Effectiveness and safety of monoclonal antibodies against amyloid-beta vis-à-vis placebo in mild or moderate Alzheimer's disease

Backgrounds and objectives Currently, no consensus has been reached on the therapeutic implications of monoclonal antibodies against amyloid-beta (Aβ) in Alzheimer's disease (AD). This study aimed to examine the effectiveness and safety of monoclonal antibodies against Aβ as a whole and also to determine the superiority of individual antibodies vis-à-vis placebo in mild or moderate AD. Methods Literature retrieval, article selection, and data abstraction were performed independently and in duplicate. Cognition and function were appraised by the Mini-Mental State Examination (MMSE), Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), Disability Assessment for Dementia (DAD), and Clinical Dementia Rating Scale-Sum of Boxes (CDR-SB). Effect sizes are expressed as standardized mean difference (SMD) with a 95% confidence interval (CI). Results Twenty-nine articles involving 108 drug-specific trials and 21,383 participants were eligible for synthesis. Of the four assessment scales, only CDR-SB was significantly reduced after using monoclonal antibodies against Aβ relative to placebo (SMD: −0.12; 95% CI: −0.2 to −0.03; p = 0.008). Egger's tests indicated a low likelihood of publication bias. At individual levels, bapineuzumab was associated with a significant increase in MMSE (SMD: 0.588; 95% CI: 0.226–0.95) and DAD (SMD: 0.919; 95% CI: 0.105–1.943), and a significant decrease in CDR-SB (SMD: −0.15; 95% CI: −0.282–0.018). Bapineuzumab can increase the significant risk of serious adverse events (OR: 1.281; 95% CI: 1.075–1.525). Conclusion Our findings indicate that monoclonal antibodies against Aβ can effectively improve instrumental activities of daily life in mild or moderate AD. In particular, bapineuzumab can improve cognition and function, as well as activities of daily life, and meanwhile, it triggers serious adverse events.


Introduction
Alzheimer's disease (AD) is a chronic neurodegenerative disease with insidious clinical presentation, and it is characterized by progressive impairment of memory and cognitive function. Approximately 50 million people are suffering from dementia globally, and the number elevates by 10 million annually, as per the 2020 report of the World Health Organization (WHO) (1). By 2050, the cases of dementia are expected to triple (2). As revealed by a systematical analysis in 2020, the overall prevalence of AD was 3.2% in Chinese individuals over 60 years, and its annual prevalence was predicted to increase from 3.81 to 6.17% within the next 5 years (3). In India, the incidence rate of AD per 1,000 person-years was 11.67 for those aged ≥ 55 years (4). Patients diagnosed with AD often experience slow and variable clinical courses, and their original survival ability gradually decreases, eventually leading to death due to complications (5). It is, hence, clinically meaningful to retard, prevent, or even reverse neurological and functional impairment through early and effective pharmacologic treatment.
Alzheimer's disease is a multifactorial disorder involving interactions among genetic, environmental, and lifestyle factors, which open new avenues for the development of tailored therapeutics in the era of precision medicine (6). It is widely recognized that dementia is the underlying cause of AD, and it accounts for 60% of cases (7). AD progresses rapidly, yet treatment options are very limited. Some approved drugs targeting AD, such as donepezil, galantamine, rivastigmine, and memantine, can only help relieve patients' symptoms and suppress the psychological and behavioral symptoms of dementia. Several theories existed for the pathophysiology of AD, including the amyloid cascade hypothesis, degeneration of neuronal cells, and aggregation of tau proteins within the cell (8,9). Thereof, the amyloid cascade hypothesis is widely accepted, and it proposes that the neurodegeneration and resultant dementia of AD occur as a result of the formation and accumulation of toxic, soluble amyloidbeta (Aβ) oligomers, formed by the misfolding of Aβ monomers (10). In the literature, different therapeutic strategies to clear Aβ from the brain were developed, and monoclonal antibodies against amyloid-beta (Aβ) have aroused growing concerns (11). There is clinical evidence that immunotherapy with monoclonal antibodies is effective for the treatment of patients at earlier AD stages before the emergence of dementia (12). Bapineuzumab is the first N-terminus-directed anti-Aβ antibody tested in humans. Subsequently, several anti-Aβ monoclonal antibody drugs were tested by clinical trials (13). Moreover, aducanumab, a human Ig monoclonal antibody, is recognized as being "risen from the grave, " and it acts in Aβ clearance and curtailing calcium defects in AD (14). Other treatment potentials, such as the immune response generating active immunotherapy and passive immunotherapeutic approaches targeting monoclonal antibodies toward Aβ aggregates, were also proposed (10). Of all anti-Aβ regimens, passive immunization with anti-Aβ antibodies is recognized as being safe and well-tolerated (15), whereas no consensus has been reached upon the therapeutic implications of monoclonal antibodies against Aβ in AD. Fortunately, metaanalysis can provide an opportunity to help derive more reliable estimates.
We aimed to examine the effectiveness and safety profiles of monoclonal antibodies against Aβ as a whole and also to determine the superiority of individual monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of patients with mild or moderate AD.

