Translational considerations in injectable cell-based therapeutics for neurological applications: concepts, progress and challenges

Significant progress has been made during the past decade towards the clinical adoption of cell-based therapeutics. However, existing cell-delivery approaches have shown limited success, with numerous studies showing fewer than 5% of injected cells persisting at the site of injection within days of transplantation. Although consideration is being increasingly given to clinical trial design, little emphasis has been given to tools and protocols used to administer cells. The different behaviours of various cell types, dosing accuracy, precise delivery, and cell retention and viability post-injection are some of the obstacles facing clinical translation. For efficient injectable cell transplantation, accurate characterisation of cellular health post-injection and the development of standardised administration protocols are required. This review provides an overview of the challenges facing effective delivery of cell therapies, examines key studies that have been carried out to investigate injectable cell delivery, and outlines opportunities for translating these findings into more effective cell-therapy interventions.


INTRODUCTION
Significant progress has been made during the past decade towards the clinical adoption of cell-based therapeutics. Preclinical studies have translated into clinical trials for conditions of the central nervous system (CNS), including Parkinson's disease (PD), 1 Huntington's disease, 2 amyotrophic lateral sclerosis (ALS) 3 and stroke. 4,5 Clinical trials have focused on the delivery of purified cellular suspensions, for example, in spinal cord injuries and stroke. [6][7][8] However, existing cell-delivery approaches have shown limited success, with numerous studies showing fewer than 5% of injected cells persisting at the site of injection within days of transplantation.
One of the main translational challenges to the implementation of injection-based cell therapy is the need to determine suitable delivery protocols to ensure sufficient accuracy, improved cell survival and reproducibility in administering cells for therapeutic efficacy. 9 In this review, we identify critical considerations for the various stages of cell administration, outline studies that have measured functional performance of injected cells and discuss criteria for designing cell-delivery devices for minimally invasive cell therapy. The various approaches used to attempt to maximise cell viability and functionality in high accuracy cell-therapy applications are also described. We suggest that if the variables linked to optimal cell, survival can be recognised, cell loss may be reduced and efficacy of cellular therapies can be improved.

CELLS AS THERAPEUTIC AGENTS: TRANSLATIONAL BARRIERS IN NEUROLOGICAL APPLICATIONS
Three stages make up a typical cell-therapy procedure: (1) in vitro preparation of cell suspensions; (2) injection procedure; and (3) retention of the administered cells post-injection. 10 Focusing on one stage only can yield optimised settings that are not favourable to the entire procedure, and therefore it is essential that a systematic investigation considers all three stages to outline optimal transplantation parameters ( Fig. 1).
Cell loss has been reported to be observed post-transplantation, 11,12 with quantified survival rate of transplanted cells as low as 1%. 13 Moreover, a large number of cells that have been originally retained die, possibly due to exposure of cells to the inflammatory microenvironment, washout, immune destruction, dispersion through impaired local vascular system, 14,15 apoptosis and anoikic cell death. 16 Variable clinical outcomes observed in two trials for PD 1,17 have been partially ascribed to a failure to properly distribute cells to the target site. 18 Attaining efficient delivery of an adequate number of cells without loss of functionality is therefore a key step in the development of regenerative medicine approaches.
The diverse behaviours of various cell types, choice of dosing density, administration protocol and cell viability post-injection are some of the obstacles facing clinical translation. This section will explore the various variables involved in the three stages of cell-therapy procedures.
