Porous scaffolds for bone regeneration

Globally, bone fractures due to osteoporosis occur every 20 s in people aged over 50 years. The significant healthcare costs required to manage this problem are further exacerbated by the long healing times experienced with current treatment practices. Novel treatment approaches such as tissue engineering, is using biomaterial scaffolds to stimulate and guide the regeneration of damaged tissue that cannot heal spontaneously. Scaffolds provide a three-dimensional network that mimics the extra cellular microenvironment supporting the viability, attachment, growth and migration of cells whilst maintaining the structure of the regenerated tissue in vivo. The osteogenic capability of the scaffold is influenced by the interconnections between the scaffold pores which facilitate cell distribution, integration with the host tissue and capillary ingrowth. Hence, the preparation of bone scaffolds with applicable pore size and interconnectivity is a significant issue in bone tissue engineering. To be effective however in vivo, the scaffold must also cope with the requirements for physiological mechanical loading. This review focuses on the relationship between the porosity and pore size of scaffolds and subsequent osteogenesis, vascularisation and scaffold degradation during bone regeneration. © 2020 The Authors. Publishing services by Elsevier B.V. on behalf of Vietnam National University, Hanoi. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Tissue engineering techniques to produce biocompatible scaffolds populated with autogenous cells has recently been shown to be an ideal alternative method to provide bone substitutes [1]. Unlike many other tissues, minor bone tissue damage can regenerate by itself [2]. However, the bone's ability for self-repair of massive defects can be limited because of deficiencies in blood supply or in the presence of systemic disease [3]. Bone-lining cells are responsible for matrix preservation, mineralisation and resorption, and serve as precursors of osteoblasts [4]. However the penetration, proliferation, differentiation and migration abilities of these cells are affected by the size and geometry of the scaffold's pores and the degree of vascularisation [5].
Bone tissue engineering requires a suitable architecture for the porous scaffold. Sufficient porosity of suitable size and interconnections between the pores, provides an environment to promote cell infiltration, migration, vascularisation, nutrient and oxygen flow and removal of waste materials while being able to withstand external loading stresses [6]. The pore distribution and geometry of scaffold strongly influences cells ability to penetrate, proliferate and differentiate as well as the rate of scaffold degradation. The scaffold degradation rate needs to be compatible with the maturation and regeneration of new tissue after transplantation in vivo [7]. Therefore, materials of ultra-high molecular weight that do not degrade in the body have limited use as bone graft materials [8]. The products of the degradation process should also be nontoxic and not stimulate an inflammatory response [9]. As such the appropriate physical and chemical surface properties of the scaffold are an inherent requirement for promoting the attachment, infiltration, growth, proliferation and migration of cells [10].

Methods for the fabrication of porous scaffolds
A number of methods have been used to control the porosity of a scaffold (Fig. 1). The combination of the freeze-drying and leaching template techniques generates porous structures. In this method, the pore size can be adjusted by controlling the gap space of the leaching template, temperature changes and varying the density or the viscosity of the polymer solution concentration during freeze drying technique [11e13]. It is not yet clear whether scaffolds with uniform pore distribution and homogeneous size are more efficient in tissue regeneration than those with varying pore size distribution. Supercritical CO 2 foaming and melt processing is another method to produce porous scaffolds with different pore sizes. In this method, the molecular weight of the polymer component is changed, which affects the pore architecture [14].
Other fabrication methods for creating porous scaffolds in macroscale dimensions include rapid prototyping, immersion precipitation, freeze drying, salt leaching and laser sintering [15]. Scaffolds with high interconnectivity and heterogeneous (large and small) pores can be obtained by using melt mixing of the two polymers [16]. Of these methods, electrospinning method delivers fibres with nanometre dimensions because of the high surfacearea-to-volume ratio, a property that is exploited to ensure a suitable surface for cell adhesion. The instability of the electrostatically drawn polymer causes the jet to whip about depositing the fibre randomly [17]. The formation of ordered structures by controlling fibre placement is one of the challenges of electrospinning. The charges of the electrospun fibres can produce a firmly compressed nonwoven mesh with very small pore sizes, which prevents cell infiltration [18]. Modified patterned stainless steel collectors or the use of cubic or circular holes as the template allow for the production of macroporous architecture scaffolds with an adequately large pore size to allow cell infiltration [19]. However, the direct melt electrowriting (MEW) technique is the most appropriate candidate for generating homogeneous porous biomaterials with a large ordered pore size (>100 mm). MEW can provide a suitable substrate to enable cells to penetrate sufficiently by controlling filament deposition on a collector resulting in customisable pore shapes with specific pore size [20].
