Author + information
- Received September 21, 2017
- Revision received September 17, 2018
- Accepted October 26, 2018
- Published online January 16, 2019.
- Muhammad Imran, MBBSa,∗ (, )
- Louis Wang, MBBSa,
- Jane McCrohon, MBBBa,
- Chung Yu, MBBSa,
- Cameron Holloway, MBBSa,
- James Otton, MBBSa,
- Justyn Huang, MBBSa,
- Christian Stehning, PhDb,
- Kirsten Jane Moffat, BAppScc,
- Joanne Rossc,
- Valentina O. Puntmann, MDd,
- Vassilios S. Vassiliou, MBBSe,
- Sanjay Prasad, MBBSe,
- Eugene Kotlyar, MBBSa,
- Anne Keogh, MBBSa,
- Christopher Hayward, MBBSa,
- Peter Macdonald, MBBSa and
- Andrew Jabbour, MBBSa
- aHeart and Lung Transplant Unit, St. Vincent’s Hospital, Sydney, Australia
- bPhilips GmbH Innovative Technologies, Hamburg, Germany
- cMedical Imaging Department, St. Vincent’s Hospital, Sydney, Australia
- dInstitute for Experimental and Translational Cardiovascular Imaging, Goethe University Hospital, Frankfurt, Germany
- eCMR, Royal Brompton Hospital, Imperial College London, London, United Kingdom
- ↵∗Address for correspondence:
Dr. Muhammad Imran, Heart and Lung Transplant Unit, St. Vincent’s Hospital, 390 Victoria Street, Darlinghurst, NSW 2010, Australia.
Objectives This study aimed to determine the role of T1 mapping in identifying cardiac allograft rejection.
Background Endomyocardial biopsy (EMBx), the current gold standard to diagnose cardiac allograft rejection, is associated with potentially serious complications. Cardiac magnetic resonance (CMR)–based T1 mapping detects interstitial edema and fibrosis, which are important markers of acute and chronic rejection. Therefore, T1 mapping can potentially diagnose cardiac allograft rejection noninvasively.
Methods Patients underwent CMR within 24 h of EMBx. T1 maps were acquired at 1.5-T. EMBx-determined rejection was graded according to International Society of Heart and Lung Transplant (ISHLT) criteria.
Results Of 112 biopsies with simultaneous CMR, 60 were classified as group 0 (ISHLT grade 0), 35 as group 1 (ISHLT grade 1R), and 17 as group 2 (2R, 3R, clinically diagnosed rejection, antibody-mediated rejection). Native T1 values in patients with grade 0 biopsies and left ventricular ejection fraction >60% (983 ± 42 ms; 95% confidence interval: 972 to 994) were comparable to values in nontransplant healthy control subjects (974 ± 45 ms; 95% confidence interval: 962 to 987). T1 values were significantly higher in group 2 (1,066 ± 78 ms) versus group 0 (984 ± 42 ms; p = 0.0001) and versus group 1 (1,001 ± 54 ms; p = 0.001). After excluding patients with an estimated glomerular filtration rate <50 ml/min/m2, there was a moderate correlation of log-transformed native T1 with high-sensitivity troponin T (r = 0.54, p < 0.0001) and pro–B-type natriuretic peptide (r = 0.67, p < 0.0001). Using a T1 cutoff value of 1,029 ms, the sensitivity, specificity, and negative predictive value were 93%, 79%, and 99%, respectively.
Conclusions Myocardial tissue characterization with T1 mapping displays excellent negative predictive capacity for the noninvasive detection of cardiac allograft rejection and holds promise to reduce substantially the EMBx requirement in cardiac transplant rejection surveillance.
Acute allograft rejection remains among the most common complications in the first year after transplantation (1), with 20% of patients experiencing at least 1 episode of acute cellular rejection in this period (1,2) and 10% of patients experiencing antibody-mediated rejection (AMR) (3). Acute allograft rejection often results in allograft dysfunction and is a main determinant of mortality and morbidity in the early post-transplant period (1). Moreover, recurrent and chronic rejection leads to allograft vasculopathy, one of the most common indications for redo transplantation (4–7). Therefore, accurate and timely diagnosis of allograft rejection is paramount to enable early and effective treatment.
