Point of care approaches to 3D bioprinting for wound healing applications

In the quest to improve both aesthetic and functional outcomes for patients, the clinical care of full-thickness cutaneous wounds has undergone significant development over the past decade. A shift from replacement to regeneration has prompted the development of skin substitute products, however, inaccurate replication of host tissue properties continues to stand in the way of realising the ultimate goal of scar-free healing. Advances in three-dimensional (3D) bioprinting and biomaterials used for tissue engineering have converged in recent years to present opportunities to progress this field. However, many of the proposed bioprinting strategies for wound healing involve lengthy in-vitro cell culture and construct maturation periods, employ complex deposition technologies, and lack credible point of care (POC) delivery protocols. In-situ bioprinting is an alternative strategy which can combat these challenges. In order to survive the journey to bedside, printing protocols must be curated, and biomaterials/cells selected which facilitate intraoperative delivery. In this review, the current status of in-situ 3D bioprinting systems for wound healing applications is discussed, highlighting the delivery methods employed, biomaterials/cellular components utilised and anticipated translational challenges. We believe that with the growth of collaborative networks between researchers, clinicians, commercial, ethical, and regulatory experts, in-situ 3D bioprinting has the potential to transform POC wound care treatment.


Introduction
Being the largest organ of the human body [1,2], skin plays an integral role in day-to-day life by regulating body temperature, providing sensations of touch, eliminating waste, and protecting internal tissues [3]. The importance of these functions becomes increasingly evident following full-thickness cutaneous injury, where, despite developments in clinical care, scar tissue formation remains a roadblock to the restoration of aesthetically and functionally equivalent skin tissue at the wound site. In order to improve patient outcomes, skin substitutes which closely mimic host tissue properties are required to facilitate the regeneration of skin in-situ. Three-dimensional bioprinting is a fabrication technology which offers high levels of control over the spatial arrangement of biomaterials and cells. An abundance of research into the utilisation of this technology for skin substitute fabrication exists, however, bioprinting systems are often designed in ways which limit their capacity for use at point of care (POC). Systems often entail ex-vivo fabrication of the skin substitute, coupled with lengthy maturation periods prior to application on the wound. Following this 'print-then-transplant' approach has several shortcomings; wound closure is delayed, clinical protocol complexity is increased, length of hospital stay is extended, and additional costs are incurred for patients. Manual handling of the construct during transfer to the wound site also limits material composition and heightens the risk of contamination and infection transmission. Further, tissue regeneration may be hindered by the fact that external bioreactors used during maturation fail to entirely mimic the native physiological environment [4,5]. It is these considerations which have served to motivate the development of bedside bioprinting systems capable of depositing skin substitute structures in-situ. In this review, the current status of in-situ 3D bioprinting systems for wound healing applications is discussed. A brief description of skin injury and current clinical practice is provided. This is followed by a summary of studies involving bedside bioprinting for the treatment of full-thickness wounds, with focus being placed on the proposed delivery method and biomaterials/cellular components utilised. Finally, translational challenges are discussed and future perspectives on this field are provided.

Skin structure and function
The skin is composed of three distinct layers; from superficial to deep these are the epidermis, dermis, and hypodermis (figure 1). The non-vascularised epidermis protects against the entry of pathogens, allergens, and toxic substances, prevents water loss, and is responsible for creating skin tone [6]. The primary cell of the epidermis is the keratinocyte (KC), responsible for providing the skin with its barrier function and allowing the selective transport of molecules to maintain homeostasis in-vivo [7]. Additionally, this epithelial tissue harbours melanocytes (which produce melanin pigment and protect against UV rays), Langerhans cells (which play a sentinel role in the immune response), and Merkel cells (slowly adapting mechanoreceptors) [1]. Sitting below, an anchoring basement membrane (BM) composed of proteins and adhesion complexes provides inter-layer adhesion and acts as a diffusion barrier between the epidermis and dermis. Beyond the BM is the dermis; a highly vascularised and innervated connective tissue which accounts for the majority of human skin [8]. An extracellular matrix (ECM) composed of ground substance (glycoproteins and proteoglycans), collagen, and elastin fibres [9] surrounds fibroblasts (FBs) (the fundamental cell of the dermis), as well as macrophages (which assist in immune responses and the wound healing process), and endothelial (EC), muscle and nerve cells (which serve a critical role in overall skin functionality). The hypodermis (or subcutaneous fat) is a loose, adipocyte-rich, connective tissue beneath the dermis which connects skin to the underlying fascia of bones and muscles and plays a role in thermoregulation, energy storage, and protection from injury [2,9].

The wound healing process
Being the first line of defence against physical, chemical, and biological injuries incurred by the external environment, the integrity of skin is often compromised, resulting in a wound. In Australia, leading causes of skin wounds include thermal injuries (burns), alongside diabetic, pressure and vascular ulcerations [2]. Such injuries are often classified according to percentage of total body surface area (TBSA) effected, depth of penetration (superficial, partial-thickness, or full-thickness), and their ability/inability to heal (acute/chronic). For the purpose of this review, focus has been placed on full-thickness wounds (those which destroy both the dermal and epidermal layers of skin) as their treatment serves to benefit the most from advances in 3D bioprinting.
Briefly, full-thickness wound healing begins with haemostasis of the wound, involving localised vasoconstriction and activation of the coagulation cascade to prevent excessive blood loss [10,11]. Following cessation of bleeding, immune cells infiltrate the wound site and signal for the release of pro-inflammatory cytokines and chemokines while working to destroy bacteria and remove debris [11]. In acute cases, immune cells eventually transition into anti-inflammatory phenotypes and facilitate granulation tissue deposition, neovascularisation, and the restoration of epithelial continuity [11]. Following wound closure, maturation of the site (characterised by the transformation of granulation tissue into scar tissue rich in collagen type I) occurs over several months or even years in an attempt to restore original skin tissue strength and structure [11].
However, it is scarring which acts as a roadblock to the restoration of functional and aesthetically-equivalent skin [2,11,12]. Scar tissue is inherently rigid and may lead to restricted joint movement [13], often hypo/hyper-pigmented [14], and characteristically different in composition to healthy dermal tissue which may result in the irregular regeneration of neural networks, sensory deficits, excessive itching, or chronic pain [15,16]. Such poor patient outcomes indicate the need for clinical intervention in the management of full-thickness wounds.

Current point of care (POC) practice
POC treatment of full-thickness wounds generally begins with the excision of devitalised tissue from the wound bed which can serve as a source of nutrients for pathogens, placing immunosuppressed patients at risk of infection [17]. This is followed by a wound closure strategy. Delayed closure has been shown to increase the risk of severe hypertrophic scarring, infection, and multiple organ dysfunction syndrome [17], hence timing is critical. For wound areas which are small enough (e.g. acute surgical wounds following planned incisions) primary closure is often performed via directly suturing appositional wound edges together. As the wound area increases, skin grafting or the application of a skin substitute product is required.

