Higher phosphorylated tau levels predict cognitive decline and amyloid-related imaging abnormalities during lecanemab treatment: clinical practice data

Abstract

Background

Lecanemab was recently approved for the treatment of patients with early Alzheimer’s disease (AD) and demonstrated reduced senile amyloid plaque and less decline on the measures of cognition and function in clinical trials. However, the real-world data on its efficacy and safety remain limited. We aimed to evaluate the effectiveness and tolerance of lecanemab treatment and determine biomarkers at baseline that could predict cognitive deterioration and the occurrence of amyloid-related imaging abnormaities (ARIA) in real-world clinical practice.

Methods

To determine the indication for lecanemab, the patients were evaluated through neurological examinations, cognitive assessments, blood test, head magnetic resonance imaging (MRI), amyloid positron emission tomography, lumbar puncture, genetic testing, and clinical conferences. The Mini-Mental State Examination (MMSE) was used to assess cognition, and the MRI scans were used for safety monitoring of ARIA.

Results

Between January 2024 and October 2025, 234 patients were screened, 100 initiated lecanemab treatment. The mean age was 72.7 years, and 68 (68.0%) patients were female. Among the 71 patients surveyed via MRI prior to the 14th infusion, 12 (16.9%) had ARIA detected. Compared with those of patients without ARIA, the baseline cerebrospinal fluid (CSF)-ptau181 levels of patients with ARIA significantly increased. When the patients were divided into high and low CSF-ptau181 groups according to the cutoff value (78.6 pg/ml) which derived from ROC analysis for ARIA prediction, the MMSE scores of the high ptau group were significantly declined compared to that of the low ptau group at 6 and 12 months after baseline. The infusion-reactions occurred only in 6.0% of patients. The longitudinal observation revealed that the plasma thrombomodulin levels significantly decreased after 6 months of lecanemab treatment.

Conclusion

Lecanemab was generally well tolerated by most patients with early AD and treatment appeared to be more effective and safer in patients with low CSF-ptau181 levels.　Our results suggest an association between lecanemab treatment and reduced markers of vascular endothelial injury.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13195-025-01912-6.

Background

Lecanemab is an anti-amyloid monoclonal antibody that binds with high affinity to soluble amyloid-β (Aβ) protofibrils, reduces senile amyloid plaques, and results in less decline on the measures of cognition and function compared with the placebo in the Clarity AD trial in patients with early Alzheimer’s disease (AD) [1]. We recently observed the structural dynamics of Aβ-protofibrils via high-speed atomic force microscopy at the single-molecule level and the effects of lecanemab [2]. We also quantified Aβ-protofibrils bound by lecanemab in the cerebrospinal fluid (CSF) and reported that lecanemab-associated Aβ-protofibril levels are high in all AD continuum stages [3].

While lecanemab was released as an effective treatment for patients with early AD and was approved in Japan at the end of 2023 [4], side effects such as amyloid-related imaging abnormalities (ARIA) and infusion-related reactions have been issues. The incidence of ARIA is greater among patients who are apolipoprotein E ε4 (APOE4) homozygotes, but the pathogenesis of the underlying mechanism of ARIA remains unknown. Initial reports from other centers are now appearing [5, 6] and help us better understand the safety and effectiveness of lecanemab in real-world clinical practice. In the Clarity AD trial and open label extension trial, lecanemab has shown particular efficacy in patients with absent or low tau accumulation [7]. Tau PET is not feasible for every patient in clinical practice; measurement of CSF phosphorylated tau (ptau)181, a biomarker that correlates with cerebral tau deposition [8], could therefore be a useful surrogate marker for predicting therapeutic efficacy. Cerebral amyloid angiopathy (CAA) is defined pathologically by Aβ deposition in cerebral blood vessels and is frequently observed in patients with advanced AD [9]. CAA and ARIA share overlapping pathological mechanisms and CAA is a major risk factor for ARIA. It was found that interaction between Aβ and fibrinogen in some types of CAA [10]. In addition, vascular injury is also a feature of AD, and increased thrombosis with fibrin deposits in the AD brain has been reported [11]. These findings suggest that coagulation and fibrinolysis factors may play an important role in developing ARIA.

