Passive immunotherapy for Alzheimer ’ s disease

Alzheimer ’ s disease (AD) is the most common neurodegenerative disease characterized by cognitive impairment with few therapeutic options. Despite many failures in developing AD treatment during the past 20 years, significant advances have been achieved in passive immunotherapy of AD very recently. Here, we review characteristics, clinical trial data, and mechanisms of action for monoclonal antibodies (mAbs) targeting key players in AD pathogenesis, including amyloid-β (A β ), tau and neuroinflammation modulators. We emphasized the efficacy of lecanemab and donanemab on cognition and amyloid clearance in AD patients in phase III clinical trials and discussed factors that may contribute to the efficacy and side effects of anti-A β mAbs. In addition, we provided important information on mAbs targeting tau or inflammatory regulators in clinical trials, and indicated that mAbs against the mid-region of tau or pathogenic tau have therapeutic potential for AD. In conclusion, passive immunotherapy targeting key players in AD pathogenesis offers a promising strategy for effective AD treatment.


Introduction
Approximately 55 million people worldwide suffer from dementia.AD is the most common form of dementia, accounting for approximately 60-70% of dementia cases.The number of dementia patients is expected to rise to 78 million in 2030 and 139 million in 2050 due to an aging population.AD and other forms of dementia were the 7th leading cause of death in 2019, and in the same year, the total global societal cost of dementia was estimated at $ 1.3 trillion, and these costs are projected to exceed $ 2.8 trillion by 2030 (Organization, 2023).
The pathological features of AD mainly include amyloid plaques formed by extracellular Aβ, neurofibrillary tangles (NFTs) formed by intraneuronal phosphorylated tau aggregates, brain atrophy caused by neuronal loss, synaptic damage, and glial cell overproliferation (Guan et al., 2021;Selkoe, 2002;Tzioras and McGeachan, 2023).Although the pathogenesis of AD has not been fully elucidated, multiple lines of evidence suggest that the cascade reaction triggered by Aβ, decreased acetylcholine, abnormal phosphorylation of tau protein, and neuroinflammation may all play important roles in AD pathogenesis, eventually leading to neuronal loss and deterioration of brain function in AD patients (Gómez-Isla et al., 1997).Based on these important contributors to AD pathogenesis, many drugs have been developed and tested in clinical trials, and some of them have been approved by the U.S. Food and Drug Administration (FDA).The first class of FDA-approved medications for AD are cholinesterase inhibitors (Nordberg, 2006), including tacrine, donepezil, rivastigmine and galantamine, which can increase the content of acetylcholine between synapses.The second type is the N-methyl-D-aspartate receptor (NMDA) antagonist memantine (Xia et al., 2010), which can alleviate the excitatory neurotoxicity induced by glutamate.The third category is monoclonal antibodies (mAbs) targeting Aβ, including aduhelm and leqembi, which work primarily by promoting Aβ clearance (Budd Haeberlein et al., 2022;van Dyck et al., 2023).While the efficacy of aduhelm remains controversial, leqembi shows a marked effect on delaying cognitive decline in AD patients based on the data from clinical trial III.Although these FDA-approved drugs can slightly or moderately alleviate cognitive impairment in AD patients, none of them has been shown to be able to stop neurodegeneration and halt disease progression clinically.Therefore, it is still urgent to develop new drugs that have greater effectiveness.
Passive immunotherapy is one of the major strategies under clinical development for AD treatment.Here, we evaluate the characteristics and clinical trial data for antibodies targeting important players in AD pathogenesis, including Aβ, tau and inflammatory/immune targets.We learned from failures and positive results and sought to describe the difference in therapeutic effect between antibodies with different epitopes, including selectively modified epitopes, dose and side effects, and discuss how future approaches might address the limitations that already exist.We also provide an update on the status of currently ongoing passive immunotherapy clinical trials.

Passive immunotherapy targeting Aβ
Based on the amyloid cascade hypothesis, the accumulation of Aβ peptides and deposition of amyloid plaques is considered as the main trigger of the neurodegenerative process of AD, although controversy remains (Selkoe and Hardy, 2016).Aβ peptides are generated from proteolysis of amyloid precursor protein (APP), a type 1 transmembrane protein predominantly concentrated in neuronal synapses (Vogt et al., 2023).APP can be proteolytically cleaved by proteases in non-amyloidogenic and amyloidogenic pathways.In non-amyloidogenic pathway, APP is cut by α-secretase and generates soluble sAPPα and the C-terminal fragment C83/CTFα.In amyloidogenic pathway, APP is cleaved by β-secretase BACE1 and thereby producing the extracellular fragment sAPPβ and the C-terminal fragment C99/CTFβ.C99/CTFβ is subsequently cleaved by γ-secretase, generating Aβ and CTFγ-AICD (Goate, 2006).After generation, Aβ peptides spontaneously aggregate into oligomers, which form protofibrils and fibrils and aggregate into plaques (Zhao et al., 2020).Aβ oligomers and protofibrils may play a more important role in AD progression, as they have been shown to be more toxic to neurons than fibrils (Haass and Selkoe, 2007;Nilsberth et al., 2001;Sehlin et al., 2012).The mechanisms underlying neurotoxicity of Aβ oligomers are complex and may involve synaptic Ca 2+ disruption, mitochondrial damage, deterioration of synapses, impairment of axonal transport, and hippocampal long-term potentiation inhibition (Cline et al., 2018).In addition, oligomeric Aβ activates glial cells to trigger neuroinflammation (Leng and Edison, 2021) and accelerates tau pathology development, possibly via microglia (Clayton et al., 2021), further exacerbating the disease progress of AD.
Aβ has long been the major focus of efforts to develop passive immunotherapy for AD (Karran et al., 2011).In the brain, after the Fab fragment of the anti-Aβ antibody binds to Aβ, its Fc fragment is able to bind to Fc gamma receptors (FcγRs) on the surface of microglia and activates microglial phagocytosis (Wilcock et al., 2004).However, anti-Aβ mAbs can induce a shift of microglial state from the M2-like phenotype (anti-inflammatory) to the M1-like phenotype (pro-inflammatory), leading to neuroinflammation and a reduced ability of microglia to clear Aβ (Wilcock et al., 2011).Because FcγRs are classified into activating and inhibitory types, the affinity of the Fc domain of anti-Aβ mAbs for different FcγR classes should be considered in antibody engineering (Fuller et al., 2014).In addition, when anti-Aβ mAbs are present at suboptimal levels in the brain, Aβ-mAb complexes can bind to C1q and be taken up by microglia via C1q receptors (Webster et al., 2001).In addition to microglial phagocytosis, both in vivo and in vitro data have shown that anti-Aβ mAbs can directly regulate Aβ aggregation and clearance (Solomon et al., 1997;Taguchi et al., 2008).
Although pharmaceutical industries have encountered many failures and setbacks in developing effective anti-Aβ immunotherapies, several mAb drugs have shown beneficial effects in phase III clinical trials and have been conditionally or fully approved by the FDA.Table 1 and Fig. 1 exhibit important information with regard to anti-Aβ mAbs that have been tested in phase III clinical trials, such as antibody epitopes and targeted Aβ forms, and clinical trial information.Four of these mAbs show clearance potency for amyloid plaques, including gantenerumab, aducanumab, lecanemab and donanemab.

