HELICAL TOMOTHERAPY FOR ADVANCED ESOPHAGEAL CANCER IMPROVES TARGET CONFORMITY AND HOMOGENEITY: A COMPARISON WITH FIXED-FIELD INTENSITY-MODULATED RADIOTHERAPY

Cai Xu, M.D, Peng Huang, M.E., *Luhua Wang, M.D., Ye Zhang, M.D., Qinfu Feng, M.D., Guishan Fu, M.E., Zhihui Hu, M.E., Minghui Li, M.E., Zhiqiang Liu, Ph.D., and *Jianrong Dai, Ph.D. Departments of Radiation Oncology, Cancer Hospital/Institute, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, 100021, China Corresponding author: Luhua Wang, M.D., Department of Radiation Oncology, Cancer Hospital/Institute, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, 100021, China. E-mail: wlhwq@yahoo.com. Co-corresponding author: Jianrong Dai, Ph.D., Department of Radiation Oncology, Cancer Hospital/Institute, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, 100021, China. E-mail: dai_jianrong@163.com First author: Cai Xu, M.D., Department of Radiation Oncology, Cancer Hospital/Institute, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, China. Email address: xchdls@126.com Co-first author: Peng Huang, M.E., Department of Radiation Oncology, Cancer Hospital/Institute, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, China. Email address: cukyhuang@163.com Ye Zhang, M.D.: drzye1983@163.com Qinfu Feng, M.D.: 13811300221@163.com Guishan Fu, M.E.: guishan11@163.com Zhihui Hu, M.E.: huzhihui81@163.com Minghui Li, M.E.: liminghui82@126.com Zhiqiang Liu, Ph.D.: andyliuzhiqiang@sina.com


Introduction
Esophageal cancer (EC) is the eighth most common cancer worldwide. 1 Approximately 70% of all EC cases in the world occur in China, with squamous cell carcinoma accounting for more than 90% of Chinese EC cases. 2,3 Radiotherapy plays a key role in the treatment of localized EC, and is usually combined with concurrent chemotherapy. However, radiotherapy has the potential to cause substantial toxicity in the form of radiation pneumonitis (RP) and lung fibrosis as well as pericardial effusion, pericarditis, myocardial ischemia and heart failure.
Compared to three-dimensional conformal radiotherapy (3D-CRT), intensity-modulated radiotherapy (IMRT) enables the modification of radiation-dose distributions and improves conformal and homogeneous target coverage while potentially decreasing radiation dose to the normal tissues. [4][5][6][7][8] Therefore, IMRT may improve clinical outcomes and reduce radiation-related toxicity, 7 and itis emerging as a standard of care for EC. However, even with the use of IMRT, the incidence of radiation-induced toxicity remains high, 9,10 and definitive dose must still be limited in the case of bulky, long and complex targets in order to limit the dose delivered to the lungs, heart and spinal cord.
Helical tomotherapy (HT) is a rotational radiotherapy modality that when combined with image-guided IMRT, allows adaptive and dose-guided radiotherapy. HT has been widely used in clinical radiation oncology, since it is associated with sharper dose gradients, greater target coverage, better homogeneity and conformity, and significantly reduced doses to organs at risk (OARs). [11][12][13] Although several reports have compared HT with IMRT, few of these studies focused on EC patients; moreover, these studies generally had small sample sizes, assessed tumors located at certain primary sites and involved plans using simultaneous integrated boost techniques. [14][15][16] Large-scale studies on definitive radiotherapy in EC are very scarce. Thus, the aim of the present study was to determine the usefulness of HT in the treatment of advancedstage EC and compare target homogeneity, conformity and normal-tissue doses between HT and fixed-field IMRT (ff-IMRT).

Patients and Methods
We enrolled 23 consecutive patients who had cT3-4N0-1M0-1a thoracic EC (upper esophagus, 9 patients; middle esophagus, 6; distal, 6 and esophagogastric junction [EGJ], 2), according to the American Joint Committee on Cancer staging system (2002 and 2010), and were treated with ff-IMRT. Each patient underwent re-planning using HT and ff-IMRT by one physicist. For analysis, the patients were divided into the upper EC, middle EC and distal EC subgroups according to the location of the primary tumor, and the entire study population was termed the thoracic EC group. EGJ cancer was included in the distal EC subgroup.
