Le

MVa

b

c

a

ARRAA

KCEIE

1

cagosr

T

0h

Clinical Neurology and Neurosurgery 116 (2014) 28– 34

Contents lists available at ScienceDirect

Clinical Neurology and Neurosurgery

j o ur nal hom epage: www.elsev ier .com/ locate /c l ineuro

ow-dose intrathecal fluorescein and etiology-based graft choice inndoscopic endonasal closure of CSF leaks

atei A. Banua, Joon-Hyung Kima,∗, Benjamin J. Shina, Graeme F. Woodworthc,ijay K. Anandb, Theodore H. Schwartza

Department of Neurological Surgery, Weill Cornell Medical College, New York, USADepartment of Otolaryngology – Head and Neck Surgery, Weill Cornell Medical College, New York, USADepartment of Neurological Surgery, University of Maryland, Baltimore, USA

r t i c l e i n f o

rticle history:eceived 3 April 2013eceived in revised form 1 October 2013ccepted 9 November 2013vailable online 16 November 2013

eywords:SF leak repairndoscopic endonasal surgeryntrathecal fluoresceintiology-based repair

a b s t r a c t

Objective: Skull base cerebrospinal fluid (CSF) leaks of various etiologies are increasingly repaired throughthe natural corridor using an endoscopic endonasal approach. Characteristics of the skull base defectsignificantly correlate with etiology, which should be ascertained to guide surgical management. Theobjectives of this study were to assess the long-term outcomes of patients that underwent endoscopicendonasal repair of CSF leak using low-dose intrathecal fluorescein (ITF) and an etiology-based algorithmfor multi-layer graft closure.Methods: Patients were divided into 4 groups: A – congenital, B – post-traumatic, C – post-endonasalsurgery, D – post-craniotomy. Low-dose ITF was utilized in all case series. Long-term clinical follow-updata, including perioperative complications associated with the use of intrathecal fluorescein and leakclosure rates, were obtained retrospectively. Endoscopic visualization of fluorescein-stained CSF as wellas the method of skull base closure and graft material is detailed.Results: We identified a total of 41 patients (N = 24 in Group A, N = 4 in Group B, N = 12 in Group C andN = 1 in Group D) that underwent 50 CSF leak repairs using the endoscopic endonasal approach with anaverage follow-up of 31.6 months. Nine patients (21.9%) had undergone a previous attempt at CSF leakrepair. Lumbar drain was used intraoperatively in 26 patients (63.4%) and kept in place for an averageduration of 3.25 days. ITF successfully identified the site of leak in 80.5% of cases regardless of etiology.Leaks were successfully closed in 92% of patients. One patient (2.4%) experienced transient leg weak-

ness following lumbar drain placement. Another patient (2.4%) developed hydrocephalus requiring aventriculoperitoneal shunt.Conclusion: Low-dose ITF is a safe and useful adjunct to endoscopic endonasal repair of CSF leaks withminimal complications and successful localization of the leak in approximately 80%. An etiology-basedapproach to graft choice and duration of lumbar drain placement in CSF leak repair may optimize closurerates.

. Introduction

Skull base cerebrospinal fluid leaks require urgent repair to cir-umvent severe complications, such as meningitis, brain abscess,nd pneumocephalus [1–4]. While conservative management isenerally advocated for post-traumatic leaks, leaks of other eti-

logies and refractory cases of post-traumatic leaks require directurgical repair [2]. For both the transcranial and the endonasalepair approach, the same guiding principles apply, which include

∗ Corresponding author at: 525 East 68th Street, Starr Pavilion 651, USA.el.: +1 212 746 5620; fax: +1 212 746 2004.

E-mail addresses: schwarh@med.cornell.edu, jhkim17@gmail.com (J.-H. Kim).

303-8467/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.clineuro.2013.11.006

© 2013 Elsevier B.V. All rights reserved.

leak confirmation, precise site localization, and adequate water-tight closure [2].

Etiology of the leak should be accurately ascertained, as it affectsthe location, size and shape of the bony and meningeal defects, andthe presence of concomitant intracranial hypertension or menin-goencephalocele [4]. Hence, etiology of the leak is an importantparameter to be surveyed prior to surgical decision-making [5].Despite utilization of preoperative imaging and stereotactic nav-igation, the precise location of the defect often remains obscure.

Intrathecal fluorescein (ITF), which dyes the CSF a fluorescent

green that is visible in both white and blue light, is utilized as a use-ful adjunct for repair of CSF leaks [6–8]. While safety concerns havehistorically limited its use, administration of low-dose ITF after pre-medication has significantly reduced the incidence of associated

gy and Neurosurgery 116 (2014) 28– 34 29

srr

skftbtlaeamdtupi

2

2

wrNboTaswalaewistulcaioosnLflt

2

tptdwn

Fig. 1. Localizing the CSF leak using ITF. Fluorescein is injected into the intrathecalspace to aid with intra-operative CSF leak visualization. Blue filtered excitation lightis employed if the defect is not properly identified by white light to further enhancethe green fluorescence of the dye, thus clearly pinpointing the site of the leak. ITF

M.A. Banu et al. / Clinical Neurolo

ide effects [9,10]. Yet little has been addressed in the literatureegarding the long-term efficacy and safety of endoscopic CSF leakepair using low-dose ITF.

