Comparison of dural grafts in Chiari decompression surgery: Review of the literature

in spinal subarachnoid space, spinal compression, and intracranial Abstract Background: Decompression of Chiari malformation is a common procedure in both pediatric and adult neurosurgery. Although the necessity for some bony removal is universally accepted, other aspects of Chiari surgery are the subject of debate. The most controversial points include the optimal amount of bony removal, the use of duraplasty (and the type of material), the need for subarachnoid dissection, and the need for tonsillar shrinkage. Material and Methods: We critically reviewed the literature to elucidate the risks and bene ﬁ ts of different graft types and to clarify optimal treatment options therein. Based on our search results, 108 relevant articles were identi ﬁ ed. With speci ﬁ c inclusion and exclusion criteria, we noted three studies that directly compared two tlpes of dural substitutes in Chiari malformation surgery. Results: Our review did not support the superiority of either autologous or nonautologous grafts when duraplasty is employed. Our institutional experience, however, dictates that when the pericranium is available and of good quality, it should be utilized for duraplasty. It is non-immunogenic, inexpensive, and capable of creating a watertight closure with the dura. Conclusions: Discrepancies between the three comparative studies analyzed are likely attributable to increases in pericranial quality and thickness with maturity. Future randomized studies with large numbers and the power to resolve differences in the relatively low rates of complications in Chiari surgery are warranted.


INTRODUCTION
Decompression of Chiari malformations is commonly performed in both pediatric and adult neurosurgery. Th e surgery involves a suboccipital craniectomy and, in most cases, a C1 laminectomy, to relieve the spectrum of symptoms associated with tonsillar herniation below the foramen magnum. [1] Chiari decompression is also an eff ective treatment for syringomyelia, which is frequently associated with Chiari malformations. [2] Th e proposed pathological mechanism underlying the progressive symptoms of Chiari malformation involves tonsillar obstruction of the normal fl ow of cerebrospinal fl uid (CSF) between the spinal and cortical subarachnoid spaces. Th is disturbance can lead to syringomyelia, CSF trapped in the spinal subarachnoid space, spinal cord compression, and increased intracranial Although decompression of Chiari malformation is a proven and eff ective treatment, several variations of the procedure have been proposed. However, there is no clear consensus among the neurosurgical community regarding the most effi cacious technique. Although the necessity for some bony removal is universally accepted, other aspects of Chiari surgery are debated. Th e most controversial points include the optimal amount of bony removal, the need for duraplasty (and the type of material), the need for subarachnoid dissection, and the need for tonsillar shrinkage. [2,[4][5][6][7][8][9] Additional variables include the use of postclosure dural/graft sealants, the type of dural/graft suture, whether to tent the dural graft to the remaining bone or a suboccipital metallic plate, and the use of adjunctive syrinx shunting when a syrinx is present.
Th e proponents of bony decompression alone assert that this method aff ords protection from postoperative complications related to dural opening and entering the subarachnoid space. Although the optimal amount of bony removal remains undefi ned, [2] bony decompression without duraplasty decreases operative time and hospital stays. [5] Furthermore, brainstem auditory-evoked potentials improved the most aft er bony decompression with litt le additional gain associated with dural opening. [4,9] Some authors have advocated the use of intraoperative ultrasonography as a means for assessing tonsillar compression aft er bony removal, with persistent compression or lack of pulsation supporting the need for dural opening and duraplasty. [6][7][8] Despite the lack of consensus on which surgical methods are necessary and suffi cient to relieve Chiari pathology, most neurosurgeons perform both bony decompression and duraplasty. [2] In a recent meta-analysis that reviewed seven series with 582 patients altogether, posterior fossa decompression with duraplasty was associated with a lower risk of reoperation than bony decompression alone but with a greater risk of CSF-related complications. Th e reviewers did not stratify the data by type of graft . Interestingly, they found no statistically signifi cant diff erence between the treatments in terms of clinical improvement or resolution of syrinx. [3] Although duraplasty remains popular and has many advocates, [1,10,11] the optimal type of dural graft is debated. A number of graft types, both autologous and nonautologous, are available to the neurosurgeon. Dural substitutes used in Chiari I decompression surgery include autologous pericranium, [1,10,[12][13][14] bovine pericardium, [12] cadaveric dura, [1,12] synthetic bovine collagen matrix (Duragen, Integra Neuroscience, Plainsboro, NJ), [11] acellular human dermis allograft (AlloDerm, LifeCell Corp., Branchburg, NJ), [11,15] autologous fascia lata, [12,16] expanded polytetrafl uoroethylene (PTFE), [10,17,18] posterior atlantooccipital membrane, [19] splenius capitis muscle fl ap, [20] and porcine small intestinal submucosa (Durasis, Cook Biotech, Inc., West Lafayett e, IN). [21] A recent survey of pediatric neurosurgeons by the American Association of Neurological Surgeons estimated graft preferences: 32% preferred autologous pericranium, 32% preferred bovine pericardium, 17% preferred lyophilized cadaveric dura, 16% preferred synthetic products, 4% preferred ligamentum nuchae, and 3% preferred fascia lata. [22] When a dural opening is made for a Chiari I decompression, these graft s are used to create room for the cerebellar tonsils by expanding the potential space posterior to the hindbrain at the foramen magnum. Th eoretically, the ideal graft would provide watertight closure without promoting arachnoid scarring or provoking an infl ammatory response.
Th e current literature refl ects the debate regarding the impact of dural graft material on several clinical outcomes measures, most notable are the need for reoperation and the incidences of postoperative CSF leak, pseudomeningocele fromation, meningitis (bacterial or aseptic), wound breakdown. Adverse reactions to nonautologous (e.g., synthetic, allogenic, xenogenic) graft s include graft dissolution, encapsulation, foreign-body reaction, excessive scarring, and adhesion formation. [21] Aft er initially successful Chiari I surgery, many recurrences may be due to a foreign body reaction, scarring, intradural adhesions, and meningitis-induced hydrocephalus. [4,23,24] Th e thickness and assimilation of the graft can also aff ect its ease of manipulation, workability, and role as a watertight sealant to prevent extradural egress of CSF and ingress of blood and contaminants. [14] We performed a critical review of the literature to elucidate the risks and benefi ts of diff erent graft types and to clarify optimal treatment options.

