Second Harmonic Generation Imaging as a Virtual Biopsy for Upper Extremity Nerve Injuries: A Cadaver Study

Abstract

Background:

A timely assessment of nerve damage is needed for early clinical decision-making. Second Harmonic Generation (SHG) microscopy visualizes collagen. This study investigated: Can SHG imaging distinguish collagenous substructures in human nerves? Can SHG imaging detect structural damage in human nerves following destructive stretch injury?

Methods:

Six human upper extremities were equally divided into 2 groups: A no-injury and a load-to-failure (LTF) group. The median, radial, and ulnar nerves were surgically exposed. Arms were placed on an Instron material testing system. Eight centimeters of each nerve was secured under 2 pins. A hook was raised from beneath the nerve until complete nerve transection occurred. After the experiment, LTF nerves were excised along their full length. No-injury and LTF nerves were placed in isotonic saline under an FVMPE-RS Multiphoton Microscope using a laser wavelength of 900 nm to induce an SHG signal. Z-stack images were acquired using a wavelength of 45 nm. Nerves were then harvested for histology.

Results:

Histology sections of NI nerves exhibited normal peripheral nerve architecture. All collagenous substructures visible on histology were clearly identifiable and distinguishable on SHG images of uninjured nerves. In LTF nerves, epineurium rupture and exposed fascicles were clearly identifiable on SHG imaging. Epineurial collagen of LTF nerves appeared heavily disorganized, with short fiber fragments following no clear trajectory. Findings were consistent among nerve types.

Conclusions:

This is the first study to visualize human nervous tissue using SHG microscopy. Second Harmonic Generation imaging offers detailed visualization of all collagenous substructures of peripheral nerves and detects structural damage, like epineurial collagen-disorganization, and exposure of individual fascicles in unprecedented detail.

Introduction

Peripheral nerve injuries are highly devastating injuries that affect roughly 5% of patients admitted to level I trauma centers, and iatrogenic nerve damage is a risk across almost all orthopedic subspecialties.^1-3 Despite extraordinarily high treatment expenses, patients are plagued by poor functional outcomes and high rates of long-term disability. ⁴ Stretch injuries are among the most devastating forms of nerve injury and common scenarios include shoulder dislocations and displaced humerus fractures. The appropriate treatment depends highly on the extent of nerve damage.^5,6 Surgical interventions provide the most favorable patient outcomes when performed shortly after injury. ⁷ Thus, a timely assessment of injury severity is paramount to indicate the correct treatment and increase chances for recovery. ⁷ However, the degree of structural damage is most often unknown for in-continuity nerve injuries.

Even mild injuries with full regenerative potential are believed to cause an acute, temporary loss of nerve function (neuropraxia/preneuroclasis injuries), which makes an acute assessment of function often inconclusive.^5,8,9 While ultrasound evaluation of injured peripheral nerves has gained popularity in recent years, it lacks the ability to visualize specific substructures within the nerve, strongly limiting its capability to differentiate between distinct degrees of nerve injury.^10-12 To date, no diagnostic tool has the ability to diagnose damage to the different microstructures and connective tissues within peripheral nerves. Development of such a tool could facilitate the distinction of injuries that might recover naturally from those requiring immediate surgical intervention in the acute clinical setting.

