Preclinical dementia rating scores are associated with plasma phosphorylated tau‐217

Abstract

INTRODUCTION

Simple screening tools are critical for assessing Alzheimer's disease (AD)‐related pre‐dementia changes. This study investigated longitudinal scores from the Quick Dementia Rating System (QDRS), a brief study partner‐reported measure, in relation to baseline levels of the AD biomarker plasma pTau217 in individuals unimpaired at baseline.

METHODS

Data from the Wisconsin Registry for Alzheimer's Prevention (N = 639) were used to examine whether baseline plasma pTau217 (ALZpath assay on Quanterix platform) modified QDRS or Preclinical Alzheimer's Cognitive Composite (PACC3) trajectories (mixed‐effects models; time = age). pTau217*age interaction effects (e.g., high vs low pTau217 simple age slopes) were compared across outcomes.

RESULTS

Higher baseline pTau217 levels were associated with faster functional (QDRS) and cognitive (PACC3) decline. Effect sizes were similar between PACC3 and QDRS. Exploratory analyses showed increased risk of transitioning to impaired QDRS classifications in those with high‐baseline pTau217.

DISCUSSION

This study demonstrates the utility of QDRS for tracking pre‐dementia AD‐related decline.

1. INTRODUCTION

Current methods for identifying Alzheimer's disease (AD), including extensive cognitive batteries and advanced neuroimaging techniques, are often critiqued for being time‐consuming, expensive, and lacking generalizability to broader populations ¹ . These limitations may hinder the early detection and treatment of cognitive impairment ² , ³ , pointing to the need for valid and reliable user‐friendly screening procedures. To address these challenges, the Quick Dementia Rating System (QDRS), a brief and cost‐effective screening, tool was introduced ⁴ . The QDRS shows strong psychometric properties ⁵ , ⁶ and can approximate the Clinical Dementia Rating (CDR ⁷ ) scale ⁸ , ⁹ . While the CDR is a gold standard for staging dementia severity ¹⁰ , ¹¹ , it is also time‐consuming and therefore less practical for screening for preclinical decline.

Despite its promise, the potential of the QDRS as a screening tool remains underexplored, with few studies examining serial data and its relationship with AD‐related biomarkers. In their cross‐sectional study of 121 participants spanning unimpaired to mild dementia, Duff and colleagues ¹ found significant relationships between QDRS total scores and amyloid positron emission tomography (PET) standardized uptake value ratio (SUVR), hippocampal volumes, and apolipoprotein E (APOE) carrier status. In a follow‐up longitudinal study (n = 110) ¹² , they also found that those classified as being PET amyloid positive at baseline showed significantly increased QDRS total scores at follow‐up compared to those who were amyloid negative at baseline. While these studies demonstrated an association between QDRS and PET biomarkers in participants spanning the unimpaired to mild dementia continuum, nothing is yet known about longitudinal QDRS during the predementia phase relative to recently validated plasma biomarkers.

Plasma pTau217 has emerged as an effective biomarker for detecting AD pathology and for differential diagnosis of etiology in mild cognitive impairment (MCI) ¹³ , ¹⁴ , ¹⁵ . pTau217 can identify amyloid plaques even in cognitively normal adults ¹⁶ , ¹⁷ . Compared to other phosphorylated Tau species like pTau181 and pTau231, pTau217 shows greater fold changes, with diagnostic accuracy (area under the curve [AUC]) often exceeding 90% in distinguishing AD ¹⁴ , ¹⁸ , rivalling cerebrospinal fluid (CSF) assessments ¹⁹ . Longitudinal analyses in the Wisconsin Registry for Alzheimer's Prevention (WRAP) cohort also indicate strong associations between plasma pTau217, AD progression, and cognitive decline, further emphasizing its value in early detection and disease monitoring ²⁰ .

Thus, the primary aim of this study was to examine whether baseline plasma pTau217 levels were associated with the longitudinal progression of QDRS scores. Specifically, we investigated whether higher baseline levels of pTau217 were predictive of faster or more severe cognitive decline among those non‐demented at plasma baseline, as measured by changes in QDRS scores.

