Osteoporosis International

, Volume 16, Issue 1, pp 56–63

Calcium absorption and bone mineral density in celiacs after long term treatment with gluten-free diet and adequate calcium intake


    • Department of Medicine, Ralston Penn CenterUniversity of Pennsylvania
  • G. P. Butcher
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
  • J. M. Subhani
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
  • P. J. Finch
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
  • L. Ang
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
  • C. Collins
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
  • R. P. Heaney
    • Creighton University
  • M. Zaidi
    • Mount Sinai Bone Program, Departments of Medicine, Geriatrics and PhysiologyBronx VA Geriatrics Research Education and Clinical Center
  • J. D. Maxwell
    • Department of Gastroenterology, Endocrinology and MetabolismSt George’s Hospital and Medical School
Original Article

DOI: 10.1007/s00198-004-1641-2

Cite this article as:
Pazianas, M., Butcher, G.P., Subhani, J.M. et al. Osteoporos Int (2005) 16: 56. doi:10.1007/s00198-004-1641-2


Calcium malabsorption, hypocalcemia and skeletal demineralization are well-recognized features of untreated celiac disease. This study investigates calcium absorption and bone mineral density (BMD) after a prolonged, over 4 years, treatment with a gluten-free diet. Twenty-four adult females with treated celiac disease and twenty age- and sex-matched control subjects were studied. Mean body mass index (MBI), energy intake, serum calcium, and serum 25(OH)D concentrations in treated celiacs did not differ from controls. However, while both dietary calcium and protein intake were significantly higher in celiacs (P<0.012), fractional calcium absorption was lower (mean percentage±SD; treated 39.8±12 versus controls 52.3±10, P<0.001). Thus, after adjusting for calcium intake, the estimated amount of calcium absorbed daily was similar in both groups. Whole body, spine and trochanter BMD were significantly lower in treated celiac patients compared with controls (P<0.05). There were significant inverse correlations between: serum parathyroid hormone (PTH) and femoral neck or total body BMD (P<0.01), PTH and duration of gluten-free diet (P=0.05), and fractional calcium absorption and alkaline phosphatase (P=0.022). Increased calcium intake could potentially compensate for the reduced fractional calcium absorption in treated adult celiac patients, but may not normalize the BMD. In addition, the inverse correlation between PTH and time following treatment is suggestive of a continuing long-term benefit of gluten withdrawal on bone metabolism in celiac patients.


BMDCalcium absorptionCeliac disease25(OH)DPTH


Metabolic bone disease is a well-recognized feature of untreated celiac disease [1, 2, 3]. Many patients with celiac disease have reduced bone mineral density (BMD) at the time of presentation. Bone biopsy indicates that the low BMD is usually due to osteoporosis, but occasionally could result from osteomalacia [4, 5, 6, 7, 8, 9]. While gluten withdrawal rapidly improves small bowel histology and corrects nutritional deficiencies, the long-term effects of this treatment on fractional calcium absorption and BMD remain to be determined.

Treatment with a gluten-free diet in childhood appears to correct skeletal demineralization [10], but the results are less clear in adults. Several studies have shown that BMD in adults treated for celiac disease does not return to normal [8, 9, 10, 11]; this is probably due to suboptimal treatment [12]. Furthermore, adults followed from the time of diagnosis show an improvement (but not a normalization) in BMD maximally during the first year of treatment with a gluten-free diet [5, 8]. Other reports have indicated a normalization of BMD in those who adhere strictly to a gluten-free diet and who have had demonstrably normal small bowel villous architecture for many years [9].

Calcium malabsorption, accompanied by hypocalcemia in some patients, has been a common complication at the time of the diagnosis of celiac disease. However, only one study has reported on the effects of gluten-free diet on cation (strontium) absorption 1 year after treatment [13]. The aim of this study was therefore to investigate fractional calcium absorption and BMD status in adult female celiac disease patients treated for over 4 years with a gluten-free diet. We also assessed their dietary calcium intake, and their biochemical bone profile, including serum parathyroid hormone (PTH) and 25(OH)D levels.

