Evaluating Real-World Applicability of Alzheimer’s Biomarkers in Clinical Settings

A CTAD presentation reported that the Lumipulse G pTau217/Aβ42 plasma ratio had 100% sensitivity but only 23% specificity in a 252-person tertiary memory‑clinic cohort—yielding a high false‑positive rate that could prompt unnecessary confirmatory testing and patient distress. This profile suggests that FDA‑cleared ratio cutpoints may not transfer directly to routine clinical populations without local validation.

By contrast, the same FDA‑cleared ratio performed much better in risk‑enriched research cohorts (≈99% sensitivity and ≈85% specificity), underscoring an environment-dependent discrepancy. Likely drivers include higher pretest probability in research participants, fewer comorbidities that confound plasma markers, and referral patterns that concentrate true positives. These factors plausibly reduce specificity in real clinics, so research-cohort specificity estimates should be viewed as optimistic for general practice.

Investigators also showed that changing thresholds—specifically customized in‑house cutpoints and pTau217-only thresholds—shifted the sensitivity/specificity tradeoff toward fewer false positives. In‑house ratio cutpoints increased clinical‑cohort specificity to ~79% while maintaining 100% sensitivity and raised research‑cohort specificity to ~96%; pTau217 alone yielded ~84% specificity in the clinical cohort and ~96% in research with minimal sensitivity loss. Threshold selection can therefore materially change positive rates in routine clinics.

When clinic pretest probability departs from the research setting, discordant results are inevitable and call for pragmatic interpretation strategies. That's why it's important to interpret plasma results in context: escalate to definitive testing when it will change management, and use serial assessment for borderline or unexpected results.

Operational priorities for memory clinics are test selection and local validation: run known positive and negative samples when available, derive in‑house cutpoints from local case‑mix, document validation methods and performance metrics, and prepare patient‑facing reports that specify probability ranges, recommended confirmatory steps, and follow‑up timing. Establishing local validation and clear reporting practices should precede widescale clinical deployment.

Key Takeaways:

• The Lumipulse ratio showed 100% sensitivity but only 23% specificity in a real‑world memory‑clinic cohort.
• Memory‑clinic patients and clinicians evaluating ambiguous plasma positives are most affected by low specificity.
• Validate locally, consider alternative thresholds or pTau217 alone, and reserve confirmatory CSF/PET for cases where results will change management.