- Email: [email protected]
Micronutrients and the risk of hip fracture: Case–control study
Clinical Nutrition xxx (2015) 1e6
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Micronutrients and the risk of hip fracture: Caseecontrol study Anne C. Torbergsen a, b, *, Leiv O. Watne a, c, Torgeir B. Wyller a, c, Frede Frihagen d, Knut Strømsøe d, Thomas Bøhmer e, Morten Mowe a, b a
Institute of Clinical Medicine, University of Oslo, Norway Department of General Internal Medicine, Oslo University Hospital, Norway c Department of Geriatric Medicine, Oslo University Hospital, Norway d Department of Orthopedic Surgery, Oslo University Hospital, Norway e Laboratory of Nutritional Analyses, Department of Medical Biochemistry, Oslo University Hospital, Norway b
a r t i c l e i n f o
s u m m a r y
Article history: Received 24 March 2015 Accepted 16 December 2015
Background: Vitamin D, and possibly vitamin K, has an established association to fracture risk. Other vitamins are, however, less studied. Aim: To determine whether specific micronutrients other than 25(OH)D and vitamin K play a role in risk of hip fracture and bone turnover. Methods: In this caseecontrol study, blood was drawn for measurements of vitamins A, B6, B12, C, E, and folic acid as well as the bone turnover markers osteocalcin and bone-specific alkaline phosphatase upon admission for hip fracture in 116 patients and in 73 home-dwelling non fractured controls. Results for vitamin K1 and 25(OH)D from the same populations have been reported previously. Results: Low vitamin A, C, and E concentrations were independently associated with a risk of hip fracture. The adjusted odds ratio (95% confidence interval) per 10 mmol/L increase in vitamin A concentration was 0.74 (0.65e0.84); for 1 mmol/L vitamin C and E: 0.94 (0.92e0.97) and 0.81 (0.74e0.89) respectively. The results were principally unchanged when 25(OH)D, vitamin K1, Body Mass Index, and other potential confounders were adjusted for. All vitamins except B12 and folic acid correlated positively with total osteocalcin and negatively with bone-specific alkaline phosphatase. Conclusions: Low vitamin A, C, and E concentrations are associated with an increased risk of hip fracture, possibly mediated through bone turnover mechanisms. This caseecontrol study is registered at: ClinicalTrials.gov. NCT01738776. The patient related outcome is also registered at: ClinicalTrials.gov. NCT01009268. © 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Hip fracture Vitamins Bone turnover markers
1. Introduction Low BMI is a well-established risk factor for hip fracture. However, whether a low BMI in itself or a decrease in specific micronutrients leads to the increased risk of fracture is unknown. Bone mass depends on the rate of bone turnover and ability to mineralize newly formed bone. A high rate of bone formation, as measured by
Abbreviations: BADL, barthel activities of daily living; BALP, Bone-specific alkaline phosphatase; BMD, bone mineral density; CRP, C-reactive protein; HGS, hand grip strength; OR, odds ratio; OC, osteocalcin; OOT, Oslo Orthogeriatric Trial; PTH, parathyroid hormone; totOC, total osteocalcin; ucOC, undercarboxylated osteocalcin. * Corresponding author. Strålsundveien 5, 1672 Kråkerøy, Norway. Tel.: þ47 95262196. E-mail address: [email protected] (A.C. Torbergsen).
bone-specific alkaline phosphatase (BALP), is often found in osteoporotic patients. A high rate of bone formation combined with low active osteocalcin (OC), a Ca-binding protein, may lead to immature bone with insufficient calcification, and may be detrimental for bone health and increase fracture risk. We previously reported that low vitamin K1 and low 25(OH)D synergistically increase the risk of hip fracture [1]. However, vitamin K1 and 25(OH)D were poorly correlated with total osteocalcin (totOC) and BALP, and only a minor fraction of the variance in totOC and BALP were attributable to vitamin K1and 25(OH)D concentrations. Vitamins C and E and the vitamin B family have received attention as potentially protective agents against degenerative disease, including osteoporosis [2]. A high intake of fruits and vegetables as well as low homocysteine is reported to be associated with a lower risk of hip fracture [3]. Whether this is a direct action
http://dx.doi.org/10.1016/j.clnu.2015.12.014 0261-5614/© 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
2
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
of vitamin C, vitamin E, or vitamin B, other nutrients, or a more indirect association with fruit and vegetable intake as a marker for a healthier lifestyle, is unknown. Several possible protective mechanisms of action of vitamins C and E on hip fracture have been suggested, such as their ability to reduce oxidative stress and inhibit bone resorption. Moreover, vitamin C is necessary for the synthesis of collagen, and is important for bone quality and tensile strength [4]. In addition, deficiencies in vitamin B (B6, B12, and folic acid), along with the consequent elevation in homocysteine, are suggested to be detrimental to bone [5e10]. Epidemiological studies have found that excessive vitamin A intake is a possible risk factor for hip fracture [11e13]. However, other epidemiological studies have failed to confirm this association [14]. It has been suggested that vitamin A may inhibit bone resorption and perhaps stimulate bone formation as well and possibly act in synergy with vitamin D in bone mineralization [15]. Therefore, the aims of this caseecontrol study were to determine whether low serum concentrations of vitamins A, C, E, and B (B6, B12 and folic acid) are associated with an increased risk of hip fracture in aged patients, if a potential association is independent of 25(OH)D, vitamin K1 and general malnutrition, and to examine whether a potential association can be explained through the bone turnover markers OC and BALP. 2. Material and methods 2.1. Study population 2.1.1. Cases This caseecontrol study was a substudy of the Oslo Orthogeriatric Trial (OOT) [16]. The patients (cases) in the OOT were consecutively admitted to Oslo University Hospital, Ullevål, Norway. Inclusion criteria in the OOT were those acutely admitted for a hip fracture as result of a low energy trauma, defined as a fall from 1 m or lower. Exclusion criteria was patients who were regarded as moribund at admittance (as determined by the admitting orthopaedic surgeon based upon their clinical experience) and patients lacking a valid informed consent or assent. The study population was examined during September 2009 to April 2011 for the nutrition sub study. Of 216 eligible patients enrolled in the OOT, 116 patients had a preoperative blood sample for vitamin analysis purposes performed and could be included in the case control study. Missing cases were due to that the patient was operated for hip fracture before there was time for blood drawl for the nutrition sub study, low capacity to draw blood for project purposes at weekends, holidays, and at night. For technical reasons, we were not able to obtain all vitamin analyses in all patients. Thus, analyses are missing for: vitamin A, n ¼ 10; vitamin C, n ¼ 28, vitamin E, n ¼ 8; vitamin B6 ¼ 8, vitamin B12 ¼ 12 and folic acid, n ¼ 14. Baseline data for the 216 patients included in the OOT were registered and all underwent a pre-surgical blood test as part of the clinical routine. There were no differences in age, BMI, morbidity, Creactive protein (CRP) or albumin, between those enrolled for vitamin and bone turnover tests and those not enrolled (data not shown). 2.1.2. Controls The control group consisted of individuals with no previous hip fracture, and was drawn at random by Statistics Norway from home-dwelling inhabitants aged 60e100 years (median age, 82 years) in the census files of Oslo in 2005. The control subjects were contacted by mail and followed up by two phone calls. A total of 73 control subjects (66% women) were recruited.
