Abstract 748P
Background
Retroperitoneal liposarcoma (RPLS) is defined as a highly complex and heterogeneous disease, making its diagnosis and histologic typing difficult. Computed tomography (CT)-based radiomics has the potential to non-invasively characterize the so-called radiological phenotype, which can be exploited by machine learning. However, little is known about the ability to combine features from multi-omics data to improve the prediction of treatment response.
Methods
Eight independent cohorts of RPLS patients were enrolled from our center (n=183) and the validation cohort (n=184). To develop and validate a CT-based multi-omics model to predict response to neoadjuvant chemotherapy (n=15) and radiotherapy (n=20) by combining contrast-enhanced CT images, bulk transcriptomic data (n=100), genomic data (n=59), lipidomic data (n=50), scRNA-seq data (n=4), and multiplex immunofluorescence (mIF) data (n=39) from tumor samples to assess the immune and metabolic landscape.
Results
We developed a novel approach based on a traditional CT radiomics model to accurately predict overall survival (OS) and tumor heterogeneity of RPLS patients. Using annotations, we developed a machine learning approach to integrate multimodal features into a risk prediction model. Our multimodal model (area under the curve (AUC) = 0.782) outperformed unimodal measures, including clinical model (AUC) = 0.743), pathologic model (AUC) = 0.723), and radiomics model (AUC) = 0.661). The multimodal risk score was significantly decreased in post-neoadjuvant therapy with more mature tertiary lymphoid structures (TLS) and immune cell infiltration, such as plasma cells and B cells, compared to pretreatment, and retinol metabolism pathway was most significantly inhibited in the high multimodal risk score group with unfavorable prognosis.
Conclusions
Therefore, our study provides a quantitative rationale for using multimodal features to improve the prediction of prognosis, TIME, and response to post-neoadjuvant therapy in patients with RPLS using expert-guided machine learning with excellent performance. This may be a promising avenue if validated by further prospective randomized trials.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Scientific Research Project of Shanghai Municipal Health Commission (20214Y0087, 20204Y0409); "Young Talents" Training Plan of Shanghai TCM-integrated Hospital (No. RCPY0063); Scientific Research Project of Hongkou District Health Committee (No. 2302-02); Natural Science Foundation of Fujian province (No. 2023J011698); Natural Science Foundation of Xiamen City (No. 3502Z20227279).
Disclosure
All authors have declared no conflicts of interest.