This is a 6-month single-center 3-arm double-blind, randomized, parallel, placebo-controlled trial that compared the effect of 20 µg/day of calcifediol with 3200 IU/day of vitamin D3 and with placebo with regards to lower extremity function.

The trial was conducted at the Centre on Aging and Mobility, Waid City Hospital, Zurich, Switzerland. The trial was approved by the Cantonal Ethics Commission of Zurich (KEK-ZH 2013–0582, PB_2016-01877) and Swissmedic (2015DR2011). The trial was closely monitored by an external company (Clinipace) on behalf of the sponsor. The trial protocol and statistical analysis plan are available in the Supplementary material. The trial’s registration number at clinicaltrials.gov is NCT02527668.

Participants signed the informed consent and were postmenopausal women between the ages of 50 and 70 years with documented osteopenia (BMD by DXA T-score: <  − 1.0 and >  − 2.5 at the spine or total hip) or osteoporosis (BMD by DXA T-score: ≤  − 2.5 at the spine or total hip) and a FRAX score that did not qualify for pharmacologic treatment for osteoporosis [16, 17]. Participants were required to be community-dwelling and ambulatory without help, have serum 25(OH)D concentration < 30 ng/mL, and limit additional vitamin D3 intake to 800 IU/day and calcium intake to 500 mg/day during the trial.

Exclusion criteria were the following: vitamin D intake > 1000 IU on any day in the 4 weeks prior to enrollment, serum calcium levels > 2.60 mmol/L adjusted for albumin if albumin ≤ 35 or ≥ 45 g/L, estimated creatinine clearance < 30 mL/min, severe visual or hearing impairment, malabsorption syndrome (celiac diseases, inflammatory bowel disease), pathologic fracture (excl. fractures due to osteoporosis and stress fractures) in the last year prior enrollment, fracture due to osteoporosis in the last 10 years prior enrollment, chemotherapy or radiation due to cancer in the last year prior enrollment, treatment which has an effect on calcium metabolism (e.g., PTH, calcitonin, chronic cortisone intake > 5 mg/day for more than 4 weeks in the last year prior enrollment (except for inhalation and sporadic infiltration), current treatment with a bisphosphonate, hormone replacement therapy, and/or selective estrogen receptor modulator.

The investigational product was produced by dsm-firmenich AG, and capsules were produced and filled into bottles by Fisher Clinical Services. The same batch of product was used for all study capsules. Study capsules were blinded and labeled with a unique randomization code by the capsule producer. Capsules were analyzed by dsm-firmenich AG and analysis certificates were approved by Swissmedic. Assignment was determined from a computer-generated randomization list produced by Ferrari Data Solutions. Study personnel remained blinded to the treatment assigned to subjects throughout the study.

At the time of trial preparation, the effectiveness and safety of a daily dose of 20 µg of calcifediol in improving lower extremity function were established in our pilot study among 20 postmenopausal women (mean age 61.5 years, mean serum 25[OH]D level 13.2 ng/mL) [15]. In this pilot trial, women receiving calcifediol had 2.8 times higher odds of maintaining or improving lower extremity function compared to those taking a daily dose of 800 IU of vitamin D3.

The daily dose of 3200 IU (80 µg) of vitamin D3 was chosen as the active comparator to calcifediol based on our pharmacokinetic study, which showed calcifediol to be three times more potent than vitamin D3 [18]. This dose also aligned with the 2010 Institute of Medicine (IOM) recommendation of a safer upper limit of 4000 IU/day (100 µg/day, 1 IU = 0.025 µg) [19]. Furthermore, a safe upper intake of 10,000 IU/day (250 µg/day) of vitamin D3 was supported by a risk assessment study involving 25 RCTs, which found no relationship between oral vitamin D (up to 100,000 IU/day) or serum 25(OH)D levels (up to 257.2 ng/mL or 643 nmol/L) and mean serum calcium levels [20]. Based on these studies, we selected 3200 IU/day of vitamin D3 as the active comparator, which should raise serum 25(OH)D levels similarly to 20 µg/day of calcifediol.

Treatment allocation was stratified by (1) DXA T-score at the spine and/or total hip (DXA BMD T-score ≤  − 2.5 at spine and/or hip was stratified as osteoporosis, all others as osteopenia) and (2) participant agreement to have a muscle biopsy performed, in block sizes 3 and 6, respectively.

Participants were assigned in a 1:1:1 ratio to receive either daily 20 µg calcifediol, 3200 IU (80 µg) vitamin D3, or placebo in a blinded fashion, and were advised to take 1 capsule per day together with a meal. Participants were followed up for 6 months, and clinical visits occurred at baseline, 3, and 6 months. At follow-up visits, participants were assessed for trial outcome measures, adverse events, and compliance with the study medication and had laboratory tests performed. Study intervention ended at month 6, and a phone call at 7 months was conducted to ascertain any adverse events.

We assessed compliance with the treatment medication by pill count at each study visit, at 3 and 6 months. Treatment compliance was defined as taking at least 80% of the study medication and calculated using two different approaches: (1) The conservative approach, which considered failure to bring the pill bottle to the visit as having taken none of the study medication contained in the not returned bottle, and (2) the non-conservative approach, which considered failure to bring the pill bottle to the visit as having taken all the study medication contained in the not returned bottle.

Serum 25(OH)D concentrations were assessed at baseline and at 3 and 6 months follow-up. Serum calcium, serum creatinine, serum albumin, and urine calcium/creatinine ratio in spot urine were assessed in addition to screening and during subsequent visits. Adverse events were monitored by interviews at 3, 6, and 7 months.

