Quantitative Analysis of Marker Based Treatment Setup Uncertainty and its Correlation to Treatment Response for HDR Endorectal Brachytherapy

HDR endorectal brachytherapy (EBT) is an emerging alternative neoadjuvant treatment option for locally advanced rectal cancer. EBT is given in 4 fractions with daily dose of 6.5 Gy in four consecutive days. Reproducing planned applicator position relative to tumor location in each fraction is crucial for replicating dose distribution in this hypo-fractionated treatment regime. Radio-opaque markers are inserted into the patient prior to CT/MRI simulation to guide tumor delineation and treatment setup. In this study, we analyzed marker-based treatment setup uncertainty and its correlation to treatment response. Fifteen patients treated with EBT at our institution were analyzed. Two to four radio-opaque fiducial markers were inserted into or adjacent to tumor within a week before simulation. An 8 channel intracavitary mold applicator was used during CT/MRI simulation and HDR treatment. Three radio-opaque dummy catheters were inserted into mold applicator channels at 12, 6 and 9 o'clock positions. These dummy catheters serve as reference points for applicator longitudional position and rotation verification. AP and Lateral kV setup images acquired before each treatment were compared to scout images taken at the time of simulation. The insertion depth of the mold applicator was verified by comparing the locations of fiducial markers relative to dummy catheters in the superior/inferior (S/I) direction. The rotation of the applicator was adjusted according to orientation of the three dummy catheters in the orthogonal images. After 6-8 weeks recovery, patients underwent surgical resection. The pathology reports from surgical samples were used to divide the patients into two groups: patients with pathological complete response (pCR group) and patients with residual tumor (non-pCR group). For analysis of setup uncertainty, the fiducial markers and a fixed reference point on the dummy catheter were identified in both scout images and daily setup kV images. The 3D locations of fiducial markers and catheter reference point were reconstructed from orthogonal images. The 3D distance of marker to catheter reference point (dM_C) and marker-to-marker distance (dM_M) were calculated. The difference between sim dM_C and daily treatment dM_C was used to quantify daily applicator setup accuracy. The difference between sim dM_M and daily treatment dM_M indicated the stability of marker locations and possible tumor deformation and displacement by gas in rectum. The correlation of setup uncertainty and treatment response was investigated. The pCR group consisted of 6 patients and the non-pCR group 9 patients. The PTV dose coverage in treatment plans was similar in both groups, with 98.6±1.8% in pCR group and 98.4±1.7% in non-pCR group. The average sim to treatment difference of dM_C (±SD) was 8±3 mm in pCR group and 9±4 mm in non-pCR group, respectively. The average difference in daily S/I distance of marker to dummy catheter was within 4 mm for both groups, indicating the setup error in S/I direction is consistent across all patients and all fractions within a few millimeter accuracy. There is no significant correlation between treatment response and setup uncertainty observed in marker to applicator distance (p=0.59). The average sim to treatment difference of dM_M (±SD) was 6±1 mm in pCR group and 12±6 mm in non-pCR group, respectively. The daily difference of dM_M is significant between the two groups (p=0.02). This indicates that tumor deformation or displacement could affect treatment response due to reduced dosimetry coverage. Accuracy of marker-based treatment setup for endorectal brachytherapy was analyzed by evaluating the positions of the markers relative to a fixed reference point on dummy catheter and relative to each other from two orthogonal setup images. Consistent daily marker to applicator setup was demonstrated. Smaller marker-to-marker distance variation between simulation and each treatment correlated with better treatment response. Tumor deformation and displacement could result in marker shifts. Three dimensional volumetric image guidance and adaptive treatment are potential methods to improve dose coverage and contribute to better treatment outcome.