The performance of plasma p‐tau217 in Black middle‐aged and older adults

Abstract

INTRODUCTION

Medical conditions prevalent in Black adults within the United States have been associated with plasma tau phosphorylated at threonine 217 (p‐tau217); however, insufficient p‐tau217 research has been conducted with Black adults.

METHODS

Participants included n = 233 predominantly cognitively unimpaired adults enrolled in the African Americans Fighting Alzheimer's in Midlife study. Subsamples had creatinine (n = 137) and positron emission tomography (PET; amyloid‐PET = 65 [amyloid‐PET‐positive = 16/65]; tau‐PET = 70). We tested whether p‐tau217 (ALZPath, Inc.) varied by medical condition and amyloid‐ and tau‐PET‐positivity status and assessed the diagnostic accuracy of p‐tau217.

RESULTS

Estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m², cardiovascular disease (CVD), and amyloid‐ and tau‐PET‐positive status demonstrated higher p‐tau217. Effect sizes (r _pb): eGFR <60 group = 0.48, CVD = 0.25, amyloid‐PET‐positive status = 0.54; tau‐PET‐positive status = 0.56. Lower eGFR was related to higher p‐tau217 when adjusting for amyloid‐PET. For abnormal amyloid‐PET and tau‐PET, p‐tau217 exhibited areas under the curve of 0.90 and 0.89, respectively.

DISCUSSION

Plasma p‐tau217 showed promise as an Alzheimer's biomarker in Black adults; however, kidney function and CVD should be considered when interpreting levels.

Highlights

• Plasma tau phosphorylated at threonine 217 (p‐tau217) was tested in a sample of Black middle‐aged and older adults.

• Level of p‐tau217 was higher in impaired kidney function and cardiovascular disease.

• Obesity and diabetes were not related to p‐tau217.

• Level of p‐tau217 was higher in amyloid‐ and tau‐PET‐positive status.

• Plasma p‐tau217 showed good receiver‐operating characteristic area under the curve for abnormal amyloid‐ and tau‐PET.

1. BACKGROUND

In the United States, the prevalence rate of clinical Alzheimer's disease (AD) is higher in non‐Hispanic Black adults relative to non‐Hispanic White adults. ¹ Research and clinical trials needed to meaningfully address this disparity will depend upon the accurate performance and interpretation of blood‐based AD biomarkers. Currently, one of the most promising blood‐based biomarkers of AD pathology is tau phosphorylated at threonine 217 (p‐tau217). Increased plasma levels have been strongly related to elevated cerebrospinal fluid (CSF) and positron emission tomography (PET) biomarkers of amyloid and tau, the pathological hallmarks of AD, in both preclinical and clinical AD. ² , ³ , ⁴ , ⁵ , ⁶ Compared to other blood‐based biomarkers (e.g., p‐tau181), plasma p‐tau217 has demonstrated greater accuracy in discriminating amyloid‐ and tau‐PET‐positive from ‐negative status, showing equivalence to CSF p‐tau217 in detecting abnormal amyloid and tau‐PET. ⁵ , ⁷

However, most plasma p‐tau217 research has been conducted using samples comprised predominantly of highly educated non‐Hispanic White adults, who are physically healthy and generally free of medical conditions and diseases associated with aging. ⁸ Of the few studies investigating the performance of p‐tau217 within the context of medical conditions, chronic kidney disease (CKD), cardiovascular disease (CVD), and obesity were found to be associated with plasma levels in most ⁹ , ¹⁰ , ¹¹ , ¹² but not all studies. ¹³ , ¹⁴ These are conditions and diseases that are more prevalent in non‐Hispanic Black relative to non‐Hispanic White adults in the United States ¹⁵ , ¹⁶ , ¹⁷ ; however, to the best of our knowledge, few studies have investigated the association between these medical conditions and plasma p‐tau217 in a sample largely comprised of non‐Hispanic Black adults. We are aware of only one study ⁹ that investigated the relationship between a health condition (i.e., obesity) and plasma p‐tau217 in a sample that contained a moderate proportion (33%) of non‐Hispanic Black adults.

In studies reporting significant relationships between medical conditions and plasma p‐tau217, CKD ¹⁰ , ¹¹ , ¹² and CVD ¹² (i.e., history of myocardial infarction [MI] or stroke) have been associated with higher p‐tau217 levels. A recent study of a large Korean cohort ¹¹ found that CKD, but not coronary artery disease (a type of CVD), was associated with higher p‐tau217 levels. In these studies, associations persisted when adjusting for age and sex, and importantly, cortical amyloid assessed by PET, suggesting that the elevation in p‐tau217 may possibly occur apart from cortical amyloid pathology. Some studies ¹⁸ , ¹⁹ , ²⁰ suggest that neither vascular disease nor elevated vascular risk is associated with a greater burden of AD pathology; however, other studies have found significant associations. ²¹ , ²² , ²³ If CKD and CVD increase plasma p‐tau217 apart from amyloid pathology, individuals with these conditions who lack elevated cortical amyloid may be incorrectly identified as being amyloid positive. Higher body mass index (BMI) or obesity, in contrast to CKD and CVD, has been related to lower p‐tau217 levels. ⁹ , ¹¹ , ¹²

We conducted a cross‐sectional study to investigate the performance and accuracy of plasma p‐tau217, currently one of the most promising blood‐based biomarkers, within a sample of Black middle‐aged to older adults who were predominantly cognitively unimpaired. We specifically investigated whether individuals with impaired kidney function and CVD had higher plasma p‐tau217 levels and whether individuals characterized as obese had lower p‐tau217. We also examined whether plasma p‐tau217 levels were higher in individuals with hypertension and diabetes, well‐recognized risk factors for CVD. Because one of our goals was to identify medical conditions whose relationship with plasma p‐tau217 may not be attributable to cortical amyloid, we investigated if associations between a medical condition and p‐tau217 were significant when adjusting for amyloid PET in addition to age and sex. In a smaller subset of participants who completed amyloid and/or tau PET neuroimaging, we examined if p‐tau217 levels were higher in amyloid‐PET‐positive (n = 16/65; global ¹¹C‐Pittsburgh Compound B [PiB] index distribution volume ratio [DVR] ≥1.16) and tau‐PET‐positive (n = 12/70; MK‐6240 standardized uptake value ratio [SUVR] [temporal lobe meta‐region of interest] >1.30) participants and explored the diagnostic accuracy of plasma p‐tau217 to detect amyloid and tau PET‐positive status.

2. METHODS

2.1. Participants

Participants were enrolled in the African Americans Fighting Alzheimer's in Midlife (AA‐FAIM) study. Participants in AA‐FAIM are co‐enrolled in one of two research cohorts focused primarily on the study of AD in the preclinical phase: the Wisconsin Alzheimer's Disease Research Center (WI‐ADRC) Clinical Core or the Wisconsin Registry for Alzheimer's Prevention (WRAP). AA‐FAIM supports the recruitment and retention of Black middle‐aged and older adults into both cohorts. ²⁴ The Wisconsin Alzheimer's Institute (WAI) Regional Milwaukee Office also supports participant recruitment and retention into WRAP. Recruitment and retention approaches are culturally sensitive and incorporate a variety of programs that are designed specifically to address the needs of middle‐aged and older adults within the Black communities of Madison and Milwaukee.

