We sought to determine the differential prognosis of patients with a normal single-photon emission computed tomographic (SPECT) perfusion study by type of stress modality. Even with a normal SPECT perfusion study, patients selected for adenosine stress have a worse survival than those selected for exercise stress. In patients who are able, low-level treadmill exercise is commonly performed during adenosine infusion (“walking” adenosine). The adjusted differential prognosis of patients performing walking adenosine is unknown. Our historical cohort underwent adenosine or treadmill stress with SPECT imaging during 2003 and 2004. Adenosine studies were classified as walking or adenosine only (no low-level exercise). Patients with an abnormal single-photon emission computed tomogram or missing demographic information were excluded. All-cause mortality was determined through July 2008. In total 3,479 patients were included, of which 1,451 (41.7%) were stressed with adenosine only, 201 (5.8%) with walking adenosine, and 1,827 (52.5%) with treadmill exercise. Over an average of 4.3 ± 1.0 years of follow-up, 307 (8.8%) died. Mortality was greatest for adenosine only, intermediate for walking adenosine, and least for exercise (p <0.001 by log-rank test), even after accounting for covariates (adjusted hazard ratio for walking adenosine 0.57, p = 0.044). In conclusion, patients with a normal SPECT perfusion scan who are able to undergo an adenosine protocol in conjunction with exercise have an intermediate prognosis between those who cannot perform low-level exercise and patients able to undergo full treadmill exercise. This differential survival remains significant after adjustment, perhaps because of differences in baseline functional capacity.
The choice of stress modality for cardiac single-photon emission computed tomographic (SPECT) imaging reflects many clinical variables including recent dietary and medication intakes, pulmonary status, conduction system disease, and functional capacity. Previous work has suggested that, even with normal SPECT images, prognosis varies by the selected stress modality. Patients selected for pharmacologic stress have a worse prognosis compared to those selected for exercise stress. However, whether such a differential prognosis exists for those selected for a combination of low-level exercise during adenosine infusion (“walking” adenosine) has not been determined. We hypothesized that patients with a normal stress SPECT perfusion study selected for a walking adenosine would have an intermediate prognosis compared to standard treadmill exercise and adenosine without low-level exercise.
Methods
We studied all patients referred to the nuclear cardiology laboratory of Northwestern Memorial Hospital (Chicago, Illinois) from January 2003 through December 2004. Retrospective analysis of patient data was approved by the institutional review board.
For each patient, the following characteristics were recorded from an intake process completed before the stress test: name, gender, date of birth, Social Security number, patient’s self-reported height and weight, medical history (presence or absence of known coronary artery disease, diabetes mellitus, hypertension, and dyslipidemia), family history of coronary artery disease, current or previous tobacco use, and current medications (angiotensin-converting enzyme inhibitors, antiplatelet agents, β blockers, anticholesterol agents, and diuretics). Body mass index (BMI) was calculated as weight (in kilograms) divided by the square of height (in meters). The type of stress (treadmill exercise, adenosine with or without low-level exercise, and dobutamine infusion) was recorded. The decision to add concomitant exercise (first stage of modified Bruce protocol) to adenosine studies when not mandated by clinical indications (left bundle branch block, limiting claudication, and so on) was left to the discretion of the nuclear laboratory physician and/or exercise physiologist. Twelve-lead electrocardiograms at rest, stress, and recovery were interpreted by cardiologists using standard criteria.
The vast majority of studies were performed using a routine 1-day dual isotope protocol. Images at rest were acquired after intravenous injection of thallium-201, and images during stress were acquired after administration of technetium-99m sestamibi. A minority of studies were performed using a 2-day high-dose technetium-99m sestamibi protocol, which was recommended for patients >350 lbs to improve image quality.
Images were acquired with a dual-detector γ-camera (ADAC Vertex, ADAC Laboratories, Milpitas, California; or Siemens e.cam, Siemens Medical Systems, Hoffman Estates, Illinois) with 64 projections for 25 seconds each, covering 180° (elliptical orbit with the ADAC Vertex, body-contoured orbit with Siemens e.cam). SPECT images were interpreted using a semiquantitative 20-segment scoring system in which 0 equals normal tracer activity and 4 equals no tracer activity. Total score for the image at rest equals the summed rest score (SRS) and similarly for the stress image’s summed stress score (SSS). Transient ischemic dilatation was reported based on a combination of visual assessment and software-reported ratio. Most SPECT images after stress were electrocardiographically gated allowing for calculation of ejection fraction (EF). Arrhythmia limited gating in a minority of patients throughout.
We included patients ≥18 years of age with complete data for review. Patients missing demographic criteria (name, gender, date of birth, Social Security number) were excluded. Patients undergoing dobutamine stress single-photon emission computed tomography were excluded to limit the population to those selected for treadmill or adenosine stress. If a patient underwent multiple stress tests during this period, only the first was included. Patients with abnormal SPECT perfusion (SSS ≥4 or SRS ≥4), gated EF <50%, or transient ischemic dilatation were excluded to limit the population to those with normal studies.
All-cause mortality was determined from the Social Security Administration’s Death Master File. Length of survival was computed as the number of days from the stress test to the date of death or July 16, 2008, whichever came first.
All statistical tests were performed using R 2.12 with missing EF and BMI values imputed by the “Multiple Imputation” package 0.09–11.03. Continuous variables are expressed as mean ± SD and were compared using t test. Proportions were compared using chi-square or Fisher’s exact test. Unadjusted incidence of all-cause death used the incidence density because follow-up differed for each patient. A Kaplan–Meier survival analysis by stress modality group used the log-rank test. All applicable tests were 2-tailed, and a p value <0.05 was taken as the cutoff for statistical significance.
