Author + information
- Received February 13, 2018
- Revision received May 9, 2018
- Accepted June 22, 2018
- Published online January 7, 2019.
- Stephanie L. Sellers, PhDa,b,
- Christopher T. Turner, PhDc,
- Janarthanan Sathananthan, MBChB, MPHd,
- Timothy R.G. Cartlidge, MDe,
- Frances Sina,b,
- Rihab Bouchareb, PhDf,
- John Mooney, MD, PhDa,
- Bjarne L. Nørgaard, MD, PhDg,
- Jeroen J. Bax, MD, PhDh,
- Pascal N. Bernatchez, MDb,
- Marc R. Dweck, MBChB, PhDe,
- David J. Granville, PhDb,c,
- David E. Newby, MD, PhDe,
- Sandra Lauck, PhDd,
- John G. Webb, MDd,
- Geoffrey W. Payne, PhDi,
- Philippe Pibarot, PhD, DVMf,
- Philipp Blanke, MDa,
- Michael A. Seidman, MD, PhDb,c and
- Jonathon A. Leipsic, MDa,b,d,∗ ()
- aDepartment of Radiology, St. Paul’s Hospital and University of British Columbia, Vancouver, British Columbia, Canada
- bCentre for Heart Lung Innovation, St. Paul’s Hospital and University of British Columbia, Vancouver, British Columbia, Canada
- cDepartment of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, British Columbia, Canada
- dCentre for Heart Valve Innovation, St. Paul’s Hospital, Vancouver, British Columbia, Canada
- eBritish Heart Foundation Centre for Cardiovascular Science, University of Edinburgh, Edinburgh, Scotland, United Kingdom
- fPulmonary and Cardiology Research University Institute, Laval University, Quebec City, Quebec, Canada
- gDepartment of Cardiology, Aarhus University Hospital Skejby, Aarhus, Denmark
- hDepartment of Cardiology, Leiden University Medical Centre, Leiden, South Holland, the Netherlands
- iUniversity of Northern British Columbia, Prince George, British Columbia, Canada
- ↵∗Address for correspondence:
Dr. Jonathon Leipsic, Department of Radiology, St. Paul’s Hospital, 1081 Burrard Street, Vancouver, British Columbia V6Z 1Y6, Canada.
Objectives This study investigated processes causing leaflet thickening and structural valve degeneration (SVD).
Background Although transcatheter aortic valve replacement (TAVR) has changed the treatment of aortic stenosis, concerns remain regarding SVD, potentially related to valve thrombosis and thickening, based on studies using computed tomography (CT). Detailed histological analyses are provided to help attain insights into these processes.
Methods Explanted transcatheter heart valves (THVs) were evaluated for thrombosis, fibrosis, and calcification for quantification of leaflet thickness. Immunohistochemical and microscopy approaches were used to investigate SVD-associated mechanisms.
Results THVs (n = 23) were obtained from 22 patients (median 81 years of age; 50% male) from 0 to 2,583 days post TAVR. Maximal leaflet thickness increased relative to implant duration (ρ = 0.427; p = 0.027). THVs explanted after >2 years were thicker than those explanted after <2 years (p = 0.007). All THVs had adherent thrombus on both aortic and ventricular sides, which beyond 60 days was seen in combination with fibrosis and beyond 4 years had calcification. Early thrombus formation (<60 days) occurred despite rapid endothelialization with an abnormal hyperplastic phenotype. Fibrosis was observed in 6 patients on both the aortic and the ventricular THV surfaces, remodeled over time, and was associated with matrix metalloproteinase-1 expression. Five THVs showed overt calcification associated with adherent thrombus and fibrosis.
Conclusions There is a time-dependent degeneration of THVs consisting of thrombus formation, endothelial hyperplasia, fibrosis, tissue remodeling, proteinase expression, and calcification. Future investigation is needed to further understand these mechanisms contributing to leaflet thickening and SVD.
