Re-Defining Cardiac Antibody Mediated Rejection with Donor-Specific Antibodies and Graft Dysfunction

Abstract

Background:

Heart transplant recipients with donor-specific antibodies (DSAs) have increased risk for antibody-mediated rejection (AMR). However, many patients with graft dysfunction and DSA do not have evidence of AMR by endomyocardial biopsy (EMB).

Methods:

Participants from this prospective, multi-center study underwent serial EMB, echocardiogram, DSA, and donor-derived cell-free DNA (dd-cfDNA) evaluations. Outcomes were defined as pAMR+ (pAMR ≥ 1) or DSA+/LV Dysfunction (DSA presence + LVEF drop ≥10% to an LVEF≤50%). Cox regression evaluated the association between AMR categories and death or sustained (for 3 months) reduction of LVEF to <50%.

Results:

216 patients (29% Female sex, 39% Black race, median age 55 [IQR 47, 62] years) had 1,488 EMB, 2,792 DSA, 1,821 echocardiograms, and 1,190 dd-cfDNA evaluations. DSA were present in 86 patients (40%). 14 patients had isolated pAMR+ episodes and 8 patients had isolated DSA+/LV Dysfunction episodes; 2 patients had pAMR+ and then subsequently DSA+/LV Dysfunction with pAMR+. Median %dd-cfDNA was significantly higher at diagnosis of pAMR+ (0.63%[IQR 0.23,2.0], p=0.0002), or DSA+/LV Dysfunction (0.40%[IQR 0.36,1.24], p< 0.0001), compared to patients without these outcomes (0.01%[IQR 0.0001,0.10]). Both pAMR+ and DSA+/LV Dysfunction were associated with long-term clinical outcome of death (n=18) or prolonged LV dysfunction (n=10): pAMR+ (HR=2.8, 95%CI 1.03–7.4, P=0.043); DSA+/LV Dysfunction (HR=26.2, 95%CI 9.6–71.3, P< 0.001); composite of both definitions (HR=6.5, 95%CI 2.9–14.3, P< 0.001). Patients who developed pAMR+ or DSA+/LV Dysfunction within the first six months of transplant were more likely to die within 3 years post-transplant (HR=3.9, 95%CI 1.03–14.6, P=0.031).

Conclusion:

Expanding the characterization of AMR to include patients with DSA and concurrent allograft dysfunction identified DSA+ patients at risk for death and prolonged LV dysfunction.

Introduction

Advances in acute rejection identification and therapy have significantly improved short- and medium-term outcomes after heart transplantation; however, antibody-mediated rejection (AMR) remains a leading cause of late graft failure and mortality^1–3. Estimates of AMR incidence vary between 3% and 43%, with this wide range related to disparate historical definitions, underdiagnosis, as well as racial differences in disease incidence (with Black recipients having significantly higher rates of AMR)^4–6. Accurate diagnosis and effective treatment of AMR are important, given increased rates of coronary artery vasculopathy (CAV) and cardiovascular mortality among patients diagnosed with AMR^2,7. The 2013 ISHLT consensus statement provided histologic and immunopathologic diagnostic criteria for the endomyocardial biopsy (EMB) diagnosis of AMR⁸. However, this standard is limited by pathologist interpretation and lacks information on circulating donor-specific antibodies (DSA) which are independently associated with AMR and graft failure^9,10. Additionally, donor-derived cell-free DNA (dd-cfDNA) is a specific marker of allograft injury and has been associated with increased post-transplant mortality^11,12. As many patients with DSA do not develop AMR by EMB, an expansion of AMR’s definition to include those with DSA and ongoing graft dysfunction may allow for identification and treatment of allograft injury to improve post-transplant morbidity and mortality.

We hypothesized that expanding the characterization of AMR to include patients with positive DSA and echocardiographic graft dysfunction would identify heart transplant recipients with 1) higher levels of allograft injury via dd-cfDNA and 2) risk for long-term adverse clinical outcomes (death and prolonged LV dysfunction).

