Dosimetry Assessment in Predicting Treatment Outcomes Following Yttrium-90 Transarterial Radioembolization of Hepatic Tumors

Abstract

Purpose:

To evaluate the use of yttrium-90 (⁹⁰Y) dosimetry in predicting treatment outcomes when used following transarterial radioembolization with SIR-Spheres® (Resin ⁹⁰Y) in patients with hepatic tumors.

Materials and Methods:

This single institution retrospective analysis included 100 patients with hepatocellular carcinoma, colorectal carcinoma or other liver metastases who underwent transarterial radioembolization with resin ⁹⁰Y and had imaging follow-up within one year of treatment. Mean tumor dose and mean dose to nontumor was calculated using voxel-based dosimetry software. Descriptive statistics were reported and methods of analyses included simple and multivariable linear regression, contingency table analyses, Kaplan−Meier estimation, and Cox proportional hazards models.

Results:

Of 100 patients included, 65 demonstrated tumor shrinkage following transarterial radioembolization. Of these, 20 (30.8%) had hepatocellular carcinoma, 22 (33.8%) had colorectal carcinoma, and 23 (35.4%) had other types of metastases. There was an association between tumor shrinkage and mean tumor dose (p = 0.0285) and mean dose to nontumor (p = 0.0028) in hepatocellular carcinoma patients, but not colorectal carcinoma, or the other subgroup. For all 100 patients, time to death and mean tumor dose was associated only in the other subgroup (p = 0.0260), but not in the hepatocellular or colorectal carcinoma groups. Time to death and mean dose to nontumor was associated in hepatocellular carcinoma patients (p = 0.0421), but not the colorectal carcinoma or other subgroup.

Conclusions:

Voxel-based dosimetry assessment is a tool that may be utilized to assist in predicting treatment outcomes in responders to transarterial radioembolization.

Introduction

Transarterial radioembolization (TARE) is a treatment for unresectable liver tumors from primary hepatocellular carcinoma (HCC) and hepatic metastases. Treatment goals include slowing the progression of tumor growth, increasing survival, downsizing tumors, and providing symptomatic relief.¹ Surgical resection is possible in less than 20% with hepatic tumors, making advancements in TARE therapy paramount.²

Liver tissue is sensitive to the harmful effects of ionizing radiation, resulting in suboptimal outcomes with external beam radiation.³ TARE can deliver radiation to the cancer site with limited exposure to healthy liver parenchyma. Pretreatment dose prescription assessment for ⁹⁰Y TARE is most accurately calculated using a personalized partition model, which utilizes technetium (Tc)−99m labeled macroaggregated albumin (MAA) as a surrogate for ⁹⁰Y.⁴ This has demonstrated greater clinical utility compared to the body surface area method in its correlation with tumor-to-normal liver ratios (TNRs) and percent of tumor involvement.⁵ MAA mapping aids in calculating the ⁹⁰Y dose to reach therapeutic levels.³ MAA is not a perfect surrogate for ⁹⁰Y, meaning alternative methods to assess the dose delivered to the tumor and surrounding parenchyma in the immediate post-treatment period is warranted.

There is not an agreed method by which absorbed ⁹⁰Y is calculated following TARE. Instead, treatment outcomes are measured by imaging weeks to months following treatment and compared to baseline imaging according to guidelines for evaluation of tumor response such as mRECIST or RECIST 1.1.^6

–9 Although this evaluation is useful in long-term management, it delays the opportunity for immediate rectification of delivered radiation. Voxel-based dosimetry assessment software attempts to address this issue by utilizing post-treatment Bremsstrahlung emission computerized tomography (BECT) to calculate the dose of ⁹⁰Y absorbed by the tumor and liver parenchyma. Dosimetry assessment offers the potential for the physician to repeat TARE if the therapeutic dose was not achieved.

This study aimed to evaluate the use of dosimetry software in predicting treatment outcomes when used in the post-treatment 3D dosimetry assessment following ⁹⁰Y TARE in patients receiving resin ⁹⁰Y with primary HCC or hepatic metastases.¹⁰ The primary outcome was whether an association exists between the dosimetry assessment and tumor response. Secondary outcomes included whether an association exists between software dosimetry assessments and patient survival and damage to healthy liver parenchyma.

