Osteoporosis in inflammatory bowel disease: effect of calcium and vitamin D with or without fluoride

Authors


Dr V. Abitbol, Service de Gastroentérologie, Hôpital Cochin, 27 rue du Faubourg Saint Jacques, 75014 Paris, France. E-mail: vered@club-internet.fr

Abstract

Background:

Previous data have indicated low bone formation as a mechanism of osteoporosis in inflammatory bowel disease. Fluoride can stimulate bone formation.

Aim:

To assess the effect of fluoride supplementation on lumbar spine bone mineral density in osteoporotic patients with inflammatory bowel disease treated in parallel with calcium and vitamin D.

Methods:

In this prospective, randomized, double-blind, parallel and placebo-controlled study, 94 patients with inflammatory bowel disease (lumbar spine T score below − 2 standard deviations, normal serum 25OH vitamin D), with a median age of 35 years, were included. Bone mineral density was measured by dual-energy X-ray absorptiometry. Patients were randomized to receive daily either sodium monofluorophosphate (150 mg, n=45) or placebo (n=49) for 1 year, and all received calcium (1 g) and vitamin D (800 IU). The relative change in bone mineral density from 0 to 12 months was tested in each group (fluoride or placebo) and compared between the groups.

Results:

Lumbar spine bone mineral density increased significantly in both groups after 1 year: 4.8 ± 5.6% (n=29) and 3.2 ± 3.8% (n=31) in the calcium–vitamin D–fluoride and calcium–vitamin D–placebo groups, respectively (P < 0.001 for each group). There was no difference between the groups (P=0.403). Similar results were observed according to corticosteroid intake or disease activity.

Conclusions:

Calcium and vitamin D seem to increase lumbar spine density in osteoporotic patients with inflammatory bowel disease; fluoride does not provide further benefit.

INTRODUCTION

Osteoporosis is a common complication of inflammatory bowel disease.1, 2 Its prevalence is about 30% in cross-sectional studies, depending on the studied populations and the technique of bone density measurement.3–5 Moreover, it has been shown that the incidence of fractures among patients with inflammatory bowel disease is 40% greater than in the general population.5 The pathogenesis of osteoporosis in inflammatory bowel disease is multifactorial; its mechanism has been studied in animal models and in histomorphometric and biochemical investigations in humans. A marked decrease in bone formation was observed in animals with severe trinitrobenzenesulphonic acid-induced colitis;6 with the healing of colitis, the bone volume increased and returned to control levels. In a population of 84 patients with inflammatory bowel disease, we observed a high prevalence of decreased levels of osteocalcin in the presence of normal levels of calcium-regulating hormones.3 This clinical approach confirmed previous histomorphometric data by Croucher et al., which showed a decreased bone formation rate and a negative remodelling balance.7 Together, these data indicate low bone formation as a potential mechanism of osteoporosis in inflammatory bowel disease in humans. On the other hand, increased bone resorption has also been shown.8, 9 This is a well-known phenomenon in inflammatory diseases, such as rheumatoid arthritis10 and spondyloarthropathies.11 Thus, osteoporosis in inflammatory bowel disease may result either from an uncoupling of bone cell activity, or from a sustained decrease in bone formation.

Corticosteroid therapy and disease activity via pro-inflammatory cytokines are the main factors responsible for low bone mineral density in patients with inflammatory bowel disease.12, 13 Prospective studies have shown a decrease of up to 5–10% of the initial bone density per year in the period of flare-up and corticosteroid therapy.14 Corticosteroids hamper intestinal calcium resorption and have a direct effect on the number and activity of osteoblasts. A decrease in bone density occurs rapidly, followed by a plateau.

