A Cheap Drug Trio Turns Cancer and Senescent Cells' Broken Metabolism Against Them, and Old Mice Live Longer

[RapAdmin]

Researchers at UC Berkeley combined two inexpensive, widely available metabolic drugs, dichloroacetate and metformin, with a ten-fold reduced dose of the experimental senolytic navitoclax (ABT-263) to create a three-drug cocktail they call DMA. The core idea is that both cancer cells and worn-out senescent cells share the same metabolic weakness: leaky, inefficient mitochondria and an over-reliance on fragile ATP production. DMA pushes these already-strained cells over the edge into an energy collapse, while healthy cells with robust metabolism shrug it off. In cell culture DMA killed multiple cancer lines and several types of senescent cells, including navitoclax-resistant ones, without the platelet toxicity that has stalled navitoclax in the clinic. In old mice, short courses improved treadmill endurance, and prolonged dosing modestly extended lifespan. The work is preliminary, mechanistically interesting, and comes with significant caveats about sample size and translation.

For years the dream of treating aging and cancer with the same drug has run into the same wall: the drugs that kill senescent cells, the so-called senolytics, tend to be toxic. Navitoclax, one of the most promising, drops platelet counts so sharply that it can cause dangerous bleeding, and many cancers simply ignore it. The Berkeley team asked a different question. Instead of hitting these cells harder, could you exploit what makes them fragile in the first place?

Their answer targets metabolism. Cancer cells and senescent cells both run their mitochondria poorly. They leak protons, waste energy, and lean heavily on glycolysis and a thin margin of ATP to stay alive. The team reasoned that if you gently sabotage energy production with two old metabolic drugs, metformin and dichloroacetate, you could make these cells so energetically brittle that even a small, normally safe dose of navitoclax would finish them off. Healthy cells, with efficient mitochondria and spare capacity, would ride it out.

That is roughly what happened in the dish. The combination, DMA, wiped out senescent lung fibroblasts and killed cancer lines from cervical, breast, and colon tumors, including a breast line that resists navitoclax on its own. It did this by draining cellular ATP to near zero in the sick cells while leaving healthy neurons, liver cells, and muscle precursors largely intact. Crucially, the reduced navitoclax dose no longer crashed platelet counts.

The animal data are more modest but intriguing. Old mice given short courses ran longer on a treadmill and showed no loss of strength, balance, or increase in frailty. Their blood showed a drop in inflammatory and senescence-associated proteins, nudging their molecular profile toward that of younger animals. When dosing continued for life, treated mice lived a median of 187 days longer than controls, and their serum proteome drifted younger.

The significance is conceptual as much as practical. Rather than inventing a new toxic drug, the team repurposed cheap, familiar molecules by aiming at a shared metabolic soft spot. If it holds up in larger, better-controlled studies, it points toward affordable, low-toxicity ways to clear the cellular debris that drives both aging and cancer. That “if” is still doing heavy lifting.

Actionable Insights

The take-home message is that this is a mouse and cell study, so there is nothing here to act on directly. What it offers is a direction and a sense of scale.

The headline benefit is lifespan. Mice that started treatment at old age (around 18 months) and continued for life reached a median lifespan of 1002 days versus 815 days for untreated mice. That is 187 extra days, a 22.9 percent longer median lifespan. Averaged across all animals the gain was smaller, roughly 12 to 14 percent. Measured only over the time after treatment began, survival stretched by about 41.7 percent, but that larger-sounding number is inflated by starting the clock late in life.

The functional benefit was endurance. Old treated mice improved their month-over-month treadmill running time significantly more than controls, without losing grip strength or gaining frailty. In cells, the effect sizes were large: senescent cells were cleared by roughly 60 to 90 percent, cancer-cell apoptosis rose about tenfold, and cancer-cell division dropped about sevenfold.

Two of the three ingredients, metformin and dichloroacetate, are already human drugs, which is why this line of work matters. But the third is not benign, the doses and delivery were tuned for mice, and no human has taken this combination. Treat it as a research signal, not a protocol.

