Interobserver variability leads to significant differences in quantifiers of prostate implant adequacy.

Abstract

To compare the prostate gland volumes and subsequent quantifiers of implant adequacy determined by five separate brachytherapists on postimplant CT images performed 1 day after prostate brachytherapy. Ten consecutive patients underwent a CT scan 1 day after prostate brachytherapy, using 125I. Four experienced prostate brachytherapists were recruited to participate in this study in addition to the lead author. Each reviewer was asked to independently contour the prostate on sequential axial CT images for all 10 patients using a commercially available treatment planning system. Prostate volumes were calculated along with commonly reported quantifiers of implant adequacy (minimal dose received by 90% of the prostate gland [D(90)] and percentage of prostate volume receiving 100% of prescribed minimal peripheral dose [V(100)]). The mean prostate volume (SD) was significantly different according to the individual reviewers: Rev 1, 50.89 cm(3) (10.6); Rev 2, 55.42 cm(3) (15.79); Rev 3, 40.02 cm(3) (9.95); Rev 4, 56.70 cm(3) (12.16); and Rev 5, 44.52 cm(3) (10.95); p <0.0001. Significant differences were also observed for the mean V(100): Rev 1, 83.4% (5.5); Rev 2, 85.4% (7.6); Rev 3, 89.9% (5.4); Rev 4, 77.7% (6.3); and Rev 5, 81.5 (4.3); p <0.0001; and for the mean D(90), reported as the percentage of the prescription dose: Rev 1, 87.9 (8.9); Rev 2, 92.5 (15.4); Rev 3, 102.6 (16.7); Rev 4, 75.1 (11.3); and Rev 5, 81.4 (8.1); p <0.0001. Reproducibility was measured using the intraclass correlation coefficient (ICC). The reproducibility of the prostate volume was fair (ICC = 0.639) but was poor for the dosimetric quantifiers (V(100) ICC = 0.344 and D(90) ICC = 0.275). Significant interobserver differences in prostate volume defined on postimplant CT scans were observed. These differences led to significant differences in commonly used dosimetric quantifiers according to the individual reviewer. Until a greater degree of agreement among reviewers can be established, inferences concerning postimplant CT-based dosimetric analysis will be problematic at best. Additional efforts to increase interobserver agreement are required.