Supplementation with calcium and vitamin D has been shown to maintain bone density in Crohn's disease.15 However, its efficacy has not been proven in the prevention of bone loss in recent corticotherapy and in the treatment of corticosteroid-induced osteoporosis.16–18

Fluoride is a pharmacological agent capable of stimulating bone formation through a direct effect on osteoblasts; it has been investigated in post-menopausal women19, 20 and in corticosteroid-induced osteoporosis,21, 22 with a beneficial effect on bone mineral density. This was the rationale for the study of Von Tirpitz et al., in a small population of patients with inflammatory bowel disease, which showed that fluoride had a positive effect on bone density.23 Therefore, we conducted a prospective, randomized, double-blind, parallel and placebo-controlled study to assess the effect of fluoride on the lumbar spine bone mineral density in osteoporotic patients with inflammatory bowel disease treated with calcium and vitamin D.

MATERIALS AND METHODS

This study is a multicentre, randomized, double-blind, parallel, controlled trial comparing the 1-year change in lumbar spine bone mineral density in osteoporotic patients with inflammatory bowel disease receiving either fluoride or placebo with calcium and vitamin D.

Patients

In this multicentre study, patients with inflammatory bowel disease referred to gastroenterological units were recruited. The diagnosis of Crohn's disease or ulcerative colitis was established from histological, endoscopic, radiological and clinical criteria before the study. Patients were eligible for entry into the study if they had a lumbar bone mineral density lower than two standard deviations below the mean value of young healthy adults (T score ≤ – 2 s.d.), as assessed by dual-energy X-ray absorptiometry. Exclusion criteria included levels of serum 25OH vitamin D below the normal range of the biochemistry laboratory, pregnancy or lactation, severe underlying disease, such as cancer or renal insufficiency, and evidence for disorders known to affect bone and mineral metabolism (excess alcohol consumption, osteomalacia, primary hyperparathyroidism, hyperthyroidism, primary Cushing's or Paget's disease). Patients were also excluded if they had abnormalities on the spinal radiographs that precluded accurate measurements of bone mineral density by dual-energy X-ray absorptiometry, or pre-existing osteoporotic femoral neck fracture. Patients were also excluded if they had received or were receiving treatment with calcium, vitamin D, sodium fluoride, bisphosphonates, calcitonin, oestrogens or other drugs, except corticosteroids, affecting bone metabolism directly.

Clinical assessment at baseline included the duration of the disease, ileal involvement established from endoscopic and histological criteria, site and length of digestive resections and body mass index. Physical activity was assessed according to the need for confinement to bed during the day. Clinical disease activity was evaluated by the Truelove and Best indices for ulcerative colitis and Crohn's disease, respectively. Hormonal status was clinically assessed by menstrual history. The cumulative lifetime steroid dose was expressed in grams of prednisone. The daily dose was calculated as the ratio between the cumulative dose and the duration of the disease. Other current therapies were documented. Serum concentrations of calcium, phosphorus, alkaline phosphatases, albumin, osteocalcin, C-reactive protein and parathyroid hormone and the erythrocyte sedimentation rate were measured at inclusion.

Trial design

Treatment and randomization. Patients were randomized per centre and per stratum according to corticosteroid intake in the last 3 months (stratum A) or not (stratum B) (block size, four). They received daily two tablets of either the trial drug or the control treatment designated as placebo for 12 months. Each effervescent tablet of the studied treatment (Yamanouchi Pharma, France) contained 75 mg of bi-sodium monofluorophosphate and 1250 mg of calcium carbonate (equal to 9.9 mg fluoride and 500.5 mg Ca2+ per tablet). Placebo tablets contained 1250 mg of calcium carbonate only. Treatment and placebo tablets were identical in appearance and consistency. The randomization code was kept in sealed envelopes until the end of the trial. All patients were asked to take a daily supplement of 800 IU colecalciferol (vitamin D).

Follow-up. After inclusion, each patient was seen at 3, 6, 9 and 12 months. Any deviation from the protocol was recorded. Patients were supplied with study medication and unused medication was given back to the physician. Compliance, expressed as the ratio of observed intake over expected intake, was calculated at the end of the trial from the tablet counts.

