Balancing benefit and risk in early Alzheimer's disease: the European Medicines Agency (EMA) assessment of lecanemab and donanemab

Summary

Alzheimer's disease (AD) remains the leading cause of dementia in Europe, with limited therapeutic options despite decades of research. Recent regulatory approvals of lecanemab and donanemab in the EU mark a significant milestone. Both antibodies, developed within the amyloid hypothesis, demonstrated statistically significant effects on cognitive and functional outcomes in patients with early AD, though the clinical meaningfulness of these effects remains debated. There are also important safety concerns, particularly amyloid-related imaging abnormalities (ARIA), which resulted in difficult benefit-risk assessments. The Personal View describes some of the lines of evidence considered by the European Medicines Agency in the assessment of efficacy of these medicines, and some of the measures to ensure their safe use. While these therapies do not halt disease progression, and have substantial risks, they represent a step toward more transformative treatments. Future research, including real-world evidence studies, should refine patient selection, optimise monitoring, and explore combination approaches and new targets to achieve improved outcomes.

Introduction

Alzheimer's disease (AD) is the leading cause of dementia, affecting nearly 8 million people in the European Union as of 2021,¹ and this number is increasing.² It is reported as the most feared disease for people over the age of 50³ and, despite decades of research, treatments that slow down the progression of AD have proven elusive.⁴ Patients with AD go through a pre-symptomatic phase where pathological changes accumulate,⁵ progressing to a stage of mild cognitive impairment (MCI) that then worsens into dementia with increasing, and ultimately complete, loss of independence and of the ability to interact socially. Behavioural and psychological symptoms heavily increase the disease burden for both patients and caregivers, and emotional and economic burdens on patients, caregivers and society are massive.^6,7

The dominant research programme^8,9 studying AD pathophysiology has been the amyloid hypothesis.10, 11, 12 While it is possible to give a general formulation of the amyloid hypothesis as the view that amyloid deposition is a central event in the pathogenesis of the disease, causing further changes and ultimately clinical manifestations, its several auxiliary hypotheses, especially on the potential mechanisms mediating this causation, lead to potentially distinct predictions on efficacy of different therapeutic approaches. While a part of the scientific community has historically been and continues to be highly critical of the amyloid hypothesis,13, 14, 15, 16, 17 most molecules recently investigated for AD have been developed within the programme.^18,19

While some of the molecules have had mixed²⁰ or negative21, 22, 23, 24, 25 results, study results for lecanemab²⁶ and donanemab²⁷ have been positive. While broadly developed within the amyloid hypothesis programme, the two monoclonal antibodies have different targets: lecanemab preferentially targets soluble aggregated Aβ protofibrils,²⁸ while donanemab primarily targets the plaques.²⁹

In the European Union (EU), lecanemab and donanemab have been authorised in 2025—after re-examination procedures (see Fig. 1)—for the treatment of mild cognitive impairment or mild dementia due to Alzheimer's disease (early Alzheimer's disease) in patients who are not homozygotes for ApoE ε4. Globally, lecanemab and donanemab have been approved for use by several other regulatory authorities. In this paper we summarise the elements—more comprehensively stated in the European Public Assessment Reports^29,30—that led to their EU marketing authorisations and shortly outline some considerations on the future of the treatment of AD.

In the EU, medicines developers can apply for a central marketing authorisation, which when granted allows them to market their product in all EU Member States. Such an application is evaluated by EMA's Committee for Medicinal Products for Human Use (CHMP), who assess the data package (dossier) submitted by the developer to determine if the medicine meets the requirements for quality, safety and efficacy and if the benefits of the medicine outweigh its risks. A positive recommendation requires that an absolute majority of CHMP's members considers that the benefits of the medicine outweigh its risks.

EMA received marketing authorisation applications for the treatment of adults with early Alzheimer's disease with confirmed amyloid pathology in January and July 2023 for—respectively—lecanemab (Leqembi) and donanemab (Kisunla).

Both applicants had previously asked in multiple procedures for Scientific Advice, in which EMA's Scientific Advice Working Party (SAWP) provided non-binding advice on the most appropriate methodology for generating robust clinical evidence to support the application.

