Prevalence of compromised ⁸²Rb PET bolus injection integrity assessed by automated quality control algorithms

Objectives: ⁸²Rb PET rest/stress protocols permit determination of a wide variety of left ventricular function & myocardial perfusion parameters, including absolute myocardial blood flow & myocardial flow reserve (MFR) assessed from first pass transit of injected tracer bolus activity. Accurate MFR computations require a rapid, cleanly-delivered bolus, with activity below levels that could cause count saturation of scanner electronics. We wished to determine the prevalence & character of bolus injection issues, & determine if these issues affected MFR computations.

Methods: Data were analyzed retrospectively for 256 rest & regadenoson-stress studies of 128 pts evaluated for known or suspected CAD. A commercially available automated injector was used to deliver pre-determined activities of ⁸²Rb to pts undergoing PET imaging. Data were collected in gated list mode & subsequently rebinned (into 20 3-sec cinematic frames, 5 12-sec frames, & 7 30-sec frames), with the first-pass dynamic portion of the data used to compute MFR. A cardiologist monitored activity delivered to pts during infusion using a beta probe for both rest & stress ⁸²Rb bolus injections to ensure that bolus delivery was sufficiently rapid to provide valid first-pass information. Data were considered only if the cardiologist verified the bolus was delivered within expected time frames. Automated quality control (QC) algorithms subsequently analyzed potential bolus delivery problems as part of measuring MFR. The algorithms flagged potential problems for rest & stress studies in 8 categories: inconsistent frame duration, scanner saturation, inability to detect blood curve peak, inappropriate blood peak width, flat blood curve tail, gradual pt motion & abrupt pt motion. Results: The QC algorithms identified problems for 9 of the 128 pts. During stress there were 4 instances of scanner saturation, 2 blood peak detection, 1 blood peak width problem, 1 gradual motion & 3 abrupt motion; at rest there were 2 instances of blood peak width problems & 2 abrupt motion. In examining whether there were associations of pt characteristics with technical problems, none were found for age (logistic regression p = 0.31), gender (p = 0.63), relative perfusion total severity score (p = 0.27), ejection fraction (p = 0.64) or asynchrony (p = 0.38). However, associations were significant between the occurrence of any technical problem & MFR (χ² = 8.7, p = 0.003), such that MFR was lower for pts with technical problems (1.5±0.5 versus 2.1±0.7, p = 0.01), even though EF was similar (p = 0.70), as was asynchrony bandwidth (p = 0.30), thereby suggesting that MFR computation accuracy was adversely affected by bolus injection technical errors.

Conclusions: It is important to verify integrity of the injected ⁸²Rb bolus in order to assure the quality of MFR computations performed as part of CAD pt evaluation.