| |
Association of Pioglitazone Treatment with Decreased Bone Mineral Density in Obese Premenopausal Patients with Polycystic Ovary Syndrome: A Randomized, Placebo-Controlled Trial
|
| |
| |
The Journal of Clinical Endocrinology & Metabolism May 2008 Vol. 93, No. 5 1696-1701
Dorte Glintborg, Marianne Andersen, Claus Hagen, Lene Heickendorff and Anne Pernille Hermann
Department of Endocrinology and Metabolism (D.G., M.A., C.H., A.P.H.), Odense University Hospital, DK-5000 Odense C, Denmark; and Department of Clinical Biochemistry (L.H.), Aarhus University Hospital, DK-8000 Aarhus C, Denmark
"Pioglitazone treatment was followed by decreased lumbar and hip BMD and decreased measures of bone turnover in a premenopausal study population relatively protected from bone mineral loss..... During 16 wk of pioglitazone treatment, the average decline of BMD levels was 1.1 and 1.4% at the lumbar spine and femur neck, respectively..... The present study included patients with high levels of testosterone, estradiol, and insulin, all factors known to be protective against bone mineral loss..... According to the changes in bone markers, the mechanism for bone mineral loss in the present and previous short-term studies seems to be altered bone balance rather than bone turnover. Bone loss may, therefore, continue throughout the treatment period..... ALP and PTH levels decreased significantly during treatment with pioglitazone, suggesting decreased osteoblast activity. No significant changes were measured in osteocalcin, ICTP, or 25OHD values."
Abstract
Objective: Our objective was to investigate the effect of pioglitazone on bone mineral density (BMD) and bone turnover markers in polycystic ovary syndrome (PCOS).
Design and Setting: We conducted a randomized, placebo-controlled study at an outpatient clinic at a university hospital.
Patients: Thirty premenopausal patients with PCOS and 14 age- and weight-matched healthy females participated.
Interventions: Pioglitazone (30 mg/d) or placebo was given for 16 wk.
Main Outcome Measures: Measurements of BMD [hip (neck and total) and lumbar spine (L2-L4)], bone metabolic parameters [alkaline phosphatase (ALP), 25-hydroxyvitamin D, C-telopeptide of type I collagen (ICTP), osteocalcin, and PTH], endocrine profiles (testosterone, estradiol, and insulin), and body composition (waist to hip ratio, body mass index, and whole-body dual-energy x-ray absorptiometry scans) were performed.
Results: Patients with PCOS had significantly higher levels of ICTP, fasting insulin, and testosterone than controls, whereas no differences were measured in ALP, PTH, body composition, or BMD. Pioglitazone treatment was followed by reduced BMD [geometric means (-2 to +2 SD)]: lumbar spine 1.140 (0.964-1.348) vs. 1.127 (0.948-1.341) g/cm2 (average decline 1.1%) and femoral neck 0.966 (0.767-1.217) vs. 0.952 (0.760-1.192) g/cm2 (average decline 1.4%), both P < 0.05. Both ALP and PTH decreased significantly during pioglitazone treatment, whereas no significant changes were measured in 25-hydroxyvitamin D, ICTP, osteocalcin, sex hormones, and body composition.
Conclusion: Pioglitazone treatment was followed by decreased lumbar and hip BMD and decreased measures of bone turnover in a premenopausal study population relatively protected from bone mineral loss.
Introduction
Polycystic ovary syndrome (PCOS) is a common but heterogeneous disease characterized by anovulation, hyperandrogenism, and/or polycystic ovaries (1). Previous studies found normal or increased bone mineral density (BMD) in patients with PCOS compared with controls (2, 3, 4, 5, 6, 7). Several factors may contribute to the conserved BMD in PCOS. Patients with PCOS have relatively high levels of estradiol and are characterized by abdominal obesity and insulin resistance (1). Abdominal obesity and increased visceral fat mass are seen in overweight (8) and normal-weight patients with PCOS (9). Adiposity is known to be positively associated with BMD (10). Furthermore, more than 50% of patients with PCOS are insulin resistant, and previous studies suggested that hyperinsulinemia, independent of body mass index (BMI), may protect against the development of osteoporosis (11, 12). Only few studies found positive associations between testosterone and BMD in PCOS (7, 13, 14), but this may in part be explained by the use of imprecise testosterone assays in most studies (15, 16).
