Effectiveness of lumbar drains on recurrence rates in endoscopic surgery of cerebrospinal fluid leaks Silviu Albu, M.D., Ph.D.,1 Enzo Emanuelli, M.D.,2 Veronica Trombitas, M.D.,1 and Ioan S. Florian, M.D., Ph.D.3
Y P
ABSTRACT
Background: In cerebrospinal fluid (CSF) leakage endoscopic repair, lumbar drains (LDs) were used in an attempt to increase success rates. To critically assess the relationship between use of LDs and recurrent leaks, we embarked on this randomized prospective study. Methods: Patients undergoing CSF leak repair between 2000 and 2012 were randomized into two groups: 75 patients were managed without LDs and in the other group of 75 patients an LD was always placed. Different parameters were analyzed to identify their relationship with failures: occurrence of increased intracranial pressure (ICP), body mass index, smoking habits, existence of diabetes, chronic corticosteroid use, previous sinus surgery, etiology (traumatic, iatrogenic, or spontaneous leak), site, and size of the CSF leak. Results: Success rate was 93% for the whole group. Patients managed with LDs attained 95% success rate and those without LD attained 92%; the difference is not significant (p ⫽ 0.2). The only factor predictive of recurrence is increased ICP: 77% success rate versus 97% for traumatic leaks and 96% for iatrogenic leaks. Recurrence rates were identical in the two groups with increased ICP, regardless of the use of a LD (23%). Conclusion: In this study, success rates of CSF repair were not associated with the use of LDs. However, the small number of cases with high success rates precludes appropriate statistical analysis. (Am J Rhinol Allergy 27, e190 –e194, 2013; doi: 10.2500/ajra.2013.27.3986)
E
ndoscopic surgery has become, over the last decades, the treatment of choice in the management of nasal cerebrospinal fluid (CSF) leaks.1–5 The endonasal endoscopic approach has minimal morbidity with excellent success rates, generally ⬎90% according to most published series.1–7 Among other adjunctive measures introduced to enhance success rates, the perioperative lumbar drain (LD) placement has been encouraged.7 Alleged advantages associated with the use of LDs are enhancement of graft apposition to the repaired skull base defect during the immediate postoperative period.7,8 Accelerated initial healing presumably will lessen the odds of recurrent fistula a few months later.9 Lessening pressure on the site of skull base repair is accomplished by diverting CSF. However, LD placement is connected with significant complications such as pneumocephalus, spinal headache, meningitis, or even cerebral herniation.10–13 Therefore, the indications for LD placement are poorly defined, and the choice regarding LD use is commonly left to surgeon preference.9 Despite frequent LD use, it is uncertain whether their application is associated with enhanced skull base defects closure rates.9,14 To critically assess the effect of LD placement on recurrent CSF leaks after surgical repair, we embarked on this randomized prospective study.
O D
MATERIALS AND METHODS This is a prospective, randomized study and it was approved both by the Institutional Review Board of the Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca and by the Institutional Review Board of the Azienda Ospedaliera-Universita` of Padua, Italy, in accordance with the Guidelines for Protection of Human Subjects. All of the patients, adults ⬎18 years of age, were provided with thorough explanations and signed an informed consent form before entering From the 1Second Department of Otolaryngology, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Romania, 2Department of Ear, Nose, and Throat and Otosurgery, University Hospital of Padua, Padua, Italy, and 3Department of Neurosurgery, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Romania The authors have no conflicts of interest to declare pertaining to this article Address correspondence to Silviu Albu, M.D., Department of Otolaryngology, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Str. Republicii nr. 18, 400015 Cluj-Napoca, Romania E-mail address: [email protected] Copyright © 2013, OceanSide Publications, Inc., U.S.A.
e190
the study. Included in the study were only those patients where the CSF leak was repaired by the use of an exclusively endoscopic approach and a follow-up period of at least 6 months. Excluded from the study were the following patients: skull base defects managed by either a purely open or a combined open/endoscopic approach and high-flow CSF leaks subsequent of open cisterns or ventricles in endoscopic enlarged tumor removal. All consecutive patients meeting the inclusion criteria operated on between 2000 and 2012 at the aforementioned tertiary referral centers were incorporated into the study. Preoperatively, a combination of nasal endoscopy, high-resolution computed tomography scan, and magnetic resonance imaging were used to assist in localizing skull base defects. Surgery was performed by two surgeons following the same principles described in the literature.1–9 Following previous published guidelines15 the induced intraoperative increase of the intracranial pressure (ICP) helped the surgeon in precise localization of the CSF leak. As a result, the intraoperative intrathecal fluorescein test was not performed during this study. Once leakage was highlighted, the surrounding mucosa around the bony defect was circumferentially removed. Bipolar cautery was used to reduce the herniated brain tissue and several layers of grafts were positioned. In large bony defects or increased ICP, cartilage or bone was inserted “underlay.” Free mucosal grafts from septum or inferior turbinate were used. The reconstruction was supported by a loose packing of gauze embedded in antibiotic ointment. During the course of the study we witnessed the advent of the vascular pedicled nasoseptal flap.16 Although this flap is the work horse for many skull base defects and has been shown to be very effective in repairs,16 we did not include cases managed with this technique in the present series to avoid potential biases. All patients, irrespective of the use of LDs, were maintained on prophylactic antibiotics with activity against sinus flora for a total of 6 days. Data concerning demographic details, body mass index (BMI), smoking habits, presence of diabetes, chronic corticoid use, previous sinus surgery, etiology, site and size of the CSF leak, surgical procedures, and length of follow-up were all recorded. Moreover, postoperative complications related to the use of LDs, recurrence of the CSF leak, and site of recurrence were also noted. In cases of spontaneous CSF leaks the ICP was measured: lumbar punctures were performed under fluoroscopic guidance with the patient in a prone position. After the tip of the needle entered the subarachnoid space, ICP was measured using a standardized ma-
T
O N
O C
November–December 2013, Vol. 27, No. 6
nometer. Identification of increased ICP was based on currently accepted criteria: documented elevated ICP ⬎ 25 cmH2O or ICP ⬎ 20 cmH2O if concurrent partial or total empty sella on magnetic resonance imaging.17–19 In an effort to avoid biases, in patients with elevated ICP, postoperative diuretics or ventriculoperitoneal (VP) shunts were not used after the first closure attempt. Patients included in the study were randomly allocated into two groups according to the use of LDs. In group A patients, an LD was placed before the induction of general anesthesia. In these patients, LDs were clamped throughout the whole surgical procedure.8 Group B included patients managed without an LD. All LDs were placed by the anesthesiologist. After defect closure, drainage of CSF was maintained at 5–10 mL/hr for the rest of the procedure and in the early postoperative period. The LD was kept open especially during patient extubation and transport.8 CSF removed was recorded hourly by the intensive care unit personnel. According to Lee et al.20 and Carrau et al.21 LDs were kept for 72 hours. Afterward, the catheter was examined to make certain complete removal and an occlusive bandage was positioned over the puncture site. Randomization was performed in blocks of two using random numbers generated by SAS function UNIFORM (SAS Institute, Cary, NC). Randomization was stratified according to the existence of increased ICP. Subjects with normal ICP were assigned randomization numbers in ascending sequential order using the lowest numbers available. Subjects with increased ICP were assigned randomization numbers in descending sequential order. The surgeons were informed about the assignment of an LD by the principal researcher 2 days before the surgery itself. In patients with increased ICP allocated to group B, LDs were removed 2 days before surgery. Sample size was calculated to detect differences of 20% in the probability of primary end point (the probability of successful CSF leak closure for group A was assumed to be 95%, whereas it was assumed to be 75% in the group B,) with a type I error of 5% and a statistic power of 80%. For each group, 60 subjects were required. To increase the power and compensate for the dropouts, the number was raised to 75 within each group. Statistical analysis was performed using SPSS for Windows, Version 10.0 (SPSS, Inc., Chicago, IL). For variables with frequency data only, differences between groups were determined by means of chisquare test. When chi-square results produced low (⬍5) expected frequencies, statistical significance was established by Fisher’s exact test. Group differences for continuous variables were assessed by analysis of variance (ANOVA test). Data were expressed as mean ⫾ SD. A value of p ⬍ 0.05 was considered significant.
