Introduction

Prostate cancer is the second leading cause of cancer-related deaths in men over the age of 40 [1]. In this malignancy, metastatic spread to the skeleton is the most frequent cause of morbidity and up to 50% of patients may demonstrate skeletal involvement at presentation [2]. Although most patients respond favourably to androgen deprivation by near-complete regression of osseous metastases, the majority will develop hormone refractoriness, universally associated with the re-appearance of skeletal metastases. Conventional X-rays are on the whole inadequate for the diagnosis of these metastases so that confirmation of metastatic involvement of the skeleton has traditionally been achieved by bone scintigraphy, which is also used to assess response of the tumour to therapy. In prostate cancer, metastases to the vertebral column predominantly involve the lumbar spine [3], but those involving the thoracic spine are associated with the highest morbidity because of their significantly increased risk for spinal cord compression [4, 5, 6, 7]. The question of whether bone scintigraphy could be of value in predicting spinal cord compression after the advent of hormone refractoriness has never been addressed.

The aim of this study was to establish whether, compared with the original tumour staging and the prevailing serum prostate-specific antigen (PSA) concentrations, bone scintigraphy performed at the time of hormone refractoriness is of any additional prognostic value regarding survival and skeletal event-free survival (particularly survival without spinal cord compression).

Materials and methods

We studied 84 patients with histologically documented prostate cancer. All patients had been androgen deprived by orchidectomy (n=42), the use of the LH-RH agonist goserelin (n=33) or the use of anti-androgens (n=9). Nineteen patients were also treated with estramustine phosphate. The diagnosis of hormone refractoriness was established on the basis of clinical progression as suggested by progressively worsening bone pain, biochemical evidence for tumour progression in the form of rising serum PSA concentrations and serum alkaline phosphatase activity, and the appearance or re-appearance of skeletal metastases on bone scintigraphy. Because of side-effects, hormonal treatment was discontinued in 23 patients and estramustine phosphate in eight. The majority of patients received palliative treatment in the form of radiotherapy (n=4), the radionuclide 89Sr (n=33) or the nitrogen-containing bisphosphonate olpadronate administered intravenously as a single dose of 20 mg (n=41). Six patients responded to conventional analgesia and required no further palliative treatment. Survival was documented in all patients.

Evaluation of tumour load

Serum PSA was measured using an MEIA (IMX) assay. The upper limit of the normal laboratory reference range is 4 μg/l. Serum alkaline phosphatase activity was measured using routine multi-channel analysis. The normal laboratory reference range is 40–120 U/l.

Tumour grading

Transrectal needle biopsy of the prostate was available in all patients. Histological sections were re-examined and tumour restaged by two independent observers (V. S-M. and H.B.) using the Gleason grading system, which is based on the degree of architectural glandular differentiation [8, 9, 10].

Spinal cord compression

The diagnosis of spinal cord compression was established clinically on the basis of progressive impairment of motor and sensory function and the finding of a segmental level of neurological impairment. The diagnosis was confirmed radiologically by MRI or CT scan of the spine and the level of the lesion was carefully recorded. Spinal cord compression is the most frequently encountered skeletal-related event and the one associated with the highest morbidity. In our study, we defined skeletal event-free survival as survival without spinal cord compression.

Bone scintigraphy

All patients underwent bone scintigraphy at the time of diagnosis of hormone refractoriness using 99mTc-labelled methylene diphosphonate (MDP). Employing this technique, high-resolution whole-body images of the skeleton with simultaneous anterior and posterior images were obtained by multi-head gamma cameras. Bone scans were scored using the five-grade validated Soloway scoring method, which evaluates severity of overall skeletal involvement with metastases, as follows: grade 0, no metastases; grade 1, less than 6 metastases; grade 2, 6–20 metastases; grade 3, more than 20 metastases; grade 4, superscan (diffuse increased uptake in the axial skeleton without a focal lesion, or more than 75% of the skeleton affected by the metastatic process) [11]. In addition, we analysed bone scintigrams by evaluating areas of metastatic involvement for extent of involvement of individual vertebrae—total (Fig. 1) or partial (Fig. 2)—and related our findings to subsequent development of spinal cord compression. Bone scintigrams were independently assessed by two of the authors (V.S-M. and J.A.), one of whom is a physician specialising in nuclear medicine (J.A.).

