A geroscience motivated approach to treat Alzheimer’s disease: Senolytics move to clinical trials

Abstract

The pathogenic processes driving Alzheimer’s disease (AD) are complex. An incomplete understanding of underlying disease mechanisms has presented insurmountable obstacles for developing effective disease-modifying therapies. Advanced chronological age is the greatest risk factor for developing AD. Intervening on biological aging may alter disease progression and represents a novel, complementary approach to current strategies. Toward this end, cellular senescence has emerged as a promising target. This complex stress response harbors damaged cells in a cell cycle arrested, apoptosis-resistant cell state. Senescent cells accumulate with age where they notoriously secrete molecules that contribute to chronic tissue dysfunction and disease. Thus, benefits of cell survival in a senescent fate are countered by their toxic secretome. The removal of senescent cells improves brain structure and function in rodent models at risk of developing AD, and in those with advanced Aβ and tau pathology. The present review describes the path to translating this promising treatment strategy to AD clinical trials. We review evidence for senescent cell accumulation in the human brain, considerations and strategies for senescence-targeting trials specific to AD, approaches to detect senescent brain cells in biofluids, and summarize the goals of the first senolytic trials for the treatment of AD (NCT04063124 and NCT04685590). This article is part of the Special Issue - Senolytics - Edited by Joao Passos and Diana Jurk.

1. Introduction

Cellular senescence is an elaborate stress response that involves initiating and maintaining a stable cell cycle arrest. It culminates as a change in cell fate often accompanied with a toxic proinflammatory secretory phenotype (Gorgoulis et al., 2019). Though initially defined as a finite replicative artifact of cells in culture (Hayflick and Moorhead, 1961), an appreciation for the role of senescence in driving aging and disease in vivo has emerged. Recent studies, including by our group (Musi et al., 2018), have demonstrated that postmitotic cells are vulnerable to senescence as well (von Zglinicki et al., 2021). Collectively the data indicate that mitotically competent cells and tissues, as well as those with limited regenerative potential, are vulnerable to the negative impacts of senescent cell accumulation. Specific to the brain, senescent cells contribute to poor brain aging and Alzheimer’s disease (for review, (DiBattista et al., 2020; Saez-Atienzar and Masliah, 2020). Senescence also plays a role in normal tissue repair and development, for recent reviews: (Rhinn et al., 2019; Wilkinson and Hardman, 2020). However, due to their notorious secretome, senescent cells become detrimental when they accumulate with advanced age. They secrete high levels of pro-inflammatory cytokines, chemokines, exosomes, and extracellular matrix-degrading proteins that negatively impact the extracellular environment and induce spread of senescence to non-senescent cells (Acosta et al., 2013; Coppe et al., 2010; Coppé et al., 2008; Kuilman and Peeper, 2009). This toxic secretome, referred to as senescence-associated secretory phenotype (SASP), contributes to inflammation, tissue degradation and ultimately chronic disease (He and Sharpless, 2017).

Several studies have chronicled senescent cell accumulation in the brains of rodents with aging (Jurk et al., 2012), tauopathy (Bussian et al., 2018; Musi et al., 2018), amyloid-β (Aβ) accumulation (Zhang et al., 2019) and other AD-associated risk factors (Table 1). Rodent models have the advantage of using senescent cell reporters (Bussian et al., 2018; Zhang et al., 2019), providing rapid access to fresh tissues for appropriately evaluating senescence-associated β-galactosidase (SA β-gal) (Jurk et al., 2012; Musi et al., 2018), and enabling longitudinal and cross-sectional studies of genetically identical individuals (Bussian et al., 2018; Jurk et al., 2012; Musi et al., 2018; Zhang et al., 2019), all of which are highly beneficial for identifying senescent cells in tissue. Studying senescent cells in the human brain presents obvious challenges. Identifying and tracking rare subpopulations of poorly defined cells in the ~100 billion brain cellular environment is not trivial. Archival tissue is not permissive to SA β-gal enzymatic assays (Debacq-Chainiaux et al., 2009) and its correlation to brain cell senescence is not clear given that some neuronal subpopulations express SA β-gal activity at young ages (Jurk et al., 2012; Musi et al., 2018). Similarly, lipofuscin, lysosomal biproducts indicative of senescence in some cell types, occurs even in young neurons in the human brain (Goyal, 1982). Moreover, the identity of the parent cell type and upstream signals may affect the post-senescence phenotype (Hernandez-Segura et al., 2017a). The resulting heterogeneity has presented challenges for identifying, defining, and studying senescent cells in vivo. While this is especially true for post-mitotic tissues such as the brain, much progress has been made in recent years, for review: (Gillispie et al., 2021; Sah et al., 2021).

Table 1.

Preclinical studies investigating cellular senescence in relation to AD risk factors and AD neuropathology.

