Accuracy and Safety of Scout Dose Resin Yttrium-90 Microspheres for Radioembolization Therapy Treatment Planning: A Prospective Single-Arm Clinical Trial

Abstract

Purpose:

To compare the accuracy and safety of 0.56 GBq resin yttrium-90 (⁹⁰Y) (scout⁹⁰Y) microspheres with those of technetium-99m macroaggregated albumin (MAA) in predicting the therapeutic ⁹⁰Y (Rx⁹⁰Y) dose for patients with hepatocellular carcinoma (HCC).

Materials and Methods:

This prospective single-arm clinical trial (Clinicaltrials.gov: NCT04172714) recruited patients with HCC. Patients underwent same-day mapping with MAA and scout⁹⁰Y. Rx⁹⁰Y activity was administered 3 days after mapping. Using paired t test and Pearson correlation, the tumor-to-normal ratio (TNR), lung shunt fraction (LSF), predicted mean tumor dose (TD), and nontumoral liver dose (NTLD) by MAA and scout⁹⁰Y were compared with those by Rx⁹⁰Y. Bland-Altman plots compared the level of agreement between the TNR and LSF of scout⁹⁰Y and MAA with that of Rx⁹⁰Y. The safety of scout⁹⁰Y was evaluated by examining the discrepancy in extrahepatic activity between MAA and scout⁹⁰Y.

Results:

Thirty patients were treated using 19 segmental and 14 nonsegmental (ie, 2 contiguous segments or nonsegmental) therapies. MAA had weak LSF, moderate TNR, and moderate TD linear correlation with Rx⁹⁰Y. Scout⁹⁰Y had a moderate LSF, strong TNR, strong TD, and very strong NTLD in correlation with those of Rx⁹⁰Y. Furthermore, the TNR and LSF of scout⁹⁰Y had a stronger agreement with those of Rx⁹⁰Y than with those of MAA. In the nonsegmental subgroup, MAA had no significant correlation with the TD and NTLD of Rx⁹⁰Y, whereas scout⁹⁰Y had a very strong correlation with both of these factors. In the segmental subgroup, both MAA and scout⁹⁰Y had a strong linear correlation with the TD and NTLD of Rx⁹⁰Y.

Conclusions:

Compared with MAA, scout⁹⁰Y is a more accurate surrogate for Rx⁹⁰Y biodistribution for nonsegmental therapies.

Most patients with hepatocellular carcinoma (HCC) are unable to undergo liver transplantation or curative resection (1), necessitating the use of liver-directed therapies, including yttrium-90 (⁹⁰Y) radioembolization.

A critical factor in optimizing radioembolization therapy is the delivery of a sufficient radioactive dose to the tumor. A recent study by Gabr et al (2) examining explanted livers indicated that in patients treated with glass ⁹⁰Y microspheres (Therasphere; Boston Scientific, Marlborough, Massachusetts), complete pathological necrosis is more likely to occur when a dose of >400 Gy is delivered to the tumor. Garin et al (3) also demonstrated that personalized dosimetry with single-photon emission computed tomography (SPECT) or computed tomography (CT) with a goal of ≥205 Gy to a targeted HCC tumor with glass ⁹⁰Y resulted in prolonged survival. Additionally, a post hoc dosimetric analysis of the SorAfenib versus Radio-loembolization in Advanced Hepatocellular Carcinoma (SARAH) trial demonstrated significantly prolonged overall survival among patients with tumors treated with ≥100 Gy compared with that among patients treated with other doses (4). The aforementioned studies and other emerging data have led to increased emphasis on prospective treatment planning with adequate target dose in recent years.

Despite the routine use of technetium-99m (^99mTc) macroaggregated albumin (MAA) for planning purposes, evidence suggests that this radiotracer is a suboptimal surrogate for ⁹⁰Y microspheres (5). Several studies (6-9) have demonstrated that ^99mTc MAA is an inconsistent or inaccurate predictor of ⁹⁰Y microsphere biodistribution, with a significant discrepancy in a substantial percentage of patients owing to size differences, particle number differences, and clumping during delivery. Because tumor dose (TD) directly affects patient outcomes, predictive dosimetric inaccuracies with MAA may result in worsened tumor response, survival, and toxicity, particularly in lobar setting.

Low-dose “scout” ⁹⁰Y microspheres administered during angiographic mapping may serve as a more accurate marker for ⁹⁰Y biodistribution and as a bioidentical surrogate. In this prospective single-arm clinical trial, the accuracy and safety of scout⁹⁰Y administration were compared with those of ^99mTc MAA administration during ⁹⁰Y preprocedural mapping.

