Evaluation of the intra‐ and interfractional tumor motion and variability by fiducial‐based real‐time tracking in liver stereotactic body radiation therapy

Abstract Purpose Tumor motion amplitude varies during treatment. The purpose of the study was to evaluate the intra‐ and interfraction tumor motion and variability in patients with liver cancer treated with fiducial‐based real‐time tracking stereotactic body radiotherapy (SBRT). Methods Fourteen liver patients were treated with SBRT using a CyberKnife. Two to four fiducial markers implanted near the tumor were used for real‐time monitoring using the Synchrony system. The tumor motion information during treatment was extracted from the log files recorded by the Synchrony system. Logfile‐based amplitudes in the superior–posterior (SI), left–right (LR) and anterior–posterior (AP) directions were compared to the 4DCT‐based amplitudes. The intra‐ and interfraction amplitude variations and the incidence of baseline shifts were analyzed for 66 fractions administered to 14 patients. Results The median (range) logfile‐based liver motion amplitudes for all patients were 11.9 (5.1–17.3) mm, 1.3 (0.4–4) mm and 3.8 (0.9–7.7) mm in the SI, LR and AP directions, respectively. Compared with the logfile‐based amplitude, the 4DCT‐based amplitude was underestimated (P < 0.05). The median (range) intra‐ and interfraction liver motion amplitude variations were 4.3 (1.6–6.0) mm (SI), 0.5 (0.2–2.2) mm(LR) and 1.5 (0.3–3.3) mm (AP) and 1.7 (0.5–4.6) mm (SI), 0.3 (0.1–3.0) mm (LR) and 0.7 (0.3–2.7) mm (AP), respectively. Baseline shifts exceeding 2 mm, 3 mm and 5 mm were observed in 27.3%, 7.6% and 3% of the measurements, respectively, within 10 min, and in 66.7%, 38.1% and 19%, respectively, within 30 min for the square root of the sum of the squares of the distances in the SI, LR and AP directions (3D). The tumor motion amplitude was found to be correlated with the baseline shift. Conclusions Most patients showed significant intra‐ and interfraction liver motion amplitude variations over the entire course of radiation. More caution is needed for patients with large tumor motion amplitudes.


| INTRODUCTION
Radiation-induced liver disease (RILD) has the potential to lead to liver failure, and the risk of RILD is correlated with the mean dose to the liver. 1 With the use of stereotactic body radiation therapy (SBRT) for the treatment of liver tumors, it became feasible to achieve high rates of tumor control while minimizing the irradiation of the surrounding uninvolved liver, thereby reducing the risk of RILD. 2,3 SBRT needs to precisely deliver a highly conformal dose to the target in fewer fractions. Hence, tumor motion evaluation and its use in treatment planning are important for liver SBRT.
There are several methods for determining liver tumor motion in a simulation process, including four-dimensional computed tomography (4DCT), 4 inhale/exhale breath-hold CT and cine magnetic resonance imaging (MRI). 5 Due to the density difference between a tumor and the surrounding normal tissue, liver tumors are generally only visible in contrast-enhanced CT scans. A synchronized contrast injection method during 4DCT simulation has been reported to account for the problem. 4,6 However, because intravenous contrast is not routinely used during pretreatment, it is difficult to recognize the tumor in cone beam CT (CBCT) images, and accurate patient alignment becomes a challenge in liver SBRT. Fiducials implanted near the liver tumor have been shown to be effective surrogates and are highly recommended. 2 Bertholet et al. compared four CBCT-based setup strategies for liver tumor. They concluded that marker-based setup was substantially better than bony-anatomy setup. 7 Tumor motion patterns may change during the treatment, with either intra-or interfraction changes. Case et al. investigated intraand interfractional tumor motion variability using CBCT 8 or respiratory-correlated CBCT. 9 However, CBCT or respiratory-correlated CBCT only represents the tumor motion over multiple respiratory cycles during image acquisition. Real-time data from the treatment can provide information with more detail about the tumor motion.
Using real-time data, Malinowski et al. 10

2.B | Simulation, prescription and treatment planning
All patients were implanted with 2-4 fiducials near the tumor under CT guidance. 11 Simulation was typically administered 7-10 days after the implantation, which can minimize the risk of marker migration. The patients were supine with their arms along their sides and were immobilized with a customized vacuum body mold. Exhalation breath-hold contrast-enhanced planning CT was used for treatment planning. An additional 4DCT was used to evaluate tumor motion in all patients in case dynamic tracking was not possible during treatment.
The gross target volume (GTV) was expanded by a 5-mm radial and 8-mm craniocaudal margin for the planning target volume (PTV).
The radiation dose was prescribed according to the isodose surface covering the PTV, typically 75% to 85% of the maximum PTV dose. SBRT was planned and delivered using the CyberKnife under free breathing.

2.C | Synchrony respiratory tracking system
The primary concept of the tracking treatment using the Synchrony system is to build a correlation model between the internal tumor location and the external marker position. 12 The internal tumor location was determined using the centroid of fiducials, which were implanted near the tumor before simulation, and the external optical marker was attached to a vest worn by the patient during the treatment. The correlation model is updated with each new X ray acquisition (typically every 60 s), thus accounting for the change of motion pattern during treatment. More detailed information can be found in the reference manual. 13 The accuracy of the correlation model has been evaluated in a previous publication. 14 The mean correlation model errors were less than 0.3 mm in their study, indicating that use of the correlation model to predict the tumor position is reliable. The internal tumor position throughout the treatment was estimated using the correlation model and was stored in log files (Modeler.log).

2.D | Tumor motion analysis
Tumor motion amplitudes, baseline drifts and intra-and interfraction amplitude variability during treatment were evaluated. The intrafraction amplitude variability was defined as the standard deviation (SD) of the peak-to-peak distance (the peak-to-peak distance was defined as the distance of each inhale peak to the next exhale peak) during one treatment fraction.      Table 2 presents the incidences of baseline shifts exceeding 2 mm, 3 mm and 5 mm in the three time blocks for the SI, LR, AP, and 3D directions. An apparent time trend increase in the incidence of baseline shift can be observed. Interestingly, the incidence of baseline shifts >2 mm in the SI direction for the first fraction was 35.7%, which is larger than the incidences of the 2nd and 3rd fractions, which were 21.4% and 14%, respectively.

| DISCUSSION
The intra-and interfraction amplitudes clearly vary. Recently, realtime observation of tumor movement during treatment is becoming more common, 16,17 and new apparatuses integrating this function are gradually being used by clinical facilities. 18,19 CyberKnife is the earliest commercial device for real-time monitoring and tracking.
Logfile data from the Synchrony system record the tumor motion during the treatment, and the amplitudes determined by the logfile data may be closer to reality than 4DCT or CBCT methods. Eccles et al. 20  T A B L E 2 Incidence of baseline shifts exceeding 2 mm, 3 mm, and 5 mm for the SI, LR, AP, and 3D directions.

CONF ILICT OF INTEREST
The authors declared that they have no conflicts of interest to this work.