an experimental study on the perineurial window

Journal of the Peripheral Nervous System 7:104–111 (2002)

© 2002 Peripheral Nerve Society, Inc.

104

An experimental study on the perineurial window

Yoshihisa Sugimoto, Shinichiro Takayama, Yukio Horiuchi, and Yoshiaki Toyama

Department of Orthopaedic Surgery, School of Medicine, Keio University

Abstract

Neurological symptoms of herniated nerve fibers resulting from limitedperineurial injury from sharp materials such as needles have become a recent topic in clin-ical practice. However, the mechanism of this disorder, which is known as a perineurialwindow, has not been clarified. To investigate the mechanism of nerve damage in theperineurial window, we designed small (1-mm length) and large (5-mm length) perineurialwindows using tibial nerves of Wistar rats. In the 1-mm group, a marked hernia of the en-doneurial contents developed soon and decreased in size with time, but protrusion ofnerve fibers was still observed after 12 weeks. Nerve fibers in both the herniated portionand under the edge of the window were damaged. Even after 12 weeks, regeneration ofthe nerve fibers and the perineurium was incomplete. In contrast, in the 5-mm group, theinitial endoneurial edema was remarkable, but herniated nerve fibers were not seen after12 weeks. Neurological impairment in the 5-mm group was marked in the early stage butrapidly recovered. The repair of the perineurium and nerve fibers in the 1-mm group wasslower than in the 5-mm group. Persistent neurological symptoms in the perineurial win-dow appeared to be more closely associated with entrapment of nerve fibers at the win-dow edge rather than with disruption of endoneurial homeostasis.

Key words:

perineurial window, perineurium, blood nerve barrier, needle, entrapment

neuropathy

Introduction

A partial perineurial defect is called a perineurial win-dow, and the endoneurial contents become herniatedthrough the defect. It was named by Spencer et al. (

1975

)and is well known as a model of focal demyelination ofnerve fibers. However, the mechanism of nerve damagein the perineurial window is still of research interest.

Recently, clinical cases of nerve injury associatedwith a perineurial window have been identified, and thisphenomenon has attracted the attention of surgeons

(Sugimoto et al., 1998a)

. To help establish effective treat-ment for the clinical problem, we have undertaken astudy to explore the consequences of different sizes ofperineurial windows and to examine the relationship be-tween the size of the perineurial window and nervedamage.

A preliminary model in which the perineurial defectwas 0.5 mm in length was too small for endoneurialcontents to bulge through the window. We found thatmore than 1 mm of window length was needed to pro-duce the endoneurial protrusion, and that the hernia inlarger windows sloped more gently. The experimentalmodels presented below were designed on the basis ofthis preliminary study.

Materials and Methods

One hundred forty-five female Wistar rats weighing200 g to 250 g were used in this study. The animalswere handled according to the animal research guide-lines of the Keio University School of Medicine. We de-signed 2 experimental models of different sizes: 1 mmas a small window model and 5 mm as a large windowmodel. The animals were anesthetized by peritoneal in-jection of Nembutal (35 mg/kg body weight). An inci-sion was made on the lateral aspect of the left thigh,and the left sciatic nerve was exposed. Then, the left

Address correspondence to:

Yoshihisa Sugimoto, Department ofOrthopaedic Surgery, Shimizu Municipal Hospital, 1231 Miyakami,Shimizu-shi, Shizuokaken, 424-8636 Japan. Tel:

�

81-543-36-1111;Fax:

�

81-543-36-1127; E-mail: [email protected]

Sugimoto et al. Journal of the Peripheral Nervous System 7:104–111 (2002)

105

tibial nerve was identified using a surgical microscope,and a perineurial window was established in the middleof its distribution. The method of preparing windowswas as follows. First, the epineurium was dissectedand the left tibial nerve was exposed. A 23-gauge sy-ringe needle was inserted into the perineurium of thetibial nerve; the perineurium was incised using springscissors in the longitudinal direction until the desiredlength was obtained. The epineurium of the sciaticnerve was exfoliated to expose the tibial nerve in thecontrol animals, but no additional surgical maneuverwas performed on the nerve. In these models, func-tional assessment using walking track analysis, andmacroscopic and histological examinations were car-ried out up to 12 weeks. Barrier functions of the periph-eral nerve were also examined.

