Visual Abstract
Abstract
Our objective was to compare 3 different therapeutic particles used for radioembolization in locally advanced intrahepatic cholangiocarcinoma. Methods: 90Y-glass, 90Y-resin, and 166Ho-labeled poly(l-lactic acid) microsphere prescribed activity was calculated as per manufacturer recommendations. Posttreatment quantitative 90Y PET/CT and quantitative 166Ho SPECT/CT were used to determine tumor-absorbed dose, whole-normal-liver–absorbed dose, treated-normal-liver–absorbed dose, tumor-to-nontumor ratio, lung-absorbed dose, and lung shunt fraction. Response was assessed using RECIST 1.1 and the [18F]FDG PET–based change in total lesion glycolysis. Hepatotoxicity was assessed using the radioembolization-induced liver disease classification. Results: Six 90Y-glass, 8 90Y-resin, and 7 166Ho microsphere patients were included for analysis. The mean administered activity was 2.6 GBq for 90Y-glass, 1.5 GBq for 90Y-resin, and 7.0 GBq for 166Ho microspheres. Tumor-absorbed dose and treated-normal-liver–absorbed dose were significantly higher for 90Y-glass than for 90Y-resin and 166Ho microspheres (mean tumor-absorbed dose, 197 Gy for 90Y-glass vs. 73 Gy for 90Y-resin and 50 Gy for 166Ho; mean treated-normal-liver–absorbed dose, 79 Gy for 90Y-glass vs. 37 Gy for 90Y-resin and 31 Gy for 166Ho). The whole-normal-liver–absorbed dose and tumor-to-nontumor ratio did not significantly differ between the particles. All patients had a lung-absorbed dose under 30 Gy and a lung shunt fraction under 20%. The 3 groups showed similar toxicity and response according to RECIST 1.1 and [18F]FDG PET–based total lesion glycolysis changes. Conclusion: The therapeutic particles used for radioembolization differed from each other and showed significant differences in absorbed dose, whereas toxicity and response were similar for all groups. This finding emphasizes the need for separate dose constraints and dose targets for each particle.
Intrahepatic cholangiocarcinoma (ICC) is the second most common liver malignancy, with surgery as its only curative option (1). Most patients (68%) cannot undergo surgery at diagnosis, because of unresectable (70%) or metastatic (30%) disease (1). Radioembolization is a promising treatment for locally advanced ICC (2–6). During radioembolization treatment, microspheres are injected into the hepatic artery and will preferentially lodge in and around the tumor, providing local irradiation. Currently, 3 types of microspheres are commercially available: 90Y-glass (TheraSpheres; Boston Scientific), 90Y-resin (SIR-Spheres; Sirtex Medical), and 166Ho-labeled poly(l-lactic acid) (QuiremSpheres; Quirem Medical B.V.) microspheres. Their physical characteristics are listed in Table 1.
Prescribed activity calculation methods vary. For 90Y-glass and 166Ho microspheres, a MIRD monocompartment model is typically used, aiming for an average absorbed dose in the target volume. The prescribed activity formula for 166Ho microspheres is based on a 60-Gy average absorbed dose to the whole liver (7). For 90Y-glass microspheres, the recommended dose varies with liver volume, ranging from 80 to 150 Gy, as determined by the treating physician’s evaluation (7). For 90Y-resin microspheres, the body surface area method is most commonly used, in which the prescribed activity is based on body surface area and fraction of tumor involvement (7). However, the actual dose distribution is heterogeneous. The so-called partition model takes into account differences in activity concentration in different compartments, including the tumor-to-nontumor ratio. Nonetheless, heterogeneous activity distribution within the specified compartments is not considered (8).
Dose–response and dose–toxicity studies improve personalized treatments by establishing tumor-absorbed dose targets and normal-liver–absorbed dose limits. Despite recommendations for ICC treatment in radioembolization guidelines, these recommendations are not supported by consistent evidence (7). In the literature, a wide variety of dose thresholds is reported (9–12). The reason for this variety includes differences in response criteria, absorbed dose calculations, scan modalities, tumor types, and microsphere types (8). This study compared the 3 radioembolization particles to understand the differences in dosimetric thresholds, efficacy, and toxicity for ICC patients.
MATERIALS AND METHODS
Treatment Procedures and Planning
Patients with unresectable, chemorefractory locally advanced ICC, with an Eastern Cooperative Oncology Group performance score of 2 or less and a Child–Pugh score of less than B8, treated with radioembolization after pretreatment [18F]FDG PET/CT, were included in this retrospective study. The medical ethics committee of the University Medical Center Utrecht waived the need for informed consent.
