Working Paper · Nutritional Biochemistry
Sunlight is not one exposure but eight, stacked in a single beam — infrared, red, green, blue, and three bands of ultraviolet, each driving a different pathway on a different tissue. The UVB that raises skin cancer risk through direct DNA damage is the same UVB that drives vitamin D synthesis and immune modulation linked to lower rates of over a dozen other cancers. Untangling which claim rests on which band, and on what kind of evidence, is the point of this paper.
“Sun exposure” is not a single variable. Ultraviolet B (280–315nm) drives both the strongest cancer-causing pathway in this paper (direct DNA mutation in skin cells) and the strongest cancer-protective one (vitamin D and immune modulation) — through two different molecular routes in the same tissue. Ultraviolet A (315–400nm) and visible light drive a separate, better-tolerated pathway: cutaneous nitric-oxide release that lowers blood pressure. Visible and infrared bands drive circadian entrainment, mitochondrial and keratinocyte signaling largely independent of the UV cancer question entirely. This paper works through the evidence band by band, then through three specific human studies — a melanoma-mortality cohort, a 29,000-person Swedish sun-exposure cohort, and an ecological review of UVB and 24 cancer types — each of which is weaker, and tells a narrower story, than its popular retelling.
This is the companion paper to Cold Exposure & Metabolic Health: both cover a non-dietary environmental input the body evolved under and that modern indoor, sun-avoidant life has mostly removed.
Six parts
UVB is the load-bearing, two-edged band. Every other part of this paper either sits upstream of it (Part I's map), tests its net effect in a real population (Parts II–IV), or covers a mechanism that doesn't depend on it at all (Parts V–VI).
Part I is a reference map, not an argument — use it to see which band a later claim is actually about. Parts II–IV each cover one human study in depth, and each carries a real methodological limitation that changes what it can and cannot support; read the caveat attached to each as part of the finding, not an afterthought. Parts V–VI cover mechanisms that don't carry cancer risk the way UV does. Caveats are collected in Limitations rather than scattered through the text.
Strong Moderate Animal–in vitro Inferred Contested Speculative
Popular writing about “sunlight” tends to treat it as one exposure with one dose-response curve. It isn't. Sunlight reaching skin and eyes spans roughly 280 to over 1000 nanometers, and different bands within that range are absorbed by different chromophores, in different tissues, triggering different signaling cascades. Infrared warms mitochondria. Blue light resets the circadian clock through the eyes. UVB, absorbed almost entirely in the outer few hundred microns of skin, does two nearly opposite things in the same cells depending on which molecule it hits first — DNA directly, or 7-dehydrocholesterol just beneath it. The table below is a reference map compiled from a published heliotherapy reference chart13 cross-checked against the mechanism literature cited through the rest of this paper; treat band boundaries as approximate, since UV and visible light are a continuum, not eight hard-edged categories.
Moderatethe physiological pathways (nitric oxide release, vitamin D synthesis, melanopsin/circadian signaling, POMC) are individually well characterized; Inferredfor many of the specific downstream disease outcomes listed, which come from correlational or ecological data discussed in Parts II–III, not a trial testing that exact band against that exact outcome.
Reading straight down the UVB row, the chart lists both a cancer-causing pathway (DNA mutation) and a cancer-protective one (vitamin D, immune modulation) under the same 280–315nm band. That is not a contradiction in the chart — it's an accurate summary of a real, unresolved tension in the literature, and the next three Parts are an attempt to work through it using three specific studies rather than the summary chart alone.
William Grant's 2012 review in Anticancer Research compiled ecological studies — comparisons of cancer incidence and mortality rates across geographic regions with different levels of ambient solar UVB — drawing on three US studies, one each from Australia, China, France, Japan, and Spain, and eight multicountry studies spanning more than 100 countries.1 The review found consistent inverse correlations between solar UVB and 15 cancer types: bladder, breast, cervical, colon, endometrial, esophageal, gastric, lung, ovarian, pancreatic, rectal, renal, and vulvar cancer, plus Hodgkin's and non-Hodgkin's lymphoma. Nine more — brain, gallbladder, laryngeal, oral/pharyngeal, prostate, and thyroid cancer, leukemia, multiple myeloma, and, notably, melanoma itself — showed weaker evidence.