Second Harmonic Generation (SHG) imaging has previously been investigated as a novel assessment of nerve damage in rats.^13,14 Second Harmonic Generation describes the secondary nonlinear optical phenomenon that allows label-free microscopic visualization of collagen, and thus many biological tissues without the need for tissue biopsy and histological processing.^15-17 When 2 photons of equal energy strike a noncentrosymmetric molecule like collagen, a photon of double the frequency and half the wavelength is emitted as visible light, enabling visualization of this specific wavelength without capturing background light.^18,19 Peripheral nerves contain large amounts of collagen, making SHG microscopy a promising tool for visualizing a nerve’s collagenous microarchitecture.^20,21 Collagen fibers provide the framework along which damaged axons regenerate. A high degree of endoneurial collagen fiber disorganization could potentially be predictive of poor recovery and neuroma formation. A potential benefit of SHG microscopy over conventional biopsy and histopathological assessment could be its “noninvasive” nature in regards to nerve integrity, as SHG does not require incision of the nerve and has been performed in-vivo in rodents. ¹³ Previous studies demonstrated the ability to visualize epineurial collagen fibers via SHG microscopy in rats.^13,14,22 However, to the best of our knowledge, no study to date has demonstrated the use of SHG imaging in human peripheral nerves. Moreover, no study has demonstrated the ability to visualize the individual collagenous microstructures of both healthy and damaged nervous tissue using SHG microscopy in human nerves.

This study therefore sought to answer 2 research questions: (1) Can noninvasive SHG imaging identify and distinguish the individual collagenous substructures in human radial, median, and ulnar nerves? (2) Can SHG imaging detect structural damage in human nerves following destructive in-situ stretch injury?

Methods

Six fresh-frozen human upper extremities (from the humeral head to finger tips) were thawed at room temperature for 12 hours and equally divided into 2 groups: A no-injury (NI) group and a load-to-failure (LTF) group (n = 3 arms each). The median and ulnar nerves of all 6 specimens were surgically exposed and gently separated from surrounding tissue in the forearm, while the radial nerves were exposed in the dorsal upper arm through a posterior skin incision. Load-to-failure nerves (n = 9) were stretched until full nerve disruption occurred, and NI nerves (n = 9) were not subjected to tension. Each nerve was excised and placed under a multiphoton microscope to generate SHG images. Uninjured nerves were histologically processed for comparison with SHG images. Excision of peripheral nerves from cadaveric upper extremities prior to SHG nerve imaging was performed to ensure compliance with biohazard regulations at our institution.

Nerve Exposure Surgery

Following an incision of the skin and subcutaneous tissue in the ventral forearm, the median and ulnar nerves were identified at the level of the carpal tunnel. Both nerves were gently separated from surrounding tissue from the carpal tunnel to the elbow using microdissection tools. To expose the radial nerve, an incision along the dorsal upper arm was followed by an incision of the triceps muscle to find the underlying radial nerve along its course around the dorsal humerus shaft.

Load-to-Failure Nerves

To induce LTF stretch injury in human peripheral nerves, we scaled the neuroclasis stretch injury model from the rat median nerve to a human-sized pin-and-hook apparatus.^9,23,24 To our knowledge, this represents the first use of a pin-and-hook system to induce LTF stretch injury in human cadaver nerves. In LTF nerves, the proximal ends of the ulnar, median, and radial nerves in the axilla were surgically exposed and secured in place using sandpaper-coated sandwich clamps attached to an immobilized metal bar. Arms were then tied down onto an aluminum bar which was placed under an Instron material testing system (model 8872, Instron, Norwood, Massachusetts). Eight centimeters of each exposed nerve was positioned under 2 blunt metal pins to recreate anatomic fulcrums.^9,25 Mimicking a displaced fracture edge, a metal hook was positioned beneath the nerve and raised at a speed of 0.5 mm/s until complete nerve transection occurred (Figure 1). ²⁶ Median and ulnar nerves were stretched in the approximated middle of the distance from the carpal tunnel to the elbow, while radial nerves were stretched at the approximated middle of the humerus shaft. After the experiment, nerves were excised along their full length and placed in isotonic saline.

Figure 1.

Nerves were loaded to failure using 2 pins 8 cm apart securing the nerve proximal and distal to the hook stretching the nerve, which was elevated at 0.5 mm/s until full nerve rupture occurred.

Note. The pins aim to mimic anatomic fulcrums, such as the elbow and carpal tunnel, whereas the hook aims to recreate a displaced fracture edge stretching the nerve.

