Abstract
Frozen cryopreservation (FC) with the vapour phase of liquid nitrogen storage (−135°C) is a standard biobank technique to preserve allogeneic heart valves to enable a preferable allograft valve replacement in clinical settings. However, their long-term function is limited by immune responses, inflammation and structural degeneration. Ice-free cryopreserved (IFC) valves with warmer storage possibilities at −80°C showed better matrix preservation and decreased immunological response in preliminary short-term in vivo studies. Our study aimed to assess the prolonged performance of IFC allografts in an orthotopic pulmonary sheep model.
FC (n = 6) and IFC (n = 6) allografts were transplanted into juvenile Merino sheep. After 12 months of implantation, functionality testing via 2-dimensional echocardiography and histological analyses was performed. In addition, multiphoton autofluorescence imaging and Raman microspectroscopy analysis were applied to qualitatively and quantitatively assess the matrix integrity of the leaflets.
Six animals from the FC group and 5 animals from the IFC group were included in the analysis. Histological explant analysis showed early inflammation in the FC valves, whereas sustainable, fully functional, devitalized acellular IFC grafts were obtained. IFC valves showed excellent haemodynamic data with fewer gradients, no pulmonary regurgitation, no calcification and acellularity. Structural remodelling of the leaflet matrix structure was only detected in FC-treated tissue, whereas IFC valves maintained matrix integrity comparable to that of native controls. The collagen crimp period and amplitude and elastin structure were significantly different in the FC valve cusps compared to IFC and native cusps. Collagen fibres in the FC valves were less aligned and straightened.
IFC heart valves with good haemodynamic function, reduced immunogenicity and preserved matrix structures have the potential to overcome the known limitations of the clinically applied FC valve.
INTRODUCTION
Human allogeneic heart valves have been described as almost perfect heart valve replacements [1]. First introduced in 1962 [2], they have optimal haemodynamic properties, outstanding resistance to infections and require no anticoagulation [3]. Their clinical application is, however, limited due to a worldwide organ donor shortage and long-term failure because of degeneration. Due to logistical issues, valves were cryopreserved with controlled-rate freezing and stored in the gas phase of liquid nitrogen [4], enabling a long-term storage time of up to 5 years. Frozen cryopreservation (FC) has been the choice of preservation of human heart valves worldwide for the last 50 years [5].
Other major shortcomings of cryopreserved allograft heart valves are their cost-intensive vapour phase liquid nitrogen storage and their limited long-term function due to immune responses, inflammation and subsequent structural deterioration [6–8]. Allograft failure is seen more frequently in paediatric patients with increased calcium metabolism resulting in early reintervention procedures [9, 10]. An alternative ice-free cryopreservation (IFC) method was developed on the basis of vitrification principles avoiding ice crystal formation by using high concentrations of cryoprotectants and circumventing the economical disadvantages of liquid nitrogen storage. In a preliminary in vivo animal study, the initial preservation of the extracellular matrix (ECM) was verified after 7 months with superior haemodynamic properties compared to those achieved with standard cryopreservation [11].
The objective of this study was to evaluate the prolonged (1 year) performance of implanted IFC and FC pulmonary valves in an orthotopic pulmonary sheep model, including the integrity of the ECM. We applied multiphoton imaging to detect ECM damage caused by interstitial ice formation in FC heart valves [12, 13]. Furthermore, Raman spectroscopy was used as a novel analysis tool for in depth analysis of collagen structures [14].
METHODS
Tissue preparation
Hearts of 12 adult sheep (Dorset cross, ∼1 year old, obtained from a slaughterhouse in Carlson Meats, Grove City, MN, USA) were obtained using aseptic conditions, rinsed with ice-cold lactated Ringer’s solution (B. Braun, Melsungen, Germany) and placed in sterile Dulbecco’s modified Eagle medium (DMEM, Invitrogen, Carlsbad, CA, USA) for overnight shipment to the processing laboratory (Cell & Tissue Systems, North Charleston, SC, USA). Pulmonary heart valves were excised and stored for 24 h at 4°C in 100 ml antibiotic solution (DMEM containing 4.5 g/l glucose with 126 mg/l lincomycin, 52 mg/l vancomycin, 157 mg/l cefoxitin and 117 mg/l polymyxin). Following the antibiotic treatment, the valves were randomly assigned to the FC and IFC groups. The individual valves of the FC group were cryopreserved, stored in the vapour phase of liquid nitrogen (−135°C) and thawed according to the American Association of Tissue Banks guidelines and as previously described by Lisy et al. [11]. For the IFC group, valve preservation, storage at −80°C in a mechanical freezer and rewarming were conducted via the protocol of Brockbank et al. [15]. After rewarming, the valves were stored in ice-cold lactated Ringer’s solution until implantation (B. Braun).