Skin grafting
Grafting refers to the process of surgically transplanting healthy skin onto the wound site to achieve wound closure. The transplanted skin can be strictly epidermal, split-thickness (including the epidermis and a portion of the dermis), or full-thickness (including the epidermis and dermis). Allogeneic and xenogeneic skin grafts can be harvested, cryopreserved, and vigorously tested prior to their use to minimise the risk of disease transmission. However, this approach is inherently limited by immune rejection [18,19]. Immunosuppressant strategies exist to delay this response; however, chronic graft rejection and constant monitoring of adverse side effects limit this wound closure strategy [20]. Therefore, allogeneic and xenogeneic grafts usually serve as a temporary means of wound coverage, guarding the underlying tissue from pathogen invasion while awaiting a means of permanent wound closure [21]. Unlike allografts and xenografts, autologous skin grafts do not pose a risk of immune rejection or disease transmission. The graft is able to form anastomoses with the surrounding endogenous capillary network to ensure its survival through sufficient nutrient delivery [17], while the donor site will either self-regenerate through gradual re-epithelialisation and KC migration (for split-thickness autografts) or undergo manual primary closure (for full-thickness autografts). Using an autologous split-thickness skin graft (STSG) is currently considered the gold standard treatment for full-thickness skin injuries [17,22]. Unfortunately, STSGs often result in a poor aesthetic match due to variation in skin colour and texture across the body and can incur painful donor sites which may introduce their own wound healing complications. Also, STSG take rate (ability of the graft to stay in place and survive) varies significantly, being between 70% and 90%; a success rate which further decreases with higher TBSA percentages [23]. STSGs also become impractical in cases where a patient suffers skin loss exceeding 60% TBSA and donor site availability is minimal [18]. Although graft size can be expanded through meshing techniques, this results in mesh-pattern scarring of the skin, greater graft contraction and delayed healing [17,19].

Skin substitute products
The limitations of skin grafting sparked a worldwide surge in the development of skin substitutes. A multitude of products have been approved for use at POC (table 1), causing a paradigm shift in full-thickness wound treatment over the past decade. Initial efforts were focused on the development of interim skin substitutes such as Biobrane ® (Smith and Nephew, Hull, England) and TransCyte™ (Advanced BioHealing Inc., California, USA) which could temporarily restore the barrier function of skin while waiting for autologous donor tissue to become available. As the criticality of donor site morbidity continued to be stressed by clinicians, this quickly transitioned towards the development of skin substitutes which could permanently integrate with host tissue such as Integra ® (Integra Life Sciences, New Jersey, USA) and Matriderm ® (MedSkin Solutions, Dr Suwelack AG, Billerbeck, Germany). However, the use of these products is hindered by the requirement for a secondary STSG procedure, delaying return to normal life and increasing the cost of treatment. Striving towards a one-step treatment and the eradication of grafting from POC practice altogether, bilayer skin substitutes such as Apligraf (Organogenesis, Massachusetts, USA) and Permaderm™ (Cambrex Corporation, New Jersey, USA) have emerged. However, their use in the clinic is often limited by high cost, short product shelf life, difficulties in handling, and long in-vitro culture times [17,24]. Also being explored is the combined delivery of dermal and epidermal skin substitute products already approved for use at POC. Gaining significant traction in the clinic, ReCell™ (Avita Medical Ltd, Cambridge, England) is an autologous cell harvesting, processing and delivery device that involves spraying non-cultured epidermal cells onto a wound site as soon as 30 min from biopsy harvesting [25]. Although not indicated for the treatment of full-thickness wounds in its standalone form, in-vivo studies showed that when ReCell™ was combined with Integra ® , the self-organisation of autologous cells accelerated re-epithelialisation of full-thickness wounds and mitigated the need for a STSG procedure [26,27]. To date, this integration of products is perhaps nearest to the vision of an immediate, single-stage, graft-free treatment strategy for full-thickness wounds.
However, the manual spray delivery of cells results in a non-homogeneous distribution which can potentially hinder tissue regeneration. Further, the dermal skin substitutes in table 1 are fabricated as pre-defined flat sheets [28]; requiring manual manipulation and handling in the operating theatre to ensure an adequate fit to the wound bed, increasing procedure time and contamination risks. Additionally, compared to grafts, these products do not wholly mimic the dermal ECM and, as a result, the formation of scar tissue remains a prominent challenge [29]. Inaccurate replication of the dermal ECM is a consequence of the techniques traditionally utilised for skin substitute fabrication: lyophilisation, casting, or electrospinning. These processes offer minimal control over pore shape and interconnectivity; factors which directly influence cell behaviour and the skin regeneration process in-vivo. This apparent lack of control over cell and biomaterial placement presents as a significant factor hindering the performance of skin substitutes today.

Harnessing bioprinting at POC for wound healing applications
Bioprinting can be considered the missing piece of this clinical puzzle, offering exquisite control over cell and biomaterial placement, while also facilitating a patient-centred approach to wound treatment. Bioprinting is a specialised subset of 3D printing; a technique developed by Charles Hull in 1984 which has evolved significantly over recent years and paved the way for many innovative medical treatments. The process aims to enable the tailored fabrication of functional, living 3D tissue constructs through the layer-by-layer deposition of biomaterial which is suitable for introduction into the human body. The biomaterial passed through the printing system is referred to as the ink and, through the addition of cells, can be transformed into a bioink to directly print 3D cell-laden constructs. In recent years, the clear-cut benefits of bioprinting skin substitute structures for wound healing purposes has driven an increase in research activity. However, the majority of proposed strategies to date have been built around ex-vivo delivery protocols which typically require lengthy in-vitro cell culture periods and the use of bioreactors for construct maturation (figure 2). Recognising the shortfalls of this approach, some have begun to focus on the development of POC bioprinting systems which can print in-situ, directly into a wound defect site. In this section, the bioprinting modality and clinical delivery method (table 2), and biomaterial(s)/cellular component(s) (table 5) utilised in these works will be discussed, focusing on the efficacy of deploying these systems at POC.