Transcatheter aortic valve replacement (TAVR) has revolutionized the treatment of symptomatic severe aortic stenosis (1–4). However, valve thrombosis and leaflet thickening have been reported to contribute to restricted leaflet motion in transcatheter heart valves (THVs) and may relate to long-term complications such as structural valve degeneration (SVD) (5,6). To date, reports of valve thrombosis, leaflet thickening, and SVD are mostly limited to post-implantation imaging by using computed tomography (CT) and echocardiography (5–11), and although recent reports of explanted THVs provide standard histological analysis (12-14), in-depth histological analyses of valve thrombus and degeneration, as well as elucidating potential causative mechanisms, are lacking. Therefore, there is a great deal of uncertainty regarding the significance of leaflet thickening and appropriate clinical management and mechanisms upon which to base future investigations, treatments, and innovations on valve design. This study investigated correlates of leaflet thickening and SVD in a systematic analysis, using a multimodal ex vivo analysis of explanted THVs.
Patients and explanted valves
Explanted THVs from TAVR patients were obtained from the Cardiovascular Tissue Registry at St. Paul’s Hospital (Vancouver, Canada). THVs were obtained during either autopsy or surgical explantation. Patient and procedural data were collected from retrospective chart reviews. The Research Ethics Board at the University of British Columbia approved the study.
Sample preparation and processing
THVs were fixed and stored in 10% (v/v) neutral buffered formalin. Leaflets were then removed from the surrounding device frame, bisected, and cut into 2- to 4-mm sections and embedded in paraffin (Supplemental Figure 1), resulting in 7 to 8 tissue cross-sections analyzed per leaflet. Paraffin sections (4 μm) were used for histology and immunohistochemistry (IHC) as described below. All analyses were assessed by a blinded cardiac pathologist.
Hematoxylin and eosin (H&E) staining was used to assess overall pathology. Subsequently, Movat’s pentachrome staining was used to differentiate thrombus and fibrosis as well as the presence of elastic fibers. Given that the term pannus has differing implications for surgical bioprosthetic valves at a macroscopic and histological level, the term fibrosis is used throughout the manuscript for clarity. Von Kossa staining was used to assess for the presence of calcification.
Image analysis was performed of high-resolution slide images generated using an Aperio slide scanner, using ImageScope software (Leica Biosystems, Wetzlar, Germany). The overall presence of thrombus, fibrosis, and calcification as well as maximal leaflet thickness was assessed on a per-patient basis (Supplemental Figure 2A). Valves explanted due to infective endocarditis were excluded from analysis of leaflet thickness. Images were used to assess thrombotic cross-sectional areas relative to leaflet cross-sectional areas (Supplemental Figure 2B) as well as the accumulation of thrombus predominantly toward the leaflet base in valves implanted for <30 days.
IHC was performed to evaluate expression of protein and cell surface markers, using manual staining or the Bond Rx system (Leica). A subset of valve leaflets representative of the overall population was selected for IHC analysis (Supplemental Tables 1 and 2). To avoid potential confounding factors, THVs from valve-in-valve procedures or infected valves were excluded from selection. The omission of the primary antibody served as negative controls (CD31, CD45, matrix metalloproteinase [MMP]-1) (Supplemental Table 3). Briefly, CD31 was used as a marker of endothelial cell(s) (EC), and CD45/leukocyte common antigen (LCA) was used as a pan-leukocyte marker to assess inflammatory infiltrates. Cross-sections (3 to 8) were studied per valve for CD31 and CD45 with 2 THVs unavailable for IHC for logistical reasons, which were excluded from analysis (Supplemental Table 1). Endothelial hyperplasia and activation were scored on a qualitative scale of 0 through 3 based on CD31 staining, where 0 = none, 1 = mild, 2 = moderate, and 3 = extensive (Supplemental Table 4). Protein expression levels of human proteoglycan decorin and human MMP-1 were assessed in a selected subset of 8 patients (Supplemental Table 2) as an indication of extracellular matrix assembly and remodeling or degradation, respectively.
Second harmonic generation
Analysis of collagen and elastin components was completed with a second-harmonic generation (SHG) signal analysis performed using a multiphoton confocal approach on deparaffinized, unstained slides, as described previously (15). Briefly, slides were scanned using the Leica confocal microscope, using an 800-nm excitation laser and a 417-nm dichroic filter. Confirmation of collagen versus elastic fibers was confirmed using Movat pentachrome stain.
