The aim of this study was to develop a deep learning-based method for segmentation of bones in CT scans and test its accuracy compared to manual delineation, as a first step in the creation of an automated PET/CT-based method for quantifying skeletal tumour burden. Convolutional neural networks (CNNs) were trained to segment 49 bones using manual segmentations from 100 CT scans. After training, the CNN-based segmentation method was tested on 46 patients with prostate cancer, who had undergone 18F-choline-PET/CT and 18F-NaF PET/CT less than three weeks apart. Bone volumes were calculated from the segmentations. The network's performance was compared with manual segmentations of five bones made by an experienced physician. Accuracy of the spatial overlap between automated CNN-based and manual segmentations of these five bones was assessed using the Sørensen-Dice index (SDI). Reproducibility was evaluated applying the Bland-Altman method. The median (SD) volumes of the five selected bones were by CNN and manual segmentation: Th7 41 (3.8) and 36 (5.1), L3 76 (13) and 75 (9.2), sacrum 284 (40) and 283 (26), 7th rib 33 (3.9) and 31 (4.8), sternum 80 (11) and 72 (9.2), respectively. Median SDIs were 0.86 (Th7), 0.85 (L3), 0.88 (sacrum), 0.84 (7th rib) and 0.83 (sternum). The intraobserver volume difference was less with CNN-based than manual approach: Th7 2% and 14%, L3 7% and 8%, sacrum 1% and 3%, 7th rib 1% and 6%, sternum 3% and 5%, respectively. The average volume difference measured as ratio volume difference/mean volume between the two CNN-based segmentations was 5-6% for the vertebral column and ribs and ≤3% for other bones. The new deep learning-based method for automated segmentation of bones in CT scans provided highly accurate bone volumes in a fast and automated way and, thus, appears to be a valuable first step in the development of a clinical useful processing procedure providing reliable skeletal segmentation as a key part of quantification of skeletal metastases.

Objective: To compare the diagnostic accuracy of the following imaging techniques in the detection of spine metastases, using magnetic resonance imaging ( MRI ) as a reference: whole‐body bone scintigraphy ( WBS ) with technetium‐99m‐ MDP , [18 F ]‐sodium fluoride ( NaF ) positron emission tomography ( PET )/computed tomography ( CT ) and [ ¹⁸ F ]‐fluoromethylcholine ( FCH ) PET / CT . Patients and Methods: The study entry criteria were biopsy‐proven prostate cancer, a positive WBS consistent with bone metastases, and no history of androgen deprivation. Within 30 days of informed consent, trial scans were performed in random order. Scans were interpreted blindly for the purpose of a lesion‐based analysis. The primary target variable was bone lesion (malignant/benign) and the ‘gold standard’ was MRI . Results: A total of 50 men were recruited between M ay 2009 and M arch 2012. Their mean age was 73 years, their median PSA level was 84 ng/mL, and the mean G leason score of the tumours was 7.7. A total of 46 patients underwent all four scans, while four missed one PET / CT scan. A total of 526 bone lesions were found in the 50 men: 363 malignant and 163 non‐malignant according to MRI . Sensitivity, specificity, positive and negative predictive values and accuracy were: WBS : 51, 82, 86, 43 and 61%; NaF ‐ PET / CT : 93, 54, 82, 78 and 81%; and FCH‐PET / CT : 85, 91, 95, 75 and 87%, respectively. Conclusions: We found that FCH‐PET / CT and NaF ‐ PET / CT were superior to WBS with regard to detection of prostate cancer bone metastases within the spine. The present results call into question the use of WBS as the method of choice in patients with hormone‐naïve prostate cancer.

