A proposal for the stratification of the risk of locoregional failure after surgical resection, perioperative high dose rate brachytherapy, and external beam irradiation: The University of Navarre predictive model

Abstract

To develop a simple clinical model predictive of locoregional failure after complete surgical resection followed by perioperative high-dose-rate brachytherapy (PHDRB) and external beam irradiation (EBRT). Patients (n=166) enrolled in several PHDRB prospective studies conducted at the University of Navarre were analyzed. PHDRB was given to total doses of 16 Gy/4 b.i.d. or 24 Gy/6 b.i.d. treatments for negative or close/positive margins along with 45Gy of EBRT. After a median followup of 7.4 years (range, 3-12+), 50 patients have failed and 116 remain controlled at last followup. Tumor size, with a cutoff point set at 3cm (p=0.041) and margin status (positive and <1mm vs. negative ≥1mm, p=0.0001) were independent predictors of locoregional control. These two parameters were used to develop a four-tiered, hierarchical scoring system that stratified patients into low-risk (negative ≥1mm margins and size ≤3cm), intermediate-risk (negative ≥1mm margins, and size >3cm), high-risk (positive and <1mm margins and size ≤3cm), and very high-risk categories (positive and <1mm margins and size >3cm). This classification yields 5-year locoregional control rates of 92.3%, 78.0%, 65.5%, and 48.0% for low-, intermediate-, high-, and very high-risk categories, respectively. The predictive ability of the model is highly significant (p=0.0001) with an area under the curve of 0.72 (0.64-0.81). The risk of locoregional failure after combined surgical resection, PHDRB, and EBRT is mainly determined by the number of residual clonogens, which is inversely proportional to the status of the surgical margins and directly related to the size of the resected tumor. These two parameters generate a four-tiered predictive model that seems to be valid for a number of different common tumors and clinical settings.