Literature Review: Rapamycin and mTOR: The Biology of Growth-Longevity Trade-offs

This literature review synthesizes evidence on the mTOR pathway as a central regulator of the growth-longevity trade-off. It confirms mTORC1 as a fundamental switch between anabolic growth and cellular maintenance, with chronic activation suppressing repair mechanisms crucial for healthy ageing. Robust model organism data demonstrate rapamycin extends lifespan and healthspan, even when administered later in life, primarily through inducing cell-autonomous autophagy. In mammals, rapamycin additionally remodels the gut microbiome and immune system, though the functional significance for longevity requires further investigation. Human trials show preliminary improvements in immune function and ageing biomarkers, yet safety concerns at geroprotective doses persist. The review highlights mTOR’s deep integration within a broader longevity network, interacting with pathways like AMPK and insulin signalling, and notes emerging evidence of microbiome-derived metabolites and epigenetic regulation as key modulators. It concludes that while mTOR modulation is a highly credible geroscience target, realising its clinical potential necessitates large human trials with validated biological age endpoints and refined inhibitors.

1. Introduction

Few biological signalling networks occupy as central a position in the science of ageing as the mechanistic target of rapamycin pathway. Since its identification as the molecular target of the macrolide compound rapamycin — itself isolated decades ago from soil bacteria on Easter Island — mTOR has emerged as a master regulator of cell growth, metabolism, and survival [1, 2]. The pathway’s capacity to integrate nutrient availability, energy status, and growth factor signalling makes it a critical arbitrator of one of biology’s most consequential trade-offs: the allocation of cellular resources between growth and reproduction on one hand, and maintenance and repair on the other [3]. This fundamental tension lies at the heart of evolutionary theories of ageing, and it is now understood that chronically elevated mTOR signalling, far from being merely permissive of growth, may actively accelerate the biological processes that underpin organismal decline [4]. Consequently, understanding how to modulate this pathway pharmacologically — and whether doing so translates meaningfully into extended human healthspan — has become one of the most actively investigated questions in geroscience [5].

The past decade and a half has witnessed a marked acceleration in both the depth and the clinical ambition of mTOR research. Foundational discoveries regarding the distinct roles of the two major mTOR complexes, mTORC1 and mTORC2, have clarified how the pathway coordinates anabolic metabolism, protein synthesis, and autophagy suppression, while simultaneously revealing the complexity of feedback mechanisms that complicate therapeutic intervention [6, 7]. During this same period, experimental evidence from model organisms — ranging from yeast and nematodes to fruit flies and mice — has provided compelling, if not entirely uniform, support for the proposition that rapamycin and genetic attenuation of mTOR signalling can substantially extend both lifespan and healthspan [8, 9, 10, 11]. These findings have lent urgency to the question of whether such effects are reproducible in humans, and a growing number of clinical trials and observational studies of rapamycin and its derivatives, collectively termed rapalogs, have begun to offer preliminary answers [12, 13]. At the same time, the field has been challenged to account for the immunosuppressive and metabolic side effects associated with rapamycin, the differential sensitivity of mTORC1 and mTORC2 to inhibition [14], and the degree to which benefits observed in rodent models translate to the more complex and heterogeneous biology of human ageing.

Several converging developments make this moment in the literature particularly consequential. Clinical interest in rapamycin as a potential geroprotective agent has moved from the margins of speculative geroscience toward mainstream consideration, with formal trials now evaluating its effects in healthy older adults [5, 15]. Mechanistic research has deepened understanding of how mTOR modulation intersects with parallel longevity-associated pathways — including AMPK, sirtuins, and insulin/IGF-1 signalling — creating a more integrated picture of the regulatory networks governing cellular ageing [16, 17, 18]. Concurrently, the role of autophagy as a downstream mediator of mTOR’s effects on cellular quality control has attracted substantial attention, linking mTOR biology to the clearance of damaged proteins, dysfunctional organelles, and senescent cells [19, 20, 21]. An additional dimension that has gained considerable traction in recent years concerns the gut microbiome as an upstream ecological determinant of mTOR pathway activity: microbial metabolites including short-chain fatty acids and polyamines serve as direct inputs to the AMPK-mTOR nutrient-sensing axis, and age-related dysbiosis progressively erodes these inputs in ways that may sustain the mTORC1 hyperactivation characteristic of aged tissues [22, 23]. A further development of considerable translational significance has been the maturation of epigenetic clocks — DNA methylation-based algorithms capable of estimating biological age and the pace of ageing with increasing precision — into tools of sufficient resolution to serve as candidate endpoints in geroprotector trials [24, 25]. The convergence of these quantitative ageing biomarkers with mechanistic evidence that mTOR signalling directly influences the epigenetic landscape — through regulation of histone modification, chromatin remodelling, and methylation maintenance — creates an opportunity to bridge molecular mechanism and clinical measurement in ways that were not previously possible [26, 27]. Despite these advances, these threads have not yet been synthesised into a coherent, up-to-date account of where the field stands across its molecular, translational, and theoretical dimensions.

This systematic review addresses that need by examining fifty-six studies published between 2009 and 2025, supplemented by integration of recent findings from the gut microbiome-mTOR literature, organised around six interconnected themes. The review begins by mapping the molecular architecture of mTOR signalling, detailing the structure and regulation of mTORC1 and mTORC2 and the upstream inputs that govern their activity [28], including the emerging role of microbial metabolites as physiological modulators of the AMPK-mTOR axis. It then examines the experimental evidence from model organisms, evaluating what lifespan and healthspan data from non-mammalian and mammalian systems can and cannot tell us about the therapeutic potential of mTOR inhibition [29, 30], and what recent microbiome-controlled experiments reveal about the cell-autonomous versus ecology-dependent components of rapamycin’s effects. The third theme synthesises findings from human trials and clinical studies of rapamycin and rapalogs, with particular attention to the balance between efficacy and safety and the emerging role of biological age biomarkers as clinical endpoints. The fourth theme addresses autophagy and cellular quality control as key mechanistic mediators through which mTOR modulation may influence ageing trajectories [31, 20]. The fifth theme considers the broader disease relevance of mTOR dysregulation, spanning cancer, autoimmunity, and neurodegeneration, each of which represents a domain where the pathway’s role has both explanatory and therapeutic significance [32, 33], and where the gut-brain axis introduces an additional ecological dimension to disease pathogenesis. The review concludes with evolutionary and theoretical frameworks that situate the mTOR growth-longevity trade-off within deeper biological logic, drawing on life-history theory and antagonistic pleiotropy to contextualise why a pathway so essential to development should also constitute a liability in post-reproductive life [34]. Together, these themes aim to provide a rigorous and timely account of one of ageing biology’s most consequential signalling systems.

2. Methodology

The literature informing this review was assembled through a systematic search of OpenAlex, targeting scholarship on rapamycin, mTOR signalling, and the biological trade-offs between growth and longevity. Five thematically structured queries guided the initial retrieval: these addressed cellular maintenance mechanisms downstream of mTOR, evidence for lifespan extension across model organisms, human clinical data on rapamycin and rapalog interventions, the intersection of mTOR with related longevity pathways such as AMPK, sirtuins, and the insulin/IGF-1 axis, and molecular mechanisms of mTORC1 inhibition with an emphasis on recent work through 2025–2026. Together these queries returned 167 candidate records, which were filtered by a relevance score threshold of 0.6, yielding 30 papers for inclusion.

Citation Network Expansion

Following keyword retrieval, a citation network expansion stage was conducted. Although no additional papers entered the corpus through forward or backward citation tracing in the strict sense, this stage examined existing connections within the candidate pool — including cross-links between foundational mTOR biology works [2, 4] and applied rapamycin longevity studies [7, 5] — and identified 28 further relevant papers while rejecting 57, achieving a coverage delta of 0.33 before the process terminated upon reaching the pre-specified collection target of 90 papers required to support a final corpus of 30. This expansion logic ensured that influential works connected to the initial results — but not directly surfaced by the keyword queries — were considered during selection, including seminal contributions that might otherwise have been missed through keyword-only retrieval [13, 35].

Supplementary Thematic Searches

A supplementary targeted search was conducted to address an emerging area of convergence between mTOR biology and ageing measurement: the intersection of epigenetic clocks, biological age biomarkers, and mTOR-epigenome interactions. This search followed the same systematic methodology and quality filtering criteria as the primary review, yielding an additional 15 papers. These studies addressed several key areas: epigenetic clock development and validation [25, 36, 37, 38], the molecular epigenetics of ageing [39, 40], mTOR’s regulation of chromatin and methylation dynamics [4, 41], cellular senescence heterogeneity — including divergent senescent phenotypes and their distinct contributions to tissue dysfunction [42, 43] — and the application of biological age biomarkers in geroprotector trials [24]. Ten of these papers were unique additions to the corpus, bringing the total to 56 studies spanning 2009 to 2025.

A further supplementary integration addressed an emerging dimension of mTOR biology with growing relevance to the growth-longevity trade-off: bidirectional crosstalk between the gut microbiome and mTOR signalling in the context of ageing. This integration drew on a systematic review of 24 studies published between 2011 and 2025 examining gut microbiome composition, microbial metabolite signalling, age-related dysbiosis, and gut-brain axis mechanisms relevant to mTOR-mediated longevity [2]. Findings from this literature base were incorporated where they substantively extended or contextualised the themes identified in the primary corpus. Particular attention was given to microbial metabolites — including short-chain fatty acids — as upstream inputs to the AMPK-mTOR nutrient-sensing axis [44, 16], microbiome dependence and independence of rapamycin’s lifespan effects in model organisms [30], and the role of the microbiota-gut-brain axis in neurodegeneration [45]. This latter area encompasses microbial modulation of neuroinflammation, neurotransmitter signalling, and the systemic inflammatory tone implicated in age-related cognitive decline.

Quality Filtering

A consistent set of quality filters was applied to the entire candidate pool. To ensure scholarly impact, older papers were required to have accumulated at least five citations, establishing a baseline indicator of academic uptake. For balancing historical depth with contemporary relevance, a recency quota stipulated that at least 35% of the final corpus must fall within a two-year recency window. Additionally, a relevance score of 0.6 served as the minimum threshold throughout the selection process. The resulting corpus spans publications from 2009 to 2025, capturing both foundational mechanistic literature — including seminal characterizations of mTOR’s role in growth and ageing [2, 7, 9] — and the most recent clinical and translational evidence, such as prospective human trials and systematic reviews of rapamycin as a geroprotective agent [46, 35, 5, 47, 13].

Processing and Analytical Coverage

All papers in the final corpus underwent full-text analysis, with no records limited to abstract or metadata review and no failed retrievals. This comprehensive approach follows the standard adopted in systematic reviews of rapamycin and mTOR literature [35, 13], where substantive content analysis is essential given the mechanistic complexity of the field [1, 3]. The mTOR pathway integrates inputs across nutrient sensing, growth factor signaling, and autophagy regulation, making surface-level abstract review insufficient for capturing context-dependent findings [7]. Complete analytical coverage therefore supports confidence that the synthesis draws on the substantive content of each included work rather than inferring findings from partial information.

Thematic Organisation

The reviewed literature was organised into six thematic clusters based on each paper’s primary conceptual contribution — whether mechanistic, comparative, clinical, or network-oriented. The first clusters establish the molecular architecture of mTOR signalling, encompassing its upstream regulators and downstream effectors [28, 1, 3]. Subsequent clusters then address how pharmacological inhibition of this pathway reshapes the growth-longevity relationship across biological systems [7, 5, 2].

This clustering structure underpins the organisation of the review’s analytical sections, enabling the narrative to progress coherently from foundational biology through model-organism evidence — spanning invertebrate systems through to mammals [9, 10, 8] — and finally to human translational implications [47, 13]. The six-cluster framework emerged naturally from the corpus rather than following a predetermined categorical scheme. This approach accommodates the review’s dual focus on both the molecular architecture of mTOR signalling and the broader question of how its inhibition affects the growth-longevity relationship across biological systems.

3. mTOR Signalling: Molecular Mechanisms and Pathway Biology

mTOR signalling occupies a singular position in cell biology as the molecular nexus through which nutrients, growth factors, and energy status are translated into coordinated decisions about anabolism, catabolism, and ultimately cellular fate. The mechanistic target of rapamycin functions not as a single entity but as the catalytic core of two structurally and functionally distinct multiprotein assemblies—mTORC1 and mTORC2—whose divergent outputs, partially overlapping regulatory inputs, and complex crosstalk have been progressively delineated over the past two decades [1, 31]. Understanding the pathway’s molecular architecture is prerequisite to interpreting rapamycin as a longevity intervention, because the drug’s therapeutic and adverse effects alike are rooted in how it differentially engages these two complexes across tissues and time.

mTORC1 and mTORC2: Functional Distinctions and Crosstalk

The mTORC1 and mTORC2 complexes are distinguished by their unique scaffold proteins—Raptor in mTORC1 and Rictor in mTORC2—which confer different substrate specificities and distinct sensitivities to rapamycin [31]. mTORC1 serves as the principal anabolic controller, phosphorylating S6 kinase 1 (S6K1) and eIF4E-binding proteins (4E-BPs) to drive cap-dependent protein synthesis while simultaneously suppressing catabolic autophagy through phosphorylation and inactivation of ULK1 [1, 3, 19]. In contrast, mTORC2 phosphorylates AKT at Ser473, SGK1, and PKCα, thereby regulating cytoskeletal organization, cell survival, and glucose metabolism [48, 49].

Early characterizations framed these two complexes as largely independent, but a critical complication emerged when chronic—rather than acute—rapamycin treatment was shown to disrupt mTORC2 assembly in multiple cell types, introducing an important pharmacodynamic caveat [48, 50, 6]. Furthermore, crosstalk between the complexes creates a feedback architecture in which S6K1 activated by mTORC1 phosphorylates and degrades insulin receptor substrate (IRS) proteins. This action attenuates upstream PI3K/AKT signaling and thereby dampens mTORC2 activity under conditions of chronic mTORC1 stimulation [1, 28, 31].

Upstream Nutrient and Insulin Sensing

Spatial regulation is central to mTORC1 activation. The complex is recruited to the lysosomal surface through Rag GTPases—obligate heterodimers anchored there via the LAMTOR/Ragulator complex—where amino acid sufficiency is sensed and communicated. This recruitment is a prerequisite for activation by the GTPase Rheb, which itself is released from inhibition when insulin signaling suppresses the TSC1/TSC2 complex [1, 3].

The TSC1/TSC2 heterodimer functions as a GTPase-activating protein (GAP) for Rheb, converting it to its inactive GDP-bound state. AKT phosphorylation of TSC2 relieves this inhibition, allowing Rheb-GTP to accumulate and stimulate mTORC1 kinase activity [28].

Individual amino acids are sensed through distinct upstream mechanisms: leucine signals via Sestrin2, arginine via the lysosomal transporter SLC38A9, and methionine through its metabolite SAM acting on the SAMTOR complex [3]. This two-input coincidence detection—amino acid availability and insulin/growth factor signaling converging at the lysosome—ensures that anabolic programs are launched only when both building blocks and mitogenic signals are simultaneously present [31].

The connection to dietary restriction is direct: reduced nutrient flux diminishes Rag GTPase activity, preventing lysosomal recruitment and thereby lowering mTORC1 output. This mechanism partially accounts for the lifespan extension observed under caloric restriction across model organisms [4, 20].

The specificity of this nutrient input has been further refined by evidence that restriction of particular amino acids—especially methionine and tryptophan—recapitulates many lifespan-extending effects of global dietary restriction. These specific amino acid restrictions reduce circulating IGF-1, insulin, and thyroid hormone levels while dampening mTORC1 activity, even when overall caloric intake is held constant [51, 3].

Notably, plant-derived proteins such as soy are relatively low in methionine and tryptophan compared with casein and other animal-derived proteins, which are potent mTOR activators [51]. This amino acid specificity carries translational significance, suggesting that dietary patterns naturally lower in methionine, such as plant-forward diets, may engage mTOR-mediated longevity mechanisms without requiring severe caloric deficit.

AMPK as a Direct Negative Regulator of mTORC1

Complementing the nutrient and growth factor inputs that activate mTORC1, a parallel energy-sensing axis provides potent inhibitory regulation. AMP-activated protein kinase (AMPK), the cell’s principal energy sensor, is activated when the AMP:ATP and ADP:ATP ratios rise during energetic stress, as occurs during exercise, fasting, and caloric restriction [52, 53]. AMPK is a conserved heterotrimeric complex whose γ subunits carry adenine nucleotide-binding CBS motifs that competitively sense AMP, ADP, and ATP; the primary upstream kinase is the tumour suppressor LKB1, which is constitutively active, meaning nucleotide binding to AMPK itself is the principal regulatory switch [54].

Once activated, AMPK suppresses mTORC1 through two mechanistically distinct and complementary routes. First, AMPK phosphorylates TSC2 at sites distinct from those targeted by AKT, but with the opposite functional consequence: AMPK-mediated phosphorylation activates the TSC1/TSC2 GAP complex, thereby accelerating Rheb inactivation and suppressing mTORC1 [53, 28]. Second, AMPK directly phosphorylates Raptor at Ser722 and Ser792, inducing 14-3-3 protein binding that disrupts the structural integrity of mTORC1 independently of the TSC axis [52, 53]. This dual mechanism ensures that energy deficit rapidly and robustly overrides nutrient and growth factor signals that would otherwise sustain anabolic output, establishing AMPK as a master brake on mTORC1 activity.

The significance of this AMPK-mTORC1 regulatory axis for longevity biology is substantial. Metformin, one of the most widely studied candidate geroprotectors, exerts its anti-ageing effects in part through AMPK activation — itself triggered by metformin’s inhibition of mitochondrial ATP synthesis and the consequent rise in cellular AMP/ADP levels — and consequent mTORC1 suppression [55, 56, 54]. AMPK also directly phosphorylates ULK1 and ULK2 at activating sites, providing a route to autophagy induction that operates in parallel with the relief of mTORC1-mediated ULK1 inhibition [53, 54]. Indeed, mouse hepatocytes deficient in either ULK1 or AMPK display marked mitochondrial accumulation and impaired mitophagy, underscoring the physiological importance of this pathway [54].

