Additive manufacturing (AM), also known as three-dimensional (3D) printing, is a disruptive technology that is revolutionizing many industries. It is gaining considerable attention, particularly in the medical field as it renders the possibilities of building new devices or modifying existing devices to match a patient's anatomy and to produce anatomically exact models, supporting health professionals with diagnostics and surgery preparation. In addition, the free-form building capability of AM allows the designer to have a complete control over the internal architecture of the device, along with tailored mechanical properties, such as compression strength, stiffness, and many surface features. As the processes of AM become well-understood, there is more control over the consistency and quality of the printed parts, positioning this technology for medical applications. With more and more medically approved 3D-printed devices entering the market, the purpose of this paper is to give an overview of the regulatory pathway to the Food and Drug Administration approval of a medical device, along with common AM processes used in the medical industry. To conclude, medical devices that are enabled by AM technology and associated companies will be highlighted.
Additive manufacturing (AM) is considered a disruptive technology that allows customization and rapid production of components that range from the automotive and aerospace industry to medical applications. AM has evolved considerably since its initial application for prototyping purposes several decades ago; it has recently gained increasingly more advantages over traditional manufacturing methods (such as machining, computer numerical control (CNC) machining, injection molding (IM), and forging) due to the advancements in processing technology and material properties . Traditional manufacturing processes need special tools and molds and have limited abilities to fabricate internal features. AM, however, has the possibility of a single-step production of highly complex free-form geometries, with intricate internal features that would have been impossible to manufacture with traditional methods. Examples of internal features include pores, alveola, lattice, and particularly tailored channel networks for fluid flow, spanning from macro to microscopic dimensions [2,3]. Over the years, parts of greater complexity and better functionality have been designed and produced thanks to AM technology. In addition to complex geometric features obtained by AM, they also confer improved mechanical properties, strength-to-weight ratio, and allow for cell attachment and tissue growth—critical properties for weight-bearing implants and tissue engineering (TE) .
Additionally, from the supply chain point of view, AM offers the possibility of building parts on demand, thus reducing delays and waste due to inventory [4,5]. Moreover, reduced production costs of small batches (< 1000 units) can be realized since no molds or special tools are required [6–8]. In a study comparing the production costs per parts of 2 AM processes with the traditional IM method, it shows that IM is less costly for large volumes as the cost for the mold is distributed in the large production . However, the opposite stands true for small batches (below 100,000 units) where AM is cheaper because the mold cost per part would play a more important role than the machine/printer cost per part. This advantage makes AM particularly important to the medical industry, where it sees small productions of specific devices. Single prostheses, implants, or guides can be created for a specific patient at a reduced cost and better fit with AM technology.
Combined with medical imagery (computed tomography scans, magnetic resonance imaging (MRI)), AM is also transforming the medical industry with its ability to render a 3D model of a patient's heart, liver, vertebra, or any organ or bone, enabling better diagnostics or the rehearsal of a surgical procedure [9,10]. Of course, AM's benefits go beyond producing visualization tools; its ability to produce complex and highly precise parts from a digital data file in a matter of days makes AM a revolutionizing process in the production of patient-matched devices with higher clinical success rates  and lowered cost .
However, since AM processes and techniques are highly varied, there is a need for stringent control of the manufacturing processes and materials, as well as approval of 3D-printed devices by health agencies. Although many external devices such as dental splints, hearing aids, and retainers which pose less risk to patients have been commercialized, only a few implants have been judged safe for marketing by the Food and Drug Administration (FDA), and most of the published reports are in the form of case studies.
The objective of this paper is to offer an overview of the AM technology specific to the medical industry. The most commonly used AM processes in the medical industry are described following an order by the materials, from polymers, to metals, and to biomaterials. Furthermore, a summary of the FDA regulatory steps in the development and marketing of a medical device will be given, along with an example of the marketing procedure for a spine fusion cage. This paper further highlights the device classes that could benefit from AM and addresses current developments in bioprinting.
2 Additive Manufacturing Technologies and Systems
AM, also known as 3D-printing, is a process in which a part is created by adding layers upon layers of material, each corresponding to a computer-aided design (CAD) model's cross-sectional area . There are several different AM processes and material combinations leading to different surface finishes, varying internal structures, and unique physical and mechanical properties of the final device. Since the choice of material and process together determine the properties of the final product, ranking of different AM technologies should be taken into consideration of material types and process parameters. The decision to select a particular process should be based on the geometry, the material selected, the dimensional precision, and the scale of production [3,13].
As defined by ASTM, AM processes can be classified into seven categories : binder jetting (BJ), directed energy deposition (DED), material extrusion (ME), material jetting (MJ), powder bed fusion (PBF), sheet lamination (SL), and vat photopolymerization, but hybrid processes and other distinct processes are developed, such as droplet printing, arc-based AM, and cold spraying (CS) .
In the next subsections, common materials and associated AM processes specific to medical applications will be described.
2.1 Processes for Polymer Devices.
Polymers are the most widely used material in AM because they are affordable, versatile, and can be easily formed or machined. Additionally, combined with the correct AM process, polymers can confer a wide range of mechanical properties such as toughness, compression strength, and flexibility . The most commonly used AM processes for polymer in the medical field are ME, powder bed fusion (PBF), and vat photopolymerization/stereolithography (SLA) . Figure 1 outlines the principal polymer AM processes and their classifications based on the process and material feedstock.
2.1.1 Material Extrusion.
Solid-based ME, or fused deposition modeling (FDM), is the most widely used AM technology across many industries . It is readily accessible at a cost of under $500 and will allow design and print with an open access software [2,3]. The printing principle is quite simple; it employs a nozzle that deposits fused material line by line to form layers and then builds on top of each previous layer. Another advantage of ME is that multiple materials or colors can be deposited simultaneously and material change over can be accomplished much quickly than that in a powder bed or resin tank-based process . However, ME produces parts with lower resolutions (100 µm to 1 cm) and higher anisotropy (weaker bonding between the layers) compared to PBF or SLA . Since it needs support material to print overhanging architecture, compatibility of certain geometry and material must be taken into consideration. Additionally, the viscosity and flowability of the melted material greatly affect the resolution. Warping of the final part can occur depending on the nozzle size, material type, and cooling rate .
The main material types used for ME are amorphous thermoplastic polymers because the process works best when polymers are deposited as viscous paste. It is easier to control their viscosity as it decreases when temperature increases, such that printing parameters can be varied according to design specifications. The most well-known polymer used in FDM is ABS, but many other polymers and blends have been used. However, for medical applications, only a few biocompatible polymers have been approved by the FDA. Examples are polylactic-co-glycolic acid (PLGA), polylactide (PLA), polymethyl methacrylate (PMMA), glycol-modified polyethylene terephthalate (PETG), polyvinylidene fluoride (PVDF), polypropylene (PP) and polyetheretherketone (PEEK), to name a few. These FDA-approved polymers have been well characterized in terms of their mechanical properties and printability [19,20]. PEEK is a semi-crystalline, high melting temperature polymer that has been 3D-printed via FDM for medical applications although PBF (2.1.2) method is more suitable for PEEK due to its higher temperature capability. Another polymer that has been approved by the FDA for specific applications, such as suture or drug delivery, and has shown potential opportunities for the AM industry is polycaprolactone (PCL). PCL scaffolds with controlled porosity and mechanical properties were successfully printed via FDM .
