Determination of tumour regression rates during radiotherapy for cervical carcinoma by serial MRI: comparison of two measurement techniques and examination of intraobserver and interobserver variability

Tumour regression rates of 11 patients with cervical carcinoma were estimat ed during external beam radiotherapy (EBRT) using serial MRI (average time interval 7 days; range 3-15 days). An average of five investigations (range 4-8) was performed per subject, Tumour volume was measured by two observer s using the Cavalieri method of modern design stereology in combination wit h (a) planimetry and (b) point counting. The mean precision of all the volu me estimates obtained by manually tracing the outline of the tumour was 6.6 %. The mean precision obtained by counting an average of 176 points per inv estigation on the same transects was 6.7%. The intraobserver repeatability of planimetry, interobserver reproducibility of planimetry and point counti ng were excellent with no significant difference between the volume estimat es obtained using either technique. Based on the planimetry measurements, i nitial tumour volumes ranged from 6.5 to 222 cm(3) (mean 63 cm(3), median 4 4 cm(3)). Based on the point counting measurements, initial tumour volumes ranged from 7.2 to 235 cm(3) (mean 68 cm(3), median 46 cm(3)). Tumour regre ssion began within a few days of commencing EBRT and showed an exponential relationship with time (p<0.01). There was good agreement between the regre ssion rates obtained by planimetry and those obtained by point counting. No significant correlation was found between initial tumour volume and tumour regression rate for either planimetry or point counting. Planimetry measur ements were, on average, obtained in about half the time taken for point co unting (ie. 30 min and 50 min, respectively). Although point counting is ge nerally likely to be the more efficient approach, planimetry may be the pre ferred approach for estimating tumour volume when a purpose built track bal l is available and the tumour morphology is relatively simple. Volume measu rement should be obtained using the Cavalieri method to ensure that the est imates are unbiased and that their precision can be predicted. The measured tumour regression rates may have important implications for improving loca l tumour control, optimum timing of brachytherapy and minimizing the risk o f radiation damage.