Methods

Top Methods Results Discussion Author & Article Info References

Patients

Our study included white postmenopausal women with lumbar (L2 to L4) BMD of the spine below the 90th percentile of the distribution of BMD of the spine seen in Belgian women with osteoporosis [10, 11]. This degree of bone loss corresponded to a T-score of –2.5,in accordance with the operational definition of osteoporosis recently proposed by a World Health Organization study group [12]. Patients were included in the study regardless of whether they had previously had vertebral or nonvertebral fractures; most of the patients were thought to be free of vertebral fractures at enrollment. Previous hip fracture was an exclusion criterion. All patients were free of other causes of osteoporosis, such as diseases or medications known to interfere with bone metabolism; none had been treated with any drug for postmenopausal osteoporosis; and no such treatment was allowed during the study. Hormone replacement therapy was continued, for ethical reasons, in women for whom it had been prescribed before enrollment for purposes other than bone therapy. Randomization was not stratified with respect to hormone replacement therapy. Patients with bone diseases other than osteoporosis, renal insufficiency, hypochlorhydria, or severe chronic disorders that could have interfered with the study were excluded.

Study Design

Patients were randomly assigned in a blinded manner to one of two therapeutic groups. Every day for 4 years, they received either two chewable tablets that each contained 76 mg of MFP (10 mg of equivalent fluoride [fluoride ion]) and 1250 mg of calcium carbonate (500 mg of equivalent calcium) or two chewable tablets that each contained 1250 mg of calcium carbonate alone and were similar in appearance to the MFP-plus-calcium tablets. Total daily dosages, therefore, were 20 mg of equivalent fluoride plus 1000 mg of calcium in the MFP-plus-calcium group and 1000 mg of calcium in the calcium-only group.

The two tablets were taken at different meals. We determined compliance at each study visit by asking each patient for the number of days on which she had not taken the tablets and by counting the unused tablets. Compliance was expressed as the percentage of tablets taken (100% if the patient had taken all of the tablets).

Patients received randomization numbers sequentially. Randomization was computer generated in blocks of four according to a strict standard operating procedure by persons who had no contact with the persons in the center who assigned patients to study groups. The randomization code was kept at the study sponsor's facility under secure conditions that were detailed in writing. The clinical research center was given opaque, sealed envelopes, each of which contained the code for one patient. Treatment assignment and other relevant information were thus concealed and were to be revealed only in the case of a medical emergency.

Blinding was achieved by using the following procedures. First, the persons who did the visual readings of the spine radiographs saw the codes only after the results were analyzed. Second, the data were analyzed under blinded conditions: that is, a first analysis was done with groups "A" and "B"; the analysts did not know which group had received which treatment.

Efficacy Evaluation Criteria

The primary end point was the number of patients with new vertebral fractures during the 4-year treatment period, in accordance with recently published guidelines for the evaluation of drugs to be registered in Europe for the prevention or treatment of osteoporosis [13].

Standardized lateral radiographs of the thoracic and lumbar spine were obtained at enrollment and at each year of follow-up, for up to 4 years, in a single radiology center. The radiographs were sent to an independent assessor. Films were digitized, and the anterior, middle, and posterior heights of each vertebral body from the fourth thoracic (T4) to the fifth lumbar (L5) were determined (accuracy of the digitizer, 0.025 mm) by a computer program. This was done by persons with no knowledge of treatment assignment or film sequence. A new vertebral fracture (incident fracture) was defined as a reduction of at least 20% and an absolute decrease of at least 4 mm in any height of at least one vertebral body between enrollment and the latest follow-up film. All fractures, including borderline cases, were confirmed by visual reading. This definition was applied to vertebrae that were not fractured at enrollment, whereas fractures present at enrollment were determined on the digitized enrollment radiographs by using the Melton-Riggs 25% unadjusted algorithm [14]. The possible progression of such baseline lesions was assessed with the Vertebral Deformity Index obtained for each vertebra (T5 through L5); these were then summed to obtain the Spine Deformity Index, a continuous measure of vertebral deformities [15]. The Spine Deformity Index was also used as a secondary variable in patients with incident or progressing vertebral deformities, in whom it was expressed as the mean change between the last observed value and baseline. Bone mineral density was measured by using dual-energy x-ray absorptiometry on the same densitometer (Hologic QDR 1000, Waltham, Massachusetts) at 6-month intervals at the lumbar spine (L2 to L4) and the nondominant hip (total hip) after previously described and validated procedures were performed [11, 16]. In our hands, the long-term coefficients of variation of dual-energy x-ray absorptiometry are 0.8% for BMD of the spine and 1.1% for BMD of the total hip [17].