Citation	Aging/Neurodegenerative Disease Model	Senotherapeutic Treatment and Dosing Strategy	Senescent Cell Types Investigated	Brief Summary of Outcomes
(Streit et al., 2009)	Human postmortem AD brain tissue	None	Microglia	Morphologically senescent-like, not activated, microglia were found near Aβ plagues and preceded tau tangles in human brains.
(Geng et al., 2010)	Male Sprague-Dawley rats (6-, 18-, 24-month old) & in vitro hippocampal neurons	None	Neurons	SA β-gal staining increased in the CA3, but not DG of the hippocampus with aging (51 % by 24 months)
(Bhat et al., 2012)	Postmortem human brain & in vitro astrocytes	None	Astrocytes	p16 positive astrocytes were increased with age and in AD brain tissue.
(Jurk et al., 2012)	Male C57Bl/6 mice (4- & 32-month old)	None	Neurons	Age and telomerase-related senescent phenotypes in neurons were mediated by the DNA damage response, slightly attenuated by caloric restriction, and fully rescued by the knockout of Cdkn1a.
(He et al., 2013)	Male C57BL/6 and APP/PS1 (9-month-old)	None	Neural stem cells	Aβ42 exposure increased senescent phenotypes in NSCs in the dentate gyrus of APP/PS1 mice.
(Al-Mashhadi et al., 2015)	Postmortem human brains with or without white matter lesions	None	Oligodendrocytes	Senescent phenotypes were identified in oligodendrocytes of aged brains, but were not associated with white matter lesions.
(Kang et al., 2015)	Postmortem human brain tissue from young and aged adults without neuropathology	None	Astrocytes; neurons; oligodendrocytes	Gata4, a SASP regulator, was highly associated with age and p16 expression in astrocytes, pyramidal neurons, and oligodendrocytes in humans.
(Turnquist et al., 2016)	Postmortem brain tissues from AD and ALS patients	None	Astrocytes	Greater senescent cell burden in AD and ALS tissue compared to non-diseased, age-matched and pediatric control tissue
(Bussian et al., 2018)	Male and female PS19;ATTAC mice (weanling to 12-month-old)	2 mg/kg AP20187 twice weekly from weaning (long term) or 10 mg/kg for five consecutive days at 6 months of age (short-term)	Astrocytes; microglia	Clearance of senescent cells prevented gliosis, hyperphosphorylation of tau, and neuronal degeneration
		50 mg/kg ABT263 for 5 consecutive days followed by 16 days of rest from weaning until 6 months
(Musi et al., 2018)	Male and female rTg4510Mapt+/+, rTg4510Mapt0/0, rTg21221 Mapt+/+, 3xTgAD mice (2- to 23-month-old)	5 mg/kg dasatinib and 50 mg/kg quercetin for 5 consecutive days followed by 9 days rest for 6 treatment cycles	Neurons	Dasatinib + quercetin treatment reduced total NFT density, SASP gene expression, neuron loss, and ventricular enlargement.
	Postmortem human brain: Alzheimer’s disease (AD) Progressive Supranuclear Palsy (PSP)	None	Neurons	AD: Elevated senescent gene expression in NFT-bearing neurons compared to neurons without NFTs.
				PSP: Elevated CDKN2D expression compared to age-matched controls that negatively correlated with cognition.
(Chow et al., 2019)	Male and female C57BL/6 J mice (3- and 22–24 month old) +/− insulin resistance	None	Neurons	Chronic insulin exposure led to senescence phenotype in neurons via dysregulation of hexokinase 2
(Moreno-Blas et al., 2019)	Male Wistar rat brains (4- or 25-month-old)	None	Astrocytes; neurons	Astrocytes and neurons showed a wide range of senescent phenotypes, which were attenuated or exacerbated in vitro via stimulation and inhibition of autophagy, respectively.
(Norton et al., 2019)	Postmortem human brain without AD and with or without cerebral small vessel disease	None	Mural cells	Senescent phenotypes were observed in mural cells of aged individuals, but were less prevalent in patients with cerebral small vessel disease.
(Ogrodnik et al., 2019)	Male C57BL/6 INK-ATTAC mice (8-month-old) +/− high fat diet; INK-ATTAC; db/db mice	10 mg/kg AP20187 for 3 days every 2weeks for a total of 8–10 weeks	Astrocytes; microglia	Clearing obesity-induced senescent glial cells improved neurogenesis and alleviated anxiety-related behavior.
	db/db mice	5 mg/kg dasatinib and 50 mg/kg quercetin for 5 days every 2 weeks for 8 weeks
(Ritzel et al., 2019)	Male C57BL/6 mice (3- & 18-month-old) +/− TBI	None	Microglia	Aging and TBI increased microglia senescent phenotypes including Bcl-2, p16INK4a, p21CIP1a, lipofuscin, and H2AX.
(Schwab et al., 2019)	Postmortem human brains with or without a history of TBI	None	Astrocytes; ependymal cells; neurons; oligodendrocytes	Brains from patients with a history of TBI showed extensive dysfunction, DNA damage, and other cellular senescence phenotypes in a variety of cell types.
(Stojiljkovic et al., 2019)	Male C57BL/6J mice (3- & 24-month-old)	None	Microglia	Various markers of microglial senescence distinct from microglial activation increased with age, although aged microglia in vivo may have been quiescent rather than senescent.
(Tominaga et al., 2019)	Male C57BL/6 mice (10-weeks-old) +/− TBI	None	Astrocytes; microglia; neurons	Cell cycle initiation increased for the first four days post-TBI, while markers of senescence were elevated on days 4, 7, and 14 with some cell type-specific differences in expression of senescence.
(Turnquist et al., 2019)	Postmortem human brain tissue from patients that had received radiation and age-matched controls	None	Astrocytes	Astrocyte senescence was the most prominent cell type in radiation exposed human tissues.
(Zhang et al., 2019)	Male and female APP/PS1, WT mice and APP/PS1/ZsGreen mice (2.5- to 8-month-old)	12 mg/kg dasatinib and 50 mg/kg quercetin, once daily for 9 days (short-term) or 11 weeks (long-term)	Oligodendrocyte progenitor cells	Aβ induced senescence in oligodendrocyte progenitor cells and senolytic treatment reduced neuroinflammation, Aβ burden, and cognitive deficits.
( Arun et al., 2020 )	Male Sprague-Dawley rats (9–10-week-old to 1-year +/−TBI)	None	Did not investigate	Single and repeated blast exposures increased markers of cellular senescence, particularly at 1 month, in several neuroanatomical structures.
(Bryant et al., 2020)	Intact microvessels from postmortem human AD brain tissue	None	Endothelial cells	Senescence-related genes were significantly upregulated in endothelial cells.
( Gao et al., 2020 )	Neural stem cell spheroids from 3-week-old and 23-month-old C57BL/6 mice	100 pg/mL Ribes meyeri anthocyanins extract	Neural stem cells (NSCs)	NSCs of aged mice showed increased senescence phenotypes and treatment with Ribes meyeri anthocyanins or Nar reduced senescence and improved cognition.
	Female C57BL/6 mice (12-mo-old)	100 mg/kg Ribes meyeri anthocyanins extract for 2 months or 20 mg/kg Nar for 1 month
( Hu et al., 2021 )	Male SAMP8 mice (2-, 6-, 12-month-old)	1010 embryonic stem cell-derived extracellular vesicles (ESC-sEVs) twice weekly for 6 months	NSCs	ESC-sEV treatment alleviated senescence in hippocampal NSCs and improved age-related declines in cognition.
( Ogrodnik et al., 2021 )	Male INK-ATTAC and WT mice (4- & 24-month-old)	10 mg/kg AP20187 for 3 consecutive days once every two weeks for 8 weeks	Microglia; oligodendrocyte progenitor cells	Senolysis reduced p16INK4a expression, microglial activation, and SASP factors and improved cognitive function.
		5 mg/kg dasatinib and 10 mg/kg quercetin for 3 consecutive days once every two weeks for 8 weeks
( Schwab et al., 2021 )	Male C57BL/6 mice (7–9-week-old) +/− repeated mild TBI	None	Did not investigate	Following an initial DNA damage response at 24 h, senescent cell phenotypes were observed in the brain 7 days after traumatic brain injury.
( Yousefzadeh et al., 2021a )	Male and female mice with hematopoietic cell Ercc1 knockout (2- to 24-months old)	None	Did not investigate	Whole-body immune senescence increased p16 expression in the brain, but showed no differences in p21 expression or age-associated lesions via H&E staining.

Abbreviations: amyloid-beta (Aβ); Alzheimer’s disease (AD); Amyotrophic lateral sclerosis (ALS); embryonic stem cell derived small embryonic vesicles (ESC-sEV); lipopolysaccharide (LPS); neural stem cell (NSC); neurofibrillary tangle (NFT); senescence-associated β-galactosidase (SA-βgal); senescence-associated heterochromatin foci (SAHF); senescence-associated secretory phenotype (SASP); traumatic brain injury (TBI).

Rodent models and postmortem human brain tissue have been invaluable in establishing the preclinical evidence necessary to begin translating senescence targeting strategies to the clinic (Tchkonia and Kirkland, 2018). Specifically, our group revealed an upregulation of genes and SASP factors in brain regions with tau-containing neurofibrillary tangles (NFTs) in four distinct transgenic mouse models of tau-pathogenesis and post-mortem human brain tissue (Musi et al., 2018). Transcriptomic comparisons between neurons with or without NFTs from postmortem human brains revealed a senescence signature in the NFT-bearing neurons. A causal role of NFT-mediated senescence was explored with pharmacological ablation of senescent cells in tau transgenic mice. The intermittent treatment with senolytic agents, dasatinib and quercetin (D + Q), selectively cleared senescent cells (Musi et al., 2018) which was independently reproduced in other models of brain aging (Ogrodnik et al., 2021) and pathogenesis relevant to AD (Ogrodnik et al., 2019; Zhang et al., 2019). Across studies, clearing senescent cells has been associated with diminished AD pathology (i.e., tau and Aβ deposition), reduced SASP and inflammation, decreased white matter hyperintensities (WMH), restored cerebral blood flow, and improved cognitive behavior (Bussian et al., 2018; Musi et al., 2018; Ogrodnik et al., 2019; Zhang et al., 2019). With these promising data, similar strategies are moving toward clinical testing in the context of amnestic mild cognitive impairment/early AD (NCT04063124 and NCT04685590). Overall, the present contribution aims to provide an accessible, updated summary of evidence that supports translating this therapeutic approach to older adults with amnestic MCI or early AD. Considerations unique to trials in AD, including potential challenges faced are also discussed.

