Dosimetric comparison of stereotactic body radiotherapy using 4D CT and multiphase CT images for treatment planning of lung cancer: Evaluation of the impact on daily dose coverage

Abstract

Purpose

To investigate the dosimetric impact of using 4D CT and multiphase (helical) CT images for treatment planning target definition and the daily target coverage in hypofractionated stereotactic body radiotherapy (SBRT) of lung cancer.

Materials and methods

For 10 consecutive patients treated with SBRT, a set of 4D CT images and three sets of multiphase helical CT scans, taken during free-breathing, end-inspiration and end-expiration breath-hold, were obtained. Three separate planning target volumes (PTVs) were created from these image sets. A PTV_4D was created from the maximum intensity projection (MIP) reconstructed 4D images by adding a 3 mm margin to the internal target volume (ITV). A PTV_3CT was created by generating ITV from gross target volumes (GTVs) contoured from the three multiphase images. Finally, a third conventional PTV (denoted PTV_conv) was created by adding 5 mm in the axial direction and 10 mm in the longitudinal direction to the GTV (in this work, GTV = CTV = clinical target volume) generated from free-breathing helical CT scans. Treatment planning was performed based on PTV_4D (denoted as Plan-1), and the plan was adopted for PTV_3CT and PTV_conv to form Plan-2 and Plan-3, respectively, by superimposing “Plan-1” onto the helical free-breathing CT data set using modified beam apertures that conformed to either PTV_3CT or PTV_conv. We first studied the impact of PTV design on treatment planning by evaluating the dosimetry of the three PTVs under the three plans, respectively. Then we examined the effect of the PTV designs on the daily target coverage by utilizing pre-treatment localization CT (CT-on-rails) images for daily GTV contouring and dose recalculation. The changes in the dose parameters of D₉₅ and D₉₉ (the dose received by 95% and 99% of the target volume, respectively), and the V_p (the volume receiving the prescription dose) of the daily GTVs were compared under the three plans before and after setup error correction.

Results

For all 10 patients, we found that the PTV_4D consistently resulted in the smallest volumes compared with the other PTV’s (p = 0.005). In general, the plans generated based PTV_3CT could provide reasonably good coverage for PTV_4D, while the reverse can only achieve 90% of the planned values for PTV_3CT. The coverage of both PTV_4D and PTV_3CT in Plan-3 generally reserves the original planned values in terms of D₉₅, D₉₉, and V_p, with the average ratios of 0.996, 0.977, and 0.977, respectively, for PTV_3CT, and 1.025, 1.025, and 1.0, respectively, for PTV_4D. However, it increased the dose significantly to normal lung tissue. Additionally, the plans generated using the PTV_4D presented an equivalent daily target coverage compared to the plans generated using the PTV_3CT (p = 0.953) and PTV_conv (p = 0.773) after setup error correction. Consequently, this minimized the dose to the surrounding normal lung.

Conclusion

Compared to the conventional approach using helical images for target definition, 4D CT and multiphase 3D CT have the advantage to provide patient-specific tumor motion information, based on which such designed PTVs could ensure daily target coverage. 4D CT-based treatment planning further reduces the amount of normal lung being irradiated while still providing a good target coverage when image guidance is used.

Introduction

It is well known that thoracic tumors exhibit significant intra-fractional motion due to respiration [1], [2], [3], [4], [5]. Moreover, respiratory motion during CT acquisition can introduce severe motion artifacts, resulting in distortion of tumor shape, size and inaccurate assessment of tumor location [6], [7], [8], [9], [10], [11]. Therefore, the conventional approach for defining target volumes employs large margins, based on typical site-specific motion amplitudes (on the order of 1–2 cm). These margins are applied to a gross tumor volume (GTV) generated from a free-breathing CT data set to create a planning target volume (PTV). This increases the amount of normal tissue in the field, which could lead to an increased toxicity. It also limits dose escalation, which is necessary to improve local control for an early stage non-small cell lung cancer [12], [13], [14], [15]. Recently, some investigators have suggested that individualized margins should be generated based on patient-specific tumor motion [16], [17], [18], [19], [20], [21], [22], especially when using high dose, hypofractionated SBRT [23].

Early methods to account for patient-specific tumor motion use multiphase CT scans taken during free-breathing, end-inspiration and end-expiration breath-hold. These scans are then fused for treatment planning [15], [24]. This approach assumes that images taken during the extremes of respiration define the extremes of target motion, thus accounting for tumor position during all respiratory phases. Recent advances in imaging technology using four-dimensional (4D) CT have provided us with a more reliable tool to study patient-specific motion [9], [25], [26], [27], [28], [29], [30], [31]. The 4D CT technique synchronizes image acquisition with respiratory phase. By sorting a sequence of 3D images according to the phase of the respiratory cycle during which they were acquired, one can reconstruct multiple CT sets (generally 10 sets for 10 respiratory phases) at different phases of respiratory cycle. Thus, respiration-correlated CT data contain information about temporal changes in tumor position and shape.

Directly using 4D CT images for treatment planning requires that the GTV be contoured in each of the 10 phases to generate an internal target volume (ITV). This creates a tremendous workload for clinicians, dosimetrists and physicists. Thus, investigators have proposed using other post-processing tools [26], [32]. One such tool is the maximum intensity projection (MIP) reconstructed 4D data set. Briefly, MIP projections reflect the highest data value encountered along the viewing ray for each pixel of volumetric data, giving rise to a full intensity display of the brightest object along each ray of the projection image. Underberg et al. [32] have studied the use of MIP for target volume definition in 4D CT scans for lung cancer and verified the accuracy of MIP in a motion phantom. They concluded that MIP is a reliable clinical tool for generating ITVs from 4D CT data sets. Another study [33] also confirmed that MIP is superior in defining an internal target volume.

In our clinic, we are now using MIP reconstructed 4D data sets to define an ITV. Prior to 2006, when we purchased the GE Light Speed CT scanner™ (General Electric Healthcare Technologies, Waukesha, WI) and its 4D CT image processing and reconstruction software, we used three multiphase CT images for ITV definition. In this study, we compare the target volume, treatment planning, and daily tumor coverage resulting from different PTV designs using 4D MIP images and multiphase 3D images. Considering that the conventional approach using helical CT images with population-based margins is still widely used clinically, we also include this approach in our comparison study. Our purpose is, by studying the impact of different target definitions on daily target coverage, to determine which approach more accurately accounts for patient-specific tumor motion during SBRT for lung cancer.