Rust, Copper and the Closing Tap: Menopause, Iron, Endothelin-1 and Bone

Why "oestrogen goes down, bone goes down" is not the whole story

Postmenopausal bone loss is usually told as one sentence: oestrogen falls, osteoclasts are released from restraint, bone is resorbed. That sentence is true, but it is not complete. Menopause also changes three other systems at the same moment, and each of them reaches bone.

• Iron. The monthly iron outflow stops, and body iron climbs for decades afterwards.
• Copper. Oestrogen drives the liver's output of ceruloplasmin (caeruloplasmin is the British spelling), the copper enzyme that moves iron out of tissues. When oestrogen falls, that support goes too.
• Endothelin-1 (ET-1). Oestrogen holds down ET-1, the body's most powerful vasoconstrictor. When oestrogen falls, the balance tips towards constriction, including in the small arteries that feed bone.

This piece follows those threads through the primary literature. The organising lens is Douglas Kell's work on poorly liganded iron (free iron), the chemistry of hydroxyl radicals and what serum ferritin really measures. Fluoride is added at the end because it is the clearest proof that bone density and bone strength are not the same thing.

A note on standards. This is not a rehearsal of guidelines. It is also not a sales pitch for an alternative protocol. Where the evidence is strong it says so; where it is animal-only, correlational, or contradicted, it says that too. A hypothesis that cannot survive its awkward data is not worth holding.

Kell's framework: it is the liganding of iron, not just the amount

Douglas Kell (Manchester, later Liverpool) has spent nearly two decades arguing one point. What damages tissue is not iron as such, but iron that is poorly bound. His 2009 review "Iron behaving badly" runs to 79 pages and more than 2,000 references [1].

The chemistry

Normal aerobic metabolism makes superoxide and hydrogen peroxide all the time. Kell's point is that these are not very reactive on their own. They become dangerous when they meet iron that is not properly held by a protein or chelator. That iron catalyses the Fenton and Haber-Weiss reactions and produces the hydroxyl radical, which attacks almost anything nearby [1].

Superoxide also reduces ferric iron back to ferrous iron, so a small amount of loose iron can redox-cycle again and again. That is why Kell describes "runaway" reactions that are hard to stop unless several points are tackled together [1].

Three consequences follow from this view:

1. Antioxidants can backfire. In the presence of poorly liganded iron (free iron), a nominal antioxidant can reduce iron and act as a pro-oxidant. Vitamin C is the clearest example: it mobilises iron stores and raises free iron, which can worsen iron toxicity and impair cardiac function in iron overload [94] (see "Vitamin C: it's the dose" below). Kell uses this to explain the mixed results of antioxidant trials. He argues that controlling redox stress needs both antioxidants and effective iron chelators, and that some polyphenols do both jobs [1].

2. Iron sits upstream of inflammation. Kell links redox stress to inflammatory cytokines through the NF-κB signalling pathway, and places iron's oxidative effects upstream of IKK, the IkB kinase [1]. In plain terms: NF-κB is the cell's master switch for inflammatory cytokines such as IL-6 and TNF-α. In a resting cell it is held in check by a partner protein, IkB ("inhibitor of κB"). IKK is the enzyme that tags IkB for destruction, which releases NF-κB to switch inflammatory genes on. So if free iron acts before IKK, it sits near the top of the inflammatory cascade: a trigger, not just a by-product.

3. Speciation matters more than totals. Total body iron tells you little. Where the iron is, and what it is bound to, tells you much more.

Vitamin C: it's the dose

Vitamin C is the clearest everyday example of Kell's point that an antioxidant can turn against you when iron is loose. High-dose oral vitamin C and IV vitamin C drips are now widely sold. The question that matters is dose.

More is not better. In the NIH depletion-repletion study, blood levels rose steeply between 30 and 100 mg a day, and plasma was completely saturated at 1,000 mg a day [93]. At single doses of 500 mg and above, absorption fell and the extra was excreted [93]. The authors concluded that doses above 400 mg a day have no evident value [93].

Where vitamin C feeds free iron. Vitamin C mobilises iron stores. That increases chelatable iron, which helps the chelator deferoxamine work, but the same rise in free iron can worsen iron toxicity and impair cardiac function [94]. Patients on deferoxamine once received 15–30 mg/kg a day of ascorbic acid; reports of sudden death ended the practice [95]. As the Brigham and Women’s iron overload service puts it, the total quantity of iron matters less than the unbound, "toxic" iron [95], which is Kell’s argument in a clinical setting. The dose decides the outcome. At 100 mg a day, vitamin C helped chelation lower iron burden in vitamin C-deficient children with thalassaemia, with no adverse events [96].

The takeaway: vitamin C is essential and correcting deficiency matters. But megadosing, by tablet or drip, can push free iron up, and has limited evidence of benefit in people who are not deficient. That applies with extra force to anyone with haemochromatosis, raised transferrin saturation or iron overload [94, 95], and to a postmenopausal woman whose iron is accumulating. If supplementing, keep it modest.

Serum ferritin is mostly a leakage product

Kell and Etheresia Pretorius made a second, very practical argument in 2014. Ferritin is made inside cells, not in serum, so why is it in blood at all? Their answer: serum ferritin mostly leaks from damaged cells. It is therefore a marker of cell damage. Crucially, that leaked ferritin has already dumped most of its iron, and unliganded iron (free iron) is highly toxic [2].

This resolves a paradox. Ferritin correlates with iron stores, with hydroxyl-radical markers and with the severity of many diseases. On Kell and Pretorius’s view these are expected on simple chemical kinetic grounds, not separate puzzles [2].

Why this matters for menopause: postmenopausal ferritin rises substantially (see the next section). On the standard reading that is "more stores". On Kell's reading it is also partly a readout of tissue damage, with free iron released along the way.

The Iron Dysregulation and Dormant Microbes (IDDM) hypothesis

In 2018 Kell and Pretorius widened the argument. Chronic inflammatory diseases, they wrote, need external causes. They proposed two: stress-induced iron dysregulation, and that free iron’s ability to wake dormant, non-growing microbes already living in the host [3].

Once woken, those microbes shed small but potent amounts of inflammatory molecules such as lipopolysaccharide (LPS) and lipoteichoic acid [3]. Downstream come blood-clotting abnormalities, amyloid formation and cell death [3]. In their account, whether such bacteria stay dormant or wake often depends on how much free iron is available [4].

This has not been tested in osteoporosis specifically. If the IDDM model holds, rising free iron after menopause could add a microbial inflammatory load to bone. That is a testable prediction, not an established fact.

The closing tap: iron accumulation after menopause

Urine, sweat, bile and digestive juices carry essentially no iron out of the body. Before menopause, menstrual blood is a woman's main iron exit [7]. Close that tap and iron rises slowly for years. The Soochow group (You-Jia Xu) and Xi Huang's group at NYU have built this into the "iron accumulation" hypothesis of postmenopausal osteoporosis [5, 6, 8].

Their definition matters. "Iron accumulation" means serum ferritin above the normal range but below 1,000 µg/L [7]. That is not haemochromatosis. It is a common process in ageing women, which may help explain why women get osteoporosis more often than men [6]. But this definition rests on serum ferritin, and that needs a caveat. Ferritin is made inside cells, not in blood. Kell and Pretorius argue that serum ferritin is mainly a leakage product from damaged cells, and so chiefly a marker of cell and tissue damage rather than a simple measure of iron stores [2]. The ferritin that leaks out has already dumped most of its iron, releasing unliganded (free) iron, which is highly toxic [2]. Ferritin can still correlate with body iron stores, but Kell and Pretorius show this is expected on chemical kinetic grounds rather than meaning ferritin is a store gauge [2]. So a rising ferritin after menopause is best read as a sign that cells are being damaged and free iron released, not as reassurance that the iron is safely stored.

Oestrogen and hepcidin: the hormonal valve

Hepcidin is the liver hormone that shuts iron absorption by degrading the iron exporter ferroportin. Two independent groups found the same thing in 2012: there is a functional oestrogen-response element in the hepcidin gene, and oestradiol suppresses hepcidin [9, 10]. Ovariectomised mice had higher liver hepcidin transcription [9].

Huang's group reads this as a menstrual compensation: oestrogen lowers hepcidin to boost iron uptake while iron is being lost monthly [10]. After menopause, oestradiol falls by more than 90%, while body iron climbs slowly through steady absorption over the years [12].

There is a subtlety here that many popular accounts miss. In late-postmenopausal women, serum hepcidin is higher, alongside higher ferritin and iron in arterial plaques [13]. So it is not simply that hepcidin stays low. The accumulation is driven mainly by the loss of outflow, and the high hepcidin then traps iron inside macrophages and tissues rather than releasing it.