2. Translating preclinical evidence to clinical trials

2.1. Evidence for senescent cell accumulation in the human brain with aging and neurodegeneration

Cell cycle activity in postmitotic brain cells in human AD was first reported, to our knowledge, in the late 1990s (Arendt et al., 1996; Luth et al., 2000; McShea et al., 1997).The elevated expression of cell cycle regulators, including senescence hallmark p16, was hypothesized to represent aberrant cell cycle activity leading to apoptosis in neurodegenerative diseases. Later work revealed that elevated p16 protected against, not induced, neuronal cell death (Park et al., 1998). While these early studies were investigating cell cycle activity, not senescence, several groups have since performed hypothesis-driven experiments focused on the role of cellular senescence in the brain (Table 1). Senescent cells indisputably accumulate in the brain, however the consequences of their accumulation and long-term survival in the context of human aging and disease remains poorly understood. In this section, we review evidence for human brain cellular senescence, and discuss whether the response represents a physiological benefit (i.e., alternative to cell death), or instead may contribute to chronic inflammation and degeneration as they are notoriously known.

2.1.1. Senescent cell accumulation associated with brain pathologies of advanced age

The privilege of growing old often comes with a cost of failing health. Common brain pathologies in non-demented older adults include vascular and white matter lesions (WML) (de Leeuw et al., 2001; Han et al., 2018) and cerebral small vessel disease (cSVD; alternatively termed arteriolosclerosis). In older adults, the prevalence of WML is >80 % (Longstreth et al., 1996); lesions associated with cSVD in older populations ranges from 8 % to 33 % for lacunes (Vermeer et al., 2007), 3 % to 35.7 % for cerebral microbleeds (Greenberg et al., 2009; Hilal et al., 2017; Poels et al., 2010), and 39 % to 96 % for WMHs, as reviewed in other studies (Prins and Scheltens, 2015). While vascular lesions and WML do not always co-occur, longitudinal magnetic resonance imaging (MRI) data suggest that small-vessel ischemic disease could contribute to white matter changes in advanced age (Erten-Lyons et al., 2013). Emerging evidence indicates cellular senescence may be involved with these pathologies, and in some cases, connect them.

Tissue-wide oxidative stress and DNA damage are more likely to be associated with WML in the presence of senescent cells (Al-Mashhadi et al., 2015). A neuropathological hallmark of cSVD is fibro-hyaline thickening of deep penetrating arteries resulting in elongated and tortuous vessels with narrowed lumina and thickened vessel walls. In rodent studies, this phenotype has been observed near NFT-bearing neurons (Bennett et al., 2018), cells which display features of cellular senescence (Musi et al., 2018). A follow-up study led by Bennett using postmortem human brain tissue from patients with AD found an association between cerebrovascular senescence and tau pathology (Bryant et al., 2020) (discussed below). While these data suggest that cellular senescence may contribute to cSVD in individuals with dementia, a separate study that used tissue from non-demented individuals found an inverse correlation between senescent myocytes and cSVD diagnosis. Specifically, Norton et al. investigated the prevalence of myocyte senescence in 60 adults, aged 80–96 years-old, with minimal AD pathology (i.e., Braak stage 0-II), and its association with neuropathological diagnosis of cSVD (Norton et al., 2019). Their study relied on histological analyses using antibodies to detect smooth-muscle myocytes, nuclear trimethylated Lys9 in histone-3 (H3K9me3) staining as a surrogate for senescence-associated heterochromatic foci (SAHF) and phosphorylated Ser139 on histone-2 (γH2AX) to detect DNA damage foci. The data indicated more abundant H3K9me3 and γH2AX positive myocytes in individuals without cSVD than those with a confirmed neuropathological diagnosis of cSVD. Total mural cell density (e.g., vascular smooth muscle cells and pericytes) was also higher in control cases, which is consistent with the depletion of vascular myocytes in cSVD (Pantoni, 2010). Since the loss of myocytes is closely associated with impaired cerebral blood flow and consequent ischemic lesions, it is tempting to speculate that retaining myocytes, even if senescent, may be preferred over their apoptosis. Similarly, given that inflammatory markers of the SASP were not evaluated in the aforementioned studies, and that not all senescent cells develop SASP (Coppe et al., 2011), the toxicity profile of senescent brain cells remains unknown. In this way, senescence may play a protective role for retaining brain vascular myocytes, especially if they do not adopt a toxic secretome. The intriguing inverse relationship between senescent myocytes and cSVD diagnosis warrants further consideration as senescent cell clearing therapies move to clinical trials.

2.1.2. Senescent cells and Alzheimer’s disease-associated neuropathologies

AD is the most common cause of dementia, and is responsible for 60–80 % of all known cases (2021a). The disease defining features include the accumulation of extracellular Aβ in the form of plaques and intracellular hyperphosphorylated tau forming NFTs (Jack et al., 2018). The pathogenic mechanisms connecting Aβ and tau protein accumulation with neurodegeneration and dementia, however, remain unclear. Many promising lines of investigation are underway, including a focus on the drivers of biological age, like cellular senescence. The formal investigation of senescence in the context of AD is a relatively recent field (DiBattista et al., 2020). However, as mentioned above, reports of elevated expression of cyclin dependent kinase (CDK) inhibitors p16 and p21, which are hallmarks of the senescence stress response, were found co-localized to NFT-bearing neurons in postmortem AD brain decades ago (Luth et al., 2000; McShea et al., 1997). Aberrant neuronal cell cycle re-entry was hypothesized to contribute to the slow death of neurons (Barrio-Alonso et al., 2018; Yang et al., 2001, 2003); however, recent data indicates this atypical neuronal phenotype resembles cellular senescence (Chow et al., 2019; Musi et al., 2018). The body of evidence now suggests that postmitotic cells are susceptible to becoming senescent. Neurons, for example, exist in quiescence. This reversable dormant cell state protects cells against stress but renders them vulnerable to enter senescence through lysosome-mediated mechanisms (Fujimaki et al., 2019). In this way it is tempting to speculate that neurons are primed for senescence given their quiescent state and susceptibility to age-associated lysosomal dysfunction (von Zglinicki et al., 2021). While it remains to be determined whether neurons slip directly into senescence, or abort cell-cycle re-entry induced apoptosis (Raina et al., 2003), brain cell senescence has emerged as a cell stress response to risk factors known to increase AD, including advanced age, Aβ, tau and traumatic brain injury (Table 1). Hypothesis-driven studies to investigate cellular senescence in postmortem human brains have revealed that multiple brain cell types may engage a senescence stress response (Bhat et al., 2012; Bryant et al., 2020; Musi et al., 2018; Zhang et al., 2019). Mechanistic experiments in mice have demonstrated that tau aggregation drives cellular senescence in neurons (Musi et al., 2018) and microglia (Bussian et al., 2018), whereas extracellular Aβ has a similar effect on oligodendrocyte precursor cells (Zhang et al., 2019). Similarly, in postmortem human AD brain, neurons (Musi et al., 2018), astrocytes (Bhat et al., 2012), oligodendrocyte precursor cells (Zhang et al., 2019), endothelial cells (Bryant et al., 2020) and microglia (Streit and Xue, 2014) display features of cellular senescence. The relative sensitivity of each of these cell types to senescence-targeted therapies, and outcomes of their removal remain to be determined in clinical trials.

