Long-term changes in bone mineral density and predicted fracture risk in patients receiving androgen-deprivation therapy for prostate cancer, with stratification of treatment based on presenting values


Mr N.J. Parr, Department of Urology, Ward 14, Wirral University Teaching Hospitals, Arrowe Park, Upton, Wirral, Merseyside, CH49 5PE, UK.
e-mail: nigelparr@dsl.pipex.com



To study the long-term effects of androgen-deprivation therapy (ADT) using luteinizing hormone-releasing hormone (LHRH) agonists or antiandrogen therapy with bicalutamide on bone mineral density (BMD) of selected groups of patients with newly diagnosed advanced prostate cancer, stratified by BMD at presentation and to predict alterations in fracture risk.


In all, 618 men with a mean (sd, range) age of 73 (7.1, 49–94) years, initiating ADT for prostate cancer were prospectively recruited and followed from October 1999 to January 2007. BMD was measured by forearm dual-energy X-ray absorptiometry (DEXA) before ADT and repeated annually. Patients with osteoporosis (T-score ≤−2.5) were commenced on bicalutamide; patients with osteopenia (T-score between −1.0 and −2.5) and normal BMD (T-score > −1.0) were commenced on an LHRH agonist. Patients with osteopenia and osteoporosis received calcium and vitamin D supplements.


Over 7 years, 1690 DEXA scans were performed. In all, 41% of patients with newly diagnosed prostate cancer were osteoporotic, 39% were osteopenic and 20% had normal BMD. In the normal group, treated with an LHRH agonist, there were significant decreases in BMD (1 year 1.2%; 2 year 3.7%; 3 year 6.5%; 4 year 8.9%; 5 year 9.9%; 6 year 12.7%), which also occurred in the patients with osteopenia with 60% developing osteoporosis after 2 years (1 year 1.8%; 2 year 5.1%; 3 year 8.0%; 4 year 8.2%; 5 year 11.5%; 6 year 14.1%). By contrast, the osteoporotic group maintained BMD (1 year 0.5%; 2 year 0%; 3 year +1.2%; 4 year 0.5%; 5 year 1.7%; 6 year 2.2%).


Patients treated with an LHRH agonist have significant and sustained decreases in BMD, whereas bicalutamide maintains BMD. We advocate routine assessment of BMD before ADT, with surveillance thereafter.


bone mineral density


androgen-deprivation therapy


dual-energy X-ray absorptiometry


quality of life.


Prostate cancer is an androgen-dependent tumour, with ablation being the mainstay of treatment for advanced disease. Furthermore, androgen-deprivation therapy (ADT) is increasingly used in nonmetastatic prostate cancer, for both locally advanced disease and biochemical failure. However, LHRH agonists, rapidly decrease serum testosterone, disturbing bone metabolism and causing osteoporosis [1].

Osteoporosis is asymptomatic, only clinically manifesting itself when an osteoporotic fracture occurs. The incidence of osteoporotic fractures in patients receiving LHRH agonists is far higher than the incidence of pathological fractures in patients with metastatic prostate cancer [2]. For each sd decrease in bone mineral density (BMD; a decrease in T-score of 1), fracture risk increases up to three-fold [3]. Measurement of BMD predicts fracture risk as accurately as blood pressure predicts stroke and better than serum cholesterol for cardiovascular disease [4]. Distal forearm BMD is the strongest predictor of overall fracture risk in men [5].

Antiandrogens, such as bicalutamide maintain serum testosterone, offering comparable survival outcomes and potential quality of life (QoL) benefits [6]. Bicalutamide has been shown to significantly increase BMD, compared with LHRH agonists and may be suitable for individuals with lower BMDs at presentation of prostate cancer [7,8]. However, these studies contained few patients with relatively short follow-ups.

Our aim, in 1999, was to study the long-term effects of LHRH agonist and antiandrogen therapy on BMD of selected groups of patients with newly diagnosed advanced prostate cancer, stratified by BMD at presentation and to predict alterations in fracture risk.


In all, 618 men with newly diagnosed prostate cancer initiating ADT were prospectively recruited from a single urology institution and followed from October 1999 to January 2007.

At baseline, tumour stage, metastatic status, Gleason score and PSA level were determined. Metastatic status was defined as: M0, patients with no evidence of metastatic disease, confirmed by a negative bone scan; Mx, patients with no evidence of metastatic disease, but no confirmation by a bone scan; M1, patients with definite scintigraphic, radiological or other evidence of metastatic disease.

