Introduction

Androgen deprivation therapy (ADT) is not only associated with an improved prognosis for patients with metastatic prostate cancer, but it has also been administered to a wide range of patients with localized cancer [1, 2]. However, various adverse events have been documented, including decreased bone density, fragility fractures, reduced sexual function, muscle weakness, and abnormal lipid metabolism associated with androgen removal [3,4,5]. Guidelines emphasize the importance of preserving bone mineral density (BMD) during hormonal therapy, with dual-energy X-ray absorptiometry (DXA) being the recommended bone densitometry technique [6, 7].

The significance of both bone quality and density in fortifying bone strength and resistance to fractures has been underscored in recent years [8, 9]. The trabecular bone score (TBS) has emerged as a convenient measure with which to assess the microstructure of trabecular bone in the first through fourth lumbar vertebrae (L2–4) [10]. This assessment can be concurrently conducted with DXA, the gold standard bone densitometry technique, offering a more expeditious procedure with reduced radiation exposure. The TBS is considered to provide more accurate assessment of the fracture risk [11].

Purpose

No studies to date have focused on changes in the TBS during ADT for prostate cancer. In this single-center, prospective, observational study, we measured the TBS in addition to periodic BMD and examined the changes in these parameters during ADT.

Materials and methods

BMD and TBS measurements were conducted following the Declaration of Helsinki principles on consecutive patients who had prostate cancer without bone metastases at Kobe University Hospital from March 2020 to December 2021. The patients underwent initial histological diagnosis, bone scintigraphy, and thoracoabdominal computed tomography to confirm the absence of bone metastases. Throughout the therapeutic course, the patients underwent prostate-specific antigen measurement every 3 months and were followed up with a computed tomography scan of the chest and abdomen approximately once a year. Patients whose condition progressed to castration-resistant prostate cancer within 2 years after administration of hormone treatments were excluded from the study. All patients continued to receive bicalutamide as a combined androgen blockade for 2 years, along with a Luteinizing Hormone-Releasing Hormone agonist or antagonist.

BMD was measured at the femoral neck, total hip, and lumbar spine (L2–4) using a Hologic Horizon A device (Hologic, Inc., Marlborough, MA, USA) before hormone treatment and at 6, 12, 18, and 24 months thereafter. Measurements were allowed to be taken within 1 month before or after the scheduled date. TBS measurements were simultaneously performed in the lumbar spine (L2–4). DXA was performed using a Hologic Horizon device with APEX software version 13.6.0.7 (Hologic, Inc.), and the TBS was measured using TBS iNsight® software version 3.0.3.0 (Medimaps SASU, Pessac, France).

Osteoporosis was diagnosed and treated in accordance with the 2012 revision of the diagnostic criteria for primary osteoporosis [12]. The diagnostic criteria were involvement of the lumbar spine or proximal femur, presence of another fragility fracture, or BMD equal to or below − 2.5 SD. Each patient was evaluated by an orthopedic specialist (N.H.) for the presence of fragility fractures prior to and during treatment, and their medical history and two-way X-rays of the lumbar region were acquired. Patients diagnosed with osteoporosis were treated with bisphosphonate and anti-receptor activator of nuclear factor-kappa B ligand antibody. Patients who were diagnosed with osteoporosis before and within 2 years after the start of ADT were excluded from the study.

Prophylactic treatment for secondary osteoporosis was not administered in any of the cases.

Statistical analysis

All statistical analyses were performed with EZR version 1.55 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [13], and a p value of < 0.05 was considered statistically significant. Multiple regression analysis, with bone density as the criterion variable, was performed to examine the impact of each explanatory variable on bone density. The variables in this analysis were age (continuous), body mass index (BMI) (continuous), habitual drinking, smoking, diabetes, Gleason grade, prostate-specific antigen concentration (continuous), bone-type alkaline phosphatase concentration (continuous), collagen type I trimeric cross-linked peptide concentration (continuous), tartrate-resistant acid phosphatase 5b concentration (continuous), 25-hydroxyvitamin D concentration (continuous), testosterone concentration (continuous), TBS (continuous), BMD at each site (continuous), and estimated glomerular filtration rate (continuous). Model selection was guided by the Bayesian information criterion. Pearson correlation analysis was performed to calculate the correlation coefficients.