Computed tomography (CT) was performed with a 5-mm slice thickness, and the CT data was imported to a treatment-planning system. The gross tumor volume (GTV), clinical target volume (CTV), planning target volume (PTV), nodes and OARs (lungs, heart, spinal cord, liver and kidneys) were contoured by the treating physician as per the Radiation Therapy Oncology Group 0436 protocol. 17 The CTV was defined as the GTV with 3-5 cm superior-inferior margins and 1 cm lateral and anterior-posterior margins, positive nodes with 1-cm uniform margins and the involved lymph node drainage area. The PTV was delineated with additional 0.5-cm margins around the CTV. All PTVs and OARs were contoured using the Pinnacle planning system, and the contoured volumes were transferred to the HT planning system. D e c e m b e r 0 8 , 2 0 1 5 All patients received 60 Gy to the PTV in 30 fractions. The dose uniformity requirement was Dmin(PTV) > 95% and Dmax(PTV) < 110% of prescription dose. The doses to the OARs were as follows: the V20 for both lungs (volume of both lungs receiving a dose of at least 20 Gy) was ≤30%; the mean lung dose (MLD) was ≤18 Gy; the V30 and V40 for the heart were ≤40% and ≤30%, respectively; and the maximum dose (Dmax) delivered to the spinal cord was ≤45 Gy. The treatment-planning software used in this study was HiART TomoPlan 4.1.2. The process of HT planning was similar to that previously described. 18,19 We used a field width of 2.5 cm, a pitch value of 0.287 and a mean modulation factor of 2.0. A normal dose-calculation grid (256 × 256 pixels) with a convolution-superposition algorithm was used for dose calculation. 20 The Philips Pinnacle Planning System 9.0 was used for ff-IMRT planning. The direct machine parameter optimization module, which was adopted for the planning, used 5-7 angles to evenly separate coplanar fields. The minimum segment area was set to 10 cm 2 , and the minimum segment MU was 10 MUs. A collapsed-cone convolution algorithm was used to calculate dosage, with a dose grid resolution of 4 mm.
Based on the International Commission on Radiation Units report no. 83, 21 dosimetric parameters were evaluated quantitatively. For the PTV, the parameters were D98% (near-minimum dose), D2% (near-maximum dose), mean dose, dose standard deviation, conformity index (CI), homogeneity index (HI) and target coverage (TC). The CI was defined as where VPTV is the volume of the PTV, TVPV is the volume of the PTV covered by the prescribed isodose line, and VTV is the treated volume of the prescribed isodose line. 22 The 95% isodose was selected as the prescribed isodose line. Possible CI values ranged from 0 to 1, with values closer to 1 indicating better conformity. The HI was defined as where D2 and D98 are the doses at 2% and 98% of the PTV, and DP is the prescribed dose. 23 Theoretically, a lower HI value represents a plan that provides a more homogenous dose distribution. TC was defined as where TVPD and TV are the target volume receiving the prescribed dose and the PTV, respectively. A value closer to unity represents better TC. 24 For each patient, analyses were performed to determine the MLD, the V5, V10, V15, V20, V25 and V30 of the lungs, the mean heart dose (MHD), the V20, V30 and V40 of the heart, Dmax to the spinal cord, maximum dose delivered to 1 cm 3 of the spinal cord (D1cc), V5 and V10 of normal tissue (NT; body excluding the PTV), V25 and V35 of the liver, V10 and V15 of each kidney and the mean dose to each kidney. Delivery time was calculated from beam-on to beam-off; it included the time required for radiation delivery and gantry rotation but not the time required for patient setup and image comparison.
Statistical analysis was performed using the Wilcoxon matched-pair signed-rank test and IBM SPSS Statistics (version 19.0.0). P values <0.05 were defined as statistically significant.

Results
The axial, coronal and sagittal views of the dose distributions in the HT and ff-IMRT plans for upper, middle and distal ECs are shown in Figure 1that showed the sharper dose gradient in HT.The HT plans were better than the ff-IMRT plans in terms of PTV coverage, conformity and homogeneity, but the 5-Gy isodose line of the HT plans showed slightly larger lung and NT volumes. Figure 2 presents the average dose-volume histograms (DVHs) for the PTV and OARs in the entire study population (thoracic EC group) and the three subgroups. The PTV line in the DVHs showed that HT was associated with better TC and homogeneity. The curves showed that in the low-dose regions (e.g., V5) of the lungs, HT was worse than ff-IMRT; however, in the relatively high-dose regions (e.g., V20), HT was significantly superior to ff-IMRT, especially for upper ECs. In terms of the V30 and V40 of the heart, HT was superior to ff-IMRT. The dose distribution in the PTV and OARs, and the delivery time for the entire study group and the three subgroups are listed in Tables 1 and 2, respectively. The data is presented as the averages over all patients and the errors at the standard deviation level.
The data for the PTV showed that compared with ff-IMRT, HT significantly improved homogeneity for thoracic, upper, middle and distal ECs (P < 0.05), with reductions of 38%, 31%, 36% and 33%, respectively, in HI. In addition, HT significantly improved conformity for thoracic, upper and middle ECs (P < 0.05), with increases of 9%, 9% and 18%, respectively in CI. However, in the case of distal ECs, similar conformity was achieved with HT and ff-IMRT (P = 0.125). The use of HT improved TC by 1% in the entire study group and the three subgroups (P < 0.05).