Materials and methods

Patients and controls

The study was restricted to female subjects, as the method of fractional calcium absorption has been validated in women [14, 15]. Twenty-four Caucasian women attending a gastroenterology clinic with treated celiac disease (average age: 46 years, range: 25–74 years) and 20 healthy Caucasian female volunteers (average age: 48 years, range: 26–74 years) were studied. Local research ethics committee and ARSAC (Administration of Radioactive Substances Advisory Committee) committee approvals were granted.

The diagnosis of celiac disease was based on clinical presentation, demonstration of villous atrophy in distal duodenal or jejunal biopsies, and clinical and hematological response to gluten withdrawal. We estimated the number of years before the diagnosis of the disease was reached, by taking into account the number of years the patient had symptoms such as anemia, diarrhea and weight loss. Following the diagnosis of celiac disease and gluten withdrawal, patients subsequently had remained on a gluten-free diet for a mean duration of 4.7 years (SD: 3.1 years; range: 1.0–12.0 years). Repeat small intestinal biopsies obtained 3–6 months after initiation of gluten free diet demonstrated variable improvement in histology. However, we did not obtain further intestinal biopsies at the time of calcium absorption studies. None of the patients was on vitamin or hormone therapy. Control subjects matched for age were recruited from healthy hospital staff volunteers and medical outpatients. Exclusion criteria included the presence of any disorder or drug therapy known to affect bone metabolism, and alcohol consumption in excess of recommended limits (14 units per week). None of the patients or controls had significant skeletal symptoms or muscle weakness. The distribution of smokers, ex-smokers and non-smokers was similar between the two groups.

Weight (kg) and height (m) were recorded at the time of each BMD measurement in patients and controls. Body mass index (kg/m2) was calculated using weight (kg) and square of height (m2). An experienced dietician assessed the dietary intake of calcium, protein and total energy in celiac patients and normal controls using a comprehensive dietary recall technique in combination with a food frequency questionnaire completed for a 4-day period including a weekend day. Nutrient ingestion was then analyzed using a food database package (Comp-eat 4; Nutritia Systems, London).

Bone density measurement

BMD was measured by dual energy X-ray absorptiometry (DXA) using a Lunar DPX densitometer on the day of the calcium absorption study. BMD was measured at lumbar spine (L2–L4), femoral neck, trochanter and total body. Results are expressed as an absolute value (g/cm2) and as a T- and Z-score using normal standard values derived from a female British population. The reproducibility (CV%) of BMD measurements was 1% for the lumbar spine and whole body and 1.5% for the femoral neck and trochanter (M. Pazianas, unpublished data).

Fractional calcium absorption

This was performed after the method of Heaney [14, 15]. Following an overnight fast, a gluten-free breakfast was given which comprised two slices of gluten free toast, corn flakes and 150 ml of soya milk to which 0.6 Mbq of 45Ca had been added. The calcium content of the soya milk was fortified with tri-calcium phosphate and the calcium content was 140 mg per 100 ml. Controls and celiac patients underwent the fractional calcium absorption using exactly the same methodology with the same products, and therefore this would not affect the comparison of the results in this study. However, because the soya milk was externally labeled, the fractional absorption values could be artificially raised [16]. Therefore, absolute values for absorbability cannot be inferred from these data. Subjects were allowed 150 ml of water. Blood was drawn at 0 and 5 h and serum 45Ca determined by liquid scintillation counting (Wallac 1211 liquid scintillation counter). In four patients and seven controls, further samples were taken at 2 and 24 h to compare the absorption kinetics of the radiolabeled calcium. Empirical measures of isotope concentration were obtained after correcting for serum calcium concentration, height and weight [14].