Histological analysis of an endomyocardial biopsy (EMBx) is the current gold standard for the diagnosis of cardiac allograft rejection. Many transplant centers have intense surveillance programs with recipients undergoing 12 to 15 biopsies in first 12 months. An EMBx is an invasive procedure associated with a 3% to 6% risk of serious complications, including carotid artery puncture, pneumothorax, tricuspid regurgitation, cardiac arrhythmias, and cardiac tamponade (8–11). Furthermore, EMBx and histological analysis are neither very accurate nor reproducible, with high false positive and false negative results reported because of sampling error and interobserver variability (12–14). Therefore, an alternative diagnostic strategy is required that is less invasive and more accurate.
Cardiac magnetic resonance (CMR) is noninvasive and has the capacity to assess regions of myocardium not accessible to EMBx. T1 and T2 mapping imaging sequences enable accurate and reproducible detection of myocardial interstitial edema and fibrosis (15–19). Their role in the detection and monitoring of acute myocarditis and myocardial ischemia is well established (15–18). Studies have also demonstrated the role of T1 mapping in the detection of interstitial myocardial fibrosis in scleroderma and rheumatoid arthritis and in the post–myocardial infarction setting (20–22). Interstitial edema is also an important marker of acute allograft rejection (23,24), whereas interstitial fibrosis is a hallmark of low-grade chronic rejection (25). We hypothesized that CMR would detect acute cardiac allograft rejection and studied the role of T1 mapping in the diagnosis and management of allograft rejection in heart transplant recipients.
Patients and study design
In this observational, prospective, cross-sectional study, all patients undergoing heart transplantation from April 1, 2014, to December 31, 2015, at a single center (St. Vincent’s Hospital, Sydney, Australia) were screened. Exclusion criteria were standard contraindications to CMR, including claustrophobia and inability to lie supine. The study was approved by the St. Vincent’s Hospital Human Research Ethics Committee (HREC/13/SVH/66). All patients gave written informed consent. All CMR studies were performed within 24 h of routine surveillance cardiac biopsies at 6, 8, 10, 12, 20, 24, 32, and 52 weeks after transplantation. If patients had clinically significant rejection as determined by histology, CMR was repeated along with the repeat biopsy after a course of treatment with pulse immunosuppressive therapy. Six patients in the cohort also underwent a nonroutine cardiac biopsy for clinical suspicion of rejection by an independent transplant physician and were also recruited into the study.
Serum high-sensitivity troponin T (hsTnT) and pro-B-type natriuretic peptide (pro-BNP) were also measured within 24 h of cardiac biopsies.
Nontransplant patients without any background cardiac history and a low pre-test probability of cardiomyopathy who were referred to our center for CMR for nonspecific symptoms and were reported to have a normal CMR by an independent specialist were used for comparison.
CMR studies were performed on a 1.5-T scanner (Achieva, Philips Medical Systems, Best, the Netherlands) equipped with a 32-channel coil. Steady-state free precession (SSFP) cine images in standard long-axis (4-, 3-, and 2-chamber) and short-axis views were performed. (26–28). An SSFP, single breath-hold, modified Look-Locker inversion recovery (MOLLI) sequence was used to acquire T1 maps in a single midventricular short-axis plane (flip angle 50°, voxel size 1.8 × 1.8 × 8.0 mm, 8 images from 2 adiabatic pre-pulse induced inversions (3,(2),5)) (29,30). Intravenous gadobutrol (Bayer Healthcare, Leverkusen, Germany) was administered in dose of 0.1 mmol/kg body weight, according to the local clinical protocol only to those patients whose estimated glomerular filtration rate was greater than 40 ml/min/1.73 m2. T1 maps were also acquired 20 min after contrast administration. The hematocrit was measured within 24 h of CMR to calculate extracellular volume (ECV), as previously described (31,32).
All CMR images were analyzed using commercially available software (CMR42, Circle Cardiovascular Imaging, Inc., Calgary, Alberta, Canada). Regions of interest were drawn along the interventricular septum, as well as circumferentially in the midventricular short-axis slice to acquire septal and global T1 values (26–28,33) (Figure 1A). Care was taken to avoid the endocardium and epicardium. After manual motion correction of each individual image, septal and global T1 values were calculated by fitting an exponential model to data from 8 images at different inversion times (26–28,30,33). Myocardial partition coefficients (λ) and ECV fractions were calculated as previously described: λ = ΔR1myocardium / ΔR1blood; ECV = λ (1 − hematocrit) (32).
