Setup errors in radiation therapy for thoracic tumor patients of different body mass index

Abstract Purpose To assess the setup errors in radiation therapy for thoracic tumors patients of different somatotypes, and to seek an individualized mathematical basis for defining the planning target volume (PTV). Methods Sixty patients with thoracic tumors were divided into four somatotypes according to their body mass index (BMI), and their body positions were setup by two groups of technicians independently. CT simulations were performed and the reconstructed radiography was digitally generated as reference images for location verification and error measurement. By setting positioning error ranges, the within‐range positioning correction rate was compared among groups. Results Position setups for patients in the emaciated group, moderate group, and overweight group were relatively stable (with minor setup error differences between the two groups of technicians). In emaciated group, moderate group, overweight group, and obese group, setup errors in the right–left direction (R‐L) were 2.2 ± 1.3 mm, 2.2 ± 1.6 mm, 3.9 ± 3.1 mm, and 8.8 ± 3.5 mm, respectively; whereas the setup errors in the four groups in the superior–inferior(S‐I) direction were 2.4 ± 1.8 mm, 2.1 ± 1.9 mm, 3.2 ± 2.6 mm, and 5.4 ± 3.5 mm, and in the anterior–posterior (A‐P) direction were 2.2 ± 1.7 mm, 1.9 ± 1.9 mm, 3.2 ± 2.9 mm, and 6.2 ± 4.2 mm, respectively. Moreover, in the moderate group, the positioning correction rate in the three directions (R‐L, S‐I, and A‐P) was 20%, 9%, 8% within the error range of 5–10 mm, and 3%, 0%, 1% with a more than 10 mm error range. However, in overweight group and obese group, the positioning correction rate in these three directions (also with a more than 10 mm error range) was 23%, 27%, 19% and 21%, 16%, 23%, respectively. Conclusions In radiation therapy for patients with thoracic tumors, the definition of PTV should be individualized. Meanwhile, with the increase in BMI, positioning correction rate has a tendency to rise too.


2.C | Equipment and materials
Varian Clinac CX linear accelerator was used for the radiotherapy treatment. Eclipse planning system (Varian Inc., Palo Alto, CA, USA) was used for the formulation of various radiation treatment plans.
The 64-slice CT simulator (Siemens, Munich, Germany) and all-digital X-ray simulator (Shandong Xinhua, China) were used for simulation positioning and planning validation. And we used the fixed body frame (MEDTEC, USA), thermoplastic sheet (MEDTEC, USA) and electrode paste (type I, Hangzhou Tianyi Medical Devices Co., Ltd.) to fix patients' position.

2.D | Positioning and measurement method of setup errors and definition of positioning correction rate
Patient took a supine position on the body frame, and was fixed with a spoon-shaped headrest, with their hands placed on the forehead cross-armed. A thermoplastic sheet was taken out from the thermostatic water tank and spread evenly on the patient body surface while fixed onto the body frame. After cooling, the shaped sheet was removed to drill three non-collinear holes (with diameters of about 5 mm, which was the same size of the metal head of the electrodes) where the sheet was closest to the skin (indicating small skin movement). Then it was again covered on the patient and the locations of the three holes were marked to attach electrode paste (no skin allergies were found in our study). Before treatment, the holes on the sheet would be aligned with the metal head of the electrode paste by technicians, and patient position was adjusted with the sagittal laser line so the sheet was fixed naturally. Before treatment, there were five times of consecutive validation for each patient, and setup was completed under simulator by two independent groups of technicians. Then a total of 300 sets of data were obtained from the AP and lateral validation images along the central axis. Because of the good visibility of bony structures on the 2D images, the repeatability and stability were better, so the sternum or vertebral body closest to the tumor was selected as reference point. The distance from the reference point to each boundary of the radiation field was measured. Then the absolute value of the difference of measurement-on-planning-system minus measurementon-simulator-validation-images was taken as the setup errors during repeating positioning. Thus the mean value of setup errors in the three directions (R-L, A-P, and S-I) was calculated. The determination of reference points and the measurement of actual distance were completed by an attending doctor. The error range was set between 3 and 10 mm, which was divided into four groups 2.E | Statistical analysis SPSS 17.0 software was used for statistical analysis. And group t-test or one-way ANOVA was used in comparison between groups, whereas LSD test was used in pairwise comparisons between groups. P < 0.05 indicates statistical significance.

