The joint use of 99mTc-MAA-SPECT/CT and cone-beam CT optimizes radioembolization planning

Purpose To determine which imaging method used during radioembolization (RE) work-up: contrast-enhanced computed tomography (CECT), 99mTc-MAA-SPECT/CT or cone beam-CT (CBCT), more accurately predicts the final target volume (TgV) as well as the influence that each modality has in the dosimetric calculation. Methods TgVs from 99mTc-MAA-SPECT/CT, CECT and CBCT were consecutively obtained in 24 patients treated with RE and compared with 90Y PET/CT TgV. Using the TgVs estimated by each imaging modality and a fictitious activity of 1 GBq, the corresponding absorbed doses by tumor and non-tumoral parenchyma were calculated for each patient. The absorbed doses for each modality were compared with the ones obtained using 90Y PET/CT TgV. Results 99mTc-MAA-SPECT/CT predicted 90Y PET/CT TgV better than CBCT or CECT, even for selective or superselective administrations. Likewise, 99mTc-MAA-SPECT/CT showed dosimetric values more similar to those obtained with 90Y PET/CT. Nevertheless, CBCT provided essential information for RE planning, such as ensuring the total coverage of the tumor and, in cases with more than one feeding artery, splitting the activity according to the volume of tumor perfused by each artery. Conclusion The joint use of 99mTc-MAA-SPECT/CT and CBCT optimizes dosimetric planning for RE procedures, enabling a more accurate personalized approach.


Background
In radioembolization (RE), the definition of the target volume (TgV)-including tumoral and non-tumoral areas-that will receive the treatment, is decisive in many dosimetric aspects: for single-compartment medical internal radiation dose (MIRD) model because it assumes a uniform activity distribution within the TgV; for modified body surface area (mBSA) or partition model methods, because TgV is incorporated in the formulas [1]; for 3D voxel-dosimetry, because the dosimetric calculations derive precisely from the predicted TgV.
Current practice of assessing TgV is based on contrastenhanced computed tomography (CECT) or magnetic resonance (MR), which reflect the standard anatomical venous segmentation as defined by Couinaud. However, this approach may be inaccurate in different clinical settings, such as in selective arterial (segmental and subsegmental) administrations, in central tumors without a pure lobar or segmental distribution, or in patients with anatomical variations-whether innate or related to tumorigenesis [2], among others.

Open Access
*Correspondence: mrodriguez@unav.es Rodríguez-Fraile et al. EJNMMI Res (2021) 11:23 Other imaging methods performed in the routine RE work-up have also been used to assess volumetric analysis. These include 99m Tc-macroaggregated albumin ( 99m Tc-MAA) SPECT/CT ( 99m Tc-MAA-SPECT/CT) or C-arm cone-beam CT (CBCT). The use of the 99m Tc-MAA-SPECT/CT as a method to calculate TgV was first described by Garin et al. [3], demonstrating its accuracy in hepatocellular carcinoma (HCC) [4] and in cholangiocarcinoma (CC) [5]. Likewise, CBCT has been proposed as a useful method for defining the TgV in total or lobar administrations [6]- [8]. Rangraz et al. [7] demonstrated that using CBCT-instead of CECT-results in a difference in volumetric parameters However, none of the abovementioned studies has been performed in segmental or subsegmental administrations (treatment via direct tumor-feeding vessel) [9], where the evaluation of the TgV in the CBCT without clear anatomical limits may be more challenging.
Once the treatment is administered, both bremsstrahlung SPECT/CT (BS) or Yttrium-90 ( 90 Y) PET/CT are generally used to verify the final distribution of the microspheres. Nevertheless, 90 Y PET/CT has been shown to be superior to BS for the assessment of target activity [10], helping in the accurate quantification of the total delivered activity [11] and to perform dose estimation [12,13]. Moreover, 90 Y PET/CT-based dosimetry after RE with resin microspheres has been shown to predict outcome in patients with liver metastases from colorectal cancer (CC) [14]. Hence, 90 Y PET/CT is a robust and reliable tool for the estimation of the 90Y-microspheres deposition.
The primary aim of this study was to determine which of the imaging methods available at the time of the initial evaluation of RE (CECT, 99m Tc-MAA-SPECT/CT or CBCT) predicts more accurately the TgV, having the 90 Y PET/CT final TgV as the reference parameter. A secondary objective was to evaluate the influence that the differences in the estimated TgV for each technique has in the dosimetric calculation.
Finally, since in order to reduce the risk of RE-induced liver disease (REILD) [15], it is highly recommended to minimize the irradiation of the non-tumoral tissue [16,17], RE administrations are becoming increasingly selective. In this sense, the contribution of 99m Tc-MAA-SPECT/CT and CBCT to the standard images (CECT) for a better dosimetric planning, especially in segmental or subsegmental approaches, was also analyzed [15].

