Clinical Utility of SPECT in the Heart Transplant Population: Analysis from a Single Large-Volume Center

Abstract

Background

Survival after heart transplant has greatly improved, with median survival now over 12 years. Cardiac allograft vasculopathy (CAV), has become a major source of long-term morbidity and mortality. Single photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) is used for CAV surveillance, but there is limited data on its prognostic utility.

Methods

We retrospectively identified patients undergoing SPECT MPI for CAV surveillance at a single, large-volume center. Images were assessed with semi-quantitative visual scoring (summed stress score [SSS] and summed rest score [SRS]) and quantitatively with total perfusion defect (TPD).

Results

We studied 503 patients (mean age 62.5, 69.3% male) at a median of 9.0 years post-transplant. During mean follow-up of 5.1 ± 2.5 years, 114 (22.6%) patients died. The diagnostic accuracy for significant CAV (ISHLT grade 2 or 3) was highest for SSS with an area under the curve (AUC) of 0.650 and stress TPD (AUC 0.648), with no significant difference between SSS and stress TPD (p=0.061). Stress TPD (adjusted hazard ratio 1.07, p=0.018) was independently associated with all-cause mortality, while SSS was not (p=0.064). The prognostic accuracy of quantitative assessment of perfusion tended to be higher compared to semi-quantitative assessment, with the highest accuracy for stress TPD (area under the receiver operating curve 0.584).

Conclusions

While SPECT MPI identified a cohort of higher risk patients, with quantitative analysis of perfusion demonstrating higher prognostic accuracy. However, the overall prognostic accuracy was modest and alternative non-invasive modalities may be more suitable for CAV surveillance.

INTRODUCTION

Survival after heart transplant has greatly improved, with median survival now at 12.5 years.^1,2 While acute rejection and infectious complications are important cause of early mortality,² cardiac allograft vasculopathy (CAV) has become a major source of long-term morbidity and mortality.^2–4 In the 2019 Registry of the International Society for Heart and Lung Transplantation Report, CAV was found in 7.7% of heart transplant patients within the first year and in 46.8% of heart transplant patients within 10 years.² Additionally, CAV is a leading cause of death in patients who survive the first year following transplant.³

CAV is characterized by diffuse lesions, affecting both the epicardial vessels and the microvasculature.⁵ Patients with CAV often lack classic symptoms of ischemia due to allograft denervation;⁴ therefore many transplant programs perform routine surveillance. Invasive angiography, often with intravascular ultrasound (IVUS), is frequently used for diagnosing CAV but is associated with procedural risk.^4–8 Other methods used for CAV surveillance include stress echocardiography,^6,7 cardiovascular computed tomography angiography (CCTA),^9,10 positron emission tomography myocardial perfusion imaging (PET MPI),^11–15 stress cardiovascular magnetic resonance¹⁶ and single-photon emission computed tomography (SPECT) MPI.^5,17 Studies of the diagnostic accuracy of SPECT MPI have reported variable sensitivity and specificity, ranging from 14-90% and 33-98% respectively, due to differences in patient populations and ascertainment of CAV.^18–22

Similarly, studies assessing the prognostic accuracy of SPECT MPI have included relatively small populations and variable methods for assessing perfusion. Several smaller studies, including 47 to 166 patients, have assessed the prognostic accuracy of SPECT MPI post-transplant but included variable definitions of abnormal perfusion.^{18 20,23 24 25}

Accordingly, this study was performed to better define the potential role for SPECT MPI as a method for CAV surveillance using standard interpretation methods and to identify any potential improvements from using the latest quantitative methods. We evaluated the diagnostic accuracy for CAV and prognostic accuracy for all-cause mortality of SPECT MPI with visual semi-quantitative scoring with a standard 17-segment model compared to quantitative analysis in patients following cardiac transplant.

