ÌÇÐÄvlog

Key Points

QuestionÌý What are the risk factors for, characteristics of, and clinical consequences of amyloid-related imaging abnormalities-edema (ARIA-E) in phase 3 trials of gantenerumab in early Alzheimer disease (AD)?

FindingsÌý Severity of amyloid-beta (Aβ) neuropathology and comorbid cerebrovascular pathology may be important baseline risk factors for ARIA-E, in addition to apolipoprotein E ε4. ARIA-E had no long-term impact on cognitive/functional scale performance in most participants.

MeaningÌý These results demonstrate that clinicians may wish to consider potential risk factors for ARIA-E (eg, severity of Aβ neuropathology and comorbid cerebrovascular pathology) when prescribing anti-Aβ monoclonal antibodies for early AD and developing individualized safety monitoring plans.

ImportanceÌý Data from 2 phase 3 studies of gantenerumab, GRADUATE I/II, and their open-label extensions represent a resource to further characterize amyloid-related imaging abnormalities (ARIA), including long-term sequelae.

ObjectivesÌý To describe the characteristics of ARIA and risk factors and clinical consequences of ARIA-edema (ARIA-E).

Design, Setting, and ParticipantsÌý Secondary data collection from the GRADUATE I/II phase 3 randomized, double-blind, placebo-controlled, 116-week parallel-group studies and their open-label extensions, including PostGraduate, with up to 210 (mean, 125) weeks of total gantenerumab treatment were conducted between 2018 and 2023. The study included multicenter trials at 288 sites across 30 countries. GRADUATE I/II enrolled 985 and 980 participants, respectively, with early symptomatic Alzheimer disease (AD) and amyloid-beta (Aβ) pathology who were aged 50 to 90 years. PostGraduate enrolled 1382 participants (671 previously randomized to gantenerumab). Data were analyzed from November 2, 2022, to October 10, 2023.

InterventionsÌý GRADUATE I/II participants were randomized 1:1 to gantenerumab or placebo. Nine-month uptitration was used to mitigate ARIA risk.

Main outcomes and measuresÌý Postbaseline safety monitoring, including brain magnetic resonance imaging (MRI) findings, and adverse events and cognitive assessments.

ResultsÌý The safety-evaluable MRI population of GRADUATE I/II comprised 1939 participants (mean age, 71.7 years; 1105 female [57.0%]). Severity of AD–related Aβ neuropathology (lower cerebrospinal fluid [CSF] Aβ42, hazard ratio [HR] for CSF Aβ42: 0.4; 95% CI, 0.2-0.7) and comorbid cerebrovascular pathology (Fazekas score: HR, 1.6; 95% CI, 1.3-2.0; total superficial siderosis count: HR, 1.9; 95% CI, 1.3-2.6; total microhemorrhage count: HR, 1.3; 95% CI, 1.0-1.5) may be important baseline risk factors for ARIA-E, in addition to apolipoprotein E (APOE) ε4 status (APOE ε4 heterozygous carrier: HR, 2.0; 95% CI, 1.4-2.8 and APOE ε4 homozygous carrier: HR, 4.7; 95% CI, 3.2-6.7). At the group level, ARIA-E did not impact long-term cognitive and functional performance (relative difference in adjusted means for Clinical Dementia Rating–Sum of Boxes was −9% in pooled GRADUATE analysis at week 116 and when censored at first ARIA-E). While taking gantenerumab, ARIA-E and ARIA-hemosiderin occurred in 24.9% (247 of 993) and 22.9% (227 of 993) participants, respectively; first ARIA-E occurred by week 64 in 86.2% (213 of 247) of participants with ARIA-E. Narratives are provided for all serious symptomatic ARIA-E cases.

Conclusions and RelevanceÌý These results show that in addition to APOE ε4 allele count, severity of Aβ neuropathology and comorbid cerebrovascular pathology may be relevant for clinicians prescribing anti-Aβ monoclonal antibodies for early AD and developing individualized safety monitoring plans. Evaluation of these risk factors in other anti-Aβ monoclonal antibodies is recommended.