Same-day RE protocol Patients
All patients treated with resin 90 Y-microspheres (SIRspheres ® , SIRTex Medical Limited) in our center from October 2018 to April 2019 were consecutively studied.
After being considered as a candidate for RE by the hepatobiliary multidisciplinary team (MDT), a same-day planning and treatment was performed in all cases. Both the aim and the approach of the treatment (total, lobar, segmental or subsegmental) were always defined by the MDT.

Pre-treatment angiography
After the oral administration of 600 mg sodium perchlorate to block free 99m Tc-pertechnetate uptake by stomach [18], a 4F catheter was advanced via common femoral artery, and a selective angiography of both the superior mesenteric artery and the celiac trunk was performed. Coil embolization was performed, if necessary, to prevent the delivery of particles to the non-target tissue. The interventional radiologist (IR) performed in all cases the angiographic simulation to cover the entire tumoral tissue while preserving as much volume of non-tumoral parenchyma as possible. When multiple extra or intrahepatic vessels feeding the TgV were detected, a selective catheterization of each one was carried out. Thus, sameday flow redistribution was performed, when deemed necessary, to treat the complete tumoral area reducing the number of injection points [19,20].
Diagnostic angiography and endovascular intervention were performed using the robotic digital subtraction angiography system (Artis Zeego Q, VE 40 A, Siemens Healthineers, Forchheim, Germany). CBCT was routinely performed immediately after the angiography to determine the best arterial access. It consisted of an unenhanced rotation (mask run) and contrast-enhanced rotations. Rotation time was 4 s. Parameters of CT acquisition were: tube voltage, 90 kV; 248 frames; 0.8º per frame; pixel size, 616 µm; acquisition time, 12 s. Once the selected arterial access was defined, 111-185 MBq 99m Tc-MAA was injected to mimic the future distribution of 90 Y-microspheres. TgV from CBCT image was delineated by a technician in radiology and supervised by the IR using a volume calculation software (Syngo DynaCT, Siemens Healthineers). The reconstruction used was as follows: voxel size 0.5 mm3 (full); slice matrix, 512 × 512; kernel type, HU (W 1400; C 550); 0.5 mm slice thickness; image characteristics, normal; reconstruction model, Nat Fill; viewing preset, Syngo Dyna CT. A 0.5 mm slice thickness was employed. Images were windowed to emphasize liver parenchyma (W 1400; C 550). Regions of interest (ROIs) were manually drawn every two axial images and then interpolated. The obtained volume was reviewed in sagittal and coronal dimensions and corrected if needed. For target volume delineation, MIP (maximum intensity projection) datasets were also employed. Edge enhancement was included in the volume determination. Tumor volume was more clearly visualized with MIP representation (6 mm).
The TgVs of CECT studies were obtained using Syngo. via software (Siemens Heathineers). CECT TgV was defined on cross-sectional images by a radiologists, using a fixed slice thickness (3 mm). The volumes of each slice were summed, independent of anatomical landmarks. Region of interest (ROI) were manually drawn in each slice involving the target/tumor volume. A unified window level (W, 300 HU) and window width (C, 40 HU) was determined. In all cases, tumoral volumes were also assessed by CECT images.
TgV in 99m Tc-MAA-SPECT/CT was defined using the multimodality reading software Syngo.via for MI (Siemens Healthineers). Using the "VOI + isocontour" tool, a volume of interest (VOI) in the target liver (including tumor and non-tumor) was drawn by a nuclear medicine (NM) physician and by means of the isocontour definition, the "molecular tumor volume or MTV" in milliliters (ml) obtained was used as the 99m Tc-MAA-SPECT/CT TgV. The isocontour threshold was visually adjusted to include the 99m Tc-MAA uptake volume into the VOI (Fig. 1).