The BMD of the nondominant forearm (ultra-distal region) was measured within 1 week of prostate cancer diagnosis, before ADT, by dual energy X-ray absorptiometry (DEXA), using the Osteometer DTX-200 scanner (Osteometer Meditech, Hawthorne, CA, USA). Osteoporosis was defined using WHO criteria as a BMD measurement of ≤−2.5 sd below the mean for a young adult population, known as T-score of ≤−2.5. Osteopenia was defined by a T-score between −1.0 and −2.5, and a normal BMD defined as a T-score of ≥−1.0. DEXA was repeated annually, providing longitudinal data.

A clinical protocol was designed to select the choice of ADT for each patient. Patients with osteoporosis at presentation were commenced on bicalutamide 150 mg daily (Casodex; AstraZeneca Pharmaceuticals, London, UK). The remainder (osteopenics and those with a normal BMD) were commenced on an LHRH agonist. Additionally, patients with osteoporosis and osteopenia were commenced on daily calcium and vitamin D preparation (Calcichew D3 forte, 12.6 mmol calcium and 400 units cholecalciferol). Lifestyle advice including regular weight-bearing exercise, avoidance of excessive alcohol, caffeine and smoking cessation was given. An interim protocol amendment was made to include baseline posteroanterior and lateral radiographs of the thoracolumbar spine. During the study a few patients were commenced on oral bisphosphonates by their primary care physicians. Also, a few of the patients with osteopenia who had developed osteoporosis, with a previous history of osteoporotic fractures, were offered oral bisphosphonates. These patients were excluded from the statistical analyses.

Patients were regularly reviewed in a designated clinic, with PSA level measurements every 3 months. Those who failed to respond or escaped treatment with hormone monotherapy were managed according to the clinical situation with second-line therapy. The protocol was approved by the Local Ethics Research Committee. This study is registered with ClinicalTrials.gov number NCT00536653.

Two-tailed t-tests were used for comparing continuous data and Fisher’s exact test for categorical data between groups; P < 0.05 was considered to indicate statistical significance.


The characteristics of the 618 men with a mean (sd, range) age of 73 (7.1; 49–94) years, are shown in Table 1. In all, 41% of patients with newly diagnosed prostate cancer were osteoporotic, 39% were osteopenic and 20% had normal BMDs. There were no significant differences between the three BMD groups at presentation for tumour stage, grade, PSA level, and other biochemical measurements. Age correlated significantly with BMD (P < 0.001).

Table 1.  Baseline characteristics of the 618 patients
  • *

    Bone metastasis status, determined by isotope bone scan; BMI, body mass index; ALP, alkaline phosphatase.

Mean (sd)   
 Age, years 70.8 (7.5) 74.0 (7.0) 77.2 (6.3)
 Height, cm175.2 (6.8)173.4 (10.1)171.1 (6.7)
 Weight, kg 88.25 (14.1) 81.9 (13.0) 74.7 (12.1)
 BMI, kg/m2 28.7 (4.1) 27.4 (4.6) 25.4 (3.6)
Median (range) serum levels   
 PSA, ng/mL 30 (4.4–3697) 37 (0.4–5599) 36.3 (1.7–2418)
 ALP, U/L 85 (41–1739) 83 (31–2775) 84 (21–3360)
 Adjusted Ca, mmol/L  2.33 (2–3.4)  2.34 (1.81–2.59)  2.35 (1.91–2.35)
 Urea, mg/L  6.4 (0.7–17.7)  6.1 (2.4–21)  6.1 (2.7–14.5)
 Creatinine, mg/L103 (69–103)104 (56–275)108 (73–108)
N (%)   
 Clinical stage   
  Unclassified  2 (1.6)  3 (1.2)  2 (0.8)
  T1 20 (16.1) 32 (13.3) 29 (11.5)
  T2 30 (24.2) 77 (32) 91 (36)
  T3 65 (52.4)109 (45.2) 113 (44.6)
  T4  7 (5.6) 20 (8.3) 18 (7.1)
  M0 46 (37.1) 69 (28.6) 97 (38.3)
  M1 29 (23.4) 57 (23.7) 50 (19.8)
  Mx 49 (39.5) 115 (47.7)106 (41.9)
 Unclassified  2 (1.6)  4 (1.7)  2 (0.8)
  2–4  7 (5.6) 18 (7.5) 17 (6.7)
  5–7 58 (46.8)121 (50.2)135 (53.4)
  8–10 57 (46) 98 (40.7) 99 (39.1)

In all, 299 men underwent baseline thoracolumbar radiographs. In 284 patients (95%) there were degenerative changes, including osteophytes and disc space narrowing and 61 vertebral fractures were identified (20% prevalence). Presenting fracture rates were related to BMD, being 7% (four of 43) in those with normal BMD, 18% (19/105) with osteopenia and 26% (39/151) in those with osteoporosis. When stratified for age, the number of patients with vertebral fractures increased with advanced age: age group <60 years (one of 14), 60–69 years 11% (five of 55), 70–79 years 19% (29 of 151), and ≥80 years 24% (19 of 79).