To overcome the lack of required mechanical properties of 3D-printed polymer parts, polymers can be blended with other materials to improve the mechanical properties during FDM . By mixing microscale iron or copper microparticles into ABS, it has been shown that improved thermal, electrical, and mechanical properties could be obtained . ABS with carbon nano fibers and graphite flakes can be made into a printable conductive filament; details as small as 0.81 × 0.81 mm have been achieved . Moreover, fiber reinforcement in the polymer can greatly enhance the properties of the resulting AM parts . Short carbon fibers have been shown to increase the tensile and shear modulus of PLA/CF specimens printed by ME , while continuous fibers increase the flexural rigidity of the part [26,27].
MJ, or liquid-based extrusion, is another AM process whose use has been reported by many hospitals for the production of full-color 3D anatomical models [28,29]. The process involves the selective deposition of liquid photopolymer droplets which are then ultraviolet (UV) cured in a polymerization process similar to stereolithography (2.1.3). The main advantage of MJ is its ability to print a 3D model with multiple colors, including transparent, and with flexible material, in a single step with a layer resolution approaching 0.014 mm . Despite its low cost and high deposition rate, there are a limited number of available materials with adequate viscosity suitable for the MJ process and even fewer existing materials with biocompatibility. Therefore, currently, MJ is mostly used for anatomical models. There has been some development of medically approved polymers for MJ, but the manufactured devices are only for short-term skin or mucosal membrane contact, such as dental surgical guides .
2.1.2. Powder Bed Systems.
Powder bed fusion (PBF) systems, also known as selective laser sintering (SLS) or electron beam melting (EBM), were one of the first commercialized AM processes for polymers . PBF uses selective melting or sintering of a powdered material to achieve the bonding of particles while building layers upon layers. The powder, preplaced on the build surface, can be sintered by a laser or electron beam, or it can be gravity fed and simultaneously melted. There are many variations of this process based on (1) either complete or partial melting of the powder, (2) utilization of pure or a mixture of materials with different melting characteristics, or (3) material types. Complete melting of a powder leads to denser parts, but less precise in terms of dimension as the surrounding powder fuses to the printed part in a process known as part growth . This usually leads to further post-processing.
An important feature of PBF over ME is that there is no need for support material when an overhanging feature is built as the unfused powder supports the next layer. Another advantage of this process is its high geometric resolution (50–100 µm) [17,18], rendering it particularly useful for complex architectures. Also, this process enables parts to exhibit similar properties as those obtained from traditional IM, hence potentially stronger and tougher than that obtained by other AM processes such as ME or VP . Additionally, a wide variety of materials can be used in this process, including metal, polymers, and ceramic/metal/polymer blends as described in the next sections [2,3,33]. Furthermore, PBF allows the user to precisely control porosity and internal features, making it possible to create structures with scaffolds and interconnections .
However, the main limitations of PBF are the surface roughness due to part growth (between 10 µm and 20 µm ) under higher process temperature and higher porosity compared to other processes such as MJ and SLA, which are capable of producing fully dense parts. Fortunately, in medical applications, especially for implants or TE, porosity is often a desired feature as it increases cell attachment and growth, such that the production of fully dense parts is often not a primary concern .
Because of the unique nature of PBF, essentially all polymers or polymer mixtures can be processed via PBF , making it possible to print also materials found in nature, such as gelatin, cellulose, or calcium phosphate ceramics. For example, starch–cellulose and cellulose acetate, which are biocompatible and suitable for biomedical applications, were reported to show a higher degree of sintering and to have promising mechanical properties for TE when combining particles (106–125 µm in dia.) in an SLS process . In addition, PBF printed scaffolds with PCL and hydroxyapatite (HA) have been observed to have the ability to stimulate bone regeneration when being seeded with osteoblast-like cells [34,37]. In another study, a PCL/HA SLS printed scaffold has been found to promote bone regeneration more favorably than pure PCL . The PCL matrix was in fact replaced by fibrous tissue and new bone.
BJ is another AM process that builds upon a powder bed principle. In contrast to PBF, BJ selectively deposits a liquid binder on the powder bed to bind particles together to form layers. BJ has similar advantages to PBF, but faster process time. However, BJ requires a post-printing sintering process to densify the parts, and the availability of polymer powder for this process is often limited .
2.1.3 Vat Photopolymerization.
Vat photopolymerization (VP) is a process in which a liquid resin or photopolymer is selectively cured by light (UV mostly, but some photopolymers react with a specific bandwidth of the visible spectrum) or radiation . The process requires, apart from a light source, a photoinitiator to generate a free radical upon absorbing light, either a reactive monomer or oligomer and a reactive diluent to control viscosity in some instances.
VP has a higher special resolution than ME and PBF, ranging widely from 0.1 to 100 µm [17,40], making it ideal for manufacturing highly complex polymer scaffold architectures. From the material usage point of view, the unspent resin can be reused rather than recycled as in PBF processes, but parts printed via SLA show lower mechanical stability and durability than their PBF counterparts, so some post-curing steps are to be undertaken in order to increase their mechanical properties . Biocompatibility issues (cytotoxicity) have been widely reported for both photoinitiators and photopolymer with zebrafish essays [41,42]. Washing the printed part or using more natural photoinitiators can improve biocompatibility .
Under this category, there are also processes termed stereolithography (SLA), digital light processing (DLP), and two-photon photopolymerization (2PP). Variations of the SLA process include (1) scanning-based SLA, where a laser essentially cures a layer one “pixel” at a time, and (2) mask projection-based SLA, where one entire layer can be cured all at once . In DLP, which is a variation of mask projection-based SLA, the parts are printed upside-down as the building platform is raised instead of lowered. The Bioglass®/acrylate scaffolds printed with this method have shown a high resolution and comparable flexural strength to the traditional Bioglass® scaffolds . However, post-printing sintering processes need to be optimized in order to yield higher compressive strength. Advances in VP technology have resulted in yet another process, Continuous Liquid Interface Production (CLIP). This process is very fast, with a printing rate reported to be up to hundreds of mm/h  (versus a printing speed below 10 mm/h for SLA ().
Materials typically used for VP are acrylates, epoxides, and other blended polymers. Ceramic or carbon fibers can be added to improve the final part's mechanical properties . An advantage of VP over ME and PBF is its ability to orient reinforcing material during the printing process. One study used a novel magnetic DLP technique to print acrylate-urethane samples with specifically oriented alumina particles mimicking architectures found in nature, such as the concentric architecture found in mammalian osteon cells as shown in Fig. 2 . They also reproduced the layered nacreous shell of abalones and the cholesteric architecture of the dactyl club found in mantis shrimps. These architectures increase tensile properties and hardness compared to simply aligned or nonaligned samples. There exists the opportunity to tune the fracture behavior of the part according to fibers’ alignment. An electrically assisted SLA was used to orient carbon nanotube (CNT) in polymer meniscus, resulting in superior mechanical properties compared to native menisci . The CNT were aligned in the same way as the collagen fibers in a native meniscus and the tailored artificial meniscus with CNT showed increased radial and circumferential moduli. A few examples of bio-applications of polymers used in SLS can be found in Table 1.
2.2 Additive Manufacturing for Metallic Parts and Devices.
In medical applications where polymer or polymer composite materials cannot fulfill the stringent mechanical property requirement, metallic parts/devices (stainless steels, cobalt-based, and titanium-based alloys) are considered. The main metal AM processes are powder bed fusion (PBF), direct energy deposition (DED), and SL. Other processes, such as CS, BJ, and bound metal deposition (BMD), are from time to time added to the list [3,15]. A controlled atmosphere is often needed to print parts for medical applications, either inert gas or vacuum, to prevent oxidation and contamination of the material [15,53]. Figure 3 outlines the most common metal AM technologies used in medical applications.