Here we report a series of 41 patients that underwent endo-copic endonasal repair of CSF leaks using low-dose ITF. To ournowledge, this study has the longest follow-up of over 30 monthsor CSF leak closure in the literature to date. We examined results ofhe endoscopic endonasal approach in identifying and closing skullase defects by a neurosurgeon and an ENT surgeon whose closureechniques evolved over time from simple packing to a multi-ayer approach with vascularized mucosal flaps. Operative nuancesccording to the etiology of the skull base defect are described, in anffort toward standardizing graft choice based on predicted char-cteristics of the suspected breach. We propose an etiology-guidedanagement algorithm that can aid in surgical and post-operative

ecision-making, including graft choice for skull base reconstruc-ion, pre-operative and intra-operative leak localization, and these of post-operative lumbar drain placement to augment graftatency. Long-term outcomes of the cases will be analyzed accord-

ng to the depicted algorithm.

. Methods

.1. Inclusion criteria

A prospectively acquired database was screened for patientsho underwent endoscopic endonasal surgery where CSF fistula

epair was the primary goal. All cases were performed at theew York Presbyterian Hospital – Weill Cornell Medical Center,etween October 2004 and December 2012. Eight patients fromur previous reports on ITF [10,11] were included in this study.he patients were treated by the senior authors of this paper (T.H.Snd V.K.A) as a collaborative effort of the departments of Neuro-urgery and Otolaryngology. The inclusion criteria for this studyere as follows: (1) The primary goal of the procedure was to stop

CSF leak; (2) Intrathecal fluorescein use for intraoperative CSFeak detection, injected either via a lumbar puncture or through

lumbar drain placed at the time of surgery. No patients werexcluded solely based on etiology. However, patients in whom ITFas not used were excluded. These cases were primarily at the

nception of the series when ITF was not used routinely by theenior author. Outcome was determined through review of hospi-al and office records or telephone survey with targeted questionssed to assess recurrence or persistence of symptoms as well as

ong term complications, such as meningitis, recurrent seizures,hronic sinusitis, pulmonary embolus, deep venous thrombosis,rachnoiditis, hydrocephalus or the need for post-operative shunt-ng. Specific details of the surgical procedure included the successf ITF in identifying the leak intra-operatively, use of an intra-perative and post-operative lumbar drain, and graft choice: gasketeal with Medpore (Porex Corp., Newnan, GA) or vomer, an abdomi-al fat inlay, DuraGuard (Synovis, St. Paul, MN) or DuraGen (IntegraifeSciences, Plainsboro Township, NJ), or a vascularized nasoseptalap. The Institutional Review Board approved both the observa-ional study and the follow-up survey.

.2. Preoperative evaluation and fluorescein protocol

A wide range of diagnostic methods was employed to confirmhe existence of a CSF leak and precisely locate the skull base defectrior to surgery. Patients clearly leaking CSF on physical examina-

ion were imaged in an attempt to find the location of the skull baseefect. Patients with ambiguous leaks had specimens collected,hich were submitted to �2 transferrin analysis for definitive diag-osis. Regardless of presenting symptom, all patients additionally

can successfully delineate the skull base breach regardless of etiology and can safelybe employed increasing closure outcomes.

underwent imaging studies consisting of a fine cut coronal CT andan MRI scan. If the location of the defect was not evident, a CT cis-ternogram was performed. Prior to surgery, all patients receivedITF to determine the location and extent of the defect as well as thevolume of the leak for an effective subsequent watertight closure. Adetailed informed consent was obtained from each patient prior tothe procedure for the use of fluorescein. Our standardized protocolfor intrathecal fluorescein has been described in previous articles[10,11]. Briefly, after general anesthesia induction and intravenouspremedication with 10 mg dexamethasone and 50 mg diphenhy-dramine, the patient was placed in lateral decubitus for lumbarpuncture. 10 mL CSF withdrawn from the patient were mixed with0.25 mL 10% fluorescein solution (Akorn Inc., Buffalo Grove, IL) andreinjected slowly into the intrathecal space. If a lumbar drain wasused it was clamped for the duration of the procedure. After open-ing the sphenoid and ethmoid sinuses, both white and blue filteredlight, for fluorescent excitation, were used intraoperative to visual-ize the fluorescein tagged CSF, using a wavelength of 465–495 nm.First, white light was used. If the leak was not identified, filteredexcitation light was used to highlight the fluorescence (Fig. 1). Ifthe leak was still not visualized, a blocking filter of 515–555 nmwas added to diminish background illumination and further high-light the fluorescence. 0◦, 30◦ and 45◦ lens rigid 4-mm endoscopes(Karl Storz GmbH & Co., Tuttlingen, Germany) were used to explorethe defect and locate the CSF leak depending on location.

