Leading Opinion On Biomaterial Mechanisms Of Biocompatibility (PDF)

Biomaterials 29 (2008) 2941–2953 Contents lists available at ScienceDirect Biomaterials journal homepage: www.elsevie...

Biomaterials 29 (2008) 2941–2953 Contents lists available at ScienceDirect Biomaterials journal homepage: www.elsevier.com/locate/biomaterials Leading Opinion On the mechanisms of biocompatibilityq David F. Williams a, b, c, d, * a Emeritus Professor, University of Liverpool, UK b Visiting Professor, Christiaan Barnard Department of Cardiothoracic Surgery, University of Cape Town, South Africa c Visiting Professorial Fellow, Graduate School of Biomedical Engineering, University of New South Wales, Australia d Guest Professor, Tsinghua University, Beijing, China a r t i c l e i n f o a b s t r a c t Article history: The manner in which a mutually acceptable co-existence of biomaterials and tissues is developed and Received 17 March 2008 sustained has been the focus of attention in biomaterials science for many years, and forms the foun- Accepted 11 April 2008 dation of the subject of biocompatibility. There are many ways in which materials and tissues can be Available online 28 April 2008 brought into contact such that this co-existence may be compromised, and the search for biomaterials that are able to provide for the best performance in devices has been based upon the understanding of Keywords: all the interactions within biocompatibility phenomena. Our understanding of the mechanisms of bio- Foreign body response compatibility has been restricted whilst the focus of attention has been long-term implantable devices. Inflammation Implant In this paper, over 50 years of experience with such devices is analysed and it is shown that, in the vast Scaffold majority of circumstances, the sole requirement for biocompatibility in a medical device intended for Biodegradation long-term contact with the tissues of the human body is that the material shall do no harm to those tissues, achieved through chemical and biological inertness. Rarely has an attempt to introduce biological activity into a biomaterial been clinically successful in these applications. This essay then turns its attention to the use of biomaterials in tissue engineering, sophisticated cell, drug and gene delivery systems and applications in biotechnology, and shows that here the need for specific and direct interactions between biomaterials and tissue components has become necessary, and with this a new paradigm for biocompatibility has emerged. It is believed that once the need for this change is recog- nised, so our understanding of the mechanisms of biocompatibility will markedly improve. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction and sustained has been of interest to biomaterials scientists and users of medical devices for many years. It has become clear that The single most important factor that distinguishes a bio- there are very many different ways in which materials and tissues material from any other material is its ability to exist in contact can interact such that this co-existence may be compromised, and with tissues of the human body without causing an unacceptable the search for biomaterials that are able to provide for the best degree of harm to that body. The manner in which the mutually performance in devices has been based upon the acquisition acceptable co-existence of biomaterials and tissues is developed of knowledge and understanding about these interactions. These are usually discussed in the broad context of the subject of biocompatibility. q Editor’s Note: This Leading Opinion Paper is based upon a series of presentations Biocompatibility is a word that is used extensively within bio- given by the author at the University of Washington Summer Workshop in August materials science, but there still exists a great deal of uncertainty 2003, a keynote paper at the World Biomaterials Conference in Sydney, 2004, the about what it actually means and about the mechanisms that are Gordon Research Conference on Biomaterials, Biocompatibility and Tissue Engi- subsumed within the phenomena that collectively constitute bio- neering, New Hampshire, USA in 2005, the Ratner Symposium in Maui, 2006 and the compatibility. As biomaterials are being used in increasingly di- Founders Award Presentation, Chicago, 2007. It forms the first of a series of essays verse and complex situations, with applications now involving that will be published, in different journals, on the subjects of the principles of biomaterials’ selection. Since the author is Editor-in-Chief of the journal, the paper tissue engineering, invasive sensors, drug delivery and gene has been refereed by four of the Associate Editors and revised on the basis of their transfection systems, the medically oriented nanotechnologies and reports. The opinions expressed in the review are, however, the sole responsibility of biotechnology in general, as well as the longer established im- the author. It should also be noted that the reference list cannot represent the to- plantable medical devices, this uncertainty over the mechanisms tality of literature on biocompatibility but points to some of the more significant literature that reflect the clinical outcomes concerned with biocompatibility. of, and conditions for, biocompatibility is becoming a serious im- * Ave de la Foret 103, Brussels, B-1000, Belgium. pediment to the development of these new techniques. This review E-mail address: [email protected] of biocompatibility attempts to address some of these uncertainties 0142-9612/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2008.04.023 2942 D.F. Williams / Biomaterials 29 (2008) 2941–2953 and provides a proposal for a unified theory of biocompatibility of biocompatibility can be devised. We do so here with reference to mechanisms. the evidence that has accumulated over the last 50 years through experiment and clinical experience. 2. The evolution of current concepts of biocompatibility 3. The agents of biocompatibility Biocompatibility has traditionally been concerned with im- plantable devices that have been intended to remain within an The paradigm of biocompatibility outlined in this paper involves individual for a long time. To those who were developing and using the separate, but potentially interrelated, responses of the two the first generation of implantable devices, during the years be- phases of the biomaterial–tissue complex and the interfacial phe- tween 1940 and 1980, it was becoming increasingly obvious that nomena that come into play when they meet. Probably the most the best performance biologically would be achieved with mate- important underlying principle is that the mechanisms by which rials that were the least reactive chemically. Thus, within metallic materials and human tissues respond to each other are not unique systems the plain carbon and vanadium steels, which demon- to this particular use but are merely variations of natural processes strated overt corrosion, were replaced by increasingly superior that occur within materials and biological sciences. Thus, in gen- stainless steels, then by the strongly passivated cobalt–chromium eral, the response of a material to implantation in the human body alloys, titanium alloys and the platinum group metals. With poly- will not involve totally new mechanisms not found in other envi- mers, the readily available and versatile nylons and polyesters were ronments, and the cellular and humoral responses of the body do replaced by the more degradation resistant PTFE, PMMA, poly- not involve the cellular and extracellular constituents performing ethylene and silicones. Consistent with this approach, the selection in ways which are entirely non-physiological. The key to un- criteria for implantable biomaterials evolved as a list of events that derstand biocompatibility is the determination of which chemical, had to be avoided, most of these originating from those events biochemical, physiological, physical or other mechanisms become associated with the release of some products of corrosion or deg- operative, (and why), under the highly specific conditions associ- radation, or additives to or contaminants of the main constituents ated with contact between biomaterials and the tissues of the body, of the biomaterial, and their subsequent biological activity, either and what are the consequences of these interactions. locally or systemically. Materials were therefore selected, or occa- Before discussing these mechanisms and the various causal re- sionally developed, on the basis that they would be non-toxic, non- lationships in biocompatibility, it is worth noting that there are immunogenic, non-thrombogenic, non-carcinogenic, non-irritant several mediators of the biocompatibility of a material other than and so on, such a list of negatives becoming, by default, the defi- the material itself. Of great significance is the nature and quality of nition of biocompatibility. the clinical intervention that places the material into contact with Three factors initiated a re-evaluation of this position. The first the tissues. For implantable medical devices, the characteristics of was that it became obvious that the response to specific individual the individual in or on whom the device is placed are also of con- materials could vary from one application site to another. Thus siderable importance and it is to be anticipated that wide patient- biocompatibility could not solely be dependent on the material to-patient variability will be seen. Age, sex, general health and characteristics but also had to be defined by the situation in which concurrent disease, physical mobility, lifestyle features and phar- the material is used. Secondly, an increasing number of applications macological status all contribute to this variation. The design of required that the material should specifically react with the tissues the device and the physical relationship between the device and rather than be ignored by them, as required in the case of an inert the body play significant roles , as do the presence or