The catchment area for the cases and controls was the city and suburbs of Oslo, Norway. 2.2. Data collection Designated project staff collected the data in cases and controls. In cases, weight was measured using a class 3 chair scale as soon as possible post-surgery. Patients wore light clothing. Height was either measured using a tape measure against a wall or calculated from measured knee height [17]. The knee was flexed so that it was bent at 90 and measurements were taken from the anterior surface of the thigh near the patella to under the heel. Residence status was registered as home dwelling or institutionalized; alcohol consumption in total abstainers and non-abstainers, and smoking habits in current smokers or non smokers. The number of prescription medications used was recorded. Activity of daily living was measured using the Barthel Activities of Daily Living (BADL) Index [18]. Handgrip strength (HGS) was examined by hand dynamometry (Jamar, Germany: three repetitions per examination) in the dominant arm. Patients were examined daily preoperatively and until the fifth postoperative day. The best handgrip test was used. In the controls, weight was measured using a class 4 standing scale, and they wore light clothing. Standing height was measured with a tape measure towards a wall. Smoking and alcohol consumption were recorded and categorized as for the patients. The number of medications was recorded and the same BADL form was used for measuring daily activity. For HGS, the best result of three repetitions was used. 2.3. Preparation and analysis of blood samples In patients, blood was collected by venipuncture shortly after admission for hip fracture prior to the operation. In controls, blood was collected in the morning by venipuncture following an overnight fast. All samples were clotted for 30 min at room temperature and serum was separated by centrifugation. Aliquots were immediately stored at 80 C, and later analyzed for serum vitamins A, C, E, B6, B12, and folic acid. The bone turnover markers totOC, BALP, and parathyroid hormone were assayed in serum. Analysis of vitamin C was performed within 14 days after sampling, according to Zannoni et al. [19]. Vitamins A, E and B6 were continuously analyzed within 2 months after sampling. Vitamin B12, and folic acid were analyzed in one batch at the end of the study. Laboratory assays of retinol (vitamin A) were performed using the Bio-Rad Laboratories kit (Munich, Germany). Alpha-tocopherol (vitamin E) was assayed by radioimmunoassay also from Bio-Rad Laboratories. High pressure liquid chromatography was used for assays of pyridoxal-5’-phosphate (vitamin B6) by Chromsystems (Munich, Germany). Folic acid and cobalamin (vitamin B12) were assayed with a Hitachi 717 Modular multianalyzer (Boehringer Mannheim, Germany). Analysis of vitamin K1, 25(OH)D, and bone turnover indicators was performed as described previously [1]. BALP was quantified in E/L with 1E ¼ 1 mmol hydrolyzed p-NPP/min (p-NPP is a monoclonal anti-bone-ALP antigen). Vitamins C, E, and A, and the B vitamins were analyzed in the Department of Clinical Chemistry or Nutrition laboratory, department of Medical Biochemistry at Oslo University Hospital, Norway. The coefficients of variation for the analyses of the vitamins A, C, E and B6 ranged from 2.5% to 4.5%. The coefficient of variance for vitamin B12 and folic acid was 6.8 and 9.7 respectively. These coefficients of variation remained stable over time. None of the methods were changed and we used the same laboratories for cases and controls during the entire project period.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
Standard blood analyses for CRP and albumin were carried out in all patients (n ¼ 216) as well as in controls according to routines of the hospital laboratory. 2.4. Sample size and statistical analysis The study was originally powered to be used for analyses regarding vitamin K1 and vitamin D and the risk of hip fracture [1]. Thus, we had insufficient statistical power to run all vitamins and confounders in a multivariate regression model simultaneously. Normally distributed data are shown as means ± SD, non normal distributed data are shown as median (interquartile range), and categorical data as numbers (percentages). Student's t-test was used to compare continuous data and the ManneWhitney U-test was used for unevenly distributed variables. The chi-square test was used to compare categorical data and Spearman's rho or Pearson's correlation coefficients were calculated to examine correlations between continuous variables. Due to statistical power considerations and high correlations between vitamin concentrations, we initially built two binary logistic regression models. In model I, all statistically significant explanatory factors other than the vitamins of interest for this paper were included and then removed in a backward manner, removing the least significant variable first and keeping all variables at p 0.05. In this model we also included vitamin K1 and 25(OH)D as we have previously reported these to be a risk factor for hip fracture [1]. In model II, all vitamins studied in this paper were included. The variables were removed in a backward manner like in model I, in order to study which of the vitamins that were significantly associated with fracture risk independently from the other vitamins at stake. Finally, we built three logistic regression models, one for each of the vitamins that had demonstrated an independent and statistically significant association with fracture risk in model II. In these final models, we initially included all potential confounders that were independent and significant in model I, and then removed the variables in the same backward manner as described above. Before the multivariate models were built, positively skewed data (CRP, vitamin B6, folic acid and vitamin K1) were log transformed. BADL was highly skewed and was dichotomized at cut off ¼ 18. All p values are two-tailed. Statistical analyses were performed using SPSS 21 (SPSS Inc, Chicago, IL, USA) for Windows. 2.5. Ethics The Regional Committee for Medical Research Ethics, the Data Inspectorate, and the Directorate for Health and Social Affairs approved the study protocol. The study was performed according to the Helsinki Declaration of 1975 as revised in 1983. 3. Results 3.1. Patients' characteristics The controls (n ¼ 77) were predominantly home-dwelling, whereas 41% of the patient group (n ¼ 116) was institutionalized. Patients and controls were similar regarding age, sex and smoking habits. BMI was significantly lower in patients compared with controls (p ¼ 0.01). Patients had lower albumin and higher CRP concentrations (p < 0.001), used a higher number of medications (p < 0.001) and were more often totally abstinent from alcohol (p < 0.001) compared with controls. The patients had lower muscle strength (HGS, p ¼ 0.01) and reduced function (BALP, p < 0.001), compared with controls. Vitamins A, B6, C and E were significantly
3
lower in patients than in controls (all p < 0.001). There were no differences between patients and controls with respect to vitamin B12 or folic acid (Table 1). Missing variables are noted in the table. 3.2. Risk of hip fracture Initially, two separate backward regression models are presented in Table 2. Model I comprises all potential confounders and model II comprises all vitamins addressed in this paper. A strong association between vitamins A (p < 0.001), C (p < 0.001) and E (p ¼ 0.001) and the risk of hip fracture remained (model II), and was principally unchanged in the three final adjusted models (Table 3). A low BMI (p ¼ 0.03) and a BADL score ¼ 18 or below was also associated with increased hip fracture risk in the final multivariate models (Table 3). A predefined stratified analysis for institutionalized patients versus home-dwelling patients was also performed. Vitamin D and vitamin B6 were significantly lower in institutionalized patients compared with home-dwelling patients (p ¼ 0.02 and p ¼ 0.01, respectively). However, the odds ratios for hip fracture were almost identical in the two strata (data not shown). 3.3. Correlations between vitamins, bone turnover markers, physical function, and BMI The vitamins associated with hip fracture all showed low but statistically significant positive correlations with totOC, and negative correlations with BALP. The strongest correlation was that between vitamin C and totOC (p < 0.001). Vitamin B12 and Folic acid did not correlate with bone turnover markers. Daily functioning measured by the BADL Index correlated with all micronutrients. HGS and BMI were weakly or not correlated with micronutrients (Table 4).
Table 1 Characteristics of patients and controls.
Institutionalized Age (y)a Sex (% female)b BMI (kg/m2)a Albumin (g/L)a CRP (mg/L)c Number of medicationsa Handgrip strength (kg)a BADL Indexc Total abstainer from alcoholb Current smokersb Vitamin A (mmol/L)a Vitamin C (mmol/L)a Vitamin E (mmol/L)a Vitamin B6 (nmol/L)c Vitamin B12 (nmol/L)c Folic acid (nmol/L)c
Controls (n ¼ 73)
Patients (n ¼ 116)
p
47 (41) 82 ± 8 48 (66) 26 ± 4 42 ± 3 2 (1e6) 3±2 26 ± 9 20 (19,20) 34 (47) 10 (14) 2.0 ± 0.4 62 ± 20 34 ± 8 34 (24e65) 342 (247e491) 18 (13e30)
3 (4) 83 ± 9 87 (75) 24 ± 5 39 ± 4 6 (2e40) 5±3 22 ± 9 18 (15e20) 82 (71) 19 (17) 1.3 ± 0.5 30 ± 19 22 ± 8 17 (11e35) 352 (263e531) 15 (9e37)
<0.001 0.5 0.2 0.03 <0.001 <0.001 <0.001 0.01 0.01 <0.001 0.5 <0.001 <0.001 <0.001 <0.001 0.4 0.07
Data are missing for some variables: BMI, n ¼ 41 patients; Albumin, n ¼ 1 patient and n ¼ 1 control; CRP, n ¼ 3 patients and n ¼ 3 controls; number of medications, n ¼ 1 control; marital status, n ¼ 1 patient and n ¼ 2 controls; education, n ¼ 17 patients; HGS, n ¼ 18 patients and n ¼ 1 control; BADL, n ¼ 4 patients and n ¼ 2 controls; alcohol consumption, n ¼ 1 patient; smoking, n ¼ 7 patients; Vitamin analysis is missing for some patients: vitamin A, n ¼ 10; vitamin C, n ¼ 28; vitamin E, n ¼ 8; vitamin B6 ¼ 8, vitamin B12 ¼ 12 and folic acid, n ¼ 14. BADL ¼ barthel activities of daily living, CRP ¼ C-reactive protein, HGS ¼ hand grip strength. a Data are mean ± SD. Patients were compared with controls using the Student's t-test. b Data are n (%). Patients were compared with controls using the c2 test. c Data are median (IQR). Patients were compared with controls using the ManneWhitney U-test.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
4
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
Table 2 Risk factors for hip fracture, logistic regression models. Unadjusted models OR (95% CI) P Vitamin A (10 mmol/L) Vitamin B6, log transformed Folic Acid, log transformed Vitamin C (mmol/L) Vitamin E (mmol/L) Vitamin K1, log transformed 25(OH)D (nmol/L) Age (y) Sex, female BMI (kg/m2) Number of medications CRP, log transformed Albumin (g/L) HGS (kg) BADL score cutoff ¼ 18 Total abstainer from alcohol
0.75 (0.69e0.82) <0.001 0.3 (0.2-0.5) <0.001 0.4 (0.1e1.2) 0.09 0.93 (0.91e0.95) <0.001 0.83 (0.78e0.88) <0.001 0.2 (0.1e0.4) <0.001 0.96 (0.95e0.97) <0.001 1.01 (0.98e1.05) 0.50 2 (1e3) 0.11 0.9 (0.8e0.99) 0.02 1.3 (1.1e1.5) 0.001 1.5 (1.2e1.8) <0.001 0.8 (0.7e0.9) <0.001 0.95 (0.92e0.99) 0.01 6 (3e11) <0.001 3 (2e6) <0.001
Adjusted model I OR (95% CI) P
Adjusted model II OR (95% CI) P 0.74 (0.65e0.85) <0.001
0.95 (0.92e0.97) <0.001 0.89 (0.83e0.96) 0.001 0.38 (0.20e0.73) 0.01 0.97 (0.95e0.98) <0.001
0.88 (0.78e0.99) 0.03
0.84 (0.71e0.98) 0.03 3 (1e9) 0.01
The potential confounders at p < 0.1 form the univariate analyses were adjusted for, removing one variable at the time in a stepwise backward manner in the model, removing the least significant variable first and keeping the variable in the model if p < 0.05. Potential confounders were analyzed two models: background variables in multivariate model I, and vitamins addressed in this paper in multivariate model II. Statistically significant variables from models I and II were then added in the final models presented in Table 3. BADL ¼ barthel activities of daily living, CRP ¼ C-reactive protein, HGS ¼ hand grip strength, OR ¼ odds ratio.
