Journal of Plastic, Reconstructive & Aesthetic Surgery (2010) 63, 213e217

Muscle-flap salvage of prosthetic dural repair

Connie Chung, Antonio J.V. Forte, Reza Momeni,Quratulain Fatima Masood, Deepak Narayan*

Departments of NeuroSurgery, Section of Plastic and Reconstructive Surgery,Yale University School of Medicine, New Haven, CT, USA

Received 21 November 2007; accepted 18 September 2008

KEYWORDSMuscle flaps;Dural repair;Cerebrospinal fluidleaks;Bovine pericardium

* Corresponding author. Section of P7313; fax: þ203 785 5714.

E-mail addresses: quratulain.fatim

1748-6815/$-seefrontmatterª2009Britdoi:10.1016/j.bjps.2008.09.023

Summary Objective: A critical element in the prevention of wound and cerebrospinal fluid(CSF) infections after craniotomies is the prevention of postprocedural CSF leaks. The salvageof infected prosthetic dural material in this milieu is not adequately addressed in the literatureand is the subject of this study.Methods: We performed a 7-year retrospective review of the Yale-New Haven Hospital patientrecords to identify successful salvage strategies in patients with relentless CSF leaks. Twenty datapoints were collected, including original diagnosis, nature of the procedure, presence of duralgraft, definitive treatment of the leak, culture results and pre- and postoperative antibiotics.Results: Thirteen patients experienced post-craniotomy CSF leaks that required surgical inter-vention. The most common cause of the original craniotomy (54% of patients) was an oncologicalaetiology, followed by ruptured aneurysms or haemorrhage in 31% of the patients. Of the 13patients experiencing CSF leaks, 76% involved the posterior skull base, and therefore a trapeziusmuscle flap was used in 38% of the cases. The Bovine pericardial graft (10 our of 13), a nonautolo-gous graft, was left intact, and CSF drainage procedures were employed in most patients GrowthofGram-positiveorganismsoncultureswas found in76%of thecases.Themost frequent offenderswere Staphylococcus aureus (five of the 13), coagulase-negative staphyloccocal species (two outof 13), and methicillin-resistant S. aureus (two out of 13). Vancomycin was administered in allcases preoperatively. All 13 patients who underwent open surgery for CSF leak had complete reso-lution of the leak without need for additional reconstructive surgical intervention.Conclusion: Comprehensive method of treating CSF leaks in conjunction with the salvage ofbovine pericardial dural grafts may be a viable clinical option.ª 2009 BritishAssociation of Plastic, ReconstructiveandAesthetic Surgeons. Published by ElsevierLtd. All rights reserved.

lastic Surgery, 330 Boardman Building, 330 Cedar Street, New Haven, CT 06520, USA. Tel.: þ203 785

a@gmail.com (Q. Fatima Masood), deepak.narayan@yale.edu (D. Narayan).

ishAssociationofPlastic,ReconstructiveandAestheticSurgeons.PublishedbyElsevierLtd.All rightsreserved.

214 C. Chung et al.

Wound infections are a devastating complication of crani-otomies, especially when associated with cerebrospinalfluid (CSF) infection. Fortunately, the incidence isextremely low, with a stated rate of 3e4%.1,2 A criticalelement in the prevention of wound and CSF infectionsafter craniotomies is the prevention of postprocedural CSFleaks. The use of artificial dural (non-autologous) materialsto provide a watertight seal, thus preventing CSF leaks, iswidely practised in neurosurgery. A wide spectrum of non-autologous materials have been used in the repair of duraldefects ranging from gold foil in the past to the more recentuse of cadaveric dura, bovine pericardial grafts (BPG) andGoretex (Polytetra fluoroethylene PTFE) among others. Theuse of collagen matrix has been successfully reported asa natural and less time-consuming option for repair ofspinal durotomy.3 The BPG as a prosthetic dural patch hasbeen increasingly adopted over cadaveric dura over fears oftransmitting the neurodegenerative CreutzfeldteJacobdisease.4 These non-autologous dural substitutes, however,constitute a concern in the presence of wound and CSFinfections.

