Imaging in the repair of peripheral nerve injury

Abstract

Surgical intervention followed by physical therapy remains the major way to repair damaged nerves and restore function. Imaging constitutes promising, yet underutilized, approaches to improve surgical and postoperative techniques. Dedicated methods for imaging nerve regeneration will potentially provide surgical guidance, enable recovery monitoring and postrepair intervention, elucidate failure mechanisms and optimize preclinical procedures. Herein, we present an outline of promising innovations in imaging-based tracking of in vivo peripheral nerve regeneration. We emphasize optical imaging because of its cost, versatility, relatively low toxicity and sensitivity. We discuss the use of targeted probes and contrast agents (small molecules and nanoparticles) to facilitate nerve regeneration imaging and the engineering of grafts that could be used to track nerve repair. We also discuss how new imaging methods might overcome the most significant challenges in nerve injury treatment.

Graphical abstract

Peripheral nerve injuries are common and debilitating conditions that occur, for example, in an estimated 2–3% of the more than 3 million annual upper extremity traumas in the USA [1]. They are usually classified according to the extent of axonal and structural damage to the nerve, as described by Sunderland [2] and Seddon [3]. These classification systems attempt to describe the severity of nerve injury by the degree of disruption of the internal architecture of the nerve and roughly indicate the likelihood of regaining function.

Neurapraxic or low grade axonotmetic injuries are treated conservatively with observation, physical therapy to maintain joint mobility, and, sometimes, electrical stimulation. In these injuries, maintenance of the perineural sheath allows for organized regeneration of fascicles and increases the potential for spontaneous nerve recovery. On the other hand, disruption of the perineural sheath (Sunderland grade IV), or, in more severe cases, disruption of the outer epineurium (Sunderland grade V), leads to a high grade axonotmetic injury or neurotmetic injury, respectively. Prognosis for recovery of these injuries is poor, and high grade axonotmetic and neurotmetic injuries usually do not spontaneously recover. They require surgical intervention for exploration, resection of damaged nerve and coaptation with the goal of restoring function and sensation to the traumatized limb. Direct epineurial or grouped fascicular repair is possible when the zone of injury is small with resulting defects of only a few millimeters [4]. An example of such repair for a partially transected nerve is shown in Figure 1.

Figure 1. . Partially transected digital nerve from a kitchen knife in a 12-year-old boy, (left) before and (right) after repair.

Injuries in which a sufficiently long segment of the nerve is irreversibly damaged and removed cannot be directly repaired without inducing longitudinal tension, which can be detrimental to healing [5]. Reconstruction of such segmental defects can be accomplished using nerve implants to span the gap. These include nerve autografts, non-nerve (e.g., vein) autografts, decellularized nerve allografts and artificial nerve guidance conduits [5] (discussed later in section: Nerve implants: source & materials). The success rate of nerve engraftment procedures varies inversely with gap length. Importantly, there is evidently a critical gap length – approximately 3 cm in humans and 1.5 cm in rats – at which the success rate for all implant types drops sharply and above which autografts are the only implant type with an appreciable success rate [6,7].

An awareness of the size, location, severity and shape of the zone of injury to the nerve is critical for surgical planning purposes. In addition, knowledge of the spatial extent of regrowth, whether normal, adverse or lacking, can inform surgical follow-up. This perspective highlights the critical role that imaging of peripheral nerves can contribute to diagnosis, operative treatment and monitoring of postoperative outcomes. Future imaging research may provide the capability to visualize nerve trauma and monitor progression of nerve regeneration. Furthermore, preclinically, imaging techniques can offer novel methods to help develop better grafts and test new approaches to accelerate recovery. Currently, little is done in either of these directions. The first part of this perspective describes the mechanism of nerve damage, the state of the art of methods to assess nerve injury and grafts to repair the nerve. The second part reviews the existing imaging modalities that have been utilized for nerve imaging. The third part discusses potential imaging approaches to monitor the graft and the nerve restoration process.

Mechanism of nerve damage

In response to traumatic injury of a peripheral nerve, the distal portions of the damaged axons undergo Wallerian degeneration in a process that relies on a host of signaling factors [8]. In turn, denervated Schwann cells (SCs) (i.e., those ensheathing-degenerated axon portions) dedifferentiate, yielding a nonmyelinating state [9]. These SCs subsequently proliferate and release chemokines that recruit macrophages into the degenerating nerve [10]. Dedifferentiated SCs and macrophages have the job of endocytosing the cellular debris resulting from degeneration. Clearance of both axonal and myelin debris is necessary for successful regeneration of peripheral axons. Following this debris clearance, extracellular matrix (ECM) is remodeled and deposited, after which SC-based bands of Büngner form [11], providing a framework for the regenerating axons originating from the proximal nerve stump (Figure 2). The rate of the subsequent axonal regrowth averages 1–3 mm/day in humans [12]. While regeneration of whole peripheral nerve over gaps longer than the critical length has virtually never been achieved, once axons have made it past the distal end of a lesion into healthy endoneurium, they alone can regenerate over very long lengths (>1 m). Restoration of sensation and/or muscle contraction occurs once the sensory neuron peripheral processes and/or motor neuron axons reach and innervate their distal target(s).

Figure 2. . Schematic of major cell components in nerve regeneration following injury.

Following an injury, the nerve environment distal to the injury undergoes significant changes. Schwann cells dedifferentiate, proliferate and migrate from both the proximal (A) and distal stumps (C) into the bridge that connects them, then release chemokines that recruit macrophages and fibroblasts (D). Both Schwann cells and macrophages perform myelin clearance. Then, extracellular matrix is remodeled and deposited, providing endoneurial structure, after which Schwann cells take position on this extracellular matrix to form bands of Büngner (E). This framework facilitates regenerative axonal growth from the proximal nerve stump, forming regeneration units (A) which eventually attain normal morphology. Ingrowth of blood vessels (B) supports this process.

Reproduced with permission from [13] © John Wiley and Sons (2019).