What the human bone data show

The key study is Kim and colleagues (2012), from Seoul. They followed 1,729 healthy Koreans (940 postmenopausal women, 789 men) for an average of three years [14]. Compared with the lowest ferritin quartile, people in the upper quartiles lost bone significantly faster at the total femur and femoral neck [14]. The authors called this the first clinical evidence that higher body iron could independently accelerate bone loss, even in healthy people [14].

Two details are worth noting:

• Average bone loss in the women was about 1.1-1.5% a year at the hip sites, versus 0.3-0.4% in men [14].
• Although women's ferritin was lower than men's, the effect of iron stores on bone appeared more striking in women [15].

The same group later found a linear link between ferritin and vertebral fracture prevalence in women, but not men [6].

Mechanisms at the bone cell level

• Oxidative stress and Wnt. Accumulated iron may suppress the Wnt/β-catenin pathway (the main bone-building signal) through oxidative stress. That cuts bone formation and raises resorption via the OPG/RANKL balance (RANKL, receptor activator of nuclear factor-κB ligand, is the signal that switches on bone-dissolving osteoclasts; OPG, osteoprotegerin, is the decoy that blocks it, so the ratio between them sets how much bone is broken down) [5].
• Direct cell effects. Iron accumulation has been reported to inhibit osteoblast formation and function, and to stimulate osteoclasts directly [5]. Huang's group showed iron inhibits pre-osteoblast differentiation and alkaline phosphatase activity [10].
• Ferroptosis of osteocytes. A 2024 study in Advanced Science showed that oestrogen withdrawal causes iron to build up in the skeleton and triggers ferroptosis (iron-dependent cell death) in osteocytes [16]. The dying osteocytes then drive osteoclast formation by changing methylation of the RANKL gene promoter [16]. Osteocytes are the master regulators of bone remodelling, so this is a significant finding.
• Animal proof of principle. Raising hepcidin genetically in mice protected against iron-driven bone loss [11]. Iron chelation and hepcidin have both been proposed as treatments [6].

The oestrogen-receptor twist

A 2023 eLife study offers a striking link to the "timing hypothesis" of HRT. In women's arterial plaques, oestrogen receptor-α levels were negatively associated with iron content [13]. The authors propose that age-related iron accumulation explains why HRT loses its benefit in late postmenopause, and suggest early HRT plus appropriate iron chelation [12]. If iron degrades the receptor oestrogen acts through, then iron load would blunt oestrogen's protection of bone as well as arteries. That extension to bone is a hypothesis, not yet shown.

Copper: the metal that moves iron and cross-links bone

Copper enters this story twice: once as the catalyst that lets iron leave tissues, and once as the cofactor that gives bone collagen its strength.

Copper is the exit key for iron

Iron leaves cells through ferroportin as ferrous iron (Fe²⁺). To load onto transferrin it must first be oxidised to ferric iron (Fe³⁺). That step is done by the multicopper ferroxidases: ceruloplasmin in plasma and hephaestin in the gut wall [21]. Both need copper to work.

The animal genetics are blunt:

• Knock out ceruloplasmin in mice and iron accumulates in multiple tissues [22].
• Knock out both ceruloplasmin and hephaestin and mice show both tissue iron overload and severe iron-deficiency anaemia at once [21]. Iron is present, but stranded.
• Dietary copper deficiency lowers ceruloplasmin, its GPI-anchored form (a version of ceruloplasmin tethered to the outside of cell membranes by a glycosylphosphatidylinositol, or GPI, anchor, which helps cells such as macrophages release their iron [20]), and hephaestin. Copper shortage seems to reduce the stability and turnover of these enzymes rather than their synthesis, which shows up as impaired iron flow [20].

That last paper adds a clinical warning: zinc-induced copper deficiency is now an important concern, and it needs proper diagnosis so that ineffective and potentially harmful treatment with iron or B12 is not given [20]. In other words, iron tablets for a copper-deficiency anaemia push more iron into a system that cannot move it.

The oestrogen-ceruloplasmin connection

Oestrogen regulates the liver's long-term production of ceruloplasmin, probably by stabilising its mRNA [23]. Oral ethinyl oestradiol in postmenopausal women raised plasma copper, about 80% of it as ceruloplasmin-bound copper [24].

Put those facts beside the iron data and a hypothesis follows. At menopause, oestrogen's support for ceruloplasmin falls at the same time as iron outflow stops. Less ferroxidase capacity plus a rising iron load is exactly the combination the knockout mice show produces stranded tissue iron. As far as I can find, no study has directly measured ceruloplasmin ferroxidase activity across the menopause transition alongside bone iron. That is a gap worth filling.

One caution: total serum copper and ceruloplasmin protein are poor markers of copper status. Both rise with oestrogen and with inflammation, since ceruloplasmin is an acute-phase protein. The ratio of ceruloplasmin enzyme activity to ceruloplasmin protein has been proposed as more sensitive [23]. A "normal" serum copper says little about whether the enzyme is actually loaded and working. So these markers should not be read alone, but alongside others that change how they should be interpreted: inflammatory markers, because inflammation raises ceruloplasmin regardless of copper status [23]; a full iron panel, because ceruloplasmin's job is moving iron [20, 21] and ferritin itself reflects cell damage as well as stores [2]; retinol, because retinol drives ceruloplasmin production [51]; and red-cell magnesium, because magnesium status shaped how the body handled minerals and copper in Nielsen's boron studies [63, 64].

Copper builds the collagen scaffold

Copper is the cofactor for lysyl oxidase, the enzyme that makes the first cross-links in collagen so it forms stable fibrils [17]. In copper-deficient animals, bone loses mechanical strength because collagen and elastin are poorly cross-linked, and this occurs even when bone calcium is unchanged [30]. Mineral density can look fine while the scaffold under it is weak.

In humans, frank copper deficiency causes osteoporosis that responds to copper [29]. The question is whether milder, common shortfalls matter after menopause. The human evidence, honestly stated:

The pattern is consistent with copper mattering most for women with low intake, over years rather than weeks. It does not support a claim that copper alone reverses osteoporosis.

Klevay's copper-deficiency theory

Leslie Klevay (University of North Dakota, USDA) spent 50 years arguing that ischaemic heart disease is largely a copper-deficiency disease. He lists more than 80 anatomical, chemical and physiological similarities between copper-deficient animals and people with heart disease, including high cholesterol, glucose intolerance, high uric acid and hypertension [25]. He also argues that copper intakes calculated from food tables are inflated by an average of 77% [25]. Controlled copper depletion in men and women raised cholesterol and blood pressure [26].

Klevay's work matters here because bone and arteries fail together after menopause, and his theory gives one shared cause. Copper deficiency impairs lysyl oxidase in both collagen and elastin, and, as the next section shows, it damages nitric-oxide signalling in the endothelium.

Endothelin-1: the constrictor oestrogen keeps in check

The endothelium balances two signals: nitric oxide (NO), the main dilator, and endothelin-1 (ET-1), the most potent constrictor it makes [35]. Oestrogen pushes that balance towards NO. Oestrogen is associated with lower ET-1, a constrictor and pro-inflammatory peptide [34].

What happens to ET-1 at menopause

The trial data on oestrogen and ET-1 are consistent:

• Ylikorkala et al. 1995. In 26 healthy postmenopausal women, HRT cut plasma ET-1 from 1.28 to 1.05 pmol/ml at 6 months, and it stayed down at 12 months [31].
• Best et al. 1998. Six months of oestradiol lowered ET-1 by a mean 3.9 pg/mL and raised the NO-to-ET-1 ratio from 2.0 by 1.2 [32].
• 24-month comparison. In 51 postmenopausal women, ET-1 fell about 18–23% on tibolone, oral or transdermal HRT. In untreated controls it rose 36.6% [33].

That last number is the menopause trajectory in miniature: without oestrogen, ET-1 climbs.

Endothelial function declines with menopause independently of chronological age, and the decline is linked to oestrogen deficiency [34]. It starts early, in the perimenopause, and one NIH-funded programme proposes ET-1 activation and loss of ET-B-mediated dilation as primary drivers [35]. The picture is not simple, though. A University of Delaware study found lower endothelial-cell ET-1 and ET-B receptor expression in postmenopausal women than in young women, with no difference in constriction to added ET-1 [37]. The system changes, but not only by "more ET-1".

Why a vascular peptide matters for bone

Bone is a highly vascular organ, and blood flow through it falls with age. Reduced bone and marrow blood flow has been reported in healthy people over 60 and is associated with osteopenia [36].

Rhonda Prisby and Michael Delp tested the link directly in female rats [36]:

• In young rats, ovariectomy reduced endothelium-dependent dilation of the femur's main nutrient artery from 82% to 61%. Oestrogen restored it to 90%.
• Trabecular bone volume followed the same pattern: 28.7% intact, 16.3% after ovariectomy, 25.7% with oestrogen.
• In old rats with low oestrogen, artery dilation correlated with bone volume (R² = 0.63).