Biochemical analyses

Fasting total serum calcium and alkaline phosphatase measurements were performed by standard methods (Technicon Dax analyzer). Calcium values were corrected for serum albumin using the formula: corrected calcium=measured calcium−0.02 (albumin g/l−43) mmol/l, with a normal range taken as 2.17–2.60 mmol/l. Fasting serum intact PTH levels were measured using a two-site immunoradiometric assay (Incstar Corp., Minn., USA), which employed two different polyclonal antibodies to detect the specific 1–34 and 39–84 regions of PTH. 25(OH)D was measured by a radioimmunoassay utilizing 125I-labeled as tracer (Incstar radioimmunoassay) [17]. In view of the seasonal variation of vitamin D, patients and controls were sampled at equivalent time points throughout the year. Serum samples were frozen for subsequent batch analysis of PTH and 25(OH)D.

Data analyses

Fractional calcium absorption results are expressed as percentages of the test dose. Statistical significance of differences between groups was calculated using the Mann-Whitney test. Linear regression was performed to examine the relation between variables. The correlation between different variables was determined using Spearman’s rank correlation. A probability level <0.05 was considered significant.


Nutritional assessment

There was no difference between treated celiac patients and controls in BMI. Energy consumption was similar in celiacs and in controls. However, mean calcium intake of celiac patients was significantly greater than that of controls (P=0.009). Celiac patients also had significantly greater estimated protein intake than controls (P=0.012). Finally, calcium intake was correlated inversely with the duration of the disease prior to its diagnosis and treatment (r=−0.551, P=0.033) (Table 1).
Table 1

Summary of biochemical and nutritional data




Mean (SD)


Mean (SD)


Serum calcium (mmol/l)

2.305 (0.12)


2.309 (0.06)



Serum alkaline phosphatase (IU/l)

67 (30)


65 (21)



PTH (pmol/l)

4.53 (2.1)


3.95 (0.95)



25(OH)D (nmol/l)

46 (38)


42 (21)



Body mass index (kg/m2)

22.7 (2.6)


24.3 (4.1)



Diet calcium (mg/day)

1129 (333)


829 (185)



Energy intake (kcal/day)

1830 (425)


2040 (363)



Protein intake (g/day)

86 (18.2)


68 (16.8)



Fractional calcium absorption (%)

39.8 (12.1)


54.7 (10.0)



*P<0.05, **P<0.01

Fractional calcium absorption (Table 2)

Isotope studies showed that the kinetics of distribution of radio labeled calcium at 2, 5 and 24 h was broadly similar in treated celiac patients and the control group (Fig. 1). Fractional calcium absorption, calculated on samples taken at 5 h post-ingestion of tracer, was significantly lower in celiac patients compared with controls (P<0.001) (Fig. 2). We should also underline the fact that the corrected serum calcium was practically identical in both groups (P=0.806).
Fig. 1

45Ca specific radioactivity (% of 5-h level) in controls and patients at 2 and 24 h post-ingestion of tracer

Fig. 2

Box whisker plot of fractional calcium absorption at 5 h post-ingestion of tracer in patients (mean percentage+SD; 39.8±12) and controls (52.3±10) (P<0.001). Boxes show the median and interquartile range, and whiskers the 10th and 90th percentiles. Outliers are also shown

There was a significant inverse relationship between fractional calcium absorption and alkaline phosphatase but no significant relationship (P>0.05) with BMD (at any site), age, BMI, histology at the time of the diagnosis of the disease, dietary calcium intake, energy, protein, years before treatment, or duration of gluten withdrawal (Table 2).
Table 2

Relationship between fractional calcium absorption and biochemical, BMD, clinical and dietary parameters

Fractional calcium absorption (%) in celiacs

Correlation coefficient


Serum alkaline phosphatase



25(OH)D in all celiac patients



25(OH)D in celiacs with FxCaAbs >30% and 25(OH)D <36 ng/m






BMD (g/cm2) at any site








Histology at the time of the diagnosis of the disease



Dietary calcium intake









Years before treatment



Duration of gluten withdrawal (years)




Biochemical measurements (Table 1)

