Neurotherapeutics (2022) 19:209–227

https://doi.org/10.1007/s13311-022-01201-2

REVIEW

Disease‑Modifying Therapies for Alzheimer’s Disease: More Questions

than Answers
Todd E. Golde1 

Accepted: 9 February 2022 / Published online: 28 February 2022

© The Author(s) 2022

Abstract
Scientific advances over the last four decades have steadily infused the Alzheimer’s disease (AD) field with great optimism
that therapies targeting Aβ, amyloid, tau, and innate immune activation states in the brain would provide disease modifi-
cation. Unfortunately, this optimistic scenario has not yet played out. Though a recent approval of the anti-Aβ aggregate
binding antibody, Aduhelm (aducanumab), as a “disease-modifying therapy for AD” is viewed by some as a breakthrough,
many remain unconvinced by the data underlying this approval. Collectively, we have not succeeded in changing AD from
a largely untreatable, inevitable, and incurable disease to a treatable, preventable, and curable one. Here, I will review the
major foci of the AD “disease-modifying” therapeutic pipeline and some of the “open questions” that remain in terms of
these therapeutic approaches. I will conclude the review by discussing how we, as a field, might adjust our approach, learn-
ing from our past failures to ensure future success.

Keywords  Alzheimer’s disease · Therapeutics · Amyloid · Tau · Inflammation · Prevention · Disease modification

Introduction Recent reviews have catalogued the therapeutic AD pipe-

line [6–8]. Rather than rehash these scholarly and detailed
AD is the most common form of dementia, accounting manuscripts, I will focus this review on three major thera-
for ~ 70% of all dementias in those over 60. Worldwide, peutic categories, Aβ-, tau-, and immune-targeting thera-
it is predicted that the estimated 35–40 million individu- pies, that are being developed for AD and related disorders.
als currently affected by AD will grow to at least 100 mil- Within each section, I will begin with a high-level over-
lion individuals with AD in 2050 [1]. AD prevalence varies view of the scientific rationale supporting the therapeutic
depending on the population studied and criteria used to approach, follow this with a concise summary of pivotal
define the disease, but prevalence rises with increasing age human clinical trial data, if such data is available, then
[1, 2]. Thus, with increased life span comes an increased address a series of “open questions” that reflect critical
risk for developing AD. Despite the recent Food and Drug knowledge gaps regarding that form of therapy. The review
Administration (FDA) approval of Aduhelm (aducanumab), will thus focus on small molecule and genetic and biologic
hailed by some as the first disease-modifying therapy for therapies, which are designed to be disease modifying by
AD, there remains a great unmet medical need [3]. Indeed, targeting some aspect of the pathophysiology of AD. I will
even if one takes the most optimistic “cherry-picked” view discuss the need for reinvestment in development of symp-
of the Aduhelm data from the phase 3 trials, the average tomatic therapies at the end of this review, but will not criti-
clinical impact in terms of slowing cognitive decline is quite cally review the ongoing efforts to develop novel sympto-
modest [4, 5]. matic therapies.

* Todd E. Golde
tgolde@ufl.edu Aβ‑ and Amyloid‑Targeting Therapies
1
Departments of Neuroscience and Neurology, Center
for Translational Research in Neurodegenerative Disease, Overall Rationale  Genetic, human biomarker, pathologic,
Evelyn F. and William L. McKnight Brain Institute, Norman and experimental modeling studies strongly support the
Fixel Institute for Neurological Diseases, University hypothesis that Aβ aggregate accumulation in the brain is
of Florida, Gainesville, FL, USA

13
Vol.:(0123456789)
210 T. E. Golde

the triggering causal event in AD [9]. Though there remain of postmortem brains collected from long-term follow pro-
many gaps in our understanding of how the slow accumu- vides suggestive support that in some individuals, there were
lation of Aβ aggregates triggers the complex downstream regional reductions in amyloid loads [23, 25].
cellular pathophysiology that are characteristic of the symp-
tomatic phase of AD, targeting Aβ with the goal of altering What Was Learned from This Study?  The top-line lessons
amyloid deposition has been the mainstay of AD therapeutic are that (i) immunizing humans with a self-antigen and
development [10, 11]. novel adjuvant warrants a great deal of caution as there
is a reasonable potential for unexpected and unwanted
immune responses that are not observed in preclinic stud-
Immunotherapies Targeting Aβ in AD ies and [2] in elderly humans, it is not easy to generate a
robust humoral response to a self-antigen. Furthermore, the
Rationale  Disruptive studies, initially conducted by Shenk study highlighted the need for better biomarkers of AD to
and colleagues during the late 1990s and early 2000s, dem- [1] ensure trial participants truly had AD and [2] to assess
onstrated both active and passive immunotherapies targeting target engagement and efficacy in effective targeting of the
Aβ could reduce amyloid accumulation in preclinical mouse pathology. Such biomarkers were just beginning to emerge
models of Aβ amyloid deposition [10, 11]. These preclinical near the end of this trial and, therefore, are not included in
studies showing reduction of amyloid by Aβ immunothera- the trial design.
pies have now been reproduced in hundreds of independent
preclinical studies (reviewed in [12, 13]). The studies were Open Questions from the AN‑1792 Study
disruptive as they overturned dogma that the limited central
nervous system (CNS) exposure of an antibody therapeu- 1. What caused the meningoencephalitis? The severe
tic would render this approach ineffective. Indeed, multiple meningoencephalitis observed in some individuals in
studies show that ~ 0.1% of a peripheral antibody crosses the the phase 2 trial in many cases responded to steroid
blood–brain barrier, limiting central nervous system (CNS) treatment suggesting that it was likely a T-cell medi-
exposure [12, 14, 15]. This limited CNS exposure has many ated response, a finding consistent with post hoc post-
implications for anti-Aβ therapy that we now understand are mortem studies and analysis of mRNA transcripts in the
important for development and selection of the most effi- blood [21, 26]; however, as peripheral cellular immune
cacious anti-Aβ (at least with respect to CNS targeting of responses were not monitored during the trial, definitive
amyloid). Multiple possible mechanisms for antibody medi- data that the meningoencephalitis was attributable to a
ated reductions and clearance of Aβ have been proposed and T-cell response and more specially a T-cell response to
supported by various preclinical studies (reviewed in [12, Aβ is lacking. Nevertheless, the concept that this was
16, 17]). How human studies have informed these preclinic a T-cell response to Aβ, and in particular to a T-cell
mechanistic studies will be discussed below. epitope within the carboxyl-terminus of Aβ, guided
future vaccine development.
2. Did the use of QS-21 contribute to the unexpected side
Active Vaccination Targeting Aβ: AN1792 effects observed? The saponin QS-21 is one of the active
fractions of the bark of Chilean tree, Quillaja saponaria,
Clinical Studies  Initial studies reported on the efficacy of and is an acylated 3, 28-bisdesmodic triterpene glyco-
an active vaccination with fibrillar Aβ42 reducing amyloid side [27]. It is one of the most potent immunological
deposition in mice with no apparent ill-effects [18], rapidly adjuvants that has been used in humans and dose in most
led to the first human trial of an anti-Aβ vaccine, AN-1792. patient populations is limited by toxicities. Indeed, it has
This vaccine was fibrillar human Aβ42 in conjunction with only recently been approved for human use as one of a
a novel adjuvant QS-21. Clinical studies of AN-1792 were multi-component adjuvant in a few vaccines [28, 29].
halted during phase 2 due to a 6% rate of aseptic meningoen- Given its potent immune stimulating nature, the contri-
cephalitis with only an ~ 20% rate of “adequate” humoral bution of this adjuvant to the side effects is unclear, but
responses to Aβ [19, 20]. Over time, a number of studies should not be discounted.
have emerged from follow-up of those enrolled in the tri- 3. Was there really amyloid reduction and, if so, what
als [21–25]. These studies show possible hints of “amyloid response to the vaccine caused the reduction? PET
plaque reduction” in postmortem tissue and possible func- amyloid tracers were not available at the time of the
tional benefit in responders [21, 23, 25, 26]. T-cell infil- AN-1792 trial. Thus, post hoc postmortem studies
tration was described in the autopsy from the patient with claiming “clearance” of amyloid plaques rely on a num-
meningoencephalitis [26]. Brain volume reductions were ber of circumstantial findings to claim actual amyloid
also noted based on structural MRI. Non-systematic analysis reduction. The original case report on an individual who

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 211

died from the meningoencephalitis was intriguing in that several of these vaccines has been discontinued or is inac-
the “clearance” was very circumscribed within select tive, whereas three vaccines (AbVac40, phase 2; ACI-24,
regions of the brain [26]. These “patchy” effects are phase 2; and UBI-311, phase 3) remain in later phase clinical
reminiscent of T-cell-mediated CNS disease that occur trials [30–34]. Building off data gleaned from the AN1792
in disorders like multiple sclerosis. Additional studies trial, these vaccines have generally been developed using a
on larger cohorts of AN-1792 vaccinated individuals restricted B-cell epitope of Aβ and some alternative helper
also claim reductions in amount of amyloid deposited T-cell epitope. Many are claimed to produce robust anti-Aβ
in some areas of the brain, relative to what might be humoral responses, but as noted above, the lack of bench-
expected [23, 25]; however, this data is again circum- marking and standardization of such titers prevents a truly
stantial. Without definitive data showing amyloid status rigorous assessment of the adequacy of response.
prior to receiving the drug, it cannot be proven amyloid
was there and then subsequently reduced. Certainly, Open Questions for Second‑generation Vaccines
the data is provocative. Though a correlation between Targeting Aβ
anti-Aβ antibody titers and extent of the claimed amy-
loid reductions is shown, several individuals with little 1. Safety looks more promising, but is the humoral response
amyloid at autopsy showed no, or almost no, anti-Aβ against Aβ sufficient to impact disease? Available data
titer. This data could be explained in a number of ways. suggest that most, if not all, of these second-generation
At face value, such data would suggest that the effects vaccines avoid the meningoencephalitis induced by
on amyloid in these trials were not due to a humoral AN1792. Some also seem to generate more consistent
response to Aβ, but either a non-specific T-cell or an Aβ/ humoral immune responses to Aβ; however, whether the
amyloid targeting T-cell response. Alternately, but less titers and the quality of the antibody responses to Aβ are
likely, would be that titering protocol missed some type sufficient to engage Aβ in the brain and, more particu-
of conformational antibody response to Aβ that medi- larly Aβ aggregates, is simply not known [30–34].
ated the efficacy. Finally, it is also quite possible these 2. What can we infer from the lack of data on amyloid PET
individuals really did not have AD at the time the trial ligand reduction? As opposed to data for select mono-
was initiated, so that there was little Aβ to be cleared. clonal antibodies, there is no published data available
4. What is a sufficient titer for an active vaccine and to that demonstrates reduction, or slowing of the increase,
what form of Aβ? Titering of antibodies simply based on in amyloid PET ligand signal in the brain following
dilution is not a particularly precise method to determine treatment with any active vaccine targeting Aβ. Such
the response to a given vaccine. Such titering allows data, or data showing some major impact on a biomarker
for one to rank order responses among individuals, but of neurodegeneration, would certainly reignite enthusi-
does not allow comparison across trials. An anti-Aβ titer asm for an active vaccination approach targeting Aβ.
converted to actual molar concertation of the antibody 3. At least with respect to amyloid reduction, could the
against Aβ would be much more useful, but even then, T-cell to Aβ response be important? Based on the
standardization of the Aβ peptides and assays used to AN1792 trial, a T-cell response to Aβ has been hypoth-
assess the titer would be needed to benchmark and stand- esized to be something to be avoided [26]; however, this
ardize across trials. As discussed below, both preclinical hypothesis was not tested in the AN1792 trial, nor has it
and clinical studies with monoclonal anti-Aβ antibod- ever been fully evaluated in preclinic models. Genera-
ies demonstrate that epitope, selectivity for aggregates, tion of chimeric antigen receptor (CAR) T-cells target-
affinity, and, to some degree, effector functions appear to ing Aβ and testing of these in animal models of Aβ dep-
be important aspects of a given monoclonal antibody’s osition might provide insight into the role that T-cells
ability to impact amyloid deposition [17]. Whether any might play in active vaccination paradigms [35]. Indeed,
vaccine targeting Aβ can generate a humoral response the assertion that the meningoencephalitis observed in
with a preponderance of antibodies of appropriate affin- the AN1792 trial was an autoreactive T-cell response to
ity and specificity remains unclear. Aβ could be formally tested in such studies. Alternately,
such an approach could show unexpected efficacy, as it
Second‑generation Active Vaccinations Targeting Aβ is well established that T-cells survey the brain and play
a role in regulating immune activation states in the CNS.
Clinical Studies  A number of additional active vaccine can-
didates targeting Aβ have been developed and moved into Passive Immunotherapy Targeting Aβ
clinical trials. A continuously updated summary of these and
other classes of AD therapeutics can be found on AlzForum Rationale  Numerous preclinical studies have shown the
(https://​www.​alzfo​r um.​org/​thera​peuti​cs). Development of ability of peripherally administered antibodies targeting