Methods:

Multicenter cohort and study design

Heart transplant recipients 18 years of age or older were enrolled in the Genomic Research Alliance for Transplantation study (GRAfT, NCT 02423070) while on the waitlist before transplant and followed serially after transplant; patients undergoing repeat or multi-organ heart transplantation were excluded. Patients were recruited at five regional transplant centers: Inova Schar Heart and Vascular, Johns Hopkins Hospital, University of Maryland Medical Center, Virginia Commonwealth University, and Medstar Washington Hospital Center. The full study methodology, including center immunosuppression and surveillance protocols has been previously published¹¹. Data were collected between 2015 and 2020. The institutional review boards of all centers and the NHLBI approved the study, with patients providing their informed consent prior to study enrollment. This study adheres to the principles of the Declaration of Helsinki and the ISHLT statement on Transplant Ethics. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Post-Transplant Evaluations

Patients underwent endomyocardial (EMB) evaluations, which were graded by the individual center’s pathologist utilizing the 2013 ISHLT AMR grading system, with AMR defined as pAMR 1 (H+), 1 (I+), 2, or 3⁸. Biopsies were obtained both as routine surveillance and if there was clinical suspicion for rejection; the incidence of for-cause biopsies was low (<1%). Evaluations of dd-cfDNA were sent at the time of EMB or identification of allograft dysfunction. Luminex™ (Luminex, Austin, TX) multiplex bead assays were used to evaluate the presence, phenotype, and quantity of DSA, with an MFI threshold of 1,000 used to determine a positive DSA evaluation. At all centers, clinical practice included universal treatment of patients with pAMR2 and those with DSA+/LV dysfunction; among pAMR1i or pAMR1h patients, however, treatment was reserved for patients having LV dysfunction, clinical compromise, or recurrent rejection. Center-specific AMR therapy agents are outlined in Table S1.

Outcome Measures

Three outcomes were defined: 1) pAMR+ (pAMR ≥ 1), 2) DSA+/LV Dysfunction (simultaneous positive DSA evaluation and LVEF decrease by ≥10% from previous to an LVEF ≤50%), and 3) composite (patients who met either or both definitions). Separate pAMR+ episodes either had an intervening normal EMB or an intervening 28-day time interval, accounting for follow-up time at each center to determine pAMR+ resolution. Separate DSA+/LV Dysfunction episodes had an intervening 28-day time interval. Death events within the first six months were censored, and survival outcome was censored at the last follow-up. Control patients were defined as those without pAMR+, or DSA+/LV Dysfunction throughout the study period. Prolonged LV dysfunction was defined as LVEF <50% for more than 90 days. A long-term clinical outcome was defined as having either prolonged LV dysfunction or death.

%dd-cfDNA measurement

%dd-cfDNA measurement included genotyping of donor and recipient DNA to identify informative single nucleotide polymorphisms (SNPs). After transplantation, serial plasma samples were used for cfDNA isolation, library construction, and paired-end shotgun sequencing. Sequence reads were analyzed to identify and compute the %dd-cfDNA as percentage of reads with donor single-nucleotide polymorphisms to reads of donor plus recipient single-nucleotide polymorphisms¹¹. dd-cfDNA measured within 30 days of meeting pAMR+ or DSA+/LV Dysfunction criteria was evaluated. The dd-cfDNA measurements for controls were identified as time-points without AMR, DSA, or acute cellular rejection (ACR) on histopathology. Given known early post-transplant dd-cfDNA elevation (associated with transplant surgery) and subsequent exponential decay, dd-cfDNA measurements <30 days post-transplant were excluded. Elevated %dd-cfDNA was defined as ≥0.25%, as has been validated by previous study¹¹.

Statistical Methods

Continuous variables are reported as the median (interquartile range [IQR], 1^st to 3^rd quartile) and categorical variables as the number (percentage). Baseline patient characteristics were compared with Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for categorical variables. A linear mixed model was used to compare the log-transformed dd-cfDNA levels for patients meeting outcome measures and controls (patient subgroup as a fixed effect) with a random intercept in the model, to account for individual subject-specific deviations and the within-subject correlations between repeated measures. Patient characteristics were correlated with long-term outcomes by the univariable Cox proportional hazards models. Multivariable Cox regression models were used to assess the relationship of the outcome measures (as a binary time-dependent covariate) with the outcomes of prolonged LV dysfunction and death, adjusted for patient’s age, sex and race. Kaplan-Meier method was used to estimate the cumulative incidence of the composite outcome for all patients and the rates of death and prolonged LV dysfunction among 6-month survivors in the landmark analysis, with event rates compared with the log-rank test. A two-sided P value < 0.05 was considered statistically significant throughout the analyses. Statistical analysis was performed using R version 4.2.2 (R Foundation for Statistical Computing) and SAS 9.4 (SAS Institute, Cary, NC).

