Abstract
2807
Introduction: Selective internal radiation therapy (SIRT) with 90Y radioembolization aims to selectively irradiate liver tumors by administering radioactive microspheres through hepatic arteries, while sparing healthy tissue. The hypothesis is that the pre-therapy injection of 99mTc labeled macro-aggregated albumin (Tc-MAA) provides an estimation of the 90Y microspheres biodistribution, which is not always the case. There is a growing interest in personalised radionuclide therapy with an increasing number of clinical trials striving to unify protocols and mitigating uncertainties. In this context, a robust relationship between the delivered and pre-treatment radiation doses is required. In this work, we aim to investigate the predictive value of dose metrics calculated from Tc-MAA compared to those obtained from 90Y post-therapy SPECT/CT.
Methods: Eight patients with hepatocellular carcinoma (HCC) who underwent 90Y SIRT with glass microspheres were retrospectively included in this study. A total of eight lesions were analyzed. SPECT/CT imaging was performed on a Symbia T Series camera after simulation and therapy. Treatment planning was designed based on single partition model. Lesion and lobar segmentation were manually performed by an experienced nuclear medicine physician on the diagnostic images (contrast-enhanced CT or MRI). Simulation and therapy SPECT/CT images were registered on the diagnostic images using rigid transformation. 3D voxel-level dosimetry based on self-calibration and local energy deposition approach was conducted, where the mean absorbed dose (MAD) and dose volume histograms (DVHs) were computed for three regions of interest: tumoral liver (TL), non-tumoral liver target (NTLt, healthy liver within the perfused lobe) and non-tumoral whole liver (NTLw, whole healthy liver). From the DVHs, we calculated D50 and D70 for tumoral and non-tumoral liver (minimum dose received by 50% and 70% of the ROI), V120 of tumoral and V20 of non-tumoral liver (percentage of volume receiving at least 120 and 20Gy, respectively). Pearson’s correlation coefficient of these metrics between simulation and therapy was analyzed.
Results: The mean tumour dose estimated from Tc-MAA was 228.8 ± 139 Gy whereas it was about 157.9 ± 78 Gy from 90Y with Pearson’s correlation coefficient of 0.84 (p-value < 0.01). Smaller lesions present higher MAD values compared to larger lesions, but with a grater uncertainty of dose prediction. Consistent with the literature, in NTLt and NTLw, the values of MAD from Tc-MAA (66.6 ± 35.7 Gy and 40.6 ± 20.2 Gy) and 90Y (67.2 ± 28.8 Gy and 45.5 ± 18 Gy) showed high correlation (0.98, p-value < 0.001; and 0.95, p-value < 0.001; respectively). It is also shown that the tumor volume is correlated with DVH parameters, since larger volumes present lower D70 values, and this trend has also been observed for V120 and D50. Besides, D50, D70 and V120 derived from TL DVH illustrated good agreement between Tc-MAA and 90Y dosimetry, but according to Pearson’s factor, is higher for V120 (0.94, p-value < 0.001) than for D50 (0.85, p-value < 0.01) and D70 (0.82, p-value < 0.02). For NTLt and NTLw, the correlation factor of D50 (0.96, p-value < 0.001 and 0.97, p-value < 0.001) and D70 (0.97, p-value < 0.001 and 0.93, p-value < 0.001) present larger values compared to V20 (0.89, p-value < 0.005 and 0.90, p-value < 0.005).
Conclusions: Mean absorbed doses between Tc-MAA and 90Y show high correlation for NTLt and NTLw. For large tumors, there is good agreement between MAD estimated from simulation and therapy, while small tumors showed larger deviations. These results agree with other studies comparing dose profiles obtained from simulation and post-therapy imaging. V120 may be a better estimator of tumoral liver dose distribution than D50 and D70, whereas D50 and D70 may be better regarding non-tumoral liver instead of V20.
Acknowledgment: Funded by EURATOM 2019-2020 (SINFONIA Project GA N. 945196)