Alzheimer's disease diagnostic progression is associated with cerebrovascular disease and neuroinflammation in adults with Down syndrome

Abstract

INTRODUCTION

Despite having few vascular risk factors, people with Down syndrome (DS) have MRI evidence of cerebrovascular disease (CVD) and neuroinflammation that worsens with Alzheimer's disease (AD) severity. We investigated whether markers of CVD and inflammation are associated with AD‐related diagnostic progression in people with DS.

METHODS

We included 149 participants (mean age [SD] = 44.6 [9]) from the Alzheimer's Biomarkers Consortium–Down Syndrome who had two (n = 24) or three follow‐up visits (n = 125). We derived white matter hyperintensity (WMH) volume and plasma biomarker (glial fibrillary acidic protein [GFAP], amyloid beta [Aβ]42/Aβ40, hyperphosphorylated tau‐217 [p‐tau217], and neurofilament light [NfL]) concentrations at baseline and examined their association with progression in clinical diagnosis.

RESULTS

Higher baseline WMH volume and higher GFAP were associated with a greater likelihood of diagnostic progression. Combining WMH and GFAP with p‐tau217 improved clinical conversion classification accuracy over AD biomarkers alone. Among individuals with evidence of amyloidosis, both WMH and GFAP were associated with clinical progression.

DISCUSSION

In DS, markers of CVD and inflammation are independently and synergistically associated with clinical AD progression.

Highlights

• Higher baseline white matter hyperintensity (WMH) volume and plasma glial fibrillary acidic protein (GFAP) concentration were associated with a higher likelihood of progressing from cognitively stable to either mild cognitive impairment or clinical Alzheimer's disease in Down syndrome.

• WMH volume and GFAP concentration discriminated between those who progressed and those who did not.

• Models including the independent and interactive effects of WMH and GFAP more accurately discriminated between participants who progressed diagnostically from those who did not.

• Individuals with evidence of amyloid pathology were more likely to progress if they also had elevated WMH or GFAP.

1. BACKGROUND

Ninety percent of individuals with Down syndrome (DS) accumulate amyloid beta (Aβ) plaques and tau neurofibrillary tangles (NFTs) by the age of 40 years, ¹ , ² and most develop dementia by the age of 60 years. ³ Today, DS is considered a genetic form of AD, ⁴ and although AD pathogenesis has long been attributed to the role of the triplication of the 21st chromosome in promoting amyloid pathology, ⁵ there is increasing evidence that, like in the neurotypical population, ⁶ cerebrovascular and inflammatory factors are critical in disease pathogenesis and course. ⁷ , ⁸ , ⁹ , ¹⁰

Our recent magnetic resonance imaging (MRI) studies indicate that cerebrovascular pathology, ¹¹ in the form of white matter hyperintensities (WMH), cerebral microbleeds, infarcts, and enlarged perivascular spaces, is a prominent feature in adults with DS that differs among those characterized as cognitively stable, having mild cognitive impairment (MCI), and having Alzheimer's disease (AD) dementia. ¹¹ Like primary AD pathophysiology, cerebrovascular pathology increases in an age‐dependent fashion, with evidence that it precedes or emerges contemporaneously with markers of amyloid and tau pathology, measured with positron emission tomography (PET). ¹² People with DS have few systemic vascular risk factors, such as hypertension or type 2 diabetes, ¹³ despite having evidence of prominent cerebrovascular disease (CVD) on MRI scanning. We thus hypothesized that there is an “endogenous” cerebrovascular component that represents a “core feature” of AD ¹⁴ not attributable solely to vascular comorbidity, which reflects or interacts with AD‐related neuroinflammation.