DSM laboratories measured serum 25(OH)D by means of a sensitive and selective assay based on liquid chromatography coupled to tandem mass spectrometry detection (HPLC–MS/MS). The samples were centrifuged and stored at − 80 °C at the Centre on Aging and Mobility before and shipped to DSM laboratories for analysis at the end of the trial.

The primary endpoint was a composite measure of lower extremity function, assessed at baseline, 3, and 6 months. This composite measure, predefined in the study protocol, included four tests: gait speed (over a distance of 10 m), knee flexor and extensor strength, and repeated sit-to-stand test. The primary endpoint was defined as the probability of success (improvement or maintenance from baseline) in any of the eight tests, four tests at 3 months and four tests at 6 months. Therefore, the primary endpoint was binary and took on a value of 1 if the participant improved or maintained on a test component, and a value of 0 if declined in one test.

For gait speed, women were asked to walk 10 m at their usual pace. The test was repeated twice, and the faster of the two measurements was used. Women were allowed to use a cane or other walking aid if needed. Knee flexor and extensor strength were measured in a seated position, with hips and knees at a 90° angle and the feet approximately 30 cm above the floor. The foot was placed in a foot strap, which was connected to a wall-mounted dynamometer through a chain. Arms were folded across the chest. For the knee flexor strength test, participants were asked to bend their knees as much as possible and pull as hard as possible against the resistance of the fixture. Knee extensor strength was measured using the same setting, but with the participant asked to straighten the leg and push against the resistance of the fixture. The test was repeated twice for each leg, with a 30-s break between attempts. Knee flexor and extensor strength were assessed as the mean strength of the right and left knees, determined from the maximum of two repeated strength tests on each knee. The repeated sit-to-stand test was measured as the time to complete five repetitions from an initial seated position and was completed when the participants reached the standing position of the fifth repetition. Women were asked to stand up from a chair five times as quickly as they could without stopping in between. Arms had to stay folded across their chest, and they had to sit with their back against the back of the chair.

Secondary outcomes included the individual components of the composite lower extremity battery and TUG at a normal and fast pace. TUG was measured by asking women to stand up from a chair, walk 3 m, walk around a mark on the floor, return to the chair, and sit down again. Women were asked not to use armrests (unless needed) while standing up and sitting down and were not allowed to be assisted by another person. Additionally, women were allowed to stop during the test but could not sit down. Women were given a trial run without timing prior to the actual measurements. Two measures were conducted: (1) women walking at their normal pace and (2) women walking as fast as they could.

Other assessments included cognitive function, assessed using the Montreal Cognitive Assessment (MoCA) [21]; quality of life and self-rated health, measured with the EuroQoL 5 Dimensions 3 Levels instrument [22]. Bone mineral density was measured using Dual-energy X-ray Absorptiometry (DXA; Lunar iDXA, GE Healthcare machines) and analyzed using enCORE software (Version 13.60.033).

Safety endpoints included serum levels of parathyroid hormone (PTH), calcium, creatinine, albumin, urine calcium/creatinine ratio, blood pressure, pulse rate, adverse events (AEs), and serious adverse events (SAEs). Except for AEs and SAEs, safety endpoints were assessed at baseline, 3, and 6 months.

AEs were assessed for causality related to the study intervention and categorized as “not related” when there is no reasonable possibility that the study intervention may have caused or contributed to the occurrence of the adverse event and “related” when there was a reasonable possibility that the intervention may have caused or contributed to the occurrence of the adverse event.

Based on our pilot study [15], we anticipated a 5% dropout rate and estimated that 40% of participants in the placebo group would maintain or improve their lower extremity function. With a sample size of 50 participants per group, the trial had 87% power to detect an odds ratio of at least 2.79 at a significance level of 0.05. Regarding the secondary endpoints, and also based on our pilot study [15], the study had more than 85% power to detect a mean difference between calcifediol and placebo groups of 42 Newtons (SD 38.3) in knee extensors strength and of 1.0 s (SD 1.6) in the repeated sit-to-stand test.

All assessments of treatment effects were based on the intent-to-treat principle. For the primary endpoint, a generalized estimating equations (GEE) model was used to compare the probability of success (improvement or maintenance) in any of the eight tests, four tests at 3 months and four tests at 6 months in the calcifediol group vs the vitamin D3 and placebo groups over 6 months. Changes over time in the secondary outcomes (Δ from baseline) were also compared between treatment groups using GEE.

Models were adjusted for age, BMI, bone status (osteopenia vs osteoporosis; stratification variable), baseline serum 25(OH)D levels, and time. For the secondary outcomes, models were additionally adjusted for the baseline outcome measure. An interaction between time and treatment group was tested. The main effects across all 6 months were derived from models without the interaction term for time, as interactions were not statistically significant (Supplementary file 1).

Additionally, the association between quartiles of achieved 25(OH)D levels and probability of success over 6 months was investigated in a post hoc analysis. The exposure of quartiles of achieved 25(OH)D levels was calculated based on the mean achieved 25(OH)D levels at months 3 and 6. A GEE model was used to assess the association of quartiles of achieved 25(OH)D levels and the probability of success adjusting for the same covariates as in the main analysis except for the treatment effects of calcifediol and vitamin D3 as both may be on the causal pathway. The interaction between the quartile of achieved 25(OH)D levels and the treatment group was tested by adding the interaction term in the model.

Statistical analyses were performed using SAS version 9.4 software (SAS Institute). All endpoints were tested at the 2-sided at 5% significance level, and no adjustments for multiple comparisons were applied.