Trial registrationsÌý ClinicalTrials.gov Identifiers: , , .

Amyloid-beta (Aβ) processing and deposition play a critical role in the Alzheimer disease (AD) pathogenesis.¹ Anti-Aβ monoclonal antibodies (mAbs) lecanemab and donanemab demonstrated significant reductions vs placebo in clinical progression of early symptomatic AD (mild cognitive impairment due to AD and mild AD) in phase 3 clinical trials, in parallel with significant Aβ plaque clearance.^2,3 These drugs are US Food and Drug Administration licensed for the treatment of early symptomatic AD.^4-6

Amyloid-related imaging abnormalities (ARIA) comprise vasogenic edema and/or sulcal effusions (ARIA-edema [ARIA-E]; best visualized on magnetic resonance imaging [MRI] fluid-attenuated inversion recovery sequences) and microhemorrhages and/or superficial siderosis (SS) (ARIA-hemosiderin [ARIA-H]; on iron-sensitive MRI sequences, such as gradient-recalled echo/T2*-weighted or susceptibility-weighted imaging).^7,8 While ARIA may develop spontaneously in relation to AD or cerebral amyloid angiopathy (CAA), it is also the main adverse effect of anti-Aβ mAbs (absolute rates vary between molecules)^2,3,9-15 and was reported for a microglial-receptor triggering receptor expressed on myeloid cells 2–binding mAb in AD.¹⁶ ARIA pathophysiology hypotheses include Aβ clearance-related temporarily increased cerebral vascular permeability^10,12; reactive immune cell–mediated inflammation,¹⁷ and complement cascade activation by anti-Aβ mAb-vascular Aβ complexes.¹⁸

Since ARIA represents a class effect of first-generation anti-Aβ mAbs, understanding ARIA risk factors and symptomatology may inform future clinical decision-making and MRI monitoring. It is also important for ensuring individuals with AD are informed of anti-Aβ mAbs risks.

Key efficacy and safety outcomes from 2 global phase 3 gantenerumab studies, GRADUATE I and II (/), and their open-label rollover studies, including PostGraduate (), have been reported previously.^19-21 Although gantenerumab was not associated with a significant reduction in clinical decline in participants with early AD (difference in Clinical Dementia Rating–Sum of Boxes [CDR-SB] at week 116 vs placebo: GRADUATE I, −0.31; 95% CI, −0.66 to 0.05 and GRADUATE II, −0.19; 95% CI, 0.55-0.17; favoring gantenerumab), these studies provide a rich resource to further characterize ARIA and its longer-term consequences.²² While in GRADUATE I/II incidence of ARIA-H was higher in gantenerumab than placebo arms, incidence of new isolated ARIA-H was comparable,²² similar to observations for other in-class mAbs^3,13; the higher incidence of ARIA-H in the active arms is driven by ARIA-H concurrent with ARIA-E. As such, while we report key ARIA-H characteristics, we focus on the risk factors, descriptive characteristics, and clinical consequences of ARIA-E.

The design of the identical phase 3 randomized, double-blind, placebo-controlled, parallel-group GRADUATE I and II trials has been previously published.²² Briefly, GRADUATE I and II enrolled 985 and 980 participants, respectively, with early symptomatic AD and confirmed Aβ pathology who were aged 50 to 90 years. More than 5 microhemorrhages and/or focal SS areas (though also more than 3 focal SS areas), overall Fazekas score of 3, and anticoagulation constituted exclusion criteria. Single- or dual-antiplatelet therapy was allowed. PostGraduate () was an open-label, multicenter, rollover study evaluating the long-term safety, tolerability, and efficacy of gantenerumab in participants from the GRADUATE I/II trials. Trial conduct followed the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use E6 guideline for Good Clinical Practice and the Declaration of Helsinki and applicable laws and regulations. An independent data monitoring committee provided oversight. All trial participants provided written informed consent.