Pre-delivery factors: scaling up pre-clinical models to human therapy To overcome low cell transplantation efficiency, one popular approach to translational scale-up has been to deliver a large number of cells to a single site 19 with doses ranging up to hundreds of millions of cells. 19 This makes cell-therapy approaches technically complex and expensive, as well as offering limited control over site-specificity, as cells will potentially migrate to other sites. 20 Determination of cell density. In cell transplantation studies, cell concentration within suspensions is often reported, with suspensions of over 100,000 cells/μL considered highly concentrated. 21 As well as higher costs of cell processing, these suspensions can be viscous and may cause needle clogging and uneven injection flow. Moreover, as cell size varies widely, depending on the site or species of origin, the maximum concentration suspended within a certain volume will therefore vary. Hence, it is more accurate to express the cellular component as a volume fraction, which is the volume occupied by the cell fraction in a certain volume of suspension vehicle, as described by Rossetti et al. 10 High-density cell suspensions may lead to cell death, attributable to limited oxygen and nutrient diffusion. 22 The limited capacity to generate appropriately large numbers of standardised cells routinely, within a limited time frame, whereas adhering to quality standards has also led to small clinical trials. In addition, large doses of administered cells also pose higher safety risks of tumorigenesis 23 and micro-embolism. 24 Moreover, higher injected cell concentrations result in exposure to increased shear forces, 25 yet may have a reduced tendency for sedimentation. 10 No agreement exists regarding the optimal cell number to be transplanted, although this is likely to vary depending on cell type, disease and administration route. For example, 3-5 × 10 7 cells/kg mesenchymal stem cells (MSCs) were administered per multiple sclerosis patient, 26 whereas in spinal cord injury, 5-6 × 10 6 cells/kg have been transplanted intrathecally. 27 Taguchi et al. 28 reported that the higher dose of 3.4 × 10 8 cells gave improved neurologic outcomes than the lower dose of 2.5 × 10 8 cells administered intravenously. A lower percentage loss of MSCs was exhibited with increasing concentrations being administered, possibly due to the finite number of cells being able to attach to the available inner surface of the device. 29 However, a phase II trial using 5-10 million NT2N cells in stroke patients showed that patients receiving the lower dose performed better. 4 A recent study also demonstrated that a suitable cell dose, rather than a higher one, can better aid the repair of injured tissue in stroke patients. 30 Cell suspension vehicle. Suspension vehicles have been found to affect the viability of cells pre-delivery and their survival upon implantation. 10,31,32 Preparing a cell suspension that maintains a homogeneous distribution and viability is essential to ensure effective clinical translation. Results suggested that MSCs viability is reduced to levels significantly below the permitted limit of 70% in a short time when stored in parenteral solutions, with other biological functions being slightly affected. 31 Cells at different temperatures will also have different requirements for storage solutions. 33 Injection volume. The human brain is 800-2300 times larger than that of rodents used for pre-clinical research. 34 To enable scale-up to larger target volumes, cell distribution can be increased by making multiple, lower volume injections for improved engraftment. 14 The adjustment of the needle/catheter for adequate cell distribution can lead to multiple needle tracks and entry points. Multiple transcortical brain penetrations have been used for a range of clinical trials. 1,17,35 This is a cause for concern, as each penetration carries a risk of intracranial haemorrhage 22 and damage to white matter tracts. The shortage of human studies with lesion volume calculations, such as occur in human spinal cord injuries, 36 make it challenging to decide on optimal injection volumes. A study by Gutierrez et al. 37 evaluated the spinal cord's tolerance to varying numbers and volumes of cell injections in Göttingen minipigs. Complete functional recovery was achieved by 2 weeks, even when injection volume and numbers were increased. However, histological analysis revealed tissue damage when large volumes (50 µL) of cell suspension were injected per site. Although increased numbers of injections did not cause an increase in tissue damage there was an optimal number of injections required to achieve the best engraftment.
In many clinical trials, doses are extrapolated from data of preclinical animal studies. Depending on how they are calculated, doses of 100 million or more cells may result in substantial volumes being required. Thus, it is vital that detailed dosing studies are carried out in animal models to establish the minimum effective and maximum tolerated doses. 38 Cell-delivery challenges Cell injector system design requirements and challenges. The main delivery platform for cell-based therapeutics has traditionally been a needle and syringe, 39 with cell preparations delivered either systemically or directly. 40 Although many clinical trials have used syringe/needle systems without cannulas, [41][42][43][44] CNS cell transplantation trials have typically utilised a frame-based platform for the insertion of a stereotactically guided straight cell-delivery cannula or needle. 1,2,17,22,35,[44][45][46] There is a growing recognition that conventional needle-based and catheter-based cell transplantation tools have considerable inadequacies that may affect clinical translation. 47,48 Insufficient pre-clinical testing of surgical tools and methods for cell delivery to the human brain and spinal cord may result in the failure of cell transplantation trials, despite the reliability of the basic biological concepts. [49][50][51] Key considerations for clinical translation of cell-delivery devices include ease of loading and use, reproducibility of delivery, possibility of sterilisation, freedom from leachable and/or extractable contaminants, and ensuring no visual obstruction through a surgical microscope in high accuracy applications.