absence of material. Thirdly, and in a similar context, some applications re- micro-organisms and endotoxins. This review concentrates quired that the material should degrade over time in the body on the material-derived processes but these always have to be rather than remain indefinitely. considered in the context of the totality of biocompatibility. It was therefore considered that the very basic edict that bio- In Table 1, the major material characteristics that may conceiv- compatibility, which was equated with biological safety, meant ably influence the host response are listed. These can be divided that the material should do no harm to the patient, was no longer into characteristics of the bulk material and those of the surface. a sufficient pre-requisite. Accordingly, biocompatibility was re- The majority of these characteristics are self-evident although ob- defined in 1987 as follows: viously some subsume a number of features. The elastic constants, for example, include Young’s modulus, shear and bulk moduli and Biocompatibility refers to the ability of a material to perform with Poisson’s ratio. Crystallinity in polymers includes the degree of an appropriate host response in a specific situation. crystallinity and the nature of the molecular symmetry, whilst in This definition, which clearly places the word in the category of metals it includes crystal structure, preferred orientations and a concept rather than a practical descriptor of a process, is based on grain size. the three tenets that a material has to perform and not simply exist in the tissues, that the response which it evokes has to be appro- Table 1 priate for the application, and that the nature of the response to a Major material variables that could influence the host response specific material and its appropriateness may vary from one situ- Bulk material composition, micro- (or nano)-structure, morphology ation to another. Crystallinity and crystallography It could be argued that this definition is so general and so self- Elastic constants Water content, hydrophobic–hydrophilic balance evident that it is not of any real help in advancing knowledge of Macro-, micro-, nano-porosity biocompatibility, and indeed it is true that it has not led to a greater Surface chemical composition, chemical gradients, surface molecular mobility understanding of specific mechanisms and individual processes, or Surface topography to the innovation of new biomaterials. Moreover, it is likely that one Surface energy Surface electrical/electronic properties concept cannot apply to all material–tissue interactions that per- Corrosion parameters, ion release profile, metal ion toxicity (for metallic materials) tain to widely varying applications, ranging from a drug eluting Degradation profile, degradation product form and toxicity (for polymeric materials) stent to a tissue engineering cartilage construct, a joint replacement Leachables, additives, catalysts, contaminants and their toxicity (for polymeric prosthesis or an invasive biosensor. It is with this diversity in mind, materials) and the wide ranging potential mechanisms of interactions based Dissolution/degradation profile, degradation product toxicity (for ceramic materials) Wear debris release profile both on materials science and on biology, that a different paradigm  D.F. Williams / Biomaterials 29 (2008) 2941–2953 2943 Table 2 Table 3 Major characteristics of the generic host response to biomaterials State-of-the-art in materials’ selection for long-term implantable devices Protein adsorption and desorption characteristics Material Applications Generalised cytotoxic effects Titanium alloys Dental implants, femoral stems, pacemaker Neutrophil activation cans, heart valves, fracture plates, spinal cages Macrophage activation, foreign body giant cell production, granulation tissue Cobalt–chromium alloys Bearing surfaces, heart valves, stents, pacemaker leads formation Platinum group alloys Electrodes Fibroblast behaviour and fibrosis Nitinol Shape memory applications Microvascular changes Stainless steel Stents, orthopaedic implants Tissue/organ specific cell responses (e.g. osteoclasts and osteoblasts for bone, Alumina Bearing surfaces endothelial proliferation) Calcium phosphates Bioactive surfaces, bone substitutes Activation of clotting cascade Carbon Heart valves Platelet adhesion, activation, aggregation UHMW polyethylene Bearing surfaces Complement activation PEEK Spinal cages Antibody production, immune cell responses PMMA Bone cement, intraocular lenses Acute hypersensitivity/anaphylaxis Silicones Soft tissue augmentation, insulating leads, Delayed hypersensitivity ophthalmological devices Mutagenic responses, genotoxicity Polyurethane Pacemaker lead insulation Reproductive toxicity Expanded PTFE Vascular grafts, heart valves Tumour formation Polyester textile Vascular grafts, heart valves SIBSa Drug eluting stent coating a Poly(styrene-block-isobutylene-block-styrene). When placed in or on the tissues of the body, a number of re- actions to a material may be seen over time, and these are listed in being aimed at maximising relevant mechanical properties (in- Table 2. Some of these, as examined later, may constitute important cluding fatigue strength, creep strength, toughness and wear re- determinants of the host response, whilst others are of greater sistance), minimising material deterioration (including corrosion, importance in the functioning of the device. Within the host, in the degradation and wear resistance) and facilitating long-term in- majority of circumstances, we may envisage a sequence of events, corporation of the device into the musculoskeletal system (in- potentially involving the interaction between proteins and other volving, for example, either cements or bioactivity). Specifically in physiological macromolecules with the material surface, the initi- relation to biocompatibility, the materials are required to optimise ation of inflammatory and/or immune responses, and then the the rate and quality of bone apposition to them, to minimise the rate repair and/or regeneration processes that may lead to stable equi- of release of corrosion or degradation products and the tissue re- librium between material and host. This is the classical bio- sponse to them, to minimise the rate of wear debris release and the compatibility paradigm that has been discussed in one form or tissue response to this debris, and to optimise the biomechanical another over the last couple of decades. environment in order to minimise disturbance to homeostasis in In determining the actual significance of each potential agent of the bone and surrounding soft tissue. Experience has shown that biocompatibility, and how they are able to control the individual the optimal balance of mechanical properties with metallic com- features of the sequence outlined in this simple paradigm, we may ponents is best achieved with either titanium alloys or cobalt– discuss some of the situations in which materials come into contact chromium based alloys and no generalised biocompatibility with human tissues and consider the evidence concerning the advantage has ever been achieved outside of these alloys or through mediation of biocompatibility in these situations. The purpose of modifications within the specifications of these alloys, including this essay is to develop a more comprehensive paradigm of bio- surface modifications. In spite of many claims to the contrary, there compatibility using the examples of long-term implantable devices, are no specific biocompatibility characteristics that are dependent intentionally degradable implantable systems, intravascularly in- on the alloying elements in titanium alloys, nor, with the exceptions vasive short term medical devices and tissue engineering products. of idiosyncratic hypersensitivity responses, on the precise compo- sition of cobalt–chromium alloys. The only characteristic that con- 4. The long-term implantable medical device trols the host response to these alloys is the rate of metal ion release. When these alloys are placed, un-cemented, in direct contact Recognising that the most trusted data on the biocompatibility with bone, the precise chemical composition within each group of a material must come from the actual use of that material in does not influence the eventual strength of attachment to the bone, practical clinical examples in humans, we shall review first the nor, to any clinically significant extent, the rate of bone apposition. generic evidence concerning some well known clinical procedures, Titanium alloys do give better and faster attachment to bone than taken from a spectrum of conditions involving both hard and soft cobalt–chromium alloys , but the precise surface chemistry does tissues and blood contact. It is noted, of course, that data on many not appear to make any difference. This is in contrast to the devices may not always be definitive with respect to materials since surface texture, where surface roughness and/or porosity does in- more than one material may be involved in a device, and since fluence the response of the bone [9,10]. There is evidence that some other factors, for example biomechanical and haemodynamic fac- metallic elements, particularly nickel, can stimulate the immune tors, may well be as important as the material characteristics, but system. it is relatively straightforward to establish the main features of With respect to polymers, again the critical factor in bio- clinical biocompatibility. All of the examples in this section involve compatibility is the balance of mechanical properties and degra- devices that are intended for long-term performance for the re- dation resistance, and the issues of polyethylene wear debris have placement of damaged or diseased tissues. Table 3 summarises the been widely discussed. The precise mechanisms of particle-induced state-of-the-art in materials’ selection for such devices. osteolysis have perhaps not been fully resolved with respect to the interplay between inflammatory cells and the osteoblast–osteo- 4.1. Total joint replacement