In GRADUATE I/II, participants were randomized 1:1 to subcutaneously administered gantenerumab or placebo for up to 116 weeks with a 9-month uptitration regimen to the 510-mg every 2 weeks target dose to mitigate dose-dependent ARIA risk.^23,24 Safety MRIs were performed at baseline, prior to each dose uptitration, regularly on target dose, and as determined by investigators and evaluated by a single central neuroradiologist (Figure 1). ARIA dosing intervention rules are provided in eFigure 1 in Supplement 1. Participants followed an identical dosing regimen and MRI frequency regardless of APOE ε4 status. ARIA-E radiological severity was determined using Barkhof Grand Total Score ([BGTS] range, 0-60).²⁵

After 116 weeks of double-blind treatment, eligible participants could receive open-label gantenerumab, initiated under the GRADUATE protocol or after directly entering PostGraduate (eFigure 2 in Supplement 1) with up to 210 (mean, 125) weeks of gantenerumab treatment in participants randomized to gantenerumab in the double-blind phase. Otherwise, 14- and 50-week postdosing follow-up was required.

Safety outcomes included adverse events (AEs), serious AEs, and MRI findings. Since only selected ARIA MRI findings were reportable as AEs (ie, symptomatic, leading to dosing intervention, or otherwise clinically significant), ARIA analyses reported are based on the MRI safety-evaluable dataset. Symptomatic ARIA-E was defined by central nervous system (CNS) AE(s) temporally associated with ARIA-E MRI findings, regardless of the investigator’s causality attribution.

Efficacy scales, including CDR-SB, the Alzheimer’s Disease Assessment Scale, Cognition Subscale 13 (ADAS-Cog13), Alzheimer’s Disease Cooperative Study Group Activities of Daily Living (ADCS-ADL), and Functional Activities Questionnaire (FAQ), and amyloid positron emission tomography, volumetric MRI, and plasma and cerebrospinal fluid (CSF) biomarkers were collected and reported. Measures were implemented to mitigate ARIA-related bias in efficacy assessments.²²

As GRADUATE I/II were identically designed with efficacy and safety results concordant,²² the pooled dataset from the double-blind study periods was used that included all participants who received at least 1 study drug dose, followed by at least 1 MRI (safety-evaluable MRI population). Postbaseline data covered the double-blind period from the first study drug dose up to 14 weeks after the last dose, but no later than the day before the first open-label gantenerumab dose for participants who entered the open-label period. Data from the open-label period and PostGraduate were used to repeat the assessment of time to first ARIA-E event in participants in the gantenerumab arm in the double-blind period. Analyses were prespecified in a statistical analysis plan, except the analyses of baseline risk factors. Additional descriptive ARIA analyses are provided beyond those reported previously,²² including stratification by APOE ε4 allele count or baseline microhemorrhage/SS presence. We used SAS version 9.4 (SAS Institute) r 4.3.1 (The R Project for Statistical Computing) for analysis.

For the analysis of risk factors for ARIA-E/ARIA-E with temporally concurrent new ARIA-H, a broad set of baseline variables for testing was developed (eMethods in Supplement 1). For the assessment of the long-term impact of ARIA-E on CDR-SB (pooled GRADUATE dataset, reported previously by study¹⁷), ADAS-Cog13, ADCS-ADL, and FAQ, a mixed-effect model of repeated measures was used, with all outcome data included as well as with outcome data censored upon first ARIA-E.