Biochemical determinations of bone remodeling were made at 6-month intervals. Bone formation was assessed by radioimmunoassay of serum bone-specific alkaline phosphatase (Ostase, Hybritech, San Diego, California). For bone resorption, we measured the ratio of urinary hydroxyproline to creatinine on the second fasting urine spot (2-hour morning urine) (Hypronostikon kit, Organon Technika, Oss Boxtel, the Netherlands).

All peripheral (nonvertebral) fractures that occurred during the study were recorded independently of the nature and severity of the trauma that may have determined them.

Statistical Analysis

All analyses were done according to the intention-to-treat approach: that is, all patients who had at least one valid measurement after randomization were considered in the analysis, whether they were still taking the study drug or not. In the case of drop-out and, thus, discontinuation of therapy with the study medication, the patient was invited to return to the clinic at annual intervals (for 4 years, if possible) so that the radiography necessary to record outcome could be done.

An exact-significance chi-square test was done to compare the number of patients with new vertebral fractures in the two groups. We calculated 95% CIs for vertebral fracture rates in each study group and for the difference in rates between the two groups, along with the point estimates of these rates. These rates were also expressed in terms of the number needed to treat for 4 years to prevent one fracture, including values for the lower and upper bounds of the 95% CIs. Changes in the Spine Deformity Index in patients with incident or progressing vertebral deformities were compared by analysis of variance.

Analysis of variance for repeated measurements was used to compare the absolute values for BMD of the spine over the course of the study in the two groups. Analysis of variance for repeated measurements was also done to compare BMD of the total hip and percentage changes in biochemical markers of bone remodeling throughout the study. All P values are two-tailed.

All statistical analyses were done with the SPS/WIN 6.2 statistical package (SPS, Inc., Chicago, Illinois).

Role of Study Sponsor

The trial was approved by the Ethical Committee of Liege University (registration no. 90/43-1262 of 14 May 1990), and all patients gave full informed consent before inclusion. The Rotta Research Group, which markets MFP and calcium combinations in Germany, Italy, and other countries, provided the drugs and funding for the study. Scientists from the Rotta Research Group were directly involved in the design, monitoring, and data management of the study and agreed to be listed as authors. However, the Rotta Research Group as a corporate entity had no control over the decision to approve or submit the manuscript for publication.

Results

Top Methods Results Discussion Author & Article Info References

Two hundred patients were enrolled in the study. The characteristics of the entire patient sample (100 patients were assigned to each group) at enrollment are shown in Table 1. The two groups were similar with regard to demographic data, baseline BMD of the spine, and biochemical markers of bone remodeling, and no differences were seen between the evaluable and the nonevaluable subpopulations. About 10% of patients in each group received hormone replacement therapy during the study; the overall proportion of patients who had had ovariectomy was 20% to 25% at enrollment. The following estrogen products and daily doses were used: 0.625 mg of oral conjugated estrogens, 50 µg of transdermal estradiol, 1.5 g of percutaneous estradiol, or 1 to 2 mg of oral estradiol valerate. In women with an intact uterus, a progestin was also used; no differences between groups or patients were seen in type and dosage of drug. The duration of hormone use in individual patients in the two groups is shown in Table 2. The average duration of hormone replacement therapy did not differ between the two groups. The study sample consisted mainly of patients without vertebral fractures because only 4 patients in the MFP-plus-calcium group and 3 patients in the calcium-only group had a single vertebral fracture at enrollment. Compliance with study medication was good and was identical in the two groups throughout the study; on average, evaluable patients in both groups took 87% of their tablets. It was not necessary to open the sealed envelopes that contained the randomization codes, and no envelopes were opened accidentally; this was verified at the end of the study and several times during the monitoring procedures. Therefore, the persons who actually conducted the intervention never had access to the randomization codes during the study. This was also true for those who read the dual-energy x-ray absorptiometry results and those who did the biochemical analyses.