The authors propose an endothelium-dependent "vascular coupling" mechanism linking how well bone's arteries dilate to how bone remodels [36]. A constricted bone circulation means less perfusion, less interstitial fluid flow through the bone matrix and fewer of the signals bone cells use to regulate themselves [36].

ET-1 is not simply "bad" for bone, however. It acts directly on bone cells: it has been shown to inhibit osteoclast bone resorption by reducing osteoclast motility [40]. Its harm is more plausibly vascular (choking blood supply) than a direct push on resorption.

Where copper and iron meet ET-1

This is where the threads join. Both metals tip the NO/ET-1 balance the same way that oestrogen loss does.

Copper deficiency starves the endothelium of NO. In copper-deficient rats, NO-mediated dilation of small arteries was blunted, and adding superoxide dismutase (SOD) restored it [38]. The explanation: copper is needed for Cu,Zn-SOD, the enzyme that clears superoxide. Without it, superoxide destroys NO before it can act [38]. USDA work also found that copper deficiency raised blood pressure by reducing NO, and promoted vascular inflammation independently of NO [39].

Iron overload drives oxidative stress in vessels. Excess iron produces non-transferrin-bound iron, which generates reactive oxygen species and lipid peroxidation in blood vessels [41]. Endothelial ageing includes reduced NO production and increased ET-1 release, and ET-1 itself raises adhesion molecules and NF-κB [41]. This is Kell's hydroxyl-radical chemistry acting on the endothelium.

So the menopausal endothelium may take three hits at once:

1. Loss of oestrogen's direct suppression of ET-1 and support for NO.
2. Rising tissue iron generating superoxide-driven, iron-catalysed oxidative stress.
3. Any copper shortfall weakening SOD and ferroxidase defences against both.

No single study has measured all three in the same women against bone blood flow. Each link is supported separately; the combined model is a synthesis.

Fluoride: denser bone that breaks more easily

Fluoride is the clearest demonstration in all of bone medicine that bone mineral density is not bone strength. At high doses this is proven by randomised trials. At drinking-water doses the evidence is mixed but no longer reassuring.

The high-dose trials: a natural experiment in bone quality

Fluoride stimulates osteoblasts, unlike most osteoporosis drugs, which slow resorption [44]. In the 1980s that made it look like the ideal bone builder. Then came the Mayo Clinic trial.

Riggs et al., NEJM 1990. 202 postmenopausal women with osteoporosis and vertebral fractures received sodium fluoride 75 mg/day or placebo, all with 1,500 mg calcium, for four years [42]. The results:

• New vertebral fractures: 163 on fluoride versus 136 on placebo, not significantly different [42].

• Non-vertebral fractures: 72 on fluoride versus 24 on placebo (P<0.01) [42]. The relative risk was 3.2 [43].

• Hip fractures: six in the fluoride group (three complete), one in the placebo group [43].

• The authors concluded fluoride increases cancellous but decreases cortical bone density, and increases skeletal fragility [42]. (Cancellous, or trabecular, bone is the spongy honeycomb inside vertebrae and the ends of long bones. Cortical bone is the dense outer shell that makes up most of the shafts of long bones such as the thigh bone and forearm, and carries much of their strength.) So fluoride thickened the spongy bone the spine scan measures, while thinning the hard outer shell, which fits the rise in non-vertebral fractures.

The extended follow-up made the paradox sharper. Spinal BMD on fluoride rose linearly for six years, a median 8.7% a year [45]. Vertebral fracture rate fell as BMD rose to about 1.2 g/cm², then doubled as BMD climbed further to 1.6 g/cm² [45]. More mineral, past a point, meant more breaks.

In the parallel Henry Ford trial, 17% of fluoride-treated patients had mineralisation defects on bone biopsy, against none in controls [43]. The 2000 Cochrane review pooled the evidence: fluoride raises spinal BMD but does not reduce vertebral fractures, and at four years non-vertebral fractures rose (RR 1.85), especially at high doses and with non-slow-release forms [44].

Skeletal fluorosis: the toxicology

Endemic skeletal fluorosis shows what chronic excess does. Fluoride is a cumulative toxin. Its bone lesions combine osteosclerosis, osteomalacia and osteoporosis in varying degrees, sometimes with secondary hyperparathyroidism [49]. Toxicity features include higher bone turnover and impaired bone collagen synthesis [49], which echoes the copper-lysyl oxidase point above: the scaffold matters as much as the mineral.

A 2021 Washington University case review explains the mechanism neatly. Fluoride is directly anabolic on bone; it forms fluorapatite crystals that resist degradation, which drives secondary hyperparathyroidism when dietary calcium is low; and the excess osteoid laid down then fails to mineralise properly [48]. Severity tracked cumulative fluoride exposure, kidney function and calcium and vitamin D status [48].

Of particular relevance to British readers: of 19 non-endemic cases with a known fluoride source, 11 came from heavy black tea and/or fluoride toothpaste consumption [48].

Toothpaste: the fluoride source in every bathroom

How much is in it. Toothpaste above 1,500 ppm fluoride can only be supplied on prescription; prescription pastes contain 2,500 or 5,000 ppm [97]. A part per million is a milligram per kilogram, so a 1,500 ppm paste carries 1.5 mg of fluoride in every gram.

How much is swallowed. It is not uncommon to swallow about 25% of the toothpaste put on the brush [98]. Fluoride from dental products such as toothpaste and mouth rinse can add considerable fluoride to the diet, often approaching or exceeding the intake from food and water [99].

Where it goes. About 99% of the body's fluoride is held in calcified tissues, and in healthy young and middle-aged adults about half of absorbed fluoride is retained in the skeleton; the other half leaves in urine [99]. Fluoride enters bone easily but leaves it slowly: release of fluoride bound as fluorapatite requires osteoclastic bone resorption, and the whole-body half-life once in bone is around 20 years [100]. A 1985 analysis warned that daily intakes considered beneficial for developing teeth may, if continued throughout adult life, lead to skeletal fluorosis of varying degrees in a significant proportion of the population, and that skeletal fluoride accumulation had been ignored as a possible cause of bone disorders common in middle and old age [101].

The timing matters for women. In a study of human ribs from 119 people, fluoride rose steadily with age in men, but in women it levelled off until 55 and then increased markedly [102]. That is the decade after menopause.

The jaw. The maxilla and mandible sit closest to toothpaste exposure, but direct human data on jaw fluoride are scarce. Animal studies show the jaw is a strong fluoride store:

• In rats given fluoridated drinking water for seven months, jawbone fluoride rose to about 8,400 mg/kg, against about 6,100 mg/kg in the femur. The authors concluded that, compared with the femur, the jawbone showed significantly high fluoride accumulation [103].
• In rat mandibles, fluoride rose in proportion to the fluoride in drinking water and the length of exposure. When fluoridation stopped, the surface layers lost fluoride, fastest at the alveolar crest, the bone around the tooth sockets [104].

Whether toothpaste adds fluoride to the jaw locally, rather than through the bloodstream, has not been shown.

When it goes wrong. In the case series reviewed earlier, 11 of 19 non-endemic skeletal fluorosis cases came from heavy black tea and/or fluoride toothpaste [48]. Two examples:

• A 53-year-old woman presented with a metatarsal fracture and a spine bone density T-score of +11. Investigation found very high serum, urine and tissue fluoride from excessive tea and toothpaste [98]. Her toothpaste contained 1,691 mg/kg fluoride,

similar to other fluoridated pastes [98]. Her bones were extremely dense and still broke.

• A 52-year-old man with severe neck immobility and osteosclerosis was diagnosed with skeletal fluorosis; nearly a decade of follow-up documented considerable correction after removal of the putative source, toothpaste [105].

The takeaway for bone: toothpaste belongs in the total fluoride count alongside tea and water [99]. A scan showing unexpectedly high bone density in a heavy tea drinker or toothpaste user is a reason to check fluoride, not a reason for reassurance [98].

Drinking-water levels: the question is open

Helte et al. 2021 (Swedish Mammography Cohort). 4,306 postmenopausal women exposed to drinking water at or below 1 mg/L were followed to 2017 [46]. The highest versus lowest third of exposure had hazard ratios for hip fracture of 1.50 for urine fluoride and 1.59 for dietary fluoride [46]. Baseline BMD was slightly higher in the highest-exposure women [46]. Restricting to women with consistent long-term water exposure strengthened the urinary association [46].

The authors concluded that high consumption of water at around 1 mg/L may increase both BMD and skeletal fragility in older women, consistent with the trials [46]. That is the same "denser but weaker" signature as the high-dose trials, at doses that WHO considers safe.

Against that: a nationwide Swedish register study of 473,277 people born 1900–1919 found no association between chronic drinking-water fluoride and hip fracture [47]. It relied on estimated drinking-water exposure rather than measured dietary intake or urine fluoride [47].

