Vitamin D and K Status Influences Bone Mineral Density and Bone Accrual in Children and Adolescents with Celiac Disease

ORIGINAL ARTICLE Vitamin D and K status influences bone mineral density and bone accrual in children and adolescents with celiac disease

DR Mager1,2, J Qiao1 and J Turner1,2

1Department of Agriculture, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada and 2Department of Pediatric Gastroenterology and Nutrition, University of Alberta, Edmonton, Alberta, Canada

Background/Objectives: Children with celiac disease (CD) are at risk for decreased bone mineral density (BMD) because of fat-soluble vitamin malabsorption, inflammation and/or under-nutrition. The study objective was to determine the interrelationships between vitamin K/D status and lifestyle variables on BMD in children and adolescents with CD at diagnosis and after 1 year on the gluten-free diet (GFD). Subjects/Methods: Children and adolescents aged 3–17 years with biopsy proven CD at diagnosis and after 1 year on the GFD were studied. BMD was measured using dual-energy X-ray absorptiometry. Relevant variables included: anthropometrics, vitamin D/K status, diet, physical activity and sunlight exposure. Results: Whole-body and lumbar-spine BMD-z scores were low (pÀ1) at diagnosis (10–20%) and after 1 year (30–32%) in the children, independent of symptoms. Whole-body BMD-z scores (À0.55±0.7 versus 0.72±1.5) and serum levels of 25(OH) vitamin D (90.3±24.8 versus 70.5±19.8 nmol/l) were significantly lower in older children (410 years) when compared with younger children (p10 years) (Po0.001). Forty-three percent had suboptimal vitamin D status (25(OH)-vitamin D o75 nmol/l) at diagnosis; resolving in nearly half after 1 year on the GFD. Twenty-five percent had suboptimal vitamin K status at diagnosis; all resolved after 1 year. Conclusions: Children and adolescents with CD are at risk for suboptimal bone health at time of diagnosis and after 1 year on GFD; likely due in part to suboptimal vitamin D/K status. Therapeutic strategies aimed at optimizing vitamin K/D intake may contribute to improved BMD in children with CD. European Journal of Clinical Nutrition (2012) 66, 488–495; doi:10.1038/ejcn.2011.176; published online 5 October 2011

