Introduction

Cardiac allograft vasculopathy (CAV) is unique to heart transplant recipients, affecting both epicardial coronary arteries and the microvasculature. It is marked by intimal thickening and fibrosis, tapering of epicardial vessels, and decreased myocardial blood flow resulting in restrictive physiology. Unfortunately, CAV is common among heart transplant recipients, with 30% to 45% of patients having some degree of epicardial CAV by 5 years post-transplant and 50% to 65% at 10 years.^1,2 In addition to being a prevalent condition, it carries a significant risk of mortality. In the International Society for Heart and Lung Transplantation (ISHLT) registry, CAV is one of the most common causes of death, and when graft failure is included (listed as a distinct mode of death, although often caused by CAV), it is the most common.¹

Evaluation for CAV historically has been centered on invasive coronary angiography and stenosis severity, and the current ISHLT definition of and grading criteria for CAV does not extend beyond epicardial angiography.³ The microvasculature, where abnormalities have been linked with adverse events independent of epicardial CAV,⁴ is not comprehensively assessed with angiography. Physiologic assessment of CAV, both invasive and noninvasive, has recently been demonstrated to provide greater discrimination. Invasively measured fractional flow reserve and markers of microvascular dysfunction (coronary flow reserve and index of microcirculatory resistance) have both been demonstrated to predict death or cardiac retransplantation.^5,6 Noninvasive assessment of myocardial flow reserve with rubidium-82 positron emission tomography (PET) myocardial perfusion imaging also has demonstrated reduced survival with reductions in myocardial blood flow reserve.^7,8 The goal of this analysis was to assess the prognostic ability of total myocardial blood flow reserve (MBFR) and isolated microvascular CAV measured by ¹³N-ammonia PET on post-transplant outcomes.

Methods

Study population

In 2016, our center changed the CAV screening protocol from biannual coronary angiography with intravascular ultrasound (IVUS) alternating with dobutamine stress echocardiography (DSE), replacing DSE with ¹³N-ammonia PET because of our and others contemporary experience with DSE (eg, limited sensitivity and frequency of nondiagnostic studies).^9,10 Patients with an abnormal PET were referred for coronary angiography with IVUS if renal function was acceptable. This observational retrospective cohort study included all adult heart transplant recipients who underwent ¹³N-ammonia PET myocardial perfusion imaging from June 2016 through September 2017, targeting 5 years of follow-up for clinical outcomes. Final study follow-up date was December 31, 2021. Demographic and clinical data were collected from the electronic medical record. CAV was defined and graded according to the ISHLT angiographic criteria.¹¹ Right heart catheterization was performed at the same time as the angiogram. Angiograms were graded by 1 of 6 board certified interventional cardiologists who were not aware of the PET results. Angiograms were included if they were performed within 3 months before the PET or in the subsequent 12 months to maximize the number of patients with contemporary angiographic data (median time between studies: 322 days).

Study definitions

Patients were classified into 2 groups according to the total MBFR: >2.0 and ≤2.0. An MBFR value of 2.0 was chosen as the cutoff based on the definition of microvascular dysfunction according to standardized COVADIS (Coronary Vasomotion Disorders International Study Group) diagnostic criteria,¹² prior data demonstrating the prognostic significance of a PET-derived MBFR below 2.0 in atherosclerosis,^13,14 and a receiver-operator characteristic curve analysis demonstrating an optimal cutoff of 1.96 (Supplemental Figure 1). Patients were secondarily classified, and epicardial CAV was defined by PET-derived ischemia (summed difference score [SDS] ≥2) or ISHLT CAV 1 or greater angiographically, and an MBFR >2.0. Microvascular CAV was defined as an MBFR ≤2.0 with no evidence of epicardial CAV (PET SDS <2 or ISHLT CAV grade 0 angiographically). Patients with mixed CAV had both ischemia (SDS ≥2 or ISHLT CAV 1 or greater) and an MBFR ≤2.0.