MATERIALS AND METHODS

Patient population

This was a retrospective analysis of transplant patients that underwent SPECT MPI for CAV surveillance at Cedars-Sinai Medical Center between February 2010 and June 2018. Only the first SPECT MPI study was included if multiple studies were performed. Most patients at our center undergo screening with SPECT or positron emission tomography MPI every other year starting at 6 years post-transplant. Patients with chronic kidney disease are also preferentially followed with annual SPECT MPI instead of invasive coronary angiography during the first 5 years post-transplant. In total, 503 patients were included in the study. Demographics were obtained at the time of SPECT MPI and included age, time post-transplant, sex, body mass index (BMI). Medical history including hypertension, diabetes, dyslipidemia was prospectively assessed with questionnaires and review with patients at the time of imaging. Chronic kidney disease was defined as estimated glomerular filtration rate < 30ml/min. Donor age, history of CAV, and history of treated acute cellular rejection (ACR) or antibody-mediated rejection (AMR) were retrospectively ascertained. History of CAV was based on the most recent invasive coronary angiogram result, which occurred a mean of 2.5 ± 3.1 years prior to SPECT MPI and was assessed using the ISHLT CAV grading system by an interventional cardiologist at the time of the procedure.²⁶ Patients with known CAV1 or greater disease were considered to have a history of CAV.

Stress Protocol

Patients underwent exercise or pharmacologic stress at the discretion of supervising personnel after reviewing referring physician requests and patient characteristics. Exercise stress testing (n = 79) was with a symptom-limited Bruce protocol. Heart rate, systolic and diastolic blood pressure measurements, and ECG tracings were obtained at rest and monitored during and after stress testing. Pharmacologic stress was performed with regadenoson (n = 304), adenosine (n = 119), or dobutamine (n = 1). Radiopharmaceutical injection was performed at near maximal stress. Stress imaging began 15-30 minutes after exercise stress and 30-60 minutes following pharmacologic stress.

Image Acquisition and Interpretation

Images were acquired with four camera systems over the study period (D-SPECT (n = 387) [Spectrum-Dynamics, Caesarea, Israel], E-Cam (n = 75) [Siemens, Munich, Germany], and Forte (n = 36) [Philips, Amsterdam, Netherlands], and Bold (n =5) [Siemens, Munich, Germany]). Patients underwent either stress and rest ^99mTc sestamibi (n = 495, 98.4%) or dual isotope ²⁰¹Tl^/99mTc sestamibi (n = 8, 1.6%).

Experienced clinicians scored perfusion abnormalities using a 4-point scale according to the 17-segment model.^27–29 Scores were combined as summed rest score (SRS), summed stress score (SSS) and summed difference score (SDS). Percent myocardium involved was calculated from semi-quantitative scores as segmental score / total possible score for stress and rest images.³⁰ Reversible perfusion defects were quantified as (% myocardium involved at stress - % myocardium involved at rest) and fixed perfusion defects as (% myocardium involved at rest).³¹ Perfusion was also quantified automatically using total perfusion deficit (TPD), which reflects the extent and severity of defects.³² Stress and rest TPD were calculated as % myocardium involved. Ischemic TPD was calculated as stress TPD – rest TPD. LV ejection fraction (LVEF) was measured on all stress studies using Quantitative Gated SPECT (QGS) software (Cedars Sinai Medical Center, Los Angeles, CA) and was defined as low if the calculated LVEF was <50%.^33,34 Rest gating was available in 323 (64.2%) of patients and was used to calculate rest LVEF and delta LVEF (stress LVEF – rest LVEF).

Statistical Analysis

The primary outcome was all-cause mortality which was determined through follow-up with the cardiac transplant clinic. The secondary outcome was one-year all-cause mortality. Continuous variables were summarized by their mean (standard deviation [SD]). Variables which were normally distributed were compared using a Student’s t-test and if not were compared using a Mann-Whitney U test. Categorical variables were summarized as numbers (proportions) and compared using a chi-square test or Fisher exact test as appropriate.

Incidence of the primary outcome was compared using Kaplan-Meier survival curves, with differences determined with the log-rank test. Prognostic significance of stress, rest, and ischemic perfusion abnormalities were assessed with Cox-proportional hazards analyses. Multivariable analyses were adjusted for rest and stress perfusion as well as age, sex, time post-transplant, medical history (hypertension, diabetes, dyslipidemia, chronic kidney disease), known CAV, and stress LVEF. Quantitative perfusion (with stress TPD and rest TPD) were assessed in a separate model from semi-quantitative assessment (with SSS and SRS). Ischemic TPD and SDS were excluded from the model due to inclusion of stress and rest values. The proportional hazards assumption was assessed using Schoenfeld residuals.

