Interest in the effects of mild primary hyperparathyroidism on the skeleton has recently been increasing. Some investigators, but not all, have reported that the disorder is associated with osteopenia, which primarily affects cortical bone [8-15]; other reports have noted that the disorder is associated with an increased risk for osteoporotic fracture [16, 17]. Thus, asymptomatic primary hyperparathyroidism is considered to be one of the four indications for measurement of bone mineral density [18], and osteopenia in persons with primary hyperparathyroidism is considered an indication for surgical intervention [19]. However, medical therapies for osteopenia have not been rigorously assessed in patients with primary hyperparathyroidism.

In eucalcemic elderly postmenopausal women, estrogen-progestin therapy (hormone replacement therapy) increases bone mineral density [20-23] and reduces the incidence of osteoporotic fractures [22, 24]. This treatment may have the same effects in women with primary hyperparathyroidism, but only limited data on this condition are available. Although high doses of estrogen suppress biochemical markers of bone turnover and reduce serum calcium levels in women with this disorder [25-27], cross-sectional studies of bone mineral density in affected women treated with estrogen have had conflicting results [28, 29]. In an uncontrolled study of postmenopausal women with primary hyperparathyroidism [30], norethisterone preserved bone mineral density of the forearm. However, because this agent has progestogenic and androgenic as well as estrogenic activity, these data are difficult to interpret.

Because skeletal status now has an important influence on the management of persons with primary hyperparathyroidism, the efficacy of potential therapies in preventing bone loss must be clearly delineated. We report the results of a 2-year, randomized, double-blind, placebo-controlled study of the effect of hormone replacement therapy on bone mineral density in postmenopausal women with mild primary hyperparathyroidism.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Seventy-three postmenopausal women with mild primary hyperparathyroidism were identified from the Auckland Hospital Endocrinology Clinic, to which most patients with this condition who live in the Auckland area are referred. In each patient, hypercalcemia was detected incidentally on routine blood testing, and primary hyperparathyroidism was confirmed by the presence of a concomitant elevation of serum ionized calcium and intact parathyroid hormone levels. No patient had a family history of hypercalcemia. Exclusion criteria were concurrent systemic illness; untreated thyroid disease; hepatic or renal dysfunction; undiagnosed genital bleeding; estrogen therapy within the past 6 months; any history of bisphosphonate or fluoride therapy; and current use of a glucocorticoid drug, an anticonvulsant agent, or a thiazide diuretic. Three patients were initially excluded on the basis of these criteria (two patients had histories of breast cancer and uterine cancer, and one patient was currently using glucocorticoid drugs). Information about the study was mailed to the remaining 70 women: Seven of these women could not be contacted, and 3 were ineligible for the study because they were taking estrogen. Of the 60 women who could be contacted and were eligible, 42 (70%) agreed to participate. Because no other selection criteria were applied, the patients are representative of postmenopausal women with mild hyperparathyroidism. None of the five patients receiving thyroid hormone replacement therapy had a suppressed serum thyroid-stimulating hormone level. Two patients who had previously received hormone replacement therapy had discontinued this treatment 18 and 11 years, respectively, before study entry.

Study Protocol

Patients were randomly assigned to receive either continuous combined therapy with conjugated equine estrogens, 0.625 mg/d, and medroxyprogesterone acetate, 5 mg/d, or identical placebo tablets for 2 years. Patients who had had hysterectomy were not given progestin. One woman in the hormone replacement therapy group, who developed substantial mastalgia while receiving 0.625 mg of conjugated estrogens per day, completed the protocol by consuming 0.3 mg/d. Randomization was done manually by an independent researcher who followed a predetermined strategy. The researcher documented the randomization and had no knowledge of the patients' clinical details. This person then supervised the dispensing of the appropriate medication into bottles labeled with the patient's study number. Treatment codes could be accessed only by this researcher and only after a patient was withdrawn from the study.

No attempt was made to influence dietary habits during the study period. The women were seen 1, 2, 3, and 6 months after the study began and every 6 months thereafter until the study ended. The patients' medical history and compliance with trial medication (assessed by tablet counts) between visits were recorded at each 6-month visit. Patients with vaginal bleeding were referred to an independent clinician, who was then notified of the patient's treatment allocation. Placebo recipients with vaginal bleeding were to have endometrial biopsy; hormone recipients were routinely investigated only if bleeding occurred after they had been receiving therapy for 12 months. This practice is consistent with that used in our institution.

