- Email: [email protected]
Bone mineral density in postmenopausal women treated with L-thyroxine
Bone Mineral Density in Postmenopausal Women Treated with L-Thyroxine E. VICTOR ADLIN, M.D., ALAN H. MAURER, M.D., ALLAN D. MARKS, M.D., BERTRAM J. CHANNICK, M.D., Philadelphia, Pennsylvania
PURPOSE: To determine if bone mineral density is decreased in postmenopausal women treated with l-thyroxine, and, if any decrease is observed, whether it is related to overtreatment with thyroid hormone, to deficiency of calcitonin, or to other factors. PATIENTSANDMETHODS: Thestudyconsistedof 19 postmenopausal women between 50 and 75 years of age treated with l-thyroxine for 5 years or longer, and 19 matching control subjects with no thyroid disease. Bone mineral density of the spine and hip was measured by dual-photon absorptiometry. Plasma calcitonin concentrations and serum thyroid hormone levels were determined by radioimmunoassays. RESULTS: The l-thyroxine-treated women had lower bone density in the lumbar spine (1.013 g/ cm2 [95% confidence interval, 0.945 to 1.0811 versus 1.134 g/cm2 [1.026 to 1.2421, p = 0.043); in the femoral neck (0.736 g/cm2 [0.694 to 0.7781 versus 0.809 g/cm2 [0.747 to 0.8721, p = 0.040); in Ward’s triangle (0.576 g/cm2 [0.530 to 0.6231 versus 0.694 g/cm2 [0.617 to 0.7701, p = 0.011); and in the trochanteric area (0.626 g/cm2 [0.581 to 0.6721 versus 0.722 g/cm2 [0.651 to 0.7941, p = 0.027). The maximal increase in calcitonin following calcium infusion was 1.37 rig/L (95% confidence interval, -0.44 to 3.17) in the l-thyroxine-treated patients versus 18.8 rig/L (95% confidence interval, 10.0 to 27.5) in normal women, p
supraphysiologic. Seven of the 19 patients had a history of hyperthyroidism in the distant past; these patients, considered separately, had significantly reduced bone density in the hip. The other 12 patients, considered separately, did not have a statistically significant loss of bone density. CONCLUSIONS: Long-term l-thyroxine therapy is associated with decreased density of the spine and hip. Since subclinical hyperthyroidism, decreased calcitoniu responsiveness, and a history of hyperthyroidism were demonstrated in some or all of these patients, these factors must be considered as possible causes of the decreased bone density.
T
hyroid hormone stimulates osteoclast activity and bone resorption [l], and hyperthyroidism, whether endogenous or exogenous, is a well-established cause of osteoporosis [2]. Recent observations have suggested that, additionally, clinically euthyroid patients treated with l-thyroxine, either as replacement therapy for hypothyroidism or suppressive therapy for thyroid nodules or goiter, have decreased bone density. Coindre and co-workers [3] found a reduction in trabecular bone volume in iliac bone biopsy specimens in hypothyroid patients given replacement doses of l-thyroxine and triiodothyronine .(Ts). Two additional studies have demonstrated decreased bone density, in the radius and hip, in premenopausal women receiving l-thyroxine [4,5]. In both these studies, the majority of patients, although clinically euthyroid, had an increased serum concentration of thyroxine (T4); the authors attributed the skeletal changes to thyroid hormone excess. Speculation has also centered on calcitonin deficiency as a potential cause of demineralization in patients with hypothyroidism. Calcitonin is a hormone produced by the parafollicular cells of the thyroid that acts to suppress bone resorption, but that has never been shown conclusively to have a physiologic role in the conservation of bone mass [6]. Calcitonin levels have been found to be decreased in patients with primary hypothyroidism
From the Department of Medicine (EVA, ADM, BJC) and the Department of Nuclear Medicine (AHM). Temple University School of Medicine, Philadelphia, Pennsylvania. Presented in part at the 63rd Meeting, American Thyroid Association, September 28, 1988, Montreal, Canada, and at the 21st European Symposium on Calcified Tissues, Jerusalem, Israel, March 13, 1989. This study was supported in part by a grant from the U.S. Public Health Service, Research Grant No. 2 MO1 RR 349 from the National Institutes of Health, General Clinical Research Centers Branch. Requests for reprints should be addressed to E. Victor Adlin, M.D.. Temple University School of Medicine, 3401 North Broad Street, Philadelphia, Pennsylvania 19140. Manuscript submitted April 26, 1990, and accepted in revised form October 1, 1990.
360
March
1991
The American
Journal
of Medicine
Volume
90
BONE DENSITY
[7] as well as in patients who have had a thyroidectomy [8,9] or radioiodine therapy [lo]. Decreased density of the radius in a group of thyroidectomized patients has been attributed to calcitonin deficiency [ll], but in another group of patients with low calcitonin levels following thyroidectomy, bone mineral density was not decreased [12]. Any effect of thyroid hormone excess or calcitonin deficiency on enhancing bone mineral loss might be expected to be most evident in postmenopausal women, a group that is especially susceptible to osteoporosis. We have studied bone density at the spine and hip in postmenopausal women treated with l-thyroxine for at least 5 years. We evaluated thyroid hormone levels and basal and stimulated calcitonin levels, to clarify the roles of these factors in any abnormalities of skeletal density that might be present in’these women.
AND L-THYROXINE
/ ADLIN
ET AL
TABLEI Characteristicsof Patientsand Controls Controls (n = 19) Age (years)*
61.6 ‘5yf;~o)
Race Height (cm)*
Patients (n=19)
Value ’
60.9 (5fp&’
9 white 161.3 (158$%;64.3)
9 white 161.0 (158%1863.3)
(65.&;5.5)
(62.:;977.6)
NS NS* NS§
Weight (kg)”
NS
Percent ideal body weight*
NS
Years since menopause* Bilateral oophorectomy Use of calcium supplement Use of thiazide diuretic Exercise program Smokers (ever)
W;;281
wy;32)
‘9.;7M;.4’
(9j2jly
6/19 10/19 5/19 6/19
NS§ NS NS* NS NS* NS*
4119 8/19 7/19 5119
PATIENTS AND METHODS
* Mean and 95% confidence intervals. + Wilcoxon rank sum test. t Fisher’s exact test. 3 Paired t-test.