Materials and Methods

Study design and patient population

Institutional Review Board (IRB) approval was obtained and this study was conducted in compliance with institutional and national research ethics standards. This single center retrospective cohort study included patients who were 18 years or older, diagnosed with HCC or liver metastases, underwent TARE with resin ⁹⁰Y SIR-Spheres® between June 2015 and May 2019, and had pretreatment and at least one post-treatment CT or MRI imaging available. Patients were excluded from the study if they: received transarterial chemoembolization (TACE) or microwave ablation (MWA) for the lesion of interest before TARE, had missing treatment data or imaging records, underwent split-dose treatment (i.e., ⁹⁰Y delivered in multiple infusions), or had complex lesions that could not be measured using RECIST 1.1 or mRECIST guidelines.

Clinical data collection

Patient demographics, clinical characteristics, and treatment history were collected and managed using Research Electronic Data Capture.¹¹ The data collected included age, gender, ethnicity, race, BMI, date of diagnosis, primary pathology, presence of chronic liver disease, prior local or systemic treatment, and Eastern Cooperative Oncology Group (ECOG) status.¹² Laboratory values were collected along with dialysis and encephalopathy status from the closest time point prior to treatment and the closest time point at one-year post-treatment. The treatment information included date of treatment, lobe treated, percent lung shunting, prescribed activity (GBq), received activity (GBq), activity delivered (GBq and %), and residual activity in delivery apparatus (GBq). Patient outcomes included post-treatment hepatic resection, TACE, MWA, date of last contact, and date of death.

Response assessment

Imaging assessments were completed using Picture Archiving and Communications System (PACS).¹³ If a patient had multiple hepatic lesions, the largest was identified as the target. If the target lesion had received prior MWA or TACE, the next largest lesion was selected. The involved liver segment(s) was recorded, and TARE treatment(s) were identified for subsequent longitudinal analysis by dosimetry software where patients had multiple treatments on the target lesion. For each baseline and follow-up image, these variables were recorded: date of scan, imaging modality, percentage of liver replaced by tumor, presence of bilobar disease, presence of vascular invasion, presence of extrahepatic involvement, severity of ascites, and three dimensions for volumetric calculations.

Tumor shrinkage was evaluated according to mRECIST criteria (for HCC) or RECIST 1.1 (for nonHCC),^7,9 which defines shrinkage as at least a 30% decrease in the sum of diameters of viable (enhancing) target lesions, taking the baseline sum of diameters as the reference. Complete shrinkage was defined as the disappearance of any intratumoral arterial enhancement in all target lesions.

Dosimetric analysis

Radiation dosimetry assessment was conducted using Varian Velocity 4.0 RapidSphere™ software. Baseline CT or MRI imaging and immediate post-treatment ⁹⁰Y Bremsstrahlung SPECT-CT (BECT) were imported into the software. Liver and tumor margins were defined on the baseline images using the software’s contouring function, which filled liver and tumor volumes based on defined margins. Baseline liver margins were utilized to register the baseline images with the BECT images, and tumor margins and volume were transferred from the baseline to the BECT images. The rationale for performing imaging registration and tumor propagation was related to the poor resolution of BECT images, which prevents direct visualization of the tumor. The liver margins on the BECT images were subsequently defined and the volumes filled utilizing the contouring function. The propagated tumor volume and defined liver volume on the BECT images were used to run the dosimetry assessment. Received activity and percent lung shunting were defined as input entities for software calculations. The software’s dosimetry assessment outputs included a heat map denoting radiation dose distribution within the liver volume, a dose volume histogram comparing lesion and liver, and the minimum, mean, and maximum radiation dose absorbed by the lesion and liver (Fig. 1).

FIG. 1.

Dosimetry assessment output data. (A) Example of radiation dose heat map of liver volume relative to tumor volume. (B) Example of line graph with percent volume against radiation dose comparing liver (red) and lesion (pink).