The standard radioembolization work-up included [18F]FDG PET/CT and multiphasic liver CT, followed by pretreatment planar scintigraphy and SPECT/CT with administration of about 150 MBq of 99mTc-labeled macroaggregated albumin ([99mTc]MAA) (TechneScan LyoMAA [Mallinckrodt Medical B.V.] or Pulmocis [Curium Pharma]) to assess lung shunt fraction (LSF) and exclude extrahepatic depositions. The treating physician determined the microsphere type on the basis of [99mTc]MAA SPECT/CT and target volume. The prescribed activity was based on manufacturer guidelines. The target region dose for 90Y-glass microspheres was based on visual assessment of activity accumulation in the tumor on [99mTc]MAA SPECT: 120 Gy for good targeting, 80–100 Gy for moderate to poor targeting, or 200 Gy for radiation segmentectomy. In all cases, a target dose of 60 Gy to the whole liver was planned for 166Ho microspheres. The target dose for 90Y-resin microspheres was determined using the body surface area method.
Posttreatment Imaging
Posttreatment quantitative imaging of 90Y-glass and 90Y-resin was performed within 24 h after microsphere administration. 90Y PET images were acquired on a Biograph mCT PET/CT scanner (Siemens Medical Solutions). An acquisition time of 15 min/bed position at 2 bed positions was used. Images were reconstructed using an iterative algorithm (4 iterations and 21 subsets) including (relative scaling method) scatter correction, resolution recovery, and low-dose CT-based attenuation correction (40 mAs, 100–120 kV). A gaussian postreconstruction filter of 5 mm was applied, and a 200 × 200 matrix was used, resulting in a pixel size of 4 × 4 × 3 mm.
Quantitative imaging of 166Ho microspheres was performed with SPECT/CT on a Symbia T16 (Siemens Medical Solutions) at 3–6 d after microsphere administration. The acquisition time was 15 s/view, and the number of projections was 120. A Monte Carlo–based reconstruction algorithm (Utrecht Monte Carlo System) was used (13). Technical details with respect to the reconstruction were described by Elschot et al. (13).
Three months after treatment, all patients underwent multiphasic CT, and 16 of them also underwent [18F]FDG PET/CT on the same scanner. The [18F]FDG PET/CT acquisition and reconstruction were according to the method of the European Association of Nuclear Medicine Research Ltd., version 1.0 (14).
Intrahepatic Dosimetry
Tumors were segmented on baseline [18F]FDG PET/CT using an SUVmean (normalized for lean body mass) higher than 2 times that in the aortic blood pool, excluding volumes under 5 cm3. For each tumor, metabolic activity was recorded as total lesion glycolysis (TLG), computed by multiplying metabolic tumor volume by tumor SUVmean normalized for lean body mass. Tumor volumes of interest (VOIs) were transferred from [18F]FDG PET/CT to 90Y PET/CT or 166Ho SPECT/CT using rigid registration based on low-dose CT scans (15). Whole-liver and treated-liver segments were manually delineated using 3D Slicer (Fig. 1) (16).
VOIs: tumor (red) segmented by [18F]FDG PET/CT thresholding, transferred to 90Y PET/CT (or 166Ho SPECT/CT). Lungs (blue) are segmented by thresholding of associated CT of [18F]FDG PET/CT for lung volume measurement and by associated CT of 90Y PET/CT (or 166Ho SPECT/CT) for dosimetry where lung portions within 2 cm of liver boundary are excluded (light blue). Liver (green) and treated liver (in this example same as liver) are manually delineated on associated CT. For response assessment, tumor (red) is segmented by [18F]FDG PET/CT thresholding (follow-up).
Reconstructed 90Y and 166Ho microsphere activity concentrations were converted into absorbed dose maps through PET and SPECT calibration properties using the local deposition method, which we preferred because 90Y is a 99.9% β-emitter and there are limited γ-photons from 166Ho (17,18). The mean absorbed dose was recorded for all tumors, whole-normal-liver tissue (defined as whole-liver VOI minus all tumor VOIs), and treated-normal-liver tissue (defined as treated-liver VOI minus all tumor VOIs). For tumor and liver tissue, a density of 1.04 g/cm3 was assumed (19). The tumor-to-nontumor ratio was defined as tumor-absorbed dose divided by treated-normal-liver–absorbed dose.