The morphology of the scaffold is a key aspect that affects the migration of cells [21]. The key parameters to consider when optimising this scaffold morphology to create a scaffold with balanced biological and physical properties include the total porosity, pore morphology, pore size and pore distribution in the scaffold [22].

Homogeneous pore size
The size of osteoblasts is on the order of 10e50 mm [23], however osteoblasts prefer larger pores (100e200 mm) for regenerating mineralised bone after implantation. This allows macrophages to infiltrate, eliminate bacteria and induce the infiltration of other cells involved in colonisation, migration and vascularisation in vivo  [24]. Whereas a smaller pore size (<100 mm) is associated with the formation of non-mineralised osteoid or fibrous tissue [24,25].
Early studies demonstrated significant bone formation in 800 mm scaffolds. Smaller pores were filled with fibroblasts while bone cells preferred to be located in larger pores suggesting a pore size of 800 mm was more appropriate to provide adequate space for cell ingrowth [26]. Similarly, Cheng et al. [27] using magnesium scaffolds with two pore sizes of 250 and 400 mm, showed that the larger pore size leads to greater formation of mature bone by promoting vascularisation. This is due to newly formed blood vessels which supply sufficient oxygen and nutrients for osteoblastic activity in the larger pores of implanted scaffolds which leads to the upregulation of osteopontin (OPN) and collagen type I and subsequent generation of bone mass [27].
Lim et al. however reported that pore sizes in the range of 200e350 mm was optimal for osteoblast proliferation whereas a larger pore size (500 mm) did not affect cell attachment [28].
Smaller pores are suitable for controlling cell aggregation and proliferation [29] however the exogenous hypoxic state associated with these scaffolds stimulates endothelial cell proliferation [30].
Also, proinflammatory cytokines such as tumour necrosis factor a and interleukin 6, 10, 12 and 13 are secreted at higher levels in larger pores and can trigger bone regeneration response [31].
In contrast to macropores, micropores provide a greater surface area, favourable protein adhesion and cell attachment on the scaffold in vitro [20]. O'Brien et al. suggested that the best pore size for initial cell adhesion was 95 mm in vitro [32]. Murphy et al. reported that a pore size of 100e325 mm was optimal for bone engineering scaffolds in vitro [33]. Previous studies have shown that although a pore size >50 mm (macropores) has beneficial effects on osteogenic quality, cell infiltration is restricted by small pore size in-vitro. Pore size <10 mm (micropore) creates a larger surface area that stimulates greater ion exchange and bone protein adsorption [34,35].

Heterogeneous pore size
Natural variation in bone density occurs in the axial direction of long bones, displaying a gradient in porous structure from cortical bone to cancellous bone [36]. This suggests bone implants made of porous gradient biomaterials that can mimic the properties of natural bone with a porosity-graded structure, may perform significantly better in bone regeneration applications.  showed a more porous layer imitated light spongy cancellous bone which had greater cell growth and transport of nutrients and waste in the highly porous region. Whereas, a compact and dense layer simulating the stiff cortical human bone was favourable for external mechanical loading [37]. Therefore, scaffolds with a gradient in porosity may be a good candidate for bone regeneration. According to Luca et al., gradient PCL scaffolds improved the osteogenic differentiation of human mesenchymal stem cells (MSCs) in vitro by increasing the calcium content and ALP activity because of the better supply of oxygen and nutrients in larger pores [38]. Sobral et al. evaluated cell-seeding efficiency of a human osteosarcoma cell in 3D poly (ε-caprolactone) scaffolds with two gradient pore sizes; 100e700e100 mm and 700e100e700 mm.
The pore-size gradient scaffolds exhibited better seeding efficiency, which increased from about 35% in homogeneous scaffolds to about 70% in the gradient scaffolds under static culture conditions [39].

