image-guided endoscopic repair of cerebrospinal fluid rhinorrhea by the bath plug grafting technique

The LaryngoscopeVC 2011 The American Laryngological,Rhinological and Otological Society, Inc.

Image-Guided Endoscopic Repair of Cerebrospinal Fluid Rhinorrheaby the Bath Plug Grafting Technique

Hazem Saleh, MD; Sameer Al Bahkaly, MD

Objectives/Hypothesis: Several techniques of endoscopic repair of cerebrospinal fluid (CSF) rhinorrhea were pre-scribed. Defect localization was mandatory for all.

Study Design: Prospective study.Methods: A prospective study was conducted on nonaccidental CSF rhinorrhea. The sensitivity and specificity of image-

guided navigation in localizing the defect were studied. The feasibility and reliability of the bath plug technique in repairingthe defect were assessed. Six cases were included.

Results: CFS rhinorrhea was spontaneous in four and iatrogenic in two. Two defects were in the fovea ethmoidalis, twoin the cribriform plate, and two in the sphenoid. The repairing graft was successfully positioned in five cases. In one case asphenoid obliteration was done. Image-guided navigation proved useful in localizing the defect with a sensitivity and specific-ity of 100%. All patients were available for follow-up (mean ¼ 19 months). The overall success rate was 83%, and the meannumber of surgeries per patient was one.

Conclusions: The bath plug was feasible for repairing the anterior skull base defects but not for all the sphenoiddefects. It proved reliable as a single management in cases with normal intracranial pressure. The image-guided navigationoverrides, in sensitivity and specificity, other methods of localization of the anterior skull base defects.

Key Words: Cerebrospinal fluid, rhinorrhea, repair, navigation system, graft.Level of Evidence: 4.

Laryngoscope, 121:909–913, 2011

INTRODUCTIONCerebrospinal fluid (CSF) rhinorrhea denotes a

skull base fistula connecting the subarachnoid space tothe nasal cavity. Defects commonly occur in the ethmoidroof, the cribriform plate, and the sphenoid, and less com-monly in the frontal sinus posterior table.1 Because of thesize and structure of the defects, the degree and nature ofdural disruptions, the intracranial pressure (ICP) gradi-ent, and the presence of meningoencephalocoeles varyaccording to the aetiology. Schlosser and Bolger2 etiologi-cally classified CSF rhinorrhea into five categories:accidental trauma, surgical trauma, spontaneous, congeni-tal, and neoplastic leaks. Lopatin et al.3 classified theminto primary (spontaneous) and secondary CSF rhinor-rhea, whereas Gendeh et al.1 suggested only threeetiologic categories: congenital, acquired, and spontaneous.

Of cases of CSF rhinorrhea, 70% to 80% are due tohead trauma fracturing the skull base. Accidental andiatrogenic skull base defects are surrounded by healthybone and have normal ICP with favorable repair out-

comes. Normal ICP is also a feature of neoplastic andcongenital CSF leaks.1 Congenital skull base defects aregenerally rare, and 63% of them occur in the foramencaecum.4 The prevalence of spontaneous CSF leaks indifferent series vary from 3% to 36%.3,5 High ICP isassociated with 63% to 88% of spontaneous leaks.4 Thecribriform plate is a common site, although Gendeh et al.found 40% to 56% of spontaneous leaks in the sphenoid.1

Spontaneous CSF leaks are associated with a 50% to100% incidence of encephalocele and a 25% to 87% inci-dence of recurrence.2 This is not only due to elevated ICP,but also to associated meningoencephaloceles and attenu-ated bony skull base. A high incidence of recurrence alsocharacterizes CSF rhinorrhea due to neoplasms or to verylarge defects. Seventy percent of defects due to accidentaltrauma close spontaneously. However, 9% to 50% ofuntreated CSF rhinorrhea are complicated by meningitis,a risk that increases with duration.1,3

