Adding the amyloid/tau/neurodegeneration (A/T/N) model of dementia to a clinical model may give an incremental but still significantly increased ability to predict cognitive decline over nearly 5 years, according to findings from a longitudinal cohort study of patients without dementia at baseline.
Although the A/T/N model is still intended only for research purposes, the study came to another important conclusion: About 50% of the memory change associated with normal aging was, in fact, caused by changes associated with Alzheimer’s disease, Clifford R. Jack Jr., MD, and colleagues wrote in JAMA.
The three groups with the fastest rates of memory decline all had abnormal amyloid and either abnormal tau and/or imaging signs of neurodegeneration. “This illustrated a dominant association of memory decline with amyloidosis but only when present in combination with tauopathy, neurodegeneration, or both,” Dr. Jack of the Mayo Clinic, Rochester, Minn., and coauthors wrote.
A/T/N, also known as the National Institute on Aging and Alzheimer’s Association Research Framework, is based on objective amyloid and tau biomarkers and imaging markers of neurodegeneration and is intended to more accurately differentiate Alzheimer’s from other dementias and, potentially, to stage the disease and predict and track decline. It generates eight clinical profiles that can identify Alzheimer’s, rule it out, or include it as a possible diagnosis.
The study comprised 480 elderly individuals enrolled in the Mayo Clinic Study on Aging. Median age of the participants ranged from 67 years in one of the eight clinical profiles (A–/T–/N–) to 83 years in another (A+/T+/N+). Most (92%) were cognitively normal; the remainder had mild cognitive impairment (MCI). They were followed for a median of 4.8 years.
Both amyloid and tau were measured with PET imaging; neuropathology was represented by MRI scans of cortical thickness. Most (n = 140) were negative for all biomarkers (A–/T–/N–). The group positive for all markers (A+/T+/N+) had the largest proportion of MCI subjects (30%). The apolipoprotein E epsilon 4 (APOE4) genotype was more common among the A+ groups than it was among the A– groups (40% vs. 21%).
The individual cognitive decline trajectories varied considerably by age and within each classification group. Only 7% of the A–/T–/N– group were 80 years or older, and only 2% of the A+/N+/T+ group were younger than 70 years.
In a clinical model, age and APOE4 status were significantly associated with faster rates of memory decline. Sex, education, and a cardiovascular/metabolic model were not, however.
“The estimated rate of memory decline in a 75-year-old individual who was an APOE4 noncarrier was –0.04 z-score units per year,” the authors wrote. “An 85-year-old individual who was also an APOE4 noncarrier could be expected to have a decline of –0.08 units per year, while a 75-year-old E4 carrier could be expected to have a decline of –0.08 units per year.”
Every 10 years of additional age was associated with a significant median worsening of 0.4 on z score for memory. A 4-year difference in education was associated with a 0.6-unit higher memory score, while APOE4 carriers had a 0.3-unit lower memory score.
The addition of the A/T/N model significantly improved the prediction of cognitive decline and memory score, although the rates of decline were still considerably variable. All of the A+ groups had the fastest decline rates.
“To place the predictive utility of biomarkers in clinical context, the decline in rates of memory for A+/T+/N–, A+/T–/N+, A+/T+/N+ [abnormal amyloid plus tau or neurodegeneration] were of similar magnitude to a 20-year increase in age and were twice that associated with APOE4 carriership,” they wrote.
A total of 88 participants had a second imaging visit at a median of 15 months. Most (n = 72) had no change in the A/T/N classification. A and T classifications were more stable (98% and 97%, respectively) than was N classification (84%).
A secondary analysis compared this model with generally accepted clinical and biomarker characteristics. Prior research has shown that prevalence of abnormal A/T/N biomarker groups increased with age in the Mayo Clinic Study on Aging. The mean annual memory z-score in this cohort at 60 years was 0.02, which dropped to 0.11 by age 90.
“Forty-six percent of this increase in decline rate [–0.06] was partitioned to the increasing prevalence of abnormal A/T/N profiles, while the remaining decline [–0.07] was partitioned to age,” the investigators reported.
While A+ subjects were most likely to decline, the A+/T–/N+ group presents a conundrum, the team wrote. “A possible explanation is that these individuals have early Alzheimer’s disease [denoted by A+T–] plus neurodegeneration due to comorbid non–Alzheimer’s disease neuropathic changes.”
This is an important point because the cognitive decline of Alzheimer’s is thought to be largely associated with tauopathy, not amyloidosis. “One possible explanation is an effect of subthreshold tau in A+/T–/N+ individuals, but this is speculative. Clearer understanding of the neuropathologic bases for the A+/T–/N+ group, as well as other A/T/N groups, awaits future biomarker-autopsy correlation studies.”
SOURCE: Jack CR et al. JAMA 2019;321:2316-25.