Study Type – Diagnostic (exploratory cohort) Level of Evidence 2a What's known on the subject? and What does the study add? Staging of patients with prostate cancer is the cornerstone of treatment. However, after curative intended therapy a high portion of patients relapse with local and/or distant recurrence. Therefore, one may question whether surgical lymph node dissection (LND) is sufficiently reliable for staging of these patients. Several imaging methods for primary LN staging of patients with prostate cancer have been tested. Acceptable detection rates have not been achieved by CT or MRI or for that matter with PET/CT using the most common tracer fluoromethylcholine (FCH). Other more recent metabolic tracers like acetate and choline seem to be more sensitive for assessment of LNs in both primary staging and re‐staging. However, previous studies were small. Therefore, we assessed the value of [ ¹⁸ F]FCH PET/CT for primary LN staging in a prospective study of a larger sample and with a ‘blinded’ review. After a study period of 3 years and >200 included patients, we concluded that [ ¹⁸ F]FCH PET/CT did not reach an optimal detection rate compared with LND, and, therefore, it cannot replace this procedure. However, we did detect several bone metastases with [ ¹⁸ F]FCH PET/CT that the normal bone scans had missed, and this might be worth pursuing. OBJECTIVES: To assess the value of [ ¹⁸ F]fluoromethylcholine (FCH) positron emission tomography/computed tomography (PET/CT) for lymph node (LN) staging of prostate cancer. To evaluate if FCH PET/CT can replace LN dissection (LND) for LN staging of prostate cancer, as about one‐third of patients with prostate cancer who receive intended curative therapy will have recurrence, one reason being undetected LN involvement. PATIENTS AND METHODS: From January 2008 to December 2010, 210 intermediate‐ or high‐risk patients had a FCH PET/CT scan before regional LND. After dissection, the result of histological examination of the LNs (gold standard) was compared with the result of FCH PET/CT obtained by ‘blinded review’. Sensitivity, specificity, positive (PPV), and negative predictive values (NPV) of FCH PET/CT were measured for detection of LNe metastases. RESULTS: Of the 210 patients, 76 (36.2%) were in the intermediate‐risk group and 134 (63.8%) were in the high‐risk group. A medium (range) of 5 (1–28) LNs were removed per patient. Histological examination of removed LNs showed metastases in 41 patients. Sensitivity, specificity, PPV, and NPV of FCH PET/CT for patient‐based LN staging were 73.2%, 87.6%, 58.8% and 93.1%, respectively. Corresponding values for LN‐based analyses were 56.2%, 94.0%, 40.2%, and 96.8%, respectively. The mean diameter of the true positive LN metastases was significantly larger than that of the false negative LNs (10.3 vs 4.6 mm; P < 0.001). In addition, FCH PET/CT detected a high focal bone uptake, consistent with bone metastases, in 18 patients, 12 of which had histologically benign LNs. CONCLUSIONS: Due to a relatively low sensitivity and a correspondingly rather low PPV, FCH PET/CT is not ideal for primary LN staging in patients with prostate cancer. However, FCH PET/CT does convey important additional information otherwise not recognised, especially for bone metastases.

Background Physical activity has been shown to mitigate the unwanted psychological and physiological side effects of prostate cancer treatments, but sustainable exercise possibilities are limited. Objective Our objective was to examine whether football in a real-world setting (i.e., local football clubs) was safe and feasible in practice and could improve quality of life, mitigate decline in muscle mass and bone density, and increase fat mass in patients with prostate cancer. Methods In this pragmatic, multicentre, parallel randomized controlled trial, men diagnosed with prostate cancer were recruited from five Danish urological departments. Men (N = 214) diagnosed with prostate cancer were randomly allocated, using random generated lists (block size 4–8) stratified for center and androgen-deprivation therapy status, to either 1 h of football twice weekly in a local football club or to usual care, which was a 15- to 30-min telephone session covering their options for physical activity or free-of-charge rehabilitation delivered as standard in Denmark. Allocation was concealed from the trial investigator performing the randomization, but—given the nature of the intervention—this was not possible for personnel and participants. Assessments were performed at baseline, 12 weeks, and 6 months. The primary outcome was mean change difference in prostate cancer-specific quality of life at 12 weeks. Secondary outcomes were body composition, 12-Item Short Form Health Survey (SF-12) physical and mental health, and safety—reported as fractures, falls, and serious adverse events. Results Attrition was 1 and 3% at 12 weeks, and 5% and 5% at 6 months for the usual care and football groups, respectively. Prostate cancer-specific quality of life was equal between groups at 12 weeks (mean difference + 1.9 points, 95% confidence interval [CI] –1.0–4.8; P = 0.20) and at 6 months (+ 0.5 points, 95% CI –2.8–3.8; P = 0.76). Fractures were equally distributed, with two fractures in the usual care group and one in the football group. Likewise, body composition outcomes were equal. Mental health improved after 6 months of football (mean difference + 2.7 points, 95% CI 0.8–4.6; P = 0.006). Conclusions In this trial, community-based football was a feasible exercise strategy for men with prostate cancer. Football did not improve prostate cancer-specific quality of life but did improve mental health; the clinical significance of this is unclear. Trial registration ClinicalTrials.gov: NCT02430792.

