Associations of static anatomical geometry with planned dosimetric endpoints in prostate radiotherapy: a multicenter...

To investigate associations between static anatomical geometry and planned dosimetric endpoints in prostate radiotherapy using a unified multicenter workflow, with particular emphasis on OAR dose metrics.

We analyzed 37 patients from four centers with planning CT and DICOM-RT objects (RTSTRUCT/RTPLAN/RTDOSE). ROI masks were rasterized from RTSTRUCT. Dose grids were reconstructed from RTDOSE, resampled to the CT grid for voxel-wise dose sampling, and DVH metrics were recomputed using a unified implementation. We extracted geometric features of targets and organs at risk (OARs), including volume and longitudinal extent, and evaluated their associations with key planned dosimetric endpoints, with particular emphasis on OAR dose metrics. Target coverage metrics were analyzed as secondary exploratory endpoints. A systematic QC audit was performed for atypical dose results. In addition, an independent institutional DICOM cohort was used to assess workflow applicability in a real-world clinical setting.

In the main cohort, bladder volume and longitudinal extent showed consistent, directionally aligned associations with multiple bladder DVH endpoints. Prescription-normalized sensitivity analyses in cases with explicit RTPLAN target prescription-dose information showed similar bladder-related patterns. Near-zero bladder D_2cc values were interpreted as reflecting geometric separation from the high-dose region rather than evident errors in dose reconstruction or DVH computation. The same processing workflow was successfully applied to the independent institutional DICOM cohort.

Under contemporary planning and PTV-margin practices, inter-patient anatomical geometry was consistently associated with planned OAR dosimetric endpoints within a static planned-dose (RTDOSE-based) DVH evaluation.