A prospective real‐world study of the efficacy and safety of aducanumab in China: Focus on early‐onset and autosomal dominant Alzheimer's disease

Abstract

INTRODUCTION

Aducanumab is the first anti‐amyloid therapy used for Alzheimer's disease (AD) in China.

METHODS

This 12‐month, single‐center, real‐world study enrolled 12 participants with early AD receiving aducanumab 10 mg/kg every 4 weeks (ChiCTR2200066153). The primary outcomes were changes in Clinical Dementia Rating‐Sum of Boxes (CDR‐SB) scores at 12 months. Secondary outcomes included amyloid clearance, brain structure measures, and biomarker assessments.

RESULTS

No amyloid‐related imaging abnormalities occurred. The mean CDR‐SB score increased by 0.88 at 12 months, with cortical and hippocampal atrophy, enlarged choroid plexus, and ventricular volumes. Two autosomal dominant AD patients exhibited transient amyloid burden elevation at 6 months and subsequent reduction at 12 months, alongside increased serum glial fibrillary acidic protein (GFAP) levels. In the remaining 10 patients, the mean amyloid clearance reached −34.93 Centiloids, alongside decreased serum GFAP levels.

DISCUSSION

Aducanumab showed good tolerability and favorable biological outcomes, with different responses in autosomal dominant mutation carriers.

Highlights

• We summarized the first real‐world efficacy and safety profiles of aducanumab treatment in China, with a particular focus on its use in early‐onset Alzheimer's disease (AD) patients, including autosomal dominant AD.

• We reported the youngest recorded patient with autosomal dominant AD to receive anti‐amyloid therapy to date.

• The biological outcomes of anti‐amyloid therapy may differ in autosomal dominant AD.

1. BACKGROUND

The incidence of Alzheimer's disease (AD) is continuously increasing in China with a rapidly aging population. ¹ Pathological accumulation of amyloid‐β (Aβ) in the brain is an early and core event in the course of AD that occurs decades before the onset of cognitive symptoms, and initiates a cascade of pathological processes including tau pathology, synaptic dysfunction, and neuronal loss. ² , ³ In the past, therapeutic agents for AD only transiently improved symptoms rather than slowing disease progression. However, recent clinical trials have confirmed that anti‐amyloid therapy can delay clinical decline in patients with early AD (mild cognitive impairment or mild dementia due to AD). To date, three anti‐amyloid monoclonal antibodies (aducanumab, lecanemab, donanemab) have been approved by the U.S. Food and Drug Administration (FDA). These agents target different amyloid binding sites and have demonstrated amyloid reduction and clinical benefits in relevant trials. ⁴ , ⁵ , ⁶ , ⁷

Aducanumab, a human monoclonal antibody targeting amyloid pathology, specifically interacts with aggregated soluble and insoluble forms of Aβ species. ⁸ It received its first approval for clinical use by the U.S. FDA in 2021, based on its capability of amyloid clearance. ⁹ In China, aducanumab was first approved for clinical AD therapy in Hainan Boao Lecheng International Medical Tourism Pilot Zone (Boao Research Hospital) in December 2022, marking the first available anti‐amyloid therapy for early AD in the country. Prior to this approval, no safety data on anti‐amyloid therapy for the Chinese population had been reported.

Most clinical trials enroll participants with sporadic AD, with the average ages typically exceeding 65 years. In real‐world practice, however, patient profiles are considerably more heterogeneous. Autosomal dominant Alzheimer's disease (ADAD) represents a distinct AD subtype, characterized by a strong family history and pathogenic mutations in one of three known genes (amyloid protein precursor [APP], presenilin [PSEN] 1, or PSEN2). ADAD typically manifests with early‐onset symptoms (before age 65) and a more rapid disease progression. ¹⁰ The Dominantly Inherited Alzheimer Network (DIAN) has conducted several anti‐amyloid trials in ADAD populations, demonstrating the capacity of anti‐amyloid agents to engage their designated Aβ targets and clear brain amyloid. ¹¹ Nevertheless, conclusive evidence regarding their efficacy in delaying cognitive decline remains limited. Overall, it is still unclear whether real‐world responses to anti‐amyloid therapy differ in individuals with confirmed autosomal dominant mutations.