Context and Source

• Open Access Paper: Selective targeting of cancer and senescence via shared metabolic shifts extends lifespan of old mice.
• Institution and country: Department of Bioengineering and QB3 Institute, University of California, Berkeley, USA. Senior author Irina M. Conboy. Note a declared conflict of interest: the senior author is co-founder and Chief Scientific Officer of Generation Lab, though the paper states the work was conducted independently of the company.
• Journal: Aging (also known as Aging-US), published by Impact Journals
  Impact evaluation: The most recent Journal Impact Factor is about 3.9 (2023 edition), trending down to roughly 3.2 in the 2024 edition, from a historical peak near 6.4 in 2014. Using the current figure: The impact score of this journal is approximately 3.9, evaluated against a typical high-end range of 0 to 60+ for top general science journals, therefore this is a Low-to-Medium impact journal. Two reputational caveats belong alongside that number: the journal appeared on Beall’s list of potentially predatory open-access journals in 2015, and it has faced criticism over self-citation practices. This does not invalidate the science, but it lowers the prior on stringent peer review.

[RapAdmin]

Lifespan of Controls:

In this study the control median lifespan was 815 days, and the pooled control average was about 851 days. That sits below the 850 to 950 day target band, so these controls are mildly short-lived by the 900-day standard, roughly 85 days under the 900-day midpoint. That will inflate the apparent extension to some degree. [Confidence: Medium-High.]

One important nuance cuts the other way. The DMA-treated median of 1002 days actually exceeds the 900-day threshold. A treated group living beyond the gold-standard control lifespan is more persuasive than the common failure mode where a “long-lived” treated group merely reaches what a healthy control should have been. So the result is not simply an artifact of sick controls, though shortish controls do amplify the headline percentage. [Confidence: Medium.]

Lifespan and Biomarker Data (Effect Size Extraction)

Lifespan, prolonged dosing (Figure 4G):

Median lifespan rose from 815 days (control) to 1002 days (DMA). That is an absolute gain of 187 days and a relative gain of 22.9 percent in median lifespan.

Average total lifespan rose from about 851 days to about 972 days, a gain of roughly 12 to 14 percent, which matches the paper’s stated 12 percent.

Split by sex, the effect is carried mostly by males: male average 993.8 versus 809.3 days (plus 184.5 days, plus 22.8 percent), female average 945.75 versus 872.5 days (plus 73.2 days, plus 8.4 percent). The authors concede the sex-specific analyses did not reach statistical significance because of small numbers. This is a real weakness. The impressive pooled number leans heavily on five male responders. [Confidence: Medium.]

Post-treatment survival extension was reported as 102.6 days, or 41.7 percent. This larger percentage is an artifact of measuring only the remaining lifespan after dosing began near 18 months of age, so it should not be compared directly with the 22.9 percent median figure. [Confidence: High.]

Maximum lifespan: Not explicitly tabulated. The survival curve runs to roughly 450 to 500 days post-treatment, but the paper gives no formal maximum-lifespan statistic (for example, mean of the longest-lived decile), so a maximum-extension effect size cannot be calculated. This is missing data. [Confidence: High that it is absent.]

Standardized effect size for survival: The survival difference was significant by log-rank (Mantel-Cox) test at p less than 0.05. The paper does not report a hazard ratio. From the median shift and curve separation, the mortality hazard ratio is plausibly in the region of 0.4 to 0.55 (roughly a 45 to 60 percent lower instantaneous risk of death in treated mice), but with n = 9 per group the confidence interval around any such estimate would be very wide and likely cross into non-significance for either sex alone. Treat the HR as an approximation, not a reported value. [Confidence: Low.]

Biomarker and physiological effect sizes. A caution first: the paper reports these as bar graphs with mean plus or minus SD or SEM but does not tabulate the underlying standard deviations, so formal Cohen’s d values cannot be computed exactly. The figures below are relative effect sizes (fold-changes and percentages) read from the data, with estimated standardized magnitudes flagged as such.

Cancer-cell apoptosis (TUNEL, MCF-7, Figure 2C): about 52 percent TUNEL-positive with DMA versus about 5 percent control and about 11 percent with navitoclax alone. That is roughly a 10-fold increase over vehicle and about 5-fold over navitoclax. Given the wide separation and modest error bars, the standardized effect is large, estimated Cohen’s d well above 2. [Confidence: Medium, estimate.]

Cancer-cell proliferation (EdU, MCF-7, Figure 2E): about 6 percent proliferating with DMA versus about 45 percent control, roughly a 7.5-fold reduction. Estimated large effect. [Confidence: Medium, estimate.]

Senescent-cell clearance: about 60 percent ablation of damage-induced senescent cells under standard culture, rising to about 90 percent under physiologic 3 percent oxygen. Selectivity held against navitoclax-resistant contexts. [Confidence: Medium-High.]