Implantation/explantation
Twelve juvenile Merino sheep (average age 39 ± 3 weeks, weight 47 ± 4 kg) underwent pulmonary heart valve replacement with either IFC (n = 6) or FC (n = 6) allogeneic valves according to a previously described protocol [16]. Deaths due to perioperative complications due to valve implantation were excluded. All animals were kept in an indoor housing facility and received humane care in compliance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication no. 85-23 revised 1985). Animals were euthanized after 12 months of implantation using an intravenous application of embutramide. The valves were explanted and, following inspection for gross morphology, they were further processed.
Cardiovascular imaging studies
Echocardiography
Two weeks prior to explantation, each animal underwent a Doppler echocardiographic (CX50, Philips, Hamburg, Germany) examination with an S5-1 (iPx-7) probe to evaluate valve function. A blinded 2-dimensional echocardiographic-Doppler examination was performed of the right ventricular outflow tract, conduit and distal main pulmonary artery. In addition, the competence of the pulmonary valve was evaluated using colour-flow Doppler mapping.
In vivo computed tomography and cardiac magnetic resonance imaging
Two representative anaesthetized animals from both groups were used for in vivo computed tomography (CT) and cardiac magnetic resonance imaging (MRI) scans to visualize possible calcification and evaluate the valve function in vivo. Settings and equipment are described in detail in Supplementary Materials Methods.
Histological analysis
After being implanted for 12 months, the heart valve explants were embedded in paraffin (Merck, Darmstadt, Germany). For general morphological analysis, representative 3 μm sections were stained with haematoxylin-eosin, Elastica van Gieson and Movat pentachrome Verhoeff stains. Van Kossa stain was used to detect calcium deposits. CD3 (DCS, Hamburg, Germany) and chloracetate esterase stain were performed to identify T cells and leukocytes, respectively.
Near-infrared multiphoton imaging
One leaflet of each explant was placed individually in 100-ml volumes of ice-cold DMEM with 4.5 g/l glucose and antibiotic mix (1.2 g/l amikacin, 3 g/l flucytosine, 1.2 g/l vancomycin, 0.3 g/l ciprofloxacin, 1.2 g/l metronidazole). Tissues were analysed within 72 h after explantation. A custom-build multiphoton laser system (JenLab GmbH, Jena, Germany) as previously described [17] was used for simultaneous autofluorescence and second harmonic generation imaging. Photomultiplier tube settings for each measurement were 1000 V for contrast and 52.7% for brightness. The attenuator power of the laser was set to 20 mW and the chosen excitation wavelength was 760 nm. The leaflets were placed in a glass-bottom dish (ibidi GmbH, Martinsried, Germany) with phosphate-buffered saline (PBS) and put on the microscope sample holder (40× oil immersion objective; N.A. 1.3; Carl Zeiss, Jena, Germany). Three spots in the ventricularis and fibrosa layers of each tissue were imaged. Crimp amplitude and period were determined for collagen fibre characterization in the fibrosa.
Raman spectroscopy
A customized Raman spectrometer [18] was used to obtain spectra of all leaflets that were used for multiphoton imaging obtained within 48 h after explanation as previously described [14]. In addition, native control pulmonary leaflets and a non-transplanted leaflet of IFC and FC were measured. Leaflets were dissected from the pulmonary root and placed in a glass-bottom dish (ibidi GmbH) in phosphate-buffered saline, with the fibrosa layer facing the bottom. Measurements were taken with a 60× water immersion objective (N.A. 1.2; Olympus, Japan). Collagen structures were targeted by the confocal brightfield imaging function of the microscope (cellB; Olympus Soft Imaging Solutions, Muenster, Germany) and 30 spots per tissue of the collagen fibres were randomly selected. Laser power was set to 85 mW, and spectra were acquired at an exposure time of 100 sec. Spectral data were processed prior to multivariate data analysis by spectroscopy software OPUS 4.2 (Bruker Optik GmbH, Ettlingen, Germany).