Keywords: vitamin K and D; bone health; children; celiac disease

Introduction in the absence of GI or indeed any symptoms (Turner et al., 2009). Screening for CD in the presence of risk factors, such Celiac disease (CD) is the most common autoimmune- as family history or type I diabetes, short stature, iron mediated gastrointestinal (GI) disorder; occurring in appro- deficiency or idiopathic osteoporosis is warranted. Our ximately 1/100 individuals (Fasano and Catassi, 2001). previous clinical experience indicates that between 40 and Genetically susceptible individuals exposed to gluten pep- 55% of symptomatic and asymptomatic children with CD tides have an inappropriate immune response that damages had low bone mineral density (BMD) at diagnosis and that the small intestine mucosa (Fasano and Catassi, 2001). While recovery after 1 year of treatment on the gluten-free diet the classical presentation of CD is malabsorption and failure (GFD) was incomplete; a worrisome finding given that to thrive in an infant or toddler, more recently it has become critical bone accrual occurs during childhood and adoles- recognized that CD can be present in individuals of all ages cence (Bailey, 1997; Turner et al., 2009; Rajani et al., 2010). The etiology of poor bone health in childhood CD is likely Correspondence: Dr DR Mager, Department of Agriculture, Food and Nutritional multi-factorial; relating to inadequate intake (vitamin D and Science, Department of Pediatrics, 4-126 Li Ka Shing Centre for Research K, calcium), malabsorption, inflammation and other lifestyle Innovation, University of Alberta, Edmonton, Alberta, Canada T6G OK2. variables (weight-bearing physical activity, medication E-mail: [email protected] Received 24 June 2011; revised 16 August 2011; accepted 29 August 2011; use, sunlight exposure). The role of vitamin D deficiency in published online 5 October 2011 the etiology of low BMD in adults with CD remains Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 489 controversial. Some studies report vitamin D levels to be in CD by the gastroenterologist (JT) in the Multidisciplinary the normal range, however, clearly some individuals had Celiac Clinic at the Stollery Children’s Hospital, Edmonton, vitamin D levels that would be considered a compromise to Alberta, Canada, before intestinal biopsy were recruited bone health (Ciacci et al., 1997; Mautalen et al., 1997; Lerner over a 10- to 12-month consecutive period. Out of these et al., 2011). Secondary hyperparathyroidism, presumably 54 patients, 52 children and their responsible caregivers due to intestinal malabsorption of calcium and/or vitamin agreed to participate. Only those with intestinal biopsies D, seems to have an important role in the pathogenesis of (n ¼ 43) diagnostic of CD, according to modified Marsh bone loss in CD (Pazianas et al., 2005). One of the larger criteria (Marsh, 1992; Oberhuber et al., 1999), were enrolled prospective studies of bone density in adults, conducted over 3 in this study at time of diagnosis and after 1 year (n ¼ 33) on years, showed more than one-quarter of subjects had second- the GFD (January–November, 2009–2010). Children with a ary hyperparathyroidism at baseline, correlated with BMD, history of co-morbid conditions known to influence vitamin and this group was least likely to recover BMD at follow-up K status (pancreatic insufficiency, short gut syndrome, (Valdimarsson et al., 2000). They also tended to have low hematologic or bone disorders) or to affect bone metabolism vitamin D levels and some persistent intestinal damage despite including endocrine, GI, renal or rheumatoid disorders (for reported compliance with a GFD. Ongoing minor intestinal example, type I diabetes, hyperparathyroidism, renal/liver damage, possibly from inadvertent gluten exposure, resulting disease) and/or taking medications known to interfere with in persistent calcium and vitamin D malabsorption and either vitamin K (for example, warfarin) or bone metabolism secondary hyperparathyroidism, is clearly a risk factor for (for example, phenytoin, non-steroidal anti-inflammatory low BMD in CD. In addition, it must be considered that peak drugs, oral glucocorticoids and oral contraceptives) were bone mass accrual may have been compromised because of excluded from this study. Informed consent/assent was early onset, potentially silent disease, during childhood. obtained from all subjects and/or their responsible caregivers The role of vitamin K deficiency in the etiology of low BMD before study entry. Ethics approval was obtained from in CD has not been considered. This is despite our knowledge the Human Research Ethics Board of the University of that a risk of vitamin K deficiency exists in CD (Cavallaro et al., Alberta (UA). 2004; Cashman, 2005). Vitamin K has an important role in bone health, being essential for the carboxylation of proteins Anthropometric variables including the bone matrix protein osteocalcin (Booth, 1997; Weight was measured by an upright scale (Sunbeam Cashman, 2005). In those at risk, vitamin K deficiency Products, Inc., Pelstar LLC, Alsip, IL, USA) to the nearest increases fracture risk, while supplementation is protective 0.1 kg. Height was measured, without shoes, by using a wall (Cockayne et al., 2006). There is increasing evidence that mounted stadiometer (Holtain Ltd, Crymych, Dyfed, UK) vitamin K has positive effects on calcium metabolism both to the nearest 0.1 cm. Body mass index was calculated for independent and synergistic with vitamin D and this may each patient according to the following equation body mass be very relevant for the inadequately explained and persi- index ¼ weight (kg)/(height (m))2. Weight-for-age, height- stent abnormalities in calcium metabolism observed in CD for-age, body mass index z-scores were calculated using (Koshihara et al., 1996; Kobayashi et al., 2002). software Epi Info 3.5.1 (Atlanta, GA, USA) using Centre for Prospective studies of bone mass accrual in children before Disease Control standards. Tanner staging was determined diagnosis of CD and subsequent to the initiation of a GFD according to standard criteria by the gastroenterologist are lacking. Current knowledge is based principally on (Tanner and Whitehouse, 1976). studies of adults and hence the pathogenesis of low BMD in CD in childhood remains incompletely understood. A few studies conducted in children with CD suggest that low BMD Laboratory variables occurs in untreated CD and while bone may normalize on a Blood was collected at baseline (at time of intestinal GFD, this is more likely the earlier a diagnosis was made endoscopy) and 1 year later for measurement of routine and the longer the duration of a GFD during childhood clinical blood work: 25(OH)-vitamin D, parathyroid hor- (Scotta et al., 1997; Kavak et al., 2003; Turner et al., 2009). mone (PTH), calcium, magnesium, phosphorus and anti- The objective of this study was to examine the potential tissue transglutaminase, and complete blood cell count after interrelationships between vitamin D and K status and bone an overnight fast. These measurements were performed in accrual and BMD at time of diagnosis and after 1 year on the the Core Laboratory at the UA Hospital according to GFD in children and adolescents with CD. standard methodologies. For purposes of analysis, any serum value of 25(OH)-vitamin D o50 nmol/l was considered deficient, 50–75 nmol/l as suboptimal and above 75 nmol/l Subjects and methods sufficient vitamin D status (Holick, 2007). An additional blood sample (B5 ml) was collected for assay of protein- Subjects induced in vitamin K absence-II (PIVKA-II) and bone In all, 54 children and adolescents (35 females and 19 males) turnover markers (bone-specific alkaline phosphatise (BAP), between the ages of 3 and 17 years were screened for risk of total osteocalcin and serum N-telopeptide type 1 collagen