The convergence of AMPK and mTOR signalling on shared downstream targets — including ULK1/ULK2 and the TSC complex — positions these kinases as opposing arms of a nutrient-energy rheostat whose balance determines whether the cell commits to growth or maintenance, a theme that recurs across the longevity biology reviewed in subsequent sections.

Microbial Metabolites as Upstream Inputs to the AMPK-mTOR Axis

Beyond the cell-autonomous nutrient and energy inputs described previously, an increasingly recognized dimension of mTOR pathway regulation involves signals originating from the gut microbial ecosystem. Short-chain fatty acids (SCFAs)—principally butyrate, propionate, and acetate—produced by commensal bacteria during dietary fibre fermentation function as physiological AMPK activators. These microbial metabolites modulate mTORC1 activity through the same dual TSC2/Raptor phosphorylation mechanism engaged by canonical energy stress signals [22, 45, 44]. Butyrate in particular activates AMPK in colonocytes and, following systemic absorption, in hepatic and immune cell compartments. This activation attenuates mTORC1 output and promotes autophagic flux in a manner mechanistically analogous to metformin or caloric restriction [22, 57, 44]. Polyamines represent a second class of microbially derived mTOR modulators. Spermidine, produced by gut commensals including Prevotella species, has been independently established as a lifespan-extending autophagy inducer [23, 58, 20], and its production depends on gut microbial community composition.

This microbial input layer carries substantial implications for understanding age-related mTOR hyperactivation. Age-associated dysbiosis is characterized by progressive loss of SCFA-producing taxa and reduced microbial diversity. This decline diminishes the tonic AMPK-activating signals that counterbalance growth factor-driven mTORC1 stimulation [22, 59, 44]. Consequently, the AMPK-mTOR rheostat shifts toward sustained anabolic signalling—not because cell-autonomous nutrient sensing has failed, but because an ecological input to the system has been eroded. Dietary factors that sustain SCFA-producing communities, including adequate fibre intake and micronutrient sufficiency, may therefore modulate mTOR activity through a mechanism upstream of and complementary to direct pharmacological inhibition. Vitamin D sufficiency in particular correlates with greater microbial diversity and enrichment of butyrate-producing taxa such as Akkermansia muciniphila and Faecalibacterium prausnitzii [59]. This ecological dimension of mTOR regulation is not yet integrated into standard pathway models but is increasingly supported by evidence reviewed in subsequent sections on model organism lifespan studies and disease relevance.

Autophagy Regulation: The ULK1, VPS34, and TFEB Axes

mTORC1 exerts profound control over autophagy through three distinct but coordinated inhibitory mechanisms. First, active mTORC1 phosphorylates ULK1/2 at inhibitory sites (Ser757 in mammals), preventing the initiation of phagophore formation [19]. Second, mTORC1 phosphorylates and disrupts the VPS34 lipid kinase complex, which generates the phosphatidylinositol-3-phosphate (PI3P) required for autophagosome nucleation [19, 31]. Third, mTORC1 phosphorylates the transcription factor TFEB at Ser211, leading to its cytoplasmic sequestration via 14-3-3 binding and blocking transcriptional upregulation of lysosomal biogenesis genes [31, 3].

When mTORC1 is inhibited—whether by rapamycin, nutrient withdrawal, or energy stress—these three inhibitory mechanisms are released in a coordinated fashion. This simultaneous release enables both acute autophagosome formation and longer-term lysosomal expansion [31, 1]. The functional importance of this regulatory axis for longevity is underscored by evidence that lifespan extensions achieved through disrupted insulin-like signalling, caloric restriction [60], or pharmacological intervention [58] are abolished when core autophagy genes are genetically silenced. This finding establishes autophagy not merely as a correlate but as a mechanistic requirement for longevity-promoting mTOR inhibition [20]. Consistent with this mechanistic role, nuclear activation of TFEB’s C. elegans orthologue HLH-30 is itself sufficient to extend lifespan, directly linking the mTORC1–TFEB axis to longevity control [61].

Epigenetic Regulation: mTOR’s Influence on the Chromatin Landscape

Beyond its well-characterised roles in protein synthesis and autophagy, mTORC1 exerts increasingly recognised influence over the epigenetic machinery that governs gene expression and chromatin organisation [4, 3]. This regulatory dimension has become particularly consequential as the field has sought to explain how mTOR activity shapes the DNA methylation landscape that underpins epigenetic clock estimates of biological age.

Several mechanistic routes connect mTORC1 to epigenetic regulation. S6K1, the principal mTORC1 effector kinase, phosphorylates substrates involved in chromatin remodelling, and mTORC1 activity modulates the expression and activity of histone-modifying enzymes that govern the balance between activating and repressive chromatin marks [26, 1, 40]. Age-associated changes in histone modification—particularly the progressive loss of repressive marks such as H3K9me3 and H3K27me3 at heterochromatic loci—have been consistently documented across model organisms and human tissues [62], and these changes are coupled to the heterochromatin erosion that permits retrotransposon reactivation and genomic instability in aged cells [26, 27, 40]. Critically, mTORC1 hyperactivation in nutrient-replete conditions promotes a histone acetylation and methylation profile characteristic of aged chromatin, while mTORC1 inhibition by rapamycin partially preserves or restores repressive chromatin architecture [27, 4].

TFEB, whose cytoplasmic sequestration by mTORC1 suppresses lysosomal biogenesis as described above, also functions as a transcriptional regulator whose nuclear translocation upon mTOR inhibition reprogrammes gene expression in ways that extend beyond autophagy to encompass metabolic and stress-responsive loci [31]. This TFEB-mediated transcriptional reprogramming represents a route through which mTOR inhibition alters the broader transcriptional landscape of the cell, producing epigenomic consequences that are captured—at least in part—by the DNA methylation patterns upon which epigenetic clocks are trained [63, 64].

The connection between mTOR activity and DNA methylation maintenance is further supported by evidence that the metabolic substrates required for methyltransferase activity—particularly S-adenosylmethionine (SAM), generated through one-carbon metabolism—are themselves regulated by mTOR-dependent metabolic pathways [39, 3]. Because mTORC1 stimulates anabolic flux through pathways that draw on methionine cycle intermediates, its activity directly conditions the availability of the universal methyl donor upon which both DNA and histone methyltransferases depend [39, 4]. The writer-eraser-reader paradigm governing methylation dynamics, in which methyltransferases deposit marks, demethylases remove them, and methyl-binding proteins interpret them, depends on metabolic and signalling inputs that mTORC1 is well positioned to coordinate [39]. Age-related disruption of this regulatory balance—manifest as the bidirectional pattern of global hypomethylation at repetitive elements and focal hypermethylation at CpG island promoters [25, 63]—may thus reflect, in part, the chronic mTORC1 hyperactivation characteristic of aged and calorically replete tissues [26, 27]. This mechanistic framework positions mTOR not only as a regulator of protein synthesis and autophagic flux but as an upstream determinant of the epigenetic drift that accumulates with age and that is quantified by the biological age biomarkers now entering clinical application.

FKBP12 and FKBP51 as Pharmacodynamic Determinants

Rapamycin does not inhibit mTOR directly; it binds the cytosolic chaperone FKBP12, and the resulting drug–protein complex then docks onto the FKBP12-rapamycin binding domain of mTOR, sterically disrupting Raptor association and thus mTORC1 activity [14, 48, 28]. While acute rapamycin treatment is largely mTORC1-selective, prolonged exposure can also disrupt mTORC2 assembly and attenuate Akt/PKB phosphorylation in a cell-type-dependent manner [6, 7]. The basis for this variable mTORC2 sensitivity across cell types was unclear until it was demonstrated that a closely related immunophilin, FKBP51, competes with FKBP12 for rapamycin binding [14]. Across five cell lines spanning prostate cancer to muscle to kidney epithelium, the ratio of FKBP12 to FKBP51 expression predicted mTORC2 inhibition by rapamycin with remarkable precision (R² = 0.8962), and knockdown of FKBP12 in otherwise rapamycin-responsive cells abolished mTORC2 inhibition entirely without affecting mTORC1 [14]. This finding reframes inter-tissue and inter-individual variability in rapamycin response as, at least partly, a function of immunophilin stoichiometry—a mechanistic insight with direct implications for interpreting the complex metabolic phenotypes reported in chronic rapamycin studies, including the hepatic insulin resistance and glucose intolerance now attributed specifically to mTORC2 disruption rather than mTORC1 inhibition [50, 1]. It also motivates the development of mTORC1-selective analogs designed to preserve mTORC2-dependent Akt signaling and thereby decouple longevity benefit from metabolic liability [65, 66].

Tissue-Specific mTOR Functions and Metabolic Consequences

mTOR signalling produces profoundly different physiological outputs depending on tissue context. In skeletal muscle, mTORC1 activity must be carefully titrated: insufficient activity impairs protein synthesis and produces atrophy, while excessive activity paradoxically causes myopathy through a distinct mechanism, demonstrating that the pathway’s anabolic function operates within a narrow optimal range [1]. In the liver, mTORC2—via AKT Ser473 phosphorylation—is essential for insulin-mediated suppression of gluconeogenesis, which explains the finding that chronic rapamycin induces hepatic insulin resistance primarily through mTORC2 disruption rather than through mTORC1 inhibition per se; liver-specific Rictor knockout mice phenocopy this glucose intolerance, decoupling it from effects on longevity pathways [50, 48]. In the immune system, mTORC1 hyperactivation drives pro-inflammatory T cell subsets and is measurably elevated prior to disease flares in systemic lupus erythematosus, while rapamycin selectively suppresses these cells while expanding regulatory T cell populations [32]. Importantly, long-term rapamycin treatment in aging mice functions more as an immune modulator than a classical immunosuppressant: it reduces exhausted PD-1+ T cells, preserves naïve B and T cell phenotypes, and shifts T cells from memory toward naïve phenotypes—effects distributed across approximately 2,000–3,000 differentially regulated genes per immune cell population [67]. These tissue-specific findings collectively argue against treating mTOR as a monolithic target and instead demand a nuanced appreciation of which complex, in which cell type, is mediating each biological outcome. Emerging evidence further underscores that sex-dependent differences in nutrient-sensing pathway architecture add an additional layer of complexity: in Drosophila and mice alike, rapamycin extends lifespan in females but not males, a dimorphism rooted in cell-autonomous enterocyte sexual identity that governs basal autophagy flux through the histone H3/H4–Bchs axis, and which is conserved across multiple mouse tissues including the intestine, brown adipose tissue, and skeletal muscle [68, 69].

Evolutionary Conservation of TOR Signalling: Grounding the Genetic Toolkit

The conservation of TOR signalling across eukaryotes is not merely a phylogenetic curiosity but a mechanistic foundation that justifies the experimental strategies used to dissect pathway function. In budding yeast, Drosophila, and Caenorhabditis elegans, TOR orthologs regulate growth, nutrient sensing, and lifespan through recognisably homologous molecular mechanisms—homologous enough that genetic manipulations in these organisms reliably illuminate the causal architecture of pathways that operate in mammalian cells [4, 20, 2]. This mechanistic fidelity is what licenses the inference from invertebrate genetic experiments—loss-of-function screens, tissue-specific knockouts, epistasis analyses with autophagy genes—to mammalian pathway biology. In Drosophila, TOR signalling intersects directly with growth and ageing programmes in a manner that closely parallels mammalian biology [9], while in C. elegans, rapamycin-mediated TOR inhibition extends lifespan partly through SKN-1/Nrf and DAF-16/FoxO transcription factors [10], illustrating the depth of mechanistic conservation at the level of downstream effectors. The reach of this conservation extends even to photosynthetic microeukaryotes: microalgae employ TOR to promote protein synthesis via S6K-RPS6 phosphorylation while suppressing autophagy under nutrient-rich conditions, and variable rapamycin sensitivity across algal species maps to differences in FKBP12 affinity rather than to structural divergence in TOR itself [70]. The implication is that the core regulatory logic—TOR as an anabolic-catabolic switch calibrated by nutrient and energy inputs—has been under strong selective constraint across an enormous evolutionary span, reflecting its indispensable role in coordinating cellular growth decisions [3]. For the purposes of the present review, this conservation means that model organism evidence reviewed in Section 4 can be read not merely as comparative biology but as mechanistic evidence bearing directly on the molecular targets engaged by rapamycin in human tissues.

Taken together, the mechanistic picture that has solidified over the past decade situates mTOR not as a simple growth-promoting kinase but as a multi-layered integration hub whose outputs—anabolic, catabolic, immune, and epigenomic—are context-dependent, complex-specific, and strongly modulated by the pharmacodynamic microenvironment in which rapamycin operates. Critically, mTORC1 does not operate in signalling isolation: its activity is continuously modulated by opposing inputs from AMPK and growth factor pathways [53, 16], as well as by ecological inputs from the gut microbial community that feed into AMPK activation [44, 71], embedding it within a broader nutrient-energy-ecology sensing network whose coordinated regulation is essential for understanding both the therapeutic promise and the metabolic liabilities of mTOR-targeted longevity interventions. The sections that follow examine how this molecular architecture plays out across biological systems and clinical contexts—from lifespan effects in model organisms and the growth-longevity trade-offs they illuminate (Sections 4 and 5), through the disease-relevant consequences of mTOR dysregulation (Section 6), to the epigenetic and microbiome dimensions of pathway function (Sections 7 and 8)—building toward an integrated account of rapamycin’s potential as a human geroprotector.

4. Rapamycin and Lifespan Extension in Model Organisms

The pharmacological inhibition of mechanistic target of rapamycin (mTOR) signalling has emerged as one of the most reproducible and cross-species-validated strategies for extending lifespan in laboratory organisms. From unicellular yeast to inbred and outbred mice, genetic or pharmacological suppression of mTORC1 activity consistently delays age-associated pathology and extends both mean and maximum lifespan [2]. What began as a mechanistic curiosity surrounding a soil bacterium-derived macrolide has, over the past two decades, matured into a rigorous experimental programme encompassing timing, dosing, sex stratification, combinatorial strategies, and large-scale cross-species screening.

Landmark Evidence and the Late-Life Intervention Paradigm

The pivotal contribution establishing rapamycin as a bona fide geroprotector in mammals came from the National Institute on Aging’s Interventions Testing Program (ITP), which reported in 2009 that feeding rapamycin to genetically heterogeneous mice beginning at 600 days of age—roughly equivalent to 60 human years—significantly extended lifespan [8]. Based on age at 90% mortality, females gained approximately 14% and males approximately 9% in maximal lifespan, a finding replicated independently across three sites. This result carried enormous conceptual weight: it demonstrated that pharmacological slowing of aging remained possible even after the majority of the lifespan had elapsed, challenging any assumption that an effective intervention window was restricted to early life [7]. Notably, subsequent ITP work demonstrated that initiating rapamycin treatment at an earlier age (9 months) yielded still greater lifespan gains, underscoring that the late-life efficacy observed in the original study represented a floor rather than a ceiling [72]. Early review syntheses quickly situated this finding within a broader cross-species framework, noting that mTOR inhibition extended lifespan in yeast, Caenorhabditis elegans, Drosophila melanogaster [9], and mice, often when initiated late in life, and that these effects co-occurred with delays in cancer incidence, cognitive decline, and cardiac aging [2].

Theoretical work in parallel argued that rapamycin’s lifespan-extending effects operate by decelerating the aging process itself rather than through aging-independent mechanisms such as cancer suppression alone [12]. This distinction matters for translational reasoning: if rapamycin merely prevented one lethal age-related disease, the benefit would be narrow; if it genuinely slows biological aging, the effects should manifest across multiple organ systems and pathologies simultaneously. Subsequent empirical reviews lent support to the broader interpretation, cataloguing rapamycin’s antineoplastic effects across diverse cancer models while also documenting protective effects in cardiovascular, neurological, and immunological aging [11] [47]. More recent characterisation of rapalogs as broad-spectrum therapeutics for age-related diseases has further reinforced this multi-system framing, emphasising that mTORC1 inhibition simultaneously attenuates the senescence-associated secretory phenotype, reduces stem cell exhaustion, and modulates mitochondrial dysfunction—effectively targeting multiple hallmarks of ageing through a single upstream node [73] [4].

Sex Differences in Lifespan Response

A persistent and biologically informative finding across ITP studies is that female mice frequently exhibit larger lifespan gains from rapamycin than males, a pattern echoed in the original 2009 report and confirmed in subsequent analyses [8, 11]. This asymmetry is not trivial in magnitude: female mice in several ITP cohorts have shown median lifespan extensions roughly double those observed in males at equivalent doses. The mechanistic basis for this sexual dimorphism remains incompletely resolved but may involve differential baseline mTORC1 activity, hormonal modulation of mTOR signalling, or sex-specific differences in drug pharmacokinetics [7]—a concern given that even within the same formulation, blood rapamycin levels display marked inter-individual variability that does not consistently track with sex, BMI, or duration of use [74]. Notably, intermittent rapamycin administration in aged female C57BL/6J mice has also been shown to significantly extend both median and maximum lifespan while preserving glucose and insulin tolerance, suggesting that dosing regimen interacts with sex in ways relevant to both efficacy and safety [29]. Recent work on longevity-associated enzymes has reinforced the broader significance of sexual dimorphism in nutrient-sensing pathways, demonstrating that even genetically identical interventions can produce distinct hepatic metabolic reprogramming in males and females through sex-dependent subcellular localisation and inter-organ signalling mechanisms [69]. Understanding this asymmetry is not merely a biological curiosity—it has direct implications for dose selection and risk-benefit calculations in any eventual human application.

Dosing Regimen Optimisation: Intermittent Administration

A critical practical question emerging from early continuous-dosing studies was whether the metabolic side effects of chronic rapamycin—including glucose intolerance and hyperlipidaemia—could be mitigated without sacrificing lifespan benefit. Mechanistically, these metabolic harms are now understood to arise not from mTORC1 inhibition per se, but from chronic rapamycin disrupting mTORC2 assembly, which impairs hepatic insulin signalling and drives gluconeogenesis [50]; chronic exposure also produces hyperlipidaemia and new-onset diabetes-like phenotypes [7]. This motivated investigations into intermittent dosing strategies. A 2016 study demonstrated that administering rapamycin at 2 mg/kg every five days, beginning at 20 months of age in female C57BL/6J mice, significantly extended both mean and maximum lifespan, with 29% of treated mice outliving the longest-lived vehicle controls [29]. This finding suggested that pulsatile mTOR inhibition could recapitulate or even optimise the longevity benefit of continuous dosing while potentially reducing adverse metabolic consequences—a distinction with obvious translational relevance given concerns about chronic immunosuppression in aging human populations [35].