Despite the withdrawal of aducanumab from the market in 2024, its real‐world safety and efficacy profile in the Chinese population, particularly in the treatment of less common ADAD patient subgroup, warrants detailed summary and discussion. Here, we present the results and clinical experiences from our real‐world study of aducanumab in China (ChiCTR2200066153), along with a comparison to the results from previous clinical trials. Notably, to our knowledge, our study also includes the youngest ADAD patient treated with anti‐amyloid therapy reported to date worldwide.

2. METHODS

2.1. Trial design and outcome measures

This was a 12‐month, single‐center, single‐arm, real‐world prospective study. The study aimed to evaluate the safety and efficacy of aducanumab in early AD among the Chinese population. The trial was registered in the Chinese Clinical Trial Registry (ChiCTR2200066153, https://www.chictr.org.cn/). All procedures involving human subjects were approved by the Ethics Committee of Ruijin Hainan Hospital. The approved ethics committee number was KY2022‐004. All included subjects provided written informed consent.

Patients received aducanumab via intravenous infusion every 4 weeks over a 48‐week period, for a total of 12 doses. The dose was titrated as follows: 1 mg/kg for the first two doses, increased to 3 mg/kg for doses 3 and 4, followed by 6 mg/kg for doses 5 and 6, and finally reaching the target dose of 10 mg/kg for doses 7 through 12 (Figure 1). Brain magnetic resonance imaging (MRI) was performed at baseline and prior to the 5th, 7th, 9th, and 12th doses for safety monitoring of amyloid‐related imaging abnormalities (ARIA). ¹² , ¹³ ARIA can manifest as brain edema or sulcal effusion (ARIA‐E), or as hemosiderosis in the brain parenchyma (ARIA‐H microhemorrhages) or on the pial surface (ARIA‐H superficial siderosis). The primary outcome measure was the Clinical Dementia Rating–Sum of Boxes (CDR‐SB). ¹⁴ Key secondary clinical outcome measures included the Alzheimer's Disease Assessment Scale's cognitive subscale (ADAS‐Cog), ¹⁵ Alzheimer's Disease Cooperative Study—Activities of Daily Living Scale (ADCS‐ADL). ¹⁶ These clinical outcome measures were assessed at baseline, week 26, and week 50. Key secondary biological outcome measures included amyloid burden measured by 18F‐florbetapir positron emission tomography (PET), and serum biomarkers including glial fibrillary acidic protein (GFAP), neurofilament light chain (NfL), phosphorylated (p)‐tau 181, p‐tau 217, Aβ42, Aβ40. Longitudinal positron emission tomography (PET) scans and blood samples were collected at baseline, week 26, and week 50. In addition, multi‐modal MRI, including 3D T1‐weighted imaging and Diffusion tensor imaging (DTI), was performed at baseline and week 50 to explore brain structural changes. The flow of this trial is depicted in Figure 1.

FIGURE 1.

The trial design and outcomes measures. PET, positron emission tomography; MRI, magnetic resonance imaging.

RESEARCH IN CONTEXT

1. Systematic review: Recent trials have shown that anti‐amyloid therapy can delay clinical decline in early Alzheimer's disease (AD). However, its clinical and biological efficacy, particularly in early‐onset AD and autosomal dominant AD, remains to be established in Chinese clinical practice. Besides, there is still a lack of robust clinical evidence supporting the efficacy of anti‐amyloid therapy for autosomal dominant AD.

2. Interpretation: In this study, we summarized China's first real‐world efficacy and safety profiles of aducanumab treatment. Notably, the biological outcomes of anti‐amyloid therapy differed in autosomal dominant AD.