Controversy dogs the management of these grafts in thepresence of an infection, and the literature does not offerany firm guidelines for the management of this specificproblem. Anson and Marchand,1 in reporting their experi-ence with BPGs for dural closure, mention one case of a CSFleak associated with wound infection and meningitis. Theystated that the dural graft was not a significant outcomefactor. Besides, they described the repair of a fenestrationin the graft, but no further details of management wereoffered. Anecdotal evidence, based on individual surgeons’experiences, suggests that removal of the artificial materialmay be a good option. This modality of treatment,however, does not take into account the technical difficultythat may be encountered in removing the dural patch andits replacement in infected and inflamed tissue. In view ofthis prevailing uncertainty, our study was undertaken tocollate and analyse data from our experience with thisproblem and to develop a surgical algorithm for itsmanagement.

Materials and methods

We performed a 7-year retrospective review of the recordsof Yale-New Haven Hospital, a tertiary neurosurgicalreferral centre, to identify successful salvage strategies inpatients with CSF leaks refractory to non-invasive measuresand leading to open surgery. Patients with leaks controlledby conservative measures such as head elevation, decom-pression or endoscopic measures were excluded. Twentydata points were collected, including age, sex, originaldiagnosis at initial craniotomy, emergent or elective natureof the procedure, presence or absence of dural graft,removal or retention of graft, latency of infection andonset of post-craniotomy CSF leak, definitive treatment ofthe leak, presence of lumbar or ventriculoperitoneal (VP)shunts, culture results, pre- and postoperative antibiotics,length of antibiotics treatment after definitive procedureand risk factors for wound complications, such as malig-nancy, diabetes, obesity, tobacco use, chemotherapysteroids and history of radiation therapy.

Results

Thirteen patients experienced a post-craniotomy CSF leakthat required surgical intervention for control (Table 1).The patients (age range: 1e60 years; mean: 33 years)included seven males and six females. Four patients (31%)were under 10 years of age. The original craniotomy, in themajority of patients (seven out of 13, 54%), was performedfor oncological aetiology (meningiomas, ependymomas,glioblastomas and medulloblastoma). A vascular cause(ruptured aneurysms or haemorrhage) was the second mostcommon cause of original craniotomy (four out of 13, 31%).

Nine of the craniotomies were non-emergent, and theremaining four craniotomies were emergent. The time fromthe craniotomy to definitive repair ranged from 9 to 393days (mean: 49 days). Most patients (11 of 13, 85%) pre-sented with a leak within 6 weeks of the original crani-otomy. Definitive repair consisted of surgical exploration,thorough debridement and culture of necrotic material andmuscle flap (local, pedicled or free). Most patients (10 outof 13, 76%) had been recipients of BPGs placed for duralclosure at the time of the original craniotomy. Ten of the 13leaks (76%) involved the posterior skull base, and a ‘trape-zius’ muscle flap was used in five of these cases. Theremaining cases were closed by mobilisation and approxi-mation of adjacent cervical musculature over the graft(three patients), ‘Pectoralis major’ muscle flap (onepatient), and ‘rectus’ free flap (one patient). The rationalefor the use of muscle flaps in these cases was the impos-sibility of approximating woody, indurated, muscle tissueover an infected prosthetic material and a CSF leak. In allpatients with a BPG (10 out of 13), this non-autologous graftwas left in place and was not removed. The CSF drainageprocedures (lumbar drain, VP shunt or ventriculostomy), inisolation or in combination, were employed used in mostpatients (10 out of 13 patients) around the time of defini-tive repair.

At our institution, it is customary for our neurosurgicalcolleagues to administer preoperative prophylactic antibi-otics, and vancomycin is the usual agent. This was admin-istered in all cases preoperatively. The patients weretreated with broad-spectrum intravenous antibiotics untilthe identification of an organism, after which the antibioticregimen was adjusted accordingly. Positive bacterialgrowth on cultures was the rule, with Gram-positiveorganisms being identified in 76% of the cases. The mostfrequent offenders were Staphylococcus aureus (five out of13), coagulase-negative staphyloccocal species (two out of13) and methicillin-resistant S. aureus (two out of 13).Gram-negative organisms were isolated in the remainingthree patients. The duration of antibiotic treatment rangedfrom 10 days to 8 weeks (mean treatment duration was 25days). The patient who was treated for 10 days did not haveany growth demonstrable in the final cultures. The patienton the other end of the spectrum had a very complicatedcourse against a background of chronic high-dose steroiduse, development of infective pancreatitis and require-ment for a second drainage procedure using a ventriculo-pleural shunt.

