In advanced breast cancer, it is important to understand not only symptoms due to organ metastasis but also complex symptoms caused by systemic pathology. In this study, we used JMP Pro to perform latent class analysis (LCA) on data from 466 advanced breast cancer patients to attempt to stratify patients by pathology. Based on discussions from last year, we switched from conventional factor analysis to LCA. The analysis identified seven pathology clusters, and two groups, particularly those with pleural effusion, suggested distinct underlying pathologies. One group was a "lymphangitis carcinomatosa type" caused by lymphatic embolism, while the other was a "pleuritis carcinomatosa type" accompanied by pleurisy. Clear differences were also observed in CT images and LDH levels. Furthermore, in parallel with this study, we have established numerous organoids from breast cancer patients and are currently verifying the biological validity of the findings obtained through LCA by comparing them with pleural effusion-derived organoids. This study is an example of reverse translational research that connects clinical data analysis with basic research using JMP Pro. Future work aims to integrate this with AI analysis of pleural effusion images. In this presentation, I would like to report on the usefulness of stratification models using JMP Pro and future research developments, and explore the possibility of collaboration with researchers in related fields.