Astrocytosis manifesting as elevated glial fibrillary acidic protein (GFAP) level is emerging as a component of a neuroinflammatory cascade that may be critical for AD pathogenesis and progression, including increasing phosphorylated tau and neurofilament light (NfL) levels. ⁷ , ⁹ , ¹⁵ , ¹⁶ In the DS population, GFAP concentrations discriminate accurately between symptomatic and asymptomatic individuals, ¹⁷ and they partially mediate the relationship between amyloid and tau pathology, assessed with PET. ¹⁸ Furthermore, our previous work suggests that among individuals with DS, CVD promotes neurodegeneration via astrocytosis tau pathology, and that CVD and astrocytosis interact synergistically such that the relationship between CVD and markers of tau pathology becomes stronger with greater evidence of astrocytosis. ⁷

In the current study, using longitudinal data from the Alzheimer's Biomarker Consortium–Down Syndrome (ABC‐DS), we hypothesized that CVD and astrocytosis are independently and interactively associated with clinical AD progression in adults with DS. We tested whether WMH volume and plasma GFAP, hyperphosphorylated tau‐217 (p‐tau217), NfL, and Aβ42/40 concentrations at baseline are associated with incident MCI and dementia diagnoses.

2. METHODS

2.1. Participants

Participants in this study were part of the ABC‐DS (U19 AG068054), a multisite, observational study that examines the biomarker, clinical, and genetic factors associated with AD in adults with DS. ¹⁹ The sample in this study included 149 individuals with trisomy 21 from two studies—Neurodegeneration in Aging Down Syndrome (NiAD; U01 AG051406) and Biomarkers of Alzheimer's Disease in Adults with Down Syndrome (ADDS; U01 AG051412)—both of which are now part of ABC‐DS. Participants undergo evaluations at baseline and then at 16‐month intervals, where they are administered neuropsychological tests and undergo blood draws and MRI and PET imaging. Apolipoprotein E (APOE) genotyping was obtained from blood samples, and participants were classified as APOE4+ based on the presence of at least one ε4 allele. In addition, premorbid intellectual disability (ID) levels were determined using standardized assessments of intellectual functioning documented in medical records or administered during the ABC‐DS study, which included the Stanford‐Binet Intelligence Scales, Fifth Edition Abbreviated Battery (SB5 AB) ¹⁸ or the Kaufman Brief Intelligence Test, Second Edition (KBIT‐2), ¹⁹ prior to any signs of dementia. These assessments are interpreted alongside adaptive functioning information provided by study partners. Mental age equivalent scores were used to classify ID severity due to the limited range of standard scores on the SB5 AB and KBIT‐2. The classifications based on mental age equivalents were as follows: mild (≥9.0 years), moderate (4.0–8.9 years), and severe/profound (≤4.0 years).

For the current analysis, we selected participants who had available baseline MRI data and plasma biomarkers of interest for at least two time‐points and for whom clinical consensus diagnosis data were also available (i.e., cognitively stable, mild cognitive impairment [MCI‐DS], and Down syndrome AD dementia [DS‐AD], or unable to be determined; N = 181). Those with a baseline diagnosis of DS‐AD (N = 23) and/or diagnoses that were unable to be determined (N = 5) were not included. Individuals whose diagnoses changed from one reflecting more impairment to one reflecting less impairment were excluded (N = 4). Diagnoses were made through a consensus conference that reviewed clinical and neuropsychological data, as described previously. ¹⁹ Individuals were classified as progressing diagnostically if their diagnosis changed from cognitively stable or MCI‐DS to MCI‐DS, DS‐AD, or unable to be determined, at one of the available follow‐up visits.

RESEARCH IN CONTEXT

1. Systematic review: By age 40, most individuals with Down syndrome (DS) develop Alzheimer's disease (AD) pathology and progress to dementia by age 60. Adults with DS have evidence of cerebrovascular dysfunction and neuroinflammation, but it is unclear whether these factors contribute to the clinical progression of AD.

2. Interpretation: Among individuals with DS, markers of cerebrovascular disease and neuroinflammation are associated with AD‐related diagnostic progression. The interaction of cerebrovascular disease and inflammation together with markers of phosphorylated tau more strongly predicts incident diagnosis of mild cognitive impairment or clinical AD than AD biomarkers alone. In the context of primary AD pathology, cerebrovascular dysfunction and neuroinflammation may play essential roles in the clinical progression of AD in people with DS.