Role of mechanical forces The mechanical forces that cells experience as they pass through the injection device is a factor influencing their subsequent viability and functionality posttransplantation. To comprehend the fluid dynamics in action, we must explore the mechanical forces exerted on the cells.
While flowing through a needle, cells may experience several types of mechanical forces, comprising shear forces characteristic of linear shear flow, a pressure drop across the cell and extensional (stretching) forces. 52 The nature of flow, whether laminar or turbulent, should be confirmed at the ejection rate and syringe/needle size used for the transplantation procedure. This can be verified through the calculation of Reynold's number (R e ), which determines flow conditions (transitional level to turbulence is R e = 2100): where ρ is the carrier fluid density (water at room tempera-ture=999.97 kg/m 3 ), Q is volumetric flow rate (mL/min), D is needle diameter and η is dynamic viscosity of the medium. Given that the flow is laminar, the velocity profile is parabolic across the diameter (Fig. 2a), with maximum velocity at the centre of the lumen. Cells and fluid in the middle of a cannula flow at a different velocity to those at the walls. This difference in velocity exposes cells to shear stress. 25 Changes in shear rate and shear stress have been suggested to affect cell viability and function. 52 Shear stress (τ) is The velocity profile across the diameter is parabolic. Shear stress (τ) is zero at the centre and increases linearly to its maximum value at the wall. As a cell flows from the syringe to the needle, it will experience increasing velocities along its length, causing it to stretch. b Plug flow behaviour-flow velocities are almost equal across the whole diameter. Shear-thinning materials display this behaviour when their flow in a capillary reaches a steady state calculated by Poiseuille's equation, where τ max is shear stress (dyn/cm 2 ); Q is flow rate (cm 3 /s); η is dynamic viscosity of the medium; and R denotes needle radius. The magnitude of shear stress is maximal at the walls of the syringe/needle, zero at the centre and changes linearly with distance between those two. Even low levels of shear stress (10 dynes/cm 2 ) have been stated to have a major influence on the activation of molecular cascades. 53,54 Any change in the system's geometry, such as the sudden tapering of a syringe to the needle, can also result in cells experiencing extensional flow, an increase in velocity and, consequently, high shear. 55 The range of shear stress values generated by clinicians may exceed physiological values. As a useful reference, average wall shear stress is 1-6 dyn/ cm 2 for venous circulation and 15 dyn/cm 2 for arterial circulation. 56,57 However, previous reports have also stated that low shear forces of 3.5 58 and 15 dyn/cm 2 59 can influence cells. Previous work has concluded that cell damage is based on the extent of shear stress as well as exposure time to zones of shear. 25,60 Damage may also occur due to collisions with the stationary surfaces of the device. 61 Complete damage of the cell may not necessarily be the only adverse result. Investigations carried out on erythrocytes have shown that excessive stretching or deformation of the cell membrane might result in loss of function. 62 Extensional flow also causes cells to experience stretching and deformation, leading to cell death. 63,64 A larger difference between diameters of the syringe and needle will result in larger extensional forces, whereas a longer needle will increase the time a cell is exposed to extensional forces. In addition, cell aggregation may intensify shear stress experienced by cells during delivery. Forces acting on cells during their administration may have two effects: cell destruction along with stretch preconditioning. 65 Biomaterial-assisted cell delivery (reviewed in detail in 'Biomaterial-assisted delivery' section) has the potential to mediate the impact of mechanical forces during cell delivery. Shear-thinning hydrogels are injectable materials that exhibit a viscosity decrease under shear strain; the viscosity returns to normal when the shear is removed. Shear-thinning materials typically display plug flow behaviour when their flow within a capillary reaches a steady state. They exhibit a central wide region, where material and cells experience little or no shear rate and a narrow zone of shear adjacent to the walls (Fig. 2b).