Results

Patient Population

A total of 216 patients who underwent heart transplantation between July 2015 and September 2020 were included in the analysis (Figure 1). The median age was 55 years (IQR 47, 62), 29% of patients were of female sex, 39% were of Black race, and the majority of patients (60%) had nonischemic cardiomyopathy before transplantation. Those meeting composite outcome (pAMR+ and/or DSA+/LV Dysfunction) were more likely to be of Black race and have a primary diagnosis of ischemic cardiomyopathy (both p=0.025, Table 1, Table S2).

Figure 1.

Flow diagram of patients evaluated and those meeting pAMR+ or DSA+/LV Dysfunction outcomes

DSA, donor-specific antibody; DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Table 1.

Demographic and baseline clinical characteristics for patients with and without pAMR+ or DSA+/LV Dysfunction^* (N=216)

Characteristic†	No pAMR+ or DSA+/LV Dysfunction N = 192	pAMR+N = 16	DSA+/LV Dysfunction N = 10	P ‡
Age (yrs)	55 (47,62)	54 (46,62)	50 (48,55)	0.54
Sex
Female	58 (30%)	3 (19%)	1 (10%)	0.17
Male	134 (70%)	13 (81%)	9 (90%)
Race				0.086
Black or African American	70 (36%)	11 (69%)	6 (60%)
White	108 (56%)	4 (25%)	4 (40%)
American Indian or Alaskan native	1 (0.5%)	0 (0%)	0 (0%)
Asian	4 (2.1%)	1 (6.3%)	0 (0%)
Other	9 (4.7%)	0 (0%)	0 (0%)
Race				0.014
Black or African American	70 (36%)	11 (69%)	6 (60%)
Non-Black or African American	122 (64%)	5 (31%)	4 (40%)
Ethnicity				0.37
Latino or Hispanic	17 (8.9%)	0 (0%)	0 (0%)
Not Latino or Hispanic	170 (89%)	16 (100%)	10 (100%)
unknown	5 (2.6%)	0 (0%)	0 (0%)
Smoking history				0.99
No	119 (62%)	11 (69%)	4 (40%)
Yes	72 (38%)	5 (31%)	6 (60%)
BMI				0.82
<30	131 (69%)	11 (69%)	7 (70%)
≥30	60 (31%)	5 (31%)	3 (30%)
Ejection fraction (%)||	65 (60, 68)	62 (55, 70)	57 (55, 61)	0.44
Primary Diagnosis				0.025
Nonischemic cardiomyopathy	119 (62%)	7 (44%)	4 (40%)
Ischemic cardiomyopathy	46 (24%)	7 (44%)	5 (50%)
Other	26 (14%)	2 (13%)	1 (10%)
Pre-transplant LVAD	56 (29%)	4 (25%)	3 (30%)	0.67
UNOS Status				0.74
1A	149 (78%)	11 (69%)	9 (90%)
1B	42 (22%)	5 (31%)	1 (10%)
Diabetes	49 (26%)	6 (38%)	3 (30%)	0.22
Creatinine	1.2 (0.9,1.5)	1.2 (1.0,1.5)	1.3 (1.0,1.6)	0.38

^*Two patients had both pAMR+ and DSA+/LV Dysfunction

^†Data are Median (IQR) or n (%).

^‡Comparing patients with composite pAMR+ or DSA+/LV Dysfunction to patients without pAMR+ or DSA+/LV Dysfunction, with Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for categorical variables, bold indicates statistical significance