3. Future directions: Future work will focus on understanding the vascular–inflammation interface to identify targets for therapeutic intervention or prevention.

2.2. MRI

MRI scans were obtained at both ADDS and NiAD participating sites. The NiAD sites performed two dimensional (2D) T2‐weighted fluid‐attenuated inversion recovery (FLAIR) scans (repetition time [TR]/echo time [TE]/inversion time [TI] = 5000/386/1800 ms, voxel size = 0.4 × 0.4 × 0.9 mm³), whereas the ADDS sites used 3D T2‐weighted FLAIR scans (TR/TE/TI = 4800/119/1473 ms, voxel size = 0.9 × 0.9 × 0.5 mm³).

White matter hyperintensity volume was quantified with in‐house software that has been described in detail previously. ⁷ Briefly, FLAIR images were first reconstructed to a 256 × 256 × 256 matrix with a 1 mm³ voxel size, and then reoriented to Montreal Neurological Institute (MNI)152 space, skull‐stripped, and bias field‐corrected. ²⁰ , ²¹ A custom module extracted percentile thresholds from the intensity histograms of each image, defining transitions between dark, bright, and brightest voxel intensities. Next, a white matter segment was created using a convolutional neural networks tool. ²² , ²³ , ²⁴ The percentile threshold initialized a Gaussian mixture model and expectation‐maximization algorithm ²⁵ to separate hyperintense and non‐hyperintense voxels in the white matter segment. To account for image quality variations, an inter‐percentile range was calculated and a relaxed threshold was applied. Probability distribution maps for WMH were generated, followed by edge detection to remove non–white matter voxels. ²⁶ The labeled voxels were summed and multiplied by voxel dimensions to calculate the total WMH volume in cubic centimeters. Figure 1 shows the voxel‐wise distribution of WMH across all participants.

FIGURE 1.

Frequency map of white matter hyperintensities (WMH) in adults with Down syndrome. A voxel‐wise frequency map of WMH was created by summing voxels labeled across all 149 individual 3D masks and dividing by 149. Each voxel's value represents the percentage each voxel was labeled as WMH across the 149 masks from lower frequency (blue) to higher frequency (red).

2.3. Plasma samples and analysis

Concentrations of Aβ42, Aβ40, p‐tau217, NfL, and GFAP were measured for each participant from plasma samples as described previously. ⁷ The plasma samples were sent to the University of North Texas, where Aβ42, Aβ40, and NfL concentrations were quantified with single‐molecule array (Simoa) assays (Quanterix). The ratio of Aβ42 to Aβ40 was calculated as a biomarker for amyloid pathology. Plasma samples from the same cohort were then shipped to Lund University for the measurement of p‐tau217 and GFAP concentrations; p‐tau217 was quantified via immunoassay on a Meso Scale Discovery platform, as per published protocols developed by Lilly Research Laboratories, ²⁷ , ²⁸ and GFAP concentrations were assessed using Simoa assays (Quanterix).

2.4. PET

We hypothesized that higher levels of WMH, GFAP, p‐tau217, and NfL are associated with a greater likelihood of diagnostic progression, but based on our previous work we did not expect to see meaningful associations with plasma Aβ42/40 concentrations, possibly due to relatively lower accuracy of plasma measures of amyloid pathophysiology. Here, we used amyloid PET measures available for a subset of participants to examine if higher levels of CVD and astrocytosis moderate the association between amyloid pathology and diagnostic conversion. A subset of 104 participants (mean age [SD] = 40.4 [7]; 50% women) underwent amyloid PET imaging using either [¹¹C]PiB (Pittsburgh compound B; 15 mCi nominal, scan performed 50–70 min after injection with 5 min frames) or [¹⁸F]florbetapir (aka AV45; 10 mCi nominal, scan performed 80–100 min post‐injection with 5 min frames). All PET data were corrected for factors including attenuation, detection dead time, scanner normalization, scatter, radioactive decay, and interframe motion. Centiloid (CL) values were derived from Aβ standard uptake value ratios (SUVRs)and standardized across a scale from 0 to 100 as described previously. ²⁹ A cutoff of 18 CL was used to indicate “amyloid positivity” based on previous work. ²⁷ , ³⁰

