With the proliferation of consumer-grade VR devices, remote meetings within virtual environments are becoming increasingly popular. Meeting segmentation can swiftly offer users a valuable, advanced understanding of past meeting discourse while enhancing team communication efficiency in virutual environments. Detecting topic Transitions within VR-based meeting, however, is an extremely difficult task due to the hinderance of non-verval communication in virtual environments. This paper explores the correlation between topic transitions, acoustic features, and changes in bodily posture in virtual environments, and proposes a novel approach that combines acoustic features and posture data for dialogue topic segmentation.