MATERIALS AND METHODS

Study Design and Patient Selection

This study was a prospective single-arm clinical trial (clinicaltrials.gov identifier: NCT04172714). The study was approved by the Emory Institutional Review Board. Written consent was obtained for each participant.

All patients recruited were those with treatment-naïve, surgically unresectable HCC and deemed to benefit from ⁹⁰Y radioembolization as either a potential curative or palliative therapy by a multidisciplinary institutional tumor board. The inclusion criteria involved adult patients with the following criteria: (a) life expectancy of ≥6 months, (b) presence of HCC confirmed by Liver Reporting and Data System (LIRADS) on CT or magnetic resonance (MR) imaging (10), (c) ≤3 lesions confined to a single lobe, (d) a targeted lesion measuring ≥2 cm and ≤8 cm, (e) absence of extrahepatic metastatic disease, (f) and an Eastern Cooperative Oncology Group (ECOG) performance status of <2. The required laboratory values included a bilirubin level of ≤2 mg/dL for nonsegmental (ie, 2 contiguous segments or lobar) therapies and ≤3 mg/dL for segmental therapies, an albumin level of ≥3 g/dL, a prothrombin time/international normalized ratio of <2, aspartate aminotransferase and alanine transaminase levels of ≤3 times the upper limit of the normal range, and a platelet count of >50,000/μL. The exclusion criteria included patients with an elevated lung shunt fraction (LSF) by MAA, which would result in dose modification, thereby resulting in an inadequate TD.

Baseline Demographics

From December 2019 to January 2021, 30 patients with 33 HCCs were recruited and treated (Table 1). The mean tumor volume was 44.9 cm³. The mean age was 66.4 years, with 90.3% of patients being men. The most common etiology of HCC was hepatitis C (54.8%), followed by alcoholic liver cirrhosis (22.6%). The mean baseline model for end-stage liver disease and albuminbilirubin scores were 10 and −2.62, respectively. The ECOG performance status was 0, and the Child-Pugh score was A in 77.4% of patients (Table 1). Overall, 19 tumors were treated with radiation segmentectomy, and 14 tumors were treated in a nonsegmental fashion. The mean tumor volume in the segmental subgroup was 33.7 cm³ (median, 21.8 cm³; SD ± 30.3 cm³) and that in the nonsegmental subgroup was 58.4 cm³ (median, 16.8 cm³; SD ± 92.7 cm³).

Table 1.

Baseline Characteristics of the Entire Cohort

Variable	Frequency	Percentage
Age (y)
Mean	66.4
Median	68.2
Male sex	28	90.3
Etiology of HCC
Hepatitis B virus	4	12.9
Hepatitis C virus	17	58.8
EtOH	7	22.6
NAFLD	4	12.9
MELD
Mean	10
Median	10
Child-Pugh class
A	24	77.4
B	6	19.4
ALBI
Mean score	−2.65
Grade
1	17	54.8
2	13	41.9
ECOG PS
0	24	77.4
1	6	19.4
2	0	0.0
3	0	0.0
4	0	0.0
Radioembolization type by tumor
Segmental	19	58
Nonsegmental	14	42
Tumor distribution
Solitary	27	90
Multifocal	3	10
Scout90Y activity (GBq)
Mean	0.56
Median	0.56
90Y therapeutic activity (GBq)
Mean	1.17
Median	0.78
BCLC Stage
A	16	53.3
B	9	30.0
C	5	16.7

ALBI = albumin-bilirubin; BCLC = Barcelona Clinic Liver Cancer; ECOG PS = Eastern Cooperative Oncology Group performance status; EtOH = ethanol; HCC = hepatocellular carcinoma; MELD = model for end-stage liver disease; NAFLD = non-alcoholic fatty liver disease.

Two patients required coil embolization of the right gastric artery before MAA administration. The mean scout⁹⁰Y activity administered was 0.56 GBq, and the mean therapeutic ⁹⁰Y (Rx⁹⁰Y) activity was 1.17 GBq (Table 1). The mean lung dose for the entire cohort was 6.8 Gy (median, 5.8 Gy; SD ± 4.7 Gy). The mean lung dose for the segmental subgroup was 7.2 Gy and that of nonsegmental subgroup was 6.3 Gy.