Table 2 shows the mean BMD at each interval from baseline, percentage change and statistical significance of the change from baseline, using the paired t-test. In both the normal and osteopenic groups there were significant reductions in BMD from baseline whereas there was no significant change in the osteoporotic group.

Table 2.  The mean BMDs, percentage change from baseline and statistical analysis of changes
BMD groupBaselineYear 1Year 2Year 3Year 4Year 5Year 6
  • *

    Paired t-tests.

Normal (LHRH)
Mean BMD (% change), g/cm20.5970.590 (−1.2)0.575 (−3.7)0.558 (−6.5)0.544 (−8.9)0.538 (−9.9)0.521 (−12.7)
 P for % change* 0.018<0.001<0.0010.0020.1430.054
Osteopenic (LHRH)
Mean BMD (% change), g/cm20.5120.503 (−1.8)0.486 (−5.1)0.471 (−8.0)0.470 (−8.2)0.453 (−11.5)0.440 (−14.1)
 P for % change* <0.001<0.001<0.0010.2890.7500.812
Osteoporotic (bicalutamide)
Mean BMD (% change), g/cm20.4060.404 (−0.5)0.406 (0.0)0.401 (+1.2)0.404 (−0.5)0.399 (−1.7)0.397 (−2.2)
 P for % change* 0.3620.8090.2000.6090.1300.807

Table 3 shows the mean T-score per year in each BMD group. The number of patients scanned at yearly intervals depended on the time-point at which they were recruited to the study and upon survival. The mean T-score curves for each of the three groups with time are shown in Fig. 1. Over 7 years, 1690 BMD measurements were made.

Table 3.  The mean T-scores and patient numbers at annual scan date
VariableBMD group
Normal(LHRH)Osteopenic(LHRH)Osteoporotic (bicalutamide)
 T-score −0.225 −1.719 −3.523
Year 1   
 N 84175179
 T-score −0.386 −1.871 −3.543
Year 2   
 N 58114121
 T-score −0.634 −2.146 −3.517
Year 3   
 N 38 69 69
 T-score −0.892 −2.386 −3.526
Year 4   
 N 24 40 42
 T-score −1.163 −2.400 −3.548
Year 5   
 N 10 17 18
 T-score −1.240 −2.694 −3.778
Year 6   
 N  2  8  4
 T-score −2.400 −2.900 −3.650
Figure 1.

Longitudinal changes in mean T-score at yearly intervals in each BMD group.

The proportion of patients who had a reduction in T-score by 1.0 is shown in Fig. 2, together with the proportion of patients with normal BMDs becoming osteopenic or osteoporotic with time (A) and the proportion of patients with osteopenia developing osteoporosis (B). Of all patients who were osteopenic at presentation, 35% developed osteoporosis after 1 year and 60% after 2 years of treatment with an LHRH agonist.

Figure 2.

A, Percentage of the normal group becoming osteopenic/osteoporotic (solid line) and the proportion of patients who had decreased their T-score by 1.0 from baseline (dashed line). B, Percentage of the osteopenic group becoming osteoporotic (red solid line) and the proportion of patients who had decreased their T-score by 1.0 from baseline (green dashed line). A decrease in the T-score of 1.0 triples the fracture risk.

In all, 241 deaths (39% overall mortality) occurred during the study, with 114 (47%) from prostate cancer. Most commonly reported adverse events for the bicalutamide group were gynaecomastia (49%) and breast pain (43%), while for those on LHRH agonists was bothersome hot flushes (57%).


The present study shows that a significant proportion of patients requiring ADT already have osteoporosis/osteopenia and vertebral crush fractures. Castration with an LHRH agonist causes significant and sustained reductions in BMD. By contrast, patients on bicalutamide maintain BMD. This suggests bicalutamide may avoid the increased fracture risk associated with LHRH agonists.

Testosterone acts as a growth factor for prostate cancer cells. Therapies that decrease serum testosterone to castrate levels are used successfully to increase longevity and decrease morbidity from advanced prostate cancer. Such therapies include orchidectomy, oestrogen and LHRH agonists. They decrease serum testosterone by >95%, but serum oestradiol, which is responsible for maintenance of BMD in both men and women, is also decreased by >80%[9]. Oestrogens cause thromboembolic and cardiac toxicity. Studies suggest that the use of ADT earlier in the course of prostate cancer may be beneficial [10]. Although such an approach may prolong survival, it extends the duration of hypogonadal state, increasing the risk of osteoporosis.