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 490 (NTX)). Plasma PIVKA-II concentration (abnormal values consumption of vitamin supplements or for variations indicative of suboptimal vitamin K status 42 ng/ml) was in nutrient content of ‘brand-name’ products. Vitamin measured by an enzyme-linked immunosorbent assay kit K content of individual foods was determined from the (Diagnostica Stago, Asnie`res, France). The intra-assay and manufacturer food label, where available, and/or using the inter-assay coefficient of variation for this assay were USDA database. Assessment of physical activity (Adolescent 8.8% and 10.3%, respectively (Mager et al., 2006). Serum Physical Recall Questionnaire and FELS Questionnaire for concentrations of BAP (Metra BAP; Quidel, San Diego, CA, children aged 6–12 years) and sunlight exposure was USA), total osteocalcin (Quidel) and NTX (Osteomark NTX, determined using validated questionnaires at diagnosis and Princeton, NJ, USA) were measured using commercial after 1 year (Crocker et al., 1997; Termorshuizen et al., 2002; enzyme-linked immunosorbent assay kits. The intra-assay Treuth et al., 2005). and inter-assay coefficients of variation for these kits were 4.4% and 5.7%, 2.5% and 1.8%, and 6.2% and 14.1% for BAP, total osteocalcin and NTX, respectively. Statistical analysis Data are expressed as mean±s.d. Univariate and multivariate analyses were performed to assess relationships between BMD, bone mineral content (BMC) and bone age potential risk factors (age at diagnosis, presence of GI Whole-body and lumbar-spine BMD and BMC were symptoms, bone turnover, vitamin K, D, calcium status/ measured using dual-energy X-ray absorptiometry scan intake, weight-bearing activity, sunlight exposure) on by using Hologic QDR 4500A and Apex System 2.4.2 BMD and BMC. Variables shown to be associated with a (Hologic Inc., Walham, MA, USA). BMD was defined by low BMD was assessed using multivariate logistic regression World Health Organization standards (BMD-z score pÀ2 models to assess risk for development of poor bone health in indicative of osteoporosis) (Lewiecki et al., 2008). For children and adolescents with CD. Non-parametric tests purposes of analysis, BMD-z scores pÀ1 were defined as were utilized for variables demonstrating skewed distribu- increased risk for poor bone health. Bone age was measured tions. Analysis of covariance was performed to adjust for according to standardized methodologies (Greulich and any variables influencing primary outcome variables. Data Pyle, 1959). analysis was completed using the SAS 9.0 statistical software (SAS, Version 9; SAS Institute, Cary, NC, USA). A value of Po0.05 was considered significant. Lifestyle variables: dietary intake, physical activity and sunlight exposure Three-day food intake records (2 weekdays, 1 weekend day) Results were collected. Subjects and/or their responsible caregivers were instructed how to record portion sizes, food type on Anthropometric and demographic data standardized diet intake records. Micro- and macronutrient Anthropometric and demographic data are presented in intake was determined using Food Processor (SQL 10.6 ESHA Table 1. A majority of children were 10 years of age or less Research, Salem, OR, USA). Additional nutrient information (n ¼ 28 at diagnosis; n ¼ 17 at follow-up). Most children were from food labels were assessed to account for individual growing at age-appropriate rates at diagnosis and 1-year

Table 1 Anthropometric and demographic data

Variable Diagnosis (n ¼ 43)a 13M/30F* 1-Year follow-up (n ¼ 32)a 8M/24F*

Age (years) 9.4±4.2 (3 to 17) 10.7±4.4 (4 to 18) Height (cm)* 136.4±24.9 (98.6 to 191) 141.2±24.4 (106.4 to 191.2) Height for age z-score 0.11±1.18 (À2.87 to 2.95) 0.08±1.14 (À2.91 to 2.97) Weight (kg)* 36.0±20.9 (14.3 to 98) 40.1±21.8 (16.3 to 97.3) Weight for age z-score À0.08±1.26 (À5 to 2.68) 0.07±1.12 (À2.49 to 2.29) BMI (kg/m2) 17.7±4.4 (12.0 to 30.3) 18.6±4.7 (12.0 to 30.3) BMI z-score À0.17±1.26 (À4.27 to 2.03) À0.06±1.36 (À4.51 to 1.83) Tanner staging 2.0±1.6 (1 to 5) 2.3±1.9 (1 to 5) GI symptoms 20 4 Abdominal pain 21 3 Family history CD 9 — Growth concerns 11 1

Abbreviations: BMI, body mass index; CD, celiac disease; F, females; GI, gastrointestinal; M, males. aValues are shown as mean±s.d. (range). *Po0.01.

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 491 Table 2 Bone age, BMD and BMC at time of diagnosis and after 1 year on GFD

Variable Diagnosisa % Of low BMD (zoÀ1) Follow-upa % Of low BMD (zoÀ1)

Bone age 10.1±4.9 (n ¼ 31) 11.1±5.0 (n ¼ 25) Whole-body BMD z-score 0.14±1.24 (n ¼ 28) 10.7% 0.25±1.52 (n ¼ 24) 20.8% Whole-body BMD (g/cm2) 0.84±0.15 (n ¼ 28) 0.88±0.14 (n ¼ 24) Whole-body BMC (g) 1130 ±659 (n ¼ 28) 1249±648 (n ¼ 24) Lumbar-spinal BMD z-score À0.37±1.10 (n ¼ 31) 32.3% À0.34±1.06 (n ¼ 25) 32.0% Lumbar-spinal BMD (g/cm2) 0.67±0.21 (n ¼ 31) 0.71±0.21 (n ¼ 25) Lumbar-spine BMC (g) 30.9±19.2 (n ¼ 31) 33.4±20.2 (n ¼ 25)

Abbreviations: BMC, bone mineral content; BMD, bone mineral density; GFD, gluten-free diet. aValues are shown as mean±s.d.

Table 3 Biochemical parameters related to micronutrient status, bone turnover and celiac disease

Variable Diagnosisa (mean±s.d.; range) Follow-upa (mean±s.d.; range)

Total 25(OH)D (nmol/l)b 77±22 (38–119)a 88±27 (32–138)b PIVKA-II (ng/ml)c 1.55±0.86 (0.2–4.67)a 1.25±0.52 (0.36–2.72)b PTH (pmol/l)d 6.5±4.0 (1.4–23.2)a 3.6±2.0 (1.4–9.9)b Calcium (mmol/l)e 2.30±0.08 (2.11–2.45)a 2.38±0.07 (2.27–2.52)b Magnesium (mmol/l)f 0.85±0.04 (0.73–0.92)a 0.89±0.06 (0.79–0.99)b Phosphorus (mmol/l)g 1.5±0.18 (1.07–1.87)a 1.45±0.24 (0.92–1.95)b BAP (U/l)h 128.8±55.4 (22.8–258.1) 110.2±47.4 (15.8–179.8) Total osteocalcin (ng/ml)i 19.5±8.1 (4.0–47.9) 23.0±13.0 (5.9–67.8) j N-telopeptide of type 1 collagen (nM BCE/l) 85.7±61.7 (14.2–430.3)a 57.9±37.4 (15.7–177.3)b ATTG (U/ml)k 519±1019 (4.6–4400)a 40±161 (1–900)b