Gantenerumab
Gantenerumab, a human monoclonal IgG1 antibody developed by phage display technologies, recognizes conformational epitopes consisting of the N-terminal (amino acids 3-11) and central portions of Aβ (amino acids 18-27) and binds Aβ fibrils with subnanomolar affinity (Bohrmann et al., 2012;Crespi et al., 2015).This mAb was initially tested in a phase III clinical study named Scarlet Road (NCT01224106) (Ostrowitzki et al., 2017).In Scarlet Road, 797 patients with prodromal AD were randomized and received 105 or 225 mg gantenerumab or placebo subcutaneously every 4 weeks.This study was halted due to futility when 50% of the patients had completed the treatment.In the 225-mg dose group, only a 4.8% reduction in brain amyloid load from the baseline was observed.Patients in the 105-mg dose and placebo groups showed 0.72% and 1.09% alterations of amyloid load from the baseline, respectively (Ostrowitzki et al., 2017).Subsequently, both Scarlet Road (NCT01224106) and another phase III trial Marguerite Road (NCT02051608), which enrolled patients with mild AD and failed in futility analysis, were converted into an open-label extension (OLE) study to investigate the safety and efficacy of higher dosages of gantenerumab (Klein et al., 2019).To decrease the risk of amyloid-related imaging abnormalities with edema or effusion (ARIA-E) in the OLE study, gradual uptitration schemes were used to reach the target dose of 1200 mg/month.Finally, 67 of 89 initially enrolled patients had scans at 52 weeks with mean amyloid levels reduced by 39 centiloids (CL), and 39 patients had scans at 104 weeks with mean amyloid levels reduced by 59 CL.At 104 weeks, the amyloid levels of 51% of patients were below the positivity threshold.These data suggest a potential clinical benefit of high-dose gantenerumab (Klein et al., 2019).Later, two additional phase III studies, Graduate I (NCT03444870) and II (NCT03443973), were completed by Roche and Genentech between 2018 and 2022 in patients with prodromal to mild AD.These two trials recruited 1965 participants who were randomized to receive gantenerumab or placebo.Gantenerumab used 9-month gradual uptitration schemes to target dose, 1020 mg/month, and was given as 510 mg every two weeks subcutaneously in 116 weeks, regardless of ApoE ε4 allele status.Regrettably, gantenerumab failed to meet the primary endpoint of Clinical Dementia Rating Sum of Boxes (CDR-SB) to slow cognitive decline at 116 weeks in Graduate phase III trials (Releases, 2022).Gantenerumab-treated patients slowed but nonsignificant cognitive decline by − 0.31 and − 0.19 on CDR-SB from the baseline in Graduate I and Graduate II studies, respectively.For safety, the ARIA-E in the gantenerumab group was 25%, and the vast majority of patients presented as asymptomatic.On underlying pathology, gantenerumab decreased mean amyloid levels by 21.1-24.1 CL after one year, and 46.8-57.6CL after two years in a substudy of 383 people.Only 27% of patients in the gantenerumab group became amyloid-negative after 116 weeks, in contrast to 51% in the OLE study.Cerebrospinal fluid (CSF) biomarkers, including phosphorylated-tau181 (P-tau181), total tau, and neurogranin, were significantly changed in the expected direction upon gantenerumab treatment.In summary, gantenerumab shows weak efficacy in delaying AD progression.

Aducanumab
Aducanumab (formerly BIIB037) is a humanized IgG1 mAb targeting amino acids 3-7 of the Aβ peptide, which can specifically bind soluble Aβ oligomers and insoluble Aβ fibrils (Sevigny et al., 2016).According to a phase Ib study (PRIME, NCT01677572), ARIA-E associated with aducanumab treatment was dose-dependent and more common in ApoE ε4 carriers (Sevigny et al., 2016).To reduce the incidence of ARIA-E, high-dose and low-dose groups were designed in phase III clinical trials based on patients carrying the ApoE ε4 gene.To test the efficacy of aducanumab, two phase III trials, EMERGE (NCT02484547) and ENGAGE (NCT02477800), were conducted, involving 1638 patients and 1647 patients, respectively, who were diagnosed with mild cognitive impairment (MCI) due to AD or mild AD dementia.Patients received X. Guo et al. high-dose aducanumab (6 mg/kg for ApoE ε4+ or 10 mg/kg for ApoE ε4-), low-dose aducanumab (3 mg/kg for ApoE ε4+ or 6 mg/kg for ApoE ε4-), or placebo randomly (1:1:1) and intravenously every 4 weeks for 76 weeks (Budd Haeberlein et al., 2022).Both EMERGE and ENGAGE were halted early in March 2019 for futility analysis, which included data from 49% of EMERGE patients and 57% of ENGAGE patients who completed a 78-week treatment.Although these two trials were designed identically, futility analysis showed different results, with an 18% CDR-SB treatment difference in favor of high-dose aducanumab in EMERGE and a 15% CDR-SB treatment difference in favor of placebo in ENGAGE (Budd Haeberlein et al., 2022).Subsequent efficacy analyses included data from the start of the clinical trials until halted.While the primary endpoint was met in EMERGE which showed 22% reduced cognitive decline on CDR-SB in the high-dose group compared to the placebo group (p = 0.12), ENGAGE showed a 2% increase in cognitive decline in the high-dose group (Budd Haeberlein et al., 2022).In addition, at the end of the 78-week study, the high-dose group of EMERGE reduced mean amyloid levels to 21 CL from the baseline (85 CL), with 64 CL of amyloid removed, while ENGAGE reduced mean amyloid levels to 37 CL from the baseline (90 CL), with 53 CL of amyloid removed.In both trials, a dose-and time-dependent change in CSF and plasma biomarkers of AD was observed.The most common adverse event was ARIA-E, consistent with other findings relevant to anti-Aβ mAbs (Ostrowitzki et al., 2012;Salloway et al., 2014;Sperling et al., 2012).In the high-dose EMERGE group, the incidence of ARIA-E was 35%, particularly in patients with the ApoE ε4 allele, whereas the incidence of ARIA-E was only 2% in the placebo group.The incidence of amyloid-related imaging abnormalities with hemosiderin deposits (ARIA-H) was 33% in the high-dose group (Salloway et al., 2022).
Different results from EMERGE and ENGAGE may be due to an amendment of the protocol made on March 24, 2017, during the clinical trial.At that time, to maximize the dose-dependent effect of aducanumab, the dose for ApoE ε4+ patients was increased from 6 to 10 mg/kg in the high-dose group (Budd Haeberlein et al., 2022).Because EMERGE started later and more patients were enrolled after this amendment, more patients received high doses in EMERGE than ENGAGE, which potentially resulted in better efficacy in EMERGE.
In 2021, although the FDA's advisory committee voted against approval, the FDA conditionally approved aducanumab to treat AD patients with the requirement of a confirmatory phase IV trial.Aducanumab became the first drug approved by the FDA to treat AD in 2003.However, the efficacy of aducanumab remains controversial due to inconsistent results between EMERGE and ENGAGE (Budd Haeberlein et al., 2022;Knopman et al., 2021).Only those patients who enrolled in continuing clinical trials of aducanumab can be reimbursed for their treatment, according to the Centers for Medicare & Medicaid Services, resulting in few insurers having covered aducanumab (Kritz, 7 Apr, 2022).

Lecanemab
In July 2023, the FDA fully approved lecanemab (formerly BAN2401) for the treatment of AD.Lecanemab is a humanized IgG1 antibody of murine mAb158 with an undisclosed conformation epitope but targets between amino acids 1 and 16 of the Aβ peptide.It binds to soluble Aβ protofibrils with high affinity (Englund et al., 2007), which have been shown to be more neurotoxic than Aβ monomers and insoluble fibrils (Haass and Selkoe, 2007;Lublin and Gandy, 2010;Sehlin et al., 2012).In September 2022, a phase III clinical study of lecanemab, Clarity AD (NCT03887455), was completed by Eisai and Biogen.This phase III trial included 1795 patients with early AD who were randomized 1:1 to receive 10 mg/kg lecanemab or placebo intravenously once every 2 weeks over 18 months.Strikingly, the Clarity AD trial had achieved the primary end point and all key secondary end points (van Dyck et al., 2023).
After 18 months of treatment, lecanemab significantly slowed cognitive decline on CDR-SB, the primary end point, with a 27% decline compared with the placebo (p < 0.001).The lecanemab group showed a significant difference from the placebo group as early as 6 months after the treatment, suggesting that early treatment can provide early benefit.In terms of safety, the incidence of ARIA-E in patients with lecanemab treatment was 12.6%, and 2.8% of patients were symptomatic, whereas the incidence of ARIA-E was 9.9% in the placebo group, without symptomatic patients.The incidence of ARIA-H was 17.3% in the lecanemab group, compared to 9.0% in the placebo group.In the lecanemab group, the incidence of ARIA-E and ARIA-H was apparently affected by the ApoE ε4 allele.ApoE ε4 homozygotes showed a higher frequency of ARIA-E and ARIA-H than ApoE ε4 heterozygotes and noncarriers.Although the incidence of lecanemab-associated ARIA appears to be numerically lower than that in other clinical trials testing Aβ mAbs, its safety still causes great discussion, as three patients died in the lecanemab group after experiencing bleeding or swelling in the brain (Mast, 2022;Pillar, 2022aPillar, , 2022b)).

