OtolaryngologyHead and Neck Surgery (2009) 140, 826-833

ORIGINAL RESEARCHSINONASAL DISORDERS

Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years

Caroline A. Banks, MD, James N. Palmer, MD, Alexander G. Chiu, MD, Bert W. OMalley, Jr., MD, Bradford A. Woodworth, MD, and David W. Kennedy, MD, Philadelphia, PA
Sponsorship or competing interests that may be relevant to content are disclosed at the end of this article. ABSTRACT OBJECTIVE: Endoscopic repair of cerebrospinal uid (CSF) leaks has become a routine approach. This study describes endoscopic closure of a large series over 21 years, focusing on management, surgical technique, and long-term outcomes. STUDY DESIGN: Chart review. SUBJECTS AND METHODS: CSF leak patients treated by the senior author and at the University of Pennsylvania from 1987 to 2008. The data included body mass index (BMI), etiology, defect location, graft material, presence of encephalocele, lumbar drain, history of meningitis, intracranial pressure, recurrence, and follow-up. RESULTS: A total of 193 cases were identied. Follow-up ranging from 1 month to 9 years (mean 21 months) was available on 166 patients. The etiology was spontaneous in 77 patients (40%), traumatic in 109 (56%), and congenital in 7 (4%). The average BMI of spontaneous CSF leak patients (35) was greater (P 0.001) than both traumatic (30) and congenital patients (23). Defects were most commonly located in the sphenoid (n 62, 32%) and ethmoid (n 60, 31%) The initial success rate was 91 percent (n 176) and overall success rate was 98 percent (n 190). CONCLUSION: The overall success rate (98%) and low morbidity in this large series support endoscopic approach as standard of care for CSF leak closure. 2009 American Academy of OtolaryngologyHead and Neck Surgery Foundation. All rights reserved.

erebral spinal uid (CSF) leak is a serious condition occurring spontaneously or as a result of head trauma, surgery, neoplastic invasion, inammatory erosion of the skull base, or congenital malformations. The etiologies can be classied broadly as traumatic, including surgical and accidental trauma, and nontraumatic. The nontraumatic category can be subdivided into patients with high intracranial pressure and normal-pressure leaks. With the exception of some leaks arising from blunt trauma, surgical management of CSF leaks is typically required to repair the defect and to prevent serious complications, such as meningitis and ab-

scess. Historically, the closure of these defects was approached intracranially using classic open skull base surgical procedures. Although the open intracranial approach provides a large surgical eld and allows for direct visualization, it is associated with high morbidity, including intracerebral hemorrhage, cerebral edema, frontal lobe decits, lengthened hospital stay, anosmia, and high recurrence rates.1,2 Over the past fty years, CSF closure techniques have undergone signicant evolution. Minimally invasive approaches, characterized by improved success rates and decreased morbidity, have gained increasing favor over the traditional open craniotomy repairs. Dohlman in 1948, Hirsch in 1952, and Vrabec and Hallberg in 1964 described successful naso-orbital, transeptal, and endonasal approaches, respectively.3 Advancements in endoscopic technology and improved surgical techniques have led to the development of the transnasal endoscopic approaches for CSF leaks. Endoscopic closure was rst mentioned in a paper on endoscopic ethmoidectomy by Wigand in 1981,4 and a small series was subsequently published by Mattox and Kennedy in 1990.5 Subsequently, endoscopic repair has become an accepted standard of care for the operative management of most CSF leaks. Multiple publications site the low incidence of complications and low rates of recurrence associated with a minimally invasive endoscopic approach.6 This study explores endoscopic CSF leak repair over the past 21 years, and reports on long-term follow-up and patient outcomes.

METHODS
The study was approved by the University of Pennsylvania Institutional Review Board. A retrospective chart review was preformed on patients diagnosed with a CSF leak and admitted to the Hospital of the University of Pennsylvania (HUP) or operated on by the senior author (D.W.K.). The series included cases operated on between 1987 and 2008.

Received September 26, 2008; revised December 15, 2008; accepted December 24, 2008.

0194-5998/$36.00 2009 American Academy of OtolaryngologyHead and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2008.12.060

Banks et al

Endoscopic closure of CSF rhinorrhea: 193 cases . . .

