WO2021009515A1

WO2021009515A1 - Scaffold for bone ingrowth

Info

Publication number: WO2021009515A1
Application number: PCT/GB2020/051715
Authority: WO
Inventors: Daniel BARBA CANCHO; Enrique MARTINEZ ALABORT; Roger Charles REED
Original assignee: Oxford University Innovation Limited
Priority date: 2019-07-17
Filing date: 2020-07-17
Publication date: 2021-01-21
Also published as: GB201910233D0

Abstract

A scaffold for bone ingrowth comprising: a monolithic structure; and at least a portion of the monolithic structure having a plurality of open cells including a plurality of first open cells intermixed with a plurality of second open cells throughout the portion; wherein the plurality of open cells in the monolithic structure have a bimodal cell size frequency distribution with one mode being associated with the plurality of first open cells and the other mode being associated with the plurality of second open cells.

Description

SCAFFOLD FOR BONE INGROWTH

FIELD

The present invention relates to three-dimensional scaffolds for implant structures and a method of manufacturing a scaffold.

BACKGROUND

Bone is a complex porous bio-material. Implants are used when bone tissue needs to be replaced by synthetic medical devices due to biological problems (e.g. osteoporosis, cancer, fracture).

Implants can comprise a scaffold which is a porous mesh with open cells designed to promote bone integration with the implant. Faster bone integration into a scaffold is desirable to allow: (1) a reduction in the resting time that patients need, (2) a reduction in the current high rate of failed implants due to poor osseointegration and (3) an increase in the activity of the patients due to stronger fixation of the implant.

The idea of using additive manufacturing to manufacture a scaffold is known.

It is desirable to improve bone integration with a scaffold structure.

The present invention provides a scaffold for bone ingrowth comprising: a monolithic structure; and at least a portion of the monolithic structure having a plurality of open cells including a plurality of first open cells intermixed with a plurality of second open cells throughout the portion; wherein the combined plurality of open cells in the monolithic structure have a bimodal cell size frequency distribution with one mode being associated with the plurality of first open cells and the other mode being associated with the plurality of second open cells.

Such a structure can be optimised for both initial bone cell attachment of embryonic bone tissue as well as for bone posterior vascularisation of the embryonic bone to form fully functional bone tissue leading to improved cell attachment and the whole bone growth process reducing patient resting times, improving success rates and increasing activity of patients due to stronger fixing of the implant to bone.

The present invention will be described below by way of example only with reference to the following drawings in which:

Figure 1 illustrates on the left an image of a scaffold according to the present invention and on the right a cell size frequency distribution according to the present invention.

Figure 2 illustrates strut patterns for a unimodal and a bimodal cell size frequency distributions.

Figure3 illustrates schematically the meaning of the term cell size.

Figure 4 illustrates schematically how an aspect ratio can be introduced into a scaffold structure.

Figure 5 is a graph of yield stress vs elastic modulus for scaffolds of the present invention compared to scaffolds with a unimodal cell size frequency distribution.

Figure 6 shows that the fatigue strength of a scaffold according to the present invention is greater than the fatigue strength of unimodal cell size frequency distribution scaffolds.

Figure 7 shows a two-dimensional example of use of the Voronoi method in which fire stations are super imposed over a map of Barcelona.

Figure 8 is a schematic illustration of a cubic space in which the Voronoi method has been used to generate a mesh structure with a plurality of open cells separated by struts.

Figure 9 illustrates two desired cell size distributions matched by actual Voronoi structures generated using Voronoi optimisation methods.

Figure 10 is a flow diagram illustrating a Voronoi method of generating a bimodal cell size frequency distribution within a space.

Bone is a complex porous bio-material with non-homogenous anisotropic properties which supports organs, muscles and other body tissues. Its structure and therefore its mechanical properties are the result of its required functionality: porosity and fibre direction are controlled by biological processes that tailor the topology of the local tissue to the mechanical requirements. This leads to a wide range of porosities, mechanical properties and anisotropy within the same bone all of them optimised for its optimal macroscopic functionality. This macroscopic structure is in local mechano-biological equilibrium: when local bone tissue is not stressed or stimulated, the local flow of nutrients is stopped. This leads to retraction of the bone tissue. Local mechanical stimulus is therefore needed to keep bone tissue alive and healthy. This can only be achieved when the vascular system hosted by the bone porous structure transports a continuous flow of nutrients.