O D
Table 1 Study group characteristics, etiology, location, and size of the CSF leak
Between 2000 and 2012, 150 patients fulfilled the inclusion criteria of the present analysis: 75 patients in group A and 75 cases in group B. The characteristics of the study group are shown in Table 1. The follow-up period ranged from 6 months to 10 years, with a mean of 4 years. Among iatrogenic CSF leaks, the most common cause was prior endoscopic endonasal surgery, occurring in 27 of the 51 patients. Other cases of iatrogenic CSF leaks include nasal septum surgery (6 cases) and CSF leaks after neurosurgical tumor resection. Pituitary tumors comprised the majority of tumor resections associated with CSF leakage (16 of 18). Two patients presented delayed CSF leaks after a transcranial frontal meningioma removal. Although vascularized nasoseptal flaps have been used in repair of leaks during pituitary surgery,16 we did not use them to avoid biases. Moreover, these flaps lengthen the operation and alter the nasal anatomy, thus making difficult repeated surgery.22 Concerning the defect location, the most common defect was in the sphenoid sinus (26%) and the ethmoid roof (27%). The baseline characteristics of the two treatment groups (age, gender, BMI, presence of diabetes, smoking habits, chronic corticoid use, prior sinus surgery, etiology, and location and size of the defect) are
American Journal of Rhinology & Allergy
n (%)
Age (yr), mean ⫾ SD Gender Males Females BMI (kg/m2), mean ⫾ SD Diabetes Smokers Chronic corticoid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess Size of CSF leak (mm), mean ⫾ SD
35.6 ⫾ 17.4 89 (59%) 61 (41%) 27.6 ⫾ 10.3 20 (13%) 67 (45%) 18 (12%) 10 (7%) 63 (42%) 51 (34%) 36 (24%) 10 26 22 (15%) 17 (11%) 41 (27%) 39 (26%) 31 (21%) 7.3 ⫾ 6.4
Y P
O C
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
T
O N
RESULTS
Patient Data (150 cases)
displayed in Table 2. As shown in the Table 2, randomization gave similar groups regarding demographics. A total of 10 patients presented with recurrent leak, giving a success rate of 93% for the whole cohort. Recurrences are presented in Table 3. From the 10 failures, 3 recurred at the same site and 7 recurred at a different location. The average time to recurrence was 14.9 months. Patient 1was involved in another head trauma after 1 month. The first two
Table 2 Baseline characteristics distribution between the two treatment groups Characteristic
Group A n ⴝ 75
Group B n ⴝ 75
p Value
Age (yr), mean (⫾SD) Gender, males/females BMI (kg/m2), Mean ⫾ SD Diabetes Smokers Chronic corticosteroid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess Size of CSF leak (mm), mean ⫾ SD
35.2 (⫾10.5) 43/28 26.5 ⫾ 9.5 8 37 13
33.8 (⫾11.6) 42/33 29.1 ⫾ 10.1 12 30 7
0.27 0.28 0.09 0.16 0.12 0.07
4 36 30
6 27 21
0.25
4 13 13 7 18 21 18 5.7 ⫾ 7.8
6 13 9 10 23 18 13 8.1 ⫾ 6.8
0.66
0.59
0.2
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
e191
leak closure (see Table 5). However, elevated BMI is related to increased ICP. The other analyzed factors did not reach statistical significance. No postoperative deaths occurred in this series. Postoperative meningitis was reported in three patients (2%): two in patients without LDs and one in a patient with LD. Among these three cases, two had a traumatic leak and one had an iatrogenic leak. In the postoperative period four patients developed acute uncomplicated sinusitis. Epistaxis was recorded in two patients. Headaches were recorded in 21 patients (14%), 17 (81%) of whom had an LD. Seizures in the postoperative period were reported in three cases. There also were two patients with deep vein thrombosis and two cases with pleural effusion. Fluid collection around the LD site occurred in two cases. No other LD-associated complications were recorded.