Vertebral fracture rates were evaluated on the basis of analysis of the radiographs from 164 of the 200 patients (82%): 80 in the calcium-only group and 84 in the MFP-plus-calcium group. Overall, 36 patients were not included in the fracture rate analysis because of early drop out and no possibility of radiologic follow-up (30 patients) or because their radiographs could not be evaluated (6 patients). Fewer patients in the MFP-plus-calcium group (2 of 84; 2.4% [95% CI, 0.3% to 8.3%]) than in the calcium-only group (8 of 80; 10% [CI, 4.4% to 18.8%]) had new vertebral fractures (P = 0.05). The difference between the rates was 7.6 percentage points (CI, 0.3 to 15.0 percentage points), and the number needed to treat for 4 years to prevent one fracture was 13 (CI, 7 to 356).

Two patients receiving hormone replacement therapy (1 in the calcium-only group had been receiving it for <12 months and 1 in the MFP-plus-calcium group had been receiving it for <6 months) were among those who had a new vertebral fracture during the study. When patients receiving hormone replacement therapy were excluded from analysis, fewer patients in the MFP-plus-calcium group (2 of 71; 2.8% [CI, 0.4% to 9.8%]) than in the calcium-only group (8 of 69; 11.6% [CI, 5.2% to 21.6%]) had a new vertebral fracture. The difference between the rates was 8.78 percentage points (CI, 0.3 to 17.3 percentage points), and the number needed to treat for 4 years to prevent one fracture was 11 (CI, 6 to 334).

Patients had only single vertebral fractures, with the exception of one patient in the calcium-only group who had fractures of two different vertebrae at the same time point and one patient in the MFP-plus-calcium group who had two sequential reports of new fractures.

New vertebral fractures were identified in the calcium-only group after 12 months (5 fractures), 24 months (1 fracture), 36 months (1 fracture), and 48 months (1 fracture). In the MFP-plus-calcium group, these patients were identified after 24 months (1 fracture) and 48 months (1 fracture). However, the patient in the MFP-plus-calcium group who had the fracture after 48 months had discontinued treatment with MFP plus calcium after 24 months. All patients with new fractures had no fractures at baseline; no further fractures occurred in the few patients who had had one vertebral fracture at enrollment. The Vertebral Deformity Index/Spine Deformity Index analysis outlined no progression of prevalent fractures in these patients. Overall, the evaluable mean change in the Spine Deformity Index in patients with incident or progressing vertebral deformities throughout the study showed a nonsignificant (P = 0.14) difference between the two groups (0.1 ± 0.03 with MFP plus calcium compared with 0.2 ± 0.04 with calcium only in 28.6% and 33.3% of patients, respectively).

The results of BMD measurement are shown in Figure 1. Patients receiving MFP plus calcium had a progressive increase in BMD of the spine to 10.0% ± 1.5% above baseline at the end of the study; the calcium-only group had a change of –0.4%± 0.7% below baseline (P < 0.001). For BMD of the total hip, the final increase with MFP plus calcium was 1.8% ± 0.06%; this was not statistically significantly different from the increase seen with calcium only (0.7% ± 0.7%) (P = 0.07).

View larger version (10K): [in this window] [in a new window]   Figure 1. Changes in bone mineral density (BMD) of the lumbar spine (left) and total hip (right) during the 4-year trial. Data are expressed as mean percentage changes from baseline with 95% CIs. MFP = sodium monofluorophosphate.

