Journal Logo

Original Basic Science—General

Label-free Identification of Antibody-mediated Rejection in Cardiac Allograft Biopsies Using Infrared Spectroscopic Imaging

Uraizee, Imran MD1; Varma, Vishal K. PhD2; Sreedhar, Hari BS3; Gambacorta, Francesca BS3; Nazeer, Shaiju S. PhD3; Husain, Aliya MD1; Walsh, Michael J. PhD3,2

Author Information
doi: 10.1097/TP.0000000000002465


Antibody-mediated rejection (AMR) remains a major obstacle to the long-term success of cardiac transplantation. Antibody-mediated rejection is associated with chronic allograft vasculopathy (CAV)1-3 and poorer outcomes, including graft dysfunction and decreased survival.4-6 Pretransplant history of panel-reactive antibodies (PRA); presence of donor-specific antibodies (DSA), namely anti-HLA antibodies; cytomegalovirus (CMV) positivity; prior treatment with and antibodies against muromonab-CD3; prior mechanical circulatory support; and multiparity are among the key known risk factors.3,7-9 AMR results from binding of specific alloantibodies to donor antigens on the vascular endothelium of the allograft, leading to activation of the complement cascade and an inflammatory response characterized by endothelial cell activation, cytokine release, and recruitment of macrophages to the microvasculature.10,11

The 2013 revision of the International Society of Heart and Lung Transplantation (ISHLT) working formulation for the standardization of nomenclature for the pathologic diagnosis of AMR reflects the continuing evolution in our understanding of the underlying pathologic processes responsible for AMR and the resulting histologic and immunologic manifestations. Histopathologic criteria include intravascular accumulation of activated mononuclear cells in graft capillaries and endothelial swelling. Immunopathologic criteria include multifocal/diffuse endothelial staining of complement 4d (C4d) by immunohistochemistry (IHC) on formalin-fixed paraffin-embedded (FFPE) right ventricular (RV) endomyocardial biopsies (Figure 1) or a similar pattern of C4d, complement 3d (C3d), or anti-HLA-DR by frozen section immunofluorescence (IF).12 Although these criteria are intended to improve diagnostic consistency and aid data collection for future guidelines, early reports suggest there may be a decrease in the number of active AMR diagnoses with increases in the rate of silent/subclinical AMR diagnoses.13

A schematic of the process of obtaining an endocardial biopsy through to taking the sample for regular histology for C4D staining and examination by the pathologist (left side) and for IR imaging for automated diagnoses (right side). IR, infrared.

Current histopathologic and immunologic diagnostic criteria, including detection of capillary endothelial swelling, intravascular macrophages, and interstitial edema, demonstrate poor sensitivity for AMR or may indicate a more advanced stage of AMR,14,15 are infrequently observed in the absence of corresponding immunopathologic findings, and may vary between institutions per quality of biopsies, fixation/processing techniques, and staining methods.12 Diffuse C4d capillary endothelial deposition as detected by IHC and IF correlates with circulating DSA10,16 and thus reflects active AMR even in the absence of clinical symptoms suggestive of graft dysfunction.1 However, asymptomatic AMR is still associated with CAV17 and circulating DSA may be present several months to years before clinical evidence of antibody-mediated graft failure.16,18 Thus, C4d immunostaining may also fall short in the goal for earlier detection or even prevention of AMR and subsequent CAV.

Infrared (IR) spectroscopy has emerged as a potentially valuable tool for nondestructive label-free biochemical imaging of tissue specimens to aid in histologic diagnosis (Figure 1).19,20 Infrared spectral images are obtained by the quantitative variations in absorbance of mid-IR wavelengths resulting from differential vibrational frequencies among various associated functional groups of different biomolecules, such as proteins, lipids, DNA, RNA, and carbohydrates within regions of interest (ROIs) of a tissue sample, allowing for the determination of a characteristic biochemical fingerprint.19-21 Rapidly improving image resolutions now permit visualization of histologic structures down to 1 μm, and each pixel can offer an entire IR spectrum comprised of many biochemical channels compared with conventional visible light microscopy alone. Acquired spectral data provides inputs for multivariate data analysis and advanced computational classifiers to predict the presence or stage of a disease process.19 There are a number of other label-free microspectroscopic methods, such as different types of Raman spectroscopy; however IR imaging is a good compromise between allowing for rapid imaging of large tissue sections and also acquiring a wide spectral range. There are a number of recent reviews that discuss the applications of IR spectroscopy to biomedical problems.21-24

