Metabolic Tumor Control Probability vs. absorbed dose in the Selective Internal Radiation Therapy of liver metastases with resin 90Y-microspheres

Purpose: This study had the purpose to explore a possible relationship between tumor absorbed dose and metabolic response in the treatment of in liver metastases by Selective Internal Radiation Therapy (SIRT) with resin 90Y-microspheres. Metabolic response often anticipates morphologic response in the SIRT of liver metastases, thus PERCIST criteria based on 18FDG-PET-CT parameters were considered to assess response in place of the standard CT based RECIST criteria.

Methods: Inoperable patients with chemo-refractory liver metastases from solid tumors were recruited for SIRT after evaluation of 18FDG-PET-CT. The activity to be administered was established by previsional dosimetry evaluated from 99mTc-MAA (75-111 MBq) SPECT acquired 2 weeks before the treatment. The constraint was a mean absorbed dose to non tumoral liver of 40 Gy. Non tumoral liver and tumors were manually segmented on contrast CT and the regions of interest copied to SPECT. Voxel dosimetry was applied with S voxel values from the web site www.medphys.it and mean absorbed doses to normal liver and tumors were derived. Six to eight weeks after SIRT, all the patients repeated the 18FDG-PET-CT examination to assess tumor response according to PERCIST. The variations (%) of PET parameters versus basal examination were evaluated to establish Complete Response - CR (visual disappearance), Partial Response - PR (SUL reduction higher than 30%; metabolic volume reduction higher than 75%), Stable Disease -SD (SUL reduction between -30% and 30%; metabolic volume reduction between -75% and 75%), Progressive Disease - PD (SUL increase higher than 30%; metabolic volume increase higher than 75%).

Results: Twenty-two patients and 29 lesions were suitable for analysis. Patients had hepatic lesions from colon-rectal (11), breast (7), ovary (1), endometrial (1), parotid (1) cancer, cholangiocarcinoma (1). All pts received a single treatment of RE, with a median activity of 1.7 GBq (range 0.6-2.9) of 90Y-microspheres. Median (range) tumor absorbed doses was 100 (30-443) Gy; average was dose 130 Gy with standard deviation of 100 Gy. Metabolic response rate of lesions assessed with PERCIST: CR=31%; PR=28%; SD=24%; PD=17%. The mean tumor absorbed doses (standard variations) of the two groups i) PD and SD and ii) PR and CR were significantly different (T- test, p= 0.0001): i) 70(25) Gy; ii) 200(110) Gy. For tumor absorbed doses higher than 200 Gy only CR were observed. Considering PR or CR as endpoint, a significant TCP curve (p lower than 0.01) was obtained by probit regression with TCP of 50%, 75%, 100% for absorbed doses of 105, 120, 165 Gy, respectively. Considering only CR as endpoint, another significant TCP curve (p lower than 0.01) was obtained by probit regression with TCP of 50%, 75%, 100% for absorbed doses of 135, 160, 225 Gy, respectively. Conclusion: Drawbacks of this study are the variety of primary tumors, the limited cohort of patients, and the uncertainty of the previsional dosimetry due to the uncertain reproducibility of 90Y-SIRT as compared to 99mTc-MAA distribution. Nevertheless, the preliminary data provided evidence of correlation between response based on PET-CT parameters and tumor absorbed doses. More ample dataset would be necessary to confirm these encouraging results, possibly with differentiation depending on tumor type. Other PET-CT parameters for response such as the metabolic tumor volume, and tumor lesion glycolysis are considered for comparison with PERCIST and possible correlations.