Table 3 Characteristics of patients presenting with recurrent CSF leaks Patient
Time to Recurrence (mo)
Etiology
Location
Identical with Primary Site
1 2 3 4 5 6 7 8 9 10
1 4 7 11 14 15 17 20 24 36
Trauma Increased ICP Increased ICP Trauma Iatrogenic Iatrogenic Increased ICP Increased ICP Increased ICP Increased ICP
Frontal sinus Lateral sphenoid Lateral sphenoid Sphenoid sinus Ethmoid roof Ethmoid roof Cribriform plate Cribriform plate Lateral sphenoid Lateral sphenoid
Yes Yes Yes No No No No No No No
CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure. Table 4 Predictors of success in surgical management of CSF leaks Parameter Group A Group B Gender Males Females Diabetes Smokers Chronic corticoid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess
Patients 150
Success Rate n (%)
p Value
75 75
71 (95%) 69 (92%)
0.2
89 61 20 67 18
85 (96%) 55 (90%) 17 (85%) 61 (91%) 16 (89%)
0.16
8 (80%) 61 (97%) 49 (96%)
0.26
10 (100%) 20 (77%) 21 (95%) 13 (76%) 39 (95%) 37 (95%) 30 (97%)
0.002
10 63 51 10 26 22 17 41 39 31
O D
0.13 0.24 0.34
CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
cases with lateral sphenoid leaks approached through a midline route recurred early and were repaired using the transpterygoid approach. The other patients with increased ICP recurred at a different site. Patient 5 with iatrogenic fistula developed signs of raised ICP and a recurrent fistula 14 months after surgery. Patient 6 with iatrogenic fistula presented 15 months later with recurrent fistula with undetectable cause. The results of the univariate analysis are presented in Table 4. In patients with LDs, success rate was 95%, and in the control group success rate was 92%; the difference lacks statistical significance. The only factor significantly associated with recurrent CSF leaks was the presence of elevated ICP: success rates were 77% in the elevated ICP group when compared with 97 and 96% for traumatic and iatrogenic leaks, respectively. Recurrence rates were identical in the two groups with increased ICP, regardless of the use of a LD: 23%, 3 of 13 patients. Although leaks located in the lateral recess of the sphenoid sinus are prone to recurrence when compared with other sites, the difference did not attain statistical significance (see Table 3). Increased BMI is also associated with increased failure rate of endoscopic CSF
e192
After skull base repair, in the early postoperative period LDs avoid the sudden ICP increase associated with cough or strain during extubation and vomiting.8,23 Moreover, LDs may decrease the incidence of postoperative meningitis.24 Theoretically, CSF drainage for a short period limits graft movements around the recipient site and promotes early healing, thus decreasing the likelihood of recurrence months later.9,14 However, Schlosser et al.8 cautioned that the tensile strength of the repair does not progress during the 1st postoperative week, given that fibrosis commences some weeks later. Thus, the greater part of the structural support is resultant of the underlay graft lying on the bony support and from the intranasal packing compressing the overlay graft. As an end result, LD indications and length of drainage are controversial issues and left to surgeon preference.9,14,21,23 There are surgeons recommending routine use of LDs,1 while others doubt their efficacy.7 A U.S. survey reported that 67% of surgeons use LDs routinely after CSF leakage repair.25 Nevertheless, several studies report high success rates without LD use.26,27 In their metaanalysis, Hegazy et al.7 found no advantage in using LDs. Recently, Caballero et al.14 found no benefit associated with LDs. However, the authors acknowledge their limits: retrospective review with its inherent selection bias, multiple surgeons having diverse skills, and using different techniques.14 In a recent retrospective study, using multivariate regression analysis, the authors found that LDs do not affect success rates.28 Our investigation is the first prospective, randomized study on the usefulness of CSF diversion as an adjunct in increasing success rates. In our cohort of 150 patients, success rates were similar in patients regardless of LD use. There are conflicting results regarding the relationship between defect size and outcome: some authors29 report suboptimal results in large defects, while others do not support such an association.30,31 However, in these retrospective studies the precise defect dimensions were not always reported. We found no association between skull base defect and success rate. We evaluated the relationship between outcomes and factors that might impair wound healing, such as history of prior sinus surgery, diabetes, chronic corticosteroid use, and smoking habits.31 Wise et al.31 found that previous sinus surgery predicts failure, because these patients may have increased skull base scarring, reduced vascular
T
O N 0.06
Y P
DISCUSSION
O C
Table 5 Predictors of success in surgical management of CSF leaks Variable
Success n ⴝ 135
Failure n ⴝ 15
p Value
Age (yr), mean ⫾ SD BMI (kg/m2), mean ⫾ SD
33.7 ⫾ 16.3 27.4 ⫾ 9.5
36.5 ⫾ 15.8 32.3 ⫾ 7.4
0.52 0.02
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid.
November–December 2013, Vol. 27, No. 6
supply, and excessive intraoperative bleeding. However, none of the aforementioned factors were associated with increased failure rate in our study, possibly related to the small number of patients. Failure to precisely localize the skull base defect predicts recurrence.1,2,31 Recognition of dural dehiscence can be demanding, because in traumatic and spontaneous CSF leaks, multiple bony defects may subsist. Intrathecal fluorescein is effective in identification of defects.1,2 Nevertheless, because its use implies an LD, we did not use fluorescein to avoid a potential bias. Previous studies have reported high failure rates (25–80%) in spontaneous leaks compared with ⬍10% for other etiology.1–3,31–36 Spontaneous leaks are usually seen in middle-aged, obese women and generally have elevated ICP.16,17,31–36 In our cohort of spontaneous leaks, 10 patients had normal ICP. However, normal ICP in spontaneous the group (as reported in Table 2) could be caused by the fact that ICPs were measured at the time of leak repair and that ICP measured in actively leaking patients may be artificially lower until the leak is repaired and homeostasis of CSF production is restored. Our results concur with earlier reports,29,33 showing a relationship between failure and elevated BMI, lateral sphenoid leaks, and raised ICP. It is claimed that long-term success in spontaneous leaks depend on the detection and treatment of increased ICP.1–3,17,18,33 In this group, the use of acetazolamide or VP shunts after repair is credited with a recurrence rate of 11–12.8%.17,36 Caballero et al.14 hypothesized that LD use may act as a substitute for a temporary VP shunt. However, recurrence rates were identical in spontaneous leaks regardless of LD use.14 To avoid biases, we avoided the use of acetazolamide or VP shunts. With LD acting as the only independent factor, we could not show any relationship between LD use and recurrence rates: 23% in both arms. Nevertheless, LDs are important, especially during the first 24 postoperative hours. Because LDs provide only short-term alleged protection against dehiscence and our failures occurred on average of 14.6 months after surgery, we concur with Caballero et al.14 to question the protective function of LD in the development of recurrent leaks in patients with increased ICP. In our series we report 10 failures, with 3 recurrences at the same site. A recurrent leak at a second site or presenting several months postoperatively is unlikely to be caused by technical issues at the original repair and is also unlikely to be improved with a perioperative LD for a few days. Thus, it is likely that only patient 1 and possibly patient 2 would have had different outcomes from use of LD. In contrast, delayed recurrent leaks or those occurring at a secondary site are likely because of failure to treat underlying elevations in ICPs. The fact that the failure rate in spontaneous leaks (23% in our series) is significantly higher than other reports in the literature17,34,36,37 strongly suggests that treatment of elevated ICPs with acetazolamide would improve these long-term failures. Severe complications associated with LDs have been described: pneumoencephalos, retained parts, and brain herniation.10–12 None of these occurred in our study. Postoperatively, the most frequent complaint was headache. The larger number of patients with headache had an LD (81%). We agree with other studies that LDs may be accountable for this complaint; however, the relationship is blurred by the confounding effect of the surgery itself.14 A serious complication was meningitis. However, the occurrence of meningitis may be better associated with the etiology of CSF leak and the surgery itself, rather than the use of LD. Ransom et al.9 suggested an algorithm for prospective LD use, in transplanum–transtuberculum approaches, given the high incidence of CSF leaks encountered. LDs are used also in patients with worsening clinical signs related to increased ICP unless a concurrent VP shunt is planned.9 However, our study is confronted with several limits: repairs were unevenly distributed over the course of the study, with more patients being operated on in the last period when compared with the beginning of the study. This would speak to whether cumulative evolving techniques and progressive surgeon experience would have altered
O D
the reported outcomes. Another flaw is represented by the small study population, particularly the high-risk group, those with increased ICP. Furthermore, the low failures rates preclude the achievement of statistical significance. The most important limit comes from the heterogeneous character of our group, because it is hard to compare spontaneous leaks, traumatic and iatrogenic. The neurosurgical cases should have been analyzed separately. However, to increase the power of the study and in conformity with other well-credited studies,35–38 we included them in the analysis. On the other hand, excluded were high-flow leaks after endoscopic endonasal tumor removal, including only inadvertent low-flow leaks during pituitary surgery and two delayed leaks after transcranial surgery. Future, prospective multicenter trials may help avoid these biases and help facilitate more powerful analysis.