These two aspects (mechanical stimulus and nutrient support for osseoinduction) are important for optimal bone implant design. Also important is initial cell attachment to ensure fast integration of the implant and thereby quick patient recovery.

A scaffold forms at least part of an implant, for example a scaffold may be formed as an outer layer of an implant or the whole implant may itself be formed of a scaffold. The scaffold comprises a plurality of open cells.

Prior art scaffolds have used open cells of a unimodal distribution with the size of the open cells typically being determined empirically and often to mimic natural bone structure with a unimodal open cell size. However the present inventors have postulated that this is the wrong approach. Bone tissue is not just bone but is bone with blood vessels. Additionally, the aim is not simply to mimic bone but to host growing bone. Therefore in the present invention the scaffold is provided with at least a portion which has a plurality of open cells. The open cells include a plurality of first open cells intermixed with a plurality of second open cells throughout the portion. This means that the first open cells and the second open cells are intermixed in the same region of the scaffold rather being formed in discrete locations within the scaffold.

The present inventors have realised that by providing the plurality of open cells to have a bimodal cell size frequency distribution, such as illustrated in Figure 1, it is possible to optimise the first open cells to have a different functionality to the second open cells. This different functionality is affected primarily by making the mean cell size of the first open cells to be different to the mean cell size of the second open cells. For example, the first open cells can have a mean cell size designed primarily to encourage initial bone cell attachment whereas the second open cells can have a mean cell size designed primarily to encourage later vascularisation. Typically the optimal size of an open cell associated with initial cell attachment is smaller than the open cell size associated with later

vascularisation. As seen in figure 1, the bimodal cell size frequency distribution is bimodal in all three orthogonal planes (i.e. in all directions). Whether this is the case or not can be observed by taking cross-sectional samples from a specimen in different directions and observing microscopically. A bimodal cell size frequency distribution in all three orthogonal planes is advantageous over distributions which are bimodal in only one or two planes (as would be the case, for example, where the scaffold is comprised of stacked layers where each layer is one open cell thick) because it is thought that such structures could exhibit undesirable variations in properties (both mechanical and bone growth) in different directions.

In an embodiment the first open cells are distributed with uniform number or volume density throughout the portion. The plurality of second open cells are also distributed with uniform number or volume density throughout the portion.

In the present invention the first open cells and second open cells are mixed in the portion in which they are located. This means that the first open cells do not only occupy a first part of the portion and the second open cells only occupy a second part of the portion. The first and second open cells are mixed together. The first open cells are uniformly distributed in position throughout the portion and the second open cells are also uniformly distributed in position throughout the portion. By uniform distribution of the open cells what is meant is that the chance of finding an open cell in one particular area of the portion with the bimodal cell size frequency distribution is equal to any other area of that portion.

The relative positions of the first open cells and second open cells can be regular or random. The relative positions of the first open cells and second open cells can be regular or random in all three orthogonal directions. In an embodiment the scaffold has a repeating cellular unit. In an embodiment, within each cellular unit the position of the first open cells and second open cells is random. In an embodiment, within each cellular unit the position of the first open cells and second open cells is random in all three orthogonal directions. The way in which such a cellular unit or a portion can be generated will be described below with reference to the Voronoi method which leads to a Voronoi tessellation structure. Thus a portion of the scaffold with the bimodal cell size frequency distribution may be periodic.

It is possible to devise a completely non-random distribution of first open cells and second open cells using regular geometrical shapes. In two dimensions this can be thought of as placing octagons next to each other with squares positioned between the meeting point of four octagons. Such techniques are limited in the ratios both in terms of cell size and proportion of total volume fraction of first open cell to second open cell and the comparative sizes of the two cells. Therefore the present inventors have used well known Voronoi techniques to generate the design of the scaffolds which are then manufactured by additive manufacturing.

Although in the past it has been suggested to use a“graded” lattice design, such designs consist of large cell/pore sizes in the outer region and smaller cell/pore sizes in an inner region so that the two types of cell are not intermixed. This is said to help to encourage transportation of nutrients and oxygen deeper into the implant where cells attach and grow (see X.P.Tan et al. Material Science and Engineering C76 (2017) 1328-1343). The present invention differs from this in that the first open cells and second open cells are both present throughout the same portion of the scaffold and are intermixed in the same portion.

However there may be advantages to having different cell sizes in different areas of the scaffold of the present invention so that different portions of the monolithic structure may have different cell size frequency distributions. At least one of the portions has a bimodal cell size frequency distribution. Such an embodiment is described in more detail below.