The cardiac EMBxs were performed from a right internal jugular approach. Biopsies were stained with hematoxylin and eosin and were reported by an anatomic pathologist blinded to CMR results. The biopsies were graded as 0, 1R, 2R, or 3R, in accordance with International Society of Heart and Lung Transplantation (ISHLT) criteria (24). Patients with a diagnosis of AMR on the basis of a combination of clinical suspicion, allograft dysfunction, and serological and tissue markers (C4d staining on immunohistochemistry and elevated donor-specific antibodies) were also included. Patients who were treated for clinically diagnosed rejection by an independent physician because of a high clinical index of suspicion on the basis of signs and symptoms of rejection, including chest pain, palpitations, dyspnea, weight gain, and ankle edema, along with a decline in echocardiography-derived left ventricular ejection fraction (LVEF) despite unremarkable biopsies, were also included.
In asymptomatic patients, pulse immunosuppressive therapy is generally indicated for ISHLT grade 2R, 3R, or AMR (34); therefore, the cohort was grouped into 3 pre-specified groups: group 0, ISHLT grade 0; group 1, ISHLT grade 1R; and group 2, grade 2R, 3R, AMR, and clinically diagnosed rejections.
Cardiac biopsies were also stained with Masson’s trichrome stain to quantify fibrosis by using semiautomated purpose-designed local software (Figures 1B and 1C).
Statistical analysis was performed using GraphPad Prism 7.00 (GraphPad Software, Inc., La Jolla, California) and Microsoft Excel (Microsoft, Redmond, Washington). Categorical data are expressed as number and percentage, whereas continuous variables are expressed as mean ± SD. A p value of <0.05 was considered significant. For the comparison of normally distributed variables, unpaired Student’s t-test, paired t-test, analysis of variance, and Tukey’s multiple comparisons tests were used as appropriate. Correlation was measured using Pearson’s coefficient. A total of 19 randomly selected CMR studies were used for interobserver variability by using Bland and Altman plots and coefficients of variance. A receiver-operating characteristic curve was used to measure sensitivity and specificity.
A total of 169 scans were performed (118 biopsy-matched CMR scans in 34 heart transplant recipients and 51 CMR scans in nontransplant healthy control subjects) (Table 1). Logistic limitations (scanner access) prevented some patients from having concurrent CMR scans. Six scans were excluded: 1 for significant motion artifact on CMR; 1 for inadequate tissue sampling on biopsy; 1 because of the presence of severe mitral regurgitation; and 3 because patients could not lie flat long enough to perform T1 mapping. A total of 89% of the scans were performed within the first year after transplantation. A total of 12 scans (11%) were performed more than 12 months post-transplantation along with clinically indicated biopsies.
Of 112 biopsies with simultaneous CMR, 60 (54%) demonstrated no rejection (group 0 [ISHLT grade 0]), 35 (31%) demonstrated only mild rejection (group 1 [ISHLT grade 1R]), and 17 (15%) demonstrated clinically significant rejection requiring pulse immunosuppression (group 2 [2R, 3R, clinically diagnosed rejection, AMR, and 2R on a repeat biopsy after recent pulse immunosuppression]). Of 34 patients, 7 patients had clinically significant rejections. The majority of rejections occurred in the first 12 months (n = 12), whereas 2 occurred in 1 patient in the second year after transplantation (3R and 2R), and 3 rejection episodes (2 AMR and 1 clinically diagnosed) occurred in another patient more than 5 years post-transplantation (Patients #5 and #1, respectively, in Online Figure 1).
Normal T1 values in heart transplantation
To establish normal native T1 values in heart transplant recipients, patients with both ISHLT grade 0 biopsies and an LVEF >60% were selected (Table 2). Their mean T1 value was marginally higher compared with nontransplant healthy control subjects (983 ± 42 ms vs. 974 ± 45 ms; p = 0.30), with no statistically significant difference (Figure 2A). Their septal T1 values were higher than global T1 values (983 ± 42 ms vs. 969 ± 39 ms; p < 0.0001); the difference was small, although statistically significant (Figure 2B). There was no significant correlation between native T1 values and donor’s age, sex, time post-transplantation, or transplant procedure ischemic time (Figures 2C and 2D).