Abbreviations: ATTG, anti-tissue transglutaminase; BAP, bone-specific alkaline phosphatise; BCE, bone collagen equivalents; PIVKA-II, protein-induced in vitamin K absence-II; PTH, parathyroid hormone. aValues followed by different alphabets are significantly different at Po0.05. bNormal reference range: X80 nmol/l (4, 5); (n ¼ 34 at diagnosis; n ¼ 31 after 1 year). cNormal reference range: o2 ng/ml; (n ¼ 43 at diagnosis; n ¼ 30 after 1 year). dNormal reference range: 1.1–6.8 pmol/l; (n ¼ 38 at diagnosis; n ¼ 30 after 1 year). eNormal reference range: 2.1–2.6 mmol/l; (n ¼ 38 at diagnosis; n ¼ 31 after 1 year). fNormal reference range: 0.7–1.0 mmol/l; (n ¼ 38 at diagnosis; n ¼ 31 after 1 year). gNormal reference range: 0.8–1.9 mmol/l; (n ¼ 38 at diagnosis, n ¼ 31 after 1 year). hNormal reference range: 60–139 U/l (3–14 years); (n ¼ 43 at diagnosis; n ¼ 29 after 1 year). iNormal reference range: 10–100 ng/ml (1–18 years) (66); (n ¼ 43 at diagnosis; n ¼ 29 after 1 year). j Normal reference range: 6–215 nM BCE/l (34); (n ¼ 43 at diagnosis; n ¼ 30 after 1 year). kNormal reference range: o7U/ml(n ¼ 38 at diagnosis; n ¼ 31 after 1 year). follow-up (Tanner and Whitehouse, 1976). Only two children (P40.05). Younger children had significantly higher whole- had height-for-age and weight-for-age z-scores p2atdiagnosis; body BMD-z scores (p10 years; z: 0.7±1.4) than older resolving in one child after 1 year (Tanner and Whitehouse, children (410 years; z: À0.5±0.7) (Po0.001) at time of 1976). Approximately, 53% (n ¼ 23) were symptomatic at time diagnosis and after 1 year. No differences in lumbar-spine of presentation; all but three had complete resolution of GI BMD-z scores were observed between younger and older symptoms after 1 year on GFD (Po0.01). children (P ¼ 0.35). Only one patient had a lumbar-spine BMD-z score oÀ2 at baseline and after 1 year. This patient BMD, BMC and bone age had elevated anti-tissue transglutaminase levels and was Data regarding bone age, BMD and BMC are shown in positive for anti-endomysial antibodies. Table 2. Whole-body and lumbar-spine BMD (g/cm2) and BMC (g) increased over the 1-year follow-up (P40.05). Laboratory data related to vitamin status and bone Although the percentage of children having a low whole- turnover markers body BMD-z scores increased (10.7 to 20.8%), this was Biochemical data related to vitamin K, vitamin D, PTH, not a significant difference (P ¼ 0.78). No significant calcium, magnesium, phosphorus and bone turnover differences were noted in whole-body BMD/BMD-z/BMC or markers are presented in Table 3. Twenty-three percent lumbar-spine BMD/BMD-z /BMC between symptomatic and (10/43) of children had serum PIVKA-II concentrations non-symptomatic patients at diagnosis or after 1 year indicative of vitamin K deficiency (42 ng/ml) at diagnosis; (P40.05). No significant relationships were found between all normalized after 1 year. There were no significant the Marsh Criteria and these factors at diagnosis were noted differences in PIVKA-II levels between children with or