Evaluation 1: walking track analysis using 25 rats

Evaluation consisted of analysis of neuromuscularfunction using the rats’ footprints. The walking trackanalysis was assessed using the tibial functional index(TFI) described by Hare et al. (

1992

). In accordance withthis method, the hind limbs of the rats were soaked in adeveloping solution, and the rats were made to walk onphotographic paper sloped at an angle of 10

�

. The pho-tographic paper was then soaked in a fixing solutionand dried, and the footprints were measured. For thephotographic paper, developing solution, and fixing so-lution, Kodak RC-3 black and white photographic paper,Kodak D-76 developer and Fuji Fix were used. Ratswere excluded from the analysis if they died or devel-oped infection, or if accurate measurement was impos-sible because of a lack of toe prints. Only data from ratsmeasured weekly until week 8 were evaluated in the1-mm (10 rats), 5-mm (10), and control groups (5). Accord-ing to Hare’s formula,

�

8.8 (not 0) denotes the absenceof paralysis and absence of difference in footprints be-tween the right and left sides. Data were tabulated usingMicrosoft Excel. All recorded values were expressed asthe mean

�

standard deviation. Analysis of data wasperformed at each time interval by the 2-group t-test. Inall statistical analyses, p

�

0.05 was considered statisti-cally significant.

Evaluation 2: macroscopic and histological examinations using 40 rats

Evaluation consisted of visual assessment of theherniation and histological examination. Animals weresacrificed at 1, 4, 8 and 12 weeks after operation. Thenerve was excised and fixed in 2.5% glutaraldehyde for2 hours. For postfixation, 1% osmic acid was used, and24 hours later the nerve specimen was dehydrated andembedded. The transverse sections of the hernia re-gion were stained with toluidine blue and observed un-der a light microscope. Ultrathin sections were double-

stained with uranium and lead salt and observed undera transmission electron microscope (HU-12AS).

Evaluation 3: examination of barrier function using80 rats

The barrier function of the peripheral nerve was ex-amined at 1, 4, 8 and 12 weeks after the operation byRydevik’s method

(Rydevik and Lundborg, 1977)

. Thepermeability of the endoneurial blood–nerve barrier wasexamined by intravenous injection of Evans blue labeled al-bumin (EBA). Thirty minutes after EBA injection in the rightfemoral vein (1 cc/100 g body weight), the nerve was re-moved and thick frozen longitudinal sections were viewedunder a fluorescent microscope to examine leakage of thedye. For the permeability of the perineurial barrier, the sec-tions of nerves were examined after local application ofEBA. The left tibial nerve, in which a perineurial windowhad already been established, was exposed. After immer-sion of the exposed nerve and its surrounding tissues in 2cc of EBA for 2 hours, the nerve was removed. The nervesamples were fixed and frozen in the same manner as ratswho were administered EBA intravenously.

Following local application of EBA in the animals, 4animals each from the 1-mm and 5-mm groups, and 2animals from the control group were examined at theend of 1, 4, 8 and 12 weeks of the study. The totalnumber of animals examined was 40. A total of 40other animals which were administered EBA intrave-nously were also examined in the same manner.

Results

Evaluation 1

In the 1-mm group, the mean value of TFI decreasedto –20 one week after the operation, and this was sus-tained until week 8. In contrast, in the 5-mm group themean TFI value markedly decreased to

�

40 at week 1,but recovered to the normal range within 4 weeks (Fig.1). Significant differences (p

�

0.05) were observed atweeks 4 and 5, and significant differences (p

�

0.01)were observed at and after week 6 between the smalland large window groups. Although a significant differ-ence was noted between the 5-mm group and the con-trol group at the end of week 1 of the study (p

�

0.05),the difference was no longer seen in the subsequentweeks. On the other hand, significant differences wereseen between the 1-mm group and the control groupfrom week 1 to week 8 of the study (p

�

0.05).

