Background
Intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) techniques are able to provide very high dose conformity for cancer radiotherapy; thus, the surrounding normal tissue and organs can be well protected when high-dose radiation is delivered to the target volume. However, many uncertainties exist in the treatment planning and operation process that can lead to deviations of the IMRT or VMAT dose distribution. Therefore, it is necessary to verify the irradiation dose distribution that is delivered by the accelerator before such kinds of treatments [
1]. Until recently, a hybrid plan has been adopted most often in 2-dimensional (2D) planar dose measurement verifications. The precision of irradiating doses is evaluated and verified by comparing the planned dose distribution, calculated by the treatment planning system (TPS), to the measured results [
2]. However, due to the lack of information about the correlations between the verification measurement results and the anatomical structure of patients, as well as the resulting lack of information about the actual irradiation doses to different target volumes and organs at risk (OARs), it is difficult to identify the geometric locations where dose errors occur during plan implementation, thus leading to inadequate clinical evaluation information [
3]. Recently, some 3-dimensional (3D) dose verification tools that provide patient anatomical structure information were applied clinically. These tools can provide important information such as the dose deviations, the pass rates and the locations of the dose deviations in the patients’ target volumes and organs, as well as identification of the error origins [
4,
5]. In this study, we adopted a 3D anatomical dose verification (3D-ADV) based on measurements of delivered dose fluence and patients’ anatomical images. Meanwhile, the traditional 2D phantom verification (2D-PDV) using an ionization chamber array with angular response correction applied an in-house software, were used to compare the efficacies of dose verification for IMRT and VMAT of nasopharyngeal carcinoma (NPC). As a result, the differences between the two verification methods and their clinical significances in evaluations of irradiating dose deviations were analyzed and clarified.
Discussion
An effective evaluation of the treatment operation in clinical practice should be able to reflect the true delivery condition of the treatment and any errors occurred in the planned parameters. The traditional 2D-PDV QA, especially the SGAC measurement which used only a single fixed incident angle to avoid the existence of angular response errors, can result in inconsistencies between the QA and clinical therapy conditions [
14], and not able to provide information regarding the relationship between the dose error and the anatomical structures. This leads to decreasing of QA abilities in clinical evaluations [
3]. In our study of 2D-PDV of 20 NPC IMRT/VMAT plans, the gamma pass rates obtained from the SGAC verification were significantly higher than the results from the MGAC measurements done with the same gantry angle of therapy. This might be explained by the fact that the SGAC verification cannot reflect the effects on the radiation dose caused by gravity-induced changes in the multi-leaf collimator (MLC) position, or the output dose angle dependence from the accelerator under different gantry angles. The results indicated that the pass rates corrected by MGAC verification with the MatriXX build-in angle correction factor were lower than those corrected by our in-house correction software. The reason could be that the correction factor of the former was determined according to the correction angle of the central detector and then used to correct all of the other chambers without considering the differences in the incident angles of each detector. The angle correction from the in-house software considered the influence of different incident angles of each detector in the 2D-IC array; therefore, the pass rates in the verification were significantly superior to the results without the independent incident angle correction.
3D-ADV can provide us with information such as the pass rates (the global, each target and OAR volumes in the measured area), the statistical results of deviations in the dose-volume histogram (DVH) parameters (including the dose volume and the volume dose) of each organ and the anatomical positions that correspond to the dose deviations. The 3D-ADV results from the 20 nasopharyngeal carcinoma patients, who received IMRT/VMAT irradiation, revealed that the mean global gamma pass rate was 99.75% ± 0.21%. Lee et al. [
15] reported that the mean gamma pass rates (3%, 3 mm) of 2D-PDV were 98.2% ± 1.3% and 98.5% ± 1.3% in nasopharyngeal carcinoma patients treated with IMRT and VMAT, respectively. Our results from the 2D-PDV with SGAC and MGAC on a similar group of patients were 99.55% ± 0.83% and 92.41% ± 7.19%, respectively. The global gamma pass rate from the 3D dose verification was similar with that from the 2D-PDV of the SGAC method but was higher than that from the 2D-PDV of MGAC measurement. The reason for the higher global pass rate in the 3D dose verification could be that, compared to 2D dose verification, the dose pixels evaluated in the 3D dose verification included all points within the CT scanning area, thus resulting in a relatively lower ratio of pixel numbers at which dose deviations occurred to the overall pixel number. Additionally, during the implementation of MGAC 2D-PDV, the pass rates might decrease due to the non-uniform phantom density that can result from the detector arrangement when the incident angles are parallel to the detector plane [
16]. The gamma pass rates of each PTV and OAR decreased at different levels in all cases when compared to the global pass rate, indicating that higher gamma pass rates might be caused by improper evaluation strategies in which some errors in the delivered dose distribution were disguised and ignored due to the use of the global pass rate in the evaluation.
