Faber J, Galløe AM. Changes in bone mass during prolonged subclinical hyperthyroidism due to l-thyroxine treatment: a meta-analysis. Eur J Endocrinol 1994;130:350–6. ISSN 0804–4643
l-Thyroxine (l-T4) in the treatment of thyroid disease resulting in reduced serum thyrotropin (TSH) has been associated with reduced bone mass and thus the potential risk of premature development of osteoporosis. However, several recent studies have failed to show such a detrimental effect. These disagreements are probably due to only a small number of patients taking part in each study, decreasing the change of finding significant differences and increasing the risk of missing a real difference (type 1 and 2 errors, respectively). We therefore performed a meta-analysis on the available papers (N = 13), in which bone mass was measured in the distal forearm, femoral neck or lumbar spine in a cross-sectional manner in women with suppressed serum TSH due to l-T4 treatment and in a control group. The women were divided according to their pre- and postmenopausal state, because preserved estrogen production plays a protective role against irreversible bone loss. Based on the number of measurements performed on the different sites of the skeleton, a theoretical bone composed of 30.4% distal forearm, 28.8% femoral neck and 40.8% lumbar spine could be constructed in premenopausal women (441 measurements). A premenopausal woman at an average age of 39.6 years and treated with 164 μg l-T4/day for 8.5 years, leading to suppressed serum TSH, had 2.67% less bone mass than controls (NS), corresponding to an excess annual bone loss of 0.31% after 8.5 years of treatment (NS). The risk of not detecting an excess bone loss of at least 1% per year (type 2 error) was p < 0.15. Similarly, a postmenopausal woman with a bone consisting of 11.3% distal forearm, 42.0% femoral neck and 46.7% lumbar spine (317 measurements) at an average age of 61.2 years and treated with 171 μg l-T4/day for 9.9 years had 9.02% less bone mass than controls (2p < 0.007), corresponding to a significant excess of annual loss of 0.91% after 9.9 years of treatment. Eighteen papers with a mean of 18 patients showing no difference between postmenopausal patients and controls would have to be published or found before this difference could turn into a non-significant finding (the file drawer problem). In conclusion, the meta-analysis on the available crosssectional studies did not find any significant reduction in bone mass during prolonged l-T4 treatment resulting in reduced serum TSH in premenopausal women. The risk of the present meta-analysis missing a clinically relevant annual loss of at least 1% in premenopausal women was less than 15%. In contrast, l-T4 treatment in postmenopausal women in a dosis leading to reduced serum TSH resulted in a significant excess of annual bone loss of 0.91%/year after 9.9 years in comparison to control women.
Jens Faber, Department of Endocrinology E, Frederiksberg Hospital, DK-2000 Frederiksberg, Denmark