Second Harmonic Generation Nerve Imaging

No-injury nerves and LTF nerves were placed in isotonic saline below a custom-made 3 × 4 cm glass cover slip and positioned under an FVMPE-RS Multiphoton Microscope for SHG imaging (Figure 2). Anatomic landmarks, such as the elbow and carpal tunnel, were marked via tissue dye along NI nerves, which were then imaged at approximately the same location where LTF nerves were stretched. Load-to-failure nerves were imaged from the site of full nerve transection to multiple centimeters proximal and distal of the transection site. Second Harmonic Generation images were acquired with Olympus Fluoview (FV31 S-SW) software on an Olympus FVMPE-RS Multiphoton Microscope (Olympus, Tokyo, Japan). Imaging was performed using a 10x/0.6NA TruResolution XL Plan N objective lens (XLPLN10XSV, Olympus, Tokyo, Japan); frame size was set to 512 × 512 pixels (X/Y) for a lateral pixel size of 2.26 µm. The tunable laser line of the Insight X3 Dual line multiphoton laser was set to 900 nm to obtain an SHG signal at 450 nm in a 410- to 460-nm band pass filter. ²⁷ Z-stacks, referred to as SHG images throughout this manuscript, were obtained with a z-step size of 8 μm/step including 30 steps per stack to obtain a stack depth of 240 μm. Imaris Viewer 10.2.0 software was used to generate 3D reconstructions.

Figure 2.

For Second Harmonic Generation (SHG) imaging, nerves were placed in isotonic saline below a custom-built glass cover. The immersion lens was in contact with ultrasound gel to facilitate SHG imaging.

Histology

Following SHG imaging of NI nerves, approximately 4-cm-long sections were excised and fixed in 10% zinc-formalin, upon which they were embedded in hydrophobic acrylic resin for high-fidelity histology. ²⁸ Sections were cut longitudinally, deplasticized, and stained with hematoxylin & eosin (H&E) for general tissue morphology. Nerve sections were evaluated for presence of normal histological features including the epineurium, perineurium, endoneurial tubes, axons, and adipose tissue. Images were obtained using a Hamamatsu NanoZoomer S210 slide scanner (https://icahn.mssm.edu/research/resources/deans-cores/biorepository-and-pathology/services-and-equipment).

Results

No-Injury Nerves: Histology

Hematoxylin & eosin–stained histology sections of NI nerves exhibited normal nerve microarchitecture. ²⁹ The superficial layers of the epineurium were composed of loose collagen fibers, whereas densely packed, highly organized collagen layers were found deeper within the epineurium, which was pervaded by longitudinal threads of adipose tissue.^29,30 Nerve fascicles were surrounded by a thin perineurial layer and contained nerve fibers enclosed in endoneurial collagen tubes. Axons of uninjured nerves followed a physiological wave pattern, which is often visible as bands of Fontana.^31,32

No-Injury Nerves: SHG Imaging

All anatomic substructures visible on histology were clearly identifiable and distinguishable on noninvasive, label-free SHG imaging of uninjured nerves (Figure 3). Second Harmonic Generation microscopy provided longitudinal cross-sectional z-stack images, allowing the operator (CAS) to “scroll” through the nerve in a z-direction. Superficial, loose collagen fibers appeared disorganized on SHG imaging, whereas deeper, densely packed epineurial collagen had a well-organized appearance with a collagenous wave pattern. Notably, longitudinal threads as well as large accumulations of adipose tissue were clearly visible as vesicular structures. On SHG imaging, the perineurium of individual fascicles was easily discernible, revealing a weave-like perineurial sheath formed by obliquely arranged fibers that separated it from the surrounding epineurial collagen. Similar to histological findings, endoneurial tubes within nerve fascicles demonstrated a characteristic wave pattern.^31,32 Second Harmonic Generation imaging provided insights into the nerve’s microarchitecture similar to conventional histology, demonstrating its use as a noninvasive virtual nerve biopsy.