Table 3 Risk of hip fracture, final models. Model with vitamin A OR (95% CI) p Vitamin A (10 mmol/L) Vitamin C (mmol/L) Vitamin E (mmol/L) 25(OH)D (nmol/L) Vitamin K, log transformed BMI (kg/m2) BADL <18
Model with vitamin C OR (95% CI) p
Model with vitamin E OR (95% CI) p
0.74 (0.65e0.84) <0.001 0.94 (0.92e0.97) <0.001 0.97 (0.95e0.99) 0.01
0.98 (0.96e0.99) 0.03
0.86 (0.76e0.97) 0.01 5 (2e15) 0.01
0.79 (0.68e0.92) 0.01 4 (1e11) 0.03
0.81 (0.74e0.89) 0.97 (0.95e0.99) 0.30 (0.13e0.68) 0.84 (0.73e0.98) 7 (2e25) 0.01
<0.001 <0.001 0.01 0.02
Statistically significant explanatory variables from models I and II (Table 2) were combined in three final models, one for each of the vitamins A, C and E. All the variables were initially forced into the model and then removed in a stepwise backward manner, removing the least significant variable first and keeping the variable when significant at p ¼ 0.05. BADL ¼ barthel activities of daily living, CI ¼ confidence interval, CRP¼C-reactive protein, HGS ¼ hand grip strength, OR ¼ odds ratio.
Table 4 Correlations between micronutrients, bone turnover, physical function, and BMI. TotOCa Vitamin Vitamin Vitamin Vitamin
A C E B6
***
0.28 0.41*** 0.24*** 0.25**
PTHb
BALP 0.22 0.212** - 0.151* 0.321***
BADL Indexa ***
1***
0.20**
0.28 0.39*** 0.25*** 0.35***
HGSa
BMI
*
0.16 0.19*
*** ¼ p < 0.001. ** ¼ p ¼ 0.01. * ¼ p ¼ 0.05. BADL ¼ barthel activities of daily living, BALP ¼ bone-specific alkaline phosphatase, HGS ¼ hand grip strength, PTH ¼ parathyroid hormone, totOC ¼ total osteocalcin. a Spearman’s rho. b Pearson's correlation coefficient.
4. Discussion We found that low levels of serum vitamins A, C, and E were associated with increased risk of hip fracture. The association remained statistically significant when adjusting for vitamin K1 and 25(OH)D as well as other potential confounders. Vitamin B6 was also associated with an increased risk of hip fracture when other confounders were adjusted for, but not when adjusting for the other vitamins. This indicates that a sufficient supply of all the vitamins A, C, D and E and perhaps vitamin B6 and K1 may be important to avoid hip fracture. Although vitamin D is an established mediator in
optimizing BMD and preventing fractures [20], our data suggests that supplementing with vitamin D alone may be insufficient in preventing hip fracture. Low BMI is associated with increased risk of hip fracture, but our findings indicate that low levels of vitamin A, C, E and 25(OH)D are independent risk factors, and associated with hip fractures in both underweight, normal and overweight patients. The patients were frailer than the controls, 40% being institutionalized. As expected, they had a lower BMI, used more medications, had a lower HGS, and scored lower in the BADL Index compared with controls. The vitamins A, B6, C and E, as well as the previously reported 25(OH)D and vitamin K1 [1], were all positively correlated with totOC and negatively correlated with BALP. This suggests that these vitamins may act directly in the metabolism of bone. Vitamin C showed the strongest correlation with totOC, suggesting that vitamin C may be more important for the production of the OC than previously assumed. Johnson et al. recently reported that the vitamins D and C synergistically increase OC promoter activity and synthesis of undercarboxylated osteocalcin (ucOC) [21]. This supports our assumption that vitamin C may act through up regulating the synthesis of OC. In vitro cell culture studies have shown bone growth, and an increase in synthesis of collagen I and OC in cells supplemented with vitamin C [22]. During carboxylation, vitamin K drives the reaction of inactive ucOC to active OC, and a vitamin K epoxide is formed [23]. Vitamin B6 is thought to be necessary for regenerating active
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
vitamin K [24]. Vitamins with anti-oxidative properties, such as vitamins A, C and E, might also be important for this process. The nuclear receptor for vitamin D forms heterodimers with the nuclear receptor X (RXR) in the promoter region for the gene that codes for OC, and may up regulate the synthesis of ucOC [25]. Vitamin A also acts through a nuclear receptor, retinoic acid receptor, which forms heterodimers with nuclear receptor RXR [15,25]. This indicates that vitamin A may act in synergy with vitamin D in bone mineralization. Based on the literature and our findings, we have suggested a mechanism of interaction between nutrients and the ability to mineralize bone, as shown in Fig. 1 [16,21,25,27]. This mechanism is so far speculative, and further research is required. High homocysteine has been associated with an increased risk of hip fracture [6,8,9]. Among the vitamins that take part in the homocysteine metabolism, vitamin B6 was lower in the hip fracture patients compared with controls. However, vitamin B6 was not associated with increased hip fracture risk when adjusting for vitamin C and vitamin A. This may be due to lack of statistical power as the variance for vitamin B6 was large. Review papers throughout the last decade have suggested that a high dietary intake of vitamin A is associated with an increased risk of fracture [15]. Barker et al. found a positive correlation between retinol and bone mineral density, but no relation between serum retinol and BALP or bone resorption [14]. The discrepancy between our results and those of Barker et al. may partially be explained by the fact that serum retinol in the cases in their study was at the same level as in our non-fractured control group. A considerably lower vitamin A concentration, as in our cases, may be associated with high bone turnover and compromised bone quality. In a national survey, vegetables were found to be the best contributor of vitamin A in the Norwegian diet, and vitamin A is reported to be adequate in the general population [26]. However, a large proportion of our patients had vitamin A deficiency or
5
inadequate vitamin A status as defined by serum retinol concentrations <0.7 mmol/L and <1.05 mmol/L, respectively. No patients had serum retinol concentrations above 4.2 mmol/L, which is regarded as the upper limit for serum vitamin A [15]. The patients in our study used a greater number of medications than did controls. This obviously reflects a higher degree of disease and inflammation in patients, which may have also caused increased oxidative stress and contributed to the association with hip fracture. However, we adjusted for the number of prescription medications used, and the association of an increased risk of hip fracture in patients with low vitamin A, C, and E concentrations remained unchanged. This association also remained unchanged when CRP, a proxy measure for inflammation, was adjusted for. Some authors have suggested that an effect of low vitamin C and E concentrations is only seen in smokers [3,27]. In our population, only 18% of patients and 14% of healthy controls were current smokers, and differences in smoking could not explain the association between low vitamin C, E and hip fracture. A limitation of our study is that we lack detailed data on comorbidity in the controls and we were unable to correct for comorbidity. Some medications may have influenzed on vitamin status i.e. warfarin. Five patients were using warfarin, but controls may also have been taken warfarin. Therefore, warfarin users were included in the analysis. The number of previous falls were not registered in neither cases nor controls, and a different risk of falling may in part explain the risk of a hip fracture in the two groups. However, the Norwegian Epidemiologic Osteoporotic studies, like our findings, suggest that low serum 25(OH)D and vitamin E is associated with an increased risk of hip fracture, whereas no association was found for vitamin A. That study did not report findings for the water soluble vitamins or vitamin K [28]. Height was measured using standing height in the controls, whereas knee height was measured in some (n ¼ 29) patients as a proxy measurement in those unable to stand. Knee height