Pore geometry
Another feature that influences the rate of bone regeneration is the geometry of the porous scaffold [40]. Most scaffolds designed for tissue engineering have different pore morphologies as a result of the differing methods i.e. salt leaching, gas foaming, freeze drying, rapid prototyping (RP) and 3D printing techniques [12,41]. Differences in pore width and curvature of the surface have been shown to lead to variations in tissue morphology and growth rate. A high growth rate associated with higher curvature. In other words, more tissue is formed because of the smaller vertical spaces between the struts [42]. Similar results were reported with greater cell proliferation occurring at the short edges of rectangular pores than at the long edges [15,42].
Tissue formation favours concave surfaces compared with flat and convex regions. Concave surfaces provide room for cell alignment, whereas convex surfaces delay tissue growth [42], as there is greater cell stress and density of actin and myosin fibres in concave areas that advances the cells migration [43]. Larger pores have a larger perimeter and less curvature (Fig. 2) [42]. There are no experimental data to support the hypothesis that minimum cell stress increases bone regeneration. However, this hypothesis is based on the stability and equilibrium of the cells to minimise their energy on a minimum surface area [42,44]. This reflects the natural tendency of molecules to minimise their energy. Therefore, cells try to reduce their surface energy to the lowest possible level by reaching the most stable state on the corners of the pore to have more contact with other cells. At the corners of a pore, the small angle of struts provides a suitable environment for cells to interact and to minimise their residual energy, whereas cells at the pore centre have the highest level of energy and are in an unsteady state [45].
According to Van Bael et al., Ti 6 Al 4 V scaffolds with hexagonal pores showed the highest cell growth, and decreased with rectangular pores and decreased further with triangular pores (Fig. 3). The reason for these differences is the higher number of corners and the short distance between the two arches in the corners, particularly in hexagonal pores. This means that cell bridging occurs faster in hexagonal pores compared to rectangular and triangular pores whose struts are further apart. However, they found out the regulation of osteogenic differentiation of the cells was independent to their proliferation and ALP activity increased in triangular pores [46]. Xu et al. reported that parallelogram and triangular shaped 3Dprinted macroporous nagelschmidtite (NAGEL, Ca 7 Si 2 P 2 O 16 ) bioceramic scaffolds exhibited greater proliferation than the square morphology. The parallelogram morphology had the highest ALP activity in the NAGEL scaffold compared with the other pore morphologies [47]. Yilgor et al. designed and constructed four complex structures of 3D printed porous PCL scaffold by changing the configuration of the deposited fibres within the architecture (basic, basic-offset, crossed and crossed-offset) (Fig. 4) [48]. Greater mesenchymal stem cells (MSCs) cell proliferation was observed for the basic offset scaffolds compared with higher cell differentiation and ALP activity in crossed scaffolds. These findings suggest that the basic-offset scaffolds (homogeneous structure) allowed cells to grow homogeneously because of the higher number of anchorage points. Interconnected struts created the angles, which differed from those in basic scaffolds and increased differentiation [48].
In a similar study, Yeo et al. fabricated various PCLeb-TCP (20 wt %) scaffolds with a square pore shape, but with five pore sizes of different offset values (0%, 25%, 50%, 75% and 100%). They found superior cell differentiation and proliferation efficacy for calcium deposition and ALP activity (up to 50%) for scaffolds with offset values of 50% and 100% [49]. These findings suggest that designing the architecture with different offset values can alter the cell behaviour, proliferation and differentiation.

Role of porosity in scaffold permeability
Higher permeability improves the amount of bone ingrowth and inhibits the formation of cartilaginous tissue in the regenerated site [50]. Permeability depends on porosity, orientation, size, distribution and interconnectivity of the pores. A larger pore size is preferred for cell growth and proliferation because the pores will be occluded later than smaller pores during progressive growth and will therefore provide open space for nutrient and oxygen supply and further vascularisation in newly formed bone tissues [51]. However, O'Brien et al. reduced the permeability by decreasing the pore size of collageneGAG scaffolds in vitro [52]. Hence, the greatest seeding efficiency is obtained by using the smallest pore size [53]. The interconnectivity of pores must also be considered when trying to create sufficient permeability and prolong pore occlusion [54]. The interconnectivity of porous scaffolds needs to be large enough for cell infiltration. For instance, ceramic-based coralline scaffold has a pore size of 500 mm, which showed optimal cell penetration [55]. The highly open pore architecture allows the cells to pass though the length of scaffold and settle at the bottom of scaffold without binding between the cells and the surface-adsorbed proteins [56]. On the other hand, restricted pore size and lack of space for infiltration forces cells to differentiate instead of proliferation [55]. Therefore, pores with smaller dimensions may not be appropriate for encouraging bone formation because they may create a hypoxic state and stimulate chondrogenesis instead of osteogenesis [57].