2.2.1 Powder Bed Systems.
Powder bed system in metal AM follows the same principles as described in 2.1.2 for polymers. The power source can be a laser or electron beam, single or multiple beams [55,56]. Advantage PBF has over other metal AM processes is that PBF provides better resolution and finer surface finish, and it allows for more complicated architectures than those provided with the DED and SL technologies . The resolution of metal PBF technologies ranges from 50 to 100 µm . However, there is often a need to carry out sintering, HIP and residual stress relief treatment as part of the post-process procedures . Fully dense commercially pure titanium parts have demonstrated greater mechanical properties than traditionally manufactured parts . As with all other metal manufacturing processes, the mechanical properties of the final part are highly dependent on the printing parameters  and other associated processing steps.
In addition to fully melting or sintering of the metal particles in the powder bed, there are also “indirect” PBF processes which include a polymer-coated metallic powder (or a polymer-metal powder mix), and during the AM process, polymer is melted to bind the metal particles together forming a green part. A debinding and a sintering processes remove the polymer and coalesce the metal particles while at the same time a significant shrinkage is resulted .
Essentially any metal that is weldable can be used for PBF; however, common metals used in medical devices are restricted to titanium, stainless steels, and cobalt-chromium alloys due to their proven biocompatibility . Titanium (and titanium alloys) is being extensively used in the medical industry because of its high specific strength, fatigue resistance, biocompatibility, and bone ingrowth performance [6,15,35]. CoCr cardiovascular stents have been made with a PBF process, which have demonstrated promising clinical results, providing future opportunities for AM of cardiovascular devices . Biodegradable magnesium scaffolds were produced with PBF for use as a biodegradable implant; they showed encouraging results when being tested in rabbits . However, magnesium is known to produce hydrogen gas during corrosion such that the degradation process must be tailored through material and process modifications so that hydrogen production is reduced .
BJ is a process that selectively deposits binder droplets on metal powder; it has the advantage of being a non-heated process, hence no residual stresses from the printing process . However, a post-process treatment is needed to consolidate metal parts, which adds considerable time to the production and also leads to shrinkage. Lastly, BJ has the potential to be much faster than other powder bed systems as multiple printheads can be used to print large volumes quickly .
2.2.2 Directed Energy Deposition.
DED is a process in which metal powder or wire is fed directly into a molten metal pool created by lasers, electron beam, plasma arc, or electric arc [15,53]. While DED does not render the same resolution and accuracy as PBF because of the nature of melting and solidification of droplets, it is more suitable for larger parts and can deliver higher deposition rates . Moreover, it is also suitable for the repair or remanufacture of a damaged part .
Another key advantage of DED over PBF is that it is one of the AM processes allowing for multiple metal depositions within the same build . Homogeneous or heterogeneous parts can easily be built in a single processing step. It can also be used to coat devices with specific material and thickness to improve their durability and lifetime, or add features or material to an existing part to improve its characteristics . However, the support structure is needed to build overhanging architectures, such that the complexity of the final part is not as high as that of PBF. Multiaxial deposition heads and improved path control will soon allow for the fabrication of complex structures. In a recent study reported by LENS (a DED process), various Ti6Al4V scaffolds were manufactured to assume consistent porosity but different elastic modulus by varying the pore geometry from spherical to irregular . This was achieved by varying printing deposition parameters, a procedure not easily achievable with PBF processes. Lastly, DED fabricated parts need stress relief and final machining to obtain the desired microstructure and dimensions.
2.2.3 Other Systems.
One of the other AM processes capable of producing porous scaffolds for TE is a direct writing (DW) process, similar to ME, but without heating during the printing process . Inks (mix of solvent and metal powder) of high viscosity are deposited in the same manner as ME. Post-AM sintering consolidates the parts and gives the parts their mechanical properties. A recent work demonstrated the capability of DW to produce scaffolds with high compressive strength and good cell viability (Fig. 4). However, a shrinkage upon sintering was found to be near 40 vol%, as well as inferior dimensional accuracy to other AM processes for metals .
CS is a low-temperature process that has seen increasing interest in the biomedical field. Originally developed as a coating technology, it has recently found application in AM [64–66]. The process involves gas atomized particles accelerating in a specially designed nozzle (converge and then diverge) with an inert gas. When these supersonic particles collide with the substrate or a previously deposited layer, plastic deformation occurs resulting in the interlocking of the particles. An advantage of this technique is that the printed part or coating layer assumes similar properties as the feedstock as it is not substantially heated [15,66]. In the biomedical field, however, applications are still limited to the functionalization of surfaces, such as coating and fabrication of nanoporous architecture to increase bioactivity and osseointegration of an implant . Table 2 outlines the advantages and disadvantages of common metal AM technologies in the biomedical industry.
2.3 Printed Biomaterials.
Three-dimensional (3D) printed biomaterials are gaining increasing interests as they have the potential to offer sophisticated pattern/architecture/material combinations for research in metabolism, toxicity, or testing drugs. Highly complex scaffolds, sometimes embedded with cells, allow for native cells to populate and replace the scaffold. Using ME processes, human tissues can be printed using, for example, PCL as the matrix and a composite cell-laden hydrogel composed of gelatin, fibrinogen, hyaluronic acid, and glycerol . Biomaterials can be printed via 3 notable processes: direct-write printing (DW), ME, and SLA .
In DW printing, a nozzle deposits liquid droplets or a continuous layer using thermal or acoustic forces to form the desired pattern . After deposition, however, some inks must be cured or crosslinked to achieve the desired stiffness. Since only inks of low viscosity can be used in DW, the resolution is limited but allows for rapid printing of skin, cartilage, bone, and even blood vessels . The ME and SLA processes use the same principles as described in 3.1, but the materials and process parameters differ from DW to enable cell viability and attachment [69,70]. These processes can also mimic cell architectures found in organs, such as livers and kidneys, making them a promising direction for further studies [71,72]. The most challenging tasks in bioprinting arise from the shear stress induced on the cells during deposition with DW and ME processes and laser-induced damage in SLA as these can decrease cell viability.
Materials used in bioinks for DW process are often comprised of cell-laden hydrogels and cell slurries. Hydrogels for scaffold-based are commonly derived from natural and synthetic polymers, such as alginate, gelatin, fibrin, chitosan, or polyethylene glycol (PEG) . For scaffold-free bioinks, cells are deposited on a medium in such a way that they cohere together and form the desired structure . Bioprinting is considered an emerging technology but has mostly been used in vitro, such that there is a lack of in vivo data .
2.3.1 Further Areas of Research in Bioprinting.
Successfully developing 3D-printed biostructures with the potential to increasingly replace or repair naturally occurring structures such as bones, soft tissues, or complete organs requires an increase in multidisciplinary research collaboration along with continued research in several specific focus areas.
Producing functional tissues and their complex biostructures using 3D-printing is a multimaterial, multi-process challenge requiring the use of ever more multidisciplinary research teams. More than ever, engineers, chemists, biologists, medical staff, regulators, and patients must work together to develop highly engineered and targeted feed materials, improve processing technologies to enable varying architectures and compositions in the bioconstruct, repeatable and optimized process designs, and a clear streamlined path to certification/licensure to enable faster commercialization and use in patient care. Because of the very complex nature of human biology in general, along with the potential risk to patients and the high cost of failure, expertise in multiple areas is essential to move this field forward.
