Paracelsus

Paracelsus could have been talking about fluoride. At high doses, it can be lethal. At low doses (1 to 2 mg/d), it prevents dental caries. At intermediate doses (8 to 80 mg/d, which might be encountered in geographic regions with high fluoride levels in well water or in settings of industrial exposure to cryolite dust), skeletal fluorosis develops. Taking advantage of this osteosclerosing effect, which is toxic in a different setting, clinical investigators have used fluoride to increase bone mass in patients with osteoporosis for the past 30 or more years. Although it is now an approved treatment in several European countries, it may remain an investigational drug in the United States, mainly because, as a nonpatentable inexpensive agent, it has no commercial patron.

Recognizing that problem, the National Institutes of Health (NIH), in 1980, launched a two-center, 48-month trial of fluoride efficacy and safety. Only one center recruited enough participants to produce analyzable results and when the outcomes were reported [1], they were construed as negative. Fluoride, although it dramatically increased bone mass, failed to decrease spine fracture frequency and possibly increased the risk for extremity fractures [1]. The preparation used also produced gastric irritation, bleeding, and a peculiar periarticular pain syndrome in the feet and legs. By the time the trial began, problems were noted in the specified protocol: wrong dose (too high) and wrong chemical form (plain sodium fluoride rather than an enteric-coated or slow-release preparation) [2]. The gastrointestinal side effects observed in the trial were, therefore, not surprising.

However, the apparent lack of antifracture efficacy was a surprise because of the generally positive results from previous and still continuing open trials. This apparent difference in outcome was attributed to the NIH study's more rigorous design. In fact, however, the data were not that different. Although the authors did not emphasize it, the NIH trial data [1] showed a statistically significant decrease in spinal fractures in the treated group for the final 3 years of the trial. This finding was partially obscured by including first-year fractures in the comparison of the two treatment groups; the finding is even more impressive because the authors used an intention-to-treat analysis and because about one third of the "fluoride-treated" group had stopped receiving therapy.

This complicated history of fluoride treatment is necessary for understanding the study by Pak and colleagues [3], reported in this issue of Annals. In a rather unusual move, Pak and colleagues did a blinded interim analysis of their data more than halfway through a 56-month randomized trial using a smaller effective fluoride dose. The absolute dose was one-third less than had been specified in the NIH trial, and the slow-release formulation was probably slightly less well absorbed than plain sodium fluoride. The authors [3] reported that, as expected, vertebral bone mass increased by about 5% per year in the fluoride-treated group but, more importantly, that vertebral fractures decreased by more than 50%. Bone mass in the appendicular skeleton was maintained, and no hint was apparent of an increase in appendicular fractures. Finally, with the slow-release preparation, no difference was noted in gastrointestinal side effects between the groups.

The apparent antifracture efficacy of fluoride in the study by Pak and colleagues is thus congruent with the actual data of the NIH trial. It also confirms the finding by Farley and collaborators [4] of a strong correlation between an increase in bone mass and a decrease in fractures. In more than 300 fluoride-treated patients, they [4] found that those persons who had the greatest gain in bone mass had the fewest new fractures. Riggs and colleagues [5], in a reanalysis and extension of the NIH trial data, recently reported a similar pattern.

This is an important observation for two reasons. Because the analysis was confined to the treated group, it sidestepped the problem of treatment bias that always occurs with nonrandomized trials. Also, it directly addressed the concern that although axial bone mass unquestionably increased with fluoride therapy, the new bone might not be "good" bone. That notion was reinforced by the reported failure to decrease fracture frequency in the NIH trial. Although it is true that the high crystallinity of the mineral in fluoride-treated bone makes the structure somewhat more brittle, the data from the study by Farley and coworkers [4], the NIH trial [5], and the study by Pak and colleagues [3] all indicate that the net effect on bone strength is positive.

Pak and colleagues [3] appropriately stress the importance of co-therapy with adequate calcium. Whether the citrate preparation they used is better absorbed than other calcium sources, as they claim, is uncertain. Nevertheless, the osteogenic response to fluoride in the spine requires a great deal of calcium, and if it is not taken by mouth, the body removes it from the appendicular skeleton. Dure-Smith and colleagues [6], for example, found extraordinary bone hunger in patients successfully treated with fluoride. Obviously, the better the osteogenic response, the greater is the need for calcium. Some of the reported negative effects of fluoride on the appendicular skeleton probably resulted from inadequate calcium intake.

In this connection, the small increase in hip bone density found by Pak and colleagues [3] is also encouraging. Concerns have arisen that fluoride might increase the risk for hip fracture, although Kanis [7], summarizing reports on this matter, concluded that the aggregate relative risk was no different than 1.0 for this important potential side effect.

Because the number of persons with osteoporosis will approximately double during the next few decades, expanded treatment options are urgently needed. Estrogen and calcitonin are the only agents now approved in the United States for treatment of osteoporosis, and the approval status of calcitonin is currently under review by the Food and Drug Administration.

Although fluoride is probably useful mainly for the vertebral form of osteoporosis, it is the only agent with proven osteogenic potential certified for use anywhere in the world. The study by Pak and his colleagues [3] will probably not satisfy all the skeptics, but its positive findings and its strong investigative design should at least rekindle official interest in fluoride, interest that was dampened after the first report from the NIH trial [1]. Steps need to be taken to resolve any remaining uncertainties. If, as seems likely, fluoride offers promise, we need to make this inexpensive agent available to North American physicians. Yet another NIH-sponsored trial seems to be in order—this time with the correct dose and with a suitable slow-release preparation.