Participants are included in AA‐FAIM if they self‐identify as Black and/or African American and do not have dementia (Clinical Dementia Rating [CDR] <1) at study entry. Participants are excluded if they have a significant neurologic disorder (e.g., major ischemic or hemorrhagic stroke, poorly controlled epilepsy, or major traumatic brain injury within the year prior to enrollment) or have current/uncontrolled major psychiatric disorder, or a significant medical illness affecting cognition or ability to participate.

AA‐FAIM participants were included in the current study if they had at least one plasma p‐tau217 value and at least one assessment for a medical condition of interest (i.e., impaired kidney function, CVD, obesity, hypertension, or diabetes). Table 1 describes demographic and sample characteristics. All available data were used; sample sizes varied depending on the medical condition tested (see Table 1). A smaller subsample of participants (n = 137) had blood previously assayed for creatinine, used to assess kidney function. Additional smaller subsamples had amyloid‐PET (n = 65) and tau‐PET (n = 70); n = 57 had both amyloid‐ and tau‐PET imaging.

TABLE 1.

Demographic and health characteristic data presented as means (SD) or counts (%).

Participant characteristics	Main n = 233	eGFR subsample n = 137	Amyloid‐PET subsample n = 65	Tau‐PET subsample n = 70
Age, years	64.5 (9.3)	66.7 (9.2)	67.2 (7.6)	66.2 (8.7)
	n = 231
Sex, no. (%) self‐identifying as male	71 (30.7)	44 (32.1)	20 (30.8)	21 (30.0)
	n = 222	n = 136	n = 64	n = 68
Education, years	14.9 (2.5)	14.8 (2.3)	15.4 (2.5)	15.3 (2.4)
Black/African American, no. (%)
Primary race	225 (96.6)	135 (98.5)	61 (93.8)	65 (92.9)
Secondary or tertiary race	8 (3.4)	2 (1.5)	4 (6.2)	5 (7.1)
Cognitive status, no. (%)
Cognitively unimpaired	192 (82.4)	106 (77.4)	56 (86.1)	57 (81.4)
Impaired‐other	17 (7.3)	12 (8.8)	2 (3.1)	6 (8.6)
Mild cognitive impairment	21 (9.0)	16 (11.7)	5 (7.7)	6 (8.6)
Dementia	3 (1.3)	3 (2.2)	2 (3.1)	1 (1.4)
Plasma p‐tau217, mean, pg/mL	0.36 (0.30)	.40 (0.33)	0.38 (0.30)	0.41 (0.36)
Plasma p‐tau217, median, pg/mL	0.25 (IQR: 0.20 to 0.37)	0.27 (IQR: 0.22 to 0.41)	0.30 (IQR: 0.22 to 0.39)	0.29 (IQR: 0.21 to 0.44)
	n = 222	n = 115	n = 64	n = 67
CVD, no. (%), yes	23 (10.4)	8 (7.0)	4 (6.3)	5 (7.5)
	n = 232
BMI	33.3 (7.9)	32.9 (8.0)	31.9 (7.5)	32.1 (8.4)
Obesity category, a no. (%)
Normal (BMI 18.5–24.9)	30 (12.9)	19 (13.9)	11 (16.9)	14 (20.0)
Overweight (BMI 25–29.9)	61 (26.3)	39 (28.5)	19 (29.2)	18 (25.7)
Obesity class I (BMI 30–34.9)	57 (24.6)	33 (24.1)	16 (24.6)	15 (21.4)
Obesity class II (BMI 35–39.9)	45 (19.4)	26 (19.0)	10 (15.4)	13 (18.6)
Obesity class III (BMI ≥40)	39 (16.8)	20 (14.6)	9 (13.8)	10 (14.3)
	n = 220	n = 124	n = 64	n = 62
Diabetes, no. (%), yes	75 (34.1)	45 (36.3)	26 (40.6)	24 (38.7)
	n = 231
SBP, mm Hg	129.6 (19.8)	130.9 (19.0)	128.9 (19.0)	129.0 (18.9)
	n = 231
DBP, mm Hg	78.8 (9.9)	78.5 (8.7)	77.3 (7.8)	77.6 (8.6)
Blood pressure category, b no. (%)
Normal	69 (29.9)	39 (28.5)	21 (32.3)	22 (31.4)
Elevated	24 (10.4)	13 (9.5)	6 (9.2)	7 (10.0)
Stage 1 hypertension	112 (48.5)	63 (46.0)	31 (47.7)	33 (47.1)
Stage 2 hypertension	26 (11.2)	22 (16.1)	7 (10.8)	8 (11.4)
Unique subsample characteristics start here:
			n = 38	n = 43
Creatinine, mg/dL		0.96 (0.37)	1.01 (0.50)	0.99 (0.48)
			n = 38	n = 43
eGFR, mL/min/1.73 m2		80.5 (19.3)	77.1 (19.6)	78.2 (19.3)
Kidney function category, c no. (%)
Normal, eGFR ≥90		51 (37.2)	10 (26.3)	12 (27.9)
Mildly decreased, 60 ≤ eGFR < 90		68 (49.6)	24 (63.2)	27 (62.8)
Mild to severely decreased, eGFR <60		18 (13.1)	4 (10.5)	4 (9.3)
		n = 38		n = 57
Amyloid‐PET DVR		1.12 (0.19)	1.13 (0.19)	1.13 (0.20)
Amyloid‐PET positive, d no. (%)		9 (23.7)	16 (24.6)	15 (26.3)
		n = 43	n = 57
tau‐PET SUVR		1.15 (0.24)	1.12 (0.20)	1.13 (0.23)
tau‐PET positive, e no. (%)		7 (16.3)	9 (15.8)	12 (17.1)

Abbreviations: BMI, body mass index; CVD, cardiovascular disease; DBP, diastolic blood pressure; DVR, distribution volume ratio; eGFR, estimated glomerular filtration rate; IQR, interquartile range; PET, positron emission tomography; SUVR, standardized uptake value ratio; SBP, systolic blood pressure; SD, standard deviation.

^a CDC 2024.

^b Whelton et al., 2017: Normal = systolic blood pressure (SBP) <120 mm Hg and diastolic blood pressure (DBP) <80 mm Hg, Elevated = SBP 120–129 mm Hg and DBP <80 mm Hg; Stage 1 hypertension = SBP 130–139 mm Hg or DBP 80–89 mm Hg; Stage 2 hypertension SBP  ≥140 mm Hg or DBP ≥90 mm Hg.

^c Levey et al., 2015.

^d Amyloid‐PET positive = global Pittsburgh Compound B (PiB) index distribution volume ratio (DVR) ≥1.16 (Cody et al., 2024).

^e Tau‐PET positive = MK‐6240 standardized uptake value ratio (SUVR) >1.30 bilateral temporal lobe meta‐region of interest (Cody et al., 2024).

RESEARCH IN CONTEXT

1. Systematic review: Medical conditions more prevalent in non‐Hispanic Black relative to White adults in the United States have been associated with tau phosphorylated at threonine 217 (p‐tau217); however, most research has been conducted using samples of predominantly White adults. Specifically, chronic kidney disease (CKD) and cardiovascular disease (CVD) were associated with higher p‐tau217 levels, and obesity was associated with lower levels. In a sample of Black adults, we examined if medical conditions and amyloid‐ and tau–positron emission tomography (PET)–positive status were associated with p‐tau217 and assessed the accuracy of p‐tau217.