Donanemab
Donanemab is a humanized IgG1 antibody that recognizes the Nterminal pyroglutamate Aβ epitope (amino acids 3[pE]− 7) that is present only in established Aβ plaques (Demattos et al., 2012;Kuo et al., 1997;Saido et al., 1995).In the phase II clinical study Trailblazer-Alz (NCT03367403), patients were given placebo or 700 mg donanemab for the first three doses intravenously every 4 weeks followed by 1400 mg donanemab for 72 weeks.Donanemab performed better than placebo in the primary endpoint on the Integrated Alzheimer's Disease Rating Scale (iADRS).Treatment with donanemab reduced mean amyloid plaque levels from 107.6 CL baseline to 22.54 CL and resulted in amyloid negativity in 67.8% of patients (Mintun et al., 2021).Very recently, donanemab has demonstrated positive results in the phase III Trailblazer-Alz2 study (NCT04437511), which enrolled 1736 patients with MCI or mild dementia due to AD (Sims et al., 2023).Primary analysis showed that donanemab met the primary endpoint, and all secondary endpoints in the low/medium tau population (n = 1182) over 18 months, with iADRS, the primary endpoint, showed 35.1% slowed clinical decline, and CDR-SB, an important secondary endpoint, showed 36.0%slowed clinical decline.When patients (n = 552) with high tau expression were added, which represents a later stage of AD progression, combined analysis of iADRS and CDR-SB slowed the clinical decline by 22.3% and 28.9%, respectively.At 76 weeks, in the low/medium tau population, donanemab reduced 88.0 CL brain amyloid plaque, with 80.1% of the population reaching amyloid clearance, whereas the placebo group increased 0.2 CL amyloid plaque.In the combined population, while 87.0 CL brain amyloid plaque decreased in the donanemab group, with 76.4% of the population reaching amyloid clearance, only 0.67 CL amyloid plaque decreased in the placebo group.However, PET imaging showed that alterations in tau deposition in the frontal cortex X. Guo et al. from baseline to week 76 did not significantly differ in patients with low/medium tau populations or in all populations.Plasma phosphorylated-tau217 (P-tau217) was significantly reduced from baseline in donanemab-treated patients compared to the placebo group.For safety, ARIA-E occurred in 24.0% of participants in the donanemab group, 6.1% were symptomatic, and the incidence of ARIA-H was 31.4% in the treatment group.Unfortunately, three patients with serious ARIA events died during or after the trial.Currently, another phase III clinical trial, Trailblazer-Alz3 (NCT05026866), is underway, which will enroll 3300 patients with preclinical AD to further evaluate the safety and efficacy of donanemab.

Clinical efficacy appears to be positively correlated with clearance of amyloid plaques
The phase III clinical trials showed that high-dose aducanumab (in EMERGE), lecanemab and donanemab (in the low/medium tau population) slowed 22%, 27% and 35.1% cognitive decline in patients at early stages of AD, respectively.Although all of these treatments lowered the amyloid level below the threshold for amyloid positivity of 24.1 CL (Navitsky et al., 2018), they showed differential abilities in clearing amyloid.While high-dose aducanumab (in EMERGE) reduced the mean amyloid level from baseline to 21 CL, lecanemab and donanemab (in the low/medium tau population) decreased amyloid to 22.44 CL and 14.4 CL, respectively.In contrast, gantenerumab and high-dose aducanumab (in ENGAGE) only reduced the mean amyloid level to 34.2-51.1 CL and 37 CL, respectively, and both failed to significantly slow cognitive decline in AD patients in the phase III trials.At the 15th CTAD, researchers compared the CDR-SB scores of people whose amyloid levels were below the positive amyloid threshold with those of people who were not, and found that the former notched a cognitive benefit over the latter in the gantenerumab treatment group.Group-level analysis shows that the effect of aducanumab treatment on CDR-SB is also associated with brain Aβ plaque levels (Budd Haeberlein et al., 2022).In addition, the results of exploratory post hoc analyses support that complete removal of plaque is the key to slow AD-associated cognitive decline.Indeed, other anti-Aβ mAbs developed earlier failed to show efficacy on cognition and removal of amyloid, including bapineuzumab (Salloway et al., 2014), solanezumab (Honig et al., 2018) and crenezumab (Ostrowitzki et al., 2022).

There is a dose-and time-dependent relationship between amyloid clearance and the mAbs
In EMERGE and ENGAGE (Budd Haeberlein et al., 2022), amyloid reduction in the brain upon aducanumab treatment is dose-and time-related.In the phase III Scarlet Road trial (Ostrowitzki et al., 2017), exploratory analyses of 225 mg and 105 mg gantenerumab and placebo groups showed a dose-dependent antibody effect on clinical efficacy, amyloid clearance and biomarker change (Ostrowitzki et al., 2017).In a phase II trial (NCT01767311), lecanemab was given in five different doses and showed dose-and time-dependent reductions in brain amyloid after 12 and 18 months of treatment (Swanson et al., 2021).In a dose-escalation study of donanemab (NCT01837641), only the highest-dose (10 mg/kg) group showed alterations in amyloid PET (Lowe et al., 2021).The association between better efficacy and higher dosage may be due to the low efficiency of mAbs crossing the blood-brain barrier (BBB).It is known that only a small fraction of mAbs can enter the human brain when they are administered intraperitoneally or intravenously, reaching levels in the brain equivalent to only 0.01-0.11% of that in the plasma (Banks et al., 2002;Levites et al., 2006;St-Amour et al., 2013).Therefore, high-dosage mAbs are needed to achieve clinical efficacy.However, treatment with high-dose mAbs may result in side effects such as ARIA, serum sickness, acute anaphylaxis and cardiotoxicity (Hansel et al., 2010) It is worth developing novel antibody drugs with better potency to cross the BBB, such as single-chain variable-fragment antibodies (scFvs) and nanobodies, which have smaller molecular weights.Another strategy to enhance antibody delivery into the brain is constructing chimeric antibodies that contain Fab fragment of an anti-transferrin-receptor monoclonal antibody (TfR-Fab) that can bind to transferrin-receptor to increase receptor-mediated transcytosis.For instance, an anti-Aβ mAb mAb31 fused with TfR-Fab has shown increased concentration in the brain fifty-five fold than the parent antibody (Niewoehner et al., 2014).
In addition, exosomal membrane-coated nanosystems may offer an effective pathway for antibody delivery into the brain.Exosomal membrane-coated nanosystems are enriched with lipid rafts and proteins responsible for membrane transport and fusion, and can facilitate direct membrane fusion between exosomes and target cells.It has been shown that exosomalmembrane-coated nanosystems combined with rabies virus glycoprotein peptide, a neuron-targeting ligand, can penetrate through the BBB and transport the loaded cargo into neuronal cells (Lopes et al., 2023).

