Geroscience‐guided repurposing of FDA‐approved drugs to target aging: A proposed process and prioritization

Abstract Common chronic diseases represent the greatest driver of rising healthcare costs, as well as declining function, independence, and quality of life. Geroscience‐guided approaches seek to delay the onset and progression of multiple chronic conditions by targeting fundamental biological pathways of aging. This approach is more likely to improve overall health and function in old age than treating individual diseases, by addressing aging the largest and mostly ignored risk factor for the leading causes of morbidity in older adults. Nevertheless, challenges in repurposing existing and moving newly discovered interventions from the bench to clinical care have impeded the progress of this potentially transformational paradigm shift. In this article, we propose the creation of a standardized process for evaluating FDA‐approved medications for their geroscience potential. Criteria for systematically evaluating the existing literature that spans from animal models to human studies will permit the prioritization of efforts and financial investments for translating geroscience and allow immediate progress on the design of the next Targeting Aging with MEtformin (TAME)‐like study involving such candidate gerotherapeutics.


| INTRODUC TI ON
Geroscience represents a novel paradigm whereby biological aging is recognized as the major modifiable driver of age-related diseases and other late-life conditions (Burch et al., 2014;Kennedy et al., 2014;López-Otín et al., 2013). Widespread clinical use of geroscience-guided interventions could transform the public health landscape because the ability to target biological aging as a risk factor could simultaneously delay the onset and progression of multiple conditions, thereby enhancing health, function, and independence in late-life. A corollary is that targeting this biology will affect human healthspan (the portion of lifespan free of major disease and disability) most profoundly, and with a better prognosis than the current model of addressing one disease at a time. Geroscientists agree on several criteria defining distinct yet interrelated molecular and cellular hallmarks of aging that include (1) the measurable systemic or tissue-specific biological process should be altered during aging; (2) its disruption should have negative consequences in both lifespan and healthspan; and (3) when positively modulated, it should extend both lifespan and healthspan in preclinical models (Huffman et al., 2016;. Currently approved and accepted tools to target aging in humans are restricted to diet and exercise, which seem to influence multiple hallmarks of aging. On the contrary, advances in geroscience are spurring the development of gerotherapeutics (pharmacological interventions that promote health by directly or indirectly targeting the aging hallmarks, in addition to the disease itself) which show efficacy toward multiple age-related conditions. Interventions that influence these hallmarks at the cellular level toward more youthful function appear to lead to similarly favorable functional changes at the organ and systemic level.
While aging is unequivocally the major risk factor for age-related diseases, regulatory bodies around the world, such as the FDA or EMA, do not yet recognize geroscience-guided clinical outcomes as a path to regulatory approval. This is in part because the processes for validating specific compounds or combinations of compounds for their ability to delay the onset and progression of multiple chronic diseases have not yet been delineated in humans. Without regulatory approval, insurers will not pay for such treatments, which disincentivizes pharmaceutical companies from developing geroscience-guided approaches, simply because there is no path for them to develop a viable business plan. Therefore, there is an urgent need to demonstrate, in a well-designed clinical trial, that a cluster of age-related diseases can be significantly delayed by repurposing existing or developing novel gerotherapeutics.
Targeting Aging with MEtformin (TAME) is such a study that has been under development for the last few years, and whose basic principles have been developed in consultation with the FDA . Metformin was initially used in the 1950s to prevent influenza and malaria, but it was also noted to lower glucose levels in people with diabetes without triggering hypoglycemia, leading to it becoming the treatment of choice for type 2 diabetes (T2D).
In addition to its well-documented benefits on diabetes, clinical and observational studies have linked metformin to beneficial effects on multiple age-related diseases, including cardiovascular disease (CVD), cancer, Alzheimer's disease, and mild cognitive impairment, as well as reduced mortality Campbell et al., 2018;Novelle et al., 2016). TAME was designed by a group of geroscientists as a robust proof-of-principle study, whereby the primary outcome of the placebo-controlled, double-blinded trial will be the "time to event" of a cluster of age-related diseases consisting of cardiovascular events, cancer, cognitive decline, dementia, and death. In this manner, TAME is planned to provide a template for the design of future studies evaluating other FDA-approved drugs that are repurposed as gerotherapeutics (Espeland et al., 2017;.
We believe that efforts to test and repurpose existing, safe gerotherapeutics should be extended beyond TAME, not only to increase the number of drugs potentially available to target aging in humans but also to mitigate the risks to the field, should any such trials fail to reach their desired outcome. The purpose of this paper is to gather available evidence in the literature that supports and ranks potential gerotherapeutics and to help clinical investigators, geroscientists, not-for-profit foundations, and governments to accelerate research and clinical trials to test their efficacy and safety. Importantly, our mission is for such drug candidates to become clinically available relatively soon, and thus, we limit our analyses only to potential gerotherapeutics that are already FDA-approved for other clinical indications.
We sought to identify such FDA-approved drugs or classes of drugs that had at least one publication showing extension of lifespan in rodents and data in humans suggesting the highest chance of success if tested in a well-controlled TAME-like clinical trial. We developed a 12-point prioritization scale that assigns equal points for the preclinical and clinical evidence for each of these candidates. Points on the preclinical side were assigned for effects on the hallmarks of aging, improvement in healthspan and extension of lifespan in rodents as part of the NIA's Interventions Testing Program (ITP), a wellcharacterized, multicentered study to evaluate gerotherapeutics, as well as non-ITP rodent lifespan studies Nadon et al., 2017). On the clinical side, points were assigned for observed healthspan outcomes extending beyond the diseases targeted by the drug, and mortality from any cause or off-target diseases, with a scale where interventional studies weigh more than observational studies. Here, we outline the nature of the process and priority scoring designed to assess the likelihood that a specific gerotherapeutic may be successful in a clinical study.