2.3. Etiology-based method of closure

Choice of closure was based on the anatomical location and sizeof the defect, dependent on the initiating factor. Our technique hasbeen refined over time using a variety of graft materials, althoughmore recently we have settled on a standard protocol, describedbelow (Table 1). This standard protocol was adopted in 2008, whichencompasses the latter half of our series. We also provide a sum-mary of graft types and the rationale behind specific choices basedon intraoperative findings [12]. In all cases, intraoperative naviga-tion was utilized, most commonly CT-navigation, to locate the siteof the previously identified or suspected breach in the skull base.The appropriate sinus, generally the sphenoid, ethmoid or a com-bined corridor, was opened to expose the defect. If the site of thedefect could not be identified pre-operatively, a more extensivebut gradual opening of the sinuses was performed for exploration.

Patients could be divided into 4 groups based on the etiology ofthe leak; A – congenital (i.e. encephalocele, meningocele or menin-goencephalocele); B – post-traumatic CSF leaks not responsive toconservative therapy; C – iatrogenic post-endonasal surgery (either

30 M.A. Banu et al. / Clinical Neurology and

Table 1Etiology-based algorithm for endoscopic endonasal closure of CSF leak in the skullbase.

Closure techniques Lumbar drain

Group A (congenital)

DuraGuard

3 daysAutologous fatVascularized nasoseptal flapDuraseal

Group B (post-traumatic)

DuraGuard

3 daysAutologous fatVascularized nasoseptal flapDuraseal

Group C (post-endoscopy)

Remove previous closure

1 dayAutologous fatMedpor or vomer±Nasoseptal flapDuraseal

Group D (post-craniotomy)

Autologous fat

1 day

Medpor or vomer±Nasoseptal flapDuraseal

fg

tmTmfvDiflMsphip

Tciv

FiAbp

If defect > 1 cm, gasket seal closureUsing autologous fascia lata

unctional sinus surgery or skull base procedures) and D – iatro-enic post-craniotomy.

In Group A, herniating brain tissue was first resected along withhe meninges. This maneuver invariably aggravated the leak. Ulti-

ately, however, a small defect in the skull base could be identified.he bony edges of the defect were cleared of all mucosal tissue. Aultilayer closure was employed, generally consisting of an inlay

ollowed by an onlay of graft material. Although we have usedarious materials over the years, we currently prefer an inlay ofuraGuard (Synovis, St. Paul, MN), an onlay of autologous fat, which

s then covered in more recent cases by a vascularized nasoseptalap invariably followed by Duraseal (Confluent Surgical, Waltham,A) (Fig. 2). This sturdy graft construct is designed to best with-

tand the stress induced by pulsatile hydrostatic forces [2]. Theseatients often have elevated body mass index (BMI), and likelyave unrecognized increases in intracranial pressure such as benign

ntracranial hypertension. For this reason, a lumbar drain is left inlace for a period of approximately 3 days at a rate of 5 ccs/h.

Group B consisted of patients with post-traumatic CSF leaks.hese patients often obtain resolution of their leaks through a

onservative approach consisting of bed rest with or without thensertion of a lumbar drain. However, failure to respond to conser-ative measures would lead to primary endonasal closure. Closure

ig. 2. Etiology based graft choice. Endoscopic repair of meningoencephalocelesnvolves an inlay of DuraGuard and an overlay of a vascularized nasoseptal flap.n onlay of abdominal fat may be added to the graft construct but its use haseen limited lately due to the increased risk of infection and interference withost-operative imaging studies.

Neurosurgery 116 (2014) 28– 34

technique was similar to congenital leaks, since the defects areoften quite small.

Group C consists of patients who have iatrogenic leaks after priorendonasal surgery, either for sinus disease or for resection of pitu-itary or other skull base tumors. Following sinus surgery, the leakmust be identified and since it is often a minute defect, the closure issimilar to that used for congenital meningoceles. However, lumbardrainage is only continued for 1 day since patients usually do nothave increased intracranial pressure. If prior endonasal skull basesurgery was performed, the location of the leak is generally knownbased on the location and nature of the initial pathology. In this sit-uation, the previous closure is completely taken down and redone.For pituitary tumors we utilize an inlay of fat with a rigid buttresssuch as Medpor (Stryker, Kalamazoo, MI) or vomer covered eitherby Duraseal (Confluent Surgical, Waltham, MA) or more recentlyby a nasoseptal flap in conjunction with Duraseal.

Group D consists of patients who had prior craniotomy forskull base tumors where the skull base was breached from aboverequiring a secondary “salvage” repair from below. The locationof the defect is often unclear and ITF is invaluable for intraopera-tive inspection. These leaks generally have a large intracranial deadspace from the initial tumor resection. We have employed an inlayof fat followed by a rigid buttress such as Medpor (Stryker, Kalama-zoo, MI) or vomer followed either by Duraseal or more recently bya nasoseptal flap. A lumbar drain is left in place for 24 h. For largedefects usually over 1 cm wide with bony edges around the defect,we employ a closure called the gasket seal [10,13]. In this situationwe harvest autologous fascia lata that is larger than the defect andplace it as an onlay, which is further anchored by a countersunkpiece of Medpor that acts as a rigid buttress. The edges of the fascialata emerge from around the Medpor and create a watertight sealagainst the bone. The construct is then covered with a nasoseptalflap and Duraseal.