Evaluation 2

Immediately after the perineurial incision, the endo-neurium herniated into the epineurial space in both the1-mm and 5-mm groups. In the 1-mm group, remark-able hernia of the endoneurial contents was observed(Fig. 2). The hernia was spherical and remarkable nerve


106

fiber constriction was detected at the edge of the 1mm window. In the hernia region, vasodilatation andedema were noted (Fig. 3). One week after the opera-tion, darkly stained axons were seen in the surfacelayer of the hernia and under the edge of the window(Fig. 4). In the rest of the fascicle there was no signifi-cant nerve fiber disturbance, although there was slightendoneurial edema.

The degree of hernia decreased gradually over time,but remained until 12 weeks. Macroscopically, the her-niated nerve was covered with an epineurium-like thinlayer. Under the surgical microscope, the border of theperineurium was clearly observed (Fig. 5).

Edema and the normal appearance of most of thenerves were observed within the endoneurium at 12weeks (Fig. 6), but nerve fibers with thin myelin and

some fibroblasts were observed at the site of the her-nia. The thickness of perineurium was not completelyrecovered (Fig. 7).

In the 5-mm group, the hernia was dome-shapedand protruded slightly through the window at 1 week(Fig. 8). Marked endoneurial edema was observedthroughout the whole fascicle (Fig. 9). Swollen axonsand darkly stained axons were seen at the herniatedarea at 1 week (Fig. 10). On electron microscopic ob-servation, the formation of minifascicles was foundaround the regenerating nerve fibers at week 4 in the5-mm window (Fig. 11). The degree of herniation de-creased progressively over time. Swelling of the endo-neurium disappeared, and the boundary of the windowwas not distinct at 12 weeks. Macroscopically, the fas-cicle appeared normal (Fig. 12). The number of minifas-cicles increased over time, their size gradually increased,and the hernia surface was covered with perineurium-like cells. Progressive regeneration of nerve fibers andperineurium was observed in the herniated region (Fig.

Figure 2. Macroscopic findings of the 1-mm group at 1week. Remarkable hernia was observed. The arrow showsthe perineurial window.

Figure 3. Light microscopic findings of the 1-mm group at1 week. The hernia was spherical and nerve fiber constric-tion was detected at the edge of the 1-mm window. The ar-row shows the window, and the region above the arrow isremarkable hernia of the endoneurial contents (�200).

Figure 1. TFI (tibial functional index). The damage of the1-mm-long perineurial window was less severe but sus-tained over a long period. In contrast, the damage to the5-mm-long perineurial window was severe, but it recoveredquickly.


107

13). Flat cells with many pinocytotic vesicles, whichwere surrounded by basement membranes, appearedand became tightly joined to each other. In addition to

this, several layers were formed and were consideredto be regenerated perineurium (Fig. 14).

Evaluation 3

The permeability of the blood–nerve barrier in theperipheral nerve near the window increased abnormallyin both groups. At 1, 4, 8 and 12 weeks, intravenous in-jected EBA was spread into the endoneurium, which in-dicated damage of the blood–nerve barrier. Remarkableleakage of the dye was observed near the window, andit spread 3-mm proximal and distal to the margins ofthe windows (Figs. 15A, B).

In the 1-mm group, the perineurial barrier functiondid not return to normal throughout the study. EBA didnot spread into the endoneurium in the 5-mm group at12 weeks, indicating recovery of the barrier function(Figs. 16A, B).