Because of the non-random assignment of patients to a stress modality, propensity analyses were performed. All variables were included in logistic regression analysis to predict choice of exercise stress (propensity score for exercise). All variables apart from those related to treadmill exercise were included in logistic regression analysis to predict choice of low-level exercise of those undergoing adenosine stress (propensity score for walking adenosine).
A Cox proportional hazards model used stress modality group as an indicator variable. More complex model building proceeded by first identifying all important (p <0.25) univariate predictors for inclusion in a preliminary effects model. The stress modality group was also included because of its centrality to our hypothesis. This multivariable model was then decreased by backward retention of variables that remained significant (p <0.05) in the composite model, in addition to the stress modality group. The 2 propensity scores were included as continuous variables in all multivariable models. In addition, stress modality was tested with all other variables for interaction in the final model. Only 1 of 17 interaction terms was significant at a p value <0.05, roughly what would have been expected by chance. Therefore, no interaction terms were kept in the final model.
Schoenfeld residuals tested the assumption of proportional hazards and found no systematic significant deviations. Plots of beta coefficients from logistic regression and univariable Cox proportional hazards models were visually examined by quintile of each clinical continuous variable (age, BMI, exercise duration, and EF) for departures from linearity. Only BMI appeared visually to have a nonlinear effect, which was confirmed by the significance of its squared term in a 2-variable Cox proportional hazards model. Therefore, BMI and its square were entered simultaneously in the preliminary and final models.
Results
From January 2003 through December 2004, our nuclear cardiology laboratory performed 6,368 stress studies with SPECT imaging. Of these, 2,219 (34.8%) were excluded because of abnormal SPECT perfusion (SSS ≥4 or SRS ≥4), 224 (3.5%) were excluded because of a previous study within this period, 116 (1.8%) were excluded because of selection of dobutamine stress or lack of demographic or clinical information, and 330 (5.2%) were excluded because of EF <50% and/or transient ischemic dilatation.
The final cohort included 3,479 patients as presented in Table 1 . Patients selected for walking adenosine had, in general, intermediate clinical predictors between adenosine-only studies and treadmill exercise. Over an average of 4.3 ± 1.0 years of follow-up, 307 patients (8.8%) died. Table 2 lists survival characteristics of these patients by stress modality. Figure 1 shows Kaplan–Meier survival curves by stress modality. Survival was least for adenosine only, intermediate for walking adenosine, and greatest for exercise (p <0.001 by log-rank test). Comparing just patients who underwent adenosine stress, survival was shorter for those undergoing adenosine only compared to walking adenosine (p = 0.001 by log-rank test).
| Variable | All | Adenosine Only | Walking Adenosine | Exercise | p Value |
|---|---|---|---|---|---|
| Number of patients | 3,479 | 1,451 | 201 | 1,827 | N/A |
| Exercise time (minutes) | N/A | N/A | N/A | 9.5 ± 3.9 | N/A |
| Age (years) | 59.4 ± 13.7 | 63.8 ± 14.4 | 55.8 ± 13.2 | 56.2 ± 12.1 | <0.001 |
| Men | 1,813 (52.1%) | 632 (43.6%) | 91 (45.3%) | 1,090 (59.7%) | <0.001 |
| Body mass index (kg/m 2 ) ⁎ | >28.6 ± 6.8 | 28.9 ± 7.7 | 28.7 ± 6.8 | 28.3 ± 5.9 | 0.007 |
| Diabetes mellitus | 633 (18.2%) | 378 (26.1%) | 38 (18.9%) | 217 (11.9%) | <0.001 |
| Family history of coronary disease | 886 (25.5%) | 278 (19.2%) | 54 (26.9%) | 554 (30.3%) | <0.001 |
| Dyslipidemia | 1,488 (42.8%) | 555 (38.2%) | 88 (43.8%) | 845 (46.3%) | <0.001 |
| Hypertension | 1,760 (50.6%) | 888 (61.2%) | 132 (65.7%) | 740 (40.5%) | <0.001 |
| Tobacco use | 674 (19.4%) | 334 (23.0%) | 39 (19.4%) | 301 (16.5%) | <0.001 |
| Known coronary disease | 359 (10.3%) | 143 (9.9%) | 18 (9.0%) | 198 (10.8%) | 0.565 |
| Antiplatelets | 1,355 (38.9%) | 573 (39.5%) | 73 (36.3%) | 709 (38.8%) | 0.683 |
| β Blockers | 1,145 (32.9%) | 599 (41.3%) | 99 (49.3%) | 447 (24.5%) | <0.001 |
| Angiotensin-converting enzyme inhibitors | 718 (20.6%) | 311 (21.4%) | 49 (24.4%) | 358 (19.6%) | 0.172 |
| Anticholesterol agents | 1,287 (37.0%) | 495 (34.1%) | 87 (43.3%) | 705 (38.6%) | 0.005 |
| Other antihypertensives | 617 (17.7%) | 336 (23.2%) | 44 (21.9%) | 237 (13.0%) | <0.001 |
| Diuretics | 715 (20.6%) | 403 (27.8%) | 59 (29.4%) | 253 (13.8%) | <0.001 |
| Ischemic electrocardiographic changes | 386 (11.1%) | 50 (3.4%) | 17 (8.5%) | 319 (17.5%) | <0.001 |
| Gated ejection fraction (%) † | 66.7 ± 7.8 | 66.8 ± 7.9 | 65.1 ± 8.3 | 66.8 ± 7.5 | 0.746 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
