Identification and Characterization of Trajectories of Cardiac Allograft Vasculopathy After Heart Transplantation

Abstract

Background: Cardiac allograft vasculopathy (CAV) is a major contributor of heart transplant recipient mortality. Little is known about the prototypes of CAV trajectories at the population level. We aimed to identify the different evolutionary profiles of CAV and to determine the respective contribution of immune and nonimmune factors in CAV development. Methods: Heart transplant recipients were from 4 academic centers (Pitié-Salpêtrière and Georges Pompidou Hospital, Paris, Katholieke Universiteit Leuven, and Cedars-Sinai, Los Angeles; 2004–2016). Patients underwent prospective, protocol-based monitoring consisting of repeated coronary angiographies together with systematic assessments of clinical, histological, and immunologic parameters. The main outcome was a prediction for CAV trajectory. We identified CAV trajectories by using unsupervised latent class mixed models. We then identified the independent predictive variables of the CAV trajectories and their association with mortality. Results: A total of 1301 patients were included (815 and 486 in the European and US cohorts, respectively). The median follow-up after transplantation was 6.6 (interquartile range, 4–9.1) years with 4710 coronary angiographies analyzed. We identified 4 distinct profiles of CAV trajectories over 10 years. The 4 trajectories were characterized by (1) patients without CAV at 1 year and nonprogression over time (56.3%), (2) patients without CAV at 1 year and late-onset slow CAV progression (7.6%), (3) patients with mild CAV at 1 year and mild progression over time (23.1%), and (4) patients with mild CAV at 1 year and accelerated progression (13.0%). This model showed good discrimination (0.92). Among candidate predictors assessed, 6 early independent predictors of these trajectories were identified: donor age ( P <0.001), donor male sex ( P <0.001), donor tobacco consumption ( P =0.001), recipient dyslipidemia ( P =0.009), class II anti–human leukocyte antigen donor-specific antibodies ( P =0.004), and acute cellular rejection ≥2R ( P =0.028). The 4 CAV trajectories manifested consistently in the US independent cohort with similar discrimination (0.97) and in different clinical scenarios, and showed gradients for overall-cause mortality ( P <0.001). Conclusions: In a large multicenter and highly phenotyped prospective cohort of heart transplant recipients, we identified 4 CAV trajectories and their respective independent predictive variables. Our results provide the basis for a trajectory-based assessment of patients undergoing heart transplantation for early risk stratification, patient monitoring, and clinical trials. Registration: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT04117152.

Donor‐derived cell‐free DNA is associated with cardiac allograft vasculopathy

“…Immune activation events in the early post‐transplant course are known risk factors for CAV development. Recently, ACR ≥2R and class II DSA during the first post‐transplant year have been identified as independent predictors of CAV trajectories 42 . Consequently, serial early assessments of ddcf‐DNA possibly reflecting these immune activation events could be useful to further inform about an individual’s risk of CAV later in the post‐transplant course pending future longitudinal studies.…”

Low-Density Lipoprotein Cholesterol Level Trends and the Development of Cardiac Allograft Vasculopathy After Heart Transplantation

“…20 Their findings also suggest that proliferation signal inhibitors such as sirolimus have a greater impact in reducing the risk of CAV irrespective of LDL-C levels. In contrast, a large multicentre observational analysis by Loupy et al recently suggested an association between a 1-year LDL-C cut off ≥1.0 g/L (2.6 mmol/L) and risk of CAV progression 21 . We evaluated LDL-C at 1-year post HT as a continuous variable to better characterize the association with angiographic CAV as well as earlier signs of CAV using IVUS data.…”

Profiles of patients’ self-reported health after acute stroke

“…Binary univariate regression was performed in advance, and parameters associated with class membership with p < 0.05 were included in multivariable multinominal logistic regression models. The NIHSS was used in categories of no (0), mild (1-4), moderate (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15), moderate to severe (16)(17)(18)(19)(20)(21) and severe (22-42) symptoms.…”

“…To this end, a systematic assessment of patient reported outcomes measures (PROMs) has been shown to deliver information beyond routine examinations by physicians and to identify disabling deficits in every-day life after stroke [17,18]. A possible strategy towards understanding the development of different outcomes and towards risk stratification is classifying patients by the results of their outcomes and analyzing class characteristics, as shown for diabetes [24] and heart transplantation [20]. In mild to moderately affected stroke patients from an outpatient setting, a retrospective analysis showed promising results building distinct profiles or classes using the Patient-reported Outcomes Measurement Information System (PROMIS) [16].…”