PET protocol

PET myocardial perfusion imaging was conducted using a Siemens PET-CT mCT 64-slice scanner. For rest images, 8 to 12 mCi of ¹³N-ammonia was injected intravenously before acquisition of cardiac perfusion images. For the stress portion of the examination, patients underwent pharmacological stress with dipyridamole (0.56 mg/kg), adenosine (140 μg/kg/min), or regadenoson (0.4 mg), at which time they received another 8 to 12 mCi of ¹³N-ammonia after a delay of 50 minutes (5 half-lives) from rest imaging. All patients were closely monitored for transient conduction system abnormalities. After all images were obtained, the reconstructed perfusion images were analyzed using Invia software (4DM) according to standard-of-care methods. Myocardial ischemia or infarction was assessed using described semiquantitative assessment of a 17-segment model by board-certified nuclear cardiologists.¹⁵ Summed rest, summed stress, and summed difference (stress-rest) scores were calculated. Stress and rest left ventricular ejection fraction was measured from gated images. Myocardial blood flow (in milliliters per minute per gram of myocardial tissue) was calculated at rest and stress using validated 2-compartment models. MBFR was calculated as the ratio of stress to rest myocardial blood flow and adjusted for rate pressure product using a reference value of 9,000. Low-dose computed tomography was used for attenuation correction and allowed for visual estimation of coronary artery calcium, which has previously been validated.¹⁶ Average radiation effective dose for each PET study was 2 to 3 mSv, less than the annual background radiation in the United States.¹⁷

Study outcomes

The primary study endpoint was death or retransplantation, ascertained from medical records. Major secondary endpoints included cardiovascular death or retransplantation and cardiovascular hospitalization, which involved heart failure hospitalization, acute coronary syndrome, percutaneous coronary intervention, and acute rejection.

Statistical analysis

Demographic and clinical variables were expressed as mean ± SD or median (IQR) for continuous variables depending on normality and count (%) for categorical variables. Group comparisons were made with the chi-square, Fisher exact, and the Mann-Whitney U test where appropriate. Kaplan-Meier survival analysis and Cox proportional hazards regression were performed to compare outcomes between the groups. Cardiovascular mortality was assessed using a Fine and Gray subdistribution hazard model to account for the competing risk of noncardiovascular mortality. A multivariable Cox model was generated to assess for significant predictors and confounders of death or retransplantation. Variables considered included those listed in Table 1. The adjusted model was generated using Akaike Information Criterion model selection to find the best-fit model. A secondary model was generated with backward selection, including all variables and removing the least significant variables until all included variables had a P value <0.10. The difference in post-PET survival between groups was determined using restricted mean survival time.²⁰ Restricted mean survival time is analogous to the area under the survival curve and, in this analysis, represents the mean event-free survival from PET scan to the end of study follow-up. A 2-tailed P value <0.05 was considered significant. Analyses were performed using SAS version 9.4 (SAS Institute, Inc). The study was approved by the Columbia University Irving Medical Center Institutional Review Board.