Bioprinting modalities
Bioprinting modalities utilised for skin substitute fabrication can be categorised into inkjet, laser-assisted, digital light processing (DLP), or extrusion-based techniques; each of which are summarised in table 3. Inkjet bioprinting uses a thermal or piezoelectric-driven mechanism to trigger the ejection of droplets toward a substrate to form 3D structures layer-by-layer. Compared to other bioprinting modalities, this technology is relatively fast and utilises inexpensive hardware. However, cells are at risk of high thermal and mechanical stress [18]. As the name suggests, laser-assisted bioprinting uses a pulsed laser beam to trigger the propulsion of microdroplets from an ink film layer toward a substrate. Although this offers high resolution and is a nozzle-free technology which eliminates stress on cells, it is slow and the printing hardware is relatively expensive [42]. DLP is a subset of stereolithography-based printing, in which a stationary light source and digital photomask are utilised to project dynamic 2D images towards a photo-curable ink reservoir. A traversing deposition stage allows layer-by-layer crosslinking of ink to build a 3D structure. This technique also mitigates the risk of stress-induced cell death however limits construct composition to a singular photo-curable ink [42]. Extrusion-based technologies use either mechanical or pneumatic actuation to extrude continuous ink filaments onto a substrate. Resolution capabilities are generally inferior compared to the aforementioned bioprinting methods, however, affordable printing hardware and the ability to handle high viscosity materials has made it the most widely employed technique for bioprinting with cells [42,43]. Although DLP and laser-assisted systems have made their debut as prospective in-situ bioprinting tools to regenerate cartilage [45] and bone [46], it is inkjet [40,47,48] and extrusion-based techniques [8,[36][37][38][39]41] that have been investigated to date as potential bedside strategies for in-situ skin substitute delivery. This is likely due to the ease of upscaling to increase volumetric output and cater for larger wound areas. The compatibility of extrusion-based systems with a much wider suite of higher viscosity biomaterials is likely the driving force behind its use in the majority of works in table 2. Further, extrusion bioprinters are simple, akin to the use of a syringe and needle in regular clinical practice, are able to integrate with existing in-situ delivery systems in the clinic (e.g. minimally invasive surgery tools), and are easily miniaturised to become handheld devices [49]. This is not to say a POC bioprinting system should be limited to a single bioprinting modality. In fact, the distinct anatomical variation between skin layers suggests that a multimodal bioprinting system may be able to fabricate skin substitutes which more closely resemble this complex stratified tissue. One of the reviewed works combined extrusion bioprinting with spraying and electrospinning techniques in a customised bioprinting device [37]. The protocol entailed preparatory spraying of a phantom wound bed, followed by extrusion of a dermal matrix material and subsequent electrospinning of an outer epidermal membrane layer. This multimodal approach, however, comes at the cost of increased protocol complexity and knowledge requirements from the clinical operator at POC.