Aβ epitopes and forms recognized by mAbs are important
There are many Aβ species, including monomers, oligomers, protofibrils, and insoluble fibrils in plaques.Among these species, Aβ oligomers and protofibrils have been shown to be highly toxic to neurons (Haass and Selkoe, 2007;Nilsberth et al., 2001;Sehlin et al., 2012).In addition, posttranslationally modified Aβ forms have been identified recently.For instance, N-terminus truncated Aβ with pyroglutamate modification (AβpE3) is found in plaques and can trigger neurodegeneration in mice (Bayer, 2022).mAbs targeting different Aβ forms may have different effects on Aβ aggregation and plaque clearance.The mAbs developed earlier, including bapineuzumab, solanezumab and crenezumab, mainly target monomeric, oligomeric or fibrillar forms of Aβ.All of these mAbs failed to show efficacy in the removal of amyloid and cognition, probably due to their weaker ability to inhibit Aβ aggregation (Honig et al., 2018;Linse and Scheidt, 2020;Ostrowitzki et al., 2022;Salloway et al., 2014).In comparison, mAbs, including gantenerumab, aducanumab, and lecanemab, preferentially bind to fibrillar Aβ forms, such as soluble protofibrils and insoluble fibrils.In particular, aducanumab and gantenerumab tend to bind fibrils more than protofibrils, whereas lecanemab binds protofibrils tenfold more strongly than fibrils (Söderberg et al., 2022).The two former mAbs showed less effect on plaque clearance and cognition clinically compared to lecanemab (Budd Haeberlein et al., 2022;Releases, 2022;van Dyck et al., 2023).Donanemab mainly recognizes plaque-specific AβpE3 (Demattos et al., 2012;Kuo et al., 1997;Saido et al., 1995) and has shown the best efficacy in plaque removal and cognition to date (Sims et al., 2023).Therefore, the binding affinities of mAbs for different Aβ species may play a decisive role in clinical effectiveness.

ARIA and brain atrophy cannot be ignored
ARIA was the most common side effect observed in patients treated with anti-Aβ mAbs.In particular, lecanemab and donanemab face significant safety concerns despite their effect on alleviating cognitive decline in clinical trials.Three patients with lecanemab treatment died after experiencing bleeding or swelling in the brain in the phase III trial (NCT03887455) (Mast, 2022;Pillar, 2022aPillar, , 2022b)).This is thought to be associated with ARIA, ApoE ε4, brain inflammation, concomitant use of antithrombotic drugs, and cerebral amyloid angiopathy (Piller, 2023).Likewise, three participants in the donanemab treatment group died due to serious ARIA events in the phase III trial (NCT04437511) (Sims et al., 2023).
The mechanisms of ARIA have yet to be elucidated but may be relevant to inflammatory responses or cellular immunity induced by Fc domain activation (Fuller et al., 2014).Activating FcγRs of microglia following therapeutic antibodies may be responsible for a pro-inflammatory response, including the release of cytokines such as TNFα and IL-1β as well as nitric oxide, all of which have potent vasculature-and neuron-damaging properties (Le et al., 2001;Ross et al., 2003;Yang et al., 2002).The inflammatory response is affected by antibody subclasses.The Fc domain of human IgG1 has a higher affinity for FcγRs and induces a stronger proinflammatory response than human IgG4 (Bruhns et al., 2009).To reduce the risk of ARIA, an IgG4 backbone isotype that does not activate microglia was selected in the development of crenezumab (Table 1).Although crenezumab treatment showed lower incidence of ARIA than other anti-Aβ mAbs, it showed no efficacy in the removal of amyloid in a phase III clinical trial (Ostrowitzki et al., 2022).This may be due to the Aβ species targeted by crenezumab or the lack of effector domains to activate the immune system to clear amyloid plaques.In addition to FcγR mediated-mechanisms, complement pathway may also contribute to ARIA.When treated with lgG1 anti-Aβ mAbs in the APP/PS1 transgenic mouse model, both CR3 and FcγRIIb were activated via the mAb-Aβ-complement complex, resulting in increased expression of inflammatory factors such as TNF-α, IL-6 and IL-1β in the microglia (Sun et al., 2023).To avoid the cascade of complement activation brought by the Fc fragment of the antibody, scFvs may represent an alternative tool for the development of AD treatment (Chia et al., 2018).In a recent study, researchers made a chimeric fusion protein, αAβ-Gas6, which contains scFv of aducanumab and a receptor binding domain of growth arrest-specific 6 (Gas6).Compared with aducanumab, αAβ-Gas6 substantially cleared Aβ through TAM (TYRO3, AXL, and MERTK) receptor-dependent phagocytosis as well as by activating phagocytosis of microglia and astrocytes in a 5×FAD mouse model, without inducing TNF, IL-6, and IL-1β and NF-κB-mediated inflammatory responses or reactive gliosis (Jung et al., 2022).These findings suggest that modification of antibody structures may be able to overcome some side effects, such as ARIA of traditional antibodies.
In addition to ARIA, alterations in brain atrophy are important phenomena associated with anti-Aβ mAb treatment.Structural magnetic resonance imaging (MRI) showed that whole brain and cortical thickness were smaller in the lecanemab group than in the placebo group.In contrast, atrophy of the hippocampal region was slowed upon lecanemab treatment (NEWS, 2022(NEWS, , 2023)).Similarly, whole brain volume was greater decreased and hippocampal volume was less reduced in patients treated with donanemab compared with the placebo group.In addition, changes in brain volume were observed in patients treated with other anti-Aβ mAbs (Alves et al., 2023).Further investigations are needed to clarify how anti-Aβ mAbs differentially affect brain atrophy.

Passive immunotherapy targeting tau
In addition to amyloid plaques, NFTs formed by aggregated tau protein are another primary pathological feature of AD.Tau is a microtubule-associated protein encoded by the MAPT gene.In the human central nervous system (CNS), there are six isoforms of tau, the longest of which has 441 amino acids.These isomers differ in the Nterminal acidic domain and C-terminal microtubule binding region but are conserved in the proline-rich domain (PRD) at positions 151-243 (Wang and Mandelkow, 2016).In the normal brain, tau is highly soluble and predominantly expressed in neuronal cells (Shin et al., 1991), where it binds to tubulin to promote microtubule assembly, regulate microtubule stabilization, and participate in axonal transport (Wang and Mandelkow, 2016).Under AD pathological conditions, the phosphorylation level of tau increases significantly, leading to the dissociation of tau from microtubules and gradually aggregate to form oligomers, fibrils, filaments, and eventually NFTs (Medeiros et al., 2011).Phosphorylated tau is hypothesized to play a key role in driven tau accumulation, synaptic dysfunction and neuronal loss (Wang and Mandelkow, 2016).The mechanism of tau hyperphosphorylation has not been fully elucidated and may be related to imbalanced activity of kinases, such as glycogen synthase kinase-3β (GSK-3β) and cyclin dependent kinase 5 (CDK5), and phosphatases, such as protein phosphatase protein phosphatase 2 (PP2A) (Medeiros et al., 2011).It has been shown that several phosphorylated tau species such as P-tau181, P-tau217 and P-tau231, are increased in the CSF and blood of AD patients 20 years before the onset of symptoms (Barthélemy et al., 2020;Suárez-Calvet et al., 2020).Interestingly, positron emission tomography analyses identified a stronger relationship between tau accumulation and dementia severity than amyloid deposits (Hanseeuw et al., 2019;Rubinski et al., 2020).
Multiple lines of evidence support that tau mediates Aβ-induced neurotoxicity.Tau (1− 368) and APP (586− 695) fragments, cleaved by δ-secretase, additively drive BACE1 upregulation and Aβ generation (Zhang et al., 2021).Reducing endogenous tau blocked excitotoxin-induced neuronal dysfunction in hAPPJ20 AD model mice, but did not affect Aβ deposition (Roberson et al., 2007).Tau deletion rescued neuronal loss and synaptic loss as well as reduced amyloid plaque burden in APP/PS1 mice (Leroy et al., 2012).In addition, the inhibitory effect of Aβ on axonal transport was dependent on tau (Vossel et al., 2010).Taken together, downregulation of tau levels may alleviate the neurodegenerative pathology of AD.However, it should be noted that tau has normal physiological functions.Tau-knockout mice develop age-dependent neurological problems (Lei et al., 2012), although they do not have significant growth and developmental problems.
Passive immunotherapy targeting tau protein is another attractive strategy for AD intervention.Development of anti-tau mAbs has focused on blocking tau aggregation, seeding and spreading (Albert et al., 2019;MERCKEN et al., 2018).Although tau is primarily localized inside neurons, recent studies have indicated that tau can be released into the extracellular space, and secreted tau may contribute to propagation of tau pathology (Kim et al., 2010;Yamada et al., 2011).Therefore, mAbs against tau may reduce tau pathology by targeting both intracellular and extracellular tau.Membrane-mediated endocytosis may be involved in antibody internalization by neurons.For instance, anti-P-tauS422 antibody MAb86 bound to membrane-associated P-tauS422 and was then endocytosed into neurons and eventually cleared intracellularly via the lysosomal pathway (Collin et al., 2014).In addition, it has been shown that tau-mAb complexes in the extracellular space can be taken up by neurons via FcγRII/III receptor-mediated clathrin-dependent pathway and cleared by the endosome/autophagosome/lysosome system (Congdon et al., 2013;Gu et al., 2013).Also in the extracellular, tau-mAb complexs was readily taken up and degraded by murine microglia via Fc receptors and functional lysosomes (Andersson et al., 2019;Luo et al., 2015).Receptors inside neurons also contribute to the effect of anti-tau mAbs.For example, TRIM21 can be rapidly recruited to tau-mAb complexes in neurons and neutralize tau seeds depending on the proteasome and valosin-containing protein (McEwan et al., 2017).
The development of anti-tau mAbs in clinics is still in the early stages.To date, four anti-tau mAbs have been tested in phase II trials, including semorinemab, tilavonemab, gosuranemab and zagotenemab.Table 2 and Fig. 2 show key information relevant to tau mAbs that have been examined or are in clinical trials.