Diagnostic accuracy for significant CAV (grade 2 or 3) was assessed only in the population of patients with reference coronary angiography. A sensitivity analysis was performed only in patients with coronary angiography within 12 months of SPECT MPI. Prognostic utility, assessed in the overall population, of semi-quantitative visual scoring (SRS, SSS, SDS) and quantitative (stress, rest, and ischemic TPD) were assessed with received operating characteristic (ROC) curves. In the primary analysis we assessed accuracy for identifying all-cause mortality at any time during follow-up. We performed a secondary analysis of the prognostic accuracy for one-year all-cause mortality. Records were reviewed to determine cause of death, with sufficient information to determine cause in 64 patients. Death due to myocardial infarction, graft failure, and sudden circulatory death were considered circulatory death. Area under the ROC curve (AUC) were compared using Delong’s method.³⁵ Sub-group analyses were performed across important subgroups including: time post-transplant (<5 years vs. ≥ 5years), mode of stress, and known CAV.

All statistical tests were two-sided, with a p-value < 0.05 considered significant. All analyses were performed using Stata version 13 (StataCorp, College Station, Texas). The study protocol conformed with the Declaration of Helsinki and was approved by the institutional review board at Cedars-Sinai Medical Center.

RESULTS

Patient population

In total, 503 patients were included with mean age 62.5 ± 13.4 years and with SPECT MPI performed at a median of 9.0 (interquartile range 7.0 – 12.0) years post-transplant. Patient characteristics are presented in Table 1. Patients who died were older (67.4 years vs 61.1 years, p < 0.001) and less likely to undergo exercise stress (4.4% vs 23.4%, p < 0.001).

Table 1.

Variable	Patients who survived (n=389)	Patients who died (n= 114)	p-value
Age, mean ± SD	61.1 ± 13.5	67.4 ± 11.7	<0.001
Years post-transplant, mean ± SD	9.6 ± 5.3	10.5 ± 5.4	0.060
Male, n(%)	272 (69.9)	77 (67.5)	0.506
BMI, mean ± SD	26.9 ± 5.2	26.5 ± 5.5	0.150
Hypertension, n(%)	292 (75.1)	86 (75.4)	0.512
Diabetes, n(%)	130 (33.4)	45 (39.5)	0.264
Dyslipidemia, n(%)	272 (69.9)	78 (68.4)	0.814
Chronic Kidney Disease, n(%)	33 (8.5)	16 (14.0)	0.104
Donor Age, mean ± SD	31.9 ± 12.4	32.9 ± 13.5	0.509
Known CAV, n(%)	55 (14.1)	21 (18.4)	0.298
Previous Treated ACR, n(%)	31 (8.0)	8 (7.0)	0.844
Previous Treated AMR, n(%)	24 (6.2)	4 (3.5)	0.357
Proliferation Signal Inhibitor, n(%)	160 (41.1)	55 (48.3)	0.197
Statin, n(%)	300 (77.1)	77 (67.5)	0.049
Exercise Stress, n(%)	91 (23.4)	5 (4.4)	<0.001
Resting HR, mean ± SD	83.9 ± 11.4	85.0 ± 12.4	0.180
Resting systolic BP, mean ± SD	139.7 ± 18.5	144.6 ± 21.8	0.006
Exercise Peak HR, mean ± SD	142.7 ± 17.6	143.8 ± 16.8	0.890
Exercise Peak systolic BP, mean ± SD	166.5 ± 17.7	169.0 ± 15.2	0.780
Pharmacologic Peak HR, mean ± SD	111.0 ± 17.2	101.2 ± 17.9	<0.001
Pharmacologic Peak systolic BP, mean ± SD	130.5 ± 20.7	129.1 ± 19.5	0.528

Baseline population characteristics. BMI – body mass index, ACR – acute cellular rejection, AMR – antibody mediated rejection, BP – blood pressure, CAV – cardiac allograft vasculopathy, HR – heart rate, SD – standard deviation.