The safety-evaluable MRI population of the GRADUATE trials comprised 1939 participants (placebo, 946; gantenerumab, 993). Baseline characteristics were comparable between treatment groups, with a mean age of 71.7 (SD, 7.7) years (eTable 1 in Supplement 1). Most participants were female (1105 of 1939 [57.0%]), had mild cognitive impairment (1062 of 1939 [54.8%]), and received symptomatic AD treatment (1240 of 1939 [64.0%]). Half of the participants were heterozygous APOE ε4 carriers (964 of 1939 [49.7%]), followed by noncarriers (645 of 1939 [33.3%]) and homozygous carriers (330 of 1939 [17.0%]). At baseline, 97 of 993 participants (9.8%) in the gantenerumab arm had microhemorrhages and/or SS (80 of 993 [8.1%] had 1 to 5 microhemorrhages and 25 of 993 [2.5%] 1 to 3 focal areas of SS). Participant flow and baseline demographics/characteristics for PostGraduate are provided in eFigure 3 and eTable 2 in Supplement 1.

In univariate models, higher APOE ε4 allele number, higher Fazekas score, and lower CSF Aβ42 were associated with increased risk of ARIA-E (Figure 2A and B) and ARIA-E with concurrent ARIA-H (Figure 2C), and remained significant following Bonferroni correction. In proportional hazards modeling, higher number of SS at baseline was associated with higher risk of ARIA-E (Figure 2A). In models of ARIA-E with concurrent new ARIA-H, larger baseline numbers of microhemorrhages and/or SS and microhemorrhages alone were associated with increased risk (Figure 2C; descriptive statistics for all 57 potential baseline risk factors in eTable 3 in Supplement 1).

Multivariate models were built from variables with an unadjusted P value of less than .05 to optimize model fit; retained variables contain additional information on ARIA-E risk relative to other variables in the model. Variables remaining in multivariate models of ARIA-E risk following stepwise selection are shown in Figure 3. For multivariate model results for ARIA-E with concurrent ARIA-H, see eTable 4 in Supplement 1.

In the GRADUATE gantenerumab dataset, ARIA-E and ARIA-H occurred in 247 of 993 (24.9%) and 227 of 993 (22.9%) participants, respectively.²² Overall incidences of ARIA-E and symptomatic ARIA-E at the study-population level increased with the APOE ε4 count.²² This pattern was also observed for ARIA-E with new concurrent ARIA-H and recurrent ARIA-E (Table 1). However, within individuals who developed ARIA-E, there was no association between the number of APOE ε4 copies and symptomatic or serious symptomatic ARIA-E rate. The characteristics of serious symptomatic ARIA-E cases are presented in eTable 5 and eMethods in Supplement 1.

In the double-blind period of the GRADUATE trials, the mean time to first ARIA-E in gantenerumab-treated participants was 35.7 (SD, 13.0) weeks and 49.2 (SD, 21.4) weeks in participants with and without baseline microhemorrhages and/or SS, respectively. A numerically shorter time to first ARIA-E onset and increased mean radiological severity, resolution time, and likelihood of recurrence in those who had an opportunity for recurrence (ie, initial episode’s resolution followed by gantenerumab dosing and MRI) were observed in APOE ε4 homozygotes compared with noncarriers and heterozygotes (Table 1). In those who developed ARIA-E, homozygotes were more likely to develop concurrent microhemorrhages but not SS compared with heterozygotes and noncarriers (Table 1). Most gantenerumab-treated participants with ARIA-E had their first ARIA-E episode by week 64 (213 of 247 [86.2%]). However, first ARIA-E occurred up to week 116 and, including the open-label period, first ARIA-E occurred up to week 168, albeit at lower incidence than in the double-blind period (eFigures 4 through 7 in Supplement 1 including by APOE ε4 stratifications).

In the GRADUATE gantenerumab dataset, concurrent new ARIA-H was reported for 142 of 324 participants (43.8%) with asymptomatic ARIA-E and 36 of 56 (64.3%) with symptomatic ARIA-E. The mean radiological ARIA-E severity of the most severe episode per participant in BGTS was 9.0 (SD, 6.7), 17.5 (SD, 12.0), and 21.3 (SD, 8.3) for asymptomatic, nonserious symptomatic, and serious symptomatic ARIA-E, respectively. Of participants with ARIA-E who had an opportunity for ARIA-E recurrence in the double-blind period, 15 of 21 (71.4%) with and 80 of 173 (46.2%) without baseline microhemorrhages and/or SS had recurrent ARIA-E.

