Cold Exposure & Metabolic Health

I.1The acute mechanism: insulin-independent glucose disposal

Cold exposure activates the sympathetic nervous system, and norepinephrine release drives two tissues to pull glucose out of circulation without needing insulin signaling at all. In skeletal muscle, shivering contraction activates AMPK, which moves GLUT4 glucose transporters to the cell membrane through the same pathway exercise uses — a route that bypasses the insulin/PI3K-Akt signaling that insulin resistance disrupts. In a 10-day human cold acclimation trial, cold-induced glucose uptake in brown fat was paralleled by increased basal GLUT4 localization at the muscle cell surface, confirmed by biopsy.¹

Stronghuman biopsy data, direct mechanism.

Brown adipose tissue (BAT) runs the same play through a different mechanism. Norepinephrine hits β3-adrenergic receptors on brown adipocytes, activating PKA, which drives both lipolysis and glucose uptake to fuel UCP1-mediated heat production. Cold exposure via adrenergic stimulation activates BAT, which combusts substantial amounts of blood glucose and free fatty acid to produce heat.² Active BAT is, functionally, a second glucose sink working alongside muscle — disposing of glucose the way working muscle does during exercise, on a schedule set by ambient temperature rather than movement.

I.2The adaptive mechanism: batokine signaling to other tissues

This is the part that explains why insulin sensitivity improves systemically rather than only in the tissues doing the shivering. BAT secretes several signaling molecules — batokines — that act on distant tissues.

Brown fat produces a lipid mediator called Maresin 2 when cold-stimulated, and this molecule resolves the chronic low-grade inflammation that drives insulin resistance in obese tissue, improving insulin sensitivity and glucose tolerance in animal models.³ Chronic inflammation is one of the upstream drivers of insulin resistance, so this pathway removes a cause rather than compensating for it.

Cold exposure also releases the lipokine 12,13-diHOME from BAT, which increases fatty acid uptake and lipolysis in a way that improves systemic lipid handling — less circulating and ectopic fat means less lipid-driven interference with insulin signaling in liver and muscle.⁴ BAT-derived FGF21 and adiponectin act on the liver to suppress excess glucose production and improve hepatic insulin sensitivity through a related route.

Animal–in vitrofor the Maresin 2 and 12,13-diHOME mechanisms specifically; Moderatefor the systemic human outcome they're proposed to explain.

The systemic reach of this signaling shows up directly in the acclimation data: BAT activity was negatively related to age and body fat percentage, and cold-induced glucose uptake in BAT was positively related to glucose uptake in nearby visceral white adipose tissue — BAT activation appears to pull surrounding fat tissue toward better glucose handling, not just improve its own metabolism in isolation.¹

I.3What the human trials show

The clearest human evidence comes from a 10-day cold acclimation protocol (6 hours daily at 14–15°C, subjects sedentary in shorts, a T-shirt, socks, and shoes) in eight overweight male type 2 diabetic patients, which improved insulin sensitivity by roughly 43% as measured by glucose infusion rate.¹ A systematic review of human cold-exposure trials found the accumulated evidence points the same direction: cold exposure activates BAT, and BAT activation is consistently linked to improved insulin sensitivity and glucose tolerance across the reviewed studies.⁵

Strongfor the BAT-activation–insulin-sensitivity link across human trials; Moderatefor the specific 43% figure, which comes from a single protocol and cohort.

II.1Uncoupling mitochondria creates heat, not ATP

UCP1 (uncoupling protein 1) sits in the inner mitochondrial membrane and lets protons leak back across it outside the normal ATP-synthase route, converting the proton gradient directly into heat instead of ATP.⁶ It's expressed at high levels in brown and beige fat and only marginally in skeletal muscle. Its presence in white fat matters even at low levels: UCP1-deficient mice show increased fat synthesis (de novo lipogenesis) in white adipose tissue at standard housing temperature, indicating UCP1 normally restrains this process rather than only generating heat.⁷

Strongfor the proton-leak thermogenic mechanism; Animal–in vitrofor the lipogenesis-restraint finding specifically.