Lung Dosimetry
The posttreatment lung-absorbed dose (LD) was determined on 90Y PET/CT or 166Ho SPECT/CT by segmenting the lungs on the associated CT scan (excluding the basal lung regions < 2 cm from the liver; Fig. 1), assuming a lung density of 0.3 g/cm3 (19). Posttreatment LSF, defined as total lung activity divided by total-liver–plus–lung activity, was determined by extrapolating activity in the part of the lungs inside the field of view to the lung volume as measured on the pretreatment [18F]FDG PET/CT scan and using activity in the previously delineated liver.
For comparison, the estimated LSF and LD based on pretreatment planar [99mTc]MAA scintigraphy was calculated. The clinically used estimated LSF was based on the geometric mean of posterior and anterior planar images of [99mTc]MAA. This estimated LSF was converted to a pretreatment estimated LD by multiplying the estimated LSF by the net administered activity (corrected for residual activity measurement), assuming a lung mass of 1 kg.
Response and Toxicity
Per-lesion response was determined using pre- and posttreatment multiphasic CT according to RECIST 1.1 (20). For patients with follow-up [18F]FDG PET/CT, metabolic response was defined per-lesion by percentage change in TLG between pre- and posttreatment [18F]FDG PET/CT, adhering to PERCIST. Over a 45% reduction in TLG was regarded as a partial response, over a 75% increase in TLG was regarded as progressive disease, and anything in between was regarded as stable disease (21). The radioembolization-induced liver disease classification by Braat et al. was used for hepatotoxicity classification (Table 2) (22).
Hepatotoxicity Classification (22)
Statistics
Microsphere groups were compared using the Kruskal–Wallis test, and significant metrics were further analyzed using the Mann–Whitney U test. A 5% significance level was assumed. [99mTc]MAA-based lung dosimetry and posttreatment lung dosimetry relationships were tested using linear regression. CIs for low numbers were determined using the Agresti–Coull formula.
RESULTS
From June 2011 to March 2020, 23 patients with advanced ICC received treatment with different microspheres: 7 with 90Y-glass, 8 with 90Y-resin, and 8 with 166Ho. One 166Ho patient was excluded because the activity at the time of the SPECT acquisition was too high for adequate quantitative imaging (1,739 MBq) (23), and 1 90Y-glass patient was excluded because of an [18F]FDG-negative tumor. Table 3 shows the characteristics of the remaining patients. Three patients received first-week 90Y-glass microspheres: 2 at 4 d and 1 at 3 d after calibration. Another 3 patients were treated with second-week microspheres: 1 at 9 d and 2 at 11 d after calibration. Significant group differences were observed in the type of radioembolization treatment, between 90Y-glass and 90Y-resin (P = 0.01), and in median administered activity and median activity at posttreatment imaging, between the 166Ho group and both 90Y groups (P = 0.01 each).
Patient Characteristics
Figure 2 shows the tumor-to-nontumor ratio, tumor-absorbed dose, treated-normal-liver–absorbed dose, and whole-normal-liver–absorbed dose for each group of microspheres. 90Y-glass microspheres differed significantly from 90Y-resin and 166Ho microspheres for tumor-absorbed dose (mean of 197 Gy vs. 73 Gy and 50 Gy, respectively; P = 0.006) and treated-healthy-liver–absorbed dose (mean of 79 Gy vs. 37 Gy and 31 Gy, respectively; P = 0.001). No microspheres differed for whole-normal-liver–absorbed dose (mean of 41, 37, and 28 Gy, respectively) or tumor-to-nontumor ratio (mean of 2.5, 2.1, and 3.0, respectively).
Microsphere posttreatment dosimetry. Numbers above box plots indicate number of included tumors or patients. *P < 0.05. G = 90Y-glass; H = 166Ho; R = 90Y-resin; T/N = tumor-to-nontumor activity concentration ratio.
All patients had an LD under 30 Gy and an LSF under 20% (Fig. 3). LD and LSF differences among groups were insignificant on planar [99mTc]MAA scintigraphy but were significant on posttreatment imaging. For LD, all microsphere groups were significantly different from each other (mean, 0.1 Gy for 166Ho vs. 5.0 Gy for 90Y-glass vs. 1.3 Gy for 90Y-resin; P = 0.001). For LSF, all patients in both the 90Y-glass and the 90Y-resin groups exhibited an LSF greater than 0% after treatment. However, in the 166Ho microsphere group, 71% (5/7) patients displayed an LSF above 0%. The mean LSF of 166Ho microspheres differed significantly from that of 90Y-glass and 90Y-resin microspheres (mean of 0.2% for 166Ho vs. 3.7% for 90Y-glass and 1.5% for 90Y-resin; P = 0.002). None of the particles showed a linear trend with [99mTc]MAA prediction—either for LD or for LSF (all 95% CIs for slope include 0 and all R2 < 0.5).
