This article is mentioned in:

Introduction

Gastric cancer is the fourth most common cancer and the second leading cause of cancer-associated mortality worldwide. Every year, ~1 million individuals develop gastric cancer and >700,000 succumb to the disease (1). It is estimated that ~50% of these patients are from China, and the majority have been diagnosed at locally advanced or advanced stage (2,3). Radiotherapy has been an important component of adjuvant treatment for locally advanced gastric cancer. The upper abdomen, the radiotherapy target region, is one of the regions influenced most by respiration. The displacement of organs due to respiration is the main source of uncertainty in the volume of the radiation target. A study by Kim et al (4) measured the displacement of upper abdominal organs during respiration, revealing that the displacement of the liver, kidneys and spleen ranged between 8.9 and 13.0 mm. Another study by Hallman et al (5) quantified respiration-induced motion of clinical target volume (CTV) in patients with liver and pancreatic cancer, demonstrating that the mean distance of motion in these patients was 9.7 and 5.0 mm, respectively. These studies indicated that the target volume varied during respiration depending on the target location. The introduction of novel radiation technologies, including intensity-modulated radiotherapy (IMRT), has led to the decrease of the radiation toxicity in normal tissue, but the problem of target motion induced by breathing remains.

Since the treatment targets in gastric cancer adjuvant radiation include several subsites with different respiratory motion amplitudes, the characteristics of the target motion of all subsites should be considered in the process of treatment planning. However, little is known about the detailed displacement pattern of the target volume in gastric cancer. To guarantee effective dose coverage of the target volume, a wide margin has to be added onto the CTV in order to account for the displacement induced by respiration, increasing the radiation toxicity of organs at risk (OARs).

Previously, significant toxicity associated with adjuvant chemoradiation has been a serious concern (6). To ensure safety during radiation, compromises on the coverage and dosage of the target in the final treatment plan have to be made in numerous cases. Although accurate radiotherapy has become increasingly widely implemented, there is an increasing risk of underdosing sections of the target area. Therefore, accurate radiotherapy requires more precisely delineated areas of the target and normal organ tissue, increasing the demand for accurately describing the displacement pattern of regions of interest (ROIs).

The 4-dimensional computed tomography (4D-CT) technique is one of the most reliable approaches to accurately describe the displacement caused by respiration during radiation (7). The technique has been widely used in accurate thoracic radiotherapy, particularly in lung cancer treatment (8–10). Nearly all of the associated studies reached the conclusion that a more uniform dose distribution in the target volume and a smaller internal target volume (ITV) can be obtained through the application of 4D-CT. The majority of the applications of 4D-CT in upper abdominal radiotherapy are from research on liver and pancreatic cancer (11). There are currently no reports on the application of 4D-CT in the treatment planning of gastric cancer.

The present study was designed to examine the displacement patterns of the ROIs associated with target volume in the gastric cancer adjuvant radiation based on 4D-CT, and to investigate its value in treatment planning. Based on the analysis of ROI displacement patterns by 4D-CT, a recommendation of asymmetric expanding margins to account for target displacement in radiation for gastric cancer is proposed. The feasibility of using an easier and more reliable way to acquire accurate ITV3D based on the displacement data from 4D-CT is also investigated.

Patients and methods

Samples and clinical data

A total of 10 patients with gastric adenocarcinoma receiving adjuvant radiotherapy were enrolled in this study between November and December 2013. The median age was 58 (range, 49–67). The study protocol was approved by the Ethics Committee of Zhongnan Hospital (Wuhan, China; approval no. 2013018) and informed consent was obtained from all participants. The main characteristics of the patients are listed in Table I.

Table I. Main characteristics of the patients.

Table I.

Main characteristics of the patients.