This study aimed to evaluate the treatment course and side effects of lecanemab treatment in patients with early AD at Kanazawa University Hospital. We investigated the CSF-phosphorylated tau marker, MRI findings, APOE4 status, and other demographic variables to identify baseline biomarkers predictive of subsequent cognitive decline and the development of ARIA. As an exploratory analysis, we examined whether coagulation and fibrinolysis test results were associated with the occurrence of ARIA.

Methods

Study population

We performed a retrospective review of patients treated with lecanemab at the anti-amyloid treatment clinic at Kanazawa University Hospital. Eligible patients were diagnosed with early AD [mild cognitive impairment (MCI) due to AD or mild AD dementia] on the basis of the National Institute on Aging-Alzheimer’s Association (NIA-AA) clinical criteria [12, 13] with confirmed amyloid pathology. Patients with a Mini-Mental State Examination (MMSE) score of < 22 or a clinical dementia rating (CDR)—a global score of 2 or higher—and those with normal cognitive function were excluded. For patients who started lecanemab, the MMSE and the CDR were used to assess cognitive decline at baseline and at 6, 12, and 18 months. This study was approved by the Medical Ethics Review Board of Kanazawa University (approval number 114,625) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from patients who visited after the study protocol was approved by the Medical Ethics Review Board of Kanazawa University. For patients who visited before that, consent was obtained by opt-out on the website.

Head MRI and amyloid PET

All patients underwent head MRI and amyloid PET imaging before starting treatment with lecanemab. The baseline structural MRI was conducted via a 1.5- or 3-T scanner (Signa Artist, 1.5T, GE Healthcare; Signa Premier, 3.0T, GE Healthcare; Ingenia 1.5T CX, Philips; Ingenia 3.0T, Philips). The protocol comprised diffusion-weighted imaging, fluid-attenuated inversion recovery, T2*-weighted imaging, and high-resolution three-dimensional (3D) T1-weighted imaging. The 3D T1-weighted images were analysed for volumetric evaluation of medial temporal lobe atrophy via a specific voxel-based regional analysis system for Alzheimer’s disease (VSRAD Advance 2; Eisai, Japan). Leucoaraiosis severity was rated via the visual rating scale proposed by Fazekas et al. [14]. For periventricular hyperintensity (PVH), the scores corresponded to the following characteristics: 0 = no changes; 1 = a cap or a pencil-thin lining; 2 = smooth halo; and 3 = irregular changes extending into the deep white matter. For deep white matter hyperintensity (DWMH), the scores corresponded to the following characteristics: 0 = no lesion; 1 = punctate foci; 2 = beginning confluent foci; and 3 = confluent changes. The extent of leukoaraiosis was determined by FLAIR imaging. MRIs for ARIA monitoring were performed before the fifth, seventh, and 14th infusions, as described in the appropriate use recommendation guidelines for prescribing information. Amyloid PET imaging was performed using a Vizamyl (flutemetamol F¹⁸ injection) tracer, and the results were visually positive or negative. The centiloid values were calculated via the VIZCalc method [15]. We excluded patients with more than four microhemorrhages, a single macrohemorrhage, an area of superficial siderosis, vasogenic edema, and multiple lacunar infarcts on MRI and/or amyloid negativity on the amyloid PET scan.

CSF and blood collection and assays

All patients underwent blood tests and lumbar punctures before the first lecanemab inclusion and approximately 6 months later. CSF samples were collected, centrifuged, and frozen at − 80 °C. Lumipalse G assays were employed to determine CSF-ptau181 (phospho-tau [181p]; SRL, Tokyo, Japan), as previously described [16]. When we investigated the CSF-ptau181 concentrations in patients treated with lecanemab, the baseline CSF-ptau181 levels of patients with ARIA significantly increased and the cutoff value that derived from ROC analysis for ARIA prediction was 78.6 pg/ml; therefore, patients with CSF-ptau181 concentrations above 78.6 pg/ml were classified into the high ptau group, whereas those with concentrations below 78.6 pg/ml were categorized into the low ptau group. The relationships between CSF-ptau181 levels and cognitive test results after lecanemab treatment were examined. Plasma was analysed for prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen (Fbg), fibrin/Fbg degradation products (FDP), FDP-D-dimer (FDP-DD), α2 plasmin inhibitor (α2-PI), plasminogen (Plg), thrombin–antithrombin complex (TAT), α2 plasmin inhibitor–plasmin complex (PIC), thrombomodulin (TM), and von Wilebrand factor (vWF). The PT and APTT were measured via the coagulation method. Fbg was measured via the thrombin method. FDP and FDP-DD were measured via latex turbidimetric immunoassay. The TAT, the PIC, and the TM were measured by a chemiluminescence immunoassay. The α2-PI and Plg were measured via the synthetic substrate assay. vWF was measured through the fixed platelet agglutination assay. The reference ranges were based on the criteria at our institution.

