Analysis of Sites of Recurrence for Upper Third Oesophageal Squamous Cell Carcinoma

Introduction

To assess geometric agreement of volumes derived from pre-treatment diagnostic FDG-PET and post-treatment CT for patients with upper third oesophageal cancer recurrence. High agreement may allow the delineation of radio-resistant sub-volumes to be targeted for dose escalation in radiotherapy treatment planning, possibly increasing overall survival by reducing recurrence and toxicity risks.

Methods

Validations: two methods were employed to determine the percentage of Maximum Standard Uptake (%SUVmax) used to generate PET volumes; Deformable Image Registration (DIR) methodology was assessed; rigid registration error was determined. PET and post-treatment CT data from one patient was deformably registered onto their planning CT, %SUVmax volumes generated and post-treatment CT recurrence volume contoured; these structures were deformably mapped onto the planning CT and overlap metrics generated. Recurrence was classified using the MD Anderson method.

Results and Discussion

43%SUVmax and 45%SUVmax values delineated PET volumes to known volumes most accurately. These and 50%SUVmax were used to generate volumes on patient data. The DIR methodology demonstrated parity with literature comparisons. Patient data showed a moderate overlap and a type B failure (MD Anderson classification). Geometric uncertainties from large registration deformation (superior/inferior data extents) may significantly affect the overlap.

Conclusion

The study aim was partially achieved. A workflow was constructed and tested using data from one patient, indicating moderate agreement between %SUVmax and recurrence volumes. Adjustments to the DIR methodology and additional data and would enable more statistically significant conclusions to be drawn.