From the perspective of 3D-printing equipment and process capacities, while there has been significant progress made over the last decade, many limitations remain in terms of dimensional precision and the ability to render bioconstructs mimicking organs and tissues. Because many feedstocks are tailored for specific process/equipment, a technology dilemma is facing the field. Either materials have a high strength (metal or high strength polymer) but lack biocompatibility and biofunctionality or they have biocompatibility and biofunctionality but lack shape fidelity and durability. Therefore, research into innovative technologies and multi-scaled processes is needed to enable the design and production of scaffolds with different controlled architectures, such as pore geometry and density, while at the same time enabling the addition of biomaterials either into or onto these complex 3D structures.
Such innovation is illustrated in a recent publication where multi-channel 3D plotted osteochondral plugs incorporate a cell-laden hydrogel and a partly mineralized calcium phosphate cement resembling articular cartilage on a support layer of interconnected porous calcified cartilage and subchondral bone . Similarly, it is conceivable that an AM process for bone implant production could be developed using multiple processes and materials. For example, a porous multi-compartment of metallic, ceramic, or polymer biocompatible scaffold structure could be printed first using a process such as SLS, SLA, or DMD. This structure could then be postprocessed to stabilize the matrix for cell adhesion and proliferation. Subsequently, a bioprinting process could be used to incorporate cells, fluids, or other biomaterials to biofunctionalize the scaffold .
Combining materials and technologies, and in particular finding ways to fully integrate these into bio-AM production systems, is key to producing better structural and functional bioconstructs. In the case of bone and articular cartilage replacements where strength requirements are particularly important due to repeated loading, 3D-printed metal is often the choice for the scaffold due to its known fatigue resistance and multiaxial loading capability. With the combination of bioprinting, one can simultaneously incorporate ceramic material to mimic bone composition (e.g., HA or bioglass) into a porous metal scaffold using a dual material injection feed system. Indeed, a multi-compartment scaffold with the subsequent addition of cells has already demonstrated success in an in vitro test . Similarly, researchers have been working on printing extracellular matrix (ECM) with poor mechanical properties onto a polycaprolactone (PCL) frame . In another study, a powder bed AM process printing calcium polyphosphate bioceramic with polyvinyl alcohol was coupled with a UV-based syringe deposition technique that deposited sacrificial photopolymer into the bioceramic construct. Post-printing sintering was then used to create micro-sized channels in the bioceramic for improved bone and osteochondral tissue regeneration . With continued research focus in this area, a significant innovation that combines multiple technologies and processes to create biostructures that better replicate both structurally and functionally the biological tissues and organs they strive to replace can be expected in the coming years.
Similarly to finding new ways to combine AM materials and technologies, continued advances in bioprinting processes and the development of new bioinks are required. Among the bioprinting technologies, the inkjet printing method has the greatest long-term potential due to its ability to print living cells, growth solutions, and other biomaterials using bioinks derived from native tissues and organs (e.g., bone marrow and liver) . Although a 3D scaffold could conceivably be printed with inkjet technology, as reported in Ref. , as with all types of bioprinting it faces challenges in formulating the ink to possess simultaneously non-cytotoxicity, adequate viscosity, dimensional stability, and more importantly the environment for continuous cell growth . In a recent development, a multiple channeled AM method was developed to fabricate functional human tissues and organs in which a core-shell structure (blood vessel or multilevel fluidic channels made with hydrogel) and macro/micro encapsulation were combined to render the construct biochemically and biomechanically functional . However, the calcium–alginate hydrogels used in this technique were unable to produce long vascular networks since the bioprinted structure is inherently weak when subjected to stress.
It is clear that development of functional and durable materials in liquid printable form requires continued research into engineered composite feedstock materials combining both natural and synthetic bioinks to achieve better structural and biofunctional properties. It also requires the development of bioinks with live cells capable of surviving UV light during the crosslinking used in some processing technologies. The incorporation of mechanical reinforcement to hydrogels through secondary networks offers a way to strengthen hydrogels, an essential component in encapsulating cells during printing . Bioinks based on engineered proteins with defined biochemical and biophysical properties such as silk fibroin is one such option . Bioglass or calcium phosphate can also be added to hydrogels to increase bioactivity and mechanical properties. In vascular engineering, grafts need to exhibit both the biological and mechanical properties of native arteries, demanding materials development at the nano-chemistry and nano-composite material levels combined with an understanding of the fatigue behavior of manmade biomaterials that cannot self-repair . Research into smart transformable and biocompatible materials that change properties and geometry under stress or certain chemical and biochemical conditions shows great promise to address many of the current biomaterial shortcomings. Similarly, engineering new biomaterials should include a sharper focus on developing those with the greatest immune privilege potential . Finally, perhaps the greatest challenge of all is developing AM biomaterials with self-healing capabilities using encapsulated cells in the AM construct, the achievement of which will close the gap between printed tissues and organs and the living ones they aim to replace.
While most of the research into AM biostructures is focused on the development and production of “replacement parts” that can be later implanted into patients, in situ tissue or organ repair  is considered by many to be a better long-term goal, whereas in vivo AM would become an integral part of the surgical operation. Such technologies, leveraging the patient's own tissues and organs needing repair rather than replacement, would allow for the use of scaffold-free (no support material) printing methods, eliminating many of the current disadvantages associated with the use of bioinks. For example, thin collagen films with corneal mesenchymal stem cells could perhaps one day be printed in situ to offer functional corneal repairs. Bioprinting has also been proposed to apply tissue directly to wounds for future treatment in space .
With continued research and the simultaneous development of AM technologies and biocompatible and biofunctional materials, it is evident that engineered tissues and organs will become more complex in structure and composition, approaching that of living tissues and organs. As the technology matures, it is within reason to expect that full tissue and organ replacement production, as well as in situ repair in the clinic, will become a reality. Additive manufacturing is well-positioned to improve medical treatment outcomes, improving both quality of life and life expectancy for many. The biological and engineering challenges in this emerging field are daunting, but the possibilities associated with AM bioprinting are truly limitless.
3 Devices and Implants Enabled by Additive Manufacturing
Due to the versatility of AM technology, a broad range of medical devices can now be manufactured with a high complexity in geometry and a freedom of combining different materials as homogenous composites or in a layered configuration. Biomedical engineers and researchers around the world are devoting a great amount of effort to create medical devices that have the potential to improve treatments and surgery outcomes. In this chapter, we will summarize several key breakthroughs.
3.1 Custom-Made Implants and Surgical Devices With Targeted Performance
3.1.1 Patient-Specific Implants.
Printing patient-specific bone implants becomes vital when the size of stock hardware misfits the patient's anatomy, increasing the risk of nerve damage and eventual implant failure . It is particularly important during craniomaxillofacial surgery, where geometry is complex but also critical to restore a patient's facial symmetry and restore anatomical aesthetics. These types of implants are usually designed using computed tomography (CT) imagery and Boolean subtraction so that the implants match the geometry of the healthy side of the patient's face . Additionally, patient-specific implants can eliminate or reduce post-surgery adjustment and remodeling . Several case studies (not all have FDA-approved devices) highlight the importance and the success of AM in the medical industry.
