NTCP Modeling of Xerostomia Related to Parotid Dose from Whole-Brain Radiation Therapy

Abstract

We recently reported an association between parotid dose and xerostomia in patients receiving whole brain radiation therapy (WBRT). The purpose of this analysis is to perform normal tissue complication probability (NTCP) modeling to further explore the parotid dosimetric indices associated with patient-reported xerostomia. Patients receiving WBRT (25-40 Gy in 10-20 fractions) for any diagnosis were enrolled on a prospective observational study. WBRT was delivered with 3D-conformal radiation therapy using standard opposed lateral fields covering C1. The parotids were not prospectively delineated, but retrospectively delineated and considered one structure for analysis. Patients completed the University of Michigan Xerostomia Questionnaire (Xerostomia score, scaled 0-100, higher is worse) and a 4-point dry mouth “bother” score (1-Not at all, 2-A little, 3-Quite a bit, 4-Very much) at baseline, conclusion of WBRT, 2 weeks, 1 month (primary endpoint), 3 months, and 6 months post-WBRT. The clinical data were fitted by the Lyman-Kutcher-Burman (LKB) and Relative Seriality (RS) NTCP models using outcomes at 1 month post-RT. We defined toxicity as a ≥20 point worsening in xerostomia score, or a ≥2 point worsening in bother score. 55 patients who reached the primary time point of 1 month were eligible for analysis. Median WBRT dose was 30 Gy in 10 fractions. Median parotid mean dose was 17Gy (Range, 11-28Gy) and median parotid V20Gy was 48% (Range, 24-84%). Xerostomia score increased by ≥20 in 19 patients (35%), and bother score increased by ≥2 points in 11 patients (20%). Parotid V5Gy-V20Gy were found to correlate best with toxicity, with AUC 0.68 for xerostomia score and 0.69 to 0.71 for bother score. AUC values for parotid mean dose were 0.64 and 0.65 for xerostomia score and bother score respectively. The values for the D50, m and n parameters of the Lyman-Kutcher-Burman model were 22.3Gy, 0.84 and 1.0, respectively for xerostomia score and 28.4Gy, 0.55 and 1.0 for bother score. Similarly, the values for the D50, γ and s parameters of the Relative Seriality model were 23.5Gy, 0.28 and 0.0001, respectively for xerostomia score and 32.0Gy, 0.45 and 0.0001 for bother score. A statistically significant toxicity Odds Ratio of 8.6 and 9.1 was observed in patients with parotid V20Gy > 48% for xerostomia score and bother score, respectively. Clinically significant xerostomia was fairly common after WBRT and correlated well with parotid V5Gy-V20Gy. The dose-response for xerostomia could be determined by fitting the clinical data with the LKB and RS NTCP models. These data support the need for minimization of parotid dose in patients receiving WBRT. Additionally, they can be very useful in prospective treatment plan optimization and evaluation.