Progressive Reinvention or Destination Lost? Half a Century of Cardiovascular Tissue Engineering

Abstract

The concept of tissue engineering evolved long before the phrase was forged, driven by the thromboembolic complications associated with the early total artificial heart programs of the 1960s. Yet more than half a century of dedicated research has not fulfilled the promise of successful broad clinical implementation. A historical account outlines reasons for this scientific impasse. For one, there was a disconnect between distinct eras each characterized by different clinical needs and different advocates. Initiated by the pioneers of cardiac surgery attempting to create neointimas on total artificial hearts, tissue engineering became fashionable when vascular surgeons pursued the endothelialisation of vascular grafts in the late 1970s. A decade later, it were cardiac surgeons again who strived to improve the longevity of tissue heart valves, and lastly, cardiologists entered the fray pursuing myocardial regeneration. Each of these disciplines and eras started with immense enthusiasm but were only remotely aware of the preceding efforts. Over the decades, the growing complexity of cellular and molecular biology as well as polymer sciences have led to surgeons gradually being replaced by scientists as the champions of tissue engineering. Together with a widening chasm between clinical purpose, human pathobiology and laboratory-based solutions, clinical implementation increasingly faded away as the singular endpoint of all strategies. Moreover, a loss of insight into the healing of cardiovascular prostheses in humans resulted in the acceptance of misleading animal models compromising the translation from laboratory to clinical reality. This was most evident in vascular graft healing, where the two main impediments to the in-situ generation of functional tissue in humans remained unheeded-the trans-anastomotic outgrowth stoppage of endothelium and the build-up of an impenetrable surface thrombus. To overcome this dead-lock, research focus needs to shift from a biologically possible tissue regeneration response to one that is feasible at the intended site and in the intended host environment of patients. Equipped with an impressive toolbox of modern biomaterials and deep insight into cues for facilitated healing, reconnecting to the "user needs" of patients would bring one of the most exciting concepts of cardiovascular medicine closer to clinical reality.

Keywords: cardiovascular tissue engineering; clinical needs; history; misleading animal models; refocusing translation.

Figure 1

Schematic presentation of the fundamental difference in cellularization and healing of prosthetic cardiovascular implants in humans (A) and in animal models (B). In humans, transanastomotic outgrowth hardly exceeds a few millimeters even after years of implantation. Continual fibrinogen and platelet replenishment from the blood leads to a compacted surface thrombus in the luminal interstices of the scaffold that increasingly becomes hostile toward capillary penetration similar to the wall thrombus of aneurysms. Over time, this compacted acellular material in the luminal layers of a scaffold becomes prohibitive for transmural endothelialisation (Insert A) even if scaffold structure and/or degradability would facilitate capillary penetration. The rapid trans-anastomotic outgrowth of adjacent endothelium and its subintimal cells in the vast majority of animal models (B) also mitigates transmural vascularization (Insert B) while actively recruiting cells from the circulation. As such, the entire healing pattern in most animal models from surface endothelium to intramural cell population is non-predictive for the tissue response in patients. For transmural endothelialisation to be successful (C), models need to be chosen where the presence of a surface endothelium is not pre-empted. Only this allows to study the antagonistic dynamics between ingrowth spaces, accelerated angiogenesis and the build-up of increasingly impenetrable, compacted thrombus in the luminal interstices of a scaffold.

Figure 2

Schematic comparison of graft lengths typically used for clinical peripheral bypass grafts (A) and experimental grafts implanted in animals (B). In >90% of all large animal experiments grafts were shorter than 6 cm. In most animal models, trans-anastomotic endothelial outgrowth (red) leads to complete surface endothelialisation within weeks making it impossible to study transmural or blood-born endothelialization. In clinically implanted bypass grafts, in contrast, ingrowth stoppage permanently leaves over 90% of the graft non-endothelialized (blue). Therefore, animal studies generally investigate a mode of surface endothelialization that is irrelevant in humans. To study transmural- or fall-out endothelialisation from the blood stream experimental grafts need to be welded between sufficiently long low-porosity grafts to clearly see a non-endothelialized zone between the progressing margins of transanastomotic outgrowth and endothelium originating from the investigated mid segment (C). From (102) with permission.

Figure 3

Typical straight infra-renal interposition graft in rats usually ± 10 mm in length (A) and loop graft of ± 100 mm length with an experimental segment welded into the mid-region of low-porosity ePTFE (B). While straight infrarenal interpositions usually lead to.trans-anastomotic endothelialisation before trans-mural ingrowth can occur the “isolation” segments in the loop graft enable the investigation of transmural endothelialisation without interference from transanastomotic endothelialisation. By also sealing the interposition segment against the adventitia, a model for fall-out healing is created. From (110) with permission.