prostheses clast relationship, but the overall situation is clear. As wear debris is released, inflammatory cells, most significantly macro- The biomaterial requirements of total joint replacements have phages and giant cells, respond and, through normal cell signalling become clearer over the 40 years since their first introduction, processes, may stimulate osteoclasts to effect a degree of bone 2944 D.F. Williams / Biomaterials 29 (2008) 2941–2953 resorption, leading to loosening of the prosthesis. The chem- each term covers a multitude of specific types. At this time there is istry of the polymer is not relevant to the inflammatory process, little to choose between these with respect to the biocompatibility, and indeed osteolysis and prosthesis loosening have been observed although clinical performance will vary as a function of factors with polymers other than polyethylene , and the only factors of design, clinical technique and patient variables. Both types are the rate of wear debris release and the physical form and of material are very resistant to degradation and leachables are dimensions of the particles. Although there have been many effectively absent. Variations within each class are normally con- attempts to improve the polyethylene, none have been clinically cerned with either chemical structure or surface modifications that relevant other than those which reduce the wear rate, principally control either or both the hydrophobic/hydrophilic nature and through cross-linking and control of sterilisation procedures. It the flexibility, mostly in relation to the clinical techniques and is possible that the host response can be marginally modulated implant functionality. The cataract literature rarely discusses bio- pharmacologically, through the use of bisphosphonates or vitamins compatibility issues in conventional designs and patients, but is for example , most likely to be given systemically. far more concerned with extending the technique, for example Ceramics are involved in joint replacements in two different towards accommodating and multifocal lenses and blue light ab- circumstances. Inert oxide ceramics may be used as bearing sur- sorbing lenses to protect against age-related macular degeneration, faces, where their hardness improves wear resistance and therefore and to the use in more difficult patient groups, including paediatric minimises osteolysis, with an increasing use of alumina–alumina cases, diabetics and those with other eye conditions. Although, combinations. There are no other biocompatibility consider- in theory, the surface characteristics of the lenses should affect the ations with ceramic bearing surfaces other than minimising deg- local tissue response, including protein adsorption, epithelial cell radation or wear. On the other hand, bioactive ceramics and glasses, overgrowth, inflammation and so on, in the vast majority of situ- principally hydroxyapatite, various calcium phosphates and bio- ations this is of no clinical relevance. At one time it was considered active glasses, are used as coatings to improve bone bonding important to control inflammation by surface modifying with [20–22]. More will be said in a later section about the use of such heparin, for example, but no significant improvements are now materials in tissue engineering and drug delivery situations, but found and in fact the heparin modified surfaces may be associated their role in long-term implantable devices is rather restricted. with greater levels of posterior capsule opacification. It may There is evidence of some clinical utility as coatings on un- well be that in certain patients with either localised (e.g. uveitis) or cemented joint prostheses, especially with respect to the kinetics of systemic (diabetic) conditions, some lens material features will be bone adaptation, where the performance is based on the balance relevant, for example the flexibility which determines the size of the between the facilitation of bone formation and the resorption or incision through which the lens is inserted, but this is not generally stability of the materials [23,24]. the case. Indeed it seems likely that the performance of intraocular As a result of this analysis based upon clinical experience and lenses has very little to do with the precise nature of the material and recent experimental studies, the only materials significantly used much more to do with clinical technique, including inflammation in joint prostheses are titanium and cobalt–chromium alloys, cross- and infection control, and the general state of health of the patient. linked UHMWHD polyethylene, alumina, PMMA for cement and Although far less advanced, the area of corneal replacement is hydroxyapatite for a bioactive coating. The only criteria for bio- somewhat similar, where biostable transparent polymers, including compatibility are the need to minimise the release of any degra- hydrogels such as poly(hydroxyethylmethacrylate), appear to give dation or wear particles or corrosion products, and to maximise the good performance, in this case the main challenges being concerned rate and efficiency of bone adaptation. With the possible exception with prosthesis anchorage. of improving bone bonding through bioactivity, where the mag- nitude of the effect is probably only marginal, there is no benefit 4.3. Devices for cardiac rhythm management from seeking characteristics of the materials other than they are as inert as possible. There are many examples of attempts to increase Cardiac pacemakers and implantable cardioverter defibrillators performance through materials ‘improvements’ but virtually all (ICDs) provide an exceptionally successful technology platform for have actually led to a diminution of performance, including porous the management of a wide variety of cardiac rhythm disorders. metal backed acetabular components , alternative polymers to They are complex implantable devices, but performance is largely polyethylene , alternatives to alumina such as transformation independent of the nature of the biomaterials used in their con- toughened zirconia for bearing surfaces, and alternative forms struction now that basic lessons have been learnt. Almost univer- of acrylic cements. Provided the choice of materials is confined sally, the active components are encapsulated in a hermetically to this very narrow range, the materials themselves play very little sealed titanium can, whose biocompatibility characteristics are role in determining the outcome of the procedures, where the most controlled by the corrosion resistance of this metal. Leads transmit important determinants of performance are surgical and nursing pulses for both sensing and delivery purposes from the can to the skills, patient compliance and infection control. electrode placed at the relevant site on the heart. The functional properties specified for these components are clearly related to the 4.2. Intraocular lenses and other ophthalmological devices electrical performance and most systems use one of a small number of high electrical conductivity, fatigue resistant and corrosion re- Within the eye there are several devices with a reasonably long sistant alloys such as those of the platinum group metal alloys or record of use, including intraocular lenses and artificial corneas, or the cobalt–chromium group such as Elgiloy for the leads and keratoprostheses, where functionality is largely determined by the electrodes , with an insulating sleeve covering the lead. optical properties, the physical compatibility with the relevant Well known problems with the cracking of polyurethane insulating tissue and the insertion into and retention within the desired leads, including stress cracking and metal ion-induced oxidation location. Biocompatibility is determined by the need to minimise , have largely been solved and polyurethane and silicones are the extent of the tissue reaction in order to avoid compromising the now standard. As far as the electrode is concerned, the main re- light transmission. Intraocular lenses are remarkably successful quirement is concerned with the delivery of the electrical impulse implantable devices, primarily used in patients with cataracts without the inducement of excessive fibrosis, which could raise the following removal of the affected lens. Two broad classes of threshold stimulus to clinically unacceptable levels. This is an im- material have been used, silicones and acrylics, although the de- portant biocompatibility issue, since the function of the electrode is scriptive literature on these materials is usually very general and to deliver impulses that involve the transfer of ions across the  D.F. Williams / Biomaterials 29 (2008) 2941–2953 2945 interface and, by definition, this function must be capable of influ- distribution of various silicone components, including oligomers, encing the tissue response. There is no clear resolution to this, with fillers and catalysts, and a resulting cascade of clinical consequences parallel attempts to minimise the ion transfer through coatings such including the development of autoimmune conditions , in- as iridium oxide or titanium nitride and to minimise the cluding scleroderma, lupus and rheumatoid arthritis, peripheral host response through the delivery of an anti-inflammatory agent neuropathy and a wide variety of degenerative diseases, and to the tissue. The situation is, therefore, that inertness is a also self-reported conditions without specific diagnosis. There principle specification for the biomaterials used in these implant- can be no doubt that the vast majority of claims about the detri- able electronic devices, with the caveat that the functional charac- mental affects of silicone entities have proven unfounded, following teristics of the leads, insulators and electrodes may impact on the both epidemiological, clinical and experimental studies. Al- host response and it may be beneficial to intervene in this response though there are still some studies showing a strong T-cell immune in order to optimise the clinical outcomes. The vast majority of response