The most commonly FDA-approved AM medical devices on the market are dental devices, including dentures, retainers, and surgical guides. For instance, Envisiotec (Germany) is registered with FDA and has commercialized FDA-approved patient-specific dental guides and denture materials . Formlabs (USA), another technology supplier, use SLS and FDA-approved SLA technologies to produce patient-specific dental devices . Formlabs systems have in fact been used by many dental industry leaders, such as Exocad, Fullcontour, Digital Smile Design, Open Implants, and 3 Shape. Another leader in the medical AM industry is EOS, the company where direct metal laser solidification technology (DMLS®) originated . EOS has two metal PBF printers which have been awarded the European certification (MDD 93/42/EEC) to manufacture cobalt chrome dental devices (medical class IIa) such as dental crowns. However, their medical successes do not stop at dental devices. EOS also produces surgical tools, and patient-specific highly porous and permeable cranial implants that allow for greater cell attachment and bone growth .
Another big player in the medical AM industry is General Electric Additive. GE Additive has received FDA clearance in 2010 and CFDA clearance in 2015 for the AM of orthopedic implants. It has since produced over 100,000 hip implants . GE also has developed an ultrasound system, Voluson E10, with built-in 3D-printing capability. Using ultrasonic images, a fetus model can be printed and visualized by doctors and parents. This system helps doctors and parents understand fetal development and congenital defects if any . Recently, GE partnered with Formlabs in an effort to promote and expand the development of 3D-printing in the medical industry .
Lastly, a British leader in medical AM technologies is Renishaw. Renishaw alone or in association with FDA-registered Croom Precision Medical (CPM, Ireland)  and using their PBF systems and associated software, several success case studies on lifesaving custom implants have been reported. These include a ribcage prosthesis, a dog maxilla prosthesis, and hip and face reconstruction implants, as shown in Fig. 5 [97,98].
3.1.2 Surgical Tools and Diagnostics.
AM has also been used in the manufacture of surgical guides. AM can significantly reduce the time required to manufacture customized cutting guides thanks to desktop-type printers and open-source software [12,100]. Furthermore, orthoses and braces manufactured with AM technologies have shown promising function, aesthetics, and comfort improvements [101–103]. Examples of manufacturers using AM to produce patient-specific implants are the Belgian business Materialise and the German EOS. Materialise manufactures patient-specific implants and surgical guides, from which some have been granted FDA 510(k) clearance. For instance, Materialise TRUMATCH® maxillofacial implants have received FDA clearance for distribution in the United States, along with several surgical guides as well as knee and hip implants as shown in Fig. 6 .
Another important example where precision is critical is for surgery on the central nervous system. In a case study by EOS, laser sintering of polyamide was used to fabricate precision fixtures for Deep Brain Stimulation (DBS) surgery . Combining the information from MRI and CT, a FORMIGA P100 system was able to complete the fixture within 48 h, which is significantly shorter than turnaround times for traditional tools.
Finally, AM is being used increasingly as a means of surgery rehearsal and diagnostics as it allows for exact reconstruction of a patient's anatomy, especially for organs subject to unpredictable anatomy or complicated cases. Surgeries were accomplished more efficiently after surgeons could see the problematic organ from all angles on a 3D-printed replica [106–108]. Materializing has since obtained its FDA-approved printing software for diagnostics Mimic® . To date, eight AM systems have been certified for point-of-care 3D-printing with Mimic®  and are used in 113 facilities in USA/Canada, 48 in EU, and 34 in Japan , illustrating the ever-increasing presence of AM technology in the medical field. Anatomical models as shown in Fig. 7, allows for better understanding of the dynamics within an organ and between a possible anomaly and the surrounding tissues. Since 3D-printed models provide surgeons with a clearer idea of the required procedures, the actual operation time can be significantly reduced, leading to increased time and money savings, as well as reduced stress for the patient and better recovery.
3.1.3 Targeted Structure.
Thanks to AM, various complicated architectures can be built in a single step. In TE, pore size and interconnectivity are critical for fluid exchange, waste transport, as well as cell attachment, growth, and propagation. Small pores allow for good cell adhesion, but larger and interconnected pores are needed for vascularization, liquid transport, and tissue ingrowth. Pore sizes of 100–400 µm were found to be beneficial for mature bone formation , while 50–200 µm would be better for smooth muscle growth [112,113]. Moreover, a pore size of 215 µm was found to be superior to 320 and 515 µm in cell proliferation and extracellular matrix of mesenchymal stem cell production in a meniscus scaffold . In bone TE, more specifically, implant strength and stiffness present challenges in that if an implant is stiffer than the native bone, stress shielding will occur resulting in bone resorption, loosening of the implant, and subsequently implant failure . With AM, it is now possible to produce scaffolds with controlled interconnected porous structures that allow efficient stress distribution across the scaffold as well as variable modulus of elasticity in the implants based on the shape and the size of the pores  . Surface functionalization is also enabled by AM; by adding nanosized structures or bioactive coatings of controlled porosity, cell attachment and viability are improved. Antibacterial coatings have been developed as well to reduce post-surgery infections [65,116]. With AM, implants now have the potential to allow better bone growth and implant attachment and to reduce the incidence of implant failure.
3.2 Drug Release and Drug Testing.
Because of the possibility to build scaffolds with controlled pore size, multiple materials, and varied structures, AM is now being studied for drug-delivery TE. Examples include anti-cancer drug-loaded scaffolds; the results demonstrated a sustained drug release, a promising option for intra-tumoral cancer treatment . By mixing Fe3O4 nanoparticles in the matrix, the same scaffold could generate heat through the superparamagnetism effect, which would assist cancer treatment through local hyperthermia. AM-fabricated scaffolds in TE can combine multiple functions, such as tissue repair, drug delivery, and heat production.
Transdermal drug delivery is a painless technique that already exists among a few drugs, such as nicotine patches, but those drugs are limited to small-weight molecules as the stratum corneum (outer layer of the skin) prevents the transport of larger molecules, such as protein, nucleic acids, and large hydrophobic molecules . Microneedles are devices that allow transdermal delivery of larger molecules through the skin and are benefited the AM technologies as they possess the ability to control the size, shape, and composition of the needles for the purpose of delivering drugs . A variety of microneedle shapes are shown in Fig. 8. Researches have been carried out on the microneedle array fabrication and characterization, but none have been approved by the FDA so far.
AM also offers new opportunities to manufacture more effective tablets. To date, though, only one 3D-printed tablet has been approved by FDA, Spritam®, a treatment for certain types of epilepsy . Since the AM process does not rely on compressing powder materials, the tablets can be made to be highly porous and dissolve rapidly, e.g., 1000 mg of the drug can be released within seconds of its ingestion. The many possibilities brought by AM to 3D-printing of tablets also include combining treatments in a single tablet with appropriate doses, controlling the drug release to optimize its effect by adjusting the tablet structure, and limiting waste if printing at the point of care .
Better physiological models for drug testing can be achieved with AM and bioprinting. Organovo, USA, is involved in bioprinting and has developed the NovoGen Bioprinter® platform, an extrusion-based bioprinter. This technology is used to create tissues with precise cellular layering that mimics human tissue architectures for drug testing and other metabolic studies . Organovo's technology can be scaled up to a capacity of 6000 tissues per month, rendering better vehicles in drug testing than using live animals.
Moreover, bioprinting technology has become so important and useful that the FDA has been using EnvisionTEC's 3D bioplotter for research and regulation purposes since 2016 .