Increased insulin sensitivity in patients with PCOS is followed by improved ovulatory function (17, 18). Thiazolidinediones, such as rosiglitazone and pioglitazone, were recently used as insulin sensitizers in patients with PCOS and effectively improved insulin sensitivity and ovulatory function (17). Thiazolidinediones stimulate the peroxisome proliferator-activated receptor (PPAR)-y receptors in the cell nucleus and, thereby, activate the transcription of genes that affect glucose and lipid metabolism mediating decreased peripheral adipocyte lipolysis, decreased free fatty acid levels, and decreased visceral fat mass (19, 20, 21). PPARy stimulation may, however, affect the regulation of the pluripotent mesenchymal stem cells and stimulate differentiation into adipocytes in preference over osteoblasts (22, 23). In previous animal studies, PPARy agonist treatment was followed by increased bone loss and impaired osteoblast function (24, 25). Few studies evaluated the effects of PPAR agonist treatment on BMD and measures of bone turnover in humans. In a recent study in patients with type 2 diabetes, increased numbers of peripheral fractures were reported in middle-aged women treated with the thiazolidinedione rosiglitazone (26). In a randomized study in healthy postmenopausal women, rosiglitazone treatment was associated with significantly decreased BMD and decreased markers of bone formation (27).
Animal studies found that bone loss was primarily seen in ovariectomized rats treated with rosiglitazone, suggesting a protective effect of estradiol on BMD during treatment with PPAR agonists (28). A study in young as well as older mice found that PPAR agonist treatment had adverse effects on BMD only in aged animals (29). It is to be determined whether premenopausal women are protected from the adverse effects of PPAR agonist treatment.
Previous studies found no increased fracture rate after PPAR agonist treatment in male patients with type 2 diabetes (26, 30). These findings may suggest a protective effect of high testosterone.
The aim of the present study was to evaluate the effect of the PPAR agonist pioglitazone on BMD and measures of bone turnover in a randomized study of obese, hyperandrogenemic, insulin-resistant premenopausal women with PCOS.
Results
Sex hormones, insulin, and body composition (Table 1)
Patients with PCOS (n = 30) had significantly higher levels of fasting insulin, estradiol, and free testosterone than controls (pooled patient data not shown). Absolute lower-extremity fat mass was decreased in patients with PCOS compared with controls. Otherwise, no significant differences were found in body composition variables in patients vs. controls.
Treatment with pioglitazone decreased levels of fasting insulin, whereas no significant changes were observed in testosterone or estradiol levels. A minor, but significantly decreased lower-extremity lean body mass was measured during treatment with pioglitazone. The results on insulin and testosterone have been presented previously (32).
BMD and bone metabolic parameters (Tables 2 and 3 and Fig. 1)
No significant differences were measured in BMD between patients and controls. ICTP levels were significantly higher in patients with PCOS compared with controls, but otherwise, no significant differences were measured in bone metabolic parameters.
BMD of L2-L4 and hip (neck and total) were significantly decreased during treatment with pioglitazone vs. placebo. ALP and PTH levels decreased, whereas no significant changes were measured in osteocalcin, 25OHD, and ICTP levels.
Baseline correlations
Metabolic bone markers In patients with PCOS and controls, osteocalcin was correlated significantly and positively with ICTP (PCOS, r = 0.40, P = 0.04; controls, r = 0.51, P = 0.05).
BMD In patients with PCOS, total hip BMD significantly correlated with BMI (r = 0.41; P = 0.02), trunk fat mass (r = 0.41; P = 0.02), trunk lean body mass (r = 0.46; P = 0.01), and lower-extremity lean body mass (r = 0.43; P = 0.02). Borderline significant (P = 0.05-0.1) correlations were found between femoral neck BMD and body composition measures. No significant correlations were found between BMD and estradiol or testosterone.
In controls, lumbar spine BMD correlated with lower-extremity lean body mass (r = 0.59; P = 0.03). Femoral neck BMD correlated positively with BMI (r = 0.58; P = 0.03) and inversely with estradiol (r = -0.61; P = 0.03).
No significant correlations were found between metabolic bone markers and BMD levels in patients with PCOS or controls.
