Purpose: For treatment of lung cancer, dose heterogeneity corrections and subsequent prescription alteration remain controversial. Previous dosimetry studies based on slab geometry with a single beam geometry do not represent the clinical situation. A circumscribed tumor within lung poses a more complex problem. Energy choice also remains controversial.
Methods and materials: An anthropomorphic phantom was modified by replacing lung cylinders (2.5 and 5.0 cm diameters by 5.0 cm length) with muscle-equivalent cylinders. The phantom was scanned on a CT simulator. Gross, clinical, and planning target volumes (GTV, CTV, PTV1 including tumor and regional nodes, PTV2 including tumor only) were designated slice-by-slice. Three-dimensional planning was performed with large fields (AP/PA/RPO) covering PTV1 and boost fields optimized for each PTV2, for 6 and 18 MV photons. Homogeneous, Ratio-Tissue-Air-Ratio (RTAR), and convolution-adapted RTAR (CARTAR) calculation algorithms were tested. Film was placed between phantom slices at the "tumor" levels. The phantom was irradiated with monitor units corresponding to homogeneous calculations, based on a homogeneous prescription. Measured and calculated doses were compared by isodoses and dose volume histograms. Ionization chambers and TLDs were also used for some test cases.
Results: The measured minimum dose covering PTV2 was within 5% of the homogeneous prescription dose of 70 Gy for 6 MV photons, while a lower dose (89% of prescription dose) was measured for 18 MV. The algorithms overpredicted the minimum dose to PTV2 by 6-18%. If the monitor units had been reduced according to simplistic heterogeneous calculations, the small PTV2 would have only been covered by 58 Gy for 18 MV irradiation. Based on this, a clinician may opt to actually increase the prescribed dose, thereby offsetting decreased monitor units. None of the algorithms predicted the diffuse penumbra associated with 18 MV photons in lung.
Conclusion: Before adjusting dose prescriptions based on heterogeneity corrections, realistic phantom studies must be performed. The accuracy and effect of the corrections must then be assessed. The deficient coverage of PTV2 by the 18 MV beam compares unfavorably with the slight increase (5%) in hot spots associated with 6 MV. Our studies support strong caution before reducing dose prescriptions based on simple algorithms.