Enzyme-linked immunosorbent assay
Segments of each leaflet were used for the enzyme-linked immunosorbent assay (ELISA). ETDA-free protease inhibitor cocktail (complete mini, Roche Diagnostics, Mannheim, Germany) was added to the radioimmunoprecipitation assay lysis buffer (Thermo Fisher Scientific, Waltham, MA, USA) in the ratio 1:10. Tissue and lysis solution (15 µl/mg tissue) were put in innuSpeed lysis tubes (Analytik Jena AG, Jena, Germany) and processed for 2 × 4 min in a SpeedMill Plus homogenizer (Analytik Jena AG). After 2 h of incubation on ice, lysates were centrifuged, and supernatants were frozen at −80°C. The supernatants were utilized for sheep collagen I ELISA (Neobiolab, Cambridge, MA, USA).
Statistical analysis
Statistical analyses were conducted with Prism 6 (GraphPad, La Jolla, CA, USA). Data are represented as the median and the interquartile range due to the small sample size. Data were checked for normality. Unpaired parametric t-tests or the non-parametric Mann–Whitney tests were performed for 2-sample comparisons. Differences between the 3 groups were assessed with an ordinary non-matched analysis of variance or the Kruskal–Wallis test. Pairwise comparisons were done with the Bonferroni correction for multiple comparisons. All statistical tests were 2-sided, and P-values of 0.05 or less were considered statistically significant.
Spectral data analysis to identify matrix-specific peak shifts and spectral differences was performed by principal component analysis (PCA) using the multivariate data analysis software, The Unscrambler® (CAMO, Oslo, Norway). Data were vector normalized, and PCA calculated 7 principal components (PCs) based on a non-linear iterative partial least squares algorithm. Each PC defines certain spectral information describing spectral similarities or differences. The score values and the loadings plot were compared to define the most relevant PC for ECM-related spectral characteristics.
RESULTS
Twelve successful pulmonary heart valve transplantations were performed. After 7 months with the implanted heart valves, 1 animal of the IFC group was excluded due to bacterial endocarditis. Six animals of the FC and 5 animals of the IFC group survived the entire 12 months.
Gross morphology
For both groups, thin and translucent leaflets were obtained (Fig. 1A and C). Minor inflammation in 1 valve of each group was detectable on the arterial side of the leaflet (Fig. 1B and D).
Gross morphology of representative ice-free cryopreserved (A, B) and frozen cryopreservation (C, D) heart valves after explantation. The leaflets in both groups were thin with translucent leaflets (A, C). One heart valve from each group showed a slightly thickened and inflamed cusp (indicated with arrows) (C, D).
Cardiovascular imaging results
No valvular stenosis with subsequent right heart failure was observed in either group. All animals developed minimal pulmonary insufficiency (50 ± 6% weight gain between implantation and explantation). Elevated values for the maximum flow velocity out of the right ventricle outflow tract, for peak pressure gradients, and for the right ventricular end-diastolic diameter were detected for the FC group. However, only right ventricular end-diastolic diameter values were increased with statistical significance compared to the IFC group (Fig. 2A–D, Supplementary Material, Table S1). Additional individual in vivo cardiac MRI scans confirmed the maximum flow velocity out of the right ventricle outflow tract and very small regurgitation fractions, indicating a minimal or rather a physiological pulmonary insufficiency [19] for the representative animals of each group.
Haemodynamic and histomorphological data of IFC (n = 5) and FC (n = 6) valves. Haemodynamic in vivo function was evaluated with Doppler echocardiography (CX50, Philips, Hamburg, Germany). The maximum blood flow through the RVOT vmax (A), the peak pressure gradient (B), the RV ED diameter (C) and the graded PI (D) (categories: 0 = none, 1 = minimal, 2 = mild, 3 = moderate, 4 = severe) are displayed. For the histomorphological analysis, the area of the leaflet with the general and CD3 positive cell count of each representative valve was determined with AxioVision Software for routine bright-field light microscopy (Axio Observer Z1, Carl Zeiss, Göttingen, Germany). The general (E) and CD3 (F) cellularity is presented as cells per 1000 μm2. *P-values of 0.05 or less were considered statistically significant. FC: frozen cryopreservation; IFC: ice-free cryopreserved; PI: pulmonary insufficiency; RV ED: right ventricular end diastolic; RVOT vmax: maximum flow velocity out of the right ventricle outflow tract.