3. Future directions: The therapeutic effects of anti‐amyloid therapy on early‐onset AD and autosomal dominant AD will continue to be investigated and monitored in clinical practice.

2.2. Study participants

All participants were established patients in the Memory Clinic of the Department of Neurology at Ruijin Hospital, and a convenience sampling strategy was employed for recruitment. Participants were diagnosed as mild cognitive impairment (MCI) or mild dementia according to 2011 National Institute on Aging‐Alzheimer's Association (NIA‐AA) criteria, with evidence of pathological amyloid deposition on 18F‐florbetapir PET. ¹⁷ , ¹⁸ They were all under 80 years of age, both male and female, and their CDR global scores should be 0.5 or 1. Detailed inclusion and exclusion criteria were listed in Methods S1 in supporting information.

2.3. Data collection and analysis

18F‐florbetapir PET was performed on a 3.0T hybrid PET/MR scanner (Biograph mMR; Siemens Healthcare, Erlangen, Germany) with a 12‐channel phase‐array head coil. Each participant received an intravenous injection of 18F‐florbetapir at an average dose of 370 MBq. After 40 to 60 min post‐injection, the static PET images were acquired. The PET images were first visually assessed by nuclear medicine physicians, followed by processing via a validated pipeline to quantify amyloid burden. Preprocessing was performed on MATLAB 2022b using the Statistical Parametric Mapping 12 (SPM12, https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) (Methods S2). The global cortical target was composed of multiple brain regions with high amyloid deposition, including the precuneus, frontal, temporal, parietal, insular cortices, and the anterior striatum. The whole cerebellum was used as the reference region for calculating the standardized uptake value relative ratio (SUVR). 18F‐florbetapir SUVR was then converted to the Centiloid (CL) scale following the formula: CL = 179.64*SUVR_Florbetapir‐186.95, which has been validated in Chinese population. ¹⁹

MRI data including 3D T1‐weighted imaging and DTI were acquired using a 3.0T uMR‐890 MRI scanner (Shanghai United Imaging Healthcare, Shanghai, China) with a 64‐channel head coil. MULTIPLEX imaging technology was employed. The 3D T1‐weighted scanning parameters were as follows: repetition time (TR) = 7.60 ms; echo time (TE) = 3.40 ms; inversion time (TI) = 1100 ms, flip angle = 7°; field‐of‐view (FOV) = 220 × 240 mm; matrix = 293 × 320; voxel size = 0.75 mm× 0.75 mm× 0.75 mm; scanning time = 235 s with a total of 360 slices each. Longitudinal T1‐weighted structural analyses were performed on FreeSurfer 7.3.2 (https://surfer.nmr.mgh.harvard.edu) using the longitudinal processing pipeline to calculate changes in cortical thickness and subcortical volume (Methods S3) from baseline to 6‐month and 12‐month follow‐up. The multi‐shell DTI scanning parameters were as follows: TR = 3800 ms; TE = 59 ms; flip angle = 90°; FOV = 256 mm × 256 mm; matrix = 128 × 128; voxel size = 2 mm × 2 mm × 2 mm; slice thickness = 2 mm; 100 diffusion encoding directions with 4 b = 0 images, 32 b = 1000 s/mm² images, 64 b = 2000 s/mm² images. DTI analysis along the perivascular space (ALPS) index was calculated to estimate the glymphatic function using a validated semi‐automated pipeline performed on FSL 6.0 (https://fsl.fmrib.ox.ac.uk/fsl/) (Methods S4). ²⁰

A 6 mL venous blood sample was collected from all participants at each visit (baseline, week 26, week 50). The concentrations of serum Aβ40, Aβ42, p‐tau181, p‐tau217, GFAP, NfL were measured using the Single Molecular Immunity Detection (SMID) platform (AST‐Sc‐Lite; ASTRABIO, China) (Methods S5).