Native T1 across rejection groups
Comparing across various rejection groups, the native T1 values were significantly higher in patients with clinically significant rejection: group 2 (1,066 ± 78 ms) versus group 0 (984 ± 42 ms; p = 0.0001) and versus group 1 (1,001 ± 54 ms; p = 0.001) (Figure 3A, Table 3). There was no statistically significant difference between group 0 and group 1. On comparison of individual rejection grades and rejection types, T1 values were significantly higher in both AMR (1,137 ± 25 ms) and grade 2R/3R (1,091 ± 97 ms) compared with grade 0 (984 ± 42 ms; p = 0.0001) and 1R (1,001 ± 54 ms; p = 0.001) (Figure 3B). The values were also significantly higher in patients with clinically diagnosed rejection (1,052 ± 11 ms) compared with grade 0 (p = 0.01), but this difference was not statistical significant compared with grade 1R. The patients who still had grade 2R rejection on biopsy within 1 to 3 weeks of treatment with pulse immunosuppressive therapy for grade 2R/3R had lower T1 values (969 ± 10 ms) compared with grade 2R/3R (p = 0.001) and AMR (p = 0.0001).
Comparing septal and global T1 values within each individual rejection group, there was no statistically significant difference between measurement techniques (Figure 3C).
Analysis of repeated observations was made in the small subset of patients who experienced significant rejection. To account for this, we also analyzed the data by selecting only the first episode of each rejection grade within individual patients. This reduced the total number of biopsies to 55 in 34 patients with simultaneous CMR. A total of 27 (49%) biopsies demonstrated no rejection (group 0 [ISHLT grade 0]), 18 (33%) demonstrated only mild rejection (group 1 [ISHLT grade 1R]), and 10 (18%) demonstrated clinically significant rejection requiring pulse immunosuppression (group 2 [2R, 3R, clinically diagnosed rejection, AMR, and 2R on a repeat biopsy after recent pulse immunosuppression]). The native T1 values were still significantly higher in patients with clinically significant rejection: group 2 (1,088 ± 87 ms) versus group 0 (977 ± 46 ms; p < 0.0001) and versus group 1 (1,007 ± 56 ms; p = 0.003) (Online Figure 2).
Tracking response to pulse immunosuppression with T1 mapping
Six patients with significant rejection underwent serial studies. Native T1 mapping values were increased during acute rejection episodes (1,093 ± 76) and reduced significantly after pulse immunosuppressive therapy in all but 1 patient (996 ± 43) (p = 0.02, paired t-test) (Figure 3D, Online Figures 2 and 3).
Comparison of extracellular volume fraction with rejection groups and interstitial fibrosis
Intravenous contrast medium was administered in 21 patients with 51 CMR scans to measure ECV fraction. Numbers were significantly smaller than the cohort size as a result of renal impairment, which occurs commonly after heart transplantation (albeit often transiently secondary to medication side effects). There was no statistical difference among rejection groups; however, interpretation is significantly limited by the small sample size (only 4 CMR scans were performed with contrast during periods of significant rejection).
The degree of fibrosis was measured in tissue sample using Masson’s trichrome stain and semiautomated software. There was no significant correlation between ECV and the degree of fibrosis (r = 0.20, p = 0.22) (Figure 4A). Comparing native T1 values with the degree of fibrosis, there was a weak, nonstatistically significant correlation (r = 0.21, p = 0.051) (Figure 4B).
Comparison of native T1 with different parameters
Both LVEF and mass were weakly correlated with native T1 (LVEF vs. T1: r = 0.22, p = 0.02; left ventricular mass vs. T1: r = 0.22, p = 0.02). There was a modest correlation between log-transformed native T1 and hsTnT (r = 0.34, p = 0.001), as well as pro-BNP (r = 0.61, p < 0.0001) (Figures 4C and 4E). After excluding patients with an estimated glomerular filtration rate <50 ml/min/m2, this correlation improved (log T1 vs. log hsTnT: r = 0.54, p < 0.0001; logT1 vs. log pro-BNP: r = 0.67, p < 0.0001) (Figures 4D and 4F).