Patient no.	Sex	Age, years	Primary tumor location	Surgery pattern	Staging (AJCC 6th)
1	Male	60	Proximal	D1	pT3N2M0
2	Male	62	Proximal	D2	pT4bN3M0
3	Female	61	Distal	D2	pT4aN3aM0
4	Male	49	Distal	D2	pT2N1M0
5	Male	67	Body and proximal	D1	pT3N+M0
6	Male	57	Distal	D2	pT4bN2M0
7	Male	49	Distal	D2	pT4NxM0
8	Female	58	Proximal	D2	pT3N2M0
9	Male	57	Distal and body	D2	pT2N3M0
10	Female	60	Body	D2	pT3N1M0

[i] AJCC, American Joint Committee on Cancer; D1, Dissection of all the Group 1 nodes; D2, Dissection of all the Group 1 and Group 2 nodes (12).

The inclusion criteria were as follows: i) Histologically confirmed gastric adenocarcinoma, curative gastrostomy with D2 lymph node dissection and R0 resection; ii) patients aged between 18 and 75 years at diagnosis; iii) stage T3 to T4 and/or N1 to N3 [2010 American Joint Committee on Cancer (AJCC) staging system 7th edition (11)]; iv) Eastern Cooperative Oncology Group (ECOG) performance status of 0–2; and v) adequate bone marrow function (hemoglobin ≥90 g/l, neutrophil count ≥1.5×109/l, platelet count ≥100×109/l); adequate liver function (serum bilirubin ≤1.5× upper limit of normal (ULN), aspartate aminotransferase (AST), and/or alanine aminotransferase (ALT) ≤3×ULN); and adequate renal function (serum creatinine ≤0.106 mmol/l, and calculated creatinine clearance ≤50 ml/min).

The primary exclusion criteria included the following: i) Receiving neoadjuvant treatment; ii) hepatic, renal, pulmonary or cardiac dysfunction; iii) severe comorbidities, such as uncontrollable diabetes mellitus, uncontrollable hypertension, myocardial infarction or unstable angina pectoris within 6 months of surgery; iv) severe postoperative complications, such as anastomotic fistula and pancreatic fistula; and v) diagnosis with another carcinoma before operation.

CT data acquisition

The patients were immobilized in a supine position with a vacuum foam pad and were advised to breathe normally. Pressure sensors (AZ-733V; Anzai Medical Co., Ltd., Tokyo, Japan) were fixed on the upper abdomen to record respiration signals. CT data were acquired using multi-slice large aperture CT (Somatom Sensation 64; Siemens Healthineers, Erlangen, Germany). The respiration cycle was spilt into 10 phases according to the signal received by the sensors. The phases included: EX0% (beginning of exhalation), EX25%, EX50% (middle of exhalation), EX75%, EX100% (end of exhalation), IN0% (beginning of inhalation), IN25%, IN50% (middle of inhalation), IN75% and IN100% (end of inhalation). Every patient underwent a regular 3D-CT scan, followed by a 4D-CT scan. CT data were integrated with the respiration cycle data with the MIM software (version 6.0; MIM Software Inc,. Cleveland, OH, USA). Overall, 10 series of CT images were reconstructed for each phase of respiration. All CT data were transferred to the Oncentra brachytherapy planning system (version 4.1; Elekta Instrument AB, Stockholm, Sweden). The CT image at EX50% was treated as baseline and the CT images from the remaining respiration phases were fused and reconstructed.

Definition of ROIs

Based on recurrence patterns and lymphatic drainage regulations, 3D anatomy markers were drawn up for target volume delineation in the postoperative radiation of patients with gastric cancer. The radiation target volume of the regional lymph nodes (LNs) was delineated according to the consensus of adjuvant radiation for gastric adenocarcinoma in the Hubei Cancer of Zhongnan Hospital.

CTV

The CTV is composed of 3 areas: Tumor bed, anastomosis and draining LN basin. The CTV3D was delineated upon simulation CT. The CTV4D was generated by combining the CTV of all 10 aforementioned breathing phases. In the present study, 7 representative ROIs were selected as the focus for displacements analysis. These were as follows: Anastomotic staples, station no. 9 LNs (LNs around the celiac artery), no. 10 LNs (splenic hilar LNs), no. 12P LNs (hepatoduodenal ligament LNs), no. 13 LNs (LNs on the posterior surface of the pancreatic head), no. 14V LNs (LNs along the superior mesenteric vein) and no. 16a2 LNs (LNs around the abdominal aorta between the upper margin of the celiac trunk and the lower margin of the left renal vein). The LN stations were defined according to the Japanese Gastric Cancer Association (12). The ROIs in the present study were composed of the main area of the CTV for adjuvant radiation of gastric cancer. Each ROI was delineated separately.