Search criteria
Th e search terms "Chiari Malformation AND Duraplasty," "Chiari Malformation AND Pericranium," "Chiari AND Graft AND Duraplasty," and "Chiari Malformation AND Graft " were separately entered in a Medline search in PubMed (November 7, 2009).

Study selection
Th e inclusion criteria were as follows: articles that involved Chiari I/II malformation surgery, articles that addressed 'duraplasty' in the title, and articles that evaluated two separate dural substitutes.
Th e exclusion criteria were as follows: articles that did not specifi cally report rates of resolution of reoperation or complications including meningitis, CSF leaks, and wound infections; articles comparing outcomes obtained with bony decompression alone versus those obtained with bony decompression and duraplasty; articles that reported outcomes in fewer than 10 patients; and articles writt en in a language other than English.

Statistical analysis
Data analysis was performed using the Fisher-exact test and Chi-square test for independence of categorical variables. All probability values were two-sided, and a P-value less than 0.05 was considered to be statistically signifi cant.

Search results
Based on the search results, 108 articles were identifi ed. Aft er applying the inclusion and exclusion criteria, we identifi ed three studies that compared two types of dural substitutes in Chiari malformation surgery. A fourth study [25] comparing pericranium versus synthetic graft was identifi ed. [14] However, it dealt with brain tumor resection in addition to Chiari decompression and was therefore excluded. A fi ft h study comparing autologous and nonautologous graft material was excluded because the full text was only available in Chinese (abstract in English). [26]

Studies included and characteristics
Th e three studies [Tables 1 and 2] compared pericranial autograft with synthetic allograft (GORE PRECLUDE MVP Dura Substitute (expanded PTFE)), [10] pericranial autograft with cadaveric dura, [1] and acellular human dura (AlloDerm (LifeCell Corp, Branchburg, NJ)) with bovine collagen matrix (Duragen, Integra Neuroscience, Plainsboro, NJ). [11] Th e studies were all published in peer-reviewed journals between 1997 and 2009 and were reported from single academic centers. Two of the articles were retrospective studies [10,11] and one employed a sequential, prospective analysis of the second cohort of 13 patients aft er switching from a nonautologous graft to an autologous graft aft er the fi rst set of 13 patients. [1] Outcomes measured Th e rates of superfi cial wound infections, bacterial meningitis, aseptic meningitis, CSF leaks, pseudomeningocele formation, and need for reoperation were evaluated. However, not all measures were reported in all three studies.