Histology
Histological analysis revealed FC valves to be more frequent (3/6) and more severely inflamed compared to IFC valves. Although the IFC heart valve leaflets were virtually acellular, FC inflamed leaflets were infiltrated, and a mild to moderate pannus formed on the ventricularis side (Fig. 3A–P). Leaflet tip thickening was detected in the FC group, whereas the IFC leaflet tip diameter was reduced (leaflet tip diameter: native: 207.4 ± 4 μm; FC: 233.8 ± 96.3 μm; IFC: 96.3 ± 43.5 μm). The thickened FC leaflets had a less condensed pars fibrosa and pars ventricularis. In contrast, the IFC leaflets had a more condensed pars fibrosa and pars ventricularis with condensed elastin fibres (Fig. 3F, G, J, K).
Histomorphological data of representative native control, IFC and FC leaflets. The leaflet belly (A–L) with the arterial and ventricular (v) side and the hinge region (M–P) were displayed. Haematoxylin-eosin (A–D), movat pentachrome (E–H, M–P) and elastica von Gieson (I–L) staining were performed. IFC valves showed thin, acellular leaflets (B, F, J, N). Three out of 6 FC valves had cell-free, thin leaflets with sporadic cells on the borders (C, G, K, O). The other 3 leaflets (D, H, L, P) showed thickened and severely infiltrated regions. A mild to moderate pannus formation, correlating to the intensity of leaflet infiltration, was detected for these valves (P: dashed rectangle). IFC leaflets were more condensed, whereas FC leaflets had thickened leaflets with looser structures. Scale bar equals 200 µm. FC: frozen cryopreservation; IFC: ice-free cryopreserved.
CD3 staining was performed to identify T cells (Fig. 4). Comparable to the native control (Fig. 4A), no CD3-positive T cells were found in the acellular IFC leaflets (Fig. 2E, F; Fig. 4B; Supplemental Table 1). Moderate T-cell infiltration was detected in the thickened areas of the infiltrated FC valves with sporadic T cells present in the pannus (Fig. 4D). Sporadic T cells were observed in the hinge region near the former muscle band and the sutures for both groups (Fig. 4E–H). T cells were present in the conduit in the tunica externa with signs of neovascularization. Infiltration by leukocytes was analysed using chloracetate esterase stain. No granulocytes or monocytes were found in valve leaflets of either group. Sporadic monocytes with a granulated cytoplasm were detected near the former suture points for both groups, as shown in the representative IFC section (Fig. 4I).
CD3 (A–H) and chloracetate esterase (I) representative sections of the native control (A) and the IFC (B, E, F, I) and FC (C, D, F, H) valves. T cells were abundantly present in the leaflet belly (A–D) of the FC-inflamed valve (D). Both groups had similar levels of T-cell infiltration in the hinge region (E, F) and near the sutures lines (G, H). No granulocytes, but sporadic monocytes near the suture lines, were detected for both groups (I). Scale bar equals 50 µm. FC: frozen cryopreservation; IFC: ice-free cryopreserved.
Neither group had calcifications in the leaflet or hinge region. Minor calcifications were detected next to the sutures. In vivo CT scans confirmed these findings (Supplementary Material, Fig. S1).
Multiphoton imaging
Multiphoton laser-induced autofluorescence imaging and second harmonic generation microscopy were used to image collagen and elastic fibres (Fig. 5). In comparison to microscopic brightfield images (Fig. 5A–C), the main ECM components were clearly distinguishable. In all tissues, the fibrosa was characterized by parallel alignment of collagen bundles with disorganized structures in the ventricularis. Native and IFC leaflets showed similar matrix structures. Less aligned, deteriorated and straightened collagen fibre structures were found in FC-treated tissues (Fig. 5D–F). The fibrosa of the FC leaflets showed significant differences in the crimp period and amplitude, which were comparable for native and IFC leaflets (Fig. 5K and l). The ventricularis layer revealed a more disorganized overall pattern in FC than in IFC (Fig. 5G–I).