Treatment regimen

Lecanemab was administered intravenously every 2 weeks as described in the prescribing information. Infusion reactions occurred in 26.4% of the participants in the Clarity AD study [1]; hence, we administered premedication up to the first three treatments to suppress symptoms. The premedication regimen included (1) oral administration of 500 mg of acetaminophen and (2) intravenous injection of 5 mg of chlorpheniramine, 20 mg of famotidine, and 9.9 mg of dexamethasone.

Statistical analysis

Clinical characteristics according to the low- and high- ptau groups or with and without ARIA groups were evaluated through a t-test of the continuous variables and a chi-squared test including residual analysis for the categorical variables. Generalized linear models (binomial distribution and logistic regression model) were utilized to analyse the independent effects of high ptau group on ARIA. Model 1 was adjusted for age and sex, and Model 2 was further adjusted for APOE4 carrier status. Models 3, 4, and 5 were adjusted for the global CDR score, amyloid PET centiloid value, and number of microbleeds on baseline MRI in addition to the covariates in Model 2, respectively. We used generalized linear mixed effects repeated measure analysis adjusted for APOE4 carrier status to evaluate the effect of high ptau group on changes in MMSE and CDR-sum of boxes (SB) scores from baseline to 6-, and 12-month follow-ups. A comparison of baseline characteristics between early and late enrollees among participants who completed the 6- or 12-month follow-up were also evaluated through a t-test and a chi-squared test. Paired t tests were used to analyse the changes in the coagulation and fibrinolysis tests and CSF-ptau181 between baseline and 6 months. The problem of multiple comparisons was addressed by controlling Benjamini–Hochberg false discovery rate (FDR). The FDR-adjusted p-values (i.e., q values) below 0.05 were considered statistically significant. A p value < 0.05 was considered to indicate statistical significance. Statistical analyses were performed via the Statistical Package for the Social Sciences software (version 28; SPSS, Inc., Chicago, IL, USA).

Results

Between January 2024 and October 2025, 234 patients visited our anti-amyloid treatment clinic, and 216 completed the evaluations. A total of 78 patients did not meet the eligibility criteria for the following reasons: six patients had normal cognition; four patients had moderate dementia; 16 patients had MRI contraindications; 24 patients were amyloid negative; seven patients were diagnosed with another type of dementia (e.g., mixed dementia or dementia with Lewy bodies); 18 patients changed their mind and no longer wanted to undergo anti-amyloid antibody treatment; and three patients developed unstable medical conditions after visiting our memory clinic. Thirty-eight patients had pending/other reasons not to receive lecanemab treatment: two patient was scheduled to begin lecanemab treatment in the next month; seven patients were APOE4 homozygous; seven patients were taking anticoagulants or antiplatelet drugs; and 22 patients chose different anti-amyloid therapies. Eventually, 100 patients underwent at least one lecanemab infusion (Fig. 1). All patients were Japanese. Table 1 presents the demographics of the patients who initiated lecanemab treatment. A total of 10 out of 100 patients (10.0%) discontinued lecanemab treatment. Four patients discontinued treatment due to ARIA, three of whom developed asymptomatic moderate ARIA-E, and one developed more than six microhemorrhages since the initiation of lecanemab treatment. All of them opted to discontinue. Two patients discontinued treatment because of infusion reactions. The other reasons for discontinuation were as follows: one patient stopped due to hospitalization for depression; one stopped due to agitation; and two patients chose to discontinue without a specified reason (Fig. 1).

Fig. 1.

Participant flow

Table 1.