2. Interpretation: Impaired kidney function, CVD, and amyloid‐ and tau‐PET‐positive status were associated with higher p‐tau217. Plasma p‐tau217 demonstrated promise as an accurate Alzheimer's disease biomarker in Black adults.

3. Future directions: Whether the association between medical conditions and p‐tau217 is amyloid independent requires additional examination. Strategies for isolating false‐positive cases for follow‐up need to be investigated in larger samples of adults from diverse racial and ethnic groups.

2.2. General procedures

AA‐FAIM participants have regular research visits, either annually if enrolled in the WI‐ADRC Clinical Core or biennially if enrolled in WRAP. Full examinations include cognitive functioning assessments, blood collection, and height, weight, and blood pressure measurements. A clinician and/or clinical coordinator collects a health history and obtains information on the presence of medical conditions and diseases. Study partners are asked to corroborate medical history for those participants who are cognitively impaired. Clinicians also perform a brief medical exam and review previous medical history and laboratory tests to further characterize medical diagnoses. A multidisciplinary consensus review team evaluates relevant medical and cognitive data to determine a diagnostic status of cognitively unimpaired, mild cognitive impairment, or dementia based on National Institute on Aging–Alzheimer's Association (NIA‐AA) criteria. ²⁵ , ²⁶ , ²⁷ , ²⁸ A subsample of WI‐ADRC Clinical Core and WRAP participants have amyloid‐ and tau‐PET at least once, either during a regular research visit or as part of a PET sub‐study. PET imaging is collected using a common acquisition protocol. ²⁹ , ³⁰

All study procedures were approved by the University of Wisconsin–Madison Institutional Review Board. Participants provided written informed consent consistent with ethics standards.

2.3. Medical conditions

2.3.1. Impaired kidney function

Estimated glomerular filtration rate (eGFR) and categories derived from eGFR values ³¹ were tested as measures of kidney function. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI) creatinine refit, ³² an equation that excludes race and is recommended by the National Kidney Foundation and the American Society of Nephrology. The equation incorporates age and sex to account for the influence of these factors on creatinine levels. ³² Creatinine entered into the eGFR calculation was assayed using the enzymatic method that is based on an IDMS (isotope dilution mass spectrometry) traceable calibration. eGFR values represent the milliliters of blood filtered by the kidney per minute per body surface area (mL/min/1.73 m²). Kidney function categories derived from eGFR values ³¹ were as follows: normal (eGFR ≥90), mildly decreased (60 ≤ eGFR < 90), and mild to severely decreased kidney function (eGFR <60). We were unable to ascertain chronic kidney disease, which is defined as eGFR <60 for >3 months, ³¹ because some of our participants (n = 114) had only one creatinine value.

2.3.2. Cardiovascular disease

History of CVD (yes vs no) was defined as clinician or self‐report of MI, stroke, and/or congestive heart failure (CHF). Numbers of participants with either MI, stroke, and/or CHF were as follows: MI alone = 10, stroke alone = 3, CHF alone = 4, MI and stroke = 1, MI and CHF = 2, stroke and CHF = 2, MI, stroke, and CHF = 1.

2.3.3. Obesity

BMI and BMI‐categories were used to assess obesity. BMI categories ³³ were as follows: normal weight (BMI = 18.5–24.9), overweight (BMI = 25–29.9), obesity class I (BMI = 30–34.9), obesity class II (BMI = 35–39.9), and obesity class III (BMI ≥40).

2.3.4. Diabetes

History of type 2 or type 1 diabetes (yes vs no) was determined by clinician review of medical history and medications, or self‐report of either type 1 or type 2 diabetes.

2.3.5. Hypertension

Systolic blood pressure (SBP) and categories of blood pressure (BP) were tested as measures of hypertension. Four BP categories ³⁴ were derived based on SBP and diastolic blood pressure (DBP): normal (SBP < 120 mm Hg and DBP < 80 mm Hg), elevated (SBP = 120–129 mm Hg and DBP <80 mm Hg), Stage 1 hypertension (SBP = 130–139 mm Hg or DBP = 80–89 mm Hg), and Stage 2 hypertension (SBP ≥140 mm Hg or DBP ≥90 mm Hg).

2.4. Plasma p‐tau 217

Ethylenediaminetetraacetic acid (EDTA) plasma was analyzed for p‐tau217 in the Alzheimer's Disease Research Center (ADRC) Fluid Biomarker Lab at the University of Wisconsin–Madison using the single molecule array (SIMOA) immunoassay developed by ALZPath, Inc. (Carlsbad, CA). The ALZPath assay has a dynamic range of 0.0070–30.0 pg/mL (with 3X dilution). EDTA plasma was assayed in duplicate on an HD‐X Analyzer and the average of the two p‐tau217 values was utilized in statistical analyses. The coefficient of variation (CV) of duplicates ranged from 0.00%–14.0%. Manufacturer controls were analyzed in duplicate with every batch to assess batch‐to‐batch reproducibility. The within‐plate CV ranged from 0.1% to 18%, and the between‐plate CV was 12%. In cohorts comprising predominantly White adults, ALZPath p‐tau217 was shown to discriminate amyloid and tau biomarker positive from negative status, exhibiting areas under the curve (AUCs) >0.90. ²

Except for creatinine used to calculate eGFR, most (91.3%–92.7%) medical condition data were collected at the same visit or within 1 month of blood collection for p‐tau217; 97.8%–99.5% of these data were collected within 1 year prior to blood collection for p‐tau217. Blood tested for creatinine was collected within a mean of 3.5 years (SD = 3.3) prior to blood collection for p‐tau217 (creatinine age minus p‐tau217 age: median = –4.0; interquartile range [IQR] = −6.1 to 0; full range = −10.0 to 0). Because eGFR values have been shown to decline on average 0.4 (± 3.6) mL/min/1.73 m² per year in older adults, ³⁵ a sensitivity analysis (see Section 2.6.1.1) was conducted to test eGFR calculated from creatinine collected within 1 month of p‐tau217.

2.5. Amyloid‐ and tau‐PET imaging

Cortical amyloid beta (Aβ) and aggregated tau were quantified, respectively, using PiB and ¹⁸F‐MK‐6240 PET imaging. Information on radioligand synthesis, image acquisition, processing, and analysis of MRI and PiB, and MK‐6240 PET images has been previously published. ²⁹ , ³⁰ Cortical PiB DVR was averaged across eight bilateral regions using Logan graphical analysis and cerebellum gray matter as the reference region. ³⁶ A PiB DVR ≥1.16 (equivalent to 17.1 Centiloids) was used as the cut‐point for amyloid‐PET positivity. ³⁷ Employing the inferior cerebellum as a reference region, ²⁹ tau burden was determined using the average MK‐6240 SUVR (70−90 min) from a temporal lobe meta‐region of interest (meta‐ROI) (bilateral parahippocampal gyrus, fusiform gyrus, inferior and middle temporal gyrus, and amygdala). ³⁸ MK‐6240 SUVR >1.30 was used as the cut‐point for tau‐PET positivity. ³⁷

Amyloid‐PET was collected on average 0.75 (SD = 2.07) years after blood collection for p‐tau217 (amyloid PET age minus p‐tau217 age: median = 0.4; IQR = 0 to 1.37; full range = −4.24 to 5.8). Tau‐PET was collected on average 0.66 (SD = 2.05) years following blood collection for p‐tau217 (tau‐PET age minus p‐tau217 age: median = 0.34; IQR = 0 to 1.20; full range = −4.5 to 6.33).