Study 1
In a report from the Johns Hopkins group, Att enello et al. [10] retrospectively compared the use of GORE PRECLUDE MVP dural substitute (expanded polytetrafl uoroethylene [ePTFE] graft ) with pericranium in 67 patients with a mean age of 11 years (standard deviation 5 years); 40 of these patients received an autologous pericranial graft . Variables of interest [Tables 3 and 4] included reoperation, wound infection, aseptic meningitis, pseudomeningocele, and CSF leakage. Indications for reoperation were not provided.
No patient who received an ePTFE graft experienced a CSF leak or symptomatic pseudomeningocele; the rate of incisional

Study 2
Th e second study [15] compared two types of allograft : DuraGen, a type I collagen matrix from bovine Achilles tendon, and AlloDerm, an acellular human dermis allograft . All patients treated with Alloderm (n=45) were treated before any patients were treated with a DuraGen onlay (n=56). No patient received both allograft s. Th e study was retrospective and performed primarily in the pediatric population. Th e DuraGen was not sutured to the dura, but rather was placed on top of the dural defect. Despite this approach, the rates of clinically apparent pseudomeningoceles and incisional CSF leaks were lowest in the DuraGen group (not signifi cant, Tables 3 and 4). Asymptomatic pseudomeningoceles were not routinely diagnosed because MRI was not performed routinely. Th e mean operative time was 37 min less in the DuraGen group (P <0.01). Rates of aseptic meningitis were not reported. Because harvesting pericranium can be diffi cult in children, the authors did not use it in this study although they stated that they considered it to be associated with the lowest risk of complications.

Case example
History A 41-year-old man presented with multiple complaints suggestive of a Chiari malformation. His most bothersome complaint was a 2-month history of increasing suboccipital headaches that worsened when he bent over. He reported radiating neck pain, intermitt ent numbness in his distal arms, increasing diffi culty grasping objects and coordinating his fi nger manipulations, and intermitt ent dysphagia and dysarthria. On examination, he had hyperrefl exia in his lower extremities and a unilateral Hoff man's sign. His sensation perception was decreased equally over his distal upper extremities in a nondermatomal patt ern. MRI of his brain [ Figure 1] showed a tonsillar herniation to the midportion of the C1 ring. Th e imaging of his brain and cervical spine revealed no other signifi cant pathology.

Surgical technique
Th e patient underwent a suboccipital craniectomy and laminectomy of C1. Th e midline incision spanned from approximately 2 cm cephalad of the superior aspect of the inion caudally to the spinous process of C2. Care was taken not to disrupt the underlying fascia of the occipital musculature or pericranium. Th e overlying tissues were undermined to allow a transverse incision to be made approximately 1 cm below the superior nuchal line. Th is strategy left a cuff of fascia and muscle for later use to recreate the muscular tension band and to allow a watertight closure of the fascia. A fascial incision starting at the midpoint of this transverse line was carried caudal to the superior aspect of C2, forming a T [ Figure 2]. In this way, the inion is never visualized, only felt. Att ention was then shift ed to the pericranium above the inion.
Using a handheld retractor/rake, the occipital scalp was undermined. A scalpel and electrocautery were used on "cut" to create the triangular graft (approximately 2.5 cm in height with a 3-cm base). A cautery tip bent at a 45° angle was utilized to maximize the graft size under the retracted scalp. Care was taken to keep the base of the triangular graft just above the superior nuchal line (approximate 0.5 cm above). Th e graft was freed from the cranium using a periosteal elevator tool with care being taken not to "butt onhole" it [ Figure 3]. Th e graft was then kept in antibiotic-instilled saline.
Aft er the suboccipital bone and posterior C1 arch were cleared, a wide bony suboccipital decompression with a C1 laminectomy was undertaken [ Figure 4]. Th is decompression extended to the posterior aspect of the occipital condyles. Th e dura was then incised in a Y fashion. Th e arachnoid was kept intact. Th e tonsils were visible though the arachnoid and no adhesions were visualized [ Figure 5]. Th e graft was sewed in place with the pericranial side facing the arachnoid using a running 4-0 Nurolon. Th e graft was tacked at each apex to assure one side would not be shorted [ Figure 6].
Watertight closure was tested with a Valsalva maneuver. Th e graft and exposed dural were coated with a single application of DuraSeal. Th e muscles were reapproximated, and the fascia was closed in running, watertight fashion.