A work-up should establish the diagnosis and local-ize the leak because the physician cannot rely on historyalone. Supine endoscopic examination shows pooling ofsecretions in the posterior choana below the sphenoidostium. b2 transferrin’s sensitivity and specificity fordiagnosing CSF rhinorrhea are 99% and 97% respec-tively. The presence of cirrhosis or congenital proteinanomalies increases the false positives. b trace proteinsensitivity and specificity reach 100%.4 Computed to-mography (CT) cisternogram with metrizamide (or withless toxic contrast materials) is useful for both diagnosisand localization. Its sensitivity ranges from 48% to 96%

From the Ear, Nose, and Throat Services Department, King AbdulAziz Medical City, Riyadh, Saudi Arabia.

Editor’s Note: This Manuscript was accepted for publicationAugust 31, 2010.

The authors have no funding, financial relationships, or conflictsof interest to disclose.

Send correspondence to Hazem Saleh, MD, 10 El Messaha Square,Al Mesreyin Tower, Apt 1401, Dokki, Guiza 12311, Egypt.E-mail: [email protected]

DOI: 10.1002/lary.21355

Laryngoscope 121: May 2011 Saleh and Bahkaly: Repair of CSF Rhinorrhea

909

depending on the flow rate and site. However, it canmiss intermittent leaks. It is contraindicated with highICP and holds low acceptability. Its interpretation is op-erator dependent.6 CT cisternogram was positive in only33% of Gendeh et al.’s cases.1 Magnetic resonance imag-ing (MRI) cisternogram (T2 or fast spin echo with fatsaturation and image reversal) can also miss intermit-tent leaks. Its accuracy is 89%, sensitivity 85% to 92%,and specificity 100%. However, Hegarty and Millar7

reported up to 42% false positives with this method.Intrathecal gadolinium MRI cisternography also exists.7

To localize the defect, high resolution CT (HRCT) has asensitivity and specificity of 87% because the dural tearmay not correspond to the shown bony defect.1 HRCTwas positive in only 25% of Gendeh et al.’s cases. The ra-dioactive cisternogram holds a localizing sensitivity of62% to 76%. Its false positives may reach 33%. Often, itcan only identify the side of the leak.2 Preoperative fluo-rescein is helpful in iatrogenic leaks after sinus surgery.Its accuracy ranges from 81%1 to 96%.2 It helps to con-firm a watertight repair at the end of the procedure.Intraoperative intranasal topical fluorescein was alsodescribed.8,9

Techniques of repair evolved from craniotomy to en-doscopy. The success rate after craniotomy was nearly80%. Endoscopic repair success rate exceeds 90%.2 Otheradvantages of endoscopic approaches are preservation ofsmell, shorter hospital stay, absence of external scarringor brain retraction, and low risk of bleeding, seizures,and infections (<1%). The techniques of endoscopicrepair depend on surgeons preferences. Consensus existson a few guidelines: to strip the mucosa around thedefect, to provide composite rather than simple graftsfor defects larger than 1.5 cm, to coagulate and transectthe pedicle of a meningoceles,4 to prescribe antibiotics(ceftriaxone) for 7 to 10 days postoperatively, and finally,if high ICP is present, to manage it either before or afterrepair.1,2,10 The bath plug technique, described by Wor-mald in 1997, is a method for repairing anterior skullbase defects.11 A fat plug, three times the width of thedefect, is positioned in underlay in the epidural or sub-dural space. It theoretically overcomes the increasedICP by preventing the high pressure from displacing thegraft. Sanderson et al.4 stated that the bath plug tech-nique fits and conforms the shape of the defect, isresistant to infection, and is strongly adherent to bone.However, the bath plug technique is not recommendedin defects next to important structures. It leads to diffi-cult interpretation of soft tissue shadows after repair,and its back herniation may prevent osteoneogenesis.2

In one series, 90% of repairs by the bath plug techniquewere successful, compared to 87% for the whole endo-scopic repair group.1

In this study, we aimed to assess the feasibility andreliability of the bath plug repair as a single manage-ment for CSF rhinorrhea from nonaccidental anteriorskull base defects. The sensitivity and specificity of animage-guided navigation system in localizing the defectswere also studied. We designed a prospective non-randomized pilot study over 3 years. We included allcases of endoscopic repair of nonaccidental CSF rhinor-

rhea. The navigation system was used to localize thedefect, and the bath plug technique was attempted torepair it. Our setting was an accredited tertiary carecenter in Riyadh, Saudi Arabia.