^||First echocardiogram after transplant

BMI, body-mass index; DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; LVAD, left ventricular assist device; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines; UNOS, United Network for Organ Sharing

pAMR+, DSA+/LV Dysfunction, and Composite Outcomes

During a median post-transplant follow-up of 13 months (IQR 6, 23), patients had 1,488 EMB, 2,792 DSA, 1,821 echocardiograms, and 1,190 dd-cfDNA evaluations. In total, 24 patients (11%) met composite outcome. 16 patients had 37 EMBs positive for AMR (median time to 37 positive EMBs of pAMR ≥ 1, 0.9 months [IQR 0.5, 10]), including 20 EMBs of pAMR 1 (7 pAMR1h and 13 pAMR1i), 17 EMBs of pAMR 2, and 1 EMB of pAMR3. The 37 EMBs with AMR were classified as 22 distinct AMR episodes since some patients demonstrated persistent AMR in successive EMBs. 10 patients had 13 episodes of treated DSA+/LV Dysfunction (median time to event 10 months [IQR 5, 16], Figure 2). pAMR+ occurred in 2 patients who subsequently met both pAMR+ and DSA+/LV Dysfunction criteria simultaneously; no other patients met both definitions and no other DSA+/LV Dysfunction patients had simultaneous pAMR+ (Figure 3). 4 of the 16 pAMR+ patients had recurrent pAMR+ requiring treatment; the remainder had one isolated AMR episode, with resolution of AMR by the next EMB (all within 28 days). There were 17 episodes of treated pAMR ≥2 among 8 patients. Only patients with pAMR2 (no patients with pAMR1i or pAMR1h as highest AMR grade) had LV dysfunction. Among pAMR+ patients, two patients had mixed rejection (ACR and AMR). Among DSA+/LV Dysfunction patients, two patients also had ACR at the time of this categorization.

Figure 2.

Cumulative Incidence and Donor-Derived Cell Free DNA Values of pAMR+ and DSA+/LV Dysfunction Outcomes

A) Cumulative incidence curves for antibody-mediated rejection. Within the study period, 16 patients had DSA+/LV Dysfunction, 10 patients had pAMR+, and 24 patients met composite of having pAMR+ and/or DSA+/LV Dysfunction. 2 patients had pAMR+ alone and subsequently had simultaneous AMR by both EMB and DSA+/LV Dysfunction.

B) Violin plots showing median, IQR, range, and distribution of donor-derived cell-free DNA levels in patients with pAMR≥1 by EMB (first group), DSA+/LV Dysfunction (second group), DSA+ without pAMR≥1 or LV dysfunction (third group), and DSA- without pAMR≥1 or LV dysfunction (fourth group).

* P<0.001 comparing dd-cfDNA levels of the first and second groups to the combined patients of the 3^rd and 4^th groups. P-values were based on the linear mixed models to compare mean levels of log-transformed dd-cfDNA, using a fixed effect for the patient subgroups of the three outcomes and a random intercept to account for individual subject-specific deviations and within-subject correlations for repeated measurements.

DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Figure 3.

Donor-derived cell-free DNA trends of 4 representative patients with pAMR+ and/or DSA+/LV Dysfunction. The dashline indicates positive %dd-cfDNA > 0.25%. The arrows represent treatment for AMR. Subjects 1 and 2 were the only patients meeting both pAMR+ and DSA+/LV Dysfunction outcomes. Subject 1 had decline of dd-cfDNA with post-EMB AMR treatment. Subject 2 had a higher dd-cfDNA at diagnoses of both pAMR+ and DSA+/LV Dysfunction as opposed to pAMR+ alone. Subjects 3 and 4 had isolated pAMR+ and DSA+/LV Dysfunction, respectively. dd-cfDNA detected both outcomes.

dd-cfDNA, donor-derived cell-free DNA; DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Donor-specific antibody elevation was detected in 86 patients (40%): 69/192 of those not meeting composite outcome, as compared to 9/16 of the pAMR+ (p=0.12), 10/10 of the DSA+/LV Dysfunction (p<0.0001), and 17/24 of the composite outcome (p=0.0016) groups. DSA were predominantly Class II (81%) and DQ-DSA (65%). All patients meeting DSA+/LV Dysfunction criteria had Class II DSA and half (5/10) of these patients also had Class I DSA (Table 2, Tables S3 and S4, Figure S1). LVEF was significantly lower in the DSA+/LV Dysfunction group (mean 37±13) as compared to the pAMR+ (mean 58±12) and no rejection (DSA+: mean 62±6 and DSA-: mean 56±10) groups, all P <0.001 (Table S5, Figure S2).

Table 2.