Fig. 1. Proposed mechanism for the influence of vitamins upon bone mineral density (BMD). Bone mineral density (BMD) is dependent of the amount of calcium retained in bone, which is possibly dependent on several micronutrients: The bone-specific protein osteocalcin (OC) binds calcium to bone, but osteocalcin is formed in an inactive form, undercarboxylated osteocalcin (ucOC). Vitamin K acts as a catalyst in the activation to form carboxylated osteocalcin, the active form of OC. In this process, a vitamin K epoxide is formed. Vitamin B6 is thought to be necessary to regenerate vitamin K. The antioxidants, vitamins A, C,D or E, may also play a role. The promoter for the prepeptide ucOC contains a vitamin D responsive element and a vitamin C responsive element that regulate ucOC production. Vitamin A might regulate the promoter for the ucOC gene by forming heterodimers between RAR and RXR, which are the nuclear receptors for vitamin A and 1,25(OH)D, respectively [15,24,25,29]. RAR ¼ retinoic acid receptor, RXR ¼ receptor X receptor, VDR ¼ vitamin D Receptor, VCRE ¼ vitamin C responsive element, VDRE ¼ vitamin D responsive element.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
6
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
underestimated height somewhat in the patient group (data from the OOT) and may have given false high BMI in the 29 patients measured using knee height. We were not able to obtain data on weightloss prior to fracture; a measure that indicates nutritional risk better than BMI and lack good data for risk of malnutrition may influence upon our results. In order to prevent issues around multivitamin supplementation, oral nutrition-, enteral- or parenteral support and intravenous fluids during hospitalization, blood was drawn as soon as possible after informed consent in the was given in the intensive care unit and prior to surgery. Case-control studies can only identify associations and not prove causality. Intervention studies are required to confirm the associations found in our study. 5. Conclusion Low vitamin A, C, and E concentrations were associated with an increased risk of hip fracture, also when 25(OH)D, vitamin K1 and BMI were adjusted for. The association is possibly mediated through bone turnover mechanisms. Supplementation with vitamins K and D may not be sufficient to optimize bone health and prevent hip fracture. Sources of support This study was supported by a research fellowship grant from Oslo University Hospital, Medical Clinic Norway, and the Sophies Minde Ortopedi Foundation (to ACT). Conflict of interest None of the authors declare a conflict of interest. Acknowledgments We thank Thomas Gundersen at AS Vitas, Kari Julien at the Hormone Laboratory, and Anne Hove at the Nutritional Laboratory, Oslo University Hospital, for performing the biochemical analyses. We also thank Elisabeth Fragaat, Grete Wang, Gunnhild Bærdal, and Gurkirpal Singh for participating in data collection, as well as Lien Diep for statistical advice. TBW, FF, KS, TB, and MM designed the research. FF and KS were the principal orthopedic surgeons in charge of including hip fracture patients and healthy controls, respectively. ACT and LOW conducted the research. ACT analyzed the data and wrote the manuscript. MM had primary responsibility for final content. All authors read and approved the final manuscript. References [1] Torbergsen AC, Watne LO, Wyller TB, Frihagen F, Stromsoe K, Bohmer T, et al. Vitamin K1 and 25(OH)D are independently and synergistically associated with a risk for hip fracture in an elderly population: a case control study. Clin Nutr 2015;34:101e6. [2] Levis S, Theodore G. Summary of AHRQ's comparative effectiveness review of treatment to prevent fractures in men and women with low bone density or osteoporosis: update of the 2007 report. J Manag Care Pharm 2012;18:S1e15. [3] Lanham-New SA. Fruit and vegetables: the unexpected natural answer to the question of osteoporosis prevention? Am J Clin Nutr 2006;83:1254e5.
[4] Sahni S, Hannan MT, Gagnon D, Blumberg J, Cupples LA, Kiel DP, et al. Protective effect of total and supplemental vitamin C intake on the risk of hip fractureea 17-year follow-up from the Framingham Osteoporosis Study. Osteoporos Int 2009;20:1853e61. [5] Ahmadieh H, Arabi A. Vitamins and bone health: beyond calcium and vitamin D. Nutr Rev 2011;69:584e98. [6] Dhonukshe-Rutten RA, Pluijm SM, De Groot LC, Lips P, Smit JH, van Staveren WA. Homocysteine and vitamin B12 status relate to bone turnover markers, broadband ultrasound attenuation, and fractures in healthy elderly people. J Bone Min Res 2005;20:921e9. [7] Gjesdal CG, Vollset SE, Ueland PM, Refsum H, Meyer HE, Tell GS. Plasma homocysteine, folate, and vitamin B 12 and the risk of hip fracture: the Hordaland homocysteine study. J Bone Min Res 2007;22:747e56. [8] Leboff MS, Narweker R, LaCroix A, Wu L, Jackson R, Lee J, et al. Homocysteine levels and risk of hip fracture in postmenopausal women. J Clin Endocrinol Metab 2009;94:1207e13. [9] Sato Y, Honda Y, Iwamoto J, Kanoko T, Satoh K. Homocysteine as a predictive factor for hip fracture in stroke patients. Bone 2005;36:721e6. [10] Sato Y, Honda Y, Iwamoto J, Kanoko T, Satoh K. Effect of folate and mecobalamin on hip fractures in patients with stroke: a randomized controlled trial. JAMA 2005;293:1082e8. [11] Melhus H, Michaelsson K, Kindmark A, Bergstrom R, Holmberg L, Mallmin H, et al. Excessive dietary intake of vitamin A is associated with reduced bone mineral density and increased risk for hip fracture. Ann Intern Med 1998;129: 770e8. [12] Feskanich D, Singh V, Willett WC, Colditz GA. Vitamin A intake and hip fractures among postmenopausal women. JAMA 2002;287:47e54. [13] Opotowsky AR, Bilezikian JP. Serum vitamin A concentration and the risk of hip fracture among women 50 to 74 years old in the United States: a prospective analysis of the NHANES I follow-up study. Am J Med 2004;117: 169e74. [14] Barker ME, McCloskey E, Saha S, Gossiel F, Charlesworth D, Powers HJ, et al. Serum retinoids and beta-carotene as predictors of hip and other fractures in elderly women. J Bone Min Res 2005;20:913e20. [15] Conaway HH, Henning P, Lerner UH. Vitamin A metabolism, action, and role in skeletal homeostasis. Endocr Rev 2013;34:766e97. [16] Watne LO, Torbergsen AC, Conroy S, Engedal K, Frihagen F, Hjorthaug GA, et al. The effect of a pre- and postoperative orthogeriatric service on cognitive function in patients with hip fracture: randomized controlled trial (Oslo Orthogeriatric Trial). BMC Med 2014;12:63. [17] Chumlea WC, Guo SS, Wholihan K, Cockram D, Kuczmarski RJ, Johnson CL. Stature prediction equations for elderly non-Hispanic white, non-Hispanic black, and Mexican-American persons developed from NHANES III data. J Am Diet Assoc 1998;98:137e42. [18] Mahoney FI, Barthel DW. Functional evaluation: the barthel index. Md State Med J 1965;14:61e5. [19] Zannoni V, Lynch M, Goldstein S, Sato P. A rapid micromethod for the determination of ascorbic acid in plasma and tissues. Biochem Med 1974;11: 41e8. [20] Bischoff-Ferrari HA. Optimal serum 25-hydroxyvitamin D levels for multiple health outcomes. Adv Exp Med Biol 2014;810:500e25. [21] Johnson NA, Chen BH, Sung SY, Liao CH, Hsiao WC, Chung WK, et al. A novel targeting modality for renal cell carcinoma: human osteocalcin promotermediated gene therapy synergistically induced by vitamin C and vitamin D(3). J Gene Med 2010;12:892e903. [22] Urban K, Hohling HJ, Luttenberg B, Szuwart T, Plate U. An in vitro study of osteoblast vitality influenced by the vitamins C and E. Head Face Med 2012;8: 25. [23] Vermeer C, Knapen MH, Schurgers LJ. Vitamin K and metabolic bone disease. J Clin Pathol 1998;51:424e6. [24] Reynolds TM. Vitamin B6 deficiency may also be important. Clin Chem 1998;44:2555e6. [25] Sierra J, Villagra A, Paredes R, Cruzat F, Gutierrez S, Javed A, et al. Regulation of the bone-specific osteocalcin gene by p300 requires Runx2/Cbfa1 and the vitamin D3 receptor but not p300 intrinsic histone acetyltransferase activity. Mol Cell Biol 2003;23:3339e51. [26] Helsedirektoratet. Norkost 3. En landsomfattende kostundersøkelse blant menn og kvinner i Norge i alderen 18-70 år. 2011. [27] Melhus H, Michaelsson K, Holmberg L, Wolk A, Ljunghall S. Smoking, antioxidant vitamins, and the risk of hip fracture. J Bone Min Res 1999;(14): 129e35. [28] Sogaard AJ, Meyer HE, Emaus N, Grimnes G, Gjesdal CG, Forsmo S, et al. Cohort profile: Norwegian epidemiologic osteoporosis studies (NOREPOS). Scand J Public Health 2014;42:804e13. [29] Vermeer C, Theuwissen E, Vitamin K. Osteoporosis and degenerative diseases of ageing. Menopause Int 2011;17:19e23.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Micronutrients and the risk of hip fracture: Caseecontrol study Anne C. Torbergsen a, b, *, Leiv O. Watne a, c, Torgeir B. Wyller a, c, Frede Frihagen d, Knut Strømsøe d, Thomas Bøhmer e, Morten Mowe a, b a