Serum calcium (corrected) and alkaline phosphatase levels showed no significant difference between the groups. Mean serum PTH (1–84) was not significantly different in celiac patients compared with the control group. There was a strong inverse correlation between PTH and whole body BMD (Fig. 3) or femoral neck (r=−0.447, P=0.029), but in the spine it did not reach statistical significance (r=−0.345, P=0.099). Finally, there was a trend for lower PTH levels with increasing time since diagnosis, with a significant inverse correlation between PTH and duration of gluten withdrawal (r=−0.516; P=0.05). There was also a significant inverse relationship between alkaline phosphatase and 25(OH)D levels (r=−0.527, P=0.03), between serum calcium and dietary calcium intake (mg/day) (r=−0.591; P=0.005) or years of disease before treatment (Fig. 4).
Fig. 3

Scattergram showing an inverse relationship between total body BMD (g/cm2) and PTH (pmol/l) levels in celiac patients (r=−0.441; P=0.035)

Fig. 4

Scattergram showing the relationship between corrected serum calcium (mmol/l) and years before treatment (r=−0.746; P<0.001)

In view of the known seasonal variation in vitamin D levels, celiac patients and controls were sampled at equivalent time points throughout the year. The mean serum 25(OH)D level was not significantly different between the groups (P=0.79) (Table 1). There was no relationship between serum 25(OH)D and fractional calcium absorption (Table 2). However, analysis of the data of the subgroup of patients with 25(OH)D levels below the reported threshold of 32–36 ng/ml (80–90 nmol/l) [18] and fractional calcium absorption higher than 30%, the relationship became significant (P=0.012). Also, there was no relationship between serum 25(OH)D and BMD (any site: P>0.5).

Bone density measurements

Mean BMD values were lower for patients with celiac disease than controls in lumbar spine, trochanter, and total body but not in the femoral neck. There was a positive trend towards normal spinal BMD values with increasing time before the disease was diagnosed. This, however, did not reach statistical significance (r=0.422; P=0.081) (Table 3).
Table 3

Summary of bone densitometry data




Mean (SD)


Mean (SD)


BMD L2–L4 spine


1.038 (0.2)

0.658 to 1.338

1.142 (0.13)

0.952 to +1.465



−1.350 (1.6)

−4.51 to 1.15

−0.474 (1.06)

−2.07 to +2.21



−0.625 (1.1)

−2.21 to 1.37

−0.096 (1.2)

−1.61 to +3.20


BMD femoral neck


0.899 (0.145)

0.554 to 1.145

0.943 (0.16)

0.711 to 1.252



−0.569 (1.26)

−3.55 to 1.38

−0.306 (1.32)

−2.24 to 2.26



0.296 (0.83)

−1.76 to 1.61

0.249 (1.16)

−1.57 to 2.74


BMD trochanter


0.740 (0.11)

0.504 to 0.933

0.839 (0.126)

0.638 to 1.044



−0.46 (0.96)

−2.61to 1.30

0.443 (1.08)

−1.38 to 2.31



0.036 (0.77)

−1.54 to 1.54

0.638 (1.04)

−1.05 to 2.47


BMD whole body


1.035 (0.12)

0.780 to 1.25

1.142 (0.09)

0.960 to 1.36



−0.482 (1.07)

−4.38 to 1.58

0.211 (1.21)

−2.09 to 2.97



−0.497 (1.09)

−2.40 to 1.75

0.518 (1.09)

−1.26 to 3.20


*P<0.05, **P<0.01


Calcium malabsorption has been documented in untreated celiac disease [1]. Previous studies have provided evidence, albeit indirect, to suggest that calcium absorption remains impaired in a proportion of treated celiac patients. There is a single report of normalization of a strontium absorption test in a group of 18 female celiacs after 12 months on a gluten-free diet [13]. Beyond the authors’ interpretation of the results, careful analysis of their findings provides mechanistic insights into the disordered calcium absorption in the patients they studied. After 1 year on a gluten-free diet, mid-molecule PTH decreased significantly from 53.43 pmol/l to 23.72 pmol/l (normal range 11.5–77.1). Interestingly, during the same period, mean 1,25(OH)2D was elevated above normal; urinary 24-h calcium and phosphate was also increased by >50% as a direct effect of the changes in serum PTH and 1,25(OH)2D. It is not clear from the data included in the paper whether the patients were on 1,25(OH)2D treatment. However, based on these findings, treatment with the active vitamin D metabolite might be an interesting therapeutic approach in normalizing calcium absorption in patients with celiac disease. Since this publication, no other group has attempted to verify their findings or extend the observations beyond the first year of treatment. To our knowledge, our study provides the first evidence that calcium absorption is impaired in treated adult celiacs even after prolonged, more than 4 years, gluten withdrawal.