3.3 Microfluidic Devices.
Microfluidic devices, also known as lab-on-a-chip, organ-on-a-chip, or physiome-on-a-chip when multiple chips are interconnected, are revolutionizing biomedicine. Such chips as developed by the Wyss Institute of Harvard University (Fig. 10) containing microchannels of different lengths and sizes of human cells can mimic the physiochemical environments found in human tissues better than what could be achieved in Petri dishes or in animals . Because of their microscopic size, microfluidic devices are hard to manufacture using traditional processes. With the advent of AM, they can now be fabricated with transparent materials embedded with cells. Other features can be embedded as well thanks to AM. These include pressurized channels and chambers allowing the application of controlled mechanical stress on the chip's epithelium, similar to what happens in peristalsis or breathing. The controlled, rhythmic stresses allow for a close representation of the cells' environment and produce the desired phenotype . These AM-manufactured biological models help to better understand the complex interactions between different organs, and organs and the immune systems . It is also made possible to study specific pathologies or patient-specific interactions using cells collected directly from the patient . Studies have observed that using microfluidic devices strong correlations between the chip's “blood levels” of nicotine and clinical data were established [123,126]. The microfluidic devices are believed to have the potential to increase the efficiency in drug trials as they can closely mimic human fluid exchanges and specific interactions between drugs and cells, and these interactions can be in turn modeled across different organs with the help of different chips. These changes can be automated, decreasing the biohazard risks when handling toxic material or pathogens .
3.4 Self-healing Capability.
Self-healing materials are currently being explored as they have great potential to increase the durability and longevity of a device. Self-healing, in a biomimetic way, offers the potential to repair itself in situ by healing cracks and defects [127,128]. However, available self-healing materials are extremely limited, and exploitation of this ability is carried out in the construction industry with polymeric materials [127–130]. Self-healing in medical applications has not been reported, but because medical implants must assume a longer lifetime than traditional products, it is foreseeable that research in this direction will intensify as AM technology matures and new materials are being developed with time.
Another way to look at self-healing would be to use an implant in a way that would allow injured or malformed tissues to heal themselves where they would usually not. Examples of such applications are provided by the University of Michigan and Novopedics. Researchers from the University of Michigan and University Hospital have developed airway splints to treat children with tracheomalacia. They used Materialise's Mimic® software and EOS’ PBF technology to produce bioresorbable splints (Fig. 10), designed to be resorbed and replaced by the patient's tissues after a few years [131,132]. To date, five children have been saved by this treatment; the first child has healed completely, and the splint was completed and reabsorbed. The device had gained FDA's EUA clearance, but they are not yet widely commercialized.
Novopedics, USA, is using the EnvisionTEC's 3D-Bioplotter in their development of a biodegradable, implantable meniscus scaffold that would allow total meniscus replacement and prevent surgery-induced arthritis . They have obtained promising results in vivo trials on large animals (sheep) and developed a better second generation of the meniscus that was granted FDA Breakthrough Device Designation in September 2020 .
3.5. Four-Dimensional Printing.
4D Printing (4DP) refers to 3D-printed materials that can change shape and/or function over time or based on an environmental stimulus (Fig. 11) , similar to a protein that changes conformation according to pH level or upon bonding with a molecular signal. Mostly polymers in this category are hydrogels, shape memory polymers (SMP), and elastomers, and they are often mixed with fibers for the structure to gain the desired properties and shapes .
For instance, combining 4DP with SMP allows cardiovascular stents to be made with complex architecture and the ability to “remember” their previous shape. By “programming” the stent to have a smaller diameter before insertion, a minimally invasive surgery can take place; and the stent can be triggered by heat to achieve its designed shape once in place .
Another development termed “smart” stents employs PLA and Fe3O4 nanoparticles, making them magnetic-inducible . Upon stimulation in an alternating magnetic field, the compressed coil expands from 1 mm to 2.7 mm in diameter (Fig. 12) within 10s. This behavior makes it possible to re-expand contracted vessels due to clots, atherosclerosis, or other pathologies with minimally invasive surgeries. Moreover, the addition of magnetic particles could also permit the device to be remotely directed to its final position.
Drug delivery could also be benefited from 4DP. Using proper blends of rigid and flexible materials that can change shape according to temperature, pH, or water presence, along with origami concepts, it is possible to design and produce folding devices that can adopt the desired shape in a designated region of the body. For example, as in Fig. 13(a), 4DP biodegradable “theragrippers” with polypropylene fumarate (a rigid, hard polymer) and poly N-isopropylacrylamide-co-acrylic acid (a heat-responsive polymer) can fold and grip the gastrointestinal epithelium to release embedded drugs locally upon being heated to 37 °C .
Another research reported biodegradable polyvinyl alcohol (a polymer with water-induced shape memory) strip intended to be inserted in the bladder to release drugs for a prolongated period of time, shown in Fig. 13(b) . The strip was “programmed” to be straight such that it could be inserted via a catheter; then upon contact with water in the bladder, it would curl up to remain inside the bladder up to complete dissolution. Finally, 4DP enables other possibilities to create more localized and precise drug delivery, therefore reducing side effects and decreasing the drug dose needed to attain the prescribed physiological response to treatment.
In conclusion, while not all implants need to be made patient-specific, the advantage of AM over traditional manufacturing methods is the single-step production of porous implants encouraging cell attachment and bone growth. As described in the previous sections, AM has been proven to produce better implants without the need for subsequent steps to add coatings and modify the shape or structure of the implant. Furthermore, research on AM of medical devices and market openings are enabled by partnerships between technology suppliers such as SLM Solutions (Germany) and Canwell Medical (China) . SLM Solutions is based in Germany and specializes in selective laser melting (SLM) technologies. They have a wide array of AM implants, from acetabular cups to knee replacements, to dental prosthetics and surgical instruments.
4 Specifics for Medical Devices
Medical devices are highly regulated in order to ensure public safety. This applies to a wide range of devices, from simple tongue depressors and needles to artificial tissues and implants. Medical device development typically follows five phases [11,140], as shown in Fig. 14.
These steps are in fact iterative and overlapping in the sense that the devices must be kept up to date and conform to the most recent regulations. Since AM technologies are evolving rapidly, processes must be re-evaluated regularly to ensure that the devices are always in the most updated states. The implementation of design controls in the development process allows the inclusion of feedback and reviews in all steps of the development of the device.
Medical devices are highly regulated by federal agencies with specific rules in each country; among which the FDA is considered to be the strictest medical device regulatory agency. Several agencies, including the FDA, Health Canada, and those from Europe, China, and other countries, have come together to accelerate and harmonize the regulation process of medical devices in an International Medical Device Regulators Forum (IMDRF) . The IMDRF issues guiding documents that are then reviewed and often implemented in each country with regard to their specific regulations.
The following sections give an overview of the FDA regulation according to the five steps in medical device development.
4.1 Initiating Development, Opportunity, and Risk Analysis.
The first step of medical device development is a need or problem identified and documented. Planning of the research and development of the device is then initiated, along with research of the current market to find if the device has already an equivalent already in place, followed by necessary risk analysis . In this step of the development of the medical device, the laboratory, establishment, or organization must register with the FDA, designate a US agent if the organization is from another country, and pay the annual registration fee [143,144].
It is important for the organization to obtain an overview of and understand the FDA regulations and to find the appropriate classification for the device so that the required premarketing application type is determined . For instance, the FDA has categorized medical devices into three classes depending on the risk imposed on the patient and the degree of control required to ensure device safety and effectiveness. The classification depends on the device itself and the intended use as well as the indications for use. For example, the indication for use of a scalpel is “to cut tissues.” Subcategories of this indication can be identified for more specialized devices, for example, “making incisions in the cornea.” The classes are briefly described in Table 3 . The FDA Device Classification Panels further regroup most medical devices into 16 panels, such as anesthesiology, cardiovascular, dental, or neurology devices . It is critical for the organization developing a new device to identify similar, already marketed devices and compare them with that still in development.