Results

Baseline characteristics

A total of 206 consecutive heart transplant recipients who underwent a PET scan were assessed, of which 181 patients comprised the study cohort (17 were excluded for high resting myocardial blood flow [>1.1 mL/min/g], 5 had technical difficulties, and 3 were repeat studies for patients previously included) (Figure 1). Baseline characteristics of the study cohort are presented in Table 2. Overall, 181 patients were enrolled, of whom 110 had an MBFR >2.0 and 71 had an MBFR ≤2.0. Median follow-up was 4.7 years (IQR: 4.0-5.2 years). The median age of individuals with reduced MBFR was greater at the time of the PET scan (65 years [IQR: 56-69 years] vs 59 years [IQR: 47-67 years]; P = 0.02), and there were trends toward a greater median donor age (34 years [IQR: 25-48.5 years] vs 29 years [IQR: 21-44 years]; P = 0.07) and a greater median time since transplant (9 years [IQR: 5-12 years] vs 7 years [IQR: 4-10 years]; P = 0.11). Patients with an MBFR ≤2.0 were significantly more likely to have had an ischemic cardiomyopathy as indication for heart transplant, have had a prior stroke, had smoked before transplantation, and had diabetes mellitus; they were less likely to be taking aspirin. They also had significantly more prior acute cellular rejection, antibody mediated rejection, and chronic kidney disease (Table 2). Among patients who had rejection, the time from AMR to PET was less among those with an MBFR ≤2.0, whereas time from ACR to PET was greater. No patient fulfilled the ISHLT criteria for restrictive physiology, and right atrial pressure, pulmonary capillary wedge pressure, and cardiac index were similar between groups. Individual hemodynamic abnormalities were associated with the risk of death or retransplantation (Supplemental Table 1).

PET characteristics

Dipyridamole was the predominant coronary vasodilator and was utilized for 95% of the studies with no difference between groups (Table 3). There were no significant conduction disturbances following vasodilator administration in this cohort. At rest, total myocardial blood flow was similar between the 2 groups (0.94 mL/min/g [IQR: 0.85-1.03 mL/min/g] vs 0.90 mL/min/g [IQR: 0.82-1.0 mL/min/g]; P = 0.08), whereas there was a significant difference in total stress myocardial blood flow (2.31 mL/min/g [IQR: 2.03-2.62 mL/min/g] vs 1.47 mL/min/g [IQR: 1.22-1.69 mL/min/g]; P < 0.0001), which drove the difference in the MBFR (2.60 [IQR: 2.20-3.0] vs 1.62 [IQR: 1.35-1.80]; P < 0.0001). Rest and stress left ventricular ejection fraction were lower in the MBFR ≤2.0 group; however, both remained within the normal range (Table 3). Ischemia was more common among those with an MBFR ≤2.0, with significant ischemia (>5% of myocardium) accounting for most of the difference. Coronary calcium was more prevalent in the MBFR ≤2.0 group (37.5% vs 19.8%; P = 0.01); however, visually estimated coronary artery calcium score categories did not significantly differ (Table 3).

Clinical outcomes

Patients with an MBFR ≤2.0 experienced more than a 7-fold increased risk of death or retransplantation (HR: 7.05; 95% CI: 3.2-15.6; P < 0.0001) (Central Illustration). The survival difference manifested soon after the PET (2-year absolute decrease 9.7%; 95% CI: 4.1%-14.5%; P < 0.0001) and expanded with time (5-year absolute decrease 33.8%; 95% CI: 26.5%-39.9%; P < 0.0001) (Figure 2). This translated into a 5-year restricted mean survival that was 302.8 days less (95% CI: 260.2-345.4 days; P < 0.0001) (Figure 3). Peak stress myocardial blood flow was equally prognostic (peak myocardial blood flow <1.8: HR: 7.12; 95% CI: 3.11-16.34; P < 0.0001). The risk of cardiovascular hospitalization was 10 times greater among those with MBFR ≤2.0 (HR: 10.1; 95% CI: 3.43-29.9; P < 0.0001). Similarly, the risk of cardiovascular death was 12-fold greater for those with an MBFR ≤2.0 (HR: 12.0; 95% CI: 2.64-54.12; P = 0.001). Although a binary MBFR cutoff of 2.0 was thoughtfully selected, the association between MBFR and an increased risk of death or retransplantation demonstrated a gradient of effect. We separated the cohort into tertiles (Figure 4), and when compared with individuals with an MBFR >2.0, patients with an MBFR <1.5 had a risk of death or retransplantation that was more than 8 times greater (HR: 8.45; 95% CI: 3.49-20.45; P < 0.0001), whereas individuals with an MBFR between 1.5 and 2.0 had a 6-fold increased risk (HR: 6.10; 95% CI: 2.56-14.59; P < 0.0001). When treated as a continuous variable, the risk of death or retransplantation increased 19% (HR: 1.19; 95% CI: 1.11-1.27; P < 0.0001) for each 0.1 decrease in MBFR.

