A Blood Test for Alzheimer’s Disease: It’s about Time or Not Ready for Prime Time?

In October 2020, C2N Diagnostics announced the availability of PrecivityAD™, a blood-based diagnostic test for Alzheimer’s disease (AD) [1]. This test uses immunoprecipitation and mass spectrometry to measure plasma levels of amyloid beta 42 and 40 (Aβ₄₂ and Aβ₄₀) and to detect apolipoprotein E (APOE) isoforms in blood (an equivalent to determining APOE genotype). Preanalytical handling of samples has been well standardized and the analytical pipeline is highly sensitive. The Aβ_42/40 ratio is combined with age and APOE genotype in a proprietary algorithm to yield an Amyloid Probability Score (APS). The APS ranges from 0 to 100. A high APS (58–100) signifies a high likelihood of a positive amyloid positron emission tomography (PET) scan result, and a low APS (0 to 35) is associated with a negative result. The algorithm was developed using data from 686 patients age 60 or older (mean age 73), all of whom had cognitive impairment felt to be potentially due to AD. Participants all underwent amyloid PET; 55% demonstrated a positive result. High and low APS cutpoints were established to optimize positive and negative predictive values. In a symptomatic population with 60% amyloid PET positivity, the positive and the negative predictive values of PrecivityAD™ are both expected to be 86%. An estimated 14% of blood tests in such a population will yield an APS in the indeterminate range (36–57).

PrecivityAD™ is the first but will not be the last blood test for AD. Blood tests bring new opportunities for patients, families, clinicians, and researchers compared to current gold standard imaging and cerebrospinal fluid (CSF) biomarkers. While amyloid PET imaging has a high concordance with postmortem brain amyloid pathology [2], it is expensive (thousands of US dollars) and not currently reimbursed by Medicare or other insurers. It also has some risk due to radiation exposure and is limited to centers with both PET scanner and tracer availability. Furthermore, the clinical read of an amyloid PET scan is qualitative and depends on the expertise of an interpreting Nuclear Medicine specialist or another certified clinician. CSF assays are variably covered by insurers and require a lumbar puncture, which may be contraindicated in some patients (e.g., those on anticoagulation) and technically challenging in others. In addition, the accuracy of CSF amyloid measurements is affected by sampling technique, particularly the type of tube used for CSF collection. Thus, blood tests have advantages in providing quantitative AD biomarker information at a reduced cost, with fewer procedural barriers and risks, and with widespread geographic availability. Such advantages could yield wider uptake of available tests, as well as increased opportunities for monitoring through serial assessments.

It is important to acknowledge some limitations of the new test(s). While the PrecivityAD™ test is less expensive than an amyloid PET scan, at $1250, the cost is not insignificant. And, as yet, PrecivityAD™ is not covered by insurance. An estimated 14% of symptomatic persons undergoing the PrecivityAD™ test will have an indeterminate result. This is comparable to the percentage of indeterminate results anticipated when using CSF profiles that include the Aβ_42/40 ratio [3]. However, the percentage of indeterminate results with amyloid PET is substantially lower. Only 11 of 13,444 (0.08%) amyloid PET scans in the Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) study were uninterpretable [4]. It is unknown how often the PrecivityAD™ APS will be indeterminate when the test is used in general practice, with more racially and ethnically diverse patients who may have greater comorbidities, including more mixed neurological presentations. A longitudinal study of people with indeterminate results is needed to know if repeat testing after a defined period of time is likely to be informative.

The data used to develop the APS algorithm are yet unpublished. The high and low APS cutpoints were set to optimize positive and negative predictive values and should be replicated in an independent sample. In addition, the positive and negative predictive values were optimized based on a symptomatic population with 60% amyloid PET positivity. If PrecivityAD™ is used in a symptomatic population with a lower AD prevalence (e.g., a community clinic that includes vascular or mixed dementia or a specialty clinic enriched in non-AD degenerative dementias), the positive predictive value may be lower than 86%. Even if the positive and negative predictive values are reliably 86%, then 14% of persons tested will receive a wrong prediction of amyloid PET status. In addition, a high APS is not likely to distinguish AD dementia from other disorders, e.g., Lewy body dementia, that are accompanied by amyloid pathology. Furthermore, amyloid positivity increases with age, and may reach 25% or higher among Caucasians (the most widely studied population) above the age of 70 [5]. The presence of a positive amyloid biomarker may be a true positive, in the sense that it indicates the presence of amyloid pathology in the brain, but amyloid may be only one of several factors contributing to cognitive impairment, particularly in people age 80 and older, where multimorbid brain pathology commonly exists [6] and clinical assessment of vascular brain changes is important.

