Animal models suggest a key role for dihydroxylated vitamin D metabolites in fracture healing, as evidenced by increases in serum concentration of 24R,25-dihydroxyvitamin D (24R,25[OH]2D) after long bone fracture. Human studies investigating the kinetics of serum concentrations of 24R,25[OH]2D, 1,25-dihydroxyvitamin D (1,25[OH]2D) and their parent metabolite 25-hydroxyvitamin D (25[OH]D) are lacking. We, therefore, conducted a longitudinal study to determine whether total, free, or bioavailable concentrations of these vitamin D metabolites fluctuate in humans after long bone fracture. Twenty-eight patients with cross-shaft (diaphyseal) long bone fracture presenting to an emergency department in London, UK, were studied. Serum concentrations of 25(OH)D, 24R,25(OH)2D, 1,25(OH)2D, vitamin D binding protein, albumin, and calcium were determined within 48 hours of fracture and again at 1 and 6 weeks postfracture. Concentrations of free and bioavailable vitamin D metabolites were calculated using standard equations. No changes in mean serum concentrations of 25(OH)D or 24R,25(OH)2D were seen at either follow-up time point versus baseline. In contrast, mean serum 1,25(OH)2D concentration declined by 21% over the course of the study, from 68.5 pmol/L at baseline to 54.1 pmol/L at 6 weeks (p < 0.05). This decline was associated with an increase in mean serum corrected calcium concentration, from 2.32 mmol/L at baseline to 2.40 mmol/L at 1 week (p < 0.001) that was maintained at 6 weeks. No changes in free or bioavailable concentrations of any vitamin D metabolite investigated were seen over the course of the study. We conclude that serum 1,25(OH)2D concentration declines after long bone fracture in humans but that the serum 24R,25(OH)2D concentration does not fluctuate. The latter finding contrasts with those of animal models reporting increases in serum 24R,25(OH)2D concentration after long bone fracture.
Up to 10% of patients with long bone fracture experience delayed union or nonunion, which is associated with significant morbidity and socioeconomic cost. A better understanding of factors involved in fracture healing is needed to inform rational development of therapies for preventing nonunion. Studies conducted in chicks suggest a key role for dihydroxylated vitamin D metabolites in fracture healing. Receptors for 1,25-dihydroxyvitamin D (1,25[OH]2D) and 24R,25-dihydroxyvitamin D (24R,25[OH]2D) are expressed in fracture healing callus,[4, 5] and co-administration of 24R,25(OH)2D and 1,25(OH)2D after fracture improves bone mechanical strength.[5, 6] In nonsupplemented chicks, these two metabolites accumulate in fracture healing callus. These changes are associated with decreases in serum 1,25(OH)2D concentration, postulated to arise as a result of consumption at the fracture site, and with early (10-day postfracture) increases in serum concentrations of 24R,25(OH)2D, associated with increased renal vitamin D 24-hydroxylase activity. Similar changes in serum concentrations of 1,25(OH)2D and 24R,25(OH)2D have been reported in rats and dogs after long bone fracture.[9, 10]
In contrast to the wealth of data from animal models, data on kinetics of change in vitamin D metabolites in humans postfracture are sparse and conflicting. Three pertinent studies have been conducted to date. The first investigated 13 young patients with long bone fracture at baseline and 6 weeks postfracture: Plasma concentrations of 24R,25(OH)2D increased by 41% over the course of the study, but concentrations of 1,25(OH)2D were not determined. The second study, conducted by the same group, investigated 41 elderly patients with long bone fracture at baseline and 8 weeks: Serum 1,25(OH)2D concentration decreased by 17% over the course of the study, and serum concentration of 24R,25(OH)2D did not fluctuate. The third, and largest, study investigated 205 white women aged >65 years with proximal femoral fracture at baseline, and at 10, 60, 180, and 360 days thereafter; serum 1,25(OH)2D concentration fell at 10 and 60 days postfracture, before recovering to baseline levels at 360 days; serum concentrations of 24R,25(OH)2D were not determined in this study. These studies suffer from two key limitations. First, none has determined serum 24R,25(OH)2D concentrations within 2 weeks of fracture; this is significant because in animal models, serum 24R,25(OH)2D concentrations peak at 10 days postfracture and normalize thereafter. Second, none has measured serum concentrations of vitamin D binding protein (DBP) and albumin. These plasma proteins bind vitamin D metabolites in serum and their concentration modifies free and bioavailable concentrations of these metabolites, with implications for their actions on bone. Concentration of these plasma proteins has long been recognized to fluctuate after trauma and surgery,[15-18] thus, physiologically relevant free and bioavailable concentrations of vitamin D metabolites might fluctuate postfracture, even if total concentrations do not change. We, therefore, conducted a study to determine kinetics of change in total, free, and bioavailable concentrations of dihydroxylated vitamin D metabolites, and of their parent vitamin D metabolite 25-hydroxyvitamin D (25[OH]D), in humans immediately post-long bone fracture and at 1 and 6 weeks thereafter.