2.5. Statistical analysis

We conducted a series of logistic regressions to examine the association of WMH volume and AD plasma biomarker concentrations with diagnostic progression. For each model, we extracted the odds ratio (OR) and corresponding 95% confidence interval (CI) to quantify the strength and direction of the association between the predictor variables and diagnostic progression. We then used the “pROC” package in R ³¹ to generate receiver‐operating characteristic (ROC) curves and the corresponding area under the curve (AUC), F1 scores, balanced accuracy, and Matthew's correlation coefficient (MCC) values to examine how well baseline WMH volume and AD plasma biomarker concentrations distinguished individuals who progressed from those who did not. These models were chosen and run systematically, first with age alone as a reference model, then with WMH and individual biomarkers as predictors, and finally, testing our hypothesis that the interaction between WMH and GFAP improved overall classification accuracy. Confusion matrices were generated for each model to detail the distribution of true positives, true negatives, false positives, and false negatives for each corresponding ROC curve (Figure S1). These analyses were adjusted for study site, sex/gender, number of days between visit dates, age, and run in R version 2023.12.1+402.

Next, we examined the association of WMH, GFAP, and amyloid pathology with diagnostic progression. For the first analysis, we divided participants into “low WMH” and “high WMH” (based on their median value) and amyloid‐positive and amyloid‐negative groups, yielding 4 Aβ/WMH groups: 29 (27.9%) were in the Aβ–/low WMH group; 25 (24%) were in the Aβ–/high WMH group; 23 (22.1%) were in the Aβ+/low WMH group; and 27 (26%) were in the Aβ+/high WMH group. We then repeated this classification with GFAP and divided the sample into four Aβ/GFAP groups: 41 (39.4%) were in the Aβ–/low GFAP group; 13 (12.5%) were in the Aβ–/high GFAP group; 11 (10.6%) were in the Aβ+/low GFAP group; and 39 (37.5%) were in the Aβ+/high GAFP group. We compared the proportion of individuals who progressed diagnostically as a function of the two sets of four groups with chi‐square analysis.

3. RESULTS

Twenty‐four participants had longitudinal data from two visits, and 125 had data from three longitudinal visits. Fifteen progressed from cognitively stable to MCI‐DS, 17 from MCI‐DS to DS‐AD, 3 from cognitively stable to DS‐AD, 2 from cognitively stable to an undetermined diagnosis, and 1 from MCI‐DS to an undetermined diagnosis. Diagnoses for three participants changed to “unable to be determined.” We included them in the analyses as individuals who progressed to a more impaired diagnosis. However, we ran sensitivity analyses after excluding them, and the results of the study did not change. Participant characteristics are reported in Table 1. Individuals who did not progress were younger than those who progressed, but there were no differences in the percentage of women, APOE genotyping, or premorbid functioning between the groups. Confusion matrices are displayed in Figure S1.

TABLE 1.

Sample characteristics by diagnostic progression status.

		Stable diagnosis	Diagnostic progression	Whole sample	Test statistic	p‐value
n		102	47	149	—	—
Age, mean (SD), years		42.3 (9)	51.5 (6)	44.6 (9)	t = 7.4	<0.0001
Women, n (%)		50 (49)	13 (28)	63 (42)	χ2 = 1.0	0.33
APOE4+, n (%)		20 (20)	13 (72)	33 (22)	χ2 = 0.79	0.38
Premorbid functioning level	Mild intellectual disability, n (%)	60 (59)	22 (47)	82 (55)
	Moderate intellectual disability, n (%)	33 (29)	22 (47)	55 (37)	χ2 = 2.9 0.23
	Severe intellectual disability, n (%)	9 (8)	3 (6)	12 (8)

Abbreviation: APOE, Apolipoprotein E.