Moderatethe correlational pattern across studies is real and consistently reported; Inferredthat UVB/vitamin D is the causal driver, rather than something else that also tracks with latitude — see the pushback below.
The mechanistic case behind the correlation is genuine cell biology, not just a geographic pattern. UVB converts 7-dehydrocholesterol in skin to previtamin D3, which becomes calcitriol (1,25-dihydroxyvitamin D), the active hormone. Calcitriol binds the vitamin D receptor, expressed in most tissue types including immune cells, and regulates genes involved in cell differentiation, apoptosis, and inflammation — the same processes disrupted in cancer's uncontrolled proliferation.2 Local, tissue-level vitamin D metabolism, independent of circulating blood levels, is also implicated: several of the tissues in the protected-15 list (colon, breast) express the enzyme that activates vitamin D locally, giving them a potential paracrine protective signal that a blood test alone wouldn't capture.
Strongfor calcitriol's role in cell differentiation and apoptosis generally; Inferredthat this specific pathway explains the geographic cancer pattern in II.1.
An ecological study correlates a population-level exposure (regional UVB) with a population-level outcome (regional cancer rate) — it does not measure either variable in the same individual. Everything else that varies by latitude and country varies right along with UVB: diet, obesity prevalence, smoking rates, alcohol consumption, cancer-screening infrastructure, and general healthcare access all differ systematically between, say, Scandinavia and equatorial countries, and any of them could independently move 15 different cancer rates in the same direction as sunlight without sunlight causing any of it. Several of the ecological studies underlying this review are Grant's own earlier work, and he directs the Sunlight, Nutrition and Health Research Center (SUNARC), a research group organized specifically around this hypothesis — not disqualifying, but worth weighing alongside independent replication. The vitamin D mechanism in II.2 is real cell biology; whether it's the dominant explanation for the geographic pattern in II.1, versus one contributor among several confounded variables, is not something an ecological design can settle.
Randomized vitamin D supplementation trials are the design that can isolate the vitamin D variable from everything else that moves with latitude, since supplementation can be randomized independent of where someone lives. That literature exists and is genuinely mixed: several large trials have found reduced cancer mortality without a matching reduction in cancer incidence — suggesting, if real, a role in slowing progression in people who already have cancer rather than preventing it from starting, a narrower and different claim than II.1's incidence-correlation implies.3
Contestedsupplementation trials are the right design to isolate this question, and the incidence-versus-mortality split across them is a real, unresolved pattern in this literature — not yet a settled answer.
Melanoma was the one cancer in II.1's list with only weak inverse correlation to solar UVB — an odd result if UVB/vitamin D is simply protective across the board. A 2009 hypothesis paper offers a specific explanation. Cutaneous melanoma has risen at a steady exponential rate in fair-skinned indoor workers since before 1940, despite indoor workers getting three to nine times less total solar UV than outdoor workers — a direct paradox if total sun exposure were the driver.4 The paper's two-part hypothesis: indoor UVA — the band that passes through ordinary window glass, unlike UVB, which glass blocks almost entirely — both causes mutations directly and degrades any vitamin D3 already synthesized in skin; separately, melanoma cells themselves can convert vitamin D3 to calcitriol, which drives growth inhibition and apoptosis in melanoma cells in culture and in animal models, so inadequate cutaneous vitamin D removes a brake on cells that have already begun to transform. The authors measured indoor solar UVA at roughly 25% of outdoor irradiance (about 60× a typical fluorescent fixture) and confirmed previtamin D3 synthesis essentially requires being outdoors, since window glass blocks the UVB it depends on. Framed this way, the hypothesis is a promotion mechanism layered on top of the established initiation mechanism from Part I — the authors explicitly still attribute melanoma's initiation to intense, intermittent outdoor UV overexposure and sunburn, not to indoor UVA instead of it.