The honest summary: high-dose fluoride causes fragile bone; that is settled. Whether long-term intake around 1 mg/L from water plus tea raises fracture risk in postmenopausal women is unsettled, and the best individual-level study suggests it might. Because DXA scores rise with fluoride, a reassuring DXA result in a heavy tea drinker or fluorosis-area resident may overstate how strong her bones are.

Putting it together: a metals-and-vessels model of postmenopausal bone loss

Pulling the threads together gives a model in which oestrogen loss is the trigger, but iron, copper and the endothelium decide how much damage follows.

1. The tap closes. Menstrual iron loss stops. Body iron rises steadily for years [7, 12].

2. The exit narrows. Oestrogen’s support for ceruloplasmin is lost [23]. Any copper shortfall (low intake, high-dose zinc, malabsorption) cuts ferroxidase activity further [20]. Iron becomes stranded in tissues, as in ceruloplasmin-knockout mice [21, 22].

3. Hepcidin traps it. Late postmenopausal hepcidin rises [13], so iron is held in macrophages and tissue stores rather than recycled.

4. Poorly liganded iron (free iron) does Kell’s chemistry. Loose iron plus normal superoxide and peroxide makes hydroxyl radicals, activating NF-κB upstream [1]. Damaged cells leak ferritin that has dumped its iron, feeding the cycle [2].

5. Bone cells die by ferroptosis. Skeletal iron triggers osteocyte ferroptosis, which in turn drives osteoclast formation through RANKL [16]. Iron suppresses osteoblasts directly and via Wnt [5, 10].

6. The bone’s blood supply constricts. ET-1 rises without oestrogen [33]; NO is destroyed by superoxide when copper-dependent SOD falls short [38] and by iron-driven oxidative stress [41]. Bone artery dilation tracks bone volume [36].

7. The collagen scaffold weakens. Low copper means less lysyl oxidase cross-linking [17, 30]. High fluoride adds poorly mineralised, brittle bone and impaired collagen synthesis [48, 49].

8. Oestrogen’s own receptor erodes. Iron load is associated with lower ER-α [13], which may explain why HRT started late helps less [12].

9. Possible microbial amplifier. If Kell’s IDDM hypothesis holds, free iron wakes dormant bacteria, which shed inflammatory molecules such as LPS [3].

What the model predicts (and how you would test it)

None of these have been answered in a definitive human trial. That is the plain state of the field.

What may help: copper, retinol and boron

If the model above is broadly right, the targets are clear: keep the copper enzymes loaded and working, support collagen cross-linking, and support the hormones and minerals bone depends on. Three nutrients map onto those targets. Each comes with real evidence and real cautions.

Copper: refilling the ferroxidases and lysyl oxidase

Copper is the one nutrient that sits under every arm of the model: ceruloplasmin and hephaestin (moving iron out of tissues) [20, 21], lysyl oxidase (cross-linking collagen) [17], and Cu,Zn-SOD (protecting nitric oxide) [38]. Dietary copper modulates lysyl oxidase activity directly in rats [58].

A striking rat result ties copper to where iron ends up. Copper restriction lowered iron in the tibia and femur and lowered haemoglobin, but raised iron in the liver [57]. Low copper did not reduce iron; it misplaced it.

Practical points:

• Doses that worked in trials were modest: 2.5 mg/day [18] and 3 mg/day on top of the diet [17], over two years. Klevay argues real intakes are often lower than food tables suggest [25].
• Watch zinc. In rats, high zinc with adequate copper lowered ceruloplasmin oxidase activity [52], and zinc-induced copper deficiency is a recognised clinical problem [20].
• Cautions: Wilson’s disease is a rare inherited disorder of copper accumulation, with symptomatic disease in roughly 1 in 30,000 people [62, 106]. It is uncommon, but in anyone who has it, extra copper is unsuitable. Because oestrogen and inflammation raise serum copper, a single serum copper is a weak guide [23].

Retinol from food: building and exporting ceruloplasmin

The chain is: retinol supports the making and release of ceruloplasmin; copper fills it and makes it work; working ceruloplasmin moves iron out of tissues. Copper separately powers lysyl oxidase. Everything here refers to preformed retinol from food, not supplements.

How ceruloplasmin gets its copper. The liver makes ceruloplasmin as a copper-free protein. In the trans-Golgi, the copper pump ATP7B loads it with six copper atoms before it is secreted into blood [61, 62]. Loading is cooperative: all six sites must fill for the protein to reach its final working shape, with no order of priority between them [61].

Where retinol comes in. Three lines of evidence:

• In vitamin A-deficient rats, giving retinoic acid significantly raised both ceruloplasmin synthesis and activity [51].
• In copper-deficient rats, retinoic acid did not raise ceruloplasmin activity unless copper was given too [51]. Retinol needs copper to produce working enzyme.
• In liver cells from people with Wilson’s disease, where ATP7B is faulty, ceruloplasmin secretion was consistently low. A drug screen found that all-trans retinoic acid and approved retinoids rescued ceruloplasmin secretion [62].

So the evidence firmly places retinol in the production and export of ceruloplasmin, with copper essential for it to work. ATP7B physically loads the copper. Whether retinol acts on that loading step itself, or on the protein’s synthesis and secretion around it, has not been pinned down. The practical conclusion is the same either way: low retinol or low copper alone each leaves ceruloplasmin, and therefore iron export, short.

Food, not pills, and still not unlimited. In the Nurses’ Health Study, women not taking vitamin A supplements who ate 1,000 µg/day or more of retinol from food had a higher hip-fracture risk than those eating under 400 µg/day (RR 1.69) [56]. The association was weaker in women taking oestrogen [56]. Beta-carotene from vegetables did not raise risk significantly [56]. But beta-carotene is not retinol, and for many women it is a poor substitute.

Beta-carotene is not retinol. Beta-carotene has to be cleaved by the enzyme BCMO1 into retinal, the precursor of retinol [85]. How well people do this varies widely: about 40% of healthy subjects are low responders, and stable-isotope studies show up to 8-fold differences between the best and worst converters [85]. Much of this is genetic:

• Leung et al. (Newcastle, 2009) found two common BCMO1 variants in female volunteers, with allele frequencies of 42% and 24% [83]. Carriers converted beta-carotene less well: 32% less with one variant and 69% less with both, with fasting beta-carotene 160–240% higher because it was not being converted [83].

• Three further variants upstream of the gene cut BCMO1 activity in women by 48-59% [84].

• About 44% of the Western population carry the 379V variant, so a large share may not reach adequate vitamin A if beta-carotene is their main source [85].

The researchers flag vegetarians and others with low retinol intake as most at risk [85]. For this piece the implication is direct: a woman relying on carrots and greens for vitamin A may be short of the retinol that ceruloplasmin production depends on [51], even on a diet that looks adequate on paper.

Copper and lysyl oxidase need no retinol. Lysyl oxidase is a copper-dependent enzyme that makes the first collagen cross-links [17, 54], and dietary copper directly modulates its activity [58]. Retinoic acid does regulate the lysyl oxidase family, but in tissue-specific and sometimes opposite directions [53, 55].

Boron: the trace element that touches every arm of the model

Lara Pizzorno's 2015 review, "Nothing Boring About Boron", argues that boron should be treated as an essential micronutrient [50]. Looked at closely, boron is the one nutrient here that reaches oestrogen, copper and ceruloplasmin, magnesium and calcium, vitamin D, and fluoride all at once. That makes it unusually relevant to the model above.

1. Boron raises oestradiol in postmenopausal women

The founding study is Nielsen and colleagues at the USDA Grand Forks centre (1987) [63]. Twelve postmenopausal women ate a low-boron diet (0.25 mg/day) for 119 days, then received 3 mg/day [63]. That 3 mg is about what a diet rich in fruit and vegetables supplies [63].

The results were large. Pizzorno summarises the hormone data [50]:

Urinary calcium fell by 44% [50]. Calcium saved was 52 mg/day on the low-magnesium diet and 22 mg/day on the adequate one [50]. Urinary magnesium also fell [63]. The authors concluded these changes were consistent with preventing calcium loss and bone demineralisation [63].

In the context of this piece, that is striking: a dietary amount of boron roughly doubled circulating oestradiol in women past menopause.

2. Boron raises ceruloplasmin and copper-SOD

This is the link most accounts miss, and it ties boron directly to the iron-copper axis. In Nielsen's follow-up study, older men and postmenopausal women (some on oestrogen) ate a low-boron diet, low in magnesium and only marginal in copper, then had boron restored [64]. During boron repletion, erythrocyte superoxide dismutase, serum enzymatic ceruloplasmin and plasma copper were all higher than during depletion [64].

Two further findings matter:

• Oestrogen therapy raised plasma copper and oestradiol, and boron deprivation blunted both rises [64]. Boron appears to be needed for oestrogen's effect on copper.
• Immunoreactive ceruloplasmin and SOD were higher during boron repletion in all subjects, not only those taking oestrogen [64].