Titanium alloys are the preferred option in long-term load-bearing osseointegrated bone implants due to a combination of relatively low elastic modulus, good biocompatibility, exceptional corrosion resistance, and suitable fatigue strength when compared

with other biomaterials such as cobalt-alloys, magnesium alloys, steels or PEEK polymers

(which may all be used with the present invention). However, the stiffness of solid Ti alloys, although lower than other alloys, still exceeds the stiffness of bone by an order of magnitude. This leads to the so-called stress shielding problem: when surrounding bone is not stressed/stimulated due to a stiffness mismatch between the implant and the bone, the flow of nutrients is interrupted, and bone tissue retracts. This leads to implant loosening leading to the need for surgical revision or even catastrophic failures. In addition to the shielding problem, the impossibility of

vascularisation in solid implants leads to a deficit of nutrients to the surrounding bone.

The present invention allows more latitude in tuning the stiffness of the implant to match the circumstances thereby addressing this difficulty.

Advanced manufacturing of porous latticed implants can circumvent many of these problems. The internal structure of pores of these meta-materials allows one

simultaneously to: (1) lower the stiffness to levels suitable for bone applications and (2) host the network of nutrient needed to keep the bone healthy in a loadbearing state.

Several techniques have been developed recently for lattice manufacturing. For example, freeze-casting, 3D fiber deposition, space holder techniques or powder sintering processes. However, the use of these techniques often introduces critical limitations such as small porosity levels, coarse control of size, shape, and distribution of the pores, or even presence of impurities.

Recently, additive manufacturing (AM) has allowed accurate control of the internal pore structure of porous architecture, thus allowing complex geometries to be manufactured with repeatability. These kinds of additive manufactured lattices are gaining a great interest in the biomedical field. Among the different additive manufacturing techniques, selective laser melting (SLM) can be regarded as a very promising route for medical devices due to its versatility, high-precision and accuracy, surface finish, and proven structural integrity.

In the present invention additive manufacturing and in particular selective laser melting is preferably used to manufacture the scaffold for bone ingrowth. This means that the scaffold is comprised of a monolithic structure meaning that each portion of the monolithic structure is integral with all other portions of the monolithic structure. As detailed elsewhere, a scaffold of the present invention may be joined to other components of a medical device for example by fasteners (bolts, screws etc.) or diffusion bonding, welding etc.

Among all the many possible lattice geometries, three main groups can be identified which are based upon the principles of computational geometry. First, lattices based on beams or struts (see for example X. Cheng, S. Li, L. Murr, Z. Zhang, Y. Hao, R. Yang, F. Medina, R. Wicker, Compression deformation behavior of Ti-6Al-4V alloy with cellular structures fabricated by electron beam melting, Journal of Medical Behavior of Biomedical Materials 16 (2012) 152-162; P. F. Egan, V. C. Gonella, M. Engensperger, S. J. Ferguson, K. Shea, Computationally designed lattices with tuned properties for tissue engineering using 3D printing, PloS ONE 12 (2017) 1-20.; J. Kadkhodapour, H. Montazerian, A. Darabi, A. Anaraki, S. Ahmadi, A. Zadpoor, S. Schmauder, Failure mechanisms of additively manufactured porous biomaterials: Effects of porosity and type of unit cell, Journal of the Mechanical Behavior of Biomedical Materials 50 (2015) 180-191). Second, lattices based on triple periodical minimal surfaces (TPMS) with added thickness (see for example F. Bobbert, K. Lietaert, A. Eftekhari, B. Pouran, S. Ahmadi, H. Weinans, A. Zadpoor, Additively manufactured metallic porous biomaterials based on minimal surfaces: A unique combination of topological, mechanical, and mass transport properties, Acta Biomaterialia 53 (2017) 572-584.; O. Al-Ketan, R. Rowshan, R. K. Abu Al-Rub, Topology- mechanical property relationship of 3D printed strut, skeletal, and sheet based periodic metallic cellular materials, Additive Manufacturing 19 (2018) 167-183). Finally, there are lattices derived from TPMS in a skeleton fashion (see for example A. Alabort, D. Barba, R. Reed, Design of metallic bone by additive manufacturing, Scripta Materialia (2019).; F. Liu, Z. Mao, P. Zhang, D. Z. Zhang, J. Jiang, Z. Ma, Functionally graded porous scaffolds in multiple patterns: New design method, physical and mechanical properties, Materials and Design 160 (2018) 849-860 ).