Angiographic Mapping Procedures

All angiographic procedures were performed by a fellowship-trained interventional radiologist (N.K.) with at least 5 years of experience. All patients underwent prospective angiographic mapping as described in previous literature (11), using coaxial arterial access with a 5-F base catheter and a 2.4–2.8-F microcatheter. This included the use of cone-beam CT with real-time 3-dimensional reconstruction performed in all cases to evaluate tumoral supply to the targeted tumor(s) and ensure complete perfusion of the tumor. From the microcatheter location(s) perfusing the entirety of the tumor, 148 MBq of ^99mTc MAA was administered for nonsegmental therapy and 74 MBq of ^99mTc MAA was administered for segmental therapy. If 2 segmental administrations were needed, 74 MBq was administered in each supplying vessel. The catheters were then removed. The access site was secured and left in place with a running heparinized saline. The patient was then transferred to the nuclear medicine department to undergo ^99mTc MAA planar and SPECT/CT scans. After scanning, the patient was brought back to the interventional radiology department for the second mapping procedure. A typical patient flowchart used in the study is depicted in Figure 1.

Figure 1.

Study flowchart for each recruited patient. ⁹⁰Y = yttrium-90; CBCT = cone-beam computed tomography; CT = computed tomography; f/u = follow-up; HCC = hepatocellular carcinoma; MAA = macroaggregated albumin; MR = magnetic resonance; PET = positron emission tomography; SIRT = selective internal radiation therapy; SPECT = single-photon emission computed tomography; Tc99 = technetium-99m; w/ = with.

Using the aforementioned techniques, the second mapping procedure was performed using resin ⁹⁰Y microspheres (SIR-Spheres; Sirtex, Woburn, Massachusetts). From the same microcatheter location and using the same type of microcatheter that was used in MAA mapping, 0.56-GBq 3-day precalibrated ⁹⁰Y microspheres were administered as the scout dose (scout⁹⁰Y). This equated to 3.24 million ⁹⁰Y microspheres administered at mapping for a single dose. If there was dual blood supply to the tumor, 0.37 GBq of ⁹⁰Y was administered to each artery, totaling to 4.32 million microspheres for patients who required a split dose. The 0.56 GBq of scout activity was determined on the basis of a previously conducted phantom study (12) demonstrating that 0.48 GBq of ⁹⁰Y activity was needed for accurate imaging on both SPECT and positron emission tomography (PET).

Catheters and sheaths were then removed. Hemostasis was achieved with a transradial band or a femoral artery closure device. The patient was then transferred to the nuclear medicine department for ⁹⁰Y PET with CT (Fig 1), after which they were discharged home.

LSF and Dosimetry Analysis

All imaging evaluation was performed by nuclear medicine physicians and physicists (D.B. and I.S.) with more than 10 years of experience. The LSF of ^99mTc MAA was assessed using planar and SPECT/CT. Attenuation-corrected images of the liver and lungs obtained using SPECT/CT were used for treatment planning, including LSF calculation, over the classically performed planar scintigraphy, given its higher accuracy. Briefly, semiautomated segmentation of the liver and lung parenchyma with manual manipulation, if needed, was performed with MIM, version 6.9 (MIM Software, Beachwood, Ohio) (13). MAA tumor-to-normal liver ratio (TNR) was also calculated using SPECT/CT. The perfused area of the liver and the targeted tumor(s) was contoured. Using the respective volume and activity count in each of the compartment, the TNR was calculated.

The LSF and TNR for the scout⁹⁰Y and Rx⁹⁰Y were assessed with PET/CT using the aforementioned method. PET/CT was selected over ⁹⁰Y SPECT/CT because of its ability to provide high-resolution images, as validated by several studies (13,14).

Prospective Dosimetry Treatment Planning

The therapeutic ⁹⁰Y activity to be administered was calculated based on MAA TNR and LSF data from SPECT/CT as the current standard of care. A partition model was used to calculate ⁹⁰Y dosages in all patients who underwent nonsegmental therapies with the goal of delivering a >200 Gy dose to tumors and <70 Gy dose to the perfused nontumoral liver (15). A single-compartment dose of 200 Gy to the segment was used for segmentectomies. A dose of the prescribed ⁹⁰Y activity minus the scout (.56 GBq) activity was then administered within 3 days from the mapping procedures for Rx⁹⁰Y. All dosimetry planning was performed using MIM SurePlan (MIM Software) (13).

Therapeutic Selective Internal Radiotherapy Procedure

Patients underwent Rx⁹⁰Y 3 days after the mappings as an outpatient. Using the aforementioned technique and from the same microcatheter location where MAA and Scout⁹⁰Y were administered and using the same type of microcatheter, the prescribed activity was administered using 3-day precalibrated resin ⁹⁰Y for segmentectomies or 2 day precalibrated resin ⁹⁰Y for nonsegmental therapies. The patient was then transferred to the nuclear medicine department for ⁹⁰Y PET/CT to confirm adequate tumor coverage and assess for extrahepatic activity (Fig 1).