Materials and Methods
Patients with a diaphyseal long bone fracture presenting to the Accident and Emergency Department, Royal London Hospital, London, UK, were assessed for eligibility to participate in the study. The following exclusion criteria were applied: age <16 years; open physes on X-ray; fracture sustained >48 hours pre-enrollment; taking medication affecting vitamin D metabolism (carbamazepine, phenobarbital, phenytoin, or primidone); taking >20 µg supplemental vitamin D per day; concomitant abdominal visceral injury, pleural injury other than pneumothorax, spinal cord injury, or Glasgow coma scale (GCS) <15 at presentation to hospital; known Paget's disease, osteopetrosis, metastatic bone cancer or primary bone cancer; currently in prison, unable to provide informed consent, or taking part in another clinical research project. Demographic and clinical details were recorded at baseline, and blood samples were drawn at enrollment and at 1 and 6 weeks postfracture for separation and freezing of serum at –80°C pending biochemical assays. Written informed consent was obtained from all participants. The study was approved by Outer South East London Research Ethics Committee (ref 10/H0805/6).
Serum concentrations of 25(OH)D and 24R,25(OH)2D were assayed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) as described by Wagner and colleagues; limits of detection were 0.24 nmol/L for 24R,25(OH)2D and 0.25 nmol/L for 25(OH)D. ELISA were employed to measure concentrations of 1,25(OH)2D (Immunodiagnostic Systems, Boldon, UK; sensitivity 6 pmol/L), DBP (R&D Systems, Minneapolis, MN, USA; sensitivity 0.65 ng/mL) and fibroblast growth factor 23 (FGF23) intact hormone and C-terminal fragment (Immutopics, San Clemente, CA, USA; sensitivity 1.5 RU/mL). Concentrations of phosphate, total serum calcium, albumin, and parathyroid hormone (PTH, sensitivity 0.32 pmol/L) were determined using an Architect ci8200 analyzer (Abbott Diagnostics, Chicago, IL, USA). Calcium concentration was corrected for serum albumin concentration using the equation: corrected calcium (mmol/L) = total calcium (mmol/L) + 0.02 × (40–albumin [g/L]). Concentrations of free and bioavailable vitamin D metabolites were calculated using equations developed by Bhan and colleagues. The following values for affinity constants between vitamin D metabolites and plasma proteins were utilized: for vitamin D binding protein, 2 × 108 M−1 for 25(OH)D and 24R,25(OH)2D, and 1 × 107 M−1 for 1,25(OH)2D, as reported by Kawakami and colleagues, and for albumin, 6 × 105 M−1 for 25(OH)D and 5.4 × 104 M−1 for 1,25(OH)2D, as reported by Bikle and colleagues.[22, 23] In the absence of published data regarding the affinity constant between 24R,25(OH)2D and albumin, this was assumed to be 6 × 105 M−1, as for 25(OH)D, on the grounds that these metabolites had similar affinity for DBP.