| Identification of potential gerotherapeutics among FDA-approved drugs
We conducted mining of the DrugAge (Build 3) database for drugs that extend lifespan (Barardo et al., 2017;detailed in Figure 1).
Results were then filtered for rodents (mice and rats) in studies reporting statistically significant increased median and/or maximum lifespan as the inclusion criteria (n = 41 drugs or supplements). This approach meant that drugs that have shown some clinical benefits but have not shown extension of lifespan in preclinical rodent models were excluded from further analysis. These findings were then updated with drugs tested in the ITP since 2019 (DrugAge Build 3 was released in July 2019). Nutraceutical/supplemental products were excluded, and only FDA-approved drugs were included (n = 9 drugs). Search was then split into preclinical (more detail on lifespan, measures of healthspan and hallmarks of aging) and clinical (healthspan and mortality) using the search terms and datasets as described in Figure 1. The search terms for clinical outcomes were based on the non-communicable diseases among the 10 leading causes of death in persons age 65 and older in the US (CDC, 2017

| Ranking gerotherapeutics
An ordinal 12-point scale was equally divided between basic and clinical studies (6 points each). Thereby, promising gerotherapeutics that have significant basic geroscience rationale, yet currently lack supportive data from human subject studies were not penalized.

| Preclinical scoring
Of the 6 points evaluating basic or preclinical factors, up to 2 points were assigned each for targeting hallmarks of aging, improving preclinical healthspan and preclinical lifespan each, with the breakdown as follows: (i) 1 point assigned for less than 3 hallmarks, 2 points assigned for 3 or more hallmarks; (ii) 2 points assigned for effect on healthspan parameters; (iii) 1 point assigned for lifespan tested outside ITP, 2 points assigned for a significant increase in lifespan within ITP (López-Otín et al., 2013).

| Clinical scoring
Out of the 6 points assessing clinical considerations, up to 3 points were assigned to healthspan and 3 points for mortality data, as follows: (i) Healthspan: drug needed to demonstrate that it targeted at least one age-related disease/pathologic process which it was not intended to treat, with 1 point assigned for observational studies and 3 points for interventional, randomized controlled trials (RCTs); (ii) mortality: drug needed to demonstrate that it reduced all-cause mortality or death from a disease which it was not intended to treat, with 1 point assigned for observational studies and 3 points assigned for RCTs.