Recent advances in quantum cascade lasers (QCLs) offer advantages over traditional Fourier Transform-based IR spectroscopic imaging systems based on their high brightness and their ability to tune to specific spectral ROI.25-27 The ability to select a restricted number of spectral frequencies as opposed to all frequencies as in FT-IR may allow for real-time diagnoses to be made. Recently, discrete-frequency IR imaging has been demonstrated to allow for rapid classification of cell types, notably with QCL-IR applied to heart transplant biopsies28 and scanning near-field optical microscopy coupled with an IR free electron laser applied to cervical cytology.29 Although these discrete-frequency techniques may facilitate rapid information for clinical decision making in certain time-sensitive contexts (eg, intraoperative biopsy), QCL-IR is also beneficial in less time-sensitive operations such as when FFPE sections are routinely collected for postcardiac transplant monitoring due to potentially increasing the laboratory workflow speed and thus reducing costs.

IR spectroscopy may potentially be a valuable tool for identifying or even predicting AMR in cardiac allograft biopsies before the immunologic or histologic features are distinguishable. We hypothesize that there is a unique biochemical signature in heart tissue that is unique to AMR positive biopsies and that may potentially appear even before immunohistological presentation. IR imaging has demonstrated that a unique spectral signature can be extracted from tissues to predict recurrence of diabetic nephropathy30 and rapid progression of fibrosis31 in renal transplant patients with no histological signs present. However, no previous studies attempt to determine whether IR spectroscopy can detect cardiac AMR. Therefore, we examined endomyocardial biopsies with demonstrated diffuse C4d capillary endothelial staining by immunohistochemistry to identify a unique biochemical signature for AMR using IR spectroscopy.


Study Sample

We reviewed RV endomyocardial biopsy reports from patients who underwent heart transplant rejection surveillance at the University of Chicago between July 2003 and January 2015 (IRB protocol number: IRB15-0030-CR002). Results were searched chronologically beginning with the most recent biopsies for strong, diffuse endothelial C4d-positive staining by IHC suggestive of active immunopathologic AMR (pAMR1 I+)12 but without evidence of concomitant acute cellular rejection (ACR) or Quilty lesions. Biopsies negative for C4d and ACR were identified in a similar manner. In total, 14 biopsies with immunopathologic AMR and 16 biopsies negative for any rejection were selected for this study. In addition, a collection of 14 native heart biopsies without major diagnostic abnormality apart from mild myocyte hypertrophy was included as a nontransplant control group. We recorded the age, sex, indication for transplantation, results of pretransplant PRA, donor CMV status, and immunosuppressive regimen at time of biopsy for each transplant recipient. Cardiac filling pressures and hemodynamics obtained during right heart catheterization for the endomyocardial biopsy were also noted if available (Table S2, SDC,

Tissue Preparation and Analysis

C4d IHC was performed using affinity-purified antihuman C4d rabbit polyclonal antibody (Catalog No. 12-5000; American Research Products, Inc., Waltham, MA). FFPE RV biopsies were initially retrieved for 40 minutes in a water bath at 95°C using target retrieval solution (Cat. No. S1699; DakoCytomation, Carpinteria, CA). Immunostaining was performed with an autostainer (BenchMark-XT; Ventana Medical Systems, Inc., Tucson, AZ). Sections were incubated for 2 hours at 37°C with a 1:25 dilution of anti-C4d antibody, followed by visualization with a horseradish peroxidase (HRP) diaminobenzidine (DAB) detection kit (Cat No. 760-500, Ultra-View HRP DAB Kit; Ventana Medical Systems, Inc.). Sections were counterstained with Harris hematoxylin, dehydrated in alcohol, and mounted in medium (Tissue-Tek Glass Mounting Medium; Sakura Tissue-Tek USA, Inc., Torrance, CA). The C4d stain was classified as positive if >50% of capillary endothelia were involved (Figure 1). Additionally, serial sections from the FFPE tissue block stained with hematoxylin and eosin (H&E) were used to identify and select ROIs for subsequent spectroscopic analysis.