The study was powered to detect a 20% change in serum 24R,25(OH)2D concentration at follow-up versus baseline with 80% power at the 5% significance level, assuming mean baseline concentration of 24,25-dihydroxyvitamin D 4 nmol/L, standard deviation of 1.3 nmol/L, and 30% loss to follow-up. Repeated measures analysis of variance with Bonferroni correction was applied to compare mean serum concentrations of vitamin D metabolites and other parameters between time points. Spearman's r was calculated to determine whether baseline concentrations of vitamin D metabolites correlated with delay from time of fracture to time of baseline blood sampling. P values < 0.05 were considered statistically significant.
Thirty-three participants were recruited to the study between April and October 2010, of whom 30 attended 1-week follow-up and 28 attended 6-week follow-up. Baseline characteristics of the 28 participants attending all three study visits, whose samples entered the analysis, are presented in Table 1. Of these, 14 (50%) were vitamin D deficient (serum 25[OH]D <50 nmol/L), 5 (18%) had vitamin D insufficiency (serum 25[OH]D 50–74 nmol/L), and 9 (32%) were vitamin D replete (serum 25[OH]D ≥75 nmol/L) at presentation. The median time from sustaining fracture to first blood sample was 24.6 hours (interquartile range 5.4 to 34.2 hours). No correlation between time to sampling and baseline concentration of any of the vitamin D metabolites investigated was seen (for 25[OH]D, Spearman's r = 0.08, p = 0.69; for 1,25[OH]2D, r = 0.15, p = 0.45; and for 24R,25[OH]2D, r = 0.08, p = 0.70).
|Variable||Mean (SD)/n (%)a|
|Age (years)||53.5 (24.3)|
|Menopausal status (females only)|
|White British||24 (86%)|
|Open reduction, internal fixation||10 (36%)|
|History of osteoporosis|
|Current smoker||6 (21%)|
|Body mass index (kg/m2)||23.8 (4.0)|
|Total 25(OH)D (nmol/L)||51.4 (31.8)|
|Total 24R,25(OH)2D (nmol/L)||12.5 (5.4)|
|Total 1,25(OH)2D (pmol/L)||68.5 (39.4)|
|Free 25(OH)D (pmol/L)||49.1 (41.2)|
|Free 24R,25(OH)2D (pmol/L)||12.0 (8.3)|
|Free 1,25(OH)2D (pmol/L)||0.94 (0.58)|
|Bioavailable 25(OH)D (nmol/L)||16.1 (13.3)|
|Bioavailable 24R,25(OH)2D (nmol/L)||3.9 (2.6)|
|Bioavailable 1,25(OH)2D (pmol/L)||29.4 (18.3)|
|Vitamin D binding protein (mg/L)||259.8 (118.3)|
|Albumin (g/L)||36.2 (5.6)|
|Uncorrected calcium (mmol/L)||2.24 (0.14)|
|Corrected calcium (mmol/L)||2.32 (0.11)|
|Phosphate (mmol/L)||1.22 (0.28)|
|Parathyroid hormone (pmol/L)||3.65 (1.79)|
|Fibroblast growth factor 23 (RU/mL)||14.01 (24.87)|
Total serum concentrations of vitamin D metabolites are presented in Fig. 1. Mean total serum 1,25(OH)2D concentration declined by 21% over the course of the study, from 68.5 pmol/L at baseline to 54.1 pmol/L at 6 weeks (95% confidence interval [CI] for difference 0.5 to 28.1 pmol/L; p < 0.05; Fig. 1A). By contrast, no changes in mean total serum concentrations of 24R,25(OH)2D or 25(OH)D were seen at either follow-up time point versus baseline (Fig. 1B, C). No change in mean 24R,25(OH)2D: 1,25(OH)2D ratio was observed over the course of the study (p = 0.62). The decline in mean total serum 1,25(OH)2D concentration postfracture was associated with a statistically significant increase in mean serum corrected calcium concentration, from 2.32 mmol/L at baseline to 2.40 mmol/L at 1 week (95% CI for difference 0.04 mmol/L to 0.13 mmol/L; p < 0.001); this increase was maintained at 6 weeks (Fig. 2A). These changes were not associated with fluctuations in serum