Role of porosity in scaffold vascularisation
Insufficient vascularity in complex or thick tissues such as bone limits spontaneous regeneration of these parts [58]. A fracture in natural bone produces a hypoxic environment, which leads to upregulation of angiogenesis and eventually creates a vascular network [59]. This process is followed by the differentiation of (MSCs) located in the medullar cavity to cartilage [60]. The newly formed cartilage is then calcified and hardened into bone. Because of the inability of the impermeable inner cartilage to transport nutrients, the cartilage cells start to die, which creates cavities and allows the vessels to invade the cavities and the vascular mesh to develop. Osteoclasts, osteoblasts, lymphocytes and nerve cells also penetrate into the cavity, and the remaining cartilage start to collapse after secretion of osteoid by osteoblasts and osteoclasts, which form the spongy bone [61]. The hypoxic zones actuate the tips of endothelial cells, which behave like oxygen sensors and migrate toward the oxygendeficient area. Stalk cells begin to sprout and branch to create vessel channels [62].
One strategy for creating in vivo preformed vessels is a two-step surgery involving implantation of a cell scaffold into a wellvascularised spot such as beneath the panniculus carnosus muscle before the next implantation at the injury site [63,64]. Another bone tissue engineering approach induces prevascularisation and osteogenesis by combining endothelial cells and osteoblasts, which will display synergistic communication and integration of VEGF, bFGF, PDGF into the biomaterials [65]. These pro-angiogenic growth factors can be supplemented within the scaffold by loading or simple coating to promote endothelial cell proliferation and vessel maturation [66]. The normal speed of neovascularisation is 1 mm per day [67]. The dual delivery of two growth factors in combination speeds the maturation of the vascular network towards full development even in larger constructs compared with single-factor delivery [68]. Multiple drug delivery requires the coculture of two cell types that require different growth factors to proliferate and to mature into blood vessels [69]. For example, the incorporation of MSC-derived osteoblasts as the bone cells and EPCs as the blood cells which induce a greater vascular formation to support early osteogenesis [70].
Another important point for angiogenesis is the high cell density needed for vasculogenic differentiation. This in turn is a function of the size of the construct which will depend on the size of the defect. Larger constructs require a greater supply of oxygen and nutrients and if these are inadequate, spontaneous vascularisation will be insufficient and the vascular network will not penetrate into the implant [71]. The optimal pore size for vascularisation during osteogenesis was noted to be 400 mm [72]. The cell population should be adequate to cover the porous structure according to the shape and dimension of the scaffold [73].
Maintaining capillarity and providing a consistent capillary force to stimulate cell diffusion and vascularisation after implantation also needs to be considered for bone engineering. Because the macro-and microporous scaffolds which are inserted into the defect site may already be filled with biomolecules and endogenous cells from physiological fluid in the early stages of implantation, this may prevent or slow continued flow of liquid [74]. According to Rustom et al., biphasic calcium phosphate scaffolds with a micropore (<20 mm) size of 2.2 mm and macropores (>300 mm) with a size range of 650e750 mm ensures a homogeneous cell distribution and bone volume fraction throughout the scaffold via the capillarity mechanism. This study reported that the capillarity process increased the bone distribution uniformly and incorporated a variety of vascular cells in the empty dry micropores which were not occupied by submersion in fluid after implantation. This is significant as better bone distribution improves the load-bearing of the repaired bone defect consequently [75].