associated with the surrounding tissue in some patients pacemakers do not employ either mechanism and it may well be , the widespread claims of major immune responses to these that the ICDs, which are of more recent origin and now increasing implants have been discredited and there are no significant lessons in popularity, would benefit more. It is also noticeable that for biocompatibility from these cases. Indeed, the converse may these rhythm control devices are being used in more cases in chil- well be true since the precautionary regulatory disapproval of sili- dren and there appear to be more persuasive arguments that this cone breast implants prompted some attempts to use alternative intervention in biocompatibility should be considered in these fillers for implants, including cellulosic and lipid derivatives paediatric cases. where limited clinical studies confirmed the rather delicate nature of the breast implant scenario. The use of a soya-bean lipid 4.4. Soft tissue reconstruction and augmentation gel within a silicone elastomer has in fact revealed a sequence of biocompatibility disasters, including the potential mutagenic The use of implantable devices in soft tissue replacement and properties of the molecules resulting from the peroxidation of the augmentation has a long and varied history, and this has generated lipid, and the degradation of the silicone produced by prolonged a series of biocompatibility controversies. The story of breast im- contact with the lipid. plants is far too long and complex to discuss in any detail here, but the lessons are profound. A brief summary of the issues is as 4.5. Heart valves follows. Implants are used for the replacement of breast tissue following mastectomy or for the augmentation, or enlargement or The replacement of diseased heart valves has been possible for re-shaping of the female breast. The functional requirements are well over 40 years. Taking first the mechanical heart valves, the that the materials and device designs should allow the replication materials have changed very little in this time, and neither has the of the physical characteristics of breast tissue, which is composed of rationale for their selection. The vast majority of valves used during fatty and glandular tissue, with long-term maintenance of shape this time have had an annulus/frame made of an alloy, usually and volume. This is not a trivial specification since no homogeneous titanium alloy or cobalt–chromium, the annulus being covered by solid synthetic material has these characteristics. The vast majority a fabric sewing ring, almost exclusively of either expanded PTFE or of breast implants follow the concept of using an elastomeric shell polyethylene terephthalate textile, with an occluder or leaflets that contains a gel, the latter to give the required consistency and made of a carbon substrate with a pyrolytic carbon surface. The the former to encapsulate the gel and give size and shape. For many alloys have been chosen to minimise corrosion whilst maintaining years, the optimal functional performance was provided by a com- adequate mechanical properties. A few well known valve failures bination of a silicone gel and a silicone elastomer shell. have been attributed to issues with the mechanical properties of The first problem to arise with these silicone breast implant was these alloy components, including fatigue , but none to corro- the development, in a significant number of patients, of an exces- sion and metal ion release problems. The choice of sewing ring sively thick fibrous capsule, the so-called constrictive fibrosis or material has largely been based on the ease of suturing to the capsular contraction, which was highly clinically significant in many cardiac muscle coupled with minimal degradation and minimal patients because of the distortion and pain associated with the interference with the healing response and there has been no contraction. In spite of many attempts to do so, there has been no perceived benefit from deviating from this choice. The selection of correlation between the fibrosis and any chemical or toxicological pyrolytic carbon for the critical occluder component is based on the feature of the silicone materials, or the silica filler that is present in need to minimise the tendency to initiate blood clots, coupled with the elastomer. On the other hand, there is good evidence to mechanical robustness. This is an immensely important aspect suggest that this fibrosis is due to mechanical irritation and cellular since the inherent haemodynamic characteristics of a mechanical stimulation associated with the micro-movement of the implant– valve imply that blood clots are very likely but it is extremely tissue interface, bearing in mind that these devices are amongst the improbable that any different materials would make this situation largest of all implants, and breast tissue is naturally subjected to any better. With all mechanical valves, this issue is addressed by significant movement. systemic anti-coagulation. Exactly as with joint replacement The desire to produce implants with more natural consistency prostheses, provided the choice of materials is confined to a very led to devices with much thinner elastomer shells and, without any narrow range, the materials themselves play very little role in doubt, these did experience a finite incidence of mechanical failure, determining the outcome of the heart valve replacement, where the so-called implant rupture. In spite of claims to the contrary, the most important determinants of performance are surgical and this has not been the result of ageing or degradation of the silicone nursing skills, patient compliance with their anti-coagulation, elastomer, which would be a significant and important bio- other concurrent cardiovascular disease, and infection control. compatibility factor, but is purely mechanical. On the other With the latter situation, endocarditis, which probably affects 1% of hand, there is no doubt that some silicone components could dif- patients is often fatal, and is preferably controlled by prophylactic fuse through the silicone elastomer shell, the so-called gel bleed, systemic antibiotics. The one situation in which localised and there have been significant questions about how much diffu- antibacterial activity was attempted, using silver, did not prove sion takes place, where the diffusing molecules go and what are clinically effective. the consequences. The controversy has generated criticisms of sil- The possibility of thromboembolic events arising from me- icones, which have involved claims of widespread release and chanical valves has led to an increasing use of bioprosthetic valves 2946 D.F. Williams / Biomaterials 29 (2008) 2941–2953 over the last 20 years. These have, in general, fallen into three inflammatory responses, calcification and structural deterioration. groups, the porcine xenograft, the bovine pericardial valve, and the The situation is obviously different to that with mechanical valves, human aortic valve allograft, also known as a homograft. The ho- but the underlying principles with respect to biocompatibility are mograft has very good performance but with problems of limited just the same. If it were not for the advantage with respect to supply and the possibility of transmission of bovine spongiform haemocompatibility, it is unlikely that the biocompatibility of encephalopathy has essentially eliminated bovine derived bio- xenograft valves would be considered sufficiently good to justify prostheses, so we may confine our comments to the porcine xeno- their use. grafts. These valves, being of natural design and construction, give good haemodynamic performance. As far as their biocompatibility 4.6. Intravascular stents is concerned they do have an advantage over the mechanical valves since they are not intrinsically thrombogenic and do not necessitate The situation with intravascular stents represents a powerful anti-coagulation therapy. They do, however, offer two challenges, reminder of the fragility of the material–tissue interface. Balloon based on the susceptibility of the collagen, upon which they are angioplasty, introduced a couple of decades ago, was a remarkably largely based, to denature and degrade, and to calcify, and also successful addition to the methods available for the treatment of possibly stimulate an immunological response or transmit infection atherosclerotic occlusion of arteries, especially of the coronary. The early, and indeed still the most popular, type of porcine arteries. However, the mechanical interference with the en- bioprosthesis has addressed all of these issues through one simple dothelium during this procedure led, in many patients, to the re- procedure, that of a chemical pre-treatment with the cross-linking currence of the stenosis as the endothelium and smooth muscle agent glutaraldehyde which simultaneously minimises infection cells reacted to this transient injury. The answer to this dilemma risk, reduces the immunogenicity and enhances resistance to en- was the intravascular stent, in which an expandable tubular stent zymatic and chemical degradation of the collagen. However, was deployed within the lumen of the vessel, thereby physically several concerns arise from this process, the first being that the holding the vessel open. Because of the need for the appro- glutaraldehyde, and several of its derivatives, are both leachable and priate mechanical characteristics compatible with stent de- cytotoxic, and a great deal of effort has been spent in attempts ployment, the stents have typically been made of stainless steel to develop less cytotoxic and more effective cross-linking agents. , a shape memory nickel–titanium alloy (such as Nitinol) , or This is a clear example where methods designed to counter one a cobalt–chromium alloy. These bare-metal stents have served biocompatibility deficiency often introduce other deficiencies. The well, but not well enough in most cases, as in-stent re-stenosis often second is that the already noted susceptibility to calcification is