O C
In this prospective, randomized study comprising patients with CSF leaks of different etiologies, increase of success rates was not accomplished with the use of LDs. However, the small number of cases with high success rates precludes appropriate statistical analysis. Therefore, future studies with increased number of cases are required.
REFERENCES 1.
Banks CA, Palmer JN, Chiu AG, et al. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. Otolaryngol Head Neck Surg 140:826–833, 2009. Harvey RJ, Parmar P, Sacks R, and Zanation AM. Endoscopic skull base reconstruction of large dural defects: A systematic review of published evidence. Laryngoscope 122:452–459, 2012. Lee DH, Lim SC, and Joo YE. Treatment outcomes of endoscopic repairs of sinonasal cerebrospinal fluid leaks. J Craniofac Surg 22: 1266–1270, 2011. Mattox DE, and Kennedy DW. Endoscopic management of cerebrospinal fluid leaks and cephaloceles. Laryngoscope 100:857–862, 1990. Schick B, Ibing R, Brors D, and Draf W. Long-term study of endonasal duraplasty and review of the literature. Ann Otol Rhinol Laryngol 110:142–147, 2001. Gassner HG, Ponikau JU, Sherris DA, and Kern EB. CSF rhinorrhea: 95 consecutive surgical cases with long term follow-up at the Mayo Clinic. Am J Rhinol 13:439–447, 1999. Hegazy HM, Carrau RL, Snyderman CH, et al. Transnasal endoscopic repair of cerebrospinal fluid rhinorrhea: A meta-analysis. Laryngoscope 110:1166–1172, 2000. Schlosser RJ, and Bolger WE. Nasal cerebrospinal fluid leaks: Critical review and surgical considerations. Laryngoscope 114:255–265, 2004. Ransom ER, Palmer JN, and Kennedy DW. Assessing risk/benefit of lumbar drain use for endoscopic skull-base surgery. Int Forum Allergy Rhinol 1:173–177, 2011. Anand VK, Murali RK, and Glasgold MJ. Surgical decisions in the management of cerebrospinal fluid rhinorrhoea. Rhinology 33:212– 218, 1995. Guido LJ, and Patterson RH Jr. Focal neurological deficits secondary to intraoperative CSF drainage: Successful resolution with an epidural blood patch. Report of two cases. J Neurosurg 45:348–351, 1976. Pepper JP, Lin EM, Sullivan SE, and Marentette LJ. Perioperative lumbar drain placement: An independent predictor of tension pneumocephalus and intracranial complications following anterior skull base surgery. Laryngoscope 121:468–473, 2011. Forsythe A, Gupta A, and Cohen SP. Retained intrathecal catheter fragment after spinal drain insertion. Reg Anesth Pain Med 34:375– 378, 2009. Caballero N, Bhalla V, Stankiewicz JA, and Welch KC. Effect of lumbar drain placement on recurrence of cerebrospinal rhinorrhea after endoscopic repair. Int Forum Allergy Rhinol 2:222–226, 2012. Wormald PJ, and McDonogh M. The bath-plug closure of anterior skull base cerebrospinal fluid leaks. Am J Rhinol 17:299–305, 2003. Hadad G, Bassagasteguy L, Carrau RL, et al. A novel reconstructive technique after endoscopic expanded endonasal approaches: Vascular pedicle nasoseptal flap. Laryngoscope 116:1882–1886, 2006.
T
O N
American Journal of Rhinology & Allergy
Y P
CONCLUSIONS
2.
3.
4. 5.
6.
7.
8. 9.
10.
11.
12.
13.
14.
15. 16.