II.2BAT increases T3 conversion, which drives browning

BAT expresses high levels of type II deiodinase (D2), the enzyme that converts thyroxine (T4) into the active hormone triiodothyronine (T3) — a conversion, not new hormone synthesis.⁸ Sympathetic stimulation from cold exposure sharply increases D2 activity, creating what researchers describe as a localized state of thyrotoxicosis inside brown adipocytes, with T3 receptor saturation running far higher in activated BAT than anywhere else in the body.⁸

StrongD2-driven local thyroid hormone amplification is well established.

That locally-generated T3 is itself one of the drivers of browning: it induces UCP1 expression in white fat depots and increases the beige-fat fraction embedded throughout white adipose tissue. Cold exposure independently drives the same browning program through sympathetic signaling, and the two pathways converge on the same transcriptional targets (PGC1α, PPARγ, UCP1) rather than acting as separate mechanisms — D2-generated T3 and direct sympathetic signaling are two inputs into one browning program.

Moderate–to–Strongfor thyroid and cold converging on UCP1-driven browning.

II.3Mitochondrial renewal: biogenesis and selective clearance in parallel

Cold exposure drives two mitochondrial quality-control processes simultaneously in BAT. PGC1α-mediated biogenesis increases mitochondrial mass and replicates the existing pool of mitochondrial DNA to expand thermogenic capacity.⁹ Running alongside it, PINK1/Parkin-mediated mitophagy selectively tags and removes mitochondria damaged by the reactive oxygen species that thermogenesis itself generates — blocking this clearance pathway measurably impairs BAT's oxidative capacity.¹⁰ The two processes run coordinated rather than sequentially: cold-challenged tissue shows both markers elevated at the same time, with net mitochondrial mass still increasing despite the active turnover.⁹

Animal–in vitrohuman BAT biogenesis is inferred from PET-measured tissue mass changes, not directly measured at the mitochondrial level.

III.1Catecholamine response

Norepinephrine spikes 200–530% and dopamine 200–250% during cold exposure, sustained for two to three hours afterward — the mechanism behind the alertness and mood lift people report.¹¹ A 20-week trial of regular cold water swimming improved WHO-5 well-being scores from 39.2 to 54.0, one of the better-controlled outcomes in this literature.¹²

Strongfor the norepinephrine/dopamine surge and its duration; Moderatefor mood/well-being improvement with regular practice.

III.2Immune modulation, not simple “boosting”

Cold water immersion increases neutrophil counts (innate immunity) while suppressing lymphocyte and monocyte counts (adaptive immunity) — a shift in immune activity, not a uniform boost.¹³ A meta-analysis of over 3,000 participants found cold-water immersion significantly increased inflammation immediately and at one hour post-exposure, the opposite of the general “anti-inflammatory” framing the practice often gets.¹⁴

Contestedas commonly framed — modulation, not uniform enhancement; acute inflammation actually rises.

III.3Cortisol: an acute spike that partially resolves same-day

Cortisol follows a similar biphasic pattern to inflammation. Cold water immersion induces a marked acute release of cortisol alongside epinephrine and norepinephrine, with increases of roughly 30–100% above baseline in the first 15–30 minutes — the standard sympathetic/HPA-axis stress response.¹⁷ One controlled study of 16 healthy adults found serum cortisol unchanged 30 minutes after a single 15-minute, 10°C immersion, but down 47% by 180 minutes, alongside reduced negative affect at the same timepoint.¹⁸ The same meta-analysis cited above for inflammation tracked a separate “stress” outcome across timepoints and found a significant reduction at 12 hours post-exposure, with no effect immediately, at 1 hour, 24 hours, or 48 hours.¹⁴

Strongfor the acute cortisol spike; Moderatefor the same-day delayed reduction (3–12h) — consistent direction, limited studies.