Those who clinically progressed had higher baseline WMH (2.87 cm³ vs 1.53 cm³, t = 5.43, p = 0.02) and GFAP concentrations (258.12 pg/mL vs 130.98 pg/mL, t = 4.81, p < 0.0001). Higher baseline WMH volume (OR 1.08, 95% CI: 1.01–1.81), and higher GFAP (OR 1.006, 1.003–1.01) and p‐tau217 (2.06, 1.49–2.88) concentrations were associated with a higher odds of clinical progression. Baseline NfL concentration (1.02, 0.99–1.06) and Aβ42/40 (OR <0.0001) were not associated with diagnostic progression.

Baseline WMH volume (AUC = 0.83, F1 = 0.70, balanced accuracy = 0.79, MCC = 0.54) and GFAP concentration (AUC = 0.84, F1 = 0.67, balanced accuracy = 0.77, MCC = 0.52) differentiated between those who clinically progressed and those who did not with similar AUC values to Aβ42/Aβ40 (AUC = 0.82, F1 = 0.66, balanced accuracy = 0.76, MCC = 0.51) and NfL (AUC = 0.83, F1 = 0.67, balanced accuracy = 0.77, MCC = 0.50). Baseline p‐tau217 concentration discriminated between those who progressed and those who did not, with slightly greater accuracy than the other individual markers (AUC = 0.88, F1 = 0.71, balanced accuracy = 0.80, MCC = 0.57). Notably, the fit statistics for the WMH by GFAP interaction were greater than Aβ42/Aβ40, NfL, or WMH and GFAP alone (AUC = 0.85, F1 = 0.70, balanced accuracy = 0.79, MCC = 0.54). The classification accuracy of p‐tau217 improved when WMH (AUC = 0.88, F1 = 0.74, balanced accuracy = 0.81, MCC = 0.62) and GFAP (AUC = 0.88, F1 = 0.72, balanced accuracy = 0.80, MCC = 0.58) were added to the models. However, the largest increase in classification accuracy was observed with the addition of the interaction between WMH and GFAP (AUC = 0.88, F1 = 0.75, balanced accuracy = 0.83, MCC = 0.63) to the p‐tau217 model and the most accurate classification model included AD plasma biomarker (Aβ42/Aβ40, p‐tau217, and NfL) combined with the interaction between WMH and GFAP (AUC = 0.89, F1 = 0.77, balanced accuracy = 0.84, MCC = 0.66). Age was used as a reference predictor and was a relatively poor fit compared to the other predictors (AUC = 0.83, F1 = 0.68, balanced accuracy = 0.78, MCC = 0.53) (Figure 2).

FIGURE 2.

Receiver‐operating characteristic (ROC) curves demonstrate how well each model discriminates between participants who progressed diagnostically from those who did not. Aβ, amyloid beta; WMH, white matter hyperintensities; p‐tau217, phosphorylated tau‐217; GFAP, glial fibrillary acidic protein; NfL, neurofilament light.

Next, we examined diagnostic progression in the PET subsample. The four Aβ/WMH groups did differ in age (F(3,148) = 34.79, p < 0.0001). Although Aβ+ was clearly associated with diagnostic progression (Figure 3), 60% more participants progressed diagnostically if they were classified as Aβ+ with high WMH than if they were classified as Aβ+ with low WMH. Notably, the Aβ+/Low WMH and Aβ+ /High WMH groups did not differ in age (p = 0.11). The four Aβ/GFAP groups also did not differ in age (F(3,148) = 41.81, p < 0.0001). Ninety percent more individuals who were Aβ+ with high GFAP concentrations progressed diagnostically than those who were Aβ+ with low GFAP (Figure 3). The Aβ+/Low GFAP and Aβ+/High GFAP groups did differ in age (p = 0.02).

FIGURE 3.