Our secondary aim was to understand observed plasma pTau217 effect sizes for QDRS outcomes in the context of effect sizes for the Preclinical Alzheimer's Cognitive Composite (PACC; a cognitive outcome designed to be sensitive to preclinical decline). This will provide valuable insights into the sensitivity of different QDRS scores in detecting early cognitive impairment and will further investigate the potential of plasma p‐tau217 as a predictive biomarker for AD‐related cognitive decline.

2. METHOD AND MEASURES

2.1. Study participants

Participant data were drawn from the May 31, 2024 data freeze for WRAP, a longitudinal study to identify midlife factors associated with the development of AD ²¹ . Enrollment of WRAP participants began in 2001, with the first follow‐up visit occurring 2 to 4 years after the baseline visit and all additional visits occurring at about 2‐year intervals thereafter. WRAP participants were free of dementia at enrollment, and the sample was risk‐enriched for late‐onset AD, with 72% of the participants having a parental history of AD dementia. The QDRS was added to the study protocol in 2015.

Eligibility requirements: participants needed to have at least one plasma pTau217 (pg/mL) measurement concurrent with a QDRS assessment and be non‐demented via consensus review at plasma baseline. Since the study focuses on QDRS trajectories, only participants with at least two visits with QDRS assessments were included, resulting in a sample size of N = 639 participants for the aims of this study. All study procedures were approved by the University of Wisconsin School of Medicine and Public Health Institutional Review Board and are in concordance with the Declaration of Helsinki.

2.2. Key assessments

At each study visit, participants typically completed a blood draw, a neuropsychological assessment, and multiple questionnaires related to a broad array of factors, such as lifestyle, modifiable risk factors, medical history, and subjective memory functioning. In the current sample, each participant also had a study partner complete questionnaires about the participant's memory functioning, general cognitive functioning, and activities of daily living. In WRAP, a study partner is a spouse, friend, or family member that knows the subject well and can answer study‐partner‐based questionnaires and interviews regarding the participant. The name and address of the study partner are identified by the participant at the time of the cognitive assessment visit, and study partner questionnaires are mailed after the visit. The study partners are required to know the participant well and were confirmed by the participant as suitable informants at each visit. For additional details on the WRAP assessment battery, please see Johnson et al. ²¹ .

2.2.1. Administration and scoring of the QDRS and the CDR

The CDR ⁷ was added to WRAP's protocol as a study partner measure in 2012, as it was widely regarded as a gold standard ¹⁰ , ²² for measuring global cognitive and functional impairment in Alzheimer's disease and other dementia syndromes. The CDR is a semi‐structured interview that takes about 30 min to complete for distinguishing cognitively normal or MCI from dementia. The QDRS emerged in 2015 as a potential alternative or complement to the CDR. The QDRS contains 10 items covering cognition and behavior and can be completed by the study partner of the person in 3–5 min ⁴ . Based on a supportive validation study ⁸ and to reduce participant, study partner, and staff burden, the WRAP protocol was adapted in late 2015 to include the QDRS, which is now included in the study partner packet of questionnaires for all WRAP participants. Per WRAP study protocol, the CDR is administered by trained and certified WRAP staff after the study visit if the QDRS global score described was greater than zero. Additionally, to partially blind CDR interviewers to cognitive status, a random sample of participants with QDRS‐global = 0 was selected for a follow‐up CDR, completed in a similar post‐visit timeframe.

In this study, five QDRS‐related scores were used for analysis: (1) QDRS total score, the sum of all 10 QDRS items (possible range: 0–30); (2) Cognitive subdomain score (QDRS‐Cog), the sum of items 1, 2, 3, and 8; (3) Behavioral subdomain score (QDRS‐Beh), the sum of items 4, 5, 6, 7, 9, and 10 ⁴ ; (4) QDRS‐SB, the sum of the first six QDRS items, and is used to obtain a QDRS‐summary score that has been shown to approximate the CDR sum of boxes (SB) score ⁸ ; and (5) Harmonized SB, which is similar to QDRS‐SB but replaces the QDRS‐SB with CDR‐SB for participants who have a CDR‐SB available for the corresponding visit ⁸ , ⁹ .