eventually appeared. The evidence would suggest that choos- potentially enhanced by the glutaraldehyde procedure, the mech- ing alloys that were as corrosion resistant as possible has been in- anisms of which have been discussed many times and still not fully sufficient here to guarantee the desirable level of biocompatibility, elucidated. It is now being increasingly accepted that in- since the irritation to the adjacent tissues has been physically rather sufficiently masked immune responses and related inflammation than chemically induced, but this does not negate the general significantly affect susceptibility to calcification and degradation principles of materials’ selection noted before.. The result is that calcification is considered an inevitability Two additional features of stents add to the complexity, but with porcine xenografts and although there are variations in the paradoxically underline these principles. The first concerns the use process, for example through the use of an ethanol pre-treatment of drug eluting stents, where, typically, the metal stent is coated and other anti-calcification agents, and indeed variations in the with a thin layer of polymer which incorporates an appropriate susceptibility of patients, it is generally agreed that 20 years is drug, the release of which causes the down-regulation of the cell a maximum that should be expected. proliferation processes that cause the re-stenosis. Not surprisingly, This has led to some profound re-evaluation of the use of so- this procedure has not been without controversy for, although the called natural tissues as materials for long-term implantation, clinical trials published thus far indicate a marked improvement especially in the context of biocompatibility and the rationale for in the patency of stents over time, there have been issues over the regeneration instead of replacement. The need for better perfor- raised levels of thrombosis, both in the early stages post-operative mance from the bioprosthetic valves has led to the introduction and later, potentially equating with the time point when all of of alternative tissue treatments, and especially those that are aimed the available drug has been eluted or related to the post-operative at specifically removing all of the cellular remnants that are con- pharmacological regime. The second feature concerns the sidered to promote calcification and immune responses, usually desire to eliminate the biocompatibility problems of stents by through processes of osmotic or enzymatic de-cellularisation. making them biodegradable. There have been some attempts to use Although the results may be variable, with some groups reporting biodegradable polymers and also a small group of so-called success, it is widely recognised that this process can have devas- bioresorbable magnesium alloys. The latter stents can be tating consequences. It was originally postulated that the decellu- designed to corrode over a time scale ranging from a few weeks to larised tissue would become repopulated with host valvular cells months. Again not surprisingly here the evidence is equivocal since. This, however, does not necessarily occur, as shown with a there is undoubtedly a greater tissue response to the corroding series of fatalities in paediatric cases with congenital valvular metal, leading to initial inflammation and hyperplasia, but then malformations in which there were severe foreign body reactions, with claims that the endothelium remodels once the metal has involving neutrophils and macrophages and, later, lymphocytes. been resorbed. Magnesium is a good choice in the sense that its There was no repopulation with the required fibroblasts and toxicity is minimal, but the degradation process inevitably leads to myofibroblasts but instead the formation of fibrous hyperplasia the release of particulate products, which are intrinsically irritant with calcific deposits, which led to both stenosis and rupture/dis- and pro-inflammatory. A recent review summarises the uses of integration. The biocompatibility implications here are clearly biomaterials in stents very well. that the collagenous structure of the porcine aortic valve material is reactive with and responsive to the human physiological environ- 4.7. Vascular grafts ment and must ultimately fail because of the ensuing interactions, most notably leading to loss of structural integrity and calcification. Vascular grafts have been in clinical use for well over four de- Attempts to ameliorate the long-term problems have led to greater cades. The current position is that although synthetic grafts, usually variability but generally with even less control of the immune and made from an expanded polytetrafluoroethylene or a polyethylene  D.F. Williams / Biomaterials 29 (2008) 2941–2953 2947 terephthalate based textile, are available, there is a limitation on drawn attention to the ability of vascular grafts to stimulate chronic the situations in which they may be used, and autologous vein inflammation and shown an inverse relationship between this grafts, such as the saphenous vein, are either mandated, such as in ability and neovascularisation within the grafts, speculating that coronary artery bypass, or often preferred, as in femoro-popliteal this is an important factor in intimal hyperplasia; certainly the bypass. The reasons for this situation are based on the ability of presence of macrophages in the connective tissue within the graft such grafts to remain patent, with appropriate haemodynamic wall can be expected to stimulate hyperplasia through their release performance, for a clinically acceptable time. This is clearly related of proliferation-enhancing cytokines. Although there were differ- to the fundamental biocompatibility characteristics of the graft, ences between different types of commercial graft, these were not although the specific effect of the graft material per se on this uniformly correlated with the graft material, but more related to biocompatibility is far from obvious. The two principal material the microarchitecture of the grafts. These views on the importance structures mentioned above appear to behave in a similar manner of inflammation compared to thrombosis are consistent with the. Soon after implantation there will be thrombus formation indications that inflammation is the key driver of re-stenosis after within the pores of the graft and the initiation of an exudative in- angioplasty in patients with peripheral artery disease, the bio- flammatory response, with a cellular infiltration, of local origin, on chemical markers of coagulation showing no correlation with ste- the outer surface of the prosthesis, and of haematogenous origin on nosis. the luminal aspect. This is followed by a reparative–proliferative Polytetrafluoroethylene and polyethylene terephthalate are not phase involving fibroblasts, with connective tissue, primarily col- normally used in other equivalent implantable situations and it is lagen, forming within the vessel wall. There is also a continued difficult to compare their general performances, but, should the maintenance of a macrophage/foreign body giant cell response. outcomes with respect to vascular grafts be dependent on their pro- Usually there is an increasing proliferation of an inner mesenchy- inflammatory nature, it is not surprising that these should be es- mal lining to the vessel without any significant endothelialisation sentially similar, at least during the short and medium term. There is and it is this so-called intimal hyperplasia that is the cause of the no reason to believe that these materials would intrinsically exert loss of patency eventually seen with these devices. The region of different stimuli to inflammation and hyperplasia in the vascular hyperplasia is composed of around 20% vascular smooth muscle graft situation during the normal time scale of clinical performance. cells that migrate through the vessel wall and which deposit an It is interesting to note that polyethylene terephthalate is ultimately extracellular matrix, and a number of inflammatory cells, including biodegradable, through either hydrolysis or enzymatic attack, as macrophages and lymphocytes. shown many years ago by the present author , consistent with In the majority of situations, the clinical outcomes with respect increasing case reports of Dacron graft degradation and breakdown, to synthetic vascular grafts are inferior to those obtained with vein often after 15 or more years (for example see Refs. [85,86]). It is also grafts. A recent systematic review of reports on clinical data con- relevant that no other material has been successfully introduced cerning above-knee femoro-popliteal bypass concluded the into vascular grafts during the last several decades. There has been five year patency rate for saphenous vein grafts was 74% but only much discussion about the potential for polyurethanes but 39% for ePTFE bypass grafts. This clearly indicates that the bio- their inherent susceptibility to biodegradation has been a limita- compatibility of the synthetic grafts is less than optimal. The tion. It has to be said that a degradable vascular prosthesis is not out question arises as to whether there is any difference in the per- of the question and several experimental studies, for example those formance of different types of synthetic graft. For some years it was of Greisler , indicate that tissue regeneration may take place generally thought that ePTFE gave better patency rates, but this is within a degrading prosthesis. not necessarily so. One recent study concluded that long-term It should be noted that there has been some apparent success outcomes with Dacron and ePTFE for femorofemoral bypass were recently with the coating of vascular grafts by heparin, Heyligers equivalent and that the preferential use of ePTFE in this situation et al. showing in vitro evidence of both anti-coagulant and was not justified. A further review showed that in femoro- anti-platelet effects with heparin-bonded ePTFE, Devine et al. popliteal bypass, secondary patency for saphenous vein grafts