Native T1 specificity and sensitivity
After plotting native T1 values on a receiver-operating characteristics curve, the area under the curve was 0.89. Using a T1 value of 1,029 ms as a cutoff, the sensitivity for detecting clinically significant rejection was 93%, with a specificity of 79% and negative predictive value of 99% (Figure 5A). A total of 19 CMR studies were randomly selected, and native T1 mapping values were measured by 2 independent CMR physicians. The interobserver variability was measured using the Bland-Altman method of comparison (Figure 5B). The coefficient of variance was 1.3%.
This study explores the role of CMR-based myocardial T1 mapping in heart transplantation and demonstrates that T1 mapping is highly sensitive for the diagnosis of clinically significant cardiac allograft rejection and is able to track recovery after pulse immunosuppressive therapy.
Most studies to date focusing on the role of CMR in heart transplantation have used less reproducible sequences to define edema and fibrosis, including short tau inversion recovery sequence (STIR), T1-weighted spin-echo sequence for global relative enhancement (gRE) and late gadolinium enhancement (LGE) (35) or recruited patients more than 6 months after transplantation, missing the most critical period to diagnose rejection (36). A study by Miller et al. (37) using T1 mapping demonstrated a trend toward increased T1 values in patients with rejection but did not achieve statistical significance. Our study used a well-validated T1 mapping sequence (17) and recruited patients from 6 weeks post-transplantation (average time from transplantation date to first CMR was 10 weeks); and the majority of the scans (n = 61, 55%) were performed in first 6 months after transplantation.
This study establishes a normal T1 mapping range for heart transplant recipients by using a widely available MOLLI sequence. We found that the normal native T1 value in heart transplant recipients with histological grade 0 rejection (and normal LVEF) was 983 ± 45 ms, which is comparable to our nontransplant healthy control subjects (974 ± 42 ms) and established normal values in published reports (30).
Native T1 values were significantly higher in patients with clinically significant rejection compared with patients with nonclinically significant rejection or no rejection. T1 mapping values in patients who have not received heart transplants are elevated in conditions with known extracellular space expansion secondary to both edema and interstitial fibrosis (15–18,21,22). Myocardial edema is an important pathological marker of acute allograft rejection (23). To help determine whether the native T1 value represented edema or fibrosis, we quantified the degree of fibrosis using Masson’s trichrome stain. Only a weak correlation was observed between T1 values and histologically quantified fibrosis (nonstatistically significant), a finding suggesting that the majority of T1 change was attributable to interstitial edema rather than fibrosis. It is also important to note that patients can develop scar tissue from repetitive biopsy of the same site, thereby reducing the accuracy of the association.
A subset of patients with ISHLT grade 1R rejection (mild) had elevated markers of myocardial injury (hsTnT and pro-BNP). These markers correlated fairly well with T1 values but not with ISHLT rejection grade. This finding suggests that some patients are experiencing low-grade active myocardial damage and edema, potentially missed by histological grading. A prospective randomized outcome study comparing T1 mapping and cardiac biopsies would help determine the clinical relevance of this discrepancy.
Using a T1 cutoff 1,029 ms, we demonstrate high sensitivity (93%) and adequate specificity (79%) with an excellent negative predictive value (99%). In our study we missed only 1 significant episode of ISHLT grade 2R rejection. In contrast to the other patients with clinically significant rejection, no change in T1 value or LVEF was observed after the administration of pulse corticosteroids in this patient, thus bringing into question the accuracy of the histological diagnosis. Comparison with the current gold standard, EMBx, was used in this study; however, future prospective randomized studies with outcome endpoints will better determine the best way to diagnose and track cardiac allograft rejection. Notwithstanding, the majority of biopsies in our cohort (85%) displayed no clinically significant rejection, and T1 mapping has promise to reduce the number of surveillance cardiac biopsies dramatically in the first year after transplantation.
All patients with AMR or ISHLT grade 3R rejection had values higher than 1,100 ms. All patients with grade 2R or clinically diagnosed rejection had values in the range of 1,030 to 1,070 ms, with the exception of the 1 patient mentioned earlier. This finding suggests that future, larger studies may be able to determine T1-based cutoff values that prompt further investigation for AMR, a condition that is notoriously difficult to diagnose and requires tailored immunosuppression.