Demographics and cognitive tests results of patients at baseline and follow-up, who initiated Lecanemab treatment

	All patients	Patients who completed 6-month follow-up	Low ptau group*	High ptau group	p Low ptau vs. High ptau at 6 months	Patients who completed 12-month follow-up	Low ptau group	High ptau group	p Low ptau vs. High ptau at 12 months
Number	100	69	35	33		37	21	16
Age at baseline, year	72.7 (7.2)	72.7 (7.6)	72.8 (8.2)	72.7 (7.1)	0.944	73.8 (7.1)	74.0 (7.9)	73.6 (6.1)	0.882
Education, year	13.2 (2.0)	13.1 (2.2)	13.3 (2.4)	12.8 (1.9)	0.396	12.6 (2.3)	12.9 (2.4)	12.3 (2.1)	0.497
Women, no (%)	68 (68.0)	45 (65.2)	20 (57.1)	25 (75.8)	0.129	27 (73.0)	15 (71.4)	12 (75.0)	1.000
Global CDR score, no (%)
0.5	74 (74.0)	54 (78.3)	29 (82.9)	25 (75.8)	0.555	30 (81.1)	16 (76.2)	14 (87.5)	0.674
1	26 (26.0)	15 (21.7)	6 (17.1)	8 (24.2)		7 (18.9)	5 (23.8)	2 (12.5)
APOE4 status, no (%)
Noncarrier	32 (32.0)	21 (30.4)	13 (37.1)	8 (24.2)	0.288	8 (21.6)	6 (28.6)	2 (12.5)	0.461
Heterozygous	52 (52.0)	38 (55.1)	18 (51.4)	20 (60.6)		22 (59.5)	11 (52.4)	11 (68.8)
Homozygous	16 (16.0)	10 (14.5)	4 (11.4)	5 (15.2)		7 (18.9)	4 (19.0)	3 (18.8)
Hypertension, no (%)	47 (47.0)	34 (49.3)	15 (42.9)	18 (54.5)	0.467	14 (37.8)	8 (38.1)	6 (37.5)	1.000
Use of medications for symptoms of dementia, no (%)	67 (67.0)	45 (65.2)	25 (71.4)	19 (57.6)	0.311	22 (59.5)	15 (71.4)	7 (43.8)	0.107
VSRAD	2.1 (1.0)	2.0 (1.0)	2.0 (1.1)	2.1 (0.9)	0.938	2.1 (1.0)	2.0 (1.0)	2.2 (1.0)	0.542
Amyloid PET, centiloids	54.5 (21.1)	53.4 (19.9)	50.1 (21.9)	57.0 (17.4)	0.155	52.0 (20.6)	47.7 (21.3)	57.5 (18.8)	0.154
CSF-ptau181 at baseline, pg/ml	92.9 (43.7)	84.2 (34.0)	57.9 (15.1)	112.1 (24.9)	< 0.001	81.1 (33.3)	58.6 (15.7)	110.6 (26.3)	< 0.001
MMSE score at baseline, points	24.6 (2.1)	24.4 (2.1)	25.2 (2.3)	23.6 (1.5)	0.002	24.0 (1.9)	24.7 (2.1)	23.1 (1.4)	0.015
CDR-SB score at baseline	2.9 (1.3)	2.9 (1.3)	2.6 (1.3)	3.1 (1.1)	0.082	3.2 (1.1)	3.2 (1.2)	3.4(0.9)	0.609
MMSE score at 6 months, points		22.1 (3.2)	23.6 (3.2)	20.8 (2.4)		22.5 (3.0)	23.9 (2.8)	20.8 (2.5)
CDR-SB score at 6 months		3.3 (1.4)	3.0 (1.4)	3.5 (1.1)		3.4 (1.3)	3.4 (1.5)	3.5 (0.9)
MMSE score at 12 months, points						22.5 (2.9)	23.4 (2.8)	21.4 (2.8)
CDR-SB score at 12 months						3.4 (1.3)	3.3 (1.5)	3.4(1.0)

*CSF could not be obtained from one patient, so the number of patients in the low/high ptau groups is one less than the total