Imaging Results

Imaging characteristics are outlined in Table 2. Patients who died during follow up had a higher mean SSS (1.33 vs 0.74, p = 0.038) and higher mean SRS (0.76 vs 0.34, p = 0.048). Those who died also had higher stress TPD (2.82 vs 1.77, p = 0.044) and rest TPD (1.92 vs 0.61, p < 0.001). Neither mean SDS (0.53 vs. 0.42, p = 0.284) nor ischemic TPD (0.94 vs. 0.93, p = 0.971) were higher in patients who died during follow-up. Most patients had no ischemia by semi-quantitative scoring (SDS = 0, 87.1%) or quantitative analysis (ischemic TPD < 1%, 81.7%). The percentage of patients with a stress LVEF less than 50% was higher in the group of patients that died during follow up than in those who survived (13.1% vs 5.4%, p = 0.01). There was no difference in rest LVEF or change in LVEF for patients who died during follow up. Patient characteristics in patients with and without abnormal stress perfusion are shown in Table S1 http://links.lww.com/TP/C214 .

Table 2.

Variable	Patients who survived (n=389)	Patients who died (n= 114)	p-value
SSS, mean ± SD	0.74 ± 2.48	1.33 ± 3.32	0.038
SRS, mean ± SD	0.34 ± 1.82	0.76 ± 2.30	0.048
SDS, mean ± SD	0.43 ± 1.52	0.53 ± 1.65	0.284
Stress TPD, mean ± SD	1.77 ± 4.9	2.82 ± 5.76	0.044
Rest TPD, mean ± SD	0.61 ± 2.66	1.92 ± 4.41	<0.001
Ischemic TPD, mean ± SD	0.93 ± 3.69	0.94 ± 3.72	0.971
Stress LVEF, mean ± SD	66.5 ± 9.3	65.0 ± 13.3	0.480
Stress LVEF < 50%, n(%)	21 (5.4%)	15 (13.1%)	0.01
Rest LVEF, mean ± SD (n=231 / n=92)	65.0 ± 9.6	62.8 ± 12.8	0.091
Delta LVEF, mean ± SD (n=231 / n=92)	0.3 ± 4.7	0.7 ± 5.1	0.483
TID ratio, mean ± SD	1.05 ± 0.12	1.04 ± 0.11	0.35
LVEDV stress, mean ± SD	69.9 ± 24.2	74.9 ± 36.9	0.82

Imaging characteristics: LVEDV – left ventricular end-diastolic volume, LVEF – left ventricular ejection fraction, SDS – summed difference score, SRS – summed rest score, SSS- summed stress score, TID – transient ischemic dilation, TPD – total perfusion deficit, SD – standard deviation.

Diagnostic Accuracy

During follow-up 329 patients underwent invasive coronary angiography at a median time of 1.0 (IQR 0.96 – 1.29) years following SPECT MPI. CAV grade 0 was seen in 189 (57.5%) patients, grade 1 in 84 (25.5%) patients, grade 2 in 32 (9.7%) patients, and grade 3 in 24 (7.3%) patients. Patient characteristics in patients with and without significant CAV are shown in Table S2 http://links.lww.com/TP/C214 , and imaging characteristics are shown in Table S3 http://links.lww.com/TP/C214 . There were no differences in SSS (1.0 ± 3.0 vs. 0.7 ± 2.1) or stress TPD (2.3 ± 5.6 vs.1.5 ± 3.9) in patients undergoing coronary angiography post-SPECT MPI.

Diagnostic accuracy of SPECT MPI parameters for significant CAV (grade 2 or 3) in patients with invasive coronary angiography are shown in Figure 1. AUC for SSS (0.650, 95% CI 0.585 – 0.716) was higher compared to SRS (AUC 0.585, 95% CI 0.534 – 0.643, p=0.006) and SDS (AUC 0.611, 95% CI 0.550 – 0.672, p=0.031). The AUC for stress TPD (0.648, 95% CI 0.582 – 0.714) was higher than rest TPD (AUC 0.586, 95% CI 0.531 – 0.641, p=0.012), but not ischemic TPD (AUC 0.610, 95% CI 0.548 – 0.672, p=0.059). There was no significant difference between SSS and stress TPD (p=0.061).