At each visit, the history of the previous 3 months, treatments and possible adverse events were reported. Bone mineral density measurements were performed at 6 and 12 months. Biological measurements were repeated at 6 and 12 months, except for serum parathyroid hormone which was measured at 6 months.

End-point. The primary end-point was the relative change in lumbar spine bone mineral density after 12 months. The secondary end-point was the relative change in femoral bone mineral density after 12 months.

Assuming that the annual relative change in lumbar spine bone mineral density follows a distribution with a standard deviation of 6.1% among patients with inflammatory bowel disease,14 the number of assessable patients to be included per treatment group is 38 in order to demonstrate, with a 90% power, an actual difference in the annual relative change of 5% between the groups (two-sided test with a 5% type I error). Assuming a 95% power efficiency of the Mann–Whitney test as compared to Student's t-test,24 and that 30% of the included patients would not be evaluated at 12 months for various reasons, 96 patients were planned to be included in the trial.

Bone mineral density and biological measurements

Bone mineral density measurements were performed by dual-energy X-ray absorptiometry at the lumbar spine (antero-posterior view) and the upper extremity of the left femur. Analyses of the scans were performed blind to the allocated treatment and followed a standardized procedure: at the lumbar spine, the mean bone mineral densities of the L2, L3 and L4 vertebrae were measured; at the upper extremity of the femur, the bone mineral densities of the femoral neck and whole region (`total hip') were assessed.

Serum concentrations of calcium, phosphorus, alkaline phosphatases, albumin and C-reactive protein and the erythrocyte sedimentation rate were measured by standard methods. Serum was stored (− 20 °C) for parathyroid hormone and osteocalcin assessment. These two measurements were performed at the end of the study, in a central facility blind to the allocated treatment, on the same batch for a given patient. Total parathyroid hormone (PTH 1–84) (normal range, 10–60 pg/mL) was measured by an immunoradiometric assay (Intact PTH, Nichols Institute, San Juan Capistrano, CA, USA). Osteocalcin (normal range, 5–25 ng/mL) was measured by a radioimmunoassay (Irma Cis-bio).

Statistical methods

Quantitative and qualitative variables were described by the mean ± s.d. (n) and frequency (n), respectively. The distribution of variables measured at inclusion was compared between the two treatment groups by the chi-squared test (Fisher's exact test, if necessary) for qualitative variables and the Mann–Whitney test for quantitative variables,24 globally and per stratum. The same tests were used to compare the distribution of variables at inclusion in patients who were assessable at 12 months and in patients who were not, globally and per treatment group. The proportion of non-assessable patients at 12 months and the distribution of these patients according to reasons for non-evaluation were compared between the treatment groups by the chi-squared test, globally and per stratum. The median relative change in bone mineral density from inclusion to 12 months was tested to zero by the Wilcoxon matched-pair signed-rank test25 within each treatment group, globally and per stratum. Distributions of relative changes in bone mineral density were compared between treatment groups by the Mann–Whitney test, globally and per stratum. The same tests were used when dealing with patients on corticosteroids during the trial. Results of the primary outcome were expressed as the 95% confidence interval (95% CI), assuming a normal distribution. All non-parametric tests performed on these outcomes globally and per stratum were confirmed by a multiple linear regression model,26 including treatment and stratum. The relationship between biological evaluations and bone mineral density measurements was tested by Spearman rank correlation (r).25 Finally, for these primary and secondary outcomes, a multiple linear regression model was used to adjust treatment comparison for inclusion factors which could be related to bone mineral density change, such as sex, age, ileitis, previous small bowel resection, treatment with corticosteroids (strata if within the previous 3 months) or immunosuppressive therapy and initial bone mineral density level. Backward selection was used,26 but treatment was forced to enter the model.

Ethics

The study protocol was approved by the Ethics Committee for Medical Research of the Cochin Hospital in Paris, France (September 1995), and the study was conducted in accordance with the 1975 Helsinki Declaration, as revised in 1983. Written and oral informed consent was obtained from all patients before inclusion.