There was an interaction between MBFR and time post-transplant (interaction P = 0.04); therefore, we analyzed outcomes by time post-transplant. As an independent variable, time post-transplant was not predictive of death or retransplantation (HR: 1.003 per month post-transplant; 95% CI: 0.998-1.007; P = 0.28). Comparing the early (1-<5 years), mid-term (5-10 years), and late (10+ years) periods post-transplant, an MBFR ≤2.0 portended a poor prognosis at all times (Supplemental Figure 2). Early post-transplant patients (n = 53) in the low MBFR group had a mean survival that was 6 months less in the subsequent 5 years (191.1 days; 95% CI: 82.0-300.1 days; P = 0.07), and midterm patients (n = 66) had similar outcomes (187.1 days; 95% CI: 72.5-301.8 days; P = 0.03). Among individuals with an MBFR ≤2.0 more than 10 years after transplantation (n = 62), the impact was substantial as patients had a mean survival that was 470 days less (95% CI: 298.9-640.7; P < 0.0001) during the following 5 years.

Using a multivariable model to adjust for between group differences, MBFR ≤2.0 remained a significant risk factor for death or retransplantation (adjusted HR: 4.04; 95% CI: 1.72-9.46; P = 0.001) (Table 1). Similar results were obtained when a multivariable model was generated by backward selection (adjusted HR: 3.64; 95% CI: 1.52-8.68; P = 0.004) (Supplemental Table 2). When the outcome was limited to cardiovascular death, the magnitude of risk was even greater for those with an MBFR ≤2.0 (adjusted HR: 7.45; 95% CI: 1.44-38.5; P = 0.02), and MBFR ≤2.0 and PET diagnosed ischemia were the only significant predictors after adjustment (Supplemental Table 3).

Outcomes by subtype of CAV

Patients were classified into 4 groups: epicardial CAV, microvascular CAV, mixed CAV, and no CAV, as described in the Methods section. When PET-only criteria were used, individuals with mixed CAV had the worst outcomes (5-year survival 25.4%; 95% CI: 4.2%-56.3%; time lost 691.0 days; 95% CI: 356.9-1,025.0 days) followed by microvascular CAV (5-year survival 64.4%; 95% CI: 51.1%-76.7%; time lost 282.0 days; 95% CI: 163.5-400.5 days) (Figure 5A). Patients with epicardial CAV with an MBFR >2 had outcomes (5-year survival 83.3%; 95% CI: 46.5%-99.9%; time lost 38.5 days; 95% CI: −30.4 to 107.4 days) that were similar to patients without CAV (5-year survival 92.5%; 95% CI: 86.4%-97.0%; time lost 43.1 days; 95% CI: 18.7-36.7 days). The findings were replicated using angiography to classify epicardial CAV (Figure 5B). Patients with mixed CAV (5-year survival 54.9%; 95% CI: 38.9%-70.5%; time lost 387.7 days; 95% CI: 221.4-554.1 days) and microvascular CAV (5-year survival 62.8%; 95% CI: 43.9%-79.9%; time lost 297.5 days; 95% CI: 132.7-462.4 days) continued to have poor outcomes, whereas individuals with isolated epicardial CAV (5-year survival 92.2%; 95% CI: 81.8%-98.5%; time lost 35.7 days; 95% CI: −6.0 to 77.3 days) and patients without CAV (5-year survival 91.6%; 95% CI: 83.1%-97.2%; time lost 47.5 days; 95% CI: −2.1 to 97.1 days) continued to have similar outcomes.

