Kenneth Nichols PhD, Presenter: Royalties, Syntermed, Inc

Andrew Van Tosh MD, Abstract Co-Author: Consultant, Pfizer Inc Consultant, Bracco Group Consultant, Cardinal Health, Inc Consultant, Ion Beam Applications, SA

Nathaniel Reichek MD, Abstract Co-Author: Nothing to Disclose

Christopher J. Palestro MD, Abstract Co-Author: Nothing to Disclose

82Rb cardiac PET analyzes data with the left ventricle (LV) in a different physiologic state at stress compared to rest, enabling detection of stress-induced ejection fraction (EF) decrease. However, in pts with reversible ischemia, accurate LV contours may be more difficult to verify on stress data than on rest data due to reduced segmental myocardial counts. Since LV mass should be the same from rest & stress data, we tested the hypothesis that observed changes from rest to stress of volume significantly exceed artifactual differences in mass.

We retrospectively reviewed data for 205 pts referred for evaluation of coronary disease who had rest & regadenoson stress 82Rb PET/CT imaging. Myocardial perfusion data were analyzed by commercially available algorithms to calculate end-diastolic volume (EDV) & end-systolic volume at rest & stress, from which LV EFs were computed; this algorithm simultaneously computed LV mass. Because of reliance of LV measurements on valve plane placement, a simple cylindrical-hemispherical LV model was fit to observed data to enable estimating the extent to which a 1-pixel change in valve plane placement produces variance of volume & mass. In this model, Volume = [(2•π•r3/3)+(N•π•r3)], & Mass = 1.05 gm/mL•[(2•π•r3/3)•((1+ r)3- r3) + N•π•r•((1+ r)2)- r2)], where cylinder height = N • the radius r of the apical hemispherical cap. N was iteratively adjusted to optimally fit observed data.

Observed mass significantly (r = 0.98, p <0.0001) fit observed EDV. Optimal LV model N was 3.0, for which model-predicted mass correlated significantly with observed mass (r = 0.97, p <0.0001). At EDV near 400 mL, the LV model predicted an 8 gm error in mass & a 10 mL error in EDV due to a 1-pixel error in valve plane placement. For pts with observed EDV near 400 mL, observed change in mass was 14 gm, but observed change in EDV from rest to stress was 48 mL. 

While observed non-zero changes in mass values from rest to stress were explained by 1-pixel uncertainty in valve plane location, observed changes in volume from rest to stress were far larger than variability associated with valve placement errors, and must represent genuine physiologic changes, not artifacts.

Attention to calculated myocardial mass can serve as a useful quality assurance check to gated 82Rb PET measurements of stress-induced LV dysfunction.

Nichols, K, Van Tosh, A, Reichek, N, Palestro, C, Validation of Cardiac PET Measurement of Exercise-induced LV Dysfunction.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14001866.html