In preparation for IR spectroscopy, FFPE RV biopsies were cut into 4-μm unstained sections and floated briefly onto a heated water bath (40–44°C) before transfer to MirrIR low-emissivity reflective slides (Kevley Technologies, Chesterland, OH). Tissue was deparaffinized in hexane for 48 hours before imaging following established protocols. H&E-stained slides were scanned with Aperio ScanScope CS (Leica Biosystems, Nussloch, Germany) and digitally analyzed using Aperio ImageScope v11.2 for histological examination and correlation with subsequent IR imaging on the reflective slides. The sections were scanned using a QCL imaging system (Spero, Daylight Solutions) in transmission mode using a 12.5× x12.5 focusing objective (NA = 0.70) with a pixel size of 1.4 × 1.4 μm over the range of 1800 to 900 cm−1.

Regions in the IR image that corresponded to cardiomyocytes were identified and endothelial cells were avoided, based on comparison to the H&E-stained serial sections. These regions consisted of many pixels that were averaged to extract a single IR spectrum from each biopsy. Regions of endothelium were avoided as the current resolution did not permit accurate extraction of spectra from these regions and could potentially confound the signature. Preprocessing was performed by baselining, removing low quality pixels, and removing outliers using MATLAB (median absolute deviation outlier removal). Sequential forward feature selection was performed using 66% of the data, and these features were used to create classifiers with 5-fold cross-validation as previously described.32 The machine learning models trained were naive Bayes, linear discriminant analysis (LDA), random forest (RF), and support vector machines (SVM). PCA-LDA was also performed on both the whole spectral range and using just the 13 selected spectral features to visualize separation.


Our study sample included a total of 44 RV endomyocardial biopsies from 39 patients with a mean age of 51.0 ± 16.2 years. Seventy-three percent of the patients were male. Thirty biopsies were from posttransplant hearts and 14 from native hearts. Of the posttransplant biopsies, 14 were diffusely positive for C4d and 16 were diffusely negative for C4d. There was no statistically significant difference in age or sex (Table S2, SDC, Nonischemic and ischemic cardiomyopathies were the most common indications for transplantation, whereas most native heart biopsies yielded a diagnosis of mild-moderate hypertrophy and 2 revealed no diagnostic abnormality. Patients with C4d-positive biopsies tended to have a higher percentage of PRAs than C4d-negative patients, but there was no difference in CMV status, pretransplant LVAD use, graft ischemic time, or time elapsed between transplantation and biopsy (Table S2, SDC,

Available invasive hemodynamic data demonstrated elevated cardiac filling pressures in patients with native heart biopsy and C4d-positive posttransplant patients compared with those with C4d-negative biopsies, respectively, namely right atrial (P = 0.010, P = 0.018) and pulmonary capillary wedge pressures (P = 0.004, P = 0.004; Table S2, SDC, RV end-diastolic pressure was elevated in the native heart biopsy group compared with the C4d-negative group (P = 0.022), but the difference between the C4d-positive group and the native biopsy and C4d-negative groups did not reach statistical significance, respectively (P = 0.43, P = 0.08). Cardiac index was markedly reduced in the native heart biopsy group (P < 0.001) compared with both posttransplant groups, which in turn did not differ from one another (P = 0.94).