The time course of biochemical markers of bone remodeling is shown in Figure 2. Serum bone-specific alkaline phosphatase significantly increased throughout the study in the MFP-plus-calcium group (to 11.5% ± 4.2% over baseline) and nonsignificantly decreased in the calcium-only group (change, –16.1%± 2.7%) at 1 year, decreasing to –2.3%± 4.0% at the end of the study (P = 0.002). Changes in the ratio of urinary hydroxyproline to creatinine varied and were not statistically significant in either group.

View larger version (12K): [in this window] [in a new window]   Figure 2. Changes in biochemical markers of bone turnover, expressed as mean percentage changes from baseline with 95% CIs. Left. Changes in serum bone-specific alkaline phosphatase (S-BAP). Right. Changes in the ratio of urinary hydroxyproline to creatinine (U-OHPr/Cr). MFP = sodium monofluorophosphate.

Peripheral fractures (Table 3) occurred with a similar incidence in the two groups. Of the 100 patients randomly assigned to receive calcium only, 11 had a total of 13 fractures during 324 patient-years of follow-up. Twelve patients receiving MFP plus calcium had a total of 15 fractures during 335 patient-years of follow-up. The difference between the groups was not significant. Hip fracture occurred in only one patient in each group.

Table 4 shows the number of and the reasons for drop out in the entire patient sample, including the early drop outs that accounted for most of the nonevaluable patients. The proportion of patients who discontinued treatment with the study drug was nearly identical in the two groups, and reasons for drop out were similar. The pattern of reasons for drop out did not differ in the patients who could not be included in the fracture rate analysis compared with the evaluable patient sample (data not shown).

Gastrointestinal problems were the most frequently reported adverse reaction in both groups. No differences were seen between the MFP-plus-calcium group and the calcium-only group with regard to the incidence, type, or severity of any adverse event. Lower limb pain was reported by 4.7% of patients receiving calcium only and 3% of patients receiving MFP plus calcium.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study shows a statistically significant reduction in the number of women with moderately low BMD of the spine who acquire new vertebral fractures when they receive continuous low-dosage (20 mg/d) fluoride, given as MFP, plus a calcium supplement. Similar results were previously reported [7] when 22.6 mg of fluoride per day, given as 50 mg of enteric-coated sodium fluoride, was compared with the therapies usually prescribed in France in women with at least one vertebral crush fracture.

More recently, 50 mg of slow-release sodium fluoride per day, given for 14-month cycles (12 months of sodium fluoride therapy followed by 2 months of no therapy), significantly reduced individual fracture rates, increased the fracture-free interval, and increased the time to new fracture in women with prevalent nontraumatic fractures [6].

In contrast, Riggs and colleagues [5] reported that 75 mg of plain sodium fluoride per day increased cancellous BMD but decreased cortical BMD at some skeletal sites without significant effects on fracture rates. Later, these results were reevaluated on the basis of the dose of sodium fluoride actually taken (and, therefore, the amount of fluoride absorbed). A bimodal effect of sodium fluoride was suggested, with low doses reducing fracture rates and higher doses (>60 mg/d) having a deleterious effect on bone structural resistance [18]. It should also be emphasized that the bioavailability of the plain sodium fluoride used by Riggs and colleagues [5] was 60% higher than that of the enteric-coated sodium fluoride used in the French study [7], with high plasma peaks of fluoride ion [3]. In another trial that was done in parallel using a similar study design, Kleerekoper and coworkers [9] concluded that sodium fluoride (75 mg/d) was no more effective than calcium only in retarding the progression of spinal osteoporosis.