Accuracy of ^99mTc MAA versus Scout⁹⁰Y in Predicting Dosimetry after Therapeutic ⁹⁰Y Radioembolization

Using voxel-based dosimetry, the prescribed activity and biodistribution of MAA and scout⁹⁰Y, which included the predicted mean TD and nontumoral liver dose (NTLD), were calculated with SurePlan. The TNR and LSF from MAA and scout⁹⁰Y were used in the partition model using the cumulative prescribed ⁹⁰Y activity (scout + therapeutic) to calculate the respective predicted TD and NTLD.

Safety of Scout⁹⁰Y versus ^99mTc-MAA

The safety of the scout dose resin microsphere was evaluated by determining nontarget embolization seen on scout⁹⁰Y PET that was not detected and mitigated during mapping cone-beam CT and evaluating for a discrepancy in extrahepatic activity between MAA SPECT/CT and scout dose ⁹⁰Y PET/CT. All adverse events were reported using Common Terminology for Clinical Adverse Events (CTCAE), version 5 (16). Furthermore, the potential embolic effect of scout⁹⁰Y was evaluated, as outlined in Appendix A (available online on the article’s Supplemental Material page at www.jvir.org).

Statistical Analysis

The mean TNR and LSF and dose predicted by MAA (MAA TD and MAA NTLD) and scout⁹⁰Y (scout TD and scout NTLD) were compared with those of Rx⁹⁰Y using the paired Student t test. The linear correlation coefficients between these variables were compared using the Pearson correlation coefficient. The degree of linear correlation was defined as follows: 0–0.3, very weak; 0.3–0.5, weak; 0.5–0.7, moderate; 0.7–0.9, strong; and 0.9–1.0, very strong. Subanalyses for all dosimetric factors were then performed for patients who underwent segmentectomy versus those treated with nonsegmental treatments. Additionally, Bland-Altman plots were created to evaluate the degree of agreement between the TNR and LSF of Rx⁹⁰Y and those of MAA and scout⁹⁰Y. Bland-Altman analysis was performed using R software, version 4 (R Foundation for Statistical Computing, Vienna, Austria) (Table E1, available online at www.jvir.org). All other statistical analyses were performed using SPSS Software, version 28 (IBM, Armonk, New York).

RESULTS

Tumor-to-Normal Liver Ratio

For the entire cohort, the mean TNR for MAA, scout⁹⁰Y, and Rx⁹⁰Y was 2.67, 2.75, and 2.74, respectively (for all, P > .05). There was a moderate linear correlation between the TNRs of MAA and Rx⁹⁰Y (r = 0.53; P = .002) and a very strong linear correlation between the TNRs of scout⁹⁰Y and Rx⁹⁰Y (r = 0.904; P = .001) (Table 2 and Fig 2a, d)

Table 2.

Tumor-to-Normal Liver Ratio and Lung Shunt Fraction Data: Entire Cohort and Nonsegmental and Segmental Groups

Variable	Mean TNR	Paired t test with Rx90Y (P value)	Correlation with Rx90Y (r)	P value	Mean LSF (%)	Paired t test with Rx90Y (P value)	Correlation with Rx90Y (r)	P value
Entire cohort: MAA	2.67	.796	0.53	.020	5.68	.418	0.39	.031
Entire cohort: Scout90Y	2.75	.893	0.90	<.001	5.79	.058	0.76	<.001
Entire cohort: Rx90Y	2.74				5.21
Nonsegmental: MAA	2.65	.312	0.62	.019	5.67	.653	0.53	.062
Nonsegmental: Scout90Y	2.38	.015	0.98	<.001	6.31	.023	0.89	<.001
Nonsegmental: Rx90Y	2.21			5.25
Segmental: MAA	2.69	.136	0.57	.013	5.68	.511	0.19	.471
Segmental: Scout90Y	3.04	.525	0.86	<.001	5.39	.598	0.43	.083
Segmental: Rx90Y	3.15				5.18

⁹⁰Y = yttrium-90; BCLC = Barcelona Clinic Liver Cancer; LSF = lung shunt fraction; MAA = macroaggregated albumin; Rx⁹⁰Y = therapeutic yttrium-90; TNR = tumor-to-normal liver ratio.

Figure 2.

Correlation of tumor-to-normal liver ratio (TNR) by yttrium-90 (⁹⁰Y) mapping and ⁹⁰Y treatment type: (a) macroaggregated albumin (MAA) versus therapeutic ⁹⁰Y (Rx⁹⁰Y) entire cohort, (b) MAA versus Rx⁹⁰Y nonsegmental treatment, (c) MAA versus Rx⁹⁰Y segmental treatment, (d) Scout⁹⁰Y versus Rx⁹⁰Y entire cohort, (e) Scout⁹⁰Y versus Rx⁹⁰Y nonsegmental treatment, and (f) Scout⁹⁰Y versus Rx⁹⁰Y segmental treatment.