2.4. Statistical analysis

Descriptive statistics were used for data summarization. Clinical data were presented as mean ± standard deviation (SD). Blood biomarker values were log‐transformed prior to statistical analysis. Linear mixed‐effects models were used to compare changes in the primary and key secondary outcomes from baseline at 6 and 12 months. The dependent variables included CDR‐SB score, ADAS‐cog score, ADCS‐ADL score, amyloid levels, and blood biomarkers, with visit time as the fixed categorical effect. R package “lmer” was used based on R version 4.4. Paired t‐tests were conducted to compare brain structural values and ALPS index between baseline and 12‐month follow‐up. The p value of < 0.05 was considered statistically significant.

3. RESULTS

3.1. Demographics and baseline characteristics

A total of 15 patients received aducanumab treatment. Among them, a 43‐year‐old AD patient with the PSEN1 E184G mutation was lost to follow‐up after completing 12 doses (baseline CDR = 1); a 74‐year‐old AD patient withdrew from the study due to severe injury from a traffic accident after the 3rd dose; and a 52‐year‐old AD patient discontinued treatment for personal reasons after the 6th dose. Consequently, 12 patients from different provinces & cities of China [Shanghai (n = 3), Jiangsu (n = 3), Zhejiang (n = 2), Yunnan (n = 2), Fujian (n = 1), Jilin (n = 1)] completed the treatment and clinical assessment, and were included in the following analysis. The first patient initiated treatment in December 2022, and the last patient started in January 2024, with all treatments ending in December 2024. Demographics and baseline cognitive assessment scores were summarized in Table 1. The mean (± SD) age was 58.92 ± 13.52 years (range from 31 to 78), and eight of 12 patients were early‐onset AD (EOAD). Among them, two were diagnosed with ADAD (Figure S1), who was a 31‐year‐old woman carrying the PSEN1 Thr147Ile mutation and a 52‐year‐old man with the APP V717I mutation. Besides, other patients also underwent whole‐exome sequencing. A PSEN2 Gly34Ser variant was detected in a 59‐year‐old patient with no family history of AD. Seven patients were apolipoprotein E (ApoE) ε4 carriers: one ε2/ε4, one ε4/ε4, and five ε3/ε4. The mean (± SD) baseline CDR‐SB score was 2.71 ± 1.72. Seven patients were diagnosed with mild dementia and five with MCI due to AD.

TABLE 1.

Characteristics of all the participants at baseline.

Characteristic	Participants n = 12
Age, years, mean ± SD	58.92 ± 13.52
Sex
Male, n, %	5, 41.67%
Female, n, %	7, 58.33%
Duration of disease, years, mean ± SD	1.46 ± 0.84
Diagnosis, n
Mild dementia due to AD	7
Mild cognitive impairment due to AD	5
Clinical subgroup, n
Autosomal‐dominant AD	2
Early‐onset AD	8
ApoE ε4 status, n
Noncarrier	5
Carrier
Heterozygotes	6
Homozygotes	1
Oral medication for disease, n, %	11, 91.67%
Amyloid level, Centiloids, mean ± SD	83.37 ± 30.57
Amyloid level, SUVR, mean ± SD	1.50 ± 0.17
Global CDR score, n
0.5	9
1	3
CDR‐SB score, mean ± SD	2.71 ± 1.72
ADAS‐cog score, mean ± SD	14.82 ± 7.12
ADCS‐ADL, mean ± SD	72.83 ± 2.55
MoCA, mean ± SD	18.64 ± 4.59

Abbreviations: AD, Alzheimer's disease; ADAS‐cog, Alzheimer's Disease Assessment Scale's cognitive subscale; ADCS‐ADL, Alzheimer's Disease Cooperative Study‐Activities of Daily Living Scale; APOE, apolipoprotein E; CDR, Clinical Dementia Rating; CDR‐SB, Clinical Dementia Rating—Sum of Boxes; MoCA, Montreal Cognitive Assessment; SD, standard deviation; SUVR, standardized uptake value ratio.

