Calculation of lung mean dose and quantification of error for Y-90-microsphere radioembolization using Tc-99m-MAA SPECT/CT and diagnostic chest CT

Abstract

Purpose: Current treatment planning for 90Y radioembolization estimates lung mean dose (LMD) by measuring the lung shunt fraction (LSF) from 99mTc-macroaggregated albumin (MAA) planar imaging and assuming a 1-kg lung mass. This methodology, however, overestimates LSF and LMD and could therefore unnecessarily limit the dose to target volume(s). We propose an improved LMD calculation that derives LSF from 99mTc-MAA SPECT/CT and the patient-specific lung mass from diagnostic chest CT. Furthermore, we investigated the errors in lung mass, LSF, and LMD arising from contour variability in patient data in order to estimate the precision of our proposed methodology. Methods: Our proposed LMD (LMDnew) calculation consisted of the following steps: (a) estimate liver counts from the MAA SPECT/CT liver contour; (b) estimate total lung counts by multiplying density (counts/g) from the MAA SPECT/CT left-lung contour by the total lung mass (g) from the diagnostic CT lung contours; (c) compute LSFnew from liver and lung counts; (d) calculate LMDnew using LSFnew and the total lung mass from the diagnostic CT (Mnew). LMDnew, LSFnew, and Mnew estimates were compared to standard model values (LMDclin, LSFclin, and 1 kg, respectively) in 52 consecutive patients with hepatocellular carcinoma who underwent radioembolization using 90Y glass microspheres. The precision of our methodology was quantified by varying lung and liver contours in the same patient population and calculating the resulting relative errors in the liver count, lung count, and lung mass measurements. Results: The median Mnew was 839 g (range, 550–1178 g) for men and 731 g (range, 548–869 g) for women. The median LSFnew was 0.02 (range, 0.01–0.11), while the median LMDnew was 4.9 Gy (range, 0.3–25.5 Gy). Mnew, LSFnew, and LMDnew were significantly lower than Mclin, LSFclin, and LMDclin, with respective relative mean (±SD) differences of −20% (±16%) for Mnew, −63% (±15%) for LSFnew, and −53% (±23%) for LMDnew. The estimated 1-sigma uncertainties in Mnew, LSFnew, and LMDnew were 9%, 10%, and 13%, respectively. Conclusions: We derived a method to calculate lung mass and LSF using routinely available diagnostic chest CT and 99mTc-MAA SPECT/CT. More importantly, we systematically quantified the errors in our measurements to establish the precision of the estimated lung dose (13%). The proposed methodology provides a more accurate LMD and an estimate of its precision, which will improve treatment and retreatment planning for 90Y radioembolizations.