The averaged spectral signatures based on ROIs extracted from the myocytes from the C4d-positive, C4d-negative, and native heart groups reveal distinct IR absorbance patterns in the 1800-900 cm-1 range commonly called the fingerprint region (Figure 2). Although the spectral changes appear modest, these small spectral shifts have been demonstrated to be highly significant and robust for diagnosis in multiple diseases.19,21 Two control groups (heart transplant biopsies without evidence of AMR and native heart biopsies) were used in this study to ensure detection of an AMR signature and not of a transplant signature. The myocyte spectral data (Figure 2) shows shifts and absorption differences in Amide I and Amide II regions. Using a subset of the data, feature selection (Table S1, SDC, was performed to isolate the best features for separation of classes. The algorithm isolated bands associated with mostly proteins in the 1800-1400 cm-1 region (Table S1, SDC, Given the nature of the technique, it is not possible to isolate the exact source of these biochemical signatures as they are holistic signatures of the entire cellular biochemistry. Even with the lack of chemical specificity, these shifts and changes in IR absorption can be useful in identifying abnormal or disease-related cell response. Identifying a small subset of spectral features can allow for new QCL-based IR imaging systems to potentially make near instantaneous diagnoses.

Average IR spectra derived from the myocardial cells of each sample with evidence of C4D deposition (green) and each sample without evidence of C4D deposition, namely those with native hearts (red) and transplanted hearts (black). AMR, antibody-mediated rejection; IR, infrared.

The multivariate data analysis technique, PCA-LDA, was performed on the whole data and the 13 selected spectral features to identify whether the 3 groups separated based on their biochemistry (Figure 3). Both analyses showed good discrimination between the C4d-positive (AMR group) and the 2 control groups, C4d-negative and native biopsies. Similar discrimination patterns suggest the feature-selected wavenumbers retain most of the discriminatory information. Minimizing the number of spectral frequencies, although not necessarily offering any kind of critical speed advantage in this context, may facilitate the clinical incorporation of this technology by simplifying the measurements that feed into pathologists’ diagnostic decision-making.

Principal Component Analysis-Linear Discriminant Analysis of the IR spectroscopic data extracted from the heart biopsy samples of patients with no evidence of C4D deposition in native hearts (red) and transplanted hearts (black) and patients with evidence of C4D deposition (green). The analysis was performed on either the entire (A) 1800-900-cm region of the IR spectrum or (B) with the 13 wavenumbers that were feature selected. IR, infrared.

Given that the ultimate goal is to be able to differentiate between the control and AMR groups, we built advanced computational classifiers using the 13 features selected (Table S2, SDC, and performed 5-fold cross-validation. Average performance for 5-fold cross-validated area under the Receiver Operator Characteristic curve for LDA, SVM, naïve Bayes, and RF were 0.89, 0.84, 0.81, and 0.88, respectively. The machine learning models performed well using the selected 13 features. The performance of each of the classifiers for each fold cross-validation is shown in Figure 4.

Areas under Receiver Operator Characteristic (ROC) Curve for each of the 5-fold cross-validations for all models trained: LDA, SVM, naive Bayes, and RF. A, Average areas under the ROC curve are also shown for 5-fold. AUC, area under the curve. LDA, linear discriminant analysis; RF, random forest; SVM, support vector machine.


In this study, using a label-free, real-time chemical imaging approach with IR spectroscopy, we identify a biochemical signature for immunopathologic AMR in cardiac allografts as well as nonrejecting allografts and native hearts using single unstained tissue sections from FFPE RV endomyocardial biopsies. Cardiac allograft patients with C4d-positive biopsies in our cohort tended to have a higher percentage of pretransplant PRAs, demonstrated elevated filling pressures but normal cardiac indices compared with those with C4d-negative biopsies, and showed no histologic evidence of AMR, suggesting cross-sectional capture of subclinical AMR. Patients in the cohort with native heart biopsies showed significantly elevated cardiac filling pressures and decreased cardiac indices suggestive of heart failure.

