Interobserver variability in gross tumor volume contouring in non-spine bone metastases.

Area of Science:

• Oncology
• Radiotherapy
• Medical Imaging

Background:

• Accurate gross tumor volume (GTV) delineation is crucial for stereotactic body radiotherapy (SBRT) in non-spine bone metastases.
• The optimal imaging modality for GTV delineation in these cases remains undefined.
• Comparing computed tomography (CT), magnetic resonance imaging (MRI), and 18-fludesoxyglucose positron emission tomography (18FDG-PET) is essential for improving treatment accuracy.

Purpose of the Study:

• To evaluate interobserver agreement in GTV delineation of non-spine bone metastases.
• To compare the accuracy of GTV delineation using CT alone versus MRI and/or 18FDG-PET.

Main Methods:

• Six radiation oncologists delineated 10 GTVs from anonymized CT, MRI, and/or 18FDG-PET images of non-spine bone metastases.
• Cases included blastic and lytic lesions across various primary cancers (lung, breast, prostate, rectum, urothelial, biliary).
• GTV volumes were compared, and interobserver agreement was assessed using the Landis and Koch protocol.

Main Results:

• CT exhibited the lowest interobserver variance and standard deviation in GTV delineation.
• MRI showed fair agreement (kappa=0.36) with CT, with generally larger GTV volumes.
• 18FDG-PET demonstrated slight agreement (kappa=0.08) with CT, with higher median GTV volumes.

Conclusions:

• CT offers the lowest interobserver variance for GTV delineation of non-spine bone metastases.
• MRI provides fair agreement, while 18FDG-PET shows only slight agreement compared to CT.
• Understanding and managing interobserver variability is key to optimizing SBRT quality for bone metastases.