13
212 T. E. Golde

Aβ to attenuate amyloid deposition in the brains of amy- Overall, the studies conducted to date are consistent with
loid depositing mice (reviewed in [12, 16, 17]). Behavioral a model whereby an anti-Aβ antibody must enter the brain
improvements have also been noted in these mice following and engage deposited Aβ [17]. Subsequently, the presence
select antibody administration. It is important to note that of the antibody with an activating Fc domain is thought to
these preclinical studies reproducibly show clearance of dif- elicit a microglial response that then reduces deposited Aβ
fuse amyloid deposit, but typically show minimal impact to a degree that results in a reduction in amyloid PET ligand
on preexisting amyloid cores. “Clearance” is often loosely signals. Binding to soluble monomer with high affinity is a
used to describe the impact of immunotherapies and the liability, in sense blunting the amount of antibody that can
field should be cautious about claims of amyloid clearance bind to deposits in the brain.
as opposed to slowing or attenuating additional deposition.
Some side effects, including microhemorrhages, have been Open Questions for Passive Aβ Immunotherapies
noted in select preclinical studies as well.
A large number of passive immunotherapies targeting 1. Is the clinical efficacy in symptomatic disease suffi-
Aβ have been advanced to human studies. As mentioned cient to warrant use? Though Aduhelm has now been
previously, one of these, aducanumab (Aduhelm), is now approved by the FDA, it is clear that many, if not most,
approved for use, though as noted, the approval and support- in the field remain unconvinced of the clinical benefits
ing data remain controversial [4, 5]. of this antibody therapy [4, 5]. Indeed, both the Euro-
pean Medical Agency’s decision to decline marketing
Clinical Studies: What Has Been Learned?  Given the large approval for Aduhelm and the recent Centers for Medi-
number of antibodies tested to date, a picture of the prop- care and Medicaid Services (CMS) draft decision that
erties that are associated with potential efficacy is emerg- may limit Aduhelm coverage to clinical trials (https://​
ing. These data reveal that antibodies that selectively bind www.​a lzfo​r um. ​o rg/​n ews/​r esea ​r ch-​n ews/​c ms-​p lans-​
to deposited Aβ, either because they bind aggregates with limit-​aduhe​lm-​cover​age-​clini​cal-​trials) reflect the con-
higher selectivity than monomer (Aduhelm [36], lecanemab cerns over the FDA approval.
[37], and gantenerumab [38, 39]) or bind modified forms of   Despite clear evidence for reduction in amyloid PET
Aβ such as N3-pGAβ (donanemab[40–42]) highly enriched ligand signal in the brain, effects on slowing cognitive
in deposited Aβ, can effectively lower amyloid PET ligand decline and improvement of function (e.g., activities of
signals to control, or near control, levels in many individuals daily living) remain uncertain. Even the most optimistic
with early clinical stages of AD. There is data that has hinted take on the Aduhelm data would suggest that an ~ 20%
at slowing of cognitive decline in almost all the early phase slowing of cognitive decline might result from this treat-
trials of these antibodies and, as noted above, highly con- ment. Given that high-dose Aduhelm treatment results
troversial data of possible cognitive and functional improve- in a large increase in ARIA (41%) compared to placebo
ment in the phase 3 trials for Aduhelm. Another common (10%), with ~ 1% of those treated with Aduhelm showing
feature of these antibodies is that they contain activating serious side effects, the high cost of the drug, and the
Fc domains and to some degree induce radiographic fea- intensive treatment regiment, it is clear that a larger clin-
tures known as Alzheimer’s related imaging abnormalities ical signal is needed to convince many stakeholders that
with edema (ARIA-E) or hemorrhage (ARIA-H) [43]. Both this drug should become standard of care [4, 5, 50]. The
types of ARIA can be associated with some cognitive side other monoclonal antibodies (lecanemab, gantenerumab,
effects that are typically mild and that also typically resolve and donanemab) with the ability to lower amyloid PET
upon discontinuation or lowering of dose. Because of this ligand signal are in late phase clinical testing. We can
association, it is thought that ARIA is, in fact, a marker of only hope that data from these trials will reveal a more
target engagement. consistent picture with respect to clinical efficacy.
A number of antibodies that have advanced to phase 3 2. Are these antibodies really clearing amyloid? It is clear
studies in symptomatic AD have not shown evidence for that in many individuals receiving select Aβ aggregate
efficacy either with respect to marked lowering of the PET targeting monoclonal antibodies that treatment overtime
ligand signal, or any significant clinical impact (ponezumab reduces amyloid PET ligand signal in the brain. Indeed,
[44], solanezumab [39, 45–47], bapineuzumab [48], cren- the FDA cited this biomarker impact as a major reason
ezumab [49]). Commonalties among these antibodies are for approval; however, we still lack critical confirmation
that they do not show much differential binding between Aβ that the reduction in amyloid PET ligand signal equates
monomers and Aβ aggregates, or preferentially bind only Aβ to a reduction in deposited Aβ. Unequivocal demon-
monomers, and, in some cases, were designed to minimize stration of amyloid clearance in a postmortem brain is
the immune activation by limiting Fc engagement of activat- needed to establish this relationship and we simply do
ing immune receptors. not have such data. Though amyloid PET signal corre-

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 213

lates very well with amyloid deposition [51], it is clear will be possible to either develop anti-Aβ antibodies that
that it is not a truly selective biomarker of cored plaque are capable of reducing the amyloid PET ligand signal
pathology, as areas of the brain with only diffuse amy- without inducing ARIA, or if dosing paradigms or tim-
loid can show marked amyloid PET ligand binding [52, ing of treatment could be altered to reduce it. Indeed, it
53]. Furthermore, preclinical studies show ability of has been reported that risk for ARIA diminishes as treat-
many antibodies to alter diffuse Aβ deposited, but cored ment progresses. Again, lack of study of postmortem
plaque pathology is highly resistant to clearance [17]. tissue from individuals who developed ARIA limits both
It would seem that if the field and regulatory bodies our mechanistic understanding of it and future modifica-
are going to use amyloid ligand binding as a biomarker, tions to immunotherapies that might limit its impact.
then we should go the distance and really establish the 4. Have these studies truly tested the amyloid hypothesis?
relationship between amyloid PET ligand reduction and The simple answer to this question is no. The amyloid
impact on amyloid depositon in the brain through post- hypothesis posits that Aβ accumulation in the brain
mortem pathological studies. triggers a complex neurodegenertive cascade that over
3. Is it possible to reduce ARIA and still have meaningful decades leads to a state akin to brain organ failure [9].
impact on the amyloid PET-ligand signal? Monoclonal Simply put, the field has spent a long-time testing a
antibody therapies that result in reductions of amyloid pretty low probability event — that targeting of amy-
PET ligand signal all appear to induce some degree of loid in the symptomatic phase of disease will have a
ARIA. Incidence of ARIA is increased in those with evi- major impact on disease [10]. Though there may yet be
dence for preexisting vascular amyloid based on presence clinical benefit that emerges from ongoing symptomatic
of preexisting microhemorrhage and also in individuals trials of anti-Aβ immunotherapies, the data that we have
with E4 alleles, who typically have more cerebrovascular in hand would suggest a limited impact of these agents
amyloid deposition [36–43, 50]. Higher doses of anti- in the symptomatic stage of disease. Indeed, I and oth-
body also result in a higher incidence of ARIA. These ers have previously discussed that the only true test of
data suggest that effector functions of these antibodies the amyloid hypothesis is to prevent amyloid deposition
and binding to cerebrovascular amyloid may underlie in humans and see if that prevents the development of
this sometimes dose-limiting impact of Aβ monoclonal AD [10, 55]. Unfortunately, the antibodies that appear
antibody therapy. Though most cases of ARIA are mild capable of robust engagement of deposited Aβ have not
and spontaneously resolve, serious adverse outcomes are yet been rigorously studied in either primary or secondary
observed in some who develop ARIA. prevention studies.
  There are many theories about the biological basis of
ARIA, but we have little direct data that provides mecha- Aβ Production and Aggregation Inhibitors
nistic insight into this effect of anti-Aβ antibody admin-
istration. Spontaneous ARIA-like events are observed in Rationale  There is no evidence that monomeric Aβ is patho-
AD and have been linked to the presence of cerebro- genic and there is little convincing evidence that it has a
vascular amyloid angiopathy (CAA) and the presence of normal necessary physiologic function [56]; however, it is
auto-antibodies against Aβ [54]. Given the current data, clear that upon aggregation, accumulation, and deposition,
it is certainly plausible to hypothesize that engagement Aβ aggregates can have a plethora of pathophysiologic func-
of cerebrovascular amyloid by an antibody with immune tions [57, 58]. Given that Aβ aggregation is concentration-
activating effector functions results in a local immune dependent phenomenon, lowering levels of Aβ can reduce,
response that results in radiographic changes, and edema, slow, or even completely block aggregation. Aβ is nor-
that are consistent with local inflammatory responses. mally produced through the sequential actions of the β-and
This response could remove amyloid from the vessels γ-secretases on the amyloid β protein precursor (APP) and
and subsequently increase cortical hemorrhages. It is also inhibitors that target these proteases have been developed
possible that engagement of deposited Aβ by an antibody and shown to block production of all species of Aβ in the
in the parenchyma can also contribute to ARIA. brain [56]. Furthermore, it has been shown that longer Aβ
  Though most cases of ARIA are not severe, the rela- species (primarily Aβ42, but in some cases Aβ43), which
tively common induction of ARIA certainly complicates are produced at lower levels than the predominant Aβ40
management of patients undergoing treatment with anti- species, are required for in vivo aggregation of Aβ [59–62].
bodies that reduce the amyloid PET ligand signal. Fur- Thus, selective targeting of these species using modulators
thermore, in some individuals, ARIA represents a severe of γ-secretase activity (GSMS) has also emerged as a pos-
side effect and negatively impacts the risk benefit equa- sible way to target Aβ production and subsequently alter
tion. This raises an important question about whether it deposition [63, 64].

13
214 T. E. Golde

Given that Aβ aggregates, and not monomer, are patho- prevention [10, 55, 57]. Given both the triggering role of Aβ
genic, many efforts have been made to develop inhibitors aggregation and accumulation in disease and the long stand-
of Aβ aggregation [65–71]. Though many of these have ing and widespread damage to the brain at the time most
shown ability to block Aβ aggregation in vitro, data in pre- individuals are diagnosed with any form of cognitive impair-
clinical animal models of amyloid deposition has always ment, inhibiting production or Aβ aggregation almost cer-
been much less impressive. Nevertheless, several anti-Aβ tainly requires a prophylactic approach to be initiated a dec-
aggregation agents remain in clinical development includ- ade or more before onset of symptoms. This means the drug
ing ALZ-801 [72] and PRI-002 [73]. must be very safe and must also engage target sufficiently
to warrant the long testing necessary to evaluate potential
Clinical Studies  The first Aβ production inhibitors to be for disease modification. Furthermore, from a public health
tested in the clinical studies were γ-secretase inhibitors perspective, the drug needs to be affordable.
(GSIs) [56, 63]. Long-term GSI treatment designed to pro-
duce moderate levels of inhibition of γ-secretase has been Open Questions for Aβ Production and Aggregation
shown to be associated with unacceptable side effects and Inhibitors
lack of clinical efficacy in AD [74, 75]. Side effects appear
to be mechanisms based and, in many cases, can be linked 1. Given that these agents are only likely to have impact
to inhibition of NOTCH1 cleavage and signaling [56, 63]. on disease in primary or secondary prevention is there
Though GSIs have been repurposed for various cancers and any path forward? β-secretase inhibitors that lowered
remain under clinical investigation, development for use in CNS Aβ levels failed to impact clinical progression in
AD has been discontinued. β-secretase inhibitors have also trials in symptomatic AD. These studies, again, repre-
been extensively tested in the clinic and were once the major sented important negative data for the field, firmly sup-
hope for a small-molecule approach to treatment or preven- porting assertions that targeting Aβ production would
tion of AD [76–83]; however, numerous studies in symp- require prophylactic treatment long before appearance
tomatic AD or MCI showed no evidence for disease modi- of symptoms; however, the data showing that several
fication despite robust lowering of CSF Aβ, and numerous β-secretase inhibitors worsened cognition and caused
adverse effects that were not anticipated based on preclinical brain volume loss in cognitively normal at-risk individu-
studies (reviewed in [83]). Indeed, phase 2 and 3 trials of als has currently halted further development of these
β-secretase inhibitors showed no benefit, or were discon- agents. Though a recent perspective suggests that there
tinued, due to futility, adverse events, or some combination may be paths to revive clinical development β-secretase
of these. Unexpected adverse effects included early, mild inhibitors, the authors acknowledge that more data is
cognitive impairment that appears to be non-progressive and needed and that these efforts will be challenging [83].
reversible upon discontinuation. These adverse effects are   Ongoing efforts to conduct prevention studies in AD
almost certainly “on-target” and likely reflect the adverse are providing invaluable and ever evolving paradigms
effects of chronic and high-level inhibition of β-secretase. for how to conduct both primary and secondary preven-
Though these data have renewed arguments about possible tion studies in Alzheimer’s disease, but the paucity of
physiologic roles for Aβ [84], it is more likely that toxicity agents that meet the safe enough, sufficient targeting
of both GSIs and β-secretase inhibitors is attributable to the engagement, and rationale for disease medication cri-
net impact on signaling events mediated by the myriad of teria is worrisome [89–93]. Indeed, what seemed like
substrates each of the proteases cleave. Several small mol- a robust therapeutic pipeline for prevention studies has
ecules referred to as γ-secretase modulators (GSMs) that rapidly dwindled over the last few years.
selectively reduce Aβ42/43 levels and increase the levels of 2. Are there any approaches to targeting Aβ safely that
shorter Aβ peptides have also been developed and advanced make sense? From a theoretical standpoint, both GSMs
into clinical trials [63, 85–88]. These agents showed limited and small molecule aggregation inhibitors remain
target engagement in humans and had a narrow therapeutic attractive approaches to “safe enough” targeting of Aβ.
index which was thought to be due to off-target effects. Indeed, for GSMs, there is extensive data that support
both the safety and selectivity of this approach and the
What Have We Learned from These Studies?  The experi- potential protective action of both lowering Aβ42 levels
ence with all of these small-molecule agents targeting Aβ and increasing the level of shorter Aβ peptides; however,
highlights the difficulties of an Aβ-centric approach to AD. it has been challenging to develop GSMs that are brain
Indeed, my colleagues and I have previously discussed in penetrant and that lack what appears to be off-target tox-
multiple reviews and perspectives the dilemma of target- icity. Nevertheless, a few preclinical GSMs programs
ing Aβ in the symptomatic phase of disease as opposed to remain active and appear to have optimized many of