Postmenopausal women between 50 and 75 years of age, who had been treated with l-thyroxine for 5 years or longer, were selected from an outpatient practice of endocrinology and internal medicine. We evaluated each patient seen in our practice during the time the study was in progress; all who met the inclusion and exclusion criteria were invited to participate. An equal number of control patients were selected from the same population by the same process, with the same inclusion and exclusion criteria; the only difference was that they had never had thyroid disease or treatment with l-thyroxine, and were selected only if they were similar in age and race to one of the l-thyroxine-treated patients, with whom they were matched. Patients were excluded if they had any condition known to affect bone density, such as diseases of the kidneys, parathyroids, or adrenals, or musculoskeleta1 disease that affected mobility. Patients were excluded if they had been treated with estrogen, glucocorticoids, anticonvulsants, or vitamin D. None of the patients or controls were known to have taken estrogens in the past except for one 68-yearold control patient who took Premarin from 1969 to 1972. Because only a small proportion of the patients in our practice population fulfilled the requirements of postmenopausal status, l-thyroxine therapy, and absence of estrogen therapy and the other exclusion criteria, the 19 l-thyroxine-treated patients and 19 control subjects were drawn from a much larger group of patients. Of the 19 control subjects, 14 were being treated for hypertension, one for reflux. esophagitis, and one for migraine headaches; three had no chronic disease. Bone mineral density of the spine and hip was
measured by dual-photon absorptiometry, using a Lunar DP3 scanner (Lunar Corporation, Madison, Wisconsin). Utilizing the 44-keV and lOO-keV photons emitted from a gadolinium source, scans were obtained of the lumbar spine and femoral neck area. Results were calculated as the absolute bone mineral density in g/cm2. To calibrate the system, we measured the density and width of three standards (large, medium, and small) daily throughout the course of the study. Based on the mean values during sequential 5-day intervals during the time of the study, the coefficient of variation for the system was less than 1.5% for all three standards. In published studies, the precision of the method is 1% to 3% with a reliability of 4% to 6% [13], and the coefficient of variation of 42 measurements in 18 individuals was 2.4% in the femoral neck, 2.8% in Ward’s triangle, and 3.0% in the greater trochanter [14]. Because many of the 50- to 75-year-old women in this study had sclerotic changes in the lumbar spine that would falsely elevate the estimate of bone density, the scans of each woman and her matching control were examined individually by one of the authors (A.H.M.) who was unaware of which was the patient’s scan and which the control’s. The area of the lumbar spine that was least affected by sclerotic changes in both subjects was selected for comparison. Ll-L2 was used in five pairs of subjects, Ll in three, L2 in four, L2-L3 in two, L2-L4 in three, and L3-L4 in one. One subject’s spine could not be scanned because her girth did not permit positioning on the scanner; spinal density was therefore determined in 18 pairs of subjects and hip density in 19 pairs.
March
1991
The American
Journal
of Medicine
Volume
90
361
BONE DENSITY
ii
2
AND L-THYROXINE
/ ADLIN
ET AL
.7-
.6-
FEMORAL NECK
LsK?
WARD’S TRIANGLE
TROCHANTERIC AREA
Plasma calcitonin concentrations were measured by Pai C. Kao, Ph.D., at the Mayo Regional Laboratory, Rochester, Minnesota, using a modification [15] of the extraction-concentration radioimmunoassay method of Body and Heath [9]. Patients were studied in the General Clinical Research Center at 9 AM following an overnight fast. Calcium gluconate in a dose of 2 mg calcium per kg body weight was infused over 5 minutes, and blood for calcitonin measurement was drawn before the infusion and at 5 and 10 minutes after the start of the infusion. Serum Tq and TS concentrations and Ta resin uptake were measured by radioimmunoassay. (In five patients the T4 concentration and T-uptake were measured by fluorescence polarization assay [Abbott TDx, Abbott Laboratories, Abbott Park, Illinois]: the normal range by this method was the same as by radioimmunoassay.) Thyroid-stimulating hormone (TSH) was determined by two separate ultrasensitive radioimmunoassay methods: Allegro HS-TSH (Nichols Institute, San Juan Capistrano, California) and HTSH-EIA (Abbott Laboratories, Abbott Park, Illinois). Statistical analyses utilized the one-tailed paired t-test, Wilcoxon’s rank sum test, Fisher’s exact test, and analysis of variance. Coefficients of correlation were calculated using the Pearson product moment. 362
March
1991
The American
Journal
of Medicine
Volume
Figure 1. Bone mineral density in 19 l-thyroxine-treated patients, The rectangles represent the mean f 1 standard deviation of the bone mineral densities of the 19 control subjects. Open circles indicate the lthyroxine-treated patients who had a history of hyperthyroidism; solid circles indicate the I-thyroxinetreated patients who had never had hyperthyroidism.
We have stated confidence intervals dence intervals.
as 95% confi-
RESULTS Characteristics of the 19 l-thyroxine-treated women and the 19 bone-density controls are shown in Table I. The two groups did not differ in age, race, height, weight, percent of ideal body weight, or years since menopause, or in the percentage of subjects who had had bilateral oophorectomy, who were taking calcium supplements or thiazide diuretics, who participated in a regular exercise program, or who had ever smoked cigarettes. Bone mineral density at the spine and hip in the l-thyroxine-treated women and in the controls is shown in Figure 1 and in Table II. Compared with the 19 control patients, the l-thyroxine-treated women had decreased bone mineral density in the spine and in all three areas of the femur. The decrease was most marked (17% decrease in density) in Ward’s triangle. Serum calcitonin concentrations following calcium infusion were measured in 18 of the 19 l-thyroxine-treated patients. Serum calcium levels were 2.43 nmol/L (2.38 nmol/L to 2.47 nmol/L) before the calcium infusion and 2.64 nmol/L (2.59 nmol/L to 2.70 nmol/L) 10 minutes after the start of the 590
BONE DENSITY
minute infusion Basal and stimulated plasma calcitonin concentrations in the l-thyroxine-treated patients and in 39 normal women studied by the same assay at the Mayo Clinic are shown in Table III. The dose and timing of the calcium infusion were the same for both groups. Because the calcitonin assays in the two groups were performed at different times, and interassay variability in basal calcitonin levels is high (oral communication, Dr. Pai C. Kao, Mayo Clinic, Rochester, Minnesota), the basal levels are not comparable. However, the maximal increase in calcitonin levels following calcium stimulation was determined with all specimens from a single individual measured in the same assay; this was markedly lower in the l-thyroxinetreated patients. A twofold or greater increase occurred in 31 of 39 normal women, but in only four of 18 women with thyroid disease. The average duration of l-thyroxine treatment was 15.1 years, with a range of 5 to 37 years. Patients had taken their current dose of l-thyroxine for 3.6 years, with a range of 1 to 7 years. The average dose at the time of the study was 120 pg, and the median dose was 100 1.18of l-thyroxine (Figure 2). Serum total Tq concentration was normal in 16 of the 19 patients, and total Ta concentration and Ts resin uptake were normal in 18 of the 19 patients (Figure 3). Serum TSH concentration, however, measured by ultrasensitive radioimmunoassay, was below normal in 10 of 18 patients in the Nichols Institute assay, and in 13 of 19 patients in the Abbott Laboratories assay. The l-thyroxine-treated patients fell into four diagnostic groups: six had received radioiodine therapy for Graves’ disease, five had primary hypothyroidism, four had had subtotal thyroidectomy, and four were receiving suppression therapy for thyroid nodules or euthyroid goiter. Three of the five patients with primary hypothyroidism had been diagnosed with Hashimoto’s thyroiditis; none of these patients had ever been known to have hyperthyroidism. When compared by analysis of variance, these four groups did not differ from each other in bone density in any of the areas measured, in basal or stimulated calcitonin concentrations, or in thyroid hormone (TJ and Ts) or TSH concentrations. Seven of the 19 patients had been hyperthyroid at some time in the past: the six who had been treated with radioiodine for Graves’ disease, and one of the four patients who had had a subtotal thyroidectomy. (In the other three patients, the operation had been performed for nontoxic nodular goiter.) Four of these patients were white and three black; their average age was 58.3 years. The time elapsed since their hyperthyroidism was successfully treated ranged from 11 to 35 years (average 17.3
AND L-THYROXINE
,’ ADLIN
ET AL
TABLE II Bone Mineral Density in Patients and Controls g/cm* f SD (95% Confidence Intervals) Controls Patients All patients Lumbarspine (n = 18)* Femoral neck (n = 19) Ward's triangle (n = 19) Trochantericarea (n = 19)
1.013&0.136 (0.945-1.081) 0.736f 0.087 (0.694-0.778) 0.576 f 0.097 (0.530-0.623) 0.626 f 0.094 (0.581-0.672)
Patients with history of hyperthyroidism Lumbarspine 1.058f0.171 (n = 7) (0.899-1.218) Femoral neck 0.717f0.054 (n = 7) (0.667-0.767) Ward's triangle 0.540 f 0.052 (n = 7) (0.492-0.588) Trochanteric area 0.600f 0.099 (n = 7) (0.509-0.691)
p Valuer
Percent Decrease
1.134f 0.218 (1.026-1.242) 0.809f 0.129 (0.747-0.872) 0.6945 0.158 (0.617-0.770) 0.722 f 0.148 (0.651-0.794)
0.043
11
0.040
9
0.011
17
0.027
13
1.186~0.166 (1.033-1.339) 0.857 f0.069 (0.794-0.921 0.761 + 0.105 (0.664-0.859) 0.739 f 0.122 (0.626-0.851)
0.132
11
0.002
16
0.002 0.028
29 19
0.112
11
0.284
4
0.195
9
0.152
10
Patients without historv of hvoerthvroidism Lumbarspine 0:984r’O.lOi 1.101 f0.247 (n = 11) (0.912-1.057) (0.935-1.267) Femoral neck 0.748f 0.103 0.782 f. 0.150 (n = 12) (0.682-0.813) (0.686-0.8771 bard's triangle i).598h OIli: t).654f 0.174 (n = 12) (0.527-0.669) (0.543-0.765) Trochantericarea 0.642f 0.092 0.713 f 0.166 (n = 12) (0.583-0.700) (0.607-0.818) 7 r refers to number of pairs tested (one thyroxine-treated patrent In eacn paw). t Parred t-test.