Scanning electron microscopy and energy dispersive X-RAY spectroscopy
Scanning electron microscopy (SEM) with energy dispersive X-ray spectroscopy (EDX) was performed to assess for the presence of calcium within THV leaflets. SEM with EDX was performed as described previously (16); images of deparaffinized sections on glass coverslips were obtained using EDX to compare the calcific deposits on cell surfaces. The operating program identified peaks by position in the spectrum and shape of count distribution, matching these to a stored reference library of elemental peak spectra and their derivatives. Accurate quantification depended on identification of every peak present in the spectrum: count data were collected for all detected peaks, and calcium and phosphorus were quantified from the proportions present in the region of interest relative to the other elemental species present.
Statistical analysis was performed using Prism software version 5 (GraphPad, San Diego, California). Comparison of maximal leaflet thicknesses was performed using a two-tailed Mann-Whitney U test, with correlation between maximal leaflet thickness or thrombus area and implant duration, performed using Spearman correlation with a 1-tailed p value. Statistical significance was taken as a p value <0.05.
Patients and explanted valves
In total, 23 THVs were explanted from 22 patients (Table 1). TAVR procedures were performed between 2006 and 2016. Patients were 81.17 (IQR: 761 to 84.4) years of age with males and females evenly distributed. Two patients had known bicuspid aortic valves. THVs were predominantly implanted using a transfemoral approach (n = 17) and the remaining were implanted transapically (n = 6).
THVs were explanted between 2007 and 2017; the majority (n = 20) were post-mortem explants, and the remainder (n = 3) were explanted during surgical intervention. Valves explanted at autopsy (13 of 20) were mostly deemed to be procedure-related (<30 days), whereas in 5 of 20 patients, cause of death was associated with valve dysfunction, and in 2 of 20, were long-term explants with cause of death unrelated to THV dysfunction (Table 2). The median age of patients at explant was 81.18 (IQR: 76.1 to 88.2) years of age. Median implant duration was 14 days (range 0 to 2,583 days; IQR: 2 to 111.5 days). THVs were predominantly balloon-expandable devices (n = 19) (Edwards Lifesciences, Irvine, California) but there were also 3 self-expandable Portico THVs (St. Jude Medical, Minneapolis, Minnesota) and 1 self-expandable model 9500TFX THV (Maverick, Edwards Lifesciences). Valves were found in a variety of sizes (Table 1).
THVs implanted for more than 2 years had greater maximal leaflet thickening than those implanted for <2 years (p = 0.007) (Figures 1A and 1B). Furthermore, leaflet thickness correlated positively with duration of THV implantation (Spearman ρ = 0.427; p = 0.027) (Figure 1C).
Natural history of thrombus formation, fibrosis, and calcification
To further our analysis of leaflet thickening, we considered the composition of cellular material that contributed to leaflet thickening by looking at the formation of thrombus, fibrosis, and calcification (Figures 2A to 2C). On a per-patient basis, histological evaluation demonstrated a progressive pathology over time regardless of cardiovascular risk factors, explantation mode, or anticoagulation therapy (Table 2, Figure 2D). Moreover, fibrosis was observed only on valves with thrombus, whereas calcification was observed only in the presence of both fibrosis and thrombus.
Thrombus was present almost immediately following deployment (day 0 explant, Sapien 3 device [Edwards Lifesciences], removed surgically). Although this explant might have been affected by acute procedural complication, thrombus was found on all valves explanted, including those as early as 1 day and as late as 2,583 days. Thrombus was present on both the aortic and ventricular aspects and in both laminar and papillary patterns, suggesting that thrombus formation occurred under both laminar and turbulent blood flow patterns (Figure 2D). Analysis of thrombus area in a subset of THVs implanted for <30 days (n = 14) showed that in 50% (n = 7) had predominantly thrombus accumulation at the leaflet base could be found (Figure 2E). However, the thrombus cross-sectional area-to-valve cross-sectional area ratio in a subset of THVs showed no correlation between thrombus deposition and implantation (Spearman ρ = –0.17; p = 0.31) (Figure 2G).