2.6. Statistical approach

2.6.1. Relationship of plasma p‐tau217 to medical conditions and AD PET biomarker status

Kruskal–Wallis and Spearman's correlation were used to test unadjusted associations between categorical and continuous variables, respectively, and plasma p‐tau217; associations adjusted for age and self‐identified sex were tested using linear regression.

For example, to examine if impaired kidney function was associated with plasma p‐tau217, we obtained a Spearman rank correlation of the continuous p‐tau217 and eGFR values; we performed a Kruskal–Wallis test examining p‐tau217 levels across the three kidney function categories identified in Section 2.3.1, conducting the Dunn's Test on pairwise contrasts if the Kruskal–Wallis chi‐square test was significant; and last, in regression analyses using the same dataset, we examined parallel regression models that had log(p‐tau217) as the outcome, age at plasma p‐tau217 collection, and self‐identified sex as covariates, with continuous eGFR in one model and the three‐level kidney function category (normal = reference) in the other. This analytical approach was followed for all medical conditions defined by continuous and categorical variables. For diabetes, CVD, and AD PET biomarker status, which were all defined by two‐level categorical variables, Kruskal–Wallis tested unadjusted relationships with p‐tau217, and linear regression tested associations adjusted for age and self‐identified sex. Each individual medical condition and AD PET biomarker status variable were tested in a separate model.

In another set of analyses that tested the relationship between a health variable and p‐tau217, amyloid‐PET DVR was adjusted in addition to age and self‐identified sex. Models were adjusted for amyloid PET DVR to explore whether cortical amyloid contributed to the association between the health variable and p‐tau217. If the medical condition was related to cortical amyloid, we anticipated that the strength of the association between the medical condition and p‐tau217, a biomarker for amyloid, would be reduced after adjusting for amyloid PET DVR.

Because the subsample that contained both medical condition data and amyloid PET was smaller than the sample used to test the age‐ and sex‐adjusted association between a medical condition and p‐tau217, we tested the relationship between the medical condition and p‐tau217 using data from the smaller amyloid PET subsample, adjusting for age and sex first, prior to proceeding with adjusting for amyloid PET, to ensure that the results were consistent between the larger and smaller samples (see Table S1). To reduce model overfitting, only continuous and dichotomous (yes vs no) health variables were tested.

If a medical condition was significantly related to plasma p‐tau217, we tested that medical condition's association with amyloid‐PET DVR and tau‐PET SUVR in regression models adjusted for age and self‐identified sex.

To examine if plasma p‐tau217 was higher in amyloid‐ and tau‐PET‐positive participants, three amyloid‐ and tau‐PET status variables were tested separately in unadjusted and age‐ and sex‐adjusted analyses: (1) amyloid (A) status (A+ vs A−); (2) tau (T) status (T+ vs T−), and (3) AT statuses (A−T− vs A+T−; A−T− vs A+T+; A+T− vs A+T+). The A−T+ group was not included in the third variable because of the small number (n = 2) in that group.

The magnitude of significant between‐group differences in plasma p‐tau217 levels was estimated using the point biserial correlation coefficient. Because of the skewed distribution of p‐tau217, it was log‐transformed prior to these analyses.

Linear regression diagnostics were conducted to ensure that residuals were homoscedastic and normally distributed. Cook's distance was used to detect influential data points. To meet the homoscedasticity assumption, p‐tau217, amyloid‐PET DVR, and tau‐PET SUVR were log‐transformed when tested as an outcome variable.

Sensitivity analyses

Because blood tested for creatinine had been collected on average 3.5 years (SD = 3.3) prior to blood tested for p‐tau217 (see Section 2.4), we conducted a sensitivity analysis to investigate the association between impaired kidney function and p‐tau217 in the participant subset (n = 59) who had blood collected for creatinine within 1 month of p‐tau217. The three‐level kidney function category and continuous eGFR variables were tested as predictors of p‐tau217 in two separate linear regression models that adjusted for age and self‐identified sex. We planned additional sensitivity tests due to the comorbidity of medical conditions. Medical conditions that were found to be significantly associated with p‐tau217 were tested together in another regression model where p‐tau217 was the outcome, and age and sex were controlled.

Model structures and sample sizes for all analyses testing the relationship of plasma p‐tau217 to medical conditions and AD PET biomarker status can be found in Table S2.

2.6.2. Accuracy of p‐tau217 to discriminate amyloid‐ and tau‐PET‐positive from negative status

Receiver‐operating characteristic (ROC) area under the curve (AUC) was used to determine the ability of plasma p‐tau217 to detect amyloid‐ and tau‐PET‐positive status; confidence intervals (CIs) were calculated employing Delong's method. Youden's index was used to derive the binary cut‐point for amyloid‐PET and tau‐PET positivity. We explored the effectiveness of a two‐cut‐point strategy to isolate false‐positive and false‐negative amyloid status cases for follow‐up. A lower cut‐point with a low false‐negative rate (≈95% sensitivity) and a higher cut‐point with a low false‐positive rate (≈95% specificity) were selected; values between the two cut‐points represented an intermediate group that would be considered for follow‐up. Positive and negative percent agreement was calculated by ignoring the intermediate values. We focused mostly on the accuracy of p‐tau217 to detect amyloid‐PET‐positive status because currently available AD therapy targets amyloid (i.e., anti‐amyloid immunotherapy), and blood‐based biomarker screening tests for amyloid may soon be implemented in the clinic once fully approved by the U.S. Food and Drug Administration (FDA).

All analyses were conducted using R 4.4.0. ³⁹ Table 1 was constructed using the tableone package ⁴⁰ ; figures were developed using the ggplot2 package. ⁴¹ Two‐tailed test statistics were performed with alpha set to 0.05. Results were not corrected for multiple comparisons.

3. RESULTS

3.1. Demographic and health characteristics of participants

Participants were middle‐aged to older adults (mean age = 64.5 years, SD = 9.3; n = 233) (see Table 1). The average age of participants in the eGFR (n = 137) and amyloid‐ (n = 65), and tau‐ (n = 70) PET subsamples was 66.7 (SD = 9.2), 67.2 (SD = 7.6), and 66.2 (SD = 8.7) years, respectively. Most participants (77.4%–86.1%) were cognitively unimpaired. Eighteen (13.1%) of 137 of the participants had eGFR <60 mL/min/1.73 m²; 23 of 222 (10.4%) had a history of CVD, 39 of 232 (16.8%) had a BMI ≥40 (i.e., obesity class III), 75 of 220 (34.1%) had a history of diabetes, and 138 of 231 (59.7%) had SBP ≥ 130 mm Hg or DBP ≥80 mm Hg (i.e., Stage 1 or 2 hypertension). Within the amyloid‐ and tau‐PET subsample, respectively, 16 of 65 (24.6%) of the participants were amyloid‐PET‐positive and 12 of 70 (17.1%) were tau‐PET‐positive.