Dosimetric calculation
The activities were calculated considering the absorbed doses (Gy) by tumor and healthy parenchyma as proposed by Gil-Alzugaray [21]. These absorbed doses were defined by formula [1] using the parameters obtained in the 99m Tc-MAA-SPECT/CT and CECT studies. In general terms, in cirrhotic patients with a predicted spared volume less than 40%, the activity was estimated to produce a safe absorbed dose by the non-tumoral compartment (≤ 40 Gy). In contrast, when the predicted TgV was small (< 60%) and the patient had a preserved liver function, a tumoricidal absorbed dose (> 100 Gy) was estimated, irrespective of the dose delivered to the nontumoral tissue [21].
TNR values used for dosimetric calculations derived in all cases from 99m Tc-MAA-SPECT/CT.

Treatment administration
More than four hours after 99m Tc-MAA, resin 90 Y-microspheres were administered during a new angiographic procedure. When 99m Tc-MAA-SPECT/CT showed an adequate distribution in the tumoral area, the vascular introducer was left in the same place for Fig. 1 a Contrast-enhanced computed tomography (CECT) image in a patient with HCC located between segments IV and VIII. b Volumetric assessment of the target volume in SPECT/CT fusion images after the injection of 99m Tc-MAA through IV and VIII segments arteries. The volume was obtained using a "volume of interest and isocontour" tool, drawn in purple. c C-arm cone-beam CT (CBCT) showing contrast uptake in the tumoral lesion with no perfusion in non-tumoral parenchyma. d Volumetric assessment of the final target volume in the 90 Y PET/CT fusion images both procedures. In cases in which the arterial access selected by the IR did not optimally reach the tumoral area in the 99m Tc-MAA-SPECT/CT images, the angiography images were re-evaluated. In these cases, if a better access was identified, the site of injection between 99m Tc-MAA and RE differed. Although it is advisable to repeat 99m Tc-MAA evaluation when changes are made, it can be avoided if the modifications are minimal or do not have an impact on the safety of the treatment.

Post-treatment PET/CT
The morning after the RE treatment, a 90 Y PET/CT scan centered on the liver region (two beds, 10 min/ bed) was performed using a Biograph mCT-TrueV (Siemens Medical Solutions), which combines a 64-slice CT with a 21.8 cm field of view time-of-flight PET scanner comprised by lutetium-based crystals (LSO) detector blocks [22]. The reconstruction protocol used (one iteration, 21 subsets, a 6-mm Gaussian filter and a 200 × 200 matrix) was previously optimized by Martí-Climent et al. [11].
Final TgV in 90 Y PET/CT was defined using the multimodality reading software Syngo.via for MI (Siemens Healthneers) as previously described for 99m Tc-MAA-SPECT/CT.

Comparison of the TgV and the predicted dosimetry for each image method
TgVs calculated by 99m Tc-MAA-SPECT/CT, CECT and CBCT were compared with the TgV obtained in the 90 Y PET/CT study.
In order to evaluate the influence that the differences that TgV for each technique has in the dosimetry, the absorbed doses calculated using the different TgV imaging modalities were compared using in all cases a fictitious prescribed activity of 1 GBq. The use of a fixed amount of activity makes it easier to appreciate the impact that the use of each TgV would have had on the absorbed doses. The TNR and tumor volumes were the real ones calculated for each patient. The absorbed doses for each modality were compared with the ones obtained using 90 Y PET/CT TgV. Considering the latter as the actual ones [3,4], the percentage of change between absorbed doses ([(Gy for 90 Y PET/ CT TgV -Gy for TgV modality)/Gy for TgV modality] × 100%), was calculated for each patient. A positive value indicates a percentage increase. (Gy in 90 Y PET/CT is higher than the predicted using the TgV for each modality.) A negative value indicates a percentage decrease (Gy in 90 Y PET/CT is lower than predicted using the TgV for each modality).