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 215

the properties of the drugs [94, 95]. Given the rationale Therapeutic Approaches Targeting Tau
for selective targeting of Aβ42 along with the challenges
of targeting β-secretase, renewed efforts to develop Overall Rationale  Accumulation of tau as neurofibrillary
novel, safe GSMs and test these agents in prophylactic tangles is a hallmark AD pathology [102]. Numerous stud-
paradigms would make sense. ies link abnormal tau metabolism to neuronal dysfunction
  Aβ aggregation inhibitors have always made sense and death. Mutations in the gene (MAPT) encoding the tau
conceptually, but have been plagued by issues of potency protein that cause frontotemporal dementia with parkinson-
and evidence for in vivo efficacy even in the preclinical ism linked to chromosome 17 (FTDP-17 MAPT) demon-
setting. Given the advances in the structural details of strate that alterations in tau are sufficient to cause neurode-
Aβ assemblies and mechanisms of nucleation and fibril generation [103]. Genetic and animal modeling data from
elongation, it might be worth revisiting past efforts to try the study of FTDP-17 MAPT only directly support tau as a
to drug Aβ aggregation [96, 97]. Obviously, it is chal- therapeutic target in FTDP-17 MAPT; however, given the
lenging to revive efforts that were largely unsuccessful prominence of tau pathology in the AD brain, its correla-
in the past, but given new data and new mechanistic tion with clinical symptoms, and experimental evidence that
insights, it may be possible to overcome limitations of Aβ can drive tau pathology in vivo, tau remains an attrac-
previous approaches. Given prior efforts, general issues tive theoretical target in AD [104]. It has been somewhat
around specificity of aggregation inhibitors, and stoi- more challenging to identify tau therapeutics as opposed
chiometry of small molecules typically needed to alter to Aβ-targeting therapies [105]. Pathologically associated
aggregation of proteins, new initiatives should proceed tau is extensively post-translationally modified (PTM) with
cautiously and include screens for potency and specific- modifications including, but not limited to, extensive phos-
ity early in the discovery process. phorylation, acetylation, and O-glycosylation [106, 107].
  Other efforts such as antisense oligonucleotide Tau PTMs are altered as tau accumulates in the brain and
(ASO)-based targeting of APP or other gene therapies to there is evidence that modifying the activity of proteins that
target Aβ production or APP levels certainly are plausi- control the PTM can alter the normal cellular function of tau
ble scientific approaches, but face implementation chal- and its subcellular distribution, but also alter its aggregation
lenges due to issues around safety and population-scale [108]. Though select kinase inhibition and tau aggregation
delivery [98]. If targeting Aβ showed more evidence for inhibitors have been the focus of previous efforts, clinical
efficacy in symptomatic individuals, then these efforts data supporting these approaches remains underwhelming.
might make more sense if they could be shown to over- Nevertheless, new kinase inhibitor targets remain under
come safety issues of other approaches; however, given development. Most recent efforts have focused on immu-
available data, it seems that major technological chal- notherapies targeting tau as well efforts to alter tau though
lenges would need to be overcome before testing these alteration in O-glycosylation. In the following paragraphs,
approaches as prophylactic therapies. I will discuss the status of these and open questions arising
3. Can new model systems help? Current cellular and ani- form completed and ongoing studies.
mal models of Aβ production, clearance, and deposition
are more than adequate models when used for pharma- Immunotherapies Targeting Tau in AD
codynamic studies that assess target engagement and
effects on Aβ [99]. The informativeness of behavioral Rationale  The notion that antibody-based therapies could
readouts in rodent models of amyloid deposition is target tau pathology and tau-related neurodegeneration was
much less certain, and it is clear that these models do quite controversial following publication of initial preclini-
not reflect the neurodegenerative symptomatic phase of cal data [109, 110]. Three possible mechanisms have been
AD. Though previous studies have indicated that cross- proposed to account for the apparent efficacy in preclinical
talk between Aβ and tau pathologies can be modeled in settings. The first of these proposed mechanisms is that tau-
transgenic mice [100, 101], there is still a huge gap in targeting antibodies can bind to extracellular tau “seeds”
our understanding of what factors mediate that crosstalk. that help to spread tau pathology between cells in the brain.
Model systems that reproducibly and faithfully recapitu- Preclinic studies demonstrate that prion-like seeding of tau
late the crosstalk between Aβ and tau as well as show pathology via this extracellular seed occurs in select model
robust neurodegenerative changes would have a huge systems and can be mimicked by exogenous application of
impact on the field and likely lead to the identification of aggregated tau seeds [111, 112]. Tau antibodies can inter-
novel therapeutic strategies that could impact the course cept extracellular tau seeds and block spread of pathology by
of disease in more advanced preclinical or even sympto- binding and neutralizing the tau seed, diverting the seed to
matic stages. be phagocytosed by microglial, or enhancing export from the

13
216 T. E. Golde

brain. The second proposed mechanism is that tau-binding but for reasons described in the answer to the next ques-
antibodies are internalized into neurons through Fc receptors tion, it is not clear whether the binding properties of any
[110, 113–115]. This mechanism remains controversial as antibody tested in humans are sufficient to have tested
the evidence for neuronal cell surface Fc receptor expression the hypothesis.
is controversial. Furthermore, it is difficult to envisage how 2. Are any anti-tau antibodies tested in humans target-
an antibody binding to a cell surface Fc receptor might be ing the right domain of tau and of high enough affinity
internalized and then capable of targeting tau in the cyto- to engage the tau “seed,” even if it is present? Stud-
plasm. A third mechanism involves possible uptake of the ies of extracellular tau in human spinal fluid show that
tau seed and bound anti-tau antibody complex by neurons or most tau present in the CSF lacks the carboxyl terminal
other cells and then binding of this complex to the intracel- microtubule domains that appear to be the core amyloi-
lular Fc binding TRIM21 protein [116, 117]. TRIM21 binds dogenic region of tau [107]. Thus, most tau in the CSF
the Fc domains of IgGs with high affinity and contains an and interstitial fluid is thought to be incapable of seeding
E3-ligase domain. TRIM21 was originally shown to mediate tau aggregation. In these same studies, a much lower
intracellular clearance of antibody-coated viruses by catalyz- level of tau containing the microtubule binding domains
ing ubiquitination of antibody bound virus and degradation was detected. Indeed, these studies indicate that maxi-
via the ubiquitin proteasome system and similar data exists mal levels of extracellular tau that could theoretically
that TRIM21 can play a role in degrading internalized tau serve as a seed were approximately 1–2 pM. Further
aggregates [116, 117]. Notably, several studies show that studies of human AD CSF show that species capable
single domain variable fragments lacking Fc domain can of seeding may be present, but are present at extremely
modulate tau pathology in vivo; such data would suggest low levels [140]. From target engagement and pharma-
that Fc-mediated mechanisms are not responsible for efficy cologic perspectives, these data suggest that in order for
of tau immunotherapies [118–120]. Though most thera- a tau seed to be engaged by an antibody within the brain,
peutic approaches have focused on using monoclonal tau the tau-targeting needs to have extremely high affinity,
antibodies, active immunization strategies have also been or avidity, and be capable of targeting the carboxyl ter-
pursued. minal domains of tau that are present in the putative
seeding competent species. Indeed, antibodies that bind
Clinical Studies  Multiple monoclonal tau antibodies have the mid-domain or amino terminal regions of tau would
been tested in human studies [121–134]. In many cases, likely bind preferentially to the much more abundant
the antibodies were tested in mild AD and in separate trials amino-terminal species present in CSF and brain inter-
in the primary tauopathy progressive supranuclear palsy stitial fluid. Given that the maximal brain levels of any
(PSP). Many of these trials have been discontinued as there peripherally delivered antibody are predicted to be in
has been little evidence for efficacy. Several active vaccine the 1–5-nM range, it is hard to envision how low pM,
strategies remain under clinical development, though lit- or even sub pM concentrations, of a tau seed could be
tle data besides induction of antibody titer and safety have effectively engaged by an antibody unless the antibody
emerged [135–138]. In many of the passive immunotherapy was selective for the seed and had low pM or sub-pM
studies, there was evidence for entry of the antibody into avidity. Based on available data, it is not clear if any
the CNS and evidence that the antibody was able to bind antibody tested to date has such properties.
the amino terminal fragment of tau that is normally present   Pathologically associated tau is extensively post-
in CSF and brain extracellular fluid. Clinical evidence for translationally modified (PTM); thus, there are a pleth-
disease modification with these antibodies has been nega- ora of epitopes and disease-associated PTM epitopes that
tive to date. theoretically could be targeted using an immunother-
apy [106, 107, 141]. It remains possible that a specific
Open Questions for Tau Immunotherapies epitope uniquely defines naturally existing extracellu-
lar tau seeds. Antibodies with exquisite specificity and
1. Does Prion-like spread of tau occur in the human brain? affinity for this theoretical seed could be the magic bul-
Much of the enthusiasm to pursue tau immunotherapy lets needed to target tauopathy in AD, but our current
is based on the preclinical data that tau can spread in a knowledge about whether such tau seeds really do exist
prion-like fashion. Prion-like spread of tau in humans and whether they contain such unique epitopes is insuf-
remains a hypothetical event [139]. If it does not occur, ficient to frame efforts to design better tau antibodies.
then it would seem that efforts to target extracellular tau Nevertheless, as more information emerges, the field
have a very low probability of success. Unfortunately, should think critically about whether that data informs
showing that an antibody therapy blocked tau spread in on the desired properties of a second-generation tau-
humans is one of the few direct tests of this hypothesis, targeting immunotherapy.

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 217

3. Could an active tau immunotherapy produce a sufficient most cells bearing them would die quickly. This acute
immune response to result in efficacy? Given the likely death is clearly not the case as most tangle bearing cells
requirement of the need to induce a very high affinity are clearly intact. Tau PET ligands would also not be
tau “seed” targeting humoral immune response, it would very useful biomarkers if tau inclusion pathology did not
seem very challenging to reproducibly generate such a increase over time in the human AD brain. Our limited
response using an active vaccine approach. Again, as understanding of both tangle formation in the setting of
vaccines could elicit T-cell-mediated responses to a tau AD and the role that tangle formation or tau dysfunction
epitope, it remains plausible that such responses could plays in neurodegeneration remains a major knowledge
play a beneficial role, but as with Aβ vaccines, the like- gap and clearly limits our ability to design tau-targeting
lihood is that such responses would probably do more therapies that are more likely to show clinical efficacy.
harm than good. Recent studies also highlight that fibrillar tau inclusions
4. Is it possible to target intracellular tau using an immu- form rapidly and turnover with an appreciable half-life
notherapeutic approach? Tau pathology is primarily [149, 150]. Such data suggests that cells adapt at least
intracellular. Preclinical studies have shown that intra- in the short term to tau inclusion formation. Though the
cellular tau targeting recombinant antibody fragments concept of reducing tau pathology is attractive, it is pos-
delivered with recombinant adenoassociated viral vec- sible that tau inclusion formation is in fact an adaptive,
tors (intrabodies) can have modest impact on tau pathol- protective response, and that efforts to reduce it may
ogy and tau-induced neurodegenertive phenotypes have unintended negative impacts.
in preclinical models [120, 142]; however, given the
widespread distribution of tau pathology in the brain, Small Molecule and ASO Approaches to Targeting
advances in transgene delivery using viral vectors or Tau
other methodologies are needed in order to obtain suf-
ficient “coverage” to actually evaluate potential for dis- Rationale  Most small molecules targeting tau are designed
ease modification. Future advances in transgene delivery to directly or indirectly alter its aggregation. The notable
and antibody engineering could eventually overcome exception being ongoing development of select microtubule
these current limitations, but it is not easy to see how stabilizing agents that designed to compensate for loss of
the current obstacles to targeting tau with an intrabody- microtubule binding by hyperphosphorylated or aggregated
like approach can be overcome in the near term. tau. Modulators of tau PTMs, tau-chaperones, and tau aggre-
5. Are preclinical models of tau-pathology reliable and gation inhibitors have all advanced to clinical trials, but most
informative enough to guide human therapeutic devel- programs have been discontinued. In the paragraph below, I
opment? Preclinical models of tau pathology are really will briefly highlight the rationale for these approaches with
modeling FTDP-17 tauopathies and not AD tauopathy. a focus on approaches that remain in clinical development.
Moreover, many widely used transgenic lines show vari- As noted above, tau has extensive PTMs including phos-
able phenotypes [143] and more recently tau-induced phorylation, acetylation, and O-linked glycosylation [106,
neurodegeneration in the commonly used Tg4510 line 107, 141]. These various PTMs have been implicated as
has been shown to be influenced by transgene inser- possible sites for therapy as altering the PTMs appears to
tion events [144]. Seeded models including seeding of regulate tau aggregation and toxicity in preclinical models
endogenous mouse tau have been developed and could [151]. One of the challenges with targeting tau PTMs with
be useful preclinic models, but, again, are quite artifi- small molecules is that phosphorylation, acetylation, and
cial with respect to how they are generated [145]. More glycosylation are all reversible dynamic modifications and
generally, the interpretation of efficacy in these models multiple phosphatases, kinases, acetylases, deacetylases,
is often based solely on reduction of pathological tau glycosylases, and glycoside hydrolases regulate the extent
and not on a neurodegenertive phenotype. Furthermore, of tau PTM [106, 107, 141]. The action of these enzymes
robust reductions of tau pathology have not been associ- is not specific to tau, but regulate PTMs on many proteins
ated with major impact on neurodegenertive phenotypes and through this alter many cellular processes; thus, small-
in select tau mouse models. molecules targeting the enzymes that regulate tau PTM will
6. Do we understand tau pathobiology well enough to tar- impact numerous proteins and cellular functions. Previous
get tau successfully? More generally, the issue of how efforts to target tau by inhibition of GSK3β have been largely
tau pathology is associated with neurodegeneration discontinued [152] and most current efforts have focused on
remains poorly understood [108, 146, 147]. Provocative inhibition of O-GlcNAcase, the glycoside hydrolase enzyme
studies demonstrate that tangle bearing neurons remain that removes O-linked N-acetylglucosamine (N-GlcNAc)
functional at least for some period of time [148]. If neu- from proteins. N-GlcNAcylation of the microtubule-
rofibrillary tangles themselves were acute toxins, then associated protein tau reduces its propensity to form toxic

