Background: Bone was the most common site of metastasis in prostate cancer(PCa) patients and was correlated with poor prognosis and increasing economicalburden. Studies were limited on the prognostic prediction for metastatic PCapatients with the assistance of neural network. Methods: Four convolutional neural network (CNN) models were developed andevaluated to predict overall survival (OS) of PCa patients with bone metastasis.All the CNN models were first trained with 64 samples and evaluated with 10samples, two models used only bone scan images and two models used both bonescan images and clinical parameters (CPs). Predictions of the best models werecompared with those of two urology surgeons on 20 test samples. Results: Our best models could predict OS of PCa patients with bone metastasiswith AUC = 0.8022 by using only bone scan images and AUC = 0.8132 by usingboth bone scan image and CPs on 20 test samples. When the sensitivities(specificities) set equal to average level of urology surgeons, their specificities(sensitivities) were 0%(7.2%) and 30.77%(7.7%) higher, which showedsignificant advantages of CNN models. Conclusion: The CNN models were suitable to predict OS in PCa patients withbone metastasis using bone scan images and CPs. Our models showed betterperformance in terms of accuracy and stability than urology surgeons. Keywords: Bone Metastasis; Bone Scan; Convolutional Neural Network;Prostate Cancer; Overall Survival

Background The 5-α reductase inhibitors (5-ARIs) are the first-line drug managing benign prostatic hyperplasia (BPH). Unfortunately, some patients showed no responses to 5-ARIs therapy, even suffering from worse symptoms. Although the decreased expression of 5-α reductase type 2(SRD5A2) in BPH tissues might explain the 5-ARIs therapy failure, the mechanisms underlying SRD5A2 decreased remained unelucidated. Objectives To investigate the mechanisms of microRNA regulating the variable expression of SRD5A2 resulting in treatment failure of 5-ARIs. Materials and methods The expression of SRD5A2 and microRNAs in BPH tissues and prostate cells were detected by immunohistochemistry, western blotting, and quantitative real-time PCR (qRT-PCR). Dual-luciferase reporter assay was performed to confirm that microRNA directly combine to SRD5A2 mRNA. The apoptosis of prostatic cells was detected by flow cytometry. Results 13.6%, 28.8%, and 57.6% of BPH tissues showed negative, weak, and strong positive SRD5A2 expression, respectively. Normal human prostatic epithelial cell line RWPE-1 strongly expressed SRD5A2, whereas the immortalized human prostatic epithelial cell line BPH-1 weakly expressed SRD5A2. miR-1199-5p expression level in BPH-1 was remarkably higher than that in RWPE-1(P༜0.001), and miR-1199-5p expression was significantly upregulated in BPH tissues with negative SRD5A2 expression than those with positive SRD5A2 expression. After miR-1199-5p mimics transfection, SRD5A2 expression was decreased markedly in RWPE-1 cells, whereas after miR-1199-5p inhibitor transfection, SRD5A2 expression increased in BPH-1 cells. Dual-luciferase reporter assay showed that miR-1199-5p could bind the 3’ untranslated region of SRD5A2. Additionally, miR-1199-5p could decrease the sensibility of finasteride (100 µM) on RWPE-1 cells. Conclusion Our results demonstrate that SRD5A2 expression varied in BPH tissues and miR-1199-5p might be one of the several factors contributing to differential SRD5A2 expression in BPH patients.

Objectives Bone is the most common site of metastasis in prostate cancer (PCa) patients and is correlated with poor prognosis and increasing economic burden. Few studies have analyzed the prognostic prediction for metastatic PCa patients with the assistance of neural networks. Methods Four convolutional neural network (CNN) models are developed and evaluated to predict the overall survival (OS) of PCa patients with bone metastasis. All the CNN models are first trained with 64 samples and evaluated with 10 samples; two models use only bone scan images and two models use both bone scan images and clinical parameters (CPs). The predictions of the best models are compared with those by two urology surgeons on 20 test samples. Results Our best models can predict OS of PCa patients with bone metastasis with AUC=0.8022 by using only bone scan images and AUC=0.8132 by using both bone scan images and CPs on 20 test samples. The best Youden indexes of the two models are 0.6263 and 0.7142, respectively, which are 0.3077 and 0.3131 higher than that of the urologists’ average Youden index, which indicate that CNN models exhibit significant advantages. Conclusions CNN models are suitable to predict OS in PCa patients with bone metastasis using bone scan images and CPs. Our models show better performance in terms of accuracy and stability than urology surgeons.