Biomarker panel predicts memory decline among elders without dementia

  • Jack CR & al.
  • JAMA
  • 18 Jun 2019

  • International Clinical Digest
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.


  • A panel of biomarkers consisting of amyloid (A) assessed with positron emission tomography (PET), tau (T) assessed with PET, and neurodegeneration (N) assessed with MRI cortical thickness improved on clinical measures for predicting memory decline among older individuals without dementia.

Why this matters

  • Clinical relevance of the novel AT(N) panel is unknown.

Key results

  • 8% of participants had mild cognitive impairment; highest proportion in A+T+(N)+ group (30%).
  • Compared with clinical model, AT(N) panel improved prediction of performance (R2, 0.26-0.31; P<.001>
  • Memory decline differed across the 8 AT(N) groups (P=.002); fastest decline seen in A+T+(N)+ group, A+T+(N)– group, and A+T–(N)+ group.
  • For an 85-year-old APOE ε4 noncarrier, respective estimated of decline in these 3 groups (z score units/year):
    • –0.13 (95% CI, –0.17 to –0.09).
    • –0.10 (95% CI, –0.16 to –0.05).
    • –0.10 (95% CI, –0.13 to –0.06).

Expert comment

  • In an editorial, David Wolk, MD, and colleagues note some caveats, such as exclusion of cognitive measures from clinical variables, uncertainty around optimal biomarker cutoffs, and expense. “Despite these caveats, the study … represents an important contribution not only to advancing the conceptualization of AD, but also for putting this new framework to the test rapidly in a relatively large sample of participants,” they conclude.

Study design

  • Population-based cohort study of 480 nondemented Mayo Clinic Study of Aging participants aged ≥60 years.
  • Participants had clinical evaluation and PET, PET, and MRI cortical thickness measures, and ≥1 clinical follow-up.
  • Main outcome: composite memory score measured at 15-month intervals over median 4.8-year follow-up.
  • Funding: NIH; Alexander Family Professorship of Alzheimer’s Disease Research; GHR Foundation.


  • Test-retest variance in memory z scores.
  • Limited follow-up for assessing effect of some biomarkers.
  • Memory composite used does not capture all aspects of memory decline.