3.2. Clinical efficacy

Six patients exhibited stable CDR‐SB scores (including two ADAD patients), while the other six showed increased scores compared to baseline. The mean CDR‐SB score at baseline was 2.71. The mean change from baseline at 12 months in the CDR‐SB score was 0.88 (95%CI: 0.31 to 1.44, p = 0.006). The mean ADAS‐cog score at baseline was 14.82. The mean change from baseline at 12 months in the ADAS‐cog score was 2.21 (95%CI: 0.23 to 4.19, p = 0.041). The mean ADCS‐ADL score at baseline was 72.83. The mean change from baseline at 12 months in the ADCS‐ADL score was −3.08 (95%CI: −4.97 to −1.19, p = 0.004). The results were shown in Table 2. After adjusting for age, the mean change in cognitive scores remained consistent, and older individuals exhibited higher CDR‐SB and lower ADCS‐ADL scores during the treatment. Supplementary results for cognitive changes were presented in Table S1 and S2 in Supporting Information.

TABLE 2.

Clinical changes in overall populations.

Clinical efficacy n = 12	Baseline	6‐month	12‐month	12‐month versus baseline	95%CI	p‐value
CDR‐SB	2.71 ± 1.72	2.92 ± 1.93	3.58 ± 2.74	0.88	0.31–1.44	0.006
ADAS‐cog	14.82 ± 7.12	14.67 ± 6.94	17.03 ± 8.34	2.21	0.23–4.19	0.041
ADCS‐ADL	72.83 ± 2.55	71.58 ± 2.97	69.75 ± 4.86	−3.08	−4.97 to −1.19	0.004

Abbreviations: ADAS‐cog, Alzheimer's Disease Assessment Scale's cognitive subscale; ADCS‐ADL, Alzheimer's Disease Cooperative Study‐Activities of Daily Living Scale; CDR‐SB, Clinical Dementia Rating—Sum of Boxes; CI, confidence interval.

3.3. Biological outcomes

At 12 months, 3 out of 12 (25%) patients achieved amyloid clearance (defined as < 24.1 CL) or amyloid‐negative status confirmed by visual assessment. However, Aβ clearance efficacy differed between ADAD patients and non‐ADAD patients. Notably, the two ADAD patients showed an initial increase in amyloid levels at 6 months, followed by a decline at 12‐month follow‐up. And the amyloid levels of the patient with APP V717I mutation were still higher than the baseline. In the non‐ADAD patients, their mean amyloid levels at baseline were 83.20 CL. The mean change from baseline at 12 months were −34.93 CL (95%CI: −46.41 to −23.44, p < 0.001) (Table 3). The changes of amyloid levels in each participant were shown in Figure 2A–C.

TABLE 3.

Biological changes in non‐ADAD populations.

Biological efficacy n = 10	Baseline	6‐month	12‐month	12‐month versus baseline	95%CI	p‐value
Amyloid burden, Centiloids	83.20 ± 22.26	76.71 ± 15.95	48.27 ± 23.26	−34.93	−46.41 to −23.44	<0.001
GFAP, log‐transformed	0.78 ± 0.28	0.50 ± 0.24	0.48 ± 0.34	−0.30	−0.53 to −0.07	0.019

Abbreviations: ADAD, autosomal dominant AD; PET, positron emission tomography; CI, Confidence Interval; GFAP, glial fibrillary acidic protein.

FIGURE 2.

The changes in amyloid levels of each patient (A) and the changes in amyloid levels (B, C), serum GFAP levels (D) of two ADAD patients. ADAD, autosomal dominant Alzheimer's disease; GFAP, glial fibrillary acidic protein; MRI, magnetic resonance imaging.