was showing a short term beneficial effect of heparin-bonded Dacron 90% whilst for ePTFE it was 47% and for Dacron 60%. and both Walluscheck and Peeters et al. recording me- In terms of mechanisms by which biomaterials may influence dium term improvements of ePTFE by heparin coating in various this hyperplastic response, two factors may be considered relevant, anatomical sites. Since heparin has both anti-coagulation and related to their effects on thrombosis and inflammation, although anti-inflammatory effects, this evidence cannot distinguish mech- these two processes are linked. The complexity of biomaterial-as- anisms. However, it is of relevance that another approach to sociated thrombosis has recently been reviewed by Gorbet and modulate the processes of intimal hyperplasia involves the mole- Sefton who persuasively argue that activation of contact phase cule nitric oxide, which normally provides an effective endogenous proteins (as implied in the intrinsic pathway of blood coagulation) resistance to leukocyte adhesion and activation, platelet aggrega- is unlikely to be important in the activation of coagulation by bio- tion and the proliferation of vascular smooth muscle, and which is materials, including those used in vascular grafts. Instead, they itself inhibited by the products of vascular tissue injury. There is argue that the extrinsic pathway, involving tissue factor expression now some evidence that a nitric oxide producing or releasing sur- by a variety of cells following vascular damage, such as neutrophils face could reduce intimal hyperplasia suggesting that the and monocytes, is important. Indeed, it may well be that it is ma- sustained inhibition of inflammation, by either heparin or nitric terial-induced leukocyte activation that is the principal mechanism oxide could be very beneficial. involved in the thrombosis that occurs within a vascular graft. Thus, It is a widely held view that synthetic vascular grafts work all of the concepts about the control of thrombogenicity through reasonably well in high flow situations, but far less well under low the physico-chemical characteristics of the biomaterial surfaces flow. As reviewed recently by Cunningham and Gotlieb the and their interaction with plasma proteins and platelets may have pathobiology of atherosclerosis and the related therapy related little or no relevance to vascular grafts; instead it is the control phenomena of re-stenosis and hyperplasia, is largely dependent on of the interactions with leukocytes, and their mediation of in- blood flow induced shear stresses. In vascular grafts these stresses flammation in general, that is important. The marked difference are themselves related to the design and mechanical compliance of between the critical surface tension of PTFE (19 dynes/cm) and the graft, providing further reasons why, in the majority of cir- polyethylene terephthalate (43 dynes/cm) makes little difference to cumstances, provided the material is minimally pro-inflammatory, their performance in vascular grafts. Salzmann et al. have it actually has very limited influence on the tissue response within 2948 D.F. Williams / Biomaterials 29 (2008) 2941–2953 and around the graft. It is also relevant that in untreated blood Of some significance here are the observations that have been vessels the maintenance of laminar flow and physiologic shear made of clinically inappropriate responses with some larger de- stresses is a pre-requisite for normal vascular function and that the vices made of the same polyesters used in fracture fixation. As introduction of flow disturbances is the cause of abnormal behav- reviewed by Bostman and Pihlajamaki , a series of clinical iour , including changes to endothelial cell gene expression, the studies were reported in the early 1990s where there were signif- initiation of oxidative and inflammatory states in the endothelium icant inflammatory reactions to polyester orthopaedic and maxil- and leukocyte adhesion. In vascular grafts, it is usually the sites lofacial implants, usually occurring during the late phase of of anastomoses that cause the most significant flow disturbances degradation, sometimes many years after implantation. There and altered shear stress patterns, and it is here that the intimal are several possible causes of this phenomenon, with the release of hyperplasia is mostly found. a large volume of pro-inflammatory crystals of micron size towards The clinical performance of vascular grafts would undoubtedly the end of the process, the low pH that may be associated with be much different if they were able to form a uniform endothelial these last stages, and the potential release of residual catalysts all lining, but this rarely happens unaided in humans. There have been being invoked at various points, but the essential feature is that many attempts over 20 or so years to solve this problem by seeding either the physical presence of particulate degradation products or the grafts prior to implantation with autologous endothelial cells. the transient chemical characteristics of the degrading milieu are The fact that this can be achieved and used for routine clinical able to stimulate inflammatory cells, especially macrophages and procedures to give significantly improved patency rates has been giant cells, at any such time as their presence exceeds a certain demonstrated by Zilla and colleagues. threshold. There have been some suggestions that this is dependent In an attempt to rationalise why prosthetic vascular grafts do on the nature of the polyester used, for example with claims that it not give better clinical performance , Zilla has analysed the occurs more frequently with polyglycolides than polylactides experience over the last half century and determined that one of but this is unlikely to be a simple matter of implant chemistry the major reasons lies with the fact that the majority of grafts are so rather than the characteristics of the degradation profile. This impervious that transmural tissue ingrowth is impossible, such that profile may be indirectly related to morphological features such as none develop a neointima except for sporadically observed small the crystallinity. Interestingly Chen and colleagues in various islands of endothelium. Moreover, the build-up of the hostile bi- studies on the polyhydroxyalkenoate (PHA) family of polymers ological environment in the inner layers of prosthetic grafts inhibits have reported far less of an inflammatory response with poly(3- capillary ingrowth. Even though the temporal pattern of cellular hydroxybutyrate-co-3-hydroxyhexanoate) systems than with events around the graft is obviously important, for example in- polylactides, attributing this to a slower degradation rate and less volving the role of macrophages in the secretion of cytokines, the acidic and apparently less inflammatory products of degradation. precise functioning of the graft materials with respect to the in- Significantly, blending the polymer with polyethylene glycol sig- fluence of their intrinsic biocompatibility has not been resolved. nificantly accelerated degradation and concomitantly increased the However, there is little or no evidence that deviating from chemical inflammatory response. inertness is of any value; it is no accident that the best patency rates Many degradable polymers have been incorporated into drug are achieved with the most chemically inert polymers. delivery systems, often as microspheres and more recently as nanoscale entities. Anderson has discussed the phenomena of 5. Degradable implantable systems biodegradation and biocompatibility with the former systems on several occasions, for example with respect to polylactide and One of the first reasons for modifying the concept of bio- polyglycolides’ microspheres , and it seems very clear that, compatibility arose with the development of degradable implant- with materials such as these, the tissue response follows a able materials and systems, where a stable equilibrium was straightforward pattern. Immediately after the parenteral injection emphatically not desired, but where the degrading material had to of microcapsules, there will be acute, sub-acute and chronic in- perform a function before or during a process by which it was de- flammatory responses, largely mediated by the mechanical injury graded and eliminated from the body. Initially the focus was on of injection and the physical presence of the particles, monocytes absorbable sutures where, for many years, surgical catgut had been soon becoming the dominant cell. For the duration of the presence the only clinically acceptable material but, being derived from of the microcapsules, and to some extent depending on their size animal sources, did suffer from relatively poor reproducibility and and size distribution, there will be a macrophage/giant cell/fibro- an aggressive tissue response. A series of poly(lactic acid) and blast response, which will then be resolved once the microspheres poly(glycolic acid) based materials, and later some other aliphatic fragment and disappear, often involving phagocytosis by the mac- polyesters was introduced which generally appeared to display rophages and giant cells. Whether or not there will be a residual superior biocompatibility, implying a greater reliability in degra- fibrous capsule depends on the intensity of this process, the ki- dation rates and acceptable host responses to the degradation pro- netics of degradation and the site of injection. cess. Although such materials can be affected by tissue enzymes and The response to microspheres is not always so straightforward. other active chemical species such as superoxides and free radicals Fournier et al. have reported the results of the response of rat , it is generally agreed that the degradation occurs through brain to biodegradable poly(methylidene malonate) microspheres. hydrolysis, the ultimate degradation products being water soluble Their work follows on from the characterisation of the degradation monomer, dimers or oligomers of the respective acid. The host profile of this polymer