Donor-specific antibody presence for patients with and without pAMR+ or DSA+/LV Dysfunction

DSA Presence	No pAMR+or DSA+/LV Dysfunction N = 192	pAMR+ N = 16	P *	DSA+/LV Dysfunction N = 10	P †
Any DSA	69 (35.8%)	9 (56.2%)	0.12	10 (100%)	<0.001
Class I	31 (16.1%)	6 (37.5%)	0.043	5 (50%)	0.018
HLA-A	20 (10.4%)	5 (31.2%)		4 (40%)
HLA-B	18 (9.4%)	5 (31.2%)		4 (40%)
HLA-C	12 (6.2%)	1 (6.2%)		3 (30%)
Class II	55 (28.6%)	7 (43.8%)	0.25	10 (100%)	<0.001
DP	7 (3.6%)	1 (6.2%)		3 (30%)
DQ	42 (21.9%)	6 (37.5%)		10 (100%)
DR	20 (10.4%)	4 (25.0%)		6 (60%)

^*comparing pAMR+ patients vs. without those without pAMR+ or DSA+/LV Dysfunction

^†comparing DSA+/LV Dysfunction patients vs. those without pAMR+ or DSA+/LV Dysfunction

DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; HLA: human leukocyte antigen; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Allograft injury at AMR diagnosis

Median %ddcfDNA among all samples was 0.024% (IQR 0.001, 0.14). Median %dd-cfDNA was 0.63% (IQR 0.23, 2.0) at pAMR+, 0.40% (IQR 0.36, 1.24) at DSA+/LV Dysfunction, and 0.41% (IQR 0.27, 1.95) at composite outcome; this was compared to patients without composite outcome and DSA, having median %dd-cfDNA 0.008% (IQR 0.0001, 0.10), all P < 0.001 (Table 3). Median %dd-cfDNA for those with DSA and without pAMR+ and LV Dysfunction was 0.06% (IQR 0.0001, 0.14), Figure 2. Patients meeting composite outcome were more likely to have elevated dd-cfDNA (≥0.25%) than controls (OR=5.5, 95%CI 1.72–17.8, P = 0.004) (Table S6).

Table 3.

Donor-derived cell-free DNA evaluation for patients with each definition

Outcome	No. of Patients	dd-cfDNA Median (IQR)	dd-cfDNA Geometric Mean (GSE)	P *
pAMR+	16 (37 EMBs)	0.63% (0.23 − 2.0)	0.72% (1.62)	<0.001
pAMR1	20 EMBs	0.35% (0.16 − 0.74)	0.36% (1.66)
pAMR 2 or greater	17 EMBs	2.1% (1.9 − 12.3)	2.90% (2.13)
DSA+/LV Dysfunction	10 (13 episodes)	0.40% (0.36 − 1.24)	0.56% (1.88)	<0.001
Composite (pAMR+ or LV Dysfunction)	24	0.41% (0.27 − 1.95)	0.66% (1.46)	<0.001
No pAMR+or DSA+/LV Dysfunction	103	0.008% (0.0001 − 0.10)	0.004% (1.17)	--

^*Comparing each definition to patients without pAMR+ or DSA+/LV Dysfunction. P-values were based on the linear mixed models to compare mean levels of log-transformed dd-cfDNA, using a fixed effect for the patient subgroup and a random intercept to account for individual subject-specific deviations and within-subject correlations for repeated measurements.

pAMR+ occurred in 2 patients who subsequently met both pAMR+ and DSA+/LV Dysfunction criteria simultaneously; no other patients met both definitions and no other DSA+/LV Dysfunction patients had simultaneous pAMR+

dd-cfDNA, donor-derived cell-free DNA; DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; GSE, geometric standard error of the mean, pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Association of AMR with Long-Term Clinical Outcomes