Figure 3.

Hematoxylin & eosin–stained histology of uninjured no-injury nerves (a, c, e, g) in comparison with Second Harmonic Generation (SHG) images of similar structures (b, d, f, h).

Note. Densely packed epineurial collagen exhibited a wave pattern on histology (a) that was clearly visible on SHG imaging (b). Similarly, perifascicular collections of adipose tissue were visible as vesicular structures on both histology (c) and SHG images (d). The thin perineurium of individual fascicles (e, blue lines) demonstrated thick, discernable collagen fibers wrapped around the fascicle on SHG imaging (f, blue lines). The endoneurial wave pattern that is characteristic for uninjured peripheral nerves was also visible on both histology (g) and SHG (h) imaging.

Load-to-Failure Nerves

Second Harmonic Generation imaging of LTF nerves revealed gross connective tissue damage after destructive stretch injury. On visual inspection, LTF nerves exhibited macroscopic disruption of the epineurium multiple centimeters proximal and distal to the site of macroscopic nerve transection. This epineurial disruption exposed multiple individual fascicles lacking an epineurial cover (Figure 4). Exposed fascicles and a disrupted epineurium were also observed in the location of both hold-down pins despite macroscopic nerve continuity in this area, likely due to friction around both pins.

Figure 4.

A disrupted epineurium (arrow marks the rupture edge) was seen in load-to-failure nerves multiple centimeters proximal and distal to the site of full nerve transection.

Note. This epineurial disruption revealed individual fascicles, the microstructure of which could be assessed using Second Harmonic Generation imaging. The clinical occurrence of epineurial disruption in in-continuity stretch injuries and its implications on function and recovery have yet to be investigated.

During SHG imaging, epineurial disruption and exposed fascicles were easily identifiable and clearly visible on obtained images. Epineurial collagen of LTF nerves appeared heavily disorganized, with short fiber fragments following no clear trajectory (Figure 5). Exposed nerve fascicles lost both their epineurial and perineurial sheath but maintained a compact, fascicular structure. Second Harmonic Generation imaging facilitated the visualization of individual endoneurial tubes in damaged human peripheral nerves. Endoneurial fibers lost their characteristic wave pattern on SHG images, which indicates plastic deformation due to strain exceeding the physiological limits of stretching.^9,33,34 The epineurial SHG signal of injured nerves was less intense than that of undamaged nerves, requiring increased filter voltage for effective visualization.

Figure 5.

Second Harmonic Generation (SHG) imaging of load-to-failure (LTF) nerves.

Note. The disrupted epineurium consisted of heavily disorganized superficial collagen fibers following no clear trajectory (a). Deeper epineurial layers demonstrated a more compact and dense collagen architecture after stretch injury; however, fibers still appeared heavily disorganized (b). The epineurium and perineurium of individual fascicles were fully disrupted and retracted proximal and distal to the site of nerve transection, exposing a “naked” endoneurial core, which consisted of parallel endoneurial fibers on SHG imaging (c). Three-dimensional reconstructions of LTF nerve fascicles confirmed the lack of a perineurium and facilitated an assessment of endoneurial tube alignment (d).

Three-dimensional reconstructions of z-stack images from LTF nerves confirmed joint rupture of the epineurium and perineurium multiple centimeters proximal and distal of the disruption site of endoneurial structures (Figure 5d). Furthermore, 3D reconstructed images revealed extensive epineurial fiber disorganization within a z-direction, demonstrating a loss of fiber alignment across the different epineurial collagen layers. Findings were consistent among all 3 nerve types assessed in this study.