Nine women (4 in the hormone replacement therapy group and 5 in the placebo group) withdrew from the study during the 2-year period. In the hormone replacement therapy group, 1 woman withdrew because of mood disturbance, 1 because of the development of thyrotoxicosis, 1 because of diffuse aching in the lower limbs, and 1 because of personal reasons. In the placebo group, 2 women withdrew because of personal reasons, 1 because of seronegative polyarticular arthritis, 1 because of renal impairment, and 1 because of symptomatic hypercalcemia that required parathyroidectomy. Thus, 33 women (17 receiving hormone replacement therapy and 16 receiving placebo) completed the protocol.

The local ethics committee approved the study, and each patient gave written informed consent.

Measurements

Bone mineral density was measured using dual-energy x-ray absorptiometry (DPX-L, Lunar, Madison, Wisconsin). Scans of the whole body, lumbar spine (vertebrae L2 to L4), proximal femur, and forearm were obtained every 6 months and were analyzed using the manufacturer's software (version 1.3). The whole-body scans were used to assess the densities of the subregions as well as the density of the total body. In our laboratory, the precisions of these measurements of bone mineral density are 0.4% for the total body, 1.0% for the lumbar spine, 1.4% for the femoral neck, 2.9% for the Ward triangle, and 1.6% for the trochanter. Lateral spine radiography was done at study entry, and any deformed vertebrae were excluded from subsequent analysis.

Calcium intake was determined by a validated food-frequency questionnaire [31]. The levels of 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D (Nichols Institute, San Juan Capistrano, California) were measured at study entry. At baseline, 6 months, and 2 years, fasting blood samples were collected for measurement of serum levels of ionized calcium, phosphate, total alkaline phosphatase, and intact parathyroid hormone (Nichols Institute); second-urination fasting urine samples were also collected at these times for measurement of levels of hydroxyproline, N-telopeptides (Ostex International, Seattle, Washington), calcium, and creatinine. The methods used to measure these substances have been described previously [32].

Statistical Analysis

We used the SAS statistical package (SAS Institute, Cary, North Carolina) for all statistical analyses. We compared baseline characteristics in the treatment groups using the Student t-test for continuous variables and the chi-square test for categorical variables. Data that were not normally distributed were appropriately transformed before analysis. Within-group changes in biochemical variables (at 6 months and 2 years) and bone mineral density (at 2 years only) from baseline were assessed using the Student t-test for paired data or the Wilcoxon signed-rank test for data that were not normally distributed. Differences between the study groups in absolute changes from baseline values in bone mineral density and biochemical variables were assessed by repeated-measures analysis of variance using data from all time points. For ease of presentation, bone density results are expressed as the percentage of the baseline value. We examined several covariates that putatively affect bone mineral density (including baseline values), but none was found to significantly influence the bone density response. With the exception of serum phosphate levels, results of between-group comparisons of the biochemical data did not significantly change when we included baseline values in the model. Because all comparisons were made a priori, we made no adjustment to . Only the results from the 33 women who completed the study were included in the main analyses. To accommodate an analysis of data from all patients, including those for whom observations were missing, the "mixed" SAS procedure (version 6.11) was used to estimate random-effects variables for a repeated-measures design. Maximum likelihood was used to estimate the unknown covariance parameters for an unstructured block for each patient. Results are presented as mean ± SE unless otherwise specified; all tests were two tailed.

Results

Top Methods Results Discussion Author & Article Info References

At study entry, the patients' clinical characteristics did not significantly differ between groups (Table 1).

Biochemical Studies

The biochemical study results are shown in Table 2. At baseline, the hormone replacement therapy group had higher mean serum phosphate levels than did the placebo group, but levels of total and ionized calcium and intact parathyroid hormone were similar in the two groups. During the study, parathyroid hormone levels increased in both groups (P < 0.01 compared with baseline by both parametric and nonparametric analyses), but the change was similar in both groups (P > 0.2 by repeated-measures analysis of variance). Ionized calcium levels did not change, and total calcium levels decreased significantly in the hormone replacement therapy group, but the between-group difference was not significant. Serum phosphate levels decreased significantly in the hormone replacement therapy group (P < 0.001 compared with baseline); the between-group difference, as shown by repeated-measures analysis of variance (P = 0.03), was no longer significant when the baseline phosphate levels were included in the model.