Statistical analysis

Descriptive statistics were reported using means and standard deviation for continuous variables and frequency counts for categorical variables. Simple and multiple linear regression models were used to study the relationship between outcomes and predictors. In case of multiple linear regression, models with interaction effects and higher order quadratic terms were tried and the results from the most parsimonious models were reported. The Kaplan−Meier estimation method was used to generate survival curves for modeling time to death using a logrank test. A semiparametric Cox proportional hazards model was used to study the effect of mean dose to tumor and time to death. Regression diagnostic checks included assessing the normality of residuals for simple and multiple linear regression, and separately, the plot of Schoenfeld residuals (for continuous covariates) and log−log survival plots (for categorical covariates) for testing the proportional hazards assumption in the Cox model. TNR were calculated using Young et al.’s formula: treated lesion mean absorbed tumor dose divided by mean dose to nontumor tissue.¹⁴ Model for end stage liver disease (MELD) scores were calculated using the Organ Procurement and Transplantation Network (OPTN) policy.¹⁵ A Pearson’s correlation coefficient was used to assess the correlation between MELD scores versus mean dose to nontumor (MNTD). In addition, exploratory methods of analyses included the use of contingency tables, one way analysis of variance, and Kruskal−Wallis test. All analyses were done using the SAS 9.4 software and statistical significance was assessed at the 5% level of significance.

Results

Patient, treatment, and imaging characteristics are summarized in Tables 1 − 3. There was no evidence that after adjusting for age and gender, the association between shrinkage percent and mean tumor dose (MTD) changed by tumor type (p = 0.4106). Additionally, the distribution of tumor shrinkage for all 100 subjects did not differ by tumor type (p = 0.9867) nor when adjusted to include only the 65 that showed shrinkage (p = 0.3439).

Table 1.

Demographics and Medical History at Baseline (n = 100)

		Mean ± SD
Age at time of diagnosis		65 ± 11
BMI at time of first treatment		28.0 ± 5.8
MELD		9.8 ± 4.7
Child-pugh		6.0 ± 1.2

		n, %
Gender
Male		55, 55
Female		45, 45
Ethnicity
Hispanic/Latino		6, 6
NonHispanic/Latino		94, 94
Race
White		81, 81
Black		9, 9
American Indian/Alaskan		1, 1
Unknown		9, 9
Primary tumor
Colorectal cancer		34, 34
Hepatocellular carcinoma		28, 28
Neuroendocrine cancer		12, 12
Cholangiocarcinoma		8, 8
Pancreatic cancer		5, 5
Breast cancer		3, 3
Other		6, 6
Unknown		4, 4
Chronic liver disease^a
None		66, 66
Alcoholic liver disease		7, 21
Nonalcoholic fatty liver disease		3, 9
Cirrhosis		31, 91
Hepatitis C infection		19, 56
Other		5, 15
Treatment prior to radioembolization
Chemotherapy		67, 67
Transarterial chemoembolization		17, 17
Microwave ablation		14, 14
Bland embolization		5, 5
Hepatic resection		10, 10
ECOG
0		31, 31
1		59, 59
2		7, 7
3–5		0, 0
Unknown		3, 3

Chronic liver disease percentages do not sum to 100% as some patients had concomitant conditions.

BMI, body mass index; ECOG, Eastern Cooperative Oncology Group; MELD, model for end stage liver disease.

Table 2.

⁹⁰Y TARE Treatment Information

	Number of patients, n, %
Number of treatments
1	33, 33
2	60, 60
3	5, 5
4	2, 2
Liver segment(s) involved by lesion of interest^a
I	5, 5
II	13, 13
III	13, 13
IVA	18, 18
IVB	16, 16
V	18, 18
VI	21, 21
VII	27, 27
VIII	29, 29

	Mean ± SD
Time from diagnosis to first treatment (months)	15 ± 18
Radiation activity
Prescribed (GBq)	1.13 ± 0.39
Delivered (GBq)	1.11 ± 0.41
Delivered (%)	93.0 ± 11.8
Lung shunting	5.62 ± 4.45

Liver segment percentages do not sum to 100% as some patients had multiple segments involved.

SD, standard deviation; TARE, transarterial radioembolization; ⁹⁰Y, yttrium-90.

Table 3.