3.A | Comparisons between the positioning by two groups of technicians
Only the three-dimensional positioning errors in the obesity group showed statistically significant difference between the two groups of technicians (Table 1).

3.B | Impact of BMI on setup errors
When compared with the moderate group, the emaciated group showed no significant difference in patient setup errors. There was significant difference between overweight group and obesity group for setup errors, and the value was significant difference in R-L direction in the obesity group. The similar conclusions were drawn in the other two directions. See Table 2.
3.C | Positioning correction rate in the four somatotypes The positioning correction rate of moderate group in the three directions (R-L, A-P, and S-I) within different error range was shown (20%, 9%, and 8% within the error range of 5-10 mm, and 3%, 0%, 1% within the error range of > 10 mm) in Table 3. The results in overweight group and obesity group showed two sets of positioning correction rate of 23%, 27%, 19% and 21%, 16%, 23%, respectively (within an error range of > 10 mm, in the order of R-L, A-P, and S-I).
This suggested that for 97% of patients in the emaciated group and moderate group, an estimated CTV-PTV margin of 10 mm was quite enough to make up for the setup errors generated by daily positioning. However, in the overweight group and obesity group, even with an estimated CTV-PTV margin of 10 mm, there were still 27% of the patients needing positioning correction in the R-L direction, and 21% needing correction in the A-P direction, and 23% in the S-I direction.
3.D | Correlation between positioning correction rate and error range in the four somatotypes Figure 1 showed that positioning correction rate in the emaciated group and moderate group was mainly concentrated on the error range of 5 mm, whereas no such central tendency was seen in overweight group or obesity group. This result indicated that for patients with BMI < 24 kg/m 2 , an estimated CTV-PTV margin of at least 5 mm was enough to avoid obvious setup errors in most cases (specifically more than 77% in the R-L direction, 89% in the A-P direction, and 90% in the S-I direction). On the other hand, for patients with BMI ≥ 24 kg/m 2 , the estimated CTV-PTV margins should be individualized.

| DISCUSSION
The purpose of setup before treatment is to repeat the patient position set by simulator, so as to repeat the PTV and the spatial relationship between radiation field and organs-at-risk, which was determined during previous planning. 13 Therefore it can ensure accurate beam irradiation on the target. 14 Hunt et al. thought that precise treatment was more affected by position error compared with conventional radiotherapy. 15 The dose distribution which is   In the study of Kutcher GJ etc., 16  high. At present, there is still a big gap between China's radiotherapy machine and the other countries', 20 and not all medical institutions have such equipment. Therefore, we can screen out the patients who are in need of the individualized radiotherapy plan by BMI from the perspective of evidence-based medicine, in order to improve the accelerator utilization, reduce unnecessary human and material resources and related treatment cost. However, because of the small sample size of this study, we need to expand the sample size in clinical work for further observation and analysis so as to find the best cost-benefit population by BMI.
However, there were some limitations in our study. First, we performed the five consecutive setup validations for each patient.
We must acknowledge that the limitation of this design was that the five consecutive validation pairs were unlikely to capture inter-fractional and large intra-fractional setup variations. Second, the results were not applicable if daily imaging-based setup is used for all patients even with conformal curative therapy due to the resource limitation. Therefore, our results were useful for allocating the imaging resources to the patients who would benefit the most.
In summary, for tumor patients with different BMIs, it was not enough to adopt a uniform PTV margin. And it should be adjusted based on the individual conditions of each patient. Moreover, BMI could be used to screen out patients who need individually adapted radiotherapy, so that the utilization of the accelerator could be improved, and unnecessary cost of manpower, material, and other related treatment expenses could be reduced.

CONF LICT OF I NTEREST
The authors declare that there is no conflict of interests to be disclosed.
All patients were informed consent and this study was carried out by the approval of the Hospital Ethics Committee.

R E F E R E N C E S SUPPORTING INFORMATION
Additional Supporting Information may be found online in the supporting information tab for this article.