2.1.3. Mixed neuropathology in AD

While neurodegenerative diseases have historically been characterized by disease-specific pathologies, recent postmortem analyses have firmly established the co-occurrence of pathologies in multiple dementia syndromes (Rabinovici et al., 2017). In AD, the most common Aβ plaque and NFT co-pathologies are Lewy bodies, TDP-43 and cerebrovascular lesions. Lewy bodies are primarily composed of α-synuclein and can co-exist with prototypical AD pathology (Rabinovici et al., 2017). Parkinson’s disease (PD), the second most common neurodegenerative disease following AD, is characterized by the presence of Lewy bodies, and is associated with progressive motor symptoms, static tremor, motor imbalance, and muscle rigidity (Zeng et al., 2018). PD brain tissue displays elevated levels of senescence markers p16^INK4a, and SASP factors including IL-6, IL-1α, IL-8, and protease matrix metalloprotein-3 (MMP-3) (Chinta et al., 2018). Additionally, astrocytes in postmortem PD brain tissue were found to be lamin B1 deficient. Given that lamin B1 decline is induced by p53 activation or p16 expression and is closely associated with senescence (Freund et al., 2012), the reduction in lamin B1 in PD is suggestive of senescence (Chinta et al., 2018). The expression of nuclear lamin proteins in the brain is complex (Mahajani et al., 2017; Takamori et al., 2018; Young et al., 2014); we caution the need for careful consideration when attributing a senescence phenotype to lamin B1 expression level. Of note, senescent astrocytes have been reported in postmortem human AD, but neither lamin B1 nor Lewy body pathologies were evaluated (Bhat et al., 2012). Future studies on similarities or differences in the senescence profile in pure AD versus mixed pathologies (including others not discussed here, i.e., TDP-43) will be important for predicting outcomes in senescence-motivated therapies in future trials.

2.1.4. Partial AD pathology - pure tauopathies in the absence of Aβ

Tauopathies, including AD, are a group of neurodegenerative disorders pathologically defined by the deposition of tau. However, unlike mixed AD with multiple co-pathologies, pure tauopathies involve tau accumulation in the absence of Aβ. These diseases include progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), some frontotemporal dementias, and chronic traumatic encephalopathy among others (Orr et al., 2017). NFT-containing neurons from postmortem human AD brains exhibit a senescence-like phenotype (McShea et al., 1997; Musi et al., 2018). Moreover, tauopathy, even in the absence of Aβ, drives cellular senescence as evidenced by elevated CDKN2A in post-mortem human brain tissue with histopathologically confirmed PSP (Musi et al., 2018). To our knowledge, studies have not evaluated senescence in human brains with elevated Aβ in the absence of tau (i.e., cerebral amyloid angiopathy, CAA). These data will be critical to understanding whether patients with CAA may also benefit from therapies targeting cellular senescence, or whether the presence of tau pathology is a key feature in driving senescence in these diseases. Nevertheless, studies involving AD neuropathology [pure, mixed, or partial (i.e., tau)] all have reported a co-occurrence of senescent cells indicating a common stress response across many distinct neurodegenerative diseases.

2.2. Key considerations for translating senescence clearing therapies to AD trials

Prior AD trials have largely focused on restoring neurotransmitter levels (i.e., acetylcholine) or reducing AD protein accumulation (e.g. cerebral amyloid beta and tau). Despite strong rationale for these therapeutic targets, the trials have failed. Between 2002–2012 there was a 99.6 % failure rate of AD-drug trials (Cummings et al., 2014), and this has not significantly improved in the last decade (Rafii and Aisen, 2020). Targeting a fundamental cellular aging process, like cellular senescence, offers a fresh, distinct approach and is backed by compelling evidence. Advanced age is the strongest risk factor for AD, and data indicate a continuum of neurodegenerative changes from normal aging to dementia (Harada et al., 2013). These include brain atrophy, synaptic loss, and neuropathological protein accumulation. As the key processes of aging interact across molecular, cellular, and system levels, targeting one pathway (e.g., cellular senescence) may exert pleiotropic effects that are further beneficial for the treatment of AD (Mattson and Arumugam, 2018).

Removing senescent cells improves clinically relevant outcomes in AD mouse models Table 1, i.e., (Bussian et al., 2018; Musi et al., 2018; Zhang et al., 2019). Over 40 agents have been identified as ‘senolytic’, or capable of ablating senescent cells (Kirkland and Tchkonia, 2020; Zhu et al., 2015). A multitude of pharmacological agents have also been discovered that modulate broader key pathways relevant to senescence (Mongelli et al., 2020). Clinical trials using senolytics have been completed or are ongoing for several chronic diseases including renal insufficiency (NCT02848131), idiopathic pulmonary fibrosis (NCT0287498), and frailty (NCT03675724) (Hickson et al., 2019; Justice et al., 2019a; Kirkland and Tchkonia, 2020). Although still in early stages, clinical trials aimed at exploring cellular senescence as an innovative target for AD are imminent (NCT04063124 and NCT04685590). In the following sections, we review key critical considerations when translating promising senescence-modulating therapies to AD clinical trials that are unique to this study population, and distinct from trials focused on older adults with normal cognition.

2.2.1. General considerations for AD trials

2.2.1.1. Study partner.

Unique to AD trials, compared to other older adult study populations without dementia, is that all AD trials require two participants: the patient and a study partner. The study partner, typically the primary caregiver (e.g., spouse, adult child or other), is critical to a successful trial (Bernstein et al., 2021; Grill et al., 2013). They assist with informed consent, ensure protocol compliance, and serve as informants for cognitive, functional, and behavioral outcome measures.

2.2.1.2. Blinding and placebo effect.

Double-blinding and controlling for placebo effects are particularly critical to AD trials. Changes in cognition and biomarkers are often difficult to isolate from control groups due to limited statistical power, particularly in short trials.

2.2.1.3. Study duration.

Cognitive decline in placebo groups typically requires an 18-month period to observe (Ito et al., 2013), which is the basis for typical phase II and III AD trial duration (Green et al., 2009; Schneider et al., 2015; Schneider and Sano, 2009). Placebo lead-in of appropriate length can help identify placebo responders during the first couple weeks of the study, which could minimize trial failures (Ozawa et al., 2017). Target engagement for senolytics may be assessed with shorter trial durations but would require robust measures of senescent cell burden in the brain; such methodologies are described in a later section.

2.2.1.4. Medication considerations: interactions and stable dosing.

It is common for AD patients to suffer with co-occurring behavioral and psychological disturbance, which may be treated with antipsychotic and antidepressant medications often with more severe drug interaction profiles (Kettunen et al., 2019; Kuroda et al., 2019). Many medications, specifically commonly prescribed anti-depressants, are inhibitors or inducers of CYP450, and are often prescribed to AD patients who commonly have behavioral disturbance (30–90 %) (Garre-Olmo et al., 2010; Müller-Spahn, 2003) and depression (> 40 %) (Panza et al., 2010; Ryu et al., 2017). In considering the primary therapeutics prescribed for symptomologic maintenance in AD, there are a few specific considerations to make regarding co-administered medications. The anticholinesterase inhibitors donepezil, and galantamine (not true for rivastigmine) (Grossberg et al., 2000) are primarily metabolized by cytochrome P450 isoenzymes 34A (CYP3A4) and 2D6 (CYP2D6) (Farlow, 2003), and can interact with other anticholinesterase agents to produce synergistic effects. All three drugs have similar cholinergic effects that can increase risk of bradycardia, especially when paired with beta-blockers (2012; Grossberg et al., 2000; Tavassoli et al., 2007). Memantine, as a N-methyl-D-aspartate (NMDA) receptor antagonist, is unique amongst the other commonly prescribed medications for the symptomology of AD. Memantine inhibits glutamatergic signaling reducing excitotoxicity in AD that may lead to destruction of cholinergic neurons, and may even promote synaptic plasticity and preserve or enhance memory, but has not had long-term effectiveness in preventing disease progression (Paulison and Léos, 2010; Rogawski and Wenk, 2003). Memantine has an inhibitory effect on Cytochrome P450 (specifically of isoenzyme CYP2B6 and CYP2D6), and can also result in significant drug interactions (Micuda et al., 2004). Prior to enrolling a subject into an AD trial, participants should be maintained on a stable dose of any AD-specific medication for a minimum of 3 months. This will ensure that any cognitive changes, alternative study measures, or adverse events are due to the new investigational product (i.e., senolytic) rather than changes in prescription or dose of other medications (Tariot et al., 2004; Xu et al., 2021). Aside from the commonly prescribed medications to treat AD symptoms, patients are often also treated with a host of medications for co-occurring diseases and disorders (Chatterjee and Mudher, 2018; Janson et al., 2004; Santiago and Potashkin, 2021). Some of these medications have been suggested to have positive effects in AD (e.g., metformin (Koenig et al., 2017a) or antihypertensives (Group et al., 2019)) while others may have extensive side-effect profiles and potential for significant drug interactions (Ali et al., 2015; Ding et al., 2020; Pasqualetti et al., 2015). Therefore, especially in a population that may struggle with drug compliance and reporting of adverse events, physicians and research teams need to be aware of the increased risk of potential adverse drug interactions (Borah et al., 2010; Lima et al., 2016).