In the present case the present inventors prefer to use lattices based on struts. This allows Voronoi methods to be used to design the open cell structure and thereby to establish a design from which data can be generated to determine the way in which the monolithic structure can be built up layer by layer by additive manufacturing.

Early results indicate that for a given porosity a scaffold according to the present invention is stronger than a scaffold with a unimodal open cell/pore size distribution. Without wishing to be held by any particular theory, it is thought that this is because in the bimodal distribution large pores are surrounded by a shell of small pores. This shell works effectively as a microtruss structure (because small pores are associated with more struts per unit area). The microtruss structure is much stiffer than a solid single strut that would be present in the unimodal distribution. A schematic illustration of the postulated mechanism is given in Figure 2. Not only is this advantageous in terms of ultimate strength, but it allows more flexibility in designing the scaffold to ensure a specific elastic modulus, which is chosen for the specific application in mind; it is desirable that the stiffness of an implant is similar to that of the bone and suitable for the particular application as described above. For example a lower stiffness implant may be desirable where the implant is being used to treat osteoporosis whereas for an active patient where the implant is being used to treat a fracture, a stronger implant may be desired. In addition, the type of alloy used can be chosen to alter the stiffness of the implant. For example, instead of an alpha type titanium alloy, use may be made of a beta type titanium alloy to create an implant with a relatively lower stiffness whilst allowing thicker struts to be used and thereby achieving higher strength.

The osseo-integration capability of each lattice geometry is strongly controlled by the pore size and curvature of the geometry (A. A. Zadpoor, Bone tissue regeneration: The role of scaffold geometry, Biomaterials Science 3 (2015) 231-245. N Taniguchi, S. Fujibayashi, M. Takemoto, K. Sasaki, B. Otsuki, T Nakamura, T Matsushita, T Kokubo, S. Matsuda,

E ect of pore size on bone ingrowth into porous titanium implants fabricated by additive manufacturing: An in vivo experiment, Materials Science and Engineering C 59 (2016) 690-701). The review of Tan et al. (X. P. Tan, Y. J. Tan, C. S. Chow, S. B. Tor, W. Y.

Yeong, Metallic powder-bed based 3D printing of cellular scaffolds for orthopaedic implants: A state-of-the-art review on manufacturing, topological design, mechanical properties and biocompatibility, Materials Science and Engineering C 76 (2017) 1328- 1343) summarises the effect of the mean pore size on bone ingrowth. Two main critical biological processes are affected by the pore size: (1) the initial osteoblast cell colonisation (commonly termed cell attachment) and (2) the following vascularisation of the pre-bone tissue. The former is promoted by smaller pore sizes (faster colonisation when pores are <500 pm, the latter requires large pores to embed the blood vessels (>500 pm). The combination of these requirements is satisfied in the present invention with a bimodal distribution of small pores and larger pores where the osseo-integration capabilities of the implant are expected to be optimal. This would not have been obvious from the prior art in which the above two mechanisms for osseointegration are not formalised or considered together and there is no hint at developing implants which have different pores designed for optimising each of the identified mechanisms.

Therefore in the present invention ideally the first open cells have a mean cell size of less than 500 pm, thought to be optimal for the attachment of some cells. Preferably the mean cell size of the first open cells is less than 450 pm and more preferably less than 400 pm. The procedure to measure the cell/pore size is following the steps in US 6,684,685 B2.

The measuring device (available commercially for example from Porous Materials Inc, see www. pmi app . com) is described in Fig 3. The device comprises of a pressurizable sample chamber for holding the lattice, a membrane located at a bottom of the lattice chamber and having a set of pores, wherein the membrane pores have a size smaller than any of the sample pores of interest in the lattice, a reservoir of measuring liquid located directly below the membrane, and a penetrometer coupled to the reservoir, wherein a level of measuring liquid rises in the penetrometer when additional gas enters the reservoir so changes in volume of the reservoir can be measured. The procedure to measure the porosity after determining the contact angle of the measuring liquid with the material of the lattice is as follows: a) placing the lattice in the sample chamber, on the membrane which can be Poretics polycarbonate membrane catalog No 13705 from Osmonics Inc of