827

Table 1 Study group characteristics, etiology, and recurrence rates Spontaneous (n 77) Mean (SD) Age (y) BMI (lb/in2) 703 Height (in) Weight (lb) Sex (n) Male Female Race/ethnicity (n) White Black/African American Asian Hispanic Indian Unknown Etiology FESS Tumor Trauma Vascular (Neurosurgery) Other Recurrent CSF leak 51.4 35.4 66.2 221.7 (11.8) (7.5)* (3.8) (54.6) Traumatic (n 109) Mean (SD) 47.1 29.7 68.2 198.6 (15.3) (7.8)* (4.4) (65.8) Congenital (n 7 ) Mean (SD) 36.5 23.3 63.2 140.6 (20.9) (9.5) (10.2) (68.5)

20 (26.0)* 57 (74.0)* 30 (39.0)* 28 (36.4)* 0 2 (2.6) 1 (1.3) 15 (19.5) 50 39 16 3 1 7 (9.1)

68 (62.4)* 41 (37.6)* 66 (60.6)* 7 (6.4)* 1 (0.9) 0 1 (0.9) 34 (31.2)

3 (42.8) 4 (57.1) 4 (57.1) 1 (14.3) 0 1 (14.3) 0 1 (14.3)

10 (9.2)

*Statistically signicant difference between groups, P 0.001. Statistically signicant difference between groups, P 0.003.

Patients with a diagnosis of CSF leak were identied through operative reports, billing records, and operating room schedules. Multiple factors were recorded, including sex, age, body mass index (BMI), leak etiology, method of diagnosis, size and location of defect, type of graft, presence of encephalocele, use of lumbar drain, history of meningitis, and months of follow-up. Evidence of increased intracranial pressure was determined from lumbar drain opening pressure, clamped drain pressure, and the presence of empty sella syndrome on MRI.

tumors comprised the majority (49%) of tumor resections in which CSF leaks were encountered and repaired. The classic symptom of clear rhinorrhea was the most common presentation for all cases, and the frequency of headache

RESULTS
One hundred ninety-three patients were identied with a CSF leak during the 21-year period. The CSF leak was diagnosed by the treating physician as spontaneous in 77 (40%) patients, traumatic in 109 (56%) patients, and congenital in 7 (4%) patients. Patient characteristics are described in Table 1. A minimum of 1-month follow-up (range 1 month to 9 years, mean 21 months) was available on 166 patients. With respect to traumatic CSF leaks, the most common etiology was prior functional endoscopic sinus surgery (FESS), occurring in 50 (46%) of the 109 cases (Fig 1). Twenty-two of these patients (44%) were undergoing revision surgery. Other causes of traumatic CSF leaks include tumor resection (36%), trauma (15%), and neurosurgical procedures for vascular anomalies (3%) (Table 1). Pituitary

Figure 1 Postoperative intracranial air following FESS. Coronal CT scan demonstrates a minimal amount of intracranial air (white arrow) and defect in the left ethmoid roof adjacent to the middle turbinate insertion following FESS.

828

OtolaryngologyHead and Neck Surgery, Vol 140, No 6, June 2009 (10%) patients to maintain patency and to promote drainage from the sinus. Twenty-four (31%) patients diagnosed with a spontaneous CSF leak and 4 (4%) patients diagnosed with a traumatic CSF leak received acetazolamide to decrease intracranial pressure. A ventriculoperitoneal (VP) shunt was placed in 12 (16%) spontaneous and six (6%) traumatic cases. On physical examination, nasal polyposis was present in eight (10%) spontaneous cases, one (14%) congenital case, and 29 (27%) traumatic cases. The internal failure rate for repair was 9 percent (17 of 193) with an average follow-up of 23 months (range 0 to 64 months) (Table 1). Nine percent of both spontaneous (n 7) and traumatic (n 10) patients developed recurrent CSF leaks. Patients with a spontaneous etiology developed recurrent leaks in an average of 7 months after repair at HUP (range 4 days to 24 months). Among spontaneous CSF leak patients who presented to HUP for both the initial repair and the revision surgery, four (57%) had defects in the same area, whereas the remaining three developed other areas of dehiscence. CSF leaks recurred in 13 percent of spontaneous patients with original defects in the sphenoid sinus (n 4), 13 percent of ethmoid defects (n 2), and 6 percent of patients with cribriform defects (n 1) (Table 5). Recurrence after endoscopic repair of traumatic CSF leaks occurred in an average of 4 months, with a range of 4 days to 29 months. Among FESS cases, 11 percent of cribriform defects and 10 percent of ethmoid defects leaked again (Table 5). Five (50%) of the traumatic cases presented with recurrent CSF leak during the immediate postoperative period. The location of the defect in seven (70%) traumatic cases was the same for the primary and secondary operation.