Gosuranemab
Gosuranemab is the first anti-tau antibody tested in clinical trials.It has been shown that extracellular tau, which consists of a series of Nterminal tau fragments, could increase Aβ levels in a continuous vicious cycle (Bright et al., 2015).Gosuranemab (formerly BIIB092, IPN002) is a humanized IgG4 mAb targeting N-terminal amino acids 15-22 of tau and can bind to monomeric, fibrillar and insoluble tau with high affinity (Sopko et al., 2020).Gosuranemab significantly reduced unbound N-terminal tau fragments in the CSF of antemortem patients with AD and progressive supranuclear palsy (PSP) according to an immunodepletion assay (Sopko et al., 2020).In the phase II trial PASSPORT (NCT03068468), gosuranemab treatment did not meet the primary endpoint and secondary endpoints in PSP patients (Dam et al., 2021).The phase II trial TANGO (NCT03352557) was terminated in June 2021 because gosuranemab treatment did not meet the primary endpoint in patients with MCI due to AD or mild AD (Release, 2021).There was no statistically significant difference in Tau PET between the gosuranemab and placebo groups at week 78, indicating that gosuranemab had no X.Guo et al. clinical efficacy in decreasing tau accumulation in the brain.

Tilavonemab
Tilavonemab (formerly ABBV-8E12), a humanized IgG4 antibody recognizing amino acids 25-30 of tau, binds to extracellularly soluble  X. Guo et al. tau (Yanamandra et al., 2013).In a tau P301S transgenic mouse model, HJ8.5 (murine version of tilavonemab) significantly improved cognitive function, markedly reduced hyperphosphorylated, aggregated, and insoluble tau, and blocked the development of tau seeding activity (Yanamandra et al., 2013).To test the efficacy of tilavonemab in PSP patients, a phase II trial, ARISE (NCT02985879), was launched in December 2016.This trial was terminated for futility analysis (Höglinger et al., 2021).Similarly, tilavonemab failed to slow cognitive decline and reduce tau deposition in patients with early AD in another phase II trial (NCT02880956), although target engagement of tilavonemab was confirmed by lower free tau in CSF and higher plasma total tau in treatment groups (Florian and Wang, 2023).

Zagotenemab
Zagotenemab (formerly LY3303560) is a humanized IgG4 anti-tau antibody derived from mouse mAb MC1 (Jicha et al., 1997) that recognizes residues 7-9 and 312-322 of tau (ALBERTO et al., 2016).The affinity of LY3303560 for soluble tau aggregates (K D <220 pM) is 1000-fold that of monomers (K D ≈235 nM) (Alam et al., 2017).In P301S mice, MC1 reduced the levels of hyperphosphorylated insoluble tau protein and NFTs and enhanced microglia-mediated tau degradation (Luo et al., 2015).The efficacy of zagotenemab has been tested in a phase II trial (NCT03518073) that enrolled patients with early symptomatic AD.In October 2021, Lilly announced that the trial did not meet its primary endpoint and halted the development of zagotenemab.The results of this trial are yet to be published.

Semorinemab
Semorinemab (formerly RO7105705), a humanized IgG4 mAb recognizing amino acids 6-23 of tau, binds to both monomeric and oligomeric tau, regardless of its phosphorylation status (Ayalon and Lee, 2021).The murine version of semorinemab, muMTAU, significantly reduced tau accumulation in hTau.P301L mouse model (Ayalon and Lee, 2021).Semorinemab treatment showed target engagement in nonhuman primates and AD patients with increased plasma total tau concentrations (Ayalon and Lee, 2021).However, semorinemab did not improve outcomes in the first phase II trial TAURIEL (NCT03289143) by Genentech, which enrolled patients with prodromal to mild AD (Teng et al., 2022).In February 2019, Roche launched another phase II study, LAURIET (NCT03828747), which showed that semorinemab treatment slowed cognitive decline by 42.2% in patients with moderate AD on one of two primary endpoints, the Alzheimer's Disease Assessment Scale-cognitive 11-item (ADAS-Cog11) (Monteiro et al., 2023).This is the first positive result for tau-targeted mAbs in clinical trials of AD.However, semorinemab did not achieve another primary endpoint, the Alzheimer's Disease Cooperative Study Group-Activities of Daily Living Inventory (ADCS-ADL), and did not show any improvement on secondary end points, including the Mini-Mental State Exam (MMSE) and CDR-SB.In addition, semorinemab failed to reduce tau accumulation (Monteiro et al., 2023).To better understand why semorinemab had no effects on the coprimary endpoint and secondary end points, this trial will be converted into an open-label portion until October 2023.

BIIB076
BIIB076 is a human recombinant IgG1 monoclonal antibody derived from NI-105.6C5, which was screened from antibody sequences of healthy human B cells, targeting amino acids 125-131 in the mid-region of tau (WEINREB et al., 2013).BIIB076 recognizes monomeric and fibrillar tau forms, as well as tau derived from AD patients and healthy individuals (Czerkowicz et al., 2017).BIIB076 blocked tau seed uptake and subsequent tau aggregation in neurons and reduced neuronal tau spreading (Nobuhara et al., 2017).BIIB076 binds to recombinant human or cynomolgus monkey tau with subnanomolar affinities.Treatment of cynomolgus monkeys with high-dose BIIB076 significantly reduced the levels of total and free tau in the CSF (Czerkowicz et al., 2017).Although a phase I (NCT03056729) study had been completed, Biogen stopped the development of BIIB076 for business reasons.