For the nonsegmental subgroup, the mean TNR was not significantly different between MAA and Rx⁹⁰Y (2.65 vs 2.21; P = .312); whereas, there was a statistically significant difference between the TNRs of scout⁹⁰Y and Rx⁹⁰Y (2.38 vs 2.21; P = .015). However, in the same subgroup, the TNR of MAA was moderately correlated with the TNR of Rx⁹⁰Y (r = 0.615; P = .019), and the TNR of scout⁹⁰Y was very strongly correlated with the TNR of Rx⁹⁰Y (r = 0.982; P < .001) (Table 2 and Fig 2).

For the segmental subgroup, the TNR was 2.69 for MAA, 3.04 for scout⁹⁰Y, and 3.15 for Rx⁹⁰Y, which were not statistically different (for all, P > .05). Again, there was a moderate correlation of MAA TNR (r = 0.57; P = .013) and a high correlation of Scout⁹⁰Y TNR with Rx⁹⁰Y TNR (r = 0.86; P < .001) (Table 2 and Fig 2).

Additionally, the descriptive Bland-Altman analysis demonstrated that the TNR of scout⁹⁰Y for the entire cohort had a clearly superior agreement with the TNR of Rx⁹⁰Y (mean difference, −0.01; standard difference, 0.57) compared with that of MAA TNR with Rx⁹⁰Y TNR (mean difference, 0.07; standard difference, 1.45) (Fig E1, available online at www.jvir.org).

Lung Shunt Fraction

For the entire cohort, the LSF for MAA, scout⁹⁰Y, and Rx⁹⁰Y was 5.7%, 5.8%, and 5.2%, respectively (for all, P > .05). There was a weak linear correlation between the LSF of MAA and the LSF of Rx⁹⁰Y (r = 0.39; P = .032) and a strong linear correlation between the LSF of scout⁹⁰Y and the LSF of Rx⁹⁰Y (r = 0.76; P < .001) (Table 2; Fig E2, available online at www.jvir.org).

In the nonsegmental subgroup, the mean LSF was not statistically different between MAA (5.7%) and Rx⁹⁰Y (5.2%; P = .65) but was different between scout⁹⁰Y (5.8%) and Rx⁹⁰Y (5.2%; P = .02). However, there was no correlation between the LSF of MAA and the LSF of Rx⁹⁰Y (r = 0.53; P = .062); whereas, there was a strong correlation between the LSF of scout⁹⁰Y and the LSF of Rx⁹⁰Y (r = 0.885; P < .001) (Table 2; Fig E2, available online at www.jvir.org).

For the segmentectomies, the mean LSFs were not statistically different (MAA, 5.6%; scout⁹⁰Y, 5.4%; and Rx⁹⁰Y, 5.2%; for all, P > .05). There was no significant correlation between the LSF of MAA and the LSF of Rx⁹⁰Y (P = .47). However, there was a week correlation between the LSF of scout⁹⁰Y and the LSF of Rx⁹⁰Y (r = 0.43; P = .042) (Table 2; Fig E2, available online at www.jvir.org).

The descriptive Bland-Altman analysis demonstrated that the LSF of scout⁹⁰Y for the entire cohort had a marginally superior agreement with the LSF of Rx⁹⁰Y (mean difference, −0.01; standard difference, 0.02) compared with that of MAA LSF with Rx⁹⁰Y LSF (mean difference, 0; standard difference, 0.03) (Fig E1, available online at www.jvir.org).

Mean TD Prediction using MAA and Scout⁹⁰Y

For the entire cohort, the mean TDs predicted using MAA and scout⁹⁰Y were 474 and 470 Gy, respectively, and were not statistically different from the TD of Rx⁹⁰Y, which was 491 Gy (for all, P > .05). There was a moderate linear correlation between the TD of MAA and the TD of Rx⁹⁰Y (r = 0.74; P < .001) and a very strong linear correlation between the TNR of scout⁹⁰Y and the TNR of Rx⁹⁰Y (r = 0.900; P < .001).

For the nonsegmental subgroup, there was no significant difference between the mean TDs (396 Gy for MAA vs 404 Gy for scout⁹⁰Y vs 412 Gy for Rx⁹⁰Y; for all, P > .05). The TD of scout⁹⁰Y was very strongly correlated with the TD of Rx⁹⁰Y (r = 0.93; P < .001), whereas MAA had no correlation with Rx⁹⁰Y (r = 0.341; P = .232) (Table 3 and Fig 3).