Fig. 1
figure 1

Localisations of spinal cord compression as shown on MRI of the spine, corresponding to anatomical areas of increased uptake of the radionuclide by whole vertebrae on bone scintigraphy

Full size image
Fig. 2
figure 2

Patchy increase in radionuclide uptake in a number of vertebrae which represented no threat for spinal cord compression because of the incomplete involvement of each vertebra

Full size image

Statistics

PSA and alkaline phosphatase activity were not normally distributed and values were log-transformed before statistical analysis. Survival and spinal cord compression-free survival were calculated from date of onset of hormone refractoriness (at which time bone scintigraphy was also performed) until date of death (survival) or date of spinal cord compression (spinal cord compression-free survival). Kaplan-Meier curves were used to express survival, and patient groups were compared using the log-rank test. The prognostic value of scintigraphic parameters for survival and spinal cord compression-free survival was evaluated using the Cox proportional-hazards regression model and the value was quantified using the relative risk and associated 95% confidence interval (CI). The Cox model was also used to test whether scintigraphic parameters contained additional prognostic information over and above that offered by established risk factors such as age, serum PSA concentration and serum alkaline phosphatase activity. A P value of 0.05 or less was considered statistically significant. Results are expressed as mean±SD values unless otherwise stated.

Results

Survival and spinal cord compression-free survival

Mean survival after development of hormone refractoriness was 8.6±10.6 months. Spinal cord compression due to metastatic lesions to the vertebrae developed in 20 of the 84 patients studied (24%), 3 days to 10 months after bone scintigraphy was performed and hormone refractoriness established. Spinal cord compression occurred at the level of a thoracic vertebra in 14 patients, in four of whom it was documented at an additional level(in two at the lumbar spine and in two at the cervical spine). Six patients demonstrated compressive lesions at the lumbar spine level. Of the 20 patients who developed spinal cord compression, 18 demonstrated increased uptake in the whole of an involved vertebra, corresponding in all cases to subsequent levels of cord compression. The remaining two patients who developed spinal cord compression demonstrated incomplete involvement of a vertebra on bone scintigraphy. Ten patients treated palliatively with the nitrogen-containing bisphosphonate olpadronate and two patients treated with 89Sr who had evidence of total vertebral involvement went on to develop spinal cord compression respectively 2 days to 8 months and 9 days to 18 weeks after administration of the therapy in question. The remaining eight patients had not received any form of palliative therapy by the time they developed spinal cord compression. Only four patients who demonstrated total involvement of a vertebra did not go on to develop spinal cord compression; all of these patients had received palliative therapy in the form of olpadronate (n=1), 89Sr (n=1) or local radiotherapy (n=2). All patients with a diagnosis of spinal cord compression were treated with local radiation therapy, which was successful in reversing neurological abnormalities in all patients except for two who required additional surgical decompression by means of a laminectomy.

Predictive values of markers of tumour load

Serum PSA was elevated in all patients, with a mean value of 511±1,035 μg/l. Serum alkaline phosphatase activity was also elevated in all patients, with a mean value of 402±503 U/l. Whereas log-transformed PSA concentrations were significantly predictive for survival (RR=1.17, 95% CI 1.04–1.32, P=0.01) and spinal cord compression-free survival (RR=1.21, 95% CI 1.07–1.36, P=0.003), log-transformed serum alkaline phosphatase activity and age were not significantly associated with survival or spinal cord compression-free survival.

Predictive value of tumour grading

Mean Gleason score was 7.5, with the majority of patients having a Gleason score greater than 7. Patients with a Gleason score equal to or greater than 7 had a significantly shorter survival and spinal cord compression-free survival than their counterparts with a score of less than 7. Median survival for Gleason ≥7 vs Gleason <7 was 6.8 vs 12.7 months (P<0.03) and median spinal cord compression-free survival for Gleason ≥7 vs Gleason <7 was 6.1 vs 12.3 months (P<0.05). Relative risk of the Gleason score was RR=1.89 (95% CI 1.02–3.53) for mortality and RR=1.76 (95% CI 0.95–3.28) for spinal cord compression. Relative risks of the Gleason score remained significant after adjusting for confounders: RR=2.33 (P=0.013) for mortality and RR=2.37 (P=0.003) for spinal cord compression.