Contribution of 99m Tc-MAA-SPECT/CT and CBCT for a better dosimetric RE planning
The contribution of CBCT and 99m Tc-MAA-SPECT/CT to the standard images (CECT) for a more personalized RE planning were evaluated by an IR and a NM physician (both with more than 15 years of experience in RE). The additional information provided by both techniques was especially focused on those clinical settings in which CECT may present some limitations for dosimetric calculations, such as selective or superselective approaches, several tumoral feeding arteries, flow redistribution, etc.

Statistical analysis
To assess agreement between studies, the Lin Concordance Correlation Coefficient (CCC) and its 95% Confidence Interval (95% CI) were used. To define which study best predicts the final TgV, the determination coefficient (R 2 ) from the regression model was utilized. For each modality, the difference changes of absorbed doses between treatment approaches were analyzed with the t students test. A p value less than 0.05 was used to determine the presence of a significant difference. The data were analyzed using Statistical Package for Social Science (SPSS) software version 22.
For cases with two or more supply arteries, 99m Tc-MAA activity was divided into 25%, 50% or 75% at the IR discretion, depending on the findings obtained during mapping arteriography. CBCT was performed in 23/24 patients (in 1/24 was not possible due to lack of patient collaboration). Four CBCT studies were excluded from the TgV analysis. An insufficient CBCT technique did not allow the correct assessment of TgV.
In all cases CBCT volumetry was obtained after 99m Tc-MAA injection, so the calculation was not used to split 99m Tc-MAA activity. CECT and 99m Tc-MAA-SPECT/CT TgVs were calculated in all patients during RE planning, not knowing the final distribution of 90 Y-microspheres on 90 Y PET/CT (Table 1).
Mean HPS was 6.9% (± 3.4). Mean TNR was 2.6 (± 1.5). In the majority of cases TNR was an average of the uptake in all tumors, while in 4 cases it was calculated for each tumor as described elsewhere [23]. To simplify the results, a mean of the TNR of all tumors was calculated in these 4 patients.
In 17/24 patients (71%), RE injection were performed through segmental or subsegmental arteries. Seven patients were treated with just one infusion of 90 Y-microspheres, four were lobar (right) and three were segmental. Fifteen patients were treated from two different arteries, three of them received a whole-liver treatment (right and left hepatic arteries), two were treated through a lobar and a segmental branch, eight through two different segmental/subsegmental branches and two through the inferior phrenic artery in association with a lobar (1) or a segmental (1) artery. Finally, two patients required three different infusions through segmental and accessory arteries (Table 1).

Contribution of 99m Tc-MAA-SPECT/CT and CBCT for a better dosimetric RE planning
The information provided by 99m Tc-MAA-SPECT/CT was determinant in 17 patients (71%), due to its capability: (a) to predict 90Y-microspheres distribution in patients with segmental or subsegmental treatments (n = 8); (b) to confirm the TgV after flow redistribution (n = 7) and (c) to detect tumoral areas not covered with the selected arterial access (n = 2).
CBCT helped to define the percentage of tumor volume perfused by each artery in 10 patients (42%) in whom the tumor was fed by more than one artery. This volumetric information was used to split the activity of 90 Y-microspheres accordingly (Fig. 4). Moreover, CBCT allowed to ensure the tumor coverage in six patients (25%) and to rule out the presence of microsatellite lesions in one patient (4%). Globally, CBCT information was used for a personalized and more accurate planning in 17/24 patients (71%).