Combination Geroprotector Strategies

A natural extension of single-agent efficacy studies is the question of whether combining geroprotectors with complementary or overlapping mechanisms produces additive or synergistic lifespan extension. The most rigorously characterised combination to date is rapamycin plus trametinib, an MEK inhibitor that independently targets the RAS-ERK pathway—a parallel pro-growth and pro-survival cascade that, like mTOR, is aberrantly active in aged tissues [2]. A 2025 study using C3B6F1 hybrid mice found that trametinib alone extended median lifespan by 7% in females and 10% in males, while rapamycin alone produced approximately 17% extension in both sexes [75]. Crucially, the combination extended lifespan additively—approximately 35% in females and 27% in males—with no statistical interaction indicating either synergy or antagonism. This additive profile suggests the two compounds act through sufficiently distinct mechanisms that their benefits sum without interference, and it reinforces the biological rationale for multi-target geroprotective strategies [15]. Notably, the combination also produced additive improvements in healthspan measures, including grip strength and rotarod performance, partially dissociating the lifespan and healthspan endpoints that do not always move in parallel [75]. The success of this multi-node approach is consistent with the emerging understanding that longevity-associated pathways—including the mTOR, RAS-ERK, AMPK, and sirtuin axes—form an interconnected signalling network rather than isolated linear cascades [4, 2], such that engaging distinct nodes simultaneously can capture non-redundant benefits that single-target interventions cannot achieve alone [55, 53]. This principle is further supported by findings from systematic multi-compound screening programmes, which have consistently identified that agents with distinct molecular targets yield the most robust additive effects when combined [76].

Cross-Species Screening Programmes and Reproducibility

The concern that lifespan findings from inbred mouse strains might not generalise has driven investment in multi-strain, multi-laboratory screening infrastructure. The mammalian Intervention Testing Program (ITP), conducted across three independent NIA-funded sites using genetically heterogeneous mice, established the foundational template for this approach — most notably demonstrating that rapamycin initiated as late as 600 days of age (roughly equivalent to 60 human years) extended maximal lifespan by 14% in females and 9% in males, with survival gains from the point of first treatment reaching 38% and 28%, respectively [8]. The C. elegans Intervention Testing Program (CITP) was developed as a direct invertebrate counterpart to this model, testing 77 compounds across 730,731 animal assays and identifying 27 with statistically significant lifespan extension, including 12 that reproducibly extended median survival by 20% or more across multiple strains and laboratories [76]. Critically, variance partitioning revealed that laboratory differences accounted for only approximately 2% of total lifespan variation, while genetic background accounted for roughly 25%, underscoring both the robustness of the screening methodology and the importance of genetic diversity in assessing intervention generalisability [76]. Rapamycin and rapalogs featured prominently among the reproducibly effective compounds in both programmes, reinforcing the cross-species validity of mTOR inhibition as a longevity target [8, 7, 2].

Microbiome Dependence and Independence of Rapamycin’s Lifespan Effects

A potentially confounding variable in lifespan studies conducted in conventionally housed organisms is the gut microbiota, which independently modulates host physiology and lifespan [77, 45]. The question of whether rapamycin’s longevity benefits are mediated through, or independent of, effects on the microbial community has been directly addressed in Drosophila using germ-free experimental designs. Rapamycin extends lifespan and preserves intestinal barrier integrity equivalently in conventionally reared and germ-free flies, establishing that the drug’s pro-longevity effects are entirely microbiota-independent in this organism [30]. Critically, genetic inhibition of the autophagy gene Atg1 abolished both the lifespan extension and the barrier protection conferred by rapamycin, isolating the mechanism to cell-autonomous autophagy induction in intestinal epithelial cells rather than to any microbiome-mediated pathway [30]. This clean dissociation provides powerful evidence that rapamycin’s core geroprotective action operates through host cell-intrinsic mTOR-autophagy signalling.

In mammalian systems, however, the relationship between rapamycin and the gut microbiome is more complex. Chronic rapamycin administration in mice substantially remodels both immune cell composition—reducing exhausted PD-1⁺ T cells, shifting T cell repertoires toward naïve phenotypes, and altering myeloid and innate lymphoid cell proportions—and the gut metagenome, although the metagenomic changes were modest relative to the immune restructuring [67]. The finding that encapsulated rapamycin extended lifespan even in RAG2⁻/⁻ and IFN-γ⁻/⁻ immunodeficient mice indicates that at least some longevity benefits are immune-independent, consistent with the Drosophila evidence for cell-autonomous action [67]. Nevertheless, the concurrent immune and microbial remodelling observed in immunocompetent mice raises the possibility that rapamycin’s effects in mammals involve both direct mTOR inhibition and indirect ecological consequences. Whether these microbiome changes augment, partially counteract, or are functionally neutral with respect to lifespan extension has not been resolved, and mammalian germ-free rapamycin studies equivalent to the Drosophila work remain a critical experimental priority.

A complementary line of evidence demonstrates that the gut microbial community can itself be deliberately manipulated to extend lifespan through pathways that converge on mTOR-relevant biology. Administration of the probiotic Bifidobacterium animalis subsp. lactis LKM512 to aged female mice significantly prolonged survival, with the mechanism traced to microbial community reorganisation that elevated faecal spermidine concentrations and suppressed colonic cellular senescence [23]. Spermidine is an established autophagy inducer that extends lifespan across yeast, nematodes, and flies through autophagy-dependent mechanisms—genetic inhibition of autophagy genes abolishes its longevity effects—and operates via histone acetylation changes that are mechanistically distinct from, yet functionally convergent with, mTOR inhibition [58, 20]. This probiotic-mediated lifespan extension positions the gut microbiome not merely as a confound in longevity studies but as a tractable target whose manipulation can engage the same autophagic quality-control programmes activated by rapamycin, albeit through an ecological rather than pharmacological route.

Dissociation of Lifespan and Healthspan Endpoints

A recurring and increasingly important theme across this literature is the degree to which lifespan extension is accompanied by genuine improvements in healthspan—the period of life spent in good physiological condition. Early reviews tended to treat these endpoints as jointly achieved by rapamycin [2, 7], and many preclinical studies support this view [12, 5]. However, a systematic review of rapalog studies in humans noted that benefits were demonstrable in immune, cardiovascular, and integumentary systems but absent in endocrine, muscular, and neurological systems, cautioning against uncritical extrapolation from lifespan to comprehensive healthspan [35]. This organ-system specificity is echoed in human clinical data: the 48-week PEARL randomized controlled trial found that low-dose weekly rapamycin (10 mg) produced sex-specific and domain-specific improvements—including significant gains in lean tissue mass in women (~4.5%) and bone mineral content in men (~1.4%)—while leaving metabolic and neurological markers broadly unchanged [46]. More recent preclinical work addresses this directly by measuring both endpoints in the same experiment, finding that combination rapamycin-trametinib treatment extends functional healthspan measures additively alongside lifespan [75]. Collectively, this body of evidence argues that lifespan and healthspan, while correlated, must be measured independently and that intervention strategies should be evaluated on both dimensions before clinical translation is contemplated.

5. Clinical Translation: Human Trials and Safety of Rapamycin and Rapalogs

The translation of rapamycin and its derivatives from model organisms to human clinical settings has emerged as one of the most active and consequential fronts in geroscience. Early mechanistic work established that rapamycin’s geroprotective effects operate primarily through mTORC1 inhibition, influencing autophagy, senescence-associated secretory phenotype suppression, and epigenetic regulation, while adverse metabolic consequences were increasingly attributed to off-target mTORC2 inhibition [5, 47]. This mechanistic bifurcation created a conceptual framework for human dosing strategies: if the therapeutic benefits and the adverse effects arise from distinct molecular targets, then lower or intermittent dosing might preferentially engage mTORC1 inhibition while sparing mTORC2. Translating this hypothesis into rigorous human evidence has required traversing substantial methodological terrain, from early immunological proof-of-concept trials through pharmacokinetic characterisation to organ-system-specific efficacy studies.

Immune Rejuvenation and Early Proof-of-Concept

The earliest human signals emerged from studies of rapalogs in older adults prior to influenza vaccination. Low-dose or intermittent everolimus improved immune responses to influenza vaccination in older adults, establishing immunological rejuvenation as a tractable clinical endpoint [47]. Importantly, these studies also revealed a dose-response ceiling: higher doses produced adverse effects without providing additional immunological benefit, an early indication that therapeutic windows for geroprotection may be narrow. This work with everolimus and later RTB101 — a selective catalytic mTOR inhibitor that, unlike rapalogs, blocks mTORC1 kinase activity directly — provided foundational evidence that targeted mTOR inhibition could modulate ageing-associated immune decline in humans [5]. Mechanistically, short courses of rapalogs were shown to restore hematopoietic stem cell function, increase naïve lymphocyte production, and upregulate antiviral interferon pathways, reducing laboratory-confirmed respiratory infections without the severe immunosuppression characteristic of transplant-dose regimens [5]. The adverse metabolic effects seen at higher doses — including hyperlipidaemia and hyperglycaemia — are now understood to arise primarily from off-target mTORC2 inhibition, reinforcing the rationale for more selective compounds [65]. The systematic literature synthesising this period identified meaningful improvements in immune system parameters alongside early signals of cardiovascular and integumentary benefit, while noting that endocrine, muscular, and neurological systems showed no significant effects across the available studies [35]. This asymmetry of organ-system effects, confirmed across nineteen studies drawn from over eighteen thousand candidate articles, represents a finding of considerable clinical significance, suggesting that rapalog-mediated benefits in humans are system-specific rather than globally distributed.

The PEARL Trial: Randomised Evidence in Healthy Adults

The publication of the PEARL trial results in 2024 marked a pivotal advance in the clinical evidence base [46]. This forty-eight-week double-blind randomised placebo-controlled trial enrolled one hundred and fifteen participants aged fifty to eighty-five years, examining low-dose intermittent rapamycin across body composition and healthspan metrics. The most striking finding was a pronounced sex-specificity in the benefit profile. Females taking ten milligrams per week demonstrated an average 4.5% increase in lean tissue mass, with all women at this dose showing net improvement across body composition measures with no declines. Males taking the same dose showed significant improvements in bone mineral content averaging 1.4%, but displayed more heterogeneous responses overall. This sex-differential pattern is consistent with preclinical evidence that intermittent rapamycin preferentially extends lifespan in female mice [29], and challenges any assumption that rapamycin’s effects in humans are uniform. It underscores the need to stratify future trial designs and dosing protocols by biological sex [47]. The broader mechanistic basis for such dimorphism may relate to baseline differences in mTOR pathway activity between sexes [1], though the precise hormonal and cellular mediators in humans remain to be fully characterised. The absence of serious adverse events in this healthy cohort [46] aligns with the broader systematic finding that serious adverse events are not reported in healthy individuals receiving rapalogs, though those with pre-existing ageing-related diseases face elevated risks including increased infection rates and raised cholesterol and triglyceride levels [35]. These safety observations reinforce the importance of participant selection in designing rapamycin trials targeting healthspan [5].

Pharmacokinetics and Formulation Fidelity

A foundational but frequently underappreciated dimension of clinical translation concerns pharmacokinetics and formulation quality. A 2025 two-cohort study characterising blood rapamycin levels in real-world ageing populations revealed that compounded rapamycin demonstrated bioavailability of approximately 31% relative to commercial formulations [74]. An independent potency analysis further found that compounded capsules contained on average 26% less rapamycin per milligram than the commercial product. These findings carry direct implications for the interpretation of studies and self-experimentation using compounded preparations, which have become increasingly common in longevity-oriented populations — with estimates suggesting over 2,000 Americans were already taking rapamycin off-label prior to formal trial activity [47]. Inter-individual variability in blood levels further complicates dose standardisation, since participants receiving nominally identical doses achieve substantially different systemic exposures [74, 35]. This pharmacokinetic heterogeneity means that trials reporting dose in milligrams without blood level verification may be measuring effects across a wide and uncharacterised range of actual mTOR inhibition — a concern underscored by the systematic observation that no clear dose-response relationship has yet been established in human rapamycin trials [35]. The PEARL trial, which used compounded rapamycin throughout its 48-week intervention, subsequently determined that participants had effectively received doses approximately 3.5-fold lower than intended, rendering observed healthspan benefits more notable in retrospect but also highlighting how formulation inconsistency can silently confound trial interpretation [46]. The 2025 data from Harinath and colleagues [74] thus introduce a methodological imperative: future trials should monitor trough blood levels alongside nominal dosing, and incorporate pharmacodynamic measures of mTOR engagement [35], to enable meaningful dose-response analyses.

Disease-Targeted Protocols: Alzheimer’s Disease

Beyond healthy ageing, the ERAP phase IIa clinical study protocol represents the most advanced disease-specific translation effort to date [78]. This single-centre open-label trial targets patients with early-stage Alzheimer’s disease, drawing on preclinical evidence that rapamycin extends murine lifespan by 9–14% based on age at 90% mortality — with females gaining approximately 14% and males 9% in the landmark Harrison et al. ITP study [8] — and delays pathological features of AD including amyloid-beta aggregation, tau pathology, neuroinflammation, and blood-brain barrier breakdown [11]. The protocol employs in vivo imaging to track disease modification, positioning rapamycin as a repurposed geroprotective agent rather than a conventional symptomatic treatment. While results are awaited, the trial design itself illustrates a broader conceptual shift: the hypothesis is not that rapamycin reverses established neurodegeneration but that mTORC1 inhibition initiated at early disease stages can slow the trajectory of ageing-related pathology [78, 13]. Preclinical neuroprotective effects in both Alzheimer’s and epilepsy models have been documented [13, 11], providing biological plausibility, though the systematic review evidence base currently shows no significant rapalog effects on neurological parameters in human studies [35], indicating that this remains an unresolved empirical question.

Biological Age Biomarkers as Emerging Clinical Endpoints

A development of considerable methodological and strategic significance for the clinical translation of rapamycin is the maturation of biological age biomarkers—particularly DNA methylation-based epigenetic clocks—into tools of sufficient resolution to serve as candidate endpoints in geroprotector trials. The fundamental challenge confronting rapamycin’s clinical development as a longevity intervention is that the outcome of ultimate interest—extension of healthy lifespan—cannot be measured within the timeframe of conventional clinical trials. Surrogate endpoints that quantify the rate or extent of biological ageing are therefore essential, and epigenetic clocks represent the most developed class of such surrogates.

First-generation epigenetic clocks, including the Horvath pan-tissue clock—which identified 353 CpG sites predictive of chronological age across diverse tissue types [63]—and the Hannum blood-based clock, which applied genome-wide methylation profiling to whole blood [64], established that DNA methylation patterns at specific CpG sites could reliably predict chronological age. However, critical appraisal revealed a fundamental limitation: clocks calibrated to predict chronological age lose their capacity to differentiate individuals on the basis of biological ageing trajectories, since perfect correlation with chronological age is a methodological liability rather than a virtue for intervention assessment [24, 25]. Second-generation clocks addressed this by calibrating against mortality and morbidity outcomes rather than chronological age. GrimAge—originally a composite of eight DNA methylation surrogates for plasma proteins including PAI-1 and GDF-15, plus a smoking-pack-years proxy, all calibrated to time-to-death [79]—and its successor GrimAge2, which further incorporates methylation surrogates for high-sensitivity CRP and HbA1c and was validated across nearly 13,400 samples from nine independent cohorts [36], along with PhenoAge, demonstrate substantially stronger associations with all-cause mortality and age-related disease incidence than their predecessors [79]. The concept of epigenetic age acceleration—the residual difference between estimated epigenetic age and chronological age—has become a central analytical construct, with accelerated ageing associating robustly with mortality risk across multiple cohorts [25].

A conceptually distinct third generation of tools has moved beyond cross-sectional age estimation to capture the rate at which an individual is ageing. DunedinPACE, derived from the Dunedin birth cohort of 1,037 individuals followed from birth through age 45 and trained on longitudinal change across 19 biomarkers spanning cardiovascular, metabolic, renal, hepatic, and pulmonary systems, produces a 173-CpG scalar estimate of the pace of biological ageing—with cohort-observed variation ranging from 0.40 to 2.44 biological years per chronological year—that can be tracked within individuals over time [37, 24]. Its test-retest reliability (ICC = 0.96) substantially exceeds that of most predecessor clocks, and it retains incremental predictive value for mortality, cardiovascular disease, and disability even after adjustment for GrimAge [37]. This distinction between biological age and rate of ageing carries direct consequences for trial design: detecting a slowing in the pace of ageing may be both more biologically meaningful and more statistically tractable than demonstrating a reduction in estimated biological age from a single timepoint measurement.

Sex differences in epigenetic ageing rates have emerged as a consistent finding across clock platforms, with men systematically showing greater epigenetic age acceleration than women [25]—an observation that resonates with the sex-differential responses to rapamycin observed in both preclinical lifespan studies and the PEARL trial [8, 46]. Non-linearity in clock performance across the lifespan represents a further methodological challenge: clocks trained predominantly on middle-aged adult cohorts may lose accuracy in older age groups, precisely the population most relevant to longevity medicine [25]. A persistent European-ancestry bias in clock training datasets compounds these concerns, limiting cross-population generalisability [25, 24].

The most rigorous validation framework for biological age biomarkers as clinical endpoints requires demonstration of correlation with chronological age, association with age-related disease and mortality, responsiveness to interventions known to modulate ageing, and generalisability across populations, sexes, and age strata [24]. Expert consensus has further proposed that no single biomarker can capture ageing’s full complexity, and that robust inference requires multi-modal panels integrating epigenetic clocks with proteomic, metabolomic, functional, and digital biomarker modalities [24, 79]. Blood-based composite SASP profiles may provide complementary information about systemic senescence burden that methylation-based clocks do not fully encode [79], reinforcing the case for composite endpoint strategies. For rapamycin trials specifically, the incorporation of epigenetic clocks as pre-specified primary or co-primary endpoints—alongside pharmacokinetic monitoring and organ-system-specific functional measures—would substantially improve the interpretability and comparability of clinical findings across the expanding trial landscape.