2.2.2. Determining target engagement

One existing barrier for cellular senescence trials is the lack of specific, reliable, accessible biomarkers to establish target engagement and monitor efficacy. Many ongoing efforts are focused on innovative approaches to identify senescent cells in human brain tissue and monitor target engagement in these trials. Newly emerging strategies aim to identify senescent-specific signatures by analyzing miRNAs, epigenetics (Horvath and Raj, 2018; Kritsilis et al., 2018) and transcriptomics (Hernandez-Segura et al., 2017b) that regulate gene expression and cellular functions integral to senescence (Selbach et al., 2008). Comparing transcriptomic networks across cultured fibroblasts, melanocytes, keratinocytes, and astrocytes exposed to various senescence-inducing stressors revealed cell type and stress type differences (Hernandez-Segura et al., 2017b). Only 4 % of senescence-associated genes were consistently differentially expressed across cell types; moreover, among 1311 differentially expressed genes in senescent fibroblasts, variances were noted across specific cellular stressors (replicative, oncogene-induced, and ionizing-induced). Results from this study highlight that senescence transcriptomes are unique to cell type and stressor (Hernandez-Segura et al., 2017b). Given that senescence signatures are expected to differ across tissues and cells, a concerted effort focused on characterizing senescent cells across the human body is greatly needed to advance clinical trials (Roy et al., 2020).

2.2.3. Biofluid markers of senescence to determine eligibility and monitor target engagement

In AD brain tissue, we find that 1–3 % of brain cells are senescent (Dehkordi et al., 2021). These results overlap with the prevalence of senescent cells reported in the blood from patients with idiopathic pulmonary fibrosis (IPF) (Justice et al., 2019b) and in adipose and skin (Hickson et al., 2019; Justice et al., 2018) to suggest a similar senescent cell burden across human tissues. Despite the limited number of senescent cells within a tissue, the diffuse paracrine transmission of the SASP induces broad changes in gene expression that may be detectable in biofluids distal from the site of origin. Successfully detecting their presence would then rely on senescence-specific molecules/phenotypes to target for analyses. Toward this end, much work has focused on the inflammatory response (Hickson et al., 2019; Schafer et al., 2020). While the specificity of these molecules to senescent brain cells remains unclear, this strategy was employed in a clinical trial of IPF where targeted gene expression analyses identified core SASP factors that were responsive to oral senolytic therapy (Justice et al., 2019b). The reduction in pro-inflammatory SASP factors corresponded to improvements in clinical outcomes. Notably, these core SASP factors are detectable in blood and/or CSF of patients with AD (Table 2) suggesting that SASPs may be useful to monitor senescence in AD as well. While the IPF study provides strong evidence for the utility of senolytic guided therapies in chronic conditions, many of the SASP factors measured co-occur with numerous inflammatory responses and highlights the need for continued biomarker discovery with higher specificity to senescence.

Table 2.

Core SASP in Human AD.

In CSF and/or Blood
IL6 (Blum-Degen et al., 1995; Brosseron et al., 2014; Lai et al., 2017; Swardfager et al., 2010)
IL9 (Taipa et al., 2019)
IL1α (Italiani et al., 2018)
IL2 (Blum-Degen et al., 1995; Brosseron et al., 2014)
IL1β (Blum-Degen et al., 1995)
MMP2 (Abe et al., 2020; Lorenzl et al., 2008)
MMP9 (Lorenzl et al., 2003, 2008)
IL-1RA (Brosseron et al., 2014; Italiani et al., 2018)
GM-CSF (Taipa et al., 2019)
FGF2 (Blasko et al., 2006; Hanneken et al., 1995; Tsai et al., 2018)
Exosomes (Burgos et al., 2014; Fiandaca et al., 2015)
miRNAs (Burgos et al., 2014; Fiandaca et al., 2015; Kiko et al., 2014; Kumar et al., 2013; Tan et al., 2014)

We also provide caution that the presence of SASP factors detected in the CSF may originate from peripheral sources. Though the brain is considered to be relatively immune-privileged, peripheral factors are able to cross the blood brain barrier through bidirectional transport systems, which may lead to senescence in the brain. This principle is demonstrated by findings of peripheral inflammation inducing transient alterations of mental status in the case of delirium (Lopez-Rodriguez et al., 2021; Munster et al., 2011), and even heterochronic parabiosis studies in mice (Villeda et al., 2011; Yousefzadeh et al., 2020). Mouse studies provide further evidence that senescent peripheral immune cells are able to increase markers of senescence in the central nervous system (Yousefzadeh et al., 2021b). Further, as blood brain barrier integrity declines with age, and to a greater extent with neurodegenerative disease (Farrall and Wardlaw, 2009; Montagne et al., 2015), considerations should be made to evaluate effects of target engagement of new therapeutics in the periphery versus the brain.

Exosomes are membrane-bound small extracellular vesicles (EVs) of endocytic origin with a size range of ~30–50 nm. These nanovesicles are secreted by cell types for intercellular communication, cellular homeostasis, stress response, and to improve cell survival (Kumar and Deep, 2020a, b; Misawa et al., 2020; Takahashi et al., 2017). They are being explored as potential biomarkers for the detection of neurodegenerative disease and senescence. Exosome biogenesis involves numerous steps, including a) the formation of clathrin-coated vesicles (CCV) from specific sites in the plasma membrane, b) fusion of CCV with the endosome, c) inward budding of the membrane of endosomes, and cargo (protein, nucleic acids, lipids, glycans, and metabolites, etc.) loading, leading to the formation of multi-vesicular endosomes (MVE) or multi-vesicular bodies (MVB) with intraluminal vesicles, e) MVE/MVB move along specific cytoskeleton track towards the plasma membrane, and lastly, f) MVE/MVB fuse with the plasma membrane and exosomes are released outside. There is a growing list of molecular regulators of exosome biogenesis and cargo loading, including the endosomal sorting complex required for transport (ESCRT), Rab GTPases, Alix, ceramides, tetraspanins, sphingomyelinases, HIF1α, syndecan, syntaxin, and syntenin etc (Colombo et al., 2013, 2014; Kowal et al., 2014; Kumar and Deep, 2020a, b; Peak et al., 2019; Ventimiglia and Alonso, 2016). Besides exosomes, cells secrete several other types of EVs, including microvesicles (~100 nm to >1000 nm in size), apoptotic bodies, and oncosomes (Jakhar and Crasta, 2019), underscoring the complexity and heterogeneity of EVs.

