Effect of body mass index on set-up errors in patients treated with image guided radiotherapy for pelvic malignancies

Purpose: To retrospectively evaluate the effect of BMI on set-up errors in patients treated with image guided radiotherapy (IGRT) for pelvic malignancies whom BMI values were measured before treatment. Additionally, based on these findings, we intended to determine optimal PTV margins in pelvic IGRT for patients with similar BMI values. Materials and Methods: The datas from 73 patients who received pelvic IGRT between March 2014 and February 2015 in Department of Radiation Oncology, Dokuz Eylul University were analyzed. Height and weight measurements of all patients were done firstly at CT-Simulation, thereafter weekly during the radiotherapy course . The patients whose weight change could result BMI group change were excluded. BMI for each patient was calculated according to standart formula "BMI: Weight in kilograms (kg) / height in meters squared (m²)" . According to National Institutes of Health classification, patients were groupped as underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9) and obese (≥ 30). According to World Health Organization criteria, patients whose ages ≥ 65 were evaluated as elderly. All of patients received pelvic volumetric modulated arc therapy with Varian Truebeam STx ® linear accelerator. For all patients before each treatment, orthogonal (anterior/posterior and lateral) kV images were acquired and these images were matched with digitally reconstructed radiographs from treatment planning system using bony anatomy. After kV-kV matching CBCT images were acquired and matched manually using soft tissue with images from treatment planning system. The requisite couch shifts were made with online procedure. Shift data for each imaging modality in X, Y, Z axes and 3D vector length (V) from these shifts were recorded. Mean shifts in each axis calculated as directional shifts with considering direction of shift and as absolute shifts without considering direction of shift. Mean shifts and standart deviations for all patients in X, Y, Z axes were derived. On mean shifts in axes; effect of BMI using " Kruskal Wallis Variance Analysis ", effect of sex using "Mann-Whitney U Test", effect of age groups using "Independent Samples t-test" were assessed statistically. Estimated CTV to PTV margins for set-up uncertainties calculated seperately for each group by using "Van Herk formula". Results: The median age of all patients was 65 (range, 36-86) and % 70 of them were male. Data was evaulated from total of 513 CBCT and 2064 kV images. Mean directional and absolute shifts in X, Y, Z, V axes with kV imaging were 0.33, 1.06, 0.13, 6.11 mm and 3.39, 2.58, 2.85, 6.11 mm while with CBCT imaging -0.24, 0.74, 0.73, 6.54 mm and 3.47, 2.90, 3.22, 6.54 mm, respectively. According to BMI groupes; mean directional and absolute shifts in X, Y, Z, V axes with kV imaging were 0.30, 1.53, -0.45, 5.54 mm and 2.82, 2.67, 2.73, 5.54 mm for BMI<25; 0.36, 0.71, 0.20, 6.16 mm and 3.57, 2.28, 2.81, 6.16 mm for BMI 25-29.9; 0.27, 1.20, 0.81, 6.82 mm and 3.78, 3.14, 3.12, 6.82 mm for BMI ≥30 while CBCT imaging -0.65, 1.23, 0.41, 6.01 mm and 3.16, 2.87, 2.82, 6.01 mm for BMI<25; 0.01, 0.46, 0.83, 6.74 mm and 3.65, 2.92, 3.34, 6.74 mm for BMI 25-29.9; -0.23, 0.68, 0.95, 6.81 mm and 3.49, 2.89, 3.51, 6.81 mm for BMI ≥30 respectively. Between BMI groups, only vectorial axis shifts in kV imaging were statistically different (p:0.039). This difference is explained by sex distrubition differences in BMI groupes and significantly higher obese group ratio in females (p:0.002). Mean absolute shifts in X, Y, Z, V axes with kV imaging for females were 3.83, 3.42, 3.17, 7.16 mm while for males 3.19, 2.21, 2.71, 5.65 mm and with CBCT imaging for females 3.74, 3.23, 4.15, 7.59 mm while for males 3.35, 2.76, 2.82, 6.08 mm, respectively. In females mean shifts in all axes were greater than males and absolute shifts in Y, Z and V axes were significantly greater (p<0.05). Directional and absolute shifts in vectorial axis with CBCT imaging were statistically different between age groups and were significantly greater for ≥ 65 age group (p:0.041). In all patients, depending on directional and absolute shift data; estimated CTV to PTV margins by using "Van Herk formula" in X, Y, Z, V axes with kV imaging were 7.21, 6.20, 6.78, 5.62 mm and 4.29, 3.99, 4.52, 5.62 mm; with CBCT imaging 7.52, 6.94, 5.93, 6.80 mm and 4.71, 5.24, 4.93, 6.80 mm respectively. Conclusion: In our study we did not find any statistically significant difference in none of the axes between directional and absolute shifts according to BMI groupes. However; because of greater shifts in females and ≥ 65 age group, more attention is needed in this group of patients' set-ups and PTV margins for these groups in planning process must evaulated more detailed. In addition, it was observed that IGRT protocol being applied in our clinic and PTV margins for uncertainties were adequate. However, in spite of modern developments in IGRT, it is still impossible to reduce set-up margins to zero by completely eliminating all geometrical errors.