Figure 1.

A: Kaplan-Meier estimates of time to primary endpoint for patients with summed stress scores stratified by summed stress scores (SSS). Patients with SSS ≥ 4 were more likely to die during follow-up compared to patients with SSS of 0 (p = 0.002). B: Kaplan-Meier estimates stratified by stress total perfusion deficit (TPD). Patients with stress TPD ≥ 4% were more likely to die during follow-up compared to patients with stress TPD <1% (p = 0.002).

Comparison of diagnostic accuracy in patients early (<5 years) or late post-transplant (≥ 5 years), undergoing exercise or pharmacologic stress, and with or without known CAV are shown in Tables S4 to S6 http://links.lww.com/TP/C214 . Diagnostic accuracy in patients undergoing coronary angiography within one year (n=125) was slightly higher for all parameters (results in Table S7 http://links.lww.com/TP/C214 ).

Clinical Outcomes

During a mean follow up of 5.1 ± 2.5 years, 114 patients died including 35 patients within one year. Kaplan-Meier curves stratified by categories of SSS and stress TPD are shown in Figure 2. Survival was significantly lower in patients with SSS ≥ 4 (p = 0.002) and in patients with stress TPD ≤ 4 (p = 0.002) compared to patients with normal stress perfusion. Cause of death could be determined in 64 patients, with cardiac mortality occurring in 22 (4.4%) patients. Stress TPD was higher in patients who experienced circulatory death (4.7 ± 1.8 vs. 1.9 ± 4.9, p=0.013), but SSS was not (1.6 ± 3.6 vs. 0.8 ± 2.7, p=0.175). Additionally, 14 (5.8%) patients underwent PCI following SPECT MPI. SSS was higher in patients who received PCI (2.9 ± 7.4 vs. 0.8 ± 2.4, p=0.005) but stress TPD was not (4.2 ± 2.9 vs. 1.9 ± 4.8, p=0.100).

Figure 2:

Area under the receiver operating characteristic curve and 95% confidence intervals identification of significant cardiac allograft vasculopathy (grade 2 or 3). SSS – summed stress score, SRS – summed rest score, SDS – summed difference score, LVEF – left ventricular ejection fraction, TPD – total perfusion deficit.

Unadjusted and adjusted associations with all-cause mortality are shown in Table 3. In unadjusted models, increasing SSS (unadjusted HR 1.06, p = 0.015), SRS (unadjusted HR 1.09, p = 0.016), stress TPD (unadjusted HR 1.05, p = 0.001), and rest TPD (unadjusted HR 1.08, p < 0.001) were associated with increased all-cause mortality. The multivariable model was adjusted for rest and stress perfusion as well as age, sex, years post-transplant, medical history, donor age, known CAV, history of previously treated ACR or AMR, and stress LVEF. In the multivariable model, only stress TPD (adjusted HR 1.07, p = 0.030) was associated with all-cause mortality.

Table 3.

	Unadjusted HR (95% CI)	p-value	Adjusted HR (95% CI)	p-value
SSS (per point)	1.06 (1.01 – 1.11)	0.015	1.10 (0.99 – 1.21)	0.064
SRS (per point)	1.09 (1.02 – 1.16)	0.016	0.99 (0.86 – 1.14)	0.862
SDS (per point)	1.03 (0.94 – 1.14)	0.494	--	--
Stress TPD (per %)	1.05 (1.02 – 1.08)	0.001	1.07 (1.01 – 1.13)	0.018
Rest TPD (per %)	1.08 (1.04 – 1.12)	<0.001	1.00 (0.93 – 1.08)	0.917
Ischemic TPD (per %)	1.03 (0.98 – 1.08)	0.295	--	--
Stress LVEF* (per %)	0.99 (0.98 – 1.01)	0.523	0.99 (0.97 – 1.01)	0.457
Stress LVEF < 50%*	1.71 (0.97 – 2.99)	0.062	1.22 (0.60 – 2.48)	0.583
Rest LVEF† (per %)	0.98 (0.97 – 1.00)	0.081	0.98 (0.96 – 1.00)	0.094
Change in LVEF† (per %)	1.00 (0.96 – 1.04)	0.932	1.01 (0.97 – 1.05)	0.747

Unadjusted and adjusted hazard ratios. Multivariable analyses were adjusted for stress and rest perfusion as well as age, sex, years post-transplant, medical history (hypertension, diabetes, dyslipidemia and chronic kidney disease), known CAV, previously treated acute cellular rejection, previously treated antibody mediated rejection, and stress LVEF.