RESULTS

Ninety-seven patients were enrolled in 16 centres between November 1996 and November 1998. In two patients, serum levels of vitamin D at inclusion were abnormal, and one patient had a lumbar T score > – 2 s.d. Thus, 94 patients (49 women), with a median age of 35 years (range, 18–68 years), participated in the trial. Among the women, 10 had been menopausal for a median duration of 12 years. The median duration of tobacco consumption was 12 years. Eighty-three patients had Crohn's disease and 11 had ulcerative colitis. Physical activity was normal in 87% of patients. Sixty-four patients were being treated with immunosuppressive agents and 46 with corticosteroids at inclusion. At baseline, parathyroid hormone and osteocalcin were assessed in 69 patients. The median parathyroid hormone value was 27 pg/mL (range, 4–105 pg/mL). The median osteocalcin value was 23 ng/mL (range, 1.2–58 ng/mL). The median serum concentrations of calcium, phosphorus and alkaline phosphatases were in the normal range.

Fifty-seven patients were allocated to stratum A (corticosteroids for the previous 3 months) and 37 patients to stratum B (not on corticosteroids for the previous 3 months). Forty-five patients were randomly assigned to fluoride therapy (28 in stratum A) and 49 to placebo (29 in stratum A). Six patients were not assigned to the correct stratum (three in each stratum).

The treatment groups were similar with regard to baseline characteristics, globally (Table 1) and per stratum. Bone characteristics and biochemical serum measurements at inclusion were not different between the treatment groups (Table 2). There was no difference between the two treatment groups in the distribution of albumin, erythrocyte sedimentation rate and C-reactive protein (data not shown).

Table 1.  . Baseline characteristics of the studied patients Thumbnail image of
Table 2.  . Bone characteristics and biochemical serum measurements at inclusion per treatment group, fluoride (n=45) or placebo (n=49) Thumbnail image of

Bone mineral density changes

Complete data from 60 patients (29 and 31 in the fluoride and placebo groups, respectively) were available for protocol analysis after 12 months. The distribution of patients with incomplete data at 12 months, according to reason, was similar in the two treated groups (fluoride and placebo) (Table 3).

Table 3.  .  Number of patients with incomplete data at 12 months according to reason and per treatment group Thumbnail image of

The relative changes in bone mineral density after 12 months are shown in Table 4. The median relative changes in lumbar spine bone mineral density increased significantly after 1 year in both groups, with mean relative changes (95% CI) of 2.7–6.8% and 1.9–4.6% in patients receiving fluoride–calcium–vitamin D and placebo–calcium–vitamin D, respectively (P < 0.001 for each group). However, these relative increases in bone mineral density were not statistically different between the two groups: difference in the mean relative change (95% CI) of − 0.9–4.0% (P=0.403). The relative changes in the femoral neck bone mineral density were not statistically different between the two groups (P=0.141). Similar results were observed for the relative changes in lumbar spine and femoral neck bone mineral density in patients in each stratum. Forty-five per cent of patients received steroids during this trial: in this sub-group, the relative changes (95% CI) in the lumbar spine bone mineral density were 0.3–8.0% (P=0.04) and 0.5–3.8% (P=0.02) in the fluoride and placebo groups, respectively, and there was no difference between the two treatment groups (P=0.360). Furthermore, 37% of patients had active disease at least once during follow-up from inclusion to 9 months: the relative change in lumbar spine bone mineral density after 12 months was no different between the fluoride and placebo groups in patients with and without active disease (P=0.356 and 0.602, respectively).

Table 4.  . Relative change (%) in bone mineral density from inclusion to 12 months according to the treatment group, globally and per stratum (A, corticotherapy in the last 3 months; B, no corticotherapy) Thumbnail image of

Bone markers

Alkaline phosphatases (mean ± s.d., n) increased significantly in the fluoride group (10 ± 29 IU, n=28) and decreased in the placebo group (− 22 ± 60 IU, n=30) (P=0.018). Median parathyroid hormone values decreased in the placebo group at 12 months (− 9 ± 20 pg/mL, n=22), whereas the variation observed in the fluoride group (4 ± 19 pg/mL, n=20) did not reach significance. There was no difference in osteocalcin distribution in the fluoride group compared to the placebo group at 6 and 12 months. The osteocalcin level at 6 months was not correlated with the relative change in lumbar spine bone mineral density at 12 months (r=0.261, P=0.073).