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 492 without GI symptoms or in younger (p10 years) versus Lifestyle variables: diet, physical activity and sunlight exposure older children (410 years) at time of diagnosis or after Diets were characterized by macronutrient distributions 1 year (P40.05). within the acceptable macronutrient distributions for age Suboptimal vitamin D status was observed in 43% (14/34) and gender (Table 4). In all, 90% (n ¼ 20) of patients did not children at diagnosis and in 25% (5/25) at 1-year follow-up have dietary intakes that met the estimated average require- (Po0.05). Only four children had 25(OH)-vitamin D levels ment for vitamin D for age and gender at diagnosis; none o50 nmol/l at diagnosis, seven at the 1-year follow-up. met the estimated average requirement at 1-year follow-up Serum 25(OH)-vitamin D was significantly higher after 1 (Ross et al., 2011). In contrast, 40.9% (n ¼ 9) and 31% (n ¼ 5) year, particularly in children that were asymptomatic at had vitamin K intakes that were o50% of the adequate diagnosis (79 versus 97 mmol/l) (Po0.05). Younger children intake at diagnosis and after 1 year; the remaining had (p10 years) had significantly higher concentrations of serum intakes approximating the adequate intake for age and 25(OH)-vitamin D than older children (410 years) at both gender (Otten et al., 2006). There was a trend toward time points (Po0.001). Seasonal variations in 25(OH)- increasing vitamin K intake on the GFD (P ¼ 0.06). Dietary vitamin D levels were not a significant confounder calcium intake was less than the estimated average require- (P40.05). Serum PTH was elevated in 43% (14/34) of ment in 64% of patients at diagnosis (n ¼ 14) and in 69% children at diagnosis; resolving in all but one child after 1 (n ¼ 11) after 1 year (Ross et al., 2011). There were no year. More than 50% (8/14) of these children, had serum significant differences in intake of these nutrients between 25(OH)-vitamin D concentrations o75 nmol/l. No signifi- younger (p10 years) versus the older children (410 years) cant relationships were found between whole-body and (P40.05). Although routinely advised at time of diagnosis to lumbar-spine BMD/BMD-z/BMC and serum concentrations supplement intake with a multivitamin supplement (con- of 25(OH)-vitamin D and PTH. taining at least 250 mg of calcium and 10 mg of vitamin D), Serum concentrations of BAP (bone formation) were only 5 out of 18 patients reported taking either a multi- elevated in 22/43 patients at diagnosis; all but 7 returned vitamin or single preparation containing vitamin D and/or to normal after 1 year (P40.05). In contrast, serum calcium after 1 year on the GFD. Supplementation with concentrations of NTX (bone resorption) decreased signifi- multivitamin containing vitamin D and calcium at time of cantly after 1 year; all were within normal reference ranges, diagnosis (n ¼ 4) and after 1 year on the GFD (n ¼ 5) did not except for one child at diagnosis (Po0.05). All children had result in significant changes in overall mean intakes of elevated levels of anti-tissue transglutaminase and were vitamins D, K or calcium. Physical activity as assessed by positive for anti-endomysial antibodies at diagnosis; only FELS and Adolescent Physical Recall Questionnaire and the two children had elevated anti-tissue transglutaminase levels number of reported sunlight hours did not differ signifi- after 1 year. Concentrations of the remaining laboratory cantly at time of diagnosis and at 1-year follow-up (P40.05) variables were within normal reference ranges. (Tables 5 and 6).

Table 4 Dietary intake in children with celiac disease at time of diagnosis and after 1 year on the gluten-free diet

Variable Diagnosisa (mean± s.d.; range) (n ¼ 22) % RDA/AI 1-year follow-upa (mean±s.d.; range) (n ¼ 18) % RDA/AI

Energy (kcal/day) 1813±680 (889–3541)a 95.9±25.2 1611±318 (1052–2096)b 84.4±24.9 Protein (g/day) 61.5±27.2 (29.6–142.2)a 215.3±95.4 56.7±10.8 (34.5–73.6)a 192.1±91.3 Vitamin D (mg/day)b 4.7±4.3 (0.1–15.4)a 31.1±28.8 3.2±2.3 (0.1–7.1)a 21.2±15.1 Vitamin K (mg/day)c 35.4±19.9 (6.5–96.3)a 57.3±27.5 49.0±23.3 (19.6–91)a 78.3±35.8 Calcium (mg/day)d 994±693 (753–3064)a 88±59 839±349 (209–1653)a 70.8±31.2 Magnesium (mg/day)e 93.8±69.5 (18.6–337.9)a 46.6±28.6 97.4±56.0 (21.3–201.1)a 45 ±27.3 Phosphorus (mg/day)f 499±235 (181–1184)a 70±44 415±213 (75–959)a 49±41 Sodium (mg/day)g 2752±1396 (1197–6204)a 160.4±53.8 2104±600 (1251–3275)a 123.8±68.3 Caffeine (mg/day)h 26.6±118.3 (0–556.2)a NA 3.4±9.2 (0–37.2)a NA

Abbreviations: AI, adequate intake; NA, not aplicable; RDA, recommended daily allowance. aValues followed by different alphabets are significantly different at Po0.05. bRDA: 15 mg/day. cAI: 30–75 mg/day; P ¼ 0.06. dRDA: 1–18 years, 700–1300 mg/day. eRDA: 1–18 years, 80–360 mg/day. fRDA: 1–18 years, 460–1250 mg/day. gAI: 1–18 years, 1000–1500 mg/day; P ¼ 0.065. hNo more than 45–85 mg/day for 7–12 years; 2.5 mg/kg body weight for 13–18 years (Health Canada: http://www.hc-sc.gc.ca/hl-vs/iyh-vsv/food-aliment/caffeine- eng.php#he).