Predictive value of bone scintigraphy

The predictive value of the conventional Soloway method of grading skeletal metastases for survival and spinal cord compression-free survival is shown in Table 1. The Soloway method of grading metastases was overall significantly predictive for survival (P=0.008) and spinal cord compression-free survival (P=0.031), but this was no longer the case after adjusting for confounders such as PSA concentrations, serum alkaline phosphatase activity and age (after such adjustment, P=0.09 and 0.35 respectively). The new method of analysing bone scintigrams, which concentrates on the extent of vertebral spine involvement, was able to significantly predict spinal cord compression (P<0.0001) and its level, as illustrated in Table 2 and Fig. 1.

Table 1 Predictive value of the Soloway grading method for survival and spinal cord compression (SCC)-free survival in patients with hormone-refractory prostate cancer
Full size table
Table 2 Incidence of spinal cord compression (SCC) in patients with total or partial involvement of a vertebra on bone scintigraphy
Full size table

Discussion

Skeletal metastases occur in 30–70% of all cancer patients [12, 13] and autopsy studies demonstrate the presence of skeletal metastases in up to 84% of patients with prostate cancer [3, 14, 15, 16, 17]. More than 80% of skeletal metastases involve the axial skeleton rich in red bone marrow, 40% affect the vertebral bodies, and lesions may involve the intradural and intramedullary spaces [4, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]. Skeletal metastases are associated with an increased incidence of fractures, particularly when osteolytic, and with an increased risk of spinal cord compression, particularly when osteoblastic, as in the case of prostate cancer. Spinal cord compression is a relatively common complication of the metastatic process, reported to occur in up to 50% of patients with hormone-refractory prostate cancer. This malignancy represents the second most common cause of metastatic spinal cord compression after lung cancer [29, 30, 31, 32]. We have previously shown that the risk of developing spinal cord compression can be significantly reduced by using local radiation therapy, radionuclides or bisphosphonates [5]. Identifying patients at clear increased risk for this complication would therefore be of clinical relevance in that it would enable early institution of treatment.

A diagnosis of prostate cancer is suspected on the basis of an elevated PSA concentration. The malignancy is confirmed by histological evaluation of needle biopsy material and the tumour is subsequently staged to allow therapeutic decision-making. PSA is considered to be the most accurate and specific marker for prostate cancer as it is produced by the prostate and is specific for prostatic tissue [28]. High serum concentrations of this tumour marker are consequently highly likely to predict skeletal tumour burden and hence risk of spinal cord compression, as also confirmed in the present study. The widespread availability of PSA measurements has led to the view that bone scintigraphy is no longer necessary or relevant in the management of the patient with prostate cancer. However, not all prostate cancer patients who display large increases in PSA concentrations develop spinal cord compression in the late stages of their malignancy, and this marker is certainly not able to predict the spinal level threatened by this complication.

Bone scintigraphy using 99mTc-MDP is one of the most sensitive and widely used methods for the diagnosis of bone metastases, detecting one-third to one-half of metastatic lesions not seen on conventional radiology [33, 34, 35, 36, 37]. Plain radiographs require a lesion to be more than 1 cm in size, with 50% loss of trabecular bone, before it is readily visible [38]. By contrast, the false negative rate for the diagnosis of skeletal metastases on bone scintigraphy has been demonstrated to be only 0.08% in one of the largest studies of breast cancer patients [39].

Radionuclide uptake depends partly on local blood flow, but largely on local osteoblastic activity, so that predominantly osteoblastic metastases can be clearly distinguished from surrounding normal skeletal uptake. In prostate cancer, bone scintigraphy is thus extremely sensitive to both progression and regression of the metastatic process, so it is likely to have strong prognostic implications. It seems logical, therefore, to perform bone scintigraphy in patients with prostate cancer with new or worsening skeletal symptoms, at any stage of their disease, particularly if this investigation is able to predict the likelihood and to localise the level of impending spinal cord compression. The new extended method of evaluation of bone scintigraphy offers an advantage over the conventional Soloway method in that it concentrates on the areas of spinal metastatic involvement, which have a direct bearing on the development of complications such as spinal cord compression and, more importantly, on the level of this threat.

Our data demonstrate that in prostate cancer, bone scintigraphy performed at the time of development of hormone refractoriness is of high predictive value for the inherent risk of spinal cord compression. Whereas the value of this method of analysis remains to be put to the test in prospective studies, we believe that our findings provide the evidence that there is indeed a place for the well-established simple investigation of bone scintigraphy in the management of the patient with hormone-refractory prostate cancer.