Baseline and Follow-Up Imaging Assessments

Imaging assessments	Baseline (n = 100)	Follow-Up 1 (n = 100)	Follow-Up 2 (n = 67)	Follow-Up 3 (n = 39)	Follow-Up 4 (n = 21)	Follow-Up 5 (n = 9)
Time from treatment (months), mean ± SD	−2.0 ± 1.3	2.3 ± 1.1	5.0 ± 2.0	7.6 ± 2.3	11.4 ± 8.3	10.7 ± 2.0
Tumor volume (cm³), mean ± SD
Total (n = 100)	166 ± 407	138 ± 284	147 ± 420	50 ± 92	75 ± 104	72 ± 113
Responders (n = 65)	180 ± 477	101 ± 253	126 ± 456	40 ± 64	67 ± 108	15 ± 118
Nonresponders (n = 35)	140 ± 231	206 ± 327	202 ± 310	84 ± 151	99 ± 108	241
mRECIST, n, %
CR	NA	3, 3	4, 6.0	1, 2.6	1, 4.8	0, 0
PR	NA	9, 9	0, 0	0, 0	0, 0	0, 0
SD	NA	11, 11	8, 11.9	6, 15.4	2, 9.5	1, 11.1
PD	NA	5, 5	3, 4.5	1, 2.6	1, 4.8	0, 0
RECIST, n, %
CR	NA	0, 0	1, 1.5	0, 0	0, 0	0, 0
PR	NA	13, 13	5, 7.5	5, 12.8	1, 4.8	1, 11.1
SD	NA	41, 41	34, 50.7	14, 35.9	11, 52.4	5, 55.6
PD	NA	17, 17	11, 16.4	12, 30.8	5, 23.8	2, 22.2
Percent of liver replaced by tumor, n, %
<25%	54, 54	50, 50	39, 58.2	25, 64.1	13, 61.9	5, 55.6
25–50%	38, 38	34, 34	20, 29.9	11, 28.2	6, 28.6	3, 33.3
>50%	8, 8	14, 14	8, 1.9	3, 7.7	2, 9.5	1, 11.1
Bilobar disease present, n, %	70, 70	75, 75	52, 77.7	30, 76.9	14, 66.7	5, 55.6
Vascular invasion present, n, %	19, 19	20, 20	11, 16.4	5, 12.8	4, 19.0	3, 33.3
Extrahepatic involvement present, n, %	43, 43	48, 48	33, 49.3	19, 48.7	10, 47.6	5, 55.6
Ascites, n, %
None	81, 81	62, 62	42, 62.7	22, 56.4	13, 61.9	4, 44.4
Mild	15, 15	27, 27	20, 29.9	12, 30.8	7, 33.3	3, 33.3
Moderate	3, 3	8, 8	4, 6.0	3, 7.7	1, 4.8	1, 11.1
Severe	1, 1	2, 2	1, 1.5	2, 5.1	0, 0	1, 11.1

CR, complete response; PR, partial response; PD, progressive disease; SD, stable disease; SD, standard deviation.

Table 4 outlines tumor response by dosimetry assessments for each cancer subtype. There was no statistically significant difference in MTD means across the three disease subgroups (p = 0.4445). There was also no statistically significant difference in all patients or in any disease subgroup.

Table 4.

Tumor Response and Dosimetry Assessments

Disease type	mRECIST &RECIST 1.1category	Number of patients (n = 99)^a n, %	MTD (Gy)mean, ± SD	MTD p-value	MNTD (Gy)mean, ± SD	MNTD p-value	TNRmean, ± SD	TNR p-value
All		100, 100	59.6, 32.9		29.7, 11.1		2.5, 4.2
	CR& PR	25, 25.2	66.7, 31.7	0.2213	31.9,10.7	0.2618	2.2, 1.2	0.7064
	SD & PD	74, 74.8	57.4, 33.2	0.2213	29.0,11.3	0.2618	2.6, 4.8	0.7064
HCC (mRECIST)		28, 28	56.7, 34.3		28.9, 12.2		2.0, 1.2
	CR& PR	12, 42.9	67.3, 36.1	0.1589	34.3, 13.2	0.0405	2.1, 1.4	0.6896
	SD & PD	16, 57.1	48.7, 31.6	0.1589	24.9, 9.9	0.0405	1.9, 1.1	0.6896
CRC (RECIST1.1)		34, 34	55.9, 29.8		29.9, 11.2		1.8, 0.8^b
	CR& PR	7, 20.6	59.1, 32.6	0.7538	29.2, 9.1	0.8558	2.0, 0.7	0.5707
	SD & PD	27, 79.4	55.1, 29.6	0.7538	30.0, 11.8	0.8558	1.8, 0.8^b	0.5707
Other (RECIST1.1)		37, 37	64.9, 34.6		30.2, 10.3		2.3, 1.1
	CR& PR	6, 16.28	74.6, 23.0	0.4986	30.2, 5.7	0.9901	2.6, 1.0	0.5620
	SD & PD	31, 81.6	63.9, 36.6	0.4986	30.2, 11.3	0.9901	2.2, 1.2	0.5620

One image assessment is missing.