There were 18 deaths and 10 patients with prolonged LV dysfunction (Table S7). Black patients were at increased risk for death (HR=3.10, 95%CI 1.17–8.28, P = 0.024) and death and/or prolonged LV dysfunction (HR=4.12, 95%CI 1.81–9.35, P = 0.0007). Other baseline patient and clinical characteristics were not associated with death or prolonged LV dysfunction. In multivariable models adjusting for age, sex, and race; outcome development was associated with death: pAMR+ (HR=2.2, 95%CI 0.8–7.8, P = 0.23), DSA+/LV Dysfunction (HR=3.7, 95%CI 0.8–17.6, P=0.095), and composite outcome (HR =3.0, 95%CI 1.04–8.8, P = 0.042). Outcome development was significantly associated with subsequent prolonged LV dysfunction/death: pAMR+ (HR=2.8, 95%CI 1.03–7.4, P= 0.043), DSA+/LV Dysfunction (HR= 26.2, 95%CI 9.6–71.3, P < 0.001), and composite outcome (HR=6.5, 95%CI 2.9–14.3, P < 0.001), Table 4. In a landmark Kaplan-Meier analysis, those with composite outcome by six months post-transplant were more likely to die than those without AMR (HR, 3.88 (95%CI 1.03–14.6, P = 0.031, Figure 4, Figure S3).

Table 4.

Univariable and multivariable analysis of pAMR+ and DSA+/LV dysfunction with risk of death and prolonged LV dysfunction^*

Models	Death		Composite Prolonged LV Dysfunction†/Death
	HR (95% CI)	P	HR (95% CI)	P
Univariable Model
pAMR+	2.86 (0.83 − 9.92)	0.097	3.58 (1.36 − 9.45)	0.010
DSA+/LV Dysfunction	4.99 (1.10 − 22.7)	0.038	23.4 (8.96 − 60.9)	<0.001
Composite pAMR+ or DSA+/LV Dysfunction	4.01 (1.41 − 11.35)	0.009	7.55 (3.45 − 16.5)	<0.001
Multivariable Model‡
pAMR+	2.15 (0.61 − 7.58)	0.23	2.77 (1.03 − 7.42)	0.043
DSA+/LV Dysfunction	3.74 (0.79 − 17.6)	0.095	26.2 (9.64 − 71.3)	<0.001
pAMR+ or DSA+/LV Dysfunction	3.03 (1.04 − 8.80)	0.042	6.46 (2.91 − 14.3)	<0.001

^*In univariable Cox analysis incorporating AMR as a time-dependent covariate.

^†Prolonged LV dysfunction was defined as LVEF <50% for more than 90 days.

^‡Each multivariable model was adjusted for patients’ age, sex and race

DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Figure 4.

The Kaplan-Meier estimate of death by presence of Composite Outcome (pAMR+ and/or DSA+/LV Dysfunction) within 6 months post-transplant. Patients who died within 6 months post-transplant were excluded.

DSA+/LV Dysfunction, donor-specific antibody presence + LVEF (left ventricular ejection fraction) drop ≥10% to an LVEF≤50%; pAMR+, pathologic AMR ≥ grade 1 by 2013 ISHLT consensus guidelines

Discussion

In this multicenter, prospective cohort evaluation of heart transplant recipients, defining AMR as the presence of DSA together with echocardiographic graft dysfunction identified patients with ongoing allograft injury (via elevated dd-cfDNA) and risk for long-term adverse outcomes. Expanding the classification of AMR to include both EMB and DSA+/LV Dysfunction criteria elucidated patients at long-term risk for mortality, not identified by EMB alone. As many heart transplant recipients develop DSA, the evaluation of concurrent graft dysfunction via echocardiography may help detect patients at risk for long-term adverse outcomes.

Donor-Specific Antibody Development

Solid-phase immunoassay evaluation of DSA has become standard practice after heart transplantation, and 28–40% of recipients are identified as having DSA^13,14. The primary risk factor for the development of DSA is pre-transplant human leukocyte antigen (HLA) sensitization, which can stem from prior cardiac surgery (including ventricular assist device [VAD] placement), pregnancy, and blood transfusions. Recipients with pre-transplant sensitization and those who develop de novo DSA after transplantation are more likely to experience rejection or mortality; however, there remains a lack of consensus regarding initiation of therapy in the setting of DSA^9,15–17. A 2013 survey of heart transplant clinicians showed that 83% would provide immunologic therapy for patients without EMB evidence of AMR and with DSA presence and a decrease in ejection fraction¹⁸. There remains a lack of evidence, however, regarding the risk profile and treatment effect of such patients. Additional gaps in knowledge remain regarding the relevance of DSA, as many patients with DSA do not develop AMR. DSA subtype (Class II DSA or DQ-DSA), which were both found in a higher proportion of patients with AMR in this current study, have been shown to portend a higher risk for rejection and mortality, but DSA subtype alone is insufficient to prompt therapy^13,14. Rather, ongoing evidence of allograft dysfunction/injury via echocardiogram or dd-cfDNA along with DSA positivity may be sufficient evidence to initiate treatment in a similar fashion to AMR by histopathology.