During the study, markers of bone turnover decreased in the hormone replacement therapy group. Thus, the serum total alkaline phosphatase level, a marker of osteoblast function, decreased by 25% ± 3% after 6 months and was 22% ± 4% lower than the baseline level after 2 years (P < 0.001 compared with baseline and with the placebo group). In the hormone replacement therapy group, urinary markers of bone resorption decreased. Urine hydroxyproline excretion had declined by 33% ± 8% after 6 months (P < 0.001 compared with baseline) and by 38% ± 5% after 2 years (P < 0.001); N-telopeptide excretion had decreased by 49% ± 9% at 6 months (P = 0.0002) and by 60% ± 6% at 2 years (P < 0.001); urinary calcium excretion was 48% ± 5% lower at 6 months (P < 0.001) and 33% ± 9% lower at 2 years (P = 0.002). In the placebo group, the levels of the urinary markers of bone resorption did not change between baseline and 6 months; however, between 6 months and 2 years, excretion of N-telopeptide and calcium decreased for unknown reasons. This decrease was also seen with nonparametric statistical analysis. The decrease in urinary hydroxyproline excretion in the hormone replacement therapy group was significantly greater than the decrease in the placebo group (P = 0.02), whereas the decrease in N-telopeptide excretion tended to be greater in the hormone replacement therapy group (P = 0.06 compared with placebo).

Bone Mineral Density

At baseline, both groups had similar bone mineral densities at all sites (Table 3).

Bone mineral densities of the total body during the study are shown in Figure 1. Total-body bone mineral density increased from the baseline density in the hormone replacement therapy group (mean change, 1.3% ± 0.4%; P = 0.004) and decreased in the placebo group ( –2.3%± 0.7%; P = 0.005). The difference between the groups in the change in bone mineral density at the end of the study was 3.6% ± 0.8% (P < 0.001). Bone mineral densities in the spine and legs (which primarily comprised trabecular and cortical bone, respectively) during the study were significantly higher in the hormone replacement therapy group than in the placebo group (between-group differences after 2 years: spine, 7.3% ± 2.6% [P = 0.002]; legs, 4.1% ± 1.3% [P = 0.0008]).

View larger version (19K): [in this window] [in a new window]   Figure 1. Mean (±SE) bone mineral density (BMD) of the total body and lumbar spine in postmenopausal women with primary hyperparathyroidism who received hormone replacement therapy (black circles) or placebo (white circles) for 2 years. The results are expressed as a percentage of the baseline values. Changes in bone mineral density during the study were significantly more positive in the patients receiving hormone replacement therapy. Figure 1. Mean (±SE) bone mineral density (BMD) of the total body and lumbar spine in postmenopausal women with primary hyperparathyroidism who received replacement therapy (black circles) or placebo (white circles) for 2 years. The results are expressed as a percentage of the baseline values. Changes in bone mineral density during the study were significantly more positive in the patients receiving hormone replacement therapy.

Figure 1 shows the changes in bone mineral density of the lumbar spine. At this site, which is composed primarily of trabecular bone, bone mineral density increased from baseline in the hormone replacement therapy group (5.2% ± 1.4%; P = 0.002) and tended to decrease in the placebo group ( –1.4%± 0.8%; P = 0.09). At the end of the study, bone mineral density at this site was 6.6% ± 1.6% greater in the hormone replacement therapy group than in the placebo group (P = 0.0005).

Figure 2 shows the bone mineral densities at the three proximal femoral sites. In the hormone replacement therapy group, the bone mineral density of the femoral neck increased significantly from baseline (3.4% ± 1.5%; P = 0.05). At the end of the 2-year study period, bone mineral densities of the femoral neck and the femoral trochanter were significantly higher in the hormone replacement therapy group than in the placebo group (between-group differences, 4.8% ± 2.3% [P = 0.04] and 5.3% ± 3.2% [P = 0.02], respectively). Bone mineral density at the Ward triangle tended to be higher in the hormone replacement therapy group (between-group difference, 6.0% ± 4.6%; P = 0.06).