IR spectroscopic imaging can sensitively detect subtle biochemical perturbations before the appearance of histologic changes and provide insights into underlying processes that may be occurring within tissue. In this study, biopsies from posttransplant patients showed neither histologic evidence of AMR consistent with pAMR1 (H+), particularly endothelial swelling or presence of activated mononuclear cells within the microvasculature, nor ACR. Although this modality cannot confirm the specific identities of biomolecules present, such as cytokines or cytoplasmic proteins, it does shed light on active processes and the biochemical milieu that can be correlated with known pathophysiology or serve as a foundation for future mechanistic studies. Using PCA-LDA multivariate analysis, it was possible to separate the C4d-positive group from the C4d-negative and native biopsy groups. Additionally, when the spectral data was subjected to 4 different classifiers, good classification could be achieved between the C4d-positive group and the C4d-negative groups (both transplant and native groups).

Differences in biochemical signatures between the 3 biopsy groups in this study may be attributable to a number of different factors. Differences in the spectral signatures of native heart and C4d-negative biopsies may reflect biochemical changes present in heart failure compared with otherwise properly functioning hearts per the significantly elevated intracardiac pressures and decreased cardiac index in the former group. Activity of compensatory mechanisms, such as the downstream effects of increased B-type natriuretic peptide (BNP) levels via a cGMP-dependent signaling pathway,33 may account for the increased spectral absorbances seen in the peptide region of native heart biopsies.

The capillary endothelium represents the initial point at which circulating donor-specific recipient antibodies make contact with donor antigens, activate the complement cascade, and cause upregulation of an inflammatory response.8 Histologic findings of interstitial hemorrhage, mixed inflammatory infiltrates, and edema are more indicative of late-stage, severe AMR that can affect myocardium.8,12 Interestingly, the biochemical signature obtained in myocardial ROIs of this cohort of C4d-positive biopsies reflects a distinct difference compared with that of C4d-negative biopsies despite the absence of histologic findings of AMR and ACR. Thus, it is possible that earlier stages of AMR may have more of an effect on graft myocardium than previously thought, such as through interstitial leakage of inflammatory mediators.

As AMR has been reported to be concomitant with ACR in up to a quarter of cases34 and to exhibit an increasingly broad spectrum of findings, the identification of biochemical differences in the myocardium may suggest more direct effects on the myocardium and surrounding interstitial space in active AMR. Future studies will seek to identify the biochemical effects of ACR on spectral signatures and accordingly characterize mixed rejection biopsies as well as search for the presence of an AMR signature in negative biopsies before the first biopsy positive for immunopathologic AMR. Identification of these differences is also important to automating the process of tissue segmentation into myocardium and capillary endothelium across disease states.

Our analysis focused on immunopathologic AMR alone (pAMR1 I+) as defined by the most recent working formulation offered by the ISHLT and did not include cases with histologic findings associated with AMR, such as capillary endothelial swelling and intravascular macrophages, which could theoretically result in missed AMR cases. However, the ISHLT working group did note that the histologic diagnosis of AMR in the absence of immunopathologic findings is rather infrequent with low sensitivities.12 This institution has relied on IHC C4d staining of FFPE RV biopsies for evaluation of AMR for the last decade and only recently began routine use of IF C4d staining of frozen tissue, the latter of which has yielded a much higher rate of diffuse endothelial staining than that seen with IHC. As only biopsies positive by IHC were included in this study, it is possible that some cases of active AMR were excluded. Our analysis did not include cases of mixed rejection to avoid confounding bias of the AMR signature, but future studies will examine coincident AMR and ACR. Finally, our study focused on spectral data acquisition in myocardial ROIs only. With improved spatial resolution, we hope to include endothelial ROIs in future investigations, which can be simplified by automated tissue segmentation on the basis of spectral and morphologic features.