Notably, serum GFAP trends differed between the two ADAD patients and the non‐ADAD patients. While non‐ADAD patients showed a statistically significant decrease in serum GFAP levels at 6 and 12 months versus baseline (p = 0.019; Table 3), the two ADAD patients exhibited a numerical upward trend in GFAP levels over the same follow‐up period (Figure 2D).

3.4. Structural MRI changes

Longitudinal analysis of brain structural changes in the 12 patients revealed reductions in mean cortical thickness and hippocampal volume, as well as increases in lateral ventricular volume and choroid plexus volume between baseline and the 12‐month follow‐up (all p < 0.05) (Table 4). The ALPS index, reflecting glymphatic function, declined after the 12‐month follow‐up (p < 0.05) (Table 4). The participant carrying the PSEN1 Thr147Ile mutation exhibited the most significant increase in bilateral ventricular volume and atrophy in the left hippocampus compared to all other participants despite her youngest age, whereas the participant with the APP mutation showed fewer structural changes (Figure S2).

TABLE 4.

Structural MRI changes in overall populations.

MRI	Baseline	12‐month	t	p‐value
lh mean thickness, mm	2.36 ± 0.10	2.30 ± 0.12	3.495	0.005
rh mean thickness, mm	2.35 ± 0.12	2.28 ± 0.13	4.178	0.002
Left lateral ventricle/TIV,%	0.91 ± 0.26	1.10 ± 0.34	−4.361	0.001
Right lateral ventricle/TIV,%	0.91 ± 0.31	1.12 ± 0.44	−4.070	0.002
Left hippocampus/TIV,%	0.23 ± 0.03	0.22 ± 0.03	6.184	<0.001
Right hippocampus/TIV,%	0.23 ± 0.03	0.22 ± 0.04	7.458	<0.001
Left choroid plexus/TIV,‰	0.87 ± 0.17	0.93 ± 0.19	−2.934	0.014
Right choroid plexus/TIV,‰	0.86 ± 0.12	0.96 ± 0.14	−3.924	0.002
ALPS (n = 10)	1.34 ± 0.10	1.29 ± 0.12	2.613	0.028

Abbreviations: ALPS, analysis along the perivascular space; lh, left hemisphere; MRI, magnetic resonance imaging; rh, right hemisphere; TIV, total intracranial volume.

3.5. Safety

Neither ARIA‐E nor ARIA‐H was observed during the entire treatment period. One patient developed a headache accompanied by vomiting on the second day after infusion, leading to temporary treatment suspension, but both ARIA‐E and ARIA‐H were ultimately ruled out. Treatment was resumed after all symptoms resolved. Another patient experienced mild headache with no evidence of ARIA. None of these symptoms were attributed to infusion‐related reactions (IRRs). At the 6‐month assessment, one patient exhibited mild anxiety, which was managed with oral escitalopram oxalate tablets. The symptom had improved by the 12‐month assessment. One patient developed seizure symptoms during the interval between the 10^th and 11^th doses. Given the timing near treatment completion, this event was considered likely attributable to disease progression rather than adverse effects.

4. DISCUSSION

This is the first and only real‐world report of aducanumab use in China. No ARIA events were observed throughout the treatment period, and the infusion process was well‐tolerated by all patients. The population in this study was heterogeneous, with a wide age range and diverse genetic backgrounds. We found different biological outcomes of anti‐amyloid therapy in ADAD patients compared with other AD patients, which will be discussed in the following sections.