^*- adjusted hazard ratio from model including summed stress score (SSS) and calculated separately for continuous and categorical left ventricular ejection fraction (LVEF).

^†- adjusted hazard ratio from model including summed stress score (SSS) but not stress LVEF.

– indicates variable not included in the multivariable analysis. SSS – summed stress score, SRS summed rest score, SDS – summed difference score, TPD – total perfusion deficit, LVEF – left ventricular ejection fraction.

Kaplan-Meier survival curves stratified by the presence of abnormal stress perfusion (SSS ≥ 4) and reduced LVEF (LVEF < 50%) are shown in Figure 3. Patients with SSS ≥ 4 and LVEF < 50% were more likely to experience all-cause mortality compared to patients with SSS <4 and LVEF ≥ 50% (log rank p = 0.008). Patients with TPD ≥ 4 were more likely to experience all-cause mortality compared to patients with TPD <4 and LVEF ≥ 50% in the setting of preserved LVEF (log-rank p=0.030) or reduced LVEF (log rank p = 0.005). One-year and annualized all-cause mortality rates across groups of combinations of SSS and LVEF are shown in Table S8 http://links.lww.com/TP/C214 .

Figure 3:

Kaplan-Meier survival curves stratified by summed stress scores (SSS) and left ventricular ejection fraction (LVEF). Patients with SSS ≥ 4 and LVEF < 50% were more likely to experience all-cause mortality compared to patients with SSS <4 and LVEF ≥ 50% (log rank p = 0.008). Patients with TPD ≥ 4 were more likely to experience all-cause mortality compared to patients with TPD <4 and LVEF ≥ 50% in the setting of preserved LVEF (log-rank p=0.030) or reduced LVEF (log rank p = 0.005).

Prognostic utility

Prognostic accuracy for all-cause mortality during the entire follow-up period is shown in Figure 4. There was a trend towards higher AUC for quantitative assessment of perfusion compared to semi-quantitative assessment, with the highest accuracy for stress TPD (AUC 0.584, 95% CI 0.533 - 0.635). Rest TPD had significantly higher AUC compared to SRS (AUC 0.579 vs 0.534, p = 0.014). However, the differences between SSS and stress TPD (AUC 0.584 vs. 0.536, p = 0.078) or SDS and ischemic TPD (AUC 0.555 vs. 0.524, p = 0.403) were not significant. Prognostic accuracy for cardiac mortality is shown in Table S9 http://links.lww.com/TP/C214 . There was a trend to higher AUC for all parameters. Stress TPD had the highest AUC (0.659, 95% CI 0.546 – 0.772), but there were no significant differences between parameters. Similar results were obtained for predicting circulatory deaths or deaths due to unknown causes.

Figure 4.

Area under the receiver operating characteristic curve and 95% confidence intervals for all-cause mortality during entire follow-up. Rest TPD had higher AUC compared to SRS (p = 0.014), with no significant differences between SSS and stress TPD (p = 0.078) or SDS and ischemic TPD (p = 0.403). SSS – summed stress score, SRS – summed rest score, SDS – summed difference score, LVEF – left ventricular ejection fraction, TPD – total perfusion deficit.

A combined model including stress TPD, stress LVEF, and history of CAV achieved higher prognostic accuracy (AUC 0.607, 95% CI 0.543 – 0.670). There were no differences in the prognostic accuracy of any individual parameter in patients with a history of CAV (AUC ranging from 0.503 to 0.584) compared to patients without known CAV (AUC ranging from 0.525 to 0.601). In the subset of patients with rest gating, rest LVEF (AUC 0.555, 95% CI 0.481 – 0.629) and stress LVEF (AUC 0.529, 95% CI 0.452 – 0.606) had similar prognostic accuracy.