The ultimate outcome of nerve repair/regeneration ranges from unnoticeable to substantial functional recovery depending on numerous medical factors, including the time between injury and treatment, as well as the quality of the repair. Conditions such as intraneural scarring and fibrosis [14] as well as fascicular escape and axonal misdirection can lead to poor recovery and painful neuromas [15]. In those cases, additional surgery is necessary to remove unneeded tissue and attempt to reset the regeneration process. Unfortunately, given the long delay between when such abnormalities appear or begin to impede regrowth and when they are typically detected, their presence often signifies irrecoverable regeneration failure in the affected nerve fibers.

Methods to assess nerve injury & regeneration

Currently, assessment of nerve injuries, both pre- and postrepair, relies heavily on patient reported outcomes, manual muscle testing and electrophysiological studies. Imaging has had a limited role in evaluation of nerve injuries until relatively recently [16–18]. Unfortunately, when neurophysiological testing is not enough to satisfactorily diagnose the degree of injury, imaging usually cannot provide enough additional information to obviate the need for surgical exploration [19]. Postoperative imaging of nerve functional recovery is not routinely employed in clinical practice despite dramatic advances in imaging over the past three decades. On the other hand, in animal models of nerve regeneration, various preclinical-only imaging methods such as transgenic labeling and histomorphology are routinely utilized.

Nerve implants: source & materials

Autogenous (autologous) nerve grafts remain the ‘gold standard’ for peripheral nerve injury repair, although simpler implants, including many nerve guidance conduits, perform comparably well for small (<1–3 cm) gaps. In a typical case where a long section of nerve is missing, a sensory nerve of sufficient length (e.g., sural nerve) is harvested for use as a cable graft (Figure 3). Allografts can also be used but must be accompanied by long-term immunosuppressant use if the foreign cells are not removed from the implanted tissue. Decellularized cadaveric nerve grafts, available commercially, have been utilized in sensory as well as mixed motor and sensory nerves with good clinical outcomes [20]. They have the advantage of not being immunogenic while providing significant molecular and structural regenerative cues. Grafts of others tissues such as vein and muscle [21] have also been used successfully to repair nerve injuries but are not commonly employed in modern practice. Additionally, isografts are employed in preclinical models (Figure 4) as positive controls for nerve repair experiments.

Figure 3. . Traumatized ulnar nerve from an all-terrain vehicle accident in a 15-year-old female.

(Left) Resection of the damaged nerve to get back to healthy fascicles. (Middle) Healthy fascicles are visible in the resected nerve. (Right) The graft in place (sural nerve autograft).

Figure 4. . Reverse isograft to fill a segmental defect in a preclinical model of nerve injury (rat sciatic nerve).

Alternatives to tissue-derived implants include nerve guidance conduits (NGCs), which are artificial nonimmunogenic tubes. The nerve stumps are inserted into the ends of the NGC and attached in place just inside the openings. This spatially confines the regeneration milieu and corrals the axonal sprouts and connective tissue spreading from the proximal end. The presence of the distal stump in the conduit promotes regeneration from the proximal stump [22]. Currently available clinical NGCs are only suitable for subcritical gaps; use of artificial nerve grafts for longer gaps produces inferior outcomes, in part due to the lack of supportive SCs [23].

The plainest NGCs used in preclinical models are sterile silicone tubes, but there is no shortage of ongoing innovation in all aspects of NGC design. The most common bioderived materials are polymer hydrogels such as collagen [24] and other ECM components [25]. Engineered materials/composites (e.g., containing magnesium [26] or carbon nanotubes [27]) are also being explored. The in vivo degradability and mechanical compliance of a conduit can be tuned so that it envelops the regenerating nerve long enough for it to heal and flexibly enough not to mechanically constrain the nerve as it thickens over time. Tailoring of biophysical properties such as conduit wall porosity, which can in turn control diffusion [28] and fibroblast invasion [29], and electrical conductivity have been shown to positively impact the various physiological processes involved in regeneration and lead to better outcomes.

Intraluminal content is the most common ‘upgrade’ to NGCs. Luminal structure can be engineered to the appropriate length scales for surface receptors, axons and connective tissue, providing distally directing tropic cues to facilitate regeneration. Various methods, including electrospinning of biocompatible, often bioderived polymers, can be used to form nano- and micro-aligned luminal filling material. One of most common luminal materials is fibrin [30], which has a special place in the nerve regeneration cascade. Coagulating from fibrinogen that seeps into the nerve space during Wallerian degeneration, fibrin constitutes the initial weak network along which regenerative cells migrate to deposit more stable ECM. Fibrin is also used as a ‘tissue glue’ in various surgical procedures, including nerve repair, as an alternative to sutures, making it a rational choice for a luminal filler material.

Preclinical research in improvement of NGCs can be viewed as an application of nerve tissue engineering and broadly intersects with the development of scaffolding for neural cell culture [31]. NGC luminal surfaces can be chemically modified to guide regeneration by presenting trophic cues and attachment sites for cell-surface receptors. NGCs can include mechanisms, such as using tethered nano- or microparticles or other mainstays of tissue engineering, for multistage controlled release of drugs, growth factors [32] and immune-modulating signals [33]. In addition, NGCs with microchannels for individual axons are being developed to better guide axonal regrowth [34,35]. Furthermore, NGCs can be additively manufactured via 3D printing [36] or microstereolithography [37], facilitating rapid production, tailoring to the anatomy of the nerve and establishment of neurotrophic concentration gradients [38].

Many kinds of nerve cell- and stem cell-derived cells (e.g., SCs) have been incorporated into NGCs [39,40]. However, while cells can potentially be a great source of nerve engineering material, producing an abundant personalized supply is still an unsolved challenge. Nonetheless, efforts have been made to identify combinations of cells and other engineered factors [38] that can match autografts in regenerative performance. It is also now possible to bioprint various types of nerve pseudotissue, including hydrogel-SC mixtures [41] or even scaffold-free stem cells [42]. In summary, a vast range of NGC designs has been explored in preclinical research with the goal of developing off-the-shelf or quickly made-to-order implants that in their ideal embodiment would be autograft-like with unlimited nerve regeneration potential.