patient and one contn
TABLE Ill Basal and Stimulated Serum Calcitonin Concentrations in l-Thyroxine-Treated Patients and in Normal Women Serum Calcitonin Concentration in rig/L (95% Confidence Intervals) Thyroxine-Treated Normal Women* Patients (n = 18) (n = 39) Basalt 5 minutes 10 minutes Maximal increase
5.98(5.10-6.86) 7.02 (5.54-8.50) 6.26 (5.06-7.45) 1.37*(-0.44-3.17)
3.23 (2.21-4.25) 21.3 (12.0-30.5) 18.6 (10.2-26.9) 18.8(10.0-27.5)
*The data on normal women were supplied by Dr. Pa C. Kao, Mayo Clinic, Rochester, Minnesota. t Blood samples were collected before a 5-minute calcium InfusIon (basal), and 5 and 10 minutes after the start of the infusion. r p
years). TSH levels were suppressed in four of the seven patients. The bone mineral density of these patients and of the 12 l-thyroxine-treated patients without a history of hyperthyroidism, and their controls, is shown in Table II. The previously hyperthyroid patients, analyzed separately, show a loss of bone density in the hip that is statistically significant despite the small size of this subgroup. The reduction of density in the spine, however, and the reduction in density in the spine and hip in a separate analysis of the other 12 patients are not statistically significant.
March
1991
The American
Journal
of Medicine
Volume
90
363
BONE DENSITY
AND L-THYROXINE
/ ADLIN
ET AL
COMMENTS
~
DOSE OF L-THYROXINE (mc9)
We have found a decrease in bone mineral density in the spine and hip in 19 postmenopausal women treated with l-thyroxine, for a variety of indications, for an average of 15 years. The greatest decrease in density, 17% below the level found in a matched control group, was seen in Ward’s triangle, the central area within the femoral neck [16], whose density has been found to be related to bone strength [17] and hip fractures [18]. The 17% reduction in density in Ward’s triangle is a more marked decrease than was reported in the two studies of lthyroxine-treated premenopausal women: Ross and co-workers [4] found a 9% reduction at the radius, while Paul and co-workers [5] reported a 12.8% decrease at the femoral neck and no decrease in the spine. In the two previous studies of bone density in women treated with l-thyroxine, the average dosage of the hormone was relatively high, 171 rug in one study [4] and 175 pg in the other [5]; in both studies the average serum T4 concentration in treated women was higher than normal. Our patients received an average dose of 120 pg of l-thyroxine and a median dose of 100 pg, and the average level of serum T4 was within the normal range. Only three of our 19 patients had slightly increased serum Tq concentrations. Recent studies have estimated the optimal daily replacement dose of l-thyroxine in patients with hypothyroidism to be 112 f 19 pg with a median dose of 125 pg [19], and 127 f 39 pg [20]. Our patients therefore received an average dose that falls within the range currently recommended for physiologic hormone replacement. Nevertheless, the l-thyroxine dosage may have been supraphysiologic in many or most cases, as indicated by the suppressed TSH levels in more than half of our patients. Correlations between bone density and indexes of the extent of thyroid hormone excess and of calcitonin deficiency were examined. Although the density of Ward’s triangle was lower in patients receiving higher doses of l-thyroxine, and the density in the trochanteric area was lower in patients with TSH suppression, most correlations did not tend to confirm a causal relationship between bone density and either thyroid hormone status or calcitonin production. Paul and co-workers [5] also found no correlation between the extent of the loss of bone density and thyroid hormone levels, TSH suppression, or duration of treatment with l-thyroxine. We also found no difference in bone density or in thyroid hormone levels or calcitonin responsiveness when we compared patients who were receiving lthyroxine because of primary hypothyroidism, because of previous thyroid ablation by surgery or
200 175 150 125 IO0 75 50
00
25 Figure disease.
2. Dose
of l-thyroxine
in the
19 women
with
thyroid
Correlations between the bone mineral density and measures of the level of thyroid hormone treatment and calcitonin responsiveness are shown in Table IV. The density at Ward’s triangle was inversely correlated with the dose of l-thyroxine, and the density at the trochanteric area was positively correlated with the serum TSH concentration, but because multiple comparisons were performed the statistical significance of these correlations is uncertain. For the most part, there was no relation between bone mineral density and the extent of treatment (or overtreatment) with l-thyroxine or the basal or stimulated level of serum calcitonin. 364
March
1991
The American
Journal
of Medicine
Volume
90
BONE DENSITY
TOTAL T4 (MCG/OL) 4.01
Figure 3. Thyroid function tests in the 19 i-thyroxine-treated women. The rectangles indicate the normal range. TSH was measured by two separate radioimmunoassay methods, using kits obtained from Nichols Institute and Abbott Laboratories. To convert to SI units, multiply the total f4 by 12.87 and the total T3 by 0.01536.
TOTAL (NGIDL)
-.