Institute of Clinical Medicine, University of Oslo, Norway Department of General Internal Medicine, Oslo University Hospital, Norway c Department of Geriatric Medicine, Oslo University Hospital, Norway d Department of Orthopedic Surgery, Oslo University Hospital, Norway e Laboratory of Nutritional Analyses, Department of Medical Biochemistry, Oslo University Hospital, Norway b
a r t i c l e i n f o
s u m m a r y
Article history: Received 24 March 2015 Accepted 16 December 2015
Background: Vitamin D, and possibly vitamin K, has an established association to fracture risk. Other vitamins are, however, less studied. Aim: To determine whether specific micronutrients other than 25(OH)D and vitamin K play a role in risk of hip fracture and bone turnover. Methods: In this caseecontrol study, blood was drawn for measurements of vitamins A, B6, B12, C, E, and folic acid as well as the bone turnover markers osteocalcin and bone-specific alkaline phosphatase upon admission for hip fracture in 116 patients and in 73 home-dwelling non fractured controls. Results for vitamin K1 and 25(OH)D from the same populations have been reported previously. Results: Low vitamin A, C, and E concentrations were independently associated with a risk of hip fracture. The adjusted odds ratio (95% confidence interval) per 10 mmol/L increase in vitamin A concentration was 0.74 (0.65e0.84); for 1 mmol/L vitamin C and E: 0.94 (0.92e0.97) and 0.81 (0.74e0.89) respectively. The results were principally unchanged when 25(OH)D, vitamin K1, Body Mass Index, and other potential confounders were adjusted for. All vitamins except B12 and folic acid correlated positively with total osteocalcin and negatively with bone-specific alkaline phosphatase. Conclusions: Low vitamin A, C, and E concentrations are associated with an increased risk of hip fracture, possibly mediated through bone turnover mechanisms. This caseecontrol study is registered at: ClinicalTrials.gov. NCT01738776. The patient related outcome is also registered at: ClinicalTrials.gov. NCT01009268. © 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Hip fracture Vitamins Bone turnover markers
1. Introduction Low BMI is a well-established risk factor for hip fracture. However, whether a low BMI in itself or a decrease in specific micronutrients leads to the increased risk of fracture is unknown. Bone mass depends on the rate of bone turnover and ability to mineralize newly formed bone. A high rate of bone formation, as measured by
Abbreviations: BADL, barthel activities of daily living; BALP, Bone-specific alkaline phosphatase; BMD, bone mineral density; CRP, C-reactive protein; HGS, hand grip strength; OR, odds ratio; OC, osteocalcin; OOT, Oslo Orthogeriatric Trial; PTH, parathyroid hormone; totOC, total osteocalcin; ucOC, undercarboxylated osteocalcin. * Corresponding author. Strålsundveien 5, 1672 Kråkerøy, Norway. Tel.: þ47 95262196. E-mail address: [email protected] (A.C. Torbergsen).
bone-specific alkaline phosphatase (BALP), is often found in osteoporotic patients. A high rate of bone formation combined with low active osteocalcin (OC), a Ca-binding protein, may lead to immature bone with insufficient calcification, and may be detrimental for bone health and increase fracture risk. We previously reported that low vitamin K1 and low 25(OH)D synergistically increase the risk of hip fracture [1]. However, vitamin K1 and 25(OH)D were poorly correlated with total osteocalcin (totOC) and BALP, and only a minor fraction of the variance in totOC and BALP were attributable to vitamin K1and 25(OH)D concentrations. Vitamins C and E and the vitamin B family have received attention as potentially protective agents against degenerative disease, including osteoporosis [2]. A high intake of fruits and vegetables as well as low homocysteine is reported to be associated with a lower risk of hip fracture [3]. Whether this is a direct action
http://dx.doi.org/10.1016/j.clnu.2015.12.014 0261-5614/© 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
2
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
of vitamin C, vitamin E, or vitamin B, other nutrients, or a more indirect association with fruit and vegetable intake as a marker for a healthier lifestyle, is unknown. Several possible protective mechanisms of action of vitamins C and E on hip fracture have been suggested, such as their ability to reduce oxidative stress and inhibit bone resorption. Moreover, vitamin C is necessary for the synthesis of collagen, and is important for bone quality and tensile strength [4]. In addition, deficiencies in vitamin B (B6, B12, and folic acid), along with the consequent elevation in homocysteine, are suggested to be detrimental to bone [5e10]. Epidemiological studies have found that excessive vitamin A intake is a possible risk factor for hip fracture [11e13]. However, other epidemiological studies have failed to confirm this association [14]. It has been suggested that vitamin A may inhibit bone resorption and perhaps stimulate bone formation as well and possibly act in synergy with vitamin D in bone mineralization [15]. Therefore, the aims of this caseecontrol study were to determine whether low serum concentrations of vitamins A, C, E, and B (B6, B12 and folic acid) are associated with an increased risk of hip fracture in aged patients, if a potential association is independent of 25(OH)D, vitamin K1 and general malnutrition, and to examine whether a potential association can be explained through the bone turnover markers OC and BALP. 2. Material and methods 2.1. Study population 2.1.1. Cases This caseecontrol study was a substudy of the Oslo Orthogeriatric Trial (OOT) [16]. The patients (cases) in the OOT were consecutively admitted to Oslo University Hospital, Ullevål, Norway. Inclusion criteria in the OOT were those acutely admitted for a hip fracture as result of a low energy trauma, defined as a fall from 1 m or lower. Exclusion criteria was patients who were regarded as moribund at admittance (as determined by the admitting orthopaedic surgeon based upon their clinical experience) and patients lacking a valid informed consent or assent. The study population was examined during September 2009 to April 2011 for the nutrition sub study. Of 216 eligible patients enrolled in the OOT, 116 patients had a preoperative blood sample for vitamin analysis purposes performed and could be included in the case control study. Missing cases were due to that the patient was operated for hip fracture before there was time for blood drawl for the nutrition sub study, low capacity to draw blood for project purposes at weekends, holidays, and at night. For technical reasons, we were not able to obtain all vitamin analyses in all patients. Thus, analyses are missing for: vitamin A, n ¼ 10; vitamin C, n ¼ 28, vitamin E, n ¼ 8; vitamin B6 ¼ 8, vitamin B12 ¼ 12 and folic acid, n ¼ 14. Baseline data for the 216 patients included in the OOT were registered and all underwent a pre-surgical blood test as part of the clinical routine. There were no differences in age, BMI, morbidity, Creactive protein (CRP) or albumin, between those enrolled for vitamin and bone turnover tests and those not enrolled (data not shown). 2.1.2. Controls The control group consisted of individuals with no previous hip fracture, and was drawn at random by Statistics Norway from home-dwelling inhabitants aged 60e100 years (median age, 82 years) in the census files of Oslo in 2005. The control subjects were contacted by mail and followed up by two phone calls. A total of 73 control subjects (66% women) were recruited.