In a recent study by the Fluoride and Vertebral Osteoporosis (FAVOS) study group [8], osteoporotic women treated with a regimen of fluoride, calcium, and vitamin D (800 U/d) had no reduction in vertebral fracture rates compared with women who received only calcium and vitamin D, a significant increase in BMD of the spine notwithstanding. However, the duration of treatment was shorter in this study (2 years) than in our trial. Stimulation of bone formation and secondary mineralization of recently synthesized matrix may require a prolonged period to ensure optimal biomechanical resistance, despite a rapid increase in BMD. Therefore, a significant reduction in fracture rate might be expected only after 3 to 4 years of treatment. The FAVOS study also targeted women whose osteoporosis was more severe (women with prevalent fractures) than that of the women in our study. Severe osteoporosis is related not only to a decrease in bone mineral content (which might be partly reversed by bone-forming agents) but also to a deterioration of bone microarchitecture (loss of structural integrity), which would not be expected to respond to such substances as fluoride [19]. Bone-forming agents might be particularly efficient in the presence of relatively mild to moderate osteoporosis, when the major damage to be corrected is related to a decrease in bone mineral. The FAVOS study also compared three different fluoride regimens-50 mg of enteric-coated sodium fluoride per day, 150 mg of MFP per day, and 200 mg of MFP per day-but was underpowered to separately evaluate the antifracture effects of each regimen.

The MFP formulation used in our trial was previously shown to be a highly soluble fluoride salt readily absorbed in the duodenum and the small intestine [20]. Skeletal uptake of fluoride after ingestion of MFP seems to be similar to or greater than that after ingestion of sodium fluoride [20, 21], and MFP has also been shown to be better tolerated by the gastric mucosa than sodium fluoride [22]. Furthermore, and in contrast to sodium fluoride, MFP is compatible not only with calcium salts but also with meals [23]. Thus, when it is given with meals, as it was in our study, bioavailability is not impaired but plasma concentrations are low and persistent [23]. This avoids the high peaks seen with plain sodium fluoride [3].

The low dosage of fluoride (20 mg/d) used in our protocol resulted in a moderate but steady increase in BMD of the spine (10% after 4 years), which paralleled the pattern of increase in serum bone-specific alkaline phosphatase (11.5% after 4 years) and is consistent with the known action of fluoride on bone formation. We suggest that such a slow but progressive increase in bone synthesis and bone density maximizes the chance that newly synthesized bone has the biochemical and structural properties needed to prevent further fractures.

This pattern of response may also prevent the secondary hyperparathyroidism, related to calcium deficiency, that has been reported to occur despite calcium supplementation in patients who had a rapid, fluoride-dependent increase in BMD of the spine [24]. We did not record any deleterious effect of MFP plus calcium on BMD of the hip; this is consistent with the results of other studies that used low-dose fluoride formulations [6-8] but not with the results of a trial of high-dose fluoride [5]. Finally, general tolerance to the drug and pain in the lower limbs were similar in the two treated groups.

Certain limitations of our trial should be acknowledged, including a possible confounding effect of hormone replacement therapy and the implications of the drop-out rate. However, the number of drop outs and the reasons for these drop outs were similar in the two groups, minimizing the likelihood of bias. Most drop outs occurred early (within the first months of the study), and they were primarily responsible for the 36 patients who could not be evaluated in the fracture rate analysis. Moreover, the remaining drop outs (those that occurred in the remaining 3 years of the 4-year trial) were reasonable (<30%) for such a long-term study, and these patients were included in the intention-to-treat analysis. Therefore, the drop-out rate is unlikely to have influenced the implementation of this regimen in practice because the difficulties encountered during the trial are similar to those likely to be found with any long-term treatment in current clinical practice. Finally, the small number of outcome events limits the strength of the inference about vertebral fractures, which makes it difficult to assess the effect of potential confounders and to draw meaningful conclusions about hip fracture rates.

When these limitations are taken into account, our data suggest that fluoride, given as MFP, plus a calcium supplement decreases the incidence of vertebral fractures compared with calcium alone in postmenopausal women with mild to moderate osteoporosis when it is given continuously in low dosages for a prolonged period.

Dr. Minne: Klinik der Furstenhof, Center of Endocrinology, Gust. Pommer e.v. Am Hylligen Born 7, D31812 Bad Pyrmont, Germany.

Drs. Rovati, Giacovelli, and Setnikar: Rotta Research Laboratorium, via Valosa di Sopra 7/9, 20052 Monza, Italy.