In this study of 3D-ADV, no statistically significant difference was found in the HI of the PTVnx between the measurement based dose reconstruction and the planned value of the TPS. However, there were obvious alterations in the verification results for each patient that the HI deviation ranged from -15.62% to 19.34%, indicating that there were significant individual differences in the irradiation results. Some patients had greater HI values in the PTVnx with irradiated doses than the planned values, indicating decreased homogeneity of the dose distribution to the target volumes. Even greater deviations from the planned value were observed in the CI of the PTV1, which were the high-risk lymph nodular target volume, ranged from -32.45% to 13.97%, indicating that the dose conformity of PTV1 in some individual cases decreased a lot after the plan delivery, which might have led to inadequate dose coverage in this target volume or to elevated doses in the surrounding normal tissues. The CI of the PTV2 decreased by an average of approximately 2% (P < 0.05), thus showing a total reduction of the PTV2 dose conformity during the implementation of the plan.
Elmpt et al. [
17] reported a work of similar 3D-ADV, using the planning CT image to reconstruct the dose distribution in combination with a Monte Carlo calculation and the energy fluence of the actual treatment beams measured pre-treatment with the electronic portal image device (EPID). In their study on head and neck cancers IMRT cases, most deviations between the reconstructed delivered dose and the planned value in the PTV, including D5%, D
mean and D95%, were less than 3%, while the mean dose in the parotid gland decreased by 3.2% ± 1.2% and the maximum spinal cord dose increased by 3.1% ± 1.9%. Our 3D verification data showed that the deviations in the D2%, D98% and D95% of the PTVs and the deviations in the V100% and V95% were all within ±5%, which were similar to the previous report. In contrast, the maximum deviations of D2% in the optic nerves and the optic chiasm were 31% and 27%, respectively, suggesting that extra attention should be given to future plan verifications with regard to whether the delivered dose to these organs will exceed the clinical limits. In the reported study of Stasi [
18] and Carrasco [
19], a weak correlations or even no correlation has been found between the gamma index and the clinical impact of a delivery dose discrepancy, like the deviation on DVH for PTV and OAR volumes. The acceptance criteria for which we had the highest frequency of correlations were (3%, 3 mm), however, this criterion hid relevant clinical dose metric differences which is not clinically acceptable. Our study also shown that there might be clinically unacceptable dose discrepancy in some cases even the (3%, 3 mm) gamma pass rate was very high.
Traditional 2D-PDV is relatively simple and easy to perform. However, this method cannot provide information regarding anatomical positions and the dose volumes that correspond to the dose deviations; thus, it can only serve as a basic quality assurance tool for IMRT. using online measured results, 3D-ADV is able to reconstruct the dose distributions from patients' anatomical images and provide us with more clinical relative information, verify the delivery deviation in both dosimetric and geometric parameters in the same way of plan evaluation. In the other hand, unlike 2D-PDV devices, the 3D-ADV utilized an independent calculation algorithm in the dose reconstruction, which differed from that used in the planning system. Therefore, it might lead to additional discrepancy in the verification results, if the dose reconstruction computation of the 3D-ADV was not accurate enough. It is important that a very strict pre-commissioning and proper evaluation has been performed when the 3D-ADV system is used for the clinic treatment plan QA.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
HL performed this study as part of his Master thesis. SH assisted with the treatment planning and dosimetry measurements. XD conceived of the study, participated in its design and coordination and helped to draft the manuscript. JZ assisted with the computation program of angular correction for ion-chamber array. LC helped to perform the measurements and data analysis. All authors reviewed and approved the final manuscript.