The longest-duration cortisol data available is a 12-week trial: two groups of 10 healthy women did either winter swimming (0–2°C, 20 seconds) or whole-body cryotherapy (−110°C, 2 minutes), three times weekly, with blood drawn at weeks 1, 2, 4, 8, and 12. By weeks 4–12, plasma ACTH and cortisol measured 35 minutes after exposure were significantly lower than the same timepoint at week 1 — habituation of the response itself, not just a same-day pattern.³⁶ That's a real longitudinal finding, but it's not the same claim as “chronic practice lowers resting cortisol”: it shows the acute response to a given exposure blunting over 12 weeks, always measured around a cold-exposure session, not cortisol on an ordinary day independent of practicing at all.

Moderatefor habituation of the acute post-exposure cortisol response over 12 weeks; Inferredstill, for true resting-day baseline cortisol shifting independent of a given session — that specific claim remains untested.

Separately, “adrenal fatigue” — the popular framing that a demanding cold-exposure practice could exhaust or burn out the adrenal glands — is not a recognized medical diagnosis. A systematic review found no consistent evidence that self-reported fatigue symptoms attributed to “adrenal fatigue” correspond to abnormal adrenal hormone markers; the condition endocrinologists do diagnose from abnormal cortisol, primary adrenal insufficiency, is a distinct, rarer condition confirmed by blood testing, not a burnout syndrome from stressors like cold exposure.³⁷

Strong“adrenal fatigue” specifically is not supported as a diagnosis; this doesn't by itself establish that cold exposure is safe for anyone with a real HPA-axis or adrenal condition, which is a separate question a clinician should evaluate.

III.4Neurotrophic signaling: the least settled piece

BDNF and related neurotrophic factors respond to cold stress in animal models — mouse studies show cold-modulated BDNF and mitochondrial markers in the hippocampus, with the response depending on exposure intensity and duration.¹⁵ Human data on cold-specific BDNF elevation is thin; the strongest human neurotrophic-factor evidence in the thermal-stress literature actually comes from heat and hyperthermia protocols, not cold.¹⁶

Animal–in vitrono direct human evidence for cold-specific BDNF elevation.

III.5The pre-workout cold-plunge testosterone claim

A claim that circulates widely in cold-exposure media, popularized separately by Andrew Huberman and by Thomas Seager — an Arizona State sustainability-engineering professor and co-founder/CEO of the ice-bath company Morozko Forge: that cold plunging immediately before resistance training, rather than after (when cold water immersion is more commonly discussed as recovery), produces a larger testosterone response than either alone. Seager's version rests on his own self-tracked biomarkers and reports from his community rather than a published trial.¹⁹ Huberman's separate, narrower claim — that cold exposure to the testes specifically may raise testosterone — is presented on his podcast as biologically plausible rather than demonstrated by a cited trial.²⁰

Speculativeno controlled trial has tested this specific protocol — cold plunging immediately before a single training session — against testosterone response.

The mechanistic link usually missing from this claim is real, though. Testosterone synthesis is a mitochondrial process before it's anything else: adrenal, gonadal, and placental mitochondria house CYP11A1 (P450scc), the enzyme that converts cholesterol to pregnenolone — the rate-limiting step of the entire pathway, gated by StAR-mediated cholesterol transport into the mitochondrion.³⁸ Damage that step and output falls regardless of what happens downstream. A 2022 study found that damaging Leydig-cell mitochondria — via Atp5a1 knockdown in cultured cells, or chronic stress in rats — cut StAR, CYP11A1, and 17β-HSD expression and triggered apoptotic signaling.³² Separately, a single 30-minute local testicular heat treatment at 43°C in rats reduced serum and testicular testosterone and downregulated StAR and CYP17 in Leydig cells.³³ Heat, specifically, is the documented stressor in this literature: testes normally run 2–6°C below core body temperature, and chronic overheating is the established way this pathway gets damaged.