In gantenerumab-treated participants in the double-blind period who had recurrent ARIA-E, the mean radiological severity of the first ARIA-E was 9.8 (SD, 7.8) vs 9.3 (SD, 7.1) for the second ARIA-E episode. There was no evidence for a change in symptomatic status between first and second ARIA-E episode (eTable 6 in Supplement 1).

Of gantenerumab-treated participants, 11 of 993 had serious symptomatic ARIA-E (1%). The most common symptoms were confusion, headache, and aphasia; less common symptoms included seizures and status epilepticus (eTable 5 in Supplement 1). Eight of 11 participants were treated with corticosteroids. None received thrombolytics. There were no ARIA-related fatalities. Summaries for all participants who experienced serious symptomatic ARIA-E (involving hospitalization) are reported in the eMethods in Supplement 1. APOE ε4 status did not substantially impact the incidence of serious symptomatic ARIA-E (Table 1).

In the GRADUATE dataset, at week 116, comparable results were observed from models including all available efficacy measure outcome data vs models where available outcome data were censored at first ARIA-E, as measured by difference and relative difference in adjusted means on any of the efficacy scales tested (Table 2).

ARIA is an important adverse effect of anti-Aβ mAbs tested in AD, including gantenerumab,^2,3,9-13 with some of these agents available in clinical practice for the treatment of early symptomatic AD.⁴

Reports describing and characterizing ARIA with anti-Aβ mAbs have previously been published.^13,26-28 As noted in the Introduction, findings from studies of gantenerumab and other in-class mAbs suggest that the imbalance in ARIA-H between active treatment and placebo arms is driven by ARIA-H co-occurring with ARIA-E.^3,13,22 For this reason, we focused on the risk factors, descriptive characteristics, and clinical consequences of ARIA-E associated with gantenerumab treatment in the GRADUATE studies.

Higher anti-Aβ mAb dose, increased number of APOE ε4 alleles, and microhemorrhage presence are well established as nonmodifiable baseline risk factors for ARIA-E.^7,13,29-33 Analyses of the GRADUATE dataset, including 57 baseline variables of interest—to our knowledge, the broadest published set of baseline variables assessed—confirmed the increasing APOE ε4 number as a significant baseline ARIA-E risk factor. Additionally, the number of SS, higher Fazekas score, and lower CSF Aβ42 emerged as potential baseline risk factors for ARIA-E. Total microhemorrhage and SS number, higher amyloid burden on positron emission tomography, more cardiovascular risk factors, lower hippocampal volume, female sex, and higher CSF phosphorylated tau showed some association that, while not statistically significant when corrected for multiple comparisons, should be explored further. Overall, these findings suggest that severity of AD amyloid-related neuropathology (lower CSF Aβ42)³⁴ and comorbid cerebrovascular pathology (Fazekas score, total SS, and total microhemorrhage counts) may be important ARIA-E risk factors in addition to APOE ε4 number.

Multiple mechanisms have been proposed to explain the association between the APOE ε4 allele and increased ARIA-E risk: (1) reduced cerebrovascular integrity, (2) increased neuroinflammation and immune dysregulation, and (3) elevated levels of CAA.³⁵ Total SS is also associated with CAA,³⁶ while white matter hyperintensities may reflect reduced cerebrovascular integrity³⁷ or CAA, particularly when in a multispot pattern,³⁶ suggesting these risk factors mediate ARIA risk through shared mechanisms.

Increasing number of APOE ε4 alleles was also associated with higher incidence of ARIA-E with new ARIA-H. Numerically shorter time to first ARIA-E onset and increased mean radiological severity, resolution time, and likelihood of recurrence (in those with a recurrence opportunity) were observed in APOE ε4 homozygotes compared with heterozygotes and noncarriers; apart from similar findings for donanemab concerning ARIA-E radiological severity,² these associations between ARIA-E characteristics and number of ε4 copies have not been published for other mAbs. The apparent lack of impact of 1 APOE ε4 copy on these ARIA-E features vs the potential impact of 2 APOE ε4 copies requires further investigation.