Of the risk factors assessed among the 13 patients, six(46%) were obese, seven (54%) had received prior radiation

Table 1 Summary of patients’ characteristics

Case Gender, age in years,follow-up (f/u)

Diagnosis Duralgraft

Latency fromoriginalcranitomy(days)

Treatment Graftremoval

1 Male, 33 (f/u 4 years) Aneurysm Yes 166 Pericranial flap No2 Female, 9 (f/u 4 years) Ependymoma No 9 Direct repair

and wash outNotapplicable

3 Female, 37 (f/u 2 years) Chiarimalformation

Yes 195 Trapezius flap No

4 Male, 57 (f/u 2 years) Cerebellarhaemorrhage

Yes 28 Trapezius flap No

5 Female, 10 (f/u 4 years) Ependymoma Yes 30 Direct repairand wash out

No

6 Female, 1 (f/u 4 years) Medulloblastoma Yes 36 Layered closurewith paraspinousmuscles

No

7 Male, 40 (f/u 3 months,withdrawal of support)

Arteriovenousmalformation

Yes 43 Trapezius flap No

8 Female, 49 (f/u 2 months,died PE)

Subarachnoidhaemorrhage fromPICA aneurysm

Yes 21 Trapezius flap No

9 Male, 60 (f/u 3 years) Acoustic neuroma No 21 Pectoralis major flap Notapplicable

10 Male, 57 (f/u 1 year,support with drawn)

Meningioma Yes 393 Rectus free flap No

11 Female, 31(f/u 2 years,died of advanced tumour)

Meningioma Yes 24 Irrigation anddebridement

No

12 Male, 2 (f/u 1 year) Hemorrhage 2�

to fallNo 18 Patch and direct

closureNotapplicable

13 Male, 48 (f/u 3 monthswithdrawal of support)

Glioblastomamultiform

Yes 94 Trapezius flap No

Muscle-flap salvage of prosthetic dural repair 215

therapy, seven (54%) had received chemotherapy, four(30%) were smokers and four (30%) were on steroids.Obviously, several patients had multiple risk factors. All 13patients who underwent open surgery for CSF leak hadcomplete resolution of the leak without need for additionalreconstructive surgical intervention. One patient under-went a delayed cranioplasty for cosmetic reasons monthslater. Follow-up ranged from 2 months to 4 years witha mean of 26 months.

Discussion

Cerebrospinal fluid (CSF) leaks are a relatively commoncomplication following cranial or spinal surgery. Relentlessleaks put patients at risk for grave intracranialcomplications.

Our review of literature found no single study deter-mining a definite incidence of post-procedural CSF leaks.Any epidemiological study done to determine the rate ofCSF leaks would have to consider cumbersome variables ofsurgical approach, dural closure techniques, surgicalexpertise and the nature of the lesion itself. However,several studies have been done concentrating on theoutcomes from different neurosurgeries. A study conductedin Spain focused on post-procedural CSF complicationsfollowing particular surgical removal of posterior fossa

tumours. Hydrocephalus, CSF leak, pseudomeningocoeleand CSF infection were the primary complications investi-gated in detail. CSF-related complications were observed in31% of the cases. Tumour size was the single factor asso-ciated with the development of CSF-related complicationsafter surgery for posterior fossa tumours.5 According toanother study conducted at the University of Oregon,incidence of CSF leak following trans-sphenoidal surgeryranges from 0.5% to 15.0 %.6 It has been argued that surgicalapproaches can influence post-operative outcomes inneurosurgery. Nevertheless, surgical exposure is tailoredaccording to individual patients and the characteristics oftheir lesions. A study of cerebellopontine-tumour removalinvestigated surgical approaches and dural-closuretechniques with postoperative CSF-leak morbidity.7 Theincidence of CSF leaks has declined due to improved intra-operative techniques and meticulous closure of the dura.Nevertheless, the exact incidence of CSF leaks, particularlyin posterior fossa craniotomies, is elusive.

Conventional options for subtle dural repairs include bedrest, hydration and steroid therapy. Kitchel and colleagues(1989) have recommended the use of closed lumbardrainage for the treatment of persistent CSF leaks. Thetechnique included using a percutaneous intradural cath-eter for drainage of 200e300 cc of CSF per day for almost 4days till the condition resolved. However, this approach

Figure 1 The use of the trapezius flap as an ideal choicebecause of its adequate bulk and surface area to cover largedefects.

Figure 2 Prone intra-operative positioning offers best accessand exposure for the use of this flap for posterior fossaproblems.