Currently available artificial nerve interventions selectively perturb only a few regeneration-related processes at either the molecular, supramolecular, subcellular, cellular or tissue level, while innate biology serves to perform the complex, omni-scale activities of cellular programing and tissue remodeling involved in healing/regeneration. These processes, including the roles of transcription factor expression, neuronal surface receptors [43], neurotrophic ligand production [44] and small molecule messengers such as nitric oxide [45], are not fully understood, and further research is needed into their interplay and their relationships to regenerative success. Targeting some of the critical molecules that mediate these processes as part of preclinical imaging research could both help elucidate their role and help find the keys to guaranteed and unlimited regeneration potential.

Imaging of peripheral nerve regeneration in conjunction with repair

Peripheral nerve imaging has the potential to be extremely useful in clinical contexts, such as postsurgical repair [46], and educative in preclinical contexts. A number of imaging modalities are available [47]. However, currently available clinical imaging technologies are often still unable to diagnose injuries, track regeneration, locate problems during recovery and monitor the progress of healing with high resolution, sensitivity, and adequate intra-endoneurial contrast. Additionally, certain injuries with severe loss of function but not total disruption of connective tissue present a unique problem because conventional (electrophysiological) examination is not able to determine whether the tissue is best left to heal on its own or surgical intervention is recommended. In these cases, clinical imaging is often not able to do so, either. Similarly, imaging is typically unable to assess postrepair regenerative progress with significantly greater accuracy than electrophysiological examination can.

Since nerve regeneration is unpredictable even over relatively short distances, it would be valuable to have an imaging modality with sufficient contrast to discern the presence and functionality status of both axons and connective tissue and thus facilitate better injury diagnosis and detailed monitoring. Researchers seek to address these goals as well as to better understand the nerve regeneration process and develop better nerve regeneration technology. Accordingly, preclinical imaging studies of peripheral nerves exploit the full gamut of major available modalities and techniques, beyond those used in the clinic, in combination with innovative labeling methods and contrast agents. This section addresses the potential of the common and emerging imaging modalities to improve the diagnostics, reconstruction and rehabilitation of peripheral nerves after injury in clinical and preclinical contexts.

Examples of imaging for nerve-related applications are shown in Figure 5. In clinical imaging, MRI and ultrasound are routinely used far more than other modalities to provide information about traumatic peripheral nerve injuries beyond what can be learned from electrophysiological testing. In preclinical models, various experimental methods for detailed monitoring of peripheral nerve injury have been implemented using MRI and ultrasound [17] as well as positron emission tomography (PET) [48], optical, photoacoustic and others. A far greater range of imaging modalities can be used specifically for imaging certain neoplastic growths within peripheral nerves [49] because they develop as a manifestly ‘different’ kind of tissue and also tend to attract contrast agents. For example, x-ray computed tomography (CT) does not have sufficient contrast to discern normal peripheral nerve structure in vivo but can image peripheral nerve tumors [50].

Figure 5. . Imaging modalities for potential clinical and preclinical evaluations of nerve grafts and repair.

Color key: blue – used routinely in clinical settings; red – used in preclinical research with limited human use; yellow – ex vivo use only, not suitable for in vivo use.

CT: Computed tomography; PET: Positron emission tomography; SNI: Spared-nerve injury.

Center image reproduced with permission from [36] © PLOS ONE (2018), licensed with Creative Commons License.

MRI image reproduced with permission from [51] © Elsevier (2014).

X-ray CT image reproduced with permission from [52] © Hindawi (2019), licensed with Creative Commons License.

Ultrasound image reproduced with permission from [53] © Journal of Neurosurgery Publishing Group (2015).

Photoacoustic microscopy image reproduced with permission from [54] © SPIE (2014), licensed with Creative Commons License.

PET image reproduced with permission from [55] © Society of Nuclear Medicine and Molecular Imaging (2011).

Optical microscopy (polarization-sensitive optical coherence tomography) image reproduced with permission from [56] © SPIE (2015), licensed with Creative Commons License.

Fluorescence (transgenic fluorescent labeling) image reproduced with permission from [50] © Society for Neuroscience (2004).

High resolution is especially important for nerve imaging since part of nerve diagnostics relies on visualizing the complex architecture and functional profile of peripheral nerves. Some ‘general-use’ contrast agents in MRI and other modalities can improve imageability for peripheral nerves, but, as with most other organ systems, there are no clinically approved contrast agents that selectively target peripheral nerves. Furthermore, the design of contrast agents for peripheral nerve imaging has special challenges. Intravenously administered contrast agents intended to accumulate inside endoneurial sheaths encounter the blood–nerve barrier, which filters out most molecules larger than around 500 Da [57]. Nevertheless, contrast agents ranging from small molecules and nanoparticles to novel constructs built on a biologics platform (antibodies, mRNA, etc.) have been used in preclinical studies to image peripheral nerves, but the use of such contrast agents is less advanced in nerve repair than in fields such as cancer and cardiology.

Magnetic resonance imaging

MRI is perhaps the most versatile soft tissue clinical imaging modality as well as the most expensive. It is sensitive to the proton composition of different tissues and thus exhibits some intraneural contrast, making it useful for traumatic peripheral nerve injuries. Peripheral nerves have a characteristic appearance under each of the MRI subvariants including T1- and T2-weighted imaging [48] and are usually best viewed using fat suppression sequences [58]. MRI-based techniques, including diffusion-weighted imaging, diffusion tensor imaging and microneurography, have shown significant potential for clinically informative anatomical imaging of peripheral nerves [59], including diagnosis of peripheral nerve injury [60] and relatively detailed intrafascicular imaging in animal models [61]. Like other imaging modalities, MRI can be used to image incidental phenomena, such as edema, when the injury cannot be directly observed. Similarly, MRI of the vasculature surrounding injured nerves in preclinical models provides an indirect means to keep track of regenerative progress [62].