13
218 T. E. Golde

aggregates and is thought to compete for phosphorylation eration, we should probably temper our expectations
at those residues [153]. Indeed, increasing tau glycosyla- regarding likelihood of impact in clinical AD.
tion is proposed to stabilize tau in a non-toxic form and 2. Are studies of tau-targeting agents in primary tauopa-
several O-GlcNAcase inhibitors including Thiamet G have thies such as FTDP-17 or PSP reasonable surrogates
shown beneficial effects in preclinical models of tauopathy for studies in AD? This clinical development strategy
[154–162]. has been used without success for a number of tau thera-
Many chaperone proteins including, but not limited to, peutics; however, it is reasonable to ask whether PSP or
HSP90, CHIP, and FKBP51 have been implicated as tau FTDP-17 are truly good phenocopies of the tauopathy
chaperones [163–165]. These chaperones can, in experi- in AD. Given emerging data that different conformers
mental systems, regulate tau aggregation and accumula- of tau may predominate in different tauopathies [169,
tion; there are no active clinical studies using drugs that 170], we should be cognizant that the different tauopa-
target these chaperones in AD, though preclinical efforts thies may be distinct disorders and success or failure of
to try to harness chaperones that regulate tau proteostasis a therapeutic in one disease may not be predictive of its
remains active. At least three clinical programs evaluating success or failure in another.
tau aggregation inhibitors remain active. Two of these trials 3. Is symptomatic disease too late? Given that tau pathol-
are assessing methylene blue in different formulations and ogy correlates more closely with signs of neurodegen-
though methylene blue is touted as an aggregation inhibi- eration and clinical symptomatology, it is postulated
tor, it may have more complex effects on chaperone systems that successful targeting of tau may be more effective
[166, 167]. Another, Anle138b, is reported to be a more in symptomatic AD; however, most preclinical data on
general aggregation inhibitor and is also being evaluated in tau therapies shows modest signs of efficacy when the
the setting of Parkinson’s disease as an α-synuclein aggre- therapy is initiated prior to pathology or when minimal
gation inhibitor. ASOs targeting tau are designed to simply pathology is present. Again, lack of robust effects on
reduce tau levels and slow tau aggregation. Preclinical data late-stage preclinical models of tauopathy would suggest
supports tau ASOs as possible way to reduce tau aggregation that we should temper expectations of major benefits
and slow tau-induced neurodegeneration [168]. in symptomatic stages of disease [120]. Though amy-
loid pathology has largely plateaued when AD symp-
Clinical Studies  Most small molecule approaches targeting toms appear, tau pathology continues to progress [52,
tau are either in early stage clinical trials, have been dis- 171–173]. These relationships mean that it may be pos-
continued, or remain active despite underwhelming clinical sible to target tau in the symptomatic phase and observe
data. Three O-GlcNAcase inhibitors are in phase I or II trials both a biomarker and clinical effect; however, given
with insufficient data at this point to discuss potential effi- relatively modest impacts of most tau therapies in pre-
cacy. Methylene Blue remains in trials in a new formulation clinical studies, it is likely agents with a higher degree
with reports of some impact on cognition. Though because of target engagement and disease-modifying potential
of challenges with blinding, trial design, and methodology, will be needed in order to impact the course of disease
many in the field are skeptical of the clinical benefit. At least in the neurodegenerative and symptomatic phase.
one microtubule stabilizer remains in clinical development 4. Will tau PET ligands or other biomarkers serve as
for PSP, but other stabilizer programs have been discontin- proxies for clinical efficacy? Demonstration of highly
ued. BIIB080, a tau targeting ASO developed by Biogen and significant clinical benefit in phase 2 studies in AD is
Ionis, was shown to be well tolerated in phase 1, but data on very challenging. Thus, having biomarkers that serve
efficacy is not avaible. as potential proxies for clinical efficacy and provide
evidence for target modification is invaluable. Tau PET
Open Questions for Tau Targeting Small Molecules ligand signals appear to track well with clinical progres-
and ASOs sion, and it is certainly hoped that they can serve as an
early predictive marker of efficacy [174, 175]. In addi-
1. Is tau proteostasis so complex or so poorly understood tion, given lessons learned from trials targeting Aβ, it
that current efforts are likely to be futile? Like many would seem we should more systematically enroll trial
intracellular neurodegenerative proteinopathies, there participants in autopsy programs so that we can cor-
are many aspects of tauopathy that are poorly under- relate effects on pathology with changes observed in
stood. Despite robust efforts to alter tau aggregation and imaging studies. Of note here, is that measures of tau or
accumulation, even the best preclinical studies show phospho-tau in CSF and blood are linked more tightly to
relatively modest impact on tau aggregation and tau- amyloid deposition than tau-pathology per se [176–178].
associated neurodegenertive phenotypes. Given these Some recent reports do suggest that CSF and plasma tau
rather modest impacts on pathology and neurodegen- levels continue to increase as disease progresses, but

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 219

the biological correlates of these associations remain Open Questions for Immune Modulation in AD
enigmatic. A recent study in mice shows that increases
in extracellular tau appear to be a response to both Aβ 1. Will immune manipulation in any direction be safe
amyloid and Danish familial dementia amyloid [179]. enough in an elderly population to conduct long-term
Notably, most extracellular tau is an amino terminal studies in AD? Therapies that modulate the immune
fragment of full-length tau that lacks the carboxyl termi- system often are accompanied by untoward side effects.
nus. Such data points to crosstalk between amyloid and Such side effects are often better tolerated in younger
tau pathologies that can be independent of tau pathol- individuals or when the therapy is intermittently used.
ogy. The lack of mechanistic insight into this aspect of Clinical experience of immune activation strategies in
crosstalk between amyloid and tau is, again, a gap in our cancer highlights the many untoward side effects of
understanding that the field should address. these therapies. It is likely that immune therapies used
in AD that are sufficient to modulate immune activation
states in the brain will be accompanied by some periph-
Therapies Targeting Immune Activation eral side effects. Given the high bar for safety and lack of
States in AD negative effects of the therapy on functional and cogni-
tive outcomes, it may be challenging to find a therapeu-
Rationale  Neuroinflammation has been a long-standing, tic window which permits sufficient target engagement
but until more recently ill-defined, target for AD and many without inducing significant side effects.
other neurodegenerative diseases [180]. Genetic data that 2. Do we have good enough biomarkers to assess target
firmly links proteins that regulate microglial and other engagement and modulation of immune activation states
innate immune responses to AD risk has recently elevated in the brain? Although PET ligands that assess micro-
the interest in novel developing immune modulators as gliosis are utilized, the signal to noise ratio of these
disease-modifying agents [181, 182]. Genetic, pathologic, ligands based on binding to TSPO is relatively small
and experimental modeling data now strongly supports the especially in the elderly [195]. Furthermore, how such
notion that activation of innate immune signaling pathways ligands inform on immune activation states is unclear.
in the brain confers protection from AD and that inhibi- CSF fluid biomarkers of brain inflammation are also not
tion of these pathways may confer risk (reviewed in [180]). well validated [196]. Unless clean robust data showing
Such data suggests that therapeutics designed to activate clinical efficacy emerges in early phase studies, advances
the immune systems are likely to provide benefit in AD in biomarkers that assess brain immune activation states
and directly contradict the long-standing dogma that inhibi- will likely be needed to ensure appropriate go-no go
tion of inflammatory and innate immune pathways would decisions are made during clinical development of these
be beneficial. various agents.
I have recently extensively reviewed how the emerging 3. If manipulation of many immune targets seems to have
genetic data and previous pathological, modeling, and epi- differential impacts on amyloid and tau pathology and
demiological data lead to fundamentally distinct interpreta- variable impacts on synaptic and behavioral functions in
tions of how to manipulate the immune system for benefit in preclinical models might we do more harm than good?
AD [180]. As there is yet little consensus in the field regard- Not all immune targeting therapies are evaluated in both
ing directionality, both immune activating and immune amyloid and tau models before they enter clinical stud-
inhibitory approaches are being developed and tested clini- ies. Given numerous published examples of disparate
cally. Indeed, a CSF1R antagonist [183], TNFα antagonists impacts of an immune manipulation on tau, amyloid,
[184, 185], an NLRPL3 antagonist [186, 187], a p38 MAPK and neurodegeneration in preclinical studies, it may be
inhibitor [188], Chromyln in combination with ibuprofen that an immune targeting therapy does more harm than
[189], lenadolidomide [190], and other approaches are being good (reviewed in [180]). Indeed, both scientists and
pursued conceptually as immune inhibitory strategies in AD. regulatory bodies probably should insist on more thor-
In contrast, GM-CSF (a CSFR1 activating ligand) [191, ough preclinical assessments of AD immune therapies,
192], TREM2 agonists (both small molecule and antibody- or at least ensure that both amyloid and tau pathologies
based approaches) [193], and a CD33/Siglec3 antagonists are assessed during the course of the trial.
[194] (antibody) are being developed as immune activating 4. When during disease progression should an immune
approaches for AD. Additional therapeutics with less clear- modulatory drug be tested? Immune manipulations may
cut actions are also being evaluated. Given the large number have beneficial or harmful consequences on brain func-
of agents and the lack of any truly positive data from clinical tion if they have different impacts on amyloid, tau, and
studies, I will not summarize the clinical data and simply neuronal or brain circuit function. The possible disease
highlight the most pressing open questions. state dependence of efficacy raises many issues with

13
220 T. E. Golde

respect to clinical testing of immune modulators in AD. signal reductions, actual effects on amyloid pathology, and
Clinical testing with these agents is largely conducted some insight into the biological underpinnings of ARIA-E.
in mild AD patients and that will likely be the initial A fourth lesson is that we really do need to do better on
intent to treat population unless we have more refined the practices of good therapeutic development practices. In
biomarkers of innate immune activation states in the the absence of an unambiguous beneficial clinical signal,
brain. As these therapies may impact amyloid, tau, and compelling evidence for target engement and supporting
neurodegeneration, it is imperative that trials monitor evidence for possible disease modification should be ascer-
these pathologies using avaible biomarkers. tained during early phase trials. Without such data, large-
scale phase 3 studies have very low probabilities of success
and are not likely to inform the field as to how to move
Lessons Learned from Two Decades forward.
of Disease‑Modifying Trials A fifth lesson is that we should do our best to test agents at
the time in disease progression when they are more likely to
The first lesson is that disease modification in AD is hard. have an impact on disease. The concept of testing therapeu-
Given the slow and variable rates of symptom progression, tics at the right time was a major focus of recent perspective
it is difficult, though possible, to measure clinically, and only written by me and several colleagues [55]. Indeed, the current
recently do we have biomarkers that can enable tracking of debate over the possible efficacy of Aduhelm in symptomatic
progression in the long prodromal silent phase of the dis- AD partially deflects us from addressing the critical unan-
ease as the pathology develops. Large-scale trials designed swered questions about Aduhelm and other Aβ targeting anti-
to assess impacts on cognitive and functional decline can bodies that appear to engage and modify the target pathology
assess clinical impacts, but these trials remain expensive in the brain. That question is as follows: “Might these anti-
and are lengthy trials to conduct. Though the recent FDA bodies work in primary or secondary prevention studies?” If
approval of Aduhelm indicates a willingness to potentially Aduhelm or another antibody truly prevented or substantially
use a biomarker of pathology as a surrogate for clinical effi- reduced amyloid deposition during the long prodromal silent
cacy, the long-term outcomes of a biomarker-based approval phase of AD, then this or a similar approach may ultimately be
may not always prove to achieve a positive benefit to risk how we prevent AD. Alternatively, if it did not alter the clini-
impact. Perhaps with improvements in the power of bio- cal appearance of AD despite impacting amyloid deposition,
markers to predict future cognitive trajectories, it will be then we would have to critically rethink the role of amyloid
possible to assess efficacy of a potentially disease-modifying in AD. Unfortunately, Aduhelm was never included in any of
agent in the presymptomatic phase of disease in relatively the large-scale prevention studies that have been underway,
small cohorts [197, 198]. and we will have to wait many more years before we have
The second lesson is that use of biomarkers to both iden- an answer to the critical question of whether targeting amy-
tify and stratify intent to treat populations with dementia loid in the prodromal phase of disease impacts subsequent
or preclinical stages of dementia and to track impact on development of AD.
underlying pathology progression is almost requisite to
conduct a rigorous clinical study. Routine use of biomark-
ers, especially imaging modalities, adds significant costs Conclusions—Challenges for the Field
and participant burden to the trial. Nevertheless, without
these biomarkers, we are in many ways blind to impact on Building off the first lesson that disease modification is hard,
underlying pathologies and possible unanticipated adverse I will conclude with several challenges to the field. The first
effects. Furthermore, if one is targeting an AD pathology is that a patient suffering from AD does not really care about
and the trial includes participants without AD, not only is disease modification; they want their symptoms to improve
the trial potentially confounded, but ethically we are expos- or disease progression to be dramatically halted. A slight
ing individuals who are highly unlikely to benefit from the slowing of decline in cognition and functional ability (20%
trial to an unnecessary risk. per year) is really not something the patient or caregiver
A third lesson is that when conducting disease-modifying will be able to discern, at least in the short term. Thus, we
trials in AD, we should try our best to enroll participants need to think about ways to dramatically improve cognitive
in brain autopsy donation programs. Most AD trials enroll and functional status in the symptomatic phase or, if we
elderly individuals and obtaining and studying their brain are shooting for disease modification, aim for a much larger
following an intervention might provide key insights into the impact on functional decline.
efficacy or adverse effects of a given therapy. If this had been In order to achieve larger impacts in the clinical phase
standard of practice for the last decade, then we would likely of AD, we likely need to revisit development of sympto-
have more insight into the relationship between amyloid matic therapies designed to directly improve cognition and