We used a method for fractional calcium absorption, initially developed by Heaney, that involved administering 45Ca in the diet followed by serial measurements of plasma radioactivity. Serial measurements between 7 and 24 h performed on over 500 females [14, 15] produced an empirical equation that predicts true calcium absorption from the 5-h isotope concentration. The original data set had included celiac subjects. Nevertheless, if fractional calcium absorption in treated celiacs is simply slower, rather than being reduced in absolute terms, the empirically derived equation may be a less accurate predictor. Our kinetic study of 45Ca absorption on randomly selected patients and controls demonstrated a similar distribution pattern.

The precise molecular mechanism of the lowered calcium absorption in celiac disease and its recovery (or its partial recovery thereof) following gluten-free diet is unclear. One mechanism for the transcellular transport of calcium across the apical enterocyte membrane is dependent upon vitamin D [19]. The traditional view is that steatorrhoea and malabsorption of fat-soluble vitamins results in the reduced bioavailability of vitamin D [20]. The recognition, however, that dietary vitamin D makes a negligible contribution to body stores in comparison with the contribution from endogenous photosynthesis makes this explanation less likely [21]. It is also clear that calcium malabsorption in celiac disease does not result from the absence of intestinal vitamin D receptors that are abundantly present in small intestinal crypts [22]. In our study, there was no correlation between fractional calcium absorption and 25(OH)D. However in the subgroup of patients with 25(OH)D values below the threshold of 32–36 ng/ml (80–90 nmol/l) and optimal fractional calcium absorption (>30%), the correlation was significant. This finding confirms recent report by Heaney’s group that estimated the above mentioned values as the 25(OH)D levels above which the response does not change with further increase in calcium intake [18]. Interestingly, this subgroup of our patients was not different from the overall celiac group in any of the other parameters analyzed in our study.

Transcellular calcium transport in addition depends upon calbindin-D9k, a calcium-binding protein located in the villous enterocyte. The protein has been found to be almost undetectable in untreated celiac disease and is present at a quarter of control values in treated celiac disease [23]. The explanation underlying this loss is likely to be persisting abnormalities of small intestinal villous architecture to a variable extent even in asymptomatic patients long after the initiation of treatment [12, 21, 24, 25]. As small intestinal biopsies were not performed at the time of BMD or calcium absorption studies, we are unable to relate our findings to small intestinal histology.

In addition to its transcellular movement, calcium is also transported paracellularly, and this depends not only upon the luminal calcium concentration, but also on the integrity of the intercellular junction. A change in villous architecture as seen in celiac disease may thus attenuate paracellular transport significantly, although these junctions have not been evaluated electron microscopically in celiac disease patients. Moreover, recently identified channel proteins, such as calcium transport protein (CaT1) in the rat [26], and the epithelial calcium channels (EcaCs)in the rabbit [27] and humans [28] that have been implicated in calcium transport across the apical membrane of the enterocyte, may be down regulated in celiac disease. Further studies are therefore required to characterize the molecular defect in calcium transport in celiac disease patients.

Our group of celiac patients appeared to have shown a satisfactory clinical response to a gluten-free diet by achieving a body mass index comparable with controls, as well as by attaining a normal hematological profile, serum calcium, alkaline phosphatase and 25(OH)D levels. Interestingly their calcium intakes were significantly greater than those of the controls, and in excess of the currently recommended dietary allowance in the UK. However, despite their reduced fractional calcium absorption, the estimated total amount of calcium absorbed daily was similar in both groups: 449 mg (0.398×1129 mg/day) in celiacs versus 453 mg (0.523×829 mg/day) in controls. In general, absorption fraction varies inversely as approximately the square root of the prevailing dietary intake. Therefore, one would expect a lower absorption fraction for an intake of 1100 mg per day than for an intake of 800 mg. However, it is questionable whether the difference in intake we report in our study is large enough to “entirely explain” the difference in absorption.