Platelet safety (Figure 1F): DMA retained about 70 percent of normal platelet counts and was not statistically different from control, whereas conventional-dose navitoclax cut platelets to roughly 25 percent (p less than 0.01). The safety margin, not just the kill, is the selling point here. [Confidence: High.]

Functional endurance (Figure 4B): DMA-treated old mice showed significantly greater month-over-month improvement in treadmill run time than controls (p less than 0.05, unpaired t-test), on the order of a 20 percentage point larger gain, with no significant change in hang-test strength or frailty. Because the data are percent-change with SEM and small n, a precise Cohen’s d is not recoverable, but the effect is statistically present and functionally meaningful. [Confidence: Medium.]

Serum proteome and SASP: Antibody arrays (n = 3) showed a general downward shift in pro-inflammatory and senescence-associated secretory proteins after DMA, trending toward a younger profile. With n = 3 this is suggestive, not robust.

[RapAdmin]

Mechanistic Deep Dive

The unifying mechanism is selective ATP catastrophe through convergent metabolic pressure, exploiting the shared mitochondrial dysfunction of senescent and malignant cells.

AMPK and oxidative phosphorylation (metformin): Metformin inhibits mitochondrial complex I, attenuating oxidative phosphorylation and, through the resulting energy stress, engaging AMPK and suppressing gluconeogenesis. In this cocktail its role is to cap the mitochondrial ATP supply. [Confidence: High on known pharmacology.]

Pyruvate routing (dichloroacetate): DCA inhibits pyruvate dehydrogenase kinase, forcing pyruvate into the mitochondria and shifting cells away from glycolysis toward oxidative phosphorylation. In healthy cells this is tolerated. In senescent and cancer cells, which the paper shows have reduced glycolytic reserve, removing the glycolytic escape route while metformin throttles oxidative phosphorylation is what collapses ATP. The Seahorse data are the mechanistic core: senescent cells could not raise respiration in response to the uncoupler FCCP after DMA, indicating no spare respiratory capacity. [Confidence: Medium-High.]

Apoptotic priming (low-dose navitoclax): At one tenth the usual dose, ABT-263 inhibits BCL-2 and BCL-XL. On its own this dose is a poor senolytic, but against cells already energetically gutted by DCA plus metformin it tips them into apoptosis. Notably the effect extended to MCF-7 cells that overexpress MCL-1 (the escape route navitoclax cannot block), and the authors show DMA acts largely independently of MCL-1 levels, arguing the metabolic hit, not BCL-2 family rebalancing, is doing the work. [Confidence: Medium.]

Novelty

What is new as of this paper: the specific demonstration that combining two cheap, approved metabolic drugs (metformin, dichloroacetate) with a sub-toxic dose of navitoclax can selectively kill both senescent and cancer cells by ATP depletion, that this overcomes navitoclax resistance driven by MCL-1, that it does so without the platelet toxicity that limits navitoclax, and that prolonged dosing in already-old mice is associated with extended lifespan and a younger serum proteome. The conceptual novelty is reframing senolysis and oncology around a shared, exploitable metabolic fragility rather than around apoptosis machinery alone. The individual drugs and their anti-senescence or anti-cancer hints were known; the tuned combination and the ATP-collapse selectivity argument are the addition.

[Joseph]

FWIW…

Quick AI Overview - Dosing

This is for informational purposes only. For medical advice or diagnosis, consult a professional. AI responses may include mistakes.

In the landmark July 2026 preclinical study led by UC Berkeley researchers, the precise doses utilized for the DMA combination in animal models were as follows:

Mouse Model Dosing (In Vivo)

For the live mouse trials—which successfully extended the lifespan of aged mice—the daily dosages administered were:

** Navitoclax (ABT-263): 5 mg/kg per day. (This represents a 10-fold reduction from the conventional experimental mouse dose of 50 mg/kg, effectively bypassing blood platelet toxicity)*
.
** Dichloroacetate (DCA): 50 mg/kg per day.*

** Metformin: 50 mg/kg per day.*

The Regimen: The animals received these doses via a specific sequence: 5 consecutive days of treatment, followed by a 2-day intermission, repeated for a second week

Anyone want to do a conversion to human?

[RapAdmin]

A more in-depth and translational dosing analysis:

The Translational Protocol (Rigorous Extrapolation)

First, the dosing regimen actually used in the mice, because it is not continuous dosing and that changes everything downstream. For the lifespan arm the animals received the three drugs once daily for five consecutive days, then two days off, then five more consecutive days, then a six to eight week break, repeated until natural death. So the exposure is heavily pulsed, roughly two weeks of dosing per two-month cycle. In-mouse doses were: dichloroacetate 50 mg/kg subcutaneous, metformin 50 mg/kg subcutaneous, and ABT-263 (navitoclax) 5 mg/kg oral.

