Computation Dosimetry of Radioligand Therapy Using Spatial Transcriptomics
Abstract
Purpose
Numerous radioligand therapy (RLT) agents are under various stages of development for treating late-stage cancers. We developed a computational methodology for identifying promising targets for RLT based on spatial transcriptomics analysis of tumor tissue.
Methods
Gene expression maps were captured by digital molecular barcoding on 14 ovarian cancer specimens. Each 2D map represents a 6.5 x 6.5 mm area of tissue, in which the spatial expression profile of 6000 genes is annotated. A second dataset representing gene expression in 3D (19 slices; 10 µm per slice) within a metastatic node was also obtained. Cells within the tissue are categorized by type. Radiation dose distribution is calculated by 2D convolution of radioligand distribution (assumed to mirror target expression) with the dose-point kernel for 177Lu. Injected dose is set to achieve an average dose of 100 Gy in a reference patient. Tumor cell kill is estimated based on a linear radiosensitivity dose response model.
Results
Radiation dose (in units of Gy) was estimated at the level of single cancer cells, allowing for relative comparisons between different patients, injected doses, targets, and radioisotopes. Tumor cell kill >95% was used as a metric to identify promising targets. Mixtures of 2 or 3 ligands were also optimized to maximize response over the entire cohort. Analysis of the 3D dataset, which allowed for more accurate dose calculation, suggest that 2D methods are reasonably accurate.
Conclusion
Spatial transcriptomics data can provide a biological rationale to guide the development of RLT agents. Given the high cost of ligand discovery and drug trials, this computational approach could help narrow down the field of potential candidates, accounting for the microscopic distribution of the molecular target relative to the cancer deposits. It also provides a valuable framework for optimizing combination agent therapy or comparing the efficacy of different radioisotopes.