Minnetonka, but the important part is that the pores of the membrane are of a size that will be cleared by the gas after the pores of interest in the lattice, in the present case

measurement pores of 1 pm are used; b) wetting the sample and membrane with the measuring liquid until the measuring liquid has entered substantially all of the pores in the lattice and membrane, c) adding a quantity of gas, in the present case air at 20°C, above the lattice in the Sample chamber at a particular pressure for example up to 20 psi; and d) after equilibrium is reached, measuring amount of displaced measuring liquid with the penetrometer at a known applied pressure of gas . Measuring the amount of displaced measuring liquid at a plurality of gas pressures can be used to calculate the pore/cell size distribution by calculating the equivalent minimum cell/pore diameter D from which the measuring liquid will be forced by the gas at a given pressure for a given measuring liquid using the following the equation:

D = 4 g cos Q /P Where P is the pressure of the gas and theta and gamma are the contact angle of the measuring liquid with the lattice and the surface tension of the measuring liquid, respectively. By knowing the displaced volume of measuring liquid and the minimum pore/cell size from which the measuring liquid is displaced, it is possible to calculate a pre/cell size frequency distribution. Depending on the minimum pore size required to be measured, different fluids can be used as indicated in Fig 3B. In the case of use of a hydrophobic membrane, galwick is used as the measuring liquid and in the case of use of a hydrophilic membrane water is used as the measuring liquid. The volume of the lattice can have an effect on the accuracy of the measurements so the measurements in the present invention are taken from a cylindrical lattice with 20mm diameter and 10mm thickness.

As will become apparent, the structure consists of a plurality of open cells wherein the cells are open in the sense that there is a passage from a given cell to cells adjacent to the given cell. The minimum diameter of a passage from one cell to another is determined by the size of the struts forming the structure and the distribution of seed points from the Voronoi process. The open cells can have imaginary boundaries defined by flat surfaces between the cells. When the scaffold is made, material is only deposited at the intersection of two flat surfaces thereby to form the struts. Thus the centre of the imaginary boundaries defined by flat surfaces is not present in the final structure and forms an opening between adjacent cells.

The porosity of the scaffold can be changed by changing the position of the Voronoi seed points and also by the thickness (diameter) of the struts. In the present invention a strut thickness (measured at a central portion in the elongate direction) in terms of cylindrical cross-sectional area equivalent diameter is in the range of 50 pm and 2 mm, preferably between 75 pm and 1.0 mm.

It has been shown that it is desirable that sharp corners are eliminated in the scaffold structures. For this purpose, in an embodiment, a junction between adjacent struts has its surface smoothened to avoid sharp comers. In an embodiment a radius of curvature of a surface of a junction between attached adjacent stmts is less than 3.0 mm, preferably from 0.5 mm to 3.0 mm.

In an embodiment the first open cells have a mean cell size of 100-450 pm, preferably 100- 350 mih. This size range is expected to be optimal for hosting bone. In an embodiment the standard deviation of cell size of the first open cells is between 0.05 and 0.5 pm.

With regard to the second open cells, which are adapted for vascularisation (i.e. for vessels being formed therein to supply the bone with nutrients), the mean cell size of the second open cells is desirably 350 pm or more, preferably 400 pm or more and most preferably 500 pm or more. In an embodiment the second open cells have a mean size of 400 - 900 pm. A suitable standard deviation of the second open cell size is 0.05 - 0.5 pm.

Figure 1 illustrates a typical bimodal cell size frequency distribution of a scaffold of the present invention in which the pore size is plotted against volume fraction of the open cells. In the embodiment of Figure 1, the ratio of the total volume of the first open cells to total volume of the second open cells is about 1 : 1. In terms of number of cells, this means that there are many more cells of the first type than of the second type.

The ratio of total volume fraction of first open cells to total volume fraction of second open cells can be adjusted to meet specific requirements including, but not limited to: the specific stiffness required of the scaffold, the strength required of the scaffold; the required porosity; the expected difficulties in achieving a successful implant. For example, in certain cases, it may be apparent that difficulty in vasculisation is likely to occur in which case it may be desirable to increase the volume fraction of second open cells. Typically the ratio of total volume fraction of first open cells to total volume fraction of second open cells lies in the range of 3:7 to 7:3, preferably in the range of 2:3 to 3:2.