Safety Profile and Outstanding Concerns

Across the clinical literature, the safety profile of low-dose intermittent rapamycin in healthy adults appears substantially more benign than the profile established in transplant populations receiving continuous high-dose immunosuppression [47, 35]. Nevertheless, chronic mTOR inhibition carries documented risks — including immunosuppression, metabolic disturbances such as insulin resistance and glucose intolerance arising from inadvertent mTORC2 suppression, impaired wound healing, and hormonal disruption — that must be weighed against geroprotective benefits, particularly when considering long-term administration in healthy populations [13, 7, 50]. The elevated infection risk and dyslipidaemia observed specifically in disease cohorts [35] suggest that baseline health status is a critical moderator of the risk-benefit calculus. Taken together, the convergence of trial evidence, systematic review, and pharmacokinetic characterisation arriving between 2023 and 2025 has substantially sharpened the human clinical picture: low-dose intermittent rapalogs appear safe in healthy older adults with meaningful but organ-system-specific and sex-differentiated efficacy — exemplified by the PEARL trial’s finding that women receiving 10 mg/week gained an average 4.5% lean muscle mass while men in the same dose group achieved approximately 1.4% gains in bone mineral content [46] — while formulation standardisation and blood-level monitoring remain essential prerequisites for reliable clinical inference, as compounded rapamycin has been shown to be approximately 3.5-fold less bioavailable than commercial preparations [74, 46, 35].

6. mTOR Modulation, Autophagy, and Cellular Quality Control in Ageing

The question of how cells maintain functional integrity across decades of biological time has emerged as one of the most productive areas of ageing research. At its centre sits the mechanistic target of rapamycin (mTOR), a serine/threonine kinase that integrates nutrient availability, energy status, and growth factor signals to orchestrate anabolic and catabolic cellular programmes. Early work established that mTOR inhibition could extend lifespan across taxonomically distant organisms [4, 11], but the more consequential insight—still being fully worked out—is that this extension depends not merely on slowing growth, but on activating a suite of downstream quality-control mechanisms: autophagy, mitochondrial surveillance, proteostasis maintenance, and suppression of the senescence-associated secretory phenotype (SASP). The causal architecture linking mTOR to these processes, and in turn to organismal longevity, forms the subject of this section.

mTOR as a Master Regulator of Cellular Catabolism

mTOR operates through two structurally and functionally distinct complexes. mTORC1, the primary longevity-relevant complex, is recruited to the lysosomal surface via Rag GTPases, where its activation requires the convergence of amino acid sensing and insulin signalling [1]. In this lysosomal context, mTORC1 exerts broad suppressive control over catabolic pathways. Mechanistically, mTORC1 inhibits autophagy at multiple nodes: it phosphorylates and inactivates the ULK1/2 initiation complex—directly phosphorylating both ULK1/2 and Atg13 to suppress kinase activity—suppresses the VPS34 lipid kinase complex required for autophagosome nucleation, and prevents nuclear translocation of TFEB, the master transcription factor governing lysosomal biogenesis and autophagic gene expression [31, 19]. This regulatory axis is broadly conserved across eukaryotes, with the yeast Atg1–Atg13–TOR module representing the ancestral counterpart to the mammalian ULK1/2 complex [19]. This multi-point suppression means that even partial mTORC1 inhibition—as achieved by rapamycin, whose incomplete blockade of mTORC1 substrates and susceptibility to compensatory feedback loops limits its efficacy as a catalytic inhibitor [48]—is sufficient to derepress autophagy and initiate cellular recycling programmes. The pharmacological tractability of this axis, and the evolutionary conservation of both mTOR and autophagy machinery from yeast to humans, made it an early and attractive target for longevity research [4].

Autophagy as a Causal Requirement for Lifespan Extension

A critical conceptual shift occurred when studies moved beyond correlating autophagy induction with longevity to testing whether it is mechanistically required. Synthesising evidence across multiple model organisms, [20] demonstrated that lifespan extensions achieved through disrupted insulin-like signalling, caloric restriction, calcineurin deficiency, and pharmacological interventions are abolished when core autophagy genes—including ATG5, ATG7, and beclin-1—are knocked down or deleted. Overexpression of autophagy components, particularly ATG5 and the LC3/ATG8 family, is itself sufficient to extend lifespan in several experimental systems [20, 58]. This evidence repositioned autophagy from a biomarker of healthy ageing to a causal effector of it [20]. The implication is substantial: interventions targeting the mTOR-autophagy axis derive their lifespan-extending effects through the activation of cellular clearance, not simply through the inhibition of biosynthetic expenditure.

This causal claim has received direct corroboration in the Drosophila system, where genetic inhibition of Atg1 in rapamycin-treated flies abolished both the lifespan extension and the improvements in intestinal barrier function that the drug otherwise confers, while leaving its mTOR-inhibitory activity nominally intact [30]. The convergence between the cross-species genetic evidence assembled by Madeo and colleagues [20] and these single-organism epistasis experiments arrives at the same conclusion: autophagy induction lies on the obligate causal pathway through which mTOR inhibition extends lifespan, rather than functioning as a parallel or dispensable effector. This mechanistic specificity carries direct translational implications, since it predicts that co-interventions or physiological states that inadvertently suppress autophagy—whether through dietary composition, drug interactions, or tissue-specific pathway silencing—could negate the intended benefits of rapamycin-based regimens.

Caloric restriction (CR) serves as the paradigmatic demonstration of this principle. CR robustly reduces mTORC1 activity through diminished amino acid and insulin signalling, thereby derepressing autophagy, improving proteostasis, and reducing the accumulation of damaged macromolecules and protein aggregates—cargo that autophagy is uniquely equipped to clear due to its capacity to engulf large oligomeric structures beyond the reach of the ubiquitin-proteasome system [21, 4]. The overlap between CR-mimetic interventions and autophagy induction is not incidental—it reflects the shared mechanistic architecture through which nutrient scarcity signals translate into cellular maintenance programmes. Indeed, multiple longevity-promoting pathways, including reduced insulin/IGF-1 signalling and dietary restriction, converge on the activation of autophagic flux through transcription factors such as TFEB and FOXO [21, 3]. That pharmacological mTOR inhibitors recapitulate many lifespan effects of CR [11, 4] further supports the view that the mTOR-autophagy axis is the central molecular transducer of dietary restriction’s longevity benefits.

Sirtuin-Mediated Autophagy Regulation and NAD⁺-Dependent Quality Control

While mTORC1 inhibition and its downstream relief of autophagy suppression constitute the most extensively characterised route to cellular quality control in ageing, a parallel and partially convergent regulatory axis operates through the NAD⁺-dependent deacetylase SIRT1. The connection between sirtuins and autophagy was established when resveratrol, a polyphenolic SIRT1 activator, was shown to extend lifespan in C. elegans through a mechanism requiring both SIRT1 activity and intact autophagy genes—genetic disruption of either abolished the longevity benefit [58]. Spermidine, another pharmacological lifespan-extending compound, similarly requires functional autophagy machinery for its effects, though it acts through a distinct epigenetic mechanism involving histone hypoacetylation [58, 20]. These convergent findings—that structurally and mechanistically diverse longevity compounds each depend on autophagy as a final common effector—powerfully reinforced the centrality of autophagic flux to lifespan regulation [20].

The mechanistic link between SIRT1 and autophagy operates at multiple levels. SIRT1 directly deacetylates core autophagy proteins, including ATG5, ATG7, and LC3, enhancing their functional activity and promoting autophagosome formation [80, 81, 82]. This post-translational regulation provides a route to autophagy induction that is independent of, but synergistic with, mTORC1 inhibition: whereas rapamycin derepresses autophagy by removing inhibitory phosphorylation from ULK1, VPS34, and TFEB, SIRT1 enhances the catalytic competence of the autophagy machinery itself. The functional coupling of these two regulatory inputs means that conditions simultaneously reducing mTORC1 activity and elevating SIRT1 function—as occurs during caloric restriction, when both nutrient depletion and a rise in the NAD⁺/NADH ratio are present—produce a coordinated and amplified autophagic response that neither mechanism achieves alone [80, 58]. Notably, caloric restriction elevates NAD⁺ by upregulating NAMPT through AMPK and circadian clock signalling, providing a direct mechanistic link between energy deficit and sirtuin activation [83].

Crucially, SIRT1 activity is itself governed by intracellular NAD⁺ availability, which declines with age in multiple tissues—with some human cohorts showing NAD⁺ levels roughly half those of young adults by middle age [83, 84, 80]. This age-related NAD⁺ depletion reduces SIRT1-mediated deacetylation of both autophagy proteins and upstream targets such as LKB1, a kinase whose SIRT1-dependent activation is required for full AMPK stimulation [81]. The resulting decline in AMPK activity diminishes its inhibitory input to mTORC1, creating a feed-forward cycle in which age-related NAD⁺ loss simultaneously weakens sirtuin-mediated autophagy and permits mTORC1 hyperactivation—two processes that converge on impaired cellular quality control [81, 82]. This mechanistic architecture positions NAD⁺ metabolism as a critical upstream variable linking energy status to the autophagy-longevity axis, and provides a molecular rationale for the geroprotective effects of NAD⁺ precursors such as nicotinamide riboside and nicotinamide mononucleotide, which may restore autophagic competence in aged tissues by re-engaging the SIRT1-AMPK-mTOR regulatory circuit [81, 80]. Human trials have confirmed that nicotinamide riboside safely elevates blood NAD⁺ by 60–100% and may modestly reduce blood pressure and inflammatory markers [83], while nicotinamide mononucleotide supplementation improved skeletal muscle insulin sensitivity by approximately 25% in overweight women with prediabetes, with additional reported gains in aerobic capacity and sleep quality [83], providing early translational support for this mechanistic framework. The consistent safety profile of both precursors across these trials further supports their candidacy as agents capable of reversing the age-associated NAD⁺ deficit that underlies autophagic decline [83, 85].

Polyphenols as Multi-Target Autophagy Modulators

Beyond the canonical mTOR and sirtuin axes, plant-derived polyphenols have attracted substantial attention as naturally occurring, multi-target compounds capable of restoring autophagic activity through engagement of at least ten distinct signalling pathways, including AMPK, SIRT1, MAPK, Nrf2/HO-1, and PINK1/Parkin [86]. This mechanistic plurality distinguishes polyphenols from single-target pharmacological agents such as rapamycin — whose longevity benefits, while well-documented, are accompanied by immunosuppressive liabilities and off-target effects attributable to its selectivity for mTORC1 [13, 7] — and may account for their broad efficacy across metabolic disease contexts. Critically, the autophagy-modulating effects of polyphenols are neither uniform nor unconditional: the same compound can enhance autophagy in certain tissues while suppressing it in others, reflecting the context-dependence of autophagic signalling and the differential expression of pathway components across cell types [86]. This organ-specific patterning is not merely a complication but arguably a feature, since it suggests that polyphenol-induced autophagic responses may be appropriately calibrated to tissue-specific needs. The engagement of PINK1/Parkin-dependent mitophagy is of particular relevance here, as selective clearance of dysfunctional mitochondria represents a distinct layer of quality control beyond bulk autophagy induction [86, 87]. The intersection of polyphenol biology with the broader nutrient-sensing architecture is further evident in the overlap between AMPK and mTOR as both nutrient sensors and autophagy regulators [54], positioning polyphenol-rich dietary patterns as capable of engaging autophagic and mitochondrial maintenance pathways simultaneously [86, 88].

Proteostasis and the Balance of Synthesis and Degradation

One of mTOR’s most consequential functions is the regulation of protein synthesis, primarily through phosphorylation of S6K1 and 4E-BP1, which together govern ribosome biogenesis and cap-dependent translation initiation [1]. In aged cells, this synthetic capacity operates against a backdrop of declining autophagic and proteasomal clearance [21, 89], producing an imbalance that favours the accumulation of misfolded, oxidised, and aggregated proteins [90, 91]. Indeed, proteostasis collapse has been characterised as an early molecular event in aging, preceding many downstream pathologies [92]. mTOR inhibition addresses this imbalance from both ends: reducing the rate of new protein synthesis while simultaneously inducing autophagy to clear existing damage [31, 19]. The lysosomal positioning of mTORC1 is particularly apt in this context, allowing the kinase to survey the amino acid pool generated by autophagic degradation and thereby couple catabolic flux to anabolic resumption [1, 3]. Disruption of this coupling—whether through constitutive mTOR activation or through the age-related decline in autophagic capacity—produces the proteostatic failure characteristic of numerous age-associated neurodegenerative and metabolic diseases [15, 21, 93].

Mitochondrial Quality Control and Convergent Longevity Mechanisms

More recent work has substantially elevated the importance of mitochondrial function within this framework. Rather than treating mitochondrial dysfunction as one hallmark among many, [56] argue that mitochondria represent a primary driver of ageing, exhibiting all primary hallmarks of the ageing process and serving as both cause and amplifier of systemic decline. Within this framing, the mitochondrial surveillance function of autophagy—specifically mitophagy, the selective autophagic clearance of damaged mitochondria—becomes a critical node linking mTOR inhibition to longevity [20, 94]. Rapamycin and metformin, two of the most studied longevity-associated compounds, extend lifespan through complementary mechanisms converging on mitochondrial function via modulation of the mTOR-AMPK signalling axis [56]. AMPK, activated by energetic stress, both inhibits mTORC1 and directly activates the ULK1 autophagy initiation complex [54, 52], providing a parallel route to mitophagy induction [56]. This complementarity suggests that combination strategies targeting both kinases may produce additive or synergistic effects on mitochondrial quality control that neither agent achieves alone [75]. The AMPK-SIRT1 positive feedback loop—in which AMPK activation raises NAD⁺ levels to stimulate SIRT1, while SIRT1-mediated deacetylation of LKB1 amplifies AMPK signalling—further reinforces mitochondrial surveillance by promoting PGC-1α-dependent mitochondrial biogenesis alongside mitophagy [53], ensuring that damaged organelles are cleared while functional mitochondrial mass is replenished [81, 95].

Cellular Senescence, SASP Suppression, and Senescence Heterogeneity

The relationship between mTOR and cellular senescence represents another mechanistic dimension of the quality-control story. Senescent cells, which accumulate progressively in aged tissues, were shown to causally drive a broad range of age-related pathologies through the SASP—a pro-inflammatory secretome that degrades the tissue microenvironment and can spread senescence to neighbouring cells [15]. mTORC1 activity is required for robust SASP production; rapamycin treatment suppresses SASP expression, in part by reducing the translation of cytokine transcripts that harbour 5′ terminal oligopyrimidine (TOP) motifs and are selectively sensitive to 4E-BP1 regulation [11]. This senomorphic mechanism has been elaborated in the broader therapeutic literature: rapamycin and its analogues attenuate NF-κB-driven cytokine output and reduce the paracrine spread of senescence without obligatorily eliminating the senescent cells themselves [96]. This connection integrates the mTOR-autophagy framework with the emerging senolytic and senomorphic therapeutic landscape, suggesting that mTOR inhibitors may act not only as autophagy inducers but also as suppressors of chronic sterile inflammation [4]. Notably, SIRT1 activity also counteracts the inflammatory milieu associated with senescence: chronic inflammatory conditions are characterised by reduced sirtuin activity, while SIRT1 activation suppresses NF-κB-driven pro-inflammatory gene expression and promotes the metabolic switch from glycolysis to fatty acid oxidation that characterises inflammation resolution [84]. The convergence of mTOR inhibition and sirtuin activation on SASP suppression through complementary mechanisms—translational repression and transcriptional silencing, respectively—further illustrates how the integrated nutrient-sensing network coordinates cellular quality control at multiple regulatory levels.

An important recent development in this area is the recognition that cellular senescence is far from a uniform biological state. Systematic molecular characterisation has revealed that senescent cell populations are heterogeneous, with distinct cell types exhibiting different SASP compositions, pathway dependencies, and responses to therapeutic intervention—a framework increasingly termed “senotyping” [42]. In chronic lung diseases such as COPD and idiopathic pulmonary fibrosis, senescent cell accumulation occurs across alveolar epithelial cells, fibroblasts, airway smooth muscle cells, and endothelial cells, each with distinct secretory profiles and vulnerabilities to senolytic or senomorphic agents [42]. This heterogeneity carries direct implications for mTOR-targeted strategies: rapamycin’s senomorphic effects—suppressing SASP without eliminating senescent cells—may be differentially effective across senotypes, and blanket approaches risk disrupting the beneficial roles of transient senescence in tissue repair contexts where senescent myofibroblast clearance facilitates wound resolution [42, 96]. The conceptual distinction between senolytics, which selectively eliminate senescent cells [97], and senomorphics such as rapamycin, which modulate their secretory output, thus gains additional practical significance when senotype-specific responses are considered. The composite of circulating SASP factors—including IL-6, IL-8, and TNF-α—has been proposed as a blood-based readout of systemic senescence burden [98], potentially complementing epigenetic clock measurements [63, 25] that may capture upstream drivers of senescence accumulation rather than its downstream secretory consequences [79].

Limitations and Outstanding Considerations

Despite the mechanistic coherence of this framework, important caveats remain. The dose- and context-dependence of rapamycin’s effects—including its tissue-specific liabilities, such as the paradoxical myopathy arising from both insufficient and excessive mTORC1 activity in skeletal muscle [1], as well as mTORC2-dependent insulin resistance and immunosuppressive effects on T and B cell populations [50, 67]—illustrate that the mTOR-autophagy axis is not uniformly beneficial to activate. Female mice consistently show greater lifespan extension from rapamycin than males [11, 29], suggesting sex-specific mTOR signalling contexts that are not yet mechanistically resolved. Furthermore, while autophagy’s causal necessity has been demonstrated for several longevity interventions [20], the relative contributions of bulk autophagy, selective autophagy (including mitophagy), and proteasomal degradation to proteostatic maintenance remain incompletely partitioned [21, 91]. Taken together, the evidence supports a model in which mTOR inhibition extends healthy lifespan primarily by restoring and amplifying cellular quality-control programmes, but the precision with which these programmes must be calibrated across tissues and temporal contexts continues to define the frontier of the field.