Correlations between changes in parameters
No significant correlations were found between changes in BMD and changes in body composition, insulin, PTH, or ALP during treatment with pioglitazone.
Discussion
In the present study, we documented significantly decreased BMD levels during treatment with pioglitazone in a group of premenopausal, obese patients with PCOS. During 16 wk of pioglitazone treatment, the average decline of BMD levels was 1.1 and 1.4% at the lumbar spine and femur neck, respectively.
Previous studies in humans on the effects of PPAR agonist treatment on fracture risk (26) and BMD (35) in patients with diabetes showed that adverse effects were seen in women only. In the ADOPT study, women from 30-60 yr of age were included (26). The study by Schwartz et al. (35) included women who were 70-79 yr. Since these clinical studies, the pharmaceutical company Takeda evaluated the data from randomized trials in 8100 women treated with pioglitazone and 7400 women treated with comparator drugs and found a fracture incidence of 1.9 per 100 patient-years during treatment with pioglitazone vs. 1.1 per 100 during treatment with comparator drugs (30).
We are aware of only one previous randomized study evaluating the effect of PPARy agonist treatment on BMD levels in women. In this study, Gray et al. (27) examined 50 healthy postmenopausal women with a minimum age of 55 yr randomized to rosiglitazone or placebo for 14 wk, and patients treated with rosiglitazone had average losses of BMD of 1.9% for total hip and 1.2% for lumbar spine.
Based on the fact that men did not have increased fracture risk during PPARy agonist treatment, it was speculated that higher levels of sex steroids could protect against fractures (36). This theory was supported by previous studies in mice (25, 29, 37) and rats (28), where young (29) and well-estrogenized (28) animals seemed to be relatively protected against bone loss during PPARy agonist treatment. In the present study, the average decline in BMD was, however, comparable to the decline found by Gray et al. during treatment with rosiglitazone vs. pioglitazone. These findings suggest similar adverse effects of PPARy agonist treatment in premenopausal and postmenopausal women.
The present study included patients with high levels of testosterone, estradiol, and insulin, all factors known to be protective against bone mineral loss. Patients with PCOS tended to have higher BMD levels than controls, although this did not reach statistical significance. Previous studies that included a higher number of patients with hyperandrogenism found them to have higher BMD values than controls, suggesting that the present study lacked power to detect differences in BMD (2, 14, 38). We found significant correlations between BMD and measures of body composition, whereas only minor correlations were found between BMD and estradiol, insulin, and testosterone. In a previous study in a larger population, we reported significant positive correlations between insulin and BMD and between testosterone and BMD, which were not reproduced in the present study (14). During treatment with pioglitazone, insulin levels significantly decreased, whereas no significant changes were measured in body composition or testosterone levels. No significant correlations were found between changes in insulin and changes in BMD. It is still possible that decreased insulin levels and minor changes in body composition or sex hormone levels may have influenced BMD levels during pioglitazone treatment. Additional testing of this hypothesis would require that the effects of PPARy agonist treatment be evaluated in insulin-sensitive individuals.
The fact that patients with PCOS had normal to slightly elevated levels of estradiol and that estradiol remained unchanged throughout the treatment period suggests that treatment with sex hormones is likely to have no or only a minor effect to preserving BMD in premenopausal women. It remains to be determined whether low-dose hormone therapy may be of benefit in postmenopausal women treated with PPAR agonists as mentioned recently by Schwartz and Cummings (36).
ALP and PTH levels decreased significantly during treatment with pioglitazone, suggesting decreased osteoblast activity. No significant changes were measured in osteocalcin, ICTP, or 25OHD values. Our findings are in agreement with the study by Gray et al. (27) and another recent study determining levels of bone turnover, but not BMD, in 56 postmenopausal patients with diabetes randomized to placebo or rosiglitazone for 12 wk (39). In these studies, ALP levels significantly decreased during treatment with rosiglitazone. Berberoglu et al. (39), furthermore, noted significantly decreased bone-specific ALP levels during treatment with this agent. Gray et al. (27) reported significantly reduced osteocalcin levels during rosiglitazone treatment, which further suggested impaired function of osteoblasts at later stages of differentiation. This finding was not reproduced in the present study or in the study by Berberoglu et al. (39). In agreement with the findings in our study, markers of osteoclast activity [ICTP (27) and urine deoxypyridinoline (39)] did not change significantly during rosiglitazone treatment.