What isn't established is the reverse — that deliberately cooling the testes, or the body generally, upregulates this machinery above its undamaged baseline. The chronic-stress rat study above used repeated brief ice-water swimming as one of seven rotating stressors and found it tracked with the same mitochondrial damage as the other six; cold imposed as an unpredictable, chronic stressor was associated with worse Leydig mitochondrial function in the one study here that included cold directly, not better.³² A separate 2022 study of 21 military trainees found testosterone decreased after a 9-hour, 4°C dive — extreme, prolonged cold exposure, in the one direct human data point available, moved testosterone the wrong direction for this claim.³⁹ The honest version of Huberman's claim above is narrower than usually stated: avoiding chronic testicular overheating protects a real, mitochondria-dependent pathway; actively cooling beyond that to raise testosterone above baseline is a plausible but untested extension of the same mechanism, and the nearest real dose data points the other way at the extreme end.

Strongfor mitochondrial cholesterol-to-pregnenolone conversion as the rate-limiting step in testosterone synthesis; Moderatefor heat damaging that step in rats; Speculativethat deliberate cold exposure protects or enhances it — both real cold-specific data points (chronic ice-water stress in rats, a 9-hour cold dive in humans) point the other way.

A structural problem also runs through most of the human “evidence” cited for this claim: it's self-selected case reports collected by advocates, not a trial. One widely circulated collection presents named individuals with lab-confirmed testosterone rises after starting cold plunging, verified by personal interview — a real step up from pure anecdote, but several of the strongest-looking rises were measured in patients concurrently on exogenous testosterone replacement therapy, where total-testosterone changes track injected dose at least as much as any independent effect. The same source's read of its own counter-cases (testosterone drops after adopting the practice) attributes them to confounding medications rather than treating them as evidence the practice doesn't work — an asymmetric interpretation applied in one direction only. Readers with a positive result are also more likely to write in to begin with. None of this means cold exposure has no effect; it means self-collected testimonial series can't distinguish that effect from selection bias, concurrent TRT dosing, and motivated interpretation, which is exactly what a trial is for.

Two genuinely different mechanisms get bundled together in popular cold-exposure discourse as “cold fights cancer”: brown fat competing with tumors for glucose, and cold-shock proteins that repair DNA more accurately. Both have real, published mechanistic support — one with a strong causal chain in mice, the other with cross-species cell-culture validation. Neither has been shown to work as a human cancer treatment outside a single pilot case.

IV.1Brown fat and tumors compete for the same fuel

Tumors are unusually dependent on glycolysis for fuel even when oxygen is available — the Warburg effect — which makes them vulnerable to anything that reliably out-competes them for circulating glucose. A 2022 study found that mice housed at 4°C, rather than a thermoneutral 30°C, showed markedly slower growth of colorectal, breast, and pancreatic tumors and lived nearly twice as long.²³ The mechanism traces directly back to Part I's insulin-sensitivity story run against a tumor instead of for one: cold-activated brown fat combusts blood glucose fast enough that the tumor is left glucose-starved.

What makes this an unusually clean causal chain for this literature is that the authors tested it two ways. Removing brown fat, or knocking out UCP1 specifically, abolished the protective effect — tumors grew at the same pace as in warm-housed mice. Giving the cold-housed mice a high-sugar drink also abolished it, by refilling the glucose supply the tumor was being starved of. Mechanism, knockout, and rescue-reversal all point the same direction.

Strongfor the mouse mechanism — multiple cancer types, causally confirmed by both knockout and rescue.

IV.2The human pilot case, and what it does not show

The same paper included a single human case: an 18-year-old with Hodgkin lymphoma, already five cycles into ABVD chemotherapy, with detectable disease remaining. Exposed to 22°C for 7 days and then 28°C for 4 days in light clothing, PET–CT showed the expected brown-fat activation at the cooler temperature, and lower tumor glucose (18F-FDG) uptake during the cold phase than the warm phase.²³

That is a metabolic PET signal in one patient, not a measured reduction in tumor size — the paper does not report the tumor shrinking, only its glucose uptake changing on imaging. There is no comparison group and no replication. It is a human correlate of the mouse mechanism, not a demonstrated clinical benefit.

Animal–in vitrofor the mechanism, established in mice; Inferredfor any human therapeutic benefit, which rests on a single uncontrolled case with a PET surrogate endpoint.

