Variability of clinical target volume delineation for definitive radiotherapy in cervix cancer.

Abstract

Accurate target volume delineation is essential for radiotherapy delivery, yet significant intra and inter-observer variability is documented. We analysed the variation in cervical cancer clinical target volume (CTV) delineation. All INTERLACE participating centres completed two RTQA outlining exercises. The Trial Management Group created a consensus outline. A separate STAPLE algorithm outline was created. Using these two outlines an optimised gold standard was generated. Volume, maximum distance from DICOM centre in all directions, and Jaccard Conformity Index (JCI) were calculated and compared for each centres' outlines. Anatomical areas included within CTV were recorded to detect systematic differences. 21 outlines were compared for case 1 and 22 for case 2. Volume ranged from 340 cc to 676 cc (case 1) and from 458 cc to 806 cc (case 2). A maximum 4 cm difference between outlines was observed in one direction. JCI ranged from 0.51 to 0.81 (case 1) and 0.57 to 0.81 (case 2). Variation in anatomical areas included in CTV exists between the two cases and between centres. Significant inter-observer variation in cervical cancer delineation has been demonstrated. Ongoing efforts are needed to ensure inter-observer consistency through education, guidelines and multi-centre collaboration.