3. Results

3.1. Patient characteristics

We identified 50 procedures performed on 41 patients (Table 2and Fig. 3). There were 26 female patients (63.4%) and 15 malepatients (36.6%) with a mean age at the time of surgery of 50.3years (range 18–77). The mean BMI (kg/m2) for patients in Group A(33 ± 8.1) was significantly higher than the other groups (ANOVA,p < 0.05), which is a known risk factor for recalcitrant leaks [14].Roughly, a third of the patients presented with headache whilerhinorrhea was present in 95% of cases.

A total of 27 patients (65.9%) had undergone prior surgery,including 10 (24.4%) with at least one previous attempt to repaira CSF leaks or resect an associated meningoencephalocele. Theseprevious operations occurred at other institutions. We achieveddefinitive closure in 9 of the 10 patients with refractory, previouslyoperated leaks. Specific operative details for the nine successfulclosure techniques employed are described in Table 3. Moreover,13 (31.7%) patients had undergone prior neurosurgical procedures,mostly endonasal resections of skull base tumors, including pitu-itary adenoma, meningioma, and clival chordoma.

3.2. Localization of CSF leaks

Location of defects, as identified on pre-operative imaging stud-ies, is depicted in Table 4. The majority of the leaks occurred at

the level of the sphenoid sinus. In 3 patients (7.3%), no fistula tractcould be identified using these studies. Intrathecal fluorescein wasadministered to all patients in this series according to the describedprotocol. Visualization of fluorescein stained CSF egression was

M.A. Banu et al. / Clinical Neurology and Neurosurgery 116 (2014) 28– 34 31

Table 2Patient characteristics based on etiology of CSF leak.

Total Etiology

Congenital Post-traumatic Post-endoscopy Post-craniotomy

No of patients 41 24 4 12 1Gender

Male 15 (36.6%) 4 (16.7%) 4 (100.0%) 6 (50.0%) 1 (100.0%)Female 26 (63.4%) 20 (83.3%) 0 (0.0%) 6 (50.0%) 0 (0.0%)

Age (years) 50.3 ± 14.6 52.4 ± 14.2 38.5 ± 16.0 50.3 ± 15.3 48.0BMI (kg/m2) 31.0 ± 7.7 33.0 ± 8.1 24.9 ± 2.3 27.2 ± 7.0 31.9Presentation

Headache 13 (31.7%) 7 (29.2%) 3 (75.0%) 3 (25.0%) 0 (0.0%)Nasal Obstruction 5 (12.2%) 2 (8.3%) 0 (0.0%) 3 (25.0%) 0 (0.0%)Rhinorrhea 39 (95.1%) 24 (100.0%) 4 (100.0%) 10 (83.3%) 1 (100.0%)

F and thC

nawis

TC

N

ig. 3. Patient selection flow chart. Diagram depicting cases included in the study

SF leak closure, 9 of which were successfully treated.

ot possible in 8 patients (19.5%). In these patients, we achieved

closure rate of 87.5%. In seven of these eight cases, the defectas accurately identified using neuronavigation and peri-operative

maging. The occult CSF leak remained symptomatic despite clo-ure attempts based on the here-depicted algorithm, emphasizing

able 3losure method in patients with previous history of CSF leak repair (n = 9).

Case no. Type of closure IF detection N

1 Autologous temporalis, plate No Y2 Fat, fascia lata Yes Y3 Gelfoam, Duragen, NS flap Yes Y4 Fat, fascia lata, gasket seal Yes Y5 NS flap, DuraGuard Yes N6 Fat, fascia lata, plate, gasket seal Yes N7 Fascia lata, bone Yes N8 DuraGuard, vomer Yes Y9 Duragen, gelfoam Yes Y

S, nasoseptal; IF, intrathecal fluorescein; LD, lumbar drain; VP, ventriculo-peritoneal.

e etiology-based groups analyzed. There were 10 cases with previous attempts of

the importance of leak visualization. Five patients had protrud-

ing meningoencephaloceles that further aided in pinpointing thedefect. In another case, the misplacement of the previous graft con-struct was evident on endoscopic examination, which helped usidentify the site of probable leak. Thus, in all of these eight cases,

euro-navigation Intra-op LD Post-op LD VP shunt

es No No Noes Yes Yes Noes Yes Yes Noes Yes No Noo No No Noo Yes Yes Yeso No No Noes Yes No Noes Yes No No

32 M.A. Banu et al. / Clinical Neurology and

Table 4Location of skull base defects on preoperative imaging (n = 50). Some leaks occurredin more than one location in the same patient.

Location of defect No. (%)

Sphenoid sinus 28 (67.3%)Cribriform plate 15 (36.6%)Ethmoid sinus 17 (41.5%)

abIuloflsdds

3

sldueallktdp

3

womOcctmtem

ohcesoTc

obd

Clivus 1 (2.4%)Unidentified 3 (7.3%)

n active leak through the skull base breach was absent as evidenty the lack of clear as well as stained CSF at the level of the defect.

n this situation we try to place the patient in Trendelenburg andse the Valsalva maneuver. If no leak is identified, the region of

ikely leak (often the thinned cribriform plate or region with defectn coronal CT) can be lined with fat and covered with a nasoseptalap. One case with persistent symptoms and recalcitrant leak had amall defect on the CT that could not be identified intraoperativelyespite employment of both ITF and the Valsalva maneuver. Weecided to close the defect according to the pre-operative imagingtudy.