| RE SULTS
Using the process detailed in Figure 1, we were able to prioritize 9 drug classes (Table 1). SGLT2 inhibitors (SGLT2i), a relatively new drug class, was the only one to receive the maximum score, owing to not only its robust effects on improving rodent healthspan and lifespan (including ITP) but also strong evidence for the extension of healthspan and reduction of mortality in humans. Metformin was next on the list, and it received a submaximal score, due to negative findings for rodent lifespan extension in ITP. Acarbose, rapamycin/ rapalogs, and methylene blue (MB) all had strong preclinical data and promising findings for human healthspan (the latter being the most robust for acarbose), but sparse clinical data for human mortality.
Angiotensin-converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARBs) were found to extend preclinical healthspan and lifespan (outside of ITP) and had robust effects on extending human healthspan, but the studies on human mortality, while abundant, were predominantly negative. The last three drugs on our list, senolytics-Dasatinib + Quercetin (D + Q), aspirin, and N-acetyl cysteine (NAC), all had strong preclinical data, but their effects on human healthspan and mortality have not yet been assessed in clinical studies or appropriate doses/populations. Obviously, future studies may change the priority order for drugs that did not receive points due to the paucity of clinical data.

| Rodent lifespan
Gerotherapeutics in preclinical models were primarily evaluated for their ability to extend rodent lifespan (Table 2). Robust evidence from ITP suggests that SGLT2i (canagliflozin), acarbose, rapamycin, MB, and aspirin extend either median and maximum lifespan or median lifespan alone in genetically heterogeneous UM-HET3 mice, generally in a sexually dimorphic manner Nadon et al., 2017). SGLT2i canagliflozin extended median survival by 14% and the age for 90th percentile survival by 9% in male mice . Acarbose extended median lifespan by 22% in males and only 5% in females, while rapamycin exhibits a dose-dependent response by increasing median lifespan by 16%-26% in females and 13%-23% in males Miller et al., 2011Miller et al., , 2014Strong et al., 2016Strong et al., , 2020. Although less rigorous than ITP, in-

| Rodent healthspan
In addition to lifespan effects, we evaluated the evidence of potential gerotherapeutics on improving healthspan in preclinical models of aging and age-related diseases, where the action of the drug was attributed to their non-primary target or mechanism. Metformin displayed strong evidence in targeting conditions that were beyond its antihyperglycemic function Kulkarni et al., 2018Kulkarni et al., , 2020. In osteoarthritic mice, it moderated cartilage degeneration and chondrocyte aging . Similarly, SGLT2i and acarbose demonstrated effects beyond their glucose-lowering potential by lowering the incidence of atherosclerosis in diabetic ApoE −/− mice with the former, and that of lung tumors, liver degeneration, and glomerulosclerosis with the latter (Han et al., 2017;Harrison et al., 2019;Liu et al., 2021). In mouse models of neurodegeneration, metformin, acarbose, rapamycin, MB, aspirin, and NAC improved cognitive and motor functions and attenuated age-related memory impairments and behavioral changes (Cao et al., 2012;Chandra et al., 2018;Hosokawa et al., 2012;Kaeberlein & Galvan, 2019;Kodali et al., 2021;Lamming et al., 2012;Martínez et al., 2000;Ryu et al., 2020;Spilman et al., 2010;Tong et al., 2015;Yan et al., 2015;Zakaria et al., 2016).
Complementary to its effects on lifespan, rapamycin improved mouse healthspan by reducing the incidence of retinopathy, myocardial alterations, liver degeneration, and endometrial hyperplasia while also scavenging reactive oxygen species (ROS) in corneal endothelial cells Shin et al., 2011).
Rapamycin, metformin, and ACEi enalapril are also shown to impact indicators of frailty, including motor and physical performance, grip strength, stride length, resistance to muscle fatigue as well as decreased incidence of kyphosis index and cataracts in middle-