So boron supports the same two copper enzymes this piece identifies as critical: ceruloplasmin (iron export) and Cu,Zn-SOD (protecting nitric oxide from superoxide). It does so on a marginal copper intake, which is the realistic situation for many women.

3. Boron and magnesium

The magnesium interaction runs through every boron study. Effects on calcium loss and hormones were consistently larger when magnesium intake was low [63, 50]. A later metabolic-ward study found that boron changed calcium and magnesium absorption and retention, again depending on magnesium status [72]. Nielsen notes that the body's response to boron is markedly shaped by dietary magnesium, potassium and methionine [75]. In practice, boron and magnesium status should be considered together.

4. Boron and vitamin D

Boron improves vitamin D use. In vitamin D-deficient animals, boron stimulated bone growth and corrected the mineral-metabolism faults of deficiency [50]. Boron also helps osteoblasts mineralise by regulating mineralisation genes and the actions of oestradiol, testosterone and vitamin D [50].

5. Boron and fluoride

Boron has a long history in fluorosis research. Borate binds fluoride, and older work tested boron as an antidote in rabbits and in human skeletal fluorosis [69]. In rats drinking water with 100 ppm fluoride, added boron lowered body fluoride and relieved toxic effects; the authors concluded boron prevents skeletal fluorosis [70]. Boron with calcium also reduced fluoride in internal organs and improved SOD activity in animals [74]. This is mostly animal and older clinical work. But it suggests boron could partly offset the fluoride concerns in the previous section.

What the human bone evidence shows, and where it stops

The honest gap: no trial has yet shown that boron raises bone density or prevents fractures in postmenopausal women. The evidence is about hormones and mineral handling, not bone outcomes. A one-year, placebo-controlled phase 2 trial in postmenopausal women with osteoporosis has been registered in Iran but was not yet enrolling [71]. And the British replication by Beattie and Peace did not find Nielsen’s effect [65]. One likely explanation is baseline status: Nielsen’s effects were largest in women depleted of boron and short of magnesium. Women already eating plenty of fruit, nuts and pulses may have little to gain.

Boron has still not been formally shown to be essential for humans [68]. But people fed only 0.2–0.4 mg/day for about two months responded positively when boron was restored [68].

Getting boron from food

Worldwide, typical intakes are about 1–3 mg/day, mainly from fruit, vegetables, nuts and pulses [73]. The 3 mg/day used in Nielsen’s study is about what a diet rich in fruit and vegetables supplies [63].

Caution: upper limits for boron differ by authority. In the US the adult tolerable upper intake level is 20 mg/day [107, 109]. The European Food Safety Authority (EFSA) set a stricter 10 mg/day for adults [67, 108].

How the limits are set. Neither figure comes from harm seen in people. Both the US and European limits were based on animal data, because suitable human data were not available [109, 110]. Most expert bodies start from the same rat study (Price et al., 1996). In it, the highest dose that did not reduce the weight of developing pups was 9.6 mg of boron per kg of body weight per day [111]. That figure is then divided by a safety ("uncertainty") factor to allow for differences between rats and people, and between one person and another. EFSA divided by 60, giving 0.16 mg per kg per day [110]; the UK Expert Group on Vitamins and Minerals reached the same 0.16 mg per kg per day [111]. The UK Committee on Toxicity notes that the limits differ because bodies read the same studies differently and chose different safety factors [111]. Studies of people with high boron exposure from their environment or work have not reported effects on reproduction or development [112].

So the upper limits are precautionary estimates scaled down from animal studies, not doses at which harm has been seen in people. The 3 mg/day used in Nielsen's studies is below all of them.

A note on the other side of the ledger: iron

The model implies that lowering excess iron is as important as adding copper. Approaches discussed in the literature reviewed here include iron chelation [6, 12], and Kell's point that some dietary polyphenols act as both antioxidants and iron chelators [1]. Regular blood donation is the obvious untested option. None have been tested in a fracture trial, and none should be tried with low ferritin or anaemia.

Professional caution: these are research findings, not a treatment plan. Copper, retinol and boron all have upper limits and interactions, and the right choice depends on an individual's iron, copper, liver and kidney status, and medications.

High-dose vitamin D: when more is worse for bone

This section asks what happens when women who are not deficient take high doses or large boluses. The randomised trials now give a consistent and uncomfortable answer: beyond modest intakes, more vitamin D does not help bone, and at high doses it can harm it.

The trials

Three patterns stand out:

1. Dose-response in the wrong direction. Burt’s team expected higher doses to increase bone density. They found a negative dose-response instead [76]. Their proposed mechanism: high doses raised a marker of bone resorption (CTX) while suppressing PTH [76].

2. Bolus doses are worst. In Sanders’ trial, fall risk was 31% higher in the first three months after each dose [77]. The authors suggested the high blood levels after a large dose, the drop that follows, or both might be causal [77]. Two trials using the same total annual dose split differently had opposite outcomes, pointing to the dosing pattern rather than total dose [77].

3. No benefit in women who are not deficient. VITAL, the largest trial, found no fracture reduction [79], and Burt’s group noted that bone benefit may be confined to treating deficiency [76].

This matters for postmenopausal women specifically. Burt’s group warned that if high-dose vitamin D speeds bone loss, it could matter more in older people who already have osteoporosis [76].

Where vitamin D meets the iron-copper model

Vitamin D lowers hepcidin. Active vitamin D directly suppresses transcription of the hepcidin gene via the vitamin D receptor, and raises ferroportin [80]. In healthy volunteers, one 100,000 IU dose of D2 cut circulating hepcidin by 34% within 24 hours [80].

In a younger woman losing iron monthly, that might help. In a postmenopausal woman whose problem is iron accumulation, lowering hepcidin with high-dose vitamin D would, in principle, increase iron absorption at exactly the wrong time. That is a hypothesis, and the best test so far does not support it at moderate daily doses: in the UK BEST-D trial, 2,000 or 4,000 IU/day for a year in people aged 65+ changed neither hepcidin, ferritin, serum iron nor transferrin saturation [81]. The acute effect may matter more with large boluses than with steady daily doses: a single 250,000 IU dose cut hepcidin by 73% within a week [90].

Vitamin D and boron. Boron extends and improves vitamin D’s actions [50]. The theme is the same as everywhere in this piece: balance between nutrients matters more than pushing one to a high dose.

Libbie Wallace: the case for caution

Libbie Wallace’s clinic blog “The case for caution: rethinking synthetic vitamin D” (3 November 2026) argues that vitamin D behaves like a steroid hormone, so the goal should be correcting real deficiency rather than chasing higher blood levels [82]. On bone, it draws on the same two trials as this piece and adds several mechanisms:

• Bone density and urine calcium. In the Calgary trial, total bone density changed by −1.4%, −2.6% and −3.6% at 400, 4,000 and 10,000 IU a day. High urine calcium developed in 17%, 22% and 31% respectively [82]. That second figure is a kidney-stone signal as well as a bone one. Participants were not deficient at the start [82].
• Falls and fractures. From Sanders’ trial: falls up 15%, fractures up 26%, with the excess greatest in the three months after each dose [82, 77].
• PTH and turnover. High doses suppress parathyroid hormone, and the Calgary authors linked the bone loss to raised resorption markers with suppressed PTH [82, 76]. Wallace advises testing PTH alongside vitamin D, not 25(OH)D alone [82].
• Vitamin A. She argues vitamin D and vitamin A compete for the RXR receptor pathway [82]. She cites a Swedish human study in which vitamin A blunted the calcium response to vitamin D [82, 89]. Cod liver oil supplies vitamins A and D together, in a natural balance of roughly ten parts vitamin A to one part vitamin D [113]. Masterjohn has proposed that the two moderate each other, with vitamin A protecting against vitamin D toxicity [114]. On that view the problem is an imbalance, such as high-dose vitamin D taken alone. This is a hypothesis, not a proven ideal ratio.

It cuts both ways for the retinol argument above: retinol is needed for ceruloplasmin [51], but excess retinol and excess vitamin D each work against the other. Excess retinol is mainly a supplement problem: in resource-rich countries vitamin A toxicity most often results from supplements, and toxicity from food alone is uncommon [115]. It is another case for getting retinol from food.

• Magnesium. Magnesium is needed for the enzymes that both activate and break down vitamin D [82]. In the trial she cites, magnesium raised vitamin D levels in people who were low and lowered them in people who were high [82, 91].
• Iron. One 250,000 IU dose cut hepcidin by 73% within a week in 28 adults, with no change in ferritin [82, 90]. That strengthens the concern above that large boluses lower hepcidin, and so could raise iron absorption, in women already accumulating iron.
• Calcification. Animal studies show arterial damage and calcification with excess vitamin D3 [82]. The Calgary follow-up found no change in arterial calcification after three years of 10,000 IU a day, which she notes is a short time [82, 92].

