Multiple and flexible strategies targeting providers and participants at provider sites and within communities might be needed to enroll underrepresented populations into clinical trials.

Objective:To evaluate whether plasma biomarkers of amyloid (Aβ42/Aβ40), tau (p-tau181 and p-tau231) and neuroaxonal injury (neurofilament light chain [NfL]) detect brain amyloidosis consistently across racial groups.Methods:Individuals enrolled in studies of memory and aging who self-identified as African American (AA) were matched 1:1 to self-identified non-Hispanic White (NHW) individuals by age, APOE ε4 carrier status and cognitive status. Each participant underwent blood and cerebrospinal fluid (CSF) collection, and amyloid PET was performed in 103 participants (68%). Plasma Aβ42/Aβ40 was measured by a high-performance immunoprecipitation-mass spectrometry assay. Plasma p-tau181, p-tau231, and NfL were measured by Simoa immunoassays. CSF Aβ42/Aβ40 and amyloid PET status were used as primary and secondary reference standards of brain amyloidosis, respectively.Results:There were 76 matched pairs of AA and NHW participants (n=152 total). For both AA and NHW groups, the median age was 68.4 years, 42% were APOE ε4 carriers and 91% were cognitively normal. AA were less likely than NHW to have brain amyloidosis by CSF Aβ42/Aβ40 (22% versus 43% positive, p = 0.003). The Receiver Operating Characteristic Area Under the Curve (ROC AUC) of CSF Aβ42/Aβ40 status with the plasma biomarkers was as follows: Aβ42/Aβ40, 0.86 (95% confidence intervals [CI] 0.79-0.92); p-tau181, 0.76 (0.68-0.84); p-tau231, 0.69 (0.60-0.78); and NfL, 0.64 (0.55-0.73). In models predicting CSF Aβ42/Aβ40 status with plasma Aβ42/Aβ40 that included covariates (age, sex, APOE ε4 carrier status, race, and cognitive status), race did not affect the probability of CSF Aβ42/Aβ40 positivity. In similar models based on plasma p-tau181, p-tau231 or Nfl, AA had a lower probability of CSF Aβ42/Aβ40 positivity (Odds Ratio [OR] 0.31 [95% CI 0.13-0.73], OR 0.30 [0.13-0.71]) and OR 0.27 [0.12-0.64], respectively. Models of amyloid PET status yielded similar findings.Conclusions:Models predicting brain amyloidosis using a high performance plasma Aβ42/Aβ40 assay may provide an accurate and consistent measure of brain amyloidosis across AA and NHW groups, but models based on plasma p-tau181, p-tau231, and NfL may perform inconsistently and could result in disproportionate misdiagnosis of AA.