RESEARCH IN CONTEXT

1. Systematic review: Using primarily Google Scholar, search terms included the Quick Dementia Rating System (QDRS), plasma phosphorylated tau 217 (pTau217), and Preclinical Alzheimer's Cognitive Composite (PACC3), with a particular focus on studies investigating the (longitudinal) associations between the QDRS and pTau217, as well as between pTau217 and PACC3. 

2. Interpretation: Our findings show that, in a predominantly unimpaired sample, higher baseline pTau217 levels were associated with faster functional (QDRS) and cognitive (PACC3) decline. Effect sizes were similar between PACC3 and QDRS. Exploratory analyses showed increased risk of transitioning to impaired QDRS classifications in those with high‐baseline pTau217.

3. Future directions: Future research will explore the application of QDRS and plasma pTau217 cutoffs in more diverse populations to assess generalizability and potential cultural or educational biases. Additionally, expanding the focus to individuals with more advanced cognitive decline could provide insights into the predictive utility of these measures across different stages of AD progression.

2.2.2. Comparison cognitive outcome: PACC3 cognitive composite

To understand QDRS outcomes alongside other WRAP measures and enable comparisons of relative effect size, we selected the WRAP cognitive composite with the least intra‐individual variability ²³ and associated with preclinical decline ²³ , ²⁴ . This composite, the three‐test Preclinical Alzheimer's Cognitive Composite (PACC3 ²³ ), includes the following neuropsychological tests: the Rey Auditory Verbal Learning Test (RAVLT; Trials 1‐5 ²⁵ ), Logical Memory II ²⁶ , and the Digit Symbol Substitution Test ²⁷ . Each measure contributing to PACC3 was individually z‐scored using the mean and standard deviation within a baseline unimpaired subset, and the average of these z‐scores was used to derive the composite PACC3 score, which serves as the basis for our analysis ²³ .

2.2.3. Plasma biomarker: pTau217

At each WRAP visit, blood samples were collected whenever feasible, processed as previously described ¹⁷ , and stored at −80°C. Plasma samples preserved in ethylenediaminetetraacetic acid (EDTA) were later analyzed at the Wisconsin ADRC Biomarker Core to measure pTau217 concentration using the ALZpath Single molecule array (Simoa) assay (ALZpath, Carlsbad, CA) on a Quanterix HD‐X instrument (Quanterix, Billerica, MA). The assay itself was described previously by Ashton et al. ¹⁹ . In addition, Ashton et al. ¹⁹ introduced a three‐range approach ²⁸ based on receiver operating characteristics curve (ROC) analyses relative to amyloid PET positivity that establishes lower (< 0.4 pg/mL, representing presumed amyloid negative) and upper (> 0.63 pg/mL, indicating presumed amyloid positive) reference points for pTau217 in the WRAP cohort. Since the sample they used to establish the cut‐points was drawn from WRAP, and since the assay manufacturer is maintaining stability across kit lots to this study ¹⁹ , we have adopted these reference points to categorize baseline pTau217 levels into a three‐group variable (low, intermediate, and high likelihood of amyloid PET positivity) for secondary analysis.

2.3. Statistical methods

All statistical analyses were conducted in R (Version 2024.04.1) ²⁹ . Specifically, to conduct the linear mixed‐effect analyses, we used the nlme R package ³⁰ . For the analyses using a generalized linear mixed‐effect model, we applied the glmmTMB R package ³¹ . Sample characteristics were summarized, and relevant comparisons across the three pTau217 categories on the sample characteristics were made.