3.2. Relationship of plasma p‐tau217 to medical conditions and AD PET biomarker status

3.2.1. Impaired kidney function

Lower (worse) eGFR was significantly related to higher (worse) plasma p‐tau217 in unadjusted (r _s[135] = –0.29, p < 0.001; see Figure 1A) and age‐ and sex‐adjusted analyses (see Table 2). The three‐level kidney function category was also significantly related to p‐tau217 (χ²[2] = 18.7, p < 0.001; see Figure 2A); post hoc testing revealed that the mildly decreased (z = 3.4, p < 0.001; r _pb = 0.28 [95% CI: 0.11 to 0.44]) and the mild to severely decreased kidney function group (z = 3.8, p < 0.001; r _pb = 0.48 [95% CI: 0.28 to 0.65]) had higher p‐tau217 than the group with normal kidney function (see Figure 2A).

FIGURE 1.

Scatterplots illustrating associations of estimated glomerular filtration rate, body mass index, systolic blood pressure, and age with plasma p‐tau217. ¹Association was not significant when adjusting for age and sex.

TABLE 2.

Results from regression analyses testing associations of medical conditions and Alzheimer's disease (AD) PET positivity to log‐transformed plasma p‐tau217.

	Medical condition		Age a , years		Self‐identified sex (male)
	b (SE)	P	b (SE)	p	b (SE)	p
Models
1. eGFR, mL/min/1.73 m2	−0.01 (0.002)	<0.001	0.03 (0.005)	<0.001	0.04 (0.09)	0.66
2. Kidney function, b reference = normal
Mildly decreased	0.21 (0.09)	0.03	0.03 (0.005)	<0.001	0.04 (0.09)	0.70
Mild to severely decreased	0.57 (0.14)	<0.001
3. Cardiovascular disease, c yes	0.34 (0.11)	0.003	0.03 (0.004)	<0.001	0.05 (0.07)	0.49
4. BMI	0.003 (0.004)	0.44	0.03 (0.004)	<0.001	0.11 (0.08)	0.14
5. Obesity category, d reference = normal
Overweight	−0.003 (0.11)	0.98	0.03 (0.004)	<0.002	0.12 (0.07)	0.10
Obesity class I	0.05 (0.11)	0.69
Obesity class II	0.03 (0.12)	0.77
Obesity class III	0.13 (0.12)	0.31
6. Diabetes, yes	0.03 (0.07)	0.72	0.03 (0.004)	<0.001	0.08 (0.08)	0.27
7. SBP, mm Hg	0.002 (0.002)	0.22	0.03 (0.004)	<0.001	0.09 (0.07)	0.22
8. BP category, e reference = normal
Elevated	0.02 (0.12)	0.87	0.03 (0.004)	<0.001	0.10 (0.07)	0.19
Stage 1 hypertension	−0.03 (0.08)	0.67
Stage 2 hypertension	0.19 (0.12)	0.11
Kidney function sensitivity analysis f
9. eGFR, mL/min/1.73 m2	−0.009 (0.003)	0.002	0.02 (0.009)	0.04	0.06 (0.11)	0.58
10. Kidney function, reference = normal
Mildly decreased (n = 29)	0.20 (0.12)	0.10	0.02 (0.008)	0.007	0.05 (0.12)	0.65
Mild to severely decreased (n = 10)	0.40 (0.16)	0.02
AD PET analyses	AD PET positivity
11. Amyloid (A) PET status (positive) g	0.52 (0.12)	<0.001	0.02 (0.007)	0.002	0.05 (0.11)	0.65
12. Tau (T) PET status (positive) h	0.77 (0.17)	<0.001	0.02 (0.007)	0.005	0.11 (0.13)	0.42
13. AT PET status (reference = A−T−)
A+T−	0.44 (0.15)	0.005	0.02 (0.007)	0.04	0.06 (0.11)	0.57
A+T+	0.77 (0.15)	<0.001

Abbreviations: AD, Alzheimer's disease; BMI, body mass index; BP, blood pressure; eGFR, estimated glomerular filtration rate; PET, positron emission tomography; SE, standard error; SBP, systolic blood pressure.

^a Age at blood collection for p‐tau217.

^b Levey et al., 2015.

^c History of MI, stroke, and/or congestive heart failure.

^d CDC 2024.

^e Whelton et al., 2017.

^f Results for the subset who had creatinine collected on the same day as p‐tau217 (n = 59).

^g Amyloid‐PET‐positive status = Global PiB index DVR ≥1.16 (Cody et al., 2024).

^h Tau‐PET‐positive status = MK‐6240 SUVR >1.30 in bilateral temporal lobe meta‐region of interest (Cody et al., 2024).

FIGURE 2.

Box plots illustrate unadjusted associations between categorical measures of medical conditions and plasma p‐tau217. ¹ P‐values for post hoc tests performed following a significant (p < 0.05) omnibus test. ²Associations were not significant when adjusting for age and self‐identified sex. BMI, body mass index; HTN, hypertension.

In the age‐ and sex‐adjusted regression analysis, the two worse kidney function categories had significantly higher p‐tau217 levels than the group with normal eGFR values (see Table 2). The group with mild to severely decreased kidney function also had significantly higher p‐tau217 (b = 0.36, SE = 0.13, p = 0.006) relative to the group with mildly decreased kidney function.

In the subset with amyloid‐PET, lower eGFR was again significantly related to higher p‐tau217 when adjusting for amyloid‐PET DVR, age, and sex (n = 38 had both eGFR and amyloid PET data) (see Table 3). In contrast, eGFR was not significantly related to amyloid‐PET DVR or tau‐PET SUVR (n = 43 had both eGFR and tau‐PET data) (see Table 3). Both models were adjusted for age and sex.

TABLE 3.

Associations between estimated glomerular filtration rate (eGFR) and AD biomarkers.

	b (SE)	p
Dependent variable = log(p‐tau217) n = 38
Age, years	0.01 (0.01)	0.18
Self‐identified sex, male	0.11 (0.16)	0.48
Amyloid‐PET, standard score a	0.18 (0.08)	0.02
eGFR, standard score a	−0.22 (0.07)	0.004
Dependent variable = log(amyloid PET DVR) n = 38
Age, years	0.007 (0.003)	0.04
Self‐identified sex, male	−0.009 (0.05)	0.86
eGFR, mL/min/1.73 m2	−0.0003 (0.001)	0.79
Dependent variable = log(tau PET SUVR) n = 43
Age, years	0.002 (0.004)	0.63
Self‐identified sex, male	−0.09 (0.07)	0.19
eGFR, mL/min/1.73 m2	−0.001 (0.002)	0.48

Abbreviations: eGFR, estimated glomerular filtration rate; DVR, distribution volume ratio; PET, positron emission tomography; SE, standard error; SUVR, standardized uptake value ratio.

^a Variables were standardized to allow comparison of regression coefficients. Standard scores were calculated using the mean and SD from the subsample (n = 38) containing eGFR and amyloid PET data.

In sensitivity analyses, we investigated associations between impaired kidney function and p‐tau217 in the participant subset who had blood collected for creatinine within 1 month of blood collection for p‐tau217 (n = 59). Results from these analyses were largely the same as in the main analyses (see Table 2) that included creatinine collected within an average of 3.5 years (SD = 3.3) prior to blood collection for p‐tau217. When testing the three‐level kidney function group in an age‐ and sex‐adjusted regression analysis, the mildly to severely decreased, but not the mildly decreased, kidney function group had significantly higher plasma p‐tau217 than the normal group.