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 493 Table 5 Physical activity in adolescents (413 Years, a) and in children growth and GI symptoms along with an evaluation of (3–13 years, b) at time of diagnosis and after 1 year lifestyle factors (diet, physical activity and sunlight expo- (a) sure) known to influence overall bone status in a population of children and adolescents newly diagnosed with CD. In Adolescents PARQ Diagnosis Follow-up (n ¼ 8) CD, there are several potential explanations for the etiology (n ¼ 11) of poor bone health. These include the potential for chronic malabsorption of nutrients known to influence bone forma- Vigorously active (METsX6.0) 4 2 Adequately active (3.5–5.9 METs) 5 4 tion (vitamins D, K and calcium), decreased dietary intake Inactive 2 2 because of the presence of significant GI symptoms (diar- rhea, anorexia) or the presence of chronic inflammation, (b) which is known to influence bone metabolism by depressing FELS PAQ for children Diagnosis Follow-up bone synthesis and upregulating bone resorption (Sylvester (n ¼ 24) (n ¼ 21) et al., 2007). Other factors known to influence overall bone health include reductions in endogenous synthesis in High-level sports (METsX8.0) 7 5 vitamin D because of poor sunlight exposure, particularly Medium-level sports 15 13 (4.5oMETso7.9) in those living in northern climates and changes in weight- Low-level sports (METsp4.5) 2 3 bearing activity (Roth et al., 2005). However, little is known about the extent to which all of these factors may influence Abbreviations: MET, metabolic equivalents; PARQ, Physical Recall Question- overall bone accrual in children and adolescents with CD, naire. particularly those with seemingly normal growth living in a northern climates (Roth et al., 2005; Genuis et al., 2009). Table 6 Number of sunlight exposure in children at time of diagnosis We hypothesized that overall bone health and bone accrual and after 1 year in children with CD would be directly influenced by vitamin Categories of Diagnosis (n ¼ 35) Follow-up (n ¼ 27) K and D status, independent of symptoms. time exposure Number of patients Number of patients In this study, we showed that approximately 10–30% of the children and adolescents diagnosed with CD have low 1 h or less 7 7 lumbar-spine and whole-body BMD-z scores at diagnosis and 1–2 h 13 7 after 1 year on the GFD. This occurred in a population of 2–4 h 9 7 4 h or more 6 6 relatively young children with seemingly normal growth and 1.8±1.6a 2.4±1.7a nutritional status, independent of the presence or absence of GI symptoms at time of diagnosis. Interestingly, the risk for aValues are shown as mean±s.d. poor bone health did not change after 1 year on the GFD, but was significantly higher in older children and adolescents Relationships between lifestyle variables and BMD and bone (410 years), indicating that time to diagnosis is an turnover markers important factor to bone health in CD (Motta et al., 2009; No relationships between vitamin D and calcium intake and Blazina et al., 2010). In addition, it shows that the length of markers of bone health (BMC, BMD/BMD-z) were observed time to recover suboptimal bone accrual in children with CD at diagnosis or after 1 year on the GFD; consistent with exceeds 1 year, even with compliance to dietary treatment. the over-all low intakes of vitamin D and calcium. This occurred in the presence of low dietary intakes of In contrast, increasing vitamin K (P ¼ 0.003) and protein vitamin K, calcium and vitamin D at diagnosis and after (Po0.001) intakes were correlated with increasing whole- 1 year in both the younger (p10 years) and older children body and lumbar-spine BMC at diagnosis and after 1 year. (410 years) despite no apparent differences in dietary intake These relationships were consistent with changes in bone of these nutrients in these age groups. While seasonal turnover markers: increasing BAP and decreasing NTX variations in serum vitamin D levels were not observed, this (Po0.05). Although our subjects reported age appropriate may have contributed to the higher prevalence of decreased levels of outdoor activities with coinciding weight-bearing BMD/diminished bone accrual in the older children. These activities, no significant relationships between these para- children had persistently lower serum levels of vitamin meters and the markers of bone health were detected (o70 nmol/l); all of which would have contributed to (Tremblay et al., 2011a, b). diminished bone accrual over the 1 year on the GFD. In addition, many of these children had elevated serum PTH levels, which would have also contributed to dimin- Discussion ished bone accrual. Although most children had vitamin D and calcium This study is a comprehensive study that included intakes that were significantly lower than recommended the assessment of BMD (whole-body/lumbar-spine), bone levels, no significant relationships between these nutrients turnover, anthropometrics, clinical variables related to and the markers of bone health measured were observed