2.3.1. Primary analysis

We used linear mixed‐effects models to examine whether baseline pTau217 level modified corresponding longitudinal QDRS‐related scores and PACC3 scores (random intercepts, where appropriate based on model fit, random slopes for age; and unstructured covariance). For each outcome variable (i.e., QDRS total score, QDRS‐Cog, QDRS‐Beh, QDRS‐SB, Harmonized SB, PACC3), the fit of the base model (including covariates of sex, age at visit [centered at 65, to account for variation in follow‐up timing], years of education, study partner's relationship to the participants for QDRS outcomes, and practice effect for PACC3) was compared with models that sequentially added the following variables: (1) baseline pTau217, (2) age², (3) interaction of baseline pTau217 with age, and (4) interaction of baseline pTau217 with age². After comparing model fit indices (e.g., Akaike Information Criterion [AIC] and Bayesian Information Criterion [BIC]), we selected the best model based on statistical fit and its appropriateness for the outcomes.

2.3.2. Secondary analyses

While the primary analysis focused on understanding the relationship between pTau217 and each outcome, secondary analyses extended the investigation in two directions. The first direction utilized the above‐mentioned cut‐points identified by Ashton et al. ¹⁹ and explored whether a categorical baseline pTau217 (3 categories reflecting low [if $<0.40$ pg/mL], intermediate [if $0.40<pTau217<0.63$ pg/mL], and high [if $>0.63$ pg/mL] likelihood of being amyloid PET positivity) interacted with age and which categories of baseline pTau217 were most strongly associated with the score changes for any of the outcomes. To achieve this, we applied the same approach as in the primary analysis, replacing continuous pTau217 with categorical pTau217; we hypothesized the group with high likelihood of amyloid PET positivity would have the worst trajectories. To illustrate this, we estimated marginal means for each group at different ages (i.e., 65, 70, 75, and 80) and performed paired comparisons between these groups (using Tukey's adjustment) to report the mean difference of the marginal effects at these ages. In addition, to facilitate comparisons of effect sizes across outcomes, we depicted forest plots of these standardized mean differences (∆M) and corresponding 95% confidence interval (CI) at the noted ages by plasma category.

The second extension was considered because of the positive skew and potential zero‐inflation observed in the QDRS scores. Specifically, since the sample was predominantly unimpaired over time, it was reasonable to assume the sample was a mixture of cognitively impaired individuals (QDRS‐related scores > 0) and unimpaired individuals (= 0), each with different relationships to p‐Tau217. To account for these unique distributional properties of the data, we applied generalized linear mixed‐effects models assuming Poisson or negative binomial distributions, with and without zero‐inflation.

In addition to the above‐noted analyses, we conducted exploratory analyses to determine whether increasing baseline pTau217 categories were also associated with an increased likelihood of QDRS‐defined mild cognitive impairment (MCI) at baseline or a greater risk of progression to MCI among participants who were cognitively unimpaired at baseline. Further details about the analysis are presented in Appendix A.

3. RESULTS

Table 1 displays the descriptive statistics for the sample. Among the 639 participants, 439 had a low, 126 intermediate, and 74 high pTau217 levels. Of these participants, 25 were classified as Impaired_QDRS at baseline, while 37 progressed to Impaired_QDRS at later visits. The sample was predominantly non‐demented, includes more females than males, is primarily non‐Hispanic white individuals, and is generally well‐educated.

TABLE 1.

Descriptive and frequency statistics of participant sample overall and by baseline pTau217 group.