Notably, no ARIA events were observed during the treatment given the small sample. While most clinical trials exclude patients with more than four microhemorrhages, our study applied even stricter baseline criteria. Given that this was the first use of this anti‐amyloid therapy in China and the absence of prior regional safety data, we enrolled only individuals without any microhemorrhages and excluded those with distinct white matter lesions. This stringent selection process is therefore a critical factor to consider when analyzing our reported ARIA event rates. Future studies with larger cohorts are needed to clarify whether the baseline number of microhemorrhages correlates with a higher risk of ARIA. Besides, the mean age of participants in our trial was younger than that in other clinical trials (around 70 years old), ⁴ , ⁶ , ⁷ and these younger participants may have better overall physical health. Another key factor associated with the incidence of ARIA is ApoE ε4 carrier frequency. The ApoE ε4 carrier rate in our trial was 58.33% (7/12), which was lower than that in the overall population (67–68%) in the EMERGE and ENGAGE trial. ⁷ The differences in ApoE ε4 carrier frequency and ARIA frequency should be interpreted with caution due to the small sample size of our study. Our safety profiles are also consistent with those from other Asian subgroup studies, which show that Asian populations treated with aducanumab or lecanemab exhibited a lower incidence of both ARIA events and IRRs compared with the global population. ²¹ , ²² In the Clarity AD Asian regional analysis, the incidence of IRRs and ARIA‐E/ARIA‐H was lower in the Asian population than in the overall population, despite similar frequencies of ApoE ε4 carriers between the two groups. ²¹ These findings may suggest a clinically favorable risk–benefit ratio for anti‐amyloid therapy in Asian populations. In addition, other common side effects include headaches and nausea.

In this study, we completed anti‐amyloid therapy for the two ADAD patients, one of whom represented the youngest reported case to date (symptom onset at age 30). Both patients exhibited an initial increase followed by a subsequent decline in amyloid levels; however, the patient carrying the APP V717I mutation (known as the London mutation) showed a continuous rise in amyloid burden despite aducanumab treatment—a finding not previously reported. Longitudinal studies reported an average amyloid accumulation rate of approximately 0.06 SUVR/year (measured by Pittsburgh‐Compound‐B [PiB] PET) in pathogenic variant carriers with increasing estimated years from symptom onset (EYO). ²³ Aducanumab treatment slowed the accumulation of Aβ in both ADAD patients, yet clearance levels remained insufficient. The variability in efficacy has also been observed in previous DIAN trials. ¹¹ Gantenerumab, which has high affinity for aggregated Aβ (including oligomers, fibrils, and plaques), significantly reduced amyloid plaques. In contrast, solanezumab, which targeted soluble Aβ, failed to show benefit on downstream biomarkers. This suggests that the effectiveness of amyloid removal is closely associated with the specific target of the anti‐amyloid agent. Furthermore, amyloid clearing efficacy in ADAD patients may also be related to the pathogenic variants. Studies showed that amyloid accumulation (measured by PiB‐PET) varied widely across PSEN1, PSEN2, or APP mutations that could be grouped into 9 affected domain‐based groups. ²³ In our study we observed that the patient with PSEN1 Thr147Ile mutation, which was grouped into PSEN1 Transmembrane Domain 2, had higher amyloid levels at baseline than the APP V717I mutation carrier. So given the heterogeneity in amyloid levels across pathogenic variants, this should also be taken into consideration when analyzing the biological outcomes. For patients with higher amyloid burden at baseline, extending treatment duration or implementing personalized therapeutic strategies should be considered.

The neuropsychological scores of the two ADAD patients remained stable, necessitating extended follow‐up to assess the long‐term cognitive effects of anti‐amyloid therapy. This is particularly relevant given the mixed evidence from previous trials. For instance, while gantenerumab failed to slow cognitive decline in symptomatic ADAD patients despite significantly reducing amyloid plaques, ¹¹ a subsequent open‐label extension study suggested that three years of treatment and full amyloid removal might potentially delay symptom onset and disease progression. ²⁴ This implies that partial or short‐term clearance of amyloid plaques may be insufficient for ADAD patients.