Discussion

To the best of our knowledge, this is the first study to visualize human nervous tissue using noninvasive SHG imaging. We sought to investigate whether SHG microscopy can be used to visualize the microarchitecture of the human radial, median, and ulnar nerves, demonstrating its use as a virtual tissue biopsy without damaging the nerve. A second aim was to investigate how damage to these nerves presents on SHG imaging when stretched to failure. Label-free SHG microscopy revealed all major anatomical substructures in similar detail as histological evaluation and additionally allowed for cross-sectional visualization and 3D reconstruction of human peripheral nerves, demonstrating key advantages of SHG imaging over conventional diagnostic imaging techniques, such as magnetic resonance imaging (MRI) and ultrasound. We further demonstrated that damage to a human nerve’s different connective tissues can be identified using label-free SHG imaging. A detailed intraoperative assessment of a nerve’s collagenous structures could guide the treatment for peripheral nerve trauma and detect nerve damage that is indicative of poor outcomes and requires immediate surgical intervention, allowing for a more informed surgical decision-making process.

Intraoperative visualization of nerve structure has multiple potential advantages over other diagnostic methods for nerve damage. Even mildly injured nerves with full regenerative potential (preneuroclasis injury/neuropraxia) often demonstrate a temporary, complete loss of function, making electrophysiological assessments of nerve conductivity highly inconclusive.^5,8,9 Nerve biopsy inherently damages the nerve, and both ultrasound and MRI fail to visualize anatomic substructures and diagnose distinct degrees of nerve injury.^10,35,36 Novel advances in nerve imaging include indocyanine fluorescent imaging. ³⁷ However, this technique primarily visualizes vascular structures, limiting its utility to an evaluation of (mal-)perfusion following injury. Gluck et al ¹³ demonstrated the use of SHG imaging to visualize epineurial collagen in vivo in rats. While this proves the ability to use SHG imaging in vivo, their study was limited to a visualization of epineurial collagen. However, the endoneurial collagen microarchitecture of peripheral nerves provides the extracellular scaffold along which damaged axons regenerate, and damage to this endoneurial framework in particular can significantly reduce the nerve’s ability to recover. ³⁸ This study demonstrates the ability to visualize intraneural connective tissue of human peripheral nerves via SHG imaging, including individual endoneurial tubes, providing unprecedented insight into the nerve’s structural organization and integrity without damaging the nerve, as a biopsy inherently would. A disrupted perineurium and epineurium mark a severe impairment of integrity, potentially hindering successful nerve regeneration. When used intraoperatively, SHG imaging could detect both epineurial and endoneurial damages predictive of long-term dysfunction and poor regeneration in the acute clinical setting, warranting immediate surgical intervention. Future preclinical studies should investigate the long-term implications of perineurial and epineurial collagen damages and disruption for nerve regeneration, to identify clear hallmark signs of nonrecoverable injuries.

Potential Application of SHG Nerve Imaging in the Acute Clinical Setting

We describe SHG imaging as “noninvasive,” as no incision of nervous tissue is required and safe in-vivo use has been demonstrated in rodents, making SHG noninvasive to the nerve itself. That said, the laser needs to be in direct contact with the nerve for satisfactory tissue visualization, and SHG imaging is thus an inherently intraoperative procedure that will require at least some degree of exploratory nerve surgery. Therefore, a clinically usable SHG microscope will need to fulfill all safety and sterility requirements that are paramount for use in an open surgical field.

Potential clinical applications of SHG imaging include damage assessment during fracture repair surgery, such as in the case of a humeral shaft fracture with radial nerve palsy. It may also be employed to ensure that no supraphysiological strain was placed on a peripheral nerve during orthopedic procedures such as reverse shoulder arthroplasty.^3,39,40 Given the microscopic detail of SHG images, it can be seen as a label-free virtual biopsy. Nonetheless, limited laser penetration depth remains a major constraint of SHG imaging. Thus, only superficial fascicles receive enough laser exposure to generate an SHG signal strong enough to enable seamless visualization. For deeper tissue visualization, the laser intensity needs to be increased considerably. Therefore, although initial clinical investigations demonstrated the general safety of SHG imaging in humans, the safety of high laser intensities should be investigated in preclinical models to ensure that SHG imaging does not lead to increased scar tissue formation. Ultimately, a future intraoperative application of this technology could guide the treatment for upper-extremity nerve injuries and detect nerve damage that requires immediate surgical intervention, allowing for a more informed surgical decision-making process.