Wallace is careful to label some of her ideas as speculative, and several of her sources are abstracts, letters, animal studies or observational work [82]. For publication, the strongest bone evidence is the Burt and Sanders trials [76, 77]. The vitamin A, magnesium and hepcidin studies [89–92] are cited here through her blog and should be checked in the originals.

What this implies

For women who are not deficient, the trial evidence argues against high daily doses (4,000 IU and above) and strongly against large intermittent boluses. Where deficiency is real, correcting it gently with modest daily intakes fits the evidence better than loading doses.

Where the evidence is weak (read this before quoting any of it)

A serious alternative view has to be harder on itself than the view it challenges. These are the awkward data:

• Ovariectomised rats did not accumulate iron, but women do. In Liu et al. 2021, ovariectomy alone produced no iron accumulation; iron only worsened bone loss when it was added as ferric ammonium citrate [59]. Rodents do not menstruate, so they cannot model the human "closed tap": the loss of monthly iron excretion at menopause. In women, ferritin rises across the menopause transition (see "The closing tap" above). Iron is also added to the human diet by law. In the UK, non-wholemeal wheat flour must contain at least 1.65 mg of iron per 100 g, rising to 2.10 mg from 13 December 2026, and ferric ammonium citrate, the compound used to load the rats with iron, is one of the permitted forms [118]. Iron oxides are common colourings

in tablets, including widely prescribed drugs such as clopidogrel, lisinopril and gabapentin [119]; US rules allow up to 5 mg of elemental iron a day from a medicine’s colouring [120]. How much of that iron is absorbed has been little studied.

• Not every human cohort agrees. In a Korean national survey (KNHANES 2008–2010), higher ferritin was linked to lower BMD in premenopausal women, but not in postmenopausal women [60]. An earlier study also found osteoporosis patients slightly iron-deficient, with lower serum iron and higher transferrin [8]. But neither study had the full picture. The Korean survey rested on ferritin [60]; the earlier study on serum iron and transferrin [8]. Each marker on its own is easily misread. Ferritin is raised by inflammation and by tissue injury, not only by iron [2, 121]. Serum iron falls in inflammation even when iron stores are preserved, because inflammation raises hepcidin and traps iron inside cells [121, 122]. Inflammation typically lowers transferrin too [122], so the higher transferrin in the earlier study does point towards genuine deficiency, but without inflammatory markers that cannot be confirmed. Ferritin should not be read alone. It needs a full iron panel (serum iron, transferrin or total iron-binding capacity, and transferrin saturation) and inflammatory markers such as CRP alongside it to give a complete picture [121, 122].

• Ferritin on its own tells you very little. A normal or high ferritin cannot distinguish iron load from inflammation or cell damage, because all three raise it [2, 121]. So “high ferritin predicts bone loss” may partly mean “inflammation predicts bone loss”. That does not exonerate iron, but it means ferritin alone cannot show that iron is the cause, or even how much iron is present. A very low ferritin is its most informative reading, suggesting depleted iron stores [122], but even that is not conclusive on its own and should be confirmed with the rest of the iron panel. Every ferritin-based study in this post, including those used here to support the iron argument, carries the same limit. Ferritin needs to be read with a full iron panel and inflammatory markers [121, 122].

• Copper trials for bone are small and mixed. Positive results came with combined minerals [18] or in women with low intake [17]; short trials and observational data were null [17, 27]. In rats, copper deficiency only slightly worsened ovariectomy-induced osteopenia, and restoring copper did not necessarily reverse it [28]. That uncertainty is about bone end points, not about whether copper matters. In humans, iron cannot be recycled into new red blood cells without copper. The copper-dependent ferroxidases, ceruloplasmin and hephaestin, convert iron into the form transferrin can carry, so it can leave the gut and the cells that recycle old red cells and reach the bone marrow [123]. Copper deficiency causes anaemia in adults, and the blood changes reverse fully with copper replacement over 4–12 weeks [124]. In copper-deficient animals, iron fails to reverse the anaemia; treating copper-deficiency anaemia with iron has been likened to treating B12-deficiency anaemia with folate [123].

• ET-1 is not a one-way story. Postmenopausal women can show lower endothelial ET-1 expression than young women [37], and ET-1 directly restrains osteoclasts [40]. The vascular coupling data for bone are from rats [36]. But the ET-1 level is only half of the picture. ET-1 is the most potent constrictor the endothelium makes, and its effect is held in check by nitric oxide (NO), the main dilator [35]. Iron accumulation and low copper both attack that check. Poorly liganded (free) iron turns normal superoxide and hydrogen peroxide into hydroxyl radicals [1], and iron-driven oxidative stress depletes NO [41]. Low copper weakens Cu,Zn-SOD, leaving superoxide free to destroy NO before it can act [38]. So even where ET-1 itself is not raised, losing NO lets its constricting effect go unopposed. On this model the vessel tips towards constriction without any rise in ET-1.

• IDDM is untested in bone. No study has looked for dormant microbes or circulating LPS as a driver of postmenopausal osteoporosis.

• Fluoride at water levels is genuinely disputed. The best individual-level cohort is positive [46]; the largest register study is null [47]. But both rank people mainly by the fluoride in their drinking water, and water is only one source. Toothpaste [98, 99] and tea [48, 98] add to it, and so do some medications: the antifungal voriconazole contains three fluorine atoms, and long-term use has raised plasma fluoride and caused painful periostitis (skeletal fluorosis) that improved when the drug was stopped or reduced [116, 117]. Fluoride is also cumulative: about half of what is absorbed is kept in the skeleton, where it stays for around 20 years [99, 100]. Water level alone is therefore an incomplete measure of a person’s lifetime exposure.

• No fracture trial exists for iron reduction, copper repletion, retinol or boron in postmenopausal women.

What survives this scrutiny is still substantial. Iron does accumulate after menopause in women. Oestrogen does regulate hepcidin and ceruloplasmin. Copper is mechanistically required to move iron and to cross-link collagen. Fluoride at high doses does make bone denser and weaker. And oestrogen does suppress ET-1. The integrated model linking them is a well-grounded hypothesis that deserves trials, not a proven treatment pathway.

References

Links go to the DOI, PubMed or PubMed Central record wherever one exists.

Douglas Kell and colleagues

1. Kell DB. Iron behaving badly: inappropriate iron chelation as a major contributor to the aetiology of vascular and other progressive inflammatory and degenerative diseases.BMC Med Genomics 2009;2:2.

2. Kell DB, Pretorius E. Serum ferritin is an important inflammatory disease marker, as it is mainly a leakage product from damaged cells. Metallomics 2014;6:748–773.

3. Kell DB, Pretorius E. No effects without causes: the Iron Dysregulation and Dormant Microbes hypothesis for chronic, inflammatory diseases. Biol Rev 2018;93:1518–1557.

4. Kell DB, Potgieter M, Pretorius E. Individuality, phenotypic differentiation, dormancy and ‘persistence’ in culturable bacterial systems. F1000Res 2015;4:179.

Iron and postmenopausal bone

5. Li GF, Gao Y, Weinberg ED, Huang X, Xu YJ. Role of iron accumulation in osteoporosis and the underlying mechanisms. Curr Med Sci 2023;43:647–654.
6. Chen B, Li GF, Shen Y, Huang X, Xu YJ. Reducing iron accumulation: a potential approach for the prevention and treatment of postmenopausal osteoporosis. Exp Ther Med 2015;10:7–11.
7. Iron accumulation and its impact on osteoporotic fractures in postmenopausal women. J Zhejiang Univ Sci B 2023;24:301–311.
8. Jian J, Pelle E, Huang X. Iron and menopause: does increased iron affect the health of postmenopausal women?Antioxid Redox Signal 2009;11:2939–2943.
9. Hou Y et al. Estrogen regulates iron homeostasis through governing hepatic hepcidin expression via an estrogen response element. Gene 2012;511:398–403.
10. Yang Q, Jian J, Katz S, Abramson SB, Huang X. 17β-Estradiol inhibits iron hormone hepcidin through an estrogen responsive element half-site. Endocrinology 2012;153:3170–3178.
11. Hepcidin is an endogenous protective factor for osteoporosis by reducing iron levels. J Mol Endocrinol 2018;60:297–306.
12. Aging-related iron accumulation and ERα in postmenopausal atherosclerosis. eLife 2023 (reviewed preprint).
13. The same study, version of record: eLife 80494.
14. Kim BJ, Ahn SH, Bae SJ et al. Iron overload accelerates bone loss in healthy postmenopausal women and middle-aged men: a 3-year retrospective longitudinal study. J Bone Miner Res 2012;27:2279–2290.
15. Commentary on Kim et al.: iron and bone loss in healthy adults.
16. Jiang Z, Qi G, He X et al. Ferroptosis in osteocytes as a target for protection against postmenopausal osteoporosis. Adv Sci 2024;11:2307388.