We next assessed if mixed CAV and microvascular CAV remained associated with an increased risk of death or retransplantation in regression with multivariable adjustment. Mixed CAV (HR: 7.37; 95% CI: 3.31-15.41; P < 0.0001) and microvascular CAV (HR: 2.96; 95% CI: 1.52-5.76; P = 0.001) replaced MBFR in a multivariable model that included all covariates listed in Table 1. In this model, both microvascular CAV (aHR: 3.86; 95% CI: 1.58-9.40; P = 0.003) and mixed CAV (aHR: 11.19; 95% CI: 3.57-35.01; P < 0.0001) remained substantial risk factors for death or retransplantation.

Cause of primary endpoint

The primary study endpoint of death or retransplantation occurred in 10 (9.1%) of those with an MBFR >2.0 and 25 (35.2%) of those with MBFR ≤2.0 (P < 0.0001). The causes are listed in Table 4. Cardiovascular causes of death were more common (11 vs. 2; P = 0.0007) in the MBFR ≤2.0 group. Among the group with an MBFR >2.0, 80% of the deaths were caused by infection or malignancy.

Discussion

Cardiac allograft vasculopathy remains a cause of significant morbidity and mortality following heart transplantation and requires assiduous screening. Anatomic assessment of CAV has long been the standard; however, this study assessed the addition of noninvasive physiological measures and demonstrated the following: 1) patients with an MBFR ≤2 had a 7-fold increased risk of death or retransplantation over the subsequent 5 years; 2) microvascular CAV was independently associated with death or retransplantation; 3) reduced MBFR was associated with an increased risk of death or retransplantation at all times post-transplant, but especially after 10 years; 4) patients with mixed CAV had the greatest risk of death or retransplantation; and 5) patients with MBFR >2.0 had excellent survival (1 year 100%, 5 years 92%).

Coronary angiography has been performed since the 1960s, and it was adopted as the method to screen for CAV after it was appreciated to be a nefarious post-transplant complication as early as 1969.^18,19 The introduction of intravascular ultrasound in 1992²⁰ furthered the understanding of the underlying pathophysiology of CAV and appreciation that CAV may develop early following heart transplantation. Much like the diagnosis of coronary atherosclerosis, which evolved from angiography and stenosis severity to include intravascular imaging and now physiological measurements (eg, fractional flow reserve [FFR], resting full-cycle ratio, instantaneous wave-free ratio), physiology is now being explored for the diagnosis of CAV. Invasive pressure-temperature sensor guidewire assessment of intracoronary physiology has demonstrated increased mortality associated with a reduced FFR or evidence of microvascular disease.⁵ More recently, evidence of epicardial CAV (FFR <0.80 in the left anterior descending artery) and microvascular disease (coronary flow reserve [CFR] ≤2.0 or index of microcirculatory resistance ≥25) at 1 year post-transplant has been shown be associated with a 3- and 2-fold increased risk of death or retransplantation after 10 years. Noninvasive measurements of MBFR (the noninvasive counterpart of CFR) have been associated with poor outcomes. In a study of patients who were a median of 8.2 years post-transplant, McArdle et al⁷ demonstrated using rubidium-82 PET that individuals with a reduced MBFR (<1.75) had a 441% increased risk of MACE (14 MACE events) during 1.5 years of follow-up. A smaller study of 89 patients who were, on average, 7 years from transplant found that an MBFR <1.5 was associated with a 3-fold increased risk of death (40 deaths).⁸ Here, we report the results of our study, in which we used a different MBFR cutoff (MBFR ≤2.0) and had a longer follow-up time of nearly 5 years. Our results support the previously published data and highlight the continuous hazard of reduced MBFR post-transplant. In our study, when MBFR was considered as a continuous variable there was a consistent decrement in survival (19% increased risk of death or retransplant for each 0.1 decrease in MBFR), and patients with an MBFR of <1.5 had a more than 8-fold increased risk of death or retransplantation. Additionally technical differences between the extraction fraction of ⁸²Rb (used in the 2 other studies), ¹³N-ammonia (this study), and vasodilators (one utilized regadenoson, which has reduced hyperemia compared with dipyridamole) may have also contributed.^21,22 Nevertheless, the findings are consistent across a range of reduced MBFR cutoffs and irrespective of the radiotracer used, highlighting the generalizability of these findings and this imaging modality.