e193
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
Seth R, Rajasekaran K III, Luong A, et al. Spontaneous CSF leaks: Factors predictive of additional interventions. Laryngoscope 120: 2141–2146, 2010. Schlosser RJ, Woodworth BA, Wilensky EM, et al. Spontaneous cerebrospinal fluid leaks: A variant of benign intracranial hypertension. Ann Otol Rhinol Laryngol 115:495–500, 2006. Reh DD, Gallia GL, Ramanathan M, et al. Perioperative continuous cerebrospinal fluid pressure monitoring in patients with spontaneous cerebrospinal fluid leaks: Presentation of a novel technique. Am J Rhinol Allergy 24:238–243, 2010. Lee TJ, Huang CC, Chuang CC, and Huang SF. Transnasal endoscopic repair of cerebrospinal fluid rhinorrhea and skull base defect: Ten-year experience. Laryngoscope 114:1475–1481, 2004. Carrau RL, Snyderman CH, and Kassam AB. The management of cerebrospinal fluid leaks in patients at risk for high-pressure hydrocephalus. Laryngoscope 115:205–212, 2005. Berker M, Aghayev K, Yu¨cel T, et al. Management of cerebrospinal fluid leak during endoscopic pituitary surgery. Auris Nasus Larynx 40:373–378, 2013. Ransom ER, and Chiu AG. Prevention and management of complications in intracranial endoscopic skull base surgery. Otolaryngol Clin North Am 43:875–895, 2010. van Aken MO, Feelders RA, de Marie S, et al. Cerebrospinal fluid leakage during transsphenoidal surgery: Postoperative external lumbar drainage reduces the risk of meningitis. Pituitary 7:89–93, 2004. Senior BA, Jafri K, and Benninger M. Safety and efficacy of endoscopic repair of CSF leaks and encephaloceles: A survey of the members of the American Rhinologic Society. Am J Rhinol 15:21–25, 2001. Casiano RR, and Jassir D. Endoscopic cerebrospinal fluid rhinorrhea repair: Is a lumbar drain necessary? Otolaryngol Head Neck Surg 121:745–750, 1999. Kirtane MV, Gautham K, and Upadhyaya SR. Endoscopic CSF rhinorrhea closure: Our experience in 267 cases. Otolaryngol Head Neck Surg 132:208–212, 2005.
O D e194
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Yeo NK, Cho GS, Kim CJ, et al. The effectiveness of lumbar drainage in the conservative and surgical treatment of traumatic cerebrospinal fluid rhinorrhea. Acta Otolaryngol 133:82–90, 2013. Lindstrom DR, Toohill RJ, Loehrl TA, and Smith TL. Management of cerebrospinal fluid rhinorrhea: The Medical College of Wisconsin experience. Laryngoscope 114:969–974, 2004. Zweig JL, Carrau RL, Celin SE, et al. Endoscopic repair of cerebrospinal fluid leaks to the sinonasal tract: Predictors of success. Otolaryngol Head Neck Surg 123:195–201, 2000. Wise SK, Harvey RJ, Neal JG, et al. Factors contributing to failure in endoscopic skull base defect repair. Am J Rhinol Allergy 23:185–191, 2009. Schlosser RJ, Wilensky EM, Grady MS, and Bolger WE. Elevated intracranial pressure in spontaneous cerebrospinal fluid leaks. Am J Rhinol 17:191–195, 2003. Schlosser RJ, Wilensky EM, Grady MS, et al. Cerebrospinal fluid pressure monitoring after repair of cerebrospinal fluid leaks. Otolaryngol Head Neck Surg 130:443–448, 2004. Schlosser RJ, and Bolger WE. Spontaneous nasal cerebrospinal fluid leaks and empty sella syndrome: A clinical association. Am J Rhinol 17:91–96, 2003. Mirza S, Thaper A, McClelland L, and Jones NS. Sinonasal cerebrospinal fluid leaks: Management of 97 patients over 10 years. Laryngoscope 115:1774–1777, 2005. Woodworth BA, Prince A, Chiu AG, et al. Spontaneous CSF leaks: A paradigm for definitive repair and management of intracranial hypertension. Otolaryngol Head Neck Surg 138:715–720, 2008. Psaltis AJ, Schlosser RJ, Banks CA, et al. A systematic review of the endoscopic repair of cerebrospinal fluid leaks. Otolaryngol Head Neck Surg 147:196–203, 2012. Banks CA, Palmer JN, Chiu AG, et al. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. Otolaryngol Head Neck Surg 140:826–833, 2009. e
T 38.
Y P
O C
O N
November–December 2013, Vol. 27, No. 6
Y P
ABSTRACT
Background: In cerebrospinal fluid (CSF) leakage endoscopic repair, lumbar drains (LDs) were used in an attempt to increase success rates. To critically assess the relationship between use of LDs and recurrent leaks, we embarked on this randomized prospective study. Methods: Patients undergoing CSF leak repair between 2000 and 2012 were randomized into two groups: 75 patients were managed without LDs and in the other group of 75 patients an LD was always placed. Different parameters were analyzed to identify their relationship with failures: occurrence of increased intracranial pressure (ICP), body mass index, smoking habits, existence of diabetes, chronic corticosteroid use, previous sinus surgery, etiology (traumatic, iatrogenic, or spontaneous leak), site, and size of the CSF leak. Results: Success rate was 93% for the whole group. Patients managed with LDs attained 95% success rate and those without LD attained 92%; the difference is not significant (p ⫽ 0.2). The only factor predictive of recurrence is increased ICP: 77% success rate versus 97% for traumatic leaks and 96% for iatrogenic leaks. Recurrence rates were identical in the two groups with increased ICP, regardless of the use of a LD (23%). Conclusion: In this study, success rates of CSF repair were not associated with the use of LDs. However, the small number of cases with high success rates precludes appropriate statistical analysis. (Am J Rhinol Allergy 27, e190 –e194, 2013; doi: 10.2500/ajra.2013.27.3986)
E
ndoscopic surgery has become, over the last decades, the treatment of choice in the management of nasal cerebrospinal fluid (CSF) leaks.1–5 The endonasal endoscopic approach has minimal morbidity with excellent success rates, generally ⬎90% according to most published series.1–7 Among other adjunctive measures introduced to enhance success rates, the perioperative lumbar drain (LD) placement has been encouraged.7 Alleged advantages associated with the use of LDs are enhancement of graft apposition to the repaired skull base defect during the immediate postoperative period.7,8 Accelerated initial healing presumably will lessen the odds of recurrent fistula a few months later.9 Lessening pressure on the site of skull base repair is accomplished by diverting CSF. However, LD placement is connected with significant complications such as pneumocephalus, spinal headache, meningitis, or even cerebral herniation.10–13 Therefore, the indications for LD placement are poorly defined, and the choice regarding LD use is commonly left to surgeon preference.9 Despite frequent LD use, it is uncertain whether their application is associated with enhanced skull base defects closure rates.9,14 To critically assess the effect of LD placement on recurrent CSF leaks after surgical repair, we embarked on this randomized prospective study.