In animal studies, treatment with PPARy agonist decreased bone volume and trabecular number, whereas bone marrow fat content increased (22, 25, 28, 29, 37). In most animal studies, measures of osteoclast activity were unchanged (22, 37). Heterozygote PPAR-y deficient mice had high bone mass and increased osteoblastogenesis, whereas osteoclast and osteoblast functions were normal (40).
In the present study and in previous human studies, BMD was evaluated by DXA scans. Increased marrow fat mass may be a possible confounder for decreased BMD levels established by DXA scans. The fact that we and others reported both decreased BMD and changes in bone markers, however, suggests that decreased BMD levels during PPARy agonist treatment cannot be due to increased bone marrow fatty infiltration alone. According to the changes in bone markers, the mechanism for bone mineral loss in the present and previous short-term studies seems to be altered bone balance rather than bone turnover. Bone loss may, therefore, continue throughout the treatment period. Long-term intervention studies with BMD measurements, as well as fracture assessments, are much needed in both pre- and postmenopausal populations.
In conclusion, treatment with pioglitazone of insulin-resistant premenopausal patients with PCOS was followed by significantly decreased BMD at the hip and lumbar spine and decreased markers of bone mineral turnover. These findings suggest that pioglitazone may have adverse effects on BMD even in a study population relatively protected from bone mineral loss.
Subjects and Methods
Subjects
The patient population in this study has been previously described (31, 32). Thirty reproductive-aged Caucasian women with PCOS were included in the study. Criteria for PCOS were irregular periods with cycle length more than 35 d in combination with total and/or free testosterone above the upper limits of the reference interval (total testosterone > 1.8 nmol/liter, and free testosterone > 0.035 nmol/liter) and/or hirsutism. The study was planned before the definition of the Rotterdam criteria, and therefore, transvaginal ultrasound with a finding of polycystic ovaries was not included in the definition of PCOS. Patients included had elevated fasting insulin levels (>50 pmol/liter) and/or were overweight (BMI 30 kg/m2). At the time this study was planned, pioglitazone had not been reported to induce loss of BMD. Therefore, the number of patients recruited was based on a power calculation involving the expected changes in glucose infusion rate during the euglycemic-hyperinsulinemic clamp studies as the primary outcome variable.
Patients with diabetes (fasting plasma glucose 7.0 mmol/liter), hypertension, elevated liver enzymes, serum prolactin or TSH outside the reference interval, renal dysfunction, and congestive heart failure were not included in the study. Patients discontinued oral contraceptives for at least 3 months before evaluation, and no patient was taking any medication known to affect hormonal or metabolic parameters.
Fourteen healthy, Caucasian, premenopausal women matched to patients with PCOS for BMI and age were studied as controls. All controls had regular menses and did not have hyperandrogenism or hirsutism.
The study was approved by the local ethics committee and by the Danish Medicines Agency, and all subjects gave written informed consent.
The trial was retroactively registered at www.clinicaltrials.gov (registration number NCT00145340).
Methods
Examinations were performed during the follicular phase in patients with oligomenorrhea. Patients with amenorrhea (period length >3 months) were examined randomly according to the protocol previously described (31, 32). After initial examination, patients were randomized to pioglitazone (30 mg/d) or placebo treatment for 16 wk in a double-blind fashion. Safety parameters included liver enzymes, electrolytes, white blood cell count, and pregnancy test. Safety parameters were repeated after 1- and 2-month treatment periods and before the final visit.
Two patients were excluded from the study. One patient in the placebo group became pregnant, and one patient on pioglitazone experienced side effects (dizziness, ankle edema, and anxiety) and was excluded after 1 wk of treatment. No other patient in the pioglitazone group experienced side effects that could be related to treatment with pioglitazone.
Fasting blood samples
Fasting blood samples were analyzed for testosterone, estradiol, insulin, and bone metabolic parameters alkaline phosphatase (ALP), 25-hydroxyvitamin D (25OHD), C-telopeptide of type I collagen (ICTP), and PTH.
