Modeling Clinical Relevance and Risk in Yttrium-90 Radioembolization of the Liver: Lung Shunt Fraction Variability...

To test the hypothesis that lung shunt fraction (LSF) in yttrium-90 (⁹⁰Y) radioembolization (RE) of the liver can be modeled for clinical relevance and risk on the basis of cancer type, tumor size, and imaging modality.

This was a retrospective analysis of 354 consecutive cases from a 34-month period at a single institution. Cancer type (hepatocellular carcinoma [HCC] vs non-HCC) and maximum tumor size were tracked. LSF was computed from planar imaging (LSF_planar) and 3 macroaggregated albumin-single photon emission computed tomography (SPECT)/computed tomography (CT) methods with different corrections at the lung/liver boundary (LSF_Full, LSF_2cm, and LSF_Left). Five tumor categories were analyzed: (a) HCC measuring <3 cm, (b) HCC measuring 3-8 cm, (c) HCC measuring >8 cm, (d) cases with transjugular intrahepatic portosystemic shunt (TIPS)/vascular invasion, and (e) non-HCC.

Median LSF was significantly different (all P < .001) among LSF_planar, LSF_Full, LSF_2cm, and LSF_Left methods, with values of 6.9%, 3.9%, 1.4%, and 0.8%, respectively. There were 72% and 19% of cases with LSF_planar of >5% and >10%, respectively, but only 7% and 2% using LSF_2cm. LSF was lower for HCC measuring <3 cm relative to HCC measuring 3-8 cm and >8 cm (P ≤ .003). In the <3-cm HCC group, there were no cases with LSF_planar > 20% or LSF_2cm > 5%. Cases with TIPS/vascular invasion maintained the highest median LSF (LSF_planar, 9.0%; LSF_2cm, 2.8%) and the highest percentages of cases with LSF values of >10% and >20% across all methods.

Despite significant variations with imaging modality and correction methods, consistent trends in LSF magnitude with cancer type and tumor size were still uncovered. The relative risk of observing clinically relevant LSF can be modeled for different liver tumor categories in ⁹⁰Y-RE.