Clinical progression on CDR‐SB©: Progression‐free time at each 0.5 unit level in dominantly inherited and sporadic Alzheimer's disease populations

Abstract

INTRODUCTION

Clinical Dementia Rating Sum of Boxes (CDR‐SB) is a reliable and clinically meaningful composite for assessing treatment effects in Alzheimer's disease (AD) clinical trials. Small CDR‐SB differences at the end of a trial often lead to controversy in deriving clinically meaningful interpretations.

METHODS

We estimated progression‐free time (PFT) participants remained at each 0.5 unit CDR‐SB increment in dominantly inherited AD (DIAD) and sporadic AD populations, evaluating its potential as an alternative measure of treatment effects.

RESULTS

PFT is longer at CDR‐SB ≤ 2.0 (1–2 years) and shorter at CDR‐SB ≥ 5 (≤ 0.33) in the Alzheimer's Disease Neuroimaging Initiative cohort. The DIAD cohort showed similar but shorter times. Using PFT, continuous lecanemab treatment for 3 years is estimated to delay disease progression by 0.62 years in the sporadic population.

DISCUSSION

PFT provides a benchmark for expressing clinical progression and treatment effects and can be applied particularly during open‐label extensions and single‐arm trials without placebo comparisons.

Highlights

• We estimated the progression‐free time at each 0.5 unit Clinical Dementia Rating Sum of Boxes increment in dominantly inherited Alzheimer's disease (AD) and sporadic AD populations.

• We proposed using progression‐free time to estimate treatment effects in open‐label extension or single‐arm studies.

• If further validated, progression‐free time could serve as a benchmark for assessing clinical progression and treatment effects.

1. BACKGROUND

The Clinical Dementia Rating Scale Sum of Boxes (CDR‐SB) is widely used as a primary or secondary endpoint in phase 2 and 3 clinical trials for Alzheimer's disease (AD). ¹ , ² , ³ , ⁴ , ⁵ It serves as a US Food and Drug Administration (FDA)‐approved composite measure for assessing mild cognitive impairment (MCI) and mild dementia, simplifying trials that might otherwise require multiple or co‐primary endpoints. ⁶

In the CDR‐SB assessment, a rater uses semi‐structured interviews of the participant and an informant to assess the former's performance across six domains: three cognitive (memory, orientation, judgment/problem solving) and three functional (community affairs, home/hobbies, and personal care). Each domain is scored on an ordinal scale (0, 0.5, 1, 2, 3), indicating “no dementia,” “uncertain,” “mild,” “moderate,” and “severe” dementia, respectively, with the exception of personal care, which omits the 0.5 score. The scores across these domains are summed to yield the CDR‐SB score, which has a range of 0 to 18 with 0.5 point intervals at lower summary scores (i.e., better function).

CDR‐SB scores and the differences in change from baseline between treatment and placebo groups are prominently presented in the clinical studies sections of the prescribing information (i.e., “the label”) for the FDA approved amyloid‐targeting antibodies, aducanumab, ¹ lecanemab, ³ and donanemab, ⁴ providing patients and prescribers with insights into treatment outcomes. Although treatment–placebo differences of the CDR‐SB are statistically significant for these antibodies and supported FDA's approval of these antibodies for marketing, the actual mean effect sizes are modest, ranging from +0.03, ¹ to –0.39, ¹ –0.45, ³ and –0.70, ⁴ sparking considerable debate regarding their clinical interpretability and significance. Understanding how these numerical differences might translate into meaningful clinical outcomes is essential for clinicians, patients, and researchers alike. An Alzheimer's Association workgroup suggested that expectations of outcomes from therapeutic interventions in AD may need to be modified. ⁷

One approach is to express differences in CDR‐SB changes as a percentage reduction compared to the placebo comparator, for example, x% slowing of progression. ² , ³ , ⁴ Another is to interpret these differences as a delay in disease progression, either relative to the decline observed in the placebo group (e.g., 5.3 months for lecanemab) ⁴ , ⁸ or relative to reaching a specific disease stage milestone. ⁹ However, these methods are relative measures, vary with the change in placebo, and may not be applicable during the open‐label extension (OLE) period when all participants receive treatment. Furthermore, these approaches are trial specific and may lack comparability across trials due to differences in baseline CDR‐SB levels and change in the placebo control groups.