View larger version (18K): [in this window] [in a new window]   Figure 2. Mean (±SE) bone mineral density (BMD) of the proximal femur in postmenopausal women with primary hyperparathyroidism who received hormone replacement therapy (black circles) or placebo (white circles) for 2 years. The results are expressed as a percentage of the baseline values. Changes in bone mineral density of the femoral neck and trochanter were significantly more positive in the patients receiving hormone replacement therapy than in the patients receiving placebo. Changes in the bone mineral density of the Ward triangle tended to be more positive in the hormone replacement therapy group.

The bone mineral densities in the proximal forearm are shown in Figure 3. At this site, bone mineral density tended to increase from baseline in the hormone replacement therapy group (1.9% ± 1.1%; P = 0.1) and tended to decrease in the placebo group ( –3.5%± 1.2%; P = 0.01). The between-group difference progressively increased, reaching 5.4% ± 1.6% at the end of 2 years (P = 0.01).

View larger version (11K): [in this window] [in a new window]   Figure 3. Mean (±SE) bone mineral density (BMD) of the proximal forearm in postmenopausal women with primary hyperparathyroidism who received hormone replacement therapy (black circles) or placebo (white circles) for 2 years. The results are expressed as a percentage of the baseline values. Changes in bone mineral density were significantly more positive in patients receiving hormone replacement therapy.

The intention-to-treat analysis confirmed these findings at each assessment site and showed increased statistical significance of the treatment effect in every case.

Adverse Events

Compliance with study medications was 95% in each group. Mastalgia occurred in 14 (83%) of the women receiving hormone replacement therapy and in 6 (38%) of the women receiving placebo (P < 0.05). Vaginal bleeding also occurred more frequently in the hormone replacement therapy group (53%) than in the placebo group (0%) (P < 0.001). However, both of these events were mild and self-limiting, and neither caused a patient to withdraw from the study. Only one woman in the hormone replacement therapy group developed vaginal bleeding after the end of the first year of the study. This patient had ultrasonography and biopsy of the endometrium, which showed endometrial atrophy. The frequencies of other adverse events did not significantly differ between the groups.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study shows that hormone replacement therapy substantially increases bone mineral density throughout the skeleton in postmenopausal women with mild primary hyperparathyroidism. The greater response in the density of trabecular-rich bones, such as the spine, reflects their greater surface-to-volume ratio and consequently greater turnover rate; this response is commonly seen with hormone replacement therapy and other antiresorptive therapies [33]. The changes we saw in bone mineral density are similar in magnitude to those reported in eucalcemic elderly menopausal women receiving hormone replacement therapy [21-23]. Whether bone mineral density would increase by the same extent in younger postmenopausal women with primary hyperparathyroidism is uncertain because bone mineral density tends to increase less in younger, eucalcemic postmenopausal women than in women in their sixties. In our study, only four women who had had menopause less than 10 years before study entry were allocated to receive hormone replacement therapy. Bone mineral density of the total body at 2 years had increased by 2.0% in these women and by 1.3% in the older women in the hormone replacement therapy group. This finding suggests that the benefits occur across a wide age range. It should also be remembered that the women in our study were not selected according to age; thus, the response of bone density to this treatment is probably typical of the response in the population of postmenopausal women with this condition.

Hormone replacement therapy reduces the risk for fracture in postmenopausal women with osteoporosis [22, 24]. Our findings therefore suggest that hormone replacement therapy may offer protection against fractures in postmenopausal women with mild primary hyperparathyroidism. Although controversy exists as to whether the risk for fracture in women with primary hyperparathyroidism is greater than that in age-matched eucalcemic controls [16, 17, 34-36], the observation that primary hyperparathyroidism occurs most commonly in elderly postmenopausal women implies that the absolute risk for fracture is high in postmenopausal women with this disease. Many (>80%) of the patients in the hormone replacement therapy group completed the current protocol; this confirms our previous finding that continuous combined estrogen-progestin therapy, which produces an atrophic endometrium and amenorrhea, is an acceptable long-term therapy in elderly postmenopausal women [23].