Rejection Diagnosis and Therapy

Diagnosing AMR via EMB was standardized in the 2013 ISHLT criteria and encompasses features observed with hematoxylin and eosin (H&E) staining, and immunohistochemical or immunofluorescence staining, respectively evaluating endothelial activation, intravascular macrophages, and complement depsosition⁸. The publication of criteria for pathologic AMR held promise for 1) phenotypic identification of rejection in patients with clinically observable rejection and without evidence of ACR and 2) in situ identification of the humoral immune reaction to the cardiac allograft. To date, however, many patients with DSA and clinical evidence of allograft dysfunction by echocardiography do not have evidence of AMR (or ACR) via EMB. These biopsy-negative rejection patients likely have a similar pathologic profile to those diagnosed with AMR by EMB (pAMR+ patients). There are multiple reasons that may explain the lack of EMB positivity in such patients. First, EMB identification of AMR is subject to sampling error, which has been difficult to study but is gaining increasing evidence as magnetic resonance imaging (MRI) of rejection has identified patients with evidence of focal allograft injury and monocyte deposition in locations outside of the traditional right-ventricular EMB sampling location¹⁹. Second, it is not clear within the pathophysiologic understanding of AMR the quantity and distribution of DSA needed to elicit the complement cascade and thus a positive immunofluorescence diagnosis, and, finally, there may be significant interrater variability between pathologists in the diagnosis of AMR. The CARGO II study described only 28% agreement between pathologists in the diagnosis of clinically significant ACR²⁰. While there has not been an analogous evaluation of interrater reliability in AMR, there may be similar discordance given the increased complexity in diagnosing AMR.

The 2022 ISHLT Guidelines for the Care of Heart Transplant Recipients separately describes treatment strategies for AMR and DSA, despite a likely significant overlap in these groups²¹. Many strategies are utilized to treat AMR, which include both ‘non-humoral’ agents, such as corticosteroids, cytolytic therapies (anti-thymocyte globulin), and intravenous immunoglobulin (IVIg), as well as antibody-directed strategies, such as pheresis, rituximab (monoclonal antibody targeting CD20 on B cells), eculizumab (complement C5 inhibitor), and bortezomib (proteasome inhibitor). These strategies are mainly derived from the treatment of AMR in renal transplantation, with evaluations of their effectiveness in treating AMR and DSA in heart transplantation largely guided by expert consensus^22–24. Expansion of the diagnosis of AMR to include patients with DSA and allograft dysfunction may allow for randomized, controlled clinical trials evaluating the effectiveness of these therapies.

Multi-Modality Rejection Diagnosis

Multiple biomarkers now exist to aid in the diagnosis of cardiac allograft rejection beyond EMB, yet there remains little guidance in the incorporation of multiple markers to define a risk profile for acute rejection and long-term adverse outcomes²⁵. The present study describes the contribution of DSA to these outcomes, but DSA presence alone is likely suboptimal. In this and other evaluations, Class II DSA and DSA-DQ were present in higher proportion of those with AMR diagnosis^13,14. More recently, a study has also shown that non-HLA antibodies, such as antibodies to the angiotensin II type 1 receptor (AT1R), increase the risk of rejection and adverse post-transplant outcomes^26,27. Other biomarkers of AMR include %dd-cfDNA, as well as emerging modalities such as microRNA and cardiac MRI^28–30. The utilization of these investigations must also include the effect of known risk factors for AMR such as Black race and pre-transplant antibody sensitization^6,15,17. The compilation of these pre-transplant risk factors and the multiple available post-transplant biomarkers into an AMR risk prediction calculation may allow for diagnostic accuracy greater than that available with EMB.

