Antibody-independent Microvascular Inflammation Impacts Long-term Risk in Heart Transplantation

Abstract

Background:

Microvascular inflammation (MVI) after heart transplantation (HT) can occur with or without circulating anti-human leukocyte antigen donor-specific antibodies (DSAs). We sought to characterize the relationship between MVI, with or without accompanying DSA, and post-transplant outcomes.

Methods and Results:

We analyzed 8305 endomyocardial biopsies from 832 adult and pediatric HT recipients between July 1, 2013, and October 31, 2023. Endomyocardial biopsies were graded by consensus guidelines, with MVI defined as histological evidence of AMR grade ≥1. Rejection phenotypes were classified as no rejection, isolated cellular rejection, DSA-negative MVI, and DSA-positive MVI. Cox models with time-varying covariates were constructed to evaluate associations with incident cardiac allograft vasculopathy (CAV) and mortality, adjusting for donor and recipient age. Among 832 HT recipients, 238 developed CAV and 121 died over a median follow-up of 4 years (interquartile range 2.3–6.4 years). Compared with individuals who never experienced biopsy-proven rejection, DSA-negative MVI was independently associated with CAV (hazard ratio [HR] 1.47, 95% confidence interval [CI] 1.00–2.16, P = 0.047). DSA-positive MVI was associated with mortality (HR 1.97, 95% CI 1.07–3.64) with DSA-negative MVI demonstrating directional concordance (HR 1.50, 95% CI 0.87–2.57), independent of CAV (HR 1.71, 95% CI 1.13–2.58). These associations remained consistent when stratified by adult and pediatric subgroups and in a 6-month landmark sensitivity analysis.

Conclusions:

MVI, with or without DSA, may be harmful in HT, extending recent renal findings to thoracic transplantation. Understanding the mechanistic basis for these results will be essential for identifying novel targets for therapeutic modulation and prolonging graft survival.

Repetitive alloimmune injury is a central driver of poor long-term survival after solid organ transplantation. Although microvascular inflammation (MVI)—inflammatory vascular injury with or without overt antibody-mediated rejection (AMR)—is associated with poor outcomes after renal transplantation,¹ its importance in thoracic transplantation where vascular injury represents a major life-limiting complication remains unclear. This uncertainty is clinically significant. Cardiac allograft vasculopathy (CAV)—the major barrier to long-term survival after heart transplantation (HT)²—originates from alloimmune vascular injury, which may be subclinical or histologically subtle.² Current diagnostic criteria classify AMR-related injury by the presence of MVI, emphasizing circulating donor-specific antihuman leukocyte antigen (HLA) antibodies (donor-specific antibodies [DSAs]),^3,4 with therapies directed at reducing DSAs to limit graft injury. Yet, the extent to which MVI, particularly in DSA-negative settings, portends future CAV or graft failure remains unclear. Treatment decisions in this context remain individualized, with substantial institutional variation, owing to limited large-scale experience in tracking the relevance of MVI on outcomes.⁴

To address this gap, we leveraged a granular longitudinal transplant database at a high-volume transplant center to evaluate whether biopsy-defined MVI, with or without accompanying DSA, is associated with adverse post-transplant outcomes. Specifically, and analogous to renal transplantation, we tested the hypothesis that DSA-negative MVI is independently associated with the development of CAV, providing a histological signal of alloimmune injury that may extend beyond conventional serological definitions.

Methods

Study Cohort

We conducted a retrospective cohort study of all adults and children who underwent HT at Vanderbilt University Medical Center (VUMC) between July 1, 2013, and October 31, 2023. This time frame was intentionally selected for two reasons: (1) the International Society for Heart and Lung Transplantation (ISHLT) Working Formulation⁵ for diagnosis of AMR was published in 2013 and our center began routinely reporting both histological evidence of AMR (pAMR1-h) and immunological evidence (pAMR1-I, based on C4d deposition) on biopsies, and (2) the time window allowed for adequate longitudinal follow-up for coronary surveillance to assess CAV status.