Discussion
The results of this study show that 99m Tc-MAA-SPECT/ CT predicts more accurately the final TgV-as defined in the 90 Y PET/CT-than CBCT or even than the conventional method (CECT). This superiority is even more notable in segmental and subsegmental administrations. The differences obtained in the TgV for each method would have had a significant impact on the dosimetric calculation. Additionally, 99m Tc-MAA-SPECT/CT and CBCT information were determinant for RE planning in a significant proportion of patients. Our study therefore suggests that the joint use of both techniques optimizes dosimetry planning for RE procedures.
In this study, TgV obtained with different imaging modalities using the 90 Y PET/CT as the method of validation, which has proven to be the most accurate technique for determining the final distribution of the microspheres [11,12].
Since all consecutive patients treated with RE in our center in a period of time were studied, not only the optimal situations (identical 99m Tc-MAA and RE administrations) were included. As it occurs in daily practice, patients with intended or unintended modifications between both procedures were considered. Furthermore, flow redistribution was performed in 41.7% of the patients. Despite all these complex circumstances, the 99m Tc-MAA-SPECT/CT TgV reached maximal concordance with 90 Y PET/CT final TgV (CCC = 0.85). As expected, when the four patients with changes between Fig. 4 Same patient as Fig. 1. a Contrast-enhanced computed tomography (CECT) image: HCC located between segments IV and VIII. b 99m Tc-MAA SPECT/CT fusion image shows low uptake in the lateral part of the tumoral nodule. 99m Tc-MAA activity was split in two doses of 50% each by IR decision, based on liver and tumoral volumes. c, c1 and c2 C-arm cone-Beam CT (CBCT) volumetric assessment of the tumoral territory perfused by each artery. VIII segments artery (in green) fed only 32% of the tumoral volume while IV segment artery fed most of it (in orange). d 90 Y PET/CT fusion image after splitting the activity according to CBCT volumes shows the adequate distribution of the microspheres throughout the lesion Tc-MAA-SPECT/CT increased its concordance value (CCC = 0.97). The high concordance found between 99m Tc-MAA-SPECT/CT and 90 Y PET/CT volumetry supports the use of 99m Tc-MAA-SPECT/CT as the most reliable available tool for predicting final TgV. Although 3D voxel-dosimetry is currently recommended, the methodology followed here sustains its reliability even using a simple tool available by most groups performing RE.
In 71% of patients, RE administrations were performed through segmental or subsegmental arteries. When only these selective administrations were analyzed, 99m Tc-MAA-SPECT/CT showed to be superior to CECT (CCC = 0.5) and to CBCT (CCC = 0.67) for predicting TgV, with a substantial concordance (CCC = 0.71) with 90 Y PET/CT TgV. These findings demonstrate that 99m Tc-MAA-SPECT/CT is also an effective tool for defining the TgV in segmental or subsegmental administrations, where CECT has some limitations [2,8]. Because of the benefit to patient outcome of the parenchyma-sparing RE administrations [16,17], selective administrations are recommended when possible. Hitherto, CECT volumes have been traditionally used for these selective administrations. However, and according to the results obtained in this study, CECT volumes poorly predict the final TgV obtained with 90 Y PET/CT.
As for the reproducibility of 99m Tc-MAA-SPECT/CT and 90 Y PET/CT isocontour definition, our results are comparable to previous studies: the isocontour mode was 3% (range 1-9%) for them both. This is in accordance with Richetta et al. [24] that using a mean threshold of 3% (range of 2-4%) found a good dose agreement between 99m Tc-MAA-SPECT/CT and 90 Y PET/CT. Martí-Climent et al. [11] in a series of 10 patients found that 5% was the isocontour level that provided the lowest relative difference between reconstructed activity and activity delivered to the whole-liver ((10.2 ± 14.7)%). Moreover, Garin et al. [3], in a phantom study using an adaptative thresholding method based on SPECT/CT images, as used in our study, encountered a mean error of < 2.5% for volumes larger than 16 ml. Therefore, it seems that the visual adjustment of the isocontour level for the definition of the total treated liver is feasible and there are not significant variations on the value used between groups. It should be noted that all TgV 99m Tc-MAA-SPECT/CT were prospectively defined, not knowing the final microsphere distribution.