MATERIALS AND METHODSBetween April 2007 and February 2010, six consecutive

patients with CSF rhinorrhea (five females, one male; meanage, 43 years) were included. None was secondary to accidentaltrauma. These latter are managed by the neurosurgeons of ourcenter. After a detailed history, endoscopic examination with a0� telescope (Karl Storz, Tuttlingen, Germany) was done for allpatients under topical anesthesia. It showed leakage of a clearfluid in one nasal cavity. High resolution CT scan according toan image-guided CT protocol was done for all patients. NeitherMRI nor intrathecal contrast-enhanced CT were an integralpart of our investigation protocol. b2 transferrin detection wasnot available at our institution. The patients’ details, etiology,and site of leak are shown in Table I.

All patients gave signed consent for endoscopic repair.Under general anesthesia, the nasal cavity was packed withXylocaine and adrenaline. Endoscopic inspection was doneusing 0� and 30� Hopkins telescopes, 4 mm in diameter (KarlStorz). The repair was conducted as described by Wormald12

(Fig. 1). The defect area was exposed using conventional dissec-tion and/or powered instruments (microdebrider, Karl Storz).After exposure, the defect was localized using the image-guidednavigation system (IGNS; Medtronic-Xomed, Jacksonville, FL).Fat was taken from the ear lobule in four cases and from theabdominal wall in two cases. The plug was harvested with a 2/0Vicryl knot and then pushed through the skull base defect usinga frontal sinus seeker, a suction tip detached from the circuit,or a freer dissector. It was positioned in underlay in the epidu-ral space. Nasal mucosa from the ipsilateral inferior turbinatethen Tutoplast allograft (Tutogen Medical, GmbH, Neunkirchen,Germany) were ascended along the Vicryl thread and positionedto overlay the defect. The mounting was reinforced by fibringlue (Baxter, Wien, Austria) and a gel foam pack was positionedin the nasal cavity (Cutanplast; Mascia Brunell S.P.A, Milan,Italy). In the first case, a meningocele was present (Fig. 2). Thiswas transacted with perfect hemostasis before repair. In the fifthcase, a spontaneous fistula was present due to a defect in thelateral recess of a roomy sphenoid sinus (Fig. 3 and Fig. 4). Sphe-noid obliteration with abdominal fat was performed after minimalwidening of the sphenoid orifice. The anterior sphenoid wall wasused as a scaffold for the fat plug. Irrigation of the nasal cavitywith antibiotic was not done, nor was packing of the nose with anonresorbable material or insertion of a lumbar drain.

Postoperatively, all patients were hospitalized for 1 weekon average and put under parenteral antibiotics (cefazolin).Patients were given stool softeners and antitussives, nursed ina semisitting position for 3 days, then allowed to ambulategradually. The patients were followed in the outpatient clinic

TABLE 1.Patient Characteristics.

Age/Sex Aetiology Site

44 years/$ Spontaneous Cribriform plate

29 years/$ Spontaneous Sphenoid

39 years/# Iatrogenic Fovea Ethmoidalis

40 years/$ Iatrogenic Fovea Ethmoidalis

53 years/$ Spontaneous Sphenoid

53 years/$ Spontaneous Cribriform plate


910

for 2 weeks after discharge, then monthly for 6 months, andthen every 4 months.