In this study, we introduce an alternative methodology that standardizes the evaluation of delays in disease progression based on the CDR scale. We accomplish this by calculating the progression‐free time (PFT) (or “residence time”) a research participant remains at a given CDR‐SB level before progressing to a higher level. To enable a comprehensive and comparative evaluation of disease progression in both rare and sporadic forms of AD, we propose estimating the PFT at each 0.5 unit CDR‐SB increment for two distinct populations: participants with dominantly inherited AD (DIAD) from the Dominantly Inherited Alzheimer Network (DIAN) observational study and those with sporadic AD from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. We suggest using PFT, or time to 0.5 unit progression, as a potential measure of clinically meaningful change.

2. METHODS

2.1. Study participants

DIAN is an international, multicenter observational study focused on individuals aged ≥ 18 years who have parents with a known causative mutation in the amyloid precursor protein (APP), presenilin 1 (PSEN1), or presenilin 2 (PSEN2) genes, for AD. Mutation carriers (MCs) face an almost certain risk of developing AD, with relatively predictable symptom onset ages. At enrollment into the DIAN cohort, these participants were either cognitively unimpaired (Clinical Dementia Rating [CDR] = 0) or showed signs of early symptomatic disease (CDR = 0.5 or 1, signifying very mild or mild dementia, respectively). Research participants with the Dutch mutation were excluded in this study due to their atypical rates of clinical progression. Detailed information on the DIAN study's structure and the assessments performed can be found in previous publications. ¹⁰ , ¹¹

The Alzheimer's Disease Neuroimaging Initiative (ADNI) is a longitudinal, multicenter study designed to track the progression of AD using various biomarkers in individuals aged 55 to 90 years. ADNI encompasses participants across the spectrum of AD pathology, categorizing them into three groups at enrollment into ADNI: normal cognition, mild cognitive impairment (MCI), and AD dementia. Data included in this study were download from adni.loni.ucla.edu. 

TABLE 1.

Baseline characteristics for each study.

Characteristic	ADNI (N = 406)	DIAN (N = 384)
CDR = 0, N (%)	210 (51.7%)	242 (63.0%)
CDR > 0, N (%)	196 (48.3%)	142 (37.0%)
Age, years (SD)	72.9 ± 7.3	38.2 ± 7.3
MMSE	28.5 ± 1.9	26.5 ± 5.5
APOE ε4 negative, N (%)	267 (65.8%)	271 (70.6%)
CDR‐SB, (SD)	0.6 ± 1.0	1.5 ± 3.1
Follow‐up duration, years (SD)	5.4 (1.8)	4.1 (2.6)

Abbreviations: ADNI, Alzheimer's Disease Neuroimaging Initiative; APOE, apolipoprotein E; CDR, Clinical Dementia Rating; DIAN, Dominantly Inherited Alzheimer Network; MMSE, Mini‐Mental State Examination; SB, Sum of Boxes, SD, standard deviation.

2.2. Statistical analysis

To map the holistic trajectory of disease progression from CDR‐SB score of 0 (the lower bound) to a CDR‐SB score of 18 (the upper bound), we applied a parametric disease progression model incorporating random effects for both the endpoint and the time variable (see Supplemental Material in supporting information for details). In alignment with previous studies, ¹² , ¹³ , ¹⁴ integrating a random effect into the time variable enables the model to accurately recover the underlying disease progression pathway by adjusting the time variable: shifting it leftward for less severe stages and rightward for more severe stages of the disease. For mutation carriers in the DIAN study, the estimated years from symptom onset (EYO) serves as a relatively precise indicator of their disease stage; thus, EYO was adopted as the time variable in our model. ¹² Conversely, for participants in the ADNI study, lacking such an accurate measure, the time since enrollment in years was used as the time variable. ¹³ , ¹⁴ After the estimation of the disease progression trajectory, the PFT—the duration spent at each 0.5 unit interval of CDR‐SB progression—was calculated based on the rate of disease progression observed in each interval.

To demonstrate the potential use of PFT, we assessed the treatment effect of lecanemab during the OLE period as a time delay in disease progression. ¹⁵ This was followed by a comparison to the delay in disease progression as estimated by conventional methods, such as calculating the time based on the percentage reduction in disease progression compared to placebo, multiplied by the total duration. ⁸

RESEARCH IN CONTEXT

1. Systematic review: The authors performed an extensive literature review, using traditional sources like PubMed, alongside abstracts and presentations from scientific meetings. Through this thorough search, they successfully identified and appropriately cited relevant publications pertaining to the progression‐free time for each 0.5 unit increment in Clinical Dementia Rating Sum of Boxes (CDR‐SB).