We used a double-blind, randomized, placebo-controlled study design to minimize the likelihood that extraneous biases would influence our results. Maintaining blinding can be difficult with a therapy such as hormone replacement therapy because side effects may make treatment allocations apparent. The most frequent side effect of hormone replacement therapy in our study, mastalgia, also occurred in many placebo recipients; however, the occurrence of vaginal bleeding reliably indicated the use of hormone replacement therapy. Thus, some patients may have learned of their study assignment. This is unlikely to have influenced the conduct of the study, however, because compliance and withdrawal rates were similar in the two groups and because bone density and biochemical variables were measured by persons who were unaware of the occurrence of vaginal bleeding. Thus, the comparability of the groups, which resulted from the initial randomization, was maintained.

The finding that the mean values of biochemical markers of bone formation and resorption at trial entry were equal to or greater than the upper limit of the normal range is consistent with previous evidence that bone turnover is increased in patients with primary hyperparathyroidism [37]. In the hormone replacement therapy group, these bone turnover indices decreased significantly in the first 6 months of treatment; these findings are consistent with the known antiresorptive properties of this therapy. The reason for the smaller decline in urinary indices of bone resorption in the placebo group near the end of the study is unknown. It is unlikely that this smaller decrease is attributable to difficulties with the assays because it was seen with three independent measurements and because these variables did not change in the hormone replacement therapy group during this interval. The smaller decrease is probably a random biological fluctuation that differs in timing and magnitude from the treatment effect seen in the estrogen group. The reduction in turnover after hormone replacement therapy contributes to the beneficial effects of this intervention on bone mass but may itself provide some antifracture protection by reducing the rate of trabecular plate perforations.

Previous short-term studies of the effects of unopposed estrogen on biochemical indices in primary hyperparathyroidism have emphasized the desirability of normalizing serum calcium levels [25-27]. In some patients, this may be achieved by giving estrogen doses that are generally higher than those used in eucalcemic elderly postmenopausal women [26]. We showed that standard-dose hormone replacement therapy protects against the bone-resorbing effects of primary hyperparathyroidism without decreasing serum calcium levels. This finding is not necessarily surprising because it is the balance between formation and resorption (rather than the absolute rate of either) that determines the skeleton's effect on serum calcium. This effect is further modified by urine calcium excretion (which estrogen reduces) and intestinal calcium absorption (which parathyroid hormone modulates to maintain the serum calcium level at the predetermined level through the hormone's effect on vitamin D metabolism). This dissociation of turnover and serum calcium levels is seen in normal women when they begin to receive hormone replacement therapy [38] and with other antiresorptive agents, such as the bisphosphonates; long-term therapy with the latter agents has effects similar to those seen in our study [33]. Estrogen's greater effect on total calcium than on ionized calcium has also been noted previously and probably occurs because estrogen decreases only the complexed fraction of serum calcium [39]. However, because complexed calcium is filtered at the glomerulus, the reduction of this fraction caused by estrogen contributes to the decrease in urine calcium excretion. The difference between our findings and those of previous studies of the effects of unopposed estrogen on serum calcium in patients with primary hyperparathyroidism may result from the lower estrogen dose used in our study or from the concomitant use of progestin in most of our patients.

A decrease in serum phosphate levels after hormone replacement therapy has been observed in most of the previous studies of this intervention in primary hyperparathyroidism [25-27]; serum phosphate levels also decrease with norethindrone [30]. This decrease is associated with reduced tubular reabsorption of phosphate [26, 30]; this association implies that the decrease represents an effect of estrogen on renal tubular function. Reduced bone resorption might also contribute. Few studies that found reduced phosphate levels after hormone treatment documented changes in parathyroid hormone concentrations. Thus, alterations in the activity of this hormone on the renal tubule are unlikely to be involved.

Our findings do not support the hypothesis that estrogen-progestin treatment has a specific "antiparathyroid hormone" effect, because serum ionized calcium levels remained unchanged and the indices of bone turnover after treatment were still higher than would be expected for normal, estrogen-replete women. Furthermore, the changes in serum phosphate levels were the opposite of what would be expected if estrogen were to act in this manner. Thus, the effects of excess parathyroid hormone were still clearly shown in the women receiving hormone replacement therapy. The small increase in intact parathyroid hormone levels, which occurred in both treatment groups, was not attributable to assay drift because the assay standards were stable throughout the study period. The levels of this hormone in a cohort of normal persons who were involved in a separate study that covered the same time interval also remained stable. Other researchers [40] have shown an increase in intact parathyroid hormone levels of this magnitude in untreated patients with primary hyperparathyroidism; this increase may be an effect of aging or gradual disease progression [41].