Postoperative course
Th e patient's postoperative course was uneventful. Incisional pain resolved by the time of his follow-up visit. His preoperative head and neck pain resolved and his intermitt ent dysphagia and arm numbness resolved. He had no pseudomeningoceles, incisional problems, or infectious sequelae. His postoperative CT obtained soon aft er surgery showed that bony decompression was adequate [ Figure 7]. We do not perform postoperative MRI unless clinically indicated.

DISCUSSION
Although defi nitive data regarding the superior dural graft choice are lacking, this review shows some clinically relevant trends [Tables 3 and 4]. When available for harvest, autologous pericranium is associated with bett er rates of aseptic meningitis, wound infections, and pseudomeningocele formation compared to allograft s. In contrast, allogenic and synthetic dural substitutes are advantageous when rates of reoperation and incisional CSF leakage are compared. Att enello et al. [10] demonstrated superior performance of ePTFE versus pericranium with respect to symptom resolution, syrinx resolution, and restoration of physiologic CSF fl ow at the foramen magnum. Th ey contend that these fi ndings can be linked to the microscopically smooth barrier of the ePTFE that minimizes neural att achments while maintaining an outer surface textured for biologic fi xation. [10] However, the inherent diffi culty of harvesting thin pediatric pericranium may have infl uenced their results. No aggregate clinically relevant diff erences reached statistical signifi cance, and the limitations of this analysis are noteworthy. Nonetheless, advantages and drawbacks of both graft groups are readily apparent.
Autologous graft material has many theoretical advantages. In practice, the defects created by the securing suture are less likely to stretch or tear with use of pericranium compared to nonautologous material. Furthermore, unlike cadaveric dura or collagen, pericranium is living tissue, which enhances its function as a barrier. It more readily forms a watertight interface with dura when used with a running suture compared to synthetic materials, which are neither transformed nor vitalized. [1] Th is property should decrease rates of both pseudomeningocele formation and CSF leaks. Th e latt er, however, also heavily depends on adequate fascial closure. Th ese advantages of autograft are also supported by studies comparing harvested pericranium and synthetic substitutes for dural reconstruction aft er tumor resection. [25] Despite these benefi ts, autologous graft s can place patients at risk for additional morbidity. Harvesting a pericranial graft involves a variable degree of additional wound extension that increases risks of wound breakdown, infection, local pain syndromes, and cosmetic complications. Although not reviewed here, fascia lata graft s require an additional operative site. Harvesting nuchal ligament for graft s can weaken support in the posterior neck.
Some nonautologus graft s may perform bett er than others. Rates of aseptic meningitis and wound infection likely depend on the interaction between the specifi c graft type and the patient's immune system. A recent study by Bejjani and Zabramski [21] , however did not support this assertion. In a camparison of porcine intestinal submucosa (Durasis, Cook Biotech Inc.) to historical data on various dural substitutes, they noted similar rates of CSF leak (~1.7%), wound infection (~3.4%), bacterial meningitis (~0%), and pseudomeningocele formation (~0%) in all subgroups.
Both the search methodology and aggregate analysis of this critical review were limited. Inclusion and exclusion criteria aff ect search results via both selection bias and inadvertent omission of studies that may have been relevant to the analysis. Furthermore, not all studies included data for all outcomes evaluated, and outcome defi nitions/indications (e.g., reoperation, wound infection, meningitis) were subjective, nonspecifi c, and variable across the study centers and providers. Diff erential timing of postoperative MRI creates biases in the detection of asymptomatic pseudomeningoceles. With respect to generalizability, the study results are limited to Chiari malformation patients who present to academic medical centers. Extrapolation of the eff ect of graft s on outcome measures for supratentorial surgery or Chiari decompression in a community population is limited.

Additional study
Although it did not specifi cally investigate Chairi decompression, a study that compared to use of pericranial graft s and a synthetic subtitute in 124 patients who underwent both supra-and infratentorial brain tumor resection [25] provides additional data on rates of CSF leak and infection. In this study, internal CSF leaks (i.e. pseudomeningoceles) were signifi cantly more frequent in the Neuro-Patch nonautologous group (n = 61) than in the pericranium group (n = 63) (13% vs.

