Liver stereotactic body radiation therapy without fiducial or retained ethiodized oil guidance warrants greater than 5 mm planning target volumes

Abstract

Introduction

For liver stereotactic body radiation therapy (SBRT), the placement of fiducial markers or retained ethiodized oil by transarterial chemoembolisation (TACE) provides a landmark for consistent target localisation. TACE and fiducial markers are invasive procedures that harbour additional risks. We hypothesise that liver SBRT can be accurately delivered without the use of these invasive surrogate markers.

Methods

We retrospectively identified 50 consecutive patients who underwent liver SBRT with respiratory motion management to a single lesion which exhibited retained ethiodized oil per prior TACE delivery. For each SBRT fraction, two manual rigid image registrations were performed by the treating physician. One using the liver contour as a surrogate for the target and second aligning only to the radio‐opaque retained ethiodized oil of the treated lesion. The magnitude of the displacement vector between the two registration methods was used to assess the accuracy of target localisation if ethiodized oil was not present.

Results

For the 50 patients, a total of 244 analysable cone‐beam CTs (CBCTs) were included (six CBCTs excluded due to poor ethiodized oil visualisation). Respiratory motion management techniques consisted of active breathing control for 13 and abdominal compression for 37 patients. Forty‐two patients had peripheral lesions and eight had central lesions (<2 cm from left and right portal veins). The average target localisation offset between the two registration methods (i.e. liver contour vs. retained ethiodized oil alignment) for patients with a single peripheral or central liver lesion was 5.8 and 5.3 mm, respectively.

Conclusions

Across all patients, the average change in target position exceeded 5 mm for image registration methods based on the liver contour alone versus the retained ethiodized oil region. This suggests that margins greater than 5 mm may be required for respiratory motion‐managed liver SBRT treatments in patients who do not undergo prior TACE or fiducial placement.

Liver stereotactic body radiation therapy without the use of fiducial guidance is feasible as long as respiratory management is achieved to align to a liver contour surrogate and planning margins larger than 5 mm is utilised.

Introduction

Stereotactic body radiation therapy (SBRT) for primary and metastatic liver tumours has emerged as an effective treatment option in non‐surgical candidates with limited tumour burden. ¹ , ² , ³ SBRT delivers highly conformal and ablative doses of radiation therapy in a limited number of fractions with rapid dose fall‐off beyond the target. By way of the highly focused treatments, SBRT effectively spares the surrounding liver parenchyma, but in turn, also demands increased accuracy utilising daily on‐board cone‐beam CT (CBCT) with respiratory management techniques. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ To account for uncertainties in inter‐ and intrafraction tumour position, a planning target volume (PTV) is used which circumferentially expands the gross or internal target volume.

Unfortunately, most primary and metastatic liver tumours are poorly visualised on non‐contrast CBCTs used for daily treatment alignment; therefore a radio‐opaque surrogate marker, such as an implanted fiducial or retained ethiodized oil by transarterial chemoembolisation (TACE), is commonly used as a reliable and easily visualised landmark. ⁵ , ⁷ , ⁹ Percutaneous placement of liver fiducial markers may be associated with minor complications such as peri‐/post‐procedure pain, hematoma and infection. ¹⁰ TACE consists of transarterial delivery of ethiodized oil to the liver lesion and may be associated with minor and major complications such as access site injuries, vascular dissections, non‐target embolisation, post‐embolisation syndrome and hepatic failure. ¹¹

Due to its superior soft‐tissue contrast, MR‐guided liver SBRT is quickly emerging as an effective tool to obviate the invasive placement of these surrogate markers. ¹² For radiation therapy facilities without MR guidance, also there is interest in omitting these surrogate markers if a reliable surrogate anatomical contour (e.g. diaphragm, liver edge and portal vein) can be visualised with respiratory motion‐managed techniques.

To our knowledge, the accuracy of respiratory motion‐managed liver SBRT without the use of fiducial markers or retained ethiodized oil remains unknown. This study aims to assess the accuracy of this approach by comparing daily CBCT alignment using retained ethiodized oil within the treated liver lesion versus a liver contour surrogate for patients undergoing respiratory motion‐managed liver SBRT.

Methods

We retrospectively identified 50 consecutive patients who received TACE prior to five fraction liver SBRT for a single lesion between 2007 and 2021. This study was approved by the Institutional Review Board of Oregon Health & Science University. Forty‐two patients had a single peripheral liver lesion (>2 cm from left and right portal veins) and eight patients had a single central liver lesion (<2 cm from left and right portal veins). All lesions exhibited retained ethiodized oil on the simulation CT and daily CBCT. All patients underwent respiratory motion management using either end‐expiratory active breathing control (ABC, Elekta Oncology Systems, Crawley, U.K., n = 13) or abdominal compression (BodyFix Diaphragm Control, Elekta Oncology Systems, n = 37). All patient treatments and on‐board CBCTs were performed on an Elekta Versa HD linear accelerator with XVI on‐boarding CBCT imaging (Elekta Oncology Systems). All image and structure sets were exported to a commercial software (MIM MaestroTM Version 7.1.6, MIM Software Inc., Cleveland, OH) where image registration was performed between the planning CT and daily CBCTs. To approximate the capabilities of our clinical treatment couch (i.e. not able to apply pitch or roll) and for simplicity, only translations were applied during the rigid registrations. For each fraction, two manual rigid image registrations were performed by the attending radiation oncologist with robust liver SBRT experience (NN). The first used only the radio‐opaque retained ethiodized oil of the treated lesion. The second used the liver contour as a surrogate for the target, focusing on the lateral border of the liver nearest to the lesion. Internal structures such as gall bladder or portal vein were not utilised for registration purposes. To minimise confounding of the results, all TACE images for all 50 patients were registered before moving on to the liver contour registrations.

