Optimal FDG PET/CT volumetric parameters for risk stratification in patients with locally advanced non-small cell lung cancer: results from the ACRIN 6668/RTOG 0235 trial
Our aim was to assess feasibility and performance of novel semi-automated image analysis software called ROVER to quantify metabolically active volume (MAV), maximum standardized uptake value-maximum (SUV(max)), 3D partial volume corrected mean SUV (cSUV(mean)), and 3D partial volume corrected mean MVP (cMVP(mean)) of spinal bone marrow metastases on fluorine-18 fluorodeoxyglucose-positron emission tomography/computerized tomography ((18)F-FDG-PET/CT). We retrospectively studied 16 subjects with 31 spinal metastases on FDG-PET/CT and MRI. Manual and ROVER determinations of lesional MAV and SUV(max), and repeated ROVER measurements of MAV, SUV(max), cSUV(mean) and cMVP(mean) were made. Bland-Altman and correlation analyses were performed to assess reproducibility and agreement. Our results showed that analyses of repeated ROVER measurements revealed MAV mean difference (D)=-0.03±0.53cc (95% CI(-0.22, 0.16)), lower limit of agreement (LLOA)=-1.07cc, and upper limit of agreement (ULOA)=1.01cc; SUV(max) D=0.00±0.00 with LOAs=0.00; cSUV(mean) D=-0.01±0.39 (95% CI(-0.15, 0.13)), LLOA=-0.76, and ULOA=0.75; cMVP(mean) D=-0.52±4.78cc (95% CI(-2.23, 1.23)), LLOA=-9.89cc, and ULOA=8.86cc. Comparisons between ROVER and manual measurements revealed volume D= -0.39±1.37cc (95% CI (-0.89, 0.11)), LLOA=-3.08cc, and ULOA=2.30cc; SUV(max) D=0.00±0.00 with LOAs=0.00. Mean percent increase in lesional SUV(mean) and MVP(mean) following partial volume correction using ROVER was 84.25±36.00% and 84.45±35.94% , respectively. In conclusion, it is feasible to estimate MAV, SUV(max), cSUV(mean), and cMVP(mean) of spinal bone marrow metastases from (18)F-FDG-PET/CT quickly and easily with good reproducibility via ROVER software. Partial volume correction is imperative, as uncorrected SUV(mean) and MVP(mean) are significantly underestimated, even for large lesions. This novel approach has great potential for practical, accurate, and precise combined structural-functional PET quantification of disease before and after therapeutic intervention.