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 221

function. Many targets that were once thought to be undrug- that we can significantly impact decline in symptomatic
gable can now be successfully targeted and new targets that AD with a single agent is likely wishful thinking. We need
might improve cognitive function independent of pathol- to approach symptomatic AD more like organ failure and
ogy have been identified. Though development of cogni- develop cocktails of agents that both target underlying
tive enhancing agents has not been completely ignored and pathologies, impact cognitive function, and have some abil-
a few promising therapies are in clinical development, the ity to regenerate and repair the damage. Such combinatorial
resources behind efforts to develop symptomatic therapies therapies will be challenging to develop and must utilize
are miniscule in comparison to the resources supporting the creative non-dogmatic approaches based in rigorous transla-
largely failed efforts to develop disease-modifying agents. tional science. Such efforts will likely yield more failure than
As noted above, another key challenge is to really try successes, but unless we rigorously and systematically try to
to align clinical testing of agents that are designed to be develop such a combinatorial approach, we cannot succeed.
disease modifying in the stage of disease that might provide Indeed, we are likely at least a decade or two away from
the largest clinical signal. Many of the scientific challenges identifying safe highly effective prophylactic interventions
to conducting such trials have now been largely overcome; for AD and universally deploying them. Thus, symptomatic
we can identify individuals with underlying AD patholo- AD will continue to be a prevalent and growing problem and
gies and track progression of those pathologies long before we must take disruptive high-risk approaches to improving
clinical symptoms emerge. We can also use genetics and bio- lives of those suffering with it, or at risk for developing it,
markers to assess individuals at risk for disease even before in the future.
radiologic evidence of pathology emerges; however, testing
therapies in prevention remains challenging, as there many Supplementary Information  The online version contains supplemen-
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 13311-0​ 22-0​ 1201-2.
financial and practical disincentives to conducting such
studies. As a society, we need to think how to incentivize Acknowledgements  I would like to thank Drs. Karen McFarland and
the private sector to conduct or participate in prevention Jane Golde for proofing and editing this manuscript.
studies in a way that produces an affordable public health
solution to prevention of AD. If, for example, CMS does pay Required Author Forms  Disclosure forms provided by the author are
for the cost of Aduhelm in ongoing clinical trials, maybe availablewith the online version of this article.
there can be a more codified public–private cost-sharing in
Funding  TEG is supported by grants from the NIH (U01AG046139,
extended trial models that could help support the long-term P30AG066506, RF1AG057933, RF1AG064942, R01AG062514,
studies of therapies designed for AD prevention. This cost- RF1AG064914). TEG is a co-founder of Lacerta Therapeutics Inc.
sharing could reduce the financial risk for a private sector and Andante Biologics Inc. He has served on advisory boards for Eli
and help them to engage in the types of trials that are likely Lilly, BMS, Novartis, Abbvie, Lundbeck, Pfizer, and Promis Neuro-
sciences Inc.
to yield the biggest public health impact. Certainly, current
prevention initiatives illustrate the power of these partner- Open Access  This article is licensed under a Creative Commons Attri-
ships [90, 92, 199], but a more standardized regulatory path bution 4.0 International License, which permits use, sharing, adapta-
with appropriate financial models that is jointly developed tion, distribution and reproduction in any medium or format, as long
with the many public and private stakeholders might ensure as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
that these partnerships are lasting and not undermined by were made. The images or other third party material in this article are
changes in the regulatory and payer landscape. included in the article's Creative Commons licence, unless indicated
The bar for safety is also very high for any therapeutic otherwise in a credit line to the material. If material is not included in
used as a prevention agent. I have previously suggested such the article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
a drug needs to be at least as safe as a statin [10]. Given the need to obtain permission directly from the copyright holder. To view a
toxicities observed with most anti-Aβ therapies that have copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.
human data supporting target engagement, the field needs to
grapple with the possibility that a true prevention study that
tests the core postulate of the amyloid cascade hypothesis
that preventing amyloid position will prevent AD is unlikely
References
to occur in the near term unless we identify safer approaches. 1. 2021 Alzheimer’s disease facts and figures. Alzheimers Dement.
A final challenge is that we likely need to think about dis- 2021:17327–406.
ease modification in symptomatic AD a little differently in 2. Reitz C, Brayne C, Mayeux R: Epidemiology of Alzheimer dis-
order to achieve the larger impact needed. AD is an incred- ease. Nat Rev Neurol 2011;7:137–152.
3. The LN. A contentious FDA ruling for Alzheimer’s disease. Lan-
ibly complex disorder and by the time symptoms manifest cet Neurol. 2021;20(8):585.
almost all cell types within the brain have been altered and 4. Knopman DS, Jones DT, Greicius MD. Failure to demonstrate
pathology building for over two decades. Thus, the notion efficacy of aducanumab: an analysis of the EMERGE and

13
222 T. E. Golde

ENGAGE trials as reported by Biogen, December 2019. Alz- amyloid-beta immunization in Alzheimer’s disease. Brain.
heimers Dement. 2021;17(4):696–701. 2019;142(7):2113–26.
5. Alexander GC, Knopman DS, Emerson SS, Ovbiagele B, Kryscio 26. Nicoll JA, Wilkinson D, Holmes C, Steart P, Markham H,
RJ, Perlmutter JS, et al. Revisiting FDA approval of aducanumab. Weller RO. Neuropathology of human Alzheimer disease after
N Engl J Med. 2021;385(9):769–71. immunization with amyloid-beta peptide: a case report. Nat
6. Cummings J, Lee G, Ritter A, Sabbagh M, Zhong K. Alzheimer’s Med. 2003;9(4):448–52.
disease drug development pipeline: 2020. Alzheimers Dement 27. White AC, Cloutier P, Coughlin RT. A purified saponin acts as
(N Y). 2020;6(1):e12050. an adjuvant for a T-independent antigen. Adv Exp Med Biol.
7. Cummings JL, Morstorf T, Zhong K. Alzheimer’s disease drug- 1991;303:207–10.
development pipeline: few candidates, frequent failures. Alzhei- 28. Ragupathi G, Gardner JR, Livingston PO, Gin DY. Natural and
mers Res Ther. 2014;6(4):37. synthetic saponin adjuvant QS-21 for vaccines against cancer.
8. van Bokhoven P, de Wilde A, Vermunt L, Leferink PS, Heetveld Expert Rev Vaccines. 2011;10(4):463–70.
S, Cummings J, et al. The Alzheimer’s disease drug development 29. Fernandez-Tejada A, Chea EK, George C, Pillarsetty N, Gardner
landscape. Alzheimers Res Ther. 2021;13(1):186. JR, Livingston PO, et al. Development of a minimal saponin
9. Hardy J, Selkoe DJ. The amyloid hypothesis of Alzheimer’s dis- vaccine adjuvant based on QS-21. Nat Chem. 2014;6(7):635–43.
ease: progress and problems on the road to therapeutics. Science. 30. Wang CY, Wang PN, Chiu MJ, Finstad CL, Lin F, Lynn S, et al.
2002;297(5580):353–6. UB-311, a novel UBITh((R)) amyloid beta peptide vaccine for mild
10. Golde TE, Schneider LS, Koo EH. Anti-abeta therapeutics in Alzheimer’s disease. Alzheimers Dement (N Y). 2017;3(2):262–72.
Alzheimer’s disease: the need for a paradigm shift. Neuron. 31. Hickman DT, Lopez-Deber MP, Ndao DM, Silva AB, Nand D,
2011;69(2):203–13. Pihlgren M, et al. Sequence-independent control of peptide con-
11. Golde TE, Petrucelli L, Lewis J. Targeting Abeta and tau in formation in liposomal vaccines for targeting protein misfolding
Alzheimer’s disease, an early interim report. Exp Neurol. diseases. J Biol Chem. 2011;286(16):13966–76.
2010;223(2):252–66. 32. Muhs A, Hickman DT, Pihlgren M, Chuard N, Giriens V,
12. Citron M. Alzheimer’s disease: strategies for disease modifica- Meerschman C, et al. Liposomal vaccines with conformation-
tion. Nat Rev Drug Discov. 2010;9(5):387–98. specific amyloid peptide antigens define immune response and
13. Spencer B, Masliah E. Immunotherapy for Alzheimer’s disease: efficacy in APP transgenic mice. Proc Natl Acad Sci U S A.
past, present and future. Front Aging Neurosci. 2014;6:114. 2007;104(23):9810–5.
14. Pardridge WM. The blood-brain barrier: bottleneck in brain drug 33. Belichenko PV, Madani R, Rey-Bellet L, Pihlgren M, Becker
development. NeuroRx. 2005;2(1):3–14. A, Plassard A, et al. An anti-beta-amyloid vaccine for treating
15. Levites Y, Smithson LA, Price RW, Dakin RS, Yuan B, Sierks cognitive deficits in a mouse model of down syndrome. PLoS
MR, et al. Insights into the mechanisms of action of anti-Abeta One. 2016;11(3):e0152471.
antibodies in Alzheimer’s disease mouse models. Faseb J. 34. Lacosta AM, Pascual-Lucas M, Pesini P, Casabona D, Perez-
2006;20(14):2576–8. Grijalba V, Marcos-Campos I, et al. Safety, tolerability and immu-
16. Morgan D. Immunotherapy for Alzheimer’s disease. J Alzhei- nogenicity of an active anti-Abeta40 vaccine (ABvac40) in patients
mers Dis. 2006;9(3 Suppl):425–32. with Alzheimer’s disease: a randomised, double-blind, placebo-
17. Golde TE. Open questions for Alzheimer’s disease immuno- controlled, phase I trial. Alzheimers Res Ther. 2018;10(1):12.
therapy. Alzheimers Res Ther. 2014;6(1):3. 35. June CH, O’Connor RS, Kawalekar OU, Ghassemi S, Milone
18. Schenk D, Barbour R, Dunn W, Gordon G, Grajeda H, Guido MC. CAR T cell immunotherapy for human cancer. Science.
T, et al. Immunization with amyloid-beta attenuates Alzheimer- 2018;359(6382):1361–5.
disease-like pathology in the PDAPP mouse [see comments]. 36. Sevigny J, Chiao P, Bussiere T, Weinreb PH, Williams L, Maier
Nature. 1999;400(6740):173–7. M, et al. The antibody aducanumab reduces Abeta plaques in
19. Gilman S, Koller M, Black RS, Jenkins L, Griffith SG, Fox NC, Alzheimer’s disease. Nature. 2016;537(7618):50–6.
et al. Clinical effects of A{beta} immunization (AN1792) in 37. Swanson CJ, Zhang Y, Dhadda S, Wang J, Kaplow J, Lai RYK,
patients with AD in an interrupted trial. Neurology. 2005. et al. A randomized, double-blind, phase 2b proof-of-concept
20. Fox NC, Black RS, Gilman S, Rossor MN, Griffith SG, Jenkins L, clinical trial in early Alzheimer’s disease with lecanemab, an anti-
et al. Effects of A{beta} immunization (AN1792) on MRI meas- Abeta protofibril antibody. Alzheimers Res Ther. 2021;13(1):80.
ures of cerebral volume in Alzheimer disease. Neurology. 2005. 38. Klein G, Delmar P, Voyle N, Rehal S, Hofmann C, Abi-Saab D,
21. Vellas B, Black R, Thal LJ, Fox NC, Daniels M, McLennan G, et al. Gantenerumab reduces amyloid-beta plaques in patients
et al. Long-term follow-up of patients immunized with AN1792: with prodromal to moderate Alzheimer’s disease: a PET substudy
reduced functional decline in antibody responders. Curr Alzhei- interim analysis. Alzheimers Res Ther. 2019;11(1):101.
mer Res. 2009;6(2):144–51. 39. Salloway S, Farlow M, McDade E, Clifford DB, Wang G,
22. Uro-Coste E, Russano de Paiva G, Guilbeau-Frugier C, Sastre N, Llibre-Guerra JJ, et  al. A trial of gantenerumab or solane-
Ousset PJ, da Silva NA, et al. Cerebral amyloid angiopathy and zumab in dominantly inherited Alzheimer’s disease. Nat Med.
microhemorrhages after amyloid beta vaccination: case report 2021;27(7):1187–96.
and brief review. Clin Neuropathol. 2010;29(4):209–16. 40. Mintun MA, Lo AC, Duggan Evans C, Wessels AM, Ardayfio
23. Holmes C, Boche D, Wilkinson D, Yadegarfar G, Hopkins V, PA, Andersen SW, et al. Donanemab in early Alzheimer’s dis-
Bayer A, et al. Long-term effects of Abeta42 immunisation in Alz- ease. N Engl J Med. 2021;384(18):1691–704.
heimer’s disease: follow-up of a randomised, placebo-controlled 41. Lowe SL, Willis BA, Hawdon A, Natanegara F, Chua L, Foster
phase I trial. Lancet. 2008;372(9634):216–23. J, et al. Donanemab (LY3002813) dose-escalation study in Alz-
24. Zotova E, Holmes C, Johnston D, Neal JW, Nicoll JA, Boche heimer’s disease. Alzheimers Dement (N Y). 2021;7(1):e12112.
D. Microglial alterations in human Alzheimer’s disease fol- 42. Ayton S. Brain volume loss due to donanemab. Eur J Neurol.
lowing Abeta42 immunization. Neuropathol Appl Neurobiol. 2021;28(9):e67–8.
2011;37(5):513–24. 43. Sperling RA, Jack CR Jr, Black SE, Frosch MP, Greenberg SM,
25. Nicoll JAR, Buckland GR, Harrison CH, Page A, Harris S, Love Hyman BT, et al. Amyloid-related imaging abnormalities in
S, et al. Persistent neuropathological effects 14 years following amyloid-modifying therapeutic trials: recommendations from