Guidelines
The conduct of this meta-analysis conformed to the statement in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (16

Inclusion/exclusion criteria
Trials were eligible for inclusion if they met the following criteria simultaneously: (i) participants: patients with mild or moderate AD; (ii) intervention: monoclonal antibodies against Aβ and placebo; (iii) comparator: control; (iv) clinical outcomes: changes in one of the four scales adopted to assess the cognition and function aspects of AD, including the Mini-Mental State Examination (MMSE), Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), Disability Assessment for Dementia (DAD), and Clinical Dementia Rating Scale-Sum of Boxes (CDR-SB); (v) study design: randomized controlled trials; and (vi) formal publication in peer-review journals.
Trials were excluded if one or more of the following criteria were satisfied: (i) publication type: narrative or systematic review, meta-analysis, case report, case series, conference abstract, comment, correspondence, or editorial; (ii) duplication publication; (iii) lack of comparator; (iv) control .

FIGURE
PRISMA flowchart illustrates the selection process of qualified articles with specific reasons for exclusion in this meta-analysis.
rather than placebo; and (v) clinical outcomes rather than four assessment scales mentioned earlier. In the case of more than one article was published using the overlapped study participants, the article with the largest sample size was retained in this meta-analysis. The eligibility assessment of each retrieved trial was made by two authors (Y.H. and M.D.) independently. Any discrepancy was solved by discussion, and if necessary, was adjudicated by a third author (W.N.).

Data collection
Data from each qualified article were separately abstracted from each qualified article by two reviewers (Y.H. and M.D.) and were typed into a predesigned Excel file, including the surname of the first author, year of publication, ethnicity, and country where participants were enrolled, study design, trial phase, intervention drugs and doses, degree of AD, intervention period, sample size of each arm, number of responses, and dropouts during regimen treatment, and assessment scales for AD, as well as some baseline characteristics, including age, gender, weight, height, body mass index, duration of AD, use of AChEI (acetylcholinesterase inhibitors) or memantine, four assessment scales associated with the risk of AD and adverse reactions, when available.
The process of data collection was completed independently and in duplicate (Y.H. and M.D.), and the consistency of the two datasets was tested by the kappa statistic. In the case of kappa statistics less than unity, original data were checked, and if necessary, a third author (W.N.) was involved.

Quality assessment
Risk of bias for each clinical trial was assessed using the "Revised Cochrane risk-of-bias tool for randomized trials" (RoB 2) (17) from the following five aspects, that is, randomization process, bias due to deviations of intended interventions, bias due to missing outcome data, bias in outcome measurements, and bias in the selection of reported results. Individual domains of risk of bias can be categorized as "low risk, " "some concerns, " or "high risk." Quality assessment was performed by two authors (Y.H. and M.D.), and any disagreement was solved by a third author (W.N.).

Statistical analyses
Data were imported from Excel to STATA software version 16 (Stata Corp, College Station, Texas, USA), which was used to handle statistical analyses in this meta-analysis. Effect-size estimates from individual trials were pooled under random-effect models, irrespective of the presence or absence of statistical heterogeneity across trials (18). Statistical heterogeneity was measured by the I 2 metric, which ranges from 0 to 100%, with higher values representing greater degree of heterogeneity.
The changes in assessment scales for AD before and after intervention are expressed as a standardized mean difference (SMD) with a 95% confidence interval (95% CI) because different rating subscales were used, and the changes in adverse events after intervention are expressed as odds ratio (OR) with a 95% CI.
Cumulative analyses were used to measure the influence of first published trials on subsequent publications and the evolution of accumulated estimates over time. Sensitivity analyses were used to assess the influence of any single trial on pooled effect-size estimates by removing one trial at a time.
Publication bias was inspected using Begg's funnel plots and Egger's tests. The significance of Egger's tests was set at 10%. In addition, to yield more information, the Duval and Tweedie non-parametric "trim and fill" method was employed to estimate the number of theoretically missing trials and derive "unbiased" effect-size estimates.

