Radiation-Induced DNA Damage in Uveal Melanoma Is Influenced by Dose Delivery and Chromosome 3 Status.

Purpose: The purpose of this study was to analyze the extent of DNA breaks in primary uveal melanoma (UM) with regard to radiotherapy dose delivery (single-dose versus fractionated) and monosomy 3 status. Methods: A total of 54 patients with UM were included. Stereotactic radiotherapy (SRT) was performed in 23 patients, with 8 undergoing single-dose SRT (sdSRT) treatment and 15 receiving fractionated SRT (fSRT). DNA breaks in the enucleated or endoresected tumors were visualized by a TUNEL assay and quantified by measuring the TUNEL-positive area. Protein expression was analyzed by immunohistochemistry. Co-detection of chromosome 3 with proteins was performed by immuno-fluorescent in situ hybridization. Results: The amount of DNA breaks in the total irradiated group was increased by 2.7-fold (P < 0.001) compared to non-irradiated tissue. Tumors treated with fSRT were affected more severely, showing 2.1-fold more DNA damage (P = 0.007) compared to the cases after single (high) dose irradiation (sdSRT). Monosomy 3 tumors showed less DNA breaks compared to disomy 3 samples (P = 0.004). The presence of metastases after radiotherapy correlated with monosomy 3 and less DNA breaks compared to patients with non-metastatic cancer in the combined group with fSRT and sdSRT (P < 0.05). Conclusions: Fractionated irradiation led to more DNA damage than single-dose treatment in primary UM. As tumors with monosomy 3 showed less DNA breaks than those with disomy 3, this may indicate that they are less radiosensitive, which may influence the efficacy of irradiation.

Distributed Bayesian Inference Over Sensor Networks.

In this article, two novel distributed variational Bayesian (VB) algorithms for a general class of conjugate-exponential models are proposed over synchronous and asynchronous sensor networks. First, we design a penalty-based distributed VB (PB-DVB) algorithm for synchronous networks, where a penalty function based on the Kullback-Leibler (KL) divergence is introduced to penalize the difference of posterior distributions between nodes. Then, a token-passing-based distributed VB (TPB-DVB) algorithm is developed for asynchronous networks by borrowing the token-passing approach and the stochastic variational inference. Finally, applications of the proposed algorithm on the Gaussian mixture model (GMM) are exhibited. Simulation results show that the PB-DVB algorithm has good performance in the aspects of estimation/inference ability, robustness against initialization, and convergence speed, and the TPB-DVB algorithm is superior to existing token-passing-based distributed clustering algorithms.