Compliance and tolerance

The mean compliance was over 94%. Compliance was similar in the two treatment groups, globally and per stratum.

Seventeen adverse events were reported in 15 patients: six in five patients in the fluoride group and 11 in 10 patients in the control group. Ten of the 17 adverse events were considered by the investigator to be drug related. Gastrointestinal symptoms (abdominal pain, diarrhoea) were the most frequent adverse events.

Validity of the results

In order to check the validity of the results obtained on two-thirds of the included patients, the baseline characteristics of the patients, assessed or not at 12 months, were compared. There was a higher proportion of patients on immunosuppressive therapy at inclusion (75% vs. 53%, P=0.029) and a higher lumbar spine bone mineral density (mean ± s.d.: 0.803 ± 0.059 vs. 0.759 ± 0.085 g/cm2; P=0.013) among assessed patients. There was a trend for a higher proportion of women (65% vs. 45%, P=0.066) and a shorter history of corticosteroid therapy (mean ± s.d.: 69 ± 66 vs. 96 ± 75 months; P=0.053) in the non-assessed patient group. Nevertheless, when comparing fluoride and placebo groups after adjustment on stratum, baseline bone mineral density, duration of corticosteroid therapy, sex and immunosuppressive therapy at inclusion, the differences between the treatment groups remained non-significant (P=0.216).

DISCUSSION

This 1-year, prospective, double-blind, parallel and placebo-controlled study shows, for the first time, that the lumbar spine bone mineral density can significantly increase during calcium and vitamin D therapy in osteoporotic patients with inflammatory bowel disease. Fluoride does not provide any further benefit.

Most patients included in this study were osteoporotic and had particularly severe inflammatory bowel disease: 67% were taking immunosuppressive therapy, 38% had undergone intestinal resection and 65% had small bowel disease. The severity of the disease may explain the difference between our results and those reported recently,23 which showed a benefit with fluoride in a small group of non-osteoporotic patients with Crohn's disease. Moreover, this trial was carried out without a placebo control group.23

During active inflammatory bowel disease, pro-inflammatory cytokines have direct effects on bone cell activity and bone turnover.13, 27, 28 Moreover, we have previously reported a spontaneous significant increase in lumbar spine bone mineral density in ulcerative colitis patients cured of their inflammatory disease after surgery and without corticosteroids.12 Therefore, disease activity may be a confounding factor in bone mineral density changes. In this study, the relative changes in lumbar spine bone mineral density were not different between the two treatment groups, in patients with and without active disease.

In corticosteroid-induced osteoporosis, published trials with calcium (500–1000 mg) and vitamin D (≤ 400 IU) failed to demonstrate efficacy in the primary prevention of bone loss.16 In patients with rheumatoid arthritis treated with low-dose corticosteroids (mean dosage, 5.6 mg/day), Buckley et al. showed that calcium and vitamin D may prevent bone loss.29 In patients with inflammatory bowel disease treated with corticosteroids, a previous study failed to show a significant effect at 1 year of prophylactic treatment with oral calcium (1 g/day) and vitamin D (250 IU/day).17 In fact, bone mineral density remained stable after 1 year of treatment. In a recent trial conducted with alendronate, no decrease in bone mineral density was observed in the placebo group receiving 400 IU vitamin D and a daily supplement of calcium carbonate.30 We observed a significant increase in lumbar spine bone mineral density in our placebo group in patients treated with higher doses of calcium (1000 mg/day) and vitamin D (800 IU/day). It should be noted that, in our previous work,14 the results of which were used to calculate the trial sample size, the annual relative change in lumbar spine bone mineral density was − 2.3 ± 6.1% in inflammatory bowel disease patients receiving no fluoride, calcium or vitamin D, whereas, in this trial, a significant increase was observed in inflammatory bowel disease patients receiving no fluoride, but taking calcium and vitamin D. Thus, we suspect that the pharmacological effect of calcium and vitamin D therapy may be related to an early minimal deficit in calcium and vitamin D despite normal levels of calcium-regulating hormones. Vitamin D status could be an environmental factor which, by shaping the immune system, affects the development of inflammatory disease. Cantorna et al. recently tested this hypothesis in an experimental animal model of inflammatory bowel disease: vitamin D supplementation blocked the progression and ameliorated the symptoms of inflammatory bowel disease, while vitamin D deficiency induced a wasting disease and death of the animals.31