T3
AND L-THYROXINE
T3 RESIN UPTAKE (Percent)
/ ADLIN
ET AL
TSH
I
6.0
60-
2oL
radioactive iodine, or as suppressive therapy for thyroid nodules or euthyroid goiter. The number of patients in each of these categories, however, were too small for us to conclude that no difference existed. At least three conditions were present in some or all of these patients that might have contributed to their loss of bone mineral density. First, thyroid hormone replacement was excessive in many, as indicated by the suppression of serum TSH levels in more than half of the patients. If hyperthyroidism was present it was subclinical, since none of the patients was considered to be clinically hyperthyroid, l-thyroxine treatment did not exceed the currently recommended dosage, and serum Tq and T3 levels were normal in most cases. In the two previous similar studies, in which higher doses of l-thyroxine were administered and serum TJ levels were elevated, thyroid hormone treatment was considered to be the cause of the bone loss. It is not clear whether this is the case in our patients. To eliminate a history of spontaneous hyperthyroidism as a factor, we analyzed separately the 12 patients who had no history of hyperthyroidism. Bone density was decreased by 11% in the spine and by 4% to 10% in the hip, but these changes did not achieve statis-
..A. NICHOLS
“.:::u ABBOTT
tical significance. Because of this, and because there was no correlation between thyroid hormone levels and bone density, our data do not allow us to conclude that thyroid hormone over-replacement contributes to the decreased bone density in l-thyroxine-treated patients, although they suggest that this may be true. An association between subtle thyroid hormone over-replacement and bone loss, if supported by further study, would suggest that TSH suppression should be avoided in patients receiving l-thyroxine, and presently recommended doses would have to be revised downward. A second factor that must be considered as a possible cause of decreased bone density is a deficiency of calcitonin. Low concentrations of this hormone, especially after stimulation by calcium infusion, have been found in patients with hypothyroidism [7] and in patients who have had thyroid ablation by surgery [8,9] or by radioactive iodine [lo]. The patients in our study had a marked decrease in the responsiveness of calcitonin to the stimulus of a calcium infusion. Although it is not clear that deficiency of calcitonin production can cause changes in bone density, this possibility cannot be excluded, especially in the presence of other factors, such as the postmenopausal state or excess replacement of
TABLEIV Relation
of Bone Density
to Thyroid
Hormone
and Calcitonin
levels Coefficient
Dose of l-Thyroxine Density at: L2-L4 Femoral neck Ward’s triangle Trochanteric area
0.040 -0.377 -0.530% -0.339
Duration of Treatment
T4
-0.120 -0.147 -0.315 0.360
0.037 0.096 0.185 -0.155
of Correlation TSH (Nichols)
Basal Calcitonin
-0.149 -0.093 -0.267 0.536+
Maximal Increase
0.047 -0.089 -0.257 -0.078
-0.022 0.340 0.282 0.263
* p = 0.020. t D = 0.022.
March
1991
The American
Journal
of Medicine
Volume
90
365
BONE DENSITY
AND L-THYROXINE
/ ADLIN
ET AL
ACKNOiNLEDGMENT
thyroid hormone, that may predispose to bone loss. However, since we observed no correlation between calcitonin levels and bone density, we cannot conclude that this was a factor in the bone loss that was present in our patients. Another factor that may have contributed to bone loss in the l-thyroxine-treated patients is a history of hyperthyroidism in the distant past, which was present in seven of the 19 patients. It is not possible to determine the length of time these patients were hyperthyroid before successful treatment of this condition, but during this time an irreversible loss of bone density may have occurred. When these seven patients are analyzed separately, they have a marked and highly significant loss of bone density in the hip, whereas the remaining 12 patients have a less marked, and no longer significant, bone loss. A greater degree of bone loss in patients with the additional factor of prior hyperthyroidism suggests that this may be an independent factor in the loss of bone density. In summary, we have observed a decrease in bone density in the spine and hip in postmenopausal women treated with l-thyroxine. We have demonstrated the presence of three factors in some or all of these patients that may have contributed to loss of bone mineral: excess thyroid hormone replacement, calcitonin deficiency, and a history of hyperthyroidism in the distant past. Our data do not indicate how much, if any, effect can be attributed to each of these potential causes, but suggest that a history of hyperthyroidism is an independent factor. Others have suggested that excessive thyroid hormone replacement is the main cause of this bone loss, at least in patients with increased serum T4 levels. Further study is needed to clarify the relative importance of these factors. L-thyroxine dosage, unlike the other factors discussed, can be readily modified. On the basis of the present study and other available data, we suggest that thyroid hormone over-replacement, as indicated by TSH suppression, is a potential cause of decreased bone mineral density. If this is true, adjustment of l-thyroxine dosage to result in the maintenance of TSH in the normal range should be a goal of treatment in patients with hypothyroidism.
366
March
1991
The American
Journal
of Medicine
Volume
We thank Isabel Ryan, R.N., for her skilled assistance, Chris Devitt for performing the statistical analyses.
and Carl Smalley and
REFERENCES l.Auwerx J, Bouillon R. Mineral and bone metabolism in thyroid disease: a review. Q J Med 1986; 60: 737-52. 2. Fallon MD, Perry HM Ill, Bergfeld M, Droke D. Teitelbaum SL, Avioli LV. Exogenous hyperthyroidism with osteoporosis. Arch Intern Med 1983; 143: 442-4. 3. Coindre JM, David JP, Riviere L, et al. Bone loss in hypothyroidism with hormone replacement: a histomorphometricstudy. Arch Intern Med 1986; 146: 48-53. 4. Ross DS, Neer RM, Ridgway EC, Daniels GH. Subclinical hyperthyroidism and reduced bone density as a possible result of prolonged suppression of the pituitary-thyroid axis with I-thyroxine. Am J Med 1987; 82: 1167-70. 5. Paul TL, Kerrigan J, Kelly AM, Braverman LE, Baran DT. Long-term l-thyroxine therapy is associated with decreased hip bone density in premenopausal women. JAMA 1988; 259: 3137-41. 6. Austin LA, Heath H ill. Calcitonin: physiology and pathophysiology. N Engl J Med 1981; 304: 269-77. 7. Body JJ, Demeester-Mirkine N, Borkowski A, Suciu S, Corvilain J. Calcitonin deficiency in primary hypothyroidism. J Clin Endocrinol Metab 1986; 62: 700-3. 8. Silva OL, Wisneski LA, Cyrus J, Snider RH, Moore CF, Becker KL. Calcitonin in thyroldectomized patients. Am J Med Sci 1978; 275: 159-64. 9. Body JJ, Heath H III. Estimates of circulating monomeric calcitonin: physiological studies in normal and thyroidectomized man. J Clin Endocrinol Metab 1983; 57: 897-903. 10. Body JJ, Demeester-Mirkine N, Corvilain J. Calcitonin deficiency after radioactive iodine treatment. Ann Intern Med 1988; 109: 590-l. 11. McDermott MT, Kidd GS, Blue P, Ghaed V, Hofeldt FD. Reduced bone mineral content in totally thyroidectomized patients: possible effect of calcitonin deficiency. J Clin Endocrinol Metab 1983; 56: 936-9. 12. Hurley DL, Tiegs RD. Wahner HW, Heath H Ill. Axial and appendicular bone mineral dehsity in patients with long-term deficiency or excess of calcitonin. N Engl J Med 1987; 317: 53741. 13. Kimmel PL. Radiologic methods to evaluate bone mineral content, Ann Intern Med 1984; 100: 908-11. 14. Mazess RB, Barden H, Ettinger M, Schultz EI Bone density of the radius, spine, and proximal femur in osteoporosis. J Bone Miner Res 1988; 3: 13-8. 15. Gharib H, Kao PC, Heath H Ill. Determination of silica-purified plasma calcitonin for the detection and management of medullary thyroid carcinoma: comparison of two provocative tests. Mayo Clin Proc 1987; 62: 373-8. 16. Harty M. The anatomy of the hip joint. In: Tronzo RG, ed. Surgery of the hip joint. Philadelphia: Lea and Febiger, 1973: 64-5. 17. Vose GP, Mack PB. Roentgenologic assessment of femoral neck density as related to fracturing. Am J Roentgen01 1963: 89: 1296-301. 18. Vose GP, Lockwood RM. Femoral neck fracturing-its relationship to radiographic bone density. J Gerontol 1965; 20: 300-5. 19. Fish LH, Schwartz HL, Cavanaugh J, Steffes MW, Bantle JP. Oppenheimer JH. Replacement dose, metabolism, and bioavailability of levothyroxine in the treatment of hypothyroidism: role of triiodothyronine in pituitary feedback in humans. N Engl J Med 1987; 316: 76470. 20. Hennessey JV. Evaul JE, Tseng YC. Burman KD, Wartofsky L. L-thyroxine dosage: a reevaluation of therapy with contemporary preparations. Ann Intern Med 1986; 105: 11-5.