Given imperfect accuracy, PrecivityAD™ should not be viewed as a standalone diagnostic. As with other AD biomarkers, an expert clinician will need to interpret the result in the context of the full clinical presentation. Testing should follow, not precede, a clinical history and cognitive assessment. If PrecivityAD™ or similar plasma biomarkers are used without prior clinical assessment and end up being performed in persons without cognitive impairment, the majority of “positive” results are likely to be false positives.

As has been done with other AD biomarkers used in clinical practice [7, 8], it will be important to develop Appropriate Use Criteria (AUC) for blood biomarkers. AUC can inform practice and decision making. Ideally, a multidisciplinary workgroup that includes expert clinicians, blood biomarker researchers, and neuroethicists will review the evidence on blood biomarker reliability, accuracy, and effect on patient outcomes, as well as the range of settings and populations for which evidence is available. Blood biomarker AUC should include guidance as to which patients are eligible and who can order a blood test. For example, PrecivityAD™ is intended for use in patients age 60 and older with mild cognitive impairment or dementia, in whom AD is a consideration. In this context, it is likely that PrecivityAD™ will have the greatest impact when an expert clinician has an intermediate suspicion of AD. As with CSF biomarkers, testing when the suspicion for AD is already extremely high or extremely low is unlikely to add prognostic value [9]. Therefore, a clinician with expertise in assessing individuals with cognitive impairment will best be able to place a test like this into context.

Pre-test counseling, including a discussion of test limitations, is essential so that an informed decision can be made about undergoing testing. For PrecivityAD™, this includes counseling about APOE genotyping. Pre-test counseling is particularly important if patients have direct access to laboratory results prior to formal disclosure. Post-test counseling is also critical so that patients understand their test results and implications, including any remaining uncertainty. In accordance with the recent European Academy of Neurology/European Alzheimer’s Disease Consortium position statement [10], pre- and post-biomarker counseling should be delivered by a physician who is a dementia specialist. Whether a primary care physician, generalist, or non-physician clinician with advanced training could serve in this role remains to be formally tested.

An acknowledged caveat is that there may not yet be enough data to evaluate this or other plasma tests to develop formal AUC. Data from diverse racial and ethnic populations are insufficient to develop AUC that are applicable to these communities. Data are increasing rapidly, however, suggesting that efforts to convene qualified working groups should begin soon. In the past year, remarkable data have emerged on several forms of plasma phosphorylated tau (P-tau). Assays for P-tau181, P-tau217, and P-tau231 appear to be both sensitive and specific for AD [11–13]. A Quanterix P-tau181 blood test has recently received the Food and Drug Administration Breakthrough Device designation. A P-tau blood test might be interpreted more easily than the Aβ_42/40 ratio, based on published data so far. There is no need to include age and APOE genotype into a predictive tool, for example. By virtue of not requiring genetic testing/disclosure, P-tau tests may also have greater potential for use in a primary care setting. The questions of whether there is a grey zone for P-tau markers and which cutoffs will work at scale and in diverse populations need further study. Published data from several cohorts looks highly encouraging, but community-based studies, especially with older participants, are needed. For example, investigation of plasma biomarkers, measured using the Quanterix SiMoA platform, found that age, impaired renal function, and comorbidity index were associated with levels of plasma amyloid and tau biomarkers [14]. These variables are important and will need further investigation as wider studies of plasma biomarkers are undertaken in more representative clinical populations.

Clinical context will remain important. For example, if a normal level of plasma P-tau is used to reassure someone with cognitive impairment that they do not have AD, then other causes of cognitive problems such as depression, medications, and general medical disorders, still need to be considered and may be potentially treatable, not to mention consideration of other brain disorders such as cerebrovascular disease or parkinsonian disorders. Therefore, while a blood test measuring tau could play an important part of a thoughtful workup, great care will need to be taken to ensure that it is not used as a short cut.

Neurofilament light and other markers of neurodegeneration, while less specific, may provide additional information. Depending on the clinical question, a panel of biomarkers may be most useful. Plasma amyloid, tau, and neurodegeneration biomarkers are being combined in models to provide individualized prognosis for persons with mild cognitive impairment [15]. Given the ease of obtaining blood, it is likely that a panel of biomarkers will be developed with strong clinical potential soon. Determining the sensitivity of such a panel to preanalytical factors, such as variations in blood sampling and processing across multiple collection sites, will be critical prior to widespread adoption. For example, West et al. [16] recently demonstrated a robust relationship between plasma Aβ_42/40 (measured using the immunoprecipitation/mass spectrometry method) and brain amyloid positivity (assessed via amyloid PET or CSF biomarkers) across 6 research cohorts without preanalytical standardization. However, the optimal plasma Aβ_42/40 cutoff value differed among cohorts. The field must guide the use of these promising assays and be prepared to adjust guidance as advances in testing, as well as treatment, require and permit.