O D
MATERIALS AND METHODS This is a prospective, randomized study and it was approved both by the Institutional Review Board of the Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca and by the Institutional Review Board of the Azienda Ospedaliera-Universita` of Padua, Italy, in accordance with the Guidelines for Protection of Human Subjects. All of the patients, adults ⬎18 years of age, were provided with thorough explanations and signed an informed consent form before entering From the 1Second Department of Otolaryngology, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Romania, 2Department of Ear, Nose, and Throat and Otosurgery, University Hospital of Padua, Padua, Italy, and 3Department of Neurosurgery, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Romania The authors have no conflicts of interest to declare pertaining to this article Address correspondence to Silviu Albu, M.D., Department of Otolaryngology, Iuliu Hatieganu University of Medicine and Pharmacy Cluj-Napoca, Str. Republicii nr. 18, 400015 Cluj-Napoca, Romania E-mail address: [email protected] Copyright © 2013, OceanSide Publications, Inc., U.S.A.
e190
the study. Included in the study were only those patients where the CSF leak was repaired by the use of an exclusively endoscopic approach and a follow-up period of at least 6 months. Excluded from the study were the following patients: skull base defects managed by either a purely open or a combined open/endoscopic approach and high-flow CSF leaks subsequent of open cisterns or ventricles in endoscopic enlarged tumor removal. All consecutive patients meeting the inclusion criteria operated on between 2000 and 2012 at the aforementioned tertiary referral centers were incorporated into the study. Preoperatively, a combination of nasal endoscopy, high-resolution computed tomography scan, and magnetic resonance imaging were used to assist in localizing skull base defects. Surgery was performed by two surgeons following the same principles described in the literature.1–9 Following previous published guidelines15 the induced intraoperative increase of the intracranial pressure (ICP) helped the surgeon in precise localization of the CSF leak. As a result, the intraoperative intrathecal fluorescein test was not performed during this study. Once leakage was highlighted, the surrounding mucosa around the bony defect was circumferentially removed. Bipolar cautery was used to reduce the herniated brain tissue and several layers of grafts were positioned. In large bony defects or increased ICP, cartilage or bone was inserted “underlay.” Free mucosal grafts from septum or inferior turbinate were used. The reconstruction was supported by a loose packing of gauze embedded in antibiotic ointment. During the course of the study we witnessed the advent of the vascular pedicled nasoseptal flap.16 Although this flap is the work horse for many skull base defects and has been shown to be very effective in repairs,16 we did not include cases managed with this technique in the present series to avoid potential biases. All patients, irrespective of the use of LDs, were maintained on prophylactic antibiotics with activity against sinus flora for a total of 6 days. Data concerning demographic details, body mass index (BMI), smoking habits, presence of diabetes, chronic corticoid use, previous sinus surgery, etiology, site and size of the CSF leak, surgical procedures, and length of follow-up were all recorded. Moreover, postoperative complications related to the use of LDs, recurrence of the CSF leak, and site of recurrence were also noted. In cases of spontaneous CSF leaks the ICP was measured: lumbar punctures were performed under fluoroscopic guidance with the patient in a prone position. After the tip of the needle entered the subarachnoid space, ICP was measured using a standardized ma-
T
O N
O C
November–December 2013, Vol. 27, No. 6
nometer. Identification of increased ICP was based on currently accepted criteria: documented elevated ICP ⬎ 25 cmH2O or ICP ⬎ 20 cmH2O if concurrent partial or total empty sella on magnetic resonance imaging.17–19 In an effort to avoid biases, in patients with elevated ICP, postoperative diuretics or ventriculoperitoneal (VP) shunts were not used after the first closure attempt. Patients included in the study were randomly allocated into two groups according to the use of LDs. In group A patients, an LD was placed before the induction of general anesthesia. In these patients, LDs were clamped throughout the whole surgical procedure.8 Group B included patients managed without an LD. All LDs were placed by the anesthesiologist. After defect closure, drainage of CSF was maintained at 5–10 mL/hr for the rest of the procedure and in the early postoperative period. The LD was kept open especially during patient extubation and transport.8 CSF removed was recorded hourly by the intensive care unit personnel. According to Lee et al.20 and Carrau et al.21 LDs were kept for 72 hours. Afterward, the catheter was examined to make certain complete removal and an occlusive bandage was positioned over the puncture site. Randomization was performed in blocks of two using random numbers generated by SAS function UNIFORM (SAS Institute, Cary, NC). Randomization was stratified according to the existence of increased ICP. Subjects with normal ICP were assigned randomization numbers in ascending sequential order using the lowest numbers available. Subjects with increased ICP were assigned randomization numbers in descending sequential order. The surgeons were informed about the assignment of an LD by the principal researcher 2 days before the surgery itself. In patients with increased ICP allocated to group B, LDs were removed 2 days before surgery. Sample size was calculated to detect differences of 20% in the probability of primary end point (the probability of successful CSF leak closure for group A was assumed to be 95%, whereas it was assumed to be 75% in the group B,) with a type I error of 5% and a statistic power of 80%. For each group, 60 subjects were required. To increase the power and compensate for the dropouts, the number was raised to 75 within each group. Statistical analysis was performed using SPSS for Windows, Version 10.0 (SPSS, Inc., Chicago, IL). For variables with frequency data only, differences between groups were determined by means of chisquare test. When chi-square results produced low (⬍5) expected frequencies, statistical significance was established by Fisher’s exact test. Group differences for continuous variables were assessed by analysis of variance (ANOVA test). Data were expressed as mean ⫾ SD. A value of p ⬍ 0.05 was considered significant.
