Redefining Primary Graft Dysfunction: Toward a Consensus in the New Era of Heart Transplantation

Nowadays, primary graft dysfunction (PGD) remains a challenging acute complication in heart transplantation.1 In 2014, the International Society for Heart and Lung Transplantation Consensus Working Group established a standardized definition and classification for PGD, defining it as occurring within the first 24 h posttransplant and characterized by allograft dysfunction not attributable to hyperacute rejection, pulmonary hypertension, surgical bleeding, or other identifiable causes.2 Severe PGD requires mechanical circulatory support (MCS) such as extracorporeal membrane oxygenation (ECMO), whereas mild-to-moderate PGD is identified using echocardiographic criteria and necessitates high-dose inotropes or intra-aortic balloon pump placement during hemodynamic compromise. This diagnostic standard has been widely adopted, promoting a uniform approach to PGD globally and enhancing both research and clinical management, leading to more consistent reporting of its incidence and severity. However, variability remains, with reported incidence rates approximately ranging from 2.5% to 30%,3 highlighting the ongoing challenges in achieving uniformity across different centers and clinical practices. Additionally, the past decade has seen technique advancements in transplantation and the reform in the US heart allocation system, leading to the increased use of donation after circulatory death (DCD), extended criteria donors, more frequent MCS application among recipients, and other developments, all of which could significantly influence PGD (Figure 1). For example, the United Network for Organ Sharing data show that the use of DCD hearts in the United States surged from 0 cases in 2018 to 612 cases in 2023. These advancements prompt specialists to reconsider whether the current PGD definitions remain adequate in light of these clinical changes.FIGURE 1.: This graphical abstract summarizes challenges to the 2014 ISHLT definition and classification of PGD resulting from recent developments in heart transplantation and presents proposed pathways for redefining PGD in the new era. AI, artificial intelligence; DCD, donation after circulatory death; ECMO, extracorporeal membrane oxygenation; ISHLT, International Society for Heart and Lung Transplantation; MCS, mechanical circulatory support; PGD, primary graft dysfunction.First, the current PGD severity scale relies heavily on treatment modalities, leading to inconsistencies in diagnoses and a rise in severe PGD cases. At present, there are no specific guidelines governing the selection or timing of MCS devices, often leaving the decision to physician discretion, which can result in inconsistent severity diagnoses for similar pathophysiological conditions or substantial heterogeneity within the same classified PGD cases. For instance, a recipient on ECMO is classified as severe PGD-LV, whereas another with a left ventricular ejection fraction of 20% and an inotrope score above 10, but hemodynamically stable without ECMO, is deemed moderate PGD-LV, despite potentially similar pathophysiological severities. Moreover, the incidence of PGD has increased over the years,4 likely due to current practices, such as the more frequent use of MCS in recipients preoperatively, the use of DCD hearts, older donors, and hearts with longer ischemic time—all of which are key risk factors for PGD.1,4 Despite new technologies such as temperature-controlled transport devices or hypothermic oxygenated machine perfusion reducing severe PGD rates,5-8 they have not yet offset the rising trend. A possibly more critical reason is the increasingly aggressive use of MCS, inflating the PGD diagnosis. For example, a decade ago, posttransplant ECMO use was risky due to complications. However, advances in ECMO technology and management have significantly reduced major complications, making aggressive ECMO strategies considerably safer.9 With a broader safety margin, physicians are more inclined to bypass initial intra-aortic balloon pump or high-dose inotropes and escalate directly to ECMO. Aggressive ECMO use could notably decrease mortality rates of severe PGD. Columbia University found that in-hospital mortality was 28% in the conservative ECMO group versus 5% in the early implantation group.10 However, aggressive ECMO use might increase the diagnosis rate of severe PGD and lead to overdiagnosis. A relatively conservative ECMO strategy could result in a lower diagnosis rate of severe PGD but a higher mortality rate, potentially reaching 30%,11 as these cases might represent true severe PGD pathophysiologically. Another scenario is that some cases of isolated PGD-RV also opt for ECMO9,10 and might be incorrectly classified as severe PGD-LV. Additionally, the current International Society for Heart and Lung Transplantation PGD severity scale shows a scarcity of reported mild PGD cases,12 raising questions about whether mild PGD is genuinely rare or the diagnostic criteria are insufficiently sensitive. Even if a diagnosis of mild PGD is made due to hemodynamic instability, it may not necessarily be PGD, as hemodynamic instability post–heart transplantation is common but not always indicative of PGD.13 All these factors complicate objective PGD assessments, affecting the reliability of multicenter studies and potentially biasing clinical research outcomes. Second, the current definition and classification of PGD lack reflection of its diverse etiology and pathophysiology, hindering objective and prospective treatment guidance. PGD’s etiology includes inflammatory