Proportion of subjects who diagnostically progressed. The y‐axis represents the percentage of individuals in each group who progressed. * indicates a statistically significant difference between the two proportion percentages. Aβ, amyloid beta; WMH, white matter hyperintensity; GFAP, glial fibrillary acidic protein.

4. DISCUSSION

Our study provides evidence that among individuals with DS, CVD and astrocytosis are associated with AD‐related clinical diagnostic progression. The interaction between CVD and astrocytosis, together with markers of phosphorylated tau, more reliably classified individuals who progressed than did AD biomarkers alone. Notably, these models predicted phenoconversion over the course of only 2–3 years. Although all biomarker combinations provided excellent classification accuracy that was similar to each other (Figure 2), the findings provide novel biological insight into factors that contribute to the progression of AD among individuals with DS, pointing to the importance of CVD and inflammation. In the context of amyloid pathology, CVD and neuroinflammation may provide a “second hit” necessary for the clinical manifestation and progression of AD. The results implicate CVD and its interface with inflammation as a core feature necessary for the pathophysiological progression of AD in DS.

The possibility of a “second hit” in AD suggests that although the accumulation of Aβ plaques occurs early in the disease, additional pathophysiological factors are necessary for the disease to progress and manifest clinically. ³² In adults with DS, in addition to aging, the second hit is particularly relevant due to early Aβ deposition in this population. ³³ Our study provides evidence that neuroinflammation and cerebrovascular lesions increase the risk for clinical progression in the context of amyloidosis, potentially promoting neurodegeneration and dementia. However, although the results suggest that adults with DS are more likely to progress because they have high WMH and high Aβ and not because they are older, age could be a confounding factor in the analyses that examined GFAP. Compromised blood–brain barrier (BBB) integrity, cerebrovascular dysfunction, and reduced cerebral blood flow can impair brain function and increase vulnerability to neurodegeneration, ³⁴ whereas chronic activation of microglia and astrocytes, leading to the release of pro‐inflammatory cytokines, can exacerbate neuronal damage and accelerate disease progression. ³⁵ , ³⁶ , ³⁷ Astrocytes specifically overexpress GFAP in response to brain Aβ deposition in AD. ³⁸ It is important to emphasize that our results do not suggest that amyloid pathology causes CVD or inflammation in AD among people with DS, but rather speak to the potential impact of the co‐occurrence of vascular and inflammatory factors and their interaction on clinical progression.

The observed interaction between inflammation and markers of cerebrovascular health likely reflects processes at the level of the BBB. The BBB is a highly selective barrier that is composed of a layer of endothelial cells, extracellular matrix, and pericytes. ³⁹ Adults with DS show evidence of BBB disruption, which is increasingly recognized as an important factor in the development of AD and cognitive decline in this population. ⁴⁰  In individuals with DS, disruption of the BBB occurs earlier in life than in the neurotypical population ⁴⁰ , ⁴¹ and is exacerbated by amyloid‐beta precursor protein coding gene (APP) overexpression. ⁴² , ⁴³ Blood brain barrier disruption is associated with reduced blood flow and white matter lesions, ⁴⁴ , ⁴⁵ , ⁴⁶ which are common in adults with DS. ¹¹ , ⁴⁷ These vascular changes can impair oxygen and nutrient delivery to the brain, further exacerbating neurodegeneration. ⁴⁸ , ⁴⁹ In addition, BBB dysfunction causes neuroinflammation, with the activation of microglia and astrocytes, which promotes AD pathology and incident dementia in DS, ⁵⁰ , ⁵¹ , ⁵² as we observed in this study. Furthermore, inflammation can damage the BBB, leading to more severe vascular injury. ⁵³ , ⁵⁴ In DS, there is evidence of extensive microglial activation, dystrophy, and neuroinflammation. ⁵⁵ , ⁵⁶ , ⁵⁷ These bidirectional associations may be cyclical, with cerebrovascular dysfunction promoting neuroinflammation, and neuroinflammation, in turn, worsening vascular health and leading to neurodegeneration and cognitive decline.