Table 2 Presenting signs and symptoms of CSF leak Spontaneous Traumatic (n 77) (n 109) N (%) N (%) Clear rhinorrhea Isolated headache Headache with rhinorrhea Meningitis History of meningitis 49 3 5 9 22 (64) (4) (7) (12) (29) 40 9 3 9 22 (37) (8) (3) (8) (20)

and meningitis was similar in both the traumatic and spontaneous groups (Table 2). Among all CSF leak cases, the defect was most commonly located in the sphenoid sinus (n 62) and the ethmoid roof (n 60) (Fig 2). The location of the defect was related to CSF leak etiology (Table 3). A variety of grafts were used in CSF leak repair, and the type of graft was associated with the location of the lesion (Table 4). Meningocele or meningoencephalocele was present in 70 (91%)of the 77 spontaneous CSF leak cases, 29 of the 109 (27%) traumatic cases, and 100 percent of the congenital cases (Fig 3). Lumbar drain was used therapeutically and diagnostically in 140 (73%) of the total 193 cases. A total of 113 patients received intrathecal uorescein to determine the location of the lesion. No patients had any neurological symptoms from the uorescein, but one patient developed premature ventricular contractions, apparently associated with uorescein injection. A frontal stent was placed in 20

Figure 2 Coronal CT (A) and axial T2-weighted MRI (B) in a patient with a spontaneous CSF leak and a small meningocele medial to the left middle turbinate.

Banks et al

Endoscopic closure of CSF rhinorrhea: 193 cases . . .

829

Table 3 CSF leak etiology and location of defect Etiology Spontaneous (n 77) N (%) 31 (40.3) 4 13 2 2 21 (27.3) 1 7 1 4 5 16 (20.8) 9 (11.7) 3 31 (62.0) 3 14 3 9 (18.0) 1 4 (8.0) 1 1 2 (2.6) 0 0 0 0 1 (2.0) 1 1 1 (1.3) 3 (6.0) 1 6 (15.4) 0 0 0 1 (2.6) 1 2 (5.1) 3 0 1 (2.6) 1 1 (2.6) 1 1 1 (7.7) 5 (30.8) 1 0 1 (7.7) 1 1 (7.7) 0 0 0 1 2 (28.6) 0 0 0 1 (33.3) 1 (14.3) 1 (14.3) FESS (n 50) N (%) 2 (4.0) Tumor (n 39) N (%) 20 (51.3) 2 1 1 1 8 1 8 (20.5) 4 1 1 2 (15.4) Trauma (n 16) N (%) 6 (30.8) 1 NSGY (n 3) N (%) 2 (66.7) Congenital (n 7) N (%) 1 (14.3)

Site of defect Sphenoid Central sphenoid Lateral sphenoid Medial sphenoid Opticocarotid recess Posterior sphenoid Sella Sphenoid roof Ethmoid Anterior ethmoid Ethmoid roof Medial ethmoid Posterior ethmoid Supraorbital ethmoid Cribriform Lateral lamella Frontal Frontal recess Medial frontal Posterior frontal Superior frontal Ethmoid/cribriform Sphenoethmoid Posterior ethmoid Frontal/ethmoid Anterior ethmoid Ethmoid roof Frontal recess Posterior frontal Unknown

2 (28.6)

NSGY, Neurosurgery. There is an association between the etiology of the leak and the location of the defect.

DISCUSSION
Etiology and Leak Site
Conrmation of the diagnosis, determining the site of a CSF leak, and determining the etiology are all crucial to management. Patients with allergic and chronic rhinitis can have unilateral watery rhinorrhea that mimics CSF rhinorrhea except for its chemical composition. Collection of uid for a beta-2 transferrin remains the gold standard for leak conrmation. Fine-cut coronal and axial CT scans and MR may be helpful in localization; however, on the basis of our experience, we believe that the most accurate leak site identication is performed intraoperatively with the use of blue light endoscopy after intrathecal uorescein injection. We utilize 0.1 cc of 10 percent uorescein mixed with 10 cc of the patients own CSF and inject this solution over a period of approximately 5 minutes, followed by a period of approximately 1 to 2 hours before endoscopy to allow the dye to circulate. Because intrathecal uorescein is not ap-