Speculativea named hypothesis built from the authors' own irradiance measurements plus cited melanoma cell-biology data, not a study measuring melanoma incidence against individually measured UVA exposure or vitamin D status.
Medical Hypotheses, where this paper appeared, was explicitly a hypothesis-generating journal at the time of publication (April 2009): submissions were screened editorially for interest and plausibility rather than sent out for standard external peer review — a policy Elsevier discontinued starting in 2010, following a separate, unrelated controversy over the journal's publication of AIDS-denialist material the same year. That history doesn't mean this specific paper is wrong; the underlying irradiance measurements and the calcitriol/melanoma-cell mechanism it cites are independently verifiable. But it means this piece carries less institutional vetting than the ecological review in II.1 or the cohort studies in Parts III–IV, and it should be read as a plausible, testable explanation for melanoma's exception — not as settled evidence of one.
Unrelated to the UVB/vitamin D mechanism above, a small open-label pilot study tested treating already-diagnosed squamous cell carcinoma in situ (SCCIS) — the same UVB-driven, DNA-mutation-caused cancer flagged in Part I's UVB row — with a topical antioxidant rather than surgery. Seventeen patients with 27 biopsy-confirmed SCCIS lesions applied a 30% ascorbic acid (vitamin C) solution in 95% dimethyl sulfoxide (DMSO, used as a skin-penetration enhancer) twice daily for 12 weeks. On repeat biopsy, 56% of lesions (15 of 27) showed complete histologic resolution, another 44% shrank by more than 85% in size, mean lesion area fell 71%, and 94% of patients (16 of 17) avoided surgical excision entirely; no patients discontinued for adverse effects.5
Speculativea small, open-label pilot with no control or placebo arm and no blinding — encouraging biopsy-confirmed numbers, but nothing here rules out natural lesion regression or assessment bias, and the study is currently a preprint, not a peer-reviewed publication.
This is a medRxiv preprint (posted October 2025) and carries medRxiv's own standard notice that it has not been peer-reviewed and “should not be used to guide clinical practice.” The authors (Institute for Biomedical Sciences, Boise, Idaho) disclose no funding and no competing interests, and the trial is registered (ClinicalTrials.gov NCT04279535). Independent of the study's own methodology, it reached this paper via a viral X/Twitter thread from an epidemiologist affiliated with the McCullough Foundation — a source with its own track record of amplifying contrarian medical claims — which has no bearing on the preprint's design but is worth knowing given how it was surfaced.
The Melanoma in Southern Sweden (MISS) cohort enrolled 29,518 Swedish women between 1990 and 1992 and followed them for 20 years through national mortality registers, grouped by self-reported sun-exposure habits.6 Active sun-seekers had lower cardiovascular and other non-cancer mortality than avoiders, despite higher skin-cancer incidence — the all-cause mortality gap was driven mainly by cardiovascular and other non-cancer deaths, not by fewer skin-cancer deaths among sun-seekers. Modeled out, sun avoidance was associated with an estimated 0.6 to 2.1 fewer years of life expectancy compared to high sun exposure — the basis of the widely repeated comparison to smoking as a mortality risk factor of similar modeled magnitude.
Moderatelarge, long-follow-up cohort with registry-linked mortality — the strongest human data in this paper by sample size and duration; Contestedas an argument that sunlight itself, rather than the lifestyle correlated with it, drives the effect.
The comparison to smoking is a modeled life-expectancy gap, not a claim that sun exposure offsets or cancels smoking's harm — a distinction the popular retelling of this study frequently drops. Two variables producing similarly sized effects on the same outcome measure does not mean either variable determines the other.