Abbreviations: APOE4 apolipoprotein E ε4, CDR clinical dementia rating, CSF cerebrospinal fluid, MMSE Mini-Mental State Examination, ptau phosphorylated tau, SB sum of boxes and VSRAD voxel-based specific regional analysis system for Alzheimer’s disease

By October 2025, seventy-one patients were surveyed through brain MRI prior to the fifth, seventh, and 14th infusions. Among the 71 patients, 12 (16.9%) developed ARIA (i.e., three with ARIA-E and nine with ARIA-H), and all patients were almost asymptomatic (Fig. 2). One patient with ARIA-E reported very mild dizziness [17]; another patient with ARIA-E could no longer copy the interlocking pentagons in the MMSE test, a task that she was able to do before. When we compared the patients’ demographics with and without ARIA, ARIAs were more common among the APOE4 carriers (22.0%) than among the noncarriers (4.7%, p = 0.034). A higher frequency was observed among the ε4 homozygotes (33.3%) than among the ε4 heterozygotes (18.4%) (Table 2). The residual analysis revealed that patients with ARIA-E were particularly prevalent among ε4 homozygotes (Table 2). Patients who developed ARIA-H were also significantly more likely to have two or more microbleeds and have grade 2 or more DWMH on the baseline MRI (Table 2). Next, we compared the centiloid scale of the amyloid PET, CSF-ptau181 levels, and coagulation and fibrinolysis tests at baseline between patients with and without ARIA. The baseline CSF-ptau181 levels of patients with ARIA were significantly greater than those of patients without ARIA (p = 0.046, Table 2). However, we found no significant differences in the centiloid scale of the amyloid PET or the coagulation and fibrinolysis tests. ROC analysis was performed to determine whether the baseline CSF-ptau181 data could predict the occurrence of ARIA. CSF-ptau181 levels could distinguish patients with ARIA from those without ARIA, with an area under the curve of 0.68 (95% CI: 0.52–0.84) [sensitivity: 0.83, specificity: 0.58, optimal cut-off value: 78.6]. Next, we divided the patients into high and low CSF-ptau181 groups according to the cutoff value of 78.6 pg/ml and estimated the adjusted odds ratios of incidence of ARIA according to the levels of CSF-ptau181. The associations of CSF-ptau181 levels with incidence of ARIA were significant even after adjusting for potential confounding factors (Table 3). Compared with those in patients without ARIA, the vWF levels in patients with ARIA significantly increased from baseline to 6 months (q value = 0.002, Supplementary Table 1). The other coagulation and fibrinolysis tests revealed no significant differences between the groups with and without ARIA.

Fig. 2.

Participants surveyed in terms of ARIA

Table 2.

Demographics of patients with and without ARIA

	ARIA(-)	ARIA(+)						p vs. All
		All		ARIA-E			ARIA-H
n	59		12		3	9
Demographics
Age, yr	72.4 (8.0)		72.9 (3.5)		70.0(2.0)	73.8 (3.4)		0.745
Education	13.0 (2.2)		13.7 (2.3)		14.0 (2.0)	13.6 (2.5)		0.318
Women, no (%)	39 (66.1)		7 (58.3)		3 (100)	4 (44.4)		0.742
APOE4 status, no (%)								0.034
Noncarrier	20 (33.9)		1 (8.3)		1	0
Heterozygoutes	31 (52.5)		7 (58.3)		0	7
Homozygoutes	8 (13.6)		4 (33.4)		2*a	2
Hypertension, no (%)	29 (49.2)		3 (25.0)		0	3 (33.3)		0.203
Amyloid PET, centiloids	55.0 (19.5)		48.8 (25.9)		67.3 (30.2)	42.7 (22.9)		0.351
CSF-ptau181, (pg/ml)	81.6 (33.9)		101.0 (34.1)		111.8 (43.2)	97.4 (32.8)		0.046
MMSE score	24.4 (2.0)		25.5 (2.7)		28.0 (1.7)	24.7 (2.5)		0.093
Head MRI findings
Microbleeds								0.023
0	43 (72.9)		5 (41.7)		2	3
1	11 (18.6)		3 (25.0)		0	3
> 2	5 (8.5)		4 (33.3)		1	3*b
PVH								0.360
Grade < 2	29 (49.2)		4 (33.3)		2	2
Grade > = 2	30 (50.8)		8 (66.7)		1	7
DWMH								0.060
Grade < 2	28 (47.5)		2 (16.6)		2	0
Grade > = 2	31 (52.5)		10 (83.4)		1	9*c

Abbreviations: APOE4 apolipoprotein E ε4, ARIA amyloid-related imaging abnormalities, DWMH deep white matter hyperintensity, PVH periventricular hyperintensity

*The residuals for conditions a, b, and c were statistically significant, with values of 2.4, 2.0, and 2.7, respectively

Table 3.