Clinical delivery methods
In-situ skin bioprinting can be performed in one of two ways: either by using a handheld device or a robotic printing platform. Both have their own set of advantages and areas of compromise which have been summarised in table 4. Handheld bioprinting systems require the surgeon to directly control a portable instrument and deposit ink into the defect site. This approach generally utilises hardware with a lower capital cost and allows for the surgeon to make real-time adjustments during fabrication to ensure a precise fit within the wound bed [50]. Handheld bioprinters have been investigated in a myriad of other clinical areas already. For example, a multi-material pneumatic extrusion device with an integrated light source dubbed the Biopen was developed to deliver and photo-crosslink scaffolds in-situ for chondral repair [51][52][53]. Similar hardware was employed for the in-situ delivery of scaffolds into bone defects for bone regeneration purposes [54] and also into skeletal muscle defects for the treatment of volumetric muscle loss injuries [55]. However, for wound healing applications in which defect surface area can be relatively large, relying on a singular, high resolution extrusion nozzle may not be feasible due to significantly prolonged fabrication time and delayed wound closure. With maximising volumetric output at front of mind, Hakimi et al [8] developed a handheld instrument capable of extruding planar bioink sheets directly onto a wound surface. The device consisted of an internal microfluidic cartridge and relied on the surgeon to passively guide the instrument using one hand, while actively driven wheels in contact with the adjacent in-tact skin enabled translation across the wound at a constant velocity ( figure 3(a)). The hardware was further optimised by Cheng et al [36] to enhance clinical feasibility. A key design iteration was the transition from two adjacent wheels to a singular wheel located behind the microfluidic chamber; enabling gap-free junctions between adjacent sheets ( figure 3(b)). Connection of the printhead to the instrument was also modified to include a two-axis gimbal, assisting in compensation for user-dependant inclination angles during printing. However, the device was inherently limited by operator-dependant compression of the wheel against human tissue, influencing velocity of translation. This led to the next iteration of the now dubbed 'ReverTome' delivery hardware design which included the introduction of compliant wheels to reduce stress on the wound bed, the optimisation of print-head components to render amenability to efficient sterilisation and the incorporation of temperature-controlled syringe cartridges [56]. However, placement of the wheel prevents the construction of multi-layered skin substitutes when depositing soft hydrogel layers. Further, sheet width alteration requires manual switching of ink outlet components, increasing the overall fabrication time and potentially delaying wound closure. Tianyuan et al [37] developed a customised handheld bioprinting device capable of extrusion, spraying and electrospinning (figure 3(c)). The delivery protocol entailed preparatory spraying of a phantom wound bed, followed by extrusion of a matrix material and subsequent electrospinning of an outer membrane layer. As a prelude to in-situ delivery, the scope of this work was limited to demonstrating the function of each deposition modality in-vitro and comparing the microscale morphology of inks when using a handheld and robotic delivery approach. It is worth noting that a bioprinting system capable of being assembled onto both handheld and robotic delivery hardware would be favourable in the clinic, allowing surgeons the freedom to select between delivery modes depending on the specific clinical scenario they are presented with. Although there was no identified significant difference in microscale ink morphology between the two delivery methods, it is inherent that construct geometry on the macroscale would vary between the two. The authors also noted that printing parameters were adjustable via a microcontroller, however, there are no further details on the user interface and whether it requires surgeons to become well versed in engineering and operational complexities prior to use.
The handheld delivery studies in table 2 all filled the entirety of a wound bed (or phantom wound bed) with ink. It would be of interest to investigate the effects of printing more geometrically complex architectures on wound healing speed and results. It has been shown that the incorporation of a spatially controlled porous network with optimised porosity/channel diameters not only improves the survival rate of  [8] utilised an actively-driven microfluidic extrusion device to deposit planar biomaterial sheets in-situ. Reproduced from [8] with permission from the Royal Society of Chemistry. (b) Optimisation of this device by Cheng et al [36] enabled adjacent gap-free junctions between sheets. Reproduced from [36]. © IOP Publishing Ltd. All rights reserved. (c) The development of a multi-modality bioprinting device by Tianyuan et al [37] for in-situ wound healing applications. Reproduced from [37]. © IOP Publishing Ltd. CC BY 3.0. (d) Albanna et al [40] employed an extendable robotic arm bioprinting system for the in-situ delivery of cell-laden bioinks. Reproduced from [40]. CC BY 4.0. (e) Ding and Chang [41] utilised a customised Fab@home extrusion printer to fabricate porous constructs within phantom gelatin-based wound beds. Reprinted from [41], Copyright (2018), with permission from Elsevier. (f) The algorithm flow chart employed by Zhao et al [39] for visually tracking the movement of freeform deposition surfaces in real time. Reproduced from [39]. CC BY 4.0. delivered cells via facilitating access to oxygen and nutrients, but also increases surface area for endogenous cell attachment [57]. This, in turn, would likely accelerate wound healing and skin tissue regeneration. In saying this, human error and the user-dependence of handheld delivery is expected to influence printing accuracy and construct repeatability. Hence, void-filling may have been a strategy employed by the authors to improve construct repeatability and reduce user-dependent variability.
An alternative delivery method which eradicates human error is robotic bioprinting. This method entails the use of hardware which follows tool path directions defined by computer aided design (CAD) software in order to deposit ink into a wound defect site. As shown in figure 2, this requires additional preparatory steps in the printing protocol in order to define wound bed geometry and develop a 3D skin substitute model. Imaging of the wound bed can be performed using advanced clinical technologies such as computed tomography or magnetic resonance imaging scanning, ultrasound imaging, or portable 3D scanning devices. Ultimately, the image acquisition method should offer the highest possible resolution while not compromising efficiency, nor adding significantly to the cost of treatment. Proposed wound image acquisition methods in the literature include hyperspectral imaging and colour segmentation [41], laser scanning [40], structured light scanning [39], and the use of displacement sensors [38]; each with advantages and limitations briefly outlined in table 2. It is well understood that resolution at this stage of the printing protocol will define quality of the final product, however, there is a fine balance to be maintained in order to avoid significantly compromising time to wound closure and overall system cost. For instance, Ding and Chang [41] argued that the inability of hyperspectral techniques to provide 3D depth maps could be overcome by surgical excision techniques prior to printing which result in relatively uniform wound surfaces. However, this assumption cannot be made for all patients and vitalised, healthy tissue should not be removed for the purpose of acquiring a planar deposition surface. Although less cost-and time-effective, the ZScanner™ Z700 laser scanner utilised by Albanna et al [40] and the Thunk3D light scanner employed by Zhao et al [39] offered sub-millimetre accuracy, ease-of-use, and effortless manoeuvrability. Although light scanning takes a greater number of measurements per data point which generally results in higher definition images, it