Copper

17. Baker A, Turley E, Bonham MP et al. No effect of copper supplementation on biochemical markers of bone metabolism in healthy adults. Br J Nutr 1999 (also summarises Eaton-Evans et al.).

18. Strause L, Saltman P, Smith KT, Bracker M, Andon MB. Spinal bone loss in postmenopausal women supplemented with calcium and trace minerals.J Nutr 1994;124:1060–1064.

19. Nielsen FH et al. Reported zinc, but not copper, intakes influence whole-body bone density, mineral content and T score responses to zinc and copper supplementation in healthy postmenopausal women.Br J Nutr 2011;106:1872–1879.

20. Prohaska JR. Impact of copper limitation on expression and function of multicopper oxidases (ferroxidases).Adv Nutr 2011.

21. Severe iron metabolism defects in mice with double knockout of the multicopper ferroxidases hephaestin and ceruloplasmin.Cell Mol Gastroenterol Hepatol 2018.

22. Identification of zyklopen, a multicopper ferroxidase (includes ceruloplasmin-knockout tissue iron findings).

23. Gibson RS. Principles of Nutritional Assessment: serum ceruloplasmin.

24. Chilvers DC, Jones MM, Selby PL et al. Effects of oral ethinyl oestradiol and norethisterone on plasma copper and zinc complexes in post-menopausal women.Horm Metab Res 1985;17:532–535.

25. Klevay LM. Copper nutriture and ischemic heart disease: a brief review.Med Res Arch 2022.

26. Klevay LM. Cardiovascular disease from copper deficiency - a history.J Nutr 2000;130:489S–492S.

27. Canal-Macías ML et al. The role of copper intake in bone health: a quantitative analysis in postmenopausal Spanish women. 2025.

28. Yee CD, Kubena KS, Walker M et al. The relationship of nutritional copper to the development of postmenopausal osteoporosis in rats.Biol Trace Elem Res 1995;48:1–11.

29. Copper (Advances in Nutrition update).

30. Medeiros DM. Copper, iron, and selenium dietary deficiencies negatively impact skeletal integrity.Exp Biol Med 2016.

Endothelin-1 and the vasculature

31. Ylikorkala O, Orpana A, Puolakka J, Pyörälä T, Viinikka L. Postmenopausal hormonal replacement decreases plasma levels of endothelin-1.J Clin Endocrinol Metab 1995;80:3384–3387.

32. Best PJM et al. The effect of estrogen replacement therapy on plasma nitric oxide and endothelin-1 levels in postmenopausal women.Ann Intern Med 1998;128:285–288.

33. Birkhaeuser MH et al. Tibolone and HRT reduce endothelin levels.Gynecol Endocrinol 1999.

34. Somani YB, Pawelczyk JA, De Souza MJ, Kris-Etherton PM, Proctor DN. Aging women and their endothelium: probing the relative role of estrogen on vasodilator function.Am J Physiol Heart Circ Physiol 2019.

35. Kuczmarski AV, Welti LM, Moreau KL, Wenner MM. ET-1 as a sex-specific mechanism impacting age-related changes in vascular function.Front Aging 2021; and NIH R01 HL146558.

36. Prisby RD, Dominguez JM II, Muller-Delp J, Allen MR, Delp MD. Aging and estrogen status: a possible endothelium-dependent vascular coupling mechanism in bone remodeling.PLoS One 2012;7:e48564.

37. Role of the endothelin system in vascular dysfunction in women. University of Delaware dissertation, 2021.

38. Schuschke DA, Saari JT, Miller FN. A role for dietary copper in nitric oxide-mediated vasodilation.Microcirculation 1995;2:371–376.

39. USDA ARS. Dietary copper requirements for optimal cardiovascular function and health (research project report).

40. Alam ASMT et al. Endothelin inhibits osteoclastic bone resorption by a direct effect on cell motility.Endocrinology 1992;130:3617–3624.

41. Młodziński K, Świączak M, Rohun J et al. Vascular aging and damage in patients with iron metabolism disorders.Diagnostics 2022;12:2817.

Fluoride

42. Riggs BL, Hodgson SF, O’Fallon WM et al. Effect of fluoride treatment on the fracture rate in postmenopausal women with osteoporosis.N Engl J Med 1990;322:802–809.

43. National Research Council. Health Effects of Ingested Fluoride: fluoride exposure and risk of bone fracture. 1993.

44. Haguenauer D, Shea B, Tugwell P, Wells GA, Welch V. Fluoride for treating postmenopausal osteoporosis.Cochrane Database Syst Rev 2000;CD002825.

45. Riggs BL, O’Fallon WM, Lane A et al. Clinical trial of fluoride therapy in postmenopausal osteoporotic women: extended observations and additional analysis.J Bone Miner Res 1994;9.

46. Helte E, Vargas CD, Kippler M, Wolk A, Michaëlsson K, Åkesson A. Fluoride in drinking water, diet, and urine in relation to bone mineral density and fracture incidence in postmenopausal women.Environ Health Perspect 2021;129:47005. See also the EHP commentary.

47. Näsman P, Ekstrand J, Granath F, Ekbom A, Fored CM. Estimated drinking water fluoride exposure and risk of hip fracture: a cohort study.J Dent Res 2013;92:1029–1034.

48. Cook FJ, Seagrove-Guffey M, Mumm S et al. Non-endemic skeletal fluorosis: causes and associated secondary hyperparathyroidism. Bone 2021;145:115839.

49. Skeletal fluorosis in humans: a review of recent progress in the understanding of the disease. Prog Food Nutr Sci 1986;10:279–314.

What may help

50. Pizzorno L. Nothing Boring About Boron. Integr Med (Encinitas) 2015;14(4):35–48.

51. Barber EF, Cousins RJ. Induction of ceruloplasmin synthesis by retinoic acid in rats: influence of dietary copper and vitamin A status. J Nutr 1987;117:1615–1622.

52. Sundaresan PR et al. Interactions in indices of vitamin A, zinc and copper status when these nutrients are fed to rats at adequate and increased levels. Br J Nutr 1996;75:915–928.

53. Comptour A, Rouzaire M, Belville C et al. Lysyl oxidase-like 4 involvement in retinoic acid epithelial wound healing. Sci Rep 2016;6:32688.

54. Smith-Mungo LI, Kagan HM. Lysyl oxidase: properties, regulation and multiple functions in biology. Matrix Biol 1998;16:387–398.

55. Regulation of the expression of lysyl oxidase mRNA in cultured rabbit retinal pigment epithelium cells.

56. Feskanich D, Singh V, Willett WC, Colditz GA. Vitamin A intake and hip fractures among postmenopausal women. JAMA 2002;287:47–54.

57. Iron and zinc status in rats with diet-induced marginal deficiency of vitamin A and/or copper. Biol Trace Elem Res 1993;38:83–95.

58. Rucker RB et al. Modulation of lysyl oxidase by dietary copper in rats. J Nutr 1996;126:51–60.

Contrary evidence

59. Liu LL, Liu GW, Liu H, Zhao K, Xu YJ. Iron accumulation deteriorated bone loss in estrogen-deficient rats. J Orthop Surg Res 2021;16:525.

60. Chon SJ, Choi YR, Roh YH et al. Association between levels of serum ferritin and bone mineral density in Korean premenopausal and postmenopausal women: KNHANES 2008–2010. PLoS One 2014;9:e114972.

Retinol, ceruloplasmin and boron

61. Hellman NE, Kono S, Mancini GM et al. Mechanisms of copper incorporation into human ceruloplasmin. J Biol Chem 2002.

62. Song D et al. Retinoids rescue ceruloplasmin secretion and alleviate oxidative stress in Wilson's disease-specific hepatocytes. Hum Mol Genet 2022.

63. Nielsen FH, Hunt CD, Mullen LM, Hunt JR. Effect of dietary boron on mineral, estrogen, and testosterone metabolism in postmenopausal women. FASEB J 1987;1:394–397.

64. Nielsen FH. Biochemical and physiologic consequences of boron deprivation in humans. Environ Health Perspect 1994;102 Suppl 7:59–63.

65. Beattie JH, Peace HS. The influence of a low-boron diet and boron supplementation on bone, major mineral and sex steroid metabolism in postmenopausal women. Br J Nutr 1993;69:871–884.

66. Meacham SL, Taper LJ, Volpe SL. Effects of boron supplementation on bone mineral density and dietary, blood, and urinary calcium, phosphorus, magnesium, and boron in female athletes. Environ Health Perspect 1994.

67. Rondanelli M et al. Pivotal role of boron supplementation on bone health: a narrative review. J Trace Elem Med Biol 2020;62:126577.

