Session Item

The FLAME trial showed that radiotherapy with iso-toxic focal boosting to the intraprostatic lesion(s) in patients with intermediate and high-risk prostate cancer improves disease-free survival (DFS). Due to the strict adherence to organs at risk (OAR) constraints, a focal boost of D98% > 90 Gy was reached only in 20% of the patients (median D98% 84.7 Gy). To ensure the maximum treatment benefit without increasing OAR dose, more complex techniques to improve high focal boost delivery may be needed. Therefore, it is important to identify those patients who most likely require a high boost dose to achieve control. In this study, we designed a risk model which predicts the 5-year DFS based on individual clinical characteristics and the delivered boost dose. 

The model was designed using data of the FLAME trial of 526 patients with available clinical data who were treated per protocol. The median follow-up was 72 months and clinical failure occurred in 99 patients. The considered set of model parameters consisted of the iPSA-value, cT-stage, ISUP grade group, age, use of hormonal therapy and the delivered boost dose. Penalized Cox regression was used for model development. Performance was evaluated with Harell’s C-index (discrimination), the calibration curve and calibration-in-the-large. To identify patients who would require a high boost dose to achieve control, we calculated the predicted 5-year DFS, assuming they were treated to a standard dose of 77 Gy. This was compared to predicted outcomes, assuming the same patients received D98% of 95 Gy.

The C-index of the model was 0.71 (95% CI: 0.69-0.72). The slope and intercept of calibration curve were 0.97 and 0.011, respectively. The observed versus predicted 5-year DFS was 84% vs. 83% in the control arm and 91% vs. 92% in the focal boost arm. The distribution of predicted outcomes for all patients given the actual D98% they received in the trial is shown in Figure 1. On the right side are the predictions if all would have received 77 Gy (bottom), or if all would have had a boost dose of 95 Gy (top). Figure 2 shows the predictions for separate categories of ISUP grade and T-stage. A trend towards a larger improvement with a focal boost was observed for patients with high ISUP grade disease. There was no trend observed for the impact of a focal boost for different clinical T-stages.

We developed a prediction model for DFS in patients with intermediate and high-risk prostate cancer treated with radiotherapy. The model shows reasonable discrimination and good calibration. The model can be used to identify patients who most likely require focal boosting through evaluation of their predicted probability of 5-year DFS. The improvement in the predicted DFS with a boost dose of 95 Gy, compared to the DFS with the actually delivered boost dose, suggests the possible gain if alternative techniques of high focal boost delivery are applied for these patients.