Skeletal standardized uptake values obtained by quantitative SPECT/CT as an osteoblastic biomarker for the discrimination of active bone metastasis in prostate cancer

Prostate cancer is a common malignancy and the sixth most common cause of cancer-related mortality in the year 2015 among Japanese men (Miyoshi et al., 2015). Bone metastasis is seen in more than 90% of patients with metastatic castration-resistant prostate cancer (mCRPC) and is an important cause of death (Klaassen et al., 2017). Bone scintigraphy is a prevailing diagnostic test for the detection of bone metastasis in patients with prostate cancer because of its high sensitivity, cumulative evidence for detection, and easy accessibility (Langsteger et al., 2016). Analysis of the degree and extent of multiple bone metastases, cumulatively called as the extent of disease (EOD), has been based on visual diagnosis of whole-body images (Imai et al., 1992). Recent introduction of new therapies has focused on the subjective evaluation of the status of bone metastasis. For example, a semi-quantitative measurement of the bone scan index (BSI) by automated methods has been applied to investigate bone metastasis status (Nakajima et al., 2017). A single photon emission tomography/computed tomography (SPECT/CT) technique has been specifically employed in patients with prostate cancer (Helyar et al., 2010).

Presence off skeletal degenerative changes is frequent in prostate cancer, and may result in equivocal findings for bone metastases in whole-body scan (Zacho et al., 2015). In a differential diagnosis of bone metastasis from skeletal degenerative changes by skeletal SPECT/CT, localised skeletal destructive and/or sclerotic changes suggest bone metastases, while sclerotic changes or skeletal deformation along the marginal part of vertebra or joints suggest skeletal degenerative or inflammatory diseases (Tuncel et al., 2016). At an initial phase, bone metastases show limited skeletal anatomical changes that are detectable only by using bone SPECT; this is a more sensitive technique than the planar or whole-body scans. SPECT/CT has demonstrated additional diagnostic value to the standard skeletal scintigraphy (Helyar et al., 2010), and specifically, hybrid imaging using SPECT/CT has been proven to increase specificity and positive predictive value of the bone scintigraphy (Palmedo et al., 2014).

The standardized uptake value (SUV) is a feasible semi-quantitative parameter, frequently used for assessing uptake of radionuclide tracers. The SUV measurement in a modern SPECT/CT scan is reported to have adequate accuracy for quantitation of ^99mTc-radioactive concentration, with an error rate of 3.6% in phantom and 1.1% in patients with high activity bladder (Zeintl et al., 2010). To get better quantitative information from SPECT/CT scans, a high-resolution image reconstruction method with CT segmentation, termed as ordered subset conjugate-gradient minimizer (OSCGM) method (Armstrong & Hoffmann, 2016), and a conjugate gradient reconstruction with tissue zoning, attenuation, and scatter corrections applied (CGZAS) method have been introduced in commercial SPECT/CT devices (Gnesin et al., 2016).

It is not clear whether an SUV obtained by skeletal SPECT/CT scans and analysed using the CGZAS method has any clinical and diagnostic value for the evaluation of bone metastasis in patients with prostate cancer. Muzahir et al. reported that ¹⁸F–sodium fluoride (NaF) PET/CT was able to differentiate metastatic from degenerative joint disease by SUVmax in patients with CRPC (Muzahir et al., 2015). Similarly, the skeletal SUV by bone SPECT/CT has a potential to be a good biomarker of osteoblastic metabolism, although body size, renal function, and skeletal disease extent might affect its value. We hypothesised that precise skeletal SUVs measured by modern SPECT/CT can be used by itself to improve assessment of active bone metastases so that it can be demarcated from skeletal degenerative changes. Thus, the SUV obtained would reflect the aggressive osteoblastic activity in bone metastatic lesions, which is characteristic of metastasised prostate cancer.

The aim of this study was to clarify the clinical utility of skeletal SUVs for improving prognostication of active bone metastases, which were obtained by skeletal SPECT/CT scans and analysed using the CGZAS method. In particular, we investigated the ability of skeletal SUVs to discriminate bone metastases in patients with prostate cancer, and successfully demarcate active metastatic lesions from skeletal degenerative changes.