Demographics and cognitive tests results of patients at baseline and follow-up, who initiated Lecanemab treatment

	No. of ARIA cases/patients	Model 1 OR (95%CI)	Model 2 OR (95%CI)	Model 3 OR (95%CI)	Model 4 OR (95%CI)	Model 5 OR (95%CI)
Low ptau group	2/36	1.00 (reference)	1.00 (reference)	1.00 (reference)	1.00 (reference)	1.00 (reference)
High ptau group	10/34	8.69 (1.62–46.43)	7.83 (1.44–42.59)	7.79 (1.43–42.45)	19.98 (2.43–163.81.43.81)	18.39 (1.91–176.33.91.33)

Abbreviations: CI confidence interval, OR odds ratio, ptau phosphorylated tau. Model 1 was adjusted for age and sex. Model2 was adjusted for age, sex, and APOE4 status. Models 3, 4, and 5 were adjusted for global clinical dementia rating score, amyloid PET centiloid value, number of microbleeds on baseline MRI in addition to the covarites in Model 2, respectively

Regarding the efficacy of the lecanemab treatment, the numbers of patients who received their 6-, 12-, and 18-month cognitive assessments were 69, 38, and 5, respectively. The mean MMSE score at baseline was 24.4 points, which significantly decreased to 22.1 points after 6 months among the 69 patients who completed the 6-month follow-up (p < 0.001). When the patients were divided into high and low ptau groups according to the cutoff value which derived from ROC analysis for ARIA prediction, no significant differences were found in the demographics between the two groups, except for the CSF-ptau181 value and the MMSE score at baseline. The MMSE scores at baseline in the high ptau group were significantly lower than those in the low ptau group (p = 0.002). A significant time×group interaction, time, and group effects on MMSE score from baseline to the 6-month follow-up (F = 4.630, p = 0.035; F = 65.606, p < 0.001; F = 13.743, p < 0.001, respectively, Figs. 3A and B, Supplementary Table 2). The CDR-SB scores in the high ptau group were also greater than those in the low ptau group; however, no statistically significant difference was observed from baseline to the 6-month follow-up (Supplementary Figs. 1 A and B, Supplementary Table 2). The MMSE scores of the 38 patients who completed their 12-month follow-up assessments significantly decreased over time (p < 0.001). The analysis of MMSE score change from baseline to the 6- and 12-month follow-ups revealed significant time×group interaction, time, and group effects (F = 3.504, p = 0.040; F = 17.870, p < 0.001, F = 12.992, p < 0.001, respectively, Supplementary Figs. 2 A and B, Supplementary Table 2). No statistically significant difference was found with respect to the progress of the CDR-SB scores over time from baseline (Supplementary Figs. 3 A and B, Supplementary Table 1). Among the patients who initiated lecanemab treatment, 32.0% were APOE4 noncarriers, 52.0% were heterozygous, and 16.0% were homozygous (Table 1). The analysis revealed no significant time × APOE4 interaction, time, or APOE4 effects on the cognitive test scores (MMSE or CDR-SB) from baseline to the 6- and 12-month follow-up visits (Supplementary Table 3 A and 3B). A comparison of baseline characteristics between early and late enrollees among participants who completed the 6- or 12-month follow-up revealed no significant differences (Supplementary Table 4). None of the patients who completed the 6- and 12-month follow-up visits progressed to moderate or severe dementia. Longitudinal observations revealed that CSF-ptau181 levels significantly decreased over time from baseline to the 6-month follow-up (p < 0.001, Supplementary Table 5). Moreover, the longitudinal coagulation and fibrinolysis tests revealed a significant decrease in the plasma TM levels after 6 months of lecanemab treatment (Supplementary Table 5), but no longitudinal differences in the other coagulation and fibrinolysis factors were detected.

