Test-retest repeatability of hypercapnea-stress Rb-82 PET myocardial blood flow is improved with dual-spillover correction in healthy normal subjects

Objectives Adenosine and related analogs are used commonly for pharmacologic stress testing with myocardial perfusion imaging (MPI), however adverse side-effects prompted the FDA to issue a Drug Safety Communication. Increased end-tidal partial pressure of CO2 (PetCO2), termed hypercapnea (HCAP) has been proposed as an alternative method for stress myocardial perfusion imaging (MPI). This study evaluated the test-retest repeatability of myocardial blood flow (MBF) measurements with HCAP-stress using a bi-directional (blood ↔ myocardium) dual spillover correction.

Methods Twelve healthy volunteers underwent REST+HCAP (PetCO2=60 mmHg) MPI [test] using rubidium-82 PET, followed immediately by HCAP+REST [retest] scans. Low-dose CT was acquired at normal end-expiration for attenuation correction. Dynamic images (6min total) were reconstructed using iterative reconstruction and 8mm post-filter. MBF quantification was performed with FlowQuant software using a 1-tissue compartment tracer kinetic model with 3 different image-derived input functions: 1) left ventricle cavity (LVIF), 2) left atrial cavity (LAIF), and 3) LVIF with bi-directional spillover correction (LVIF+SOC) accounting for myocardial signal contamination in the LVIF. Test vs retest correlation and Bland-Altman analysis was used to evaluate repeatability of the MBF measurements. Repeatability coefficient (RPC) values were computed as retest - test delta SD/mean x 100%.

Results MBF mean and test-retest delta/mean values values are tabulated below. There was a significant increase in MBF from REST to HCAP using all 3 input functions, demonstrating hyperemic effect of the hypercapnea method. The LVIF+SOC analysis method had lower MBF values compared to LVIF and LAIF (both p<0.05). There was no difference in correlation or test-retest RPC values at REST (ANOVA p=0.72), however during HCAP-stress the repeatability improved dramatically from RPC = 33% to 15% (p=0.02), comparing LVIF to LVIF+SOC methods.

Conclusions Hypercapnea appears to be a repeatable method to induce hyperemia for stress myocardial perfusion imaging. Quantification of myocardial blood flow is improved using the proposed dual spillover correction together with an image-derived input function in the left ventricle cavity.