In the non‐ADAD patients, Aβ clearance was observed during treatment with aducanumab. For different amyloid tracers, optimal CL thresholds for amyloid positivity against visual assessment fell within the range between 24.1 and 30 CL, ²⁵ , ²⁶ , ²⁷ which were considered to distinguish sparse/none versus moderate/frequent amyloid plaques in autopsy confirmation. ²⁸ According to this standard (< 24.1 CL) and visual assessment, 25% (3 of 12) of the patients achieved amyloid clearance, which is similar to the results in the TRAILBLAZER‐ALZ 4 trial that showed 24.6% (15 of 61) of aducanumab‐treated participants reached amyloid clearance at 12 months. ²⁹ The rate increased to 43.1% (25 of 58) at 18 months. ²⁹ These results indicate a dose‐ and time‐dependent reduction in amyloid levels. The mean change from baseline at 12 months in our cohort was −34.93 CL. It was less than the level of amyloid clearance in donanemab, which was −80.03 at 12 months and −84.22 at 18 months. ²⁹ At the individual level, some patients did not have a significant reduction in amyloid levels after treatment, which indicates that the efficacy varied among individuals.

Regarding the outcome measures, the mean change in CDR‐SB scores from baseline to 12 months was 0.88 in the overall study population. Scores remained stable in the two ADAD patients (baseline scores: 0.5 and 1, respectively) and the other 10 non‐ADAD patients exhibited a mean increase of 1.05 in CDR‐SB scores (Table S2). Previously, a longitudinal observational study of symptomatic AD with a global CDR of 0.5 or 1 reported annual CDR‐SB progression rates of 1.43 (for the CDR 0.5 group) and 1.91 (for the CDR 1 group). ³⁰ The CDR‐SB change observed in aducanumab treatment was lower than the annual progression rates. Changes in brain structure aligned with findings from previous trials reporting lateral ventricle volume increases, ⁶ , ⁷ which were particularly prominent in the patient with the PSEN1 mutation. These observations suggest that ventricular enlargement may result from genetic predisposition combined with anti‐amyloid therapeutic effects. No significant changes were observed in serum p‐tau181 or p‐tau217 levels. In contrast, reductions in p‐tau181 and p‐tau217 were detected during treatment with lecanemab and donanemab. ²⁹ , ³¹ This indicates that the magnitude of amyloid clearance may not reach the efficacy of lecanemab and donanemab therapies, or that serum biomarker levels may exhibit greater variability. Notably, GFAP levels demonstrated a significant decline in non‐ADAD patients but a rising trend in the two ADAD patients compared with other biomarkers. These results potentially indicate its enhanced sensitivity as an early indicator of astrocyte activation during amyloid clearance dynamics.

A limitation of our study was that the sample size was relatively small. The efficacy varied between patients. Besides, the manufacturer of aducanumab had withdrawn the medication from the market so we could not extend treatment. Of note, 8 of the 12 patients in our study were diagnosed with EOAD. While mutations in APP, PSEN1, and PSEN2 account for only about 11% of EOAD cases overall, ³² most genetic risks in EOAD remain unexplained, especially among those with a family history (familial EOAD). ¹⁰ In line with this, we did not identify clear pathogenic variants in our EOAD patients. Other genetic risk factors, such as the ApoE ε4 genotype, should also be considered. Therefore, the influence of potential genetic factors on anti‐amyloid therapy efficacy in EOAD remains uncertain and requires further investigation in long‐term follow‐up studies. Our team will continue to investigate the therapeutic effects of these anti‐amyloid monoclonal antibodies on familial early‐onset AD including ADAD (NCT06883019, Lecanemab for Early Onset Familial Alzheimer's Disease).

Overall, we demonstrated the safe use of aducanumab in a small‐sample, real‐world study at a single site in China. Our early clinical experience with aducanumab allowed us to develop and publish clinical recommendations for anti‐amyloid monoclonal antibody therapy in the Chinese population, ³³ which have guided us to use lecanemab and donanemab more systematically in clinical practice. Findings from our small EOAD cohort, which included two patients with confirmed AD autosomal dominant mutations, revealed some differences in biological and clinical outcomes between those with known mutations and those without. These results are intriguing and may warrant further investigation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All participants provided written informed consent.

Supporting information

Supporting Information

Supporting Information