Another variable which can be adjusted is the total porosity (the percent of volume of open space, as opposed to space filled by scaffold material, for a given volume of scaffold, in other words the percentage of space in the scaffold which could absorb water) of the portion of the scaffold which has the bimodal cell size frequency distribution. Typically the porosity of a scaffold is in the range of 15-25% by volume (measured by taking the mass W of the scaffold and taking its overall dimensions to calculate an overall volume V. From knowledge of the density p of the material from which the scaffold is made, a porosity is calculated (pV-W)/ pV). However the porosity can be adjusted, particularly in order to vary the elastic modulus of the structure. This can be important so that the elastic modulus of the implant can be matched to that of the bone of a particular patient. Ways of adjusting the porosity include: varying the strut thickness; varying the proportion of first open cells to second open cells; varying the mean cell size of the first open cells and/or second open cells; varying the material (for a given required stiffness, the stiffness of the material will determine the porosity). A more porous scaffold allows more space for bone to grow but results in a decrease in strength of the scaffold. Therefore a balance needs to be struck between successful bone integration and strength.

Some bone tissues have directional mechanical properties. It is desirable to match the directionality of the mechanical properties of the bone in the scaffold. This can be achieved by ensuring that the open cells are elongate in a given first direction (i.e. have an aspect ratio). There are two methods of achieving this. In the first method, as illustrated schematically in Figure 4, a scaffold design is produced and is then stretched in the first direction and then the stretched scaffold design is manufactured using additive

manufacturing. The structure and stiffness graph at the top of Figure 4 is for a structure without any aspect ratio. At the bottom, following stretching of the structure, the stiffness is increased in the vertical direction relative to the two horizontal directions as a result of the stretching in that direction. Another method is by selectively changing the thickness of the struts orientated along the first direction. For example, a relationship in the form of thickness of struts being proportional to one divided by the sum of a constant plus the difference in angle of the strut to the first direction. In this way struts with angles close to first direction would have a larger thickness than struts at large angles to the first direction. The constant can be defined empirically and can be changed according to the desired mechanical properties. In an embodiment the open cells have an aspect ratio of between 1 : 1.1 and 1 :3.0, preferably between 1 : 1.5 and 1 :2.0 between the elongated direction and the other two directions.

The prior art devices have included a portion of the device with a first cell size and a portion with a second, different, cell size but not intermixed. This allows properties of the scaffold to be varied and optimised for the particular location of the scaffold. For instance, it may be desirable to provide the first open cells close to the surface of existing bones and more second open cells away from existing bone. This is also possible with the present invention. For example, a first portion may include the above described bimodal cell size frequency distribution with a first set of parameters such as mean cell size, porosity and ratio of first open cells to second open cells. A second, further, portion of the scaffold may comprise open cells of only the first type or only the second type or may comprise open cells of a different, third type (e.g. mean size). The second further portion may even comprise open cells of a fourth type (e.g. mean size) in addition to cells of the first, second or third types (e.g. mean size). In an embodiment the second further portion comprises a plurality of open cells which have a bimodal cell size frequency distribution, just like the first portion. That is, the parameters of the open cells in the second further portion can be different in one or more ways to the parameters of the open cells in the first portion. Thus the first and second further portions can have different cell size frequency distributions.

That is, the difference in bimodal distribution between the first portion and the second portion may be one or more selected from: total porosity of the first portion compared to total porosity of the second further portion; ratio of total volume of first open cells to total volume of second open cells compared to ratio of total volume of third open cells to total volume of fourth open cells; first open cell mean cell size compared to third open cell mean cell size; second open cell mean cell size compared to fourth open cell mean cell size; standard deviation of cell size of the first open cells compared to standard deviation of cell size of the third open cells; standard deviation of cell size of the second open cells compared to standard deviation of cell size of the fourth open cells.

Figure 5 shows experimental results relating to a scaffold of the present invention.

Scaffolds with bimodal open cell size frequency distributions are shown in black dots and scaffolds with unimodal open cell size frequency distributions are shown in shaded dots. As can be seen, the scaffolds according to the present invention with a bimodal cell size frequency distribution show greater yield stress for a given elastic modulus (the elastic modulus mainly being controlled by porosity).

Figure 6 shows similar improvement for the present invention over prior art scaffolds in a fatigue test where for a given stress ratio the bimodal scaffold performs much closer to conventionally manufactured (e.g. rolled) TieAUV than scaffolds with unimodal distribution. Therefore the scaffold of the present invention not only has benefits in terms of being optimised for bone integration, but may also have improvements of mechanical properties compared to scaffolds with unimodal cell size distribution.