.3. Surgical results and graft choices

The surgical approach and the type of graft were cho-en based on location, etiology and severity or volume of theeak. Although procedures after 2008 followed the guidelinesescribed above, in some cases different combinations weresed to adapt to local anatomy, particularly during our earlyxperience in this series. Graft choices according to etiologyre described in Table 5. The endonasal endoscopic proceduresasted an average of 160 min (range 71–327 min). Intra-operativeumbar drain was employed in 26 patients (63.4%) and wasept in place for a mean of 3.25 days (±1.4 SD). A ven-riculoperitoneal (VP) shunt was required in one patient whoeveloped hydrocephalus; CSF leak, however, did not recur in thisatient.

.4. Long-term outcomes

The efficacy of closure was 92% (46 of 50 cases). Thereere 3 persistent leaks after surgery and one late recurrence

ut of the 50 closures. Congenital leaks proved to be theost resilient, with persistence of the leak in 8.3% of cases.ne patient with a post-endoscopy CSF leak developed hydro-ephalus requiring a VP shunt. This patient eventually hadomplete symptom alleviation, and the leak did not recur afterhe repair. The average follow-up was 31 months (range 1–82

onths; Table 6). Three patients were lost to follow-up. Short-erm complications included one patient with transient lowerxtremity weakness presumed to be caused by lumbar drain place-ent.Persistent leaks occurred in 3 patients (7.3%). Of these patients,

ne had a BMI of 49, one had gamma knife procedure, which mayave influenced graft viability, while in the third case, fluores-ein stained or clear CSF was never identified despite additionalmployment of the Valsalva manuever. In this latter case, the clo-ure consisted of Gelfoam placement at the level of the lateral recessf the sphenoid sinus and Duraseal in conjunction with Floseal.he patient refused subsequent surgical treatment for the recurrentondition.

One patient developed chronic sinusitis, treated with antibi-tics. A variety of minor neurological symptoms were describedy three patients (7.3%), ranging from slight balance problems andizziness to nystagmus or occasional headaches.

Neurosurgery 116 (2014) 28– 34

4. Discussion

Risk factors for CSF leak have been discussed extensively in theliterature [15–19]. Accidental trauma from closed head injury isthe primary inciting event (44%) [20] reported by most sources,with surgical, iatrogenic trauma a close second (29%); both scenar-ios generally have normal ICP at the time of repair [7]. Comprisedmostly of small cracks, defects in the former group often close spon-taneously with conservative treatment (80%) [21] but the risk ofinfection is not insignificant (30–40%) [22]. Presence of tumors isanother potential initiating factor (22%) causing leaks by infiltrat-ing and eroding the bone, devascularizing the mucosa or blockingCSF flow and increasing ICP [22]. Resection of the tumor furtheradds to the dural and vascular insult that discourages wound heal-ing and graft viability. Spontaneous or congenital leaks are anotherrelatively common etiology with frequent recurrence [15] becauseof their funnel shape and associated hydrocephalus or benignintracranial hypertension [3,23]. Given the various potential etiolo-gies of the leak, an understanding of these etiological factors andthe local anatomy associated with specific defects should guide themethod of closure.

ITF has been advocated as an adjunct in the endoscopicendonasal identification of CSF leaks with a reported success rate ashigh as 96% [24]. In our series, we visualized the fluorescein tintedCSF roughly 80% of the time. In cases with absent visualization, itis suspected that the fluorescein may have not adequately circu-lated to the skull base or the skull base breach may have not beenactively leaking CSF. Given that no stained CSF was detected byendoscopic examination of the bony defect in any of the 8 cases, thelatter scenario is more likely. In some cases, protruding meningoen-cephaloceles at the level of the defect further aided in localizing thebreach. Fluorescein was administered before starting the endonasalapproach, which gives a time window of roughly one hour priorto skull base inspection. Additionally, Trendelenburg positioningand Valsalva maneuver have been employed to promote visual-ization of fluorescein in some cases. If unstained, clear CSF wasidentified, the defect was closed as described in the Methods sec-tion. When no defect was identified, the region of likely leak, oftenthe thinned cribriform plate, or the region with defect on imagingstudies can be lined with fat and covered with a nasoseptal flap.Consensus on the utility of ITF in achieving a watertight closureirrespective of fistula etiology has not been reached, with certaingroups advocating for the sole use of the Valsalva maneuver [2].We consider ITF to be an effective tool in ascertaining the quality ofclosure and identifying the leak site, or lack thereof at the conclu-sion of leak repair. Thus, by injecting fluorescein at the beginningof the procedure, we improve the tools for leak visualization, anessential step in achieving a complete closure, regardless of etiol-ogy. Interestingly, we achieved successful closure in 7 of 8 patientswithout fluorescein stained CSF leakage. Thus, visualizing the leakduring the repair procedure is useful, but not entirely necessary fora successful closure.