Adult HT recipients received induction therapy with corticosteroids alone or in combination with basiliximab or antithymocyte antibody (thymoglobulin), followed by maintenance immunosuppression consisting of calcineurin inhibitor, antiproliferative agent (eg, mycophenolate mofetil), and prednisone taper per institutional protocol with planned discontinuation by 4–6 months post HT. Pediatric HT recipients uniformly received thymoglobulin for induction, followed by a 5-day steroid taper. Maintenance immunosuppression in children included a calcineurin inhibitor and antiproliferative agent for the first 3 months post HT, with a mammalian target of rapamycin inhibitor replacing the antiproliferative agent subsequently.

Surveillance for circulating DSAs was conducted at post-HT weeks 2, 4, 12, and 24, followed by at least once every 6 months or as clinically indicated. Per institutional protocol in adults, surveillance endomyocardial biopsies (EMBs) were performed coinciding with the steroid taper and were typically performed at post-HT weeks 2, 4, 8, 12, 16, and 20 or as clinically indicated. Baseline coronary angiography typically occurs between 2 and 3 months post HT, with annual angiographic surveillance initiated at 1 year post HT. This study was approved by the VUMC Institutional Review Board (#200551).

Exclusion Criteria

1. HT recipients transplanted at other centers and who transferred care to VUMC without verifiable EMB data;

2. Veterans Affairs patients who underwent HT at VUMC but received post-transplant care exclusively within the Veterans Affairs system; and

3. Absence of coronary angiography data for CAV surveillance.

Data Collection

Recipient data were extracted from VUMC’s electronic health record and donor data were abstracted from the United Network for Organ Sharing database. Recipient variables included age at transplant, sex, comorbidities, and DSAs. Donor variables included donor age and method of organ procurement (donation after brain death or circulatory death).

Rejection Grading

After EMB collection and processing, sections were transferred to the pathology department in formalin or in saline. Formalin-fixed, paraffin-embedded specimens underwent sectioning, followed by hematoxylin and eosin staining. Specimens transferred in saline were embedded in optimal cutting temperature compound and underwent sectioning and immunofluorescence staining for C4d.

All EMB specimens were graded by cardiac pathologists with expertise in HT. Acute cellular rejection (ACR) was graded according to the ISHLT criteria,⁶ with ACR grades 2R and 3R considered clinically significant rejection. AMR was graded according to the ISHLT consensus.⁵ Histological evidence of AMR (pAMR1-h) was characterized by endothelial swelling in the absence of pericapillary C4d deposition. Immunological evidence of AMR (pAMR1-i) was characterized by pericapillary C4d deposition in >50% of capillaries (excluding perimyocyte staining). The presence of both histological and immunological evidence of AMR was defined as pAMR2, with pAMR3 representing severe AMR with concomitant hemodynamic compromise.³

For primary analytical purposes, MVI was defined as the presence of pAMR1-h, pAMR1-i, or pAMR2/3. MVI was further stratified as DSA positive or DSA negative based on serological status at the time of biopsy. Individuals who never demonstrated biopsy-proven rejection over the follow-up period comprised the no rejection group and those with ACR only (without MVI or AMR) were classified as ACR only. Of note, per institutional protocol, adults with pAMR1-i, pAMR2, or pAMR3 are treated as definitive AMR with therapies targeted at reducing DSA. In contrast, pAMR1-h typically does not result in intensified immunosuppression (unless other markers of graft injury, such as graft dysfunction, are present), but, rather, is closely followed for resolution.