| Attenuation of hallmarks of aging
To assess the molecular utility of gerotherapeutics in targeting fundamental mechanisms of aging, we subsequently evaluated their actions on hallmarks of aging. By grouping similar pathways together, the seven pillars of aging were classified into four (1) Macromolecular damage and adaptation to stress; (2) epigenetic effects, stem cell renewal and regeneration; (3) proteostasis, inflammation (and senescence); and (4) metabolism (Kennedy et al., 2014;López-Otín et al., 2013). Importantly, drugs well-studied from the aging perspective such as metformin, rapamycin, and aspirin have substantial evidence of targeting hallmarks of aging, as compared to relatively less studied or newly approved drugs including SGLT2i, senolytics, and NAC.
Additionally, gerotherapeutics are shown to regulate metabolism by targeting evolutionarily conserved nutrient-sensing pathways across multiple species, primarily AMPK signaling by metformin, mTOR signaling by rapamycin and rapalogs, and insulin-IGF1 signaling by acarbose (Barzilai et al., 2012;Tong et al., 2015). Glucose tolerance was improved with SGLT2i dapagliflozin while senolytics  • ↓ Age-related behavioral and biochemical changes in SAMP8 mice  • ↓ Age-related memory impairment  • ↓ tumor burden and hematocrit in Apc +/Min mouse model of intestinal tumorigenesis at the higher dose (Dodds et al., 2020) No applicable studies

| Mortality
To assess the effects of gerotherapeutics on mortality, we focused on all-cause mortality and deaths from off-target diseases, that is, those that were not the primary target for the drug (Table 3). Rapalogs are also used to coat intravascular stents, but systemic doses delivered by this route are merely a fraction of a single daily dose used to achieve immunomodulation (Wiemer et al., 2008).
In small clinical trials, MB improved survival post cardiac surgery, owing to its vasoconstrictive effect, but it was not studied outside of the critical care setting (Levin et al., 2004). Numerous studies have investigated the role of aspirin in human lifespan and healthspan but, unlike animal studies, the administered doses were in the range at which aspirin displays anti-platelet, not anti-inflammatory activity.
The use of anti-inflammatory doses in humans would likely be prohibited by the risk of bleeding (ASCEND Study Collaborative Group, 2018; Hansson et al., 1998;McNeil et al., 2018;Ridker et al., 2005).
While abundant, clinical mortality data for ACEi/ARB in nonhigh-risk hypertensive populations (i.e., without heart failure, acute coronary syndromes, and organ transplants) are conflicted, at least in part due to marked variability in study populations, agents, and dosages used, comparators (placebo or active comparators), openlabel addition of other agents and achieved blood pressure goals (Hansson et al., 1999;Lithell et al., 2003Lithell et al., , 2004Oparil et al., 2013).

Hallmarks of Aging
• No applicable studies.

No applicable studies No applicable studies
Meta-analysis: No Δ in CRC mortality (Dai et al., 2015) •

| Human healthspan
In examining human healthspan, we focused on the ability of gerotherapeutics to prevent age-related disease or reduce the progression of off-target age-related pathologies. SGLT2i, originally intended for glucose-lowering, in RCTs demonstrated particularly strong and consistent cardioprotective and renoprotective effects, while metformin had beneficial effects across the widest range of healthspan outcomes  Hong et al., 2013;Knowler et al., 2002). Pilot trials suggest that metformin may improve memory and executive function, while observational studies and meta-analyses reported up to 81% reduction in incident dementia (Campbell et al., 2018;Cheng et al., 2014;Koenig et al., 2017;Luchsinger et al., 2017;Samaras et al., 2020;Zhou et al., 2020). By contrast, one small observational study reported impaired cognitive performance with metformin, but it did not adjust for glycemic control, duration of diabetes, or concomitant glucose-lowering therapy (Moore et al., 2013). Another case-control study reported that metformin use was associated with increased odds for prevalent Alzheimer's disease but without consistent dose-response relationship or replication of findings with metformin monotherapy (Imfeld et al., 2012). Observational evidence further suggests that metformin may reduce overall cancer incidence by 31%, including colon, liver, pancreas, lung, and ovarian cancer (Gandini et al., 2014;Lee et al., 2011;Shi et al., 2019;Tseng, 2012;Xiao et al., 2020;Zhou et al., 2016).
Other drugs have been less studied in appropriate doses/populations, except for ACEi/ARB. In RCTs, acarbose prevented T2D in individuals with glucose intolerance (Chiasson et al., 2002;Holman et al., 2017). The STOP-NIDDM RCT found a 49% reduction in major cardiovascular events with acarbose, while two RCTs in higher-risk populations reported conflicted findings (Chiasson et al., 2003;Holman et al., 2017;Yun et al., 2016). Rapalogs have shown promise for boosting the immune response to vaccination and reducing respiratory infections in older individuals (Mannick et al., 2014(Mannick et al., , 2018. Recent phase 2b and phase 3 studies of RTB101 (also named BEZ235, dactolisib) reported a reduction in laboratory-confirmed respiratory infections and upregulation of IFN-induced antiviral genes in older participants. However, since RTB101 is not an FDAapproved agent, we excluded these studies from grading (Mannick et al., 2021). MB and its derivatives show promise in reducing cognitive decline in mild and moderate Alzheimer's disease, while D+Q have shown beneficial effects in small open-label pilot clinical trials in idiopathic pulmonary fibrosis and CKD (Hickson et al., 2019;Justice et al., 2019;Wilcock et al., 2018;Wischik et al., 2015).