2. Interpretation: In clinical trials for Alzheimer's disease (AD), CDR‐SB is a reliable and clinically meaningful composite for assessing treatment effects. Small CDR‐SB differences at the end of a trial often lead to controversy in deriving clinically meaningful interpretations. To enhance the interpretation, we estimated the progression‐free time research participants remained at each 0.5 unit CDR‐SB increment in dominantly inherited AD (DIAD) and sporadic AD populations, evaluating its potential as an alternative measure of treatment effects. Progression‐free time is longer at CDR‐SB ≤ 2.0 (1–2 years) and shorter at CDR‐SB ≥ 5 (≤ 0.33) in the Alzheimer's Disease Neuroimaging Initiative cohort. The DIAD cohort showed similar but shorter times. Using progression‐free time, continuous lecanemab treatment for 3 years is estimated to delay disease progression by 0.62 years in the sporadic population.

3. Future directions: Progression‐free time provides a benchmark for expressing clinical progression and treatment effects and can be applied particularly during open‐label extensions and single‐arm trials without placebo comparisons. Future studies should explore alternative models or use different study cohorts to confirm the consistency of our findings.

All statistical analyses were conducted using SAS software, version 9.4. We used two‐sided t tests with a Type I error rate of 5% and reported the corresponding 95% confidence intervals (CIs).

3. RESULTS

3.1. PFT for DIAD and sporadic AD populations

In the DIAN study, 384 mutation carriers were included from data freeze 17 (without Dutch mutations), while in the ADNI study, 406 participants were involved. The baseline characteristics of these participants are detailed in Table 1.

Figure 1 illustrates the PFT at each 0.5 unit increment of the CDR‐SB score, up to 12.5, for the DIAD population. The PFT at CDR‐SB 0.5 is ≈ 1.65 years, which sharply declines to < 0.5 years once the CDR‐SB score reaches ≥ 2.5. Table S1 in supporting information presents the PFT for all intervals at 0.5 unit increments of the CDR‐SB score.

FIGURE 1.

Estimated progression‐free time and its 95% confidence interval for each 0.5 unit Clinical Dementia Rating Sum of Boxes (CDR‐SB) interval based on the Dominantly Inherited Alzheimer Network study.

Figure 2 presents similar data for the sporadic AD population, showing PFTs for each 0.5 unit increment of the CDR‐SB score, also up to 12.5. Here, the PFT from CDR‐SB 0.5 to CDR‐SB 1 is roughly 2.31 years, which then rapidly decreases to < 0.5 years when the CDR‐SB score surpasses 3.5. The PFT for all intervals at 0.5 unit increments of the CDR‐SB score are presented in Table S1. Recognizing that age is a significant factor in sporadic AD, though less so in DIAD, we also present PFTs in the ADNI cohort across 5 year age groups from 55 to 80 years in Table S2 in supporting information. As age increases, PFT decreases slightly. The PFT for the sporadic AD population was also estimated using National Alzheimer's Coordinating Center (NACC) data and is presented in Table S3 in supporting information. The PFTs across 0.5 point CDR‐SB increments are highly comparable between ADNI and NACC, with differences generally ≈ ≤ 0.1 years.

FIGURE 2.

Estimated progression‐free time and its 95% confidence interval for each 0.5 unit Clinical Dementia Rating Sum of Boxes (CDR‐SB) interval based on the Alzheimer's Disease Neuroimaging Initiative study at cohort mean age 72.9.

3.2. Application of PFT to assess treatment effects

According to reports from the OLE period of the phase 3 lecanemab trial, participants in the early‐start group, who began treatment from the double‐blind baseline, showed a decline of 3.09 units in the CDR‐SB score over 3 years, moving from a baseline of 3.17 to 6.26. ³ , ¹⁵ Because participants in the original placebo group transitioned to lecanemab treatment during the OLE (becoming the delayed‐start group), it is not feasible to directly assess the treatment's effect in the early‐start group by comparing it to the delayed‐start group during this period. Instead, we could evaluate the treatment effect using the estimated PFT (Figure 2) for the sporadic AD population.

Our estimation suggests that without lecanemab treatment, participants might have progressed from a CDR‐SB score of 3.17 to 6.26 in just 2.38 years. However, with continuous lecanemab treatment, this progression was extended to 3 years, effectively a time difference or delay of progression of 0.62 years.

