites have limited capacity. A trial requiring amyloid-PET at 20 centers across three countries doesn’t start enrollment overnight. Scheduling imaging for 500 potential participants, waiting for reads, then enrolling the 40% who are biomarker-positive—that’s 6 to 12 months of screening for what might be an 18-month treatment phase. Blood biomarkers (plasma p-tau, phosphorylated tau-181, p-tau217) have begun reducing this burden. A blood draw costs $200–$500, processes faster, and doesn’t require specialized imaging equipment. Trials like the AHEAD study (ADCOMS trial) used plasma p-tau-217 to stratify cognitively normal older adults at risk for amyloid and tau accumulation. This made screening cheaper and faster—but the blood biomarkers are still not perfect predictors of clinical decline, so even biomarker-selected cohorts showed variability in response.

Real-World Trial Examples and Disease Heterogeneity

The Clarity AD trial (lecanemab) enrolled only cognitively mild participants with amyloid and tau pathology confirmed by imaging. The result was a slowing of cognitive decline by 27% over 18 months—a clinically meaningful effect, but not dramatic. If the same drug had been tested in 1,200 people without biomarker screening (mixing MCI, mild dementia, and cognitively normal at-risk individuals), the effect might have been diluted to 15–18% because some participants would have non-amyloid pathology driving their cognitive loss. The DIAN-TU trial (Dominantly Inherited alzheimer Network Trial Trials Unit) took a different approach: it enrolled only carriers of known pathogenic mutations causing familial Alzheimer’s disease. These individuals have 100% certainty of developing the disease and often have known biomarker timelines. This genetic biomarker (mutation carrier status) allowed smaller, younger cohorts and longer follow-up windows before cognitive symptoms appeared.

The study could test preventive drugs in the presymptomatic stage—something impossible in a general population trial. The APOE4 genotype is another pseudo-biomarker. People with APOE4 have higher amyloid and tau burden and faster cognitive decline. Enriching for APOE4 carriers doesn’t require imaging; a single blood test identifies them. But APOE4 is imperfect—some carriers stay cognitively normal into their 90s, while some non-carriers develop early dementia. Using APOE4 alone as an enrollment criterion cuts screening cost but introduces more clinical heterogeneity than amyloid-PET selection.

Statistical Power and the Smaller-Trial Math

The sample size needed for a trial depends on three things: the effect size you expect the drug to have, the disease progression rate in your population, and your statistical power threshold (usually 80% power to detect the effect if it exists). Biomarker selection doesn’t magically reduce the number you need—but it can change the effect size. A drug slowing cognitive decline by 20% in a homogeneous, biomarker-positive population might produce a larger statistical effect than the same drug in a heterogeneous population where some people lack the target pathology. Larger effect sizes require fewer participants. If biomarker selection increases your detectable effect from 15% to 25%, you might drop from 400 to 250 participants.

But if your effect size is still 15% (because the drug just isn’t very potent), biomarker selection doesn’t shrink the sample size—it just makes your cohort more homogeneous. The risk is over-relying on biomarker selection to solve power problems. A small trial with perfect biomarker enrichment still fails if the drug doesn’t work or if disease progression is highly variable. Predictive biomarkers (those that predict who will decline) are rarer than diagnostic biomarkers (those that confirm disease presence). You can enrich for amyloid positivity easily; predicting who will decline cognitively over 18 months is much harder.

Dropout, Placebo Response, and the Hidden Costs of Small Samples

Smaller trials are vulnerable to dropout. If your 200-person trial expects 10% dropout and 20% of participants show only placebo response, you’re left with 144 people showing true drug response. Statistical noise from normal variation in cognitive testing swamps small numbers. A larger, less-selected trial of 600 people tolerates dropout better because your final analyzable population is still 480–500. Biomarker screening may paradoxically increase dropout. Participants who undergo expensive screening, receive imaging results, and learn they’re positive for amyloid accumulation or tau burden may become anxious or depressed—conditions known to affect cognitive test performance and trial adherence.

Some studies report 15–20% dropout in the first 6 months after positive biomarker feedback, especially in cognitively normal people who felt fine before being told they have brain pathology. The placebo effect in dementia trials is large and unpredictable. Cognitive tests can vary by 1–3 points just from practice effects or fluctuations in mood and sleep. In smaller trials, a few participants who improve unexpectedly (whether from placebo or random variation) can distort the results. Lecanemab’s effect—27% slowing of decline—is considered a modest win partly because it was detectable despite placebo effects in a trial large enough to absorb random noise. A biomarker-selected trial of 100 people might miss that same effect due to noise.

Presymptomatic and Prevention Trials—Where Biomarkers Are Essential

The strongest case for biomarker-selected smaller trials is in presymptomatic prevention. You cannot run a traditional trial of cognitively normal people without biomarkers—there’s no cognitive decline to measure yet, and most will never develop dementia in their lifetime. Selecting for amyloid or tau positivity creates a high-risk cohort where cognitive change can be detected in reasonable timeframes.

The AHEAD study enrolled cognitively normal older adults aged 55+ with elevated amyloid and/or tau on PET. Smaller regional cohorts became feasible because researchers knew participants had pathology. Traditional trials in this population would require thousands of cognitively normal people followed for decades to capture meaningful cognitive decline. The AHEAD design shows that biomarker enrichment isn’t just about efficiency—it’s about feasibility for prevention research.

Blood Biomarkers and the Future of Streamlined Trials

Plasma phosphorylated tau variants (p-tau181, p-tau217, p-tau-threonine) are shifting the economics of biomarker screening. A single blood draw, processed by ELISA or mass spectrometry, costs $100–$300 compared to $4,000 for PET imaging. Multiple trials now use blood p-tau to screen for amyloid and tau pathology without requiring neuroimaging. The AHEAD study used plasma p-tau-217 and amyloid-beta 42/40 ratio for enrollment stratification.

Lecanemab’s follow-up trials and next-generation anti-amyloid antibody trials increasingly use plasma biomarkers for initial screening, reserving PET confirmation for borderline cases. This accelerates enrollment and reduces site burden. A trial can screen 1,000 participants via blood biomarker in the time it takes to screen 200 via imaging. Even if blood biomarkers are 80% sensitive compared to imaging, the speed and cost efficiency often outweigh the loss of precision—especially when cognitive decline is the ultimate outcome measure, not biomarker status.