Beta2-transferrin, a protein specifically present in CSF, is thegold standard for confirming the presence of CSF [3]. However,patients must have high flow leaks to catch sufficient fluid forthis analysis. Nevertheless, controversy still hovers over the idealmethod of imaging the fistula. High resolution axial, sagittal andcoronal CT identifies the anatomical position of the bony defectwith precision, though it lacks the capacity to assess the activityof the fistula [17]. It has been the preferred method in most of ourpatients, clearly distinguishing and localizing the defect in 92.7%.Radioactive cisternograms are an invasive and rather poor map-

ping technique for skull base injuries [23]. CT cisternograms, on theother hand, appear to have a much higher sensitivity, the intrathe-cal contrast being able to distinguish meningoencephalocele sacs.However, success is limited if the patient is not actively leaking at

M.A. Banu et al. / Clinical Neurology and Neurosurgery 116 (2014) 28– 34 33

Table 5Operative characteristics for endoscopic endonasal repair (n = 41; 50 cases total).

Total Etiology

Congenital Post-traumatic Post-endoscopy Post-craniotomy

No of patients 41 24 4 12 1Lumbar drain 26 (63.4%) 16 (66.7%) 3 (75.0%) 6 (50.0%) 1 (100.0%)Duration of post-op lumbar drain (d) 3.3 ± 1.4 3.3 ± 1.4 3.0 ± 1.7 3.5 ± 1.3 2Intra-op fluorescein detection 33 (80.5%) 17 (70.8%) 4 (100.0%) 11 (91.6%) 1 (100.0%)

Type of graftGelfoam 19 (46.3%) 7 (29.2%) 2 (50.0%) 10 (83.3%) 0 (0.0%)Fat 30 (73.2%) 15 (62.5%) 4 (100.0%) 10 (83.3%) 1 (100.0%)Fascia lata 14 (34.1%) 6 (25.0%) 1 (25.0%) 6 (50.0%) 1 (100.0%)Miniplate 3 (7.3%) 0 (0.0%) 0 (0.0%) 3 (25.0%) 0 (0.0%)Bovine pericardium 4 (9.8%) 2 (8.3%) 0 (0.0%) 2 (16.7%) 0 (0.0%)DuraGuard 13 (31.7%) 8 (33.3%) 2 (50.0%) 3 (25.0%) 0 (0.0%)Duragen 4 (9.8%) 2 (8.3%) 0 (0.0%) 2 (16.7%) 0 (0.0%)Septum/vomer 13 (31.7%) 9 (37.5%) 1 (25.0%) 3 (25.0%) 0 (0.0%)Nasoseptal flap 15 (36.6%) 8 (33.3%) 3 (75.0%) 4 (33.3%) 0 (0.0%)Sealanta 41 (100.0%) 24 (100.0%) 4 (100.0%) 12 (100.0%) 1 (100.0%)Gasket seal 10 (24.4%) 4 (16.7%) 1 (25.0%) 5 (41.7%) 0 (0.0%)

Mean op time (min) 160.0 ± 63.1 172.2 ± 54.0 172.5 ± 46.0 134.0 ± 80.5 109

t[cw[tdctifwc

hsnriviitywas

avan

TE

a DuraSeal, Tisseel.

he time of the study, thus affecting the yield in intermittent leaks17]. The invasiveness of the procedure can be reduced by using MRisternography with T2 weighted images or fast spin echo sequenceith fat suppression to highlight CSF fistulas with a 89% accuracy

25]. An active leak is a sine qua non for most studies. Here, 8 ofhe 41 patients had transitory leaks that were difficult to visualizeuring the surgery. These unidentified leaks were almost entirelyongenital in nature, further adding to the difficult management ofhis category. In such cases, we rely heavily on the pre-operativemaging studies and neuronavigation tools. A fine cut CT is obtainedor all CSF leak patients to visualize the defect prior to proceedingith surgery. In cases where defect is not clearly visible, CT/MR

isternography can be additionally obtained.Yet uncommon but severe side effects associated with ITF

ave been reported in the literature [10], which ranges fromeizures and neurotoxicity [26] to headache, lower extremity weak-ess/numbness or even opisthotonus [27], primarily occurring withapid injection or lack of sufficient dilution, and simultaneousntrathecal application of contrast material [28]. Although intra-enous fluorescein can be administered in conjunction with thentrathecal delivery, the addition of systemic fluorescein does notncrease visibility while increasing the potential for side effects. Forhis reason, the senior author has stopped this practice in recentears. In this series, we experienced only one case of transient legeakness, which may have also been related to the placement of

lumbar drain. Major neurological side effects such as grand maleizures were not observed.

The choice of graft material is quite variable in the literatures are the various inlay and onlay options [14]. Bone graft har-

ested from the middle turbinate [18] placed in the epidural spacend covered by a fascial layer in an overlay fashion shows efficacyot only in leak cessation but also in long term infection control

able 6tiology-based long-term outcomes on follow-up.