Figure 4

Interposition isolation models for studying endothelialisation without the interference of trans-anastomotic endothelial out- and over-growth. (A) Rat infra-renal loop graft model. A high porosity polyurethane graft was welded into the mid-segment of an up to 100 mm long low-density ePTFE graft. The confluent mid-graft endothelium reached onto the otherwise endothelial-free ePTFE sections. Transmural ingrowth from the adventitial side was confirmed with corrosion casting. The origin of the surface endothelium was sometimes traceable to capillary openings on the blood surface. (B) Senescent baboon femoro-femoral isolation graft model. Experimental grafts were equally welded into the mid-section of low-porosity ePTFE. The experimental ePTFE graft shown possessed a dense middle layer which made it impenetrable for cells. After 6 weeks, cellularity was almost exclusively on the adventitial side, highlighting that trans-mural ingrowth from the adventitia overwhelmingly accounts for the cell population of the graft wall unless actively recruited by transanastomotic endothelium (see Figure 6). From (113) and (102) with permission.

Figure 5

Transmural endothelialisation of vascular grafts in the rat (A) and the senescent baboon (B,C). The loop graft in the rat (A) shows clearly discernible edges of the trans-anastomotically outgrowing endothelium (post-colored in red) from the infra-renal aorta with a long stretch of endothelium-free surface separating it from the trans-mural mid-graft endothelium (also post-colored in red). The presence of a long separating zone between the experimental graft and the transanastomotic outgrowth-edge is even more important in view of transanastomotic outgrowth also occurring in the opposite direction (Insert A) from the experimental graft in case of successful trans-mural endothelialisation. (B,C) Similar isolation graft in femoro-femoral position in the senescent Chacma baboon. After 6 weeks, transmural sprouting had either successfully led to confluent surface endothelialisation (C) with multiple capillary openings (Insert C) or in its absence led to the build-up of a dense fibrin matrix in the interstices near and on the surface (B). Sometimes sporadic small endothelial islets are detectable (Insert B).

Figure 6

Femoro-femoral isolation model in the Chacma Baboon (6 weeks) (102, 126). The wrapped, ingrowth-preventive 30 μm ePTFE segements [(A) near anastomosis with trans-anastomotic endothelial outgrowth; (B) beyond trans-anastomotic endothelium with compacted surface thrombus)] had a high-porosity 150μm IND experimental ePTFE graft [(C) and Insert] welded into the mid-section. The low-porosity, wrapped isolation segments highlight the need for the presence of an endothelium for the active recruitment of largely mononuclear cells from the blood stream (A,B). The very-high porosity isolated mid-graft segment showed either dense, compacted thrombus in the internodal spaces of the luminal side (C) or well-healed grafts with fully trans-murally endothelialized blood surfaces (insert C).

Figure 7

Demonstration of the detrimental effect of the compacted, dense surface thrombus on transmural endothelialisation in the baboon femoro-femoral isolation model (identical implant periods in A to C). (A) Restriction of transmural vessel ingrowth to the outer half of the graft in the presence of a compacted surface thrombus in the interstitial spaces of the luminal third of the graft wall. (B) Identical mid-graft with a sealing, oxygen-permissible silicon membrane on the blood surface. The transmural vessel ingrowth reaches through the entire wall thickness. (C) Identical graft as in A but occluded. The long-distance outgrowth of trans-mural blood vessels through the graft wall and the entire thrombus highlights the fact that fibrin clots are pro-angiogenic unless they become compacted near the blood surface as typically seen in non-endothelialised vascular grafts in humans.

Figure 8

In myocardial regeneration, the potential of a hydrogel (polyethylene glycol) to preserve space and entrap cells is demonstrated with fluorescent labeled PEG (Alexa 665 nm) seen polymerised between the cardiomyocyte bundles in an infarcted rat heart (A). The advantageous effect on stress reduction through wall remodeling was much more pronounced if the gel injection was delayed. Inset (B) shows a similarly labeled PEG hydrogel entrapping adipose derived mesenchymal stem cells (green) within the infarcted wall of a rat heart. To improve cellular retention in myocardial regeneration therapy, cellular self-assembly into 3D microtissues (3D-MTs) using the “hanging drop” method (178, 179) prior to intra-myocardial injection (C) and compact 100 μm thick myocardial muscle bundles grown from human induces pluripotent stem cells (hiPSC) (176) or human embryonic stem cells (hESC) (177) have emerged as an encouraging alternative to single cell injection “therapy” with its high cell loss due to a lack of entrapment. 3D-microtissues have been shown to significantly enhance the angiogenic activity and neovascularization potential of stem cells. From (180) and (179) with permission.