Table 3.

Tumor Dose and Nontumor Liver Dose

Variable	Mean TD (Gy)	Paired t test with Rx90Y (P value)	Correlation with Rx90Y (r)	P value	Mean NTLD (Gy)	Paired t test with Rx90Y (P value)	Correlation with Rx90Y (r)	P value
Entire cohort: MAA	474	.68	0.74	<.001	117	.003	0.93	<.001
Entire cohort: Scout90Y	470	.44	0.9	<.001	91	.163	0.97	<.001
Entire cohort: Rx90Y	491				87
Nonsegmental: MAA	396	.83	0.341	.232	67	.14	0.44	.114
Nonsegmental: Scout90Y	404	.75	0.93	<.001	62	.19	0.95	<.001
Nonsegmental: Rx90Y	412				61
Segmental: MAA	534	.71	0.87	<.001	164	.005	0.92	<.001
Segmental: Scout90Y	521	.49	0.91	<.001	117	.23	0.97	<.001
Segmental: Rx90Y	552				110

⁹⁰Y = yttrium-90; MAA = macroaggregated albumin; NTLD = nontumor liver dose; Rx⁹⁰Y = therapeutic yttrium-90; TD = tumor dose. Statistically significant values are set in bold.

Figure 3.

The correlation of mean tumor dose by yttrium-90 (⁹⁰Y) mapping and ⁹⁰Y treatment type: (a) macroaggregated albumin (MAA) versus therapeutic ⁹⁰Y (Rx⁹⁰Y) entire cohort, (b) MAA versus Rx⁹⁰Y nonsegmental treatment, (c) MAA versus Rx⁹⁰Y segmental treatment, (d) Scout⁹⁰Y versus Rx⁹⁰Y entire cohort, (e) Scout⁹⁰Y versus Rx⁹⁰Y nonsegmental treatment, and (f) Scout⁹⁰Y versus Rx⁹⁰Y segmental treatment.

For the segmental subgroup, the mean predicted TDs by MAA (534 Gy) and scout⁹⁰Y (521 Gy) were statistically similar to that of Rx⁹⁰Y (552 Gy; for all, P > .05). Additionally, MAATD had a strong correlation with Rx⁹⁰Y TD (r = 0.87; P < .001), whereas scout⁹⁰Y TD was very strongly correlated with Rx⁹⁰Y TD (r = 0.91; P < .001) (Table 3 and Fig 3).

Mean NTLD Prediction by MAA and Scout⁹⁰Y

For the entire cohort, NTLDs predicted by MAA were significantly different from those predicted by Rx⁹⁰Y (117 vs 87 Gy; P = .003), whereas the predicted NTLDs by scout⁹⁰Yand Rx⁹⁰Y were similar (91 vs 87 Gy; P = .163). The scout⁹⁰Y NTLD was very strongly correlated with the Rx⁹⁰Y NTLD (r = 0.972; P < .001). In addition, there was a very strong linear correlation between the MAA NTLD and the Rx⁹⁰Y NTLD (r = 0.933; P < .001).

In the nonsegmental subgroup, NTLDs were not significantly different among MAA, scout⁹⁰Y, and Rx⁹⁰Y (67, 62.2, and 61.1 Gy, respectively; P = .19). However, the MAA NTLD had no correlation with the Rx⁹⁰Y NTLD (r = 0.442; P = .114), whereas the scout⁹⁰Y NTLD had a very strong correlation with the Rx⁹⁰Y NTLD (r = 0.95; P < .001) (Fig E3, available online at www.jvir.org).

In the segmentectomy subgroup, the mean MAA NTLD was significantly higher than the mean Rx⁹⁰Y NTLD (164 vs 111 Gy; P < .001). There was no statistically significant difference in the mean scout⁹⁰Y NTLD and Rx⁹⁰Y NTLD (117 vs 110 Gy; P = .228) (Table 3; Fig E3, available online at www.jvir.org). Both the MAA NTLD (r = 0.92) and scout⁹⁰Y NTLD (r = 0.97) had a very strong linear correlation with the Rx⁹⁰Y NTLD (for all, P < .001) (Table 3; Fig E3, available online at www.jvir.org).

Safety

No extrahepatic nontarget embolization was seen on scout⁹⁰Y PET. Specifically, there was no discrepancy between extrahepatic activity seen on MAA SPECT/CT and scout⁹⁰Y PET/CT. Additionally, there was no nontarget embolization seen on Rx⁹⁰Y PET/CT. The clinical and laboratory adverse events at 3 and 6 months after ⁹⁰Y are reported in Table 4. The embolic effect of scout⁹⁰Y is outlined in Appendix A (Fig E4, available online at www.jvir.org).