Our BMD findings support previous reports that low BMD is frequently present in treated and untreated celiac patients [2, 9, 12, 29, 30]. The decreased BMD in untreated celiac disease could be due to high bone turnover driven by secondary hyperparathyroidism as a result of negative calcium balance [6]. The increase in BMD after gluten withdrawal may thus be due to reduced turnover and a consequent (partial) correction in the mineral deficit [5]. Our finding of a strong inverse relationship between whole body BMD and PTH lends weight to the hypothesis that secondary hyperparathyroidism may be one of the main conditions affecting the bone metabolism in celiac disease. This is consistent with the compensatory changes in the PTH and 1,25(OH)2D levels reported previously [31]. It is also consistent with the high prevalence of fractures in the appendicular skeleton, as has been reported recently [32]. However, it is unlikely that the reduced BMD seen in our patient population could be attributed to hyperparathyroidism, as none of the patients had clinical or biochemical evidence of overt metabolic bone disease.

Treatment of celiac patients with a gluten-free diet for 1 year is associated with a significant increase in BMD, although this does not reach the BMD of normal controls. In one study, for example, lumbar spine BMD in untreated celiac patients compared with paired controls increased from 80.5% to 86.1% after 12 months on a gluten free diet. A further significant increase in BMD is considered unlikely and maximal improvement may be apparent by 2 years [5, 33]. In addition, the patients with the highest BMD improvement could be those who had suffered from the disease long enough for the thickness of their trabeculi to be reduced to the bare minimum without damaging the overall bone structure. This possibility could be strengthened by our finding of a positive, close to significant, correlation (P=0.081) between spinal BMD and the length of time lapsed before the disease was diagnosed. Nevertheless, the most critical period for the normalization of the BMD should be the time of the acquisition of the peak bone mass. Indeed, a complete restoration of BMD in young adult celiacs has been reported recently after 10 years of gluten-free therapy [34]. In the light of recent genetic studies, future studies are expected to explore whether the recently identified gene for a high [35] and low [36] BMD interacts with the molecular regulators of intestinal calcium absorption, and to what extent this interaction could be affected following damage of the intestinal structure.

The low BMD in our patients, in agreement with findings from most other studies, suggests that celiacs comprise a high-risk group for developing low bone mass. Detection of individuals with low BMD at the time when they are diagnosed with celiac disease should thus identify those who are at greatest risk of fracture. Therefore, routine BMD in celiac patients has been advocated [12]. The optimal timing of DXA scans has not been determined, but a case could be made for repeat DXA scans to be performed between 1 and 2 years following initiation of a gluten-free diet on those who have reduced BMD. Regarding the sites of BMD measurement, our finding of a strong correlation between PTH levels and total body BMD that confirmed previous relevant reports, indicates the need for measuring BMD at appendicular, cortical bone-rich, sites in addition to the axial, mainly trabecular, skeleton.

Apart from the continuing long-term benefit of gluten withdrawal on bone mineral metabolism, supported also by our finding of a significant inverse relation between PTH and duration of treatment, celiac disease patients should be on calcium supplements. Considering their reduced fractional calcium absorption, their daily dose should be at least 1200 mg per day. However, there may be cases requiring more rigorous measures, such as anti-resorptive agents, which may provide additional benefit.

In summary, adult celiac patients on adequate calcium intakes, despite their clinical remission in response to a gluten-free diet, have reduced fractional calcium absorption and BMD compared with control subjects. However, we do not interpret these findings as treatment failure. Instead, in line with previous longitudinal studies, it would appear that gluten withdrawal fell short of normalizing fractional calcium absorption and BMD.


We thank Dr. A Theodossi for permission to study his patients, and Dr. J.R.F. Walters for advice on study design and helpful discussion. M.Z. acknowledges support of the National Institute on Aging (NIH, RO1 AG14917-06) and the US Department of Veterans Affairs (Merit Review and GRECC).

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2004