Valve endothelium and inflammation
Given the observation of rapid early thrombus formation, it was hypothesized that such thrombus forms before valve endothelization. However, CD31-positive endothelial-like cells were identified on all THVs, including those explanted as early as 1 day after TAVR, with thrombus observed overlaying endothelial cells (ECs). These cells demonstrated an abnormal pseudostratified columnar hyperplasia phenotype with confounding staining of platelets excluded by morphology (Figures 3A and 3B). Furthermore, inflammatory cell infiltration was evaluated using leukocyte common antigen (LCA/CD45) expression and minimal inflammatory response was found (Figure 3D), albeit with some instances of infiltration with polymorphonuclear and giant cells (Figure 3E). Semiquantitative scoring of endothelial response revealed that this abnormal hyperplasic phenotype was most severe in valves explanted between 8 and 30 days (Figure 3C), relative to both earlier (p < 0.0001) and later (p < 0.01) explants.
Thrombus in combination with fibrosis was observed in all patients with THVs implanted for >60 days (Table 2, Figure 2). All patients with fibrotic valves (n = 6), excluding those with infected THVs, had fibrosis on both the aortic and the ventricular surfaces. Leaflet fibrosis demonstrated a pattern of progressive remodeling and reorganization as indicated by histological features of immature to mature fibrosis over time (Figure 4A). Analysis of a subset of fibrotic valves (explanted at days 946, 1,611, and 2,496) showed less staining for the proteoglycan decorin in immature (earlier) than mature (later) fibrosis (Figure 4A). In addition, SHG analysis displayed diminished signal in areas of immature versus mature fibrosis (Figure 4B). Also, MMP-1 expression, which may regulate collagen remodeling in fibrosis, or act to degrade the THV leaflet pericardium, was associated with fibrosis (Figure 4A).
In the 4 THVs explanted after more than 4 years of implantation, valve calcification was observed in combination with both thrombus and fibrosis (Table 2, Figure 2). A notable exception was a THV implanted for a valve-in-valve procedure and explanted due to endocarditis; this demonstrated calcification much sooner, at 145 days after implantation.
Among the 5 cases of calcified THVs, calcification was observed within adherent thrombus and fibrosis as well as confined within the THV leaflets themselves where it was observed to disrupt valve structure (Figure 5A). In addition, the presence of calcification was confirmed using SEM-EDX (Figure 5B).
Our study describes detailed histomorphometric findings from a series of explanted THVs and the associated time-dependent changes in components of SVD, including fibrosis and calcification (17). Consistent with prior clinical imaging reports (7,9), the current findings suggest a sequential cascade of thrombus formation within hours, fibrosis after 60 days, and calcification after 4 years, which ultimately contribute to progressive leaflet thickening and SVD. These findings have important ramifications for future valve design, potential adjunctive therapies, and further determination of mechanisms driving SVD.
This study has a number of strengths. First, key histological, immunohistochemical, and SEM-EDX analyses were conducted that provide novel mechanistic insights into the natural history and SVD of THVs. In addition, a wide spectrum of implantation duration was established so that the time-dependency of findings could be explored. Finally, a range of different THVs from different manufacturers was examined to underscore the generalizability of findings.
The findings support the hypothesis of progressive leaflet thickening over time and are consistent with those of prior reports of leaflet thickening observed on computed tomography (CT), in which valve thrombus is commonly identified as being crescent shaped and thickest at the base of the leaflets. Indeed, one-half of the explants in the current study demonstrated the propensity of thrombus to accumulate at the leaflet base in this fashion. However, thrombus was found on all valves, unlike the stated prevalence of 7% to 14% on CT for surgical and transcatheter bioprostheses, which is likely to reflect selection bias because thrombus is likely more common among subjects who have died or have undergone repeated surgery. Furthermore, how histological findings of thrombus translate to those appreciable on current clinical imaging modalities and the threshold at which histological thrombus can be considered subclinical remains unknown and a point for future investigation. Despite these limitations our findings provide important insight into the potential histopathological correlates to the recently chronicled imaging findings.
