Introduction

Androgen deprivation therapy (ADT) is the standard treatment for recurrent, advanced, and metastatic prostate cancer. It is also frequently used in men with early stage disease to prevent cancer progression. In Japan, nearly half of patients with localized or locally advanced prostate cancer receive ADT without any other additional treatment [1]. Since it is possible for the duration of ADT to be prolonged in these conditions, associated long-term adverse effects should be taken into account by patients and physicians [2]. Osteoporosis has emerged as a clinically important adverse effect of ADT. Bone mineral density (BMD), a surrogate for fracture risk, decreases significantly during short-term and long-term treatment with ADT [3]. The annual loss of bone mass ranges from 2−8% at the lumbar spine and 1.8–4.1% at the femoral neck during ADT, which is five- to ten-fold higher than the 0.5–1.0% loss in the general population of aging men [4, 5]. ADT is associated with a significantly greater risk of clinical fractures [6, 7]. Clinical fractures during ADT correlate with shorter overall survival [8]. Decreased BMD and increased fracture risk in men receiving ADT are mostly reported from Western countries. The issue has not yet been investigated adequately in the Asian population. Therefore, we undertook a cross-sectional survey of investigations related to BMD of both non-metastatic prostate cancer (NMPC) patients who have not yet received ADT (hormone naive) and patients receiving prolonged ADT in Japan.

Materials and methods

Consecutive Japanese male patients with NMPC who attended our institution from January 2011 to November 2016 who were receiving continuous ADT or who were planning to receive ADT (hormone naive) were enrolled in this study. All of their prostate cancers had been proven pathologically by needle biopsy, and imaging tests (computed tomography, magnetic resonance imaging, and bone scintigraphy) revealed no metastatic lesions. ADT included gonadotropin-releasing hormone agonists (GnRHa) or combinations of GnRHa and anti-androgens. Patients receiving intermittent ADT and patients with castration-resistant prostate cancer were excluded. Moreover, patients with bone metabolic disease, including Paget’s disease, osteomalacia, hyperprolactinemia, hyper- or hypothyroidism, hyperparathyroidism, and Cushing disease, or previous or concomitant treatment with bone-modifying agents, including bisphosphonates, denosumab, parathyroid hormones, selective estrogen receptor modulators, calcitonin, and calcitriol, were excluded from the study. The study protocol was approved by the local institutional review board.

BMD of the posteroanterior (PA) spine (L2–L4) and non-dominant femoral neck was measured using dual-energy X-ray absorptiometry (DEXA). DEXA was performed using QDR-Discovery (Hologic, Inc., Marlborough, MA, USA). T-scores and young mean adult (YAM) values were calculated using the Hologic database for East Asian ethnicity. The coefficient of variation of BMD at our institution was 1.0% at both the PA spine and total hip. X-ray examinations of the PA spine and lateral spine (cervical, thoracic, lumbar) were performed to assess the presence of silent vertebral fractures. To assess patient characteristics, medical records and questionnaires were investigated to check whether they had any factors that possibly affected BMD such as body mass index (BMI), previous bone fractures of the spine or hip, any bone fractures during ADT, family history of bone fractures of the spine or hip, smoking status, alcohol excess, hypertension, diabetes, rheumatoid arthritis, chronic kidney disease (CKD) and steroid administration. Bone metabolism markers, bone alkaline phosphatase (BAP) as a bone formation marker, and urine type I collagen cross-linked N-telopeptide (urine NTx) or tartrate-resistant acid phosphatase 5b (TRACP-5b) as bone resorption markers were also measured to assess bone metabolism. Urine specimens for measuring urine NTx were obtained in the morning.

Baseline characteristics were compared using Student’s t test, ANOVA, or Pearson’s correlation analysis and Pearson’s chi-squared test, respectively, for normally distributed continuous variables and for categorical variables. Stepwise regression analysis was used to find significant variables affecting BMD. Moreover, logistic regression analysis was performed to investigate significant variables influencing the diagnosis of osteoporosis. Statistical analyses were performed using StatView® 5.0 statistical software. Values were reported as the mean ± SD [median (IQR)] unless otherwise specified. All p values were two sided, and p < 0.05 was considered statistically significant.