The lungs were better spared by HT. The MLD was significantly lower with HT than with ff-IMRT in the case of thoracic (10.9 Gy vs. 11.6 Gy) and upper ECs (11.6 Gy vs. 12 Gy; P < 0.05 for both). The V20 and higher dose volumes of the lungs were decreased by HT in all groups, but the differences were significant for thoracic, upper and distal ECs (P < 0.05), with reductions of 2.1%, 3.1% and 2.2%, respectively. The V20 and V25 of the lungs in the case of middle ECs did not significantly differ between HT and ff-IMRT, though they were better with HT. In the low-dose regions however, HT resulted in a larger V5, with increases of 3.5%, 1.5%, 7.2% and 3.2% for thoracic, upper, middle and distal ECs, respectively. The difference was statistically significant for thoracic ECs (P = 0.017; 56.6% vs. 53.1%), but not for the three subgroups (P > 0.05). In addition, V10 and V15 for both the entire group and the three subgroups did not significantly differ between HT and ff-IMRT (P > 0.05), except that V15 for upper ECs was better with HT (P < 0.05).
The sparing of the heart was also significantly better with HT than with ff-IMRT in terms of the V30 and V40 for the thoracic, upper, middle and distal ECs (P < 0.05), except in the case of V30 for middle ECs. The MHD for thoracic ECs was significantly lower with HT than with ff-IMRT (P < 0.05); this dose was also lower with HT in the three subgroups, but D e c e m b e r 0 8 , 2 0 1 5 the difference was not significant (P > 0.05).
In the case of the spinal cord, the dose constraints were always met with both HT and ff-IMRT. However, HT resulted in greater values of Dmax and D1cc of the spinal cord for thoracic, upper, middle and distal ECs.
For distal ECs, the mean dose to the left and right kidneys and the V10 and V15 of the kidneys were lower with HT, but the differences were not significant. The V25 of the liver was similar in both HT and ff-IMRT, but the V35 was significantly reduced by HT (P < 0.05).
The delivery time was slightly lengthened by HT for thoracic, middle and distal ECs (P < 0.05), but was similar for both HT and ff-IMRT in the case of upper ECs.

Discussion
In our study, both HT and ff-IMRT produced good plans and dose distributions, however, HT achieved superior HI, CI and TC for the PTV, indicating better target homogeneity, conformity and coverage than ff-IMRT, which may result in better clinical outcomes, and HT may be particularly beneficial in those patients who have target volumes in close proximity to the spinal cord because of its sharper dose gradient.Although it is generally agreed that in radiotherapy planning, improvement in conformity often occurs at the cost of worsening homogeneity, we have demonstrated that HT can provide not only better conformal TC but also more homogeneous dose distributions.
In our study, the order of priority in organ sparing was lungs, heart, and spinal cord for both IMRT and HT. We found that HT could reduce the volume of high doses of the lungs (MLD and V20 or higher), which may reduce the risk of radiation-induced lung toxicity, although HT may result in slightly larger low-dose volume of lungs, whereas, it is still very controversial whether larger low-dose lung volume could increase the risk of radiation pneumonitis. In a dosimetric study comparing IMRT with HT for ECs, Chen et al. demonstrated superior lung sparing in terms of V20 and MLD by HT. 15 Similarly, in the present study, HT significantly improved lung sparing in terms of the MLD, and V20 and higher dose volumes. Because lung V20 and MLD is generally considered the threshold dose for the lungs, [25][26][27][28][29][30] in theory, HT could reduce pulmonary complications. However, HT might result in a larger low-dose volume when rotational therapies are applied. 15 In the current study, HT resulted in a higher V5 of the lungs than did ff-IMRT, especially for thoracic ECs (P = 0.017); the difference was not significant for the three subgroups (P > 0.05). The larger low-dose volume of the lungs might result in radiation-induced lung toxicity; however, no long-term data exist on using HT for EC. Whether or not the low lung dose increasesthe risk of RP remains controversial, and this topic warrants further investigation. Wang et al. 31 reported that several DVH parameters (lung V5-V65 and MLD) were highly correlated with RP; however, V5 was the only factor significantly associated with RP. The 1-year actuarial incidences of RP of grade ≥3 in the groups with V5 ≤42% and V5 >42% were 3% and 38%, respectively (P= 0.001). 31 Lee et al. demonstrated that pulmonary complications were noted more often (35% vs. 8%, P = 0.014) when V10 was ≥40% vs. <40% and when V15 was ≥30% vs. <30% (33% vs. 10%, P= 0.036). 32 However, a later update of the study reported V5 ≥ 47% as a significant risk factor instead of V10 and V15. 33 A few studies have reported that reduction in the high-dose region is more important than a counterbalanced increase in the low-dose region.