Notably, recent reports suggest lower rates of thrombus (∼20%) of THVs at post-mortem or at surgical explantation. Reasons for this disparity are unclear, although previous reports arose from a limited number of post-mortem examinations or reflect a series from a randomized controlled trial of severe cases treated with only 1 THV type and may result from a different threshold for defining thrombus formation (12–14,18). Overall however, it does appear that thrombus formation is common and of potentially pathophysiological relevance. This suggests that future studies should consider whether short- or long-term anticoagulation may have a role in prolonging the longevity of THVs.
Although anticoagulation therapy may seem like an obvious solution to the problem of valve thrombus formation, our findings suggest that there may be opportunities to better understand the role and mechanism of endothelial dysfunction as a driver of leaflet thrombus. It has been assumed that THV thrombus occurs because of a lack of endothelialization, but the present data suggest this may not be the case as the THVs displayed rapid endothelialization and, despite this, thrombus formation was present on all valves. In some cases, ECs had a markedly abnormal pseudostratified columnar hyperplasia phenotype. This suggests that, rather than a lack of endothelialization, endothelial dysfunction may contribute to valve thrombus and that mechanisms driving such dysfunction may be potentially therapeutic targets to prevent thrombus. Moreover, given the atypical endothelial morphology and lack of CD45 staining, this underscores the need for incorporating IHC into future investigations. Study of EC surface markers would also be of benefit to further elucidate the phenotype and function of these cells before fixation, and en face staining of ECs would define overall leaflet surface coverage.
Although the present analysis cannot prove a direct link between thrombus and fibrosis, fibrosis was present only in combination with thrombus, suggesting a stepwise progression of valve pathology. Fibrosis was also found on both aortic and ventricular sides of THVs, which is an unanticipated finding based on historical reference to surgical aortic valves wherein fibrosis is commonly reported as an ingrowth from the sewing ring (21), a situation that is not possible in THVs. Therefore, the origin of cells depositing extracellular matrix should be investigated and may provide a target to prevent fibrosis.
Fibrotic remodeling during valve thrombus from an immature to mature fibrosis was characterized by a lack of cellularity and dense, linear organization as well as decorin expression and increased SHG signal which are expected in the presence of organized mature collagen arrangement. This may be of importance as mature fibrosis would present an extreme challenge to remove from a surface such as a THV in the event it impedes valve function. This is further complicated by the fact that CT cannot, at present, distinguish thrombus and mature fibrosis. In addition, cellular mechanisms involved in fibrosis and fibrotic remodeling are known to have the potential to contribute to SVD; THV leaflets are made of a pericardium which contains primarily collagens types I and III, both substrates of MMP-1. We found expression of MMP-1 in THVs with fibrosis, and cleavage of THV pericardium by MMP-1 is a potential mechanism by which SVD may occur.
Finally, determining means to prevent thrombus and fibrosis may be important as they may serve as a potential substrate for calcification, a process we observed early post-TAVR using SEM-EDX. In the future, imaging this pathway could potentially be addressed using techniques such as fluorine-18-labeled sodium fluoride positron emission tomography/CT as has been done in native aortic valve calcification (19,20). This would allow for evaluation of avidity in areas of anatomical thickening on CT suggesting the potential for areas of microcalcification much earlier than the overt calcification can be demonstrated on standard clinical CT imaging.
Given that the explanted THVs used in this study were obtained from autopsy or surgical explantation (i.e., samples of convenience), we cannot be certain that the histopathological findings are representative of the broader spectrum of THV pathology and early SVD. Moreover, this population naturally yields a non-normal distribution of extreme distributions and, thus, correlation methods will likely yield inflated results. This method of obtaining valves as well as the historic nature of many of the cases makes us unable to provide insight into factors which may contribute to thrombus formation and/or SVD, including valve expansion and positioning (12,18). Moreover, the sample size is limited and represents THVs from a single tissue registry. This limited sample size also does not provide a comparison of pathology between different valve types (e.g., make, size, or self-expanding vs. balloon expandable). Limited valve numbers also creates the need to set arbitrary cut points for analysis. Finally, as with all histological studies, we do not have true longitudinal data and can only assess the explanted valve at a single time point. Despite these limitations, our histological analysis has provided invaluable initial insight to leaflet thickening, valve thrombosis, and SVD.