CAV surveillance was performed by invasive coronary angiography, considered the gold standard by current ISHLT consensus.² In a small number of cases where individuals underwent redo HT or autopsy, CAV grading was adjudicated by direct pathological examination of the coronary arteries. Briefly, CAV-1 represents mild angiographically evident disease in the coronary vasculature, CAV-2 denotes ≥70% in a single major vessel or severe branch disease in 2 coronary distributions, and CAV-3 includes left main coronary disease or severe multivessel involvement.² Additionally, CAV-3 could be diagnosed in the context of graft dysfunction with underlying CAV-1 or CAV-2.² Clinically significant CAV was considered CAV grade ≥1, which is associated with an increased risk of graft loss and mortality after transplantation.^2,7–9

Outcomes

The primary outcomes were incident CAV and all-cause mortality.

Statistical Analyses

Continuous variables were summarized as median (interquartile range [IQR]) and categorical variables were described as number (%). To account for time-varying exposures, Cox proportional hazards models were constructed with rejection phenotype and DSA status treated as time-varying covariates. In the mortality analysis, CAV was also treated as a time-varying covariate. DSA status was also modeled as a separate variable, because it may be present independent of rejection and can fluctuate over time irrespective of biopsy findings. This approach allows for granular alignment of evolving immunological status (eg, rejection states, presence or absence of circulating DSA) with individual-level risk. Rejection status and DSA exposure were updated at each biopsy, and person-time was recalculated accordingly. The CAV outcome was based on the last date of coronary evaluation.² Patients without CAV were censored at their last coronary interrogation. For all time-to-event models, time zero was the date of the HT.

Adjusted hazard ratios (HRs) and their 95% confidence intervals (CIs) are reported. Kaplan–Meier curves were constructed to visualize unadjusted time-to-event distributions. A 6-month landmark sensitivity analysis was conducted to address early post-HT immune dynamics and surveillance intensity and to assess the robustness of our primary model. Within the landmark analysis, rejection phenotypes were categorized into their most severe rejection episode during the 6-month post-HT period (hierarchically: DSA-positive MVI > DSA-negative MVI > ACR only > no rejection). All analyses were performed in R version 4.4.0 (R Core Team).

Results

We identified 832 adults and children who underwent HT between July 1, 2013, and October 31, 2023. The median recipient age at HT was 51 years (IQR 33–62 years) and 273 recipients (33%) were female. The median donor age was 28 years (IQR 19–36 years). In total, 8305 EMBs were included in the analysis, spanning the spectrum of rejection phenotypes (Figure 1). The first EMB post HT was performed at a median of 11 days (IQR 7–15 days). Among the cohort, 287 individuals (34%) developed circulating DSAs, first detected at a median of 187 days (IQR 16–720 days) post HT. Full demographics are outlined in Table 1. Figure 2A and 2B depict the Kaplan–Meier curves for incident CAV and all-cause mortality for the entire cohort.

Figure 1.

The distribution of rejection phenotypes across 8305 endomyocardial biopsies. ACR, acute cellular rejection; AMR, antibody-mediated rejection (defined here as pAMR2 or pAMR3); MVI, microvascular inflammation (defined here as pAMR1-h or pAMR1-i).

Table 1.

Baseline Cohort Demographics (N = 832)

Characteristic	Median (Q1, Q3) or No. (%)
Recipient age at transplant (years)	51 (33, 62)
Recipient sex (female)	273 (33)
Donation type
Circulatory death	135 (16)
Brain death	697 (84)
Donor age (years)*	28 (19, 36)
Days to first biopsy	11 (7, 15)
Died during follow-up	121 (15)
CAV	238 (29)
Donor-specific antibodies	287 (34)

^*N = 831.

Figure 2.

CAV-free survival and all-cause mortality following heart transplantation. (A) Kaplan–Meier curve for CAV-free survival after transplantation in the entire cohort. Individuals were censored at the last coronary assessment (eg, angiography, pathology of explanted heart at time of redo transplant or autopsy). (B) Kaplan–Meier curve for all-cause mortality after transplantation in the entire cohort. CAV, cardiac allograft vasculopathy.