is also more sensitive to lighting conditions and use in the operating theatre may require a more rigorous scanning protocol to prevent inaccuracies caused by illumination. It is also important to consider how wound bed geometry and localisation may change with patient position, and the user-dependency of scanning devices. Although the aforementioned works must be credited for emphasising the importance of integrated scanning systems, the development of an accompanying scanning protocol which ensures consistent positioning to minimise error will also be necessary in the future. The information obtained from wound scanning can then be loaded into a suitable computer-aided design/manufacturing software to develop a patient-specific 3D model through the detection of negative space. This model is then sliced into two dimensional (2D) cross-sections to generate layer-by-layer tool paths for the 3D bioprinter. Existing commercial software packages may be utilised for this process (e.g. Autodesk ArtCAM [40], Cura3D [39], TSIM [38], etc), or alternatively, a custom algorithm and user interface may be developed to accomplish this task in the clinic with minimal input required from the surgeon. Again, it is both the processing speed and ease-of-use which are of high importance during this phase of the delivery protocol in order for the printing platform to be feasible at POC. Following the completion of scanning and tool path generation, robotic in-situ printing can then ensue.
A modified Fab@home 3 degrees of freedom (DOFs) extrusion printer was utilised by Ding and Chang [41] in a feasibility study to print directly into phantom wound moulds in-vitro. The increased accuracy of robotic printing enabled the direct deposition of five-layer cross-hatched constructs with regular pore shape and size (figure 3(e)). However, although used as a proof-of-concept tool, the Fab@home printing platform offers a small build volume (20 × 20 × 20 cm) and would require significant upscaling and modification to increase the working area and enable positioning of a patient beneath the extrusion module. Further, 3 DOF platforms provide poor control over construct architecture when depositing on a non-planar, irregular surface such as a wound bed. Compared to a manoeuvrable handheld device which interfaces with the human body, a 3 DOF system inhibits orientating the tip perpendicularly to inclined planes. This results in rolling or sliding of filaments due to both downward extrusion forces and torque elevation caused by lateral translation of the filament centroid [39]. By supplementing the axes of movement, a 6 DOF system such as that utilised by Albouy et al [38] can improve print fidelity when depositing onto complex wound topographies.
Albouy et al employed a customised BioAssemblyBot to extrude skin patches directly into porcine wounds [38]. To overcome the restraints imposed by original build volume capacity, a safety enclosure accessory was developed to increase working area. Despite the capability of this printing platform to fabricate complex geometries, continuous filament was extruded to form non-porous structures which filled the wound volume entirely. This may have been to provide physical support to adjacent filaments and prevent excessive lateral spreading which could have occurred in the absence of a crosslinking bath and cooled deposition substrate which was employed in their previous in-vitro studies [28]. Nevertheless, the authors have demonstrated the ability to customise existing commercial printing platforms in order to meet the conditions imposed by surgical theatres.
A shared limitation of the two robotic approaches discussed so far is the inability to account for gross and microscale movement of the human body throughout the printing process. In fact, the majority of reported in-situ robotic bioprinting systems perform 'calibrate-then-print' operations, restricting their use to static surfaces [39]. Ding and Chang introduced a 'safety gap' in the printing protocol, reducing the scanned wound contour by 5% to inhibit unintended contact between the extrusion tip and surrounding wound edge ( figure 3(e)). This somewhat assists in minimising the risk of damage to in-tact skin tissue, however, fails to account for movement in the plane perpendicular to the extrusion tip caused by respiration. The extent of this movement will vary depending on proximity of the wound to the lungs, however, strategies must be implemented in order to guarantee patient safety during treatment. As such, there has been an increase in efforts towards the development of adaptive bioprinting techniques which are able to compensate for the unexpected movement of deposition surfaces. For instance, Zhao et al [39] incorporated visual feedback into a robotic bioprinter by mounting a binocular camera on the end-effector. By exploiting the distinguishable boundary between a wound and surrounding healthy skin, wound bed coordinates in space were tracked visually using colour-based segmentation of real-time images. As depicted in figure 3(f), these coordinates then served as an input for a complex series of identity relations which, when solved, were able to detect spontaneous movement of the deposition surface and compensate for this via real-time modification of the pre-set printing path. Zhao et al [39] validated the printing system by extruding directly into murine wounds and stressed the criticality of incorporating such safety mechanisms to mitigate the risk of damage to healthy human tissue; a factor particularly important when seeking the regulatory approval of a bioprinting system for clinical use. However, closed-loop 3D printing systems are in their infancy; often limited by reduced printing speed and overall increased cost.
It is this challenge amongst others which has likely driven the development of inkjet printing technologies for intraoperative skin substitute delivery in recent years. Compared to extrusion, inkjet systems facilitate an increased distance (albeit in the order of millimetres) between the ink outlet and wound bed, reducing the risk of unintended contact between the two. Skardal et al [47,48] employed a pneumatically driven inkjet printer with interchangeable bioink cartridges to print directly into murine wounds. A similar approach was utilised by Albanna et al [40]; who mounted eight pneumatically driven inkjet nozzles onto a robotic arm system ( figure 3(d)). The extendable arm was able to achieve a full reach of 127 cm (suitable for covering the torso of an average sized patient) and was used to accurately fill both murine and porcine wounds with bioink. Markedly, the aforementioned inkjet systems were devised with additional features which serve to drive their clinical adoption. Both were mounted on mobile frames small enough to enable ease of transportation to a patient's bed or operating theatre rather than relying on movement of the fragile patient. Albanna et al [40] also included specialised locks to engage patient tables in place to minimise unintended movement during printing. Detachable print heads compatible with autoclaving and built-in commands to flush the system with ethanol for sterilisation are further valuable features of the developed platforms. Lastly, the incorporation of multiple printheads in both instances not only enabled the fabrication of multi-material constructs, but also increased volumetric throughput of the delivery system, in turn reducing the overall delivery time; particularly beneficial for wound healing applications.
Finally, it should be noted that all in-vivo studies in the reviewed works have treated a wound area which under-represents that of which would occur following traumatic skin injury. In fact, data from the Burns Registry of Australia and New Zealand indicated that between 2009 and 2013, 75% of patients admitted for severe burns sustained a wound area which exceeded 50% TBSA [58]. Using the average Australian adult TBSA of 1.96 m 2 [59], this corresponds to a total wound area of 0.98 m 2 . Comparatively, the largest reported wound area treated via in-situ bioprinting was 0.01 m 2 [40]. Although wound size will depend on the animal species utilised for in-vivo studies, it is critical that future works prioritise fabrication speed during POC system development, highlighting the potential to deliver to larger TBSAs in a timely manner.