4.2 Financial Feasibility, Prototype, and Concept Formulation.
In this stage, a proof of concept (POC) model will be built, prior to prototype production. The funding for the development, trials, costs, and expected revenues will also be assessed. Funding usually comes from sponsors, research grants, or associations formed between technology suppliers and design companies to develop the device jointly.
When designing the POC model, it is important to consider the basic requirements of medical devices: compatibility, mechanical properties, and manufacturing . In terms of compatibility, obviously biocompatible materials must be used but the device itself must be designed to be biocompatible, and care must be taken when specifying a material for use in a specific region of the body. For mechanical properties, the designer must choose materials with appropriate performances to fulfill their purpose, such as appropriate stiffness, fatigue, and creep resistance  in addition to other chemical properties.
Subsequently, the manufacturing process will be designed. It is crucial for a medical device designer to understand how a part can be manufactured as manufacturing has a direct effect on mechanical and surface properties, as well as sterilizability of the device . To follow the FDA safety and quality requirements, the organization must establish and follow Quality Systems Regulation (QSR), also known within the FDA as Current Good Manufacturing Practice (CGMP) . Because of the variety of existing and in-development medical devices, the CGMP are flexible, acting as guidelines to help organizations develop and follow a quality system framework. To ensure that the quality systems are consistent with international standards, the FDA's CGMP is created to be in accordance with the International Organization for Standards (ISO) 9001:1994 “Quality Systems—Model for Quality Assurance in Design, Development, Production, Installation, and Servicing” and ISO/CD 13485 “Quality Systems—Medical Devices—Supplementary Requirements to ISO 9001”. Device-specific guidelines are also stipulated at this stage.
Within the QSR, it is necessary to validate the manufacturing process, monitor, and control each step, ensuring that the device will function as intended per regulation 21 CFR 820.75 (a) and (b) (QSR—process validation). Because it is anticipated that a given device within a batch will eventually fail during quality testing, it is important to establish a set of rejection criteria before testing and to apply at the same time statistical analysis of the manufacturing process to avoid bias. Test methods must also be validated per 820.72(a) (QSR—Inspection, Measuring, and Test Equipment) and 820.250(a) (QSR—Statistical Techniques).
When the QSR is put in place, the prototypes can then be built. Manufacturing process, testing, and characterization of the parts associated with the particular device must be well documented as this information will be examined by the FDA upon submission of the device. Materials must be chosen so they are biocompatible according to ISO-10993 (Biological Evaluation of Medical Devices Part 1: Evaluation and Testing within a Risk Management Process).
4.3 Validation and Verification of Design and Prototype.
The step following the POC stage is to build a functional prototype and test it. Once the working prototype is built, it is necessary to assess its performance in order to ensure that they meet the safety requirements. ASTM International has established standardized tests for implants, which include shear, bending and fatigue testing for a number of materials .
It is also essential to test the prototype in an environment simulating the physiological conditions of the application environment to characterize its biocompatibility. Microfluidic devices and 3D-printed tissues (see chapter 4) are advantageous in drug testing, and in vivo animal models are often used to test drugs and the biocompatibility of the implant materials. For implants, nonclinical testing is carried out using in vitro, in vivo, and/or in situ human or animal tissues and human cadavers .
For AM custom or patient-fitted implants, the effectiveness of the fit is tested using cadaveric specimens, where an implant is produced specifically for the test subject from CT data and installed to confirm the fit and trial the implantation of the device. As mentioned previously, performance and rejection criteria must be identified and documented.
4.4 Final Validation Before Launch of the Product, Tested and Approved by Competent Authority.
Once the QRS is validated and the prototype characterized, refined, and deemed ready for commercialization, the organization must submit a premarket notification 510(k) or premarket approval (PMA) to the FDA. These are lengthy processes in which scientific documentation and clear communications are vital. All new devices and those that do not fall within Class I or Class II are considered Class III until sufficient information is made available to allow for reclassification of the device to Class I or II.
The 510(k) shall be made to the FDA “to demonstrate that the device to be marketed is as safe and effective, that is, substantially equivalent, to a legally marketed device (Sec. 513(i)(1)(A) FD&C Act). And the device is compared to said marketed devices with significant evidence to support the claim” . Upon “clearing” the device, the FDA issues the organization a letter that declares the device to be substantially equivalent and that it can be marketed in the United States.
The FDA determines the safety and the performance of the device by reviewing the methods used in its characterization and results from the performance testing. Therefore, gathering precise data and documenting all steps of the production and testing is crucial for FDA clearing.
A device is said to be substantially equivalent to a legally marketed device if it :
has the same intended use as the predicate; and
has the same technological characteristics as the predicate; or
has the same intended use as the predicate; and
has different technological characteristics and does not raise different questions of safety and effectiveness; and
the information submitted to FDA demonstrates that the device is as safe and effective as the legally marketed device.
Any organization holding a 510(k) for a device must also carry design control documentation throughout the development process according to article 21 CFR 820.30 of the Code of Federal Regulations and make it available for FDA review. Any change in the manufacturing process or device specification must be reported, and this may be subjected to a new 510(k) submission.
4.4.2 Premarket Approval Application.
The PMA is a scientific review process for the assessment of the safety and effectiveness of the device . It is the strictest FDA review process and thus all new (not substantially equivalent to existing 510(k) cleared devices) or class III devices must be submitted through a PMA. Documentation must include technical sections containing non-clinical laboratory studies and clinical investigations data (additional permissions required for clinical trials of the device will be described in the next subsection).
The non-clinical laboratory studies must include all laboratory testing (biocompatibility, toxicology, animal, mechanical tests, etc.). All non-clinical evaluations must be performed according to the FDA's Good Laboratory Practices for Nonclinical Laboratory Studies. For better communication with the FDA, it is best if the PMA contains fully detailed formal reports including :
objective of the test;
data analysis plan;
test results; and
The clinical investigation section includes all clinical investigation data, namely protocols, safety data, adverse reactions, complications and failures, patient complaints, and statistical analyses, amongst other things. It also comprised a Certification of Compliance with the Requirements of Clinical Trials form (FDA-3674).
4.4.3 Investigational Device Exemption.
When the device is to be the subject of clinical studies in support of a PMA before commercialization, an Investigational Device Exemption (IDE) must be submitted to the FDA . The IDE allows the organization to collect device safety and effectiveness data on human subjects. Again, documenting the study and recording all data is required.
The informed consent from all patients must be obtained ahead of the study. An investigational plan must be fully detailed and approved by an Institutional Review Board (IRB). A convention of labeling of the device must clearly indicate that it is for investigational use only. The investigation process itself must be properly monitored. Lastly, the organization must maintain specified records and prepare full formal reports according to the requirements described in the previous subsections to the FDA and IRB.
4.4.4 Emergency Use Authorization.
Under life-threatening conditions, the FDA can issue, on an exceptional basis, an emergency use authorization (EUA) for a device with the potential of preventing, diagnosing, or treating the condition . For example, in the case of five children with tracheomalacia, thankss to the availability of FDA EUA, they were treated with specially 3D-printed bioresorbable airway splints . Additionally, ventilators and related devices to treat COVID-19 have benefited from EUA during the 2020 pandemic.
