Changes in pulmonary function after incidental lung irradiation for breast cancer: A prospective study

PURPOSE: The aim of this study was to analyze changes in pulmonary function after radiation therapy (RT) for breast cancer. METHODS AND MATERIALS: A total of 39 consecutive eligible women, who underwent postoperative irradiation for breast cancer, were entered in the study. Spirometry consisting of forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1), carbon monoxide diffusing capacity (DLCO), and gammagraphic (ventilation and perfusion) pulmonary function tests (PFT) were performed before RT and 6, 12, and 36 months afterwards. Dose-volume and perfusion-weighted parameters were obtained from 3D dose planning: Percentage of lung volume receiving more than a threshold dose (Vi) and between 2 dose levels (V(i-j)). The impact of clinical and dosimetric parameters on PFT changes (Delta PFT) after RT was evaluated by Pearson correlation coefficients and stepwise lineal regression analysis. RESULTS: No significant differences on mean PFT basal values (before RT) with respect to age, smoking, or previous chemotherapy (CT) were found. All the PFT decreased at 6 to 12 months. Furthermore FVC, FEV(1), and ventilation recovered almost to their previous values, whereas DLCO and perfusion continued to decrease until 36 months (-3.3% and -6.6%, respectively). Perfusion-weighted and interval-scaled dose-volume parameters (pV(i-j)) showed better correlation with Delta PFT (only Delta perfusion reached statistically significance at 36 months). Multivariate analysis showed a significant relation between pV(10-20) and Delta perfusion at 3 years, with a multiple correlation coefficient of 0.48. There were no significant differences related to age, previous chemotherapy, concurrent tamoxifen and smoking, although a tendency toward more perfusion reduction in older and nonsmoker patients was seen. CONCLUSIONS: Changes in FVC, FEV1 and ventilation were reversible, but not the perfusion and DLCO. We have not found a conclusive mathematical predictive model, provided that the best model only explained 48% of the variability. We suggest the use of dose-perfused volume and interval-scaled parameters (i.e., pV(10-20)) for further studies.