Human Equivalent Dose (HED), FDA body-surface-area normalization

Formula (FDA guidance, 2005): HED (mg/kg) = Animal dose (mg/kg) x (Animal Km / Human Km), where mouse Km = 3 and human Km = 37. The Km ratio is 3 / 37 = 0.0811, equivalent to dividing the mouse mg/kg dose by 12.33.

Metformin: 50 mg/kg x (3/37) = 4.05 mg/kg HED. For a 60 kg adult: 4.05 x 60 = 243 mg/day. For 70 kg: 284 mg/day. For 80 kg: 324 mg/day.

Dichloroacetate: 50 mg/kg x (3/37) = 4.05 mg/kg HED. For 60 kg: 243 mg/day. For 70 kg: 284 mg/day. For 80 kg: 324 mg/day.

Navitoclax: 5 mg/kg x (3/37) = 0.405 mg/kg HED. For 60 kg: 24 mg/day. For 70 kg: 28 mg/day. For 80 kg: 32 mg/day.

Sanity check against real clinical doses. The metformin HED (about 284 mg/day at 70 kg) is roughly 14 percent of a standard 2000 mg/day diabetic dose, so the mouse metformin exposure is sub-therapeutic by human metabolic standards. The navitoclax HED (about 28 mg/day at 70 kg) is roughly 9 percent of the 325 mg/day oncology dose, consistent with the paper’s “ten-fold reduced dose” framing and the whole point of the design. The DCA HED (about 4 mg/kg/day) is well under the 12.5 to 50 mg/kg clinical antihyperlactatemia range. [Confidence: High on the arithmetic; Low that BSA scaling predicts an efficacious human dose, since BSA-HED is a starting-dose safety tool for toxicology, not an efficacy predictor, and two of the three drugs were dosed subcutaneously in mice while any human protocol would be oral.]

Important extrapolation caveat: BSA-HED assumes the pharmacologically relevant exposure scales with surface area. It does not account for the pulsed schedule, the subcutaneous-to-oral route change (which matters for metformin, whose oral bioavailability is only about 50 to 60 percent, and less for DCA at near 100 percent), or species differences in target biology. Treat these HED numbers as an order-of-magnitude anchor, not a prescription.

Pharmacokinetics (PK/PD)

Metformin: oral bioavailability approximately 50 to 60 percent, decreasing with dose. Plasma half-life approximately 4 to 9 hours (longer in erythrocytes). Not metabolized; excreted unchanged by the kidney via OCT2 and MATE transporters. PD marker: reduced fasting glucose and insulin, and AMPK activation.

Dichloroacetate: oral bioavailability approximately 100 percent. Half-life is dose-dependent and unusual: under one hour on the first dose, rising to several hours with repeated dosing because DCA inhibits its own metabolizing enzyme (GSTZ1, glutathione transferase zeta 1). This causes accumulation with chronic dosing, which is directly relevant to its neuropathy risk. PD marker: plasma lactate falls by more than 60 percent at 35 to 50 mg/kg, a clean readout of pyruvate dehydrogenase activation (target engagement).

Navitoclax (ABT-263): terminal plasma half-life approximately 15 hours. Oral, with absorption increased by a high-fat meal. It is a CYP3A4 substrate and highly protein-bound. PD marker: platelet count decline, which is the on-target consequence of BCL-XL inhibition. [Confidence: High, from clinical literature.]

Safety and Toxicity

Metformin: NOAEL / LD50: oral LD50 in rodents is high (on the order of 1000 mg/kg or greater), so acute lethality is low. The clinically meaningful risk is not acute toxicity. Signals: metformin-associated lactic acidosis (rare but serious, primarily in renal impairment), vitamin B12 deficiency with chronic use, and gastrointestinal intolerance. Contraindicated at eGFR below 30, dose-limited below 45. CYP450: negligible. Metformin is not a CYP substrate, inhibitor, or inducer. Interactions are transporter-mediated (OCT2/MATE): cimetidine, dolutegravir, ranolazine, and iodinated contrast (renal) raise levels.