Results

A total of 230 patients with NMPC were evaluated. Patient characteristics are summarized in Table 1. Of the 230 patients, 151 (65.7%) were receiving ADT, and 79 (34.4%) had not yet received ADT (hormone naive). The mean duration of ADT among the 151 patients who were receiving ADT was 37.4 ± 30.7 [median 31 (IQR 13.5, 52.5)] months. BMD was measured in 79 hormone-naive patients (34.4%), 63 patients (27.4%) who received <2 years of ADT, 47 patients (20.4%) who received from 2−4 years of ADT, 21 patients (9.1%) who received from 4−6 years of ADT, and 20 patients (8.7%) who received ≥6 years of ADT.

Table 1 Patient characteristics
Full size table

The results of DEXA are shown in Table 2. As the duration of ADT increased, lumbar spine BMD and femoral neck BMD decreased gradually (p = 0.0005 and p = 0.0014, respectively). Univariate analyses revealed that significant variables positively affecting lumbar spine BMD were diabetes and BMI; significant variables negatively affecting lumbar spine BMD were spinal fracture on radiography, duration of ADT, BAP, urine NTx, and TRACP-5b. Moreover, significant variables positively affecting femoral neck BMD were diabetes and BMI, and significant variables negatively affecting femoral neck BMD were age, spinal fracture on radiography, duration of ADT, BAP, urine NTx, and TRACP-5b (Table 3). On the other hand, any other factors of fracture—including previous bone fracture of the spine or hip, fracture during ADT, and family history of bone fracture of the spine or hip, and any other factors of lifestyle and lifestyle diseases including smoking status, alcohol intake, hypertension, CKD, and rheumatoid arthritis—were not associated with lumbar spine and femoral neck BMD.

Table 2 Comparison of BMD and bone metabolism markers according to ADT duration
Full size table
Table 3 Comparison of lumbar spine and femoral neck BMD according to patient characteristics
Full size table

We analyzed whether there was a relationship between BAP and the duration of ADT, between urine NTx and the duration of ADT, and between TRACP-5b and the duration of ADT. Based on the results of the study, BAP and TRACP-5b were weakly related to the duration of ADT (r = 0.180, p = 0.0062 and r = 0.211, p = 0.0086, respectively). We also found that BAP was significantly higher in patients with previous bone fractures of the spine or hip, and both BAP and TRACP-5b were significantly higher in patients with spinal fracture on radiography than in patients without any fractures (Table 4).

Table 4 Association between factors of fracture and bone metabolism markers
Full size table

A multivariate forward stepwise regression analysis was calculated to predict lumbar spine BMD based on diabetes, BMI, spinal fracture on radiography, the duration of ADT, and BAP. A significant regression equation [F (5, 224) = 12.925, p < 0.0001, with an R2 of 0.224] found that the duration of ADT, BMI, diabetes, BAP, and spinal fracture on radiography were all significant predictors of lumbar spine BMD. Furthermore, a multivariate forward stepwise regression analysis was calculated to predict femoral neck BMD based on diabetes, BMI, age, spinal fracture on radiography, the duration of ADT, and BAP. A significant regression equation [F (4, 225) = 23.406, p < 0.0001, with an R2 of 0.294] found that the duration of ADT, BMI, BAP, and spinal fracture on radiography were significant predictors of femoral neck BMD (Table 5). As a precaution, urine NTx and TRACP-5b were excluded from the multivariate analysis because each was only measured as a bone resorption marker in the patients. In summary, stepwise regression analyses revealed that the duration of ADT was a significant variable of both lumbar spine BMD and femoral neck BMD.