Fig. 3.

Longitudinal change in the MMSE scores according to the low and high ptau groups. The median CSF-ptau181 concentration was 78.6 pg/ml in our patients; therefore, we defined the patients with CSF-ptau levels above 78.6 pg/ml as the high ptau group and those with levels below 78.6 pg/ml as the low ptau group. The blue and red colors denote the low and high ptau groups, respectively. Panel A shows the group changes, whereas Panel B shows the individual changes at 6 months

During this period, six patients (6.0%) experienced infusion-related reactions, five of which were mild and occurred only after the first and/or second dose. One patient developed moderate skin swelling and redness after the first and second lecanemab doses. The symptoms improved within 1 week, but the patient requested the discontinuation of further treatments. Another patient experienced mild headache and fever due to infusion-related reactions and requested discontinuation (Fig. 1).

Discussion

This study revealed the following four points: (1) the CSF-ptau181 level at baseline was associated with cognitive decline and ARIA occurrence during lecanemab administration; (2) the plasma vWF levels increased in patients with ARIA, and changes in the coagulation cascade occurred; (3) the plasma thrombomodulin levels decreased after lecanemab administration, suggesting that lecanemab may have reduced vascular endothelial damage; and (4) the incidence of infusion-related reactions was relatively low, and premedication in all patients may have helped alleviate infusion-related reactions.

The patients treated with lecanemab had significantly lower MMSE scores at the 6- and 12-month follow-ups than at baseline. Compared with the low ptau group, the high ptau group at baseline had significantly lower MMSE scores. The CSF-ptau181 level is a predictive marker of cognitive decline in individuals with MCI due to AD and mild AD dementia [18–20]. For the first time, we showed herein that elevated CSF-ptau181 levels at baseline can predict future cognitive decline in patients with AD who are receiving lecanemab treatment. CSF-ptau181 is an amyloid-reactive tau marker that becomes abnormal at the same time as the amyloid PET and before the tau PET [21]. However, higher CSF-ptau181 levels predict a subsequent increase in tau accumulation, and CSF-ptau181 levels are correlated with tau accumulation progression on tau PET [8]. In other words, lecanemab may be more effective in slowing cognitive decline in patients with early AD than in those with advanced tau pathology.

We are also the first work to demonstrate that elevated CSF-ptau181 levels at baseline can predict the occurrence of ARIA within 6 months of lecanemab initiation. As reported previously, ARIA-E is associated with longitudinal changes in CSF-ptau, total tau, and Aβ40 levels in a phase 3 bapinezumab clinical trial [22]; however, no predictive biomarker for the occurrence of ARIA has been determined in advance. CSF-ptau181 levels are associated with soluble triggering receptors expressed on myeloid cell-1, which are provided by microglia and promote neuroinflammation [23]. Additionally, CSF-ptau181 levels are significantly associated with chitinase-3-like protein 1 and glial fibrillary acidic protein [24]; hence, CSF-ptau181 is thought to be a marker of glial hyperreactivity and neuroinflammation. ARIA was possibly more likely to occur in the brain of individuals where glial hyperreactivity and neuroinflammation are shown, as reflected by the high levels of CSF-ptau181.

The ARIA in patients with AD shares many pathophysiological features with CAA. The APOE4 carrier status, anti-amyloid antibody dosage, and the presence of any microbleeds on the baseline MRI are major risk factors for ARIA [25]. In our study, APOE4 homozygotes were significantly associated with ARIA-E, whereas a greater number of microbleeds was significantly associated with ARIA-H, which is consistent with previous reports [26]. Additionally, ARIA-E and ARIA-H were frequently distributed in the posterior circulation, similar to the findings of a previous phase 2 bapinezumab trial [27]. Baseline DWMH scores of 2 or higher were more common in patients with ARIA than in those without ARIA (Table 2), but the difference did not reach statistical significance. The ARIA rate in our patients (All: 16.9%; ARIA-E: 4.2%; ARIA-H: 12.6%) was similar to Asian regional analysis of Clarity AD (ARIA-E: 6.2%; ARIA-H: 14.4%) [28], and also similar to that reported in the real-world findings of Bregman N. et al. [6] (All: 18.6%), although they did not treat APOE4 homozygous individuals because of safety concerns.