concentrations of PTH, phosphate, or FGF23 (Fig. 2B–D), and no correlation was observed between corrected serum calcium concentration and serum 1,25(OH)2D at baseline (Pearson's r = 0.18, p = 0.36), 1 week (Pearson's r = –0.18, p = 0.37) or 6 weeks (r = –0.20, p = 0.32). An upward trend in DBP concentration was observed over the course of the study, although this did not attain statistical significance (mean DBP concentration 259.8 mg/L at baseline versus 312 mg/L at 6 weeks, p > 0.05; Fig. 3A). Mean serum albumin concentration dipped initially (from 36.2 g/L at baseline to 34.4 g/L at 1 week) but subsequently increased again to re-attain baseline levels at 6 weeks (p < 0.01; Fig. 3B). No associated changes in mean serum concentrations of free (Fig. 4) or bioavailable (Fig. 5) concentrations of any vitamin D metabolite studied were seen over the course of the study. Subgroup analyses did not reveal any fluctuation in 24R,25(OH)2D concentrations over the course of the study in the 7 patients with fractures of the femoral or tibial shafts; in the 17 patients whose fractures were managed nonoperatively; in the 14 patients with baseline serum 25(OH)D concentration ≥50 nmol/L; or in the 14 patients aged ≤50 years.
We report results of the most comprehensive prospective study of vitamin D metabolism in humans after long bone fracture conducted to date. We observed a decline in serum 1,25(OH)2D concentrations during the study, associated with an increase in total and corrected calcium concentration. No fluctuations in serum 25(OH)D or 24R,25(OH)2D concentrations were seen.
Our observation that serum 24R,25(OH)2D concentrations did not increase postfracture contrasts with those of studies in young animals and humans, which have demonstrated significant increases in circulating concentrations of this metabolite after long bone fracture.[8, 10, 11] We propose four potential explanations for the contrasting findings that we report. First, the volume of callus produced by our patients may have been less than that in the studies referenced above, either because the largest fractures in the current study were managed by open reduction and internal fixation (ORIF), or because the median age of our study population was 50.5 years. Both direct healing and increasing age are associated with decreased callus formation.[25, 26] If fluctuations in serum 24R,25(OH)2D concentration are mediated by a soluble factor produced by healing callus, as has been suggested, they might be attenuated in the current study as compared with studies in which fractures occurring in younger animals/people were managed nonoperatively. Second, production of dihydroxylated vitamin D metabolites by patients in the current study might have been limited by availability of 25(OH)D substrate, whose concentration was <50 nmol/L in 50% of participants. However, no fluctuations in 24R,25(OH)2D were seen in the subgroup of patients with serum 25(OH)D concentrations ≥50 nmol/L. Third, fluctuations in circulating concentrations of vitamin D metabolites might not have been observed in the present study because many of the fractured bones were relatively small; however, when analysis was restricted to patients with fractures of the femoral or tibial shaft only, no fluctuations in serum 24R,25(OH)2D were seen. Finally, it may be that the contrast between results of human and animal studies represents a fundamental difference in the pathophysiology of fracture healing between different species.