| DISCUSS ION
The primary objective of the nascent field of geroscience is to translate the discoveries from basic research on the biology of aging to human studies and ultimately clinical care (Sierra et al., 2021).

N-Acetyl Cysteine
No applicable studies. No applicable studies.

TA B L E 3 (Continued)
Current approaches to achieving this goal range from continuing robust research in preclinical models to testing candidate gerotherapeutics in clinical trials against individual age-related diseases.
Vigorous efforts into the development of possible phase-III trials are needed and repurposing already FDA-approved drugs represents a promising avenue offering accelerated evidence for geroscience at only a fraction of the cost and a faster rate of return than validating new compounds. A critical issue, therefore, is the identification of the best possible candidates for this purpose. Here, we identify several drugs that, like metformin, are relatively low-cost molecular entities with a well-established safety profile, representing possible candidates for inclusion in a geroscience-guided clinical trial modeled after TAME, and aimed at probing whether a predefined cluster of age-related diseases or well-characterized loss of function can be significantly delayed using an approach based on targeting aging biology.
We provide a rigorous assessment of the literature concerning both preclinical and clinical status for several FDA-approved potential gerotherapeutics. We developed a scoring scale that equally values (6 points each) the preclinical and clinical evidence supporting the claims for each of the nine potential gerotherapeutics selected.
Surprisingly, with a score of 12/12, the best evidence was found for a relatively new class of drugs, the SGLT2i, which showed better promise than other known potential gerotherapeutics such as metformin, rapamycin, or acarbose.
The profound effects of SGLT2i on healthspan and lifespan in both animal models and humans may result from the pleiotropic effects beyond renal glucose and sodium handling, including improved mitochondrial function, restoration of autophagy, and promotion of ketogenesis-dependent dampening of mTORC1 hyperactivation (Durak et al., 2018;Korbut et al., 2020;Tomita et al., 2020). We also identify several drugs that, like metformin, are relatively lowcost molecular entities with a well-established safety profile, representing possible candidates for inclusion in a geroscience-guided clinical trial modeled after TAME, and aimed at probing whether a predefined cluster of age-related diseases or well-characterized loss of function can be significantly delayed using an approach based on targeting aging biology. Moreover, our geroscience-guided literature-driven approach is concordant with previously proposed data-driven drug repurposing approaches using evidence from gene expression or model organism studies, with the overlap of rapamycin and senolytics identified as top candidate drugs across these (Dönertaş et al., 2018;Ziehm et al., 2017).
For our analysis, we decided to assign an equal number of points to knowledge derived from geroscience (i.e., preclinical) and to findings from humans (i.e., clinical-both observational and interventional where their action is clearly not related to their usual therapeutic use in osteoporosis. So, while these and possibly other FDA-approved drugs might indeed show great potential as gerotherapeutics, they have not met our inclusion criteria, and their analysis is beyond the scope of this paper. The available data suggests that both colchicine and bisphosphonates should be rapidly tested in the ITP or in other centers, to confirm improved healthspan and lifespan in mammals. In addition to the classes of drugs discussed above, another type of molecular entity that was not considered in our analysis deserves special mention; we did not consider nutraceuticals, several of which have indeed been shown to positively affect both healthspan and lifespan (e.g., NAD + replenishment). They not only lack robust pharmacological and clinical data, but also fail to serve our main purpose of identifying FDA-approved drugs with the goal of repurposing them as gerotherapeutics that can reduce the rate of deterioration accompanying aging.
One important point that was also not considered in our analysis involves potential sexually dimorphic effects of gerotherapeutics.