As for all applications evaluated by the CHMP, Rapporteurs were appointed—following objective criteria which aim to ensure the use of the best available expertise—to perform the scientific evaluations and present their positions to the Committee. All other member states contributed to the assessment by providing comments (in writing and/or in the plenary). The evaluations of lecanemab and donanemab followed a defined process whereby the CHMP asked the applicants for both medicines to clarify certain aspects of their application or provide additional analyses. To further address uncertainties about the benefits and risks of the medicines, the CHMP also consulted the Pharmacovigilance Risk Assessment Committee (PRAC) who provided the assessment of aspects concerning safety and risk management, in particular regarding ARIA. CHMP also requested specialised advice from the Neurology scientific advisory group (SAG-N), a group of independent European experts (comprising a core group to which additional experts can be added if specific expertise is needed) appointed by the CHMP to provide an expert position on procedures that require external specialist input. In the case of lecanemab and donanemab, the experts were asked about the clinical relevance of treatment benefits seen in the studies and the management of risks. Because CHMP had remaining major objections at the end of its evaluation for both products, the applicants were invited to address these directly with the CHMP in an oral explanation. The applicants of both lecanemab and donanemab provided oral explanations to address outstanding concerns.

On completing its initial evaluation, the CHMP issued a recommendation to the European Commission to refuse a central marketing authorisation for both lecanemab (in July 2024) and donanemab (in March 2025), considering at that time that the demonstrated benefits of the medicines did not outweigh the risks.

It is possible for applicants to appeal the CHMP's opinion, and accordingly the applicants for lecanemab and donanemab asked for CHMP to re-examine its initial recommendation before the European Commission issued a binding decision. The applicants provided specific grounds for the CHMP to reconsider its opinion during re-examination, addressing concerns regarding the clinical relevance of measured benefits and the risk of ARIA.

During the re-examination, the CHMP, in a process led by new sets of Rapporteurs, assessed the grounds provided by the applicant. The CHMP also considered proposed risk minimisation measures, strategies to manage the risk of ARIA in patients taking the medicine. These measures included educational materials for healthcare professionals and patients to inform them of the risks and how to mitigate them, and a controlled access programme implemented by EU Member States wherein treatment is only provided by specific centres in full accordance with the clinical particulars laid out in the summary of product characteristics to ensure safe and effective use in the indicated population only.

Following re-examination of both lecanemab (in November 2024) and donanemab (in July 2025), considering the risk minimisation measures presented, the CHMP considered that the benefit-risk balance of both medicines was favourable, and recommended granting of a marketing authorisation subject to certain conditions.

Details of the evaluation and re-examination of the marketing authorisation applications for lecanemab and donanemab are presented in much greater detail in the European public assessment reports published on the website of the European Medicines Agency.^29,30 For all medicines evaluated by EMA, clinical study reports and protocols, as well as other information supporting the applications are published on the EMA Clinical Data website.³¹

Fig. 1.

The authorisation of new medicines in the European Union. The schematic illustrates the standard 210-day evaluation process (blue trajectory), starting from the submission and validation of the extensive dossier submitted and led by CHMP members appointed as Rapporteurs. This phase can be followed by a 60-day re-examination procedure (red trajectory) leading to CHMP opinion. The European Commission then adopts a decision (grey). The schematic includes the Pharmacovigilance Risk Assessment Committee's evaluation and the optional consultation with Scientific Advisory Groups.

The efficacy of lecanemab and donanemab

Both lecanemab and donanemab have been studied in single pivotal studies, recruiting respectively 1795 and 1736 patients with early AD. For both medicines, efficacy was characterised on scales that measure cognitive functions such as memory, language and praxis (ADAS-Cog), level of functioning on the activities of daily living (ADCS-MCI-ADL and ADCS-iADL) and scales that combine measures of the two aspects (CDR-SB and iADRS). While some of the scales have been recently constructed or updated, the assessment of efficacy in early AD still relies on scales that were developed without the current understanding of psychometric methods, for more advanced stages of AD, and not for the purpose of measuring a slowing of decline. This is an important limitation for the interpretation of the efficacy results. See Table 1 for a description of the main outcome measurement tools used.

Table 1.

Description of important outcome measurement tools used in the two pivotal studies.