responses, ischemia/reperfusion injury, and activation of the Kallikrein-Kinin system, among other factors.14 Studies show that the risk factors for mild or moderate PGD differ from those for severe PGD, suggesting different underlying causes. For instance, amiodarone use, beta-adrenergic receptor antagonists, and extended ischemic times increase the risk for all PGD types. Donor death causes and recipient diabetes are linked to mild-to-moderate PGD, whereas intraoperative blood product use is associated with severe PGD.1 These findings emphasize the causal link between cause and severity of PGD. Clarifying the cause may improve PGD treatment efficacy; for example, recent findings suggest that locus-specific calculated panel-reactive antibodies for HLA-A significantly predict PGD severity, indicating that interventions such as plasma exchange or IVIG could mitigate PGD risk.15 Furthermore, with the increasing use of DCD hearts, it is vital to assess whether current PGD grading criteria apply to DCD hearts, as DCD and donation after brain death (DBD) hearts have different pathophysiological processes and preservation methods. DCD donors experience warm ischemia, hypoxemia, and increased right ventricular afterload, often using technologies such as normothermic machine perfusion or normothermic regional perfusion, whereas DBD donors undergo a catecholamine surge due to brainstem death and typically use cold static preservation, leading to different etiological causes of PGD. Research shows DCD recipients have more severe biventricular PGD but shorter posttransplant MCS durations and hospital stays than DBD recipients, indicating differences in donor characteristics and recovery mechanisms.16 A quicker resolution of dysfunction in DCD hearts suggests the need for a distinct grading scale compared with the prolonged dysfunction seen in DBD hearts. Therefore, a precise, standardized PGD definition and grading system based on etiology is crucial to facilitate more targeted and effective treatments, especially with the growing use of DCD hearts. The new era of heart transplantation necessitates redefining PGD. Based on the issues we identified, we propose focusing on the following 3 areas: First, maximizing the standardization of initiation thresholds for ECMO is crucial to minimize variability in clinical practice. Although complete standardization may be challenging in practical settings, reaching a consensus on some key thresholds for initiating ECMO is feasible. In addition, we suggest raising the diagnostic threshold for PGD by eliminating the mild PGD grade. The new definition may consider the minimum criteria for PGD to begin from the existing moderate grade and further subdivide the severe PGD grade by incorporating specific ECMO metrics or other MCS characteristics, such as duration of support and recovery trajectory, to more accurately reflect PGD severity. It is believed that early successful weaning from MCS is likely indicative of rapid cardiac recovery and a less severe initial PGD, suggesting that specific MCS metrics could be valuable indicators of PGD severity, although further validation is required. Second, it is imperative to advance both basic and clinical research to elucidate the cause and pathophysiology of PGD at molecular, cellular, and organ levels. Special attention should be given to the different characteristics of PGD in DCD versus DBD. Identifying a combination of highly sensitive biomarkers is a promising direction. Research should focus on biomarkers in the donor heart and the recipient, using high-throughput omics technologies to identify biomarkers for different causes and stages of PGD. These biomarkers could include combinations indicating myocardial injury, cardiac metabolism, and immunoinflammatory responses. This approach will help differentiate causes, reduce the current overdependence on MCS usage, and foster a better understanding of graft conditions and tailored treatment. For instance, the recently identified immune-related molecule CLEC4C and several myocardial metabolites are valuable for grading PGD severity or distinguishing between PGD characteristics in DBD and DCD hearts,17,18 although they still require further validation in larger patient cohorts. Third, artificial intelligence (AI) offers significant advantages in analyzing, predicting, and diagnosing PGD, which results from complex interactions among multiple factors. AI can process high-dimensional and intricate data, capturing complex patterns and relationships that traditional statistical methods often miss. Additionally, AI can dynamically learn and update with new data, continuously improving performance and accuracy. For example, an artificial neural network model can predict PGD risk by incorporating 33 significant predictors from 77 variables, emphasizing factors such as heart failure diagnosis, ischemia time, and sex.19 This demonstrates the potential of AI to enhance PGD management. However, current AI models face limitations due to missing data, data imbalances, and uneven variable distribution in data sets.20 An urgent need for AI development is the establishment of standardized multicenter databases that include comprehensive clinical and biosample data. Despite these challenges, advancing AI in this field is a crucial and promising direction for the future. In conclusion, redefining PGD is crucial to align with recent advancements in heart transplantation, with efforts aimed at achieving greater diagnostic consistency and accuracy while capturing the biological complexities involved, thereby further improving heart transplantation outcomes.