This is an important issue as we try to move geroscience into the clinical realm since many of the positive results published by the ITP are sex-specific, increasing longevity either solely, or at least preferentially in one sex. For example, the ITP has clearly shown that acarbose has convincingly male-specific effects on longevity (Nadon et al., 2017). Much more work is required in this domain, but we did not consider this variable because, in most human studies, data are adjusted for sex, so that potential differences are obscured, and in many cases, the size of the studies precludes the ability to perform a separate bona fide sex-specific analysis. Another issue that is relevant for translating findings from geroscience-guided animal studies into humans is the question of dosing. For instance, the doses of rapamycin or rapalogs used both in mice (ITP studies) and human (Mannick et al, 2014;Mannick et al, 2018;Mannick et al, 2021) studies are several times lower than doses commonly used in clinical practice to induce immunotolerance in transplant patients, or as an adjuvant to therapy in stage IV cancer. Furthermore, in human studies aimed at geroprotection (Mannick et al, 2014;Mannick et al, 2018;Mannick et al, 2021), these drugs have been used intermittently and/or for short periods, unlike in clinical practice where they are used chronically. As a result, especially in human studies, different dosage regimens result in different outcomes, that is, stimulation of the immune response in gerotherapeutic studies with intermittent exposure to low doses vs.
immunosuppression in clinical settings with chronic administration of high doses. Investigating these dose-specific differences will allow for minimizing side effects in gerotherapy clinical trials.
Recent advances in geroscience and aging biology research show that the hallmarks and pillars of aging described almost a decade ago interact strongly with each other, often in an additive or even a synergistic manner (Kennedy et al., 2014;Kulkarni et al., 2020;López-Otín et al., 2013). It is therefore possible to imagine that future gerotherapies will encompass combinations of various gerotherapeutics (perhaps in lower doses and thus with better safety profile) or, in cases where a patient is already afflicted by a disease, a combination of a gerotherapeutic along with a disease-specific intervention. For example, elimination of senescent cells in mice can improve the response to chemotherapy, both by further reducing tumor number and size, but also by decreasing side effects such as fatigue. Further research will be needed; however, since while some combinations (e.g., rapamycin +metformin) might result in further benefits than each drug alone, other combinations could result in either blunting of established geroprotective strategies (e.g., metformin or rapamycin with exercise) or worsening risk of toxicity often seen with polypharmacy in older adults (Strong et al., 2016;Walton et al., 2019).
Delaying aging to prevent multiple age-related diseases is an obviously more exciting and promising approach than attacking one disease at a time, where because of comorbidities, the outcome is a no-gain exchange of one disease for another. The benefits for health and fulfillment are obvious. However, there are also powerful political, economic, and societal gains to be attained, as framed in The Longevity Dividend (Olshansky, 2016). The economic value that can be derived from applying geroscience principles to healthcare has been recently reassessed showing that a slowdown in aging that increases life expectancy by just one year is worth $38 trillion a year, due to complementary gains in health and longevity (Scott et al., 2021). This is a very conservative estimate since the available data suggest that, for example, metformin has the potential to increase life span by ~3 years Scott et al., 2021). Thus, targeting aging with gerotherapeutics offers potentially larger economic gains than eradicating individual diseases, and for both quality of life and economic reasons, we cannot afford to forfeit this opportunity for a truly transformational approach to healthcare.

ACK N OWLED G M ENT
This work was supported by The Nathan Shock Center of Excellence for the basic Biology of Aging (P30AG038072) (NB) and the Einstein-Paul Glenn Foundation for Medical Research Center for the Biology of Human Aging (NB), and NIH grants P30AG067988 and R33AG061456 (GAK).

CO N FLI C T O F I NTE R E S T
ASK is an employee of AbbVie, but his recent employment did not influence the work reviewed or presented in this manuscript. SA, DB, FS, GAK, and NB declare no competing interests.