When using the liver contour to align the two images, retained ethiodized oil was blinded by using the ‘fill contour’ function in MIM for the PTV in the simulation CT and daily CBCT images. This ensured that no retained ethiodized oil was visible in either scan when aligning the two images based on the liver contour. Once each fraction was manually aligned, the ‘analyze fusion alignment’ function in MIM was used to produce the x, y and z translations in the Digital Imaging and Communications in Medicine (DICOM) standard patient coordinate system. ¹³ Next, the Euclidean norm was used to calculate the magnitude in the displacement vector (i.e. distance) between the two registration methods in order to assess the accuracy of target localisation if ethiodized oil was not present. This provided a single, positive absolute value for the displacement vector of each fraction, which was averaged across all five fractions for each patient and then across all patients.

Results

A total of 50 patients were included in the study, amounting to 244 analysed CBCTs. Two patients had two CBCT excluded and two patients had one CBCT excluded due to poor ethiodized oil visualisation. The average gross or internal tumour volume was 35.8 mm³ (range: 0.6–225.3 mm³). When aligning to the liver contour as compared to retained ethiodized oil, the average target localisation offset was 5.8 mm (SD = 2.47) and 5.3 mm (SD = 1.72) for peripheral and central lesions, respectively. There was no significant difference in the magnitude of the localisation offset between the two groups based on central versus peripheral lesion location (P > 0.05).

Figure 1 displays individual patient treatment offsets. Regardless of lesion location, 26/50 (52%) of patients were found to have an average target localisation offset greater than 5 mm. Additionally, 43/50 patients showed a target localisation offset greater than 5 mm in at least one of the five fractions. Only one patient in the ABC group showed a target localisation offset of less than 5 mm for all five fractions, with all other patients in this group having at least two fractions with greater than 5 mm offset. The maximum and minimum target localisation offset was 17.5 and 0.2 mm, respectively.

Figure 1.

Box and whisker plots of offsets for all SBRT fractions for each patient visualised in box and whisker format. The diamond dot represents the mean offset for each individual patient. The dashed blue line demarcates patients treated with abdominal compression (above blue line) and active breathing control (below blue line).

All fractions were pooled to analyse the distribution of target localisation offset between the two registration methods for which 86/244 (35.2%) of the fractions had a target localisation offset greater than 6 mm. When comparing to our standard 5 mm isotropic margin, we found that 114/244 (46.7%) of fractions exhibited greater than 5 mm localisation offset. Furthermore, 59 fractions (24.2%) had a vector change between 5 and 7 mm, 24 (9.8%) between 7 and 10 mm, 14 (5.7%) between 10 and 12.5 mm, 11 (4.5%) between 12.5 and 15 mm and six (2.5%) between 15 and 17.5 mm.

We then analysed the target offset localisation in each anatomical direction between the two registration methods (Left–Right, Superior–Inferior, Anterior–Posterior). We found an average offset of 2.83 mm in the Left–Right, 3.38 mm in the Anterior–Posterior and 4.15 mm in the Superior–Inferior directions. There was no significant difference in offset localisation for each anatomical direction when accounting for lesion location. When comparing motion‐management technique (ABC vs. abdominal compression), there was no significant difference in the overall target offset localisation or offset localisation in each anatomical direction (P > 0.05).

Discussion

To our knowledge, this is the first study to evaluate the accuracy of respiratory motion‐managed liver SBRT using a liver contour surrogate rather than radio‐opaque markers for image guidance. In this study evaluating 244 SBRT fractions for 50 patients with retained ethiodized oil, we found that the average target localization offset was greater than 5 mm if retained ethiodized oil was not visualised and image guidance was based only on the liver contour. These findings were independent of respiratory management technique (ABC vs. abdominal compression) and tumour location (central vs. peripheral). These findings are especially important in patients with fiducial failure (i.e. migration), poorly visualised retained ethiodized oil on inter‐fractional CBCT, or for patients who are not candidates for invasive surrogate marker procedures (percutaneous fiducial placement or TACE). Furthermore, eliminating the need for such procedures would decrease patient healthcare costs and any associated risks or discomfort from such procedures. In our practice, if a patient is unable to tolerate respiratory motion‐management techniques, a radio‐opaque surrogate is compulsory.

A 2016 study of six patients receiving free‐breathing liver SBRT with implanted fiducial markers assessed the inter‐fractional motion using fiducial markers as compared to the organ motion of the liver and the relative position within the liver volume. ¹⁴ These results are largely consistent with our study, identifying a greater than 5 mm offset. However, their methodology of normal anatomy alignment surrogacy utilising free‐breathing CT and CBCT may introduce additional alignment errors as the liver contour is largely distorted given the respiratory blur. A potential hypothesis regarding the most likely source of discrepancy in our study between using TACE versus liver surrogate alignment is related to liver anatomy distortion in inter‐fractional imaging, even with respiratory management techniques. The radio‐opaque TACE is easily visible and less affected by motion blur compared with the liver edge which is less detailed and more difficult to align to the planning CT and subject to inter‐ and intra‐user variability.

Conclusion

This study showed the accuracy of using a liver contour surrogate for daily SBRT image guidance for respiratory‐managed patients with an isolated liver lesion. These findings suggest that greater than 5 mm isotropic PTV margins should be applied when delivering respiratory‐managed liver SBRT without the assistance of fiducial markers or retained ethiodized oil.

Funding Information

Support for this project was funded by the generosity of Robin and Roald Pettersen.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.