RESULTSThe image-guided navigation was able to detect the

exact site of all six defects during surgery with no falsepositives or negatives (100% sensitivity and specificity).We were able to localize all defects, and the areas local-ized by the image-guided navigation system were thoserepaired in all cases. The success of repair was assessedby the absence of postoperative watery rhinorrhea. Nopatient reported watery discharge in the immediate andearly postoperative period. The mean number of sur-geries per patient was one. Our mean follow-up periodwas 19 months (range, 4 to 38 months). CSF rhinorrhearecurred in one patient 6 months later. This patient hada history of a treated brain tumor for 20 years with alarge encephalomalacia (Fig. 5). Subsequent investiga-tion revealed the presence of high ICP.

DISCUSSIONA pitfall of some etiologic classifications of CSF

rhinorrhea is to consider leaks associated with tumors,congenital skull base malformations, and delayed trau-matic leaks as secondary spontaneous leaks.1 Wesuggest classifying CSF rhinorrhea as traumatic ornontraumatic. Both groups differ in the defect site, inci-dence of complications, prognosis, and association withmeningoencephalocele. Traumatic causes can then besubdivided into accidental or iatrogenic and nontrau-matic ones into spontaneous, congenital, or neoplastic.

The transnasal endoscopic repair of CSF leaks, firstdescribed by Wigand in 1981,13 is considered the treat-ment of choice for most anterior cranial and sphenoidCSF leaks. The surgical technique does not affect theoutcome.10 The main factor in choosing the surgical

Fig. 1. The bath plug technique forrepair of anterior skull base defects.(1) Mucosal stripping around thedefect. (2) Harvesting an oblong fatplug three times the width of thedefect. (3) A knot is tightened at thetop of the fat plug. The suture ispassed through the whole length ofthe plug. (4) The fat plug is gentlymanipulated through the defectwhile the suture is still attached toit. (5) Traction on the suture toexpand the plug. (6) Reinforcementof the reconstruction with overlaymucosal graft ascended along thesuture.

Fig. 2. Case 1. Coronal computed tomography showing a defectin the left cribriform plate with meningoencephalocele.

Fig. 3. Case 5. Coronal computed tomography showing defect inthe left sphenoid lateral recess.


911

repair technique is the familiarity of surgeons withit.4,10,14 This was the case in the present study.

In accordance with Zweig et al.,10 the results of thepresent study show that no technique should be adoptedas an empirical repair for all anterior skull base defects.The indications and suitability of the bath plug graftingtechnique differs already among authors. Some recom-mend it for large defects, others for small defects withhigh ICP and herniating brain or meninges.4 Although

the bath plug technique was theoretically designed towithstand high ICP, in the present study it was not reli-able as the sole management for anterior skull basedefects with high ICP.

Sphenoid obliteration is the most common mode ofrepair of sphenoidal CSF leaks.3,15 It was used in 58% ofTosun et al.’s cases.5 The predators of sphenoid mucosalstripping admit it is never complete and suggest it ispractically impossible and rather risky,2 therefore we didnot rely totally on it. We are rather subjecting ourpatients with sphenoid obliteration to a life-long closefollow-up to screen any mucocele development.

A lumbar drain can be useful in patients withincreased ICP, elevated body mass index, revisionrepairs, and large skull base defects.8 However, theinsertion of a lumbar drain is not essential for success.5

The absence of evidence of necessity kept us from insert-ing lumbar drains in our patients. Casiona and Jassirreported a 97% success rate of endoscopic repair of CSFleak without lumbar drainage (33 cases over 7 years).16

They did only lumbar tap and concluded that a normalopening pressure in a patient with an active CSF leakdoes not exclude an increase in ICP once the defect isclosed.

Because all of our defects were 1.5 cm or less, wedid not use any composite or bony grafts. We were ableto repair them with the bath plug alone (except in onecase with the sphenoid lateral recess defect). Our successrate (83%) is less than that reported for endoscopicrepairs.1,2 This is due to the small population number ofthis pilot study. Our single failure represented 17% of thepopulation. This failure was due to longstanding highICP with patient refusal of ventriculo-peritoneal shunt.

