Experimental validation of absolute SPECT/CT quantification for response monitoring in breast cancer

Recent developments in iterative image reconstruction enable absolute quantification of SPECT/CT studies by incorporating compensation for collimator-detector response, attenuation, and scatter as well as resolution recovery into the reconstruction process (Evolution; Q.Metrix package; GE Healthcare, Little Chalfont, UK). The aim of this experimental study is to assess its quantitative accuracy for potential clinical ^99m Tc-sestamibi (MIBI)-related SPECT/CT application in neoadjuvant chemotherapy response studies in breast cancer.

Two phantoms were filled with MIBI and acquired on a SPECT/CT gamma camera (Discovery 670 Pro; GE Healthcare), that is, a water cylinder and a NEMA body phantom containing six spheres that were filled with an activity concentration reflecting clinical MIBI uptake. Subsequently, volumes-of-interest (VOI) of each sphere were drawn (semi)automatically on SPECT using various isocontour methods or manually on CT. Finally, prone MIBI SPECT/CT scans were acquired 5 and 90 min p.i. in a locally advanced breast cancer patient.

Activity concentration in the four largest spheres converged after nine iterations of evolution. Depending on the count statistics, the accuracy of the reconstructed activity concentration varied between -4.7 and -0.16% (VOI covering the entire phantom) and from 6.9% to 10% (8.8 cm ⌀ cylinder VOI placed in the center of the phantom). Recovery coefficients of SUV_max were 1.89 ± 0.18, 1.76 ± 0.17, 2.00 ± 0.38, 1.89 ± 0.35, and 0.90 ± 0.26 for spheres with 37, 28, 22, 17, and 13 mm ⌀, respectively. Recovery coefficients of SUV_mean were 1.07 ± 0.06, 1.03 ± 0.09, 1.17 ± 0.21, 1.10 ± 0.20, and 0.52 ± 0.14 (42% isocontour); 1.10 ± 0.07, 1.02 ± 0.09, 1.13 ± 0.19, 1.06 ± 0.19, and 0.51 ± 0.13 (36% isocontour with local background correction); and 0.96, 1.09, 1.03, 1.03, and 0.29 (CT). Patient study results were concordant with the phantom validation.

Absolute SPECT/CT quantification of breast studies using MIBI seems feasible (<17% deviation) when a 42% isocontour is used for delineation for tumors of at least 17 mm diameter. However, with tumor shrinkage, response evaluation should be handled with caution, especially when using SUV_max .