PNT001
PNT001 is a human IgG4 mAb from Pinteon Therapeutics that recognizes the cis-isomer of tau phosphorylated at T231 (Cis pT231).Cis pT231 is detectable in the brain tissues of AD patients and is able to promote tau aggregation but is resistant to dephosphorylation and degradation (Lu et al., 2016;Nakamura et al., 2012).PNT001 has a high affinity for the cis-pT231 peptide with 0.3 to 3 nM K D and specifically binds to NFT-like structures in tauopathic patients (Foster et al., 2023).An in vitro study showed that murine PNT001 (mPNT001) could lower the seeding ability of tau derived from brain homogenates of rTg4510 mice, which overexpress the pathogenic human tau 0N4R P301L mutant (Santacruz et al., 2005).In addition, mPNT001 treatment reduced the mRNA levels of neuroinflammatory markers, improved synaptic and behavioral outcomes, and decreased serum NfL and the formation of mature NFTs in rTg4510 mice (Foster et al., 2023).Data from a phase I clinical trial (NCT04096287) released by Pinteon showed that PTN001 produced dose-linear blood and CSF concentrations and was well tolerated.No phase II clinical trials are currently underway.

RG7345
RG7345 is a humanized mAb recognizing tau phosphorylated at the S422 site (pS422).Phosphorylated-tau422 (P-tau422) has been detected in paired helical filaments (PHF) and intraneuronal NFTs but not in naive tau protein (Bussière et al., 1999;Caillet-Boudin and Delacourte, 1996;Hasegawa et al., 1996).(Augustinack et al., 2002;Collin et al., 2014).Treatment with the anti-P-tau422 antibody MAb86 reduced the accumulation of tau in the brains of triple transgenic TauPS2APP mice, which overexpress mutant APP, PSEN2, and MAPT (Collin et al., 2014).A phase I trial (NCT02281786) testing the safety of RG7345 in healthy participants was completed in 2015.However, the relevant data have not been released to the public.

Bepranemab
Bepranemab (formerly UCB0107) is a humanized IgG4 mAb derived from antibody D targeting amino acids 235-250 of tau, recognizing monomeric and PHF tau (Courade et al., 2018).An in vitro study showed that antibody D exhibited the best potency in blocking tau seeding when compared with antibody A (against tau 15-24), antibody B (against tau 25-30), antibody C (against tau pS202 + T205), antibody E (against tau pS422) and antibody F (against tau 312-322) (Courade et al., 2018).In Tg30tau mice and hTau.P301L transgenic mice, bepranemab neutralized pathological tau forms and blocked the spread of tau pathology, whereas an antibody targeting the N-terminus of tau showed less effect on tau pathology (Albert et al., 2019).A phase II study (NCT04867616) of bepranemab was started in 2021 to test the efficacy of this mAb in patients with MCI or mild AD and is expected to be finished in 2025.

JNJ-63733657
The microtubule-binding domain (MTBR) of tau is considered to be necessary for functional tau seeding (Falcon et al., 2015) and constitutes the protease-resistant core of terminal filamentous tau deposits (Fitzpatrick et al., 2017).JNJ-63733657, a humanized IgG1 monoclonal antibody generated by Janssen, binds MTBR of tau with high affinity for tau phosphorylated at residue 217 (pT217) between G204-K225 and for phosphorylated PHF tau (27 pM) (MERCKEN et al., 2018).JNJ-63733657 was able to eliminate tau seeding ability in vitro and to inhibit the spreading of tau pathology in hTau.P301L mouse model X.Guo et al. (MERCKEN et al., 20 Sep 2018).There was a dose-dependent relationship between decreased levels of pTau in CSF and JNJ-63733657 in the phase I study (Bijttebier et al., 2021).The phase II trial (NCT04619420) of JNJ-63733657 in patients at early stages of AD is undergoing and planned to be completed in 2025.

E2814
E2814 is a humanized IgG1 monoclonal antibody recognizing an HVPGG epitope in the MTBR of tau and binds to pathological tau structures in the AD brain (Roberts et al., 2020).Functionally, E2814 inhibited tau aggregation in vitro and immunodepleted a variety of tau species containing MTBR.E2814 treatment reduced the accumulation of insoluble tau in a mouse model of tau seeding and transmission (Roberts et al., 2020).A phase I/II trial (NCT04971733) of E2814 was launched by Eisai in 2021, which enrolled patients with mild to moderate cognitive impairment due to dominantly inherited AD and was planned to be finished by 2024.

Lessons learned from anti-tau mAbs
In recent years, tau-targeted passive immunotherapy for AD has received intensive attention.Anti-tau mAbs developed in earlier years all failed to show efficacy in patients with early AD in clinical studies, including semorinemab, tilavonemab, gosuranemab and zagotenemaball (a.k.a., STGZ).It remains uncertain why these mAbs are noneffective clinically.One possibility is that they are not able to effectively stop the spread of tau pathology.A recent study has shown that the N-terminus is largely truncated in tau secreted by induced pluripotent stem cell (iPSC)-derived neurons or in human CSF (Sato et al., 2018).Because the epitopes of STGZ mAbs are located in the N-terminal region of tau, it is likely that these mAbs cannot efficiently bind and clear tau seeds truncated in the N-terminus.Indeed, mAbs against the tau N-terminus showed less effect on tau propagation in mice than mAbs targeting the mid-region of tau (Albert et al., 2019).Industries have now focused on developing mAbs targeting the tau mid-region and phosphorylated tau, while most of these mAbs are still in the early stages of clinical trials.It should be noted that naive tau has physiological function, and therefore, it is worth developing mAbs that specifically target pathogenic forms of tau, such as tau oligomers and fibrils, in the future.
The effector function of mAbs can also affect efficacy.STGZ are all IgG4 backbone isotype that has a lower affinity for FcγRs (Bruhns et al., 2009) and is deliberately engineered not to recruit immune microglia to reduce the incidence of side effects observed with anti-Aβ mAbs.Similar to the failure of crenezumab in targeting Aβ, this design may affect the antibody's ability to remove tau from the brain.In contrast, new generations of anti-tau mAbs are almost all IgG1 (Table 2).As different FcγRs play different roles in neurodegeneration by a number of mechanisms (Fuller et al., 2014), it is critical for Fc domain designs to balance activation and inhibitory effect of FcγRs to obtain higher levels of clearance of tau accumulation with fewer side effects.
During preclinical evaluation of mAbs, it should be noted that mouse models cannot fully recapitulate the pathogenesis of AD in humans.For instance, tau ubiquitination and acetylation are weak in mouse models (Wenger et al., 2023).Tau transgenic mice do not overexpress all tau isoforms seen in humans.More importantly, most AD mouse models mimic familiar AD with dominant mutations but not sporadic AD.Therefore, it may be helpful for evaluating mAbs in additional models such as brain organoids induced from patient-derived iPSCs in the preclinical stage.

Passive immunotherapy targeting neuroinflammation
Neuroinflammation is an immune response activated by glial cells in the CNS that usually occurs in response to stimuli such as nerve injury, infections and toxins or in response to autoimmunity.Microglia are the primary immune cells in the CNS.In the physiological state, microglia can remove metabolites and toxic substances, maintain the hemostasis of the microenvironment in the brain.Under AD-associated pathological conditions, on one hand, microglia can phagocytose and degrade Aβ by receptors such as triggering receptor expressed on myeloid cells 2 (TREM2) (Udeochu et al., 2018); on the other, sustained accumulation of Aβ and tau induces an inflammatory response in microglia that releases large amounts of inflammatory factors, leading to sustained neuronal damage, which in turn leads to continuous deposition of Aβ plaques and NFTs, thereby accelerating the course of AD (Leng and Edison, 2021).
Targeting microglial phagocytosis may be a viable treatment option for AD.The phagocytic activity of microglia in pathological states depends on the activation of signaling pathways mediated by TREM2, which is expressed on the surface of microglia, and TREM2 deletion increases neuronal atrophy in PS2APP mice (Meilandt et al., 2020).The binding of ligands to TREM2, such as lipids, ApoE and Aβ, can trigger its downstream signaling pathway involving the phosphorylation of spleen tyrosine kinase (SYK), thereby regulating microglial survival, proliferation, phagocytosis, and the immune response (Atagi et al., 2015;Wang et al., 2015;Yeh et al., 2016;Zhao et al., 2018).GWASs have also identified numerous risk genes for late-onset AD that are highly expressed in microglia, highlighting the therapeutic potential of modulating the microglia-mediated immune response in AD (Lewcock et al., 2020;Wightman et al., 2021).Among these microglial risk genes, TREM2 has been intensively investigated in recent years.TREM2 variants such as R47H significantly increase AD risk (Guerreiro et al., 2013;Jonsson et al., 2013), probably due to partial loss of function (Carmona et al., 2018).In contrast, activation of the TREM2 pathway by TREM2 agonistic antibodies demonstrated beneficial effects in multiple AD models (Cheng et al., 2018;Fassler et al., 2021;Price et al., 2020;Schlepckow et al., 2020;Wang et al., 2020;Zhao and Xu, 2022).The development of TREM2 agonists is ongoing, and two of them have entered clinical trials.