Parameter	Overall N = 639	Low* baseline pTau217 N(%) = 439 (68.7)	Intermediate* baseline pTau217 N(%) = 126(19.7)	High* baseline pTau217 N(%) = 74(11.6)	p‐value
Age at baseline, mean (SD)	64.9(6.6)	64.1(6.5)	65.9(6.7)	68.2(5.8)	<0.001
Years of QDRS follow‐up, mean (SD)	4.1(1.8)	4.2 (1.8)	4.0 (1.7)	3.6 (1.7)	0.029
Female, n (%)	426(66.7)	297(67.7)	75(59.5)	54(73.0)	0.110
non‐Hispanic White, n (%)	629(98.4)	429(97.7)	126(100)	74(100)	0.114
Education level $\ge$ BA, n (%)	405(63.4)	270(61.5)	83(65.9)	52(70.3)	0.232
Baseline pTau217 concentration, median [Q1, Q3]	0.30 [0.23, 0.44]	0.25 [0.20, 0.31]	0.48 [0.44, 0.55]	0.84 [0.71, 1.01]	<0.001
Baseline QDRS total score, median [Q1, Q3]	0 [0.0, 0.5]	0 [0.0, 0.5]	0 [0.0, 0.5]	0 [0.0, 0.875]	0.063
Baseline ImpairedQDRS (QDRS total score > 2.5), n (%)	25(3.9%)	11(2.5%)	7(5.5%)	7(9.5%)	0.009
Ever ImpairedQDRS, n (%)	62(9.7%)	33(7.5%)	12(9.5%)	17(23.0%)	< 0.001
QDRS Progressors n (%*)	37(5.8%)	22(5.0%)	5(4.2%)	10(13.5%)	0.012

Note: Low*, intermediate*, and high* baseline pTau217 refer to the three groups (low, intermediate, and high likelihood of amyloid PET positivity) based on the cut‐points for pTau217 established in the WRAP cohort ²⁰ .

QDRS Progressors includes participants whose baseline QDRS total score was $\le$2.5 but progressed to >2.5 in later visits, and the corresponding sample size of QDRS Progressors is N* = 614. %* = n/N*. To derive corresponding p‐values, we used Kruskal‐Wallis tests or chi‐squared tests (or Fisher's exact test when cell sizes were small).

Abbreviations: BA, Bachelor of Arts; QDRS, Quick Dementia Rating System.

3.1. Primary analyses

We tested models with both random intercepts and random slopes for age at visit across all outcome variables. This structure performed well for some outcomes (e.g., QDRS total score, QDRS‐Beh, and PACC3), but for others (e.g., QDRS‐SB, Harmonized SB, and QDRS‐Cog), models with random slopes either failed to converge or showed poorer fit. For these outcomes, we retained random intercept‐only models to ensure stability and interpretability. Model comparisons using AIC and likelihood ratio tests (Appendix B) consistently supported inclusion of the interaction between age and baseline pTau217, as well as the quadratic age term. As shown in Table 2, the non‐linear effects of age on the six outcomes indicated that cognitive decline accelerated with increasing age. Furthermore, the significant interaction between age and baseline pTau217 across QDRS‐related scores and PACC3 indicated that higher baseline pTau217 amplified the impact of aging on dementia‐related trajectories. Note that directionality of “worse” of the PACC3 score was opposite to that of the QDRS‐related scores (i.e., higher PACC3 indicates less cognitive impairment, whereas lower QDRS‐related scores indicate less cognitive impairment), so the signs of the coefficients were reversed.

TABLE 2.

Model output for the linear mixed effects best fitting model for each outcome based on continuous baseline pTau217.

Parameter	QDRS total score	QDRS‐Cog	QDRS‐Beh	QDRS‐SB	Harmonized SB	PACC3
Age	−3.76e‐3	−3.46e‐4	−5.43e‐3	−1.45e‐4	−4.94e‐3	−5.18e‐2***
Age2	1.23e‐3 *	3.89e‐4$·$	7.20e‐4$·$	6.30e‐4*	4.46e‐4$·$	−2.16e‐3***
Baseline pTau217	0.278	0.137$·$	0.148	0.199	0.205*	−0.614***
Age $\times$ pTau217	0.066**	0.030**	0.038*	0.039**	0.042***	−0.069***
Sex (male)	0.040	−0.012	0.042	−0.011	0.019	−0.603***
Education level $\ge$ BA vs. $<$ BA	−0.039	−0.017	−0.021	−0.030	−0.053	0.538***
Relationship
Child vs. spouse	0.103	0.061	0.081	0.097	0.112
Sibling vs. spouse	−0.095	−0.097	−0.004	−0.100	−0.087
Others vs. spouse	−0.248	−0.121	−0.154	−0.148	−0.075
# of practice						0.075***