Exosomes have a lipid bilayer that protects their cargo from degradation, making these vesicles a useful tool for local and distant communication. These vesicles are found in all biofluids and exhibit certain common proteins, even in exosomes from different cellular origin. However, exosomes are also loaded with certain unique cargoes, including proteins, miRNAs, and metabolites, that could potentially relate to the cellular source of their origin and their molecular, physiologic, and metabolic state (de Jong et al., 2012; Eissa et al., 2016; Kumar and Deep, 2020a, b; Ramteke et al., 2015; Schlaepfer et al., 2015). Several studies have reported that exosomes released by cells under stressful physiological conditions or pathologic states have certain unique molecular features compared to those secreted under normal conditions (Chistiakov et al., 2016; de Jong et al., 2012; Hedlund et al., 2011; Kumar and Deep, 2020a, b; Panigrahi et al., 2018; Ramteke et al., 2015; Sheller et al., 2016). For example, under hypoxia (low oxygen condition), cells usually secrete higher amounts of exosomes loaded with relatively unique cargo and could transmit hypoxic phenotypes to recipient cells (Jung et al., 2017; Kumar and Deep, 2020a, b; Panigrahi et al., 2019; Ramteke et al., 2015; Zonneveld et al., 2019). Exosomes released under oxidative stress are loaded with certain unique proteins reflecting a pro-oxidative physiological state of the parent cells (de Jong et al., 2012; Eissa et al., 2016; Hedlund et al., 2011). Broadly, under certain conditions, the composition of exosomes could be a snapshot of the cell at the time of their biogenesis and secretion. Due to relative ease with which they can be obtained, less invasive access in various biofluids compared to tissue biopsies, and their unique cargo reflecting the pathophysiological state, exosomes have been extensively characterized for the diagnosis and prognosis of various pathologies including cancer, AD, and PD (Asai et al., 2015; Chistiakov et al., 2016; Eissa et al., 2016; Kalluri, 2016; Kumar and Deep, 2020a, b; Panigrahi and Deep, 2017; Soria et al., 2017; Winston et al., 2016).

Numerous studies have now established that exosomes secreted by various brain cells (e.g., neurons, astrocytes, and endothelial, etc) could be present in biofluids (blood, CSF, saliva, and urine), and could be useful to better understand molecular underpinnings of neurological disorders and neurodegenerative diseases including AD (Cha et al., 2019; Fiandaca et al., 2015; Goetzl, 2020; Goetzl et al., 2018, 2015; Goetzl et al., 2016a, b; Kapogiannis, 2020; Kapogiannis et al., 2015; Mustapic et al., 2019; Nogueras-Ortiz et al., 2020; Vella et al., 2016; Winston et al., 2016). Specific brain cell-derived exosomes could be isolated from blood plasma based upon their unique surface proteins and characterized for established AD biomarkers, and used to discover novel biomarkers. Fiandaca et al. reported that the level of established AD biomarkers (p-T181-tau, p-S396-tau, and Aβ42) in neuron-derived exosome (NDE), isolated from blood plasma, could predict the development of AD up to 10 years before the clinical onset (Fiandaca et al., 2015). Similarly, abnormal plasma NDE levels of p-Tau, Aβ42, neurogranin, and repressor element 1-silencing transcription factor precisely predicted the MCI conversion to AD (Winston et al., 2016). Goetzl et al. reported that astrocyte-derived exosomes (ADE) from AD patients exhibited higher levels of BACE-1 (β-site amyloid precursor protein-cleaving enzyme 1) and sAPPβ (soluble amyloid precursor protein β), and lower GDNF (glial-derived neurotrophic factor) compared to ADE from healthy controls. Few recent studies have also reported the usefulness of endothelial-derived exosomes (EDE) in better understanding cerebral vascular disease and cognitive function. For example, Abner et al. reported that high EDE levels of Aβ40, Aβ42, and p-Tau in patients with WMHs indicate the presence of small cerebral vascular disease in the pre-clinical stages of AD (Abner et al., 2020).

Senescent cells secrete not only inflammatory biomolecules but also release higher amounts of EVs with unique compositions, a phenomenon mainly regulated by p53 (Buratta et al., 2017; Carracedo et al., 2019; Jakhar and Crasta, 2019; Lehmann et al., 2008; Misawa et al., 2020; Saheera et al., 2020; Takahashi et al., 2017; Urbanelli et al., 2016). EVs could carry SASP factors and have been suggested to play a role as senescence effectors (Jakhar and Crasta, 2019). EV secretion by senescent cells is a defense mechanism to maintain cellular homeostasis (Misawa et al., 2020; Takahashi et al., 2017). EVs secreted by senescent cells could potentially contribute to detrimental effects (e.g., pro-tumorigenic effects, accelerate aging, and various age-associated pathologies) or beneficial outcomes (tissue regeneration and wound healing) (Misawa et al., 2020; Saheera et al., 2020; Takahashi et al., 2017; Urbanelli et al., 2016). Considering the importance of senescence in AD pathology, various brain-cell derived exosomes or EVs in the blood (or other biofluids) could offer important information regarding the senescent cell type. Furthermore, the expression of senescence-related biomarkers in brain cell-derived exosomes could potentially serve as novel and early diagnostic biomarkers for incipient AD.

Overall, the availability of easily accessible and less invasive exosomes in plasma has shown immense promise to better understand AD pathogenesis. Though, several challenges persist, such as lack of standard protocols for the collections and storage of biofluids as well as isolation and characterization of pure exosomes. For example, based upon currently available methods, it is impossible to isolate pure exosomes from biofluids and separate exosomes from other EVs of similar characteristics. Another key challenge is the specificity of exosomal surface proteins currently being used to isolate brain cell-derived exosomes from plasma. For example, L1CAM (L1 cell adhesion molecule) is widely used to isolate NDE from plasma; however, questions have been recently raised regarding the neuronal specificity of L1CAM and the presence of L1CAM on the NDE surface (Norman et al., 2021). It is also clear that more specific exosomal surface biomarkers are needed as currently used biomarkers do not provide any information on the brain region from which they originate. Isolation of exosomes from blood using surface markers unique to specific anatomical sites of the brain could enable a greater understanding of the biology of these sites (including senescence) and their relative contribution in AD. Though this research area is understudied and the role of exosomes secreted by senescent cells during AD pathogenesis is not well characterized, it holds great promise and warrants further investigation.

2.2.4. Relevant outcome measures for AD trials – considering pathogenesis and symptoms

For the purpose of research, AD is defined by the presence of Aβ plaques and tau-containing NFTs (Jack et al., 2018). While the mechanisms linking plaques and tangles to the behavioral and cognitive symptoms are not clear, strategies to reduce these protein aggregates are hypothesized to alter disease trajectory. Until recently, the FDA approved drugs for MCI and AD have only offered symptom management. For example, acetylcholinesterase inhibitors (donepezil, rivastigmine, galantamine) and a N-methyl-d-aspartate receptor antagonist (memantine) modulate cholinergic and glutamatergic signaling, respectively (Howard et al., 2012; Raina et al., 2008; Rogawski and Wenk, 2003). While they effectively modulate symptoms and behaviors, they are unable to reverse or delay the course of AD progression (Raina et al., 2008). However, in June 2021, the Aβ specific antibody, aducanumab was announced as the first FDA approved drug for altering AD disease trajectory (2021b; Sevigny et al., 2016). Aducanumab is unique compared to the historically utilized medications indicated for use in MCI or AD. Its purpose is to remove Aβ plaques and slow disease progression, whereas other FDA approved drugs modulate disease symptoms. Similar to aducanumab, the compounds described in Section 3 of this review, aim to target biological processes responsible for AD pathogenesis. Outcomes for senescence AD trials, therefore, need to consider the underlying pathogenesis (Aβ and tau), the downstream effects on cognition, and symptoms associated with disease progression, in addition to assessing markers of senescence and its consequences.