Eligible articles
By using the prespecified key terms, the literature search of three public databases retrieved a total of 140 publications. After applying predesigned inclusion and exclusion criteria, only 29 articles published in English from 2009 to 2021 were eligible for the final analysis , involving 108 drug-specific trials and 21,383 participants. Figure 1 illustrates the process of article selection for this meta-analysis. Table 1 shows the trial characteristics in this meta-analysis. Five trials involved patients with mild AD, 95 trials involved patients with mild or moderate AD, and five trials involved patients with prodromal or mild AD. Forty-six trials were in phase I, 26 in phase II, two in phase II-III, 26 in phase III, and eight in unreported phases. Trial duration ranged from 12 to 208 weeks. In terms of risk of bias, all clinical trials involved in this meta-analysis were classified as "low risk" or "having some concerns" due to missing necessary information. Table 2 shows the detailed targeting information of monoclonal antibodies against Aβ under evaluation. Specifically, eight monoclonal antibodies against Aβ were available, including aducanumab (BIIB037), bapineuzumab (AAB-001), bapineuzumab modified (AAB-003), crenezumab (MABT5102A), donanemab (LY3002813), gantenerumab (R1450/RO4909832), ponezumab (PF04360365), and solanezumab (LY100/LY2062430). Comparison with placebo was available for aducanumab in 13 trials, for bapineuzumab in 35 trials, for bapineuzumab modified in five trials, for crenezumab in nine trials, for donanemab in seven trials, for gantenerumab in three trials, for ponezumab in 13 trials, and for solanezumab in 23 trials.

FIGURE
Forest plots of four assessment scales for monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of mild or moderate Alzheimer's disease.
Overall estimation Figure 2 provides the forest plots of four assessment scales for monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of AD. Of four assessment scales, only CDR-SB was significantly reduced after using monoclonal antibodies against Aβ relative to placebo (SMD: −0.12; 95% CI: −0.2 to −0.03; p = 0.008), indicating that monoclonal antibodies against Aβ can effectively improve instrumental activities of daily life. Statistical heterogeneity across trials for each assessment scale was significant (I 2 > 90%; p < 0.001).

Cumulative and influential analyses
Supplementary Figures 1, 2 separately show the cumulative and influential analyses of four assessment scales for monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of AD. Figure 3 presents the filled funnel plots of four assessment scales for monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of AD. There were separately one, six, five, and four theoretically missing studies required to make the funnel plots symmetrical for MMSE, ADAS-Cog, DAD, and CDR-SB. Egger's test indicated a low likelihood of publication bias, with the corresponding probabilities being 0.687, 0.434, 0.880, and 0.282.

Subsidiary estimation
As different monoclonal antibodies against Aβ might exert a diverse impact on assessment scales, drug-specific subsidiary analyses were done accordingly (Table 3). To control potential bias from small-scale estimation, only subgroups involving three or more trials are displayed. Specifically, bapineuzumab was . /fneur. .  Regarding solanezumab, there was a significantly reduced risk for nervous system disorders (OR: 0.808; 95% CI: 0.713-0.916). For ponezumab, the risk of headache was reduced significantly (OR: 0.542; 95% CI: 0.297-0.991). In contrast, gantenerumab was associated with a significantly increased risk of amyloid-related imaging abnormalities (OR: 13.145; 95% CI: 5.215-33.136).

Adverse events
Rare adverse events associated with monoclonal antibodies against Aβ vis-à-vis placebo in the treatment of mild or moderate AD are presented in Supplementary Table 2.