Study Type – Diagnostic (case series) Level of Evidence 4 OBJECTIVES: To evaluate prospectively [ ¹⁸ F]‐fluorocholine positron‐emission/computed tomography (FCH PET/CT) for lymph node staging of prostate cancer before intended curative therapy, and to determine whether imaging 15 or 60 min after radiotracer injection is preferable. PATIENTS AND METHODS: In all, 25 consecutive patients with newly diagnosed prostate cancer (Gleason score >6, and/or a prostate‐specific antigen level of >10 ng/mL, and/or T3 cancer) were scanned before lymphadenectomy. Each patient was assessed twice with imaging, at 15 and 60 min after the injection with FCH. Images were compared with the results of histopathological examination of the surgically removed lymph nodes. Maximum standardized uptake values (SUV _max ) at 15 and 60 min were also compared. RESULTS: Histopathologically, metastases were present in removed lymph nodes from three patients. FCH PET/CT showed a high radiotracer uptake in four patients, the former three and a fourth. The sensitivity, specificity, positive and negative predictive value of FCH PET/CT for patient based lymph node staging of prostate cancer were 100%, 95%, 75% and 100%, respectively; the corresponding 95% confidence intervals were 29.2–100%, 77.2–99.9%, 19.4–99.4% and 83.9–100%, respectively. Values of SUV _max at early and late imaging were not significantly different. CONCLUSIONS: This small series supports the use of FCH PET/CT as a tool for lymph node staging of patients with prostate cancer. Values of SUV _max at early and late imaging did not differ. However, larger prospective studies are needed to validate these findings.

Objective: To evaluate the Bone Scan Index ( BSI ) for prediction of castration resistance and prostate cancer‐specific survival ( PCSS ). In this retrospective study, we used novel computer‐assisted software for automated detection/quantification of bone metastases by BSI . Patients with prostate cancer are M‐staged by whole‐body bone scintigraphy ( WBS ) and categorised as M0 or M1. Within the M1 group, there is a wide range of clinical outcomes. The BSI was introduced a decade ago providing quantification of bone metastases by estimating the percentage of bone involvement. Being too time consuming, it never gained widespread clinical use. Patients and Methods: In all, 88 patients with prostate cancer awaiting initiation of androgen‐deprivation therapy due to metastases were included. WBS was performed using a two‐headed γ‐camera. BSI was obtained using the automated platform EXINI bone ( EXINI Diagnostics AB , Lund, Sweden). In Cox proportional hazard models, time to castration‐resistant prostate cancer ( CRPC ) and PCSS were modelled as the dependent variables, whereas prostate‐specific antigen ( PSA ) level, Gleason score and BSI were used as explanatory factors. For Kaplan–Meier estimates, BSI groups were dichotomously split into: BSI <1 and BSI ≥1. Discrimination between prognostic models was explored using the concordance index (C‐index). Results: The mean (range) age of the patients was 72 (52–92) years, the median (range) PSA level was 73 (4–5 740) ng/mL, the mean (range) Gleason score was 7.7 (2–10), and the mean (range) BSI was 1.0 (0–9.2). During a mean (range) follow‐up of 26 (8–49) months, 48 patients became castration resistant and 15 had died; most (13) from prostate cancer. In multivariate analysis including PSA level, Gleason score and BSI , only prediction by BSI was statistically significant. This was true both for time to CRPC (hazard ratio [ HR ] 1.45, 95% confidence interval [ CI ] 1.22–1.74; C‐index increase from 0.49 to 0.69) and for PCSS ( HR 1.34, 95% CI 1.07–1.67; C‐index increase from 0.76 to 0.95). Conclusion: BSI obtained using a novel automated computer‐assisted algorithm appears to be a useful predictor of outcome for time to CRPC and PCSS in patients with hormone‐sensitive metastatic prostate cancer.