7. mTOR in Disease: Cancer, Autoimmunity, and Neurodegeneration

mTOR dysregulation has emerged as a unifying pathological thread across some of the most clinically significant age-related diseases, linking oncogenesis, autoimmunity, neurodegeneration, and metabolic dysfunction within a shared molecular framework. Understanding how aberrant mTOR signalling contributes to these diverse conditions — and how its pharmacological modulation can be exploited therapeutically — has been a central preoccupation of biomedical research for over two decades, with the picture growing considerably more nuanced as evidence has accumulated.

PI3K/AKT/mTOR Oncogenesis and Targeted Therapy

Among the most extensively documented pathological roles of mTOR is its position as a downstream effector of the PI3K/AKT/mTOR (PAM) pathway in cancer. Early work recognised that this pathway governs cell survival, proliferation, and metabolism, but the full scale of its oncogenic relevance became apparent only as genomic data accumulated. Comprehensive analysis has since revealed that the PAM pathway is dysregulated in approximately 50% of human cancers, with PIK3CA activating mutations and PTEN loss constituting among the most prevalent alterations across tumour types [99]. These genetic lesions produce constitutive mTORC1 activation, driving anabolic metabolism and suppressing apoptosis in a manner that confers strong selective advantage on malignant cells.

The translational response to this understanding proceeded through successive generations of inhibitors. First-generation rapalogs — rapamycin and its analogues everolimus and temsirolimus — achieved regulatory approval in renal cell carcinoma, pancreatic neuroendocrine tumours, and advanced breast cancer [100, 33], but their clinical impact was constrained by incomplete mTOR inhibition — rapalogs block S6K1 phosphorylation yet fail to fully suppress 4EBP1, leaving cap-dependent translation largely intact — and the paradoxical upregulation of AKT signalling that resulted from the relief of mTORC1-mediated negative feedback on insulin receptor substrate proteins [99, 11, 33]. This feedback loop, a recurrent problem in PAM pathway therapeutics, effectively negated single-agent efficacy in many contexts. Resistance mechanisms also emerged through PIK3CA secondary mutations and KRAS co-activation, as well as through alterations in cap-dependent translation machinery involving 4E-BP1 and eIF4E that allow tumour cells to maintain protein synthesis through alternative kinases, underscoring the pathway’s redundancy and adaptability [99, 100].

Subsequent strategies sought to address these vulnerabilities through dual PI3K/mTOR inhibitors and isoform-selective PI3K inhibitors. The selective agents — including alpelisib, duvelisib, and idelalisib — have demonstrated more favourable clinical profiles than pan-PI3K inhibitors, primarily because isoform specificity reduces off-target toxicity while maintaining therapeutic efficacy [99]. ATP-competitive mTOR kinase inhibitors (TORKIs), such as AZD8055 and sapanisertib, achieve more complete blockade of both mTORC1 and mTORC2 and can induce apoptosis and reverse rapalog resistance, though dose-limiting toxicities have complicated their clinical translation [100, 33]. More recently, next-generation bivalent inhibitors termed RapaLinks, which simultaneously engage both the FKBP12-rapamycin binding domain and the kinase domain of mTOR, have shown the capacity to overcome resistance that arises through mutations in either target domain individually [99]. Combination strategies pairing mTOR inhibitors with MEK inhibitors, CDK4/6 inhibitors, or immunotherapy represent a current frontier, grounded in the recognition that single-pathway blockade is rarely sufficient given the crosstalk that characterises advanced malignancies [99, 11, 100]. The dual role of AMPK in cancer biology — where its tumour-suppressive function through mTORC1 inhibition and lipogenesis suppression is countered by its capacity to support tumour cell survival under hypoxic and nutrient-deprived conditions [55, 101] — further complicates therapeutic strategies targeting the broader nutrient-sensing network in oncological contexts.

Rapamycin as an Immunomodulator: Autoimmune Disease

A distinct but mechanistically coherent body of work has established mTOR as a regulator of immune cell fate with direct relevance to autoimmune pathology. Seminal synthesis of this field recognised that mTORC1 hyperactivation serves as both a predictive biomarker and a mechanistic driver of disease flares in systemic lupus erythematosus (SLE) [32]. In lupus patients, aberrant mTOR activity promotes the expansion of proinflammatory double-negative T cells (CD3⁺CD4⁻CD8⁻) — a pathogenic lymphocyte population that resists conventional immunosuppression — while simultaneously impairing the development and function of regulatory T cells (Tregs) that would otherwise constrain autoreactive immune responses [32].

Rapamycin’s immunological profile makes it particularly well suited to this context. Unlike conventional immunosuppressants that broadly suppress lymphocyte proliferation, rapamycin’s mTORC1-preferential inhibition selectively contracts effector T cell populations while paradoxically expanding Tregs, owing to the distinct sensitivity of these cell populations to mTOR signalling levels [32, 48, 5]. Early clinical observations in lupus cohorts were sufficiently promising to sustain interest in rapamycin repurposing for autoimmune indications. The broader point — that mTOR inhibition can be immunomodulatory rather than simply immunosuppressive — has since influenced thinking about the drug’s utility across inflammatory and autoimmune conditions beyond lupus, including multiple sclerosis models where mTOR-regulated metabolic reprogramming of T cells has been implicated in disease pathogenesis [32, 5]. Chronic rapamycin administration in murine models has provided further characterisation of this immunomodulatory profile, demonstrating that prolonged treatment reduces total splenic immune cell numbers while disproportionately contracting exhausted PD-1⁺ T cell populations and shifting the residual T cell pool from memory toward naïve configurations — phenotypic changes consistent with immune rejuvenation rather than blanket suppression [67]. That rapamycin extended lifespan even in immunodeficient RAG2⁻/⁻ and IFN-γ⁻/⁻ mice indicates that immune remodelling represents one of multiple parallel mechanisms contributing to longevity, rather than the sole operative pathway [67].

Neuroprotective Effects of mTOR Inhibition and the Gut-Brain Axis

Perhaps the most clinically urgent translational application concerns the potential of rapamycin to modify the course of neurodegenerative disease. Preclinical evidence across multiple model systems established that rapamycin reduces amyloid-beta burden, attenuates tau pathology, and preserves cognitive function in mouse models of Alzheimer’s disease (AD), with the proposed mechanism centring on the restoration of autophagic clearance that becomes impaired with ageing and mTOR hyperactivation [7, 11]. The landmark ITP study demonstrated that rapamycin feeding initiated late in life—at 600 days of age—extended median lifespan by approximately 9% in males and 14% in females in genetically heterogeneous mice, establishing that mTOR inhibition can meaningfully alter the trajectory of mammalian ageing even when administered after midlife [8]. Similar neuroprotective effects have been reported in models of Parkinson’s disease and Huntington’s disease, where the accumulation of misfolded protein aggregates can be mitigated by mTOR-induced autophagy upregulation [11, 5].

These findings collectively generated the hypothesis that mTOR-driven suppression of autophagy represents a convergent mechanism in neurodegeneration, and that pharmacological mTOR inhibition could delay or partially reverse proteotoxic burden. Translation to human disease has, however, proceeded cautiously given rapamycin’s complex systemic effects and the challenge of delivering adequate CNS exposure. A critical development in this regard is the ERAP phase IIa clinical trial, initiated to investigate rapamycin as a repurposed treatment for early-stage AD using in vivo neuroimaging as a primary outcome measure [78]. Enrolling fifteen patients with early-stage AD on a weekly rapamycin dosing schedule, ERAP represents the first structured effort to determine whether the lifespan extension and attenuation of AD pathology observed preclinically can translate to measurable benefit in humans [78]. Its outcomes are anticipated to be definitive for the field’s near-term direction.

An emerging dimension of neurodegeneration research that intersects directly with mTOR biology is the gut-brain axis—the bidirectional communication network linking the gut microbial ecosystem to central nervous system function through immune, neural, and metabolic pathways [45]. Gut microbiota regulate a remarkable range of neurological parameters, including blood-brain barrier integrity, microglial activation states, hippocampal neurogenesis, and complex behaviours, with germ-free animal models and antibiotic-treated preparations demonstrating that microbial communities are not merely correlated with but required for normal brain development and homeostasis [45]. The metabolic arm of this communication operates substantially through short-chain fatty acids: butyrate in particular regulates microglial morphology and reactivity, influences blood-brain barrier tight junction protein expression, and modulates hippocampal gene expression relevant to neuroplasticity [45, 22].

The relevance of this axis to age-related neurodegeneration is both mechanistic and translational. Age-associated dysbiosis, characterised by loss of SCFA-producing taxa such as Faecalibacterium prausnitzii and Bifidobacterium species and expansion of pro-inflammatory species, diminishes the neuroprotective metabolite signals that sustain blood-brain barrier function and constrain neuroinflammation [22, 62]. Since SCFAs activate AMPK and thereby suppress mTORC1, the progressive loss of microbial SCFA production with age may contribute to the mTOR hyperactivation and consequent autophagy suppression in neural tissue that permits proteotoxic aggregate accumulation [22, 45]. This creates a conceptual bridge between the ecological disruption of the ageing gut and the intracellular proteostatic failure characteristic of Alzheimer’s and Parkinson’s disease: dysbiosis erodes an upstream input to the AMPK-mTOR-autophagy axis, while mTOR hyperactivation simultaneously suppresses the autophagic clearance machinery on which post-mitotic neurons are uniquely dependent. The oral-gut microbial axis provides a further upstream input, as oral pathogens such as Porphyromonas gingivalis can survive gastric transit, establish gut colonisation, and contribute to systemic inflammatory burden—a pathway of particular significance given the marked increase in periodontal disease prevalence with age and emerging epidemiological associations between periodontal disease and dementia risk [102].

This integrated gut-brain-mTOR framework suggests that interventions targeting the microbial community—whether through dietary fibre to sustain SCFA production, probiotics to restore beneficial taxa, or direct SCFA supplementation—could complement pharmacological mTOR inhibition by addressing the ecological dimension of mTOR hyperactivation that drug-based approaches do not reach. Mediterranean dietary patterns, which are associated with approximately 30% reductions in both cardiovascular and depression risk and are characterised by high fibre and polyphenol content that sustains SCFA-producing microbial communities, may achieve their neuroprotective effects in part through this dual mechanism of maintaining microbial AMPK input while simultaneously providing polyphenolic autophagy activators [88, 86]. The gut microbiome itself has been shown to modulate longevity pathways independently of host genetics, as demonstrated in Drosophila models where rapamycin’s effects on tissue ageing and lifespan were found to operate partially through microbiota-dependent mechanisms [30]. Nevertheless, the systematic review evidence base currently shows no significant rapalog effects on neurological parameters in human studies [35], and the contribution of microbiome-mediated mechanisms to neurodegeneration in humans—as distinct from animal models—remains an unresolved empirical question requiring dedicated longitudinal investigation.

Vascular Ageing and Senescence-Driven Cardiovascular Pathology

Emerging evidence has positioned vascular senescence as a distinct and clinically significant domain in which mTOR-associated pathology drives age-related disease. Senescent endothelial cells and vascular smooth muscle cells accumulate progressively with age under conditions of haemodynamic stress, reactive oxygen species exposure, and telomere shortening [98], producing a SASP enriched in pro-inflammatory cytokines—including IL-6, IL-8, and PAI-1—and matrix metalloproteinases—particularly MMP-9—that directly compromise vascular integrity [103]. Senescent endothelial cells additionally lose the capacity to produce nitric oxide, impairing vasodilation and promoting thrombosis and immune infiltration, while senescent vascular smooth muscle cells exhibit reduced collagen synthesis alongside increased elastase activity [98]. SASP-derived MMP-9 degrades extracellular matrix components within atherosclerotic plaques, reducing structural integrity and increasing vulnerability to rupture, thereby mechanistically linking cellular-level senescence to acute cardiovascular events [103]. Arterial stiffness, a clinically measurable consequence of senescence-driven vascular smooth muscle dysfunction and matrix remodelling—including senescence-associated vascular calcification driven by osteogenic mediator secretion [98]—represents a particularly important translational endpoint connecting molecular mechanisms to haemodynamic burden assessable via pulse wave velocity in human populations [103]. Given that mTORC1 activity is required for robust SASP production and that rapamycin suppresses SASP expression through translational mechanisms [11, 96], vascular senescence represents a pathological context in which the senomorphic properties of mTOR inhibitors may have direct therapeutic relevance—though the challenge of achieving sufficient vascular tissue exposure with systemically tolerable doses remains unresolved.

Metabolic Disease: mTOR-Driven Insulin Resistance and Obesity

The relationship between mTOR signalling and metabolic disease introduces an important therapeutic paradox. mTORC1 hyperactivation in the context of chronic nutrient excess — as occurs in obesity — drives insulin resistance through serine phosphorylation of insulin receptor substrate-1 (IRS-1) by S6K1, thereby attenuating insulin signalling downstream of its own activation [48, 32, 1]. This feedback inhibition loop, in which mTOR undermines the very pathway that activates it, mechanistically links overnutrition to type 2 diabetes. Rapamycin would appear to offer a logical therapeutic intervention, and indeed short-term mTORC1 inhibition does improve insulin sensitivity in certain contexts [48].

However, prolonged rapamycin treatment has been shown to impair glucose tolerance and exacerbate dyslipidaemia in mice, effects attributable to inadvertent mTORC2 inhibition that disrupts AKT-mediated hepatic glucose metabolism [7, 48, 5, 50]. Mechanistically, chronic rapamycin disrupts mTORC2 assembly in metabolically important tissues, attenuating phosphorylation of AKT at Ser473 and impairing suppression of hepatic gluconeogenesis — a causal role confirmed by the observation that liver-specific deletion of the mTORC2 subunit Rictor recapitulates these defects independently of rapamycin treatment [50]. This mechanistic distinction — that beneficial metabolic effects are mTORC1-mediated while adverse effects arise from mTORC2 suppression — has become a central axis of drug development strategy, motivating the design of mTORC1-selective compounds and specific schedules of intermittent rapamycin dosing that preserve mTORC2 function [5, 29]. A notable proof-of-concept is DL001, a rapamycin analogue reported to be over 430-fold more selective for mTORC1 over mTORC2 than rapamycin itself, which in mice preserved normal glucose tolerance and lipid profiles while retaining robust mTORC1 inhibition across multiple tissues [65]. Mannick and Lamming’s synthesis of the current evidence emphasises that resolving this selectivity problem is arguably the foremost challenge for translating mTOR biology into safe long-term therapeutics for metabolic and age-related disease [5]. This tension between mTOR’s indispensable physiological roles in anabolic metabolism and its pathological hyperactivation in disease states remains one of the field’s most consequential unresolved questions, shaping both drug discovery priorities and clinical trial design across all the disease categories reviewed here.

8. Evolutionary and Theoretical Frameworks for mTOR, Growth, and Ageing Trade-offs

Understanding why mTOR inhibition extends lifespan requires more than cataloguing molecular mechanisms; it demands a theoretical scaffold that explains why evolution produced a pathway so consequential to ageing in the first place. Over the past two decades, three converging frameworks — the hyperfunction theory, life history theory, and evolutionary mismatch — have progressively clarified the conceptual logic connecting mTOR activity, growth, and longevity trade-offs. Crucially, recent synthesis work published in 2025 has drawn these threads together with greater coherence, grounding rapamycin’s geroprotective promise in evolutionary first principles while simultaneously sharpening cautions about its translational risks.

Hyperfunction Theory: mTOR Overactivity as Quasi-Programmed Ageing

The most influential theoretical contribution to this field is Blagosklonny’s hyperfunction theory, which reframes ageing not as an accumulation of molecular damage but as the pathological persistence of growth-promoting programmes beyond their developmental utility. As articulated in an early theoretical synthesis, rapamycin extends lifespan by slowing ageing itself rather than through ageing-independent mechanisms, with ageing operationally defined as an exponential increase in mortality probability in which age-related diseases serve as biomarkers of underlying biological deterioration [12]. On this account, mTOR does not malfunction in old age; it continues functioning precisely as selected — anabolically driving growth and biosynthesis — but does so in a post-reproductive context where such activity is no longer adaptive and instead becomes pathogenic [2]. The term “quasi-programmed” captures this dynamic: the pathway is not programmed for senescence, but senescence is the quasi-inevitable consequence of a programme that evolution had no strong selection pressure to switch off after reproductive maturity [12]. This framing resonates with broader evolutionary perspectives that situate mTOR hyperactivity within antagonistic pleiotropy and the declining force of natural selection in later life [34, 104].

A 2025 tribute to Blagosklonny’s legacy elaborates this framework with particular clarity, arguing that mTOR hyperactivity drives “geroconversion” — the transition of arrested cells from a reversible quiescent state to an irreversible senescent phenotype — and that rapamycin’s therapeutic benefits are predicted, rather than merely post-hoc explained, by the hyperfunction model [105]. This retrospective synthesis underscores how the hyperfunction theory has progressively unified disparate observations: why the same pathway that promotes growth in development drives pathology in ageing, and why pharmacological dampening of mTOR consistently extends healthspan and lifespan across phylogenetically distant organisms — from yeast and Drosophila to mice [9, 4, 105]. The landmark ITP study demonstrating lifespan extension in genetically heterogeneous mice fed rapamycin late in life provided particularly striking empirical support for this cross-species consistency [8]. This predictive coherence distinguishes hyperfunction theory from damage-accumulation accounts, which struggle to explain why lifespan extension follows from reducing activity in a pathway that is not itself damaged [106].

Life History Theory and Evolutionary Trade-offs

The hyperfunction framework gains deeper evolutionary grounding when situated within life history theory, which analyses how organisms allocate finite resources among competing demands of growth, reproduction, and somatic maintenance. A comprehensive 2025 review integrating evolutionary medicine and life history perspectives identifies trade-offs as one of three core frameworks for understanding ageing: traits beneficial early in life impose later costs, and selection is insensitive to those costs once reproductive fitness has been secured [107]. mTOR exemplifies this logic precisely. High mTOR activity during development and reproductive life promotes growth, immune function, and tissue repair — all fitness-enhancing — while the same activity in post-reproductive somatic tissue drives hypertrophy, senescence, and ultimately age-related pathology [107, 12, 4]. This misalignment between developmental programming and late-life physiology is characterised in the developmental theory of ageing as “hyperfunction”: the deleterious continuation of growth-promoting pathways — such as sustained TOR signalling — that enhance early fitness but drive late-life pathologies including cardiovascular disease and osteoporosis [104].