AL002
AL002 is a humanized IgG1 mAb recognizing the extracellular region of human TREM2.Administration of AL002a (a variant of AL002 that recognizes mouse TREM2) activated TREM2 signaling, recruited microglia to amyloid plaques, decreased amyloid deposition, altered the plaque-associated gene expression signature and improved cognitive function in a 5×FAD mouse model with amyloid deposition (Price et al., 2020).Prolonged administration of AL002c, a preclinical variant of AL002, induced microglial proliferation, reduced inflammatory signaling and response, attenuated neurotoxicity and normalized behavior in mouse models expressing either the R47H variant or common variant (CV) of TREM2 (Wang et al., 2020).Target engagement of AL002 was demonstrated by lower soluble TREM2 (sTREM2) and higher soluble CSF-1R (sCSF-1R) in CSF in the treatment group in a phase I study (NCT03635047) (Wang et al., 2020).Currently, a phase II trial (NCT05744401) of AL002 by Alector and AbbVie is underway in patients with early AD, with an amendment of the protocol to exclude ApoE ε4 homozygote patients due to serious adverse events observed in this population.This study will run until December 2025.

DNL919
Effective delivery of antibodies into the central nervous system has been greatly restricted by the BBB.To promote antibody transcytosis across the BBB, Denali developed a TREM2 agonist antibody containing a transferrin-receptor binding sequence in the Fc domain, DNL919, with antibody transfer vehicle (ATV) technology.The murine precursor of DNL919, 4D9, reduced the shedding of TREM2 on the microglial surface, activated the SYK signaling pathway to stimulate microglial survival and increased the uptake of Aβ (Schlepckow et al., 2020).The concentration of DNL919 in the brain is sixfold higher than that in the non-ATV version and boosted microglial activity in 5×FAD mice (van Lengerich and Zhan, 2023).A phase I study (NCT05450549) was launched in July 2022.In the phase I study, DNL919 triggered reversible hematologic effects at the highest dose tested.On August 8, 2023, Denali and Takeda terminated the development of DNL919 (Details, 2023).

Passive immunotherapy based on other inflammatory/immune targets
In addition to TREM2-based immunotherapies, several mAbs against other inflammatory/immune targets are in clinical trials, including CD38, semaphorin 4D (SEMA4D), galectin-3, and CD33.Daratumumab is an FDA-approved mAb targeting CD38 for the treatment of multiple myeloma with immunomodulatory activity against CD38-positive nonplasma cells (Xia et al., 2016).It has been shown that CD38 expression is significantly increased in CD8 + T effector memory CD45RA + (T EMRA ) cells in the blood of patients at early stages of AD, and CD8 + T EMRA cells exhibit pro-inflammatory and cytotoxic functions (Gate et al., 2020;Schindowski et al., 2007).Currently, a phase II study DARZAD (NCT04070378) of daratumumab in patients with mild to moderate AD is ongoing and is expected to be completed in June 2024.SEMA4D is a transmembrane homodimeric protein belonging to the semaphorin family.SEMA4D is traditionally known to be involved in the regulation of axonal guidance in the brain (Chapoval, 2018).The expression of SEMA4D is upregulated in neurons of AD patients or after CNS injury, where it can activate glia after association with glial receptors, resulting in the release of inflammatory cytokines and inducing the process collapse of neurons and oligodendrocytes (Evans and Mishra, 2022;Moreau-Fauvarque et al., 2003;Smith et al., 2015).Based on these findings, a phase I and II study (NCT04381468) has been launched to test the efficacy of pepinemab, a mAb against SEMA4D, in patients with mild dementia due to AD.This trial is expected to be completed in June 2024.Galectin-3 is a β-galactosidase-binding protein involved in macrophage activation and antimicrobial immune responses (Dong et al., 2018).Galectin-3 expression is increased in microglia associated with amyloid plaques in AD patients and mouse models.Deletion of galectin-3 in 5×FAD mice decreased amyloid plaque burden and improved cognitive function (Boza-Serrano et al., 2019).Based on these facts, the clinical efficacy of TB006, a humanized mAb targeting galectin-3, has been tested in a phase I/II (NCT05074498) study involving patients with mild to severe AD.This trial was completed in October 2022 and narrowly missed statistical significance on its primary outcome, as reported at the 15th CTAD.Interestingly, in a follow-up OLE study, the sponsor observed evident cognitive improvement in 47% of participants with three-month treatment of TB006.Detailed results are yet to be published.CD33 is an AD risk gene abundantly expressed in microglia.It is a type I transmembrane protein belonging to the Ig superfamily (Duan and Paulson, 2020).The expression of CD33 is increased in the AD brain (Griciuc et al., 2013).CD33 may modulate the neuroinflammatory process via cross-talk between it and TREM2 in AD (Griciuc et al., 2019).AL003 is a mAb targeting CD33 and has been evaluated in healthy participants and patients with mild to moderate AD in a phase Ib study (NCT03822208) launched in March 2019.Although AL003 was shown to be generally safe and well tolerated in this trial (Maslyar et al., 2022), Alector and AbbVie terminated the development of AL003 for unknown reasons in June 2022.Table 3 shows important information from clinical trials with regard to mAbs targeting neuroinflammation.

