Analysis of localization errors in the definition of the mantle field using a beam's eye view treatment-planning...

Reports of the treatment of Hodgkin's disease (HD) with radiotherapy using the mantle field technique have demonstrated that coverage of disease by the field blocks significantly compromises outcome. It is our hypothesis that the availability of computerized tomography images reduces the incidence of localization error, and that the use of beam's eye view treatment planning techniques may further improve localization. The purpose of this report is to assess the possible contribution of a three-dimensional treatment-planning system to tumor localization and mantle block drawing in patients with HD.

We evaluated the localization error rate of four experienced radiation oncologists as they drew the lung blocks for the mantle field. The analysis included 16 patients treated with mantle fields in our department between 1989 and 1991. In each case our computerized three-dimensional treatment planning system was used to generate a beam's eye view display of tumor volumes. Simulation radiographs for all 16 patients were overlaid with acetate film, and lung blocks were drawn by clinicians using only the simulation radiographs for reference. The process was repeated with the thoracic CT scans available for reference. The mantle block contours for each trial were then superimposed upon the beam's eye view plots of tumor volumes. The beam's eye view plot was our benchmark for the evaluation of errors of tumor localization. Localization errors were defined as touching or overlap of the shielding blocks onto tumors.

There was a high degree (p < 0.0003) of consistency in scoring across all pairing of clinicians and the results from all four were polled for the analysis. The overall error rate using the simulation radiographs alone was 18%. The rate was significantly lower (13%) when the CT images were available (p = 0.038). The axillary region had the highest localization error rate (41.7% with CT available and 27.1% with CT available) and the superior mediastinum had the lowest error rate (10.7% without CT, 8.5% with CT). Compared with a system such as beam's eye view, which could reduce the localization error rate to zero, the error rate with CT scans available is still significant [95% confidence interval (CI = 10-17.1%)]. Localization errors were more likely with increasing tumor size when CT scans were not available (p = 0.029). A similar trend was not seen when CT scans were available (p = 0.2). In a multivariate analysis, the use of CT scans predicted for reduced localization error rate (p = 0.03). Tumors in the axilla and inferior mediastinum had a greater relative risk than those in the superior mediastinum (p = 0.0001)

The availability of CT imaging offers an advantage in the outlining of the mantle field in the treatment of Hodgkin's disease. When the error rate is evaluated using a beam's eye view treatment planning system, a significant proportion of tumors may be overlapped by the outlined mantle blocks even when CT images are available for reference. The use of beam's eye view treatment planning in mantle field definition, especially for tumors in the axillary region, may reduce the incidence of geographic misses.