Needle characteristics Critical parameters in injectable cell delivery include needle characteristics and/or diameter of the tubing used. 50 Needle characteristics include inner and outer diameters, length, stiffness and bevel design. Deep subcortical target structures, such as the caudate nucleus or corpus striatum, require a long, thin needle/cannula of sufficient rigidity to penetrate to the target site without injuring the overlying structure. This cannula can be 19 cm or more. Shorter needles (8-10 cm) would require direct brain exposure, which is more invasive. 50 Various needle gauge sizes have been used for neurological applications, typically starting from 25 G. 45,66 Devices with needle sizes of 30 G have been developed for CNS applications, 51, 67 and 30-34 G are typically used for applications requiring high accuracy. 68,69 However, smaller bore size needles, e.g.: 27-34 G, are more prone to obstruction by concentrated cell suspensions, especially when successive injections are required. Smaller gauge needles cause less tissue trauma, minimise leakage through the track created by needle insertion, and have been suggested to reduce gliosis. 70 Mehta et al. 71 reported a 53% headache rate with the use of a 22 G needle in neurological patients, and suggested that this high incidence could be caused by the large-bore spinal needle, which was used to prevent mechanical damage to the infused cells. Karussis et al. 72 also reported a high incidence of headache with MSCs in multiple sclerosis and ALS patients, attributing this to the lumbar puncture itself; needle size was not reported. Care should be taken to choose the correct needle size that accounts for cell type, injection speed, site of transplantation and the viscosity of the suspension. Veraitch et al. 73 reported that 45 passages through a bore of 500 µm diameter were needed before a significant reduction in viability was detected. Longer, finer needles have resulted in the delivery of a smaller volume 50 and a lower percentage of the cell dose administered. 60 However, loss of cell viability resulting from the use of smaller needle gauges was not substantial. 25,60,74 Faster acceleration of fluid and cells within longer needles may be more prone to clogging, perhaps because there is less time for cells to dislodge from transient adhesions to other cells and the injector. 50 Bevel angle and length may also influence injectate dispersion and direction.
Material employed for construction of delivery device Recent studies have shown that cells may be retained in the delivery device, 60,75 with this effect being more pronounced in glass cannulas than in metal. 75 Whether the adherence of the cells to the walls of the syringe/cannula is due to chemical, physical or charge effects has not been studied, but all of these might possibly be manipulated to decrease adhesion and settling. 75 In addition, coating the inner surface of the needle/catheter with proteins may reduce cell adhesion to the device. 76 In addition to possible adherence to device materials, cells may encounter residual trace amounts of manufacturing agents that could induce apoptosis or undesired differentiation. The use of hydrophilic and siliconised coatings on internal walls of the cannula has also been suggested to decrease friction between the needle and tissue in vivo. 75 Although glass cannulas present several advantages, including higher precision and minimal penetration injuries, 21 glass may not be sufficiently rigid to endure injection pressure. Novel materials such as electroactive polymers and magnetorheological elastomers 77 show promise for the development of tailored needles whose rigidity could be adjusted by current, magnetic field or temperature.
Sedimentation Uniformity of cell dosing may be affected by aggregation or sedimentation of cells in suspension over time, with clumping of cells being found to affect cell viability. 78 The possibility of cell sedimentation during the course of the surgical procedure, especially in large diameter needles/cannula, 75 and the potential for inconsistent cell dosing needs to be considered in designing cell-therapy protocols, particularly given that orientation will vary depending on application. In vertical cannulas, cells sediment towards the tip of the needle, with most cells appearing in the first 10-20% of the injectate. In horizontal cannulas, settling and adherence of cells on the side caused most of the cell dose to emerge in the final 10-20% of injectate 75 ; similar effects of cell dose retention have been reported. 74 A suspension medium should preferably have a density of~2 g/cm 3 and a viscosity >3 cP to reduce cell settling at 20-40% cell volume fraction. 10 Negative impacts of cell sedimentation observed post cell therapy include graft-induced dyskinesias in PD trials which may have occurred due to formation of dopamine 'hot spots' in the brain, resulting in abnormal activation of neural circuits. 79 Moreover, some cells, such as neural progenitor cells (NPCs), have a natural tendency to form clusters that may settle at the tip of the needle. 50 Therefore, it is essential to maintain an even cell suspension using agitation or other approaches.