Table 4.

Frequencies and Percentages of Any Adverse Event Reported after Yttrium-90 Radioembolization at 3 and 6 Months

Toxicity	Frequency of any CTCAE
	3 mo	%	6 mo	%
Clinical toxicities
Fatigue	2	7	1	3
Abdominal pain	2	7	4	13
Ascites	1	3	2	7
Encephalopathy	1	3	1	3
Anorexia	0	0	0	0
Fever	0	0	0	0
Nausea	3	10	0	0
Vomiting	0	0	1	3
Death	1	3	1	3
Any CTCAE	6	20	4	13
Severe >3 CTCAE	0	0	1	3
Laboratory toxicities
Total bilirubin	5	17	3	10
ALT	1	3	2	7
AST	4	13	2	7
Alkaline phosphatase	5	17	1	3
Albumin	4	13	3	10
Creatinine	3	10	2	7
Hyponatremia	7	23	2	7
WBC	3	10	1	3
Anemia	3	10	2	7
Platelets	7	23	5	17
Any CTCAE	26	87	18	60
Severe >3 CTCAE	6	20	4	13

Note–The one patient death over the study period was secondary to transplantation complications.

ALT = alanine transaminase; AST = aspartate aminotransferase; CTCAE = Common Terminology for Clinical Adverse Events; WBC = white blood cell.

DISCUSSION

Personalized ⁹⁰Y dosimetry is critical for improving outcomes in patients with primary and metastatic liver cancers. This has led to the use of ^99mTc MAA as a surrogate marker to predict doses to the tumor, liver, and lung prospectively. However, several studies (5,6,9) have demonstrated limitations with ^99mTc MAA as a surrogate marker for Rx⁹⁰Y. For example, a study by Wondergem et al (6) comparing ^99mTc MAA SPECT and ⁹⁰Y bremsstrahlung SPECT/CT found a difference in activity of >10%, 20%, and 30% in 68%, 43%, and 32% of contoured segments, respectively. When examining activity per milliliter. In another comparison of more than 1,000 lesions of various pathologies by Ilhan et al (9), ^99mTc MAA SPECT/CT was found to have a weak correlation with ⁹⁰Y bremsstrahlung SPECT/CT TNR. A study by Villalobos et al (5) on 190 patients with HCC treated with glass microspheres showed that the MAA TNR had a moderate correlation with that of ⁹⁰Y on SPECT/CT, with a weak correlation in the segmentectomy subset and moderate correlation in lobar treatment. The result of the current study is concordant with previously published data by Villalobos et al (5), that overall, the MAA TNR has a moderate linear correlation with the Rx⁹⁰Y TNR.

There are multiple proposed mechanisms by which ^99mTc MAA falls short as a ⁹⁰Y surrogate. MAA has significant discrepancies in size, particle number, and morphology compared with ⁹⁰Y particles. Most (90%) MAA particle sizes range from 10 to 100 μm, with a mean particle diameter smaller than that of ⁹⁰Y microspheres (6). Smaller MAA particles may more easily pass through hepatic capillaries, resulting in overestimation of shunting and falsely depict nontumoral uptake (17). In addition, the significantly lower number of MAA particles delivered may not accurately simulate ⁹⁰Y flow dynamics (6). Finally, quality control issues related to the preparation of ^99mTc and albumin particles, including clumping, can also contribute to inaccuracies (18). Additionally, a potential cause of discrepancy between MAA and ⁹⁰Y in a large number of studies is the comparison of ^99mTc MAA SPECT/CT to ⁹⁰Y Bremsstrahlung SPECT/CT, which has a lower spatial resolution than that of ⁹⁰Y PET/CT (5,6,9). In this study, to eliminate the inherent inaccuracies of ⁹⁰Y SPECT/CT, the LSF and TNR of MAA SPECT/CT were compared with those of ⁹⁰Y PET/CT.

The limitations of ^99mTc MAA in accurately predicting ⁹⁰Y biodistribution have led to investigations of more accurate surrogates (14). Biomarkers with properties more similar to those of ⁹⁰Y have been proposed, such as resin or albumin microspheres labeled with positron-emitting isotopes, such as fluorine-18 and yttrium-86 (19,20). Rather than identifying other biosimilar markers, this study proposes the use of a bioidentical scout⁹⁰Y as a more optimal mapping radiotracer, similar to the Holmium-166 microsphere model (17). In addition to potentially superior dosimetry data, the use of ⁹⁰Y in lieu of other isotopes may be more convenient in centers that are already well adapted to handle and deliver ⁹⁰Y particles.