BMD loss is an important concern in ADT. Rates of loss vary by study population and type of ADT. During the first year of ADT, men lose 2–8% of spinal BMD and 1.8–6.5% at the hip, which compares to a normal male age-related reduction of 0.5–1.0% per year [11]. In contrast to previous short-term studies [12–14], the present data shows the rate of fall in BMD is ongoing and continuous, rather than occurring predominantly during the early months of ADT. Lee et al.[15] studied 65 men, some initiating LHRH agonist therapy and others on long-term LHRH agonist for a mean of 35 months. The total hip BMD decreased steadily during initial and long-term treatment by a mean of 2.0% at 1 year, which is consistent with the present findings.

Evidence suggests that advanced prostate cancer itself may result in increased bone turnover, probably due to mild parathyroid overactivity [16]. We previously reported that ≈40% of men presenting with advanced prostate cancer are osteoporotic and this is significantly greater than in age-matched controls whose mean BMD is 6.6% higher [17]. These findings are confirmed in the present larger study and are similar to another large recently reported European study [18]. BMD at presentation is not routinely measured, although we think it will become increasingly important in selecting appropriate management. Forearm BMD is the strongest predictor of overall fracture risk in men [5]. Furthermore, in men with prostate cancer undergoing ADT with an LHRH agonist, bone loss was maximal at the forearm [13]. We assessed the ultra-distal region of the forearm, composed predominantly of trabecular bone (65%), which is similar to the spine (66%) and total hip (50%). Trabecular assessment is more sensitive to causes of bone loss than cortical assessment. In older men, osteoarthritis in the spine with osteophyte formation and aortic calcifications may falsely elevate lumbar spine BMD, masking bone loss from increased age and ADT [19]. Also, in patients with prostate cancer, lumbar sclerotic metastases can corrupt BMD measurements.

We found the use of a dedicated, portable forearm DEXA in the clinic, with its compact size and potential for rapid patient throughput, greatly facilitated baseline measurement and monitoring. This can be used in routine clinical practice, with greater access to patients compared with the use of large, static and expensive axial machines, used for hip and lumbar spine measurements.

Osteoporotic fractures are an important cause of death and disability. Fractures significantly decrease QoL with an independent negative correlation for prostate cancer survival [20]. Hip and upper extremity fractures generally impact more acutely whereas vertebral fractures may cause ongoing pain and spinal deformity limiting daily activity.

Data on the effect of ADT on the risk of osteoporotic fractures is largely from retrospective studies. Daniell et al.[1] reported a fracture incidence of 28% after 7 years ADT. A study by Shahinian et al.[21] of 50 613 men on the Surveillance, Epidemiology and End Results-Medicare database showed that in men undergoing ADT and surviving for ≥5 years from diagnosis, there was a fracture rate of 19.4% compared with 12.6% in those not receiving ADT. The fracture rate increased with the number of doses of LHRH agonist received. The rate of spinal fractures was 3.2% in those on ADT, compared with 1.6% with no ADT. We think retrospective database studies underestimate the true incidence of vertebral fractures. In our experience, when investigating patients with a history of back pain, emphasis is on excluding metastatic disease by isotope-bone scanning. When negative and the PSA level is stable, patients are often reassured and not investigated further. In this situation, plain films sometimes show vertebral collapse at multiple levels and these patients should be treated to prevent further osteoporotic vertebral collapse and fragility fractures of the hip and distal forearm. In the present study, of those who underwent thoracolumbar radiographs, there was an increasing prevalence of fractures in patients with lower BMDs and older age at presentation.

In a large prospective study of 2778 men aged >55 years, without cancer, the relative risk of hip fracture was 3.0 for each sd decrease in BMD [3]. As can be seen from Fig. 2, with increasing duration of ADT, a large proportion of the present patients decreased their T-score by 1.0. After 3 years, at least half of the patients who had a normal or osteopenic BMD at presentation had tripled their fracture risk.

The best therapeutic approach to managing osteoporosis in men undergoing ADT is yet to be decided. Lifestyle changes to osteoporotic risk factors may have a role, although large studies have not shown this. Ricchiuti et al.[22] identified no association of risk factors with fracture rate, perhaps partly due to the few fractures in their study. Another study analysing association among several risk factors and male osteoporosis found that of all factors evaluated, only age, body mass index, current mobility level, and past history of fractures were independently associated with osteoporosis. Other dietary (milk and alcohol intake) and lifestyle factors (smoking, low level of physical activity) showed no association [23].