We first examined outcomes among individuals who experienced pAMR1 (defined as either pAMR1-h or pAMR1-i^3,5) during follow-up (Figure 3A). Among 154 individuals who experienced pAMR1 without progression to pAMR2, 63 (40.9%) developed CAV. An additional 10 individuals progressed from pAMR1 to pAMR2, with 3 (30.0%) subsequently developing CAV. In contrast, among those who experienced pAMR2 without prior pAMR1 (n = 9), 5 individuals (55.6%) developed CAV. Notably, 4 of 15 individuals (26.7%) who experienced pAMR2 followed by a later episode of pAMR1 also developed CAV. These findings suggest that MVI—captured at the pAMR1 stage—may represent an early and clinically meaningful precursor to vascular injury, even in the absence of progression to pAMR2/3. For comparison, 40 of 123 individuals (32.5%) in the ACR only group developed CAV and 123 of 521 recipients (23.6%) with no biopsy-proven rejection during follow-up developed CAV. These findings support the clinical relevance of MVI, particularly in the form of isolated pAMR1, as a high-risk phenotype associated with increased CAV risk. Although the number of individuals with sequential rejection phenotypes was small, the high event rate among those with isolated pAMR2 further underscores the pathogenic potential of sustained or severe vascular injury. Together, these results suggest that MVI—whether transient or persistent—may be an early and meaningful contributor to CAV, even in the absence of concurrent DSA or progression to more advanced AMR.

Figure 3.

Association between microvascular inflammation (MVI), cardiac allograft vasculopathy (CAV), and all-cause mortality. (A) Percentage of heart transplant (HT) recipients with varying rejection phenotypes—diagnosed on endomyocardial biopsy—who subsequently developed CAV. The no rejection group (n = 521) includes individuals who never exhibited cellular (ACR) or antibody-mediated rejection (AMR); pAMR2 only (n = 9) includes HT recipients with both histological and immunological (eg, C4d deposition) evidence of AMR; pAMR1 only (n = 154) includes HT recipients who had either pAMR1-h or pAMR1-i without evidence of any other rejection phenotype; ACR only (n = 123) includes recipients who experienced grade 2R or 3R ACR without any MVI or AMR phenotypes; “AMR1 transition into pAMR2 (n = 10) includes HT recipients who initially experienced pAMR1, which evolved into pAMR2 prior to the development of CAV; pAMR2 followed by pAMR1 (n = 15) includes HT recipients who temporarily experienced pAMR2, followed by pAMR1 (either pAMR1-h or pAMR1-i) before CAV development. (B) Forest plot of hazard ratios (HRs) for incident CAV (yellow) and all-cause mortality (blue) derived from Cox proportional hazard models. In the CAV analysis, rejection phenotype and circulating DSAs were treated as time-varying covariates, while donor and recipient age were treated as fixed covariates. In the mortality analysis, CAV was also treated as a time-varying covariate.

Over a median 4-year follow-up (IQR 2.36.4 years), 238 recipients developed CAV and 121 died. Among 154 individuals with pAMR1, 63 (40.9%) developed CAV compared with 23.6% of those without rejection (Figure 1A). In comparison with no rejection, DSA-negative MVI was independently associated with CAV (HR 1.47, 95% CI 1.00–2.16, P = 0.047) (Figure 1B). DSA-positive MVI was associated with mortality (HR 1.97, 95% CI 1.07–3.64) with DSA-negative MVI demonstrating directional concordance (HR 1.50, 95% CI 0.87–2.57), independent of CAV (HR 1.71, 95% CI 1.13–2.58).

We also conducted several prespecified subgroup and sensitivity analyses. First, we stratified the cohort by age to evaluate whether the associations observed in the overall cohort were consistent across adult (n = 693) and pediatric (n = 139) HT recipients. In adults, both the DSA-negative MVI (HR 1.39, 95% CI 0.94–2.06) and DSA-positive MVI (HR 1.11, 95% CI 0.65–1.89) phenotypes demonstrated directional concordance with the primary analyses regarding their association with incident CAV (Supplemental Figure 1A). Similar trends were observed for mortality. In pediatric HT recipients, the relationship between MVI and CAV was also directionally consistent with the overall cohort (Supplemental Figure 1B). However, precision was limited in this subgroup owing to the smaller sample size and fewer CAV events, resulting in wider CIs.