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 223

the Alzheimer’s Association Research Roundtable Workgroup. 63. Golde TE, Koo EH, Felsenstein KM, Osborne BA, Miele L.
Alzheimers Dement. 2011;7(4):367–85. gamma-Secretase inhibitors and modulators. Biochim Biophys
44. Landen JW, Andreasen N, Cronenberger CL, Schwartz PF, Acta. 2013;1828(12):2898–907.
Borjesson-Hanson A, Ostlund H, et al. Ponezumab in mild-to- 64. Weggen S, Eriksen JL, Das P, Sagi SA, Wang R, Pietrzik CU,
moderate Alzheimer’s disease: randomized phase II PET-PIB et al. A subset of NSAIDs lower amyloidogenic Abeta42 indepen-
study. Alzheimers Dement (N Y). 2017;3(3):393–401. dently of cyclooxygenase activity. Nature. 2001;414(6860):212–6.
45. Siemers ER, Sundell KL, Carlson C, Case M, Sethuraman G, 65. Walsh DM, Klyubin I, Fadeeva JV, Rowan MJ, Selkoe DJ. Amy-
Liu-Seifert H, et al. Phase 3 solanezumab trials: secondary out- loid-beta oligomers: their production, toxicity and therapeutic
comes in mild Alzheimer’s disease patients. Alzheimers Dement. inhibition. Biochem Soc Trans. 2002;30(4):552–7.
2016;12(2):110–20. 66. Cherny RA, Atwood CS, Xilinas ME, Gray DN, Jones WD,
46. Honig LS, Vellas B, Woodward M, Boada M, Bullock R, Borrie McLean CA, et al. Treatment with a copper-zinc chelator mark-
M, et al. Trial of solanezumab for mild dementia due to Alzhei- edly and rapidly inhibits beta-amyloid accumulation in Alzhei-
mer’s disease. N Engl J Med. 2018;378(4):321–30. mer’s disease transgenic mice. Neuron. 2001;30(3):665–76.
47. Doody RS, Thomas RG, Farlow M, Iwatsubo T, Vellas B, Joffe 67. McLaurin J, Golomb R, Jurewicz A, Antel JP, Fraser PE. Inositol
S, et al. Phase 3 trials of solanezumab for mild-to-moderate Alz- stereoisomers stabilize an oligomeric aggregate of Alzheimer
heimer’s disease. N Engl J Med. 2014;370(4):311–21. amyloid beta peptide and inhibit abeta -induced toxicity. J Biol
48. Salloway S, Sperling R, Fox NC, Blennow K, Klunk W, Raskind Chem. 2000;275(24):18495–502.
M, et al. Two phase 3 trials of bapineuzumab in mild-to-moderate 68. McLaurin J, Yang D, Yip CM, Fraser PE. Review: modulat-
Alzheimer’s disease. N Engl J Med. 2014;370(4):322–33. ing factors in amyloid-beta fibril formation. J Struct Biol.
49. Yoshida K, Moein A, Bittner T, Ostrowitzki S, Lin H, Honigberg 2000;130(2–3):259–70.
L, et al. Pharmacokinetics and pharmacodynamic effect of cren- 69. Wang Q, Yu X, Li L, Zheng J. Inhibition of amyloid-beta aggrega-
ezumab on plasma and cerebrospinal fluid beta-amyloid in patients tion in Alzheimer’s disease. Curr Pharm Des. 2014;20(8):1223–43.
with mild-to-moderate Alzheimer’s disease. Alzheimers Res Ther. 70. Yang F, Lim GP, Begum AN, Ubeda OJ, Simmons MR, Ambe-
2020;12(1):16. gaokar SS, et al. Curcumin inhibits formation of amyloid beta
50. Salloway S, Chalkias S, Barkhof F, Burkett P, Barakos J, Purcell oligomers and fibrils, binds plaques, and reduces amyloid in vivo.
D, et al. Amyloid-related imaging abnormalities in 2 phase 3 J Biol Chem. 2005;280(7):5892–901.
studies evaluating aducanumab in patients with early Alzheimer 71. Necula M, Kayed R, Milton S, Glabe CG. Small-molecule inhibi-
disease. JAMA Neurol. 2021. tors of aggregation indicate that amyloid beta oligomerization
51. Wey SP, Weng CC, Lin KJ, Yao CH, Yen TC, Kung HF, et al. and fibrillization pathways are independent and distinct. J Biol
Validation of an (18)F-labeled biphenylalkyne as a positron emis- Chem. 2007.
sion tomography imaging agent for beta-amyloid plaques. Nucl 72. Hey JA, Yu JY, Versavel M, Abushakra S, Kocis P, Power A,
Med Biol. 2009;36(4):411–7. et al. Clinical pharmacokinetics and safety of ALZ-801, a novel
52. McDade E, Wang G, Gordon BA, Hassenstab J, Benzinger prodrug of tramiprosate in development for the treatment of Alz-
TLS, Buckles V, et al. Longitudinal cognitive and biomarker heimer’s disease. Clin Pharmacokinet. 2018;57(3):315–33.
changes in dominantly inherited Alzheimer disease. Neurology. 73. Kutzsche J, Jurgens D, Willuweit A, Adermann K, Fuchs C,
2018;91(14):e1295–306. Simons S, et  al. Safety and pharmacokinetics of the orally
53. McDade E, Bateman RJ. Tau positron emission tomography in available antiprionic compound PRI-002: a single and multi-
autosomal dominant Alzheimer disease: small windows, big pic- ple ascending dose phase I study. Alzheimers Dement (N Y).
ture. JAMA Neurol. 2018. 2020;6(1):e12001.
54. Antolini L, DiFrancesco JC, Zedde M, Basso G, Arighi A, Shima 74. Siemers ER, Dean RA, Friedrich S, Ferguson-Sells L, Gonzales
A, et al. Spontaneous ARIA-like events in cerebral amyloid angi- C, Farlow MR, et al. Safety, tolerability, and effects on plasma
opathy-related inflammation: a multicenter prospective longitu- and cerebrospinal fluid amyloid-beta after inhibition of gamma-
dinal cohort study. Neurology. 2021;97(18):e1809–22. secretase. Clin Neuropharmacol. 2007;30(6):317–25.
55. Golde TE, DeKosky ST, Galasko D. Alzheimer’s disease: the 75. Siemers ER, Quinn JF, Kaye J, Farlow MR, Porsteinsson A,
right drug, the right time. Science. 2018;362(6420):1250–1. Tariot P, et al. Effects of a gamma-secretase inhibitor in a ran-
56. De Strooper B, Vassar R, Golde T. The secretases: enzymes domized study of patients with Alzheimer disease. Neurology.
with therapeutic potential in Alzheimer disease. Nat Rev Neurol. 2006;66(4):602–4.
2010;6(2):99–107. 76. Roberts C, Kaplow J, Giroux M, Krause S, Kanekiyo M. Amyloid
57. Golde TE. Alzheimer disease therapy: can the amyloid cascade and APOE status of screened subjects in the elenbecestat Missio-
be halted? J Clin Invest. 2003;111(1):11–8. nAD phase 3 program. J Prev Alzheimers Dis. 2021;8(2):218–23.
58. Walsh DM, Selkoe DJ. Deciphering the molecular basis of mem- 77. Lo AC, Evans CD, Mancini M, Wang H, Shcherbinin S, Lu M,
ory failure in Alzheimer’s disease. Neuron. 2004;44(1):181–93. et al. Phase II (NAVIGATE-AD study) results of LY3202626
59. Moore BD, Martin J, de Mena L, Sanchez J, Cruz PE, Ceballos- effects on patients with mild Alzheimer’s disease dementia. J
Diaz C, et al. Short Abeta peptides attenuate Abeta42 toxicity Alzheimers Dis Rep. 2021;5(1):321–36.
in vivo. J Exp Med. 2018;215(1):283–301. 78. Wessels AM, Lines C, Stern RA, Kost J, Voss T, Mozley LH,
60. Kim J, Onstead L, Randle S, Price R, Smithson L, Zwizinski C, et al. Cognitive outcomes in trials of two BACE inhibitors in
et al. Abeta40 inhibits amyloid deposition in vivo. J Neurosci. Alzheimer’s disease. Alzheimers Dement. 2020;16(11):1483–92.
2007;27(3):627–33. 79. Qiu R, Ahn JE, Alexander R, Brodney MA, He P, Leurent C,
61. McGowan E, Pickford F, Kim J, Onstead L, Eriksen J, Yu C, et  al. Safety, tolerability, pharmacokinetics, and pharmaco-
et al. Abeta42 is essential for parenchymal and vascular amyloid dynamic effects of PF-06751979, a potent and selective oral
deposition in mice. Neuron. 2005;47(2):191–9. BACE1 inhibitor: results from phase I studies in healthy adults
62. Kretner B, Trambauer J, Fukumori A, Mielke J, Kuhn PH, Kremmer and healthy older subjects. J Alzheimers Dis. 2019;71(2):581–95.
E, et al. Generation and deposition of Abeta43 by the virtually 80. Neumann U, Ufer M, Jacobson LH, Rouzade-Dominguez ML,
inactive presenilin-1 L435F mutant contradicts the presenilin Huledal G, Kolly C, et al. The BACE-1 inhibitor CNP520 for
loss-of-function hypothesis of Alzheimer’s disease. EMBO Mol prevention trials in Alzheimer’s disease. EMBO Mol Med.
Med. 2016;8(5):458–65. 2018;10(11).