90
PURPOSE: To determine if bone mineral density is decreased in postmenopausal women treated with l-thyroxine, and, if any decrease is observed, whether it is related to overtreatment with thyroid hormone, to deficiency of calcitonin, or to other factors. PATIENTSANDMETHODS: Thestudyconsistedof 19 postmenopausal women between 50 and 75 years of age treated with l-thyroxine for 5 years or longer, and 19 matching control subjects with no thyroid disease. Bone mineral density of the spine and hip was measured by dual-photon absorptiometry. Plasma calcitonin concentrations and serum thyroid hormone levels were determined by radioimmunoassays. RESULTS: The l-thyroxine-treated women had lower bone density in the lumbar spine (1.013 g/ cm2 [95% confidence interval, 0.945 to 1.0811 versus 1.134 g/cm2 [1.026 to 1.2421, p = 0.043); in the femoral neck (0.736 g/cm2 [0.694 to 0.7781 versus 0.809 g/cm2 [0.747 to 0.8721, p = 0.040); in Ward’s triangle (0.576 g/cm2 [0.530 to 0.6231 versus 0.694 g/cm2 [0.617 to 0.7701, p = 0.011); and in the trochanteric area (0.626 g/cm2 [0.581 to 0.6721 versus 0.722 g/cm2 [0.651 to 0.7941, p = 0.027). The maximal increase in calcitonin following calcium infusion was 1.37 rig/L (95% confidence interval, -0.44 to 3.17) in the l-thyroxine-treated patients versus 18.8 rig/L (95% confidence interval, 10.0 to 27.5) in normal women, p
supraphysiologic. Seven of the 19 patients had a history of hyperthyroidism in the distant past; these patients, considered separately, had significantly reduced bone density in the hip. The other 12 patients, considered separately, did not have a statistically significant loss of bone density. CONCLUSIONS: Long-term l-thyroxine therapy is associated with decreased density of the spine and hip. Since subclinical hyperthyroidism, decreased calcitoniu responsiveness, and a history of hyperthyroidism were demonstrated in some or all of these patients, these factors must be considered as possible causes of the decreased bone density.
T
hyroid hormone stimulates osteoclast activity and bone resorption [l], and hyperthyroidism, whether endogenous or exogenous, is a well-established cause of osteoporosis [2]. Recent observations have suggested that, additionally, clinically euthyroid patients treated with l-thyroxine, either as replacement therapy for hypothyroidism or suppressive therapy for thyroid nodules or goiter, have decreased bone density. Coindre and co-workers [3] found a reduction in trabecular bone volume in iliac bone biopsy specimens in hypothyroid patients given replacement doses of l-thyroxine and triiodothyronine .(Ts). Two additional studies have demonstrated decreased bone density, in the radius and hip, in premenopausal women receiving l-thyroxine [4,5]. In both these studies, the majority of patients, although clinically euthyroid, had an increased serum concentration of thyroxine (T4); the authors attributed the skeletal changes to thyroid hormone excess. Speculation has also centered on calcitonin deficiency as a potential cause of demineralization in patients with hypothyroidism. Calcitonin is a hormone produced by the parafollicular cells of the thyroid that acts to suppress bone resorption, but that has never been shown conclusively to have a physiologic role in the conservation of bone mass [6]. Calcitonin levels have been found to be decreased in patients with primary hypothyroidism
From the Department of Medicine (EVA, ADM, BJC) and the Department of Nuclear Medicine (AHM). Temple University School of Medicine, Philadelphia, Pennsylvania. Presented in part at the 63rd Meeting, American Thyroid Association, September 28, 1988, Montreal, Canada, and at the 21st European Symposium on Calcified Tissues, Jerusalem, Israel, March 13, 1989. This study was supported in part by a grant from the U.S. Public Health Service, Research Grant No. 2 MO1 RR 349 from the National Institutes of Health, General Clinical Research Centers Branch. Requests for reprints should be addressed to E. Victor Adlin, M.D.. Temple University School of Medicine, 3401 North Broad Street, Philadelphia, Pennsylvania 19140. Manuscript submitted April 26, 1990, and accepted in revised form October 1, 1990.
360
March
1991
The American
Journal
of Medicine
Volume
90
BONE DENSITY
[7] as well as in patients who have had a thyroidectomy [8,9] or radioiodine therapy [lo]. Decreased density of the radius in a group of thyroidectomized patients has been attributed to calcitonin deficiency [ll], but in another group of patients with low calcitonin levels following thyroidectomy, bone mineral density was not decreased [12]. Any effect of thyroid hormone excess or calcitonin deficiency on enhancing bone mineral loss might be expected to be most evident in postmenopausal women, a group that is especially susceptible to osteoporosis. We have studied bone density at the spine and hip in postmenopausal women treated with l-thyroxine for at least 5 years. We evaluated thyroid hormone levels and basal and stimulated calcitonin levels, to clarify the roles of these factors in any abnormalities of skeletal density that might be present in’these women.
AND L-THYROXINE
/ ADLIN
ET AL
TABLEI Characteristicsof Patientsand Controls Controls (n = 19) Age (years)*
61.6 ‘5yf;~o)
Race Height (cm)*
Patients (n=19)
Value ’
60.9 (5fp&’
9 white 161.3 (158$%;64.3)
9 white 161.0 (158%1863.3)
(65.&;5.5)
(62.:;977.6)
NS NS* NS§
Weight (kg)”
NS
Percent ideal body weight*
NS
Years since menopause* Bilateral oophorectomy Use of calcium supplement Use of thiazide diuretic Exercise program Smokers (ever)
W;;281
wy;32)
‘9.;7M;.4’
(9j2jly
6/19 10/19 5/19 6/19
NS§ NS NS* NS NS* NS*
4119 8/19 7/19 5119
PATIENTS AND METHODS
* Mean and 95% confidence intervals. + Wilcoxon rank sum test. t Fisher’s exact test. 3 Paired t-test.
Postmenopausal women between 50 and 75 years of age, who had been treated with l-thyroxine for 5 years or longer, were selected from an outpatient practice of endocrinology and internal medicine. We evaluated each patient seen in our practice during the time the study was in progress; all who met the inclusion and exclusion criteria were invited to participate. An equal number of control patients were selected from the same population by the same process, with the same inclusion and exclusion criteria; the only difference was that they had never had thyroid disease or treatment with l-thyroxine, and were selected only if they were similar in age and race to one of the l-thyroxine-treated patients, with whom they were matched. Patients were excluded if they had any condition known to affect bone density, such as diseases of the kidneys, parathyroids, or adrenals, or musculoskeleta1 disease that affected mobility. Patients were excluded if they had been treated with estrogen, glucocorticoids, anticonvulsants, or vitamin D. None of the patients or controls were known to have taken estrogens in the past except for one 68-yearold control patient who took Premarin from 1969 to 1972. Because only a small proportion of the patients in our practice population fulfilled the requirements of postmenopausal status, l-thyroxine therapy, and absence of estrogen therapy and the other exclusion criteria, the 19 l-thyroxine-treated patients and 19 control subjects were drawn from a much larger group of patients. Of the 19 control subjects, 14 were being treated for hypertension, one for reflux. esophagitis, and one for migraine headaches; three had no chronic disease. Bone mineral density of the spine and hip was
measured by dual-photon absorptiometry, using a Lunar DP3 scanner (Lunar Corporation, Madison, Wisconsin). Utilizing the 44-keV and lOO-keV photons emitted from a gadolinium source, scans were obtained of the lumbar spine and femoral neck area. Results were calculated as the absolute bone mineral density in g/cm2. To calibrate the system, we measured the density and width of three standards (large, medium, and small) daily throughout the course of the study. Based on the mean values during sequential 5-day intervals during the time of the study, the coefficient of variation for the system was less than 1.5% for all three standards. In published studies, the precision of the method is 1% to 3% with a reliability of 4% to 6% [13], and the coefficient of variation of 42 measurements in 18 individuals was 2.4% in the femoral neck, 2.8% in Ward’s triangle, and 3.0% in the greater trochanter [14]. Because many of the 50- to 75-year-old women in this study had sclerotic changes in the lumbar spine that would falsely elevate the estimate of bone density, the scans of each woman and her matching control were examined individually by one of the authors (A.H.M.) who was unaware of which was the patient’s scan and which the control’s. The area of the lumbar spine that was least affected by sclerotic changes in both subjects was selected for comparison. Ll-L2 was used in five pairs of subjects, Ll in three, L2 in four, L2-L3 in two, L2-L4 in three, and L3-L4 in one. One subject’s spine could not be scanned because her girth did not permit positioning on the scanner; spinal density was therefore determined in 18 pairs of subjects and hip density in 19 pairs.
