In this issue of the International Journal of Radiation Oncology, Biology, Physics, Allen et al. from the University of Michigan consider methods of predicting radiation-induced changes in whole lung function. They report that they were unable to identify a relationship between three-dimensional dosimetric parameters (e.g., the mean lung dose [MLD], or the percent of lung volume receiving 20 Gy [V20]) and reductions of pulmonary function tests (PFTs) in 43 patients receiving radiotherapy (RT) for lung cancer (1). They conclude that additional work needs to be done to develop better methods of predicting RT-induced changes in PFTs. We agree. Their negative result is likely primarily the result of the tumor type studied, plus several methodologic limitations. Prospective studies at the Netherlands Cancer Institute (NKI) and Duke University Medical Center have also attempted to relate 3D dosimetric parameters to changes in quantitative pulmonary function tests (2, 3). The NKI group reported, for 81 patients with breast cancer and Hodgkin’s lymphoma, a good correlation between the mean lung dose and reductions of the diffusion capacity (TL,CO) and forced expiratory volume in 1 second (FEV1) (r 0.58 and 0.74, respectively) at 3 months postradiotherapy (2). For TL,CO, the correlation improved after excluding patients that received chemotherapy. At Duke, only weak correlations were found (r 0.2–0.4), but the study group also included patients with lung cancer (3). Predicting changes in PFTs after RT for lung cancer is complicated, because there are many confounding factors. Often, tumor-related reductions in PFTs are present at baseline (4–8). Thus, RT-induced tumor shrinkage might actually lead to an improvement in PFTs. The post-RT PFTs, therefore, reflect both improvements in function resulting from tumor shrinkage and declines in function because of damage to normal lung. It seems logical, therefore, that the correlation would be improved if patients without central lung tumors (that frequently cause adjacent hypoperfusion or atelectasis) are excluded. In the Duke study, the correlation coefficient between 3-D dosimetric parameters and declines in PFTs was improved when patients with “central tumors and adjacent hypoperfusion” were excluded (r 0.3–0.9) (3). In the study from Allen et al., their correlations did not improve when the 6 patients with pre-RT “atelectasis involving at least a lung lobe” were excluded. Twenty-four of their 43 evaluated patients had Stage III disease and, presumably, had central mediastinal disease. A large fraction of these patients would be expected to have some tumorinduced reduction in lung function, such as reduced perfusion (5, 8–10), but very few of these were excluded from their subset analysis. The absence of atelectasis of an entire lobe with routine radiography does not exclude significant tumor-induced physiologic changes within the lung. Single photon emission computer tomography (SPECT) perfusion scans are more sensitive in assessing such regional functional heterogeneities than is computed tomography (10, 11). In the series from the NKI involving patients with breast cancer and lymphoma, there were good correlations between 3-D dosimetric parameters and declines in PFTs (2). This is consistent with the concept that tumor-induced functional changes make prediction of PFTs a particularly challenging problem in patients with lung cancer. Moreover, the exacerbation of coexisting pulmonary disease in lung cancer patients undoubtedly complicates prediction of PFTs. Interestingly, several surgical series demonstrated generally excellent correlations between declines in PFTs and the estimated percent of lung resected (r 0.6 to 0.9) (12–16, review see 17). This supports the concept that the sum of the regional injuries can be used to predict changes in whole organ function in parallel-architecture type organs such as the lung. In the Duke series, the correlation between the dosimetric parameters and the subsequent decline in PFTs was strongest in patients with a larger number of post-RT PFTs (i.e.,
Read full abstract