Several MRI contrast agents are clinically available, with many more being tested preclinically, but even relatively small contrast agents are normally excluded from the endoneurium due to the low molecular weight cutoff of the blood–nerve barrier. Accordingly, the obligatory bulkiness of MRI contrast agents due to inclusion of metal chelation moieties ordinarily presents an additional constraint when designing them to distinctly target nerves. However, endoneurium reversibly self-permeabilizes upon sustaining trauma or disease [63]. Thus, some preclinical contrast agents can freely diffuse through injured endoneurium and ‘stain’ lesions/disrupted fascicles but not healthy, intact fascicles [64]. Notably, at least one such MRI contrast agent, Gadofluorine M, has an apparent affinity for certain fibrillary ECM components and degenerating myelin found in damaged and regenerating nerves, persisting there long enough to be imaged for many hours after administration [63]. Another avenue into the nerves is available for the MRI-active metal ion manganese(II), which can pass through calcium channels and therefore accumulate differentially in nociceptive and non-nociceptive nerves [65]. Some non-nerve-specific preclinical T1 and ¹⁹F MRI contrast agents can also enhance peripheral nerve injury imaging by highlighting immune activity [48,63].

Ultrasound imaging

Ultrasound is appropriate for imaging most regions of the arms and legs, where the most commonly traumatically injured nerves are located. Along with MRI, it is the main clinical modality for diagnostic imaging of peripheral nerve injuries. Although both ultrasound and MRI yield tomographically reconstructed 2D images of cross-sectional or longitudinal slices, ultrasound has certain significant advantages. Namely, ultrasound works in real time, allowing for dynamic examinations [64], and ultrasound machines are portable and less expensive than MRI machines. Ultrasound is also considered one of the safest forms of internal imaging because it does not use electromagnetic radiation. As with other imaging modalities, a detailed knowledge of anatomical landmarks is necessary in order to make the most of the technique.

Ultrasound can reliably distinguish (fascicular) perineurium from interfascicular epineurium in humans [66] and can be used to identify and localize disruption of nerve continuity and other gross structural anomalies [67]. Therefore, it can be used to estimate the fractions of preserved and injured fascicles in partially torn nerves, as a visual guide for invasive procedures and intra-operative exploration, and to diagnose postoperative nerve repair complications [68] such as neuromas. In addition, because it is real-time, ultrasound is used to provide visual guidance for delicate tasks such as insertion of nerve blocks [69] and anesthetic injection. A variant of ultrasound, acoustic radiation force impulse imaging, has been shown in a clinical research setting to additionally discern mechanical properties of peripheral nerves [70].

A chief disadvantage of ultrasound is that the imaging targets must be ‘accessible’ via the skin or a body cavity, in other words, it is most effective for superficial nerves. Increased depth and hyperechoic structures such as bone occlude underlying tissue, so high-resolution ultrasound is not useful, for example, for portions of nerve deep in the buttocks or calf [16]. Furthermore, ultrasound distinguishes nerve components more poorly (has lower contrast resolution) than MRI, showing comparatively less detailed black-and-white ‘honeycomb’-resembling short-axis cross-sections. However, ultrasound can have higher spatial resolution than whole-body MRI and can sometimes clearly delineate lesions that MRI cannot [53]. Like MRI, ultrasound cannot discern subfascicular details of regeneration such as endoneurial or axonal regrowth or distinguish absent or nonfunctional axons from functional axons.

Photoacoustic imaging

Photoacoustic imaging detects ultrasound resulting from the thermoelastic pulsation of tissue due to absorption of energy from optical (nonionizing laser) stimulation. It provides morphological information in living human peripheral nerves beyond that provided by ultrasound alone [71]. Its potential has further been demonstrated in in vivo and ex vivo studies [72], but it is not used for traumatic peripheral nerve injuries in humans outside of research studies. In addition, photoacoustic microscopy has been used in in vivo-like conditions to image peripheral nerves with visible axon-bundle striations [54]. A potential disadvantage of photoacoustic imaging is that strong light may harm delicate nerve tissue through overheating and cause significant discomfort in the patient or animal.

Optical imaging using fluorescent tags

Although not used widely for clinical peripheral nerve imaging, fluorescence imaging modalities show great promise based on their diverse applications in animal models and potential uses in humans, for example, for cancer detection (Figure 6). Fluorescence imaging can be advantageous because it is typically much easier to visually interpret than MRI or ultrasound and can be easily coregistered with ordinary optical imaging for additional structural information. Indocyanine green (a blood-pooling, near-IR fluorophore), fluorescein (a rapidly excreted, green fluorophore) and methylene blue (a blood-pooling, far-red fluorophore) remain the three most widely used clinical fluorescent dyes. They can be used in the clinic as intra-operative neurovascular dyes by highlighting nerve-adjacent blood vessels [73]. Tissue-targeted conjugates of these dyes, on the other hand, have rarely been used in humans.

Figure 6. . Optical imaging modalities.

BLI: Bioluminescence imaging; CT: Computed tomography; DOT: Diffuse optimal tomography; FMT: Fluorescence molecular tomography; FPT: Fluorescence protein tomography; FRI: Fluorescence reflectance imaging; HR-FRI: High-resolution fluorescence reflectance imaging; MPM: Multiphoton microscopy; OCT: Optical coherence tomography; OFDI: Optical frequency-domain imaging.

Reproduced with permission from [74] © Springer Nature (2008).

In vivo optical imaging depth is less than that of either MRI or ultrasound due to signal scattering. Thus, imaging modalities that operate in the near-IR range (700–950 nm) are favored because near-IR radiation is not strongly scattered by tissue and is nonionizing. Even better penetration is expected in shortwave infrared (SWIR) (1000–2000 nm) [75,76]. Imaging depth can then be several centimeters, sufficient for superficial nerves and for intra-operative use.