Dual-energy x-ray absorptiometry (DXA) scans
BMD DXA scans in array mode (Hologic QDR-4500) were used to measure bone mineral content of the lumbar vertebrae (L1-L4) and proximal left femur (neck and total hip). Technical performance was monitored by daily calibration scans using an anthropomorphic Hologic phantom. The coefficient of variation (CV) for replicate scans of the same individual was 0.7% for the spine and 0.9% for the proximal femur. CV was determined by repeated scanning of 10 of the study participants.
Body composition DXA in whole-body array mode (Hologic QDR-4500) was used to measure whole-body fat mass and to differentiate between peripheral (legs and arms) and central fat mass. Default software readings were used to divide the body into six compartments: head, trunk, arms, and legs. The trunk was defined by a horizontal line below the chin and vertical lines passing through colli femori. The arm regions were separated from the trunk at the levels of the shoulder joint. Body fat mass was determined for each region. Whole-body fat mass was calculated by subtracting head fat mass from total fat mass. Technical performance was monitored by daily calibration scans using an anthropomorphic Hologic phantom. The CV for replicate scans of the same individual was 0.8% for whole-body fat mass.
Assays
Serum total testosterone and SHBG were analyzed using a specific RIA after extraction as previously described (33). In this method, testosterone, dihydrotestosterone, and androstenedione are extracted before applying RIA, and overestimation of testosterone levels was avoided. This method shows a close correlation with the determination of testosterone levels by mass spectrometry. The intraassay CVs for total testosterone and SHBG were 8.2 and 5.2%, respectively. The interassay CVs for total testosterone and SHBG were 13.8 and 7.5%, respectively. Free testosterone levels were calculated from measurements of total testosterone and SHBG.
Serum levels of insulin and estradiol were analyzed by time-resolved fluoroimmunoassay using commercial kits (AutoDELFIA; Perkin-Elmer Life Sciences, Oy, Turku, Finland). Intraassay CVs for estradiol and insulin were 3.8-5.2 and 2.1-3.7%, respectively. Interassay CVs for estradiol and insulin were 3.7-8.5 and 3.4-4.0%, respectively.
ICTP was measured using a RIA kit (Orion Diagnostica Oy, Espoo, Finland) with an intraassay CV ranging from 4.8-9.4%, and interassay CV was 5.6-6.4% as determined by the manufacturer.
ALP was measured using routine catalytic methods (Modular; Roche Diagnostics GmbH, Mannheim, Germany). Intraassay and interassay CVs for ALP were 6.4 and 4.8-5.6%, respectively.
Plasma 25-hydroxyvitamin D2 and D3 (25OHD) were determined by isotope-dilution liquid chromatography-tandem mass spectrometry. Interassay CVs were 9.4-9.7% for both metabolites. ICTP was analyzed by a RIA from Orion Diagnostica (Oulunsalo, Finland), and intra- and interassay CVs were less than 6% and less than 8%, respectively. Osteocalcin (N-terminal midfragment (N-MID) osteocalcin) and intact PTH were analyzed by electrochemiluminescence immunoassays using an automated instrument (Elecsys 2010; Roche). The osteocalcin method detected intact osteocalcin and the large N-mid terminal fragment. Intra- and interassay CVs were less than 2% and less than 4% for osteocalcin, respectively, and less than 4% and less than 7% for PTH, respectively.
Statistical analysis
Fasting hormonal and metabolic variables were log-Gaussian distributed. Data were back-transformed and presented as geometric means (-2 to +2 SD) as previously described by Altman (34). Pretreatment differences between patients with PCOS in the pioglitazone and placebo group and controls were tested using unpaired t tests on log-transformed values. The effects of pioglitazone and placebo treatment were evaluated by performing paired t test on log-transformed values. Basal differences between patients randomized to pioglitazone and placebo were taken into account by comparing delta-values of hormonal and metabolic variables between the placebo and pioglitazone groups using Mann Whitney U tests. Similar significant results were found when using these two approaches. Possible mechanisms for changes in BMD during treatment with pioglitazone were evaluated correlating significant -delta values of BMD with significant delta-values of hormonal and metabolic variables using Spearman nonparametric correlation tests in pioglitazone-treated patients.
All statistics were performed using SPSS 13.0 (SPSS Inc., Chicago, IL) for calculations, and P values < 0.05 were considered significant.
|
|
| |
| |
|
|
|
