Strontium for Bone Mineral Density

Strontium ranelate, the prescription form tested at 2 g a day, raised measured bone density on DXA scans: in the 1,649-woman SOTI trial, density at month 36 was up 14.4% at the lumbar spine and 8.3% at the femoral neck (P < 0.001 for both comparisons) (Meunier 2004). That is a large rise compared with alternatives such as bisphosphonates (Blake 2007). But strontium absorbs X-rays more strongly than calcium, so a scan reads higher whenever strontium has taken calcium's place in bone. The drug's EU product information says this accounts for approximately 50% of the measured change in density over 3 years (EMA product information), and a modelling study judged 75% or more of the spine rise the most likely figure (Blake 2007).

For the strontium citrate and chloride sold as dietary supplements, for example in the United States, there is no controlled trial of strontium on its own (Westberg 2016). The only randomised trial we found tested strontium citrate inside a nightly four-ingredient combination, so it can't show what strontium contributed (Maria 2017). The prescription drug was restricted in the EU in 2014 over heart and clot risks, and its authorisation was withdrawn on 14 April 2020 at the manufacturer's request for commercial reasons; it had not been marketed in the EU since 2017 (EMA 2014; EMA Protelos page).

In treated people, strontium is mainly adsorbed onto the surface of bone crystals and only slightly replaces calcium in the crystal of newly formed bone. In iliac crest biopsies taken after up to 60 months of 2 g a day, the label reports no harmful effects on bone quality or mineralisation (EMA product information).

DXA estimates density from how much of an X-ray beam the bone absorbs, and strontium atoms absorb more than calcium atoms. In laboratory mixtures, each 1 mol% of calcium replaced by strontium made DXA read about 10% too high, with a straight-line relationship (r² above 0.99) (Nielsen 1999). Later work found the factor depends on the machine: on phantoms, 11.2 for a Hologic Discovery, 9.9 for a GE-Lunar Prodigy, 8.6 for a Demetech Calscan and 6.3 for an Osteometer Dexacare G4 (Liao 2010).

Estimates differ, and so does the evidence they rest on. STRATOS, a 353-woman trial, adjusted lumbar density for bone strontium content and still found a dose-dependent gain: the annual slope rose from 1.4% with 0.5 g a day to 3.0% with 2 g a day, above placebo (P < 0.01) (Meunier 2002). Pooling trials, the Cochrane review found a lumbar spine difference from placebo of 5.44% (95% CI 3.41 to 7.46) adjusted for strontium content and 11.29% (10.22 to 12.37) unadjusted over two years (O'Donnell 2006a). In PREVOS, the 40 early postmenopausal women taking 1 g a day for 24 months gained 5.53% in lumbar density from baseline as measured and 1.41% adjusted; the adjusted difference from placebo was about 2.4% (Reginster 2002).

A modelling study put the share of the spine rise explained by bone strontium at 75% or more as the most likely figure, with values as low as 40% possible, and warned of large uncertainties because strontium at the scan site has to be inferred from iliac crest biopsies (Blake 2007). So a real gain probably exists, but its size is uncertain. And the mechanism says nothing about fractures: whether a higher reading means stronger bone has to come from trials, not from the scan itself.

The trials used strontium ranelate as 2 g once a day, stirred into water. The label said to take it at bedtime, at least two hours after eating, because food, milk and calcium-containing medicines cut absorption by about 60-70%, and to add calcium and vitamin D only if dietary intake was inadequate (EMA product information). In STRATOS, 2 g a day was judged to give the best balance of efficacy and safety among 0.5, 1 and 2 g (Meunier 2002); in PREVOS, 1 g a day was the minimum dose that worked in early postmenopausal women without osteoporosis, who all took calcium 500 mg a day (Reginster 2002).

There is no tested dose for supplements. The two published reports that give one involve single patients: a woman who took 680 mg a day of strontium chloride for 2.5 years and showed rising scan readings (Westberg 2016), and a woman who took 680 mg of strontium a day as citrate and whose bone strontium kept rising for at least 800 days (Moise 2012). Neither measured fractures. In ten women who chose their own supplements, bone strontium varied widely after three years (finger signal from 3.15 to 28.15 on the study's scale) and had not plateaued after four years of use (Moise 2014).

For postmenopausal women with osteoporosis, the American College of Physicians recommends bisphosphonates for initial treatment (strong recommendation, high-certainty evidence), denosumab as second line for women who can't take them, and romosozumab or teriparatide, followed by a bisphosphonate, only for women at very high fracture risk (Qaseem 2023). For women over 65 with low bone mass (osteopenia), it suggests an individualised decision about starting a bisphosphonate. Other supplements have their own bone density pages: vitamin D, calcium, vitamin K2 and collagen.

SOTI randomised 1,649 postmenopausal women with osteoporosis and at least one vertebral fracture to strontium ranelate 2 g a day or placebo for three years, with calcium and vitamin D given to both groups. Density at month 36 was up 14.4% at the lumbar spine and 8.3% at the femoral neck (P < 0.001 for both comparisons), and new vertebral fractures fell by 41% (relative risk 0.59, 95% CI 0.48 to 0.73) (Meunier 2004). TROPOS randomised 5,091 women; density at 3 years was up 8.2% at the femoral neck and 9.8% at the total hip (P < 0.001), and non-vertebral fractures fell by 16% (P = 0.04) (Reginster 2005).