ITV

ITV was mainly used to account for the target displacement attributed to breathing and organ movements during radiation. In this study, the ITV4D was defined to be equal to CTV4D. ITV3D was generated by adding a margin of 1 cm radial, 1.5 cm distal and 1 cm proximal to the CTV, according to the recommendation of the European Organization for Research and Treatment of Cancer (EORTC) (13). For ITV3Dcal, a margin recommended based on the present displacement analysis was used, as described next.

Planning target volume (PTV)

The PTV encompassed ITV plus a setup error. The setup error is set as 5 mm in Zhongnan Hospital (evaluated by cone beam CT). The PTV4D was calculated by applying an expansion of 5 mm on the ITV4D. For the PTV3D and PTV3Dcal, the linear sum of all errors is an overestimate of the total errors in in the majority of cases. According to the 62nd Report of the International Commission on Radiation Units and Measurements (14), the PTV3D and PTV3Dcal were calculated based on the formula: PTV=CTV + (IM2 + SM2)1/2, where IM refers to the internal margin and SM is the setup margin.

Surgery

All patients had previously undergone curative resection with D2 LN dissection (12). This procedure involved the resection of the perigastric LNs, the left gastric artery, the common hepatic artery, the celiac artery, the splenic hilum and the splenic artery LNs. Pathological evaluation was performed on ≥15 LNs.

Treatment plans

All treatment plans were developed using the Oncentra treatment plan system. In total, 3 sets of treatment plans using IMRT technology were designed for each patient: Plan3D, Plan3Dcal and Plan4D. The prescription dose of radiation was 45 Gy in 25 fractions, and 7–9 coplanar and/or non-coplanar beams were used. The plans were optimized to obtain satisfactory dose coverage, with ≥95% of the PTV receiving the prescribed dose. The treatment plans were delivered with 6-MV photon beams.

Plan evaluation and dose limitation

A dose-volume histogram was used to analyze the coverage of the target volume, where VN was the volume of the ROI that received ≥N Gy, Dmax was the maximum dose administered to the ROI, and Dmean was the mean dose. The main dose limitations for OARs were as follows: Liver, V30<50%; kidney, Dmean<20 Gy or V20<30% and V5<65% for both; remnant gastric, V40<50% (no hot spot on it) and Dmax<54 Gy; spinal cord, V45<0.03 cm3; intestinal (delineated upon simulation CT from 2 cm above PVT in cephalic-caudal direction), Dmean<30 Gy, V50<10% (no hot spots on it) and Dmax<54 Gy. For patients with proximal gastric cancer, a dose limitation of V30<30% for the heart and V20<20% for the lungs was applied.

Displacement of ROIs

The displacement was measured in three directions: Left-right (LR) direction, anterior-posterior (AP) direction and cephalic-caudal (CC) direction. In order to compare the magnitudes of displacement in the different directions in 3D, a displacement vector (DV) was defined to quantify the displacement of all ROIs, calculated with the following formula: DV=(ΔX2 + ΔY2 + ΔZ2)1/2, where ΔX, ΔY and ΔZ represent the maximum displacement distances on the x-, y- and z-axis, respectively.

Statistical analysis

Statistical analysis was performed using SPSS version 19.0 software (IBM Corp., Armonk, NY, USA). Analysis of variance was used for comparison among groups and variables. Multiple comparisons between the groups were performed using SNK method. P≤0.05 was considered to indicate a statistically significant difference. The results are presented as the mean ± standard deviation. The 95% confidence interval (CI) of the displacement was calculated with the following formula: CI (95%) =X- ± S * zα/2, where X- refers to mean, S refers to standard deviation, zα/2=2.262.