This evolutionary perspective explains an otherwise puzzling feature of the geroscience literature: why single-gene mutations in metabolic and growth pathways produce such disproportionately large effects on lifespan. Early genetic work in C. elegans overturned the assumption that lifespan was determined by hundreds of genes with negligible individual effects, demonstrating instead that mutations in insulin/IGF-1 and mTOR signalling could extend lifespan by 40–60% [15]. Notably, evidence from C. elegans indicates that suppression of insulin/IGF-1 signalling confined to adulthood can extend lifespan without incurring reproductive costs, a pattern more consistent with hyperfunction theory than with a simple resource allocation trade-off [104]. Life history theory contextualises these findings: these are not peripheral housekeeping genes but central resource-allocation regulators whose pleiotropic effects on early fitness and late-life survival are precisely what trade-off theory predicts [107, 15]. Reducing their activity may shift the allocation balance toward somatic maintenance at modest cost to growth and reproduction — a trade-off that is invisible to natural selection but highly legible to biogerontologists.

The relationship between extrinsic mortality and the evolution of ageing has itself been refined beyond the classical prediction that high extrinsic mortality uniformly selects for faster senescence. More nuanced analyses have shown that increased extrinsic mortality can, in some ecological contexts, actually select for improved somatic maintenance and extended longevity [34, 104]. This refinement complicates simple predictions but enriches evolutionary models by highlighting that the ecological and physiological context in which an organism evolves shapes the specific molecular architecture of its ageing programme, including the sensitivity of the mTOR-mediated growth-maintenance allocation to environmental variation.

Dietary Protein Restriction as an Evolutionary Test of Life History Trade-offs

The evolutionary logic of life history trade-offs receives powerful empirical support from nutritional biology. Systematic analysis across Drosophila, rodents, and human epidemiological data has demonstrated that restriction of dietary protein — and more specifically of key amino acids such as methionine and tryptophan — recapitulates many of the longevity and healthspan benefits previously attributed broadly to caloric restriction [51]. Geometric framework studies further reveal that it is the ratio of protein to non-protein energy, rather than total caloric intake per se, that most powerfully predicts both lifespan and reproductive output across these organisms [51, 108]. This finding reframes caloric restriction not as a global energetic phenomenon but as a specific signal of protein and amino acid availability that feeds directly into the molecular machinery governing growth-versus-maintenance allocation decisions. The mechanistic logic is deeply consistent with evolutionary theory: in environments where essential amino acids are scarce, an organism that shifts investment from anabolic growth and reproduction toward somatic repair would be expected to survive longer to exploit future resource availability [51, 34, 104]. This adaptive plasticity in resource allocation is interpreted as a conditional life-history strategy shaped by natural selection under fluctuating nutritional environments [108, 34]. The dietary interventions characterised in this literature can thus be understood as experimentally imposed ecological conditions that recapitulate, at the level of individual physiology, the selective pressures that shaped the evolution of conditional life-history strategies.

Evolutionary Mismatch and Hyperfunctional Anabolic Pathways

The third framework — evolutionary mismatch — addresses why the mTOR/ageing trade-off has become particularly acute in contemporary human populations. Ancestral environments characterised by caloric scarcity and high physical activity would have maintained mTOR signalling within ranges calibrated by millions of years of selection. Calorie-rich, sedentary modern environments chronically elevate nutrient and growth-factor inputs, producing constitutive mTOR hyperactivation that amplifies the hyperfunction dynamic described above [107]. In this sense, the modern epidemiological pattern — obesity, type 2 diabetes, cardiovascular disease, and neurodegeneration as the dominant burdens of ageing — reflects an ancestral anabolic programme operating in a context for which it was never selected [107].

Rapamycin can be understood within this mismatch framework as a pharmacological corrective: by attenuating mTOR activity, it partially restores a signalling environment more congruent with ancestral conditions without requiring the environmental changes that are socially and behaviourally difficult to sustain [107, 105]. Reviews of mTOR inhibitor biology have consistently noted that rapamycin’s longevity effects resemble, but are not identical to, those of caloric restriction — both interventions reduce nutrient-sensing pathway activity, though rapamycin targets mTORC1 more specifically [5, 12, 4]. Crucially, rapamycin retains efficacy when initiated late in life: the landmark Interventions Testing Program (ITP) study demonstrated significant lifespan extension in genetically heterogeneous mice when treatment commenced at 20 months of age — equivalent to approximately 60 human years — yielding median lifespan increases of ~9% in males and ~14% in females [8]. This contrasts with caloric restriction, which loses geroprotective potency when begun in aged animals [5]. This late-life efficacy is theoretically important: it suggests that correcting mismatch-driven hyperfunction can extend healthspan even after decades of excess anabolic signalling have accumulated. The broader mismatch framework also explains why metabolic flexibility — the capacity to efficiently switch between fuel substrates in response to changing energy availability — declines with age and chronic overnutrition [95]; the constitutive anabolic signalling driven by mTOR hyperactivation in modern environments actively impairs this adaptive metabolic switching, as mTORC1 activity suppresses catabolic programmes such as autophagy and fatty acid oxidation that are essential for fuel-substrate transitions [4, 95].

Epigenetic Drift as an Evolutionary Consequence of Growth-Longevity Trade-offs

The evolutionary frameworks described above gain additional mechanistic specificity when extended to the epigenome. The progressive deterioration of epigenetic landscapes with age — manifest as global DNA hypomethylation, focal promoter hypermethylation, heterochromatin erosion, and retrotransposon reactivation [40, 39] — can be understood not as random molecular noise but as the downstream consequence of developmental programmes optimised for growth and reproduction rather than for long-term epigenomic maintenance [26, 27]. The reactivation of LINE-1 and other repetitive elements following heterochromatin loss is of particular significance, as it triggers innate immune sensing pathways and amplifies the inflammatory milieu characteristic of aged tissues [40]. Longevity-associated genetic variants, including those near FOXO3, SIRT1, and APOE, exert their protective effects in part by modulating DNA methylation states through methylation quantitative trait loci (meQTLs), suggesting that exceptional longevity is associated with greater epigenomic stability rather than merely favourable gene expression [27]. The distinctive epigenomic profiles of centenarians — characterised by preserved heterochromatin integrity, stable repetitive element silencing, and favourable patterns of histone acetylation — provide empirical support for the view that epigenetic resilience is a proximate basis of exceptional longevity, shaped substantially by the same nutrient-sensing and growth-regulatory pathways whose evolutionary logic the hyperfunction and life history frameworks describe [27]. This convergence positions epigenetic drift as a measurable molecular readout of the growth-longevity trade-off, one that is now quantified by an expanding toolkit of DNA methylation-based biological age biomarkers — including the Horvath multi-tissue clock [63], the Hannum blood-based clock [64], GrimAge [36], and pace-of-ageing measures such as DunedinPACE [37] — that are entering clinical and interventional application.

Translational Cautions and Ethical Dimensions

Theoretical elegance does not dissolve translational complexity, and recent literature has been careful to balance the evolutionary logic supporting rapamycin with substantive caution about off-label longevity use. The same mTOR activity that drives age-related pathology also underpins immune surveillance, tissue repair, and insulin sensitivity; chronic mTORC1 inhibition produces immunosuppression, metabolic disturbances, impaired wound healing, and hormonal disruption in both preclinical and clinical settings [13, 48, 35]. Mechanistic reviews have further clarified that many adverse effects arise not from mTORC1 inhibition per se but from unintended suppression of mTORC2, whose distinct substrate profile — including AKT and SGK1 — governs glucose metabolism and cytoskeletal organisation [66, 5, 50, 6]. Critically, Lamming et al. demonstrated that rapamycin-induced insulin resistance is mediated specifically by mTORC2 loss and is mechanistically uncoupled from its longevity benefit [50], a finding that both motivates and complicates the search for selective mTORC1-sparing regimens [7, 65]. This mechanistic nuance — that the therapeutic window for longevity may hinge on selective complex inhibition — remains unresolved in human applications.

The ethical dimensions of off-label rapamycin use for longevity are thus framed by a genuine tension between a theoretically robust rationale and an incompletely characterised risk profile in healthy individuals [13, 47]. Systematic review evidence underscores that data from transplant and oncology populations cannot be straightforwardly extrapolated to healthy older adults receiving low-dose intermittent regimens [35], and emerging real-world cohort studies are only beginning to characterise pharmacokinetics and tolerability in this population [74, 46]. Evolutionary and life history frameworks powerfully explain why mTOR inhibition should slow ageing, but they do not specify the dose, schedule, or population in which the benefits of correcting evolutionary mismatch outweigh the harms of disrupting physiologically necessary anabolic signalling. Resolving that tension constitutes a central challenge for the next phase of geroscience translation [47, 5].

9. Discussion

The body of evidence synthesised across these six themes reveals a field that has undergone substantial conceptual maturation in the past two to three years. Where earlier work established mTOR as a promising longevity target largely on the basis of pharmacological lifespan extension in rodents, the current literature presents a considerably more nuanced picture — one in which mechanistic specificity, context-dependence, and translational complexity have moved to the centre of scientific debate.

Perhaps the most consequential shift in recent understanding concerns the functional architecture of mTOR signalling itself. The traditional binary framing of mTORC1 as the growth-promoting, pro-ageing complex and mTORC2 as largely separate and secondary has been substantially revised. Emerging evidence demonstrates extensive cross-talk between the two complexes, with chronic mTORC1 inhibition capable of paradoxically activating mTORC2-AKT signalling in ways that may counteract some intended longevity benefits and contribute to metabolic side-effects. This mechanistic insight reframes a longstanding clinical puzzle: the immunosuppressive, dyslipidaemic, and glucose-dysregulating effects observed in early rapalog trials are now understood not merely as on-target toxicities of mTOR inhibition, but as emergent consequences of network-level feedback dynamics. Theoretical frameworks drawing on evolutionary life-history trade-offs reinforce this interpretation — mTOR occupies a position in nutrient-sensing networks that was shaped by selection for reproductive fitness under resource variability, meaning that its inhibition in calorically replete modern environments produces pleiotropic effects that are mechanistically coherent but phenotypically unpredictable.

The model organism literature, while compelling in aggregate, has itself been re-evaluated with greater methodological rigour. Lifespan extension by rapamycin is now understood to be highly sensitive to dose, timing, sex, and genetic background — findings that create both problems and opportunities for translation. The demonstration that late-life initiation of rapamycin can still confer meaningful lifespan extension in mice challenged earlier assumptions that interventions must begin early to be effective, and opened genuinely new questions about optimal treatment windows in humans. However, the tendency for model organism studies to measure lifespan under conditions of minimal pathogen exposure, controlled diet, and absent reproductive demands substantially limits their predictive validity for human healthspan. Recent work on Drosophila and C. elegans has been particularly valuable not for its direct translational relevance but for dissecting pathway architecture — clarifying, for instance, the relative contributions of S6K1 suppression versus 4EBP1 activation to longevity outcomes, and pointing toward downstream autophagy induction as a mechanistically central node. The Drosophila demonstration that rapamycin’s lifespan benefits are entirely microbiota-independent, operating through cell-autonomous autophagy in intestinal epithelial cells [30], provides a critical methodological baseline against which more complex mammalian findings—where rapamycin concurrently remodels both immune architecture and gut metagenome composition [67]—must be interpreted. Comparative analyses of longevity-associated compounds have further contextualised rapamycin’s profile, with cell-based screening revealing that structurally diverse agents—including the odd-chain fatty acid pentadecanoic acid (C15:0)—share a substantial fraction of rapamycin’s anti-inflammatory, antifibrotic, and anticancer cellular activities despite operating through fundamentally different upstream targets [109]. This convergence suggests that rapamycin’s longevity-relevant cellular effects are not unique to mTOR inhibition per se, and that dietary or nutritional compounds may partially recapitulate them with potentially more favourable safety profiles, though direct clinical comparisons remain unavailable.

The intersection of mTOR modulation with autophagy and cellular quality control now stands as arguably the most theoretically coherent mechanism linking mTOR inhibition to healthspan benefits. Evidence across themes converges on a model in which mTORC1 suppression restores proteostatic capacity, mitochondrial quality control through mitophagy, and clearance of senescent cellular material — processes that collectively deteriorate during normal ageing and that underlie pathology across cancer, neurodegeneration, and autoimmunity. Crucially, this mechanistic convergence suggests that the efficacy of mTOR modulation in any given disease context may depend substantially on the baseline autophagic tone of the target tissue, which varies considerably by age, sex, and metabolic status. This could explain why clinical trials of rapalogs in cancer and age-related immune decline have produced inconsistent results that correlate poorly with preclinical predictions. The recognition that senescent cell populations are themselves heterogeneous—with distinct senotypes exhibiting different SASP profiles, pathway dependencies, and therapeutic vulnerabilities [42]—adds a further layer of complexity to this picture, suggesting that mTOR-mediated senomorphic effects may vary substantially across tissue compartments and disease contexts.

The Integrated Nutrient-Sensing Longevity Network

A recurring but insufficiently synthesised theme across the reviewed literature is that mTOR does not operate as an isolated signalling cascade but as one node within a deeply interconnected nutrient-sensing longevity network whose other principal components—AMPK, sirtuins, and insulin/IGF-1 signalling—are linked through direct biochemical crosstalk and shared downstream effectors. The molecular architecture of this network is now sufficiently well characterised to explain a longstanding puzzle in geroscience: why interventions as pharmacologically diverse as rapamycin, metformin, NAD⁺ precursors, resveratrol, and caloric restriction produce overlapping but non-identical longevity phenotypes.

At the core of this network lies a set of reciprocal regulatory relationships. AMPK, activated by energetic stress, suppresses mTORC1 through dual phosphorylation of TSC2 and Raptor [53, 52], while simultaneously activating ULK1 to initiate autophagy independently of mTOR [53]. SIRT1, activated by rising NAD⁺/NADH ratios, deacetylates and activates LKB1, the principal upstream kinase of AMPK, thereby amplifying AMPK signalling; AMPK activation in turn elevates intracellular NAD⁺ levels, creating a self-reinforcing positive feedback loop between the two enzymes [81, 80]. Notably, computational modelling of this tripartite mTORC1–AMPK–SIRT interaction has confirmed the bistable, switch-like dynamics of this feedback architecture, providing a mechanistic basis for the abrupt metabolic state transitions observed during nutrient deprivation [16]. SIRT1 also directly deacetylates core autophagy proteins (ATG5, ATG7, LC3) and activates FOXO transcription factors that upregulate antioxidant defences and stress resistance genes [81, 58]. Insulin/IGF-1 signalling, the fourth major axis, activates PI3K/AKT to phosphorylate and inhibit TSC2, releasing Rheb to stimulate mTORC1 [28]—an input that is directly antagonised by AMPK’s opposing phosphorylation of the same complex. The result is a rheostat-like regulatory architecture in which AMPK-sirtuin signalling and insulin/IGF-1-mTOR signalling function as countervailing arms: the former promoting maintenance, stress resistance, and catabolic recycling, the latter promoting growth, biosynthesis, and anabolic investment [52, 53].

This integrated network architecture has substantial explanatory power. Caloric restriction engages multiple nodes simultaneously—reducing insulin/IGF-1 input to mTORC1, activating AMPK through energy depletion, and raising NAD⁺ to stimulate SIRT1—which accounts for its broad and robust longevity effects across species [4, 20, 81, 108]. Rapamycin, by contrast, directly inhibits mTORC1 but does not inherently activate AMPK or SIRT1, explaining why its phenotypic effects resemble but are not identical to those of caloric restriction [5]. Metformin acts primarily through AMPK activation and consequent mTORC1 suppression [55, 56], while NAD⁺ precursors engage the network through the sirtuin axis: cellular NAD⁺ levels decline by as much as 50% between young adulthood and middle age in multiple tissues [83], and restoring them via NMN or NR supplementation re-activates SIRT1-dependent LKB1 activation and downstream AMPK-mTOR regulation that is progressively lost with age [80, 81]. Resveratrol’s longevity effects, shown to require both SIRT1 activity and intact autophagy genes [58], exemplify how a sirtuin-directed intervention ultimately converges on the same autophagic quality-control output that mTOR inhibition achieves through a distinct entry point.

This framework carries direct implications for the combinatorial geroprotector strategies reviewed in Section 4. The additive lifespan extension observed with rapamycin plus trametinib [75] is consistent with multi-node network engagement, and the same logic predicts that combinations targeting truly orthogonal nodes—for instance, an mTOR inhibitor paired with an AMPK activator or NAD⁺ precursor—could achieve comparable or superior additivity by simultaneously suppressing the anabolic arm and reinforcing the maintenance arm of the network. Preclinical evidence that rapamycin and metformin extend lifespan through complementary mechanisms converging on mitochondrial function [56], and that SIRT1 and mTOR inhibition suppress the senescence-associated secretory phenotype through independent but convergent routes [84, 11], provides biological plausibility for such rational combinations. However, the same network connectivity that creates combinatorial opportunity also introduces risk: the positive feedback between AMPK and SIRT1 means that pharmacological perturbation of one node can propagate unpredictably through the network, and the dual role of AMPK in both tumour suppression and tumour cell survival under metabolic stress [55, 101]—where AMPK paradoxically enables cancer cell adaptation to nutrient-poor tumour microenvironments—illustrates that network-level interventions require careful context-dependent evaluation. The challenge for the next generation of geroprotector trials will be to move beyond single-agent dose-finding toward rational multi-target protocols informed by this network topology.

The Gut Microbiome as an Ecological Input to the Longevity Network

A dimension of the nutrient-sensing longevity network that has only recently come into focus concerns the gut microbiome as an ecological layer of mTOR regulation that operates upstream of, and in parallel with, the cell-autonomous signalling architecture described above. The evidence reviewed across preceding sections—from microbial SCFA production activating AMPK to suppress mTORC1, through polyamine-mediated autophagy induction [58, 20], to dysbiosis-driven erosion of these protective inputs—collectively positions the gut microbial ecosystem not as a peripheral modulator but as a physiologically significant input to the growth-maintenance rheostat whose balance determines ageing trajectories [22, 23, 45].

