The effectiveness of yttrium-90 selective internal radiation therapy (90Y-SIRT) in hepatocellular carcinoma may depend on more than the total radiation dose delivered to the tumour.
A study published in the European Journal of Nuclear Medicine and Molecular Imaging suggests that the spatial distribution of radiation within both the tumour and surrounding liver tissue may influence treatment response and toxicity after 90Y selective internal radiation therapy, or SIRT.
The findings indicate that different features of dose distribution may become important at different overall dose levels. While highly irradiated regions appeared to influence response in tumours receiving lower median doses, adequate coverage of less-irradiated regions became more relevant when the median dose was higher.
When the Average Dose Does Not Tell the Whole Story
90Y-SIRT is an established treatment for hepatocellular carcinoma. Its effectiveness depends on delivering sufficient radiation to the tumour while limiting exposure to non-tumoral liver tissue.
Individual treatment planning has traditionally focused largely on the mean or median absorbed dose. This approach assumes that radioactive microspheres are distributed relatively uniformly within the treated tissue.
In clinical practice, however, post-treatment 90Y PET/CT frequently shows substantial variation in radiation distribution. Tumour vascularity, perfusion, and microsphere deposition can result in some regions receiving very high doses while others receive considerably less.
The clinical importance of this heterogeneity has remained uncertain. Under-dosed areas may allow viable tumour cells to persist, but highly irradiated areas may also contribute to treatment response through dose-dependent biological effects within the tumour microenvironment.
The investigators therefore examined whether voxel-based measurements of dose distribution were associated with tumour response and severe toxicity following 90Y-SIRT.
A Voxel-by-Voxel View of Radiation Delivery
The retrospective study included 45 patients with hepatocellular carcinoma who underwent treatment with 90Y resin microspheres at the University Medical Centre Groningen between October 2021 and January 2025.
A total of 68 lesions were evaluated. Eligible lesions measured at least 20 mm and had appropriate contrast-enhanced imaging before treatment and approximately four months after therapy.
Radiation dose distributions were calculated using post-treatment 90Y PET/CT. Rather than evaluating only a single average dose, the researchers assessed several dose-volume histogram parameters.
Hot-spot measurements included D2, D10, and Dmax, representing radiation exposure within the most highly irradiated tumour regions. Cold-spot measurements included D90, D98, and Dmin, reflecting the dose received by the less-irradiated parts of the tumour.
The investigators also evaluated tumour coverage parameters, including V100, V150, and V200, which represented the proportion of tumour volume receiving at least 100, 150, or 200 Gy.
Dose heterogeneity was assessed using the interquartile range, the difference between maximum and minimum doses, and the coefficient of variation.
Tumour response was evaluated using modified Response Evaluation Criteria in Solid Tumors, or mRECIST. Severe adverse events were defined as grade 3 or higher and were classified as acute when they occurred within three months of treatment and latent when they developed later.
Complete Responses in More Than 40% of Lesions
Among the 68 treated lesions, 30, or 44.1%, achieved a complete response. An additional 18 lesions, or 26.5%, achieved a partial response.
Stable disease was reported in 16 lesions, representing 23.5% of the cohort, while four lesions, or 5.9%, developed progressive disease.
The lesions were divided into two equally sized groups according to the cohort’s median absorbed dose. Thirty-four lesions had a D50 below 128 Gy, while 34 had a D50 of at least 128 Gy.
The overall distribution of mRECIST response categories did not differ significantly between these two groups.
The median absorbed dose also did not differ significantly across the four individual response categories. However, when complete and partial responses were combined, responder lesions had a significantly higher D50 than lesions with stable or progressive disease.
Different Dose Levels, Different Drivers of Response
The relationship between dose heterogeneity and tumour response appeared to depend on the overall radiation dose delivered.
Among lesions with a D50 below 128 Gy, greater dose heterogeneity was associated with an improved response. Hot-spot measurements, including D2, D10, and Dmax, demonstrated similar patterns.
These findings suggest that when the overall tumour dose is relatively low, highly irradiated regions within the tumour may contribute meaningfully to treatment activity.
The pattern was different among lesions with a D50 of at least 128 Gy. In this group, cold-spot measurements, particularly D90 and D98, were significantly associated with a better response.
At higher median doses, treatment success may therefore depend less on achieving extremely high-dose regions and more on ensuring that the least-irradiated areas of the tumour still receive sufficient radiation.
The findings point to a dose-level-dependent relationship: hot spots may be more influential when the median dose is lower, while adequate minimum-dose coverage may become increasingly important once higher median doses are achieved.
Dose Heterogeneity and Treatment-Related Toxicity
The analysis also identified different associations between radiation exposure and the timing of severe toxicity.
Grade 3 or higher toxicity occurred in 40.0% of patients during the first three months following SIRT. Severe toxicity occurring more than three months after treatment was reported in 33.3%.
The mean radiation dose delivered to non-tumoral liver tissue was associated with acute toxicity. In contrast, spatial heterogeneity within the non-tumoral liver dose distribution was associated with latent severe toxicity.
This temporal distinction suggests that early and delayed liver injury may be influenced by different characteristics of radiation exposure.
A higher average dose to healthy liver tissue may contribute more directly to early injury, while an uneven radiation distribution may have greater relevance to severe toxicity that appears later.
At the time of analysis, 37.8% of patients had died. Most deaths were related to disease progression. Three treatment-related deaths were reported and were attributed to liver failure, variceal bleeding, and radiation-induced liver disease.
Median progression-free survival was 8.2 months. Median overall survival had not been reached after a median follow-up of 14.2 months.
Moving Toward Spatially Informed SIRT Planning
The study demonstrates that a single mean or median dose may not fully describe the biological effect of 90Y-SIRT.
Radiation delivery within a tumour is spatially complex. Two lesions receiving a similar median dose may have very different internal dose distributions, with one containing highly irradiated regions and another achieving more uniform coverage.
The results suggest that these differences may matter clinically. At lower dose levels, highly irradiated tumour regions may support response, while at higher dose levels, avoiding under-dosed areas may become more important.
Similarly, evaluating only the mean dose to non-tumoral liver tissue may not capture the full risk of delayed toxicity.
The investigators concluded that spatial dose distribution influences outcomes beyond mean absorbed dose. Their findings support further evaluation of voxel-based dosimetry as a tool for individualized SIRT planning, with the aim of improving tumour control while reducing treatment-related liver injury.
Written by Nare Hovhannisyan, MD
More articles like this are available on OncoDaily