by Le Visage et al. who showed that response to these sutures and similar devices has involved the one pathway of degradation was by hydrolysis of side chain ester presence of inflammatory cells over the course of the degradation groups leading to the release of glycolic acid and ethanol and period but with no clinically unacceptable outcomes. The bio- leaving a residual polycarboxylic acid, which would be gradually degradable polyester suture material has become the model for solubilised. Fournier et al. revealed two pathways, with direct clinically acceptable biocompatibility performance with intentionally hydrolytic scission of the polymer chain, leading to the release of degradable implant systems, bearing in mind the relatively small formaldehyde and an alkyl cyanoacrylate in addition to the side volume of material used, the 3–12 weeks’ degradation profile and the group hydrolysis. They found that microspheres of the polymer, apparently physiologically acceptable products of degradation and when implanted into the brain, elicited only a mild initial in- resorption. The question arises as to if, how and where degradable flammatory reaction, which became essentially quiescent until the systems do not show acceptable biocompatibility. microspheres began to visibly degrade at about six months, when  D.F. Williams / Biomaterials 29 (2008) 2941–2953 2949 a significant inflammatory reaction was reactivated, with a clear tip appear to be more important than the precise material. The sit- direct toxicological effect on the surrounding tissue as the degrading uation is, therefore, that there is little, if anything, to choose between polymer formed a gel. It was argued that the acidic nature of the silicones and polyurethanes for these applications and nor is there solubilising residual polymer had a direct cytotoxic effect, producing any substantive data to explain why these give better clinical per- what was described as irreversible tissue destruction, although it formance than other polymers. It is in fact likely that non-blood cannot be ruled out that the cyanoacrylate and formaldehyde were compatibility factors are involved, since the flexibility of the cath- not themselves involved. A similar biocompatibility problem has eters and the lubricity of their surfaces are of utmost importance arisen with a degradable orthopaedic adhesive intended to augment from a clinical handling perspective and almost certainly silicones the use of fracture screws. Alkylene bis(dilactoyl) methacrylate was and polyurethanes will be superior to all the other materials men- developed for this purpose, synthesised from ethylene glycol, lactic tioned in relation to these properties. acid and methacrylic acid and although initial studies showed rea- It cannot be claimed that these two families of polymers rep- sonable responses, two detailed longer term studies in large animal resent optimal biocompatibility in the context of venous catheters, models showed massive local inflammatory responses and osteol- but it is far from clear how material, or material-surface specific ysis as the polymer degraded [106–108]. Degradation mechanisms effects are involved with any thrombolic event. Many catheters do are not clear but the similarities are striking and the inflammation eventually become enveloped in a fibrin sheath, but this has not would appear to be associated with a combination of the physical mechanistically been related to any material property nor to any presence of fragmenting polymer and the residence in the local relationship between a material surface and any of the protein ad- tissue of solubilising molecular residues of the degradation process. sorption phenomena that are thought to take place at the interface. Obviously these considerations need to be borne in mind with the As discussed by Brash there are several such phenomena that development of other degradable polymer systems, such as some are potentially involved in the processes of blood–material in- polyurethanes , where the profile of side group and main chain teractions, and indeed can be demonstrated experimentally, and hydrolysis and the solubilisation and distribution of the resulting which lead to theoretical methods of minimising these interactions, molecules have to be taken into account. but there is little evidence yet to show how any of these can be of It should be noted here that significant interest has been shown help in the complexity of the clinical application. Systemic heparin to entities of very small dimensions in the development of more is known to be of help in some situations and this fact has now been precise and efficient delivery systems, especially for complex translated to the use of heparin-coated catheter materials, notably pharmaceutical and gene delivery processes. At one stage micro- polyurethanes, at least in the short term. The performance spheres were very popular but in the last five years it is the may be improved by using covalent complexes of heparin with nanoscale that has received most attention. By definition the antithrombin. nanoscale means of the order of 100 nanometres or less, and a wide range of nanoparticle based formulations have been developed and assessed. The understanding of the biocompatibility of nano- 7. Tissue engineering scaffolds particles is in its infancy, and factors such as the ease of trans- location of nanoparticles throughout the body, possibly including 7.1. Background the brain, and across membranes, including cellular membranes and the blood–brain barrier, and their ability to directly interact Notwithstanding the routine clinical successes with many of the with DNA are obviously important [110–112]. long-term implantable devices discussed in Section 4 above, there are significant limitations to the approach of using manufactured 6. Transient invasive intravascular devices prosthetic devices for the treatment of chronic diseases or injuries. Biocompatibility considerations obviously provide one category of Large numbers of patients, such as those undergoing haemo- limiting factors although we have seen that, provided certain basic dialysis, come into contact with biomaterials through the insertion rules are followed, these do not constitute difficult barriers in most of a catheter into their venous system, either for a short term de- situations. As discussed by the author elsewhere , the major livery of some substance for nutritional, diagnostic or therapeutic limitations are based on the fact that non-viable replacements for purposes, or for more long-term purposes. The intervention may be tissues and organs can largely address only their physical and either central or peripheral. For many years it has been recognised mechanical deficiencies on a long-term basis and there is simply no that these interventions are not without risk, largely related to ei- options for the use of synthetic structures to replace the biological ther infection or thrombosis and their sequelae. Because of the well (e.g. metabolic) functions. A fundamental shift in the approach to known propensity for foreign materials to induce thrombosis, therapies for many diseases has been witnessed during the last it might be assumed that the inherent blood compatibility of the decade, through the various strategies within the broad area of catheter materials was of critical importance in the selection of regenerative medicine. Usually taken to include cell therapies, gene suitable materials. Some 25 years ago, the materials selected in- therapy and tissue engineering, these are essentially aimed at cluded various forms of polyethylene, PVC, PTFE, silicone elastomers regenerating diseased tissue rather than replacing it with synthetic and polyurethanes. A few studies were performed (for example see materials. Tissue engineering has been defined conceptually as : Ref. ) which attempted to compare different materials but these ‘The persuasion of the body to heal itself, through the delivery to the were never of real value in determining mechanistically how ma- appropriate sites of molecular signals, cells and/or supporting terials performed, although through rarely documented procedures, structures’, the first three of these materials were largely discarded such that today it is mainly the silicones and polyurethanes that are used but is perhaps better seen at a more practical level as:. A comprehensive review of risk factors for deep vein throm- ‘Tissue engineering is the creation of new tissue for the therapeutic bosis related to central venous catheters has considered pa- reconstruction of the human body, by the deliberate and controlled tient characteristics and the catheter itself and there can be no doubt stimulation of selected target cells through a systematic combi- that the former (including inherited coagulation disorders and the nation of molecular and mechanical signals’ disease state, such as cancer) are likely to be as significant as the catheter features. Within the latter category, the number of lumen in It will be seen that tissue engineering is concerned with the the catheter, the puncture site and the eventual site of the catheter stimulation of cells, from wherever they are derived, to generate 2950 D.F. Williams / Biomaterials 29 (2008) 2941–2953 new tissue, often through the expression of extracellular matrix, for or sub-micron sized particles. What is urgently required is a re- the functional restoration of tissues or organs. This is not a trivial assessment of the specifications for these materials, which has to task since most of the affected cells in human adults do not innately include a deeper understanding of their biocompatibility. have this capacity, hence the emphasis on controlled and system- There will not be a single set of requirements for all of the ap- atic stimulation. Although the use of a biomaterial is not mandatory plications, and they will depend on the tissue or organ under re- in tissue engineering (which itself calls into question the boundary generation as well as the location of the regenerative process, that between tissue engineering and cell