Parathyroidectomy is currently recommended for persons with osteopenia and primary hyperparathyroidism [19], and the presence of osteopenia has become an increasingly frequent indication for surgical intervention in affected patients. Surgical cure of primary hyperparathyroidism produces an increase in bone mineral density, but the magnitude of the reported responses varies [9, 10, 12, 14, 42-46]. Differences in study populations and design may explain some of this variation in response because several of the surgical studies included symptomatic patients with more severe disease [9, 10, 14, 42, 43, 45], who might be expected to have a greater response to treatment because of higher bone turnover before therapy. It is persons with mild disease whose management is likely to be influenced by bone mineral density measurement, because symptomatic patients and those with more severe hypercalcemia will be routinely referred for parathyroidectomy. Studies that have assessed the effect of parathyroidectomy on bone mineral density in patients with mild disease [12, 44, 46] have shown increases at cortical sites that are similar to those we observed in response to hormone replacement therapy. At sites containing a substantial component of trabecular bone, the effects of parathyroidectomy are less clear. Thus, although two studies of lumbar spine bone mineral density after parathyroidectomy [47, 48] have reported increases similar to those we saw in response to hormone replacement therapy, a third study [46] reported increases of 8% at the lumbar spine and 6% at the femoral neck in the first 2 years after parathyroidectomy. The same investigators found no ongoing bone loss in postmenopausal women with untreated primary hyperparathyroidism [49]; thus, the between-group difference during a 2-year period (patients who had had parathyroidectomy compared with controls) was similar to that reported in our study. A randomized trial comparing the effects of hormone replacement therapy with those of parathyroidectomy is needed to determine whether either intervention provides superior skeletal protection. However, our data suggest that hormone replacement therapy is an acceptable alternative to surgery for the many postmenopausal women with mild primary hyperparathyroidism in whom osteopenia is the primary indication for intervention. Our findings may be particularly pertinent to the management of elderly patients or patients who have other illnesses that increase the risk for general anesthesia.

Estrogen has extraskeletal effects that may influence its use in postmenopausal women with primary hyperparathyroidism. Cardiovascular disease is the most common cause of death in postmenopausal women [50]. Epidemiologic studies suggest that estrogen reduces the risk for coronary artery disease in normal postmenopausal women by about 50% [50, 51]. This protective effect may be greater still in women with prevalent arterial disease [52]. Some evidence suggests that primary hyperparathyroidism is associated with an increased risk for death from cardiovascular disease [53]. Thus, postmenopausal women with primary hyperparathyroidism may obtain substantial cardiovascular protection from the use of estrogen. Although uncertainty exists as to whether the risk for breast cancer is slightly increased in long-term users of postmenopausal estrogen therapy [54, 55], the cardiovascular and skeletal benefits of estrogen are likely to outweigh any adverse effects on breast disease in the overall balance of risks and benefits associated with the use of estrogen in primary hyperparathyroidism.

Urolithiasis occurs more frequently in patients with primary hyperparathyroidism [2], and its presence is commonly cited as an indication for parathyroidectomy [19]. The hypercalciuria that frequently occurs in persons with primary hyperparathyroidism probably contributes to the development of urinary tract stones. Our findings show that hormone replacement therapy reduces urinary calcium excretion to a level well within the normal range, suggesting that the therapy may reduce the occurrence of urolithiasis in primary hyperparathyroidism.

Our randomized, controlled trial shows that hormone replacement therapy significantly increases bone mineral density and reduces urinary calcium excretion and bone turnover in postmenopausal women with mild primary hyperparathyroidism. The changes in bone mineral density, which occurred without suppression of serum calcium, are likely to be associated with a substantially reduced risk for fracture. Our findings suggest that hormone replacement therapy should be considered an alternative to parathyroidectomy in the many postmenopausal women with mild primary hyperparathyroidism in whom osteopenia is the reason for intervention. Our results also indicate that the efficacy of hormone replacement therapy in treating mild primary hyperparathyroidism should not be judged solely on the basis of effects on serum calcium levels.

Dr. Grey: Department of Endocrinology, Yale University School of Medicine, PO Box 208020, New Haven, CT 06520.