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 494 (Roth et al., 2005; Ross et al., 2011). This is likely due to the The major limitations in this study is the inability to fact that intake of these nutrients did not significantly differentiate how some lifestyle factors (physical activity increase on the GFD and remained persistently low, a finding and sunlight exposure) specifically contributed to poor bone that is supported within the literature, particularly in health, given the smaller sample size. In addition, given the Canada where intakes of vitamin D are generally low in absence of a population control group it is challenging healthy children and adolescents (El Hayek et al., 2010; to differentiate between the lifestyle factors that may be Ohlund et al., 2010; Wild et al., 2010; Mark et al., 2011). contributing to poor bone health, independent of the CD; Although a few children were taking a vitamin supplement requiring further investigation. This is particularly (n ¼ 5), these were consumed on an ad hoc basis, and likely true of dietary vitamin D intake; however, the additional did not contribute sufficiently to overall vitamin D/calcium risk of malabsorption in the CD population highlights a need nutriture and bone health. Recent evidence indicates that to address this risk factor for the CD population. At the vitamin D insufficiency in CD is age dependent rather than current time, we do not know if this can be achieved by a related specifically to degrees of intestinal damage; GFD alone, certainly our study would suggest that is not with increased risk for insufficiency related to increasing the case. age, particularly in adults (Lerner et al., 2011). Our data are consistent with this finding, and points to the need for careful consideration for routine daily supplementation in Conclusion vitamin D and calcium in children with CD, even in those with apparent global nutritional adequacy. This is particu- Both asymptomatic and symptomatic children and adoles- larly evident in children and adolescents living in northern cents with CD are at risk for suboptimal bone health at climates like Alberta, where vitamin D insufficiency in the time of diagnosis and after 1 year, despite apparent age general population in this age group is thought to approx- appropriate rates of growth and compliance with a GFD. imate at least 30–40% of the population (Roth et al., 2005; Suboptimal vitamin D and K status may contribute to Genuis et al., 2009). Given that all of the children had an increased risk for poor bone health in childhood CD. dietary intakes of calcium and vitamin D intakes well below Furthermore, suboptimal dietary intake of vitamins D, the recommended levels, routine supplementation at time of vitamin K and calcium are common in this population, diagnosis of CD to at least the recommended daily allowance including when on a GFD. Careful consideration should be appears warranted in children and adolescents with CD given to routine supplementation of these nutrients at time (Ross et al., 2011). of diagnosis of CD. Vitamin D is a nutrient of particular There are limited available data that relate the role of concern, particularly in northern climates such as Alberta. vitamin K and bone health in children and adults with CD. Given that childhood and adolescence are a critical period However, the role of vitamin K in bone health in healthy for bone growth and deposition, it remains imperative adults and children has been extensively studied (Booth, 1997; that monitoring of bone health in children with CD is part Valdimiarsson et al., 2000). It has been demonstrated in a of routine clinical assessment. In order to maximize recent study that better vitamin K status is related to increased bone health for children, CD early diagnosis is likely whole-body bone mass in healthy pre-pubertal children to be important and requires awareness of atypical and (Pazianas et al., 2005). One study examined vitamin K status asymptomatic presentations; particularly as risk factors for in children newly diagnosed with CD using prothrombin poor bone health are equivalent regardless of symptoms times as a marker of vitamin K status and found that presentation. approximately 35% of children were deficient by this marker (Pazianas et al., 2005). However, this may have been an underestimate of the prevalence of vitamin K deficiency as prothrombin time is a very insensitive marker of overall Conflict of interest vitamin K status (Pazianas et al., 2005). More sensitive markers of vitamin K status include serum levels of PIVKA-II and/or The authors declare no conflict of interest. undercarboxylated osteocalcin, both of which are vitamin K- dependent proteins (Booth, 1997; Valdimiarsson et al., 2000). Although over 25% the children were vitamin K deficient at Acknowledgements diagnosis, this resolved in all children after 1 year. This appeared to be due in part to improvements in vitamin K The authors gratefully acknowledge the assistance of Leanne intake on the GFD. However, a remaining one-third of Shirton RN in data collection. Funding for this project was children and adolescents continued to have vitamin K intakes provided by the JA Campbell Award, Canadian Celiac considerably lower than the adequate intake on the GFD and Association, Food and Health Innovative Initiative, Univer- had whole-body BMD-z scores below minus 1; all pointing to sity for Alberta (JQ). We also wish to gratefully acknowledge suboptimal vitamin K status as a potential contributing factor the children and parents and/or caregivers for participating to persistent poor bone health in childhood CD. in this study.