As an alternative approach to assess the treatment effect, a cohort from the ADNI study, matched as external controls, was used to benchmark disease progression from the double‐blind baseline to the 36 month mark, including 18 months within the OLE period. ¹⁵ It was observed that the matched ADNI controls experienced a decline of 4.14 units in the CDR‐SB score, which was 0.95 units greater than the decline in the early‐start group receiving lecanemab. ¹⁵ To quantify the delay in disease progression using the percentage reduction method, ⁸ we first calculated the percentage reduction in disease progression relative to the ADNI group. We then multiplied this value by the total study duration in years. This calculation yielded a delay of 0.7 years, derived from the formula $(0.95/4.14)\times (36/12)$, ⁸ which aligns well with the findings obtained using the PFT approach.

4. DISCUSSION

CDR‐SB has been demonstrated as a reliable composite measure for assessing treatment effects in AD clinical trials and its use in AD drug development is ongoing. ¹ , ³ , ⁴ Consequently, there is a great interest among clinicians, patients, and researchers to derive meaningful interpretations from the treatment effects measured by the change over time and treatment–control differences in CDR‐SB in trials. ⁷ Various approaches have been suggested to quantify these effects, including: (1) expressing the difference in CDR‐SB scores between treatment and control groups as mean differences; (2) expressing the difference in terms of percentage reduction compared to controls; ³ , ⁴ (3) translating these differences into time savings relative to the decline observed in the control or placebo group; ⁴ , ⁸ , ¹⁶ (4) calibrating the treatment effect to reflect delays in reaching critical milestones, such as the loss of independence in activities of daily living. ⁹ These methodologies have significantly enhanced our ability to interpret the outcomes of recent amyloid‐targeting antibody trials. ³ , ⁴ Building on these, our study introduces an alternative approach by calculating the PFT for each 0.5 unit increment in the CDR‐SB score for both DIAD and sporadic AD populations.

Our findings indicate that the DIAD population progresses faster than the sporadic AD population, consistent with previous reports based on cognitive endpoints. ¹⁷ Applying this PFT approach to quantify treatment effects in the Clarity AD trial during its 18 month OLE period yielded an estimated 0.62 year delay in disease progression, aligning closely with results from the existing method proposed by Wang et al. ⁸ This suggests that PFT, or “residence time,” at a given 0.5‐unit increment in CDR‐SB can provide a standardized measure of treatment effects.

The concept also offers clinical relevance: during early AD, patients with lower CDR‐SB scores (e.g., 0.5–1.0) may remain stable over 1 to 2 years, making even a 0.5 point decrease clinically significant. Conversely, at higher CDR‐SB scores (e.g., ≥ 2.5), PFT shortens to ≈ 0.5 to 0.6 years, when a 1.0 point reduction may represent a minimal clinically important difference. By quantifying the time patients remain at specific disease stages, PFT provides a useful tool for clinical interpretation of treatment effects over extended periods, particularly during OLE periods, and in single‐arm trials in which placebo comparisons are unavailable.

Our study represents an initial attempt to quantify PFT for each CDR‐SB 0.5 unit; however, several limitations should be considered. First, our estimates are based on a parametric model, which, although fitting the data reasonably well, requires further validation to assess its robustness. Future studies should explore alternative models such as time to event–based models to confirm the consistency of our findings. Second, our results need validation using different study cohorts to assess variability. Third, we did not report PFT from CDR‐SB 0 to CDR‐SB 0.5 because an individual's CDR‐SB 0 stage can last for many years, and the estimated transition time depends on when participants enter the study. Future research could explore PFT from an earlier benchmark, such as amyloid positivity in cognitively normal individuals, to CDR‐SB 0.5, providing a clearer timeline of disease progression during this stage. Fourth, using PFTs derived from ADNI as a benchmark for assessing treatment effects may introduce bias due to differences in concomitant illnesses, medications, education levels, or other characteristics between observational study participants and those typically enrolled in clinical trials or eligible for anti‐amyloid therapies. Consequently, results based on these PFTs should be interpreted cautiously. Finally, although the confidence intervals for PFTs are relatively narrow, caution is required when applying them at an individual level. The duration an individual has spent at a given CDR‐SB level must be considered when estimating progression. For example, if a patient has been at CDR‐SB 0.5 for 1 year, their expected time to progress to CDR‐SB 1 is ≈ 1.31 years (assuming a PFT of 2.31 years for those newly at CDR‐SB 0.5), rather than the full 2.31 years.

Much like oncology's use of progression‐free survival, a PFT chart could serve as a benchmark for understanding AD progression. This approach would allow clinicians to communicate the rate of cognitive decline in a more accessible and intuitive manner, facilitating clearer discussions with patients and caregivers. By framing disease progression in terms of the time delayed in disease progression by stage of illness, or the potential extension of stability offered by treatment, this method translates abstract concepts into tangible metrics that may be more interpretable by patients, families, and physicians. For patients, families, and clinicians such a chart could simplify the understanding of disease progression and treatment impacts, enhancing engagement and adherence to treatment protocols. More informed discussions about AD treatment effects may be provided by integrating PFT metrics into clinical practice, helping set realistic expectations and plan for future care needs.