3.2.2. Cardiovascular disease (CVD)

CVD history, defined as history of MI, stroke, and/or CHF, was significantly related to higher plasma p‐tau217 in unadjusted (χ²[1] = 7.6, p = 0.01; r _pb = 0.25 [95% CI: 0.12 to 0.37]; see Figure 2B) and age‐ and sex‐adjusted analyses (see Table 2).

Because eGFR and CVD are common comorbidities, we conducted a sensitivity analysis to test the relationship between CVD and p‐tau217 when adjusting for continuous eGFR. In the subset of participants who had CVD and eGFR data (n = 115), CVD history was significantly related to higher p‐tau217 (b = 0.37, SE = 0.18, p = 0.04) when eGFR, age, and sex were controlled. eGFR was also significant in this model (b = ‐0.005, SE = 0.002, p = 0.03). In this subset, 8 participants had CVD history and 16 had eGFR <60 mL/min/1.73 m². Five participants had both CVD and eGFR <60 mL/min/1.73 m².

Four participants had a history of CVD in the subset (n = 64) with both CVD and amyloid‐PET data. Prior to proceeding with the amyloid‐PET adjusted analysis, we examined if the association between CVD and p‐tau217 was still significant in this smaller subset and found that it was not significant (see Table S1). Thus, we did not have enough participants with CVD history to test if it was related to p‐tau217 in a model that adjusted for amyloid PET. In separate age‐ and sex‐adjusted analyses, CVD was not significantly related to amyloid‐PET DVR (b = 0.01, SE = 0.08, p = 0.88) or tau‐PET SUVR (b = 0.06, SE = 0.08, p = 0.44).

3.2.3. Obesity

Continuous BMI (r _s[230] = 0.03, p = 0.69; see Figure 1B) and obesity categories (χ²[4] = 1.6, p = 0.80; see Figure 2C) were not significantly related to plasma p‐tau217 in unadjusted or age‐ and sex‐adjusted analyses (see Table 2). BMI was not significantly related to p‐tau217 in an analysis adjusting for amyloid‐PET DVR, age, and sex (b = 0.002, SE = 0.007, p = 0.80; n = 65). Because impaired kidney function is comorbid with obesity and could potentially obscure a negative association between BMI and p‐tau217, we tested the association between BMI and p‐tau217 when adjusting for age, sex, and eGFR (n = 137). BMI was not significantly related to p‐tau217 in this analysis (b = 0.008, SE = 0.006, p = 0.13); however, continuous eGFR was significant (b = –0.009, SE = 0.002, p < 0.001).

3.2.4. Diabetes

History of diabetes was not significantly associated with plasma p‐tau217 in unadjusted analysis (χ²[1] = 2.2, p = 0.14; see Figure 2D), age‐ and sex‐adjusted analysis (see Table 2), and in an analysis adjusted for amyloid‐PET DVR, age, and sex (history of diabetes [yes]: b = –0.08, SE = 0.11, p = 0.49; n = 64, n = 26/64 with history of diabetes).

3.2.5. Hypertension

SBP was significantly related to p‐tau217 in an unadjusted model (r _s[229] = 0.17, p = 0.01; see Figure 1C) but not in the age‐ and sex‐adjusted analysis (see Table 2). BP category was significantly related in an unadjusted model (χ²[3] = 10.7, p = 0.01; see Figure 2E); Stage 1 hypertension (z = –2.9, p = 0.002; r _pb = ‐0.25 [95% CI: −0.09 to −0.37]) and normal BP (z = −3.2, p < 0.001; r _pb = ‐0.28 [95% CI: −0.08 to −0.45]) demonstrated significantly lower p‐tau217 than Stage 2 hypertension. However, no BP category was significantly related to p‐tau217 when age and self‐identified sex were controlled (see Table 2). SBP was not significantly related to p‐tau217 in an analysis adjusted for amyloid‐PET DVR, age, and sex (b = 0.0008, SE = 0.003, p = 0.78; n = 65).

3.2.6. AD PET biomarker positivity status

Amyloid‐ and tau‐PET‐positive status was associated with higher plasma p‐tau217 values in unadjusted models (amyloid‐PET positivity: χ²[1] = 23.6, p = < 0.001; r _pb = 0.54 [95% CI: 0.35 to 0.70]; tau PET positivity: χ²[1] = 17.4, p < 0.001; r _pb = 0.56 [95% CI: 0.38 to 0.71]) and in age‐ and sex‐adjusted analyses (see Table 2 and Figure 3). In the subsample that had both amyloid‐ and tau‐PET data (n = 57), n = 8 participants (14.0%) were A+T+, n = 7 (12.2%) were A+T−, n = 40 (70.2%) were A−T−, and n = 2 (3.5%) were A−T+ (this latter group was not included in the analysis of the subset combining A and T statuses). There was a significant main effect of AT PET biomarker status on p‐tau217 (χ²[2] = 21.7, p < 0.001) in the unadjusted analysis; A–T– status had significantly lower p‐tau217 than A+T– (z = –2.7, p = 0.004; r _pb = ‐0.44 [95% CI: –0.17 to –0.65]) and A+T+ (z = –4.2, p < 0.001; r _pb = –0.67 [95% CI: –0.47 to –0.80]) statuses (see Figure 2). A+T– status did not have significantly lower p‐tau217 than A+T+ status (z = –1.7, p = 0.16). In the age‐ and sex‐adjusted analysis, both the A+T+ and A+T– groups had higher p‐tau217 than the A–T– group (see Table 2 and Figure 3).

FIGURE 3.

Box plots illustrate unadjusted associations between amyloid‐ and tau‐PET positivity and plasma p‐tau217. Scatter labeled according to kidney function categories. Dashed lines in (A) and (B) represent the p‐tau217 cut‐point for each AD biomarker. Beneath each box plot is a cross‐tabulation table showing the count within each kidney function and PET‐positivity status category. A, amyloid; PET, positron emission tomography; T, tau.

3.3. Accuracy of p‐tau217 to discriminate amyloid‐ and tau‐PET‐positive from ‐negative status

ROC analyses revealed that plasma p‐tau217 demonstrated good diagnostic accuracy in detecting AD PET biomarker positive status: the AUC for amyloid‐PET was 0.90 (95% CI: 0.82 to 0.98) and for tau PET was 0.89 (95% CI: 0.78 to 0.99) (see Figure 4). The threshold for detecting amyloid‐PET‐positive status derived using Youden's index was 0.35 pg/mL, resulting in a sensitivity of 0.94 and specificity of 0.80. The threshold for detecting tau‐PET positivity was 0.556 pg/mL, resulting in a sensitivity of 0.75 and specificity of 0.93. Based on our sample, where 24.6% of the participants were amyloid‐PET positive, the positive predictive value (PPV) and negative predictive value (NPV) for p‐tau217 were 0.60 and 0.98, respectively. The PPV describes the proportion of positive cases (i.e., participants with p‐tau217 >0.35) who were true positives (i.e., amyloid‐PET positive), whereas the NPV describes the proportion of negative cases that were true negatives (i.e., amyloid‐PET negative).

FIGURE 4.