Institutional preference
At our institution, all surgeons who perform Chiari malformation surgery employ bony decompression and the use of duraplasty. We favor a wider extent of bony decompression and oft en remove the bone at the medial aspect of the posterior occipital condyle to allow maximal decompression of the lateral foramen magnum. [18] Confi rmation of adequate lateral decompression is obtained intraoperatively by carefully palpating the lateral foramen magnum with a No. 4 Penfi eld dissector to ensure that the remaining bone is fl ush with the occipital condyle.
We fi nd that duraplasty is essential for successful Chiari surgery because it creates an artifi cial cisterna magna where one was not previously present. [1] Opening of the arachnoid varies by surgeon. Tonsillar resection/shrinkage is seldom performed, and the choice of graft is distributed evenly between autograft and allograft (synthetic or xenogenic). For the senior author (VKHS) pericranium duraplasty is used in conjunction with subarachnoid dissection of arachnoid veils. We concur with Stevens et al. [14] that pericranium is "generally nonimmunogenic, nontoxic, rapidly integrated into native tissues, fl exible, strong, easily suturable, and inexpensive. " Great care is taken to avoid allowing hemorrhage from superfi cial tissues to enter the subarachnoid space.
In the case example presented here, a duraplasty was employed without arachnoid opening, because decompression was adequate inferior to the caudalmost aspect of the tonsillar herniation [ Figure 4]. Th is technique provides the advantage of tonsillar decompression without allowing communication between the subarachnoid space and outside compartments.
As an additional option employed by a minority of surgeons at our institution, a dissected arachnoid can be resutured to prevent CSF-related complications. We do not routinely use ultrasonography to evaluate the adequacy of decompression of the tonsils and normalization of CSF fl ow dynamics.
Authors have advocated the use of electrocautery in procuring pericranial graft s. [14] We believe that a scalpel is an acceptable alternative. We agree with Stevens et al. that pericranium can be harvested without extending the incision signifi cantly above the inion and superior nuchal line, thus avoiding injury to the occipital arteries and nerves. [14] Th e use of a hand-held retractor to develop a subgaleal plane with Metzenbaum scissors beneath the rostral limits of the incision facilitates this dissection.
We harvest a graft that is approximately 2.5 cm × 2.5 cm in dimension, perform the harvest before the bony decompression or dural opening, and soak the graft in antibiotic-saline irrigation. We also fi nd that a T incision of the fascia based below the superior nuchal line allows for more robust watertight closure of the fascia to contain possible pseudomeningocele formation. Th is fascia incision also creates an anchor point to which the occipital musculature can be fi xed, preventing loss of the occipital-to-C2 tension band.

RECOMMENDATIONS AND CONCLUSIONS
No class I or II level evidence addresses the extent of bony removal, the need for a dural opening, or the extent of intradural manipulation required in Chiari malformation surgery. Some of the available Class III data, however, can begin to guide decision making. Despite considerable variability in surgeons' practice patt erns, an increasing number of investigations and community neurosurgeons advocate the use of duraplasty in Chiari decompression. Th e reduction in reoperation rates must be weighed against the risk of CSF-related complications. In our opinion, the creation of an artifi cial cisterna magna via duraplasty is imperative. Opening the arachnoid with or without reapproximation is not well studied and is left to the discretion of the operating surgeon.
When duraplasty is employed, the literature does not strongly support the superiority of either autologous or nonautologous graft s. Th is lack of consensus was recapitulated in the three series analyzed here. Nonetheless, we believe that when the pericranium is available and of good quality, it should be utilized. It is non-immunogenic, inexpensive, and capable of creating a watertight closure with the dura. Th e quality of a pericranial graft is a relevant concern. While Vanaclocha et al. [19] demonstrated signifi cant advantages to the use of pericranial graft in adults, the superiority of ePTFE was demonstrated in the larger of the two pediatric comparison studies. [4] Th ese discrepancies are likely att ributable to increases in the quality and thickness of the pericranium with maturity. Future randomized studies with large numbers and the power to resolve diff erences in the complication rates of meningitis, wound infections, and reoperations (which are all oft en less than 5%) are warranted to establish the superiority of pericranium over other available dural substitutes.