One extreme outlier was removed.

CR, complete response; CRC, colorectal carcinoma; HCC, hepatocellular carcinoma; MNTD, mean dose to nontumor; MTD, mean tumor dose; PR, partial response; PD, progressive disease; SD, stable disease; SD, standard deviation; TNR, tumor-to-normal liver ratios.

There was no statistically significant difference in MNTD across the three disease subgroups (p = 0.8966). The only subgroup to show a statistically significant difference was in the 28 patients with HCC, where MNTD for the complete response (CR)/partial response (PR) group (mean 34.3, standard deviation [SD] 13.2) was different than MNTD for the stable disease (SD)/progressive disease (PD) group (mean 24.9, SD = 9.9) (p = 0.0405). There was no statistically significant difference in TNR across the three disease subgroups (p = 0.2093). There was also no statistically significant difference in TNR overall nor in any subgroup.

Of the 100 patients included, 65 showed tumor shrinkage following treatment with a mean tumor volume reduction of 48.5% (SD = 29.2). There was an association between shrinkage and MTD in the 65 responders (p = 0.0346). After adjusting for age, gender, and tumor type, the association between shrinkage and MTD was still statistically significant (p = 0.0479).

Table 5 outlines tumor shrinkage by dosimetry assessments for each cancer subtype. When looking at patients with shrinkage, there was an association between MTD and shrinkage (p = 0.0360). When looking at subgroups, 20 patients with HCC showed an association between MTD and shrinkage (p = 0.0285). For every 50 unit increase in mean dose absorbed by tumor, tumor volume reduced 20.2%. There was no association in colorectal carcinoma (CRC) or the other subgroup (Fig. 2).

FIG. 2.

Scatterplot showing association between tumor shrinkage and MTD in CRC, HCC, and other type of cancers. There was no association between CRC and other types of cancer, but there was a statistically significant association between tumor shrinkage and MTD in HCC patients (p = 0.0285). CRC, colorectal carcinoma; HCC, hepatocellular carcinoma; MTD, mean tumor dose.

Table 5.

Tumor Shrinkage and Dosimetry Assessments by Cancer Subtype

Disease type	Number of patients (n = 65) n, %	Mean decrease in tumor volume,%, ±SD	MTD (Gy),mean, ± SD	MTD p-value	MNTD (Gy)mean, ± SD	MNTD p-value	TNRmean, ± SD	TNR p-value
All	65, 100	48.5, 29.2	63.1, 34.3	0.0360	29.1, 10.7	0.0009	2.8, 5.1	0.9420
HCC	20, 30.8	40.9, 30.2	56.7, 34.3	0.0285	28.9, 12.2	0.0028	2.1, 1.3	0.8927
CRC	22, 33.8	49.4, 29.3	55.9, 29.8	0.9433	29.9, 11.2	0.3390	1.9, 0.7	0.9917
Other	23, 35.4	54.2, 27.9	64.9, 34.6	0.3755	30.2, 10.3	0.0751	2.6, 1.2	0.7438

CRC, colorectal carcinoma; HCC, hepatocellular carcinoma; MNTD, mean dose to nontumor; MTD, mean tumor dose; SD, standard deviation; TNR, tumor-to-normal liver ratio.

After adjusting for age, gender, and tumor type, the association between shrinkage and MNTD was statistically significant (p = 0.0014). After adjusting for age and gender, for every 10 unit increase in MNTD, expected tumor shrinkage increased by 11.0%. No other multivariate analyses among age, gender, and tumor type were statistically significant. Additionally, MTD and MNTD were moderately correlated with a correlation coefficient of 0.42 (p < 0.0001).

There was an association between MNTD and shrinkage when looking at all shrinkage patients (p = 0.0009). For the 20 patients with HCC who experienced tumor shrinkage, there was also an association between MNTD and shrinkage (p = 0.0028). For every 10 unit increase in mean dose absorbed by the nontumor, tumor shrinkage increased 18.1%. There was no association in CRC or the other subgroup (Fig. 3). There was no association between TNR and shrinkage for all patients nor for any subgroup.

FIG. 3.

Scatterplot showing association between tumor shrinkage and MNTD in CRC, HCC, and other type of cancers. There was no association between CRC and other types of cancer, but there was a statistically significant association between tumor shrinkage and MNTD in HCC patients (p = 0.0028). CRC, colorectal carcinoma; HCC, hepatocellular carcinoma; MNTD, mean dose to nontumor.

