Prospective Validation of a High-Dimensional Shape Model for Organ Motion in Intact Cervical Cancer

Abstract

Validated models are necessary for developing appropriate planning target volumes (PTVs) for intact cervical cancer, particularly when using intensity modulated radiation therapy (IMRT). A model proposed by Khan et al. (PMID 2227016) serves as the basis for a PTV recipe for several IMRT trials. Our objective was to independently validate this model using prospectively collected clinical trial data. We analyzed 37 patients with stage IB-IVA cervical cancer registered in one of two prospective IMRT trials. This validation study was a prespecified aim for the one of the trials. Patients received daily fractionated pelvic IMRT with concurrent chemotherapy. The clinical target volume (CTV) included gross tumor, cervix, uterus, parametria, upper vagina, and pelvic lymph nodes. The PTV consisted of an expansion of 5mm around nodes, 10mm around the vagina and parametria, and 15mm around the uterus, cervix, and tumor. Daily on-line kV-kV images were used for initial setup verification. We collected cone beam computed tomography (CBCT) scans before each fraction. The planning CT was rigidly registered to each CBCT. CTVs from the planning CT were cast onto each CBCT and manually redrawn, while blinded to the PTV, to account for organ motion and deformation, creating a new CTV for each fraction. For each fraction, the 95% isodose cloud from the planning CT was applied to the CBCT, and the CTV volume outside the 95% isodose cloud was computed. The primary aim was to determine the proportion of CTVs that were encompassed within the 95% isodose volume. A one-sided, one sample t-test was used to test the hypothesis that the probability of complete coverage was at least 95%. We used repeated measures analysis of variance (ANOVA) to measure inter- and intra-patient variation and paired t-tests to test differences in CTV volumes. The 95% isodose line completely encompassed 91.3% of all CTVs (95% CI 86.8, 95.8%), not significantly different from the 95% probability anticipated a priori (P = 0.052). The percentage of fractions missed per patient ranged from 0-45.8%. The overall proportion of missed CTV was small: the grand mean of the volume of CTV falling within the 95% isodose cloud was 99.9%, and 33.2% of misses were less than 1 cm3. There were no significant differences between the mean CTV volumes from the planning CT and either the first CBCT or the mean CTV from all CBCTs (P = 0.42 and 0.53, respectively). Treatment day did not affect coverage (P = 0.83). Using the model developed by Khan et al., the probability of CTV coverage was high, and the volume of CTV missed was minimal. This PTV expansion strategy is acceptable for clinical trials and practice; however, daily image guidance is recommended to avoid systematic large misses in select patients.