Administration protocols. As cell injections for neurological injuries move on to extensive clinical testing, procedural standardisation is necessary to reduce the probability of technical failures and measure the effects of interventions on patient outcome and other endpoints (e.g. injection-associated trauma, graft survival). Clinical trial design to enhance the final number of cells that integrate within the diseased tissue will require reexamining the current lack of optimisation of transplantation protocols. For a given clinical trial, optimum factors for cell implantation are normally estimated from pre-clinical research or the investigator's judgement. 80 Administration protocols will vary depending on cell type, disease and administration route. While inappropriately administered cells may in most cases result in no functional impact, in some cases, transplanted cells may result in undesirable side effects. Injection rate Injection rates employed in clinical trials for cell injection are inconsistent. For neural cell transplantation, for instance, studies have used a rate of 5 µL/min, 81 compared to 300 µL/min for spinal injury, 82 and between 10 and 1000 µL/min for stroke. 7,83,84 Rate of injection has been shown to be potentially vital in cellular delivery. 50,60,74 Although small volume injections may be made over extended periods in the laboratory, clinical injections must be made within limited surgical duration. It has been suggested that delivering the required volume over a longer time will potentially reduce mechanical forces acting on the cells and the creation of damaging pressure gradients 21 ; delivery of 44 nL in 5 min was linked to good tissue preservation. 85 However, such a rate would be impractical, with 1 μL requiring over an hour to deliver. Notably, brain microinjections in rats at a rate >1 μL/min have been linked to tissue damage. 86 Recently, intracarotid transplantation of glial-restricted precursors (GRPs) and MSCs through a microcatheter at an infusion rate of ≥1 mL/min resulted in stroke in 27/44 rats, even with a vehicle-only infusion. A lower rate (0.2 mL/min) was safe for the infusion of both vehicle and GRPs. 87 Moreover, at high flow rates, backflow along the catheter shaft may occur if the applied force is removed from the plunger and affect delivery. It was determined that backflow can be avoided in grey matter with a 32 G catheter at rates <0.5 µL/min in a rat model. 88 However, injection at a slow rate may lead to a lower percentage of the cell dose being delivered, higher apoptosis levels and influence the immunophenotype of delivered cells. 60,74 Cells employed Optimal injection parameters will vary depending on the morphological, physiological and growth characteristics of each cell type. Thus, it is not surprising that preparation and ejection protocols optimised for one type of cells are not necessarily applicable to others. 32,60,74 When studying injection rates, smaller NIH-3T3 cells had an optimum ejection rate of 150 µL/min, whereby the maximum percentage of viable cell dose was delivered, 60 whereas the relatively larger BMSCs had an optimal ejection rate of 300 µL/min. 74 Some cell types, such as MSCs, may be especially disposed to clumping, leading to blockage of flow within the needle, 49 whereas other cell types, e.g. NSCs, are sensitive to manipulation and undergo apoptosis easily. During handling, MSCs were observed to have a greater inclination than NPCs to come out of suspension and form tight clusters. 49 Tissue compliance If injections are made into cavities, larger volumes can be introduced without generating large pressure gradients if fluid drainage from the cavity is adequate, i.e. match injection to lesion volumes. 89,90 Hydrodynamic injury will occur if intraparenchymal pressure surpasses the tensile limits of the parenchyma. Given that many microsyringes are manufactured to tolerate 1000 psi or more before failure, the likelihood of producing large tissue pressures during injection exists. Larger injection volumes also worsen the reflux of infused materials along the penetration tract, 91 making cell dosing unpredictable in terms of numbers and final location, and contributing to injectionrelated injury. 21 The low elastic modulus of brain tissue also provides little resistance to reflux of infusions. 34 Post-delivery complications: functional performance of injected cells Given the number of clinical trials that use needle-based systems or catheter-based systems for cell delivery, surprisingly few studies have focused on the impact of small-bore needle injection on cell function. Transplantation studies have focused on the outcome of the trial rather than the variables that affected results. Investigating viability and potency of transplanted cells at the time of delivery is crucial, as a small level of cell death within a concentrated cell population could have a significant effect on the remaining viable portion through the release of cytotoxic agents. 92 To date, research has been carried out to evaluate the impact of injection on cell functionality. 25,29,50,52,[93][94][95][96] Studies have highlighted parameters needed to achieve an adequate cell density for therapy, such as the period between cell preparation and implantation, 97 injection rate, needle length and bore size 50 and cell concentration. 70,98 There has been wide variability in how the effects of delivery devices on cell performance are assessed. Therefore, it is vital to develop standardised assays to consistently characterise cellular therapy products. Discrepancies in the employed delivery devices and administration methodologies has complicated comparisons and led to contradictory results. This is illustrated in Table 1, with some studies demonstrating that cell manipulation through a needle did not significantly affect viability, 25,93,95 whereas other studies show that it did exert a significant effect. 94 In one study, ejection of NIH-3T3 cells at 150 µL/min exhibited the optimal percentage of dose delivered, 60 whereas another study on hMSCs 74 showed an optimal recovery at the highest flow rate under investigation (300 µL/min). A study by Heng et al. 29 showed reduced cell recovery at the higher flow rate (1600 µL/min), with large variation in their samples.