The catchment area for the cases and controls was the city and suburbs of Oslo, Norway. 2.2. Data collection Designated project staff collected the data in cases and controls. In cases, weight was measured using a class 3 chair scale as soon as possible post-surgery. Patients wore light clothing. Height was either measured using a tape measure against a wall or calculated from measured knee height [17]. The knee was flexed so that it was bent at 90 and measurements were taken from the anterior surface of the thigh near the patella to under the heel. Residence status was registered as home dwelling or institutionalized; alcohol consumption in total abstainers and non-abstainers, and smoking habits in current smokers or non smokers. The number of prescription medications used was recorded. Activity of daily living was measured using the Barthel Activities of Daily Living (BADL) Index [18]. Handgrip strength (HGS) was examined by hand dynamometry (Jamar, Germany: three repetitions per examination) in the dominant arm. Patients were examined daily preoperatively and until the fifth postoperative day. The best handgrip test was used. In the controls, weight was measured using a class 4 standing scale, and they wore light clothing. Standing height was measured with a tape measure towards a wall. Smoking and alcohol consumption were recorded and categorized as for the patients. The number of medications was recorded and the same BADL form was used for measuring daily activity. For HGS, the best result of three repetitions was used. 2.3. Preparation and analysis of blood samples In patients, blood was collected by venipuncture shortly after admission for hip fracture prior to the operation. In controls, blood was collected in the morning by venipuncture following an overnight fast. All samples were clotted for 30 min at room temperature and serum was separated by centrifugation. Aliquots were immediately stored at 80 C, and later analyzed for serum vitamins A, C, E, B6, B12, and folic acid. The bone turnover markers totOC, BALP, and parathyroid hormone were assayed in serum. Analysis of vitamin C was performed within 14 days after sampling, according to Zannoni et al. [19]. Vitamins A, E and B6 were continuously analyzed within 2 months after sampling. Vitamin B12, and folic acid were analyzed in one batch at the end of the study. Laboratory assays of retinol (vitamin A) were performed using the Bio-Rad Laboratories kit (Munich, Germany). Alpha-tocopherol (vitamin E) was assayed by radioimmunoassay also from Bio-Rad Laboratories. High pressure liquid chromatography was used for assays of pyridoxal-5’-phosphate (vitamin B6) by Chromsystems (Munich, Germany). Folic acid and cobalamin (vitamin B12) were assayed with a Hitachi 717 Modular multianalyzer (Boehringer Mannheim, Germany). Analysis of vitamin K1, 25(OH)D, and bone turnover indicators was performed as described previously [1]. BALP was quantified in E/L with 1E ¼ 1 mmol hydrolyzed p-NPP/min (p-NPP is a monoclonal anti-bone-ALP antigen). Vitamins C, E, and A, and the B vitamins were analyzed in the Department of Clinical Chemistry or Nutrition laboratory, department of Medical Biochemistry at Oslo University Hospital, Norway. The coefficients of variation for the analyses of the vitamins A, C, E and B6 ranged from 2.5% to 4.5%. The coefficient of variance for vitamin B12 and folic acid was 6.8 and 9.7 respectively. These coefficients of variation remained stable over time. None of the methods were changed and we used the same laboratories for cases and controls during the entire project period.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
Standard blood analyses for CRP and albumin were carried out in all patients (n ¼ 216) as well as in controls according to routines of the hospital laboratory. 2.4. Sample size and statistical analysis The study was originally powered to be used for analyses regarding vitamin K1 and vitamin D and the risk of hip fracture [1]. Thus, we had insufficient statistical power to run all vitamins and confounders in a multivariate regression model simultaneously. Normally distributed data are shown as means ± SD, non normal distributed data are shown as median (interquartile range), and categorical data as numbers (percentages). Student's t-test was used to compare continuous data and the ManneWhitney U-test was used for unevenly distributed variables. The chi-square test was used to compare categorical data and Spearman's rho or Pearson's correlation coefficients were calculated to examine correlations between continuous variables. Due to statistical power considerations and high correlations between vitamin concentrations, we initially built two binary logistic regression models. In model I, all statistically significant explanatory factors other than the vitamins of interest for this paper were included and then removed in a backward manner, removing the least significant variable first and keeping all variables at p 0.05. In this model we also included vitamin K1 and 25(OH)D as we have previously reported these to be a risk factor for hip fracture [1]. In model II, all vitamins studied in this paper were included. The variables were removed in a backward manner like in model I, in order to study which of the vitamins that were significantly associated with fracture risk independently from the other vitamins at stake. Finally, we built three logistic regression models, one for each of the vitamins that had demonstrated an independent and statistically significant association with fracture risk in model II. In these final models, we initially included all potential confounders that were independent and significant in model I, and then removed the variables in the same backward manner as described above. Before the multivariate models were built, positively skewed data (CRP, vitamin B6, folic acid and vitamin K1) were log transformed. BADL was highly skewed and was dichotomized at cut off ¼ 18. All p values are two-tailed. Statistical analyses were performed using SPSS 21 (SPSS Inc, Chicago, IL, USA) for Windows. 2.5. Ethics The Regional Committee for Medical Research Ethics, the Data Inspectorate, and the Directorate for Health and Social Affairs approved the study protocol. The study was performed according to the Helsinki Declaration of 1975 as revised in 1983. 3. Results 3.1. Patients' characteristics The controls (n ¼ 77) were predominantly home-dwelling, whereas 41% of the patient group (n ¼ 116) was institutionalized. Patients and controls were similar regarding age, sex and smoking habits. BMI was significantly lower in patients compared with controls (p ¼ 0.01). Patients had lower albumin and higher CRP concentrations (p < 0.001), used a higher number of medications (p < 0.001) and were more often totally abstinent from alcohol (p < 0.001) compared with controls. The patients had lower muscle strength (HGS, p ¼ 0.01) and reduced function (BALP, p < 0.001), compared with controls. Vitamins A, B6, C and E were significantly
3
lower in patients than in controls (all p < 0.001). There were no differences between patients and controls with respect to vitamin B12 or folic acid (Table 1). Missing variables are noted in the table. 3.2. Risk of hip fracture Initially, two separate backward regression models are presented in Table 2. Model I comprises all potential confounders and model II comprises all vitamins addressed in this paper. A strong association between vitamins A (p < 0.001), C (p < 0.001) and E (p ¼ 0.001) and the risk of hip fracture remained (model II), and was principally unchanged in the three final adjusted models (Table 3). A low BMI (p ¼ 0.03) and a BADL score ¼ 18 or below was also associated with increased hip fracture risk in the final multivariate models (Table 3). A predefined stratified analysis for institutionalized patients versus home-dwelling patients was also performed. Vitamin D and vitamin B6 were significantly lower in institutionalized patients compared with home-dwelling patients (p ¼ 0.02 and p ¼ 0.01, respectively). However, the odds ratios for hip fracture were almost identical in the two strata (data not shown). 3.3. Correlations between vitamins, bone turnover markers, physical function, and BMI The vitamins associated with hip fracture all showed low but statistically significant positive correlations with totOC, and negative correlations with BALP. The strongest correlation was that between vitamin C and totOC (p < 0.001). Vitamin B12 and Folic acid did not correlate with bone turnover markers. Daily functioning measured by the BADL Index correlated with all micronutrients. HGS and BMI were weakly or not correlated with micronutrients (Table 4).
Table 1 Characteristics of patients and controls.
Institutionalized Age (y)a Sex (% female)b BMI (kg/m2)a Albumin (g/L)a CRP (mg/L)c Number of medicationsa Handgrip strength (kg)a BADL Indexc Total abstainer from alcoholb Current smokersb Vitamin A (mmol/L)a Vitamin C (mmol/L)a Vitamin E (mmol/L)a Vitamin B6 (nmol/L)c Vitamin B12 (nmol/L)c Folic acid (nmol/L)c
Controls (n ¼ 73)
Patients (n ¼ 116)
p
47 (41) 82 ± 8 48 (66) 26 ± 4 42 ± 3 2 (1e6) 3±2 26 ± 9 20 (19,20) 34 (47) 10 (14) 2.0 ± 0.4 62 ± 20 34 ± 8 34 (24e65) 342 (247e491) 18 (13e30)
3 (4) 83 ± 9 87 (75) 24 ± 5 39 ± 4 6 (2e40) 5±3 22 ± 9 18 (15e20) 82 (71) 19 (17) 1.3 ± 0.5 30 ± 19 22 ± 8 17 (11e35) 352 (263e531) 15 (9e37)
<0.001 0.5 0.2 0.03 <0.001 <0.001 <0.001 0.01 0.01 <0.001 0.5 <0.001 <0.001 <0.001 <0.001 0.4 0.07
Data are missing for some variables: BMI, n ¼ 41 patients; Albumin, n ¼ 1 patient and n ¼ 1 control; CRP, n ¼ 3 patients and n ¼ 3 controls; number of medications, n ¼ 1 control; marital status, n ¼ 1 patient and n ¼ 2 controls; education, n ¼ 17 patients; HGS, n ¼ 18 patients and n ¼ 1 control; BADL, n ¼ 4 patients and n ¼ 2 controls; alcohol consumption, n ¼ 1 patient; smoking, n ¼ 7 patients; Vitamin analysis is missing for some patients: vitamin A, n ¼ 10; vitamin C, n ¼ 28; vitamin E, n ¼ 8; vitamin B6 ¼ 8, vitamin B12 ¼ 12 and folic acid, n ¼ 14. BADL ¼ barthel activities of daily living, CRP ¼ C-reactive protein, HGS ¼ hand grip strength. a Data are mean ± SD. Patients were compared with controls using the Student's t-test. b Data are n (%). Patients were compared with controls using the c2 test. c Data are median (IQR). Patients were compared with controls using the ManneWhitney U-test.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
4
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
Table 2 Risk factors for hip fracture, logistic regression models. Unadjusted models OR (95% CI) P Vitamin A (10 mmol/L) Vitamin B6, log transformed Folic Acid, log transformed Vitamin C (mmol/L) Vitamin E (mmol/L) Vitamin K1, log transformed 25(OH)D (nmol/L) Age (y) Sex, female BMI (kg/m2) Number of medications CRP, log transformed Albumin (g/L) HGS (kg) BADL score cutoff ¼ 18 Total abstainer from alcohol
0.75 (0.69e0.82) <0.001 0.3 (0.2-0.5) <0.001 0.4 (0.1e1.2) 0.09 0.93 (0.91e0.95) <0.001 0.83 (0.78e0.88) <0.001 0.2 (0.1e0.4) <0.001 0.96 (0.95e0.97) <0.001 1.01 (0.98e1.05) 0.50 2 (1e3) 0.11 0.9 (0.8e0.99) 0.02 1.3 (1.1e1.5) 0.001 1.5 (1.2e1.8) <0.001 0.8 (0.7e0.9) <0.001 0.95 (0.92e0.99) 0.01 6 (3e11) <0.001 3 (2e6) <0.001
Adjusted model I OR (95% CI) P
Adjusted model II OR (95% CI) P 0.74 (0.65e0.85) <0.001
0.95 (0.92e0.97) <0.001 0.89 (0.83e0.96) 0.001 0.38 (0.20e0.73) 0.01 0.97 (0.95e0.98) <0.001
0.88 (0.78e0.99) 0.03
0.84 (0.71e0.98) 0.03 3 (1e9) 0.01
The potential confounders at p < 0.1 form the univariate analyses were adjusted for, removing one variable at the time in a stepwise backward manner in the model, removing the least significant variable first and keeping the variable in the model if p < 0.05. Potential confounders were analyzed two models: background variables in multivariate model I, and vitamins addressed in this paper in multivariate model II. Statistically significant variables from models I and II were then added in the final models presented in Table 3. BADL ¼ barthel activities of daily living, CRP ¼ C-reactive protein, HGS ¼ hand grip strength, OR ¼ odds ratio.
