The ethics board of the hospital approved the present study, and all of the investigations were conducted in accordance with the relevant guidelines and regulations. From March 2014 to March 2016, 24 patients with rectal cancer who underwent adjuvant or neoadjuvant radiotherapy were selected. The patient characteristics are summarised in Table 1.
CT scans (3 mm thick slices) of the patients’ whole abdomen and pelvis were obtained with the treatment position on a Siemens Emotion-Duo CT simulator. Standard commercial immobilisation devices were applied. A carbon fibre frame and thermoplastic mask fixation (Pelvicast system, Orfit, Wijnegem, Belgium) was used. The patients were in the supine position with a pillow under their heads. Their knees and ankles were supported with vacuum cushions, and their arms resting on their chests. In the prone position, a belly board (Civco Medical Solutions, Coralville, IA, USA) was applied to allow the abdomen to extend into its aperture. The patients were instructed to empty the bladder an hour before CT simulation. Gastrografin solution (600 mL) was administered orally an hour before scanning to better visualise the small bowel for delineation. CT scans were subsequently imported into the treatment planning system (Pinnacle 9.0, Philips Radiation Oncology, Fitchburg, MA. USA) for target delineation and treatment planning design. After the plan was confirmed, the patients were treated at the Medical Synergy Accelerator (Elekta Synergy, Elekta Oncology Systems, Crawley, UK), and when treatment they were required to keep their bladder moderately filled similar to simulation. CT images were obtained and defined as 1W, 2W, 3W, and 4W, respectively, on the Friday of weeks 1-4 during treatment under the same scanning conditions. Subsequently, the 4 weekly CT scans were automatically co-registered to the Plan CT respectively based on pelvic bone anatomy, algorithm of Normalized Mutual Information in treatment planning system was used.
Delineation of PS and BL
Per the delineation methods of small bowel from RTOG  and Robyn B , BL and PS were delineated for each patient’s group of CT images. BL was delineated along the bowel loop’s outer surface based on the contrast effect of Gastrografin solution and excluding the colon. The upper boundary was 1 cm above the superior level of the planning target volume (PTV), and the lower boundary was delineation of the small bowel until it ended. For the PS, the anterior and bilateral boundaries were the inner surface of the abdominal muscles, the posterior boundary was the vertebral body, sacrum, or sigmoid colon. The upper boundary was 1 cm above the superior PTV level. The lower boundary was parallel to the inferior sigmoid colon level. The PS included the small bowel and colon, but did not include the bladder, ovary, and uterus. A window width of 600 and window level of 40 were selected for delineation of the BL and PS and were completed by the same senior attending physician.
Target volume definition and treatment planning design
The target volume was delineated per the RTOG and NCCN guidelines [18-19]. The clinical target volume (CTV) included the lymphatic drainage area of the perirectal lymph nodes, presacral lymph nodes, and internal iliac lymph nodes, and some patients’ external iliac lymph nodes were included. A margin of 1 cm in the cranial-caudal direction and 0.5cm in the anterior-posterior and lateral directions was given to the CTV to form the PTV. The prescription was 50 Gy in 25 fractions to the PTV. In the Pinnacle 9.0 treatment planning system, 7 field IMRT plans used the PS (V15<830 cc) dose constraints were designed . The plans used a 6 MV X-ray CC convolution algorithm and a 0.3cm computational grid. An Elekta Synergy accelerator and 40 pairs of MLCs were selected. Dose constraints of V40<50% and V50<5% were used for the bladder and bilateral femoral head respectively. The target dose coverage required more than 95% of the PTV covered by 100% of the prescription dose and a maximum dose (Dmax) <54 Gy inside and outside the PTV. Subsequently, the IMRT plans from the Plan CT were copied to the 1-4W CT images which had co-registered to the Plan CT.
Evaluation of small bowel dose volume
The absolute irradiated volume (cc) of the small bowel was described by its volume exposed to 5-50 Gy with 5 Gy intervals. Each patient’s small bowel volume (or irradiated volume) was expressed by the mean value over their CT images. All of the patients’ small bowel volumes (or irradiated volumes) during treatment were expressed as their median volume values.
Evaluation of small bowel motion
The shift% was used to describe the small bowel movements, and shift%=SD/mean . The SD and mean were the standard deviation and mean of the small bowel volume (or irradiated volume) from all of the CT images. The variations among the patients were expressed by their median values. A larger shift% signified greater motion of the small bowel during treatment.
NTCP prediction of small bowels
The Lyman-Kutcher-Burman (LKB) calculation module in Pinnacle 9.0 was used to predict chronic complications of the small bowel (called NTCPC) [21-23]. The n (volume factor), m (slope of dose response curve), and TD50 (mean dose of 50% complication probability) parameters were set to 0.15, 0.16, and 55 Gy, respectively . The complications were defined as small bowel obstructions, perforations, or fistulas. Logistic formula NTCP= (1+(V50/V)k)-1 was used to calculate the acute toxicity of the small bowel based on its V15 (called NTCPA), where V50 and k were 130 cc and 1.1, respectively . Each patient’s NTCP was expressed by the mean value over their all of the CT images. The NTCP of all of the patients during treatment was expressed by their median values. The shift% here was used to describe the NTCP variations during treatment, and shift%=SD/mean.
Safety assessment of small bowel during treatment
V15<275 cc from Robyn B et al.  and Dmax≦54Gy were used as the criteria for safety evaluation of the small bowel during treatment. The small bowel was at risk when the value exceeded these criteria.
SPSS 19.0 software was used for the data analysis. Sigma Plot 10.0 and Microsoft Excel 2007 were used for figure plotting. A paired sample t-test was used to compare the differences between the two groups’ data, and their correlation was analysed via Pearson’s correlation coefficient. A two-tailed value of p<0.05 was considered statistically significant.