By Makoto Ishii, MD, PhD
Assistant Professor of Neuroscience and Neurology, Feil Family Brain and Mind Research Institute, Department
of Neurology, Weill Cornell Medical College
Dr. Ishii reports no financial relationships relevant to this field of study.
In a population-based longitudinal study of older individuals without dementia, the inclusion of imaging biomarkers for amyloid, tau, and neurodegeneration modestly improved the ability to predict memory decline compared to a model that only used clinical data and APOE genotype.
Jack CR Jr, Wiste HJ, Therneau TM, et al. Association of amyloid, tau, and neurodegeneration biomarker profiles with rates of memory decline among individuals without dementia. JAMA 2019;321:2316-2325.
For more than 100 years, Alzheimer’s disease (AD) could be diagnosed definitively only by postmortem histopathological confirmation of amyloid-beta plaques and tau neurofibrillary tangles. With the recent development of cerebrospinal fluid (CSF) immunoassays and positron emission tomography (PET) tracers specific for amyloid-beta and tau, it is possible to detect these neuropathological hallmarks in vivo before the onset of dementia. By using specific cutpoints, an individual can be classified as abnormal (+) or normal (-) for amyloid (A), tau (T), or neurodegeneration (N). A diagnostic framework based on the AT(N) biomarker profile has been constructed that biologically defines AD by pathological changes in A, T, and N. However, it is not known if the biological information obtained by the AT(N) profile would be clinically meaningful. For example, in predementia individuals, it is not clear if memory decline varies by the AT(N) profiles or if the use of AT(N) profiles would improve the ability to predict memory decline.
Some of the originators of the AT(N) framework sought to address these questions by examining 480 nondemented subjects enrolled in the Mayo Clinic Study of Aging, a population-based study of cognitive aging in a geographically defined population of Olmsted County, MN, with a median follow-up of 4.8 years (interquartile range [IQR], 3.8-5.1). Amyloid PET, tau PET, and MRI measures of cortical thickness were obtained at baseline to evaluate for A, T, and N, respectively. The primary outcome was a numeric memory composite score based on three delayed recall memory tests. Age, sex, education, APOE genotype, and a composite cardiovascular and metabolic conditions (CMC) score were used as variables for the clinical prediction model.
Of 480 subjects, 99% of study participants were self-reported white, and 44% were women. For the AD biomarker profiles, there were 140 (29%) A-T-(N)-, 33 (7%) A-T+(N)-, 81 (17%) A-T-(N)+, 22 (5%) A-T+(N)+, 54 (11%) A+T-(N)-, 24 (5%) A+T+(N)-, 69 (14%) A+T-(N)+, and 57 (12%) A+T+(N+) subjects. The A+T+(N)+ group was older, with a median age of 83 years (IQR, 76-87) compared to the median age of 67 years (IQR, 65-73) for the A-T-(N)- group. Ninety-two percent of study participants were cognitively unimpaired, with the A+T+(N)+ group having the largest proportion of mild cognitive impairment (30%). The proportion of APOE E4 carrier was greater among the four A+ groups compared to the four A- groups (40% vs. 21%; P < 0.001).
In the clinical prediction model, age and APOE E4 status were significantly associated with a faster rate of memory decline, but sex, education, and CMC score were not. Adding the AT(N) biomarker model to the clinical prediction model led to a relatively small but statistically significant improvement in predicting memory decline (likelihood ratio test P < 0.001 with R2 increasing from 0.26 to 0.31). Furthermore, the A+T+(N)+, A+T+(N)-, and A+T-(N)+ groups had the fastest rates of memory decline compared to the other five groups (P = 0.002). Finally, an estimated 46% of memory decline in older predementia individuals was associated with abnormal AT(N) biomarker profile.
COMMENTARY
Despite the increasing use of AD biomarkers in research, the AT(N) classification remains a research construct that requires clinical validation. This important study by Jack et al is the first to use longitudinal clinical outcomes to examine the association of the AT(N) classification system with memory decline in a relatively large number of predementia individuals. As predicted, the A+T+(N+) group had the highest proportion of mild cognitive impairment, and those groups with the worst pathological changes A+T+(N+), A+T+(N-), A+T-N(+) had the fastest rate of memory decline. These results appear to validate the AT(N) profile as an appropriate framework for classifying and risk-stratifying predementia individuals. However, adding AT(N) profiles led to only a modest improvement in the ability to predict memory decline, which as the study authors noted is of uncertain clinical importance.
Additional study limitations are worth noting. First, the AT(N) profile used strict cutpoints, which may be challenging to define as these measures are part of a natural continuum. Second, the clinical prediction model that was used did not incorporate cognitive measures. It is not clear if the modest gains in predicting memory decline made with the addition of AT(N) biomarkers would be further reduced with the use of cognitive measures, which are common in clinical practice and significantly less expensive than imaging biomarkers. Third, it is not known whether the findings from this study can be generalized to other biomarkers (e.g., CSF) or to more diverse populations.
Despite these limitations, the findings from this study not only advance our fundamental understanding of AD but suggest an important role that AT(N) biomarker profiles could have in future clinical trials. A major limitation of past trials was the inability to identify those individuals with mild or no cognitive symptoms who had the highest risk for memory decline. It is plausible that past clinical trials failed in part because of the inclusion of subjects who ended up having very modest decline in memory during the short study period. This would make it difficult to discern any significant benefit from an intervention. By stratifying individuals to the different AT(N) biomarker profiles, investigators conceivably could better prognosticate and target cognitively intact or mildly impaired subjects with the highest risk for memory decline, which would significantly increase the chance for successfully finding an effective therapy for AD.