Biomaterials
The success of any bioprinting platform ultimately depends on the inks, or biomaterials, being used. For wound healing, the ideal biomaterial must meet a stringent set of printability and functionality requirements which have been extensively reviewed in previous works [60][61][62]. For a POC system, additional requirements are imposed to enable their in-situ delivery ( figure 4). Here, the inks must require minimal preparation prior to use, retain their printability status when being deposited directly onto a wound bed which sits at an elevated temperature within the range of 31 • C-33 • C [63] and not require sacrificial support structures to facilitate shape retention. Consideration should also be placed on the influence of a wet, enzyme-laden deposition environment which may influence ink and crosslinker concentrations and interfere with gelation processes [49]. Additionally, the crosslinking mechanism itself must be feasible at POC and not add significantly to overall procedure time. For instance, soaking in a crosslinking bath post-fabrication or the use of toxic or immunogenic chemical compounds are not suitable strategies for in-situ implementation.
Only one of the reviewed works has utilised a synthetic polymer-based material [37], and this system was not tested in-vivo. Although offering tuneable degradation kinetics and superior mechanical properties which can be tailored to match skin tissue strength [64,65], their application in this field is hindered by inferior replication of native skin ECM morphology and inherently low levels of (or no) cell recognition sites [65]. Comparatively, natural polymer-based inks are renowned for their capacity to accurately mimic skin tissue ECM [64,66,67] and drive the tissue regeneration process, as well as their ease of crosslinking which can be regulated by temperature change, ion, chemical and/or enzyme delivery. The employed biomaterial(s) and their associated crosslinking mechanism(s) in published in-situ skin substitute bioprinting works are summarised in table 5.
The most commonly utilised in the reviewed works is fibrinogen [8,36,38,40,47]; a naturally occurring glycoprotein found in the blood which, in the presence of thrombin, is rapidly polymerised into fibrin. Fibrin is rich with amino acid sequences which favour the attachment, proliferation, and differentiation of various cells, offers tuneable mechanical properties and is biodegradable through plasmin-mediated fibrinolysis [68], making it an ideal contender as a biomaterial for wound healing applications.
Another protein regularly utilised is collagen [8,40,47]; a prevalent component of the dermis which, in its gelled state, is known to have a microstructure similar to the porosity and anisotropy of skin ECM [69]. There are 28 different types of collagen in the human body, type I being the most prevalent in the dermal ECM [70,71], suggesting its use as an ink component for skin regeneration should be useful. Interestingly, however, Skardal et al [48] has suggested that the application of this collagen type in wounds may have detrimental effects considering the fact that it is also the primary component of scar tissue. Collagen crosslinks in response to pH and temperature reaching near physiological values (6.5 • C-8.5 • C and 20 • C-37 • C respectively) [70], an alluringly apposite characteristic for direct deposition on to a wound bed surface. However, the kinetics of this process are generally too slow to counter gravity-induced deformation, leading to the requirement for additional biomaterials to provide immediate structural integrity [61,64,70]. Further, it is well recognised that collagen gels undergo contraction which is mediated by their interaction with FBs resulting in poor print fidelity and potentially increasing wound contraction-associated scarring in-vivo [47]. The denatured form of collagen, gelatin, retains the beneficial arginylglycylaspartic acid (RGD) peptides present in collagen, while offering the benefit of lower antigenicity and high solubility [72]. It is often included in ink formulations to serve as a rheological modifier owing to its thermo-reversible nature [38,41,48], or modified in some way to enable the formation of covalent crosslinks. For instance, Skardal et al [48] employed a thiolated form of gelatin (Gelin-S ® ) which was amenable to rapid photo-crosslinking.
Also commonly cited is the natural polysaccharide, hyaluronic acid (HA) [8,36,48]; another key skin ECM component with excellent moisture retention properties [67,73]. Despite a lack of cell adhesion peptides [67], HA's popularity as an ink component for wound healing applications has been somewhat driven by the discovery of scar-free healing in foetal skin [74]. It has been speculated that compared to adult dermal ECM, increased concentrations of HA in foetal skin may be responsible for this phenomenon and assist in the regeneration of skin tissue with minimal scar formation [48].
More recently gaining traction due to reduced ecological burden and ethical concerns is alginate; a marine plant-derived polysaccharide extracted from the cell-wall and intra-cellular spaces of brown seaweeds [67]. Although alginate lacks RGD peptides critical for in-vivo cell attachment [64,66], it is able to crosslink rapidly into ECM-mimicking hydrogels with high mechanical strength in the presence of cations (e.g. Ca 2+ , Cu 2+ , Ba 2+ ) [66,75]. This has driven it is use as a structurally supportive component in many ink formulations [37,38,41]. Despite the development of novel modification strategies to combat the limitations of specific natural biomaterials, an ideal ink for wound healing applications does not yet exist. One of the underlying reasons for this is that the improvement of mechanical strength and printability properties frequently comes at the cost of compromised cytocompatibility and host tissue mimicry [76]. To combat this, unifying the desirable traits of multiple materials in a single ink to reach somewhat of a 'sweet spot' in mechanical and biological properties has been explored. For example, combining the superior printability and mechanical strength of alginate with the RGD sequences of fibrinogen, gelatin, or collagen to improve cellular response [37,41], or increasing the mechanical strength and viscoelastic properties of fibrinogen-based bioinks by adding collagen [8,40,47,61]. Although a currently underutilised strategy in-situ, in-vitro skin substitute fabrication often mitigates this trade-off by utilising two (or more) separate inks; one cell-free and devoted to replicating skin mechanical properties, and another providing a soft, cytocompatible environment to facilitate cell survival, migration, and proliferation. It is anticipated that this strategy will continue to gain heightened attraction with the parallel growth of multi-material deposition techniques (e.g. bioprinting platforms with multiple print heads or custom multi-material deposition hardware such as the coaxial extrusion nozzle [77]).
Ultimately, it is not only the ink(s) but also the crosslinker(s) which must undergo rigorous testing and evaluation in order to deem their efficacy at POC. For instance, some concerns have been raised regarding the effects of excess exogenously applied thrombin on the blood system where it is applied [78], leading to the prospect of a thrombin-free approach to fibrin formation using Dulbecco's Modified Eagle Medium [76] or alternative buffer solutions at specified pH levels [79]. Equally, there exists an instinctive reluctance to subject viable skin tissue to UV light due to potential carcinogenic effects. Although cytocompatibility of the UV crosslinking mechanism utilised by Skardal et al [48] was demonstrated on human hepatoma cells in-vivo [80], it would be of use to conduct this investigation for skin cells for an extended time frame to ensure there is no long-term damage. The delivery protocol for these crosslinker(s) must also be in-situ compatible. For example, in their in-vitro feasibility studies, Albouy et al [38] submerged alginate-laden constructs in a CaCl 2 bath for 30 min to induce crosslinking; a technique which cannot be replicated in-situ. Both Skardal et al [47] and Albanna et al [40] delivered a layer of crosslinking solution in alternation with a layer of bioink during fabrication; in fact, Albanna et al [40] implemented a 15 min gelation period per layer prior to deposition of the next. This approach would become problematic with large wound volumes due to a significant increase in total procedure time and an increased risk of cell sedimentation and death within bioink reservoirs. As such, the simultaneous delivery of bioink and crosslinking solution, as performed by Hakimi et al [8] and Cheng et al [36] would be preferential at POC. In both of these studies, thrombin was extruded above fibrinogen-based sheets concurrently to initiate fibrin formation upon contact within the wound bed.
In order for regulatory approval to be granted, reliable and certifiable sources of biomaterials with minimal batch-to-batch variability must be established. It is critical that strategies are established and implemented to mitigate inter-species disease transmission and immune rejection risks. In fact, these are some of the largest roadblocks hindering the use of collagen extracted from bovine, porcine, murine and marine sources in this field [64]. Recombinant collagen produced using synthetic biology techniques can mitigate some of these risks, however the commercial feasibility of this approach when upscaled must be kept front of mind [70]. Finally, it is important to note that human clinical trials of the proposed biomaterial and crosslinker combinations have not yet been performed. As shown in table 5, murine and porcine wound models have been used to study the feasibility of these systems, the latter more closely resembling human skin [81]. However, fundamental differences in anatomy and physiology remain, meaning that the immunogenicity, biocompatibility, and degradation rate amongst other properties of a biomaterial may differ when tested in a human model. Further, the length of in-vivo studies varied across reviewed works, ranging from 30 min [8] to 56 d [40]. In order to obtain valuable data regarding the dynamic behaviour and effects of certain biomaterials on wound healing, the duration of in-vivo experiments should span the lifetime of the biomaterial and all phases of the wound healing process.