In a sample of cardiac allograft RV biopsies with immunopathologic AMR, we observed a unique biochemical fingerprint compared with that of nonrejecting and native hearts using IR spectroscopic imaging without using a single stain. Thus, IR spectroscopy may be a valuable and feasible tool for cardiac AMR surveillance. In addition, we have shown that only a small number of spectral frequencies are required for accurate classification, potentially allowing for real-time diagnoses using new QCL-based imaging approaches. Future research will focus on whether an AMR-specific spectral signature precedes C4d deposition and could predict those patients who will ultimately develop late stage AMR. In addition, we will determine the role of concomitant ACR in cases of mixed rejection.


The authors would like to thank Dr. Ben Bird and Dr. Jeremy Rowlette from Daylight Solutions for their guidance with QCL IR imaging. The authors would like to thank the University of Chicago Department of Pathology Histology Lab for tissue sample preparation. Finally, the authors would like to acknowledge that this work was funded by the Chicago Biomedical Consortium with support from the Searle Funds at The Chicago Community Trust.


1. Rodriguez E, Skojec DV, Tan CD, et al. Antibody-mediated rejection in human cardiac allografts: evaluation of immunoglobulins and complement activation products C4d and C3d as markers. Am J Transplant. 2005;5:2778–2785.
2. Hammond EH, Hansen JK, Spencer LS, et al. Vascular rejection in cardiac transplantation: histologic, immunopathologic, and ultrastructural features. Cardiovasc Pathol. 1993;2:21–34.
3. Michaels PJ, Espejo ML, Kobashigawa J, et al. Humoral rejection in cardiac transplantation: risk factors, hemodynamic consequences and relationship to transplant coronary artery disease. J Heart Lung Transplant. 2003;22:58–69.
4. Tan C, Sokos G, Pidwell D, et al. Correlation of donor-specific antibodies, complement and its regulators with graft dysfunction in cardiac antibody-mediated rejection. Am J Transplant. 2009;9:2075–2084.
5. Fedrigo M, Gambino A, Tona F, et al. Can C4d immunostaining on endomyocardial biopsies be considered a prognostic biomarker in heart transplant recipients? Transplantation. 2010;90:791–798.
6. Casarez TW, Perens G, Williams RJ, et al. Humoral rejection in pediatric orthotopic heart transplantation. J Heart Lung Transplant. 2007;26:114–119.
7. Reed EF, Demetris AJ, Hammond E, et al. Acute antibody-mediated rejection of cardiac transplants. J Heart Lung Transplant. 2006;25:153–159.
8. Colvin MM, Cook JL, Chang P, et al. Antibody-mediated rejection in cardiac transplantation: emerging knowledge in diagnosis and management. Circulation. 2015;131:1608–1639.
9. Hammond EH, Wittwer CT, Greenwood J, et al. Relationship of OKT3 sensitization and vascular rejection in cardiac transplant patients receiving OKT3 rejection prophylaxis. Transplantation. 1990;50:776–782.
10. Smith RN, Brousaides N, Grazette L, et al. C4d deposition in cardiac allografts correlates with alloantibody. J Heart Lung Transplant. 2005;24:1202–1210.
11. Brasile L, Zerbe T, Rabin B, et al. Identification of the antibody to vascular endothelial cells in patients undergoing cardiac transplantation. Transplantation. 1985;40:672–674.
12. Berry GJ, Burke MM, Andersen C, et al. The 2013 International Society for Heart and Lung Transplantation working formulation for the standardization of nomenclature in the pathologic diagnosis of antibody-mediated rejection in heart transplantation. J Heart Lung Transplant. 2013;32:1147–1162.
13. Szymańska S, Grajkowska W, Sobieszczańska-Małek M, et al. Reclassification of C4d-positive endomyocardial biopsy (EMB) according to new International Society for Heart and Lung Transplantation (ISHLT) 2013 categories for reporting pathologic antibody-mediated rejection (pAMR): preliminary data from a polish single-center study. Ann Transplant. 2015;20:351–356.