proved by the Food and Drug Administration, a full informed consent for its utilization is recommended. The neurological complications of intrathecal uorescein, including seizures, headaches, and cranial nerve decits are well documented.7 One patient in our study developed multiple premature ventricular contractions after receiving an injection of uorescein, but none of our patients experienced neurological symptoms associated with intrathecal injection. Our study reinforces recent reports, which indicate the safety of very low-dose uorescein administered with a slow infusion rate.8 It is important to recognize that intrathecal radioactive tracer studies have been associated with a signicant false-positive rate and are best replaced with intrathecal uorescein and intraoperative blue light endoscopy as the denitive study for the presence and localization of a leak. The cause of CSF rhinorrhea is most commonly traumatic, including surgical or accidental injury. Among iatrogenic CSF leaks, FESS and neurosurgical procedures ac-

830
Table 4 Type of graft used in CSF leak repair

OtolaryngologyHead and Neck Surgery, Vol 140, No 6, June 2009

Site of defect Material used Frontal bone Mastoid bone Ethmoid bone Turbinate bone Septal bone Septal cartilage Mucoperiosteum Mucoperichondrium Temporalis muscle Temporalis fascia Nasal mucosa Abdominal fat Sphenoid (n 62) N (%)* 7 2 8 9 2 1 2 6 22 16 18 0 (11.3) (3.2) (12.9) (14.5) (3.2) (1.6) (3.2) (9.7) (35,5) (25.8) (29.0) Ethmoid (n 60) N (%)* 1 (1.7) 7 (11.7) 0 10 (16.7) 2 (3.3) 6 (10.0) 1 (1.7) 6 (10.0) 1 (1.7) 17 (28.3) 25 (41.7) 1 (1.7) Cribiform (n 28) N (%)* 0 4 (14.3) 0 4 (14.3) 2 (7.1) 2 (7.1) 4 (14.3) 4 (14.3) 0 7 (25.0) 10 (35.7) 1 (3.6) Frontal (n 22) N (%)* 2 (9.1) 2 (9.1) 0 2 (9.1) 4 (18.2) 1 (4.5) 0 2 (9.1) 1 (4.5) 6 (27.3) 9 (40.1) 4 (18.2)

Table shows the variety of grafting materials used and the relationship between location of defect and type of tissue graft. *Percentages may not add up to 100% when graft materials are used in combination.

count for the majority of cases.9 However, in terms of the overall number of FESS cases performed and head injuries received, traumatic CSF leaks are relatively uncommon, occurring in less than 1 percent of FESS cases10 and in 1 percent of all patients with head injuries.11 In our study, FESS and tumor resection were responsible for the bulk of

traumatic cases. Spontaneous CSF leaks account for a smaller percentage of cases in the literature; however, in our series, 77 of 193 cases were spontaneous. The etiology of CSF leak was based on the diagnosis made by the treating physician. Seven of the 193 cases were classied as congenital CSF leaks. It might be expected that congenital leaks would present at a young age; however, the mean age among congenital CSF leaks was 37 years. Two of the seven patients had well-documented histories of chronic CSF leak symptoms, including intermittent rhinorrhea for 25 years and a history of congenital encephalocele repair. In the cases in which chronic CSF leak symptoms were not documented, it is possible that the surgeon may have erred in diagnosis of congenital versus spontaneous etiology. Spontaneous CSF leaks are associated with idiopathic intracranial hypertension (IIH), also known as benign intracranial hypertension or pseudotumor cerebri.12 However, the raised intracranial pressure (ICP) does not become evident until the pressure is rechecked after the leak is closed.
Table 5 Site of defect in recurrent CSF leaks Site of defect Etiology Spontaneous FESS Trauma Tumor Sphenoid Ethmoid Cribiform Frontal N (%) N (%) N (%) N (%) 4 (12.9) 0 1 (16.7) 2 (10.0) 2 3 1 1 (12.5) 1 (6.2) 0 (9.7) 1 (11.1) 0 (50.0)* 1 (100) 1 (20.0)* (11.1) 0 0

Figure 3 Coronal T2-weighted MRI scan demonstrating a large, right-sided congenital encephalocele in a patient with a CSF leak and a secondary obstruction of the right maxillary sinus.