13
224 T. E. Golde

81. Sur C, Kost J, Scott D, Adamczuk K, Fox NC, Cummings JL, et al. with antisense oligonucleotides reduces toxic amyloid-beta pro-
BACE inhibition causes rapid, regional, and non-progressive volume duction. Mol Ther. 2018;26(6):1539–51.
reduction in Alzheimer’s disease brain. Brain. 2020;143(12):3816–26. 99. Dawson TM, Golde TE, Lagier-Tourenne C. Animal models of
82. Kennedy ME, Stamford AW, Chen X, Cox K, Cumming JN, neurodegenerative diseases. Nat Neurosci. 2018;21(10):1370–9.
Dockendorf MF, et al. The BACE1 inhibitor verubecestat (MK- 100. Lewis J, Dickson DW, Lin WL, Chisholm L, Corral A, Jones G,
8931) reduces CNS beta-amyloid in animal models and in Alzhei- et al. Enhanced neurofibrillary degeneration in transgenic mice
mer’s disease patients. Sci Transl Med. 2016;8(363):363ra150. expressing mutant tau and APP. Science. 2001;293(5534):1487–91.
83. McDade E, Voytyuk I, Aisen P, Bateman RJ, Carrillo MC, 101. Gotz J, Chen F, van Dorpe J, Nitsch RM. Formation of neurofi-
De Strooper B, et al. The case for low-level BACE1 inhibi- brillary tangles in P301l tau transgenic mice induced by Abeta
tion for the prevention of Alzheimer disease. Nat Rev Neurol. 42 fibrils. Science. 2001;293(5534):1491–5.
2021;17(11):703–14. 102. Lee VM, Goedert M, Trojanowski JQ. Neurodegenerative
84. Brothers HM, Gosztyla ML, Robinson SR. The physiological tauopathies. Annu Rev Neurosci. 2001;24:1121–59.
roles of amyloid-beta peptide hint at new ways to treat Alzhei- 103. Hutton M. Molecular genetics of chromosome 17 tauopathies.
mer’s disease. Front Aging Neurosci. 2018;10:118. Ann N Y Acad Sci. 2000;920:63–73.
85. Ross J, Sharma S, Winston J, Nunez M, Bottini G, Franceschi 104. Seabrook GR, Ray WJ, Shearman M, Hutton M. Beyond amy-
M, et al. CHF5074 reduces biomarkers of neuroinflammation loid: the next generation of Alzheimer’s disease therapeutics.
in patients with mild cognitive impairment: a 12-week, double- Mol Interv. 2007;7(5):261–70.
blind, placebo-controlled study. Curr Alzheimer Res. 2013. 105. Golde TE, Petrucelli L, Lewis J. Targeting Aβ and tau in
86. Morimoto B. Drug Development for Neurodegenerative Alzheimer’s disease, an early interim report. Exp Neurol.
Diseases--Second Annual marcus evans Conference. Advances 2010;223(2):252–66.
in drug development for NDD and expediting discovery 106. Dujardin S, Commins C, Lathuiliere A, Beerepoot P, Fernandes
through novel compounds and sound clinical trials. IDrugs. AR, Kamath TV, et  al. Tau molecular diversity contributes
2010;13(7):440–3. to clinical heterogeneity in Alzheimer’s disease. Nat Med.
87. Black S, Wilcock GK, Hawworth J, Hendrix S, Zavitz K, 2020;26(8):1256–63.
Christensen Db M-H, et al. Efficacy and safety ofMPC-7869 107. Sato C, Barthelemy NR, Mawuenyega KG, Patterson BW, Gordon
(R-flurbiprofen), a selective Abeta42 loering agent in mild Alz- BA, Jockel-Balsarotti J, et al. Tau kinetics in neurons and the
heimer’s disease: results of a 12-month phase 2 trial and 1-year human central nervous system. Neuron. 2018;98(4):861–4.
follow on study. Neurology. 2006;66(Suppl2):A347. 108. Wang Y, Mandelkow E. Tau in physiology and pathology. Nat
88. Green RC, Schneider LS, Amato DA, Beelen AP, Wilcock G, Rev Neurosci. 2016;17(1):5–21.
Swabb EA, et al. Effect of tarenflurbil on cognitive decline and 109. Sigurdsson EM. Immunotherapy targeting pathological tau pro-
activities of daily living in patients with mild Alzheimer disease: tein in Alzheimer’s disease and related tauopathies. J Alzheimers
a randomized controlled trial. JAMA. 2009;302(23):2557–64. Dis. 2008;15(2):157–68.
89. Khachaturian ZS, Barnes D, Einstein R, Johnson S, Lee V, Roses 110. Sigurdsson EM. Tau-focused immunotherapy for Alzhei-
A, et al. Developing a national strategy to prevent dementia: Leon mer’s disease and related tauopathies. Curr Alzheimer Res.
Thal Symposium 2009. Alzheimers Dement. 2010;6(2):89–97. 2009;6(5):446–50.
90. Reiman EM, Langbaum JB, Fleisher AS, Caselli RJ, Chen K, 111. Peng C, Trojanowski JQ, Lee VM. Protein transmission in neu-
Ayutyanont N, et al. Alzheimer’s Prevention Initiative: a plan to rodegenerative disease. Nat Rev Neurol. 2020;16(4):199–212.
accelerate the evaluation of presymptomatic treatments. J Alz- 112. Vaquer-Alicea J, Diamond MI. Propagation of protein aggre-
heimers Dis. 2011;26(Suppl 3):321–9. gation in neurodegenerative diseases. Annu Rev Biochem.
91. Mills SM, Mallmann J, Santacruz AM, Fuqua A, Carril M, 2019;88:785–810.
Aisen PS, et al. Preclinical trials in autosomal dominant AD: 113. Boutajangout A, Ingadottir J, Davies P, Sigurdsson EM. Passive
implementation of the DIAN-TU trial. Rev Neurol (Paris). immunization targeting pathological phospho-tau protein in a
2013;169(10):737–43. mouse model reduces functional decline and clears tau aggre-
92. Sperling RA, Rentz DM, Johnson KA, Karlawish J, Donohue M, gates from the brain. J Neurochem. 2011;118(4):658–67.
Salmon DP, et al. The A4 study: stopping AD before symptoms 114. Gu J, Sigurdsson EM. Immunotherapy for tauopathies. J Mol
begin? Sci Transl Med. 2014;6(228):228fs13. Neurosci. 2011;45(3):690–5.
93. McDade E, Bateman RJ. Stop Alzheimer’s before it starts. 115. Krishnamurthy PK, Deng Y, Sigurdsson EM. Mechanistic stud-
Nature. 2017;547(7662):153–5. ies of antibody-mediated clearance of tau aggregates using an
94. Ratni H, Alker A, Bartels B, Bissantz C, Chen W, Gerlach I, ex vivo brain slice model. Front Psychiatry. 2011;2:59.
et al. Discovery of RO7185876, a highly potent gamma-secretase 116. McEwan WA, Falcon B, Vaysburd M, Clift D, Oblak AL, Ghetti
modulator (GSM) as a potential treatment for Alzheimer’s dis- B, et al. Cytosolic Fc receptor TRIM21 inhibits seeded tau aggre-
ease. ACS Med Chem Lett. 2020;11(6):1257–68. gation. Proc Natl Acad Sci U S A. 2017;114(3):574–9.
95. Rynearson KD, Ponnusamy M, Prikhodko O, Xie Y, Zhang 117. Foss S, Bottermann M, Jonsson A, Sandlie I, James LC,
C, Nguyen P, et al. Preclinical validation of a potent gamma- Andersen JT. TRIM21-from intracellular immunity to therapy.
secretase modulator for Alzheimer’s disease prevention. J Exp Front Immunol. 2019;10:2049.
Med. 2021;218(4). 118. Vitale F, Ortolan J, Volpe BT, Marambaud P, Giliberto L,
96. Dobson CM, Knowles TPJ, Vendruscolo M. The amyloid phe- d’Abramo C. Intramuscular injection of vectorized-scFvMC1
nomenon and its significance in biology and medicine. Cold reduces pathological tau in two different tau transgenic models.
Spring Harb Perspect Biol. 2020;12(2). Acta Neuropathol Commun. 2020;8(1):126.
97. Ghosh U, Yau WM, Collinge J, Tycko R. Structural differences 119. Ising C, Gallardo G, Leyns CEG, Wong CH, Jiang H, Stewart
in amyloid-beta fibrils from brains of nondemented elderly indi- F, et al. AAV-mediated expression of anti-tau scFvs decreases
viduals and Alzheimer’s disease patients. Proc Natl Acad Sci U tau accumulation in a mouse model of tauopathy. J Exp Med.
S A. 2021;118(45). 2017;214(5):1227–38.
98. Chang JL, Hinrich AJ, Roman B, Norrbom M, Rigo F, Marr RA, 120. Goodwin MS, Sinyavskaya O, Burg F, O’Neal V, Ceballos-Diaz
et al. Targeting amyloid-beta precursor protein, APP, splicing C, Cruz PE, et al. Anti-tau scFvs targeted to the cytoplasm or

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 225

secretory pathway variably modify pathology and neurodegen- in patients with Alzheimer’s disease: a randomised, double-blind,
erative phenotypes. Mol Ther. 2021;29(2):859–72. placebo-controlled, phase 1 trial. Lancet Neurol. 2017;16(2):123–34.
121. Abyadeh M, Gupta V, Gupta V, Chitranshi N, Wu Y, Amirkhani 138. Kontsekova E, Zilka N, Kovacech B, Novak P, Novak M. First-
A, et al. Comparative analysis of aducanumab, zagotenemab in-man tau vaccine targeting structural determinants essential for
and pioglitazone as targeted treatment strategies for Alzhei- pathological tau-tau interaction reduces tau oligomerisation and
mer’s disease. Aging Dis. 2021;12(8):1964–76. neurofibrillary degeneration in an Alzheimer’s disease model.
122. West T, Hu Y, Verghese PB, Bateman RJ, Braunstein JB, Alzheimers Res Ther. 2014;6(4):44.
Fogelman I, et  al. Preclinical and clinical development of 139. Golde TE, Borchelt DR, Giasson BI, Lewis J. Thinking later-
ABBV-8E12, a humanized anti-tau antibody, for treatment of ally about neurodegenerative proteinopathies. J Clin Investig.
Alzheimer’s disease and other tauopathies. J Prev Alzheimers 2013;123(5):1847–55.
Dis. 2017;4(4):236–41. 140. Takeda S, Commins C, DeVos SL, Nobuhara CK, Wegmann
123. Koga S, Dickson DW, Wszolek ZK. Neuropathology of pro- S, Roe AD, et  al. Seed-competent high-molecular-weight
gressive supranuclear palsy after treatment with tilavonemab. tau species accumulates in the cerebrospinal fluid of Alzhei-
Lancet Neurol. 2021;20(10):786–7. mer’s disease mouse model and human patients. Ann Neurol.
124. Hoglinger GU, Litvan I, Mendonca N, Wang D, Zheng H, 2016;80(3):355–67.
Rendenbach-Mueller B, et al. Safety and efficacy of tilavonemab 141. Wesseling H, Mair W, Kumar M, Schlaffner CN, Tang S, Beerepoot
in progressive supranuclear palsy: a phase 2, randomised, pla- P, et al. Tau PTM profiles identify patient heterogeneity and stages
cebo-controlled trial. Lancet Neurol. 2021;20(3):182–92. of Alzheimer’s disease. Cell. 2020;183(6):1699–713 e13.
125. Hoglinger GU. Neuropathology of progressive supranuclear 142. Gallardo G, Wong CH, Ricardez SM, Mann CN, Lin KH, Leyns
palsy after treatment with tilavonemab - author’s reply. Lancet CEG, et al. Targeting tauopathy with engineered tau-degrading
Neurol. 2021;20(10):787–8. intrabodies. Mol Neurodegener. 2019;14(1):38.
126. Vaz M, Silvestre S. Alzheimer’s disease: recent treatment strate- 143. Woerman AL, Patel S, Kazmi SA, Oehler A, Freyman Y,
gies. Eur J Pharmacol. 2020;887:173554. Espiritu L, et al. Kinetics of human mutant tau prion forma-
127. Ayalon G, Lee SH, Adolfsson O, Foo-Atkins C, Atwal JK, Blendstrup tion in the brains of 2 transgenic mouse lines. JAMA Neurol.
M, et al. Antibody semorinemab reduces tau pathology in a transgenic 2017;74(12):1464.
mouse model and engages tau in patients with Alzheimer’s disease. 144. Gamache J, Benzow K, Forster C, Kemper L, Hlynialuk C, Furrow E,
Sci Transl Med. 2021;13(593). et al. Factors other than hTau overexpression that contribute to tauopa-
128. Kondo A, Shahpasand K, Mannix R, Qiu J, Moncaster J, thy-like phenotype in rTg4510 mice. Nat Commun. 2019;10(1):2479.
Chen CH, et al. Antibody against early driver of neurodegen- 145. He Z, Guo JL, McBride JD, Narasimhan S, Kim H, Changolkar
eration cis P-tau blocks brain injury and tauopathy. Nature. L, et al. Amyloid-beta plaques enhance Alzheimer’s brain tau-
2015;523(7561):431–6. seeded pathologies by facilitating neuritic plaque tau aggrega-
129. Andersson CR, Falsig J, Stavenhagen JB, Christensen S, Kartberg tion. Nat Med. 2018;24(1):29–38.
F, Rosenqvist N, et al. Antibody-mediated clearance of tau in 146. Strang KH, Golde TE, Giasson BI. MAPT mutations, tauopathy, and
primary mouse microglial cultures requires Fcgamma-receptor mechanisms of neurodegeneration. Lab Invest. 2019;99(7):912–28.
binding and functional lysosomes. Sci Rep. 2019;9(1):4658. 147. Alonso AD, Beharry C, Corbo CP, Cohen LS. Molecular mecha-
130. Bijttebier S, Theunis C, Jahouh F, Martins DR, Verhemeldonck nism of prion-like tau-induced neurodegeneration. Alzheimers
M, Grauwen K, et al. Development of immunoprecipitation - Dement. 2016;12(10):1090–7.
two-dimensional liquid chromatography - mass spectrometry 148. Kuchibhotla KV, Wegmann S, Kopeikina KJ, Hawkes J, Rudinskiy
methodology as biomarker read-out to quantify phosphorylated N, Andermann ML, et al. Neurofibrillary tangle-bearing neurons
tau in cerebrospinal fluid from Alzheimer disease patients. J are functionally integrated in cortical circuits in vivo. Proc Natl
Chromatogr A. 2021;1651:462299. Acad Sci U S A. 2014;111(1):510–4.
131. Sopko R, Golonzhka O, Arndt J, Quan C, Czerkowicz J, Cameron 149. Croft CL, Goodwin MS, Ryu DH, Lessard CB, Tejeda G, Marrero M,
A, et al. Characterization of tau binding by gosuranemab. Neu- et al. Photodynamic studies reveal rapid formation and appreciable
robiol Dis. 2020;146:105120. turnover of tau inclusions. Acta Neuropathol. 2021;141(3):359–81.
132. Dam T, Boxer AL, Golbe LI, Hoglinger GU, Morris HR, Litvan 150. Croft CL, Futch HS, Moore BD, Golde TE. Organotypic brain
I, et al. Safety and efficacy of anti-tau monoclonal antibody gos- slice cultures to model neurodegenerative proteinopathies. Mol
uranemab in progressive supranuclear palsy: a phase 2, randomized, Neurodegener. 2019;14(1):45.
placebo-controlled trial. Nat Med. 2021;27(8):1451–7. 151. Marcus JN, Schachter J. Targeting post-translational modifica-
133. Roberts M, Sevastou I, Imaizumi Y, Mistry K, Talma S, Dey tions on tau as a therapeutic strategy for Alzheimer’s disease. J
M, et al. Pre-clinical characterisation of E2814, a high-affinity Neurogenet. 2011;25(4):127–33.
antibody targeting the microtubule-binding repeat domain of tau 152. Bhat RV, Andersson U, Andersson S, Knerr L, Bauer U,
for passive immunotherapy in Alzheimer’s disease. Acta Neuro- Sundgren-Andersson AK. The conundrum of GSK3 inhibi-
pathol Commun. 2020;8(1):13. tors: is it the dawn of a new beginning? J Alzheimers Dis.
134. Albert M, Mairet-Coello G, Danis C, Lieger S, Caillierez R, 2018;64(s1):S547–54.
Carrier S, et al. Prevention of tau seeding and propagation by 153. Gong CX, Liu F, Grundke-Iqbal I, Iqbal K. Impaired brain glu-
immunotherapy with a central tau epitope antibody. Brain. cose metabolism leads to Alzheimer neurofibrillary degeneration
2019;142(6):1736–50. through a decrease in tau O-GlcNAcylation. J Alzheimers Dis.
135. Godyn J, Jonczyk J, Panek D, Malawska B. Therapeutic strate- 2006;9(1):1–12.
gies for Alzheimer’s disease in clinical trials. Pharmacol Rep. 154. Yuzwa SA, Shan X, Macauley MS, Clark T, Skorobogatko Y,
2016;68(1):127–38. Vosseller K, et al. Increasing O-GlcNAc slows neurodegenera-
136. Novak P, Zilka N, Zilkova M, Kovacech B, Skrabana R, Ondrus tion and stabilizes tau against aggregation. Nat Chem Biol.
M, et al. AADvac1, an active immunotherapy for Alzheimer’s 2012;8(4):393–9.
disease and non Alzheimer tauopathies: an overview of preclinical 155. Yuzwa SA, Shan X, Jones BA, Zhao G, Woodward ML, Li
and clinical development. J Prev Alzheimers Dis. 2019;6(1):63–9. X, et al. Pharmacological inhibition of O-GlcNAcase (OGA)
137. Novak P, Schmidt R, Kontsekova E, Zilka N, Kovacech B, Skrabana prevents cognitive decline and amyloid plaque formation in
R, et al. Safety and immunogenicity of the tau vaccine AADvac1 bigenic tau/APP mutant mice. Mol Neurodegener. 2014;9:42.