Post-treatment laboratory values were strongly associated with historical laboratory values. After adjusting for historical laboratory values, there was no association between any post-treatment laboratory values and MNTD. The same was true for Child-Pugh scores (rho = −0.225, p = 0.0717). MELD scores, however, were negatively correlated with MNTD (r = −0.277, p = 0.0254) (Fig. 4).

FIG. 4.

Scatter plot showing correlation between MNTD and MELD scores. MNTD, mean dose to nontumor; MELD, model for end stage liver disease.

Seventy-six patients died during the study period. Survival analysis from time of diagnosis to death showed a median time to death of 747 d (2.05 years). There was no association between time to death and tumor type for all 100 (p = 0.3111) (Fig. 5), nor for the 65 alone that showed shrinkage (0.1742) (Fig. 6).

FIG. 5.

Kaplan−Meier plot by tumor type for all 100 subjects. There was no association between time to death and tumor type.

FIG. 6.

Kaplan−Meier plot by tumor type for 65 subjects that showed tumor shrinkage. There was no association between time to death and tumor type.

There was no association between time to death and MTD in HCC patients (p = 0.1351), nor CRC patients (p = 0.2512). There was an association however, between time to death and MTD in the other subgroup (p = 0.0260). For every 50 unit increase in radiation dose absorbed by the tumor, instantaneous risk of death decreased by 53.0%. There was an association between time to death and MNTD in HCC patients (p = 0.0421). For every 10 unit increase in radiation dose absorbed by the nontumor, death decreased by 37.6%. There was no association between time to death and MNTD in CRC patients (p = 0.9915), nor other patients (p = 0.2954). There was no association between time to death and TNR in HCC (p = 0.6949), CRC (p = 0.1671) nor TNR patients (p = 0.1064).

Discussion

This study found tumor response and MTD were not associated in any disease subgroup. The same was true for TNR. The only associated outcome with tumor response was MNTD in the HCC group. This study also found an association in responders with HCC between MTD and shrinkage and MNTD and shrinkage. Toxicity showed an association only between MELD and MNTD. Lastly, time to death and MTD were associated in the other cancer subtypes and time to death and MNTD were associated in HCC patients.

Potrebko et al. commented on this software’s potential efficacy by describing three cases: one of complete tumor response, one of partial tumor response, and one of extrahepatic toxicity.¹⁶ However, this limited case series lacked sufficient data for formal statistical analysis and did not comment on patients whose tumor grew following ⁹⁰Y TARE treatment.

While results of this analysis were not consistent among all patient subpopulations, the findings favor the continued investigation of this software. The association between shrinkage and MTD and MNTD in HCC responders correlates with previous studies that show HCC tumors are sensitive to ⁹⁰Y TARE treatment.^17

–20 The lack of association in the CRC and other subgroup may be attributed to differences in tumor biology, software adherence, or dosing parameters. Additionally, variances in associations between survival and MTD, MNTD, and TNR may be due to different survival outcomes among cancer subtypes. Future studies should focus on cancer-specific subtypes to establish associations between response, survival, and dosimetric software outputs.

The lack of association between TNR and tumor response in the cancer subtypes is not easily explained. Given that the TNR calculation included both MTD and MNTD, one would expect associations within the HCC group to hold. Different TNR calculation methods might have shown consistent associations, which warrants further exploration.¹⁴

In this study, BECT was utilized due to its availability and established use in clinical practice. However, it is known to have limitations in image resolution and dosimetric precision.²¹ This could, at least partly, explain why dose-response relationships were observed in HCC but not in colorectal or other cancer subtypes. In contrast, ⁹⁰Y PET has been shown to provide higher spatial resolution and more accurate voxel-based dosimetry, leading to stronger correlations in dose response studies across various tumor types.²²

MNTD did not correlate with damage to healthy hepatic parenchyma, as evidenced by the lack of association with post-treatment laboratory values and Child-Pugh scores. However, an association was observed when analyzing MELD scores. Both Child-Pugh and MELD use bilirubin and INR, but differ on other factors they consider, which may explain these findings. ⁹⁰Y may have a greater impact on the variables that are part of MELD calculations. If future studies continue to validate voxel-based dosimetry software, these finding may be reassuring for the preservation of liver function and suggest that more aggressive TARE therapy could optimize treatment efficacy.