Cellular components
The cellular component of a skin substitute bioink may be derived from an autologous, allogeneic, or xenogeneic source. Allogeneic and xenogeneic cells are largely limited by the risk of disease transmission and immune rejection. Hence, their use in wound healing is often overshadowed by their autologous counterparts; cells harvested directly from the patient to be treated.
The use of autologous cells, however, does not come without challenges. In the case of traumatic skin wounds, autologous cell-laden bioinks are unable to be prepared prior to a patient presenting to the clinic (after which, time becomes a critical factor). Hence, the harvesting protocol must be as efficient as possible, while also being minimally invasive and inflicting negligible donor site damage or pain. Further, the protocol for distributing the collected cells throughout a bioink formulation must be rapid, reliable, and able to be performed in a clinical setting, guaranteeing a homogeneous cell distribution prior to in-situ deposition.
Relying on cell extraction on the day of treatment obviates the opportunity for cell isolation and expansion processes since minimum time to wound closure is critical to success. This, together with patient-to-patient biological variability, might result in bioinks with cell densities too low to significantly accelerate wound healing. Several preliminary proof-of-concept works in this field utilise high cell densities for bioprinting to accelerate the extraction of useful data. For example, Ding and Chang [41] bioprinted scaffolds in-vitro using a 12 × 10 6 cells ml −1 FB-laden ink, while Albanna et al [40] deposited a 1.875 × 10 6 cells ml −1 FB-laden ink beneath a 3.75 × 10 6 cells ml −1 KC-laden ink in-vivo. For comparison, a characterisation of the POC cell harvesting device, ReCell™, reported an average achievable yield of 1.7 × 10 6 cells cm −2 of skin biopsy tissue, a figure which includes both extracted FBs and KCs [25]. In the future, it will be interesting to evaluate skin regeneration and wound healing results using cell densities comparable to what is likely to be extracted from a patient following presentation at POC.
The use of non-cultured cells also brings to the surface concerns regarding reduced proliferative capacity. Butler et al [82] explored this idea by comparing the rate of full-thickness porcine wound healing following the delivery of cultured or non-cultured KC cells. It was demonstrated that culturing cells prior to delivery increased the proportion of dividing cells in preference to differentiated cells and in turn accelerated the formation of a confluent epidermis over a 14 d period. As this study ceased at 14 d, it is not possible to conclude whether or not the uncultured cell treatment group eventually demonstrated comparable healing outcomes. Of use in the future will be an investigation of longer duration to assist in weighing the benefits of accelerated wound healing against the benefits of efficient wound closure; as culturing cells to sub-confluency can take up to two weeks, meaning an interim treatment strategy would be required at POC and patient outcomes likely compromised [26].
As outlined earlier in this review, there are several cell phenotypes within native skin, each with a critical role and often interrelated with respect to function. However, this vast cellular population is not yet mirrored in bioprinted skin substitutes, whether fabricated in-situ or in-vitro. In fact, a 2020 review concluded that 61% of skin substitutes fabricated through extrusion printing included only dermal FBs [69]; an unsurprising finding considering FBs are responsible for the ECM secretion necessary for dermal regeneration. In a recently completed pilot study, Albouy et al [38] introduced dermal FB cells into a bioink consisting of gelatin, alginate and fibrinogen which enabled the in-situ generation of growth factors such as vascular endothelial growth factor, fibroblast growth factors, and platelet-derived growth factor in full-thickness porcine wounds; molecules which play fundamental roles in various stages of wound healing. Compared to wounds treated with the non-cell-laden ink, healing was accelerated by 10 d, scar tissue formation was minimised, and vascularisation improved over the 42 d examination period. A greater understanding of cellular crosstalk has developed in recent years; in particular, that the interaction between KCs and FBs is necessary to achieve skin homeostasis and accelerate tissue regeneration in-vivo [69]. This has resulted in the majority of more recent works incorporating both FBs and KCs into the printing protocol. As a prelude to in-situ delivery, Hakimi et al [8] utilised a novel handheld sheet extrusion device to deposit a layer of FB-laden bioink beneath a layer of KC-laden bioink in-vitro. For the in-situ treatment of full-thickness murine wounds, Albanna et al [40] jetted a layer of FB-laden bioink to fill the wound area and then subsequently jetted a layer of KC-laden bioink on top. Bioprinted cells were found to be present within the wound 6 weeks after printing, accompanied by an endogenous cell population. The formation of an organised dermis and defined epidermis was also observed by week 3, compared to cell-free ink treated wounds which lacked the aforementioned by week 6. The same procedure was used for treating porcine wounds and also included an allogeneic cell source treatment group for comparison. Structures printed using autologous cells demonstrated reduced contraction and accelerated epithelium formation in comparison to those using allogeneic cells over an 8-week examination period.
Both of the aforementioned works explored the ability to compartmentally deposit these two cell types in a stratified manner, also mimicking the typical FB:KC cell density ratio (∼1:2) in native human skin. This has been argued to accelerate wound healing as the final printed structure mimics the native anatomy of skin as closely as possible. However, what seems a simple decision in the in-vitro printing protocol may act as a significant roadblock for clinical translation. At POC, the isolation of different cell types from a single biopsy and the preparation of a greater number of individual bio-inks increases procedure complexity, time to wound closure, and equipment required. Hence, it might be of interest to explore the delivery of FBs and KCs together within one bioink. In fact, clinicians have already paved the way for exploring this notion in the absence of bioprinting technologies through the use of ReCell™; a buffer suspension composed of a mixture of autologous FBs and KCs which is sprayed onto partial-thickness or superficial wounds. To extend its applicability to the treatment of full-thickness wounds, Wood et al [26] trialled spraying this mixture of cells onto a dermal skin substitute scaffold prior to transplantation into full-thickness porcine wounds. The ability of FB and KC cells to self-assemble into distinct dermal and epidermal layers in-vivo in response to endogenous biological cues was observed over 21 d. This migration was corroborated by previous works where cultured KCs seeded onto the underside of a skin substitute were observed to migrate upwards through the dermal scaffold to form an epidermis in 2-3 weeks [82,83]. In a similar study conducted more recently, Damaraju et al [27] applied drops of ReCell™ to the underside of a skin substitute used to treat full-thickness porcine wounds. Epithelial hyperplasia (the presence of KCs migrating upwards through the dermis towards the epidermis) was resolved by 42 d and the union of products was again deemed promising as a POC treatment strategy. This invites us to question: if the human body is able to provide the correct cues to initiate the self-assembly of a mixture of disorganised skin cells in-vivo, is there sufficient reason to introduce time delays prior to printing to isolate different cell types and prepare respective bioinks which enable their stratified deposition? To answer this, further studies are required in the future to better understand the cross talk between relevant cell types and the interactions of different cell types with their 3D fabricated environment both in-vitro and in-vivo.
An alternative approach involves the use of stem cells (SCs). SCs offer a multilineage differentiation capacity (can differentiate into FBs, KCs, and other cell types), release GFs and cytokines to support angiogenesis and ECM production and can be harvested in large quantities [69]. SCs may be embryonic, adult, or induced pluripotent (embryonic-like cells engineered on the bench from tissue specific cells). Due to the ethical concerns of utilising embryonic SCs, and the infancy of induced pluripotent SC research, the use of adult tissue derived SCs (in particular, mesenchymal SCs (MSCs)) has gained the most traction. MSCs, or stromal cells, are renowned for their self-renewal, multipotency, anti-inflammatory properties and immunomodulatory capacity; the latter suggesting an allogeneic source may become suitable as an off-the-shelf product for wound healing applications in the future [47]. In wound healing applications specifically, they have been shown to accelerate wound closure, increase the rate of epithelialisation and enhance angiogenesis in-vivo [84]. MSCs can be isolated from bone marrow, adipose tissue, umbilical cord tissue, or amniotic fluid; each having a unique extraction procedure which differs in invasiveness, pain, and time required for the retrieval of a sufficient cell quantity. Skardal et al [47] compared the encapsulation of bone marrow-derived MSCs against amniotic fluid-derived MSCs in a bioink consisting of fibrinogen and collagen, each subsequently delivered to full-thickness murine wounds in-situ. Compared to cell-free ink treated wounds, both MSC types accelerated wound closure and re-epithelialisation, facilitated swift neovascularisation, and resulted in the formation of a well organised epidermal layer of skin by 2 weeks. Interestingly, both types of MSCs failed to differentiate into key skin cell phenotypes such as FBs or KCs in-vivo, and further, could not be detected in or around the wound 2 weeks after bioprinting. Their transient presence suggested it was rather the secretion of GFs which facilitated wound healing rather than the integration of cells into host tissue and it will be of interest in future work to investigate the specific fate of these cells in-vivo. In 2017, Skardal et al encapsulated amniotic fluid-derived MSCs in a bioink consisting of HA and gelatin, again observing MSC transiency following deposition into full-thickness murine wounds [48]. However, the authors articulated that the short-term presence of such cells may in fact be considered an advantage for the clinical application and regulatory approval of allogeneic SC sources, decreasing the risk of immune rejection and tumorigenic behaviour in-vivo. More recently, Cheng et al [36] encapsulated umbilical cord-derived MSCs in their bioink formulation (composed of fibrinogen and HA) prior to its extrusion as a planar sheet into full-thickness porcine wounds. As found in Skardal's work, the MSCs accelerated re-epithelialisation, reduced scarring and contraction compared to cell-free ink treated control groups. Additionally, they found that the MSCs significantly increased cell repopulation of the wound site. Regardless of recent progress and the promising potential of SCs in wound healing, the use of these cells for bioprinting skin substitutes is still considered to be in its infancy [85]. Further, when considering a POC system, it may be the harvesting procedure for some autologous SC types which could act as a roadblock to clinical adoption. For this to become feasible, extraction of cells should be both non-invasive and minimally painful, and the isolation and purification of these cells must not be time-consuming. Further research into the long-term effects of allogeneic SC inclusion and the possibility of becoming an off-the-shelf cell source will also be needed to determine clinical efficacy.
Additional insight on cell phenotype selection can be gained when examining the works of those who have rather bioprinted skin substitutes on the bench for subsequent transplantation. Some have attempted to improve vascularisation of engineered skin tissue by incorporating ECs alongside FBs and KCs [64,86,87]. Baltazar et al [87] took this notion further with the addition of pericytes; cells which wrap around ECs and line blood vessels in the body. Other works have focused on enhancing aesthetic outcomes following implantation through the incorporation of melanocyte cells [88,89] which significantly improve skin pigmentation and restore protection against light and heat [69]. However, before proceeding further with the notion of cellular diversity and its benefits in the acceleration of skin tissue regeneration, it is important to be reminded that the overarching goal in bioprinting skin substitutes for wound healing is not necessarily for the construct to wholly mimic native skin architecture and functionality immediately. Bioprinting such a construct would require numerous cell phenotypes to be sourced, isolated, cultured, and strategically placed