Table 3 Risk of hip fracture, final models. Model with vitamin A OR (95% CI) p Vitamin A (10 mmol/L) Vitamin C (mmol/L) Vitamin E (mmol/L) 25(OH)D (nmol/L) Vitamin K, log transformed BMI (kg/m2) BADL <18
Model with vitamin C OR (95% CI) p
Model with vitamin E OR (95% CI) p
0.74 (0.65e0.84) <0.001 0.94 (0.92e0.97) <0.001 0.97 (0.95e0.99) 0.01
0.98 (0.96e0.99) 0.03
0.86 (0.76e0.97) 0.01 5 (2e15) 0.01
0.79 (0.68e0.92) 0.01 4 (1e11) 0.03
0.81 (0.74e0.89) 0.97 (0.95e0.99) 0.30 (0.13e0.68) 0.84 (0.73e0.98) 7 (2e25) 0.01
<0.001 <0.001 0.01 0.02
Statistically significant explanatory variables from models I and II (Table 2) were combined in three final models, one for each of the vitamins A, C and E. All the variables were initially forced into the model and then removed in a stepwise backward manner, removing the least significant variable first and keeping the variable when significant at p ¼ 0.05. BADL ¼ barthel activities of daily living, CI ¼ confidence interval, CRP¼C-reactive protein, HGS ¼ hand grip strength, OR ¼ odds ratio.
Table 4 Correlations between micronutrients, bone turnover, physical function, and BMI. TotOCa Vitamin Vitamin Vitamin Vitamin
A C E B6
***
0.28 0.41*** 0.24*** 0.25**
PTHb
BALP 0.22 0.212** - 0.151* 0.321***
BADL Indexa ***
1***
0.20**
0.28 0.39*** 0.25*** 0.35***
HGSa
BMI
*
0.16 0.19*
*** ¼ p < 0.001. ** ¼ p ¼ 0.01. * ¼ p ¼ 0.05. BADL ¼ barthel activities of daily living, BALP ¼ bone-specific alkaline phosphatase, HGS ¼ hand grip strength, PTH ¼ parathyroid hormone, totOC ¼ total osteocalcin. a Spearman’s rho. b Pearson's correlation coefficient.
4. Discussion We found that low levels of serum vitamins A, C, and E were associated with increased risk of hip fracture. The association remained statistically significant when adjusting for vitamin K1 and 25(OH)D as well as other potential confounders. Vitamin B6 was also associated with an increased risk of hip fracture when other confounders were adjusted for, but not when adjusting for the other vitamins. This indicates that a sufficient supply of all the vitamins A, C, D and E and perhaps vitamin B6 and K1 may be important to avoid hip fracture. Although vitamin D is an established mediator in
optimizing BMD and preventing fractures [20], our data suggests that supplementing with vitamin D alone may be insufficient in preventing hip fracture. Low BMI is associated with increased risk of hip fracture, but our findings indicate that low levels of vitamin A, C, E and 25(OH)D are independent risk factors, and associated with hip fractures in both underweight, normal and overweight patients. The patients were frailer than the controls, 40% being institutionalized. As expected, they had a lower BMI, used more medications, had a lower HGS, and scored lower in the BADL Index compared with controls. The vitamins A, B6, C and E, as well as the previously reported 25(OH)D and vitamin K1 [1], were all positively correlated with totOC and negatively correlated with BALP. This suggests that these vitamins may act directly in the metabolism of bone. Vitamin C showed the strongest correlation with totOC, suggesting that vitamin C may be more important for the production of the OC than previously assumed. Johnson et al. recently reported that the vitamins D and C synergistically increase OC promoter activity and synthesis of undercarboxylated osteocalcin (ucOC) [21]. This supports our assumption that vitamin C may act through up regulating the synthesis of OC. In vitro cell culture studies have shown bone growth, and an increase in synthesis of collagen I and OC in cells supplemented with vitamin C [22]. During carboxylation, vitamin K drives the reaction of inactive ucOC to active OC, and a vitamin K epoxide is formed [23]. Vitamin B6 is thought to be necessary for regenerating active
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
vitamin K [24]. Vitamins with anti-oxidative properties, such as vitamins A, C and E, might also be important for this process. The nuclear receptor for vitamin D forms heterodimers with the nuclear receptor X (RXR) in the promoter region for the gene that codes for OC, and may up regulate the synthesis of ucOC [25]. Vitamin A also acts through a nuclear receptor, retinoic acid receptor, which forms heterodimers with nuclear receptor RXR [15,25]. This indicates that vitamin A may act in synergy with vitamin D in bone mineralization. Based on the literature and our findings, we have suggested a mechanism of interaction between nutrients and the ability to mineralize bone, as shown in Fig. 1 [16,21,25,27]. This mechanism is so far speculative, and further research is required. High homocysteine has been associated with an increased risk of hip fracture [6,8,9]. Among the vitamins that take part in the homocysteine metabolism, vitamin B6 was lower in the hip fracture patients compared with controls. However, vitamin B6 was not associated with increased hip fracture risk when adjusting for vitamin C and vitamin A. This may be due to lack of statistical power as the variance for vitamin B6 was large. Review papers throughout the last decade have suggested that a high dietary intake of vitamin A is associated with an increased risk of fracture [15]. Barker et al. found a positive correlation between retinol and bone mineral density, but no relation between serum retinol and BALP or bone resorption [14]. The discrepancy between our results and those of Barker et al. may partially be explained by the fact that serum retinol in the cases in their study was at the same level as in our non-fractured control group. A considerably lower vitamin A concentration, as in our cases, may be associated with high bone turnover and compromised bone quality. In a national survey, vegetables were found to be the best contributor of vitamin A in the Norwegian diet, and vitamin A is reported to be adequate in the general population [26]. However, a large proportion of our patients had vitamin A deficiency or
5
inadequate vitamin A status as defined by serum retinol concentrations <0.7 mmol/L and <1.05 mmol/L, respectively. No patients had serum retinol concentrations above 4.2 mmol/L, which is regarded as the upper limit for serum vitamin A [15]. The patients in our study used a greater number of medications than did controls. This obviously reflects a higher degree of disease and inflammation in patients, which may have also caused increased oxidative stress and contributed to the association with hip fracture. However, we adjusted for the number of prescription medications used, and the association of an increased risk of hip fracture in patients with low vitamin A, C, and E concentrations remained unchanged. This association also remained unchanged when CRP, a proxy measure for inflammation, was adjusted for. Some authors have suggested that an effect of low vitamin C and E concentrations is only seen in smokers [3,27]. In our population, only 18% of patients and 14% of healthy controls were current smokers, and differences in smoking could not explain the association between low vitamin C, E and hip fracture. A limitation of our study is that we lack detailed data on comorbidity in the controls and we were unable to correct for comorbidity. Some medications may have influenzed on vitamin status i.e. warfarin. Five patients were using warfarin, but controls may also have been taken warfarin. Therefore, warfarin users were included in the analysis. The number of previous falls were not registered in neither cases nor controls, and a different risk of falling may in part explain the risk of a hip fracture in the two groups. However, the Norwegian Epidemiologic Osteoporotic studies, like our findings, suggest that low serum 25(OH)D and vitamin E is associated with an increased risk of hip fracture, whereas no association was found for vitamin A. That study did not report findings for the water soluble vitamins or vitamin K [28]. Height was measured using standing height in the controls, whereas knee height was measured in some (n ¼ 29) patients as a proxy measurement in those unable to stand. Knee height