Approximately 1 in 5 individuals with ARIA-E had temporally associated CNS symptoms, most commonly headache and dizziness.²² Although incidence of any ARIA-E and symptomatic ARIA-E increased with APOE ε4 allele count, despite APOE ε4 allele count being associated with greater radiological severity, among individuals who developed ARIA-E, there was a similar rate of symptomatic cases across APOE ε4 allele counts. Notably, two-thirds of symptomatic ARIA-E events, about 50% more than of asymptomatic ARIA-E events, co-occurred with new ARIA-H. However, greater radiological severity of symptomatic ARIA-E, and ARIA-E with concurrent new ARIA-H (eTable 7 in Supplement 1) makes it difficult to determine whether the relationship with symptomatology is directly or indirectly related to greater radiological severity.

The specific ARIA incidence varies between anti-Aβ mAbs (ARIA-E, 13% to 35% and symptomatic ARIA-E, 3% to 9%; ARIA-H microhemorrhage, 14% to 27% and SS, 6% to 16%)^2,3,13,22 due to intrinsic molecular characteristics, targeted amyloid species, trial designs, and population characteristics, including baseline amyloid positron emission tomography burden and distribution, or definition of symptomatic ARIA-E. However, generally similar patterns regarding rates according to APOE ε4 status, rates of symptomatic ARIA-E within overall ARIA-E rates, or imbalances in ARIA-H driven by those with ARIA-E have been reported for these mAbs, including gantenerumab.^2,3,13,15,22

Most gantenerumab-treated participants with ARIA-E had their first ARIA-E episode within 16 months of initiating gantenerumab; however, first ARIA-E cases also occurred much later, up to 42 months from treatment initiation. This was later than for other mAbs^5,6,13,15,38 and unexpected based on previous open-label studies with a similar gantenerumab dosing regimen³⁹ and could be due to the longer titration period and slower and lower than predicted Aβ removal in the GRADUATE studies. Baseline microhemorrhages and/or SS were associated with an approximately 3-month earlier onset of first ARIA-E.

Radiological ARIA-E severity on gantenerumab was approximately 2 times higher in participants with CNS symptoms than in those without and higher with serious than nonserious symptomatic ARIA-E. However, there also was a significant overlap between the groups and large variability in radiological-clinical relationship, with some radiologically severe ARIA-E cases remaining asymptomatic. Further work is required to understand the relationship between radiological features and symptomatology, including lesion location. Radiological severity did not increase in recurrent events, although this analysis carries a risk for retention bias.

There was no evidence for ARIA-E impact on longitudinal cognitive (CDR-SB, ADAS-Cog13) and functional (ADCS-ADL, FAQ) outcomes at the group level, supporting that ARIA-E do not generally result in long-term detrimental effects on cognition or function. However, ARIA-E might have long-term sequelae in individual cases (eMethods in Supplement 1). For example, in a participant who experienced a serious symptomatic ARIA-E, a residual significant decline in activities of daily living following ARIA-E resolution was considered by the investigator as a potential ARIA-E sequela, albeit confounded by significant Mini-Mental State Examination decline trajectory that started before ARIA-E onset. The lack of impact of APOE ε4 status on serious symptomatic ARIA-E incidence in this series differs from other reports with serious events primarily in APOE ε4 carriers^2,40 and may be due to the limited number of serious events.

While serious symptomatic ARIA-E remains understudied due to its relative rarity, to our knowledge, ours is the first publication that provides case narratives for 11 such cases (eMethods in Supplement 1). Several of the cases presented with temporal focal neurologic deficits. Since the initial workup in individuals on anti-Aβ mAbs exhibiting such symptoms may focus on and lead to an inadvertent diagnosis of suspected ischemic stroke and treatment with thrombolytics, it is critical that emergency clinicians are aware that patients are receiving an Aβ-lowering antibody, ARIA is an important differential diagnosis, computed tomography is insensitive in detecting ARIA and MRI is required, and use of thrombolytics should be restricted unless ARIA is excluded. Although GRADUATE I/II were not designed to assess the safety of a treatment rechallenge post–ARIA, treatment reintroduction following severe radiological/symptomatic ARIA should be considered carefully.