10, 30, 34-36 Ahmed et al. 35 reported that 12 of 52 patients with locally advanced lung tumors treated with volumetric modulated arc therapy (VMAT) had demonstrable fibrosis on CT. The mean MLD and V20 in the patients with fibrosis were 12.3 Gy and 20%, respectively. Moreover, in 83% of all patients with evidence of lung fibrosis, the entire fibrotic region was encompassed by the ≥36 Gy isodose line. This suggests a characteristic pattern of fibrosis after VMAT. 35 However, few DVH parameters that are consistently related to RP have been reported, such as V5, V10, V15, V20 and MLD. 10, 25-36 Wang et al. reported that multiple dosimetric factors that define the shape of the DVH, rather than a single factor might affect the incidence of RP. 31 It is promising that in the current study, HT resulted in reduced V20 and higher dose volumes and MLD, comparable V10 and V15, and slightly increased V5. Considering all these factors, we think that compared to IMRT, HT for ECs might reduce the risk of RPor allow dose escalation to the tumor with acceptable lung toxicity.
The analysis of the data regarding the dose delivered to the lungs stratified according to the target location showed that compared with ff-IMRT, HT showed the best lung sparing for upper ECs, even with similar low-dose volumes, followed by thoracic and distal ECs. Patients with upper EC might benefit the most from HT, whereas those with middle EC might benefit less from HT. However, the true benefit of HT still awaits clinical results.
The effect of RT on the heart has been reported in many studies on EC. [37][38][39] It is reported that cardiac complications in this case are related to heart V30, V40 and higher dose volumes, and MHD. 40,41 HT might spare the heart better than does IMRT because of its sharper dose gradient, and might thus reduce the risk of heart diseases related to radiation. A study including 101 patients with inoperable ECs treated with chemoradiotherapy found that the risk of pericardial effusion increased significantly with a mean pericardial dose of >26.1 Gy (73% vs. 13%, P = 0.002) and a pericardium V30 of >46% (73% vs. 13%, P = 0.001). 41 Gayed et al. 40 reported a 54% rate of cardiac ischemia 3 months following chemoradiotherapy for EC; they found that most perfusion defects were located in the region of the heart receiving ≥45 Gy. The INT 0123 study reported a total of 11 (10%) treatment-related deaths in the group of patients receiving 64.8 Gy combined with chemotherapy, and 3 of these 11 deaths were cardiac related. 42 Thus, lowering the radiation dose to the heart may be potentially life-saving. In the present study, HT spared the heart better than IMRT in terms of the volume of high doses (V30, V40 and MHD), which could potentially decrease radiation-related cardiac diseases in long-term survivors. D e c e m b e r 0 8 , 2 0 1 5 Axial views pointed out how to apply the dose in the area of the spinal cord (Fig. 1). In the current study, the Dmax for the spinal cord was ≤45 Gy, with a 3-to 5-mm safety margin for setup uncertainty in all HT and ff-IMRT plans. However, HT resulted in slightly higher values of Dmax and D1cc of the spinal cord because of the lower priority than lungs. Better sparing of the spinal cord may result in a loss of sparing of the lungs. Therefore, Dmax and D1cc of the spinal cord were accepted as no more than 45 Gy in order to spare the lungs as much as possible.
Compared to IMRT, HT results in larger low-dose volume of the NT after whole pelvic radiotherapy in postoperative endometrial cancer patients. 43 In the current study, the HT technique also resulted in larger volumes of NT receiving low doses (e.g., V5, V10) as compared to ff-IMRT, which may cause radiation-induced secondary malignancy. Because most EC patients are elderly with poor prognosis, the secondary malignancy from radiation is generally not considered a significant clinical problem because such malignancies invariably develop long after radiotherapy. 44 We compared dosimetric parameters according to tumor location, and found HT could result in much better target conformity and normal tissue sparing for upper esophageal cancer.However, dosimetry plan comparison studies are very hard to perform adequately due to the many variables needing to be controlled.Some bias might exist because of the use of different calculation algorithms in the planning software systems 14,45,46 , and the order of priorityin organs sparing also greatly affects the resulting doses.

Conclusion
This study demonstrates that compared to ff-IMRT, HT can provide superior plans with improved target conformity, homogeneity, and coverage, and significantly reduce the irradiated volume of high doses of most critical structures (lungs and heart) in patients with advanced EC, which may improve survival and reduce radiation-induced toxicities. HT may be a treatment option for advanced ECs, especially upper ECs, tumors that are bulky, long and have a complex shape, and tumors in close proximity to the spinal cord. The potential role of HT in improving the clinical outcomes of patients with advanced EC must be confirmed using early and long-