Results

Displacement of ROIs

The displacement ranged widely for various ROIs. In the same direction, the anastomosis staples exhibited the largest displacement magnitude on the LR direction (X-axis; 5.5±2.5 mm), while no. 12p LNs went through the largest displacement on the AP direction (Y-axis; 5.3±2.7 mm) and the cephalic-caudal (CC) direction (Z-axis; 7.8±3.6 mm). A significant difference in the DV was observed among the 7 ROIs (P<0.001). The no. 12p LNs had the largest DV, while the no. 16a2 LNs had the smallest (Table II).

Table II. Displacement vectors of regions of interest for the 10 patients (mean ± standard deviation).

Table II.

Displacement vectors of regions of interest for the 10 patients (mean ± standard deviation).

	Displacement dimension, mm
Region of interest	Left-right (x)	Anterior-posterior (y)	Cephalic-caudal (z)	Displacement vector
No. 9 LNs	2.2±1.2	3.8±2.2	4.2±2.2	9.6±1.8
No. 10 LNs	3.3±2.5	4.2±2.3	7.6±3.1	14.1±2.4
No. 12P LNs	3.4±2.1	5.3±2.7	7.8±3.6	14.6±1.6
No. 13 LNs	2.6±1.8	4.9±2.1	7.1±2.9	12.8±1.7
No. 14V LNs	2.9±2.4	4.6±2.4	5.4±3.1	12.5±2.3
No. 16a2 LNs	1.7±1.1	3.1±1.5	4.7±2.2	9.1±1.1
Surgical staples	5.5±2.5	3.0±1.8	6.7±3.6	13.5±3.7

[i] LNs, lymph nodes.

Recommended margin for expanding the CTV

The 95% confidence interval (CI) was obtained based on the data of the displacement of the ROIs. The PTV3Dcal was calculated according to the displacement analysis data by expanding the margins of the CTV3D, according to the formula PTV3Dcal=CTV + (IM2 + SM2)1/2, IM was generated by taking the upper limit value, and SM was equal to 5 mm. Thus, the recommended safety margins for ROIs were obtained and are presented in Table III.

Table III. Recommended safety margins for the regions of interest.

Table III.

Recommended safety margins for the regions of interest.

	Margin dimension, mm
Region of interest	Left-right (x)	Anterior-posterior (y)	Cephalic-caudal (z)
No. 9 LNs	7.0	8.5	9.0
No. 10 LNs	8.5	9.0	12.5
No. 12P LNs	8.0	10.0	13.0
No. 13 LNs	7.5	9.0	11.5
No. 14V LNs	8.0	9.0	10.0
No. 16 LNs	7.0	8.0	9.0
Surgical staples	10.5	8.0	11.5

[i] LNs, lymph nodes.

Comparison of the PTVs

The 3 types of PTV were compared in each patient (Fig. 1A). The volume of PTV4D had no significant difference compared with that of PTV3Dcal (868.87±318.18 vs. 967.87±384.81 cm3; P=0.538). The PTV3D (1719.04±509.35) was the largest, significantly so compared with the PTV3Dcal or PTV4D (P<0.001; Fig. 1B). All PTV3Dcal and PTV4D were included within the PTV3D. For the majority of patients (8/10), the PTV4D was completely encompassed in the PTV3DCal (Fig. 2A-C), while the PTV3Dcal in certain sections was smaller than the PTV4D in 2 out of 10 patients (Fig. 2D-F). However, the PTV3D was larger than the PTV4D in all patients.

Figure 1. Comparison of the 3 types of PTV in the 10 patients. (A) The PTV3DCal, PTV4D and PTV3D in each patient. (B) The mean PTV3DCal, PTV4D and PTV3D in the 10 patients. PTV, planning target volume **P<0.01.

Figure 2. Representative comparison of PTV4D and PTV3DCal section by section. (A) Axial view, (B) coronal view and (C) sagittal view of the abdominal area in patient no. 7, revealing a larger PTV3Dcal compared with the PTV4D in all sections. (D) Axial view, (E) coronal view and (F) sagittal view of the abdominal area in patient 9, indicating that the PTV3Dcal is smaller than the PTV4D in certain sections. Yellow area, PTV3DCal; purple area, PTV4D. PTV, planning target volume.