In the present study, we prove that a discrimination of active bone metastasis can be established with high accuracy in patients with prostate cancer based on SUV parameters. Thus, skeletal SUVs may help the differentiation of active bone metastasis from degenerative changes in hot foci in patients with prostate cancer. A pathological osteoblastic nature of bone metastases from prostate cancer is derived by the complex molecular interactions behind prostate cancer, affecting the balanced activity of osteoclasts and osteoblasts (Florimonte et al., 2016). During aging, the inflammation and tissue remodelling in chondral tissue around bone leads to calcification and ossification (Hayes et al., 2013). The different osteoblastic mechanism may affect SUV in prostate cancer with bone metastases and degenerative changes, reflecting the pathological osteoblastic nature of prostate cancer activity in higher SUV.

We observed that the skeletal SUVmax and SUVave in visually normal vertebral bodies in patients with prostate cancer were relatively low, at 7.58 ± 2.42 and 4.62 ± 1.68 in T, and 8.12 ± 2.24 and 4.54 ± 1.38 in L regions, respectively. Our observations are in agreement with previous publications reporting a SUVmax of 7.1 ± 0.4 (Kaneta et al., 2016), or a SUVave of 5.91 ± 1.54 (Cachovan et al., 2013) in normal vertebral bodies. In our study, the SUVs in regions with degenerative changes were greater than in visually normal vertebral bodies; however, the SUVs pertaining to degenerative changes were usually much lower than the skeletal SUVs in active bone metastases. This may be primarily due to the aggressive osteoblastic characteristic of bone metastases in prostate cancer. We speculate that the re-declined trend in SUVs with EOD grade 4 as high bone metastatic tumour burden is due to the decrease in the binding concentration of bone agents; this decrease is caused by the relative decrease in the amount of ^99mTc-MDP binding to the skeletal lesions, when the mass of bone metastasis increases due to disease extent to grade 4. In the phase 3, double-blind, randomised ALSYMPCA trial, relative risk for first symptomatic skeletal event was higher in EOD grade 3 than that in EOD grade 1 (Sartor et al., 2014). This phenomenon might account for the reason why nuclear medicine therapy by bone seeking radiopharmaceuticals was less effective in patients with EOD grade 4, in addition to increased tumour burden and its advanced stage (Kraeber-Bodere et al., 2000).

We observed that SUV correlated with disease diagnosis of bone metastasis even in cases with MV less than 10 cm³. In this study, the modern SPECT/CT system demonstrated visually effective and quantitative images than those reported by ordinal SPECT/CT (Beck et al., 2016). CT attenuation correction with CT zone mapping of bone cortex and medulla gave us the precise localisation of activity coming from the CT marginal resolution. A standard chi-square minimisation method underestimated the mean counts drawn from Poisson distribution. To remedy this, we used the CGZAS method and improved the estimation by using a region-weighted chi-square minimisation method proposed by Mighell (Mighell, 1999). Gnesin et al. concluded in their publication that this system provided accurate activity recovery within 10% of correct values and its performances were comparable to current generation PET/CT systems (Gnesin et al., 2016). Thus, image reconstruction by the CGZAS method improved both visual and quantitative analysis of SPECT/CT images compared with reconstruction by an ordinal OSCGM method (Armstrong & Hoffmann, 2016). To our best knowledge, this is the first report about clinical application and validation of the CGZAS method in patients with prostate cancer.