For AD clinical studies, there are several measures, when collected together, can contribute to a comprehensive summary of AD status. Imaging and biomarker protocols, behavioral assessments, neurological, physical, and functional assessments, as well as a harmonized cognitive/neuropsychological battery has been established by the National Alzheimer’s Coordinating Center (NACC) that promotes standardization of data collection among Alzheimer’s Disease Research Centers (ADRCs) (Weintraub et al., 2009). Similarly, the Alzheimer’s Disease Neuroimaging Initiative (ADNI) has worked hard to validate imaging biomarkers for consistent use in clinical studies (Veitch et al., 2019). Imaging data is commonly collected via MRI to assess morphologic and volumetric changes. PET tracer imaging has improved the utility of imaging data to help identify molecular brain changes in specific brain regions (Alexander et al., 2002; Frisoni et al., 2010; Johnson et al., 2005). Regarding biomarkers relevant to AD, collection of CSF via lumbar puncture often tests for Aβ (Aβ40 and Aβ42), phosphorylated-tau (p-Tau) and total-tau as markers of pathologic load, but tools are being developed to measure these and other (i.e. Aβ structure, Aβ42/Aβ43, Aβ42/APP669-711, p-Tau 181/217/231, flotillin) informative markers less invasively in blood (Suárez-Calvet et al., 2020; Zou et al., 2020). With the identification and improved measuring techniques for AD biomarkers in plasma and CSF, including specificity for tau isoforms, these new tools should be incorporated into newly proposed AD clinical trials. Similar biomarker and imaging strategies to reliably identify senescent cells in vivo would significantly benefit senescence trials. Until these advances are made, levels of senescent cells in the blood or biopsied tissues, and composite measures of SASP factors remain the gold standard for senescence trials (Hickson et al., 2019; Justice et al., 2019b).

3. Clinical trials in AD that directly or indirectly target cellular senescence

3.1. Senolytic compounds

The first senolytic compounds were identified using a hypothesis-driven bioinformatics approach (Zhu et al., 2015). Senescent cells are resistant to apoptosis, which is governed through the upregulation of senescent cell anti-apoptotic pathways (SCAPs). Compounds were subsequently identified that disrupted the SCAPs, inducing death of senescent cells while leaving healthy cells unaffected. Forty-six potential senolytic agents were discovered through this process. To advance translational efforts, the majority of research has focused on agents with known safety profiles and limited off-target effects (Kirkland and Tchkonia, 2020). The best characterized senolytic agents are dasatinib, a tyrosine kinase inhibitor approved for use in humans for cancer treatment, and quercetin, a naturally occurring plant flavonoid. The agents have a synergistic effect, making their combination more potent for senescent cell clearance (Zhu et al., 2015). As senescent cells do not divide and accumulate over a period of weeks, they can be administered using an intermittent approach, which further serves to reduce the risk of side effects (Kirkland and Tchkonia, 2020). In preclinical trials, the combination of dasatinib and quercetin (D + Q) have been found to alleviate numerous chronic medical conditions including vascular stiffness, osteoporosis, frailty, and hepatic stenosis (Kirkland and Tchkonia, 2020; Ogrodnik et al., 2017; Roos et al., 2016; Tchkonia and Kirkland, 2018). The first clinical trial of D + Q was conducted as an open-labeled study in adults with IPF (Justice et al., 2019a). The treatment was generally well-tolerated with no trial discontinuation due to side effects. Although IPF is a fatally progressive disease, participants demonstrated improvements in physical functioning after three weeks of intermittent treatment. A subsequent open label trial of D + Q conducted in adults with diabetic kidney disease demonstrated effective senescent cell clearance in adipose tissue after only 11 days of intermittent treatment (Hickson et al., 2019).

Initial proof-of-concept data for clearing senescent cells as a therapeutic treatment for AD was conducted by our lab using tau transgenic mice (Musi et al., 2018) and has since been replicated by independent teams (Bussian et al., 2018; Zhang et al., 2019). In our study, oral D + Q were intermittently administered to tau transgenic mice with late-stage pathology (approximated to a 70-year-old human with advanced AD) (Musi et al., 2018). The treatment effectively reduced cellular senescence and associated SASP. The 35 % reduction in NFTs was accompanied by enhanced neuron density, decreased ventricular enlargement, diminished tau accumulation, and restoration of aberrant cerebral blood flow. A subsequent preclinical study validated the findings, reporting that intermittently administered D + Q cleared senescent cells in the central nervous system, reduced Aβ plaques, attenuated neuroinflammation, and enhanced cognition (Zhang et al., 2019).

The strong mechanistic support for cellular senescence as a novel target for AD in preclinical trials, coupled with the encouraging safety data from human studies for other conditions, provided the framework for translational research in neurodegenerative disease. The first clinical trial of D + Q for early-stage AD has completed enrollment (Gonzales et al., 2021). The primary aim of the open-label pilot study was to examine the central nervous system penetrance of D and Q in a small sample of older adults with early-stage AD (NCT04063124). In addition, two placebo-controlled trials of D + Q for neurodegenerative disease are underway (NCT04685590 and NCT04785300). One of the trials in development is a multi-site, double blind, randomized, placebo-controlled study of senolytic therapy in older adults with amnestic mild cognitive impairment (MCI) or early-stage dementia (Clinical Dementia Rating Scale (CDR) Global 0.5–1) due to AD (elevated CSF total tau/Aβ42 ratio). The treatment regimen will consist of 12-weeks of intermittently administered oral D + Q. Participants will continue to be monitored across a 9-month follow-up period to evaluate changes in senescence markers and disease trajectory. Across the study duration, safety will be closely monitored with assessment of adverse events, vital signs, physical examination, ECG, and laboratory parameters (biochemistry, hematology, liver, and renal function). Baseline, post-treatment (12 weeks), and end of study (48 weeks) changes will be examined across indices of blood and CSF senescence and AD markers, tau PET retention, cognition, functional status, brain MRI, and physical functioning. The results will shed insight on efficacy, safety, and feasibility of senolytic treatment for AD and will be used to evaluate the decision to pursue a larger trial.

3.2. Senomorphic compounds

While senolytics ablate senescent cells, senomorphics attenuate senescence without affecting the total number of senescent cells (Mongelli et al., 2020). Several senomorphics exert their effects by modulating the SASP and may extend organism lifespan (Kirkland and Tchkonia, 2020). Similar to senolytics, clinical trials of senomorphic agents for neurodegenerative disease are presently underway. The sections below highlight a few examples of senomorphics and select clinical trials targeting neurodegenerative disease. As these agents have pleiotropic effects on pathways relevant to aging (Kirkland and Tchkonia, 2020; Mongelli et al., 2020), assessment of their potential impact directly attributable to senomorphic properties will require thorough investigation.

3.2.1. Rapamycin

The mechanistic target of rapamycin (mTOR) is a serine-kinase that is integral to cellular metabolism and its dysregulation propagates multi-system degradation, potentially contributing to age-related chronic diseases (Stanfel et al., 2009). mTOR is a member of the phosphatidylinositol-3 kinases (PI3K)-related kinase (PIKK) family that is present in at least two protein complexes, mTOR complex 1 (mTORC1) and mTORC2 (Saxton and Sabatini, 2017). MTORC1 has been directly implicated in propagating the SASP through upregulation of cytokine IL1A (Laberge et al., 2015). mTORC1 is also hypothesized to contribute to the metabolic dysregulation of senescent cells by inhibiting autophagy of dysfunctional mitochondria (Carroll et al., 2017).

