A model for the prediction of a successful stress-first Tc-99m SPECT MPI

Abstract

Background

Stress-only Tc-99m MPI saves time, radiation exposure, and a normal study has a benign prognosis. However, a stress-first protocol is relatively labor intensive requiring pre-test screening for suitability and early post-stress image review to determine the need for rest imaging. The purpose of this study was to develop a simple clinical score used prior to a patient’s myocardial perfusion imaging (MPI) study to determine if they should undergo a stress-first protocol.

Methods

We reviewed all patients who underwent Tc-99m SPECT MPI over a 27-month period and divided them into derivation and validation cohorts. Patients were categorized as having a successful stress-first protocol based on a summed stress score ≤1, with or without attenuation correction. We generated a multivariable model from the derivation cohort to identify demographic and clinical correlates of successful stress-first imaging. Two validation cohorts using a CZT and a conventional SPECT camera were then used to test the performance of the model.

Results

The derivation cohort included 1,996 patients and the validation cohort consisted of 1,005 CZT SPECT patients and 2,430 conventional SPECT patients. The following variables were associated with unsuccessful (i.e., abnormal) stress-first imaging: age >65 years (1 point), diabetes (2 points), typical chest pain (2 points), congestive heart failure (3 points), abnormal ECG (4 points), male gender (4 points), and documented CAD (5 points). Emergency Department location (−2 points) was negatively associated with an unsuccessful protocol. An increasing score showed a strong association with an unsuccessful stress-first protocol in both the derivation and the validation cohorts (P < .0001) and dividing the cohorts into low (<5), intermediate (≥5 and <10), and high (≥10) risk scores accurately stratified patients based on their frequencies of unsuccessful stress-first imaging. ROC curve analysis showed excellent prediction in both the derivation and the validation cohorts with an area under the curve of 0.82 and 0.75-0.83, respectively.

Conclusions

This pre-test scoring tool accurately identifies patients who can successfully undergo a stress-first imaging protocol without the need for rest imaging and may allow for wider adoption of stress-first imaging protocols.

Introduction

In routine clinical practice, up to 60%-70% of appropriately indicated myocardial perfusion imaging (MPI) studies demonstrate normal perfusion.¹,² It is well documented that a normal SPECT MPI study carries a benign prognosis with a 1-year cardiac event rate of <1%.³,⁴ A normal SPECT MPI study is defined as a negative ECG response with adequate exercise or vasodilator stress, the absence of stress perfusion defects, and normal left ventricular systolic function. With normal stress images, rest images provide no additional prognostic or diagnostic value and eliminating the rest images becomes possible. The clear advantages of stress-only imaging are decreased length of the test, both for the patients and for the laboratory, and a lower radiation dose. A stress-only study can be completed, processed, and read in <90 minutes as opposed to the usual 3-5 hours. The radiation dose can be decreased by 30%-60% depending on the dose used for stress imaging.⁵,⁶

The benign prognosis of a normal stress-only study has now been confirmed in more than 10,000 patients and appears to be no different than that of a normal rest-stress study. Three large studies found low cardiac event rates and similar all-cause and cardiac mortality to full rest-stress studies.5, 6, 7 However, a stress-first protocol is relatively labor intensive as it requires that a reader be present to screen the patients for suitability and read the stress images upon completion to determine the need for rest imaging. Stress-first protocols are also greatly facilitated using some type of attenuation correction such as Gd-153 line sources, prone in addition to supine images, or CT correction.⁸

Currently, there are no published guidelines or models for the determination of which patients are candidates for a Tc-99m stress-first protocol. In fact, current stress testing protocols endorsed by ASNC include only 1-day (low dose-high dose) or 2-day (rest-stress/stress-rest) studies,⁹ but a recent preferred practice statement¹⁰ and clinical update¹¹ from ASNC refer to stress-only protocols but suggest that more study in this area was needed. Obviously, patients with low pre-test probability (based on age, gender, risk factors, symptoms, and rest ECG) are premier candidates. Another suitable group is patients with a high BMI (>35 kg/m²) or weight over 250 lbs. Also patients with recent (<3 years) negative noninvasive or invasive studies for the presence of obstructive CAD would seem to be good candidates. The purpose of this study was to develop a simple clinical score which could be employed at the beginning of a patient’s MPI study to determine if they should undergo a stress-first protocol. For the purposes of this study, a successful stress-first study was defined as one in which stress images could be reliably interpreted as normal without needing rest images for comparison.