We further performed a 6-month landmark sensitivity analysis to account for early post-HT surveillance intensity (Supplemental Figure 2). This analysis demonstrated results that were directionally aligned with our primary time-varying models. These findings suggest that early histopathological evidence of microvascular injury may set the trajectory for later CAV development.

Discussion

In this large, granular, longitudinal cohort of HT recipients, we found that MVI—whether or not accompanied by circulating anti-HLA DSAs—was associated with adverse long-term outcomes, including CAV and all-cause mortality. These findings extend observations from renal transplantation to thoracic transplantation and suggest that subclinical alloimmune microvascular injury may represent a shared mechanism of chronic graft dysfunction across organ systems.¹ Importantly, DSA-negative MVI was not benign, but rather, it conferred a significant risk of CAV development, independent of histological ACR or circulating DSA.

These results are particularly informative in light of the recent consensus report on AMR,⁴ which explicitly outlines the need for improved understanding of DSA and MVI. Further, the consensus statement emphasizes that center-level practice variation remains substantial, particularly in the surveillance, reporting, and treatment of MVI and AMR, often driven by individual protocols and local interpretations. Despite the widespread use of ISHLT grading systems,⁴⁶ the document acknowledges that therapeutic decision-making in pAMR1 remains highly inconsistent, and that robust, outcomes-based data are lacking to guide standardized care in this group.

Our real-world findings help to fill this critical knowledge gap, with our results indicating that MVI—regardless of DSA status—identifies patients at increased risk of CAV. In our cohort, 154 individuals experienced pAMR1 (either histological [pAMR1-h] or immunological [pAMR1-i]) without progression to pAMR2, and 63 of these patients (40.9%) subsequently developed CAV. An additional 10 individuals progressed from pAMR1 to pAMR2, with 3 (30.0%) developing CAV. Among those with isolated pAMR2 (n = 9), 5 (55.6%) developed CAV, whereas 4 of 15 (26.7%) who experienced pAMR2 followed by later pAMR1 also developed CAV. These findings strongly reinforce the hypothesis that MVI—captured at the pAMR1 stage—may serve as a critical early immunopathological trigger for vascular injury, even when formal criteria for AMR are not met. By contrast, the incidence of CAV was 32.5% among those with ACR only and 23.6% among individuals who never demonstrated histological rejection, highlighting the relative risk gradient conferred by MVI.

The progressive nature of MVI and its potential to evolve into more severe AMR raises important clinical considerations. First, these findings support the concept that pAMR1 may not represent a static phenotype but rather a transitional state within the continuum of alloimmune injury. Second, the observation that ~41% of individuals with pAMR1 (in the absence of pAMR2) developed CAV underscores the need to re-evaluate whether isolated MVI—regardless of DSA or graft dysfunction—warrants intensified surveillance or preemptive therapeutic intervention. This is especially relevant in light of the most recent ISHLT consensus,⁴ which recommends treatment primarily in cases of pAMR2/3 while offering limited guidance on the management of pAMR1. Our findings directly challenge the prevailing assumption that pAMR1, particularly when DSA negative, is benign and instead suggest that earlier recognition and management of MVI may be critical to improving long-term outcomes after HT. As clinical practice continues to evolve, these data provide real-world support for incorporating MVI more formally into post-HT risk stratification algorithms.