68. Nielsen FH, Eckhert CD. Boron. Adv Nutr 2020;11:461–462.

69. Elsair J et al. Boron as a preventive antidote in acute and subacute fluoride intoxication in rabbits. Fluoride 1980;13:129–138; and Franke J et al. Boron as an antidote to fluorosis? Fluoride 1985;18:187–197. Not read directly; cited from this reference list.

70. Xu P, Guo X, Cao H et al. Preventive effect of boron on damage to cartilage of rats with intake excessive fluoride. Acad J Xi’an Jiaotong Univ 2005;17:36–39.

71. Tabriz University of Medical Sciences. Efficacy of OsteoBor in postmenopausal osteoporosis (phase 2 trial registration).

72. Hunt CD et al. Metabolic responses of postmenopausal women to supplemental dietary boron and aluminum during usual and low magnesium intake. Am J Clin Nutr 1997;65:803–813.

73. Meacham S, Karakas S, Wallace A, Altun F. Boron in human health: evidence for dietary recommendations and public policies. Open Mineral Processing Journal 2010;3:36–53.

74. Gowda NKS, Rajendran D, Krishnamoorthy P et al. Boron and calcium chloride as possible ameliorators of fluoride toxicity in Wistar rats. Indian J Exp Biol 2017;55:864–869.

75. Nielsen FH. New essential trace elements for the life sciences. Biol Trace Elem Res.

High-dose vitamin D

76. Burt LA, Billington EO, Rose MS, Raymond DA, Hanley DA, Boyd SK. Effect of high-dose vitamin D supplementation on volumetric bone density and bone strength: a randomized clinical trial. JAMA 2019;322:736–745.

77. Sanders KM, Stuart AL, Williamson EJ et al. Annual high-dose oral vitamin D and falls and fractures in older women: a randomized controlled trial. JAMA 2010;303:1815–1822.

78. Smith H, Anderson F, Raphael H, Maslin P, Crozier S, Cooper C. Effect of annual intramuscular vitamin D on fracture risk in elderly men and women. Rheumatology 2007;46:1852–1857.

79. LeBoff MS, Chou SH, Ratliff KA et al. Supplemental vitamin D and incident fractures in midlife and older adults. N Engl J Med 2022;387:299–309.

80. Bacchetta J, Zaritsky JJ, Sea JL et al. Suppression of iron-regulatory hepcidin by vitamin D. J Am Soc Nephrol 2014;25:564–572.

81. Lamikanra AA et al. Effect of supplementation with vitamin D on biochemical markers of iron status and erythropoiesis in older people: BEST-D trial. Br J Nutr 2025.

82. Wallace L. The case for caution: rethinking synthetic vitamin D. Clinic blog, 3 November 2026.

Beta-carotene conversion

83. Leung WC, Hessel S, Méplan C et al. Two common single nucleotide polymorphisms in the gene encoding β-carotene 15,15′-monoxygenase alter β-carotene metabolism in female volunteers. FASEB J 2009;23:1041–1053.

84. Lietz G, Oxley A, Leung W, Hesketh J. Single nucleotide polymorphisms upstream from the β-carotene 15,15′-monoxygenase gene influence provitamin A conversion efficiency in female volunteers. J Nutr 2012;142:161S–165S.

85. Effect of SNPs in the BCMO1 enzyme. ClinicalTrials.gov NCT02276014 (trial registration and background).

Cholecalciferol toxicology and units

86. Peterson ME, Fluegeman K. Cholecalciferol. Top Companion Anim Med 2013;28:24–27.

87. Tauer D (reviewed by Brutlag A). Cholecalciferol (vitamin D3) poisoning in animals. Merck Veterinary Manual, reviewed November 2025.

88. Institute of Medicine. Dietary Reference Intakes for Calcium and Vitamin D, chapter 3. National Academies Press, 2011. doi:10.17226/13050.

Cited via Wallace’s blog (not read directly; check before publishing)

89. Johansson S, Melhus H. Vitamin A antagonizes calcium response to vitamin D in man. J Bone Miner Res 2001;16(10):1899–1905.

90. Smith EM et al. High-dose vitamin D3 reduces circulating hepcidin concentrations. Clin Nutr 2017;36(4):980–985.

91. Dai Q et al. Magnesium status and supplementation influence vitamin D status and metabolism. Am J Clin Nutr 2018;108(6):1249–1258.

92. Calgary Vitamin D Study arterial calcification follow-up. ClinicalTrials.gov NCT01900860.

Vitamin C

93. Levine M, Conry-Cantilena C, Wang Y et al. Vitamin C pharmacokinetics in healthy volunteers: evidence for a recommended dietary allowance. Proc Natl Acad Sci USA 1996;93:3704–3709.

94. Deferoxamine. StatPearls, updated May 2023.

95. Brigham and Women's Hospital. Iron overload and hemochromatosis (cites Nienhuis 1981 on sudden deaths).

96. Elalfy MS, Saber MM, Adly AA et al. Role of vitamin C as an adjuvant therapy to different iron chelators in young β-thalassemia major patients. Eur J Haematol 2016;96:318–326.

Toothpaste and bone fluoride

97. Fluoride intake during and after toothbrushing with toothpastes above 1,500 ppm. ISRCTN79212143 (trial registration).

98. Skeletal fluorosis due to excessive tea and toothpaste consumption. Osteoporos Int 2011;22:2557–2560.

99. Institute of Medicine. Dietary Reference Intakes: The Essential Guide to Nutrient Requirements. Fluoride. National Academies Press, 2006.

100. National Research Council. Fluoride in Drinking Water: A Scientific Review of EPA's Standards, chapters 3 and 5. National Academies Press, 2006.

101. Estimating skeletal fluoride accumulation over years of intake. Xenobiotica 1985.

102. Fluoride distribution profile in human rib bone by sex and age. J Dent Health 1991;41:607–616.

103. Yang S, Liu C, Ding Q, Yang H, Jia Y. Comparative study on fluoride accumulation in hard tissues of rats with chronic drinking-water-borne fluorosis. J Environ Occup Med 2022.

104. The effect of fluoridation and its discontinuation on fluoride profiles in the alveolar bone of rat. 2005.

105. Recovery from skeletal fluorosis (an enigmatic, American case). J Bone Miner Res 2007;22:163–170.

Wilson’s disease

106. Mohr I, Weiss KH. Current anti-copper therapies in management of Wilson disease. Ann Transl Med 2019;7(Suppl 2):S69.

Boron upper limits

107. NIH Office of Dietary Supplements. Boron: Fact Sheet for Health Professionals.

108. European authority sets safety levels for boron. FoodNavigator, 24 August 2004.

109. Institute of Medicine. Arsenic, Boron, Nickel, Silicon, and Vanadium. In: Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. National Academies Press, 2001. doi:10.17226/10026.

110. German Federal Institute for Risk Assessment (BfR). Proposed maximum levels for the addition of boron to foods including food supplements.

111. UK Committee on Toxicity. Deriving a health-based guidance value for boron to support development of UK Drinking Water Standards: evaluations by other authoritative bodies.

112. Duydu Y, Basaran N. Effects of boron exposure on human reproduction and development. Curr Opin Toxicol 2023;34:100403.

Cod liver oil: vitamins A and D

113. American Institute for Cancer Research. Is there any risk to taking cod liver oil?

114. Masterjohn C. Vitamin D toxicity redefined: vitamin K and the molecular mechanism. Med Hypotheses 2007;68(5):1026–1034. doi:10.1016/j.mehy.2006.09.051.

115. Vitamin A toxicity. StatPearls. StatPearls Publishing.

Fluoride from medication

116. Wermers RA, Cooper K, Razonable RR, et al. Fluoride excess and periostitis in transplant patients receiving long-term voriconazole therapy. Clin Infect Dis 2011;52(5):604–611.

117. Moon WJ, Scheller EL, Suneja A, et al. Plasma fluoride level as a predictor of voriconazole-induced periostitis in patients with skeletal pain. Clin Infect Dis 2014;59(9):1237–1245.

Iron in food and medicines

118. GOV.UK. Bread and flour: labelling and composition.

119. Drugs.com. Ferric oxide red (excipient).

120. US Code of Federal Regulations. 21 CFR §73.1200 Synthetic iron oxide.

Reading iron markers

121. Ganz T. Anemia of inflammation. N Engl J Med 2019;381:1148–1157.

122. Weiss G, Ganz T, Goodnough LT. Anemia of inflammation. Blood 2019;133(1):40–50.

Copper and red-cell iron recycling

123. Collins JF, Prohaska JR, Knutson MD. Metabolic crossroads of iron and copper. Nutr Rev 2010;68(3):133–147.

124. Myint ZW, Oo TH, Thein KZ, Tun AM, Saeed H. Copper deficiency anemia: review article. Ann Hematol 2018;97(9):1527–1534.