This ecological dimension carries several implications for the interpretation and design of human geroprotector trials that existing frameworks have not adequately addressed. First, inter-individual variability in gut microbiome composition represents a previously uncharacterised source of heterogeneity in rapamycin response. If SCFA-producing microbial communities provide tonic AMPK-activating input that synergises with pharmacological mTORC1 inhibition, then individuals with depleted microbiomes—a common feature of ageing and Western dietary patterns [22, 44]—may achieve less effective mTOR suppression at a given rapamycin dose than individuals with intact microbial ecology. This microbiome-dependent pharmacodynamic variability could contribute to the substantial inter-individual heterogeneity in blood rapamycin levels and clinical response already documented in longevity cohorts [74, 46], and it suggests that microbiome profiling should be incorporated as a stratification variable or exploratory endpoint in future rapamycin trials [47].

Second, the finding that rapamycin’s core geroprotective effects are microbiota-independent in Drosophila [30] but that chronic rapamycin concurrently remodels both immune cell composition and the gut metagenome in mice [67] creates an unresolved question about the mammalian context. A further complication is introduced by evidence that mTORC2 suppression also drives gut microbial community shifts in diet-induced obese mice [71], suggesting that the reciprocal crosstalk between mTOR signalling and microbial ecology may operate through multiple complex branches. Whether rapamycin-induced immune remodelling secondarily reshapes the microbial community in ways that augment, attenuate, or are functionally neutral with respect to longevity has not been determined. Mammalian germ-free rapamycin studies—equivalent to the Drosophila experiments that cleanly dissociated drug effects from microbial effects—remain a critical gap in the evidence base. Until such studies are conducted, the relative contributions of cell-autonomous mTOR inhibition, immune restructuring, and microbiome remodelling to rapamycin’s mammalian longevity benefits cannot be partitioned with confidence.

Third, the demonstration that deliberate microbiome manipulation—through probiotic administration that elevates microbial spermidine production and suppresses colonic senescence [23]—can independently extend murine lifespan suggests that the gut microbial community represents a tractable therapeutic target whose manipulation engages the same autophagic quality-control programmes activated by rapamycin [20]. This opens the possibility of combinatorial strategies in which pharmacological mTOR inhibition is paired with microbiome-targeted interventions—dietary fibre supplementation, probiotics, or direct SCFA administration [44]—to achieve more complete engagement of the AMPK-mTOR-autophagy axis than either approach alone [87]. Such strategies would be particularly relevant for aged populations in whom both microbiome depletion and mTOR hyperactivation are simultaneously present [2, 110], creating a compound deficit that single-target interventions may not fully address.

Epigenetic Clocks, Biological Age Measurement, and the mTOR-Epigenome Axis

A development that has the potential to reshape both the design and the interpretability of geroprotector trials is the maturation of epigenetic clocks into quantitative instruments capable of tracking biological ageing with sufficient resolution to detect pharmacological effects. The convergence of this biomarker technology with mechanistic evidence that mTOR signalling directly influences the epigenetic landscape constitutes one of the most consequential recent developments in the field, bridging the molecular biology of mTOR with the measurement science of ageing in ways that were not previously achievable.

The case for epigenetic clocks as clinical endpoints rests on a progression from first-generation tools calibrated to chronological age [64, 63], through second-generation mortality-predictive clocks such as GrimAge and PhenoAge, to third-generation pace-of-ageing metrics such as DunedinPACE that capture longitudinal rate of biological decline [24, 25, 79]. This progression has been driven by recognition that clocks tracking chronological age too closely are methodologically unsuited to intervention assessment—a biomarker that perfectly predicts chronological age has, by definition, no residual variance through which to detect biological ageing rate differences [24]. The conceptual distinction between estimating biological age at a single timepoint and measuring the pace at which an individual is ageing is not merely semantic: it determines whether a trial of practical duration can detect a meaningful pharmacodynamic signal from a geroprotective intervention. The DunedinPACE metric, developed from the longitudinal Dunedin cohort, addresses this directly by quantifying a single-timepoint pace-of-ageing estimate validated against multi-decade longitudinal trajectories [38].

The mechanistic basis for expecting mTOR inhibition to influence epigenetic clock readouts is now substantively grounded. As detailed in Section 3, mTORC1 activity modulates histone modification patterns, TFEB-mediated transcriptional reprogramming, and the metabolic availability of methyl donors required for DNA methyltransferase activity [26, 39, 31]. Age-associated epigenetic deterioration—including the bidirectional pattern of global hypomethylation at repetitive elements and focal hypermethylation at CpG island promoters, loss of repressive histone marks at heterochromatic loci, and retrotransposon reactivation following heterochromatin erosion—represents the measurable molecular substrate upon which epigenetic clocks are trained [26, 27]. If mTOR hyperactivation accelerates these processes, and its pharmacological inhibition arrests or partially reverses them, then epigenetic clock deceleration following rapamycin treatment would constitute not merely a correlative biomarker shift but a mechanistically interpretable pharmacodynamic signal. This conceptual upgrade—from clocks as ancillary measures to clocks as candidate pharmacodynamic endpoints—has not yet been fully operationalised in completed human trials, but the mechanistic architecture now supports it.

Several methodological challenges temper this optimism. Clock accuracy degrades in the oldest age groups, precisely the population most relevant to geroprotective intervention [25]. Pan-tissue clocks implicitly assume a common epigenetic ageing process across cell types, but this assumption fails in specific pathological contexts—tissue-specific clocks demonstrably outperform pan-tissue alternatives in conditions characterised by accelerated ageing in particular lineages [25]. A persistent European-ancestry bias in clock training datasets limits cross-population generalisability [25, 24], and the observation that longevity-associated genetic variants modulate DNA methylation states through meQTLs [27] implies that genetic background should be treated as a confounder in clock-based intervention analyses. Furthermore, while the SASP is recognised as a driver of aberrant methylation patterns in bystander cells, the degree to which rapamycin’s senomorphic effects on SASP suppression contribute to, versus are independent of, its effects on epigenetic age remains unresolved.

Perhaps the most important analytical caution concerns the potential for discordance between lifespan extension and epigenetic age modification. The evidence reviewed across preceding sections demonstrates convincingly that rapamycin extends lifespan and delays age-related pathology across model organisms, yet the extent to which these functional benefits are captured by—or even correlated with—changes in DNA methylation-based age estimates has not been systematically tested in the same experimental systems. If lifespan extension and epigenetic age deceleration prove partially dissociable, the implications for endpoint selection in human trials would be substantial: epigenetic clocks might underestimate geroprotective efficacy in some contexts while overestimating it in others. Expert consensus increasingly favours composite endpoint strategies integrating epigenetic clocks with proteomic, metabolomic, functional, and digital biomarker modalities rather than reliance on any single metric [24, 79], an approach that would provide robustness against this risk while enabling the mechanistic decomposition of intervention effects across distinct biological dimensions of ageing.

Human trial evidence, particularly from the PEARL trial and the work with RTB101 in older adults, has meaningfully narrowed the uncertainty around short-course, low-dose rapalog regimens for immune enhancement in healthy ageing populations. The PEARL trial—a 48-week double-blind, randomised, placebo-controlled study of 115 participants aged 50–85 receiving weekly rapamycin at 5 mg or 10 mg—found dose-dependent and sex-specific improvements across multiple healthspan metrics, including lean tissue mass, bone mineral density, pain, and quality of life, with no significant differences in metabolic markers or serious adverse events compared to placebo [46]. Notably, the trial was conducted using a compounded formulation subsequently found to have approximately 3.5-fold lower bioavailability than commercial rapamycin, meaning the functional doses received were considerably lower than nominal, rendering the observed benefits more remarkable and highlighting the importance of pharmacokinetic standardisation in future trials [46]. Prior evidence from short-course low-dose everolimus and RTB101 regimens had already established that transient mTOR inhibition can restore age-related immune decline—including improved vaccine responses—without the infection risk or metabolic disruption associated with chronic immunosuppressive dosing, a finding that has been comprehensively reviewed in the context of rapalog clinical translation [35, 5, 47]. Nevertheless, the field still lacks randomised controlled trial evidence for the outcome that matters most—durable extension of human healthspan rather than surrogate biomarker improvement. The gap between mechanistic plausibility and clinical demonstration remains wide, and closing it will require agreement on validated biological age endpoints that current regulatory frameworks have not yet accommodated.

Limitations and Future Directions

This review is necessarily constrained by the heterogeneity of study designs, model systems, and outcome measures across included literature, which limits the strength of cross-theme synthesis. The predominance of male subjects and C57BL/6 mice in preclinical work introduces systematic bias that recent literature is only beginning to address [29]. Most mechanistic work linking mTOR to epigenetic regulation remains based in rodent models or cell culture systems, and few human geroprotector trials have yet reached completion with biological age as a pre-registered primary endpoint [47, 35]. The integration of microbiome-mTOR crosstalk findings, while substantively grounded in model organism evidence, is limited by the absence of mammalian germ-free rapamycin studies that would enable definitive partitioning of cell-autonomous from microbiome-mediated longevity effects, and by the nascent state of standardised microbiome profiling in human geroprotector trials.

Looking forward, the most productive research trajectories appear to lie at the intersections identified here: combining mTOR modulation with senolytic, NAD⁺-boosting, microbiome-targeted, or caloric restriction mimetic strategies in a rationally mechanistic rather than additive fashion [15, 111]; developing tissue-selective or temporally gated mTOR inhibitors that exploit the now-clearer network topology [5]; and investing in longitudinal human cohort studies with deep molecular phenotyping sufficient to test whether mTOR pathway activity predicts both biological ageing trajectories and therapeutic response [24]. The integrated nutrient-sensing network framework described above suggests that monitoring not only mTOR activity but also AMPK phosphorylation status, NAD⁺ levels, sirtuin-dependent deacetylation markers, and gut microbiome composition could substantially improve patient stratification and outcome prediction in future geroprotector trials [1, 4]. The gut-brain axis in neurodegeneration represents a particularly underexplored therapeutic frontier, where mTOR-autophagy dysregulation and microbial metabolite deficiency may interact to accelerate pathology in ways that neither field has yet examined jointly in human populations [45, 22]. Head-to-head comparison of epigenetic clock metrics as pharmacodynamic endpoints in rapamycin dosing trials stratified by baseline mTORC1 activity and microbiome composition represents a particularly high-priority design [24, 47], as does the development of polygenic and epigenetic stratification tools to identify subpopulations—those with constitutively elevated mTORC1 signalling, accelerated epigenetic pace, high SASP burden, or depleted SCFA-producing microbiota—most likely to benefit from targeted geroprotection [112, 87]. The field is positioned at a point where conceptual frameworks are strong but empirical translation remains the limiting factor — a productive tension that defines where the next phase of work must focus.

10. Conclusions

This systematic review of 56 papers, organized across six thematic domains and supplemented by integration of recent findings on gut microbiome-mTOR crosstalk, synthesizes the current state of evidence on mTOR signalling as a central regulator of aging and a tractable target for longevity intervention. The conclusions drawn across the four research questions converge on a coherent and actionable picture.

mTOR as a Growth-Maintenance Switch

The evidence establishes mTOR signalling, particularly through the mTORC1 complex, as a fundamental arbiter of the trade-off between anabolic growth and cellular maintenance [3]. Elevated mTOR activity promotes protein synthesis — via phosphorylation of 4E-BP1 and S6K1 — cell proliferation, and metabolic output [28], while simultaneously suppressing autophagy, stress response pathways, and damage clearance mechanisms by phosphorylating and inactivating key autophagy initiators including ULK1 and ATG13, and by retaining the lysosomal biogenesis factor TFEB in the cytoplasm [3, 31]. This suppression of maintenance processes, when sustained across the life course, is a primary driver of the cellular dysfunction associated with organismal aging [3]. Rapamycin’s inhibition of mTORC1 effectively rebalances this switch in favour of maintenance, providing a mechanistic rationale for its pro-longevity effects — demonstrated most compellingly by the finding that rapamycin extended maximum lifespan by 14% in female and 9% in male genetically heterogeneous mice even when treatment began at 600 days of age [8].

Model Organism Evidence Is Compelling

Across invertebrate and mammalian model systems, the evidence for rapamycin extending both lifespan and healthspan is robust [11]. Critically, lifespan benefits have been observed even when rapamycin treatment begins in mid-to-late life, demonstrating that aging trajectories remain modifiable well beyond developmental stages. The landmark Interventions Testing Program study by Harrison et al. showed that feeding rapamycin to genetically heterogeneous mice from 600 days of age—roughly equivalent to 60 human years—still yielded a 14% gain in maximal lifespan in females and 9% in males, with life expectancy from the point of treatment increasing by 38% and 28%, respectively [8]. Improvements in immune function, cognitive performance—including attenuation of age-related cognitive decline and reduced Alzheimer’s-associated pathology in transgenic models [11]—and tissue homeostasis in aged animals strengthen the case that these are not simply artefacts of lifespan extension but reflect genuine improvements in biological aging [12]. Controlled experiments in Drosophila have established that rapamycin’s core lifespan-extending mechanism operates through cell-autonomous autophagy induction independently of the gut microbiota [30], while in mammals the concurrent remodelling of immune cell populations and gut metagenome composition [67] suggests that additional microbiome-dependent mechanisms may contribute to—or modulate—longevity outcomes in more complex organisms.

Human Evidence Requires Cautious Optimism

Human trials of rapamycin and rapalogs have demonstrated measurable improvements in immune function and early biomarkers of aging-related decline, with the TRITON and related studies providing early proof-of-concept [5]. A 2024 systematic review of 19 human studies found rapalogs improved parameters in the immune, cardiovascular, and integumentary systems — notably, everolimus enhanced influenza vaccine responses and RTB101 reduced respiratory infections in adults aged 85 and older — though effects on muscular, endocrine, and neurological systems were absent [35]. The PEARL trial, the largest and longest randomized controlled trial of rapamycin in healthy aging adults to date (n=115, 48 weeks), found no significant differences in metabolic markers or organ function versus placebo, and observed dose-dependent, sex-specific improvements in lean tissue mass, bone mineral density, and quality of life [46]. However, safety concerns — including metabolic dysregulation, elevated infection risk, and context-dependent adverse effects — remain unresolved at doses and durations relevant to healthy aging populations [35, 47]. The human evidence base is presently insufficient to support broad clinical adoption, and larger, longer, and more rigorously controlled trials in non-diseased populations are urgently needed.

Pathway Interactions Offer Both Opportunity and Complexity

mTOR does not operate in isolation. Its interactions with AMPK, sirtuins, insulin/IGF-1 signalling, and the senescence machinery reveal a deeply interconnected longevity network in which AMPK-mediated phosphorylation of TSC2 and Raptor directly suppresses mTORC1, SIRT1-dependent deacetylation of LKB1 amplifies AMPK signalling through a positive feedback loop, and NAD⁺ metabolism links cellular energy status to sirtuin activity and downstream autophagic quality control [53, 81, 52]. These interactions create opportunities for combinatorial intervention strategies that may achieve greater efficacy at lower individual doses, potentially improving the therapeutic window. Empirical support for this principle is accumulating: trametinib and rapamycin, targeting the RAS–ERK and mTOR axes respectively, combine additively to extend mouse healthspan and lifespan at doses that individually produce more modest effects [75], illustrating how engagement of distinct but converging network nodes can yield amplified outcomes [5]. These interactions also introduce complexity that demands careful mechanistic investigation before combination approaches are tested clinically. The observation that caloric restriction, rapamycin, metformin, and NAD⁺ precursors produce overlapping but non-identical longevity phenotypes — explicable as engagement of different entry points within the same integrated network [15, 7] — provides both a conceptual foundation for rational multi-target geroprotection and a caution that network-level interventions may propagate effects in ways not fully predicted by single-node pharmacology. Perturbation at one node can induce compensatory responses elsewhere in the network, and the history of single-target pharmacology in complex biological systems warrants humility about the predictability of combination approaches prior to empirical validation.

Taken together, the four thematic conclusions of this review form a coherent, if still incomplete, picture of the field. The mechanistic case for mTOR as a growth-maintenance switch is well established and grounded in conserved biology; the model organism evidence for rapamycin’s capacity to modify aging trajectories is among the most reproducible in geroscience; the emerging human evidence, while promising, remains constrained by short durations, surrogate endpoints, and safety uncertainties that preclude broad clinical translation; and the pathway interaction data, though conceptually rich, introduce a degree of network complexity that demands mechanistic discipline in the design of combination strategies. These four bodies of evidence are not in tension — they are sequential steps in a translational chain that has advanced substantially but has not yet closed the gap between biological mechanism and demonstrated clinical benefit in healthy human populations.

Realising the clinical potential of mTOR modulation will depend on resolving several empirical prerequisites that this review repeatedly identifies as limiting factors. Chief among these are the establishment of validated biological age endpoints — including epigenetic clock metrics such as GrimAge and pace-of-aging measures [36, 37] and integrated multi-omic signatures [24, 113] — that are acceptable to regulatory frameworks and sufficiently sensitive to detect healthspan effects within trial timescales; the development of tissue-selective or temporally gated mTORC1 inhibitors [65, 5] that can exploit the network’s now-clearer topology without the metabolic and immunological liabilities of systemic rapalog exposure; and the standardisation of microbiome profiling within geroprotector trials sufficient to distinguish cell-autonomous from microbiome-mediated longevity contributions and to enable rational patient stratification. Progress on these three fronts would not merely accelerate rapamycin’s translational trajectory — it would establish the empirical infrastructure on which the broader programme of network-level geroprotection depends.