Another potential concern is the inadequate testing of many aspects of cellular health in most studies, thereby not providing the complete picture needed to develop appropriate clinical administration protocols. Table 1

provides a summary of the investigations carried out into the effects of injection-based administration of cells on various parameters of cell health.
Conventional tests, such as propidium iodide, can reveal which cells are dead, but are not useful predictors of potential delayed damage to the cells. The transient exposure to shear forces when injected does not make membrane integrity measurements, such as trypan blue exclusion assay, a good method of uncovering delayed apoptotic and senescent responses triggered. This is evident in some studies that have only relied on the analysis of cell viability to conclude that cells were not affected by the injection parameters under investigation, as shown in Table 1. 50,95,96 In addition, some studies have utilised a relatively small sample size of n ≤ 3 for their investigations. 25 45 considered a reduction of almost 50% in viability of cells post-injection acceptable.
The aforementioned issues illustrate that an enhanced understanding of what happens to cellular therapeutics post-injection, specifically regarding vital cellular health parameters, will facilitate the development of more efficient administration and formulation approaches. This reinforces the need for defining crucial parameters and trial-specific pre-clinical good laboratory practice validation of any injection protocol before human application. One method to enhance cell functionality testing in pre-clinical and clinical studies is to assess the viability of a small aliquot of injected cells. Further investigations into proliferative capacity, phenotypic expressions, apoptotic responses and other transcription-level changes of the various cell types under clinical investigation is critical. Without this, the uncertainty of whether delivery was effective will undermine interpretations of efficacy.
Most investigations carried out on the impact of injection on cell functionality and viability are completely in vitro, and although these data are important, other issues can undermine cell viability in vivo. For example, reflux and rejection can Translational considerations in injectable cell-based therapeutics MH Amer et al. Translational considerations in injectable cell-based therapeutics MH Amer et al. eliminate grafts in animals. In vitro experiments presented in this review will need to be augmented with in vivo data. We cannot exclude the possibility that cells may experience significant biological changes on catheter/syringe/needle passage outside the conditions examined in studies cited in this review. Optimisation of injection protocols, materials from which delivery devices are constructed, injection rates and cell dose is therefore crucial to achieve higher efficacy and reduce variability using the smallest possible cell dose.

APPROACHES TO IMPROVE INJECTABILITY OF CELLS
Improvement of neurological cell transplantation protocols and tools To decrease variations encountered in manual injection, automated devices (e.g. computer-controlled syringe pumps) have been suggested to offer better control. 49, 100 Automated brain injections were more reproducible compared to manual injections, with variability 2-20-fold higher in manual techniques. 49 For cell aggregates or encapsulated cells, automated cell-delivery methods may prove useful to control the cell dose and preserve structure integrity. However, manual delivery is still often preferred to automated robot-assisted tools due to the lack of haptic feedback to the surgeon in the latter method. 101 To overcome the need for multiple injections, Mendez et al. 48 developed a two-hole cannula tip design, whereas Lim and colleagues created a system capable of radially branched deployment (RBD) of a catheter at adjustable angles. 22,34 Notably, cells at high density (6 × 10 7 cells/mL) were not damaged by transit at a high rate of delivery (50 µL/min, 99.6% viability). 34 Although it represents an improvement, RBD still does not fully repeat the cell distribution achieved in pre-clinical models.