Tool	Concept(s) of interest	Scoring methodology
Clinical Dementia Rating Scale (CDR), Sum of Boxes (sb)a and Global Score (GS)32	Cognition (memory, orientation, judgement and problem solving), and function (community affairs, home and hobbies, and personal care)	Clinician-reported outcome, based on interview with patient and informant.
Alzheimer's Disease Assessment Scale–Cognitive Subscale (ADAS-Cog)	The original 11-items version33 measures different cognitive domains including memory, language, orientation and constructional and ideational praxis. The 13-items and 14-items versions add further measures of memory and executive function34	Both performance-based and clinician-assessed elements.
Alzheimer's Disease Cooperative Study—activities of daily living inventory (ADCS-ADL)35	Activities of daily living, both basic (relating to personal care) and instrumental, such as cooking, travelling, attending to hobbies and household chores. The version for use in MCI patients36 has additional items focussing on complex activities and the functioning in extraordinary circumstances	Clinician-reported outcome, based on interview with an informant.
Integrated Alzheimer's Disease Rating Scale (iADRS)37,b	The scale is a combination of cognitive aspects (as it includes the ADAS-Cog14) and of functional aspects (from the instrumental part of the ADCS-ADL, ADCS-iADL)	Both performance-based and clinician-assessed elements.
European Quality of Life, 5 Dimensions, 5 Levels version (EQ-5D-5L)38	Quality of Life domains of mobility, self-care, usual activities, pain/discomfort, anxiety/depression.	Patient reported outcome
Quality of Life in Alzheimer's Disease (QoL-AD)39	Quality of Life domains including energy, social relationships, ability to perform chores and activities for fun.	Patient and care partner reported outcome
Zarit Burden Interview (ZBI)40	Perceived emotional, social, physical and financial burden for the care partner.	Care partner reported outcome

^aPrimary for the main lecanemab study.

^bPrimary for the main donanemab study.

While the trials for the two molecules had the same study duration and targeted broadly the same population, there are differences in the patients recruited and procedures followed. As a result, data don't allow for a direct comparison and should be read accordingly. When estimated with methods that make realistic assumptions on patients with missing data after treatment discontinuation,^41,42 the main effects at 18 months for lecanemab were a difference in terms of change from baseline compared with control on the Clinical Dementia Rating Scale Sum of Boxes (CDR-sb) of −0.535 (95% CI: −0.778, −0.293), on the ADAS-Cog14 scale of −1.512 (−2.486, −0.538) and on the ADCS MCI-ADL of 1.936 (1.029, 2.844). For donanemab, the difference in change from baseline versus control on the primary endpoint of iADRS was 2.65 (1.04, 4.26). Secondary endpoints were also in favour of donanemab with the following estimates: −0.69 (−0.95, −0.43) for CDR-sb, −1.35 (−2.19, −0.51) for ADAS-Cog13, and 1.46 (0.50, 2.42) for ADCS-iADL.

While—after careful consideration of potential sources of bias and unreliability–the effects were considered robust from a methodological and statistical perspective, opinions considerably differed on their clinical meaningfulness. Under the assumption that the group difference is also the treatment effect for each individual patient, one can use the minimal clinically important difference (MCID) to understand its clinical meaningfulness. While the estimates of the MCIDs in the literature are higher than the effect estimates for lecanemab and donanemab,^43,44 the derivation of MCIDs is not straightforward and data are also compatible with attributing clinical meaningfulness to a difference of 0.5 points on the CDR-sb. As the MCIDs for iADRS are derived—among other approaches–from anchoring to values known in the literature for other scales,⁴⁵ including CDR-sb, the uncertainty propagates to such values. Furthermore, the assumption that the treatment effect is the same (i.e. equal to the between-groups difference) for all patients is far from an obvious fact in a disease for which the rate of progression is known to be heterogeneous. A decision on clinical relevance based solely on the MCID therefore did not appear adequate in this case, hence other lines of evidence were considered.

Firstly, data from lecanemab's pivotal trial explicitly measured the effects on quality of life (EQ-5D-5L and QoL-AD) and care partner burden (ZBI).⁴⁶ While the effect sizes for those measures may in turn be debated, the presence of favourable effects on all indicators was taken as reassuring.

Secondly, both developers presented the effects of lecanemab and donanemab with summary measures different from differences in group average change scores. For example, in the lecanemab dossier, “time-saved” analysis showed that—on the primary endpoint of CDR-sb and in the population indicated—patients who receive placebo reach the level of decline seen at 18 months in patients treated with lecanemab 6.1 months earlier on average. In the dossier for donanemab a definition of deterioration was used that was based on having two consecutive CDR-GS measurements greater than the patients’ own baseline. Such a definition was met with a 38% lower risk with donanemab than with placebo.

Thirdly, the developers presented how the effects of lecanemab and donanemab would extend beyond the 18-month double-blind phase of their trials. The exercise used data from the long-term extensions of the pivotal studies and, in part, data from an external natural history cohort (ADNI).⁴⁷ For lecanemab, it was estimated that 66% of untreated patients would progress to the next stage of disease by 42 months, whereas the same proportion of lecanemab-treated patients would only have reached progression at 60 months. For donanemab, the developer presented an extrapolation on the long-term effects of treatment, depending on baseline predicted disease progression. In the three percentiles (25%, 50%, and 75%) presented, the delays to severe dementia were respectively 26.1, 10.8 and 5.5 months. Importantly, it should be noted that this exercise relied on untested assumptions and any conclusion on the efficacy profile beyond the 18 months corresponding to the double-blind phase of the trials is subject to considerable uncertainty.