O D
Table 1 Study group characteristics, etiology, location, and size of the CSF leak
Between 2000 and 2012, 150 patients fulfilled the inclusion criteria of the present analysis: 75 patients in group A and 75 cases in group B. The characteristics of the study group are shown in Table 1. The follow-up period ranged from 6 months to 10 years, with a mean of 4 years. Among iatrogenic CSF leaks, the most common cause was prior endoscopic endonasal surgery, occurring in 27 of the 51 patients. Other cases of iatrogenic CSF leaks include nasal septum surgery (6 cases) and CSF leaks after neurosurgical tumor resection. Pituitary tumors comprised the majority of tumor resections associated with CSF leakage (16 of 18). Two patients presented delayed CSF leaks after a transcranial frontal meningioma removal. Although vascularized nasoseptal flaps have been used in repair of leaks during pituitary surgery,16 we did not use them to avoid biases. Moreover, these flaps lengthen the operation and alter the nasal anatomy, thus making difficult repeated surgery.22 Concerning the defect location, the most common defect was in the sphenoid sinus (26%) and the ethmoid roof (27%). The baseline characteristics of the two treatment groups (age, gender, BMI, presence of diabetes, smoking habits, chronic corticoid use, prior sinus surgery, etiology, and location and size of the defect) are
American Journal of Rhinology & Allergy
n (%)
Age (yr), mean ⫾ SD Gender Males Females BMI (kg/m2), mean ⫾ SD Diabetes Smokers Chronic corticoid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess Size of CSF leak (mm), mean ⫾ SD
35.6 ⫾ 17.4 89 (59%) 61 (41%) 27.6 ⫾ 10.3 20 (13%) 67 (45%) 18 (12%) 10 (7%) 63 (42%) 51 (34%) 36 (24%) 10 26 22 (15%) 17 (11%) 41 (27%) 39 (26%) 31 (21%) 7.3 ⫾ 6.4
Y P
O C
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
T
O N
RESULTS
Patient Data (150 cases)
displayed in Table 2. As shown in the Table 2, randomization gave similar groups regarding demographics. A total of 10 patients presented with recurrent leak, giving a success rate of 93% for the whole cohort. Recurrences are presented in Table 3. From the 10 failures, 3 recurred at the same site and 7 recurred at a different location. The average time to recurrence was 14.9 months. Patient 1was involved in another head trauma after 1 month. The first two
Table 2 Baseline characteristics distribution between the two treatment groups Characteristic
Group A n ⴝ 75
Group B n ⴝ 75
p Value
Age (yr), mean (⫾SD) Gender, males/females BMI (kg/m2), Mean ⫾ SD Diabetes Smokers Chronic corticosteroid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess Size of CSF leak (mm), mean ⫾ SD
35.2 (⫾10.5) 43/28 26.5 ⫾ 9.5 8 37 13
33.8 (⫾11.6) 42/33 29.1 ⫾ 10.1 12 30 7
0.27 0.28 0.09 0.16 0.12 0.07
4 36 30
6 27 21
0.25
4 13 13 7 18 21 18 5.7 ⫾ 7.8
6 13 9 10 23 18 13 8.1 ⫾ 6.8
0.66
0.59
0.2
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
e191
leak closure (see Table 5). However, elevated BMI is related to increased ICP. The other analyzed factors did not reach statistical significance. No postoperative deaths occurred in this series. Postoperative meningitis was reported in three patients (2%): two in patients without LDs and one in a patient with LD. Among these three cases, two had a traumatic leak and one had an iatrogenic leak. In the postoperative period four patients developed acute uncomplicated sinusitis. Epistaxis was recorded in two patients. Headaches were recorded in 21 patients (14%), 17 (81%) of whom had an LD. Seizures in the postoperative period were reported in three cases. There also were two patients with deep vein thrombosis and two cases with pleural effusion. Fluid collection around the LD site occurred in two cases. No other LD-associated complications were recorded.
Table 3 Characteristics of patients presenting with recurrent CSF leaks Patient
Time to Recurrence (mo)
Etiology
Location
Identical with Primary Site
1 2 3 4 5 6 7 8 9 10
1 4 7 11 14 15 17 20 24 36
Trauma Increased ICP Increased ICP Trauma Iatrogenic Iatrogenic Increased ICP Increased ICP Increased ICP Increased ICP
Frontal sinus Lateral sphenoid Lateral sphenoid Sphenoid sinus Ethmoid roof Ethmoid roof Cribriform plate Cribriform plate Lateral sphenoid Lateral sphenoid
Yes Yes Yes No No No No No No No
CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure. Table 4 Predictors of success in surgical management of CSF leaks Parameter Group A Group B Gender Males Females Diabetes Smokers Chronic corticoid treatment Prior sinus surgery Traumatic Iatrogenic Spontaneous Normal ICP Increased ICP Sphenoid Lateral sphenoid Ethmoid roof Cribriform plate Frontal sinus/recess
Patients 150
Success Rate n (%)
p Value
75 75
71 (95%) 69 (92%)
0.2
89 61 20 67 18
85 (96%) 55 (90%) 17 (85%) 61 (91%) 16 (89%)
0.16
8 (80%) 61 (97%) 49 (96%)
0.26
10 (100%) 20 (77%) 21 (95%) 13 (76%) 39 (95%) 37 (95%) 30 (97%)
0.002
10 63 51 10 26 22 17 41 39 31
O D
0.13 0.24 0.34
CSF ⫽ cerebrospinal fluid; ICP ⫽ intracranial pressure.
cases with lateral sphenoid leaks approached through a midline route recurred early and were repaired using the transpterygoid approach. The other patients with increased ICP recurred at a different site. Patient 5 with iatrogenic fistula developed signs of raised ICP and a recurrent fistula 14 months after surgery. Patient 6 with iatrogenic fistula presented 15 months later with recurrent fistula with undetectable cause. The results of the univariate analysis are presented in Table 4. In patients with LDs, success rate was 95%, and in the control group success rate was 92%; the difference lacks statistical significance. The only factor significantly associated with recurrent CSF leaks was the presence of elevated ICP: success rates were 77% in the elevated ICP group when compared with 97 and 96% for traumatic and iatrogenic leaks, respectively. Recurrence rates were identical in the two groups with increased ICP, regardless of the use of a LD: 23%, 3 of 13 patients. Although leaks located in the lateral recess of the sphenoid sinus are prone to recurrence when compared with other sites, the difference did not attain statistical significance (see Table 3). Increased BMI is also associated with increased failure rate of endoscopic CSF
e192
After skull base repair, in the early postoperative period LDs avoid the sudden ICP increase associated with cough or strain during extubation and vomiting.8,23 Moreover, LDs may decrease the incidence of postoperative meningitis.24 Theoretically, CSF drainage for a short period limits graft movements around the recipient site and promotes early healing, thus decreasing the likelihood of recurrence months later.9,14 However, Schlosser et al.8 cautioned that the tensile strength of the repair does not progress during the 1st postoperative week, given that fibrosis commences some weeks later. Thus, the greater part of the structural support is resultant of the underlay graft lying on the bony support and from the intranasal packing compressing the overlay graft. As an end result, LD indications and length of drainage are controversial issues and left to surgeon preference.9,14,21,23 There are surgeons recommending routine use of LDs,1 while others doubt their efficacy.7 A U.S. survey reported that 67% of surgeons use LDs routinely after CSF leakage repair.25 Nevertheless, several studies report high success rates without LD use.26,27 In their metaanalysis, Hegazy et al.7 found no advantage in using LDs. Recently, Caballero et al.14 found no benefit associated with LDs. However, the authors acknowledge their limits: retrospective review with its inherent selection bias, multiple surgeons having diverse skills, and using different techniques.14 In a recent retrospective study, using multivariate regression analysis, the authors found that LDs do not affect success rates.28 Our investigation is the first prospective, randomized study on the usefulness of CSF diversion as an adjunct in increasing success rates. In our cohort of 150 patients, success rates were similar in patients regardless of LD use. There are conflicting results regarding the relationship between defect size and outcome: some authors29 report suboptimal results in large defects, while others do not support such an association.30,31 However, in these retrospective studies the precise defect dimensions were not always reported. We found no association between skull base defect and success rate. We evaluated the relationship between outcomes and factors that might impair wound healing, such as history of prior sinus surgery, diabetes, chronic corticosteroid use, and smoking habits.31 Wise et al.31 found that previous sinus surgery predicts failure, because these patients may have increased skull base scarring, reduced vascular
T
O N 0.06
Y P
DISCUSSION
O C
Table 5 Predictors of success in surgical management of CSF leaks Variable
Success n ⴝ 135
Failure n ⴝ 15
p Value
Age (yr), mean ⫾ SD BMI (kg/m2), mean ⫾ SD
33.7 ⫾ 16.3 27.4 ⫾ 9.5
36.5 ⫾ 15.8 32.3 ⫾ 7.4
0.52 0.02
BMI ⫽ body mass index; CSF ⫽ cerebrospinal fluid.