Discussion
The aim of this meta-analysis was to summarize data on the effectiveness and safety of monoclonal antibodies against Aβ visà-vis placebo in the treatment of mild or moderate AD. It is .
/fneur. . noteworthy that monoclonal antibodies against Aβ as a whole can effectively improve instrumental activities of daily life based on CDR-SB scores. Moreover, analysis of individual antibodies revealed that bapineuzumab can improve cognition and function, as well as activities of daily life, yet it also triggers the occurrence of serious adverse events. To the best of our knowledge, this is the largest meta-analysis thus far that has synthesized data on monoclonal antibodies against Aβ compared with placebo for mild or moderate AD. The deposit of extracellular Aβ plaques is a key feature of AD, and mounting evidence indicates that aberrant Aβ production or clearance is a potential harbinger in the pathogenesis of AD (48). Immunotherapy with monoclonal antibodies is increasingly identified as an effective therapeutic regime against AD, and dozens of clinical trials have been undertaken to explore the effectiveness and safety of monoclonal antibodies against Aβ in patients with AD (11, 49,50). However, the results of these trials are not often reproducible. For example, Doody et al. in a multicenter, randomized, placebo-controlled trial demonstrated a marginally significant increase in MMSE scores in favor of donepezil (51), and contrastingly, Rinne et al. found that bapineuzumab exerted an unfavorable effect on MMSE scores (21). The reasons for these inconsistencies are likely several-fold. One reason might be related to sample sizes, because the magnitude of changes in instrumental scores between interventions is small in most cases. Another reason is probably due to the diverse types of monoclonal antibodies against Aβ, in view of the different targeted Aβ epitopes (31,36,(52)(53)(54)(55)(56). A third reason rests with the differences in demographic and clinical characteristics, as well as genetic undergrounds across trials. Fortunately, metaanalysis offers a rational and helpful approach to dealing with inconsistencies from many studies of the same research topic. With the help of this approach and based on 29 articles and 21,383 participants, we interestingly found that monoclonal antibodies against Aβ as a whole can effectively improve instrumental activities of daily life based on CDR-SB scores in patients with mild or moderate AD, in line with the observations of many clinical trials (26, [36][37][38]. In addition, we explored the effectiveness and safety of individual monoclonal antibodies against Aβ in patients with AD. Because of the limited number of eligible trials, statistical significance was merely identified for bapineuzumab, an antibody targeted against the N-terminus of Aβ as reflected by MMSE and DAD scores, which can not only improve cognition and function but also enhance activities of daily life, as reflected by CDR-SB scores in terms of effectiveness. Simultaneously, the administration of bapineuzumab was associated with the development of serious adverse events. We agree that the safety profile is paramount, and the long-term benefits and risks of bapineuzumab treatment for mild or moderate AD are not yet known (25,41). However, we here express concerns that such warnings may discourage patients and their families from choosing bapineuzumab in practice. From another aspect, Aβ might not be the best treatment target in patients with mild or moderate AD, or monoclonal antibodies against Aβ cannot remove an important species of Aβ that plays a contributing role in the pathogenesis of AD (37). Nevertheless, we agree that more large-scale clinical trials with long-term extended follow-ups are warranted to unveil the full potential of monoclonal antibodies against Aβ in AD.
In addition to the clear strengths of this meta-analysis, including the largest sample size, comprehensive analyses, and solid observations, several limitations should be acknowledged. First, only clinical trials published in English were retrieved, which leaves selection bias an open question, as some excellent trials may be published in other languages. However, explorations on publication bias revealed a low probability. Second, the power to detect significance in some subgroups was limited, and betweentrial heterogeneity cannot be totally accounted for. Third, only the effectiveness and safety of monoclonal antibodies against Aβ .
/fneur. .     vis-à-vis placebo were examined in the current meta-analysis, and comparison between other classes of drugs targeting AD will be addressed in the future. Fourth, definitions of adverse effects evaluated in this meta-analysis differed across trials, and caution is needed when interpreting the safety profiles of monoclonal antibodies. Taken together, our findings indicate that monoclonal antibodies against Aβ as a whole can effectively improve instrumental activities of daily life based on CDR-SB scores in mild or moderate AD. Individually, bapineuzumab can improve cognition and function, as well as activities of daily life, yet it also triggers the occurrence of serious adverse events. Further functional investigations on the molecular mechanisms of monoclonal antibodies against Aβ, in particular, bapineuzumab, in the pathophysiology of AD.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.