While acknowledging that all the approaches above have significant limitations, they were part of the many lines of evidence that the Committee for Medicinal Products for Human Use (CHMP) of the European Medicines Agency considered in reaching their conclusions.

Amyloid-related imaging abnormalities and their clinical consequences

The main adverse reactions to lecanemab and donanemab are amyloid-related imaging abnormalities (ARIA), findings underpinned by haemorrhages (ARIA-H) or oedema (ARIA-E) that are believed to follow complex—and only partly understood—mechanisms triggering temporary increase in vascular vulnerability in the context of amyloid clearance.⁴⁸ ARIA is asymptomatic in most cases, but it can have symptoms from mild to severe, with potentially fatal outcomes. As reported more widely in the literature for this class of therapeutics,⁴⁹ the incidence of ARIA (including the incidence of severe and fatal cases) correlated with the presence and number of ApoE ε4 alleles.

To find a population with an acceptable safety profile, the indication was restricted to patients who are ApoE ε4 non-carriers or heterozygotes. While this is based on subgroup analyses, the restriction was in line with regulatory standards,⁵⁰ considering the context of overall positive studies, the biological plausibility,⁵¹ and the consistency of the subgroup effect across trials in the same class. Even in such restricted population, the risk of ARIA is considerable. In terms of absolute frequencies and relative risks (RR) compared to placebo, the risk of ARIA-E was increased in the pivotal data on the eligible population for donanemab from 1.8% to 20.6%, with a RR of 11.5 (95% CI: 6.6, 20.1) and for lecanemab from 1.3 to 8.9% with a RR of 6.8 (95% CI: 3.5, 13), and the risk of ARIA-H was increased for donanemab from 12.2% to 27.6%, with a RR of 2.3 (95% CI: 1.8, 2.8) and for lecanemab from 6.8% to 12.9% with a RR of 1.9 (95% CI: 1.4, 2.6).

Consequently, extensive risk minimisation measures have been put in place. Both medicines are subject to a controlled access programme, ensuring that the treatment is initiated only in patients who are eligible and to whom adequate monitoring can be offered. Contraindications to the use of the two antibodies go beyond those used in the clinical development programmes, as an attempt to further reduce the incidence and severity of ARIA. In particular, both products are contraindicated for patients who—at a baseline MRI—present with more than 4 microhaemorrhages, superficial siderosis, vasogenic oedema, or other findings suggestive of cerebral amyloid angiopathy. Patients with bleeding disorders that are not under control or those receiving ongoing anticoagulant therapy are also not eligible to initiate treatment. In addition, during donanemab's evaluation, the Committee has received and evaluated data from the study of an alternative posology—with a slower titration intended to decrease the incidence of ARIA.⁵² This became the recommended posology in the EU.

As there is a concern that continuing treatment in presence of ARIA findings of a certain nature may trigger or exacerbate symptoms, regular monitoring MRIs are mandatory for both products, and detailed guidance is given in the Summaries of Product Characteristics^53,54 about what to do in case ARIA is found. The presence of intensive MRI monitoring, together with the Controlled Access Programme, means that it is possible that not all centres will be able to offer treatment with these two molecules. While this may raise legitimate concerns about equitable access, this measure was considered essential to mitigate the risks.

As ARIA is a relatively new phenomenon, some uncertainties exist about its impact in clinical practice, especially on its long-term consequences. Registry-based studies will further characterise these events. While clinical trial data do not seem to suggest that cognitive decline is accelerated in patients with ARIA, this concern will also be addressed.

Considering outside views in EMA's decision-making

EMA's evaluation procedures incorporate ways to gather information from the public, from patients' associations, and from experts. On the one hand, a scientific advisory group, including clinicians and patients' representatives, was convened to discuss specific points of the applications for both lecanemab⁵⁵ and donanemab, and they expressed—at that stage–a negative view. On the other hand, unsolicited third-party interventions were submitted by individual patients, patients' organisations and groups of experts, and these were overwhelmingly positive.