The results of this study provide evidence of the inadequacy of EMB as the ‘gold standard’ for AMR diagnosis, as 1) most pAMR+ patients had isolated episodes with resolution by 28 days and 2) EMB-negative patients with DSA and allograft dysfunction had a similar risk profile as those with AMR by EMB. Thus, the presence of DSA and allograft dysfunction (by echocardiographic abnormalities and/or signs of heart failure) may be sufficient to initiate AMR therapy. Dd-cfDNA has been validated for the prediction of AMR both as a composite outcome with ACR and as a separate outcome by the GRAfT investigators with 88% sensitivity, 85% specificity and 0.95 AUC for AMR^11,31,32. The identification of allograft injury via dd-cfDNA in pAMR+ and DSA+/LV Dysfunction diagnoses of AMR in the current study gives further evidence of dd-cfDNA in the identification of AMR. Recent literature has described the negative predictive value of dd-cfDNA, allowing the avoidance of EMB in settings of negative dd-cfDNA^33,34. However, EMB is still routinely obtained in recipients with elevated dd-cfDNA and signs of allograft dysfunction to differentiate ACR from AMR and provide appropriate therapy. In this setting, a growing number of centers are utilizing MMDx Molecular Microscope^® Diagnostic system (Transcriptome Sciences, Alberta, Canada), which has shown evidence of AMR among EMB samples negative for pAMR graded rejection³⁵. As more is learned about the available molecular biomarkers of rejection, these modalities may prove superior to traditional EMB diagnosis and provide diagnoses for patients previously identified as having biopsy-negative rejection.

Limitations

This study’s strengths include a prospective, multicenter cohort study with serial assessments of EMB, allograft function, DSA and %dd-cfDNA. There were 10 patients identified as having AMR with the DSA+/LV Dysfunction criteria: this number is likely underpowered to evaluate the effect of this phenotype, resulting in wide confidence intervals for these patients in the prediction of the long-term outcome of prolonged LV dysfunction/death. Similarly, the low number of deaths (total n=18, 2 in patients with DSA+/LV Dysfunction) likely resulted in the lack of statistical significance in the identification of DSA+/LV Dysfunction as an independent predictor of mortality. Those meeting composite outcome had increased risk of mortality both independently and within a combined long-term adverse outcome of mortality and prolonged LV dysfunction; this latter outcome, however, is limited as it includes ejection fraction—a component of the DSA+/LV Dysfunction criteria for rejection. Future clinical validations can assess the risk profile of patients with DSA+/LV Dysfunction criteria.

Conclusions

Expanding the classification of AMR to include patients with DSA and allograft dysfunction allowed for the identification of patients at significant risk for mortality. Through the identification of patients with concurrent DSA and allograft dysfunction, we propose a new definition of AMR which is associated with allograft injury and an increased risk for long-term adverse outcomes. This work elucidates a pathway to distinguish deleterious post-transplant DSA, which can inform and expand our definition of AMR. Whether patients with DSA+/LV Dysfunction criteria benefit from AMR treatment in the absence of AMR by EMB forms the basis for future clinical investigations.

Clinical Perspective.

What is new?

• Expanding the characterization of antibody mediated rejection (AMR) to include those with donor-specific antibodies and graft dysfunction identifies patients at risk for death or prolonged ventricular dysfunction.

• Elevation of donor-derived cell-free DNA was found both in AMR defined by biopsy and in cases of donor-specific antibody positivity with graft dysfunction, showing a similar degree of allograft injury in both rejection subtypes.

• Patients who developed either AMR defined by biopsy or donor-specific antibody positivity with graft dysfunction had higher risk of death by 3-years post-transplant than those without rejection.

What are the implications?

• Treating patients with donor-specific antibodies and graft dysfunction may mitigate risk of long-term adverse outcomes.

• Patients with donor-specific antibodies and graft dysfunction may not require biopsy to initiate rejection therapy, potentially resulting in improved post-rejection outcomes.

• These results can stimulate clinical trials evaluating the effects of identification and treatment of patients with a revised characterization of AMR.

Abbreviations:

ACR

acute cellular rejection

AMR

antibody-mediated rejection

dd-cfDNA

donor-derived cell-free DNA

CAV

coronary artery vasculopathy

DSA

donor-specific antibody

EF

ejection fraction

ISHLT

International Society for Heart & Lung Transplantation

LV

left ventricle

MRI

magnetic resonance imaging

VAD

ventricular assist device