Calcium and vitamin D are recommended to help prevent bone loss in men receiving ADT. However, in the present study, calcium and vitamin D supplementation alone was inadequate to protect against further bone loss in patients with osteopenia on LHRH agonists. This is consistent with studies evaluating the use of bisphosphonates during ADT, where the placebo groups, on calcium and vitamin D, showed significant losses in BMD [9,24].

The selective use of antiandrogens in those at high risk of osteoporotic fractures was adopted in the present study. Antiandrogens inhibit the action of dihydrotestosterone and testosterone at target sites by competitive binding to the cytosolic androgen receptor in prostate cancer cells. Bicalutamide, a potent nonsteroidal antiandrogen, increases serum oestradiol levels by raising the amount of androgens available for aromatization in the extra glandular tissues and also by direct secretion of oestradiol from the testes under the influence of high gonadotrophin levels [25]. Two randomised studies (AZ 306 and 307) have compared bicalutamide 150 mg once daily with castration in 1453 patients with metastatic and locally advanced prostate cancer [6]. In the 805 patients with metastatic disease, there was a survival benefit of 6 weeks in favour of castration. However, the bicalutamide group in those studies did not routinely receive an LHRH agonist on evidence of progression. Kaisary et al.[26] reported that a significantly better QoL in the bicalutamide group supports the choice of bicalutamide 150 mg monotherapy as an option for patients with metastatic prostate cancer for whom medical castration is not acceptable. The short overall survival benefit with castration has to be weighed against the significant morbidity and mortality associated with osteoporotic fractures for those who are already osteoporotic at presentation. In addition, a further analysis of this data showed survival did not significantly differ between bicalutamide and castration for those with a baseline PSA level of ≤400 ng/mL. In the present study an LHRH agonist was added for patients who failed to respond or escaped with bicalutamide monotherapy and presenting with a PSA level of <400 ng/mL. We think that the use of bicalutamide in patients with metastatic prostate cancer with osteoporosis is a reasonable option, which should be discussed with patients. Current European guidelines support the use of bicalutamide monotherapy in certain subgroups of well-informed patients with mestatases [27].

In the present study, BMD at 2 years had significantly decreased by 5.1% in the LHRH agonist group, whereas those on bicalutamide showed no significant decrease (−0.5%). The present findings are in accordance with other studies. Sieber et al.[7] randomised 103 men with nonmetastatic prostate cancer to bicalutamide 150 mg daily or an LHRH agonist. BMD was measured at 6 month intervals. There was a progressive and sustained decrease in BMD over the study period. At 2 years, BMD of the lumbar spine and hip decreased by 5.4% and 4.4% in the LHRH agonist group, while those on bicalutamide increased by 2.4% and 1.1%, respectively. Smith et al.[8] evaluated BMD in 52 men randomized to receive an LHRH agonist or bicalutamide 150 mg. At 1 year BMD of the lumbar spine and total hip decreased by 2.5% and 1.4% in the LHRH agonist group, while those on bicalutamide increased by 2.5% and 1.1%, respectively.

Bisphosphonates inhibit osteoclast mediated resorption and prevent treatment-related bone loss in patients with prostate cancer. Smith et al.[24] randomly assigned 106 men with nonmetastatic prostate cancer and beginning LHRH agonist, to receive 3 monthly bisphosphonate zoledronic acid or placebo for 1 year. Zoledronic acid increased the mean BMD of the lumbar spine and total hip by 5.6% and 1.1%, respectively, while in the placebo group BMD decreased by 2.2% and 2.8%. However, compared with combined treatment with an LHRH agonist and bisphosphonate, bicalutamide monotherapy may provide an attractive alternative to preserve BMD in those presenting with osteoporosis, at high risk of fractures.

In conclusion, osteoporosis and vertebral fractures are common in patients requiring ADT for prostate cancer. In the present study, patients on an LHRH agonist showed significant and sustained decreases in BMD, whereas those on the antiandrogen bicalutamide maintained BMD over the long-term. The present results, containing longer follow-up data and more patients than previous studies, argue strongly for routine baseline BMD measurement before commencing ADT and for surveillance thereafter. The option of bicalutamide monotherapy should be discussed with patients who are osteoporotic at presentation.


Statistical advice was given by Dr Helen Wong, Statistician, Clatterbridge Centre for Oncology, UK. We also thank Mr R.N. Stephenson, Mr P.W. Kutarski, Mr A.M. Cliff and Dr J. Littler for referring their patients into this study. Finally, we are indebted to Dr E. Goerge, Consultant Rheumatologist, for his helpful advice with regards to the initiation and design of this study including DEXA scanning.


None declared.