The use of inorganic bioactive elements has advantages associated with their longeterm activity after implantation [1]. The instability, high cost and a short half-life of growth factors in vivo inhibit their usefulness in clinical translation [76]. Cobalt (Co) ions are used as a cofactor for metalloproteins, which are required for the formation of the HIF-1a complex, which activates and regulates vegf and numerous angiogenic genes in vitro [77]. Zhao et al. integrated Co nanograins measuring 30e60 nm at different concentrations coated on the surface of TiO 2 /TCP microporous structure with a diameter pore size of 3e4 mm. The spreading and attachment of the cells was greatly improved because of cell anchorage to the micropores of the TCP construct. Cell proliferation was best in the low Co concentration range of 10 ppm. However, a higher Co concentration (>15 ppm) caused cell cytotoxicity and reduced cell proliferation. But Co dose enhancement had positive effects on osteogenesis by increased angiogenic factors (VEGF and HIF-1a) [78]. Xu et al. reported that the release of Ca, P and Si ionic products from NAGEL, Ca7Si2P2O16 scaffolds accelerated the proliferation of human umbilical vein endothelial cells (HUVECs) in at high concentrations (12.5 mg ml À1 ) of NAGEL extracts by promoting angiogenesis and endothelial cells for bone engineering [47].

Role of porosity in scaffold mechanical properties
There is a linear relationship between the resistance to mechanical loading and bone density or toughness [79]. The complex heterogeneous and hierarchical structure of bone tissue creates variations in compressive strength and tensile values in different regions of bone [80]. A reduction in bone mass increases the susceptibility to fracture [81]. Cortical bone contains 20% porosity along the transverse axis and has a load bearing capacity of 8e20 GPa parallel to the osteon direction. Cancellous or spongy bone (>90% porosity) is found next to joints that are highly vascular with young's modulus of 100 MPa, which is lower than that in cortical bone. Therefore, cortical bone generates compact bone which is denser than cancellous bone [82].
One effective factor for regulating the mechanical properties of a scaffold is the porosity. The mechanical properties of the scaffold tend to deplete exponentially with increasing porosity [83,84]. Cell delivery requires a highly porous scaffold (>90%), and porosity >80% is not recommended for polymeric scaffold implantation into bone defects [85,86]. The polymer molecular weight can also affect the porosity, interconnectivity, pore size and mechanical properties of a scaffold [15]. Contradictions in mechanical property results between in vitro and in vivo studies may have been affected by different cell types that desire different pore sizes for localization in the scaffold after implantation. For example, fibroblasts, which prefer to be deposited in smaller pores compared with bone cells that prefer larger pores. According to the study of Roosa et al., the mechanical properties were higher in scaffolds with pore sizes of 350 mm compared to 550 and 800 mm 4 weeks after implantation.
This increase may be due to initial filling with fibroblast cells that prefer smaller pore sizes while the bone cells preferred the larger pores (550 and 800 mm). The mechanical stability of the scaffold therefore decreases over time following the addition of bone cells into the larger pores [26].
The Young's modulus and mechanical properties are affected by modification of the biomaterials. For example, calcium phosphate (CaP) scaffolds are an osteoconductive material that has been used in bone tissue engineering and influence biomaterial stiffness [87]. One of the parameters which increases the proliferation of the osteoblasts is the stiffness of the biomaterial. The submicron and nanoscale surface roughness of the pore wall promotes the differentiation and ingrowth of anchorage-dependent bone-forming cells [29]. Engler et al. confirmed that mesenchymal stem cells differentiate towards skeletal muscle and bone lineages on stiffer substrates and neural cells on softer substrates [88]. According to Gharibi et al., mechanical loading on CaP scaffolds activates transcription factors which upregulate the genes controlling osteoblast differentiation and proliferation such as ERK1/2 and RUNX2 and eventually augment mineralisation in vitro [89].
Other factors such as pore size distribution, homogeneity or heterogeneity of the pores, fibre positioning and orientation, and morphology of the pores also play an important role in determining the ultimate mechanical properties [90]. Serra et al. reported that poly (L-lactide)-b-poly (ethylene glycol) with composite CaP glass (PLA/PEG/G5) scaffolds with orthogonal structure exhibited greater compression strength than those with displaced double-layer patterns. Although the presence of glass in PLA/PEG/G5 increased the compressive modulus, the resistance to mechanical stress decreased because of the large pore sizes [91]. The construct with only one large pore size had a lower Young modulus and poorer mechanical properties [92,93].