IV.3Cold-shock proteins, DNA repair, and Peto's paradox

Bowhead whales live up to roughly 200 years and, with far more cells dividing over far longer lifespans than humans, should by simple probability accumulate far more cancer-causing mutations — yet they don't get much cancer. That mismatch between body size, lifespan, and observed cancer rates is known as Peto's paradox. A 2025 study from the University of Rochester found that bowhead whale cells carry roughly 100-fold higher levels of a cold-shock protein called CIRBP than other mammalian cells, and that CIRBP drives more accurate repair of double-strand DNA breaks — the class of damage most likely to drive cancer when repaired badly.²⁴ Adding bowhead CIRBP to human and fruit-fly cells in culture improved DNA repair in both, and extended lifespan in the flies.

Cold is the connective tissue in this mechanism, not incidental: CIRBP is a cold-shock protein by definition, and mammalian cells generally increase CIRBP output when their temperature drops a few degrees in culture. That's the closest thing to a “cold exposure improves DNA repair” pathway in mammalian cells — but it was established as a cell-culture temperature response, not as an outcome of a deliberate human cold-exposure practice. Whether a cold plunge or cold room lowers cellular temperature enough, for long enough, to meaningfully raise CIRBP output in a living, thermoregulating human — and whether that would measurably change cancer risk — has not been tested.

Strongfor CIRBP's role in bowhead DNA repair and its cross-species validation in human cell culture; Speculativefor the leap to a deliberate human cold-exposure practice lowering cancer risk.

IV.4An unresolved tension in the mouse data

Standard laboratory housing temperature for mice (20–26°C) is mild chronic cold stress relative to their thermoneutral zone (30–31°C). A 2013 study found that moving tumor-bearing mice to the warmer, thermoneutral 30–31°C markedly improved tumor control — smaller tumors, less metastasis — through more CD8+ T-cell infiltration and fewer immunosuppressive MDSCs and regulatory T cells.²⁵ On its face, that runs the opposite direction from IV.1: warming mice up helped there, while cooling mice down (to a genuinely cold 4°C, not merely sub-thermoneutral) helped here.

This isn't necessarily a contradiction so much as two different mechanisms engaged by two different protocols. The 2013 study tested a mild, chronic housing-temperature difference across the 20–31°C range, where the cooler condition looks like it produces chronic stress and immunosuppression without recruiting much thermogenic brown fat. The 2022 study tested a much colder, deliberately thermogenic protocol (4°C) engineered to drive substantial BAT activation. Neither study tested the other's temperature range, and where on that spectrum immune suppression gives way to glucose competition as the dominant effect is not established.

Contestedboth findings are real and peer-reviewed; which mechanism dominates in a given cold-exposure regimen is not established.

IV.5The ketosis question, and a contested theory underneath it

Some cold-exposure advocacy traces the tumor-suppression mechanism back further, to Thomas Seyfried's argument that cancer is fundamentally a mitochondrial and metabolic disease rather than a genetic one — that nuclear DNA damage is often downstream of mitochondrial dysfunction, not its cause. That's a genuine, published position, but it's a minority one. Most cancer biology holds that nuclear mutations are the primary driver for most cancers — well-characterized oncogenes and tumor suppressors like TP53 and BRCA1/2, and viral oncogenes that directly inactivate tumor-suppressor proteins, are hard to explain as merely downstream of a metabolic problem. This paper doesn't take a side in that broader debate; IV.1–IV.2 hold regardless of which theory of carcinogenesis is correct, since the glucose-competition mechanism they describe doesn't depend on nuclear mutations being secondary.

Underneath that theoretical debate is a narrower, testable claim: that ketogenic diets and ketone supplementation slow tumor growth. That literature is real. A 2014 study found that supplementing β-hydroxybutyrate decreased viability of metastatic cancer cells in culture and prolonged survival in mice with metastatic disease.³⁴ A 2017 realist review covering 29 animal and 24 human studies found an anti-tumor effect in 72% of the animal studies — but its own conclusion states that human evidence is “weak and limited to individual cases.”³⁵

Moderate–to–Strongfor ketone supplementation slowing tumor growth in mice; Inferredfor the same effect in humans — the review's own summary calls this weak and case-level.