Brightfield (A–C) and multiphoton (D–I) images of fibrosa and ventricularis of native and transplanted leaflets. Collagen and elastic fibres of the fibrosa showed parallel alignment (A–F); there was no ordered structure in the ventricularis (G–I). FC leaflets (C, F, I) showed deteriorated collagen structures. IFC leaflets (B, E, H) resembled the arrangement of the native matrix structures (A, D, G). (J) Crimp period (CP) and crimp amplitude (CA) were determined and showed significant differences for FC-treated leaflets; n = 5 (K, L). Green: collagen, red: elastic fibres, cells; Scale bar: brightfield equals 20 µm, multiphoton equals 60 µm. *P-values of 0.05 or less were considered statistically significant. FC: frozen cryopreservation; IFC: ice-free cryopreserved.
Elastin structures were visibly more disrupted in the FC group and better preserved in IFC-treated tissues. Increased density of elastic fibres was observed in the IFC-treated leaflets on the ventricularis side compared to the native tissue. Despite the less aligned structures, the signal intensity remained similar for the native, FC and IFC tissues. Multiphoton autofluorescence imaging only allowed a qualitative interpretation of the ventricularis integrity, revealing structural damages, especially in the FC group.
Raman analysis of fibrosa extracellular matrix structures
Raman spectroscopic analyses of the ECM structures of the fibrosa of native control pulmonary valves and of transplanted and non-transplanted IFC and FC valves were performed for an in-depth study of changes in the collagen. PCA was applied to evaluate spectral differences between the individual groups. PC2 revealed distinct clustering of overlapping native and IFC tissue data in the negative range and FC tissue data in the positive range (Fig. 6A). The mean score values of PC2 indicated significant differences between FC and native data but not between IFC and native tissues (Fig. 6B). Five peaks correlating to a certain molecular bond in the collagen molecule [20] were identified in the loadings plot (Fig. 6C) and displayed the most influencing peaks on spectral differences in the fingerprint region. Single peak shifts but no decrease in the overall ECM fingerprint intensity were detected. These results were consistent with no decreased Col I content in FC leaflets quantified by Col I ELISA (Fig. 6D).
Multivariate data analysis. (A) Principal component analysis scores plot; FC tissue data were clustered, whereas IFC and native tissue data overlapped. Each spot correlates with a single spectrum. (B) Score values of PC2. Significant difference between FC and native tissue but not between IFC and native leaflet is shown. *P = 0.05, n = 5. (C) Loadings plot of PC2 identified outstanding peaks characterizing certain molecular bonds specific in collagen. (D) Collagen I enzyme-linked immunosorbent assay (µg collagen/mg tissue) showed no significant decrease in FC leaflets compared to IFC leaflets. (E) Significant collagen peaks of the loadings plots and their correlating molecular groups. FC: frozen cryopreservation; IFC: ice-free cryopreserved; PC: principal component.
PCA of native, non-transplanted and transplanted tissues showed distinctly higher scores for the non-transplanted tissues compared to the transplanted equivalents (Supplementary Material, Fig. S2). Transplanted and non-transplanted IFC tissues showed no significant differences in their scores. The differences in the transplanted and non-transplanted FC tissues were highly significant. A general shift towards the native molecular pattern of collagen structures was seen in both transplanted groups. Importantly, the IFC group was more similar to the native group than the FC group.
DISCUSSION
Structural deterioration and inflammatory processes are long-term obstacles of the current state-of-the-art frozen cryopreserved heart valves used in clinical practice. This study demonstrated that these shortcomings can be overcome by using IFC valves due to the protection and maintenance of natural ECM structures, resulting in outstanding long-term function of acellular grafts. We implanted IFC and FC heart valves in an orthotropic sheep model, a common cardiovascular model for biological heart valves [21]. Our extended in vivo testing over 1 year correlates to approximately 40 million wear cycles [22, 23] and corresponds to roughly 18 years of implantation in humans. Different sheep strains (donor: Dorset cross sheep; recipient: Merino sheep) were used to obtain a true allogeneic model, which better simulates the naturally occurring, more extensive genetic differences among humans. According to the American Association of Tissue Banks guidelines, the standard protocol was applied for FC to ensure state-of-the-art preservation for heart valve transplants.
At the time of explantation, IFC heart valves showed well-preserved ECM integrity. IFC valves exhibited excellent function with leaflets remaining acellular and a sporadically cellularized conduit wall, indicating that recellularization was not fundamentally important for long-term function. Three out of 6 FC valves revealed early signs of long-term graft failure. Leaflet thickening and moderate T-cell infiltration were detected, both of which are known to be responsible for early degeneration [24, 25]. Deep tissue analysis of FC valves revealed no signs of remaining native cells.