Prognostic accuracy for one-year all-cause mortality with perfusion and LVEF results are shown in Figure 5. Rest TPD had the highest prognostic accuracy (AUC 0.584, 95% CI 0.496 – 0.672), followed by stress LVEF (AUC 0.578, 95% CI 0.468 – 0.687) and stress TPD (AUC 0.569, 95% CI 0.476 – 0.661). Comparison of prognostic accuracy in patients early (<5 years) or late post-transplant (≥ 5 years), undergoing exercise or pharmacologic stress, and with or without known CAV are shown in Tables S4 to S6 http://links.lww.com/TP/C214 .

Figure 5.

Area under the receiver operating characteristic curve and 95% confidence intervals for all-cause mortality within one year. There was no difference between visual semi-quantitative scoring (green) and automated assessment of perfusion with total perfusion deficit (TPD, blue). SSS – summed stress score, SRS – summed rest score, SDS – summed difference score, LVEF – left ventricular ejection fraction.

DISCUSSION

We evaluated the prognostic utility of SPECT MPI for CAV surveillance in, to our knowledge, the largest study to date with 503 cardiac transplant recipients. This is also the first study to assess the potential role of quantitative analysis compared to semi-quantitative scoring using a standard 17-segment model. Stress perfusion, using SSS or stress TPD, had the highest diagnostic accuracy for significant CAV. We also found that SSS, SRS, rest TPD and stress TPD were higher in patients who died during follow-up, but only stress TPD was independently associated with all-cause mortality. In contrast, the extent of reversible perfusion (assessed with SDS or ischemic TPD) was not associated with all-cause mortality. Greater prognostic accuracy and stratification for mortality risk was achieved by combining stress perfusion abnormality and stress LVEF. The prognostic accuracy of quantitative assessment of perfusion tended to be higher compared to semi-quantitative analysis; however, the overall prognostic accuracy was modest.

Several studies have investigated the role of SPECT MPI for diagnosis or risk-prediction in patients with known or suspected CAV.^{18–20,25,36,37} In a study including 110 patients with 14 events, stress perfusion defects on SPECT MPI involving more than 3 segments were associated with increased risk of cardiovascular death or re-transplantation.¹⁸ Wu et al. found that reversible perfusion abnormality in ≥ 6 segments was more common in patients who experienced circulatory death in a study that included 47 patients.²⁰ A study on 104 heart transplant patients showed that inhomogeneous myocardial perfusion, defined as <70% radiotracer uptake in ≥ 10% of segments in a 384 segment model, was associated with allograft dysfunction but not death or major adverse cardiovascular events.²³ Hacker et al. showed that the summed stress score (SSS), using a 20-segment model, had higher accuracy compared to summed rest score (SRS) for predicting cardiac events in a cohort of 77 patients with 10 events.²⁴ The most recent study, which included 166 patients, found that patients with normal stress SPECT MPI had lower all-cause mortality for up to 5 years and that large reversible perfusion defects were associated with increased all-cause mortality.²⁵ However, the authors used a six-segment model to calculate semi-quantitative scores. In contrast, neither SDS nor ischemic TPD were associated with all-cause mortality in our study. Importantly, our study expands on the existing literature by assessing a large, contemporary cohort of patients, using a standard semi-quantitative assessment of perfusion as well as the latest quantitative techniques.

This is the first study to assess the potential role of quantitative perfusion assessment compared to semi-quantitative assessment of SPECT MPI in patients following cardiac transplant. Quantitative assessment of perfusion with TPD has lower interobserver variability compared to expert visual interpretation.³⁸ There was no significant difference in diagnostic accuracy of stress perfusion between quantitative and semi-quantitative assessment. Additionally, we found that the unadjusted associations between stress and rest perfusion with all-cause mortality were similar between semi-quantitative and quantitative analysis of perfusion. However, only stress TPD was independently associated with all-cause mortality in the multivariable analysis. Additionally, we found that prediction of all-cause mortality during the entire follow-up period was generally higher for quantitative assessment of perfusion, with a statistically significant difference between rest TPD and SRS. We investigated several important sub-groups and noted some differences in diagnostic and prognostic accuracy. For instance, resting perfusion deficits had higher diagnostic and prognostic utility in patients undergoing pharmacologic stress. However, none of these differences were statistically significant after correcting for multiple testing. It is worth noting that the overall prognostic accuracy of either quantitative or semi-quantitative visual scoring was modest in this patient population. This limitation of SPECT MPI is likely related to the diffuse nature of CAV and an inability to identify global reductions in myocardial blood flow.^4,8