Sun-exposure habits were self-reported, not randomized, and sun-seeking behavior is entangled with a cluster of other health-relevant habits: more time outdoors generally means more walking, more incidental exercise, more social contact, and plausibly better baseline health that both causes someone to spend more time outside and independently protects cardiovascular health. The study could not separate UV dose from this broader lifestyle pattern, and no exercise data was collected at all. The mortality gap is real and well measured; whether sunlight itself is the active ingredient, versus a marker for an generally more active and socially engaged life, is exactly what an observational design like this one cannot resolve.
WHO and Swedish national sun-protection guidance — cover up and seek shade above UV Index 3 — is unchanged by this study. The finding is a population-level mortality pattern over two decades, not a demonstrated individual protocol for how much unprotected sun exposure is beneficial versus harmful for a given skin type.
Berwick and colleagues followed 528 population-based cutaneous melanoma cases for an average of more than five years, combining interview data (sun exposure history, skin self-awareness, screening behavior) with histopathology review of each lesion (solar elastosis, Breslow thickness, mitotic rate).7 In multivariable competing-risk models, two factors were independently associated with lower melanoma death after adjusting for tumor thickness, mitotic index, and head/neck location: solar elastosis, a histologic marker of chronic sun damage (HR 0.4, 95% CI 0.2–0.8, p=.009), and self-reported skin awareness (HR 0.5, 95% CI 0.3–0.9, p=.022). Tumor thickness, mitoses, ulceration, and head/neck location all independently predicted higher death risk, as expected.
Strongfor the statistical associations themselves — a well-powered cohort with histopathology-confirmed covariates and appropriate competing-risk modeling.
The authors' own reading of their result is a detection-effect story, and it fits the data better than a protective-sunlight story does. Solar elastosis is a marker that someone has accumulated substantial visible sun damage over their lifetime — exactly the kind of skin that makes a new or changing mole more noticeable to the person carrying it or to a clinician examining it. Skin awareness is a direct behavioral measure of the same thing: people who pay closer attention to their skin catch melanomas earlier, when they are thinner and more survivable, which is precisely what the thickness and mitosis covariates in the same model are already controlling for. The paper was explicitly designed to test this: whether the previously observed positive correlation between UV exposure/solar elastosis and melanoma survival reflects sunlight's biology or simply earlier detection — and the pattern (awareness and elastosis predicting survival independent of tumor severity at diagnosis) is consistent with detection, not consistent with sunlight altering the tumor's underlying biology once it has formed.
Strongas evidence for a detection/early-catch effect; Speculativeas evidence that sun exposure changes melanoma's course biologically once it exists — the study wasn't designed to distinguish that from detection timing, and its authors don't claim it does.
Part I's UVA row lists melanoma risk rising with cumulative UV exposure — that remains the operative causal claim. This study does not contradict it. It answers a narrower, different question: among people who already have melanoma, why do some survive longer — and the answer here is earlier detection, not a protective effect of the sun exposure that likely contributed to causing the melanoma in the first place.
Unrelated to the detection-effect story above, a dietary intervention has shown a direct antitumor effect on melanoma specifically. Mice fed a 5% glycine diet for just three days before subcutaneous B16 melanoma implantation grew tumors 50–75% smaller than control-fed mice, weighing nearly 65% less at 14 days — not through slowing cancer-cell proliferation directly (glycine had no effect on B16 cell growth rates in culture) but through a 70% reduction in tumor blood-vessel density, alongside dose-dependent inhibition of endothelial cell growth in vitro.8 See the Glycine nutrient page for the parallel liver-tumor finding and glycine's broader mechanism profile.
Animal–in vitroa mouse implantation model, not a human melanoma trial — an antiangiogenic diet-and-tumor-growth finding, mechanistically unconnected to the sun-exposure/detection question in IV.1–IV.2.