14. Fedrigo M, Gambino A, Benazzi E, et al. Role of morphologic parameters on endomyocardial biopsy to detect sub-clinical antibody-mediated rejection in heart transplantation. J Heart Lung Transplant. 2011;30:1381–1388.
15. Hammond MEH, Stehlik J, Renlund DG, et al. Utility of histologic parameters in screening for antibody-mediated rejection of the cardiac allograft: a study of 3,170 biopsies. J Heart Lung Transplant. 2005;24:2015–2021.
16. Colvin RB, Smith RN. Antibody-mediated organ-allograft rejection. Nat Rev Immunol. 2005;5:807.
17. Wu GW, Kobashigawa JA, Fishbein MC, et al. Asymptomatic antibody-mediated rejection after heart transplantation predicts poor outcomes. J Heart Lung Transplant. 2009;28:417–422.
18. Reed EF, Hong B, Ho E, et al. Monitoring of soluble HLA alloantigen and anti-HLA antibodies identifies heart allograft recipients at risk of transplant associated coronary artery disease. Transplantation. 1996;61:566–572.
19. Baker MJ, Trevisan J, Bassan P, et al. Using Fourier transform IR spectroscopy to analyze biological materials. Nat Protoc. 2014;9:1771–1791.
20. Martin FL, Kelly JG, Llabjani V, et al. Distinguishing cell types or populations based on the computational analysis of their infrared spectra. Nat Protoc. 2010;5:1748–1760.
21. Walsh MJ, Reddy RK, Bhargava R. Label-free biomedical imaging with mid-IR spectroscopy. IEEE J Sel Top Quantum Electron. 2012;18:1502–1513.
22. Bhargava R. Infrared spectroscopic imaging: the next generation. Appl Spectrosc. 2012;66:1091–1120.
23. Paraskevaidi M, Martin-Hirsch PL, Martin FL. Progress and challenges in the diagnosis of dementia. ACS Chem Nerosci. 2018;9:446–461.
24. Baker MJ, Byrne HJ, Chalmers J, et al. Clinical applications of infrared and Raman spectroscopy: state of play and future challenges. Analyst. 2018;143:1735–1757.
25. Schwaighofer A, Brandstetter M, Lendl B. Quantum cascade lasers (QCLs) in biomedical spectroscopy. Chem Soc Rev. 2017;46:2903–5924.
26. Bird B, Rowlette J. High definition infrared chemical imaging of colorectal tissue using a Spero QCL microscope. Analyst. 2017;142:1381–1386.
27. Pilling MJ, Henderson A, Bird B, et al. High-throughput quantum cascade laser (QCL) spectral histopathology: a practical approach towards clinical translation. Faraday Discuss. 2016;187:135–154.
28. Tiwari S, Reddy VB, Bhargava R, et al. Computational chemical imaging for cardiovascular pathology: chemical microscopic imaging accurately determines cardiac transplant rejection. PloS One. 2015;10:e0125183.
29. Halliwell DE, Morais CLM, Lima KMG, et al. Imaging cervical cytology with scanning near–field optical microscopy (SNOM) coupled with an IR-FEL. Sci Rep. 2016;6:29494.
30. Varma VK, Kajdacsy-Balla A, Akkina SK, et al. A label-free approach by infrared spectroscopic imaging for interrogating the biochemistry of diabetic nephropathy progression. Kidney Int. 2016;89:1153–1159.
31. Varma VK, Kadjacsy-Balla A, Akkina S, et al. Predicting fibrosis progression in renal transplant recipients using laser-based infrared spectroscopic imaging. Sci Rep. 2018;8:686.
32. Sreedhar H, Varma VK, Bird B, et al. Applications of QCL mid-IR imaging to the advancement of pathology. 2017.Paper presented at: SPIE Medical Imaging.
33. Levin ER, Gardner DG, Samson WK. Natriuretic peptides. N Engl J Med. 1998;339:321–328.
34. Kfoury AG, Hammond MEH, Snow GL, et al. Cardiovascular mortality among heart transplant recipients with asymptomatic antibody-mediated or stable mixed cellular and antibody-mediated rejection. J Heart Lung Transplant. 2009;28:781–784.

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.