March
1991
The American
Journal
of Medicine
Volume
90
361
BONE DENSITY
ii
2
AND L-THYROXINE
/ ADLIN
ET AL
.7-
.6-
FEMORAL NECK
LsK?
WARD’S TRIANGLE
TROCHANTERIC AREA
Plasma calcitonin concentrations were measured by Pai C. Kao, Ph.D., at the Mayo Regional Laboratory, Rochester, Minnesota, using a modification [15] of the extraction-concentration radioimmunoassay method of Body and Heath [9]. Patients were studied in the General Clinical Research Center at 9 AM following an overnight fast. Calcium gluconate in a dose of 2 mg calcium per kg body weight was infused over 5 minutes, and blood for calcitonin measurement was drawn before the infusion and at 5 and 10 minutes after the start of the infusion. Serum Tq and TS concentrations and Ta resin uptake were measured by radioimmunoassay. (In five patients the T4 concentration and T-uptake were measured by fluorescence polarization assay [Abbott TDx, Abbott Laboratories, Abbott Park, Illinois]: the normal range by this method was the same as by radioimmunoassay.) Thyroid-stimulating hormone (TSH) was determined by two separate ultrasensitive radioimmunoassay methods: Allegro HS-TSH (Nichols Institute, San Juan Capistrano, California) and HTSH-EIA (Abbott Laboratories, Abbott Park, Illinois). Statistical analyses utilized the one-tailed paired t-test, Wilcoxon’s rank sum test, Fisher’s exact test, and analysis of variance. Coefficients of correlation were calculated using the Pearson product moment. 362
March
1991
The American
Journal
of Medicine
Volume
Figure 1. Bone mineral density in 19 l-thyroxine-treated patients, The rectangles represent the mean f 1 standard deviation of the bone mineral densities of the 19 control subjects. Open circles indicate the lthyroxine-treated patients who had a history of hyperthyroidism; solid circles indicate the I-thyroxinetreated patients who had never had hyperthyroidism.
We have stated confidence intervals dence intervals.
as 95% confi-
RESULTS Characteristics of the 19 l-thyroxine-treated women and the 19 bone-density controls are shown in Table I. The two groups did not differ in age, race, height, weight, percent of ideal body weight, or years since menopause, or in the percentage of subjects who had had bilateral oophorectomy, who were taking calcium supplements or thiazide diuretics, who participated in a regular exercise program, or who had ever smoked cigarettes. Bone mineral density at the spine and hip in the l-thyroxine-treated women and in the controls is shown in Figure 1 and in Table II. Compared with the 19 control patients, the l-thyroxine-treated women had decreased bone mineral density in the spine and in all three areas of the femur. The decrease was most marked (17% decrease in density) in Ward’s triangle. Serum calcitonin concentrations following calcium infusion were measured in 18 of the 19 l-thyroxine-treated patients. Serum calcium levels were 2.43 nmol/L (2.38 nmol/L to 2.47 nmol/L) before the calcium infusion and 2.64 nmol/L (2.59 nmol/L to 2.70 nmol/L) 10 minutes after the start of the 590
BONE DENSITY
minute infusion Basal and stimulated plasma calcitonin concentrations in the l-thyroxine-treated patients and in 39 normal women studied by the same assay at the Mayo Clinic are shown in Table III. The dose and timing of the calcium infusion were the same for both groups. Because the calcitonin assays in the two groups were performed at different times, and interassay variability in basal calcitonin levels is high (oral communication, Dr. Pai C. Kao, Mayo Clinic, Rochester, Minnesota), the basal levels are not comparable. However, the maximal increase in calcitonin levels following calcium stimulation was determined with all specimens from a single individual measured in the same assay; this was markedly lower in the l-thyroxinetreated patients. A twofold or greater increase occurred in 31 of 39 normal women, but in only four of 18 women with thyroid disease. The average duration of l-thyroxine treatment was 15.1 years, with a range of 5 to 37 years. Patients had taken their current dose of l-thyroxine for 3.6 years, with a range of 1 to 7 years. The average dose at the time of the study was 120 pg, and the median dose was 100 1.18of l-thyroxine (Figure 2). Serum total Tq concentration was normal in 16 of the 19 patients, and total Ta concentration and Ts resin uptake were normal in 18 of the 19 patients (Figure 3). Serum TSH concentration, however, measured by ultrasensitive radioimmunoassay, was below normal in 10 of 18 patients in the Nichols Institute assay, and in 13 of 19 patients in the Abbott Laboratories assay. The l-thyroxine-treated patients fell into four diagnostic groups: six had received radioiodine therapy for Graves’ disease, five had primary hypothyroidism, four had had subtotal thyroidectomy, and four were receiving suppression therapy for thyroid nodules or euthyroid goiter. Three of the five patients with primary hypothyroidism had been diagnosed with Hashimoto’s thyroiditis; none of these patients had ever been known to have hyperthyroidism. When compared by analysis of variance, these four groups did not differ from each other in bone density in any of the areas measured, in basal or stimulated calcitonin concentrations, or in thyroid hormone (TJ and Ts) or TSH concentrations. Seven of the 19 patients had been hyperthyroid at some time in the past: the six who had been treated with radioiodine for Graves’ disease, and one of the four patients who had had a subtotal thyroidectomy. (In the other three patients, the operation had been performed for nontoxic nodular goiter.) Four of these patients were white and three black; their average age was 58.3 years. The time elapsed since their hyperthyroidism was successfully treated ranged from 11 to 35 years (average 17.3
AND L-THYROXINE
,’ ADLIN
ET AL
TABLE II Bone Mineral Density in Patients and Controls g/cm* f SD (95% Confidence Intervals) Controls Patients All patients Lumbarspine (n = 18)* Femoral neck (n = 19) Ward's triangle (n = 19) Trochantericarea (n = 19)
1.013&0.136 (0.945-1.081) 0.736f 0.087 (0.694-0.778) 0.576 f 0.097 (0.530-0.623) 0.626 f 0.094 (0.581-0.672)
Patients with history of hyperthyroidism Lumbarspine 1.058f0.171 (n = 7) (0.899-1.218) Femoral neck 0.717f0.054 (n = 7) (0.667-0.767) Ward's triangle 0.540 f 0.052 (n = 7) (0.492-0.588) Trochanteric area 0.600f 0.099 (n = 7) (0.509-0.691)
p Valuer
Percent Decrease
1.134f 0.218 (1.026-1.242) 0.809f 0.129 (0.747-0.872) 0.6945 0.158 (0.617-0.770) 0.722 f 0.148 (0.651-0.794)
0.043
11
0.040
9
0.011
17
0.027
13
1.186~0.166 (1.033-1.339) 0.857 f0.069 (0.794-0.921 0.761 + 0.105 (0.664-0.859) 0.739 f 0.122 (0.626-0.851)
0.132
11
0.002
16
0.002 0.028
29 19
0.112
11
0.284
4
0.195
9
0.152
10
Patients without historv of hvoerthvroidism Lumbarspine 0:984r’O.lOi 1.101 f0.247 (n = 11) (0.912-1.057) (0.935-1.267) Femoral neck 0.748f 0.103 0.782 f. 0.150 (n = 12) (0.682-0.813) (0.686-0.8771 bard's triangle i).598h OIli: t).654f 0.174 (n = 12) (0.527-0.669) (0.543-0.765) Trochantericarea 0.642f 0.092 0.713 f 0.166 (n = 12) (0.583-0.700) (0.607-0.818) 7 r refers to number of pairs tested (one thyroxine-treated patrent In eacn paw). t Parred t-test.