We observed that the plasma vWF levels increased after the occurrence of ARIA in patients treated with lecanemab. vWF is thought to be a biomarker of endothelial damage and inflammation. It works when blood vessels are injured, and bleeding occurs. As reported, plasma vWF levels increase in cardiovascular diseases [29]. Therefore, we speculated that the increased plasma vWF levels reflected the vascular endothelial damage caused by ARIA.

We also found that plasma TM levels were significantly decreased after lecanemab treatment in patients with AD. Increased thrombomodulin levels [30] were found in patients with AD, although some reports have suggested no correlation [31]. TM levels are elevated when vascular endothelial damage is present. In patients with AD, increased fibrin (ogen) deposition is present in the brain parenchyma, and activated coagulation factors and platelets are present in the AD circulation [11, 32]. Our results are consistent with a potential association between lecanemab treatment and preservation of vascular endothelial function.

In our experience, infusion-related reactions occurred in only 6.0% of patients. The incidence of infusion-related reactions in our patients was lower than that reported in clinical trials (26.4 and 12.3%) [1, 28] and other real-world reports (22.1 and 37.0%) [5, 6]. We performed premedication in all patients administered lecanemab, which is thought to have contributed to the low incidence of infusion-related reactions.

There are some limitations in this study. First, the relatively small number of patients and the absence of a control group with real-world clinical data. Second, our patients were all Japanese, and this low diversity limited the generalizability of the obtained results. Third, the International Working Group-2024 (IWG-2024) criteria [33], which requires the presence of pathophysiological biomarkers, were considered desirable for accurate AD diagnosis and research purposes. We applied the NIA-AA clinical criteria in this study, and all patients fall into the AD continuum; however, all participants were amyloid PET–positive and 85.4% of participants were CSF‑ptau181–positive, therefore, the remaining 14.6% of participants did not meet the strict pathological definition by the IWG-2024 criteria. Fourth, Cognitive function was primarily assessed using the MMSE. Although the MMSE has limited sensitivity for detecting MCI [34], it remains widely used in clinical trials, including the Clarity AD trial. In Japan, an MMSE score of 22–30 is used as an eligibility criterion for initiating lecanemab; therefore, we adopted the MMSE in the present study.

The strength of this report is the real-world data analysis, which provides insights into the safety and feasibility of lecanemab use for patients with early AD in the hospital setting.

Conclusion

In conclusion, lecanemab was tolerated by most patients with early AD, and lecanemab treatment was associated with trends toward reduced vascular endothelial injury. Lecanemab might be more effective and safer, particularly in patients with low CSF-ptau181 levels. Elevated CSF-ptau181 levels predict the occurrence of ARIA within 6 months of lecanemab initiation. Measuring CSF-ptau181 levels may be useful for selecting patients susceptible to ARIA and for appropriate side effect monitoring.

Authors’ contributions

Moeko Noguchi-Shinohara (Conceptualization; Formal Analysis; Investigation; Writing - Original draft preparation)Takahiro Yoshinobu (Conceptualization; Investigation; Writing - Original draft preparation)Taro Ozaki (Conceptualization; Investigation)Daiki Muramatsu (Investigation)Ayano Shima (Investigation)Yasuhiro Sakashita (Investigation)Yasutake Tada (Investigation)Hiroki Yamaguchi (Investigation)Junji Komatsu (Investigation)Tokuhei Ikeda (Investigation)Kenjiro Ono (Conceptualization, Writing - Review & Editing; Supervision; Project administration).

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

Data is available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Medical Ethics Review Board of Kanazawa University (approval number 114625). Written informed consent was obtained from all patients who visited after the study protocol was approved by the Medical Ethics Review Board of Kanazawa University. For patients who visited before that, consent was obtained by opt-out on the website.

Consent for publication

Not applicable. This manuscript does not contain any individual person’s data in any form.

Competing interests

T.Y., T.O., D.M., A.S., Y.S., Y.T., H.Y., J.K., and T.I. have nothing to report. M. N‒S. and K.O. report receiving honoraria from Eisai.