therapy), most attempts to is whether the process is being carried out ex vivo or in vivo, and accomplish this stimulation have involved biomaterials in one form the nature of the bioreactor system. It should be obvious that the or another, partly to impart shape to the tissue that is being scaffold should have two functions, to determine the shape of the regenerated, and partly to facilitate the stimulation via molecular regenerated tissue and to facilitate the appropriate cell behaviour, and/or mechanical signals. especially the development and/or maintenance of phenotype and Tissue engineering may be achieved through several different the expression of relevant extracellular matrix. Although cells routes but there is a basic paradigm of ex vivo tissue regeneration, seeded in a scaffold may regenerate tissue spontaneously without discussed recently which may serve as a template, in which any specific guidance from the scaffold material, it is unlikely that there is a progression from cell sourcing through cell manipulation this will be achieved with any degree of consistency and efficiency. and signalling to tissue expression and construct formation, fol- The biomaterial should be biologically active in the sense that it lowed by implantation into the host and its full incorporation into should possess, within its molecular structure, the appropriate li- that host. In the centre of this paradigm is the seeding of the re- gands that are recognisable by the relevant cells and made available quired cells into a biomaterial scaffold or matrix, wherein they to those cells with the right density and over the appropriate length produce the new tissue. Usually, although not necessarily, the bio- of time. This is obviously a significant challenge, especially if more material is required to degrade or dissolve as the new tissue forms. than one cell type is involved in every functional tissue in the body. Obviously the biocompatibility of the biomaterial is crucial in This will not be achieved with the conventional synthetic bio- this process and here we see a distinct departure from the desirable materials, but should be possible with natural biopolymers, such as characteristics of biocompatibility discussed in relation to almost certain individual proteins (collagen , elastin , silk ), all previous biomaterial scenarios. Whereas with implantable polysaccharides (hyaluronan , alginate , chitosan ), devices, drug delivery systems and intravascular invasive devices, some natural tissue derived materials [127,128] and some engi- the key to biocompatibility success with any material has been to neered forms or derivatives of such substances. It should be achieve a physical or mechanical function without eliciting any noted that there have been many attempts to confer this type of unusual response from the relevant tissue, with a tissue engineer- bioactivity to synthetic polymers by surface grafting of molecules, ing scaffold or matrix, the whole point is that the material should such as peptide or amino acid sequences. Although some be designed to actually elicit such a response. It is poignant and changes to properties in vitro are often seen, it is difficult to visualise relevant to note that the vast majority of attempts to produce how such a modified surface can maintain activity in vivo, especially scaffold-based tissue engineering products have been predicated as the underlying polymer has been designed to be hydrolysable. In on the perceived requirement that the material should have had order to facilitate the cell–surface interactions, certain properties previous regulatory approval within the context of medical devices, will be important, including the hydrophilicity , and of special a specification which may have a practical (and economic) basis, relevance will be the geometrical features of the porous structure, but which is fundamentally in error from a scientific point of view. including the pore size and size distribution , the micro- The question today is that if such a specification is erroneous, what architecture , and the degree of heterogeneity, isotropy and are better specifications for tissue engineering biomaterials with interconnectivity of the porosity within the template. respect to biocompatibility? Of equal significance is the requirement for suitable bio- degradation parameters. There is no point in designing a system 7.2. The essential material specifications for a tissue engineering that will facilitate complex tissue regeneration if that tissue is ul- scaffold timately destroyed by the influx of inflammatory cells associated with the degradation process or if the material stimulates the Let us assume for a moment that, within the basic paradigm immune system as it degrades and releases antigenic material. outlined above, we wish to use a biomaterial to support cells in an Obviously the material and its degradation products have to be ex vivo culture system. Let us further initially assume that the devoid of any potential for mutagenicity, genotoxicity, carcinoge- material is to be used as an open porous system and that the cells nicity, reproductive toxicity and other adverse systemic effects. are fully differentiated cells derived from a biopsy taken from the It is clear that some basic principles of biocompatibility still eventual host. The question naturally arises as to the nature of the apply, but that entirely different mechanisms of interaction should specifications of the optimal materials for those scaffolds. The ap- be required and achieved. parently successful use of degradable polymers in medical devices has, unfortunately, been extrapolated into the ‘potential’ for such 8. The central biocompatibility paradigm materials to be used in tissue engineering products, without an understanding of the requirements and specifications for these two We have previously defined biocompatibility in terms of the quite different applications. There is a large difference between the ability of a material to perform with an appropriate host response in requirements of a biodegradable material for a medical device, a specific situation. That was, at its inception, a powerful reminder which as we have seen should not interfere at all with any biological that biomaterials have to perform a function, and can only do so if process, and a biodegradable scaffold material, which should assist they invoke a response from the tissues, or tissue components, that in the biological processes associated with regeneration. they are in contact with, that is, at the very least, compatible with The current situation with respect to this specification is, in fact, that function, or better, actively support that function. It is neces- even more problematic since the presently used biodegradable sary, as originally envisaged, to define biocompatibility specifically materials do not always satisfy the requirement of degrading in relation to those functions, but also to have a more profound without harmful effects for, as we have seen (for example see overarching concept. It is clear from some well established situa- Ref. ), the processes of degradation can be pro-inflammatory tions, in which there is ample clinical evidence, that the principal through the release of acidic moieties, residual catalysts and micron component of a material’s biocompatibility is that, whatever the  D.F. Williams / Biomaterials 29 (2008) 2941–2953 2951 desired function, the material shall do no harm, just as the first implantable devices. Here, over 50 years of experience has de- principle of Hippocrates was that the doctor should do no harm. termined that, in the vast majority of circumstances, the sole re- Long-term implantable medical devices are the obvious here, and quirement for biocompatibility in a medical device intended for the following definition or paradigm may be proposed: sustained long-term contact with the tissues of the human body is that the material shall do no harm to those tissues, achieved The biocompatibility of a long term implantable medical device through chemical and biological inertness. Rarely has an attempt refers to the ability of the device to perform its intended function, to introduce biological activity into a biomaterial been clinically with the desired degree of incorporation in the host, without elic- successful in these applications. Only now that the focus for bio- iting any undesirable local or systemic effects in that host. materials has turned towards tissue engineering, sophisticated cell, The edict that the biomaterial shall do no harm may well be re- drug and gene delivery systems and indeed applications in bio- assuring to the recipient of an implantable device that is intended technology, has the need for specific and direct interactions be- to survive longer than them, but may not be sufficient for other tween biomaterials and tissue components become necessary. The stakeholders in advanced medical technologies, where specific portfolio of biomaterials will now include poly(ethylenimine) non- functionality, usually sooner rather than later, as well as safety is viral vectors, recombinant silk, elastin and collagen proteins, required. If we take tissue engineering scaffolds as examples, there superparamagnetic iron oxide nanoparticles, 3-D fibrinogen based is little point in using inert materials, or, even more importantly, hydrogels, RGD-polymer blends, electrospun nanofibrous com- materials that have non-specific or inappropriately directed activity. posites, DNA-based nanoswitches for reagentless sensors and pat- Here, it is suggested that a more appropriate paradigm would be: terned, doped diamond like carbon for neuroprostheses, and many more. With these new biomaterials a new paradigm for bio- The biocompatibility of a scaffold or matrix for a tissue engineering compatibility has emerged. It is believed that once the need for this product refers to the ability to perform as a substrate that will change is recognised, so our understanding of the mechanisms of support the appropriate cellular activity, including the facilitation biocompatibility w