Table shows the site of the original defect in recurrent CSF leaks according to etiology. *Patient had a leak in the ethmoid and frontal sinuses.

Banks et al

Endoscopic closure of CSF rhinorrhea: 193 cases . . .

831

IIH, characterized by ICP, has an uncertain etiology and classically affects middle-aged, obese females. Our demographic data conrms prior reports of a relationship between IIH and spontaneous CSF rhinorrhea, demonstrating an average BMI of 35, a female predominance of 75 percent, and an average age of 51 years among patients with spontaneous CSF leaks. The average BMI of spontaneous CSF leak patients was statistically greater than the average BMI of traumatic (30) and congenital patients (23). The spontaneous CSF leak group comprised a signicant percentage of African American patients (36%) compared with traumatic (6%) etiologies. This nding is likely related to an increased average BMI in this patient population. In light of the relationship between spontaneous leaks and raised BMI, it is interesting to speculate whether the incidence of spontaneous leaks will increase with the obesity epidemic in industrialized nations.

Management
The management of CSF rhinorrhea should be tailored according to etiology. Cases of CSF leak caused by craniomaxillofacial trauma have traditionally been treated conservatively with bed rest, head elevation, and strict sinus precautions prior to surgical repair. On the basis of the observation that the majority of CSF leaks arising from closed head injuries will resolve spontaneously, many authors suggest a trial of conservative management unless there is coexisting neurological decline or additional intracranial pathology. However, long-term post-injury follow up has demonstrated up to a 29% incidence of subsequent meningitis in these patients, raising the question as to whether more of the leaks should be closed endoscopically.13 Multiple studies have demonstrated the type of tissue used for leak closure does not appear to signicantly inuence success rates. Materials utilized previously included fat, fascia, acellular collagen, and pericranium to name but a few. Septal mucoperiosteum has the advantages of being thick and readily available, and was used as free graft in the majority of anterior cranial fossa and lateral sphenoid leaks in this series. More recently, improved success has been reported in the closure of large defects with the use of pedicled septal aps based on the sphenopalatine artery.14 Multilayer grafts also have conceptual advantages; these grafts combine intracranial fascia or collagen with cartilage or bone harvested from the nasal septum or mastoid cortex, and the bone is then covered with a mucosal graft. However, no signicant improvement in results has been documented from the use of multilayer grafts. On the other hand, our experience has been that adequate preparation of the defect site, so as to create a at surface to which the graft can be made to fully conform, has been critical to closure success. In cases of spontaneous CSF leak with elevated ICP, a diuretic is often indicated, and 29 percent of patients with spontaneous CSF leak in our series received acetazolamide. Schlosser et al15 reported a mean decrease in ICP of 9.9 cm water approximately 3 to 4 hours after the administration of

intravenous (IV) acetazolamide. Patients with an adequate response to IV diuretics are routinely prescribed long-term oral diuretic therapy, and our previous series of spontaneous CSF leaks emphasized the need for strict management of underlying intracranial hypertension.16 Further investigation is needed to determine whether chronic use of oral diuretics is effective in maintaining ICP within the normal range, thus preventing CSF leak recurrence. VP drains are typically reserved for patients with spontaneous leaks who have recurrences postclosure or very high intracranial pressures, or who do not respond to acetazolamide. Lumbar drains are increasingly being used diagnostically in conjunction with uorescein. However, their role as a therapeutic adjunct to closure is more controversial and remains unproven, although the concept of using a drain to lower the pressure and allow a graft to seal does appear rational. Seventy-three percent of the patients in our study (n 113) underwent placement of a lumbar drain. Most commonly, this procedure was for diagnostic purposes as a method of introducing the intrathecal uorescein. Although a combination of medical management, lumbar drain, and endoscopic repair of the CSF leak is often successful, cases of refractory elevated ICP and recurrent leakage may require more permanent diversion procedures, such as VP shunts. VP shunts consist of a ventricular catheter, a shunt valve, and a peritoneal catheter. These shunts are associated with a signicant failure rate secondary to complications, including infection, obstruction, mechanical malfunction, and over- or under-drainage. Among the 18 patients who underwent VP placement, one patient presented with spiking fevers postoperatively, and one patient had two failed shunts followed by a CSF leak recurrence. Owing to the morbidity associated with VP shunts, careful patient selection should be considered. A prior study from our institution measured ICP postoperatively and initiated a trial of diuretics. Patients with signicantly elevated ICP, or those who failed to respond to IV acetazolamide, were considered for the VP shunt procedure.15