13
226 T. E. Golde

156. Yuzwa SA, Macauley MS, Heinonen JE, Shan X, Dennis RJ, 174. Betthauser TJ. AD molecular: imaging tau aggregates with
He Y, et al. A potent mechanism-inspired O-GlcNAcase inhibi- positron emissions tomography. Prog Mol Biol Transl Sci.
tor that blocks phosphorylation of tau in vivo. Nat Chem Biol. 2019;165:107–38.
2008;4(8):483–90. 175. Ossenkoppele R, Smith R, Mattsson-Carlgren N, Groot C,
157. Yu Y, Zhang L, Li X, Run X, Liang Z, Li Y, et al. Differential Leuzy A, Strandberg O, et al. Accuracy of tau positron emission
effects of an O-GlcNAcase inhibitor on tau phosphorylation. tomography as a prognostic marker in preclinical and prodromal
PLoS One. 2012;7(4):e35277. Alzheimer disease: a head-to-head comparison against amyloid
158. Wang X, Li W, Marcus J, Pearson M, Song L, Smith K, et al. positron emission tomography and magnetic resonance imaging.
MK-8719, a novel and selective O-GlcNAcase inhibitor that JAMA Neurol. 2021;78(8):961–71.
reduces the formation of pathological tau and ameliorates neu- 176. Leuzy A, Mattsson-Carlgren N, Palmqvist S, Janelidze S, Dage
rodegeneration in a mouse model of tauopathy. J Pharmacol JL, Hansson O. Blood-based biomarkers for Alzheimer’s disease.
Exp Ther. 2020;374(2):252–63. EMBO Mol Med. 2022;14(1):e14408.
159. Lim S, Haque MM, Nam G, Ryoo N, Rhim H, Kim YK. Moni- 177. Bayoumy S, Verberk IMW, den Dulk B, Hussainali Z, Zwan M,
toring of intracellular tau aggregation regulated by OGA/OGT van der Flier WM, et al. Clinical and analytical comparison of
inhibitors. Int J Mol Sci. 2015;16(9):20212–24. six Simoa assays for plasma P-tau isoforms P-tau181, P-tau217,
160. Graham DL, Gray AJ, Joyce JA, Yu D, O’Moore J, Carlson and P-tau231. Alzheimers Res Ther. 2021;13(1):198.
GA, et al. Increased O-GlcNAcylation reduces pathological 178. Thijssen EH, La Joie R, Strom A, Fonseca C, Iaccarino L, Wolf
tau without affecting its normal phosphorylation in a mouse A, et al. Plasma phosphorylated tau 217 and phosphorylated tau
model of tauopathy. Neuropharmacology. 2014;79:307–13. 181 as biomarkers in Alzheimer’s disease and frontotemporal
161. Ding N, Ping L, Shi Y, Feng L, Zheng X, Song Y, et al. Thiamet- lobar degeneration: a retrospective diagnostic performance study.
G-mediated inhibition of O-GlcNAcase sensitizes human leuke- Lancet Neurol. 2021;20(9):739–52.
mia cells to microtubule-stabilizing agent paclitaxel. Biochem 179. Kaeser SA, Hasler LM, Lambert M, Bergmann C, Bottelbergs
Biophys Res Commun. 2014;453(3):392–7. A, Theunis C, et al. CSF p-tau increase in response to Abeta-
162. Li X, Lu F, Wang JZ, Gong CX. Concurrent alterations of type and Danish-type cerebral amyloidosis and in the absence of
O-GlcNAcylation and phosphorylation of tau in mouse brains neurofibrillary tangles. Acta Neuropathol. 2022;143(2):287–90.
during fasting. Eur J Neurosci. 2006;23(8):2078–86. 180. Golde TE. Harnessing immunoproteostasis to treat neurodegen-
163. Young ZT, Mok SA, Gestwicki JE. Therapeutic strategies for erative disorders. Neuron. 2019;101(6):1003–15.
restoring tau homeostasis. Cold Spring Harb Perspect Med. 181. Sims R, van der Lee SJ, Naj AC, Bellenguez C, Badarinarayan
2018;8(1). N, Jakobsdottir J, et al. Rare coding variants in PLCG2, ABI3,
164. Shelton LB, Koren J 3rd, Blair LJ. Imbalances in the Hsp90 and TREM2 implicate microglial-mediated innate immunity in
chaperone machinery: implications for tauopathies. Front Neu- Alzheimer’s disease. Nat Genet. 2017;49(9):1373–84.
rosci. 2017;11:724. 182. Bellenguez C, Grenier-Boley B, Lambert JC. Genetics of Alz-
165. Peak SL, Gracia L, Lora G, Jinwal UK. Hsp90-interacting co- heimer’s disease: where we are, and where we are going. Curr
chaperones and their family proteins in tau regulation: introduc- Opin Neurobiol. 2020;61:40–8.
ing a novel role for Cdc37L1. Neuroscience. 2021;453:312–23. 183. Mancuso R, Fryatt G, Cleal M, Obst J, Pipi E, Monzon-Sandoval
166. Wilcock GK, Gauthier S, Frisoni GB, Jia J, Hardlund JH, Moebius J, et al. CSF1R inhibitor JNJ-40346527 attenuates microglial
HJ, et  al. Potential of low dose leuco-methylthioninium proliferation and neurodegeneration in P301S mice. Brain.
bis(hydromethanesulphonate) (LMTM) monotherapy for treat- 2019;142(10):3243–64.
ment of mild Alzheimer’s disease: cohort analysis as modified 184. Butchart J, Brook L, Hopkins V, Teeling J, Puntener U, Culliford
primary outcome in a phase III clinical trial. J Alzheimers Dis. D, et al. Etanercept in Alzheimer disease: a randomized, placebo-
2018;61(1):435–57. controlled, double-blind, phase 2 trial. Neurology. 2015;84(21):2161–8.
167. Schirmer RH, Adler H, Pickhardt M, Mandelkow E. “Lest we forget 185. MacPherson KP, Sompol P, Kannarkat GT, Chang J, Sniffen L,
you--methylene blue...”. Neurobiol Aging. 2011;32(12):2325 e7–16. Wildner ME, et al. Peripheral administration of the soluble TNF
168. DeVos SL, Miller RL, Schoch KM, Holmes BB, Kebodeaux inhibitor XPro1595 modifies brain immune cell profiles, decreases
CS, Wegener AJ, et al. Tau reduction prevents neuronal loss and beta-amyloid plaque load, and rescues impaired long-term poten-
reverses pathological tau deposition and seeding in mice with tiation in 5xFAD mice. Neurobiol Dis. 2017;102:81–95.
tauopathy. Sci Transl Med. 2017;9(374). 186. Dempsey C, Rubio Araiz A, Bryson KJ, Finucane O, Larkin
169. Shi Y, Zhang W, Yang Y, Murzin AG, Falcon B, Kotecha A, C, Mills EL, et al. Inhibiting the NLRP3 inflammasome with
et  al. Structure-based classification of tauopathies. Nature. MCC950 promotes non-phlogistic clearance of amyloid-beta
2021;598(7880):359–63. and cognitive function in APP/PS1 mice. Brain Behav Immun.
170. Goedert M. Cryo-EM structures of tau filaments from human 2017;61:306–16.
brain. Essays Biochem. 2021;65(7):949–59. 187. Heneka MT, McManus RM, Latz E. Inflammasome signalling in
171. Wang G, Coble D, McDade EM, Hassenstab J, Fagan AM, Benzinger brain function and neurodegenerative disease. Nat Rev Neurosci.
TLS, et al. Staging biomarkers in preclinical autosomal dominant 2018;19(10):610–21.
Alzheimer’s disease by estimated years to symptom onset. Alzhei- 188. Scheltens P, Prins N, Lammertsma A, Yaqub M, Gouw A, Wink
mers Dement. 2019;15(4):506–14. AM, et al. An exploratory clinical study of p38alpha kinase
172. Jack CR Jr, Knopman DS, Jagust WJ, Petersen RC, Weiner MW, inhibition in Alzheimer’s disease. Ann Clin Transl Neurol.
Aisen PS, et al. Tracking pathophysiological processes in Alz- 2018;5(4):464–73.
heimer’s disease: an updated hypothetical model of dynamic 189. Hori Y, Takeda S, Cho H, Wegmann S, Shoup TM, Takahashi K,
biomarkers. Lancet Neurol. 2013;12(2):207–16. et al. A Food and Drug Administration-approved asthma thera-
173. Gordon BA, Blazey TM, Christensen J, Dincer A, Flores S, Keefe peutic agent impacts amyloid beta in the brain in a transgenic
S, et al. Tau PET in autosomal dominant Alzheimer’s disease: model of Alzheimer disease. J Biol Chem. 2015;290(4):1966–78.
relationship with cognition, dementia and other biomarkers. 190. Decourt B, Wilson J, Ritter A, Dardis C, DiFilippo FP, Zhuang
Brain. 2019;142(4):1063–76. X, et al. MCLENA-1: a phase II clinical trial for the assessment
of safety, tolerability, and efficacy of lenalidomide in patients

13
Disease-Modifying Therapies for Alzheimer’s Disease: More Questions than Answers 227

with mild cognitive impairment due to Alzheimer’s disease. aspects of Alzheimer’s disease pathogenesis. Cell Mol Life Sci.
Open Access J Clin Trials. 2020;12:1–13. 2019;76(10):1833–63.
191. Jim HS, Boyd TD, Booth-Jones M, Pidala J, Potter H. Granulo- 197. Schindler SE, Li Y, Buckles VD, Gordon BA, Benzinger
cyte macrophage colony stimulating factor treatment is associ- TLS, Wang G, et  al. Predicting symptom onset in spo-
ated with improved cognition in cancer patients. Brain Disord radic Alzheimer disease with amyloid PET. Neurology.
Ther. 2012;1(1). 2021;97(18):e1823–34.
192. Potter H, Woodcock JH, Boyd TD, Coughlan CM, O'Shaughnessy 198. Koscik RL, Betthauser TJ, Jonaitis EM, Allison SL, Clark LR,
JR, Borges MT, et al. Safety and efficacy of sargramostim (GM- Hermann BP, et al. Amyloid duration is associated with preclini-
CSF) in the treatment of Alzheimer’s disease. Alzheimers cal cognitive decline and tau PET. Alzheimers Dement (Amst).
Dement (N Y). 2021;7(1):e12158. 2020;12(1):e12007.
193. Ulland TK, Colonna M. TREM2 - a key player in microglial biol- 199. Bateman RJ, Benzinger TL, Berry S, Clifford DB, Duggan C,
ogy and Alzheimer disease. Nat Rev Neurol. 2018;14(11):667–75. Fagan AM, et al. The DIAN-TU next generation Alzheimer’s
194. Bradshaw EM, Chibnik LB, Keenan BT, Ottoboni L, Raj T, Tang prevention trial: adaptive design and disease progression model.
A, et al. CD33 Alzheimer’s disease locus: altered monocyte func- Alzheimers Dement. 2017;13(1):8–19.
tion and amyloid biology. Nat Neurosci. 2013;16(7):848–50.
195. Beaino W, Janssen B, Vugts DJ, de Vries HE, Windhorst AD. Publisher's Note Springer Nature remains neutral with regard to
Towards PET imaging of the dynamic phenotypes of microglia. jurisdictional claims in published maps and institutional affiliations.
Clin Exp Immunol. 2021;206(3):282–300.
196. Dhiman K, Blennow K, Zetterberg H, Martins RN, Gupta VB.
Cerebrospinal fluid biomarkers for understanding multiple

13