Correlation between target volume and electron transport effects affecting heterogeneity corrections in stereotactic body radiotherapy for lung cancer

摘要

Under Institutional Review Board (IRB) exemption, we retrospectively analyzed the treatment plans of 74 patients with lung cancer who were treated with SBRT at Indiana University School of Medicine, Indianapolis. There were 22 and 52 cases in the right- and left-lobe of the lung, respectively. Details of the prescribed doses for these patients are listed in Table 1. Every SBRT patient was immobilized using either an Elekta Stereotactic Body Frame (Elekta, Stockholm, Sweden) or a CIVCO Body Pro-Lok system (CIVCO, Kalona, IA), depending upon the patient size, comfort and suitability as decided at the time of simulation. Eclipse version 10.0 was used for the treatment planning. Plans were generated to cover 95% of the PTV with the prescribed dose, thus providing typically a 25% higher dose to the GTV. Eclipse provides two algorithms for heterogeneity correction; pencil beam convolution (PBC) and AAA. The PBC model in Eclipse is mainly for homogenous medium (water) dose calculation that gets supplemented with older models modified Batho, Batho power law and equivalent tissue air ration (ETAR) [18]. It was noted that both modified Batho and Batho power law gave almost the same results for SBRT, and hence default modified Batho with PBC was used. The ETAR option with PBC cannot be used with non-coplanar field arrangements, thus ETAR was not an option in this study as the majority of the SBRT fields were non-coplanar. Heterogeneity correction was applied to all clinically approved plans using an AAA algorithm. The AAA is considered a superior algorithm compared with older and pencil beam algorithms [7, 19–21]. Two additional plans were generated for each patient: (i) a treatment plan with no correction (NC), and (ii) PBC with modified Batho heterogeneity correction. In the PBC algorithm, the dose deposited at a point was calculated as a convolution of energy fluence, or total energy released per unit mass (TERMA), with the respective dose deposition kernel pre-calculated for a narrow beam in water [22]. The Batho power-law correction method is an empirical correction to account for both primary beam attenuation and scatter changes in heterogeneous materials. The modified Batho correction uses only the descending part of the TAR/TMR curve because the curve in the build-up region of a high-energy photon is no longer valid. However, PBC does not take into account changes in lateral electron transport. This provided us with an opportunity to differentiate between photon attenuation (PBC) versus electron transport (AAA). JMP software (ver. 9.0.2; SAS Institute, Cary, NC) was used for statistical analysis. All pairwise comparisons among the three calculation algorithms were conducted for the isocenter dose (D_Iso), D_95% and V_100% using the Steel–Dwass test. The correlation between dosimetric parameters and anatomical characteristics including ΔPL, the PTV and the distance between the PTV and the chest wall or mediastinum was assessed with Spearman's rank correlation coefficient. Statistical significance was defined as a P-value 0.05. The D_Iso relative to the prescribed dose for each plan is shown in Fig. 3a. The D_Iso (average ± SD) for the entire patient population was 110.8 ± 4.4%, 132.6 ± 4.3% and 124.2 ± 3.1% for NC, PBC and AAA, respectively. The ΔD_Iso calculated with PBC and AAA were larger than that calculated with NC by 21.8% and 13.4%, respectively, in our population, with both P 0.0001. The D_Iso of AAA was 8.4% smaller than that of PBC (P 0.0001). In Fig. 3b, the isocenter dose difference (ΔD_Iso) between the dose with and without heterogeneity corrections of PBC (ΔPBC) and AAA (ΔAAA) were plotted with ΔPL as shown in Eq 1. The slope of the lines for PBC and AAA are nearly identical. The ΔPBC and ΔAAA showed intermediate linear correlation with ΔPL (ρ = 0.60, P 0.0001 for both). Typically, patients are treated with better than 95% target coverage, as shown in Fig. 4a indicating V_100% with NC, PBC and AAA algorithms for all patients. The median V_100% was 90.6% (range, 42.3–98.7%), 100.0% (range, 92.9–100.0%) and 96.0% (range, 75.6–99.9%) for NC, PBC and AAA, respectively. The long error bars indicate the variability among the patients. The average ± SD of D_95% was 97.6 ± 5.0%, 114.6 ± 7.4% and 100.5 ± 2.8% for NC, PBC and AAA, respectively as shown in Fig. 4b. The D_95% of PBC and AAA were larger than that of NC by 17.0% (P 0.0001) and 2.9% (P 0.0001), respectively. The ΔD_95% of AAA plans was 14.1% smaller than that of PBC (P 0.0001). Figure 4c illustrates the evaluation of ΔD_95% of the PTV for each patient, representing a similar analysis to that of Fig.3b for D_Iso. The ΔPBC shows intermediate linear correlation with ΔPL (ρ = 0.58, P 0.0001), although ΔAAA showed weaker correlation with ΔPL (ρ = 0.41, P = 0.0003