Table 5 Stepwise regression analyses to predict lumbar spine BMD and femoral neck BMD
Full size table

According to lumbar spine and femoral neck BMD, the number of patients who were normal, who had osteopenia, and who had osteoporosis were 58, 120, and 52, respectively, based on WHO criteria for the diagnosis of osteoporosis [9]. The prevalence of osteoporosis was 12.7% in hormone-naive patients, 11.1% in patients with <2 years of ADT, 44.7% in patients with 2–4 years of ADT, 23.8% in patients with 4–6 years of ADT, and 45.0% in patients with ≥6 years of ADT. Univariate analyses showed that as the duration of ADT increased, the prevalence of osteoporosis increased statistically (p = 0.0002) (Fig. 1). As with the duration of ADT, other variables—BMI, age, spinal fracture on radiography, and BAP—were also significant for the prevalence of osteoporosis in univariate analyses (p < 0.0001, p < 0.0001, p = 0.0001, and p = 0.0002, respectively). When diabetes, BMI, age, spinal fracture on radiography, duration of ADT, and BAP were included as predictor variables in multivariate logistic regression analysis to predict diagnosis of osteoporosis, the duration of ADT persisted as a significant variable after controlling simultaneously for potential confounders [OR 1.020 (1.008–1.032), p = 0.0012] (Table 6).

Fig. 1
figure 1

Changes in the prevalence of osteoporosis stratified by ADT duration (biennial category); as the duration of ADT increased, the prevalence of osteoporosis increased (p = 0.0002), reaching 45% in patients with ≥6 years of ADT

Full size image
Table 6 Logistic regression analyses to predict diagnosis of osteoporosis
Full size table

Discussion

Decreased BMD in prostate cancer patients receiving ADT is well documented in Western countries [2,3,4,5, 10, 11]. The present study showed that BMD decreased both at the lumbar spine and femoral neck as the duration of ADT increased. This result is compatible with those of previous studies. Loss of BMD means an increase in fractures in NMPC patients receiving ADT. In the present study, spinal fractures on X-rays were seen in 9.1% of subjects. On the other hand, the prevalence of vertebral fractures was 13–33% in the Caucasian population [12, 13]. White race and low BMD were significantly associated with vertebral fracture. Therefore, the differences in the prevalence of vertebral fracture between our study and other studies might be explained by ethnic differences [12].

To date, only a few studies discuss BMD loss associated with ADT for prostate cancer patients in the Asian population. Almost all of the studies were reported from Japan, but were either small in number or contained both non-metastatic and bone metastatic prostate cancer patients [14,15,16,17]. Yuasa et al. [14] showed that a decrease in BMD was associated with ADT in Japanese patients. In their study, lumbar spine, total hip, and femoral neck BMD was measured by DEXA, and the prevalence of osteoporosis was compared between 70 ADT-treated patients without bone metastasis and 88 hormone-naive patients. The results showed that although ADT-treated patients without bone metastasis had significantly lower BMD values than hormone-naive patients, their ADT treatment (on average 30.7 months) did not increase the prevalence of osteoporosis. These findings are different from the results of our study which show the prevalence of osteoporosis and osteopenia were 12.7 and 36.7% in hormone-naive patients and 39.8 and 42.0% in patients receiving >2 years of ADT, respectively, and the prevalence of osteoporosis was positively associated with the duration of ADT in multivariate analysis. On the other hand, other studies from Western countries showed that prostate cancer patients undergoing ADT had a high incidence of osteoporosis. For example, Morote et al. [2] reported that 35.4 and 45.2% of hormone-naive patients had osteoporosis and osteopenia, respectively, while 42.9 and 39.3% of patients treated with ADT for 2 years suffered from osteoporosis and osteopenia, respectively. The finding that Japanese patients had a lower baseline incidence of BMD loss and osteoporosis was confirmed in our study as well as in the study by Yuasa et al. However, our study’s finding, i.e., that as the duration of ADT increased the prevalence of osteoporosis increased, was compatible with the findings for Caucasians. As Yuasa et al. mentioned, there may be some racial differences that determine the BMD between Japanese and Caucasians. However, it seems reasonable that as BMD decreases with ADT, the prevalence of osteoporosis increases. We speculated that the reasons why there was no difference in the prevalence of osteoporosis between hormone-naive patients and ADT-treated patients in the study by Yuasa et al. were because the number of ADT-treated patients was small, the duration of ADT was relatively short, it included 6 patients treated with bicalutamide monotherapy and 3 patients using estramustine phosphate, and the baseline incidence of BMD loss in Japanese patients was low. To confirm the differences among these studies in Japanese patients, larger scale and prospective studies are warranted.