The simple architecture of homogeneous scaffolds is prone to collapse under high stress. The complexity of non-uniform porous scaffolds allows them to recover after deformation and maintain their elastic state, which is critical for the effective use of implanted biomaterials and biomedical applications [39]. Ma et al. produced 3D biodegradable porous PLLA and PLGA scaffolds and their mechanical analysis showed that the maximum supported stress was achieved by using uniform small pores. Although heterogeneous porous patterns containing both small and large pore sizes produced better mechanical properties [94]. One study indicated better compressive strength and non-brittle failure for a porositygraded (200e400 mm pore diameter) calcium polyphosphate (CPP) scaffold than a homogeneous porous structure (H-CPP). The reason being increased degradation in H-CPP compared with the porosity-graded CPP [95].
The orientation of pores is another parameter that directly affects the mechanical properties of scaffolds [96]. Arora et al. reported maximum mechanical properties and a doubled Young modulus for aligned pores in vitro and when implanted into an injury site [97]. A more complex morphological architecture has greater compressive strength [98], e.g. Young's modulus was reported as 9.81 MPa for a blended PCL/PLGA bio-scaffold with a diagonal morphology, 7.43 MPa for that with a stagger morphology, and 6.05 MPa for that with a lattice morphology [99]. Other studies by Ma et al. reported that spherical pores in a PLGA scaffold had better mechanical properties than cubic pores [94].

Role of porosity in scaffold degradation rate
The pore size plays an important role in the pattern of scaffold degradation. Although greater porosity leads to further permeability, which ultimately results in faster degradation, other parameters such as the homogeneity of pores, morphology and pore size influence the degeneration of porous biomaterials [100]. For example, Wu et al. investigated the in vitro degradation rate of 3D porous scaffolds composed of PLGA85/15 (poly (D,L-lactide-co-glycolide)) with a porosity of 80e95% and pore size of 50e450 mm in PBS at 37 C for 26 weeks. The scaffolds with larger pore size and lower porosity degraded faster than those with smaller pore size and higher porosity. This finding was attributed to the effect of the higher surface area in the scaffolds with larger pore size which increased the diffusion of acidic degradation products during the incubation period and led to a stronger acid-catalysed hydrolysis [101].
Pore size and porosity regulate the rate of degradation in PLA scaffolds with a pore size of 0e500 mm from solid to highly porous scaffolds with porosity >90%. In another study, degradation occurred faster in scaffolds with a larger pore size and in solid films because the degradation products were trapped in isolated pores as a result of autocatalysed degradation. Intermediate degradation behaviour was observed in scaffolds with pore sizes between 0 and 500 mm [102]. The study of Xu et al. reported that among the different pore morphologies, the square pore provided a faster degradability and scaffold weight loss [47].

Conclusion
This review examined the importance of pore size and porosity on cell behaviour during ossification and angiogenesis, as well as how the porosity of biomaterial scaffolds determines their mechanical and degradation properties. Among the various manufacturing techniques, additive manufacturing technologies have proved more successful in fabricating 3D custom-designed scaffolds with the best configuration to control the pore size.
Macroporous (100 and 600 mm) scaffolds allow better integration with the host bone tissue, subsequent vascularisation and bone distribution. Increasing the pore size increases the permeability, which increases bone ingrowth, but small pores are more suitable for soft tissue ingrowth. Regarding the geometry of the structure, triangular, rectangular and elliptic pores support angiogenesis and cause faster cell migration because of the greater curvature while staggered and offset pores help to produce a larger bone volume compared with scaffolds with aligned patterns. The combination and ratio of endothelial cells and osteoblasts also plays a pivotal role in pre-vascularisation during osteogenesis and homogeneous bone distribution in macroporous scaffolds.
With respect to the scaffold's mechanical properties, a greater compressive modulus is associated with smaller pore sizes, a gradient porosity and staggered orientated pores. The major advantage of using gradient porosity scaffolds is their ability to maintain and recover their elastic properties after deformation, while square pores help to improve the stable mechanical strength. A faster degradation rate is attributed to a larger pore size because of the greater dispersal of acidic products during degradation.
Although several reports have shown the effects of pore size, shape and porosity on ossification, some have reported on the influence of heterogeneous porosity on degradation, mechanical properties and angiogenesis after implantation to stimulate bone healing. As a consequence, there is an extensive scope for further research in this field of bone tissue engineering.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.