Results

Of 48 patients who had prostate cancer without bone metastases, 9 were excluded because of an osteoporosis diagnosis at the initial assessment. The remaining 34 patients were enrolled in the study after accounting for deaths, transfers, progression to castration-resistant prostate cancer, and initiation of osteoporosis treatment during the 2-year follow-up. Bone densitometry was unattainable for one patient at 18 months. All patients maintained testosterone levels within the castration range throughout the study. No patients developed fragility fractures.

The patients’ background characteristics at the start of ADT are shown in Table 1. Their median age was 75 years, and their median BMI was 23.8 kg/m2. All patients underwent combined androgen blockade during the study period. The median BMD was 0.706, 0.914, and 1.083 g/cm2 in the femoral neck, total hip, and lumbar spine, respectively. The median TBS was 1.31.

Table 1 Patients’ characteristics
Full size table

The median change in BMD was − 0.013 g/cm2 (− 1.8%) for the femoral neck, − 0.015 g/cm2 (− 1.6%) for the total hip, and − 0.029 g/cm2 (− 2.7%) for the lumbar spine at 12 months, progressing to − 0.027 g/cm2 (− 3.8%), − 0.038 g/cm2 (− 4.2%), and − 0.066 g/cm2 (− 6.1%) at 24 months, respectively, indicating a temporal decline (Fig. 1a–c). The median change in the TBS was − 0.052 (− 4.0%) and − 0.218 (− 16.6%) at 12 and 24 months, respectively (Fig. 1d).

Fig. 1
figure 1

Changes in median bone mineral density of a femoral neck, b total hip, and c lumbar spine and d TBS over time. TBS, trabecular bone score

Full size image

In the Pearson correlation analysis, the lumbar spine BMD and TBS at the initiation of hormone therapy displayed a weak but nonsignificant correlation (R = 0.332, 95% confidence interval =  − 0.01 to 0.602, p = 0.055). Conversely, a robust correlation was observed between BMD and TBS in the lumbar spine 2 years after initiation of treatment (R = 0.753, 95% confidence interval =  − 0.528 to 0.879, p < 0.001). No correlation was discerned between BMD and TBS at each site throughout the 2-year period (Fig. 2a–c).

Fig. 2
figure 2

Correlation between change in bone mineral density of a femoral neck, b total hip, and c lumbar spine and change in trabecular bone score at 2 years after androgen deprivation therapy

Full size image

The multiple regression analysis of factors influencing BMD loss at 2 years showed that BMI, pretreatment femoral neck BMD, collagen type I trimeric cross-linked peptide, and TBS were influential factors for the femoral neck (multiple R2 = 0.481, adjusted R2 = 0.394, p = 0.003) (Table 2); that age was an influential factor for the total hip (multiple R2 = 0.13, adjusted R2 = 0.098, p = 0.055) (Table 3); and that BMI, testosterone, and TBS were influential factors for the lumbar spine (multiple R2 = 0.295, adjusted R2 = 0.21, p = 0.031) (Table 4). The multiple regression analysis revealed no factors predicting the change in the TBS at 2 years (data not shown). Histograms of the residuals showed that the deviations were normally distributed. The variance inflation factors were consistently < 5, signifying an absence of multicollinearity.

Table 2 Multiple regression analysis for bone mineral density loss of the femoral neck
Full size table
Table 3 Multiple regression analysis for bone mineral density loss of the total hip
Full size table
Table 4 Multiple regression analysis for bone mineral density loss of the lumber spine
Full size table

Discussion

This study is the first to investigate changes in BMD and the TBS throughout a 2-year period in patients undergoing ADT for prostate cancer. The TBS exhibited a significant decrease during this treatment period. A strength of this study is that the data were prospectively collected from consecutive patients.

In contrast to DXA, which quantifies bone mass, the TBS provides an indirect assessment of bone microstructure, which correlates with bone quality [14]. Numerous studies have suggested that the TBS can identify risks not captured by BMD when predicting fragility fractures [15,16,17]. In addition, it has been stated that the TBS of the lumbar spine is a predictor of fractures not only in the lumbar spine but also in other areas, regardless of gender [18,19,20]. In prostate cancer patients without bone metastases, who have a high potential for long-term prognosis, measuring the TBS is considered important from the perspective of bone health.