While numerous risk factors have been identified for post-menopausal osteoporosis, including age, personal or family history of fracture, Asian or Hispanic heritage, smoking and cortisone use, only a few studies have examined lifestyle factors in androgen-deprived patients with prostate cancer [18]. In the present study, some lifestyle factors associated with osteoporosis were investigated in NMPC patients with ADT, and BMI was confirmed as the strongest predictor for BMD and osteoporosis among several variables, including the duration of ADT. To date, studies examining the relationship between body composition and bone mass have found conflicting results. Although a few studies found that individuals with higher BMI levels have a higher risk of osteoporosis, the majority of studies using BMI as an indicator of adiposity have primarily found obesity to be protective against osteoporosis [19]. On the other hand, the inverse relationship between underweight and BMD is well understood, although in postmenopausal women it was clearly shown that low BMI was an important risk factor for low bone mass and increased bone loss [20]. Ryan et al. [21] showed that BMI was positively associated with Z-scores at the femoral neck and total hip after adjusting for the duration of ADT and other lifestyle factors in androgen-deprived patients with prostate cancer. However, the duration of ADT, not BMI, was the strongest predictor for BMD. The differences in the power of predicting osteoporosis between the duration of ADT and BMI in the study by Ryan et al. and the present study may be due to the difference in the distribution of BMI. In the study by Ryan et al., most patients were overweight or obese, with a median BMI of 28.8 kg/m2; however, in our study, most patients were in the normal range, with a median BMI of 23.0 kg/m2. Elderly Japanese men, who usually have a lower BMI than elderly Caucasians, may be more susceptible to the influence of BMI on BMD. It should be mentioned that BMI as well as ADT is one of the most important factors affecting bone mass and osteoporosis in Japanese prostate cancer patients with ADT. On the other hand, some lifestyle diseases are related to BMD and osteoporosis. In particular, diabetes and CKD are well known as being related to fracture risk. The present study showed that diabetes was positively associated with BMD in multivariate analyses. This finding is compatible with the results of a meta-analysis that found BMD increased in patients with type 2 diabetes [22]. To our knowledge, this is the first study to reveal a relationship between BMD and diabetes in prostate cancer patients undergoing ADT.

A limitation of the study is that it was cross-sectional and retrospective, and the number of patients was relatively small. Additionally, patients grouped according to the duration of ADT might not be comparable at the time of starting ADT. There should be some differences within each group, and long-term ADT might affect their backgrounds. To eliminate this concern, prospective longitudinal studies will be necessary. Serum testosterone was not investigated in this study, so we did not confirm whether patients receiving ADT were castrated. Since we did not measure serum testosterone levels in patients in this study, we could not confirm whether patients undergoing ADT were in castration. However, we were strictly administering GnRHa on schedule, so we considered that almost all patients undergoing ADT were castrated [23]. Moreover, we did not investigate some other factors that could possibly affect BMD, such as sex hormones (including testosterone and estrogen), factors related to calcium metabolism (including parathyroid hormone, thyroid-stimulating hormone, calcium, and vitamin D), and lifestyle factors (including calcium and vitamin D intake and amount of exercise) [24]. Additionally, this study was carried out on the basis of daily practice, so only urine NTx or TRACP-5b was measured as a bone resorption marker in each patient because of the medical insurance restriction. For this reason, bone resorption markers were not added to the multivariate analyses.

Conclusion

This study showed that ADT negatively affected lumbar spine and femoral neck BMD in Japanese patients with NMPC. As the duration of ADT increased, BMD decreased at both sites. We also observed a progressive increase in the prevalence of osteoporosis in Japanese NMPC patients with ADT. The prevalence of osteoporosis reached 45% after 6 years of ADT. In addition, we investigated patient backgrounds in detail for anything that may possibly affect BMD in the Asian population, and we showed for the first time a positive relationship between BMD and diabetes in prostate cancer patients undergoing ADT. Larger scale and prospective studies are warranted to clarify ethnic differences in the prevalence of osteoporosis accompanying ADT in prostate cancer patients of Asians and Caucasians.