- PET/CT shows higher tumour dose metrics (TD-mean, TD-V30, TD-V70) compared with SPECT/CT; differences all p < 0.001.
- SPECT/CT underestimates TD-mean by 127 Gy and overestimates NTLD-mean by 79 Gy, with wide limits of agreement suggesting systematic bias.
- Strong correlations for many endpoints (r > 0.9) but modalities are not interchangeable without modality-specific thresholds or further validation.
Cardiovasc Intervent Radiol. 2026 Sep 24. doi: 10.1007/s00270-026-04620-2. Online ahead of print.
ABSTRACT
PURPOSE: To quantitatively compare Y90 PET/CT and Y90 SPECT/CT voxel-based dosimetry after RESIN Y90 TARE in hepatocellular carcinoma (HCC). This was a secondary aim of a previously published prospective single-center trial evaluating low-dose scout RESIN Y90 for treatment planning (NCT04172714).
MATERIALS AND METHODS: Thirty patients with 33 treatment-naïve HCC lesions underwent Y90 PET/CT and SPECT/CT after scout and therapeutic RESIN Y90. Voxel-based dosimetry (local deposition method) yielded mean, minimum, and the mean dose to the highest-dosed 30% (V30) and 70% (V70) of the volume, for tumor dose (TD) and non-tumoral liver dose (NTLD). Identical PET/CT-derived contours were applied to the SPECT/CT, and agreement was assessed by Bland-Altman analysis and intraclass correlation (ICC).
RESULTS: TD-mean (494 vs. 366 Gy), TD-V30 (594 vs. 408 Gy), and TD-V70 (405 vs. 326 Gy) were higher on PET/CT than SPECT/CT, while NTLD-mean (103 vs. 182 Gy) and NTLD-V70 (77 vs. 141 Gy) were lower on PET/CT (all p < 0.001). Correlations were very strong for TD-mean, TD-min, TD-V30, TD-V70, NTLD-mean, and NTLD-V30 (r > 0.9). Relative to PET/CT, SPECT/CT underestimated TD-mean by 127 Gy (95% limits of agreement -435 to 181 Gy) and overestimated NTLD-mean by 79 Gy; ICC was low for several non-tumoral endpoints (NTLD-mean 0.57, 95% CI -0.09 to 0.86) despite strong correlation.
CONCLUSIONS: After RESIN Y90 TARE for HCC, PET/CT calculated higher TD and lower NTLD than SPECT/CT, correlating strongly but with a potentially clinically relevant systematic bias and wide limits of agreement. The modalities should not be considered interchangeable without modality-specific thresholds or further validation.
PMID:42786295 | DOI:10.1007/s00270-026-04620-2
Share Evidence Blueprint
Save to Google Notes

Search Google Scholar
Save as PDF
⭐ My Revision List