We chose to test fluoride therapy because it can stimulate bone formation through a direct effect on osteoblast activity and because several authors have observed low bone formation in osteoporotic patients with inflammatory bowel disease.3, 6, 7 In this study, we observed an increase in alkaline phosphatase values in the fluoride group, indicating a positive effect on osteoblast activity. In post-menopausal women with vertebral fractures, there is no evidence that fluoride therapy has an anti-fracture effect. However, these patients have severe deterioration of both bone quantity and micro-architecture. One can speculate that, in younger patients, without fractures, an increase in bone mineral density may have a better benefit on bone strength.32[33]–34 In this study, small doses of fluoride were used to limit side-effects and because low doses of fluoride (37.5 mg) have been shown to be more effective than higher doses (75 mg).35, 36

Our study has some limitations due to the design, and the results should be interpreted in this context. We chose an observation period of only 1 year to optimize compliance with the trial. Nevertheless, 1 year may be too short, and a longer period (2 years) may be needed to demonstrate fluoride efficacy in osteoporotic patients with inflammatory bowel disease. Moreover, the aim of anti-osteoporotic treatment in patients with inflammatory bowel disease is to prevent fractures; bone mineral density change is only an intermediate criterion in this matter. Finally, our results were obtained on two-thirds of the included patients; however, we checked that the differences between assessed and non-assessed patients did not jeopardize our conclusions.

Bisphosphonates are effective in the prevention and treatment of corticosteroid-induced osteoporosis. However, patients with inflammatory bowel disease were excluded from the largest studies because of the uncertainty of the potential relationship between the gastrointestinal adverse events of the treatment and the symptoms of the disease. Recently, alendronate has been shown to increase spine bone mineral density in a small group of 32 patients with Crohn's disease in remission; only 13% of them were osteoporotic.30 Intravenous formulations are now available for some bisphosphonates, and may be of interest in patients with inflammatory bowel disease. The efficacy of hormone replacement therapy has been clearly shown in post-menopausal women in general,37 and specifically in post-menopausal women with inflammatory bowel disease.38 Small gains in bone mineral density at the hip and spine levels can be obtained by low-impact exercise programmes in patients with Crohn's disease, but compliance with the exercise programmes was rather disappointing.39

In conclusion, evidence is lacking for fluoride benefits in this population. We can recommend physiological doses of calcium and vitamin D in the treatment of osteoporotic patients with inflammatory bowel disease.

ACKNOWLEDGEMENTS

Thanks are due to M. Benayad for expert secretarial assistance, C. Geneix for expert technical assistance and to other Groupe d'Etudes Thérapeutiques des Affections Inflammatoires Digestives (GETAID) investigators: I. Sobhani (Paris), C. Florent (Paris), H. Lamouliatte (Bordeaux), J. L. Legoux and P. Potier (Orléans), J. F. Colombel (Lille), Y. Bouhnik (Paris), J. C. Delchier (Créteil), A. Bourreille (Nantes), C. Cellier (Paris), R. Bader (Mulhouse).

This work was supported by a grant from the French National Society of Gastroenterology (SNFGE) in 1996.

Ancillary