Parallel to the temporal trends in bone density observed in men undergoing hormone therapy in the current study, the TBS exhibited a declining trajectory. Prior research demonstrated a 0.18% to 1.08% annual decrease in the TBS in older men aged 66 to 75 years without factors affecting bone metabolism [21]. Another report showed a − 0.25% annual decline in older men with a median age of 72.8 years [22]. The rate of decline in the TBS in the present study was 4.0% at 12 months and 16.6% at 24 months, which is a particularly rapid rate of decline compared with these previous reports. The rate of decline in the TBS in postmenopausal women during a 1-year period was − 0.31% [23] or − 0.16% [24], and the TBS in our study declined at a comparatively rapid rate. A similar phenomenon has been reported for changes in bone density over time. Analogously, the rate of bone density loss in patients with prostate cancer undergoing hormone therapy, generally ranging from 4 to 8% [25, 26], exceeds the typical rate in postmenopausal women (approximately 2% annually) [27].

The pivotal role of both testosterone and estrogen in sustaining bone density is well-established. In addition to its direct effect on maintaining bone density, testosterone is metabolized by aromatase to estrogen, resulting in estrogen’s ability to maintain bone density. Therefore, testosterone exerts a dual effect on bone density [28]. The findings presented herein suggest a potent impact of testosterone on bone quality as evidenced by the TBS. The study suggests that testosterone may be involved in the tendency of TBS, which reflects the trabecular bone structure, to show a delayed decline compared to BMD, which reflects the bone mass of both cortical and trabecular bones. It has been reported that due to differences in androgen receptor expression levels, cortical bone is more influenced by male hormones than trabecular bone [29]. In fact, it has been demonstrated that testosterone replacement therapy in patients with hypopituitarism significantly improves BMD compared to TBS [30]. However, one study showed that testosterone replacement in patients with low testosterone concentrations had no effect on the TBS [31]. Therefore, further investigations are needed to elucidate the underlying mechanisms.

Previous studies have shown a correlation between lumbar spine BMD and the TBS in healthy older men (R = 0.555–0.576) [32, 33], and in the present study, a weak correlation was found between these parameters prior to treatment. Conversely, no correlation was seen between the reductions in BMD and TBS throughout the 2 years following treatment initiation. This is consistent with previous reports [34, 35]. The stronger correlation observed after 2 years can be attributed to the fact that patients with higher TBS are more likely to maintain their BMD, while patients with lower TBS are more likely to experience a decline in BMD, as demonstrated by the multiple regression analysis.

For further investigation, a multiple regression analysis was performed to examine the factors affecting the 2-year changes in BMD and TBS. Although the correlation coefficient was small and did not appear to be an effective predictor, patients with a high TBS at the initial examination tended to have a slower rate of BMD loss at the femoral neck and the lumbar spine (Tables 2, 3, 4), but this was not observed at the total hip. Generally, the rate of bone remodeling varies depending on the bone site and race [36, 37], but there have been no consistent reports explaining why the total hip is less affected by TBS compared to the femoral neck and lumbar spine. Further investigation is needed to understand this phenomenon. By contrast, factors affecting changes in the TBS could not be identified. This finding explains the stronger correlation between the lumbar spine BMD and TBS at 2 years than at baseline. Although Miyazawa et al. indicated that patients who were undergoing hormone therapy for prostate cancer and had impaired renal function experienced more rapid BMD loss [38], renal function did not seem to have a significant effect in the present study. In our study, patients with higher BMD at the femoral neck before ADT treatment experienced a faster rate of BMD loss. Although there are not many reports examining the impact of pretreatment BMD, Miyazawa et al. reported that pre-ADT treatment BMD does not affect the rate of BMD loss, which contrasts with our findings. We believe these differences may be due to variations in statistical methods or measurement sites.

This study had several limitations. First, the number of patients was small and the analysis lacked statistical power. Second, the use of the Bayesian information criterion in the multiple regression analysis may not have accurately encompassed all factors. Third, an adverse event associated with ADT treatment is an increase in body fat [39], which is considered an important factor in the reduction of TBS [40]. In this study, data on changes in body weight and body fat percentage over time were not collected, so it is necessary to fully acknowledge that these factors may have influenced the results. Finally, there was a lack of attention to fracture events in this study. A decrease in the TBS has not previously been shown to lead to fractures in patients with hormone-treated prostate cancer, but associations between the TBS and fracture events have been well documented in patients with other conditions.

Conclusion

The study revealed decreases in BMD and the TBS over a 2-year period during ADT in patients who had prostate cancer without bone metastases. The factors precipitating the decline in the TBS remain unclear; however, patients with a lower pretreatment TBS tended to experience a faster rate of bone density decline.