This study has several limitations. First, the predominance of white race and non-Hispanic/Latinx ethnicity may reflect disparities in access to TARE. Future research should aim to identify and address these barriers. Second, this study also lacked an adequate sample size for expansive subpopulation analyses. Of particular interest would have been analyzing dosimetry assessments between individual cancer subtypes instead of grouping them into one category. Third, the study was limited by focusing on the largest measurable lesion per patient for dosimetric analysis, which may not fully capture the dose response relationship in patients with multiple significant lesions. Conducting lesion-level analyses in future studies could provide a more comprehensive assessment of dosimetry and improve the understanding of treatment outcomes. Additionally, this study did not examine survival among groups based on overall patient-level doses. Fourth, the use of BECT for dosimetric analysis, while practical, has inherent limitations compared to ⁹⁰Y-PET imaging, which may impact dose precision and study outcomes. Finally, to increase internal validity, patients who received split-dose infusions were excluded. These patients may have had more complex or disseminated liver disease, which could affect study results. Future studies could include this patient population if all liver lesions are analyzed, rather than focusing on a single lesion of interest.

Conclusions

This study demonstrated the utility of the voxel-based dosimetry assessments in predicting treatment outcomes, particularly regarding survival and tumor shrinkage in the immediate post-treatment period. The potential impact of this software in clinical practice may increase if predictive factors to identify responder and nonresponder subpopulations are discovered.

Footnotes

Acknowledgments

The authors would like to thank the following individuals for their technical support with Picture Archiving and Communication Systems (PACS) and Velocity 4.0 RapidSphere^TM software: Miranda Leiker, James Lee Ingalls III, Harley Creek, a Radiology IT Clinical Analyst, and Tony Chilcoat, a System Administrator II. The authors also thank radiation safety program coordinator, Stefanie Smith, and the Radiation Safety Office at the University of Kansas Medical Center for assistance in obtaining radiation treatment data. The authors would like to thank the following for their help in collecting additional data for calculating Child-Pugh and MELD scores: Peyton Ackerman.

Authors’ Contributions

Z.C., C.W., and M.P.: Study concepts and design. C.W.: Literature research. C.W., W.W., J.M., I.K., G.W., and A.D.: Data collection. M.P.: Statistical analysis. C.W., S.S., M.P., C.W., A.R., and Z.C.: Article preparation and editing.

Data Availability Statement

All data generated or analyzed during this study are included in the article. Further inquiries can be directed to the corresponding author.

Disclosure Statement

The authors of this presentation have research support from a Sirtex Medical Limited grant. Z.C., serves as a proctor and speaker for Sirtex Medical Limited.

This study did not require IRB approval as it involved only a retrospective review of deidentified patient data, with no direct patient contact or intervention, and met criteria for exemption under applicable ethical guidelines for human subjects research.

Funding Information

This study was funded by Sirtex Medical Limited grant R19-032-AMER. The funder provided guidance on the design of the study, but no other direct input in the collection, analysis, interpretation, or writing-up of the data.

References

About SIRT. Available from: https://www.sirtex.com/us/clinicians/about-sirt/ [Last accessed: July 4, 2021 ].

Salem

, Thurston

. Radioembolization with 90Yttrium microspheres: A state-of-the-art brachytherapy treatment for primary and secondary liver malignancies. J Vasc Interv Radiol, 2006; 17(8):1251–1278; doi: 10.1097/01.RVI.0000233785.75257.9A

Kennedy

, Nag

, Salem

, et al. Recommendations for radioembolization of hepatic malignancies using Yttrium-90 microsphere brachytherapy: A consensus panel report from the radioembolization brachytherapy oncology consortium. Int J Radiat Oncol Biol Phys, 2007; 68(1):13–23; doi: 10.1016/j.ijrobp.2006.11.060

Kim

, Cohalan

, Kopek

, et al. A guide to 90Y radioembolization and its dosimetry. Phys Med, 2019; 68:132–145; doi: 10.1016/j.ejmp.2019.09.236

Venkatanarasimha

, Gogna

, Tong

KTA

, et al. Radioembolisation of hepatocellular carcinoma: A primer. Clin Radiol, 2017; 72(12):1002–1013; doi: 10.1016/j.crad.2017.07.021

Miller

, Hoogstraten

, Staquet

, et al. Reporting results of cancer treatment. Cancer, 1981; 47(1):207–214; doi: 10.1002/1097-0142(19810101)47:1<207::AID-CNCR2820470134>3.0.CO;2-6