Note: p < 0.10$·$, p < 0.05 *, p < 0.01 **, p < 0.001 ***; interaction denotes the interaction between baseline pTau217 and age; age and age² centered at 65. Relationship denotes the informant's relationship to the participant and was included in QDRS‐related outcome models but not in the PACC3 model, as PACC3 was not rated by an informant. # of practice being included in the PACC3 models, as PACC3 is subject to practice effect.

Abbreviations: BA, Bachelor of Arts; PACC3, three‐test Preclinical Alzheimer's Cognitive Composite; QDRS, Quick Dementia Rating System.

3.2. Secondary analyses

3.2.1. Secondary analysis 1

Table 3 presents the results of linear mixed‐effects models examining the relationship between categorical baseline pTau217 levels and various QDRS‐related outcomes, as well as PACC3. The results largely match those obtained using continuous baseline pTau217, supporting the findings. However, the categorical approach helps characterize effect sizes between different pTau217 groups (low, intermediate, and high likelihood of baseline amyloid PET positivity). To visualize the longitudinal changes in QDRS in relation to pTau217, we presented simple slopes for each pTau217 group in Figure 1, color‐coded by baseline pTau217 group; panels A‐F correspond to the outcomes represented in the columns of Table 2, as well as a spaghetti plot shaded by baseline pTau217 group is in Appendix C.

TABLE 3.

Model output for the linear mixed effects models based on categorical baseline pTau217.

Parameter	QDRS total score	QDRS‐Cog	QDRS‐Beh	QDRS‐SB	Harmonized SB	PACC3
Age	1.49e‐2*	8.11e‐3**	5.78e‐3	9.74e‐3**	5.74e‐3$·$	−0.071***
Age2	1.35e‐3*	4.13e‐4$·$	7.93e‐4*	6.56e‐4*	4.61e‐4	−2.25e‐3***
Baseline pTau217
Intermediate vs. low	0.158	0.045	0.111	0.083	0.089	−0.024
High vs. low	0.169	0.071	0.081	0.081	0.089	−0.518***
Age $\times$ pTau217:
Intermediate vs. low	−3.50e‐3	8.99e‐4	−5.97e‐3	1.29e‐3	8.49e‐4	1.85e‐3
High vs. low	6.19e‐2**	2.91e‐2***	3.97e‐2**	4.32e‐2***	4.71e‐2***	−4.37e‐2**
Sex (male)	0.030	−0.017	0.041	−0.013	0.017	−0.604***
Education level $\ge$ BA vs. $<$ BA	−0.045	−0.022	−0.022	−0.032	−0.057	0.527***
Relationship
Child vs. spouse	0.104	0.063	0.084	0.102	0.118
Sibling vs. spouse	−0.070	−0.089	0.018	−0.082	−0.067
Others vs. spouse	−0.236	−0.114	−0.143	−0.132	−0.056
# of practice						0.081***

Note: p < 0.10$·$, p < 0.05 *, p < 0.01 **, p < 0.001 ***; interaction denotes the interaction between baseline pTau217 and age; age and age² centered at 65. Correction for multiple comparisons were applied using the FDR method. Relationship denotes the informant's relationship to the participant and was included in QDRS‐related outcome models but not in the PACC3 model, as PACC3 was not rated by an informant.

Abbreviations: BA, Bachelor of Arts; FDR, false discovery rate; PACC3, three‐test Preclinical Alzheimer's Cognitive Composite; QDRS, Quick Dementia Rating System.

FIGURE 1.

Spaghetti plots for the observed QDRS related scores and PACC3, with estimated slopes at the midpoints of each pTau217 group. PACC3, three‐test Preclinical Alzheimer's Cognitive Composite; QDRS, Quick Dementia Rating System.