The first edition of the Cochrane review found density up at all sites after two to three years, with the highest dose giving the greatest rise. At 3 years, two pooled trials gave a difference from placebo of 8.25% (95% CI 7.84 to 8.66) at the femoral neck and 9.83% (9.39 to 10.26) at the total hip, and the review rated the evidence "silver level" (O'Donnell 2006a). An update later in 2006, covering four trials, reported with 2 g a day over three years a 37% fall in vertebral fractures (relative risk 0.63, 95% CI 0.56 to 0.71) and a 14% fall in non-vertebral fractures (0.86, 0.75 to 0.98) (O'Donnell 2006b).

STRATOS and PREVOS found the gains rose with dose (Meunier 2002; Reginster 2002). In MALEO, 261 men with osteoporosis (mean age 72.9) took 2 g a day or placebo for 2 years; lumbar spine density rose 9.7% with strontium ranelate and 2.0% with placebo, a difference of 7.7% (95% CI 5.9% to 9.5%, P < .001) (Kaufman 2013).

These trials mostly report measured density. The abstracts of MALEO and the phase III trials don't say whether their headline figures were adjusted for bone strontium, so some of the measured rise is likely to be scan effect, to an unknown degree.

SOTI and TROPOS were 5-year double-blind trials, with the main analysis at 3 years; women could then join open-label extensions. In the 3-year extension, 879 women who took strontium ranelate for 8 years had density rising at the lumbar spine, femoral neck and total hip throughout (Reginster 2009). In a 10-year extension of 237 women, lumbar density rose 34.5% (SD 20.2%) from baseline, with fractures in years 6 to 10 lower than in a statistically matched placebo group over 5 years (Reginster 2012).

Both extensions were open-label and included only women who stayed in them, and the abstracts don't say whether the density figures were adjusted for strontium. They show the drug was continued and density kept rising, not how much of the rise is bone.

In a post hoc analysis of the strontium arms of SOTI and TROPOS, three-year changes in femoral neck and total proximal femur density explained 76% and 74% of the reduction in vertebral fractures, with each percentage point of femoral neck density linked to a 3% lower risk (95% CI 1% to 5%). Spine density change, the site where the scan effect is largest (Blake 2006), was not statistically associated with new vertebral fractures (P = 0.10) (Bruyère 2007). The abstract doesn't test whether the scan effect explains that difference, but it is a reason to look at the hip rather than the spine number.

More broadly, a pooled analysis of 23 placebo-controlled trials (91,779 participants) found hip density changes explained 44% to 67% of the fracture reduction for treatments such as bisphosphonates, denosumab, PTH-receptor agonists and hormone therapy (Black 2020). Strontium wasn't among the treatments listed in that analysis's abstract, so it doesn't vouch for strontium scans.

The only randomised trial we found that included a strontium supplement was MOTS: a one-year, double-blind trial in postmenopausal women aged 49 to 75 with osteopenia, testing nightly melatonin, strontium citrate, vitamin D3 and vitamin K2 (MK-7) together against placebo. Density rose 4.3% at the lumbar spine and 2.2% at the left femoral neck, with a trend at the total left hip (P = 0.069) (Maria 2017). The abstract doesn't give the strontium dose or group sizes, doesn't say whether density was adjusted for strontium, and can't separate strontium from the three other ingredients. The other human reports are single-patient or tiny: one woman whose scan readings rose on strontium chloride (Westberg 2016), and bone-strontium measurements showing that supplement strontium accumulates in bone (Moise 2012; Moise 2014). Because supplement strontium accumulates in bone, a scan effect is plausible, but none of these reports measured its size or adjusted for it. A rat study found strontium citrate raised bone strontium more than strontium ranelate until the doses were adjusted for, after which the groups did not differ significantly (Wohl 2013).

In the Cochrane review, 2 g a day increased diarrhoea (relative risk 1.38, 95% CI 1.02 to 1.87) without raising withdrawals, and the pooled data showed venous thromboembolism at 2.2% versus 1.5% on placebo (odds ratio 1.5, 95% CI 1.1 to 2.1) (O'Donnell 2006a). The EU review identified serious cardiac disorders (including heart attack), thromboembolic events, serious skin reactions (DRESS, Stevens-Johnson syndrome and toxic epidermal necrolysis), disturbances in consciousness, seizures, hepatitis and blood cell disorders. In the pooled trial population of 7,572 women (3,803 treated), the odds ratio for heart attack was 1.60 (95% CI 1.07 to 2.38, P = 0.020), with about 2 additional heart attacks and 2 additional venous clots per 1,000 patient-years, though cardiovascular and overall mortality were not increased (EMA 2014).

Studies of routine care point both ways. A Danish cohort found no excess acute coronary syndrome against bisphosphonates (adjusted HR 0.89, 95% CI 0.52 to 1.55) (Svanström 2014). A five-country study in people without contraindications found no excess heart attacks. Against bisphosphonates, current strontium ranelate use carried a non-significant 24% higher odds of venous clots (1.24, 0.96 to 1.61) and 35% higher odds of cardiovascular death (1.35, 1.02 to 1.80); against past strontium use, clot odds were 1.30 (1.04 to 1.62) but cardiovascular death odds were lower (0.68, 0.48 to 0.96), which the authors say may reflect treatment stopping near the end of life or residual confounding (Ali 2020). No one has measured these risks for supplement forms, and that absence is not reassurance.

These interactions come from the label for the drug. Strontium supplements haven't been tested, but the chemistry the label describes, divalent cations competing for absorption and binding antibiotics, would be expected to apply to them too (EMA product information). The scan effect is a different kind of interaction: it changes how a result is read, not how a medicine works.