Dosage analysis in the PTVs

The 3 treatment plans (Plan3D, Plan4D and Plan3Dcal) for all patients were compared. Following the analysis of the DVH and a slice-by-slice examination, no difference in the coverage and distribution was detected in the 3 types of treatment plan for all patients.

Dose of radiation on the OARs

The radiation doses on the OARs in the 3 types of treatment plan were compared. The Dmean and V30 of the liver showed no significant difference between the PTV3Dcal and the PTV4D treatment plans. The Dmean and V30 of the liver were significantly increased in the PTV3D treatment plan compared with those in PTV4D and PTV3Dcal treatment plans. The Dmean of the kidney (right and left) were significantly lower in PTV4D and PTV3Dcal compared with these in PTV3D treatment plan. However, the V20 of the right kidney showed no difference when PTV3D was compared with PTV4D or PTV3Dcal (Fig. 3).

Figure 3. Comparison of 3 different doses of the liver and kidney of PTV. For PTV4D, (A) the Dmean and (B) the V30 of the liver were significantly decreased compared with that in the PTV3D and PTV3Dcal. While the Dmean and V30 in liver were decreased slightly in the PTV4D compared with that in the PTV3Dcal, there was no significant difference. (C) The Dmean and (D) the V20 of the left kidney were also significantly decreased. (E) The Dmean of the right kidney was significantly decreased in PTV3Dcal and PTV4D compared with those in PTV3D groups. (F) The V20 of the right kidney showed no significantly difference among the three types of PTV. *P<0.05, **P<0.01. PTV, planning target volume; Dmean, mean dose; V20, volume of the organ that received ≥20 Gy; V30, volume of the organ that received ≥30 Gy.

Discussion

The main findings of the present study were as follows: i) Different displacement magnitudes and directions were described for each ROI in the adjuvant radiation treatment of patients with gastric cancer, implying that a uniform margin expansion for all target areas is not optimal for the generation of the PTV; ii) the PTV4D was smaller than the PTV3D; iii) the treatment plan of PTV4D was beneficial for the protection of the liver and left kidney; and iv) regarding the PTV and protection of OARs, Plan3Dcal lay between Plan4D and Plan3D, with a satisfactory PTV coverage, despite demonstrating certain PTV underestimation in ROIs with a large displacement range compared with that of Plan4D.

While a number of previous studies have addressed the displacement mode of upper abdominal organs or surgical staples (4,5,15), the present study, to the best of our knowledge, described for the first time the displacement mode of radiotherapy-associated lymphatic drainage stations of gastric cancer. The present findings revealed that different subsites of the target volume had different patterns of displacement, according to their anatomical positions. Generally, the displacement magnitude was much larger in the CC direction. The LN regions adjacent to the hepatic portal and splenic hilum exhibited a larger displacement in the CC direction, as they are located next to organs affected by the motion of breathing. In addition to the displacement in the CC direction, the anastomosis staple also demonstrated a large displacement in the LR direction, which may be due to the peristalsis of the stomach. Certain ROIs, including no. 9 and no. 12 LNs, exhibited much smaller displacement than anticipated, as they are located in the peritoneum and are fixed by the surrounding ligament. These findings should be taken under consideration in the process of expanding the margins of CTV to generate the PTV, as those areas with large displacement magnitude could be administered an insufficient radiation dosage, and wide margins should be considered to account for any displacement. However, for the areas with smaller displacement, a smaller margin is sufficient to cover the target volume and this could lead to a decrease in radiation toxicity.

The ITV was used to account for variations induced by the target displacement and deformation. These variations generally consist of two types, the interfractional and intrafractional variations. Interfractional variations comprise the changes in organ volume, including bowel/stomach variations, and weight, while the sources of intrafractional variations mainly come from respiration and peristalsis. Of all variations, the displacement induced by breathing is the main source of uncertainty in the adjuvant radiation for gastric cancer. As the main ROIs in the present study were the draining LN regions, which are less influenced by organ filling or weight changes, the interfractional variations were ignored, and the combination of CTV4D was defined as ITV4D.