In addition, Figure 2 provides further characterization of these findings, depicting forest plots of standardized mean differences for paired comparisons among the three pTau217 groups for each outcome at ages 65, 70, 75, and 80. While almost all the confidence intervals for the paired comparisons of low vs intermediate groups overlap zero, contrasts between the high pTau217 group consistently diverge from the low and intermediate groups. For example, at age 65, the high group scores less than a quarter point higher on average than the low group; by age 80, however, that difference is nearing a full point. The largest standardized mean differences were observed at older ages and in Harmonized SB, followed by QDRS‐SB, indicating that these measures may be sensitive to differences between high likelihood of amyloid PET positivity individuals and those in the lower likelihood groups at older ages. The patterns were consistent across QDRS‐related outcomes, as shown in the forest plots, and the patterns observed for PACC3 are generally aligned with these: all show increasing differences by baseline pTau217 with age, though PACC3 shows earlier divergence — between high and lower likelihood groups by a younger age. This consistency highlights the potential utility of the QDRS‐related scores as complementary measures for tracking AD‐related changes, given their ease of administration.

FIGURE 2.

Paired comparisons between baseline pTau217 risk groups at ages 65, 70, 75, and 80 on estimated QDRS related scores and PACC3. (A) QDRS total score (B) QDRS‐Cog (C) QDRS‐Beh (D) QDRS‐SB (E) Harmonized SB (F) PACC3. Note: The dots and error bars represent the standardized mean differences between the three pairs and corresponding 95% CIs. CI, confidence interval; PACC3, three‐test Preclinical Alzheimer's Cognitive Composite; QDRS, Quick Dementia Rating System.

3.2.2. Secondary analyses 2

Table D1 (Appendix D) presents the results of generalized linear mixed‐effects models (GLMMs) examining the relationship between baseline pTau217 levels and QDRS‐related outcomes. While the interaction between age and pTau217 remains a significant predictor across most outcomes, these analyses differ from the primary models by showing that the quadratic age term (age²) is no longer significant. This suggests that, when a modeling strategy that is targeted toward skewed and coarse outcomes is used to model QDRS, the need for curvature to account for frequent zeroes at younger ages disappears. Negative binomial models continue to provide a better fit than Poisson models because of overdispersion in the data, consistent with earlier findings (Tables 2 and 3). Zero‐inflated models improve fit only for QDRS‐Beh, indicating that behavioral symptoms may emerge more slowly. Despite these differences, the overall conclusions from the primary analysis remain robust, highlighting the modifying role of pTau217 across QDRS outcomes, with those in a range indicating increased likelihood of amyloid PET positivity having the greatest worsening in outcomes.

4. DISCUSSION

This study contributes to the growing literature addressing relationships between longitudinal study partner evaluations of participant functioning and AD‐related biomarkers. To our knowledge, this study is among the first to characterize the trajectory of QDRS scores over time with respect to baseline plasma pTau217 levels. Our primary analyses demonstrated that QDRS scores reflected AD‐related changes over time, as evidenced by a significant interaction between baseline pTau217 levels and age across all the QDRS‐related outcomes. Furthermore, the significance of the quadratic age term for QDRS total score, QDRS‐SB, and QDRS‐Beh indicated the acceleration of cognitive decline with growing age. Such patterns of decline emphasized the importance of early biomarker‐based screening, potentially enabling interventions for at‐risk groups before cognitive impairment progresses.

The secondary analyses corroborated and extended these findings. Using categorical baseline pTau217 levels defined by published cutoffs ¹⁹ , we found that individuals in the high likelihood of amyloid PET positivity group exhibited steeper declines in QDRS‐related outcomes over time compared to those in the low and intermediate likelihood groups. This supports pTau217 as a likelihood stratification tool for AD. Notably, the similar effect sizes observed across QDRS‐related scores and the PACC3 composite (Figure 2)—the latter being a well‐validated measure in AD research with low error variance ²³ —suggest that the QDRS may serve as an effective alternative or complement to cognitive composites in future studies. Given the QDRS's brevity and ease of administration, it holds potential for broader clinical and research applications in resource‐limited settings.

