Segmentation of the clinical target volume (CTV) in cervical cancer is a crucial step for radiotherapy. Existing methods overlook the importance of cancer progression stages and do not consider the spatial relationship between organs at risk (OARs) and the CTV, resulting in suboptimal segmentation outcomes. In this paper, we employ auxiliary information to guide the cervical cancer CTV segmentation. Patient cases are additionally annotated and classified into eight categories based on the cancer progression stages and surgical statuses. These annotations are extended to the size of the inputs and concatenated with them, allowing our network to learn more CTV information from the annotations. We simultaneously train the segmentation of OARs and the CTV, employing a shared encoder and LoRA layers to merge features from OARs and CTV segmentation. By merging the features, the spatial relationship between OARs and the CTV is leveraged. Additionally, we use a Poisson’s ratio to model the deformation of OARs under force and implement a data augmentation method by simulating these deformations. Extensive ablation studies and experiments on various baseline networks demonstrate the effectiveness of the proposed method. Our method provides a more generalized and accurate solution for CTV segmentation in cervical cancer.