Purpose: Clinical observations often reveal individual differences in the severity of lung fibrosis after definitive radiation therapy for lung cancer. Recent experimental studies suggest that the risk of developing lung fibrosis may be genetically controlled. The present study was undertaken to examine the magnitude of individual variation in the incidence and severity of lung fibrosis in a well-defined patient population treated by concurrent chemoradiation for limited small-cell lung carcinomas (LSCLC).
Materials and methods: Between 1989 and 1994, 56 patients with LSCLC were enrolled in one of two controlled prospective studies of concurrent chemotherapy and concomitant conventional (45 Gy in 25 fractions q.d. over 5 weeks) or accelerated (45 Gy in 30 fractions b.i.d. over 3 weeks) radiotherapy. Chemotherapy consisted of cisplatin and etoposide (PE) or PE plus ifosfamide and mesna (PIE). Of the 56, a group of 25 patients who had serial computerized tomography (CT) examinations of the chest and were deemed to have unequivocal radiographic complete responses were selected for this study. The severity of lung fibrosis was recorded for each patient from the CT images using an arbitrary scale (0 to 3) at 1 year after treatment. Radiographic fibrosis scores were recorded on 1-3 CT slices in 3 different dose-areas (20-30 Gy; 30-40 Gy; and >40 Gy) that were defined using the corresponding CT slices from the patient's CT treatment plan. Of these patients, 23 (92%) had at least 2 slices scored; 11 patients had all 3 slices scored.
Results: Among the clinical and treatment parameters investigated (including type of chemotherapy), only total dose and fractionation schedule were identified as significant and independent determinants of lung fibrosis. Radiographic fibrosis scores were higher in high-dose areas and among patients treated with the accelerated schedule. Using a fit of the proportional odds (PO) model based on the total dose and fractionation schedule, fibrosis score residuals were calculated for each patient. The residual for each score is defined as the difference between the observed and expected score based on the dose and treatment schedule received. Average residuals varied significantly among patients (p = 0.005, Kruskal-Willis test). Using a modified version of the PO model, the coefficient of variation in patient heterogeneity was estimated to be 10.1% (95% confidence interval: 6.2-14.9%). Inclusion of the heterogeneity factor, in addition to total dose and fractionation schedule, improved the fit of the PO model to an extremely high level of significance (p < 10(-7)).
Conclusions: Our data indicate that the risk and severity of lung fibrosis analyzed radiographically on CT images increases with total dose and with the use of an accelerated radiation schedule, for patients treated with chemoradiation for small-cell lung cancer. There was also demonstrable patient-to-patient heterogeneity, suggesting that the risk of lung fibrosis is strongly affected by inherent factors that vary among individuals.