Furthermore, patients’ preference studies have been conducted both on the specific questions associated with lecanemab and donanemab,^56,57 and to investigate the preferences of patients with AD more generally.58, 59, 60 While some of these studies have important methodological issues, they generally support the notion that the attitude of patients is heterogeneous, with at least some patients more likely than others to accept significant risks for a medicine that potentially slows down their decline.

All views expressed were considered by CHMP when reaching its final decisions on the two medicines.

Balancing benefits and risks

Balancing benefits and risks is a complex, multidimensional problem, rather than a quantitative exercise with fixed thresholds. Ultimately, CHMP had to decide whether lecanemab and donanemab are good options for at least a subset of the patients with early AD, acknowledging that patients differ in their preferences, personal situations and willingness to accept risks. Initially, CHMP issued a negative opinion for both products, considering that the observed beneficial effects on cognitive decline do not counterbalance the risk of ARIA. During re-examination procedures, the decision was positive by majority for both products, subject to the restrictions, the MRI monitoring, and other measures outlined above. In the presence of treatments with a demonstrated and precisely quantified efficacy, and with risks that can be appropriately quantified and—to some extent—mitigated—the Committee decided by majority to recommend authorisation of these medicines, that will represent new options for patients suffering from this devastating and invariably progressive disease.

Those Committee members expressing divergent positions^29,30 considered that the size of the effects is too small to outweigh the risks. This minority position emphasised the doubts on the relevance of the amyloid reduction observed, on the clinical relevance of the effects, as well as the uncertainties on the long-term safety of the products.

The future of treating Alzheimer's disease

After regulatory approval, the availability of products in the EU depends on the companies' launch strategy and Member State considerations based on cost-benefit evaluations. These are stand-alone considerations and may not necessarily reach the same conclusions as the CHMP's benefit-risk evaluation. To the extent that the products will be available, doctors and eligible patients will need to have informed discussions on their specific situation and possible benefits and risks, and decide whether a prescription is warranted given the individual circumstances and preferences of the patient. For those who will receive a prescription, lecanemab and donanemab may slow down AD progression to some extent, giving patients some more time to participate in their activities with some level of independence and satisfaction. At the same time, risks will have to be carefully managed with strict adherence to the SmPC, and while measures are in place in an attempt to minimise them, their persistence and seriousness should be acknowledged.

As real-world experience starts to accrue,61, 62, 63 increased knowledge (from this and other types of research) may provide information to further support clinicians to select patients who are likely to benefit most from these medicines (for example, building on signals of an inverse correlation between benefit and tau burden at baseline^29,30). Similarly, data accrual will allow to fine-tune the selection and monitoring of patients to ensure that fewer will suffer from adverse events. For lecanemab, the MRI schedule has already been updated post-approval in response to new analyses on the timing of occurrence of ARIA events.⁶⁴ Currently, lecanemab can be given until a patient reaches the moderate AD dementia stage, regardless of amyloid clearance, with the rationale that it preferentially targets the oligomers, while treatment with donanemab—which targets the plaques—should only be maintained until amyloid plaques are cleared. While these were the best recommendations possible at the time of approval, new knowledge may inform on the best treatment strategies in practice (and hence lead to variations of the terms of the authorisations).

The presence of two approved amyloid-targeting treatments also changes the landscape for developing new treatments, both those targeting amyloid in different ways and those with other targets. On the one hand, it will be possible to study combination strategies where other disease pathways will be targeted at the same time as amyloid. On the other hand, studying treatments that are not intended to be used in combination with the available amyloid-targeting medicines may become more difficult.

While regulatory approval of a medicine does not imply the validation of a theory, but rather relates to the phenomenological assessment of the results of clinical experiments, important interactions with theory do exist. The assessment of the clinical data confirmed the validity of two observations. First, the two antibodies have demonstrated some level of efficacy on the progression of early AD, which can be explained either—as proposed by the developers—with a lowering of (some species of) amyloid in patients' brains, or in alternative ways.⁶⁵ Second, this efficacy was not (nearly) a complete halting of the progression in all patients despite reduction in the amyloid load. Any new development within or beyond the amyloid hypothesis should account for both these observations. Increasing understanding of Alzheimer's disease will contribute to developing more transformative treatments, including treating the disease before it manifests with symptoms.

EMA will continue to support innovation in Alzheimer's disease research. It recently hosted a multistakeholder Workshop on Patient Registries for AD,⁶⁶ and is currently updating its guidelines for developers.⁶⁷

The coordinated effort of independent researchers, developers and regulators will be key to further progress the fight against this disease.

Contributor

L.G. has written the original draft. All authors have contributed to the conceptualisation and writing (review and editing).

Declaration of interests

We declare no competing interests.