The application of generalized linear mixed‐effects models in secondary analyses further validated the robustness of the primary findings. By addressing overdispersion and zero inflation in the predominantly unimpaired sample, these models confirmed the interaction between baseline pTau217 and age. Interestingly, the zero‐inflation component was only significant for QDRS‐Beh scores, suggesting that behavioral changes emerge later than cognitive declines in the progression of AD. Exploratory analyses, as shown in Appendix A, added further depth, revealing that individuals with high baseline pTau217 levels were more likely to develop from unimpaired to impaired QDRS scores. This finding highlighted pTau217's value for identifying those at heightened likelihood of future AD‐related cognitive impairment, offering potential for early intervention studies.

4.1. Strengths and limitations

This study examined one of the largest longitudinal datasets of QDRS assessments relative to AD biomarkers, including plasma pTau217, currently a high‐interest biomarker. Although the pTau217 cutoffs used were derived from assays performed in Gothenburg, Sweden, and may not fully generalize to those obtained locally, they were obtained using data from the same cohort and assay. Our data indicated clearly that the group with pTau217 in a range associated with high probability of amyloid PET abnormalities was at increased risk of worse cognitive trajectories, via both objective cognitive assessment (PACC3) and study partner ratings (QDRS), increasing our confidence in the application of these cutoff values.

One potential limitation is that approximately 27% of our participants had a change in study partner over the study period, and the extent to which contact frequency was consistent across different study partners is unknown. Future analyses should consider modeling these changes when prevalent, and future data collection efforts would benefit from recording whether the study partner cohabits with participant and, if not, the frequency and nature of their contact. Another limitation is that, while focusing on a predementia sample allowed deeper understanding of how the QDRS functions in the earliest stages of decline, these results may have limited generalizability to individuals already experiencing more advanced cognitive decline. Furthermore, the sample was predominantly non‐Hispanic White individuals and highly educated, which may further restrict generalizability of the results to more diverse populations. In addition, covariates such as health comorbidities, which may affect either cognitive functioning or the performance of the pTau217 assay, should also be considered. We explored the impact of one such factor (i.e., LIBRA) in Appendix E; however, future studies should consider incorporating a broader range of health‐related variables.

5. CONCLUSIONS

This study characterized the longitudinal trajectories of QDRS scores and highlighted their relationship to plasma pTau217 levels, with both showing potential for monitoring AD‐related cognitive and functional decline. The findings showed that QDRS scores, alongside plasma pTau217, are valuable tools for identifying individuals at heightened risk during the preclinical phase of AD. They could enhance risk stratification and monitoring in both clinical and research contexts, complementing current AD screening methodologies and laying the foundation for earlier interventions.

CONFLICT OF INTEREST STATEMENT

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, Merck Sharp and Dohme, Merry Life, Nervgen, Novo Nordisk, Optoceutics, Passage Bio, Pinteon Therapeutics, Prothena, Quanterix, Red Abbey Labs, reMYND, Roche, Samumed, ScandiBio Therapeutics AB, Siemens Healthineers, Triplet Therapeutics, and Wave, has given lectures sponsored by Alzecure, BioArctic, Biogen, Cellectricon, Fujirebio, LabCorp, Lilly, Novo Nordisk, Oy Medix Biochemica AB, Roche, and WebMD, is a co‐founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program, and is a shareholder of MicThera (outside submitted work). Sterling C. Johnson serves as a consultant to Eli Lilly, Alzpath, Enigma Biomedical, and Merck. The remaining authors have nothing to disclose. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All human subjects have provided informed consent.

Supporting information

Supporting Information

Supporting Information

Huang Q, Jonaitis EM, Studer RL, et al. Preclinical dementia rating scores are associated with plasma phosphorylated tau‐217. Alzheimer's Dement. 2025;17:e70179. 10.1002/dad2.70179