Discussion

The perineurium acts as a permeability barrier, whichselectively controls the entrance of materials into the

Figure 4. Light microscopic findings with high magnificationof the 1-mm group at 1 week. Darkly stained axons wereseen under the edge of the window (�200).

Figure 5. Macroscopic findings of the 1-mm group at 12weeks. Hernia remained at 12 weeks. The arrow shows theperineurial window.

Figure 6. Light microscopic findings of the 1-mm group at12 weeks. Edema and normal appearance of most of thenerves was observed (�40).


108

endoneurial space to create an optimal environment forthe peripheral nervous system

(Kristensson and Olsson,1971; Olsson and Kristensson, 1973)

. Moreover, vascu-

lar endothelial cells of peripheral nerves act as a blood–nerve barrier

(Ahmed and Weller, 1979; Nakao et al.,1995)

. Because of these 2 barrier functions, a positive

Figure 7. Light microscopic findings of the 1-mm group at12 weeks. The perineurium and nerve fibers did not showcomplete recovery (�200).

Figure 9. Light microscopic findings of the 5-mm group at 1week. Marked endoneurial edema was observed. The arrowshows the window (�100).

Figure 8. Macroscopic findings of the 5-mm group at 1 week.Hernia was dome-shaped and protruded slightly through thewindows.

Figure 10. Light microscopic findings of the 5-mm group at1 week. Swollen axons and darkly stained axons were seenat the herniated area at 1 week (�200).


109

pressure is maintained inside the fascicle

(Sunderland,1946)

. For this reason, when the perineurium is dam-aged, endoneurial contents herniate through the defectin the perineurium. This defect was first associatedwith demyelination by Spencer et al. (

1975

) and called aperineurial window.

Nesbitt and Acrand (

1980

) showed that there wassurprisingly little damage to the fascicular contents bystripping of the perineurium. Thus, we supposed thatthere was no severe direct damage caused by openingthe perineurium in this study. However, Nukada et al.(

1992

) measured nerve blood flow with a laser Dopplerflow meter before and after perineurial rupture and sug-gested that ischemia contributes to the process of de-myelination in the perineurial window.

In our study, the area of perineurial injury wassmall, but the hernia was nodular and markedly pro-

truded in the 1-mm group where it was noted even at12 weeks, whereas the border with the perineuriumwas still clear. In the 5-mm group, the hernia protrudedslightly through the window. In the 1-mm group, theproportion of early abnormal fibers was less than that inthe 5-mm group; however, the recovery of perineurialbarrier function and regeneration of nerve fibers wereslower than in the 5-mm group.

Neurological impairment persisted longer when theperineurium injury was smaller, suggesting that entrap-ment of the nerve fibers rather than a disruption of en-doneurial homeostasis was the pathogenic factor in

Figure 11. Electron microscopic findings in the 5-mm groupat 4 weeks. The formation of minifascicles was found aroundthe regenerating nerve fibers.

Figure 12. Macroscopic findings of the 5-mm group at 12weeks. The degree of hernia decreased; the boundary of thewindow was unclear.

Figure 13. Light microscopic findings of the 5-mm group at12 weeks. The degree of hernia in the 5-mm models de-creased consequently. Compartment formation was alreadyin progress around the regenerated nerve fibers. Theperineurium and nerve fibers recovered (�100).

Figure 14. Electron microscopic findings in the 5-mm groupat 12 weeks. Perineurial cells with many pinocytotic vesicleswere surrounded by basement membrane (arrow), and theybecame tightly adjoined (arrowhead) at 12 weeks (bar � 10micrometers).


110

perineurial window pathology

(Sugimoto et al., 1998b)

.Thus, our results indicated that the size of the windowis a factor in the degree of neurological deficit.

