Author + information
- Ihdina Sukma Dewi, MD, PhD,
- Olof Gidlöf, PhD,
- Zsuzsanna Hollander, PhD,
- Karen K. Lam, PhD,
- Mark D. Benson, MD, PhD,
- Oscar O. Braun, MD, PhD,
- Johan Nilsson, MD, PhD,
- Scott J. Tebbutt, PhD,
- Raymond T. Ng, PhD,
- Jenny Öhman, MD, PhD,
- Bruce M. McManus, MD, PhD and
- J. Gustav Smith, MD, PhD∗ ()
- ↵∗Department of Cardiology, Clinical Sciences, Faculty of Medicine, Lund University, Skåne University Hospital, SE-221 85, Lund, Sweden
Acute cellular rejection (ACR), a T-cell–mediated form of organ rejection, remains a common problem after transplantation that compromises long-term allograft survival. Surveillance for ACR has been based on endomyocardial biopsy since the 1970s, which is burdensome due to its invasiveness, risk of complications, significant cost, interobserver variability, and risk of false negative findings. Circulating biomarkers reflecting inflammation or myocardial injury, such as C-reactive protein or troponins, are not recommended for rejection surveillance due to poor specificity (1). Recent developments in affinity-based technologies have made systematic discovery of more informative biomarkers possible by allowing robust, simultaneous quantification of large sets of low-abundance proteins in blood (2). We aimed to comprehensively explore immunological protein profiles in heart transplant recipients with ACR.
We applied a proximity extension assay (2) to measure 92 immune-related proteins, selected to provide broad coverage of immune pathways (Inflammation panel, Olink Proteomics, Uppsala, Sweden), in 2 heart transplant cohorts. Protein abundance is quantified from real-time polymerase chain reaction quantification cycle values, and normalized protein expression units, expressed on the log2 scale. The discovery cohort consisted of recipients of heart transplants between February 2009 and September 2013 at 6 Canadian centers, including 22 biopsy-confirmed ACR cases (91% grade 2R, 9% 1R, median 23 days after transplantation [interquartile range: 13 to 98 days]) and 38 controls without evidence of rejection in routine surveillance biopsies, all graded by 3 blinded expert pathologists (3). In this cohort, 10 of the 92 inflammatory proteins were significantly higher in patients with ACR (Table 1). Lasso regression analysis retained 5 proteins as independently informative, which were strongly correlated with the excluded markers (Pearson pairwise r = 0.50 to 0.86), particularly the chemokines (r = 0.86 for CXCL10 and CXCL9). These 5 proteins were combined into a multimarker protein score based on normalized protein expression sums weighted by beta-coefficients, which allowed significant separation between ACR cases and controls (p < 0.0001) and discrimination of ACR (area under the curve [AUC]: 0.82). A maximally informative score cutoff point was 0.359 (maximum Youden J statistic), with 82% sensitivity and 76% specificity, and 100% sensitivity was obtained with a cutoff of 0.102.
An increase during ACR was confirmed for 4 proteins using linear mixed models (Table 1) in a validation cohort consisting of serial serum samples (before, during, and after ACR) from 10 heart transplant recipients (Skåne University Hospital in Lund, Sweden, 1997 to 2010) with biopsy-proven ACR (78% grade 3A, 11% 3B, and 11% 2, median 356 days after transplantation [interquartile range: 225 to 731]) (4). The multimarker score provided similar diagnostic accuracy as in the discovery cohort (AUC: 0.80). The score cutoffs from the discovery cohort for maximum information content and 100% sensitivity provided a sensitivity of 67% and 100%, respectively, and a specificity of 67% and 27%.
The pathway most enriched among the 5 biomarkers was related to lymphocyte signaling (canonical pathway “Altered T Cell and B Cell Signaling in Rheumatoid Arthritis” from Ingenuity Pathway Analysis Software build 430520M, Qiagen, Hilden, Germany). Four of the proteins (SLAMF1, LTA, IL-12B, and CSF-1) were included in this pathway as compared with 17 of the 92 proteins in the full assay (p for enrichment = 0.004). Ex vivo activation of human CD4+ T cells (n = 6) with CD3/CD28 beads (Dynabeads Human T-Activator CD3/CD28, Life Technologies, Carlsbad, California) resulted in increased expression of mRNA transcripts for all 5 biomarkers (p < 0.01).
In summary, circulating proteins associated with T-cell activation were increased in heart transplant recipients with ACR. A multimarker model including 5 such biomarkers provided strong and reproducible discrimination of ACR. The only noninvasive modality currently recommended for ACR surveillance is gene expression profiling of 20 mRNA transcripts in peripheral blood (AlloMap, CareDx, Brisbane, California) (1), used at many U.S. centers at a cost similar to endomyocardial biopsy. We note that our score achieves substantially higher AUC than reported for AlloMap in the CARGO II (Cardiac Allograft Rejection Gene Expression Observational II) trial (0.80 vs. 0.69) and substantially higher specificity at the 100% sensitivity cutoff (27% vs. 2%), but recognize that no direct comparison has been performed (5). The encouraging results from this study motivate a large, prospective study to determine whether plasma biomarkers reflecting T-cell activation can reduce the burden of endomyocardial biopsy in heart transplant recipients.
Please note: The authors thank Siv Svensson and Rebecca Rylance (Department of Cardiology, Clinical Sciences, Lund University, Lund, Sweden), the Prevention of Organ Failure (PROOF) Centre of Excellence team (Vancouver, British Columbia, Canada) including Janet Wilson-McManus, Paul Keown, and Robert McMaster, the Science for Life Laboratory and Olink Proteomics teams (Uppsala, Sweden) for technical assistance, and all heart transplant recipients who contributed to the study through the Biomarkers in Transplantation (BiT) consortium and at Skåne University Hospital in Lund. This work was supported by the European Research Council, the Swedish Heart-Lung Foundation, the Wallenberg Center for Molecular Medicine in Lund, the Swedish Research Council, the Crafoord Foundation, governmental funding of clinical research within the Swedish National Health Service, Skåne University Hospital in Lund, and the Scania county (to Dr. Smith). The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2017 American College of Cardiology Foundation
- Smith J.G.,
- Gerszten R.E.
- Sukma Dewi I.,
- Torngren K.,
- Gidlof O.,
- et al.