CBCT showed only a moderate concordance with 90 Y PET/CT final TgV, lower than for 99m Tc-MAA-SPECT/ CT. Although both are functional modalities, CBCT depends on the lapse of time between injection of the contrast agent and the image acquisition and also on the speed and volume of the injection. In some of our cases, this contrast volume could be insufficient to precisely demarcate the limits of the TgV. This was sometimes done deliberately to avoid contrast reflux to non-target areas or because CBCT was performed to detect other tumoral nodules-and not with volumetric purposes. Therefore, more studies are needed to discern whether CBCT moderate accuracy encountered was due to a limited capacity to define TgV in segmental/subsegmental administrations without clear anatomical limits or due to technical issues. Nevertheless, CBCT has other advantages not explored in this study such as its capability to detect extrahepatic arterial supply [25], feeding arteries not identified by CECT [26] or the exclusion of necrotic areas with no contrast uptake for a more precise volumetric assessment of the tumor.
Using the TgVs estimated by each imaging modality and a fictional administered activity of 1 GBq, the corresponding absorbed doses by tumor and non-tumor were calculated for each patient. Consistent with the results obtained for volumetry, 99m Tc-MAA-SPECT/CT showed lower differences with the values obtained with 90 Y PET/ CT TgV, than the rest of modalities. Therefore, and as described before [3,7,8], the use of 99m Tc-MAA-SPECT/ CT volumes reduces the risk of underdosing. Even so, using 90 Y PET/CT as the method to define the actual TgV, the calculated Gy in the tumor were 18% lower than those predicted using 99m Tc-MAA-SPECT/CT TgV (median of − 14 Gy). This difference was almost half for absorbed doses by non-tumor liver (median of − 5 Gy). These results are in accordance with other studies founding that 99m Tc-MAA-SPECT/CT tends to overestimate posttherapy dosimetry in tumor, being more accurate for the non-tumor liver dosimetric assessment [27,28]. Regarding the influence that the treatment approach could have in the differences in dosimetry, CBCT showed lower differences for lobar or total treatments than for selective or superselective approaches. As mentioned above, probably the delimitation of TgV in more selective administrations, without clear anatomical boundaries, can be a limitation of this technique. Therefore, more studies are needed to elucidate this.
Another important aspect of the study lies on the added utility that each modality has in RE planning: -the information obtained from 99m Tc-MAA-SPECT/ CT was determinant in 71% of the patients due to its capability to define and confirm the TgV in segmental and subsegmental treatments or after flow redistribution; it also helped to detect tumoral areas not receiving 99m Tc-MAA with the selected arterial access.
-CBCT was especially useful in 29% of the patients, ensuring the total coverage of the tumor and ruling out the presence of microsatellite lesions that would have changed the selected arterial access. Moreover, thanks to CBCT information it was possible to split the activity according to the volume of tumor perfused by each feeding artery in up to 42% of patients. This approach, which as far as we know has not been published before, enables a better coverage of the microspheres in the target area.
The strengths of this study are worth highlighting. First, all 99m Tc-MAA-SPECT/CT TgV were obtained blindly during RE work-up, not knowing the final 90Y-microsphere distribution. Second, the same IR performed RE evaluation and treatment in one day in all patients. This reduces the risk of undesired changes in catheter position and therefore the agreement between the distribution of 99m Tc-MAA and the 90 Y-microspheres is less subject to non-measurable errors.
This study has also some limitations. It is a singlecenter study involving a relatively small number of patients. Shallow breathing was allowed during SPECT/ CT and PET/CT acquisition and breathing motion was not corrected. However, as Bastiaannet et al. [29] described, healthy liver parenchyma suffered only marginally from breathing and collimator effects due to the larger volume, being individual tumors the most affected. Despite these limitations, the results herein presented are promising and can help to plan a more precise and personalized treatment with those imaging methods routinely used during RE work-up.

Conclusion 99m
Tc-MAA-SPECT/CT has shown to be a reliable tool to predict the liver volume that will be treated during RE. Its concordance with the TgV obtained with 90 Y PET/CT has demonstrated to be superior to that obtained with CBCT or CECT, used in current practice. This superior prediction persisted also for segmental and subsegmental infusions performed for a more effective and safer RE. Moreover, the use of 99m Tc-MAA-SPECT/CT TgV could have reduced the risk of underdosing with respect to the use of CECT or CBCT TgV. Nonetheless, CBCT provided essential information for a personalized RE planning, ensuring the total coverage of the tumor and, in cases with more than one feeding artery, splitting the activity according to the volume of tumor perfused by each artery. Therefore, the joint use of 99m Tc-MAA-SPECT/CT and CBCT optimizes dosimetric planning for RE procedures, enabling a more accurate personalized approach.