Figure 9

Concomitant stress reduction through gel injection in post infarction myocardial regeneration. The preservation of wall thickness in infarcted rat hearts after injection of polyethylene glycol hydrogels is clearly visible at 4 weeks (B) and 13 weeks (D) relative to untreated controls (A,C). Finite element models (185, 186) have shown that this mitigation of detrimental post-infarction remodeling dramatically reduces the ventricular mechanical stress (187) that drives the infarcted heart toward failure. Green (scarring) and purple (viable myocardium). From (183) with permission.

Figure 10

(A) Scanning electron micrograph of the midsegment of a 4 mm ePTFE graft, in-vitro enothelilised with mass-cultured, allogenic, multidonor endothelial cells after 16 days of implantation as femoral graft in a senescent Chacma Baboon. Only residual cell islands are left of an originally confluent endothelium at the time of implantation interspersed with denuded areas with densely adherent leukocytes and platelets. (B) Confluent monolayer of autologous endothelial cells 4 weeks after implantation of an in vitro endothelialised 4-mm ePTFE femoro-femoral graft into a senescent baboon. One can still recognize the underlying structure of the PTFE graft. No endothelial cell detachment was found in spite of the shear stress exposure. The antithrombogenic potential of the cultured endothelial cells was reflected by a higher patency rate and the lack of platelet or fibrin depositions. With permission from (208) and (209).

Figure 11

(A) Midgraft segment of an autologous, in-vitro endothelialized graft 41 months after implantation showing a confluent endothelium (CD 31) resting on layers of well-aligned actin-positive cells. A delicate intima was demarcated from the α-SMC actin positive cells by a well-defined internal elastic membrane (Insert: Orcein). (B) Primary patency (y-axis) over time (x-axis in years) highlighting the clinical benefit of autologous in-vitro endothelialisation in 6 and 7 mm femoropopliteal bypass grafts of 341 consecutive patients opposite a comparable patient group randomized to receive saphenous vein (“SV”) and ePTFE grafts (“ePTFE”) [with permission (37)]. The entire cohort of patients receiving an in-vitro endothelialised grafts had no saphenous vein available and as such, an ePTFE graft was an obligatory choice for each of them. Therefore, the patency of endothelialised grafts needs to be compared with that of the subgroup in the randomized Veith et al. study where an ePTFE prostheses was equally obligatory (“ePTFE ^w/_o SV”). From (80) with permission.

Figure 12

Midgraft segments of two in-vitro endothelialised femoro-popliteal grafts explanted at the time of re-operation for graft failure 41 months (A,C) and 63 months (B) after implantation. Both specimens contained other areas of more significant stenoses but the displayed pre-stenotic regions were packed with large islands of foam cells. Typically, the foam cells were wedged underneath pannus-like, cell-poor tissue that occasionally showed stretches of complete acellularity (B). From (80) with permission.

Figure 13

Radial Laser holes creating ingrowth permissible spaces in vascular grafts. (A) Laser holes in polyurethane grafts and (B) a decellularized arterial allograft. The clear-cut edges without tissue trauma are typical for very short-waved lasers (330). Trans-mural tissue ingrowth from the adventitia after 3 months of carotid interposition grafting in sheep. In spite of dense fully transmural tissue ingrowth through the laser perforation the decellularized allograft tissue surrounding the laser hole remained acellular (B1). Vimentin positive fibroblasts were restricted to the laser channel (B2). Within the laser hole, neovascularization was restricted to the outer half as capillary sprouting did not reach the graft lumen (von Willebrand staining) (B3). From (331, 332) and (329) with permission.

Figure 14

Transmural endothelialisation model in the absence of interference by trans-anastomotic endothelialisation or surface thrombus compaction. Rat subcutaneous implantation of low-porosity ePTFE-lined ingrowth permissible constructs connected to an implantable osmotic mini pump allowing well-defined administration of pro-angiogenic cues. Transmural neovascularisation is image-analytically assessed on histology and by high resolution micro-CT. From (361) with permission.

Figure 15

Leaflets of two different electrospun heart valves made of the biodegradable supra-molecular polymer bisurea-modified poly-carbonate (PC-BU) after 24 weeks of trans-catheter pulmonary valve replacement in the sheep. The leaflet cellularity generally decreased with distance to the leaflet base with minimal tissue deposition observed toward the free edges. Leaflet remodeling depended on stent integration in the surrounding tissue with transmural population of the leaflet base with polymer absorption and ECM deposition in the well-integrated valve (A) and poor cell population, surface overgrowth of tissue and a lack of collagen deposition in valve (B), which was in line with its poor stent integration and migration over time indicating the dependence of the cell population of the leaflet scaffold on transmural ingrowth. From (158) with permission.