Human skin stores a substantial pool of nitric oxide in inactive forms — nitrite and nitrosated compounds — independent of the enzymatic nitric oxide synthase pathway used elsewhere in the body. UVA wavelengths, which penetrate deeper into skin than UVB and are absorbed by different chromophores, photochemically release this stored nitric oxide into circulation. Circulating nitric oxide is a potent vasodilator, and human trials of UVA skin exposure have shown measurable blood pressure reduction accompanying the release, independent of vitamin D or any DNA-mediated pathway.9
Moderatethe cutaneous nitric-oxide store and its UVA-triggered release is documented in human skin; Inferredfor how much this specific mechanism contributes to the Part III cohort's cardiovascular mortality gap versus the lifestyle confounders flagged there.
Separately, UV exposure in skin upregulates proopiomelanocortin (POMC), a precursor protein cleaved into several active peptides including α-melanocyte-stimulating hormone (α-MSH, which drives melanin production and tanning) and beta-endorphin. This is the leading proposed mechanism behind the mood-lifting and mildly reinforcing quality some people report from sun exposure, and behind why moderate, repeated UV exposure produces habituation rather than aversion in some studies — a real, biologically plausible route distinct from vitamin D or nitric oxide, though it carries its own caveat.10
Animal–in vitrothe POMC/beta-endorphin pathway is best characterized in mouse skin; direct human behavioral confirmation is thinner.
A beta-endorphin-mediated reward pathway is also the leading proposed mechanism behind “tanning dependence” described in a small addiction-medicine literature — the same signal that makes moderate sun exposure feel good is mechanistically capable of reinforcing excessive exposure in a minority of people. This doesn't undercut the mechanism; it's a reason dosing matters here as much as it does anywhere else in this paper.
Every mechanism in this paper runs on dose and timing, not just presence or absence of light. A published heliotherapy guide organizes the practical benefits of getting this right into six areas: lower blood pressure and better cardiovascular health, improved mood and reduced stress, healthier immune function and reduced chronic inflammation, increased metabolism and mitochondrial function, better sleep, and enhanced energy, alertness, and cognition.13 Its own framing is worth keeping: these aren't six independent wins so much as one nest of interlocking systems, where bright, well-timed days and genuinely dark nights raise circadian amplitude and that amplitude improvement is what cascades into the other five.
Indoor lighting rarely exceeds a few hundred lux; overcast outdoor daylight is typically 1,000–10,000 lux, and direct sun exceeds 50,000. The blue-rich portion of that daylight signal drives melanopsin-containing retinal ganglion cells that set the circadian clock in the hypothalamus — the same pathway summarized in Part I's blue-light row.11 A weak daytime light signal is one of the more overlooked contributors to poor sleep specifically because the body generates its night-time melatonin signal in proportion to how strong a contrast it registered during the day — someone who is dim-lit all day and then also screen-lit all evening has removed the contrast entirely, not just moved it later.
The same blue wavelengths that entrain the clock by day suppress melatonin and delay sleep onset at night, because melanopsin cells don't distinguish a screen from the sun — only intensity and spectrum.12 The practical implication is spectral matching, not blue-light avoidance outright: bright, cool-toned light through the day; dimmer, warmer-toned light in the few hours before sleep, mirroring the natural shift toward red and amber wavelengths as the sun approaches the horizon that Part I's infrared/red row describes.
There's a mechanistic reason getting outside early matters beyond circadian timing. A 2024 peer-reviewed study pretreated hairless mice with 630nm red LED light (30 minutes, twice daily, for five days) prior to three days of UVB exposure (200 mJ/cm²/day) and compared them to mice given UVB alone.14 On histology, LED-pretreated skin kept visibly more organized collagen and elastin and less epidermal thickening and inflammation than UVB alone, closer to untreated control skin. On Western blot, Claudin-1 — a skin-barrier tight-junction protein UVB alone significantly reduced (p<0.001 vs. control) — trended higher with LED pretreatment, though that specific comparison against UVB alone didn't reach statistical significance.