These observations also reinforce the potential for MVI to serve as a modifiable target in post-HT care. As mechanistic studies emerge—particularly those leveraging single-cell and spatial transcriptomic profiling to delineate cell-type–specific signatures of microvascular injury^10,11— there is growing opportunity to discern transient, clinically inconsequential immune activity from sustained pathogenic microvascular injury. In parallel, the role of non-HLA antibodies (eg, anti-angiotensin II type 1 receptor antibodies),¹² natural killer cell–mediated cytotoxicity,¹³ and complement-independent endothelial injury are increasingly recognized as potential drivers of MVI, independent of conventional AMR. Dissecting these overlapping yet distinct immunological axes will be essential for developing rational, targeted interventions aimed at mitigating progressive vascular injury and improving long-term graft outcomes. Our study provides a foundation upon which these mechanistic layers can be superimposed.

Study Limitations

This study represents a large, real-world experience with HT, incorporating granular, longitudinal rejection phenotyping and angiographic CAV surveillance. Although our approach enabled a detailed characterization of temporal alloimmune microvascular injury and its relationship with long-term vascular outcomes in both adult and pediatric HT recipients, several limitations merit consideration. As with any observational cohort, these findings are subject to residual confounding and unmeasured bias despite our application of time-varying methods that account for dynamic changes in rejection phenotypes and exposure status. The use of time-varying covariates preserves temporal ordering, but does not fully account for the fact that these processes are measured intermittently and with error. A joint longitudinal–survival modeling approach could more rigorously account for these features, but was beyond the scope of the present work. As such, results related to time-varying covariates should be interpreted with caution, because modest bias related to measurement error and modeling specification is possible. The diagnosis of MVI and its classification are based on histopathological criteria and may be affected by interobserver variability.¹⁴ Because rejection phenotypes were adjudicated by a dedicated group of cardiac pathologists at our institution, interobserver variability in the diagnosis and grading of rejection phenotypes is attenuated. We further mitigated misclassification bias by using rigorous definitions—for both rejection and CAV grading—aligned with ISHLT consensus.^2,3,5,6 Although our pediatric subgroup represents one of the largest modern cohorts evaluated in this context, the relatively low event rate in this population limits statistical power and the precision of risk estimates. However, the directionality of observed associations was consistent across age strata, supporting the biological plausibility of our findings. Finally, as a single-center study, the generalizability of our findings may be influenced by institution-specific immunosuppression protocols, biopsy schedules, and DSA detection thresholds. Nevertheless, our consistent surveillance strategies, including angiographic CAV assessment by experienced interventional and transplant cardiologists and histopathological evaluation by a dedicated team of cardiac pathologists, strengthen internal validity and support the reproducibility of our analytic framework for future multicenter validation efforts.

Conclusions

Our observations suggest that MVI may be harmful in HT (with or without DSA), extending recent renal findings to thoracic transplantation.¹ Although observational confounding (eg, selection bias, biopsy frequency) are important limitations, our results reflect a large-scale real-world experience that underscores the prognostic import of MVI currently not widely treated.⁴ This strategy may be especially crucial in thoracic transplants, where progressive graft failure presaged by MVI is often catastrophic and irreversible. With novel technologies that allow access to patient-based heterogeneity in rejection at cellular resolution, understanding the mechanistic basis for this observation—and potential therapeutic routes (including beyond conventional HLA-directed therapies)—will be critical to extending graft survival across solid organ transplantation.

Lay summary

After a heart transplant, some patients develop a type of injury called microvascular inflammation, which may or may not involve their immune system producing antibodies against the transplanted heart. We studied >8000 heart biopsies from 832 adults and children after heart transplantation to understand whether microvascular inflammation (regardless of the presence or absence of these antibodies) is linked with worse long-term outcome, such as chronic rejection or death. Our findings show that microvascular inflammation, even in the absence of antibodies against the transplanted heart, could be harmful and highlight the need for ways to better detect and treat this type of injury to prevent long-term complications.

Supplementary Material

Supplementary materials

Supplementary material associated with this article can be found in the online version at doi:10.1016/j.cardfail.2025.08.021.

Biography

Kaushik Amancherla