European Journal of Clinical Nutrition Celiac disease; children; vitamin D/K; bone mineral density DR Mager et al 495 References Marsh MN (1992). Gluten, major histocompatibility complex, and the small intestine. A molecular and immunobiologic approach to Bailey DA (1997). The Saskatchewan pediatric bone mineral the spectrum of gluten sensitivity (‘celiac sprue’). Gastroenterology accrual study: bone mineral acquisition during the growing years. 102, 330–354. Int J Sports Med 18 (Suppl 3), S191–S194. Mautalen C, Gonzalez D, Mazure R, Vazquez H, Lorenzetti MP, Blazina S, Bratanic N, Campa AS, Blagus R, Orel R (2010). Maurino E et al. (1997). Effect of treatment on bone mass, mineral Bone mineral density and importance of strict gluten-free diet in metabolism, and body composition in untreated celiac disease children and adolescents with celiac disease. Bone 47, 598–603. patients. Am J Gastroenterol 92, 313–318. Booth SL (1997). Skeletal functions of vitamin K-dependent proteins: Motta ME, Faria ME, Silva GA (2009). Prevalence of low bone mineral not just for clotting anymore. Nutr Rev 55, 282–284. density in children and adolescents with celiac disease under Cashman KD (2005). Vitamin K status may be an important treatment. Sao Paulo Med J 127, 278–282. determinant of childhood bone health. Nutr Rev 63, 284–289. Oberhuber G, Granditsch G, Vogelsang H (1999). The histopathology Cavallaro R, Iovino P, Castiglione F, Palumbo A, Marino M, Di Bella S of coeliac disease: time for a standardized report scheme for et al. (2004). Prevalence and clinical associations of prolonged pathologists. Eur J Gastroenterol Hepatol 11, 1185–1194. prothrombin time in adult untreated coeliac disease. (See com- Ohlund K, Olsson C, Hernell O, Ohlund I (2010). Dietary ment). Eur J Gastroenterol Hepatol 16, 219–223. shortcomings in children on a gluten-free diet. J Hum Nutr Diet Ciacci C, Maurelli L, Klain M, Savino G, Salvatore M, Mazzacca G 23, 294–300. et al. (1997). Effects of dietary treatment on bone mineral Otten J, Pitzi Hellwig J, Meyers LD (ed) (2006). Dietary (DRI) Reference density in adults with celiac disease: factors predicting response. Intakes. Institute of Medicine, The National Academies Press: Am J Gastroenterol 92, 992–996. Washington, DC, USA. Cockayne S, Adamson J, Lanham-New S, Shearer MJ, Gilbody S, Pazianas M, Butcher GP, Subhani JM, Finch PJ, Ang L, Collins C et al. Torgerson DJ (2006). Vitamin K and the prevention of fractures: (2005). Calcium absorption and bone mineral density in celiacs systematic review and meta-analysis of randomized controlled after long term treatment with gluten-free diet and adequate trials. Arch Int Med 166, 1256–1261. calcium intake. Osteoporosis Int 16, 56–63. Crocker PR, Bailey DA, Faulkner RA, Kowalski KC, Mcgrath R (1997). Rajani S, Huynh HQ, Turner J (2010). The changing frequency Measuring general levels of physical activity: preliminary of celiac disease diagnosed at the Stollery Children’s Hospital. evidence for the physical activity questionnaire for older children. Can J Gastroenterol 24, 109–112. Med Sci Sports Exerc 29, 1344–1349. Ross AC, Manson JE, Abrams SA, Aloia JF, Brannon PM, Clinton SK El Hayek J, Egeland G, Weiler H (2010). Vitamin D status of Inuit et al. (2011). The 2011 report on dietary reference intakes preschoolers reflects season and vitamin D intake. J Nutr 140, for calcium and vitamin D from the Institute of Medicine: what 1839–1845. clinicians need to know. J Clin Endocrinol Metab 96, 53–58. Fasano A, Catassi C (2001). Current approaches to diagnosis and Roth DE, Martz P, Yeo R, Prosser C, Bell M, Jones AB (2005). Are treatment of celiac disease: an evolving spectrum. (See comment). national vitamin D guidelines sufficient to maintain adequate Gastroenterology 120, 636–651. blood levels in children? Can J Public Health 96, 443–449. Genuis SJ, Schwalfenberg GK, Hiltz MN, Vaselenak SA (2009). Scotta MS, Salvatore S, Salvatoni A, De Amici M, Ghiringhelli D, Vitamin D status of clinical practice populations at higher Broggini M et al. (1997). Bone mineralization and body latitudes: analysis and applications. Int J Environ Res Public Health composition in young patients with celiac disease. (See comment). 6, 151–173. Am J Gastroenterol 92, 1331–1334. Greulich W, Pyle S (1959). Radiographic Atlas of Skeletal Development of Sylvester FA, Wyzga N, Hyams JS, Davis PM, Lerer T, Vance K et al. the Hand and Wrist, 2nd edn, Stanford University Press: Stanford, (2007). Natural history of bone metabolism and bone CA. mineral density in children with inflammatory bowel disease. Holick MF (2007). Optimal vitamin D status for the prevention and Inflamm Bowel Dis 13, 42–50. treatment of osteoporosis. Drugs Aging 24, 1017–1029. Tanner JM, Whitehouse RH (1976). Clinical longitudinal standards Kavak US, Yuce A, Kocak N, Demir H, Saltik IN, Gurakan F et al. for height, weight, height velocity, weight velocity, and stages of (2003). Bone mineral density in children with untreated and puberty. Arch Dis Child 51, 170–179. treated celiac disease. (See comment). J Pediatric Gastroenterol Nutr Termorshuizen F, Wijga A, Garssen J, Den Outer PN, Slaper H, 37, 434–436. van Loveren H (2002). Exposure to solar ultraviolet radiation Kobayashi M, Hara K, Akiyama Y (2002). Effects of vitamin K2 in young Dutch children: assessment by means of a 6-week (menatetrenone) on calcium balance in ovariectomized rats. retrospective questionnaire. J Expo Anal Environ Epidemiol 12, Jpn J Pharmacol 88, 55–61. 204–213. Koshihara Y, Hoshi K, Ishibashi H, Shiraki M (1996). Vitamin K2 Tremblay MS, Leblanc AG, Janssen I, Kho ME, Hicks A, Murumets K promotes 1alpha,25(OH)2 vitamin D3-induced mineralization et al. (2011a). Canadian sedentary behaviour guidelines for in human periosteal osteoblasts. Calcified Tissue Int 59, 466–473. children and youth. Appl Physiol Nutr Metab 36, 59–64; 65–71. Lerner A, Shapira Y, Agmon-Levin N, Pacht A, Ben-Ami shor D, Tremblay MS, Warburton DE, Janssen I, Paterson DH, Latimer AE, Lopez HM et al. (2011). The clinical significance of 25OH-vitamin Rhodes RE et al. (2011b). New Canadian physical activity D status in celiac disease. Clin Rev Allergy Immunol; e-pub ahead of guidelines. Appl Physiol Nutr Metab 36, 36–46; 47–58. print 7 January 2011; doi:10.1007/S12016-010-8237-8. Treuth MS, Hou N, Young DR, Maynard LM (2005). Validity and Lewiecki EM, Gordon CM, Baim S, Leonard MB, Bishop NJ, reliability of the Fels physical activity questionnaire for children. Bianchi ML et al. (2008). International Society for Clinical Med Sci Sports Exerc 37, 488–495. Densitometry 2007 Adult and Pediatric Official Positions. Turner J, Pellerin G, Mager D (2009). Prevalence of metabolic bone Bone 43, 1115–1121. disease in children with celiac disease is independent of symptoms Mager DR, Mcgee PL, Furuya KN, Roberts EA (2006). Prevalence of at diagnosis. J Pediatr Gastroenterol Nutr 49, 589–593. vitamin K deficiency in children with mild to moderate chronic Valdimarsson T, Toss G, Lofman O, Strom M (2000). Three years’ liver disease. J Pediatr Gastroenterol Nutr 42, 71–76. follow-up of bone density in adult coeliac disease: significance of Mark S, Lambert M, Delvin EE, O’Loughlin J, Tremblay A, secondary hyperparathyroidism. Scand J Gastroenterol 35, 274–280. Gray-Donald K (2011). Higher vitamin D intake is needed to Wild D, Robins GG, Burley VJ, Howdle PD (2010). Evidence of high achieve serum 25(OH)D levels greater than 50 nmol/l in Quebec sugar intake, and low fibre and mineral intake, in the gluten-free youth at high risk of obesity. Eur J Clin Nutr 65, 486–492. diet. Aliment Pharmacol Ther 32, 573–581.

European Journal of Clinical Nutrition