Although the percentage of damaged nerve fiberswas very low, the gait disturbance and the dysfunctionof the barrier functions continued for up to 12 weeks.Regeneration of the perineurium and recovery of barrierfunctions were related to the functional recovery. Nervefiber injury was considered to be the cause of per-sistence of pain and numbness. Our data confirm that

perineurial cells provide footholds for nerve fiber regen-eration and readily create minifascicles of regeneratingfibers

(Thomas and Jones, 1967; Lundborg and Hans-son, 1979)

.This study showed that the perineurium formed a

compartment during the process of nerve regeneration,which extended gradually to enclose all of the fasciclesin multiple layers. The cells forming the compartment,despite still being immature, contained the same or-ganelles as in the perineurium. Conversely, the forma-tion of minifascicles proves the existence of the degen-erated nerve fibers.

The perineurium was also considered to play an im-portant role in the induction of the nerve fibers duringthe regenerative process. We presumed that perineur-ial cells facilitated nerve regeneration and formed acompartment. When the cells became mature, thecompartment expanded to such a degree as to allowcomplete regeneration of the perineurium and com-plete enclosure of all of the nerve bundles. The perineu-rium and the nerve fibers regenerated in a manner sug-gesting that one facilitated the other. The regenerativeperineurium was observed in the large-sized perineurialwindow model even when the injury was extensive.However, in the small-sized perineurial window model,the regenerative process of the nerve fibers and theperineurium was probably retarded because of the con-striction at the edge of the window.

The compartments enlarge with maturation ofnerve fibers, and regeneration of the perineurium isconsidered to be complete when the compartmentswrap the entire fascicle. Thus, perineurial cells are im-portant to the induction of nerve fibers in the regenera-tion process

(Schröder et al., 1993; Weis et al., 1994)

.It is also possible that EBA immersed around the

tibial nerve flows into the endoneurium from the sur-rounding vessels. However, this assumption may be re-futed when data from the treated and non-treated ani-mals are compared. The intensity of luminescence inthe endoneurium was weak at the end of 12 weeks inboth the 5-mm group and the control group, and therewas no significant difference between the 2 groups.These findings suggest restoration of the perineurialbarrier in the 5-mm group. Furthermore, when theperineurial barrier does not function well, it is possiblethat intravenously injected EBA enters the endoneu-rium by surrounding bleeding from the defect ofperineurium.

The blood–nerve barrier is easily damaged, even bymild compression that does not cause the degenera-tion of nerve fibers. Although the function of theperineurial barrier is not readily impaired without directinjury of the perineurium, once its function is impaired itdoes not recover sufficiently even after histological re-pair has apparently been completed.

Figure 15. The permeability of the blood-nerve barrier (BNB)in the peripheral nerve near the window increased abnor-mally in both groups. The leak of EBA is represented by redor orange color. At 12 weeks, EBA spread into the endoneu-rium in the 1-mm window model (A) and 5-mm windowmodel (B) at 12 weeks.

Figure 16. The permeability of perineurial barrier is indicated.The leak of EBA is represented by red or orange color. Nor-mal area is dark green. The left wide area is hernia region.(A) The perineurial barrier function in the 1-mm windowmodel was not returned at 12 weeks. (B) EBA did not spreadinto the endoneurium in the 5-mm window model at 12weeks.


111

This study confirmed that focal injury to the peri-neurium results in herniation of endoneurial contentsfollowed by pathological changes in herniated nerve fi-bers accompanied by persistent opening of the blood–nerve barrier. Its recovery was not complete even after12 weeks. In this study, the width of the herniatednerve or the thickness of the perineurium was not com-pletely restored even at the end of 12 weeks in the5-mm group.