CAV is known to be a pan-arterial disease with involvement of both the epicardial arteries and the microvasculature. The modalities recommended in the ISHLT guidelines to diagnose CAV (invasive: coronary angiography, intravascular imaging may be considered; noninvasive: dobutamine stress echocardiography, coronary computed tomography angiography) do not allow for careful assessment of the microvasculature. Pathological studies have demonstrated that stenotic microvascular disease was associated with nearly 3-year reduction in post-transplant survival, independent of epicardial CAV.⁴ Invasively measured microvascular CAV or dysfunction has been shown to portend a poor short-term prognosis when elevated at 1 year after transplantation.²³ A more recent multicenter study including 254 patients found that invasively measured microvascular disease at 1 year post-transplant, defined as a CFR ≤2.0 or index of microcirculatory resistance of ≥25 in the absence of significant epicardial disease, resulted in a 233% increased risk of death or retransplantation after 10 years of follow-up. These findings, in a population different from this study, in some respects parallel the findings of our study where isolated microvascular CAV was associated with nearly a 4-fold increased risk of death or retransplantation after 5 years. This further highlights the importance of assessing the microvasculature, because microvascular CAV, whether detected on pathology, via invasive pressure-temperature sensor guidewire assessment, or PET, has been consistently associated with an increased risk of death or retransplantation, independent of epicardial CAV.⁶

The prevalence of CAV increases with time post-transplantation (30%-45% at 5 years, 50%-65% at 10 years) and it also becomes a leading cause of death as time progresses.¹ In this study, patients ranged from 1 year to 28 years post-transplant (75% within the first 11 years) and MBFR was prognostic at each time point. Even early reductions in MBFR (<5 years since transplant) carry significant consequence, because patients with an MBFR ≤2.0 had 3 times the risk and a 21% absolute increase in the risk of death or retransplantation in the next 5 years. Put a different way, these patients had 6 less months alive during the 5 years follow-up than those with an MBFR >2.0. An alternative and more optimistic way of looking at these data is that an MBFR>2.0 portends an excellent prognosis: 1-year survival of 100%, 2-year survival of 99.1%, and 5-year survival of 92%.

Study limitations

The first limitation is that this is a single-center, nonrandomized cohort study of consecutive patients who underwent a PET scan, and although the findings are consistent with other external cohorts, the findings may differ among different populations. Next is that although the study included 181 heart transplant recipients and is the largest to evaluate PET-derived MBFR, the sample size remains modest. As such, multivariable models carried the risk of overfitting; however, consistent results across multiple model-building techniques helped to assuage those concerns. Third, there is no codified definition of microvascular CAV. We used a definition that was consistent with other studies,^6,12 and the findings were similar whether angiography or PET was used to define epicardial CAV; however, this definition requires further validation. Furthermore, IVUS was not performed on each patient, limiting the comparison of IVUS with MBFR; however, recent data have shown the importance of physiology even when accounting for IVUS.⁶ Because angiography and PET scans were not contemporaneous, the inclusion of angiograms that were performed up to 1 year after the PET scan may have inflated the influence of angiographic measures given the progressive nature of CAV. Last, the role of the immunosuppression protocol on MBFR cannot be fully assess in this study and will need to be evaluated in a prospective randomized study.

Conclusions

Reduced myocardial blood flow reserve on ¹³N-ammonia PET was associated with an increased risk of death or retransplantation, irrespective of the time since transplantation. Patients with epicardial CAV and reduced MBFR had the greatest risk of death or retransplantation. Microvascular CAV was associated with an increased risk of death or retransplantation, independent of epicardial CAV, and warrants screening.

Footnotes