Rapamycin, an FDA-approved immune modulator, effectively suppresses the mTOR pathway and extends lifespan in model organisms (Arriola Apelo and Lamming, 2016). Rapamycin dampens the SASP through inhibition of IL1α, which in turn reduces the transcription of inflammatory genes regulated by NF-κB (Laberge et al., 2015; Mongelli et al., 2020). In preclinical trials, rapamycin has been shown to ameliorate several age-related conditions including neurodegenerative disease. In mouse models of AD, oral rapamycin administration attenuated cerebral mTOR signaling, and reduced pro-inflammatory cytokine levels in the central nervous system (Van Skike et al., 2018). Rapamycin has also been shown to reduce the accumulation of Aβ and tau, enhance cerebral blood flow, and improve cognition in preclinical studies (Caccamo et al., 2010; Lin et al., 2013; Tang et al., 2013). A double blind, randomized, placebo-controlled Phase 2 clinical trial of rapamycin for MCI and early-stage dementia (CDR Global 0.5–1) due to AD (amyloid PET positivity) is presently underway (NCT04629495). Enrolled participants will be randomized 1:1 to rapamycin (1 mg/day) or placebo for 12-months followed by an extended 6-month observation period. Safety will be evaluated across the study duration with assessment of adverse events, vital signs, and laboratory parameters. Baseline, post-treatment (12 months) and end of study (18 months) assessments include brain MRI, FDG PET, cognitive evaluation, questionnaires, and blood and CSF markers of AD and neuroinflammation. A multicenter, phase 2 randomized, double-blind, placebo-controlled clinical trial of rapamycin for ALS is also underway (NCT03359538) (Mandrioli et al., 2018). The primary aim of the investigation is to evaluate changes in regulatory T lymphocytes (Treg), which have been previously associated with ALS disease progression (Beers et al., 2011). Secondary outcomes include broader assessments of changes in immune parameters, inflammatory markers, safety, and clinical outcomes.

3.2.2. Nicotinamide riboside

Sirtuins are a class of histone deacetylases, which play a critical role in the modulation of chromatin structure and enable gene transcription (North and Verdin, 2004; Toiber et al., 2011). They also increase anti-oxidant production and have been shown to extend lifespan in preclinical studies (Kaeberlein et al., 1999). Sirtuins attenuate cellular senescence by preventing age-related telomere attrition, maintaining chromatin integrity, and promoting DNA repair (Lee et al., 2019). Sirtuin activity requires nicotinamide adenine dinucleotide (NAD+), a coenzyme crucial for cellular metabolism (Mouchiroud et al., 2013). Nicotinamide riboside is an NAD + precursor that has been shown to increase lifespan in animal studies (Cantó et al., 2012). Nicotinamide riboside further contributes to axonal integrity and prevents neuronal death in model organisms (Schöndorf et al., 2018; Vaur et al., 2017). We recently completed a double-blinded RCT pilot study of nicotinamide riboside for individuals with MCI. The 10-week study increased blood NAD, was well-tolerated, and showed evidence of altered cerebral blood flow in disease relevant brain regions (Orr et al., 2020). Additional trials examining the efficacy of nicotinamide riboside for subjective cognitive decline, MCI, and AD are underway (NCT04078178, NCT04430517).

3.2.3. Metformin

Metformin is an FDA-approved medication for diabetes, which modulates several key pathways underlying aging (Barzilai et al., 2016). These include reducing IGF-1 and mTOR signaling, attenuating reactive oxygen species production, and decreasing cellular senescence (Barzilai et al., 2016; Batandier et al., 2006; Kickstein et al., 2010; Liu et al., 2011; Moiseeva et al., 2013). Metformin has been shown to reduce cellular senescence by stimulating autophagy and inhibiting the NF-κB pathway (Deschênes-Simard et al., 2019; Moiseeva et al., 2013). In model organisms, metformin effectively extends lifespan and healthspan (Chen et al., 2017; Martin-Montalvo et al., 2013). Observational studies have further highlighted the agent’s potential cognitive benefits. Within the Singapore Longitudinal Aging Study, metformin use was associated with a 51 % reduction in risk for cognitive impairment over and above adjustment for vascular risk factors (Ng et al., 2014). A small placebo-controlled clinical trial of individuals with major depression and type 2 diabetes reported that metformin enhanced cognition and reduced depressive symptoms relative to placebo (Guo et al., 2014). However, metformin has also been associated with increased risk of AD in population studies (Imfeld et al., 2012). A large-scale, multi-site NIH-funded initiative is underway to examine the potential of metformin for attenuating age-related diseases including cognitive decline as part of the Target Aging with Metformin (TAME) trial (Barzilai, 2017). Additional clinical trials of metformin for preventing or halting cognitive decline in healthy older adults and individuals with MCI and dementia have been conducted or are presently ongoing (NCT00620191, NCT01965756, NCT03757910, NCT04098666) with initial data supporting treatment-related improvements in cognition (Koenig et al., 2017b; Luchsinger et al., 2016). In a 16-week randomized crossover pilot trial of metformin in individuals with MCI or dementia due to AD, it was found to be safe, tolerable, and detected in the CSF of participants, indicating penetration of the blood brain barrier (Koenig et al., 2017b). Although no changes in AD biomarkers (including Aβ42, total tau, or p-Tau) were detected in the CSF, significant improvements in executive functioning and trends suggesting improvements in attention, learning, and memory were seen (Koenig et al., 2017b). While initial analyses revealed no significant differences in cerebral blood flow with metformin, post-hoc analyses revealed an improvement in orbitofrontal cerebral blood flow in the intervention group compared to placebo (Koenig et al., 2017b). Though the crossover trial did not include a washout period, included a small sample size (n = 20), and was too short to account for placebo responders, it is promising and warrants further investigation.

4. Concluding Remarks

Clinical trials focused on removing senescent cells and their toxicity in the context of AD are in early stages. These pioneering trials face daunting failure rates that have plagued the AD field and are tasked with overcoming challenges associated with determining efficacy/target engagement for a complex senescence phenotype in a notoriously difficult tissue. We also acknowledge that senescent cells contribute to both healthy physiology and disease. The early safety trials will be responsible for identifying potential negative effects of senescent cell removal. The intermittent dosing regimen planned for these trials was designed to optimally minimize potential drug side effects, reduce senescent cells and their detrimental SASP effects while preserving important function. Nevertheless, data from postmortem human brains and mechanistic studies in AD mouse models overwhelmingly support this next step. On the side of optimism, we note that many of the past failed studies focused on clinical outcomes relevant to cognition. With new, improved, sensitive imaging and biomarker assays, trials are now better positioned for success. They are able to recruit subjects with earlier disease stages, assess more sensitive outcomes that reflect the underlying biology, all while tracking changes in important cognitive outcomes (Rafii and Aisen, 2020).

We also note a potential need for interventions that target multiple pathways; senolytic trials innately test this approach. They target a hallmark driver of biological aging, and may have potential to broadly impact cellular and molecular processes across tissues. Indeed, the hallmarks of aging interact across molecular, cellular, and system levels - targeting one (i.e., senescence) may exert pleiotropic effects that may be potentially beneficial for the treatment of AD. Moreover, combining senolytics (or other agents targeting mechanisms underlying the biology of aging) with the newly approved aducanumab, or other promising protein-clearing therapies (i.e., tau reducing approaches) may be considered in future trials. Such a combination strategy could potentially clear existing pathologies and unhealthy cells to improve tissue health and function. In summary, the upcoming senolytic trials are backed by strong evidence and will provide a first glimpse into the efficacy of targeting an aging hallmark to treat AD. Regardless of outcome, these studies are critically important and will lay the groundwork for future geroscience trials in AD.