Clinically, it has been reported that sharp objectssuch as a needle or a piece of glass can damage periph-eral nerves and that the resultant perineurial windowcauses neuropathy. Perineurial windows must be dis-tinguished from partial lacerations of the peripheralnerve, inducing the lateral neuroma. Clinical problemsare caused invariably by hernia through small windows,and neuropathy persists for more than 1 year in pro-longed cases

(Sugimoto et al., 1998a)

. We assumed thatthe perineurial window is not rare in clinical situations.The window is a very small alteration to the nerve,which might easily be overlooked. Careful dissectionunder a microscope is essential for detection of thewindow. Moreover, it is advisable to consider that aperineural window might be the cause when mildsymptoms of nerve damage continue after injury. If aninjection needle is inserted into the fascicle of a smallnerve about 1 mm in diameter from a slanted angle, theopening is about the same size as the 1-mm perineurialwindow prepared in this experimental study. Neuro-pathy appears to be caused by such a mechanism.Perineurial windows should be suspected when mildneuropathy persists after trauma caused by a sharp ob-ject.

The persistence of nerve damage was associatedwith the size of the window. We suggest that not onlysuturing the perineurium but also enlarging the windowmay be useful in the treatment for perineurial window.Further experimental and clinical studies will be re-quired to establish the most effective treatment forperineurial windows.

Acknowledgements

Thanks are due to the staff of Hand Surgery at KeioUniversity for helpful suggestions.

References

Ahmed AM, Weller RO (1979). The blood nerve barrier and re-constitution of the perineurium following nerve grafting.Neuropathol Appl Neurobiol 5:469–483.

Hare GMT, Evans PJ, Mackinnon SE, Best TJ, Brain JR, SzalaiJP, Hunter DA (1992). Walking track analysis: a long-term as-sessment of peripheral nerve recovery. Plast Reconstr Surg89:251–258.

Kristensson K, Olsson Y (1971). The perineurium as a diffusionbarrier to protein tracers. Acta Neuropathol (Berl) 17:127–138.

Lundborg G, Hansson H (1979). Regeneration of peripheralnerve through a preformed tissue space. Brain Res 178:573–576.

Nakao Y, Sakihama N, Kumagami H (1995). Vascular permeabil-ity changes associated with experimentally induced facialnerve lesions in the rabbit. Eur Arch Otorhinolaryngol 252:255–257.

Nesbitt JA, Acrand RD (1980). Histopathological changes fol-lowing removal of the perineurium. J Neurosurg 53:233–238.

Nukada H, Powell HC, Myers R (1992). Perineurial window: De-myelination in nonherniated endoneurium with reducednerve blood flow. J Neuropathol Exp Neurol 51:523–530.

Olsson Y, Kristensson K (1973). The perineurium as a diffusionbarrier to protein tracers following trauma to nerves. ActaNeuropathol (Berl) 23:105–111.

Rydevik B, Lundborg G (1977). Permeability of intraneural mi-crovessels and perineurium following acute, graded experi-mental nerve compression. Scand J Plast Reconstr Surg 11:178–187.

Schröder JM, May R, Weis J (1993). Perineurial cells are thefirst to transverse gaps of peripheral nerves in siliconetubes. Clin Neurol Neurosurg 95:S78–S83.

Spencer PS, Weinberg HJ, Raine CS, Prineas JW (1975). Theperineurial window: a new model of focal demyelination andremyelination. Brain Res 96:323–329.

Sugimoto Y, Takayama S, Horiuchi Y (1998a). Perineurial win-dow on the digital nerve. A case report. Hand Surg 3:201–203.

Sugimoto Y, Takayama S, Horiuchi Y (1998b). An experimentalstudy of the perineurial window. 7th Congress of the Inter-national Federation of Societies for Surgery of the Hand.Roth JH, Richards RS (Ed). Monduzzi, Bologna, Italy, pp641–645.

Sunderland S (1946). The effect of rupture of the perineuriumon the contained nerve fibers. Brain 69:146–152.

Thomas PK, Jones DG (1967). The cellular response to nerve in-jury. J Anat 101:45–55.

Weis J, May R, Schröder JM (1994). Fine structural and immu-nohistochemical identification of perineurial cells connectingproximal and distal stumps of transected peripheral nervesat early stages of regeneration in silicone tubes. Acta Neuro-pathol (Berl) 88:159–165.

an experimental study on the perineurial window

Documents