patient and one contn
TABLE Ill Basal and Stimulated Serum Calcitonin Concentrations in l-Thyroxine-Treated Patients and in Normal Women Serum Calcitonin Concentration in rig/L (95% Confidence Intervals) Thyroxine-Treated Normal Women* Patients (n = 18) (n = 39) Basalt 5 minutes 10 minutes Maximal increase
5.98(5.10-6.86) 7.02 (5.54-8.50) 6.26 (5.06-7.45) 1.37*(-0.44-3.17)
3.23 (2.21-4.25) 21.3 (12.0-30.5) 18.6 (10.2-26.9) 18.8(10.0-27.5)
*The data on normal women were supplied by Dr. Pa C. Kao, Mayo Clinic, Rochester, Minnesota. t Blood samples were collected before a 5-minute calcium InfusIon (basal), and 5 and 10 minutes after the start of the infusion. r p
years). TSH levels were suppressed in four of the seven patients. The bone mineral density of these patients and of the 12 l-thyroxine-treated patients without a history of hyperthyroidism, and their controls, is shown in Table II. The previously hyperthyroid patients, analyzed separately, show a loss of bone density in the hip that is statistically significant despite the small size of this subgroup. The reduction of density in the spine, however, and the reduction in density in the spine and hip in a separate analysis of the other 12 patients are not statistically significant.
March
1991
The American
Journal
of Medicine
Volume
90
363
BONE DENSITY
AND L-THYROXINE
/ ADLIN
ET AL
COMMENTS
~
DOSE OF L-THYROXINE (mc9)
We have found a decrease in bone mineral density in the spine and hip in 19 postmenopausal women treated with l-thyroxine, for a variety of indications, for an average of 15 years. The greatest decrease in density, 17% below the level found in a matched control group, was seen in Ward’s triangle, the central area within the femoral neck [16], whose density has been found to be related to bone strength [17] and hip fractures [18]. The 17% reduction in density in Ward’s triangle is a more marked decrease than was reported in the two studies of lthyroxine-treated premenopausal women: Ross and co-workers [4] found a 9% reduction at the radius, while Paul and co-workers [5] reported a 12.8% decrease at the femoral neck and no decrease in the spine. In the two previous studies of bone density in women treated with l-thyroxine, the average dosage of the hormone was relatively high, 171 rug in one study [4] and 175 pg in the other [5]; in both studies the average serum T4 concentration in treated women was higher than normal. Our patients received an average dose of 120 pg of l-thyroxine and a median dose of 100 pg, and the average level of serum T4 was within the normal range. Only three of our 19 patients had slightly increased serum Tq concentrations. Recent studies have estimated the optimal daily replacement dose of l-thyroxine in patients with hypothyroidism to be 112 f 19 pg with a median dose of 125 pg [19], and 127 f 39 pg [20]. Our patients therefore received an average dose that falls within the range currently recommended for physiologic hormone replacement. Nevertheless, the l-thyroxine dosage may have been supraphysiologic in many or most cases, as indicated by the suppressed TSH levels in more than half of our patients. Correlations between bone density and indexes of the extent of thyroid hormone excess and of calcitonin deficiency were examined. Although the density of Ward’s triangle was lower in patients receiving higher doses of l-thyroxine, and the density in the trochanteric area was lower in patients with TSH suppression, most correlations did not tend to confirm a causal relationship between bone density and either thyroid hormone status or calcitonin production. Paul and co-workers [5] also found no correlation between the extent of the loss of bone density and thyroid hormone levels, TSH suppression, or duration of treatment with l-thyroxine. We also found no difference in bone density or in thyroid hormone levels or calcitonin responsiveness when we compared patients who were receiving lthyroxine because of primary hypothyroidism, because of previous thyroid ablation by surgery or
200 175 150 125 IO0 75 50
00
25 Figure disease.
2. Dose
of l-thyroxine
in the
19 women
with
thyroid
Correlations between the bone mineral density and measures of the level of thyroid hormone treatment and calcitonin responsiveness are shown in Table IV. The density at Ward’s triangle was inversely correlated with the dose of l-thyroxine, and the density at the trochanteric area was positively correlated with the serum TSH concentration, but because multiple comparisons were performed the statistical significance of these correlations is uncertain. For the most part, there was no relation between bone mineral density and the extent of treatment (or overtreatment) with l-thyroxine or the basal or stimulated level of serum calcitonin. 364
March
1991
The American
Journal
of Medicine
Volume
90
BONE DENSITY
TOTAL T4 (MCG/OL) 4.01
Figure 3. Thyroid function tests in the 19 i-thyroxine-treated women. The rectangles indicate the normal range. TSH was measured by two separate radioimmunoassay methods, using kits obtained from Nichols Institute and Abbott Laboratories. To convert to SI units, multiply the total f4 by 12.87 and the total T3 by 0.01536.
TOTAL (NGIDL)
-.
