PET/MRI of Hepatic 90Y Microsphere Deposition Determines Individual Tumor Response

Abstract

PurposeThe purpose of our study is to determine if there is a relationship between dose deposition measured by PET/MRI and individual lesion response to yttrium-90 (90Y) microsphere radioembolization.Materials and Methods26 patients undergoing lobar treatment with 90Y microspheres underwent PET/MRI within 66 h of treatment and had follow-up imaging available. Adequate visualization of tumor was available in 24 patients, and contours were drawn on simultaneously acquired PET/MRI data. Dose volume histograms (DVHs) were extracted from dose maps, which were generated using a voxelized dose kernel. Similar contours to capture dimensional and volumetric change of tumors were drawn on follow-up imaging. Response was analyzed using both RECIST and volumetric RECIST (vRECIST) criteria.ResultsA total of 8 hepatocellular carcinoma (HCC), 4 neuroendocrine tumor (NET), 9 colorectal metastases (CRC) patients, and 3 patients with other metastatic disease met inclusion criteria. Average dose was useful in predicting response between responders and non-responders for all lesion types and for CRC lesions alone using both response criteria (p < 0.05). D70 (minimum dose to 70 % of volume) was also useful in predicting response when using vRECIST. No significant trend was seen in the other tumor types. For CRC lesions, an average dose of 29.8 Gy offered 76.9 % sensitivity and 75.9 % specificity for response.ConclusionsPET/MRI of 90Y microsphere distribution showed significantly higher DVH values for responders than non-responders in patients with CRC. DVH analysis of 90Y microsphere distribution following treatment may be an important predictor of response and could be used to guide future adaptive therapy trials.

Targeted Radionuclide Therapy: An Evolution Toward Precision Cancer Treatment

“…The same group of investigators also showed that metabolic response on FDG PET/CT can predict overall survival [51]. Another group of investigators suggested that dose volume histograms derived from PET/MRI of 90 Y distribution in tumor can be helpful in predicting responders and nonresponders [52]. …”

Validation of post‐treatment PET‐based dosimetry software for hepatic radioembolization of Yttrium‐90 microspheres

“…Furthermore, post‐treatment imaging can provide precise information on the microsphere distribution. Emerging data show that it is common to have an inhomogeneous distribution of 90 Y microspheres within both individual lesions and across the entire liver parenchyma . This is contrary to the current assumptions of treatment planning, specifically in regards to activity prescription calculations for this procedure, which assume a uniform distribution of activity.…”

“…This is contrary to the current assumptions of treatment planning, specifically in regards to activity prescription calculations for this procedure, which assume a uniform distribution of activity. This inhomogeneous distribution of 90 Y microspheres often results in mixed clinical outcomes in patients with multiple lesions . Extending post‐procedural imaging beyond quality control and toward quantitative dosimetry can offer significant clinical utility to users of 90 Y microspheres.…”

“…Extending post‐procedural imaging beyond quality control and toward quantitative dosimetry can offer significant clinical utility to users of 90 Y microspheres. Numerous studies using voxel‐based dosimetry have reported that dosimetric quantities, such as D avg and D 70 , derived from post‐radioembolization PET images can predict clinical outcomes on a lesion‐by‐lesion basis . Knowledge of a lesion's potential response or progression can help guide salvage therapy much sooner than what clinical workflows currently allow…”

“…Voxel S values are DPKs translated to cubic voxel geometries for application to cumulated activity in nuclear medicine images, and are generated by EGS4 Monte Carlo code . DPK and voxel S values have been used for 90 Y PET‐based dosimetry, but until recently have only been applied with in‐house developed code at each institution . Still, Willowson et al., Lea et al., and Fowler et al.…”