Comparison of passive immunotherapy and other medications
In addition to passive immunotherapy, other strategies targeting Aβ and tau have been developed, including active immunotherapy, inhibitors to reduce Aβ generation, antisense oligonucleotides (ASOs) to reduce tau expression, GSK-3β inhibitors to inhibit tau phosphorylation, tau aggregation inhibitors and microtubule stabilizer.Different therapeutic strategies that have reached or completed clinical phase trials are shown in Table 4.
Since amyloid cascade hypothesis was proposed, strategies targeting Aβ pathology has been developed.Schenk et al. demonstrated that using of full-length human Aβ 1-42 as an immunogen attenuated Aβ deposition in the mouse brain (Schenk et al., 1999).This led to the development of the first active immunotherapeutic approach for AD.However, to date, active immunotherapies targeting Aβ have not succeeded.AN1792 is a vaccine consisted of full-length human Aβ 1-42 and was earliest tested in a phase II clinical trial (NCT00021723) in 2001.It was abandoned because of the emergence of T cell-mediated aseptic meningoencephalitis (Gilman et al., 2005).Second-generation vaccines had focused on the N-terminus of Aβ to avoid T cell epitopes at the C-terminus (Pride et al., 2008).Until now, amilomotide is the only Aβ vaccine had reached phase II/III trial (NCT02565511) began in 2015 planned to run until 2023, but was stopped by Novartis in 2019 for no reason.Another Aβ vaccine, ACI-24.060(phase Ib/II, NCT05462106), received FDA fast track designation on June 27, 2023 (Releases, 2023).As described above, because Aβ forms greatly affect the efficacy of mAbs, a potential issue for Aβ vaccine is whether and how it can promote the production of antibodies with high efficiency to clear Aβ deposits.Inhibition of secretases involved in Aβ generation is another stagey to reduce Aβ pathology.Unfortunately, none of the inhibitors for β-or γ-secretase demonstrate the efficacy for AD.For instance, treatment with β-secretase inhibitor verubecestat (NCT01739348, NCT01953601) or γ-secretase inhibitor semagacestat (NCT00594568, NCT00762411) worsened cognitive decline and showed little amyloid plaques clearance in AD patients in phase III trials (Doody et al., 2013;Egan et al., 2018).Semagacestat is even associated with more skin cancer and infections (Doody et al., 2013).These outcomes may be resulted from alterations in other downstream targets of the secretases (Hampel et al., 2021;Hur, 2022).
Active immunotherapies targeting tau have also been developed.Two tau vaccines, AADvac1 (phase II, NCT02579252) and ACI-35.030(phase Ib/IIa, NCT04445831), are currently in clinical trials with positive results (Alzforum, 2023;Novak and Kovacech, 2021).MAPT Rx , an antisense oligonucleotide targeting tau, reduced tau concentrations in CSF by over 50% from baseline after 24 weeks post-last dose in phase Ib trail NCT03186989 (Mummery et al., 2023).While antibodies stimulated by tau vaccines may recognize both total and pathogenic tau, ASOs primarily target tau mRNA but not pathogenic tau protein.Therefore, these strategies may interfere physiological function of total tau and generate the associated side effects.In addition, GSK-3β inhibitors, tau aggregation inhibitors and microtubule stabilizers for AD treatment have almost failed or need future study (Vaz and Silvestre, 2020).
Passive immunotherapy shows several advantages compared with these strategies (Table 4).First, to date, passive immunotherapy targeting Aβ is the only treatment option that can slow cognitive decline and reduce amyloid plaques in AD patients.Second, mAbs have high specificity for recognizing targets to avoid inhibiting other substrates, X. Guo et al. with relatively less side effects.Third, passive immunotherapy is more appropriate and effective for the elder people, who usually show reduced vaccine responsiveness.

Conclusions and perspectives
Aβ, tau, and neuroinflammation are immunotherapy targets focused on by industries for AD intervention.Passive immunotherapy targeting Aβ was launched decades ago and has reached milestone progress with full approval of lecanemab by the FDA very recently.While the development of mAb drugs targeting tau or immune modulators is at an early stage, several preclinical and clinical studies have shown promising results.
For the anti-Aβ strategy, it should be noted that even the mAbs (lecanemab and donanemab) only showed efficacy in patients with early AD.This may be because other factors, such as tau, have important contributions to neuronal loss in the later stages of AD.In support of this notion, donanemab only showed efficacy in AD patients with low/ medium tau pathology.Therefore, it is important to elucidate the clinical effect of anti-Aβ mAbs on neuronal loss, and it is worth testing the combined immunotherapy strategy targeting both Aβ and tau in patients with moderate symptoms or medium/severe tau pathology in the future.In addition, efficient Aβ-targeted immunotherapy is associated with a high incidence of ARIA and brain atrophy, and the underlying mechanisms need to be clarified.For anti-tau immunotherapy, most of the mAbs recognizing the N-terminal epitopes failed in clinical trials.The industries have now focused on developing mAbs targeting the tau midregion or phosphorylated tau, which may be able to stop tau seeding and spreading.
In addition to Aβ, tau, and neuroinflammation, there are multiple copathologies associated with aging in the brain that may influence the onset and progression of AD, such as TDP-43 pathology and α-synuclein pathology.TDP-43 dysfunction and cytoplasmic aggregation are the primary pathogenic mechanisms of amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) (Tamaki and Urushitani, 2022).TDP-43 accumulation is found in 75% of AD patients and may have an independent effect on the clinical progression of AD (Josephs et al., 2020).Highly expressed and misfolded α-synuclein constitutes Lewy body pathology, which is the primary pathology underlying Parkinson's disease (PD) (Atik et al., 2016) and Lewy body dementia (Sanford, 2018).Coexisting Lewy body pathology is present in more than 10% of preclinical AD patients with unimpaired cognition (Palmqvist et al., 2023) and is also observed in 24% of AD patients (Quadalti and Palmqvist, 2023), suggesting that Lewy body pathology may contribute to AD progression.A recent study found that lysosomal type II transmembrane protein 106B (TMEM106B) filaments form in an age-dependent manner in the brains of older adults with normal neurological systems, although there is currently no clear relationship with the disease (Schweighauser and Arseni, 2022).Therefore, targeting the above biomarkers together with Aβ and tau may offer a more effective therapeutic strategy for AD patients with particular copathologies.
Gene therapy is another attractive treatment approach for AD.ApoE ε4 is the strongest genetic risk factor for AD.Studies have shown that reducing ApoE ε4 levels decreased Aβ pathology in Aβ amyloidosis mouse model (Bien-Ly et al., 2012;Kim et al., 2011).Treatment with ASOs targeting ApoE ε4 prior to plaque deposition can strongly influence the onset of Aβ pathology in APP/PS1 mice (Huynh et al., 2017).In addition, treatment targeting ApoE ε4 significantly preserved synaptic density and reduced neuroinflammation in tauopathic model mice (Litvinchuk and Huynh, 2021).Recently, LX1001, an adeno-associated virus (AAV) based gene therapy targeting APOE, has entered phase I/II clinical trial (NCT03634007).According to the results released by Lexeo Therapeutics in 2022 at the 15th CTAD, expression of APOE2 in ApoE ε4 carriers altered levels of AD-associated biomarkers such as P-tau.Thus, reducing ApoE4 levels or modulating the expression of different APOE forms may offer personalized therapeutic options for AD patients carrying ApoE ε4 .
Stem cell transplantation is also a potential strategy for AD treatment.Transplanted neural stem cells (NSCs) are able to regenerate damaged cholinergic neurons.The newborn functional neurons can form new synaptic connections with the remaining neurons of the host, improving cognitive function in preclinical studies (Liu et al., 2013;Zhang et al., 2019).NSCs also have other functions, including activating microglia for immune regulation, secreting neurotrophic factors such as BDNF to increase synaptic density, and secreting growth factors and anti-inflammatory cytokines to protect neurons (Boese et al., 2020).NSCs have been transplanted into the brain of PD patients and have been shown to be safe (Madrazo et al., 2019).
In recent years, disease-modifying therapies for AD targeting neuroinflammation have gained increasing attention.In tauopathic mice, microglia play a role in recruiting T cells into the brain, which produced IFNγ that can promote the production of CD11c + microglia.These activated microglia stimulated inflammatory responses and enabled antigen presentation to CD4 + T cells through T cell receptor, at least in part augmenting tau pathology and neurodegeneration.Depletion of microglia or T cells attenuated neurodegeneration in tauopathic mice (Chen et al., 2023).In addition, Aβ peptide can be present to Th1 cells through microglia and subsequently mediating nitric oxide cytotoxicity of T cells in vitro (Monsonego et al., 2003).Future studies on the mechanisms of T cell-microglia cross-talk need to be deepened to develop the corresponding treating strategies.
In summary, the development of effective treatments for AD relies on unveiling the complicated pathogenic mechanisms.In the future, through early diagnosis, early intervention, multidrug combination and multimodal approaches may generate additive or synergistic effects in AD patients.

Fig. 1 .
Fig. 1.Epitopes of Aβ recognized by the mAbs tested in clinical trials.1: Conformational epitopes consisting of the N-terminal and central portions of Aβ. 2: Conformational undisclosed epitope, between amino acids 1 and 16. 3: The third amino acid is pyroglutamate-modified.

Table 2
Passive immunotherapies targeting Tau in development.

Table 3
Passive immunotherapies targeting inflammatory regulators in development.

Table 4
Different therapeutic strategies reached or completed clinical phase trials.
X.Guo et al.