Limitations

An inherent limitation of this study is the use of cadaveric specimens. Even in fresh cadaveric specimens, dead tissue potentially exhibits a different reaction to strain and requires different SHG signal intensities than living human nervous tissue. In addition, recovery and outcomes after nerve injury cannot be assessed in cadaver studies and could thus not be correlated to damage on SHG imaging. A quantitative analysis would primarily be useful to identify differences in the width, length, and alignment of individual collagen fibers between multiple distinct, clinically representative degrees of injury. However, the two groups in this study included a transection stretch injury model and uninjured nerves with a vastly different presentation on qualitative evaluation of SHG images. Thus, a quantitative analysis was not required to answer this study’s research questions. An in-continuity stretch injury is the clinically most challenging type of nerve injury and represents the injury type that a diagnostic tool like SHG imaging would be most useful for. ⁶ Accordingly, future studies will need to develop cadaver models that can induce clinically representative grades of in-continuity stretch injury, for which a quantitative analysis of SHG images should be performed. Similarly, future long-term animal studies should include a quantitative analysis of collagen fiber (dis-)organization at different levels of injury, to identify thresholds of epineurial and endoneurial disorganization that indicate poor long-term outcomes.

We used a transaction stretch injury model for this proof-of-concept study to show that SHG imaging can identify gross damage to all substructures. In this model, the disruption of the epineurium and perineurium were decoupled from fascicular disruption, and the endoneurial core was exposed by disruption of the epineurium. This finding is consistent with the sequence of structural failure during stretching in rat nerves, which also starts at the epineurial level. ⁹ However, the clinical occurrence and implications of epineurial disruption in human nerves have yet to be investigated.

Findings were consistent among radial, median, and ulnar nerves. However, subtle differences in collagen fiber alignment and epineurial sheath thickness are possible and would require a quantitative analysis of SHG images, which was not done in this study for the aforementioned reasons. Overall, despite these limitations, this study successfully demonstrates the first ever use of SHG imaging on human nervous tissue and its ability to serve as a virtual biopsy to visualize nerve microstructure and assess connective tissue damage.

Excision of peripheral nerves from cadaveric upper extremities prior to SHG nerve imaging was performed to ensure compliance with biohazard regulations at our institution. The excision of nerves was not an anatomical or methodological necessity to facilitate unobstructed imaging, and we believe that intraoperative in-vivo SHG imaging is methodologically and anatomically doable in humans, similar to an in-vivo application in rats as demonstrated by Gluck et al. ¹³ Nonetheless, future work in human cadavers should demonstrate a true in-vivo application of SHG imaging for in-continuity stretch injuries.

Conclusion

Label-free SHG imaging—a method that visualizes collagen fibers—can identify and distinguish the individual collagenous substructures in human radial, median, and ulnar nerves, providing details comparable to histological evaluation. Second Harmonic Generation imaging also detected structural damage following transection stretch injury, revealing disruptions of the epineurium and perineurium, disorganized epineurial fibers, and altered endoneurial tube alignment. As the first reported use of SHG imaging in human nervous tissue, this proof-of-concept highlights this technique’s potential as a label-free, intraoperative “virtual biopsy” during orthopedic surgery. Given that the collagenous microarchitecture of peripheral nerves provides the scaffold along which axons regenerate, a method capable of assessing microstructural damage in the acute clinical setting could serve as a prognostic tool for orthopedic nerve trauma. More preclinical and clinical research is needed to identify the appropriate use cases and safety of SHG imaging before it could be employed in routine surgical practice.