Clinical feasibility of quantitative holmium-166 SPECT in presence of technetium-99m colloid

Objectives: Investigate the effect of technetium-99m colloid (^99mTc) photons crosstalk on holmium-166 (¹⁶⁶Ho) dosimetry when used in combination to perform dual isotope (DI) SPECT imaging. This protocol, using ¹⁶⁶Ho-microspheres and ^99mTc to identify the healthy liver, allows for automatic compartmental dosimetry before and after radioembolization treatment.

Methods: To assess the crosstalk (the reciprocal influence of one isotope’s downscatter in the other’s main photopeak), ¹⁶⁶Ho-SPECT reconstructions of patient scans acquired before and after additional ^99mTc administration (referred to as ¹⁶⁶Ho-only and ¹⁶⁶Ho-DI respectively) were compared. The ¹⁶⁶Ho-only and ¹⁶⁶Ho-DI SPECT scans were performed with <10 min interval to minimize patient motion. Images were reconstructed using Siemens Symbia software (Flash3D). To compensate for ^99mTc-photons crosstalk, its influence in the ¹⁶⁶Ho photopeak (81 keV) window was accounted for in the DI image reconstruction using energy window based scatter correction methods. The quality of the ¹⁶⁶Ho-DI image was compared to that of the ¹⁶⁶Ho-only image, qualitatively and quantitatively. The qualitative assessment was performed by independent blinded comparison by two nuclear medicine physicians assessing 65 pairs of SPECT/CT scans. Inter-observer agreement was tested by Cohen's kappa coefficient. For the quantitative analysis, three volumes of interest (VOI) were considered. VOI_TUMOR and VOI_HEALTHY, within the liver, were manually delineated on the ¹⁶⁶Ho-only reconstruction and transferred to the co-registered ¹⁶⁶Ho-DI reconstruction. VOI_LUNGS was semi-automatically segmented in both ¹⁶⁶Ho-only and ¹⁶⁶Ho-DI reconstructions. Absorbed dose within the resulting VOIs was calculated based on the administered therapeutic activity.

Results: The qualitative assessment showed no distinct preference for either the ¹⁶⁶Ho-only or ¹⁶⁶Ho-DI SPECT (kappa = 0.093). Quantitative analysis indicated that the median absorbed dose in the VOIs was 27, 116 and 3 Gy for VOI_HEALTHY, VOI_TUMOR and VOI_LUNGS, the mean difference in absorbed dose between ¹⁶⁶Ho-DI and ¹⁶⁶Ho-only was -2.00±2.84 Gy, -5.27±8.99 Gy and 0.80±1.08 Gy respectively. These differences were regarded as clinically not relevant by treating physicians. The corresponding Pearson’s correlation coefficient between ¹⁶⁶Ho-only and ¹⁶⁶Ho-DI for absorbed dose was 0.97, 0.99 and 0.82 for VOI_HEALTHY, VOI_TUMOR and VOI_LUNGS.

Conclusions: The DI protocol enables satisfying ¹⁶⁶Ho SPECT within the clinical reconstruction framework, allowing an automatic estimation of absorbed dose in tumors, healthy liver and lungs. Research Support: Nederlandse Organisatie voor Wetenschappelijk Onderzoek