Although results from the present study and animal models are discordant with respect to changes in 24R,25(OH)2D postfracture, our observation that serum total 1,25(OH)2D concentration falls postfracture is consistent with other human studies[12, 13] and with studies conducted in chicks and rats.[7, 9] This result cannot be attributed to a decline in PTH or a rise in FGF23 postfracture, as neither fluctuated during the study. It is also unlikely to be attributable to increased calcium intake in the postfracture period in our study, as no supplemental calcium or specific dietary advice was provided to study participants. Postfracture declines in serum 1,25(OH)2D concentrations have previously been attributed to increased consumption of 1,25(OH)2D by fracture healing callus, a hypothesis supported by the observation that 1,25(OH)2D accumulates in callus in rats and humans alike.[9, 27] An alternative explanation is that decreases in serum 1,25(OH)2D are secondary to increases in serum calcium resulting from postfracture immobility; our observation that serum 1,25(OH)2D concentrations did not correlate with serum calcium concentrations at any point in the study does not favor this interpretation, however. Whatever the underlying mechanism, it is interesting to note that the decline in total 1,25(OH)2D concentration we observed was not associated with any change in either the free or the bioavailable concentration of this metabolite. Recently it has been reported that free and bioavailable concentrations of 25(OH)D correlate more strongly with bone mineral density than total concentrations. If this principle holds for 1,25(OH)2D, then the impact of a fall in total 1,25(OH)2D on bone metabolism may be limited if the free and bioavailable concentrations of this metabolite are unchanged.
Our study has several strengths: It is unique among human studies in measuring serum concentrations of 24R,25(OH)2D at 1 week postfracture. Our patient population was heterogeneous: patients with different fracture sites, ages, sexes, ethnicities, and 25(OH)D concentrations were represented. Our results are therefore generalizable to a “real world” population of patients with long bone fracture. Moreover, our study was powered to detect modest (20%) changes in 24R,25(OH)2D; the lack of fluctuation in serum concentrations that we report is therefore likely to represent a true negative result, and the chance of type 2 error is low.
Our study also has limitations. First, and inevitably for a clinical study in humans, prefracture samples were not available. However, the fact that serum concentrations of vitamin D metabolites did not correlate with the duration of delay to sampling suggests that early fluctuations in serum concentrations of the metabolites of interest have not been missed. Second, although we were powered to detect small fluctuations in serum 24R,25(OH)2D concentration in the study population as a whole, we were not formally powered to detect potential subgroup effects, eg, in vitamin D–replete patients, in those with fractures of larger bones only, in those whose fractures were managed nonoperatively, or in specific age groups. However, the lack of any trend toward a change in 24R,25(OH)2D concentration over time in these groups suggests that we have not failed to detect clinically significant subgroup effects. Third, we studied circulating concentrations of vitamin D metabolites rather than bone-specific expression of 24- and 1-alpha-hydroxylase enzymes that might mediate local, but not circulating, concentrations of 24R,25(OH)2D. Our study does not, therefore, exclude a role for 24R,25(OH)2D in fracture healing. Finally, we were unable to correlate concentrations of vitamin D metabolites with risk of delayed/nonunion because a significant proportion of fractures were treated by open reduction and internal fixation. Future studies designed for longer-term follow-up of concentrations of vitamin D metabolites in fractures treated operatively and nonoperatively may help in quantifying such risk.
In conclusion, we report that serum 1,25(OH)2D concentration declines after long bone fracture in humans but that serum 24R,25(OH)2D concentration does not fluctuate. The latter finding contrasts with results of studies in animal models that report large increases in serum 24R,25(OH)2D concentration after long bone fracture; this difference may relate to the relatively advanced age of many of the participants in our study, and further investigations in younger populations are warranted.
All authors state that they have no conflicts of interest.
Funding was provided by Barts and The London Charity, ref 295/1595.
We thank Dr Eleftherios Diamandis and Dr Vathany Kulasingam, University of Toronto, Canada, for performing 24R,25(OH)2D assays.
Authors' roles: Study design: ADMB, MR, TH, AWN, CJG, and ARM. Study conduct: ADMB, VK, BDM, MR, and TH. Data collection: ADMB and VK. Biochemical analyses: CLG, PMT, TRV, and RV. Data analysis: ARM. Drafting manuscript: ARM. Approving final version of manuscript: all authors. ARM takes responsibility for the integrity of the data analysis.