Fig. 1. Proposed mechanism for the influence of vitamins upon bone mineral density (BMD). Bone mineral density (BMD) is dependent of the amount of calcium retained in bone, which is possibly dependent on several micronutrients: The bone-specific protein osteocalcin (OC) binds calcium to bone, but osteocalcin is formed in an inactive form, undercarboxylated osteocalcin (ucOC). Vitamin K acts as a catalyst in the activation to form carboxylated osteocalcin, the active form of OC. In this process, a vitamin K epoxide is formed. Vitamin B6 is thought to be necessary to regenerate vitamin K. The antioxidants, vitamins A, C,D or E, may also play a role. The promoter for the prepeptide ucOC contains a vitamin D responsive element and a vitamin C responsive element that regulate ucOC production. Vitamin A might regulate the promoter for the ucOC gene by forming heterodimers between RAR and RXR, which are the nuclear receptors for vitamin A and 1,25(OH)D, respectively [15,24,25,29]. RAR ¼ retinoic acid receptor, RXR ¼ receptor X receptor, VDR ¼ vitamin D Receptor, VCRE ¼ vitamin C responsive element, VDRE ¼ vitamin D responsive element.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014
6
A.C. Torbergsen et al. / Clinical Nutrition xxx (2015) 1e6
underestimated height somewhat in the patient group (data from the OOT) and may have given false high BMI in the 29 patients measured using knee height. We were not able to obtain data on weightloss prior to fracture; a measure that indicates nutritional risk better than BMI and lack good data for risk of malnutrition may influence upon our results. In order to prevent issues around multivitamin supplementation, oral nutrition-, enteral- or parenteral support and intravenous fluids during hospitalization, blood was drawn as soon as possible after informed consent in the was given in the intensive care unit and prior to surgery. Case-control studies can only identify associations and not prove causality. Intervention studies are required to confirm the associations found in our study. 5. Conclusion Low vitamin A, C, and E concentrations were associated with an increased risk of hip fracture, also when 25(OH)D, vitamin K1 and BMI were adjusted for. The association is possibly mediated through bone turnover mechanisms. Supplementation with vitamins K and D may not be sufficient to optimize bone health and prevent hip fracture. Sources of support This study was supported by a research fellowship grant from Oslo University Hospital, Medical Clinic Norway, and the Sophies Minde Ortopedi Foundation (to ACT). Conflict of interest None of the authors declare a conflict of interest. Acknowledgments We thank Thomas Gundersen at AS Vitas, Kari Julien at the Hormone Laboratory, and Anne Hove at the Nutritional Laboratory, Oslo University Hospital, for performing the biochemical analyses. We also thank Elisabeth Fragaat, Grete Wang, Gunnhild Bærdal, and Gurkirpal Singh for participating in data collection, as well as Lien Diep for statistical advice. TBW, FF, KS, TB, and MM designed the research. FF and KS were the principal orthopedic surgeons in charge of including hip fracture patients and healthy controls, respectively. ACT and LOW conducted the research. ACT analyzed the data and wrote the manuscript. MM had primary responsibility for final content. All authors read and approved the final manuscript. References [1] Torbergsen AC, Watne LO, Wyller TB, Frihagen F, Stromsoe K, Bohmer T, et al. Vitamin K1 and 25(OH)D are independently and synergistically associated with a risk for hip fracture in an elderly population: a case control study. Clin Nutr 2015;34:101e6. [2] Levis S, Theodore G. Summary of AHRQ's comparative effectiveness review of treatment to prevent fractures in men and women with low bone density or osteoporosis: update of the 2007 report. J Manag Care Pharm 2012;18:S1e15. [3] Lanham-New SA. Fruit and vegetables: the unexpected natural answer to the question of osteoporosis prevention? Am J Clin Nutr 2006;83:1254e5.
[4] Sahni S, Hannan MT, Gagnon D, Blumberg J, Cupples LA, Kiel DP, et al. Protective effect of total and supplemental vitamin C intake on the risk of hip fractureea 17-year follow-up from the Framingham Osteoporosis Study. Osteoporos Int 2009;20:1853e61. [5] Ahmadieh H, Arabi A. Vitamins and bone health: beyond calcium and vitamin D. Nutr Rev 2011;69:584e98. [6] Dhonukshe-Rutten RA, Pluijm SM, De Groot LC, Lips P, Smit JH, van Staveren WA. Homocysteine and vitamin B12 status relate to bone turnover markers, broadband ultrasound attenuation, and fractures in healthy elderly people. J Bone Min Res 2005;20:921e9. [7] Gjesdal CG, Vollset SE, Ueland PM, Refsum H, Meyer HE, Tell GS. Plasma homocysteine, folate, and vitamin B 12 and the risk of hip fracture: the Hordaland homocysteine study. J Bone Min Res 2007;22:747e56. [8] Leboff MS, Narweker R, LaCroix A, Wu L, Jackson R, Lee J, et al. Homocysteine levels and risk of hip fracture in postmenopausal women. J Clin Endocrinol Metab 2009;94:1207e13. [9] Sato Y, Honda Y, Iwamoto J, Kanoko T, Satoh K. Homocysteine as a predictive factor for hip fracture in stroke patients. Bone 2005;36:721e6. [10] Sato Y, Honda Y, Iwamoto J, Kanoko T, Satoh K. Effect of folate and mecobalamin on hip fractures in patients with stroke: a randomized controlled trial. JAMA 2005;293:1082e8. [11] Melhus H, Michaelsson K, Kindmark A, Bergstrom R, Holmberg L, Mallmin H, et al. Excessive dietary intake of vitamin A is associated with reduced bone mineral density and increased risk for hip fracture. Ann Intern Med 1998;129: 770e8. [12] Feskanich D, Singh V, Willett WC, Colditz GA. Vitamin A intake and hip fractures among postmenopausal women. JAMA 2002;287:47e54. [13] Opotowsky AR, Bilezikian JP. Serum vitamin A concentration and the risk of hip fracture among women 50 to 74 years old in the United States: a prospective analysis of the NHANES I follow-up study. Am J Med 2004;117: 169e74. [14] Barker ME, McCloskey E, Saha S, Gossiel F, Charlesworth D, Powers HJ, et al. Serum retinoids and beta-carotene as predictors of hip and other fractures in elderly women. J Bone Min Res 2005;20:913e20. [15] Conaway HH, Henning P, Lerner UH. Vitamin A metabolism, action, and role in skeletal homeostasis. Endocr Rev 2013;34:766e97. [16] Watne LO, Torbergsen AC, Conroy S, Engedal K, Frihagen F, Hjorthaug GA, et al. The effect of a pre- and postoperative orthogeriatric service on cognitive function in patients with hip fracture: randomized controlled trial (Oslo Orthogeriatric Trial). BMC Med 2014;12:63. [17] Chumlea WC, Guo SS, Wholihan K, Cockram D, Kuczmarski RJ, Johnson CL. Stature prediction equations for elderly non-Hispanic white, non-Hispanic black, and Mexican-American persons developed from NHANES III data. J Am Diet Assoc 1998;98:137e42. [18] Mahoney FI, Barthel DW. Functional evaluation: the barthel index. Md State Med J 1965;14:61e5. [19] Zannoni V, Lynch M, Goldstein S, Sato P. A rapid micromethod for the determination of ascorbic acid in plasma and tissues. Biochem Med 1974;11: 41e8. [20] Bischoff-Ferrari HA. Optimal serum 25-hydroxyvitamin D levels for multiple health outcomes. Adv Exp Med Biol 2014;810:500e25. [21] Johnson NA, Chen BH, Sung SY, Liao CH, Hsiao WC, Chung WK, et al. A novel targeting modality for renal cell carcinoma: human osteocalcin promotermediated gene therapy synergistically induced by vitamin C and vitamin D(3). J Gene Med 2010;12:892e903. [22] Urban K, Hohling HJ, Luttenberg B, Szuwart T, Plate U. An in vitro study of osteoblast vitality influenced by the vitamins C and E. Head Face Med 2012;8: 25. [23] Vermeer C, Knapen MH, Schurgers LJ. Vitamin K and metabolic bone disease. J Clin Pathol 1998;51:424e6. [24] Reynolds TM. Vitamin B6 deficiency may also be important. Clin Chem 1998;44:2555e6. [25] Sierra J, Villagra A, Paredes R, Cruzat F, Gutierrez S, Javed A, et al. Regulation of the bone-specific osteocalcin gene by p300 requires Runx2/Cbfa1 and the vitamin D3 receptor but not p300 intrinsic histone acetyltransferase activity. Mol Cell Biol 2003;23:3339e51. [26] Helsedirektoratet. Norkost 3. En landsomfattende kostundersøkelse blant menn og kvinner i Norge i alderen 18-70 år. 2011. [27] Melhus H, Michaelsson K, Holmberg L, Wolk A, Ljunghall S. Smoking, antioxidant vitamins, and the risk of hip fracture. J Bone Min Res 1999;(14): 129e35. [28] Sogaard AJ, Meyer HE, Emaus N, Grimnes G, Gjesdal CG, Forsmo S, et al. Cohort profile: Norwegian epidemiologic osteoporosis studies (NOREPOS). Scand J Public Health 2014;42:804e13. [29] Vermeer C, Theuwissen E, Vitamin K. Osteoporosis and degenerative diseases of ageing. Menopause Int 2011;17:19e23.
Please cite this article in press as: Torbergsen AC, et al., Micronutrients and the risk of hip fracture: Caseecontrol study, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.12.014