Lung dosimetry. Individual patient data are connected by line. *P < 0.05. G = 90Y-glass; H = 166Ho; R = 90Y-resin.
RECIST 1.1 was assessed in 20 patients. Evaluation was impeded for 1 patient because of intrahepatic edema after treatment (Fig. 4) and for another because of surgical clip artifacts. Sixteen patients underwent 3 mo of follow-up [18F]FDG PET/CT for per-lesion response assessment (Fig. 5). The number of tumors included for RECIST 1.1 and TLG change were 6 and 6 for 90Y-glass, 22 and 9 for 90Y-resin, and 11 and 11 for 166Ho, respectively. Response differences were not significant per RECIST 1.1. TLG change showed that 166Ho differed significantly from 90Y-glass but not from 90Y-resin (median TLG change of −29% [range, 220% to −79%] for 166Ho, −84% [range, −49% to −100%] for 90Y-glass, and −84% [range, 391% to −100%] for 90Y-resin; P = 0.02). On a per-patient basis according to RECIST 1.1, 2 of 20 patients showed a partial response, 14 of 20 showed stable disease, and 4 of 20 showed progressive disease.
Extensive intrahepatic edema and necrosis after left lobar treatment with 90Y-glass microspheres. Shown are multiphase contrast-enhanced CT at baseline (A and C) and 3 mo after treatment (B and D). At baseline, primary tumor invades left liver lobe on portal venous CT (C) and shows enhancement in arterial phase (A). Posttreatment images show intrahepatic edema and necrosis, masking tumor for RECIST 1.1 assessment. However, disappearance of arterial enhancement suggests complete response according to modified RECIST.
Response according to RECIST 1.1 and TLG change. CR = complete response; PD = progressive disease; PR = partial response; SD = stable disease.
90Y-glass, 90Y-resin, and 166Ho microspheres showed no significant differences in hepatotoxicity classification with respect to whole-normal-liver–absorbed dose and treated-normal-liver–absorbed dose (Fig. 6). Fatal radioembolization-induced liver disease, grade 5, occurred only once.
Hepatotoxicity classification with respect to dose absorbed by whole normal liver (left) and treated normal liver (right).
DISCUSSION
This study compared the commercially available microspheres for radioembolization in locally advanced ICC. Despite dosimetric differences, the microspheres showed similar toxicity and response. The differences in microsphere-specific activities cause variations in the administered number of microspheres. Pasciak et al. showed that greater microsphere quantities result in a more homogeneous dose distribution, leading to receipt of a potentially toxic absorbed dose by a greater volume fraction of the treated liver (24) and potentially explaining the lesser toxicity of 90Y-glass than of 90Y-resin at the same average absorbed dose (9). Administered microsphere numbers for 166Ho resembled those for 90Y-resin but were considerably less than for 90Y-glass.
90Y-glass microspheres delivered a higher tumor-absorbed dose because of a higher total administered activity than for 90Y-resin microspheres. However, all 90Y-glass microsphere patients received partial-liver treatment, resulting in a significant difference in the choice of radioembolization treatment between the 90Y-glass and 90Y-resin microspheres. Although no significant difference in the previous treatments and Eastern Cooperative Oncology Group performance status were found, these physician-induced group differences are inherent in retrospective research and consistent with day-to-day practice.
The per-tumor response analysis revealed distinct response profiles between TLG and RECIST 1.1., partly because RECIST 1.1 does not differentiate necrotic from nonnecrotic tissue. Although there is no consensus on the preferred response criteria for ICC treated with radioembolization, and although determining this preference was outside the scope of this study, Figure 4 suggests greater suitability of modified RECIST (not used in this work), in which contrast enhancement in the arterial phase represents viable tumor parts. However, modified RECIST cannot be used for hypovascular ICC, and [18F]FDG PET/CT and TLG change can depict the viable tissue proportion in both hypervascular and hypovascular tumors. The latter method is not perfect either, however, as 1 patient in this study was excluded because of [18F]FDG-negative disease.
Consistent with prior studies on ICC and radioembolization, hepatotoxicity was modest (3,6,25,26). Grade 5 radioembolization-induced liver disease occurred in 1 patient, who received a whole-liver treatment (mean liver-absorbed dose of 81 Gy). The injection position differed from the [99mTc]MAA procedure, leading to a poor tumor dose and a high normal-liver dose.
