Purpose: To determine the relationship between the local radiation dose and the decrease in lung function associated with thoracic irradiation.
Patients and methods: Twenty-six patients treated with thoracic irradiation for lung cancer, for whom three-dimensional CT-based dosimetry was used in treatment planning, were evaluated with before and after treatment pulmonary function tests. Six patients were treated with radiotherapy alone (2.15 Gy daily fractions), and 20 patients with concurrent chemotherapy (cisplatin, etoposide) with hyperfractionated (HF) radiation therapy (1.2 Gy in twice-daily fractions). Eleven patients treated with concurrent HF chemoradiation also received the radioprotector amifostine. The normalized decrease in the diffusing capacity for carbon monoxide (DL(CO)) was used as an objective measure of the change in lung function. The dose-volume histogram (DVH) data were used to estimate the local dose-response relationship for loss of DL(CO). In each subvolume of lung, the loss in normalized DL(CO) was assumed to be a sigmoid function of dose, ranging from no loss at low doses to total loss at high doses. The whole-lung decrease in DL(CO) was modeled as the sum of the local declines in DL(CO) over all subvolumes. Nonlinear regression analysis was used to estimate the parameters of the local dose-response function.
Results: The data are most consistent with a pronounced decrease in DL(CO) when the local dose (for radiotherapy alone or HF concurrent chemoradiation) exceeds 13 Gy (95% CI, 11-15 Gy). In patients who received amifostine in addition to HF radiotherapy with concurrent chemotherapy, this stepwise loss of DL(CO) occurred above 36 Gy (95% CI, 25-48 Gy). Grade 2 or higher pulmonary symptoms were associated with a DL(CO) loss of >30% (p = 0.003).
Conclusions: The decrease in pulmonary diffusion capacity correlates with the local dose to irradiated lung. Amifostine significantly reduces the loss in DL(CO). A local dose-loss relationship for normalized DL(CO) can be extracted from DVH data. This relationship allows an estimate of the loss of function associated with a radiation treatment plan. Different plans can thus be compared without resort to an empiric DVH reduction algorithm. The very low (13 Gy) threshold for deterioration of DL(CO) suggests that it is better to treat a little normal lung to a high dose than to treat a lot to a low dose.