Therasse

, Arbuck

, Eisenhauer

, et al. New guidelines to evaluate the response to treatment in solid tumors. J Natl Cancer Inst, 2000; 92(3):205–216; doi: 10.1093/jnci/92.3.205

Bruix

, Sherman

, Llovet

, et al; EASL Panel of Experts on HCC. Clinical management of hepatocellular carcinoma. conclusions of the Barcelona-2000 EASL conference. J Hepatol, 2001; 35(3):421–430; doi: 10.1016/S0168-8278(01)00130-1

Lencioni

, Llovet

. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Semin Liver Dis, 2010; 30(1):52–60; doi: 10.1055/s-0030-1247132

10.

Fox

. Velocity RapidSphereTM: Image-Guided Dosimetry for ⁹⁰Y Microsphere Therapy. Webinar presented at: My Varian: Webinars. Varian Medical Systems; April 13, 2018. Available from: https://www.myvarian.com/s/mvwebsummary?Id=a0O4400000t0dKdEAI&lang=en [Last accessed: July 4, 2021 ].

11.

Harris

, Taylor

, Thielke

, et al. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42(2):377–381; doi: 10.1016/j.jbi.2008.08.010

12.

Buccheri

, Ferrigno

, Tamburini

. Karnofsky and ECOG performance status scoring in lung cancer: A prospective, longitudinal study of 536 patients from a single institution. Eur J Cancer, 1996; 32A(7):1135–1141; doi: 10.1016/0959-8049(95)00664-8

13.

Strickland

. Current topic: PACS (picture archiving and communication systems): filmless radiology. Arch Dis Child, 2000; 83(1):82–86; doi: 10.1136/adc.83.1.82

14.

Young

, Chen

, Flanagan

, et al. Realized tumor to normal ratios in hepatocellular carcinoma patients undergoing transarterial radioembolization: A retrospective evaluation. Eur Radiol, 2022; 32(6):4160–4167; doi: 10.1007/s00330-021-08501-9

15.

OPTN/UNOS Policy Notice. Clerical Changes for Implementation of Adding Serum Sodium to the MELD Score. Available from: https://optn.transplant.hrsa.gov/media/1575/policynotice_20151101.pdf

16.

Potrebko

, Shridhar

, Biagioli

, et al. SPECT/CT image-based dosimetry for Yttrium-90 radionuclide therapy: Application to treatment response. J Appl Clin Med Phys, 2018; 19(5):435–443; doi: 10.1002/acm2.12400

17.

Lewandowski

, Gabr

, Abouchaleh

, et al. Radiation segmentectomy: Potential curative therapy for early hepatocellular carcinoma. Radiology, 2018; 287(3):1050–1058; doi: 10.1148/radiol.2018171768

18.

Padia

, Johnson

, Horton

, et al. Segmental Yttrium-90 radioembolization versus segmental chemoembolization for localized hepatocellular carcinoma: Results of a single-center, retrospective, propensity score-matched study. J Vasc Interv Radiol, 2017; 28(6):777–785.e1; doi: 10.1016/j.jvir.2017.02.018

19.

Biederman

, Titano

, Korff

, et al. Radiation segmentectomy versus selective chemoembolization in the treatment of early-stage hepatocellular carcinoma. J Vasc Interv Radiol, 2018; 29(1):30–37.e2; doi: 10.1016/j.jvir.2017.08.026

20.

Padia

, Kwan

, Roudsari

, et al. Superselective yttrium-90 radioembolization for hepatocellular carcinoma yields high response rates with minimal toxicity. J Vasc Interv Radiol, 2014; 25(7):1067–1073; doi: 10.1016/j.jvir.2014.03.030

21.

Elschot

, Vermolen

, Lam

MGEH

, et al. Quantitative comparison of PET and Bremsstrahlung SPECT for imaging the in vivo yttrium-90 microsphere distribution after liver radioembolization. PLoS One, 2013; 8(2):e55742; doi: 10.1371/journal.pone.0055742

22.

Lhommel

, van Elmbt

, Goffette

, et al. Feasibility of 90Y TOF PET-based dosimetry in liver metastasis therapy using SIR-Spheres. Eur J Nucl Med Mol Imaging, 2010; 37(9):1654–1662; doi: 10.1007/s00259-010-1495-9