What isn't established is the link some retellings add on top: that cold exposure itself reliably induces ketosis fast enough and deep enough to matter for tumor metabolism. Seki et al. explicitly did not measure ketone levels in their mouse or human data²³ — the ketone mechanism is not something that study showed. The claim that cold exposure produces rapid ketosis traces to self-reported, unpublished continuous-monitoring observations from an ice-bath industry source, not a peer-reviewed measurement.¹⁹ It's a plausible extension — two real literatures (cold-induced BAT glucose competition, and ketone-induced tumor suppression) stacked on top of each other — but no cited study tests the combined pathway directly.

Speculativethat cold exposure induces cancer-relevant ketosis; this stacks two separately-supported findings without a study testing the combination.

Every mechanism in Parts I–IV runs through the same starting trigger: cold-sensing nerve endings in skin signaling the sympathetic nervous system. What varies across the common “how to do it” methods — a cold windy day, a face dunk, an ocean or lake plunge, an ice bath, a cold shower — is mostly how efficiently each one delivers that trigger, not a different mechanism each time. There is one real exception, covered in V.2.

V.1Intensity and surface area matter more than the label on the method

Water pulls heat from the body far faster than air at the same temperature, which is why immersion protocols dominate the acute catecholamine and BAT literature cited throughout this paper rather than air exposure at a similar number on the thermometer. In one whole-body 10°C water immersion study, plasma norepinephrine rose from a baseline of roughly 359 pg/ml to about 642 pg/ml within 2 minutes, and continued climbing to a peak of roughly 1,171 pg/ml by 45 minutes — the response is not simply a first-30-seconds jolt that plateaus.³⁰ There is also a real intensity threshold: cold-pressor protocols generally need water colder than roughly 7°C to reliably trigger the sympathetic response at all — a mild chill under that threshold often doesn't do it.³¹

Moderateintensity/threshold effects are real and documented; whether onset speed specifically, versus final temperature or total duration, is the dominant variable has not been isolated in a dedicated dose-response trial.

V.2The face is a special case, not a smaller version of a full plunge

Forehead cooling produces a faster and larger sympathetic nerve response, per unit of skin surface, than cooling the hand — the face is disproportionately effective for its size.²⁶ But face immersion also engages the mammalian dive reflex through the trigeminal nerve, a distinct, predominantly parasympathetic (vagal) pathway that produces bradycardia.²⁹ A 2024 study found that during progressive face cooling, sympathetic activation switches on at a comparatively mild forehead skin temperature (around 19°C), while the vagal, heart-rate-lowering response engages later, at a colder threshold (around 11°C).²⁷

A face dunk is therefore not simply a scaled-down plunge. It's an unusually efficient sympathetic trigger for the skin area involved, and if it's cold enough to also cross the lower vagal threshold, it layers a genuine calming, heart-rate-lowering reflex on top — something whole-body immersion doesn't produce the same way. It's a distinct tool, useful in its own right, not a lesser substitute for the whole-body mechanisms in Part I if the goal is glucose disposal or BAT recruitment specifically.

Moderate–to–Strongthe differential sympathetic/parasympathetic thresholds during face cooling are specific and consistently reported across independent studies.

V.3Modality, compared

A cold windy day with skin exposed works through the same pathway as immersion, but air's slower heat transfer means it generally needs to be colder, windier (convective heat loss adds up), and sustained longer to match what a plunge does in a few minutes at the same nominal temperature — not a lesser method, a slower one. An ocean or lake plunge and an ice bath both deliver the fastest, largest-surface-area heat pull of the common methods, which is why they dominate the acute-effect literature. A face dunk is the distinct case from V.2: a real, efficient tool, but not a substitute for whole-body immersion if the target is the metabolic mechanisms in Part I. A cold shower is typically the mildest method in practice — water contacts only part of the body at a time, and most household cold-water taps and most people's tolerance don't reach the intensity or duration described in V.1.