The scaffold of the present invention may be formed separately or integrally with other parts of a biomedical device. Such a biomedical device may be a spinal implant such as a spinal fusion cage (which is likely to be comprised completely of scaffold), a hip implant (in which the cup has a mesh on an outer surface but the stalk does not generally require a scaffold); and a shoulder implant in which the scaffold (biomedical mesh) is important because of the low surface area to which the implant is attached. The whole of the biomedical device or only a part, for example the scaffold, may be made by additive manufacture.

The Voronoi method is a method in which a given volume (e.g. the volume of the desired scaffold) is assigned a certain set of seeds at certain, for example at random, positions within the volume. The Voronoi method divides that volume into separate Voronoi cells, each associated with a seed such that every point on a boundary between adjacent Voronoi cells is equidistant from the seed points on either side of the boundary. Figure 7 is a two- dimensional example of this method in which fire stations are super imposed over a map of Barcelona on the left. Each of the fire stations is assigned as a seed and Voronoi cells are generated around each fire station such that the boundary of each Voronoi cell is equidistant from the seeds on either side of the boundary. The resulting pattern is termed a Voronoi tessellation.

Voronoi tessellations occur in many natural systems including metallic microstructures, cell arrangements, tissue arrangements, animal skin patterns and natural porous structures.

Figure 8 illustrates a typical three-dimensional Voronoi tessellation structure using struts. The example of Figure 8 is a space within a cube being assigned a plurality of seed positions and a Voronoi tessellation being formed to define Voronoi cells associated with each seed. Each of the Voronoi cells is filled with the largest sphere which fits within that particular cell. The surfaces defining each Voronoi cell are not shown. Instead, as in the scaffold of the present invention, a strut is placed at the intersection of each boundary surface defining each Voronoi cell.

In order to generate a bimodal cell size frequency distribution using the Voronoi method it is necessary to control the placement of the seeds. For randomly distributed seeds, the cell volume distribution follows a unimodal normal distribution around u=V/n where u is the average volume, V is the total volume and n is a number of seeds. However by changing the position of the seeds different distributions can be achieved including bimodal cell size frequency distributions used in the present invention. This is done using a well-known technique called Voronoi distribution optimisation.

Figure 9 illustrates two cell size frequency distribution graphs, one on the left showing a unimodal distribution with the actual distribution achieved illustrated by bars and the desired distribution with a line. On the right hand side is a bimodal distribution again with the desired distribution illustrated by a line and final distribution illustrated by bars. The way a desired cell size distribution is achieved is to move seed points to obtain the required distribution using a mathematical optimisation algorithm. There are several different known mathematical optimisations, for example those which are gradient based or so called maximum descent methods (the points are moved based on a fitness function in the direction that most optimise the distribution and iterate), Monte Carlo methods (where the seeds are moved randomly to see if the distribution which is achieved is better than earlier ones if so the move is preserved, if not, the move is reversed and a different move is used) or a generic algorithm (where a large number of seed distributions are worked and are mixed together to get the optimal distribution). Figure 10 is an example loop which can be implemented using generic algorithms and illustrated with a target unimodal cell size distribution in 2 dimensions. At the top left the process starts by generating random seed points. Standard software is used to generate the Voronoi cells (middle top of figure 10). The pore/cell size distribution is then calculated (top right) and compared to the desired pore size distribution (bottom right). The pore/cell size is estimated to be equal to the diameter of a sphere with equal volume which produces similar results to the penetrometer measurement method described above. If the desired pore/cell size distribution has been achieved, the process stops. If there is a discrepancy between the calculated pore/cell size distribution and the desired pore/cell size distribution (beyond a certain amount), the process moves to the next stage and the position of a number of the previous seed points is moved. The above described optimisation then occurs. One example of software for generating Voronoi cells is Neper which is an open source software package designed for polycrystal generation and meshing available from http://neper.sourceforget.net/ which can be accessed through the internet archive called way back machine, for example the snapshot taken on 27 January 2019.

Claims

1. A scaffold for bone ingrowth comprising:

a monolithic structure; and

at least a portion of the monolithic structure having a plurality of open cells including a plurality of first open cells intermixed with a plurality of second open cells throughout the portion;

wherein the plurality of open cells in the monolithic structure have a bimodal cell size frequency distribution with one mode being associated with the plurality of first open cells and the other mode being associated with the plurality of second open cells.