Under ideal circumstances, in vivo fluorescence imaging of peripheral nerves is superior to other techniques due to its high spatial resolution (near the light diffraction limit, i.e., approximately 0.3–1 μm depending on the wavelengths used), high sensitivity (10^-9 to 10^-12 mol/l) [77], low toxicity, low cost and usually adequate imaging depth. One disadvantage is that it usually requires a potentially unsafe or poorly clearing fluorescent tag. However, there can be considerable flexibility in the design of tags: they can be small molecules, proteins, nanoparticles, conjugates – among others. Furthermore, the contrast can be enhanced with dynamic fluorescence techniques such as fluorescence lifetime imaging [78,79]. Therefore, fluorescent probes, especially several working in concert, may be highly effective for detailed imaging of peripheral nerve regeneration.

Fluorescent probes can be optimal in vivo labeling agents when there is a clear view of the tissue of interest. Thus, one major application of fluorescence imaging in both humans and animals is guiding surgery by highlighting specific types of tissue intra-operatively in real time. Peripheral nerves have even been directly imaged in large animals [80] using dedicated intra-operative fluorescence imaging systems and tissue-specific dyes [57]. In addition, confocal fluorescence microscopy and spectral imaging can also be performed through endoscopes that can be inserted through external orifices or surgical incisions [81] for less invasive imaging of hard-to-reach spaces.

Fluorescence imaging typically uses a fluorophore conjugated to a targeting moiety such as an antibody. Efficacy of a fluorescent probe greatly depends on its targeting moiety, which ideally has selective affinity for peripheral nerve-associated cell-surface receptors or substances such as myelin. The commonly used technique of ex vivo immunofluorescent staining followed by fluorescence/optical microscopy [82] is an example of the sensitivity and high resolution of fluorescence imaging. In one application, peripheral nerves excised mid-regeneration were chemically treated to become optically clear and imaged at various time points to see unprecedented detail of the regeneration process [83]. In vivo, intraperitoneally injected fluorescently labeled neuronal ganglioside-targeted antibodies were successfully used for intra-operative imaging in mice, where the fluorescence signal accumulation peaked after 6 days [84].

Unfortunately, while dyes of various sizes with nerve-specific affinities have been developed, the blood–nerve barrier prevents all but the smallest biomolecules and conventional fluorescent imaging agents (such as Alexa Fluor dyes) from passing into the intraneural space, ordinarily precluding any accumulation. Thus, many specific, high-affinity molecules such as axonal transport tracers, myelin- [80] and Nissl substance-binding compounds [81] must be introduced directly via nerve injection or after nerve transection in order to be used for imaging [85], limiting their translatability. Fluorescent molecules that both have nerve-specific affinity (e.g., to myelin) and are small enough to cross the blood–nerve barrier have recently been identified, including ones with near-IR activity [86].

Fluorescent probe-based imaging has been used extensively to examine dynamic biological processes. A number of in vitro and in vivo fluorescent probes [87] and techniques have been described for nerve-specific molecular imaging [88,89]. Fluorescence imaging can capture complex, spatially nonuniform biological events such as reactive oxygen species activity [90] through the use of a specially designed sensing probe that is temporarily or permanently ‘switched on’ by an instance of the activity being imaged.

Transgenic labeling

In vivo optical modalities for animal peripheral nerve imaging have expanded beyond those used in humans in part because of the advent of transgenic fluorescent animals. Therein, the contrast agents are natively expressed, fluorescent fusion proteins with various advantages as imaging tags, including automatic quantitative labeling. While not suitable for clinical applications, the use of cell type- and/or phenotype-specific transgenic labeling has proven useful in understanding the biology of peripheral nerve regeneration as well as for testing of nerve regeneration therapies. For example, in Thy1-YFP rodents [91], axons expressing fluorescent proteins provide ample contrast for in vivo imaging. Transgenic fluorescent labeling of nerve-associated cell phenotypes has been performed in various animals including zebrafish [92] and rodents in order to easily track progress of nerve regeneration [93]. Another transgenic labeling modality is bioluminescence imaging (BLI), which has an even higher sensitivity than fluorescence. As an illustrative example, luciferase-labeled brain-derived neurotrophic factor can be imaged via near-IR biochemiluminescence upon the systemic injection of a luciferin analog [94].

Optical microscopy

Various optical microscopy imaging modalities can provide exquisite subcellular structural and functional detail of nerves, often without any labels or contrast agents [95] and with submicron resolution. In vivo peripheral nerve microscopy is performed on superficial or intra-operatively exposed nerves [96]. For example, high-resolution 3D images of human peripheral nerves can be obtained intra-operatively using a topical optical coherence tomography probe without any specific preparation [97]. Signal is obtained deep enough, and most nerves are narrow enough, that the whole nerve can be imaged provided that its outer surface is accessible. Polarization-sensitive optical coherence tomography can similarly be used to discern demyelination in rat nerve injuries [56]. Because the use of microscopy is significantly constrained by focus and depth penetration limits, most modalities for peripheral nerve imaging are typically better suited to in vitro cell or organoid culture and ex vivo samples. However, simultaneous application of multiple, digitally combined optical techniques (polarimetry, refractometry, multiangle scattering, etc.) is increasingly being shown to have potential use in nerve surgery.

Multimodal contrast agents

Multimodal contrast agents combining modalities such as ¹¹C-PET and MRI have been designed for peripheral nerve regeneration imaging [98] and to assess conditions involving demyelination [99]. In one example, a class of engineered nanoemulsions was used as fluorescence–¹⁹F-MRI contrast agents to discern nerve inflammation after injury and repair and help determine whether there is immune rejection [100]. This contrast agent facilitates imaging of processes such as regeneration and inflammation by targeting immune-related cells. In related work, a class of theranostic and multimodal contrast agents was shown to promote peripheral nerve regeneration [101]. Also, a multimodal imaging agent based on perfluorocarbons and near-IR dye was demonstrated to temporarily open the blood–brain barrier under laser stimulation [102] and therefore could potentially open the blood–nerve barrier and be used for peripheral nerve imaging. In another potential example, an optical–photoacoustic, cell surface-targeting contrast agent such as has been described for imaging of sentinel lymph nodes [103] could similarly be developed for nerves.