Dichloroacetate: NOAEL / LD50: this is where data thin out for chronic human use. The dose-limiting human toxicity is reversible peripheral neuropathy (length-dependent, axonal, sensorimotor), demonstrated in a randomized controlled MELAS trial that was halted for neuropathy. Neuropathy risk is age-dependent (worse in adults than children) and pharmacogenetically modified by GSTZ1 haplotype; slow metabolizers accumulate DCA and are at higher risk. Hepatic effects and, in animals, carcinogenicity signals at high chronic doses have been reported. Phase I: DCA has been through multiple small human trials (mitochondrial disease, PDH deficiency, some cancers) but has no FDA approval for any indication. CYP450: primary concern is GSTZ1-mediated self-inhibition and accumulation rather than CYP. [Confidence: High on neuropathy as dose-limiting.]

Navitoclax (ABT-263): NOAEL / LD50: not the relevant frame; the human dose-limiting toxicity is well characterized. Phase I / II safety: dose-limiting, on-target, dose-dependent thrombocytopenia. Grade 3 to 4 thrombocytopenia occurred in about 41 percent of patients in single-agent small-cell lung cancer studies at 325 mg/day. Neutropenia and gastrointestinal effects also occur. Because platelets depend on BCL-XL for survival, this toxicity is mechanistic and unavoidable at BCL-XL-inhibiting exposures, though it is dose-proportional (the paper’s whole rationale for the reduced dose). CYP450: CYP3A4 substrate. Strong CYP3A inhibitors (ketoconazole, ritonavir, grapefruit) would raise exposure and thrombocytopenia risk; strong inducers (rifampin, St John’s wort, carbamazepine) would lower it.

Safety of the DMA combination specifically in humans: Safety Data Absent. There is no human PK, no human safety study, and no toxicology of the three-drug combination. The mouse data show the reduced navitoclax dose preserved about 70 percent of platelets and no acute frailty increase, but that is a two-week mouse readout, not a human safety package.

Biomarker Verification (Target Engagement)

The paper and the drugs’ known pharmacology give a set of measurable markers that would confirm each arm is hitting its target:

Pyruvate dehydrogenase activation (DCA): plasma or serum lactate should fall. This is the most direct, cheapest target-engagement marker in the whole cocktail.

AMPK / respiratory inhibition (metformin): reduced fasting insulin and glucose, modest lactate rise from metformin alone (opposed by DCA here), and in research settings phospho-AMPK in accessible cells.

BCL-XL inhibition (navitoclax): platelet count decline is the unavoidable on-target pharmacodynamic marker. A drop confirms engagement but is also the toxicity.

Senescent-cell and SASP reduction (the actual longevity endpoint): the paper reports reduced serum SASP and pro-inflammatory proteins trending younger. Translatable markers a specialist could track: circulating IL-6, IL-1-alpha, MMP-3, TNF-alpha, and where measurable p16INK4a expression in peripheral blood T cells or tissue SA-beta-gal. The paper’s in vitro proof of engagement was cellular ATP depletion plus TUNEL-positive apoptosis in target cells, which are not practical human biomarkers.

Bottom line on verification: the cocktail has good, cheap PD markers for the two metabolic arms (lactate, glucose/insulin) and a double-edged one for navitoclax (platelets). The senescence endpoint relies on inflammatory and p16-type panels that are noisy and not standardized. [Confidence: Medium-High.]

Feasibility and ROI

Sourcing: Metformin: prescription, generic, ubiquitous. Trivial to obtain legitimately. Dichloroacetate: not FDA-approved for any use. Available as sodium dichloroacetate through compounding pharmacies in some jurisdictions and, unfortunately, widely as a gray-market research chemical marketed to cancer patients. Purity and legality vary by country. This is a research chemical for practical purposes. Navitoclax: investigational oncology drug, never approved, available to humans only inside clinical trials. Sold by chemical suppliers strictly for laboratory research, not for human consumption. This is the hard gate: there is no legitimate route for a person to obtain pharmaceutical-grade navitoclax for self-use, and using research-grade material carries real thrombocytopenia risk with no medical monitoring.

Cost versus effect: Raw ingredient cost is low. Metformin is a few dollars a month. DCA as a research chemical is inexpensive. Even navitoclax, priced as a lab reagent, is not the binding constraint. At the theoretical HEDs above (roughly 284 mg/day metformin, 284 mg/day DCA, 28 mg/day navitoclax for a 70 kg adult, pulsed two weeks in eight), monthly drug cost would plausibly be in the low tens of dollars.

The problem is not cost, it is the numerator over the denominator. The marginal gain is entirely unproven in humans and, even in mice, rests on nine animals per group with mildly short-lived controls and a male-driven effect. The navitoclax component carries a real, mechanistic bleeding risk, and DCA carries a real, cumulative neuropathy risk that is worse in adults and genetically variable.