2. A scaffold according to claim 1, wherein the first open cells have a mean cell size of less than 500pm, preferably less than 450pm and more preferably less than 400pm.

3. A scaffold according to claim 1 or 2, wherein the first open cells have a mean cell size of 100 to 450pm, preferably 100 to 350pm.

4. The scaffold of any of claims 1-3, wherein the standard deviation of cell size of the first open cells is between 0.05 and 0.5pm.

5. A scaffold according to any of claims 1-4, wherein the second open cells have a mean cell size of more than 350pm, preferably more than 400pm and more preferably more than 500pm.

6. A scaffold according to any of claims 1-5, wherein the second open cells have a mean cell size of 400 to 900pm.

7. The scaffold of any of claims 1-6, wherein the standard deviation of cell size of the second open cells is between 0.05 and 0.5pm.

8. A scaffold according to any of the preceding claims, wherein a ratio of the total volume of first open cells to total volume of second open cells is in a range of 3:7 to 7:3, preferably in a range of 2:3 to 3:2.

9. The scaffold of any of the preceding claims, wherein a total porosity in the portion resulting from the plurality of open cells is in the range of 15-25 percent by volume.

10. The scaffold of any preceding claim, wherein the portion is a strut based lattice structure.

11. The scaffold of claim 10, wherein a radius of curvature of a surface of a junction between attached adjacent struts is 0.0mm to 3.0mm, preferably 0.5mm to 3.0mm.

12. The scaffold of claim 10 or 11, wherein an average thickness of a central portion of the struts is in the range of from 50pm and 2mm, preferably between 75pm and 1.0mm.

13. The scaffold of any preceding claim, wherein at least a portion of the open cells are elongate in a first direction, the open cells having an aspect ratio of between 1 : 1.1 and 1:3.0, preferably between 1 : 1.5 and 1 :2.0.

14. The scaffold of any preceding claim, wherein the monolithic structure comprises a further portion, the further portion having a plurality of open cells including a plurality of third open cells throughout the portion.

15. The scaffold of claim 14, wherein the further portion has a plurality of fourth open cells intermixed with the third open cells throughout the portion;

wherein the plurality of open cells in the further portion have a bimodal cell size frequency distribution with one mode being associated with the plurality of third open cells and the other mode being associated with the plurality of fourth open cells, the bimodal distribution in the further portion being different to the bimodal distribution in the at least one portion.

16. The scaffold of claim 15, wherein the difference in biomodal distribution between the at least one portion and the further portion is at least one selected from: total porosity of the at least one portion compared to total porosity of the further portion; ratio of total volume of first open cells to total volume of second open cells compared to ratio of total volume of third open cells to total volume of fourth open cells; first open cell mean cell size compared to third open cell mean cell size; second open cell mean cell size compared to fourth open cell mean cell size; standard deviation of cell size of the first open cells compared to standard deviation of cell size of the third open cells; standard deviation of cell size of the second open cells compared to standard deviation of cell size of the fourth open cells.

17. The scaffold of any preceding claim, wherein the plurality of open cells in the portion have a Voronoi tessellation structure.

18. The scaffold of any preceding claim wherein the monolithic structure is additive manufactured.

19. The scaffold of any preceding claim, wherein the monolithic structure is made of a titanium alloy, preferably a beta type titanium alloy.

20. The scaffold of any preceding claim, wherein the portion comprises one or more cellular units, each cellular unit having a non-uniform placement of attachment cells and vascularisation cells.

21. The scaffold of any preceding claim, wherein the relative positions of the first open cells and second open cells in the portion are random, preferably wherein the relative positions of the first open cells and second open cells in the portion are random in three orthogonal directions.

22. The scaffold of any preceding claim, wherein the bimodal cell size frequency distribution is bimodal in all three orthogonal planes.

23. A biomedical device comprising a scaffold of any of claims 1-22.

24. A method of manufacturing a scaffold for bone ingrowth comprising:

determining a desired structure for the scaffold with at least a portion having a plurality of open cells including a plurality of first open cells intermixed with a plurality of second open cells throughout the portion; wherein the plurality of open cells have a bimodal cell size frequency distribution with one mode being associated with the plurality of first open cells and the other mode being associated with the plurality of second open cells;

additive manufacturing the structure.

25. The method of claim 24, wherein the determining includes using the Voronoi method to determine a position of struts such that the struts define boundaries of the plurality of open cells which have a Voronoi tessellation structure.