Positron emission tomography

PET can be used to assess neuroinflammation at the site of a peripheral nerve injury [55,104] since inflamed tissue has higher metabolic activity. However, it is used far less commonly for peripheral nerve imaging than for CNS imaging [105]. The resolution of PET is on the order of millimeters, which is comparatively poor and limits its usefulness. As a result, PET is most useful for nerve injury in combination with higher resolution, structural imaging such as MRI [106] in order to reference anatomical landmarks that otherwise could not be located.

X-ray computed tomography

X-ray CT for peripheral nerve injury imaging is limited to excised tissue samples because the density of peripheral nerve tissue is not sufficiently different from surrounding tissue to produce appreciable contrast without contrast agent perfusion. CT imaging of peripheral nerves in preclinical models can be performed ex vivo or upon administering contrast agents in a terminal procedure [107]. In this context, x-ray CT has been used to discern myelin microscopically [108], image angiogenesis incident to rat peripheral nerve regeneration [109] and enable nondestructive volume rendering reconstruction imaging.

Raman imaging

Spectroscopy-based imaging of Raman scattering of monochromatic light by peripheral nerves is highly sensitive and specific, able to precisely identify chemical structures. In one variant, microscopy using coherent anti-Stokes Raman scattering was used to chemically image peripheral nerves in living mice down to individual nodes of Ranvier [110]. Ex vivo, Raman spectroscopy was used to quantify biochemical composition associated with axons and myelin that spatially and temporally tracked with injury response [111]. In another example, Raman spectroscopy-based biochemical measurement of a peripheral nerve fiber was used to construct an image-like 2D array that successfully distinguished peripheral nerve, connective tissue and skeletal muscle [112]. Unfortunately, Raman imaging is limited by the inherently weak signal and background signals. However, the signal of a given target molecule can be greatly amplified with surface-enhanced Raman scattering nanoparticles (which can be tuned to the near-IR range), allowing for exciting in vivo biosensing applications currently in development [113].

Future perspective: imaging of nerve grafts & monitoring repair

Visualization of permanently labeled grafts

Imageability and imaging facilitation are underexplored functionalities in NGC design and graft technology that can be used to track the progress of the regenerating nerve. NGCs could be ‘upgraded’ to facilitate real-time in vivo visualization via incorporation of a contrast agent in the scaffold or by using an exogenous contrast agent specific to the regenerating nerve. In principle, most contrast agents, from simple dyes to complex, multimodally functionalized nanoprobes, could be added to NGCs and grafts via tissue engineering methods [114] similarly to, say, paracrine signaling molecules. Furthermore, many therapeutic molecules loaded into NGCs to promote regeneration could be augmented for imaging purposes, such as by conjugating signaling polypeptides with organic dyes before incorporating them into microparticles for controlled release.

To start, contrast agents can be incorporated into NGCs, NGC components or grafts by soaking the implant in contrast agent solution, thus achieving noncovalent binding or by applying a contrast agent gel to the inside wall. This could also be achieved by chemically conjugating the contrast agent to the scaffold via corresponding reactive moieties or strong interaction pairs such as biotin–streptavidin, NGCs whose base material is highly contrastive would already serve this purpose. An optimal solution would incorporate the contrast agent into an existing regeneration-promoting NGC design without significantly changing its other properties. For example, collagen tissue scaffolds, which have been used successfully as a base material for NGCs, can be infused with covalently and noncovalently embedded iron oxide nanoparticles to enhance contrast in MRI [115]. An additional example is perfluorocarbon-labeled tissue scaffolds [116] that can be seen with ¹⁹F-MRI imaging. Enhancement of ‘regular’ NGCs with embedded contrast agents could thus make an NGC more clearly visible under one or more modalities without worsening the imageability of its contents.

Another avenue would be to tag the luminal filling material, for example, use fluorescently labeled or contrast agent-embedded fibrin. If the luminal filling is a solid gel with matrix-bound or mechanically cross-linked contrast agent particles, then its signal will evolve, nonuniformly attenuate, and dissipate over time as contrast agent is freed from the matrix due to routine biochemically mediated degradation of the luminal filling (some of which, if it is ECM, may be reincorporated into nearby tissue); the luminal filling undergoes mechanical compression and displacement as the nascent nerve tissue expands into the NGC; and luminal filling in the vicinity of newly arrived cells is degraded and displaced due to protease secretion and local ECM remodeling. This would be a form of indirect and ‘inverted’ imaging of the nerve.

A somewhat more sophisticated approach would be to make the on/off state of the contrast agent dependent specifically on ‘confrontation’ by a nerve regeneration-associated molecule, such as a protein biomarker, metabolite, cofactor, messenger or hormone. For example, the contrast agent could be attached to the scaffold via a bond that is cleavable by regeneration-associated molecules such as phospholipase A2 or certain matrix metalloproteases. With this configuration, the signal attenuation over time would depend on the cumulative activity of a specific enzyme rather than on general degradation rate.

A possibly bigger challenge is to deliver contrast label to the peripheral nerve tissue itself inside and around the NGC. One potential way is the use of exogenously introduced fluorescently labeled proteins as tracers to monitor the activity of a target protein. For example, systemically introduced fluorescently labeled fibrinogen can be used to image the degradation of fibrin in vivo, a method which could also be applied to nerve regeneration, taking advantage of the important role that fibrin plays in that process. Similarly, systemically introduced fluorescently or radio-labeled fibrinogen can be used to detect bleeding because of its predictable accumulation at the site of a wound (some labeled fibrinogen subsequently becomes part of the fibrin in the wound clot [117]). In summary, tagged ECM or matricellular molecules clearly have potential as imaging agents to track peripheral nerve tissue reorganization during the early stages of regeneration.