Our results suggest that the skeletal SUV can function as a reliable osteoblastic biomarker for discriminating active bone metastases with feasible accuracy. In recent literature, where bone scintigraphy was applied as a diagnostic tool, the BSI was proposed to be a good index for the extent of bone metastasis (Dennis et al., 2012; Takahashi et al., 2012; Ulmert et al., 2012; Nakajima et al., 2013; Kaboteh et al., 2013). An automated analysis using BSI has shown higher accuracy of 0.957 in AUC than the accuracy of our AUCs by the SUVs in prostate cancer on a per patient basis, using a specialized Japanese database (Nakajima et al., 2013). Since the automated methods with BSI analysis also depended on data from a database discriminating high-risk lesions from ordinal bone scans, it was generally impossible to detect bone metabolic disease. Compared with BSI, SUV seems to be a more sensitive index with a wider dynamic range, reflecting osteoblastic activity for each skeletal lesion with a suitable threshold; this is because SUV was derived from precise 3-dimensional data, while BSI was derived from 2-dimensional data.

Since SUV reflects active osteoblastic nature, it might be useful for the classification of skeletal lesions to diagnose biological aspects and select suitable therapy. SUV could be applicable to the dosimetry of bone lesions by radionuclide bone-targeted therapy with bone-seeking agents, since the concentration of the bone-seeking agents would be expected to be parallel to SUV (Pacilio et al., 2016). In addition to monitoring the effect of radionuclide bone-targeted therapy, including ⁸⁹Sr-chloride and ²²³Ra-dichloride that are available in Japan, SUV might be a direct index enabling the prediction of an effect in each bone metastasis based on each radiation dose, as well as of a side-effect on the haematopoietic activity by estimating the radiation dose in the bone marrow (Florimonte et al., 2016; Kraeber-Bodere et al., 2000; Miederer et al., 2015; Alva et al., 2017).

In the current study, the SUVmax by NaF-PET/CT in degenerative joint diseases was 2.6–49.9 (mean, 6.2), which was less than the bone SUVmax (mean, 16.7) by bone SPECT/CT. In contrast, the SUVmax by NaF-PET/CT in bone metastases was 11.2–188 (mean, 160), which was higher than the SUVmax (mean, 40.9) by bone SPECT/CT (Muzahir et al., 2015). AUC of 0.964 by NaF-PET/CT was slightly higher, but comparable to AUC of 0.926 observed in the present study, after lesion-based ROC analysis. As a result, NaF-PET/CT data was 2-fold higher uptake and high target-to-background ration than ^99mTc-MDP due to faster blood clearance caused by lower protein binding and lower molecular weight (Langsteger et al., 2016). Clinical investigation using bone SUV of ^99mTc-biphosphonate analogs is rare. Thus, since SUVmax gathered using the NaF-PET technique has been reported to be helpful in assessing tumour burden and response to therapy, as well as in predicting outcome in bone metastases, skeletal bone SPECT/CT may serve as an effective surrogate for NaF-PET (Cook et al., 2011; Rohren et al., 2015). It is also well-known that the SPECT/CT method using radiopharmaceuticals with a longer half-life is cost-effective, with greater insurance coverage and reimbursement, and thus, more accessible to patients and physicians. In addition, an innovative development of detector might set high-resolution bone SPECT/CT method comparable to NaF-PET in the future (Langsteger et al., 2016). As an advantage, 3-min SPECT/CT has been reported to be sufficient for the assessment of bone metastases as add-on to planar scintigraphy (Zacho et al., 2017). Lastly, it has been demonstrated that whole-body SPECT/CT may be more accurate and cost-effective than whole-body planar plus SPECT/CT (McLoughlin et al., 2016).

This study suffered from several limitations, including the lack of scans of the chest or lumbar-pelvic portion in several SPECT/CT images. Even with the availability of whole-body scans for reference, there is a possibility that lesions with the highest uptake may have accidentally been excluded our analyses. Additionally, we omitted patients with proven bone metastases who did not undergo quantitative SPECT/CT scans. Therefore, there is a possibility of selection bias in our study.

The skeletal SUVs by ^99mTc-MDP of bone metastases were greater than that of degenerative changes in patients with prostate cancer with adequate discrimination accuracy in the hot foci. Skeletal SUVs in quantitative SPECT/CT may be useful for the prognostication of bone osteoblastic metastatic burden in patients with prostate cancer. As the most accessible and easily measured bone SUV parameter, SUVmax might be the best candidate to serve as an osteoblastic biomarker for skeletal diseases in daily clinical practice.