In the dosimetric analysis, no differences were observed in the PTV coverage and distribution among the three treatment plans, while the volume of PTV4D was the smallest among the 3 sets of treatment plans. As similar studies on gastric cancer are few in number, only certain studies addressing other upper abdominal tumors could be reviewed. Matoba et al (16) compared treatment plans based on 4D- and 3D-CT in gastric lymphoma, and concluded that Plan4D was as effective as Plan3D and could minimize the exposure of the OARs to radiation. Xi et al (17) reached a similar conclusion regarding the radiation treatment of liver cancer. It is worth noting that, although the PTV4D is significantly smaller than the PTV3D and PTV3DCal, it is closer to the PTV3DCal than to the PTV3D. The difference between PTV4D and PTV3D is >800 cm3, while that between PTV4D and PTV3Dcal is <100 cm3, indicating that the PTV3Dcal also has potential in protecting the normal organs and tissues from radiation. Furthermore, the present study confirmed that the expansion margin according to the recommendation of the EORTC was sufficient to cover the displacement induced by respiration in all patients. However, the margin may be considered too large in certain target areas, where PTV3Dcal and PTV4D are significantly superior to PTV3D in protecting normal tissues.

The results of the present study demonstrate that 4D-CT has an advantage over traditional 3D-CT with regard to obtaining a more accurate target volume in treatment planning. However, treatment planning based on 4D-CT has certain limitations. 4D-CT demands higher requirements for equipment and personnel training. Furthermore, the delineation of the CTV on all phases of the CT image and treatment planning is more time-consuming than that in regular treatment planning. We hope to develop a novel method for performing treatment planning, based on 3D-CT, which retains the advantage of a smaller and more accurate target volume. Based on the analysis of the subsite target displacement patterns of 4D-CT, a non-uniform expansion of the CTV can be calculated and applied to generate the PTV3Dcal. This integrates the main advantage of treatment planning based on 3D-CT and 4D-CT. In the present study, the feasibility of the application of PTV3Dcal in treatment planning was investigated. The main result agrees with the hypothesis that the PTV3Dcal can decrease the PTV3D and provide better protection for OARs compared with the PTV3D.

However, in an analysis of target coverage of PTV3Dcal compared with PTV4D, 2 out of 10 patients experienced target underdosing. The upper 95% CI limitation of target displacement was used as the recommended expansion of PTV and an estimated 5% target underdosing was observed. The slice-by-slice dose coverage analysis revealed that the portion of PTV4D receiving an insufficient radiation dosage was small. To reach a balance between efficacy and toxicity, a 5% PTV4D underdose may be considered acceptable. The investigation into PTV3Dcal is one of the most important aspects of the present study, as it has provided a novel approach to delineation of a more accurate target volume, although certain improvements are necessary. A more accurate target displacement calculation based on a larger sample population is required, and the establishment of a more optimized mathematical model is required in order to integrate all the expansion margins.

Overall, the present study has described the detailed displacement mode of the main region of the CTV, in particular the displacement mode of the lymphatic drainage area, in the 4D-CT-based adjuvant radiation treatment of gastric cancer. A non-uniform expansion of CTV margins accounting for the displacement due to respiration is recommended, based on the displacement analysis data of ROIs. The PTV4D and PTV3Dcal are significantly smaller than the PTV3D and the treatment plans based on these are beneficial in protecting OARs. The PTV3Dcal treatment planning procedure may provide a novel, simple approach to delineation of a more accurate target volume, although certain further improvements are required.

Acknowledgements

Not applicable.

Funding

This study was supported by grants from the National Natural Science Funds of China (grant nos. 81172129 and 81472798).

Availability of data and materials

The datasets used or analysed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

JP and JG performed the research and wrote the manuscript. JM and XW collected and analysis data. HX collected and analysis the clinical data; FZ designed the project and final approved the manuscript. JD and YZ analysis part data, and reviewed the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of Zhongnan Hospital (Wuhan, China) (approval no. 2013018) and written informed consent was obtained from all participants.

Patient consent for publication

All patients enrolled in the present study provided written consent for publication.

Competing interests

The authors declare that they have no competing interests.

References