Another set of approaches might be matrix- or conduit wall-embedded plasmonic nanoparticles for either surface-enhanced Raman scattering detection of target molecules, plasmon-enhanced Raman absorption imaging of the chemical environment or plasmon-enhanced fluorescence. These methods of Raman imaging via rationally designed NGC-bound nanoparticles may be able to show position-dependent variations in concentration of the target molecule. The spectrum of the plasmonic nanoparticles reflects the ‘average’ local chemical environment, potentially allowing the discernment of different kinds of tissues [118] (e.g., lipid-rich or protein-rich) to reveal detailed images of nerve regeneration. Unfortunately, numerous engineering challenges need to be surmounted to establish such an imaging technique, including protecting the plasmonic surface from nonspecific binding in vivo, which will muddle the signal.

Preclinical NGCs seeded with host-compatible SCs to jumpstart the regeneration process can be augmented by using genetically altered cells expressing fluorescent proteins. Specifically, fibroblasts could be collected from the connective tissue, transfected with fluorescent protein, propagated in culture, chemically induced to convert into SCs [119], and incorporated into an NGC for re-implantation. This would allow in vivo imaging within the NGC as the labeled cells proliferate, migrate and/or are replaced as the endogenous nerve tissue grows. Analogous to tracking of labeled fibrinogen within the ECM cascade, tracking of in vivo fluorescent cells can provide valuable information about the role of a given cell type within the regeneration process.

In addition to having a bright contrast agent, an ideal imageable or imaging-facilitating NGC would consist entirely of nonimmunogenic, biocompatible components that completely degrade or become safely incorporated into tissue at a time scale that matches that of nerve regeneration. Therefore, any optically active nanoparticles for in vivo sensing/imaging should be small so that they can be completely eliminated via the renal excretion pathway and should be composed and/or coated so as to minimize immunogenicity, for example, using polyethylene glycol or gold. Contrast agents, regardless of form, should be covalently attached or strongly integrated into the NGC luminal matrix or wall to ensure that the imaging functionality of the conduit will be intact for as long as the conduit maintains its structural integrity.

Imaging of regeneration using nerve-specific contrast agents

Postrepair nerve imaging research is relevant not only to simple nerve injury repair (with or without grafts), but also to more complex surgeries involving nerves, including reconstructive surgery, organ transplants and composite tissue transplants. Tracking of nerve regeneration is especially needed in transplants since destruction of donor nerve channels by recipient immune cells – rather than occupation by recipient nerves – is a significant risk. In addition, nerves for which the imaging path is obscured due to their location need to be imaged using a modality with a large maximum depth penetration, a consideration that favors near-IR fluorescence and MRI.

The motivation for a molecularly targeted nerve contrast agent imaging approach is to clearly see specific biological structures and events associated with the regeneration process. The basic nanoimaging strategy would be to identify a (uniquely) nerve-associated entity and develop a targeted probe. Cell surface-expressed or matrix-bound markers are preferable for image-based monitoring of regeneration since they provide more specific spatial information and better signal-to-background ratio. Probes are typically constructed from a pool of diverse building blocks (receptor antagonists, antibodies, nanoparticles, luminescent moieties, etc.) to create labeled constructs to amplify the signal. The signal intensity from the reporter correlates with the concentration of the biomarker. Consistent with a common theme in nanomedicine, dual-purpose probes – incorporating both therapeutic molecules and imaging agents – are even more clinically valuable.

Various biomarkers of peripheral nerve regeneration have been described [120], including secreted signaling molecules such as brain-derived neurotrophic factor, surface-expressed axonal growth cone markers transient receptor potential channel 1 (TRPC1) and short transient receptor potential channel 3 (TrpC3) [121], and neuronal damage markers cyclo-oxygenase 2 and prostaglandin E₂ receptor 1 [122]. Unfortunately, axonal regeneration itself is relatively challenging to image via biomarker targeting, as healthy axons and axonal fragments may be inadvertently labeled during ongoing Wallerian degeneration. Additionally, biomarkers of regenerating axons are frequently expressed intracellularly, making them significantly more difficult to target. However, an axon targeting approach could prove more viable for studies working on long time scales.

Cell surface biomarkers expressed on SCs are particularly compelling targets to identify axonal injury and/or growth, as SC phenotype and biomarker expression change in response to both injury and axonal regeneration. Some key biomarkers of nonmyelinating and dedifferentiated SCs have low endogenous expression that is increased upon injury and returns to low levels following axonal regeneration. These include glial fibrillary acidic protein, expressed intracellularly and p75 [123], expressed on the cell surface. Thus, p75 and other differentially expressed SC surface markers could be promising candidates to identify injury and monitor healing progress.

A more indirect regeneration-related imaging target is immune cells, which are normally rare in peripheral nerves but densely populate the immediate vicinity of injuries during healing and regeneration. Most in vivo nerve injury imaging studies to date have used immune activity to distinguish injured or diseased nerves from healthy ones. Immune cells are particularly suitable for nanoimaging because they do not necessarily require active targeting: many immune cells habitually phagocytose all types of nanoscopic foreign objects. The most extensively explored such contrast agents in preclinical peripheral nerve imaging are iron oxide nanoparticles and perfluorocarbon nanoemulsions, which are used with ¹H and ¹⁹F MRI, respectively [124], and which can both be augmented with other functionalities. Greater imaging detail may be achieved using nanoprobe functionalization that promotes binding or phagocytosis by specific types of immune cells, such as pro-inflammatory macrophages, in order to distinguish tissue in different stages of healing. A separate approach, imaging via targeting of immune activity-associated paracrine molecules, may highlight roughly the same spatial extent as imaging the immune cells themselves. While attractive, imaging of neuroinflammation less precisely reflects actual regeneration than does imaging of the nerve fibers themselves, specifically SCs and axons.