CONCLUSIONThe bath plug is feasible for anterior skull base but

not for all sphenoid defects. It proved reliable as a singlemanagement in cases with normal ICP. The image-guided navigation system overrides, in sensitivity andspecificity, other methods of localization of anterior skullbase defects. A future prospective comparative study,using a larger number of patients, will offer a higherlevel of evidence to promote the generalizability of ourpilot study results.

BIBLIOGRAPHY

1. Gendeh BS, Mazita A, Selladurai BM, Jegan T, Jeevanan J, Misiran K.Endonasal endoscopic repair of anterior skull-base fistulas: the KualaLumpur experience. J Laryngol Otol 2005;119:866–874.

2. Schlosser RJ, Bolger WE. Nasal cerebrospinal fluid leaks: critical reviewand surgical considerations. Laryngoscope 2004;114:255–265.

3. Lopatin AS, Kapitanov DN, Potapov AA. Endonasal endoscopic repair ofspontaneous cerebrospinal fluid leaks. Arch Otolaryngol Head NeckSurg 2003;129:859–863.

4. Sanderson JD, Kountakis SE, McMains KC. Endoscopic management ofcerebrospinal fluid leaks. Facial Plast Surg 2009;25:29–37.

5. Tosun F, Carrau RL, Snyderman CH, Kassam A, Celin S, Schaitkin B.Endonasal endoscopic repair of cerebrospinal fluid leaks of the sphenoidsinus. Arch Otolaryngol Head Neck Surg 2003;129:576–580.

6. Mostafa BE, Khafagi A. Combined HRCT and MRI in the detection of CSFrhinorrhea. Skull Base 2004;14:157–162; discussion 162.

7. Hegarty SE, Millar JS. MRI in the localization of CSF fistulae: is it of anyvalue? Clin Radiol 1997;52:768–770.

8. Ulualp SO. Complications of endoscopic sinus surgery: appropriate man-agement of complications. Curr Opin Otolaryngol Head Neck Surg 2008;16:252–259.

Fig. 4. Case 5. Coronal T2 magnetic resonance imaging showingcerebrospinal fulid leakage in the left sphenoid.

Fig. 5. Case 6. Axial T2 magnetic resonance imaging showingright parietal encephalomalacia leading to increased intracranialpressure.


912

9. Saafan ME, Ragab SM, Albirmawy OA. Topical intranasal fluorescein: themissing partner in algorithms of cerebrospinal fluid fistula detection.Laryngoscope 2006;116:1158–1161.

10. Zweig JL, Carrau RL, Celin SE, Snyderman CH, Kassam A, Hegazy H.Endoscopic repair of acquired encephaloceles, meningoceles, andmeningo-encephaloceles: predictors of success. Skull Base 2002;12:133–139.

11. Wormald PJ, McDonogh M. ‘Bath-plug’ technique for the endoscopic man-agement of cerebrospinal fluid leaks. J Laryngol Otol 1997;111:1042–1046.

12. Wormald PJ. Endoscopic Sinus Surgery Anatomy, Three-DimensionalReconstruction, and Surgical Technique. New York, NY: Thieme; 2005.

13. Wigand ME. Transnasal ethmoidectomy under endoscopical control.Rhinology 1981;19:7–15.

14. Hegazy HM, Carrau RL, Snyderman CH, Kassam A, Zweig J. Transnasalendoscopic repair of cerebrospinal fluid rhinorrhea: a meta-analysis.Laryngoscope 2000;110:1166–1172.

15. Mehendale NH, Marple BF, Nussenbaum B. Management of sphenoidsinus cerebrospinal fluid rhinorrhea: making use of an extendedapproach to the sphenoid sinus. Otolaryngol Head Neck Surg 2002;126:147–153.

16. Casiano RR, Jassir D. Endoscopic cerebrospinal fluid rhinorrhea repair:is a lumbar drain necessary? Otolaryngol Head Neck Surg 1999;121:745–750.


913

image-guided endoscopic repair of cerebrospinal fluid rhinorrhea by the bath plug grafting technique

Documents