There are some limitations to this study. The development, implementation, and understanding of fluidic AD‐related biomarkers are rapidly evolving, and the underlying factors that drive variance in these measures are not fully understood. For example, although p‐tau217 reliably reflects tau PET and cerebrospinal fluid (CSF) p‐tau levels, ⁵⁸ , ⁵⁹  increases in p‐tau217 concentration do capture some degree of Aβ pathology once reaching more elevated levels. ⁵⁹ , ⁶⁰ There is also no unified CL cutoff to indicate Aβ positivity ⁶¹ , ⁶² ; however, with use of a CL >18 threshold, participants in the earliest stages of Aβ+ were captured for this study. In addition, our study examined only one marker of CVD, and although we acknowledge that the etiology of WMH has been widely debated in the context of AD, the prevailing view is that WMH reflects small vessel CVD and dysfunction. ⁶³ Some argue that WMH are attributable solely to cerebral amyloid angiopathy (CAA). ⁶⁴ However, we disagree that the pattern of WMH that we observed is a manifestation of CAA because WMH that are caused by CAA are discrete, punctate lesions, ⁶⁵ not the confluent, distributed patterns of WMH that we observed. Instead, we believe that WMH in the context of AD (and in the absence of vascular risk factors) reflect primarily inflammatory changes at the level of the endothelium, which could contribute to weakening of the vessel walls and downstream tau abnormalities, but further study is needed. The emergence of WMH and amyloid occurs at approximately the same age in adults with DS ¹² and we did not observe that baseline plasma amyloid concentration predicted phenoconversion; therefore, it is unlikely that WMH are attributable solely to amyloidosis in this population. Finally, we acknowledge that the fit statistics for the different ROC curves are quite similar to each other; however, the intent of the comparison across the models is descriptive. The goal of displaying each AUC is to illustrate the factors that discriminate are related to diagnostic progression. The best fitting models fit objectively well, and some do not involve traditional AD markers, highlighting the involvement of cerebrovascular and neuroinflammatory factors in diagnostic progression among adults with DS.

Our findings contribute to a growing body of literature that implicates CVD, neuroinflammation, and the interface between the two, as critical features in the progression of AD in adults with DS. These analyses point toward potential pathways that could not only inform a potential risk profile for intervention but can expand our understanding of AD pathogenesis, course, progression, and clinical manifestation in both the DS and neurotypical populations.

CONFLICT OF INTEREST STATEMENT

Oskar Hansson has received consulting fees for AC Immune, Amylyx, ALZpath, BioArctic, Biogen, Cerveau, Eisai, Eli Lilly, Fujirebio, Merck, Novartis, Novo Nordisk, Roche, Sanofi, and Siemens. Shahid Zaman has received consulting fees from Lundbeck. Donna M. Wilcock has received consulting fees from Biohaven Therapeutics. Michael A. Yassa has received consulting fees from Eisai Cognito Terapeutics, LLC, CuraSen Terapeutics, Inc, and Enthorin Terapeutics, LLC. Elizabeth Head has received consulting fees for Alzheon and Cyclo Therapeutics. Adam Brickman receives compensation for consultation to Cognition Terapeutics and Cognito Terapeutics. He is on the Scientific Advisory Board of CogState. He is an inventor of a patent for white matter hyperintensity quantification (US Patent US9867566B2) and serves on a Data Safety Monitoring Board for the University of Illinois, Urbana‐Champaign. All other authors do not have competing interests to declare. Any author disclosures are available in the Supporting Information.

CONSENT STATEMENT

MRI scans and plasma samples were collected with the written informed consent of the patients or through their authorized representatives. All study protocols are performed in accordance with each respective ABC‐DS institution's Internal Review Board.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Edwards NC, Lao PJ, Alshikho MJ, et al. Alzheimer's disease diagnostic progression is associated with cerebrovascular disease and neuroinflammation in adults with Down syndrome. Alzheimer's Dement. 2025;21:e70726. 10.1002/alz.70726