More unconventional peripheral nerve regeneration imaging targets may be available in addition to the above-mentioned classic approaches. The mRNA specific to axonal growth cones could be targeted, for which the imaging probe would incorporate microRNA or small interfering RNA. Exosomes, types of which are specifically associated with peripheral nerve injury [125], are a class of complex membrane-bound vesicles with each particle potentially having multiple available targeting modalities. ECM may also be targeted; for example, the bioengineered polypeptide NP41 has great affinity for nerve-specific laminin [126].

The targeting moiety of a probe is usually an antibody by default, typically monoclonal and humanized when used clinically. While incredibly useful, antibodies are not ideal as their large size and chemical complexity leads to unwanted biological interactions. Fortunately, molecular recognition polypeptide technology has, for some biological targets, improved on natural extracts in terms of ‘atomic efficiency’, in other words, polypeptides have been designed with far fewer components that do not directly hug the target. Such polypeptides can be based off of an antibody’s aptamer. Various types of antibody mimics, such as affibodies and avimers, consisting of sequences with high and specific affinity that are not evolved from the target’s antibody(ies), are available. These molecules are typically developed using phage display or another iterative combinatorial selective protein-sequence amplification technique [127]. This process uses a ‘scaffold’ protein with a variable domain and ideally leads to a polypeptide with very high affinity for the particular target.

Even with sleeker targeting moieties, peripheral nerve targeting using dyes and contrast agents is still limited because of problems with delivery and localization within the body, especially when administered systemically rather than injected intraneurally or implanted. The blood–nerve barrier allows only small oligopeptides to diffuse through and actively transports only select classes of full-length proteins, constraining the per-molecule complexity and functionality of labeling agents. One possible workaround may be to constitute a contrast agent from components individually small enough to pass through the blood–nerve barrier that subsequently gain mutual affinity and self-assemble after reacting with solid- or liquid-phase enzymes or other molecules unique to the intraneural space.

Another marker-based strategy is to target enzymes secreted by regenerative cells with probes featuring their respective substrates or other enzyme-sensitive moieties. Such probes are often activatable (beacon-type): the enzyme cleaves a prequenched substrate, generating a fluorophore and the result is observed via the change in intensity or fluorescence lifetime. For example, slow and misdirected axonal outgrowth, which leads to imperfect regeneration, is associated with apoptosis of SCs [128], a process that can be detected by caspase-3-specific near-IR fluorescent probes [129].

Various molecular configurations could be used to improve in vivo imaging targeting and signal, from fluorescently labeled antibodies to increasingly sophisticated, multifunctional nanoprobes. These labeling agents can have diverse structures (e.g., quantum dots, dendrimers, etc.) and diverse functionalities (e.g., radioligands conjugated to fluorescent dyes, rapidly expanding nanobubbles). They can be coated in nonfouling, brush-like textures, red blood cell plasma membrane or other ‘disguises’ to improve circulation lifetime. They can be conjugated to biosignaling molecules such as the Tat protein sequence that allow them to be internalized by cells. And they can be made to be ‘switched on’ in response to a stimulus (e.g., enzymes, ligand–receptor bonding, etc.). These are elements of a toolbox that can be used to design peripheral nerve-specific probes that are precision-targeted, bright and safe.

Although peripheral nerve imaging is currently not performed in typical clinical cases, the combined information available from the various modalities and imaging targets will become increasingly useful in the coming years. This will benefit both injury diagnosis and follow-up to evaluate regeneration. With improvements in imaging capabilities, detailed 3D reconstruction of nerves with fascicle-level accuracy will become available, facilitating creation of custom nerve tissue scaffolding. Advances in micro-additive manufacturing will lead to the ability to replicate tissue in NGCs in detail and confer biophysical properties appropriately tailored to allow the nerve to regrow comfortably. Thus, a rapid, intraoperative protocol may be achieved in which the nerve stumps are imaged and then a custom-made conduit is produced and implanted, leading to greater repair success.

Conceivably, future tissue grafts as well as mass-produced NGCs, whether off-the-shelf or manufactured in situ, could be imbued with a contrast agent in a standard process prior to implantation, thus sidestepping the blood–nerve barrier. The contrast agent could be slowly released from depots located in the implant and then selectively bind to regenerative tissue (such as SCs and axonal processes) or otherwise approximate the spatial extent of the regenerated nerve. The nerve growth front could thus be accurately pictured through the skin in real time, enabling convenient monitoring of nerve regeneration progress. Potentially regrowth-dooming events such as axonal derailment could be seen and/or predicted much more easily and then pre-emptively treated to restore the possibility of full regeneration and reduce failure rate. If this were achieved with fluorescent agents, the nerve could be imaged using a portable and relatively inexpensive digital camera. This would, in turn, allow the regrowing nerve to be viewed on demand outside of a clinical setting, potentially giving early warning of adverse eventualities.

Executive summary.

• In the next decade, in vivo imaging of peripheral nerve regeneration processes will facilitate rapid evaluation of nerve repair materials and techniques and elucidate the delicate biological interplay that drives regeneration.

Imaging of peripheral nerve regeneration in conjunction with repair

• When high imaging depth is necessary, x-ray, nuclear, MRI, ultrasound imaging and their variants can be used to discern some structural and functional features of peripheral nerves, both with and without contrast agents, but at some cost in terms of resolution and contrast.

• None of the currently available imaging techniques are ideal, and high-resolution modalities are typically limited to intra-operative procedures.

• Optical imaging combined with fluorescent tags may hold significant unique promise as a method that can be both highly targeted to specific molecules and, using near-IR or shortwave IR optics, imageable relatively deep in the body, as well as provide good spatial resolution.

Future perspective: imaging of nerve grafts & monitoring of repair

• An ideal peripheral nerve regeneration imaging modality would be able to detect the specific signals from the nerve and the nerve graft and associated information about the ‘chemical environment’ up to several centimeters deep in an organism with single-axon (few μm) resolution.

• Innovations in the development of labeling technologies for nerve repair grafts and regeneration markers will facilitate noninvasive nerve regeneration tracking.