Animal–in vitron=5 mice per group; the histology comparisons are descriptive, and the one quantified barrier-protein result (Claudin-1) didn't reach significance against the UVB-only group specifically.
The study's own abstract describes LED pretreatment as producing “a reduction in… Nrf-2 and HO-1 expression” — the skin's two main antioxidant-response markers — which is the claim a viral social media thread repeated as “lower oxidative stress markers.” The paper's own Results section says the opposite: Nrf-2 and HO-1 were both numerically higher, not lower, in the UVB+LED group than in the UVB-only group (not a significant difference between those two groups specifically, though both were significantly elevated versus control and versus LED-alone). Read at face value, the data describes LED pretreatment priming the skin's own antioxidant defense pathway harder, not needing it less — a more mechanistically coherent preconditioning story than the abstract's wording, or the social thread repeating it, conveys.
None of this suggests an LED device substitutes for actual morning sunlight. Infrared and red wavelengths make up a larger share of the solar spectrum at low sun angles, per Part I's infrared/red row — a zero-cost version of the same preconditioning signal this study tested with a device.
A 2021 randomized trial in Egypt asked whether adding abdominal red light to a diet moved fatty-liver markers more than the diet alone.15 Sixty obese adults (roughly 65–75, ultrasound-confirmed NAFLD, elevated ALT) all followed the same Mediterranean diet with about 15% calorie restriction and biweekly dietitian sessions. Half also wore abdominal straps of 660nm LEDs for 30 minutes, twice weekly, for 12 weeks (reported radiant exposure 127.8 J/cm² per session). Both groups improved, and the device group improved more. Waist circumference fell about 15 cm (~6 in) in both men and women versus about 6.6 cm (~2.6 in) on diet alone. ALT fell from 51.6 to 32.7 U/L versus 53.4 to 45.0; AST from 50.1 to 33.1 versus 47.4 to 40.1; BMI from 35.8 to 32.0 versus 36.0 to 33.9. HDL, LDL, and triglycerides moved the same direction.
Moderatefor diet-plus-device beating diet alone on these markers in this one trial; Speculativefor any mechanism, since none was measured.
The control arm wore no device at all, so there is no sham comparison: attention, expectation, and adherence differences are all bundled into the “red light effect.” A 15 cm waist drop in 12 weeks is also a large fat-loss result, and liver enzymes generally fall with weight loss, so the trial can't separate light acting on the liver from light-associated fat loss. Liver fat itself was never re-imaged after baseline; only enzymes and lipids were followed. It is a single-center study of 30 per arm, and the reporting is loose: every between-group comparison is given as P = 0.01 (every within-group as 0.001 or 0.01) with no exact values or confidence intervals, and the baseline table lists BMI up to 39.8 and ages from 60 despite inclusion criteria of BMI 30–34.9 and age 65–75. A social-media thread summarizing the trial also credits red light with lowering the fat-synthesis genes SREBP1C, FAS, and SCD1; this trial did not measure gene expression, so that claim rests on separate work the thread doesn't cite. The direction is encouraging and the design is randomized; the size of the effect and the reason for it are both open.
Outdoor light exposure within the first hour or two of waking is the single highest-leverage change available given how much stronger daylight is than any indoor source — nothing indoors easily substitutes for it. Getting outside more broadly through daylight hours, dimming and warming indoor light in the evening, and protecting a genuinely dark night are the same three levers described across Parts II–V, applied on a daily clock rather than a lifetime-cancer-risk one.
None of this licenses unprotected midday UV exposure without regard to skin type or duration — see the flag in Part III. The dosing question for UV specifically (how much, for whom, before benefit gives way to risk) remains the least resolved part of this paper; see Limitations.
What would strengthen this paper most is a randomized trial dosing UV exposure by skin type against both cancer incidence and cardiovascular mortality in the same cohort — the current evidence base assembles these from an ecological review, one observational cohort, and one melanoma-survival study with different designs, populations, and outcome measures.