T3
AND L-THYROXINE
T3 RESIN UPTAKE (Percent)
/ ADLIN
ET AL
TSH
I
6.0
60-
2oL
radioactive iodine, or as suppressive therapy for thyroid nodules or euthyroid goiter. The number of patients in each of these categories, however, were too small for us to conclude that no difference existed. At least three conditions were present in some or all of these patients that might have contributed to their loss of bone mineral density. First, thyroid hormone replacement was excessive in many, as indicated by the suppression of serum TSH levels in more than half of the patients. If hyperthyroidism was present it was subclinical, since none of the patients was considered to be clinically hyperthyroid, l-thyroxine treatment did not exceed the currently recommended dosage, and serum Tq and T3 levels were normal in most cases. In the two previous similar studies, in which higher doses of l-thyroxine were administered and serum TJ levels were elevated, thyroid hormone treatment was considered to be the cause of the bone loss. It is not clear whether this is the case in our patients. To eliminate a history of spontaneous hyperthyroidism as a factor, we analyzed separately the 12 patients who had no history of hyperthyroidism. Bone density was decreased by 11% in the spine and by 4% to 10% in the hip, but these changes did not achieve statis-
..A. NICHOLS
“.:::u ABBOTT
tical significance. Because of this, and because there was no correlation between thyroid hormone levels and bone density, our data do not allow us to conclude that thyroid hormone over-replacement contributes to the decreased bone density in l-thyroxine-treated patients, although they suggest that this may be true. An association between subtle thyroid hormone over-replacement and bone loss, if supported by further study, would suggest that TSH suppression should be avoided in patients receiving l-thyroxine, and presently recommended doses would have to be revised downward. A second factor that must be considered as a possible cause of decreased bone density is a deficiency of calcitonin. Low concentrations of this hormone, especially after stimulation by calcium infusion, have been found in patients with hypothyroidism [7] and in patients who have had thyroid ablation by surgery [8,9] or by radioactive iodine [lo]. The patients in our study had a marked decrease in the responsiveness of calcitonin to the stimulus of a calcium infusion. Although it is not clear that deficiency of calcitonin production can cause changes in bone density, this possibility cannot be excluded, especially in the presence of other factors, such as the postmenopausal state or excess replacement of
TABLEIV Relation
of Bone Density
to Thyroid
Hormone
and Calcitonin
levels Coefficient
Dose of l-Thyroxine Density at: L2-L4 Femoral neck Ward’s triangle Trochanteric area
0.040 -0.377 -0.530% -0.339
Duration of Treatment
T4
-0.120 -0.147 -0.315 0.360
0.037 0.096 0.185 -0.155
of Correlation TSH (Nichols)
Basal Calcitonin
-0.149 -0.093 -0.267 0.536+
Maximal Increase
0.047 -0.089 -0.257 -0.078
-0.022 0.340 0.282 0.263
* p = 0.020. t D = 0.022.
March
1991
The American
Journal
of Medicine
Volume
90
365
BONE DENSITY
AND L-THYROXINE
/ ADLIN
ET AL
ACKNOiNLEDGMENT
thyroid hormone, that may predispose to bone loss. However, since we observed no correlation between calcitonin levels and bone density, we cannot conclude that this was a factor in the bone loss that was present in our patients. Another factor that may have contributed to bone loss in the l-thyroxine-treated patients is a history of hyperthyroidism in the distant past, which was present in seven of the 19 patients. It is not possible to determine the length of time these patients were hyperthyroid before successful treatment of this condition, but during this time an irreversible loss of bone density may have occurred. When these seven patients are analyzed separately, they have a marked and highly significant loss of bone density in the hip, whereas the remaining 12 patients have a less marked, and no longer significant, bone loss. A greater degree of bone loss in patients with the additional factor of prior hyperthyroidism suggests that this may be an independent factor in the loss of bone density. In summary, we have observed a decrease in bone density in the spine and hip in postmenopausal women treated with l-thyroxine. We have demonstrated the presence of three factors in some or all of these patients that may have contributed to loss of bone mineral: excess thyroid hormone replacement, calcitonin deficiency, and a history of hyperthyroidism in the distant past. Our data do not indicate how much, if any, effect can be attributed to each of these potential causes, but suggest that a history of hyperthyroidism is an independent factor. Others have suggested that excessive thyroid hormone replacement is the main cause of this bone loss, at least in patients with increased serum T4 levels. Further study is needed to clarify the relative importance of these factors. L-thyroxine dosage, unlike the other factors discussed, can be readily modified. On the basis of the present study and other available data, we suggest that thyroid hormone over-replacement, as indicated by TSH suppression, is a potential cause of decreased bone mineral density. If this is true, adjustment of l-thyroxine dosage to result in the maintenance of TSH in the normal range should be a goal of treatment in patients with hypothyroidism.
366
March
1991
The American
Journal
of Medicine
Volume
We thank Isabel Ryan, R.N., for her skilled assistance, Chris Devitt for performing the statistical analyses.
and Carl Smalley and
REFERENCES l.Auwerx J, Bouillon R. Mineral and bone metabolism in thyroid disease: a review. Q J Med 1986; 60: 737-52. 2. Fallon MD, Perry HM Ill, Bergfeld M, Droke D. Teitelbaum SL, Avioli LV. Exogenous hyperthyroidism with osteoporosis. Arch Intern Med 1983; 143: 442-4. 3. Coindre JM, David JP, Riviere L, et al. Bone loss in hypothyroidism with hormone replacement: a histomorphometricstudy. Arch Intern Med 1986; 146: 48-53. 4. Ross DS, Neer RM, Ridgway EC, Daniels GH. Subclinical hyperthyroidism and reduced bone density as a possible result of prolonged suppression of the pituitary-thyroid axis with I-thyroxine. Am J Med 1987; 82: 1167-70. 5. Paul TL, Kerrigan J, Kelly AM, Braverman LE, Baran DT. Long-term l-thyroxine therapy is associated with decreased hip bone density in premenopausal women. JAMA 1988; 259: 3137-41. 6. Austin LA, Heath H ill. Calcitonin: physiology and pathophysiology. N Engl J Med 1981; 304: 269-77. 7. Body JJ, Demeester-Mirkine N, Borkowski A, Suciu S, Corvilain J. Calcitonin deficiency in primary hypothyroidism. J Clin Endocrinol Metab 1986; 62: 700-3. 8. Silva OL, Wisneski LA, Cyrus J, Snider RH, Moore CF, Becker KL. Calcitonin in thyroldectomized patients. Am J Med Sci 1978; 275: 159-64. 9. Body JJ, Heath H III. Estimates of circulating monomeric calcitonin: physiological studies in normal and thyroidectomized man. J Clin Endocrinol Metab 1983; 57: 897-903. 10. Body JJ, Demeester-Mirkine N, Corvilain J. Calcitonin deficiency after radioactive iodine treatment. Ann Intern Med 1988; 109: 590-l. 11. McDermott MT, Kidd GS, Blue P, Ghaed V, Hofeldt FD. Reduced bone mineral content in totally thyroidectomized patients: possible effect of calcitonin deficiency. J Clin Endocrinol Metab 1983; 56: 936-9. 12. Hurley DL, Tiegs RD. Wahner HW, Heath H Ill. Axial and appendicular bone mineral dehsity in patients with long-term deficiency or excess of calcitonin. N Engl J Med 1987; 317: 53741. 13. Kimmel PL. Radiologic methods to evaluate bone mineral content, Ann Intern Med 1984; 100: 908-11. 14. Mazess RB, Barden H, Ettinger M, Schultz EI Bone density of the radius, spine, and proximal femur in osteoporosis. J Bone Miner Res 1988; 3: 13-8. 15. Gharib H, Kao PC, Heath H Ill. Determination of silica-purified plasma calcitonin for the detection and management of medullary thyroid carcinoma: comparison of two provocative tests. Mayo Clin Proc 1987; 62: 373-8. 16. Harty M. The anatomy of the hip joint. In: Tronzo RG, ed. Surgery of the hip joint. Philadelphia: Lea and Febiger, 1973: 64-5. 17. Vose GP, Mack PB. Roentgenologic assessment of femoral neck density as related to fracturing. Am J Roentgen01 1963: 89: 1296-301. 18. Vose GP, Lockwood RM. Femoral neck fracturing-its relationship to radiographic bone density. J Gerontol 1965; 20: 300-5. 19. Fish LH, Schwartz HL, Cavanaugh J, Steffes MW, Bantle JP. Oppenheimer JH. Replacement dose, metabolism, and bioavailability of levothyroxine in the treatment of hypothyroidism: role of triiodothyronine in pituitary feedback in humans. N Engl J Med 1987; 316: 76470. 20. Hennessey JV. Evaul JE, Tseng YC. Burman KD, Wartofsky L. L-thyroxine dosage: a reevaluation of therapy with contemporary preparations. Ann Intern Med 1986; 105: 11-5.
90