Modeling normal tissue complication probability from repetitive computed tomography scans during fractionated high-dose-rate brachytherapy and external beam radiotherapy of the uterine cervix.

Abstract

To calculate the normal tissue complication probability (NTCP) of late radiation effects on the rectum and bladder from repetitive CT scans during fractionated high-dose-rate brachytherapy (HDRB) and external beam radiotherapy (EBRT) of the uterine cervix and compare the NTCP with the clinical frequency of late effects. Fourteen patients with cancer of the uterine cervix (Stage IIb-IVa) underwent 3-6 (mean, 4.9) CT scans in treatment position during their course of HDRB using a ring applicator with an Iridium stepping source. The rectal and bladder walls were delineated on the treatment-planning system, such that a constant wall volume independent of organ filling was achieved. Dose-volume histograms (DVH) of the rectal and bladder walls were acquired. A method of summing multiple DVHs accounting for variable dose per fraction were applied to the DVHs of HDRB and EBRT together with the Lyman-Kutcher NTCP model fitted to clinical dose-volume tolerance data from recent studies. The D(mean) of the DVH from EBRT was close to the D(max) for both the rectum and bladder, confirming that the DVH from EBRT corresponded with homogeneous whole-organ irradiation. The NTCP of the rectum was 19.7% (13.5%, 25. 9%) (mean and 95% confidence interval), whereas the clinical frequency of late rectal sequelae (Grade 3-4, RTOG/EORTC) was 13% based on material from 200 patients. For the bladder the NTCP was 61. 9% (46.8%, 76.9%) as compared to the clinical frequency of Grade 3-4 late effects of 14%. If only 1 CT scan from HDRB was assumed available, the relative uncertainty (standard deviation or SD) of the NTCP value for an arbitrary patient was 20-30%, whereas 4 CT scans provided an uncertainty of 12-13%. The NTCP for the rectum was almost consistent with the clinical frequency of late effects, whereas the NTCP for bladder was too high. To obtain reliable (SD of 12-13%) NTCP values, 3-4 CT scans are needed during 5-7 fractions of HDRB treatments.