ROC curves for plasma p‐tau217 in detecting amyloid‐ and tau‐PET positivity. Plasma p‐tau217 cut‐points were 94% sensitive and 80% specific for amyloid‐PET‐positive status and 75% sensitive and 93% specific for tau‐PET‐positive status. Beneath each ROC curve is a cross‐tabulation table depicting the count in each plasma p‐tau217 and PET biomarker positivity status category. DVR, distribution volume ratio; PET, positron emission tomography; PiB, Pittsburgh Compound B; ROC, Receiver‐operating characteristic; ROI, region of interest; SUVR, standardized uptake value ratio.

We also examined sensitivity and specificity using a previously published ALZPath p‐tau217 threshold (0.42 pg/mL) for amyloid‐PET‐positive status (global PiB index DVR >1.2) that was determined using a predominantly cognitively unimpaired non‐Hispanic White adult sample. ² Relative to this sample, we found that sensitivity was lower (0.73 vs 0.95) and specificity was higher (0.89 vs 0.745) for amyloid‐PET‐positive status (16.9% of our sample had global PiB index DVR >1.2).

To explore the effectiveness of a two‐cut‐point strategy for isolating false‐positive amyloid status, a lower cut‐point having 94% sensitivity (i.e., p‐tau217 >0.35 pg/mL) and an upper cut‐point with 96% specificity (i.e., p‐tau217 >0.55 pg/mL) were selected. The positive percent agreement of the upper cut‐point was 87.5%, and the negative percent agreement of the lower cut‐point was 95.1%. The intermediate group included 24.6% of the whole sample and isolated 80% (8/10) of the false‐positive cases and 50.0% (8/16) of the true‐positive cases. Amyloid PET DVR for the two false‐positive cases that were not isolated in the intermediate zone was 1.14 and 1.03; the amyloid PET DVR for the remaining eight false‐positive cases ranged from 1.01 to 1.11 (median = 1.05). The lower cut‐point was the same as the binary cut‐point for amyloid‐PET‐positive status; thus, false‐negative cases, which were already low (n = 1), were not isolated in the intermediate group.

Because of missing creatinine data (n = 27 in both PET subsamples), we were unable to determine whether impaired kidney function was associated with the false‐positive rate.

4. DISCUSSION

In a Black middle‐aged and older adult sample enrolled in the AA‐FAIM study, we found that plasma p‐tau217 levels were significantly higher in individuals with impaired kidney function and CVD but not hypertension or diabetes, consistent with some ¹⁰ , ¹¹ , ¹² but not all ¹³ , ¹⁴ prior research. Contrary to previous research, ⁹ , ¹¹ , ¹² obesity and higher BMI were not significantly related to lower p‐tau217. In a smaller subset who had PET, we found that p‐tau217 demonstrated promise as a biomarker for detecting abnormal amyloid‐ and tau‐PET. We elaborate upon these findings, discuss their implications, and describe study limitations, offering suggestions for future research.

Plasma p‐tau217 was higher in participants with mild to severely decreased kidney function, defined as an eGFR <60 mL/min/1.73 m², a value diagnostic of CKD when persisting >3 months. Congruent with previous research, ¹⁰ , ¹¹ participants with lower (worse) eGFR had higher plasma p‐tau217 when adjusting for age and sex, as well as amyloid PET. Although eGFR was related to plasma p‐tau217, it was not significantly related to amyloid‐ or tau‐PET. Taken together, findings suggest that the association between worse kidney function and higher plasma p‐tau217 may be at least partly independent of cortical amyloid; however, sample sizes tested were small, and further examination is needed in a larger sample.

Most of the participants in our sample were cognitively unimpaired; thus, we were unable to evaluate whether the association between kidney function and p‐tau217 varied across cognitive statuses. Relative to unimpaired adults, cognitively impaired adults have been found to demonstrate weaker relationships between CKD and plasma p‐tau217 and greater elevations in p‐tau217 associated with amyloid positivity versus CKD. ¹⁰ Thus, the association between CKD and plasma p‐tau217 may decrease as more AD pathology accumulates in later disease.

Impaired kidney function has been associated with an increased risk for cognitive impairment in some studies. ⁴² , ⁴³ To rule out AD as a cause of impairment, clinicians may rely on plasma p‐tau217 once it is fully approved as an amyloid biomarker by the FDA. Prior to implementation, it will be important to understand how much variability in p‐tau217 is explained by CKD across different cognitive and amyloid status groups. Furthermore, methods that minimize misinterpretation of levels in individuals with CKD should be developed. For example, an alternative plasma biomarker may be the p‐tau217/non‐phosphorylated‐tau‐217 ratio (p‐tau217/np‐tau217). In a study enrolling Swedish adults, CKD was less associated with p‐tau217/np‐tau217 than with p‐tau217 alone. ¹⁰ A similar study conducted with adults from diverse racial and ethnic groups is needed.

Participants with a history of CVD had higher p‐tau217. When adjusting for eGFR in addition to age and sex, we found that CVD was still significantly related to higher p‐tau217. CVD and CKD are comorbid conditions. Although these results suggest that CVD may be associated with p‐tau217 apart from reduced eGFR, replication is needed in a sample with a larger number of participants with CVD and impaired kidney function.

In another study, MI and stroke were significantly related to higher p‐tau217 when adjusting for amyloid‐PET, age, and sex. ¹² Whether CVD is associated with p‐tau217 levels independent of cortical amyloid requires further investigation. Because of the small number of participants with CVD (n = 4) in the amyloid‐PET subsample, we were unable to examine if the association between CVD and p‐tau217 persisted after adjusting for cortical amyloid. Results from studies investigating a linkage between vascular disease or increased vascular risk and biomarkers of AD pathology have been inconsistent. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸

Obesity categories and BMI were not related to plasma p‐tau217, in contrast to previous studies finding significant inverse associations. ⁹ , ¹¹ , ¹² Accelerated weight loss has been observed at least 6 years prior to clinical AD diagnosis ⁴⁹ , ⁵⁰ and may account for the inverse associations previously found between BMI and amyloid biomarkers. Relatedly, higher BMI in older (but not midlife) adults has been related to decreased dementia risk. ⁵¹ Our results may have differed from other studies because our sample had more participants who were younger and cognitively unimpaired; individuals with preclinical AD may have been in the earliest stages of disease before appreciable weight loss typically occurs.

Diabetes was not significantly related to plasma p‐tau217. SBP and Stage 2 hypertension were related to higher p‐tau217 in unadjusted analyses but were not significantly related when controlling for age and sex, consistent with previous studies. ¹¹ , ¹² Diabetes, hypertension, and obesity are risk factors for CVD, which in our study were associated with elevated p‐tau217. Whether the association between vascular risk and p‐tau217 depends upon the severity of cardiometabolic and vascular dysfunction should be investigated.