[RapAdmin]

Part 5: The Strategic FAQ

Ten questions a skeptical longevity specialist would put to the lead author, with the best current answers from the paper and external literature.

1. Your survival cohorts were nine mice per group and sex-imbalanced, with the effect driven by males. Is the lifespan claim adequately powered, and will you replicate with sex-stratified, adequately powered groups? Answer from the data: it is not adequately powered. Median went 815 to 1002 days (plus 22.9 percent), but neither sex reached significance alone and the pooled log-rank was only p less than 0.05. The paper concedes the sample size limits sex-specific conclusions. This needs independent replication with tens of animals per sex per arm. [Confidence: High.]

2. Your control median was 815 days. Against the 900-day rule for healthy C57BL/6 controls, are your controls short-lived, and does that inflate the effect? Answer: yes, partly. 815 days sits below the 850 to 950 day target band, so the extension is somewhat inflated by shortish controls. The mitigating point is that treated mice reached 1002 days, exceeding the 900-day standard, so it is not purely a sick-controls artifact. [Confidence: Medium-High.]

3. All the anti-cancer data are in vitro. Do you have any in vivo tumor model showing DMA controls or prevents cancer? Answer: no. The oncology case is entirely cell-culture (HeLa, MCF-7, SW480). There is no tumor-bearing or cancer-prone-strain in vivo experiment. The cancer claim is mechanistically plausible but not demonstrated in an animal. [Confidence: High.]

4. What is the actual systemic exposure of each drug at your mouse doses, and how does that map to human PK given navitoclax is a CYP3A4 substrate and DCA self-inhibits its own metabolism? Answer: the paper does not report plasma PK for any component in the mice. Externally, navitoclax half-life is about 15 hours and CYP3A4-dependent, and DCA accumulates on repeat dosing via GSTZ1 inhibition. Without measured exposures, the mouse-to-human translation is guesswork. Safety Data Absent for combination PK. [Confidence: High.]

5. Navitoclax thrombocytopenia is on-target BCL-XL inhibition. You preserved about 70 percent of platelets in mice at the reduced dose, but is that margin real in humans, where 325 mg/day gives grade 3 to 4 thrombocytopenia in about 41 percent? Answer: unknown in humans. The reduced dose scales to roughly 28 mg/day HED, about 9 percent of the oncology dose, which should reduce but not eliminate platelet suppression. Mouse platelet biology is not a reliable proxy for the human bleeding threshold. [Confidence: Medium.]

6. DCA causes cumulative, age-dependent peripheral neuropathy and is GSTZ1-genotype sensitive. How do you propose to dose chronically in older humans, the exact population at highest neuropathy risk? Answer: the paper does not address this. It is a serious translational obstacle, since the target population (old adults) is precisely the group with the worst DCA neuropathy profile, and slow-metabolizer GSTZ1 haplotypes accumulate the drug. The pulsed schedule (two weeks on, six to eight off) might mitigate accumulation, but that is untested for neuropathy. [Confidence: High that this is a real concern.]

7. You attribute selectivity to ATP collapse in cells with poor spare respiratory capacity. Did you rule out that healthy human tissues with high energy demand (heart, brain, kidney) are also vulnerable? Answer: partially. Healthy neurons, hepatocytes, and myoblasts were largely spared in vitro, and mice showed no acute frailty. But cardiac, renal, and central nervous system tissue were not functionally assessed in vivo, and high-demand tissues are a legitimate theoretical concern. [Confidence: Medium.]

8. Your senescence and SASP conclusions rest partly on n equals 3 antibody arrays. How robust is the “younger proteome” claim? Answer: weak as stated. An n of 3 serum proteomic comparison is hypothesis-generating, not confirmatory. The direction (lower IL-6, IL-1-alpha and related SASP factors) is consistent with senolysis, but the effect size and reproducibility are not established. [Confidence: Low-Medium.]

9. You did not test mTOR, autophagy, or cGAS-STING. Given the SASP reductions, how do you know the benefit is metabolic ATP collapse and not, say, autophagy modulation by metformin? Answer: they do not fully separate these. The mechanistic case rests on Seahorse ATP and respiration data showing loss of spare capacity in target cells. Metformin engages AMPK and autophagy independently, and the cGAS-STING link to the observed SASP drop was not tested. Mechanism is plausible but not exclusive. [Confidence: Medium.]