Cells
• Optimised autologous cell extraction and bedside preparation procedures • Cost-benefit assessment of isolating and culturing autologous cell types prior to printing compared to immediate delivery • Methods to accelerate autologous cell proliferation at POC within a 3D structure; a time-consuming process which delays wound closure and in doing so, compromises patient outcomes. The aim at POC is rather to bioprint structures which can achieve efficient wound closure while facilitating in-vivo self-assembly (in other words, endowing structures with the capacity to regenerate aesthetically and functionally equivalent skin tissue over time). Further, as these scaffolds are deposited directly into a wound bed they may somewhat rely on neighbouring healthy tissue for nourishment and further infiltration of endogenous host cell types [65]. Although fundamental studies involving additional cell types are important for the development of a foundation of knowledge which can motivate future research directions, as we move towards the development of a bioprinting system for clinical deployment, it will be necessary to seriously question the efficacy of incorporating additional cell types for POC wound healing applications.

Translational challenges and future perspectives
Campbell and Weiss [90] were amongst the first to introduce the concept of in-situ bioprinting, expressing a lack of confidence in its feasibility considering the preference for simple, off-the-shelf solutions clinicians tend to have. However, advances in 3D bioprinting strategies and biomaterials used for skin tissue engineering in recent years present opportunities to realise this technology in the clinic. With improved patient outcomes within reach, we must now question what challenges remain and may arise throughout the clinical deployment of this technology such that we are well-equipped to overcome, if not avoid, these roadblocks. Still an early-stage concept, the bioprinting technology (and coupled in-situ delivery protocol), biomaterials, and cells utilised require further exploration in order to develop clinically and economically viable systems (table 6).
Ultimately, sustained use of a bioprinting system at POC will be contingent upon whether or not the clinical team feels proficient in its use. To achieve this, the bioprinting modality should not require the learning of complex engineering concepts and be tailored to time-poor surgeons who, in some cases, may not have an interest in understanding system technicalities. Bearing a likeness to the injection of solutions using a syringe and needle, the concept of extrusion printing is straightforward compared to alternative bioprinting modalities. Extrusion hardware is also relatively cost-effective, helping to minimise concerns regarding the introduction of healthcare inequities with expensive medical technologies and the deployment of a system which only the wealthy can afford, rather than those who may in fact need it more. However, one of the biggest challenges yet to be overcome is the absence of established strategies which guarantee patient safety throughout the printing process and mitigate unintended contact between the extrusion nozzle and viable wound bed tissue. Further developments in adaptive, closed loop printing mechanisms which cater for moving freeform surfaces will be integral to the survival of extrusion printing systems during their journey from bench to bedside. Another significant challenge to overcome will be the augmentation of printing speed. Regardless of deposition modality, it is critical that printing speed be addressed in the early stages of system development to ensure the technology is capable of achieving timely wound closure in the case of large surface area wounds commonly encountered in a clinical setting. Here, future consideration should be placed on feasible methods to increase volumetric output without compromising on printing capabilities, whether this be via custom-made microfluidic attachments, additional print heads, or other novel multi-material deposition methods.
Taking the selected bioprinting modality, the delivery method must then be robust, reliable, and simple. Both handheld and robotic approaches have been discussed in this review, however, the portability, cost-effectiveness, and simplicity of a handheld device at this point in time may serve to favour its clinical adoption. Further, the elimination of wound scanning and 3D model development from the protocol fast-tracks printing and mitigates the risk of confidentiality breaches of patient-specific digital data, potentially expediting regulatory approval. It is likely, therefore, that the deployment of handheld POC bioprinters will precede their robotic equivalents. Due to the inherent limitation of human error and decreased printing precision, as robotic delivery systems and adaptive printing strategies mature, they will serve as important tools at POC for the treatment of large and complex wounds. The level of surgeon involvement required will also need to be questioned. Offering the surgeon complete control over all printing parameters will increase system complexity, lengthen the training process, and reduce the chance of knowledge retainment and prolonged adoption of the technology. However, providing no freedom to optimise print settings whatsoever, will serve to decrease the capabilities of the bioprinting system in certain clinical scenarios. In the future, utilisation of artificial intelligence and machine learning principles to guarantee high precision and excellent print quality while maintaining a suitable level of user involvement will serve invaluable to the survival of this technology during clinical translation. Equally as important is the development of pre-and post-printing protocols for the bioprinting system. Consideration of how the system is to be assembled, manoeuvred, sterilised, maintained and repaired throughout its lifetime in a hospital setting is essential for successful clinical deployment. Such considerations may lead to the future incorporation of single use disposable components, automated sterilisation commands, and the creation and upkeep of regular maintenance checks and refresher training schedules for clinicians. It is likely that POC bioprinting platforms would also benefit from the co-deployment of personnel with expertise in biofabrication and operating principles of the system. Working alongside the surgeon, the presence of a technician would facilitate a smooth handover of the medical technology to end users in the clinic and assist in building operator confidence for ongoing use of the bioprinter at POC.
Biomaterial and cell selection are additional areas in which we have not yet reached an ideal or gold standard. Although there are numerous ink (and bioink) candidates with promising printability and biological properties for wound healing, in the future there needs to be a greater weight of consideration on whether selected biomaterials can be sourced and sterilised reliably and stored with a sufficient shelf-life to make them clinically and economically viable. A similar, more holistic analysis is also required for cell selection. It is known that autologous cells provide the most promising wound healing results from a biological perspective, however without established cell banking services in place we must, in the meantime, devote efforts towards optimising their sourcing and handling procedures at POC. It will be important to verify whether these cells truly require expansion in the clinic prior to printing to facilitate wound healing. If uncultured autologous cells at densities lower than their cultured equivalents can be utilised in a bioink to facilitate skin regeneration, this will obviate the need for additional equipment and personnel not naturally found in clinical settings, accelerate permanent wound closure, and decrease the cost of treatment. If this is not the case, efforts into the development of strategies to accelerate autologous cell proliferation at the bedside would be advantageous. It will also be beneficial in future works to investigate whether their self-assembly capabilities in-vivo may be manipulated within a bioprinted construct to eradicate the need for FB and KC isolation and purification; potentially increasing the efficiency of treatment further.
It is evident from this review that there does not yet exist the optimal bioprinting system nor ink(s) for the treatment of full-thickness wounds at POC. Moving forward, a holistic cost-benefit assessment of the bioprinting platform (including hardware, delivery protocol, biomaterials and cells utilised) will be required to ensure the development of both clinically and economically viable systems. Earnestly questioning and objectively testing whether the technology will improve patient outcomes compared to current clinical practice (e.g. the use of skin substitute products or STSGs) is imperative. In this comparison, aesthetic and functional wound healing outcomes as well as quality of life metrics and downstream impacts on the patient must come into play. Similarly, cost-effectiveness of the technology must consider both upfront capital costs and future, anticipated expenses throughout its lifetime. The goal for the POC technology is to offset the initial increase in medical spending by improving patient outcomes such that future spending down the line is lowered; short-run costs are only part of the story [93].
Ultimately, successfully traversing bioprinting from bench to bedside for wound healing will require expertise from a vast array of areas. Early established and ongoing inter-disciplinary collaborations between not only researchers and clinicians, but also ethical, regulatory and commercialisation specialists are essential to avoid falling into the dubbed translational 'valleys of death' [94]. It is this tight-knit network that will facilitate the development of POC bioprinting systems which meet the needs of the patient, surgeon, and relevant regulatory approval body, leading to a paradigm shift in POC wound treatment for the better.

Data availability statement
No new data were created or analysed in this study.