Limitations of our study include the unknown generalizability of our findings to a real-world early symptomatic AD population given the underrepresentation of racial and ethnic minorities and exclusion of certain individuals (eg, with more than 5 microhemorrhages and/or focal SS areas, though also more than 3 focal SS areas, significant white matter pathology, unstable or clinically significant cardiovascular disease, individuals on anticoagulation) from GRADUATE I/II. The number of symptomatic and serious symptomatic ARIA-E events is low; therefore, conclusions should be interpreted cautiously. Time to first ARIA-E-event analyses relied on routine MRI time points (although investigators could perform unscheduled MRIs), which might lead to delayed recording of ARIA-E onset. Additionally, PostGraduate was terminated early after the GRADUATE studies missed their primary end point. ARIA-E severity was expressed in BGTS because that determined clinical actions in the GRADUATE trials, but BGTS does not map 1:1 to other scales.⁴¹ Although ARIA is a class effect of first-generation anti-Aβ mAbs, these findings should be considered molecule specific and requiring further evaluation for other in-class molecules, particularly concerning ARIA-E baseline risk factors.

While gantenerumab lacked sufficient clinical benefit for continued development, Roche is developing trontinemab, a novel, distinct monoclonal antibody. Trontinemab uses a Brainshuttle technology to gain enhanced brain access via transferrin receptor 1–mediated transcytosis at the capillary level, in combination with a gantenerumab-derived immunoglobulin G framework as an anti-Aβ binder.⁴² It is hypothesized that the ability to use low doses and to deliver the therapy more directly into the brain parenchyma may result in additional safety benefits, such as reduced ARIA risk. This is supported by preliminary results from an ongoing phase 1b/2a study () suggesting rapid and robust amyloid lowering by trontinemab with low ARIA incidence.⁴³

Data from 1939 participants in 2 phase 3 studies with gantenerumab in early symptomatic AD allowed for characterization of ARIA-E with gantenerumab. Severity of AD–related amyloid neuropathology (lower CSF Aβ42) and comorbid cerebrovascular pathology (Fazekas score, total SS, and total microhemorrhage counts) may be important baseline ARIA-E risk factors, in addition to APOE ε4 allele count, and, in part, may increase risk because of higher baseline CAA burden. Symptomatic ARIA-E events tended to be radiologically more severe and were more frequently associated with ARIA-H vs asymptomatic cases. Although APOE ε4 allele count is associated with increased ARIA-E risk and increased radiological severity, it did not appear to influence symptomatic status among those who developed ARIA-E. In most participants, ARIA-E did not have a long-term impact on cognitive and functional performance.

These findings may be considered in clinical practice when prescribing anti-Aβ mAbs for early AD and developing individualized safety monitoring plans. Further evaluation of the identified potential ARIA-E risk factors in the context of other anti-Aβ mAbs is recommended.

Accepted for Publication: September 19, 2024.

Published Online: November 18, 2024. doi:10.1001/jamaneurol.2024.3937

Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2024 Salloway S et al. JAMA Neurology.

Corresponding Author: Stephen Salloway, MD, Warren Alpert Medical School, Brown University, 345 Blackstone Boulevard, Providence, RI 02906 (ssalloway@carene.org); Jakub Wojtowicz, MPharm, F. Hoffmann-La Roche, Pharma Development Safety Risk Management, Bldg 2, Floor 36, Grenzacherstrasse 124, 4070 Basel, Switzerland (jakub.wojtowicz@roche.com).

