The value of 99mTc-MAA SPECT/CT for lung shunt estimation in 90Y radioembolization: a phantom and patient study

Background: A major toxicity concern in radioembolization therapy of hepatic malignancies is radiation-induced pneumonitis and sclerosis due to hepatopulmonary shunting of Y-90 microspheres. Currently, Tc-99m macroaggregated albumin (Tc-99m-MAA) imaging is used to estimate the lung shunt fraction (LSF) prior to treatment. The aim of this study was to evaluate the accuracy/precision of LSF estimated from Tc-99m planar and SPECT/CT phantom imaging, and within this context, to compare the corresponding LSF and lung-absorbed dose values from Tc-99m-MAA patient studies. Additionally, LSFs from pre- and post-therapy imaging were compared. Results: A liver/lung torso phantom filled with Tc-99m to achieve three lung shunt values was scanned by planar and SPECT/CT imaging with repeat acquisitions to assess accuracy and precision. To facilitate processing of patient data, a workflow that relies on SPECT and CT-based auto-contouring to define liver and lung volumes for the LSF calculation was implemented. Planar imaging-based LSF estimates for 40 patients, obtained from their medical records, were retrospectively compared with SPECT/CT imaging-based calculations with attenuation and scatter correction. Additionally, in a subset of 20 patients, the pre-therapy estimates were compared with Y-90 PET/CT-based measurements. In the phantom study, improved accuracy in LSF estimation was achieved using SPECT/CT with attenuation and scatter correction (within 13% of the true value) compared with planar imaging (up to 44% overestimation). The results in patients showed a similar trend with planar imaging significantly overestimating LSF compared to SPECT/CT. There was no correlation between lung shunt estimates and the delay between Tc-99m-MAA administration and scanning, but off-target extra hepatic uptake tended to be more likely in patients with a longer delay. The mean lung absorbed dose predictions for the 28 patients who underwent therapy was 9.3 Gy (range 1.3-29.4) for planar imaging and 3.2 Gy (range 0.4-13.4) for SPECT/CT. For the patients with post-therapy imaging, the mean LSF from Y-90 PET/CT was 1.0%, (range 0.32.8). This value was not significantly different from the mean LSF estimate from Tc-99m-MAA SPECT/CT (mean 1.0%, range 0.4-1.6; p = 0.968), but was significantly lower than the mean LSF estimate based on planar imaging (mean 4.1%, range 1.2-15.0; p = 0.0002). Conclusions: The improved accuracy demonstrated by the phantom study, agreement with Y-90 PET/CT in patient studies, and the practicality of using auto-contouring for liver/lung definition suggests that Tc-99m-MAA SPECT/CT with scatter and attenuation corrections should be used for lung shunt estimation prior to radioembolization.