---

References

[1] Kennedy, B., Lamming, D. (2016). The Mechanistic Target of Rapamycin: The Grand ConducTOR of Metabolism and Aging. Cell Metabolism. [2] Johnson, S., Rabinovitch, P., Kaeberlein, M. (2012). mTOR is a key modulator of ageing and age-related disease. Nature. [3] Fernandes, S., Demetriades, C. (2021). The Multifaceted Role of Nutrient Sensing and mTORC1 Signaling in Physiology and Aging. Frontiers in Aging. [4] Papadopoli, D., Boulay, K., Kazak, L., et al. (2019). mTOR as a central regulator of lifespan and aging. F1000Research. [5] Mannick, J., Lamming, D. (2023). Targeting the biology of aging with mTOR inhibitors. Nature Aging. [6] Sarbassov, D., Ali, S., Sengupta, S., et al. (2006). Prolonged Rapamycin Treatment Inhibits mTORC2 Assembly and Akt/PKB. Molecular Cell. [7] Lamming, D., Ye, L., Sabatini, D., et al. (2013). Rapalogs and mTOR inhibitors as anti-aging therapeutics. Journal of Clinical Investigation. [8] Strong, R., Strong, R., Sharp, Z., et al. (2009). Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Nature. [9] Katewa, S., Kapahi, P. (2010). Role of TOR signaling in aging and related biological processes in Drosophila melanogaster. Experimental Gerontology. [10] Robida-Stubbs, S., Glover-Cutter, K., Lamming, D., et al. (2012). TOR Signaling and Rapamycin Influence Longevity by Regulating SKN-1/Nrf and DAF-16/FoxO. Cell Metabolism. [11] Selvarani, R., Mohammed, S., Richardson, A. (2020). Effect of rapamycin on aging and age-related diseases—past and future. GeroScience. [12] Blagosklonny, M. (2013). Rapamycin extends life- and health span because it slows aging. Aging. [13] Zukule, V., Iffland, P. (2025). Rapamycin for longevity: the pros, the cons, and future perspectives. Frontiers in Aging. [14] Schreiber, K., Ortiz, D., Academia, E., et al. (2015). Rapamycin-mediated mTORC2 inhibition is determined by the relative expression of FK506-binding proteins. Aging Cell. [15] Campisi, J., Kapahi, P., Lithgow, G., et al. (2019). From discoveries in ageing research to therapeutics for healthy ageing. Nature. [16] Sadria, M., Layton, A. (2021). Interactions among mTORC, AMPK and SIRT: a computational model for cell energy balance and metabolism. Cell Communication and Signaling. [17] Sharples, A., Hughes, D., Deane, C., et al. (2015). Longevity and skeletal muscle mass: the role of IGF signalling, the sirtuins, dietary restriction and protein intake. Aging Cell. [18] Marafie, S., Al‐Mulla, F., Abubaker, J. (2024). mTOR: Its Critical Role in Metabolic Diseases, Cancer, and the Aging Process. International Journal of Molecular Sciences. [19] Jung, C., Ro, S., Cao, J., et al. (2010). mTOR regulation of autophagy. FEBS Letters. [20] Madeo, F., Zimmermann, A., Maiuri, M., et al. (2015). Essential role for autophagy in life span extension. Journal of Clinical Investigation. [21] Vı́lchez, D., Sáez, I., Dillin, A. (2014). The role of protein clearance mechanisms in organismal ageing and age-related diseases. Nature Communications. [22] Theis, B., Park, J., Kim, J., et al. (2025). Gut Feelings: How Microbes, Diet, and Host Immunity Shape Disease. Biomedicines. [23] Matsumoto, M., Kurihara, S., Kibe, R., et al. (2011). Longevity in Mice Is Promoted by Probiotic-Induced Suppression of Colonic Senescence Dependent on Upregulation of Gut Bacterial Polyamine Production. PLoS ONE. [24] Moqri, M., Herzog, C., Poganik, J., et al. (2023). Biomarkers of Aging for the Identification and Evaluation of Longevity Interventions. Cell. [25] Bell, C., Lowe, R., Adams, P., et al. (2019). DNA methylation aging clocks: challenges and recommendations. Genome biology. [26] Saul, D., Kosinsky, R. (2021). 410 — This article came from a journal that’s no longer in DOAJ. International Journal of Molecular Sciences. [27] Ciaglia, E., Montella, F., Lopardo, V., et al. (2025). The Genetic and Epigenetic Arms of Human Ageing and Longevity. Biology. [28] Hay, N., Sonenberg, N. (2004). Upstream and downstream of mTOR. Genes & Development. [29] Apelo, S., Pumper, C., Baar, E., et al. (2016). Intermittent Administration of Rapamycin Extends the Life Span of Female C57BL/6J Mice. The Journals of Gerontology Series A. [30] Schinaman, J., Rana, A., Ja, W., et al. (2019). Rapamycin modulates tissue aging and lifespan independently of the gut microbiota in Drosophila. Scientific Reports. [31] Kim, Y., Guan, K. (2015). mTOR: a pharmacologic target for autophagy regulation. Journal of Clinical Investigation. [32] Perl, A. (2015). mTOR activation is a biomarker and a central pathway to autoimmune disorders, cancer, obesity, and aging. Annals of the New York Academy of Sciences. [33] Magaway, C., Kim, E., Jacinto, E. (2019). Targeting mTOR and Metabolism in Cancer: Lessons and Innovations. Cells. [34] Flatt, T., Partridge, L. (2018). Horizons in the evolution of aging. BMC Biology. [35] Lee, D., Kuerec, A., Maier, A. (2024). Targeting ageing with rapamycin and its derivatives in humans: a systematic review. The Lancet Healthy Longevity. [36] Lu, A., Binder, A., Zhang, J., et al. (2022). DNA methylation GrimAge version 2. Aging. [37] Belsky, D., Caspi, A., Corcoran, D., et al. (2021). DunedinPACE: A DNA methylation biomarker of the Pace of Aging. [38] Belsky, D., Caspi, A., Arseneault, L., et al. (2020). Quantification of the pace of biological aging in humans through a blood test: The DunedinPoAm DNA methylation algorithm. [39] Liu, R., Zhao, E., Yu, H., et al. (2023). Methylation across the central dogma in health and diseases: new therapeutic strategies. Signal Transduction and Targeted Therapy. [40] Zhu, X., Chen, Z., Shen, W., et al. (2021). Inflammation, epigenetics, and metabolism converge to cell senescence and ageing: the regulation and intervention. Signal Transduction and Targeted Therapy. [41] Ji, S., Xiong, M., Chen, H., et al. (2023). Cellular rejuvenation: molecular mechanisms and potential therapeutic interventions for diseases. Signal Transduction and Targeted Therapy. [42] Ozdemir, S., Faizan, M., Kaur, G., et al. (2025). Heterogeneity of Cellular Senescence, Senotyping, and Targeting by Senolytics and Senomorphics in Lung Diseases. Preprints.org. [43] Dodig, S., Čepelak, I., Pavić, I. (2019). Hallmarks of senescence and aging. Biochemia Medica. [44] Blaak, E., Canfora, E., Theis, S., et al. (2020). Short chain fatty acids in human gut and metabolic health. Beneficial Microbes. [45] Cryan, J., O’Riordan, K., Cowan, C., et al. (2019). The Microbiota-Gut-Brain Axis. Physiological Reviews. [46] Harinath, G., Lee, V., Nyquist, A., et al. (2024). Safety and efficacy of rapamycin on healthspan metrics after one year: PEARL Trial. [47] Konopka, A., Lamming, D., Grasso, B., et al. (2023). Blazing a trail for the clinical use of rapamycin as a geroprotecTOR. GeroScience. [48] Li, J., Kim, S., Blenis, J. (2014). Rapamycin: One Drug, Many Effects. Cell Metabolism. [49] Sun, Y., Wang, H., Qu, T., et al. (2023). mTORC2: a multifaceted regulator of autophagy. Cell Communication and Signaling. [50] Lamming, D., Ye, L., Katajisto, P., et al. (2012). Rapamycin-Induced Insulin Resistance Is Mediated by mTORC2 Loss and Uncoupled from Longevity. Science. [51] Soultoukis, G., Partridge, L. (2016). Dietary Protein, Metabolism, and Aging. Annual Review of Biochemistry. [52] Hardie, D., Ross, F., Hawley, S. (2012). AMPK: a nutrient and energy sensor that maintains energy homeostasis. Nature Reviews Molecular Cell Biology. [53] Jeon, S. (2016). Regulation and function of AMPK in physiology and diseases. Experimental & Molecular Medicine. [54] Hardie, D. (2011). AMP-activated protein kinase—an energy sensor that regulates all aspects of cell function. Genes & Development. [55] Steinberg, G., Carling, D. (2019). AMP-activated protein kinase: the current landscape for drug development. Nature Reviews Drug Discovery. [56] Radovic, M., Gartzke, L., Wink, S., et al. (2025). Targeting the Electron Transport System for Enhanced Longevity. Biomolecules. [57] Wang, G., Qin, S., Chen, L., et al. (2023). Butyrate dictates ferroptosis sensitivity through FFAR2-mTOR signaling. Cell Death and Disease. [58] Morselli, E., Galluzzi, L., Kepp, O., et al. (2009). Autophagy mediates pharmacological lifespan extension by spermidine and resveratrol. Aging. [59] Chambers, P. (2025). Vitamin D, Calcium to Magnesium Ratio, and the Gut Microbiome. Medical & Clinical Research. [60] Hansen, M., Chandra, A., Mitic, L., et al. (2008). A Role for Autophagy in the Extension of Lifespan by Dietary Restriction in C. elegans. PLoS Genetics. [61] Silvestrini, M., Johnson, J., Kumar, A., et al. (2018). Nuclear Export Inhibition Enhances HLH-30/TFEB Activity, Autophagy, and Lifespan. Cell Reports. [62] Guo, J., Huang, X., Dou, L., et al. (2022). Aging and aging-related diseases: from molecular mechanisms to interventions and treatments. Signal Transduction and Targeted Therapy. [63] Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome biology. [64] Hannum, G., Guinney, J., Zhao, L., et al. (2012). Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Molecular Cell. [65] Schreiber, K., Apelo, S., Yu, D., et al. (2019). A novel rapamycin analog is highly selective for mTORC1 in vivo. Nature Communications. [66] Lamming, D. (2016). Inhibition of the Mechanistic Target of Rapamycin (mTOR)—Rapamycin and Beyond. Cold Spring Harbor Perspectives in Medicine. [67] Hurez, V., Dao, V., Liu, A., et al. (2015). Chronic mTOR inhibition in mice with rapamycin alters T, B, myeloid, and innate lymphoid cells and gut flora and prolongs life of immune-deficient mice. Aging Cell. [68] Regan, J., Lu, Y., Ureña, E., et al. (2022). Sexual identity of enterocytes regulates autophagy to determine intestinal health, lifespan and responses to rapamycin. Nature Aging. [69] Sánchez‐Mendoza, L., González‐Reyes, J., López, S., et al. (2025). Adaptations of mitochondrial, autophagy and nutrient sensing pathways in the liver from long-lived mice overexpressing CYB5R3 are sex-dependent and involve inter-organ responses. GeroScience. [70] Mallén‐Ponce, M., Pérez‐Pérez, M., Crespo, J. (2022). Deciphering the function and evolution of the target of rapamycin signaling pathway in microalgae. Journal of Experimental Botany. [71] Jung, M., Lee, J., Shin, N., et al. (2016). Chronic Repression of mTOR Complex 2 Induces Changes in the Gut Microbiota of Diet-induced Obese Mice. Scientific Reports. [72] Wu, J., Liu, J., Chen, E., et al. (2013). Increased Mammalian Lifespan and a Segmental and Tissue-Specific Slowing of Aging after Genetic Reduction of mTOR Expression. Cell Reports. [73] Walters, H., Cox, L. (2018). 410 — This article came from a journal that’s no longer in DOAJ. International Journal of Molecular Sciences. [74] Harinath, G., Lee, V., Nyquist, A., et al. (2025). The bioavailability and blood levels of low‑dose rapamycin for longevity in real‑world cohorts of normative aging individuals. GeroScience. [75] Gkioni, L., Nespital, T., Baghdadi, M., et al. (2025). The geroprotectors trametinib and rapamycin combine additively to extend mouse healthspan and lifespan. Nature Aging. [76] Driscoll, M., Sedore, C., Onken, B., et al. (2025). NIA Caenorhabditis Intervention Testing Program: identification of robust and reproducible pharmacological interventions that promote longevity across experimentally accessible, genetically diverse populations. GeroScience. [77] Feng, Q., Chen, W., Wang, Y. (2018). Gut Microbiota: An Integral Moderator in Health and Disease. Frontiers in Microbiology. [78] Svensson, J., Bolin, M., Thor, D., et al. (2024). Evaluating the effect of rapamycin treatment in Alzheimer’s disease and aging using in vivo imaging: the ERAP phase IIa clinical study protocol. BMC Neurology. [79] Ferrucci, L., González‐Freire, M., Fabbri, E., et al. (2020). Measuring biological aging in humans: A quest. Aging Cell. [80] Salminen, A., Kaarniranta, K., Kauppinen, A. (2013). 410 — This article came from a journal that’s no longer in DOAJ. International Journal of Molecular Sciences. [81] Guan, G., Chen, Y., Dong, Y. (2025). Unraveling the AMPK-SIRT1-FOXO Pathway: The In-Depth Analysis and Breakthrough Prospects of Oxidative Stress-Induced Diseases. Antioxidants. [82] Grabowska, W., Sikora, E., Bielak-Żmijewska, A. (2017). Sirtuins, a promising target in slowing down the ageing process. Biogerontology. [83] Alpay, F. (2025). Boosting NAD+ for Anti-Aging: Mechanisms, Interventions, and Opportunities. [84] Vachharajani, V., Liu, T., Wang, X., et al. (2016). Sirtuins Link Inflammation and Metabolism. Journal of Immunology Research. [85] Odoh, C., Guo, X., Arnone, J., et al. (2022). The role of NAD and NAD precursors on longevity and lifespan modulation in the budding yeast, Saccharomyces cerevisiae. Biogerontology. [86] Wang, Z., Quan, W., Zeng, M., et al. (2023). Regulation of autophagy by plant-based polyphenols: A critical review of current advances in glucolipid metabolic diseases and food industry applications. Food Frontiers. [87] Todorova, M., Savova, M., Mihaylova, L., et al. (2024). Nurturing longevity through natural compounds: Where do we stand, and where do we go?. Food Frontiers. [88] Fekete, M., Szarvas, Z., Fazekas‐Pongor, V., et al. (2022). Nutrition Strategies Promoting Healthy Aging: From Improvement of Cardiovascular and Brain Health to Prevention of Age-Associated Diseases. Nutrients. [89] Margulis, B., Цимоха, А., Зубова, С., et al. (2020). Molecular Chaperones and Proteolytic Machineries Regulate Protein Homeostasis in Aging Cells. Cells. [90] Labbadia, J., Morimoto, R. (2015). The Biology of Proteostasis in Aging and Disease. Annual Review of Biochemistry. [91] Klaips, C., Jayaraj, G., Hartl, F. (2018). Pathways of cellular proteostasis in aging and disease. The Journal of Cell Biology. [92] Ben‐Zvi, A., Miller, E., Morimoto, R. (2009). Collapse of proteostasis represents an early molecular event in Caenorhabditis elegans aging. Proceedings of the National Academy of Sciences. [93] Buttari, B., Tramutola, A., Rojo, A., et al. (2025). Proteostasis Decline and Redox Imbalance in Age-Related Diseases: The Therapeutic Potential of NRF2. Biomolecules. [94] Sun, Y., Jin, L., Qin, Y., et al. (2024). Harnessing Mitochondrial Stress for Health and Disease: Opportunities and Challenges. Biology. [95] Smith, R., Soeters, M., Wüst, R., et al. (2018). Metabolic Flexibility as an Adaptation to Energy Resources and Requirements in Health and Disease. Endocrine Reviews. [96] Birch, J., Gil, J. (2020). Senescence and the SASP: many therapeutic avenues. Genes & Development. [97] Robbins, P., Jurk, D., Khosla, S., et al. (2021). Senolytic Drugs: Reducing Senescent Cell Viability to Extend Health Span. The Annual Review of Pharmacology and Toxicology. [98] Suda, M., Paul, K., Minamino, T., et al. (2023). Senescent Cells: A Therapeutic Target in Cardiovascular Diseases. Cells. [99] Glaviano, A., Foo, A., Lam, H., et al. (2023). PI3K/AKT/mTOR signaling transduction pathway and targeted therapies in cancer. Molecular Cancer. [100] Hua, H., Kong, Q., Zhang, H., et al. (2024). Targeting mTOR for cancer therapy. Journal of Hematology & Oncology. [101] Sanli, T., Steinberg, G., Singh, G., et al. (2014). AMP-activated protein kinase (AMPK) beyond metabolism: A novel genomic stress sensor participating in the DNA damage response pathway. Cancer Biology & Therapy. [102] Xu, Q., Wang, W., Li, Y., et al. (2025). The oral-gut microbiota axis: a link in cardiometabolic diseases. npj Biofilms and Microbiomes. [103] Picos, A., Seoane, N., Campos‐Toimil, M., et al. (2025). Vascular senescence and aging: mechanisms, clinical implications, and therapeutic prospects. Biogerontology. [104] Lemaître, J., Moorad, J., Gaillard, J., et al. (2024). A unified framework for evolutionary genetic and physiological theories of aging. PLoS Biology. [105] Barzilai, D. (2025). Mikhail ‘Misha’ Blagosklonny’s enduring legacy in geroscience: the hyperfunction theory and the therapeutic potential of rapamycin. Aging. [106] Blagosklonny, M. (2011). Hormesis does not make sense except in the light of TOR-driven aging. Aging. [107] Aronoff, J., Trumble, B. (2025). An evolutionary medicine and life history perspective on aging and disease: Trade-offs, hyperfunction, and mismatch. Evolution Medicine and Public Health. [108] Regan, J., Froy, H., Walling, C., et al. (2020). Dietary restriction and insulin-like signalling pathways as adaptive plasticity: A synthesis and re-evaluation. Functional Ecology. [109] Venn‐Watson, S., Schork, N. (2023). Pentadecanoic Acid (C15:0), an Essential Fatty Acid, Shares Clinically Relevant Cell-Based Activities with Leading Longevity-Enhancing Compounds. Nutrients. [110] Franceschi, C., Garagnani, P., Morsiani, C., et al. (2018). The Continuum of Aging and Age-Related Diseases: Common Mechanisms but Different Rates. Frontiers in Medicine. [111] Suda, M., Paul, K., Tripathi, U., et al. (2024). Targeting Cell Senescence and Senolytics: Novel Interventions for Age-Related Endocrine Dysfunction. Endocrine Reviews. [112] López-Otı́n, C., Pietrocola, F., Roiz‐Valle, D., et al. (2022). Meta-hallmarks of aging and cancer. Cell Metabolism. [113] Justice, J., Ferrucci, L., Newman, A., et al. (2018). A framework for selection of blood-based biomarkers for geroscience-guided clinical trials: report from the TAME Biomarkers Workgroup. GeroScience.