4.5 Product Launch and Post Launch Assessment.
Upon receiving the FDA approval for the marketing of the device, the device can legally be commercialized and distributed (in the United States). After being granted marketing approval (and during clinical trials throughout the development stage of the device), it is mandatory that each device has adequate labeling . The FDA defines a label as a display of written, printed, or graphic matter upon the immediate container of any article, and labeling as all labels and other written, printed, or graphic matter. Minimum labeling requirements include name and place of business, intended use, and adequate directions for usage. Some devices require further information depending on their classification.
The FDA may also require that some devices follow post-market surveillance should negative consequences occur to the patient upon failure of the devices, such as a permanent implant, a device for life-sustaining, or a device used in pediatric populations . Post-market surveillance submissions are formal reports which include a complete study of the device behavior in the patient population. However, in all cases, if a problem with the device is detected by the manufacturer, the importer, or the medical facility utilizing the device, the FDA must be informed.
4.6 Example of Food and Drug Administration Approval Pathway—Spine Fusion Cages.
Spine fusion devices are used to fuse vertebras, restore the stability of the spine and relieve compression of the nerves in the treatment of chronic back pain, radiculopathy, and other spine diseases. Bone graft (to treat traumatized bones or joint pains) spacer was introduced in the 1930s before it progressed to allogenic bone graft in the late 1940s [159,160]. However, bone graft was found to collapse or break due to insufficient mechanical strength, leading to the need for further surgical procedures . Healing difficulties were also frequently reported as donor's bone tended to heal much slower, leading to infection risks . Cages were introduced to alleviate these issues. The first modern polymer cage was invented by John W. Brantigan in the late 1980s. The carbon fiber-reinforced PEEK cage was designed to have better mechanical strength and weight-bearing capability than the bone graft, hence improved healing could be resulted [161,162]. A timeline of the device's regulatory pathway is shown in Fig. 15. Brantigan patented the device in 1988–1989. After building working prototypes, the implants were characterized using cadaver testing . Pullout and compression tests were conducted subsequently, and modes of failure were identified. Mechanical and material properties were thoroughly examined and documented, including objections from previous studies. Then, the device was implanted in Spanish goats for 2 years, before being tested in the patients who were followed up 2 years post-op .
The pre-clinical studies resulted in the success of the implant in terms of fusion levels and radiographic fusion. Subsequently, the device was subjected to IDE between 1991 and 1993 to test the safety and efficacy of the implant. Following the consistent clinical success and detailed documentation, the FDA approved the device in 1999 for commercialization in the United States and requested a 10-year follow-up from the investigators. It was shown that clinical success was achieved in 87.8% of the participants at the 10-year post-op and 93.9% of patient satisfaction was also reported. Independent reports also showed high levels of fusion and successful rehabilitation of patients .
It is to be mentioned that before the 10-year follow-up study of the original implant, the device was modified, and its design improved. Another IDE was conducted to collect more data and demonstrate that the updated implant was equally safe and efficient, and the updated device was marketed following the approval of a PMA supplement .
4.7 Differences in Medical Device Approval for Conventional and Additive Manufacturing Manufactured Devices.
Because premarket evaluations by the FDA require a complete description of the device and the verification of each step of the manufacturing process, as explained in the previous subsections, AM devices mostly follow a similar regulatory pathway as that of conventional, non-AM devices . In other words, the process and the manufactured device must be validated as one entity. However, further analysis of the process is required as part of the approval process due to the unique characteristics of AM manufacturing.
Given that the AM material's properties vary significantly depending on the system used, the process parameters, the raw material used, and the post-processing, process validation must be thoroughly documented to ensure all devices will function as intended. Also, since each set of printing parameters affects the results from subsequent processing steps, a clear understanding of the influence of each parameter must be demonstrated. Additionally, the impact of each process parameter on material's behavior must be identified and potential solutions found.
Since AM is an automated process, the software must also be validated in accordance with 21 CFR 820.70(i) (QSR—production and process controls).
To guarantee the quality of device, in-process monitoring is also required in order to validate an AM process. Parameters such as the built environment (inert gas or vacuum, for example), the temperature (of the parts or in the environment), and the power of the energy source should be monitored. Visual inspection and non-destructive evaluation are also to be implemented. Some AM printers are not able to monitor all parameters during the process. In that case, verification testing methods for process monitoring must be incorporated to guarantee quality.
Lastly, because some AM processes require the use of support material, methods to remove the support material should be developed and documented such that the function and safety of the device are not compromised during the removal of the support material.
4.7.2 Device Documentation.
One of the advantages of AM is its ability to make patient-specific devices with complex and intricate architecture. Therefore, AM devices must not have discrete features and sizes. For this reason, device documentation must include :
type of AM technology used;
range of dimensions of the device;
critical dimension and tolerances to match a patient's anatomy;
critical features (ex. porous scaffolding);
all materials involved (starting and final) including chemistry information;
o material name, Chemical Abstract Service (CAS) number, or recognized consensus material standard;
o specifications and certificates of analysis (COA); and
o characteristics (particle size, viscosity, composition, purity, etc.).
material reuse process and amount of material reuse (if any).
Most of the information regarding dimensions should also be found on the device's technical drawings.
AM devices do not differ much from conventional ones in terms of the protocol for mechanical testing. However, because of the nature of the AM process, they might present anisotropies specific to the AM process. It is recommended that when designing the devices to be manufactured with AM, the anisotropic nature of the AM material must be taken into consideration and the design be validated accordingly in this regard.
Finally, full sterilization may present challenges in reaching intricate geometries such as voids, channels, and porosities. The sterilization process must also be developed and validated, accounting for these features unique to AM.
In summary, AM devices should follow the same regulatory path as traditional medical devices; however, designers/manufacturers should include additional information to avoid any variability due to the manufacturing process and to guarantee the safety and efficiency of the device.
5 Concluding Remarks
Additive manufacturing is an immerging technology that allows the fabrication of medical devices with highly tailored architecture and controlled mechanical, chemical, and biological properties. There are many AM technologies available, each with its respective advantages and shortcomings.
In particular, 3D-printed patient-matched implants have shown better clinical success than the standard-sized counterparts. With AM, it is not only feasible to adapt an existing device with variable dimensions, but also powerful in its ability to create innovative implants, drugs, tissues, and organs. The freedom of design enabled by AM makes it possible to design porous scaffolds with outstanding osseointegration properties, along with resorbable features engineered to be replaced by autologous tissues.
As with all medical devices, the regulation process to market a medical device is lengthy and requires rigorous documentation; so far, the FDA has approved several AM devices and software, and the path is paved for companies and institutes involved in the AM of medical devices in their pursuit of FDA approval. It is envisioned that the list of FDA-approved devices will continue to grow as more development of AM medical devices comes to maturity.
However, because of the wide variabilities of AM systems, process parameters, and materials, AM is far from a standardized manufacturing process. Because of the stringent requirement of medical devices, an in-process testing/quality control will have to be incorporated into the AM system, such as real-time temperature, stress, and distortion monitoring and feedback control system. The relationship between the processes, feedstock materials, and the resulting performance of the devices must be thoroughly understood. And in this highly multidisciplinary field, continued collaborations among medical workers and engineers are indispensable.
We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC) for the financial support in the form of USRA (A. Bastin) and Discovery Grant (X. Huang). NSERC Acknowledgement Figure
Conflict of Interest
There are no conflicts of interest. This article does not include research in which human participants were involved. Informed consent is not applicable. This article does not include any research in which animal participants were involved.
Data Availability Statement
No data, models, or code were generated or used for this paper.