Non-invasive modalities other than SPECT-MPI are also used for CAV surveillance. Dobutamine stress echocardiography has a sensitivity between 70-80% for detecting angiographically significant CAV,³⁹ but limited sensitivity for detecting mild CAV.^40,41 Ischemia on dobutamine stress echo was not associated with cardiovascular outcomes in either study.^40,41 While CCTA has high diagnostic accuracy for CAV^9,10, there is limited evidence for its use in risk-stratification. In our cohort, SPECT had modest diagnostic accuracy for significant CAV and prognostic accuracy for all-cause mortality. There was a trend towards higher AUC for predicting cardiac mortality, but this analysis was limited by our ability to retrospectively confirm cause of death. In contrast, there is growing evidence for the diagnostic and prognostic accuracy of positron emission tomography (PET) for CAV surveillance. Several studies have demonstrated high diagnostic or prognostic accuracy of PET measurements of myocardial blood flow or myocardial flow reserve.^11–15 Importantly, we found that myocardial flow reserve had an AUC of 0.748 for all-cause mortality in the same referral population.¹² We also demonstrated that stress myocardial blood flow and myocardial flow reserve, but not regional perfusion, were independently associated with all-cause mortality.¹² Lastly, the radiation exposure from rest-stress SPECT MPI (up to 11 mSv) exceeds that for CCTA (~ 3 mSv) and perfusion PET MPI (1-2 mSv) ^42,43. Given the overall body of literature, transplant centers should consider a tailored approach with CCTA preferred in patients without known CAV for diagnostic purposes and PET preferred in patients with CAV for risk stratification.

Our study has a few limitations. History of CAV was based on invasive angiography which was performed during routine clinical care and there was variation in the time interval between angiography and SPECT MPI. Patients at our center who undergo SPECT MPI, and in particular those who subsequently undergo coronary angiography, represent a selected population, with a higher prevalence of chronic kidney disease. However, this does reflect how many centers utilize SPECT MPI for CAV surveillance. We considered a heterogenous group of patients undergoing different modes of stress, which would have lowered the precision of our estimates but also makes the results more generalizable. We do not have comprehensive information regarding changes in medical therapy that occurred in response to SPECT MPI results. This would be expected to lead to an underestimation of the association between SPECT results and clinical outcomes as well as prognostic accuracy. Additionally, we have no information regarding frailty which is an important consideration in patients post cardiac transplant. Lastly, we did not evaluate cardiovascular mortality in this retrospective patient population.

CONCLUSION

This represents the largest study of heart transplant recipients to evaluate the prognostic ability of SPECT MPI. Abnormal stress perfusion, with semi-quantitative or quantitative analysis, identified high-risk patients, but only stress TPD was independently associated with all-cause mortality. Quantitative analysis tended to demonstrate higher prognostic accuracy for all-cause mortality with potential improvement by integrating multiple parameters. However, the overall prognostic accuracy remained modest, suggesting that transplant programs should consider alternative modalities for CAV surveillance.

Abbreviations:

ACR

acute cellular rejection

AMR

antibody mediate rejection

AUC

area under curve

BMI

body mass index

CAV

coronary allograft vasculopathy

CCTA

cardiovascular computed tomography angiography

IVUS

intravascular ultrasound

LVEDV

left ventricular end diastolic volume

LVEF

left ventricular ejection fraction

MPI

myocardial perfusion imaging

PET

positron emission tomography

ROC

received operating characteristic

SDS

summed difference score

SPECT

single photon emission computed tomography

SRS

summed rest score

SSS

summed stress score

TPD

total perfusion deficit