Total Congenital

No of patients 41 24

Follow-up time (months) 31.6 ± 26 37.3 ± 29.7

Long term outcomeRecurrent 1 (2.4%) 0 (0.0%)

Persistent 3 (7.3%) 2 (8.3%)

VP shunt 1 (2.4%) 0 (0.0%)

[13,29]. Recent literature advocates the use of a nasoseptal flap, freeintranasal mucosal flap with or without additional fat graft for rein-forcement [30], and the gasket seal closure [31]. After employingseveral different methods, we have settled on a routine algorithm,as described in the Methods section. However, the series was accu-mulated over the course of this evolution in our approach and thusa variety of graft materials were used. Therefore, this allowed for arelative comparison of efficacy among the different techniques. Asa general rule of thumb, we believe that a type of inlay that will notbe dislodged is a critical first step. For small congenital and trau-matic leaks, we prefer DuraGuard since it is usually rigid enough tohold its shape. For larger defects, we employed the gasket seal clo-sure. A vascularized flap covered with tissue sealant was also a veryeffective final layer. Likewise, for patients with intracranial hyper-tension, a lumbar drain is quite useful in the first few days while thetissues are healing, as it diverts pressure off the closure and theo-retically aids in graft patency. On the other hand, an intracranial fatgraft has the advantage of promoting scar tissue formation and tak-ing up any dead space to prevent pooling, but it can interfere withfollow-up MR studies and can be a nidus for infection [2]. Due tothis reason, its future use may be more limited despite employingit in 73.2% of our cases.

A limitation of this study is the relatively heterogeneousgroup of patients with uneven distribution in their leak etiologiesand the graft material used for closure. Statistical analysis andfurther generalization of our results are thus limited based on thispresent cohort. The retrospective nature of the study should alsobe noted. Although the patients were entered into the databaseprospectively, long-term outcomes were obtained retrospectively.

Recall and “late look” biases are inherently present in a telephonesurvey, even by using a standardized questionnaire, as was thecase here. A more systematic and perhaps prospective study on the

Post-traumatic Post-endoscopy Post-craniotomy

4 12 18.2 ± 7.6 28.4 ± 18.0 37.1

1 (25.0%) 0 (0.0%) 0 (0.0%)0 (0.0%) 1 (8.3%) 0 (0.0%)0 (0.0%) 1 (8.3%) 0 (0.0%)

3 y and

est

5

afgscoolsia

D

ti

A

a

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

4 M.A. Banu et al. / Clinical Neurolog

tiology-based management algorithm as well as on the long-termafety and efficacy of low-dose ITF is warranted to fully evaluatehe role of low-dose ITF in endoscopic CSF leak closure.

. Conclusion

In this series, the endoscopic endonasal approach performed by neurosurgeon and an ENT surgeon successfully established leak-ree closure in 92% of cases. A vascularized flap is a viable option forrafting in complex cases. Recalcitrant leaks may require CSF pres-ure monitoring and/or VP shunt placement to ensure durable leaklosure. Careful selection of graft materials according to the etiologyf the defect and a multilayer closure using synthetic and autol-gous materials, vascularized flaps, tissue sealants and selectiveumbar drainage can increase the likelihood of successful endo-copic leak closure. Intrathecal fluorescein can identify the leak siten roughly 80% of cases regardless of etiology and is thus a usefulnd safe adjunct in endoscopic repair of skull base defects.

isclosure

The authors report no financial or material support concerninghe materials or methods used in this study or the findings specifiedn this paper.

cknowledgment

We would like to thank Matthew Holt for excellent technicalssistance in preparing Fig. 2.

eferences

[1] Kryzanski JT, Annino Jr DJ, Heilman CB. Complication avoidance in thetreatment of malignant tumors of the skull base. Neurosurg Focus 2002;12:e11.

[2] Schlosser RJ, Bolger WE. Nasal cerebrospinal fluid leaks: critical review andsurgical considerations. Laryngoscope 2004;114:255–65.

[3] Wise SK, Schlosser RJ. Evaluation of spontaneous nasal cerebrospinal fluid leaks.Curr Opin Otolaryngol Head Neck Surg 2007;15:28–34.

[4] Darakchiev BJ, Pensak ML. Cerebrospinal fluid dynamics in skull base surgery.Curr Opin Otolaryngol Head Neck Surg 2004;12:404–7.

[5] Banks CA, Palmer JN, Chiu AG, O’Malley BW, Jr., Woodworth BA, Kennedy DW,et al. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. OtolaryngolHead Neck Surg 2009;140:826–33.

[6] Senior BA, Jafri K, Benninger M. Safety and efficacy of endoscopic repair of CSFleaks and encephaloceles: a survey of the members of the American RhinologicSociety. Am J Rhinol 2001;15:21–5.

[7] Marton E, Billeci D, Schiesari E, Longatti P. Transnasal endoscopic repair ofcerebrospinal fluid fistulas and encephaloceles: surgical indications and com-

plications. Minim Invasive Neurosurg 2005;48:175–81.

[8] Sanderson JD, Kountakis SE, McMains KC. Endoscopic management of cere-brospinal fluid leaks. Facial Plast Surg 2009;25:29–37.

