Normal Tissue Complication Probability for Hematologic and Gastrointestinal Toxicity in Postoperative Whole Pelvic Radiotherapy for Gynecologic Malignancies using Intensity Modulated Proton Therapy with Robust Optimization

Abstract

Hematologic toxicity (HT) and gastrointestinal toxicity (GI) are often observed in the postoperative whole pelvic radiotherapy (WPRT) for gynecologic malignancies since the irradiation field contains large volume of the bone marrow (BM) and bowel bag (BB). Although proton beam therapy with single field uniform dose optimization has a potential of the risk reduction of HT compared with intensity modulated X-ray therapy (IMXT), the risk reduction of GI has not been previously described. Thus, the aim of this in silico study was to evaluate the risk of HT and GI using intensity modulated proton therapy with robust optimization (robust IMPT). This dosimetric planning study included 9 patients who had previously undergone hysterectomy and WPRT at our institution from 2008 to 2020. For robust IMPT planning, we used two posterior oblique fields and a 3.5% beam specific margin from the distal and proximal edges of clinical target volume (CTV) in water equivalent space to discern the range uncertainty. The prescribed dose was 45 GyE in 25 fractions for 99% of the CTV (CTV D99). We evaluated CTV coverage using following indicators, CTV D99, conformity index (CI), homogeneity index (HI); dose volume statistics of BM received 10 and 20 GyE (BM V10 and BM V20), and BB received 40 GyE (BB V40); the planning robustness for CTV by shifting each 5 mm for six axes; and the risk of HT and GI using the Lyman-Kutcher-Burman normal tissue complication probability (LKB-NTCP) model. The parameters of LKB-NTCP model were determined to calculate the risk of grade ≥ 3 acute HT and chronic GI such as obstruction or perforation. We performed the comparative evaluation of NTCP value for HT and GI between robust IMPT and IMXT prescribed 45 GyE in 25 fractions for 95% of the planning target volume with 7-fields step and shoot technique. The Wilcoxon signed rank test was used for all statistical comparisons. In the LKB-NTCP modeling analysis, the NTCP values for robust IMPT were significantly lower than IMXT (HT: 0.06±0.02 for robust IMPT and 0.19±0.03 for IMXT, p<0.05, and GI: 0.01±0.004 for robust IMPT and 0.03±0.01 for IMXT, p<0.05, respectively). Moreover, robust IMPT was excellent CTV coverage; the mean and standard deviation of CTV D99, CI, and HI was 45.63±0.21 GyE, 1.00±0.001, and 0.03±0.01, respectively. The dose volume statistics for BM V10, BM V20 and BB V40 were 61.41±4.47%, 53.12±4.75%, and 17.77±3.45%, respectively. Changes in the mean value of CTV D99 in the shifted plan were within 2%, except for the anterior shift plan (left: -1.05%, right: -0.91%, superior: -0.97%, inferior: -1.17%, anterior: -2.59%, and posterior: -0.10%). Robust IMPT for WPRT reduced the risk of HT and GI without compromising CTV coverage, compared with IMXT. These results indicate that robust IMPT has a clinical advantage in reducing the risk of HT and GI in the treatment of gynecologic malignances.