Author Contributions: Dr Voyle and Ms Rossomanno had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Wojtowicz, Voyle, Lane, Klein, Lyons, Rossomanno, Mazzo, Barkhof, Schneider, Grundman, Aldea, Boada, Smith, Doody.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Salloway, Wojtowicz, Voyle, Lane, Lyons, Rossomanno, Schneider, Grundman, Boada, Smith.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Voyle, Klein, Rossomanno.

Obtained funding: Smith.

Administrative, technical, or material support: Lane, Klein, Lyons, Barkhof.

Supervision: Bullain, Grundman, Boada, Smith, Doody.

Other - data visualization, analysis and interpretation: Aldea.

Conflict of Interest Disclosures: Dr Salloway reported grants and personal fees from Biogen, Lilly, Roche, Eisai, Novartis, and NovoNordisk, and personal fees from Prothena, AbbVie, Acumen, CognitionRX, and Kisbee during the conduct of the study. Mr Wojtowicz reported employment and shareholding from F. Hoffmann-La Roche. Dr Lane reported employment and shareholding from Roche Products during the conduct of the study. Dr Klein reported employment and shareholding from F. Hoffmann-La Roche. Dr Lyons reported employment and shareholding from Roche Products. Ms Rossomanno reported employment and shareholding from F. Hoffmann-La Roche outside the submitted work. Dr Mazzo reported employment and shareholding from F. Hoffmann-La Roche during the conduct of the study. Dr Bullain reported personal fees from F. Hoffmann-La Roche during the conduct of the study and personal fees from Biogen International outside the submitted work. Dr Barkhof reported personal fees from Biogen, Prothena, Eisai, Roche, IXICO, Combinostics, and Merck, and grants from Roche and Biogen outside the submitted work. Dr Bittner reported full-time employment and shareholding from F. Hoffmann-LaRoche. Dr Schneider reported employment and shareholding from F. Hoffmann La Roche outside the submitted work. Dr Grundman reported personal fees from Global R&D Partners. Dr Aldea reported a US patent issued for 18/140,583 and employment and stock/stock options from F. Hoffmann-La Roche. Dr Boada reported personal fees from Grifols, Araclon Biotech, Roche, Biogen, Lilly, Merck, Zambon, and Novo-Nordisk; holding advisory board memberships with Grifols, Roche, Lilly, Araclon Biotech, Merck, Zambon, Biogen, Novo-Nordisk, Bioiberica, Eisai, Servier, and Schwabe Pharma; and lecture fees from Roche, Biogen, Grifols, Nutricia, Araclon Biotech, Servier, and Novo-Nordisk outside the submitted work. Dr Smith reported employment and share options from Roche Products outside the submitted work. Dr Doody reported employment from F. Hoffman-LaRoche during the conduct of the study. No other disclosures were reported.

Funding/Support: The GRADUATE I and II and PostGraduate studies were funded by F. Hoffmann-La Roche, Basel, Switzerland. The sponsor, F. Hoffmann-La Roche, was responsible for the design and conduct of the GRADUATE I and II and PostGraduate studies.

Role of the Funder/Sponsor: The sponsor and authors shared equal responsibility for the collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We thank all the participants, study partners and their families, the investigators and site staff, and the entire study team for their time and commitment to the GRADUATE I and II and PostGraduate studies. We also thank Sophie Banzet, MD, F. Hoffmann-La Roche and Yahveth Cantero-Fortiz, MD, MSc, Ace Alzheimer Center Barcelona (no compensation from F. Hoffmann-La Roche received) for review comments, Chris Pelentrides, MSc, F. Hoffmann-La Roche and Richard Plank, BSc, Parexel International, for support with statistical output generation, Christopher Galli, MS, (employee of F. Hoffmann-La Roche at the time of the study) for curation of magnetic resonance images, and Jens Wuerfel, MD, F. Hoffmann-La Roche, for magnetic resonance imaging image review. Editorial support for the development of this manuscript was provided by Joshua Quartey, BSc, of Nucleus Global, funded by F. Hoffmann-La Roche.