This study demonstrated increased accuracy and the predictive value of a scout⁹⁰Y when compared with ^99mTc MAA particles, with a stronger correlation of LSF, TNR, TD, and NTLD to those of Rx⁹⁰Y. Specifically, for nonsegmental therapies, MAA had no correlation with the mean TD or NTLD when compared with the actual corresponding doses by Rx⁹⁰Y, whereas scout⁹⁰Y demonstrated a very strong linear correlation for both. Accurate prospective dosimetry planning is more important in nonsegmental therapies than in segmentectomies to minimize toxic dose to the nontumoral liver. Hence, the results of this study suggest the need for the use of a more accurate surrogate for treatment planning, particularly in the nonsegmental setting.

Selection of the 0.56 GBq of activity of scout⁹⁰Y was based on the result of a phantom study (12). This dose assured that, even with a hypothetical LSF of 100%, 0.56 GBq of ⁹⁰Y would cause a 27 Gy dose to the lung, which is less than the recommended limit. In 2 patients, tumors in 2 segments were treated separately. In this subset of patients, 2 separate 0.37 GBq ⁹⁰Y scout activities were used with no visible difference in the quality of image on ⁹⁰Y PET/CT.

Other phantom studies have suggested accurate imaging of scout⁹⁰Y dose with activities as low as 0.10 GBq (21). The authors acknowledge that 0.56 GBq of activity could conceivably cause treatment-related liver toxicity in patients who may not be a further ⁹⁰Y candidate because of “failure” of mapping. However, this activity was chosen to ensure adequate imaging in this proof-of-concept study. Although this study had no patients with mapping failure, a larger multi-institutional study with at least 12 months follow-up will be necessary to evaluate the true toxicity effects in patients who were scouted but not treated with ⁹⁰Y.

This study has several additional limitations. The sample size is relatively small; hence, small differences may not be appreciated. The nuclear medicine physicians calculating the MAA and ⁹⁰Y dosimetry were not blinded, potentially introducing bias. Additionally, given the propensity of ^99mTc MAA to overestimate LSF (17), subtle differences between the biomarkers may have been appreciated with a larger cohort. A larger sample size would also be useful to validate the safety profile because no cases of extrahepatic shunting were noted in this study. Furthermore, MAA may be potentially embolic, particularly in a small volume infusion. Although the authors did not observe signs of embolization after MAA injection based on visual assessment of cone-beam CT before scout⁹⁰Y, this could potentially affect the TNR and LSF, particularly in segmentectomies. Another limitation is the TD calculations, which were calculated as the Rx⁹⁰Y in addition to Scout⁹⁰Y dosage, and no residual activities were taken into account. Lastly, the results of this resin scout⁹⁰Y needs to be replicated for glass ⁹⁰Y.

In conclusion, this study demonstrated that scout⁹⁰Y is a safe and feasible surrogate for ⁹⁰Y treatment planning in patients with HCC. It demonstrated superior predictive accuracy compared with ^99mTc MAA for TNR and LSF in both segmental and nonsegmental therapies. Additionally, scout⁹⁰Y demonstrated superior dosimetry prediction compared with MAA for nonsegmental therapies. Further larger randomized (MAA vs scout⁹⁰Y) studies with a longer-term follow-up period are needed to elucidate the clinical benefit of using scout⁹⁰Y instead of MAA.

RESEARCH HIGHLIGHTS.

• In this single-arm prospective trial in which 30 patients underwent concurrent radioembolization planning using scout yttrium-90 (⁹⁰Y) and technetium-99m microaggregated albumin, scout⁹⁰Y treatment had a significantly higher correlation with the therapeutic ⁹⁰Y tumor-to-normal ratio, lung shunt fraction, mean tumor dose, and nontumor liver dose in patients undergoing nonsegmental ⁹⁰Y therapy.

• There was no discrepancy in nontarget embolization between the scout⁹⁰Y and technetium-99m microaggregated albumin treatments, suggesting a similar short-term safety profile.

STUDY DETAILS.

Study type:

Prospective, observational, descriptive study

Level of evidence:

4 (SIR-C)

ABBREVIATIONS

CT

computed tomography

HCC

hepatocellular carcinoma

LSF

lung shunt fraction

MAA

macroaggregated albumin

⁹⁰Y

yttrium-90

^99mTc

technetium-99m

NTLD

nontumoral liver dose

PET

positron emission tomography

Rx⁹⁰Y

therapeutic ⁹⁰Y

TNR

tumor-to-normal liver ratio

TD

tumor dose

SPECT

single-photon emission computed tomography