The ECM of the leaflets has to withstand extreme biomechanical demands due to major deformation processes and large sheer stresses [23, 26]. Therefore, we focused on the analysis of the architecture of leaflet ECM fibres. We hypothesized that, by maintaining the structural integrity of the ECM, the degeneration processes can be reduced or even avoided, thereby preventing long-term failure of heart valve implants [6, 27]. Multiphoton imaging demonstrated collagen and elastin degradation similar to that found in mid-term implanted (7 months) FC valves [11]. Consequently, previously described destructive effects of standard cryopreservation due to intratissue ice formation [28] were detected in the FC group even 1 year after implantation. Moreover, a severely impaired collagen crimp structure of the FC valves with a crimp altitude decrease of 2.4-fold and a 2.2-fold increase in the crimp period, indicated long-term failure and loss of biomechanical sustainability. The loss of the characteristic crimp structure of collagen Type I in the fibrosa has previously been associated with reduced flexural rigidity of tissue valves [29]. In contrast to FC leaflets, native-like matrix structures in the IFC leaflets were detected after explantation. Additional PCA results of Raman spectroscopy analysis of collagen confirmed that the IFC group had native-like collagen structures after being implanted for 12 months, whereas the FC group showed a lack of integrity of the matrix. Raman enabled the non-destructive detection of ECM changes at the molecular level. Specific changes in certain molecular bonds in collagen instead of general collagen degeneration were shown, with PCA loading plots indicating no degradation processes in the FC leaflet. Additional collagen quantification determined by applying ELISA demonstrated that the overall collagen content was not reduced after 12 months in FC allografts compared to IFC grafts.
PCA including non-transplanted leaflets showed distinct differences between native and non-transplanted preserved tissues with a shift towards native-like PC2 score values after implantation in both groups. Those results need to be analysed carefully because of the lack of statistical significance due to of the lack of non-transplanted tissues. Nevertheless, we hypothesize that this shift can be observed due to dehydration of tissue during FC and IFC cryopreservation. The tissue might not be fully hydrated after thawing. Additional passive mechanically driven rearrangements of collagen might also occur over time due to the major deformation processes and large sheer stresses. Additional Raman analyses of tissue hydration processes and potential ex vivo bioreactor experiments might offer deeper insight.
CONCLUSION
In summary, IFC offers a preferable preservation method compared to FC. Despite the small number of animals in this expensive large-animal model and limited cardiac MRI data sets for all animals, we successfully demonstrated a distinct improvement in long-term function with regard to immunogenicity and true ECM preservation. Moreover, our preservation method has additional economical, financial, logistical and processing advantages because it involves less complex and less expensive equipment. Processing and storage at only −80°C, shipment in dry ice and easier preservation protocols result in a cost effective method applicable even in developing countries. We are convinced that transplanted IFC pulmonary heart valves will be superior in a clinical setting compared to standard FC grafts. This hypothesis is supported by Neumann et al. [30], who observed favourable long-term haemodynamic results over 8 years of follow-up and no cellular immune responses over 36 months for fresh decellularized allografts in young adults and children. We postulate that the remaining cells or the repopulation of cells is not required for good long-term function of the allogeneic valves. Possible devitalization or even decellularization processes with the aid of the IFC solution should be elucidated in the future due to its ease and cost-effectiveness. IFC represents a promising new technique for maintaining a more natural ECM makeup and minimizing the immunological reactions in vivo.
ACKNOWLEDGEMENTS
We thank Harald Keller and Eva Knoepfler for technical assistance with the cardiopulmonary perfusion.
Funding
This work was supported by the German Research Foundation [Sto 359/10-1 to U.A.S., SCHE 701/10-1 and INST 2388/30-1 to K.S.-L.] and the Ministry of Baden-Wuerttemberg for Sciences, Research and Arts [33-729.55-3/214 and SI-BW 01222-91]. K.G.M.B. was supported by the National Institute of Allergy and Infectious Disease, National Institutes of Health [grant R43 AI114486].
Conflict of interest: Julian L. Wichmann received speaker fees from GE Healthcare and Siemens Healthcare. Kelvin GM Brockbank is the owner and the employee of Tissue Testing Technologies LLC.
REFERENCES
Author notes
Anna Christina Biermann and Julia Marzi authors contributed equally to this study.