Planar [99mTc]MAA scintigraphy showed similar lung dosimetry for the entire population, confirming the similarity of the groups at baseline, but posttreatment lung dosimetry showed differences between therapeutic particles, where 166Ho stood out for both LSF and LD. This observation is at least partially due to the different imaging modalities. 166Ho images were acquired on a SPECT system, combined with Monte Carlo reconstruction. 90Y images were acquired on a PET system, which suffers from a low count rate and a high random fraction. This is known to cause a positive bias in the reconstruction, especially in regions with low counts (e.g., the lungs), resulting in an overestimation of the LSF (27,28). Hence, these observed differences may be more reflective of the technical aspects of scanning and of significant differences in treatment approaches and volumes (whole liver vs. selective) than of any physiologic variations.
Contrary to other publications on dosimetry in radioembolization, calculation of absorbed dose was based on posttreatment imaging, not on pretreatment [99mTc]MAA SPECT/CT. In addition, reconstructed activity was not scaled on the basis of net administered activity. This introduces errors in 90Y PET/CT and 166Ho SPECT/CT quantification but avoids errors introduced by scaling and residual activity measurement of administered net activity (29). Although commonly used, the rigid coregistration method for transferring tumor VOIs from [18F]FDG PET/CT to SPECT/CT relies on low-dose CT, not on the molecular images themselves, introducing errors. Additionally, the use of rigid coregistration for a deformable organ is known to induce errors to some degree.
Dosimetric outcomes adhere closely to the instructions for use outlined by each manufacturer, complicating direct comparison among the 3 microsphere types. Still, following these instructions enhances the real-world applicability of our results. However, patients were treated before the release of the 2022 European Association of Nuclear Medicine guidelines (7), possibly limiting comparisons with current clinical practice.
A limitation of this study was the small population, which impedes our ability to draw generalizable conclusions. In addition, follow-up [18F]FDG PET/CT was not available for all patients, and modified RECIST could not be assessed in most patients because of hypovascular tumors. Therefore, conclusions on dose–response and dose–toxicity relationships cannot be drawn. Overall, irrespective of the particle used, and on a per-patient basis, our results were in line with the literature (3,6,26).
Another limitation was the bias regarding type of treatment. In our institute, all particles are used for whole-liver treatments, but for selective treatments 90Y-glass microspheres are preferred because of high specific activity. In the presented patients with locally advanced ICC, this difference resulted in a 90Y-glass group with only partial-liver treatments, a 90Y-resin group with only whole-liver treatments, and a mixed 166Ho microsphere group. Moreover, 90Y-glass microspheres were injected at different times after calibration, that is, with a different specific activity per sphere. This variability within the 90Y-glass group may limit comparisons of this group with the other treatment groups.
Finally, because many patients had discrepancies between the [99mTc]MAA procedure and treatment (e.g., different injection positions or additional coiling), we could not use [99mTc]MAA SPECT/CT data to compare intrahepatic dose metrics with posttreatment PET/CT or SPECT/CT. (30). We did use [99mTc]MAA planar data for lung dosimetry, as a different catheter position would barely influence these results.
CONCLUSION
In the present study, 3 available microspheres were compared in locally advanced ICC. The particles differed in physical characteristics and methods for calculating prescribed activity, resulting in significant differences in absorbed dose. However, toxicity and response were similar for all groups. This finding emphasizes the need for separate dose constraints and dose targets for each particle.
DISCLOSURE
The UMC Utrecht Department of Radiology and Nuclear Medicine receives royalties from Terumo. Marnix Lam is a consultant for Boston Scientific and Terumo and receives research support from Boston Scientific and Novartis. Arthur Braat is a consultant for Boston Scientific and Terumo and receives research support from Ariceum Therapeutics. Maarten Smits is a consultant for Philips, Terumo/Quirem Medical, and Swedish Orphan Biovitrum. He also serves as a speaker for Medtronic. No other potential conflict of interest relevant to this article was reported.
KEY POINTS
QUESTION: How do different therapeutic particles for SIRT compare in the treatment of advanced ICC?
PERTINENT FINDINGS: The therapeutic particles used for radioembolization significantly differ from each other in terms of absorbed dose, yet they show similar toxicity and response across all groups.
IMPLICATIONS FOR PATIENT CARE: The findings emphasize the need for distinct dose constraints and targets for each therapeutic particle, optimizing patient outcomes in ICC treatment.
Footnotes
Published online Jan. 4, 2024.
- © 2024 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication May 16, 2023.
- Revision received November 7, 2023.