[I163] Dosimetric planning of liver radioembolization with 90Y microspheres: Methodological problems and clinical achievements

Purpose The apparent simplicity of individualized dosimetry in radioembolization with 90 Y microspheres has pushed many centres to collect data. As a result, dosimetric papers and software advances obtained in 10 years are much more than those produced in 70 years of metabolic therapy with other radiopharmaceuticals. Aim of this lecture is to enlighten the limits of the present planning methodology and to remark the importance of the clinical results obtained. Methods Dosimetric treatment planning is based on 99m Tc albumin macroaggregate (MAA) SPECT/CT. Scatter correction is mandatory. The patient relative calibration method makes a correspondence between total 99m Tc SPECT counts and the intended 90 Y activity, using each patient as a calibration phantom for himself. Local energy deposition method more accurately represents the true activity distribution than convolution methods, but not in lung, where convolution is necessary. Breathing motion is a problem both in liver SPECT and CT. In hepatocarcinoma (HCC), segmentation of large lesions showing a necrotic core raise the problem of SPECT versus CT based segmentation. Infiltrative lesions can be evidenced with MRI only, but the two compartment model cannot be applied. Voxel dosimetry is limited by SPECT noise, which acts as intrinsic uncertainty principle. The heaviest source of uncertainty is the discrepancy between MAA and actual 90 Y microsphere biodistribution. Results Despite the above mentioned limits, dose response correlations have been reported by almost all researchers, with both kind of microspheres and for several kind of disease (HCC, colorectal metastases). Dose toxicity correlation has been less studied but it has been reported. The most striking results, never reported previously for any other therapeutic agents, is that at a multivariate analysis, lesion absorbed dose is the only variable associated with overall survival. Conclusions While limitations of the treatment planning methodology result in a poor prognostic value on some individual patient, we have strong indications of improvements of clinical outcome in terms of general parameters of the treated population (response and toxicity rate, progression free and overall survival).