Outcomes
Initial success rates of CSF leak repair ranging from 85 to 90 percent and overall success rates of 97 to 99 percent have been well documented in the literature and demonstrate the efcacy of endoscopic repair.17-19 Our initial success rate of 91 percent and overall success rate of 98 percent fall within these accepted standards, but were overall somewhat lower than those previously reported from this institution by Lanza et al.3 The higher initial success rate experienced by Lanza in 1996 may be explained by differences in CSF leak etiology among the two studies, with endoscopic sinus surgery being responsible for 33 percent of cases in the prior study compared with 26 percent of cases in our current study. The long-term success of endoscopic repair is inuenced by CSF leak etiology. Studies indicate increased failure among spontaneous CSF leak repair, especially within the subset of patients demonstrating evidence of increased ICP.18 In addition to intracranial hypertension,

832

OtolaryngologyHead and Neck Surgery, Vol 140, No 6, June 2009

commonly cited patient factors associated with recurrent CSF leak include obesity, noncompliance, comorbidities, and poor wound healing. In a review of long-term outcomes of endoscopic repair, Zuckerman et al19 focused on the timing of recurrent CSF leaks. The average time for recurrence in their series was 7 months, ranging from 1 to 25 months. We found a similar time frame in our study for failure of repair, with spontaneous leaks recurring in an average of 7 months, ranging from 4 days to 24 months, and traumatic leaks recurring in an average of 4 months, ranging from 4 days to 29 months. Our data are consistent with the conclusion of Zuckerman et al that the majority of recurrent CSF leaks occur within 2 years following surgical repair. In addition, we found that 50 percent (n 5) of traumatic recurrent leaks presented within the rst 2 weeks of the postoperative period, probably reecting a technical failure rather than a true recurrence. Meningitis is a potentially fatal complication of CSF leaks. Conservative treatment of CSF leaks alone, including bed rest, lumbar drains, and/or diuretics is associated with a high incidence of ascending bacterial meningitis.13 BernalSprekelsen et al20 found that endoscopic closure of defects prevented meningitis in both patients who had previously presented with meningitis and patients without a history of intracranial infection. We also advocate early or immediate closure of surgical CSF leaks. In our series, 46 (24%) patients had a history of meningitis before surgical closure. One patient had symptoms of recurrent meningitis postoperatively, presenting with multiple episodes of aseptic meningeal irritation. Following surgical repair, there were four new cases of bacterial meningitis and one new case of clinical meningitis with sterile cultures. These outcomes suggest that endoscopic closure of CSF leaks is highly effective in preventing primary and recurrent cases of meningitis.

AUTHOR INFORMATION
From the Department of OtorhinolaryngologyHead and Neck Surgery, University of Pennsylvania Health System. Corresponding author: David W. Kennedy, MD, Department of Otorhinolaryngology, 21 Penn Tower, 399 S. 34th St, Philadelphia, PA 19104-4385. E-mail address: David.Kennedy@uphs.upenn.edu. Presented at the Annual Meeting of the American Academy of OtolaryngologyHead and Neck Surgery, Chicago, Illinois, September 21-24, 2008.

AUTHOR CONTRIBUTIONS
David W. Kennedy, study design, coauthor; Caroline A. Banks, study research, primary author; James N. Palmer, provision of cases, manuscript review; Alexander G. Chiu, provision of cases, manuscript review; Bert W. OMalley, Jr, manuscript review and contributions, provision of some cases; Bradford A. Woodworth, case and manuscript review.

DISCLOSURES
Competing interests: David W. Kennedy, consultant and royalties: Medtronic-Xomed, consultant: Sinucare, Schering-Plough; speaker: Merck; James N. Palmer, scientic advisory board: GEMSNAV; speaker: Pri-Cara; consultant: Medtronic-Xomed, Gyrus ACMI; advisory board: NeilMed; Alexander G. Chiu, consultant: Medtronic ENT, BrainLAB. Sponsorships: None.