216 C. Chung et al.

carries a significant risk of meningitis and requires vigilantphysician monitoring. Autologous blood patches have alsobeen employed in patients with persistent CSF leaks.Despite the simplicity of this procedure, it is not suitablefor extensive post-laminectomy CSF leaks and limited todural tears measuring 1e2 mm. Hence, cohesive evidencehighlights that dural repair through surgical exploration isthe gold standard for dealing with a persistent CSF leak.Primary dural repair offers the best safeguard against leaks.Unfortunately, none of the cases selected in this study weresuitable candidates for primary dural repair necessitatingdural grafting. However, there are situations such as thoseoutlined in our study that necessitate a more aggressiveapproach.

Bovine pericardial tissue has become increasinglypopular among neurosurgeons for prosthetic dural coveragein the last decade. This was, in part, prompted by theconcerns of transmission of CreutzfeldteJacob diseaseassociated with the use of cadaveric human dura.4 Prior tothe use of BPG, cadaveric dura was the main alternative toautologous tissue in dural closure for many years. BPG is inmany ways an ideal prosthetic material, since it is inex-pensive, pliable, non-toxic, fluid-impermeable, unlikely toadhere to the meninges, easily sterilised and conducive tosuture placement. In view of our current study, we believean additional advantage in favour of BPG is its ability to berendered sterile, in the setting of active infection, bya combination of operative intervention and appropriateantibiotic treatment. This is a significant benefit, thusobviating the necessity for graft removal and all its atten-dant technical and clinical sequelae. Conventional surgicalwisdom dictates the removal of infected prosthetic mate-rial in an infected field in order to obtain a closed, stablewound. This may be a counsel of perfection in most situa-tions. However, when facing a posterior fossa CSF leak, theperils of removal and replacement outweigh attempts atsalvage. This argument is also bolstered by other studiesshowing that prosthetic material in an infected field may besalvaged without having to immediately resort to removal.Two such examples are spinal orthopaedic hardware andabdominal-wall hernia meshes in the presence of infection.The technical difficulties experienced in dural closure inthe posterior cranial fossa may explain the preponderanceof leaks in this location. The use of the trapezius flap in thissituation is advantageous, since it is an easily harvested,robust choice with adequate bulk and surface area to coverlarge defects (Fig. 1). Prone intra-operative positioningoffers best access and exposure for the use of this flap forposterior fossa problems, since adequate neck flexion inlateral decubitus position, especially in the obese, may benearly impossible (Figure 2). A useful technical manoeuvreis to harvest the fascia over the trapezius as well as a partof the lumbodorsal fascia to aid in suturing the edges inorder to achieve a watertight closure. It has been ourpractice to secure the muscle and its fascia to the boneedges by means of judiciously placed drill holes on theedges of the bony defect. This provides additional strengthand security to the closure. The use of closed suction drainsis a necessity for the harvest of the trapezius muscle.Despite best efforts to attain a watertight seal, we havefound, on occasion, that CSF leaks communicate with themuscle bed. For this reason, we avoid placing drains in

proximity to the repair, instead, placing it at a distance ina dependent position. For the same reason, we avoidplacing these tubes onto the high wall suction, since itseems to encourage the egress of spinal fluid. Reducing CSFpressure by raising the head of the bed is complemented byformal CSF-drainage procedures in our reconstructions.These modalities (lumbar drains, ventriculostomies or VPshunts) were used in close co-operation with our neuro-surgical colleagues. In general, the temporary CSF drainswere not left in for longer than 1 week. A recent animalstudy has also suggested the use of fibrin glue as a safematerial to improve appropriate closure of the dura.8

Another study proposes the use of a radial forearm free-tissue transfer for persistent CSF leaks in the patients witha failure of previous repair or in patients with excessivedead space requiring bulky repair.9

Intensive care unit monitoring, preferably in a neuro-surgical setting, is another imperative in the postoperativemanagement of these patients. The help of experienced