Although there was no statistically significant difference in ECV values among groups, the sample size was too small to draw any conclusions. Given the high prevalence of renal impairment (>50%), it is unlikely that ECV measurement, which requires gadolinium contrast administration, is a practical method for determining myocardial expansion in this cohort.
Our study also demonstrates that native T1 mapping tracks recovery from clinically significant rejection after pulse immunosuppressive therapy. T1 values fell from elevated levels in patients with clinically significant rejection after therapy. Furthermore, a hypothesis-generating finding of our study was that several patients had normalized their T1 values after pulse immunosuppressive therapy despite a persistent ISHLT grade 2R on their serial biopsy. This finding suggests that perhaps the histopathological recovery may lag behind the reduction in CMR-determined myocardial edema.
A total of 102 scans (93%) were performed within 24 h of cardiac biopsy, whereas 8 CMR studies were performed just outside the pre-specified 24-h period for logistic reasons. None of these biopsies revealed significant rejection, and none of these patients had any pulse immunosuppression or change in their immunosuppressive therapy between cardiac biopsy and CMR. The prospective, observational study was conducted over a relatively short period of 12 months. The sample size is relatively small, although not unreasonable for a group of transplant recipients. Cardiac biopsies were used as the gold standard comparator for T1 mapping; however, it is well recognized that this has poor reproducibility with high interobserver variability. Cardiac biopsy samples were taken from the right ventricular septal endocardium, whereas T1 mapping regions of interest were placed in the midwall to avoid blood pool artifact. A prospective randomized study with morbidity and mortality endpoints is required to evaluate further the safety of replacing cardiac biopsy-diagnosed rejection with a CMR-based diagnosis.
T1 mapping is highly sensitive for the diagnosis of clinically significant cardiac allograft rejection and tracks recovery after pulse immunosuppressive therapy. The technique demonstrates excellent interobserver reproducibility and holds promise to be a routine noninvasive method of cardiac allograft rejection surveillance in the first year after heart transplantation.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Our research demonstrates that the use of noninvasive and highly reproducible native T1-mapping is able to reduce the number of routine surveillance cardiac biopsies after cardiac transplantation substantially and therefore can improve overall patient care.
TRANSLATIONAL OUTLOOK: The results of this study warrants a prospective randomized controlled trial using mortality and morbidity endpoints before native T1 mapping may replace the current gold standard.
The study was funded by the National Health and Medical Research Council and St. Vincent’s Clinical Foundation, Australia. Dr. Stening has been employed by Philips Healthcare. Dr. Keogh has conducted clinical trial research for Actelion, Pfizer, United Therapeutics, Arena, Acceleron, Bayer, Respira, GlaxoSmithKline, and Gilead. Dr. Macdonald has received an institutional research grant from Novartis; has been on the advisory boards of Novartis and AstraZeneca; has received speaker honoraria from Servier; and has received travel support from Transmedics. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- antibody-mediated rejection
- cardiac magnetic resonance
- extracellular volume
- endomyocardial biopsy
- high-sensitivity troponin T
- International Society of Heart and Lung Transplantation
- left ventricular ejection fraction
- pro–B-type natriuretic peptide
- Received September 21, 2017.
- Revision received September 17, 2018.
- Accepted October 26, 2018.
- 2019 American College of Cardiology Foundation
- Deckers J.W.,
- Hare J.M.,
- Baughman K.L.
- Hinojar R.,
- Foote L.,
- Arroyo Ucar E.,
- et al.
- Ugander M.,
- Bagi P.S.,
- Oki A.J.,
- et al.
- Lima J.A.C.,
- Donekal S.
- Kali A.,
- Choi E.Y.,
- Sharif B.,
- et al.
- Ntusi N.A.,
- Piechnik S.K.,
- Francis J.M.,
- et al.
- Herskowitz A.,
- Soule L.M.,
- Mellits E.D.,
- et al.
- Puntmann V.O.,
- D’Cruz D.,
- Smith Z.,
- et al.
- Puntmann V.O.,
- Voigt T.,
- Chen Z.,
- et al.
- Puntmann V.O.,
- Carr-White G.,
- Jabbour A.,
- et al.
- Usman A.A.,
- Taimen K.,
- Wasielewski M.,
- et al.