The use of syringe/cannula rotation during the injection procedure is a strategy described in the literature. 45,75,102 In a clinical trial for Parkinson's disease, the injection protocol incorporated rotations of the cannula between deposits, with transplant survival confirmed up to 12 months after surgery. 48 Skewed distribution of cells in horizontally oriented cannulas can also be amended by rotation at regular intervals during the procedure. 75 Robot-assisted surgery, image-based needle guidance systems can also facilitate accurate delivery. These include needles that automatically stop and lock into position by sensing drops in mechanical resistance. 103 Technological improvements to platform and cannula design have reduced procedural invasiveness while improving injection accuracy, resulting in progress from single unilateral microinjections to multiple bilateral injections without long-term neurological consequences. 51 Few clinical trials thoroughly optimise and state their cell transplantation protocols, and do not specify infusion volumes, rates or duration of administration, which may lead to variability and lower transplantation efficiency (Table 2). However, recent studies are beginning to recognise the importance of optimisation of transplantation protocols, both in pre-clinical and clinical trials. 87,98,[104][105][106][107] In a model of retinal dysfunction, photoreceptor integration was 20-fold to 30-fold increased by improved transplantation procedures including single and dual injections, and optimisation of the number of cells injected per µL. 108 Another study defined optimised conditions for an autologous stem cell therapy to treat a craniofacial traumatic deficiency, regenerating 80% of the original jawbone deficiency in only 4 months vs. a minimum healing period of 6-8 months with typical protocols. 104 Biomaterial-assisted delivery Although many transplantation studies have employed a salinebased delivery vehicle, alternative cell carriers have included injectable hydrogels. Hydrogels are hydrated, polymeric networks  with great potential as cell carriers. 109,110 Alternatives to saline vehicles have included cells embedded within hydrogels or microencapsulated within polymers, attached to the surfaces of microcarriers, or injected as multicellular aggregates. 111 Injectable biomaterial scaffolds as cell carriers have demonstrated increased spatial and temporal administration compared to saline injections. 112,113 In addition, the use of biomaterials provides an opportunity to deliver growth factors alongside cells. Hydrogels such as alginate may experience 'shear banding' along inner walls of the needle, 52 whereby a layer of hydrogel shear-thins to form a fluid, acting as a lubricant, allowing the rest of the intact hydrogel to slip through the needle. The width of the plug flow region is dependent upon rigidity of the hydrogel and flow rate, 114 therefore altering the hydrogel's formulation may impact plug flow. This mechanical protection is independent of cell properties 52 and is therefore applicable to different cell types.
Aguado et al. 52 tested 1% alginate with three different molecular weights, and demonstrated protective effects with optimised mechanical properties; human umbilical vein endothelial cells had significantly lower cell viability in phosphate-buffered saline or in non-crosslinked alginate compared to in crosslinked alginates.
Considerable research on hydrogel cell carriers has focused on their role post-delivery, including cell localisation, 110 support of tissue growth 115,116 and protection from local inflammatory conditions. 117 More focus is required on their role during the injection procedure, as opposed to post-delivery, where the presence of an optimised viscoelastic material may protect cells from damaging mechanical forces. 52,60 This has been only investigated on a small scale (Table 3), and needs to be explored further.

CONCLUDING REMARKS AND FUTURE OUTLOOK
The complexity of the cell injection process has resulted in a paucity of studies where combinations of process parameters have been evaluated. Nonetheless, these few studies demonstrate that interactions between cell preparation protocols and injection procedures are significant, and may substantially alter cell delivery. Studies also suggest that standardisation of injection parameters will be a critical aspect of designing and comparing clinical studies. In addition, factors such as choice of instrument, interval between filling of the cannula/needle and fixing to the stereotaxic frame prior to injection, and angles of delivery must be considered and empirically validated before use in clinical practice.
An integrated approach to the evaluation of cell-delivery success is needed to improve the assessment of delivery efficacy and to allow for sound interpretations of clinical results. Improved cell-delivery tools are also required to streamline the delivery of cell-based therapeutics from the donor to the patient without compromising quality. Finally, pre-clinical planning and testing of the desired administration protocol with cell-type specificity is essential to achieve good clinical trial design.