Muscle-flap salvage of prosthetic dural repair 217

nursing staff to control perioperative agitation, hyperten-sion and CSF-pressure alterations need not be over-emphasised. Antibiotics play a key role in the success of ourapproach. All our patients with infected CSF cultures (12out of 13) were treated for a minimum of 14 days ofculture-guided antibiotic therapy. It is impossible to statethe precise length of time required to achieve completebacteriological control leading to graft salvage. Since ourmean treatment time was 26 days, it would seem reason-able to aim for a 3e4 week course of antibiotic therapy.The length of therapy must, however, be tailored to theparticulars of the case. It has also been reported thatcorynebacteria, mostly considered mere contaminants,may also be responsible for osteomyelitis in post-crani-otomy bone flaps.10 Propionibacterium acne has also beena Gram-positive bacteria identified in patients with duralallograft post-neurosurgery.11,12 Although the small numberof patients involved in this series makes it difficult togeneralise the risk-factor findings and stratify patients atrisk for poor wound healing, it is important to note that allpatients with non-autogenous dural grafts had full resolu-tion of their CSF leak, in the presence of infection,following open surgery. It is also instructive to note that100% dural graft salvage was achieved, despite, in manycases, a multiplicity of risk factors such as prior craniot-omies, radiation therapy, chemotherapy, steroids, obesityand tobacco abuse. For patients on long-term steroids, weinstituted pharmacological doses of vitamin A (25,000 IUdaily for 3 weeks) since there is support in the literature forvitamin A’s ability to counter the baleful effects of steroidson wound healing. Close monitoring of liver functioning andpossible drug interactions are a part of this regimen.

An additional shortcoming of the study is the length offollow-up, in that we have a mean follow-up of 2 years. It isconceivable that a longer follow-up period might reveala higher failure/re-operation rate. In summary, based onour experience, we believe that the points enumeratedbelow are important in achieving successful results.

1. Thorough debridement of all necrotic muscle and tissue2. Repeat operative washouts till wounds are clean3. Closure of leaks identified intra-operatively by a Val-

salva manoeuvre4. Use of sturdy, absorbable, monofilament sutures (e.g.,

Maxon) for securing muscle flaps to bone and for duralclosure

5. Closed drainage of muscle harvest areas with lowsuction pressures (e.g., bulb suction)

6. Neuro-ICU monitoring7. CSF decompression for at least 1 week (or permanently

if neurosurgically indicated)

8. Vitamin A (25,000 IU) daily for 3 weeks in patients onsteroids

9. Appropriate antibiotics administration for at least 3weeks.

The success of this treatment strategy suggests that thisalternate method of treating CSF leaks in conjunction withthe salvage of bovine pericardial dural grafts may bea viable clinical option.

References

1. Anson J, Marchand EP. Bovine pericardium for dural grafts:clinical results in 35 patients. Neurosurgery 1996;39:764e8.

2. Jain VK, Mehrotra N, Sahu RN, et al. Surgery of vestibularschwannomas: an institutional experience. Neurol India 2005;53:41e5.

3. Narotam PK, Jose S, Nathoo N, et al. Collagen matrix (Dura-Gen) in dural repair: analysis of a new modified technique.Spine 2004;29:2861e7.

4. Martinez-Lage JF, Rabano A, Bermejo J, et al. Creutzfeldt-Jakob disease acquired via a dural graft: failure of therapywith quinacrine and chlorpromazine. Surg Neurol 2005;64:542e5.

5. Santamarta D, Blazquez JA, Maillo A, et al. Analysis ofcerebrospinal fluid related complications (hydrocephalus,fistula, pseudomeningocele and infection) following surgeryfor posterior fossa tumors. Neurocirugia (Astur) 2003;14:117e26.

6. Shiley SG, Limonadi F, Delashaw JB, et al. Incidence,etiology, and management of cerebrospinal fluid leaksfollowing trans-sphenoidal surgery. Laryngoscope 2003;113:1283e8.

7. Cueva RA, Mastrodimos B. Approach design and closure tech-niques to minimize cerebrospinal fluid leak after cer-ebellopontine angle tumor surgery. Otol Neurotol 2005;26:1176e81.

8. Steinbok P, Singhal A, Mills J, et al. Cerebrospinal fluid (CSF)leak and pseudomeningocele formation after posterior fossatumor resection in children: a retrospective analysis. ChildsNerv Syst 2007;23:171e4.

9. Ozisik PA, Inci S, Soylemezoglu F, et al. Comparative duralclosure techniques: a safety study in rats. Surg Neurol 2006;65:42e7.

10. Weber SM, Kim J, Delashaw JB, et al. Radial forearm free tissuetransfer in the management of persistent cerebrospinal fluidleaks. Laryngoscope 2005;115:968e72.

11. Wilson IF, Candia GJ, Worthington MG, et al. Chronic osteo-myelitis due to Corynebacteria in a postcraniotomy bone flap.Clin Infect Dis 1999;28:1323e4.

12. Jallo GI, Koslow M, Hanna BA, et al. Propionibacterium asa cause of postneurosurgical infection in patients with duralallografts: report of three cases. Neurosurgery 1999;44:1138e41.