CONFLICT OF INTEREST STATEMENT

Guoqiao Wang, PhD, is the biostatistics core co‐leader for the DIAN‐TU. He reports serving on a data safety committee for Amydis Corporate, Abata Therapeutics, and is a statistical consultant for Eisai inc. and Alector Inc. Eric McDade, DO, is the Associate Director of the DIAN‐TU. He reports serving on a data safety committee for Eli Lilly and Company and Alector; serving as scientific consultant for Eisai and Eli Lilly and Company; and institutional grant support from Eli Lilly and Company, F. Hoffmann‐La Roche, Ltd. and Janssen. Chengjie Xiong, PhD, reports serving as a statistical consultant for Diadem and for FDA Advisory Committee on Imaging Medical Products. He also receives support from the following grant: NIHGrantAG067505. Randall J. Bateman, MD, is the Director of the DIAN‐TU and Principal Investigator of the DIAN‐TU‐001. He co‐founded C2N Diagnostics. Washington University and R.J.B. have equity ownership interest in C2N Diagnostics and receive royalty income based on technology (stable isotope labeling kinetics, blood plasma assay and methods of diagnosing Alzheimer's disease with phosphorylation changes) that is licensed by Washington University to C2N Diagnostics. R.J.B. receives income from C2N Diagnostics for serving on the scientific advisory board. R.J.B. has received research funding from Avid Radiopharmaceuticals, Janssen, Roche/Genentech, Eli Lilly, Eisai, Biogen, AbbVie, Bristol Myers Squibb, and Novartis. He receives research support from the National Institute on Aging of the National Institutes of Health, DIAN‐TU Trial Pharmaceutical Partners (Eli Lilly and Company, F. Hoffman‐La Roche, Ltd., and Avid Radiopharmaceuticals), Alzheimer's Association, GHR Foundation, Anonymous Organization, DIAN‐TU Pharma Consortium (Active: Biogen, Eisai, Eli Lilly and Company, Janssen, F. Hoffmann‐La Roche, Ltd./Genentech. Previous: AbbVie, Amgen, AstraZeneca, Forum, Mithridion, Novartis, Pfizer, Sanofi, United Neuroscience). He has been an invited speaker for Novartis and serves on the advisory board for F. Hoffman La Roche, Ltd. Dr. John Morris reports serving as a consultant for Barcelona Brain Research Center BBRC and Native Alzheimer Disease‐Related Resource Center in Minority Aging Research, Ext Adv Board, serving on a data safety committee for Cure Alzheimer's Fund, Research Strategy Council and LEADS Advisory Board, Indiana University.

Lon Schneider, MD, reports within a 36 month time frame, grants from NIH R01 AG054434 (Delivery of Essential Fatty Acids to the Brain in Alzheimer's disease), P30 AG066530 (USC ADRC), R01 AG062687, R01 AG051346, R01 AG055444, P01 AG02350, R01 AG053267 (DIAN‐TU), R01 AG074983 (Aβ and tau vaccines), R01 AG063826, State of California CADC 15‐10291; grants and personal fees from Eli Lilly/Avid and grants and personal fees from Roche/Genentech; grants/contracts from Eisai, Biogen, and Biohaven; grants from Washington University, St. Louis/NIA DIAN‐TU, grants from UC San Diego ADCS; personal fees from Neurim, Ltd (Israel), Cognition Therapeutics, Corium, Immunobrain Checkpoint, Ltd (Israel), Alpha‐cognition, BioVie, Lexeo, Lighthouse, AC Immune (Suisse), Athira, GW Research (UK, Jazz, USA), Merck, Otsuka (USA), Pharmatrophix, Linus Health, Lundbeck, Novo Nordisk, Muna (Denmark), Ono Ltd (Japan), Vivli.org, outside the submitted work; and from The Della Martin Foundation endowment. All the other authors reported no conflicts of interest. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

Informed consent from participants is not necessary for this study/manuscript.

Supporting information

Supporting Information

Supporting Information

Wang G, Li Y, McDade E, et al. Clinical progression on CDR‐SB©: Progression‐free time at each 0.5 unit level in dominantly inherited and sporadic Alzheimer's disease populations. Alzheimer's Dement. 2025;21:e70643. 10.1002/alz.70643