In a smaller subsample, plasma p‐tau217 showed promise as an accurate biomarker of amyloid and tau assessed by PET. Higher levels were observed for amyloid‐ and tau‐PET positivity. Levels of p‐tau217 differentiated PET biomarker status, demonstrating AUCs for amyloid (0.90) and tau (0.89) positivity only slightly lower than reported (0.93) in a recent study. ² According to guidelines published by the Global CEO Initiative Workgroup on AD Blood Biomarker Tests, ⁵² the sensitivity (94%) and specificity (80%) for amyloid‐PET positivity that we found indicated a performance level that may be acceptable for a triaging test in secondary care settings that have high capacity for follow‐up testing (PET or CSF) on individuals with positive blood tests. Without high‐capacity follow‐up, specificities ≥85% are recommended to limit the number of false‐positive cases. ⁵² A primary purpose of triaging tests is to discover individuals who likely do not have amyloid pathology; thus, triaging tests are recommended for patient populations with ≤50% pre‐test probability of amyloid positivity. ⁵² In that population, test sensitivities ≥90% would result in high NPVs. Performance standards proposed are relevant for tests that will be performed on individuals with objective evidence of cognitive impairment. ⁵² Our sample included primarily cognitively unimpaired adults; however, 24.6% of the participants were amyloid‐PET positive, which is slightly higher than the theoretical rate (20%) for low‐prevalence amyloid pathology used to calculate recommended PPVs and NPVs in the Global CEO Initiative report. ⁵² With that prevalence rate, a high‐specificity (85%) triaging test with 90% sensitivity would have a PPV of 60% and NPV of 97%. Calculated using that same rate, p‐tau217 in our sample had a PPV of 54% and a comparable NPV of 98%.

A two‐cut‐point strategy has been suggested as a method for minimizing false‐positive results and reducing costs for follow‐up testing when selecting patients for anti‐amyloid therapy. ⁵³ , ⁵⁴ With this approach, follow‐up testing is reserved for individuals with p‐tau217 values intermediate of the upper and lower cut‐point. In our study, the intermediate group contained 80% (8/10) of false‐positive and 50% (8/16) of true‐positive cases; thus, n = 16 individuals (vs n = 25 if using the binary cut‐point) would have received follow‐up testing and two false‐positive cases would not have received follow‐up. The clinical impact of the two‐cut‐point strategy requires additional investigation in larger samples of Black adults with and without medical comorbidities shown to influence p‐tau217 levels.

Our study has several limitations in addition to those noted previously. Because our study was cross‐sectional and correlational, causative claims cannot be made. Although our results suggest that reduced eGFR may increase the possibility of a false‐positive p‐tau217 test result, we were unable to definitively assess if impaired kidney function was associated with the false‐positive rate for amyloid or tau. Blood collected for creatinine occurred on average 3.5 years (SD = 3.3) prior to blood collection for p‐tau217. Because eGFR has been shown to decline with age, ³⁵ eGFR values for some participants may have been higher than they would have been at the time of p‐tau217 collection. Sensitivity analyses conducted using creatinine collected within 1 month of p‐tau217 supported the main finding that lower eGFR is related to higher p‐tau217 levels. Whether impaired kidney function and CVD elevate p‐tau217 levels independent of amyloid and/or play a role in the AD pathophysiologic process requires additional investigation. Because p‐tau217/np‐tau217 and p‐tau181/np‐tau181 have been shown to be less influenced by CKD than either p‐tau species alone, ¹⁰ it is possible that reduced kidney filtration increases p‐tau levels. CVD is a more common neuropathologic finding in CKD than amyloid plaques or tau tangles ⁵⁵ ; however, higher homocysteine, which is associated with lower glomerular filtration rate, ⁵⁶ has been related to greater tau burden. ⁵⁷ Amyloid‐ and tau‐PET should be tested as mediators of the relationship between changes in eGFR and p‐tau217 (and p‐tau217/np‐tau217) to examine the extent that kidney function is associated with p‐tau217 independent of amyloid and tau.

CVD and diabetes were identified through clinician‐ and/or self‐report, which is less valid than objective assessments of disease. Future work should examine whether elevated fasting glucose and glycated hemoglobin values indicative of prediabetes and diabetes influence plasma p‐tau217 levels. We measured kidney function according to well‐recognized eGFR categories ³¹ ; however, we did not have cystatin C to confirm eGFR in participants with values <60 mL/min/1.73 m² or urine albumin–creatinine ratio to assess kidney damage. ³¹ , ³²

Because the subsample of participants with PET data was small, we consider analyses examining the diagnostic accuracy of plasma p‐tau217 and cut‐points for amyloid‐ and tau‐positive status to be exploratory. Replication of findings and further investigation of the utility of the two‐cut‐point strategy are needed using larger samples than our own. Samples should include cognitively unimpaired and impaired adults from diverse racial and ethnic groups underrepresented in medical research and contain enough individuals with medical comorbidities to detect clinically meaningful influential effects, if present, on p‐tau‐217. This work is needed before cut‐points relevant for a racially and ethnically diverse population, as found in the United States, are established.

In conclusion, plasma p‐tau217 demonstrated promise as an accurate biomarker for detecting abnormal amyloid‐ and tau‐PET in Black middle‐aged and older adults who were predominantly cognitively unimpaired. Impaired kidney function and CVD, however, were associated with higher plasma p‐tau217 levels. Although additional investigation is needed to examine if associations are independent of AD pathology, findings suggest that the presence of medical comorbidities prevalent in Black adults in the United States should be considered when interpreting plasma p‐tau217.

CONFLICT OF INTEREST STATEMENT

Carey E. Gleason is a scientific advisor for Huntington Research Institute, is a member of Alzheimer's Disease Research Center External Advisory Boards (Stanford University; University of New Mexico), is the Chair of the Institutional Data and Safety Monitoring Board for a National Institute on Aging (NIA)–funded study (R01‐AG074971), and has been a scientific advisor for a documentary film Matter of Mind: My Alzheimer's produced in partnership with the Public Broadcasting Service. Carey E. Gleason and Megan Zuelsdroff are members of the Scientific Advisory Board for an NIA‐funded study (UH2AG083258). Megan Zuelsdroff is a Senior Associate Editor for Alzheimer's & Dementia. Sanjay Asthana was an editor of Hazzard's Geriatric Medicine and Gerontology (McGraw Hill). Sterling C. Johnson has served as a consultant to Enigma and ALZpath. Henrik Zetterberg. has served at scientific advisory boards and/or as a consultant for Abbvie, Acumen, Alector, Alzinova, ALZpath, Amylyx, Annexon, Apellis, Artery Therapeutics, AZTherapies, Cognito Therapeutics, CogRx, Denali, Eisai, Enigma, LabCorp, Merry Life, Nervgen, Novo Nordisk, Optoceutics, Passage Bio, Pinteon Therapeutics, Prothena, Quanterix, Red Abbey Labs, reMYND, Roche, Samumed, Siemens Healthineers, Triplet Therapeutics, and Wave; has given lectures sponsored by Alzecure, BioArctic, Biogen, Cellectricon, Fujirebio, Lilly, Novo Nordisk, Roche, and WebMD; and is a co‐founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program (outside submitted work). Gilda E. Ennis, Derek Norton, Rebecca E. Langhough, Diane C. Gooding, Fabu P. Carter, Rachael Wilson, and Shenikqua Bouges have nothing to disclose. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All study procedures were approved by the University of Wisconsin–Madison Institutional Review Board and were conducted in compliance with the World Medical Association Declaration of Helsinki. All participants provided written informed consent.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Ennis GE, Norton DK, Langhough RE, et al. The performance of plasma p‐tau217 in Black middle‐aged and older adults. Alzheimer's Dement. 2025;21:e70288. 10.1002/alz.70288