Dose, volume, and tumor control prediction in primary radiotherapy of non-small-cell lung cancer.
BACKGROUND: To evaluate the influence of total dose and tumor volume on local control and survival in primary radiotherapy of non-small-cell lung cancer (NSCLC). METHODS AND MATERIALS: We retrospectively analyzed the clinical course and CT-derived pre- and post-therapeutic tumor volume data of 135 patients with NSCLC undergoing primary radiotherapy at our department between 1989 and 1996. Among these, a total of 192 spatially separated tumor volumes (135 primary tumors, 1 additional intrapulmonary tumor, and 56 involved lymph nodes) were available for analysis. In all patients, treatment was planned using CT-based three-dimensional treatment planning. The dose to each tumor volume was derived from the individual dose plans. Mean total dose was 59.9 Gy (range: 30-80 Gy). All but 3 patients were followed until death. For local control analysis, each tumor was analyzed separately, and its remission status was determined in serial follow-up CT scans. A total of 784 CT scans were analyzed. Actuarial local control analysis was performed for the 192 separated tumor volumes, and survival analysis was performed for the 135 patients. Tumor control probability was calculated using a Poisson statistical model. RESULTS: Overall 1- and 2-year local control rate was 50% and 37%, respectively. The 2-year local control rate for tumors <50 ccm, 50-200 ccm, and >200 ccm was 51%, 22%, and 10%, respectively (p = 0.02). The 2-year local control rate for dose levels < or = 60 Gy and >60 Gy was 28% and 43% (p < 0.001). For the subgroup of 147 tumors smaller than 100 ccm, the local control rate increased up to 70% (1 year) and 51% (2 years) with doses of more than 60 Gy. For tumors larger than 100 ccm, no dose effect was seen. Only 2 of 45 tumors >100 ccm were controlled more than 2 years. Multivariate analysis revealed tumor volume, total dose, histopathologic type, and grading as significant and independent prognostic factors for local control. The number of delay days by split course (if used) and application of chemotherapy was not found to influence local control. Overall 1- and 2-year survival rate was 42% and 13%. Total radiation dose, chemotherapy, and T and N stage---but not tumor volume---were found to be independent and significant prognostic factors for survival in multivariate analysis. CONCLUSION: Tumor volume is an important predictor of local control in NSCLC. We found a clear dose effect for local control and survival in NSCLC. Long-term local control for a significant proportion of patients seems possible for small tumors only (<100 ccm, i.e., maximum diameter 6 cm) with doses of 70 Gy and more. Tumors of > or = 100 ccm are unlikely to be controlled long term by conventional doses up to 70 Gy. These results support dose escalation in patients with NSCLC.