THV degeneration appears to occur in a stepwise fashion with thrombus followed by fibrosis after 60 days and calcification after 4 years, with a trend toward increased leaflet thickness throughout implantation. Importantly, thrombus seems to form overlying dysfunctional ECs, whereas thrombus and fibrosis may be substrates for factors that may drive SVD including calcification and MMP-1 expression. Thus, addressing EC dysfunction, initial thrombus formation, and fibrosis to further understand mechanisms of SVD appears to be key next steps to improve long-term THV durability.
COMPETENCY IN MEDICAL KNOWLEDGE: Increasing interest in surgical and transcatheter valve thrombus has been driven by largely incidental imaging findings on CT. Our histopathological analysis provides important insight into the pattern of valve thrombus, fibrosis and calcification which will help inform future imaging studies aimed at defining the potential link between valve thrombus and SVD. Our findings also highlight a potential limitation of imaging to discriminate thrombus and fibrosis on valves during intermediate term follow-up.
TRANSLATIONAL OUTLOOK: Post-implantation CT imaging has brought the concept of clinically silent bioprosthetic valve thrombus to the field of structural heart disease. The present histopathological analysis highlights the complexity but apparent consistent pattern of valve degeneration. This helps to better inform the imaging community of both the challenges of post-implant CT and the opportunities through continued thoughtful investigation and provides the basis for further multidisciplinary investigation of mechanisms of THVs dysfunction.
Supported by Edwards Lifesciences and unrestricted grants from the Silber and Belzberg Family Foundations. Dr. Leipsic is supported by the Canadian Research Chair in Advanced CardioPulmonary Imaging; and consults for Edwards Lifesciences, HeartFlow, and Circle CVI; and holds stock options in HeartFlow and Circle CVI. Dr. Pibarot is supported by the Canadian Research Chair in Valvular Heart Disease; and provides uncompensated core laboratory services for Edwards Lifesciences and Medtronic. Dr. Newby is supported by British Heart Foundation (BHF) grants CH/09/002, RE/13/3/30183, and RM/13/2/30158; and is a recipient of Wellcome Trust Senior Investigator Award WT103782AIA. Dr. Dweck is supported by BHF grant FS/14/78/31020; and is a recipient of Sir Jules Thorn Award for Biomedical Research 2015. Dr. Blanke is a consultant for Edwards Lifesciences, Tendyne, WL Gore, and Circle Cardiovascular Imaging. Drs. Blanke and Leipsic provide uncompensated computed tomography services for Edwards Lifesciences, Medtronic, Neovasc, Aegis, and Tendyne Holdings. Dr. Webb is a consultant for Edwards Lifesciences and Abbott. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cluster of differentiation 31
- cluster of differentiation 45/leukocyte common antigen
- endothelial cell(s)
- energy dispersive X-ray spectroscopy
- hematoxylin and eosin
- matrix metalloproteinase
- platelet and endothelial cell adhesion molecule
- scanning electron microscopy
- second-harmonic generation
- structural valve degeneration
- Received February 13, 2018.
- Revision received May 9, 2018.
- Accepted June 22, 2018.
- 2019 American College of Cardiology Foundation
- Del Trigo M.,
- Munoz-Garcia A.J.,
- Wijeysundera H.C.,
- et al.
- Hansson N.C.,
- Grove E.L.,
- Andersen H.R.,
- et al.
- Leetmaa T.,
- Hansson N.C.,
- Leipsic J.,
- et al.
- Jose J.,
- Sulimov D.S.,
- El-Mawardy M.,
- et al.
- Mangione F.M.,
- Jatene T.,
- Goncalves A.,
- et al.
- Yahagi K.,
- Torii S.,
- Ladich E.,
- et al.
- Yahagi K.,
- Ladich E.,
- Kutys R.,
- et al.
- Monaghan M.G.,
- Kroll S.,
- Brucker S.Y.,
- Schenke-Layland K.
- Capodanno D.,
- Petronio A.S.,
- Prendergast B.,
- et al.
- Foroutan F.,
- Guyatt G.H.,
- Otto C.M.,
- et al.
- Dweck M.R.,
- Jenkins W.S.,
- Vesey A.T.,
- et al.
- Pawade T.A.,
- Cartlidge T.R.,
- Jenkins W.S.,
- et al.