REFERENCES
1. McCormack B, Cooper PR, Persky M, et al. Extracranial repair of cerebrospinal uid stulas: technique and results in 37 patients. Neurosurgery 1990;27:4127. 2. Tolley NS, Brookes GB. Surgical management of cerebrospinal uid rhinorrhoea. J R Coll Surg Edinb 1992;37:125. 3. Lanza DC, OBrien DA, Kennedy DW. Endoscopic repair of cerebrospinal uid stulae and encephaloceles. Laryngoscope 1996;106:1119 25. 4. Wigand ME. Transnasal ethmoidectomy under endoscopic control. Rhinology 1981;19:715. 5. Mattox DE, Kennedy DW. Endoscopic management of cerebrospinal uid leaks and cephaloceles. Laryngoscope 1990;100:857 62. 6. Martin TJ, Loehrl TA. Endoscopic CSF leak repair. Curr Opin Otolaryngol Head Neck Surg 2007;15:359. 7. Keerl R, Weber RK, Draf W, et al. Use of sodium uorescein solution for detection of cerebrospinal uid stulas: an analysis of 420 administrations and reported complications in Europe and the United States. Laryngoscope 2004;114:266 72. 8. Placantonakis DG, Tabaee A, Anand VK, et al. Safety of low-dose intrathecal uorescein in endoscopic cranial base surgery. Neurosurgery 2007;61:1615. 9. Gassner HG, Ponikau JU, Sherris DA, et al. CSF Rhinorrhea: 95 consecutive surgical cases with long term follow-up at the Mayo Clinic. Am J Rhinol 1999;13:439 47. 10. Stankiewicz JA. Cerebrospinal uid stula and endoscopic sinus surgery. Laryngoscope 1991;101:250 6. 11. Mendizabal GR, Moreno BC, Flores CC. Cerebrospinal uid stula: frequency in head injuries. Rev Laryngol Otol Rhinol (Bord) 1992; 113:4235. 12. Schlosser RJ, Woodworth BA, Wilensky EM, et al. Spontaneous cerebrospinal uid leaks: a variant of benign intracranial hypertension. Ann Otol Rhino Laryngol 2006;115:495500.

CONCLUSION
In our study, we present the largest case series of traumatic and nontraumatic CSF leaks to date, and review our experience with endoscopic repair spanning a 21-year period. On the basis of our experience, we recommend early endoscopic leak closure and the use of adjunctive measures to lower pressure in patients with raised intracranial pressure. Using this protocol, we found no signicant difference between the frequency of recurrence of spontaneous leaks compared with traumatic and congenital defects within the period of follow-up of the study. Unlike open intracranial approaches, the morbidity of endoscopic leak closure is minimal, thus signicantly recalibrating the risk-benet ratio of early leak closure versus watchful waiting. The overall success rate of 98 percent and low incidence of morbidity strongly support endoscopic closure.

Banks et al

Endoscopic closure of CSF rhinorrhea: 193 cases . . .

833

13. Bernal-Sprekelsen M, Bleda-Vzquez C, Carrau RL. Ascending meningitis secondary to traumatic cerebrospinal uid leaks. Am J Rhinol 2000;14:2579. 14. Hadad G, Bassagasteguy L, Carrau RL, et al. A novel reconstructive technique after endoscopic expanded endonasal approaches: vascular pedicle nasoseptal ap. Laryngoscope 2006;116:1882 6. 15. Schlosser RJ, Wilensky EM, Grady MS, et al. Cerebrospinal uid pressure monitoring after repair of cerebrospinal uid leaks. Otolaryngol Head Neck Surg 2004;130:443 8. 16. Woodworth BA, Prince A, Chiu AG, et al. Spontaneous CSF leaks: a paradigm for denitive repair and management of intracranial hypertension. Otolaryngol Head Neck Surg 2008;138:71520.

17. Hegazy HM, Carrau RL, Snyderman CH, et al. Transnasal endoscopic repair of cerebrospinal uid rhinorrhea: a meta-analysis. Laryngoscope 2000;110:1166 72. 18. Mirza S, Thaper A, McClelland L, et al. Sinonasal cerebrospinal uid leaks: management of 97 patients over 10 years. Laryngoscope 2005; 115:1774 7. 19. Zuckerman J, Stankiewicz JA, Chow JM. Long-term outcomes of endoscopic repair of cerebrospinal uid leaks and meningoencephaloceles. Am J Rhinol 2005;19:5827. 20. Bernal-Sprekelsen M, Alobid I, Mullol J, et al. Closure of cerebrospinal uid leaks prevents ascending bacterial meningitis. Rhinology 2005;43:277 81.