[9] Keerl R, Weber RK, Draf W, Wienke A, Schaefer SD. Use of sodium fluores-cein solution for detection of cerebrospinal fluid fistulas: an analysis of 420

[

Neurosurgery 116 (2014) 28– 34

administrations and reported complications in Europe and the United States.Laryngoscope 2004;114:266–72.

10] Placantonakis DG, Tabaee A, Anand VK, Hiltzik D, Schwartz TH. Safety of low-dose intrathecal fluorescein in endoscopic cranial base surgery. Neurosurgery2007;61:161–5, discussion 165–66.

11] Tabaee A, Placantonakis DG, Schwartz TH, Anand VK. Intrathecal fluorescein inendoscopic skull base surgery. Otolaryngol Head Neck Surg 2007;137:316–20.

12] Tabaee A, Anand VK, Brown SM, Lin JW, Schwartz TH. Algorithm for recon-struction after endoscopic pituitary and skull base surgery. Laryngoscope2007;117:1133–7.

13] Garcia-Navarro V, Anand VK, Schwartz TH. Gasket seal closure for extendedendonasal endoscopic skull base surgery: efficacy in a large case series. WorldNeurosurg 2011.

14] Nyquist GG, Anand VK, Mehra S, Kacker A, Schwartz TH. Endoscopic endonasalrepair of anterior skull base non-traumatic cerebrospinal fluid leaks, meningo-celes, and encephaloceles. J Neurosurg 2010;113:961–6.

15] Lindstrom DR, Toohill RJ, Loehrl TA, Smith TL. Management of cerebrospinalfluid rhinorrhea: the Medical College of Wisconsin experience. Laryngoscope2004;114:969–74.

16] Zuckerman JD, DelGaudio JM. Utility of preoperative high-resolution CT andintraoperative image guidance in identification of cerebrospinal fluid leaks forendoscopic repair. Am J Rhinol 2008;22:151–4.

17] Lloyd KM, DelGaudio JM, Hudgins PA. Imaging of skull base cerebrospinal fluidleaks in adults. Radiology 2008;248:725–36.

18] Ismail AS, Costantino PD, Sen C. Transnasal transsphenoidal endoscopicrepair of CSF leakage using multilayer acellular dermis. Skull Base 2007;17:125–32.

19] Tosun F, Gonul E, Yetiser S, Gerek M. Analysis of different surgical approachesfor the treatment of cerebrospinal fluid rhinorrhea. Minim Invasive Neurosurg2005;48:355–60.

20] Martin TJ, Loehrl TA. Endoscopic CSF leak repair. Curr Opin Otolaryngol HeadNeck Surg 2007;15:35–9.

21] Locatelli D, Rampa F, Acchiardi I, Bignami M, De Bernardi F, Castelnuovo P,et al. Endoscopic endonasal approaches for repair of cerebrospinal fluid leaks:nine-year experience. Neurosurgery 2006;58. ONS-246–56; discussion ONS-256–57.

22] Friedman RA, Cullen RD, Ulis J, Brackmann DE. Management of cerebrospinalfluid leaks after acoustic tumor removal. Neurosurgery 2007;61:35–9, discus-sion 39–40.

23] Connor SE. Imaging of skull-base cephalocoeles and cerebrospinal fluid leaks.Clin Radiol 2010;65:832–41.

24] Felisati G, Bianchi A, Lozza P, Portaleone S. Italian multicentre study onintrathecal fluorescein for craniosinusal fistulae. Acta Otorhinolaryngol Ital2008;28:159–63.

25] McMains KC, Gross CW, Kountakis SE. Endoscopic management of cere-brospinal fluid rhinorrhea. Laryngoscope 2004;114:1833–7.

26] Anari S, Waldron M, Carrie S. Delayed absence seizure: a complication ofintrathecal fluorescein injection. A case report and literature review. AurisNasus Larynx 2007;34:515–8.

27] Jones ME, Reino T, Gnoy A, Guillory S, Wackym P, Lawson W, et al. Identificationof intranasal cerebrospinal fluid leaks by topical application with fluoresceindye. Am J Rhinol 2000;14:93–6.

28] Wolf G, Greistorfer K, Stammberger H. [Endoscopic detection of cerebrospinalfluid fistulas with a fluorescence technique. Report of experiences with over925 cases]. Laryngorhinootologie 1997;76:588–94.

29] Gjuric M, Goede U, Keimer H, Wigand ME. Endonasal endoscopic closure ofcerebrospinal fluid fistulas at the anterior cranial base. Ann Otol Rhinol Laryngol1996;105:620–3.

30] Elrahman HA, Malinvaud D, Bonfils NA, Daoud R, Mimoun M, Bonfils P, et al.

Endoscopic management of idiopathic spontaneous skull base fistula throughthe clivus. Arch Otolaryngol Head Neck Surg 2009;135:311–5.

31] Leng LZ, Brown S, Anand VK, Schwartz TH. Gasket-seal watertight closure inminimal-access endoscopic cranial base surgery. Neurosurgery 2008;62. ONSE342–43; discussion ONSE 343.