November–December 2013, Vol. 27, No. 6
supply, and excessive intraoperative bleeding. However, none of the aforementioned factors were associated with increased failure rate in our study, possibly related to the small number of patients. Failure to precisely localize the skull base defect predicts recurrence.1,2,31 Recognition of dural dehiscence can be demanding, because in traumatic and spontaneous CSF leaks, multiple bony defects may subsist. Intrathecal fluorescein is effective in identification of defects.1,2 Nevertheless, because its use implies an LD, we did not use fluorescein to avoid a potential bias. Previous studies have reported high failure rates (25–80%) in spontaneous leaks compared with ⬍10% for other etiology.1–3,31–36 Spontaneous leaks are usually seen in middle-aged, obese women and generally have elevated ICP.16,17,31–36 In our cohort of spontaneous leaks, 10 patients had normal ICP. However, normal ICP in spontaneous the group (as reported in Table 2) could be caused by the fact that ICPs were measured at the time of leak repair and that ICP measured in actively leaking patients may be artificially lower until the leak is repaired and homeostasis of CSF production is restored. Our results concur with earlier reports,29,33 showing a relationship between failure and elevated BMI, lateral sphenoid leaks, and raised ICP. It is claimed that long-term success in spontaneous leaks depend on the detection and treatment of increased ICP.1–3,17,18,33 In this group, the use of acetazolamide or VP shunts after repair is credited with a recurrence rate of 11–12.8%.17,36 Caballero et al.14 hypothesized that LD use may act as a substitute for a temporary VP shunt. However, recurrence rates were identical in spontaneous leaks regardless of LD use.14 To avoid biases, we avoided the use of acetazolamide or VP shunts. With LD acting as the only independent factor, we could not show any relationship between LD use and recurrence rates: 23% in both arms. Nevertheless, LDs are important, especially during the first 24 postoperative hours. Because LDs provide only short-term alleged protection against dehiscence and our failures occurred on average of 14.6 months after surgery, we concur with Caballero et al.14 to question the protective function of LD in the development of recurrent leaks in patients with increased ICP. In our series we report 10 failures, with 3 recurrences at the same site. A recurrent leak at a second site or presenting several months postoperatively is unlikely to be caused by technical issues at the original repair and is also unlikely to be improved with a perioperative LD for a few days. Thus, it is likely that only patient 1 and possibly patient 2 would have had different outcomes from use of LD. In contrast, delayed recurrent leaks or those occurring at a secondary site are likely because of failure to treat underlying elevations in ICPs. The fact that the failure rate in spontaneous leaks (23% in our series) is significantly higher than other reports in the literature17,34,36,37 strongly suggests that treatment of elevated ICPs with acetazolamide would improve these long-term failures. Severe complications associated with LDs have been described: pneumoencephalos, retained parts, and brain herniation.10–12 None of these occurred in our study. Postoperatively, the most frequent complaint was headache. The larger number of patients with headache had an LD (81%). We agree with other studies that LDs may be accountable for this complaint; however, the relationship is blurred by the confounding effect of the surgery itself.14 A serious complication was meningitis. However, the occurrence of meningitis may be better associated with the etiology of CSF leak and the surgery itself, rather than the use of LD. Ransom et al.9 suggested an algorithm for prospective LD use, in transplanum–transtuberculum approaches, given the high incidence of CSF leaks encountered. LDs are used also in patients with worsening clinical signs related to increased ICP unless a concurrent VP shunt is planned.9 However, our study is confronted with several limits: repairs were unevenly distributed over the course of the study, with more patients being operated on in the last period when compared with the beginning of the study. This would speak to whether cumulative evolving techniques and progressive surgeon experience would have altered
O D
the reported outcomes. Another flaw is represented by the small study population, particularly the high-risk group, those with increased ICP. Furthermore, the low failures rates preclude the achievement of statistical significance. The most important limit comes from the heterogeneous character of our group, because it is hard to compare spontaneous leaks, traumatic and iatrogenic. The neurosurgical cases should have been analyzed separately. However, to increase the power of the study and in conformity with other well-credited studies,35–38 we included them in the analysis. On the other hand, excluded were high-flow leaks after endoscopic endonasal tumor removal, including only inadvertent low-flow leaks during pituitary surgery and two delayed leaks after transcranial surgery. Future, prospective multicenter trials may help avoid these biases and help facilitate more powerful analysis.
O C
In this prospective, randomized study comprising patients with CSF leaks of different etiologies, increase of success rates was not accomplished with the use of LDs. However, the small number of cases with high success rates precludes appropriate statistical analysis. Therefore, future studies with increased number of cases are required.
REFERENCES 1.
Banks CA, Palmer JN, Chiu AG, et al. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. Otolaryngol Head Neck Surg 140:826–833, 2009. Harvey RJ, Parmar P, Sacks R, and Zanation AM. Endoscopic skull base reconstruction of large dural defects: A systematic review of published evidence. Laryngoscope 122:452–459, 2012. Lee DH, Lim SC, and Joo YE. Treatment outcomes of endoscopic repairs of sinonasal cerebrospinal fluid leaks. J Craniofac Surg 22: 1266–1270, 2011. Mattox DE, and Kennedy DW. Endoscopic management of cerebrospinal fluid leaks and cephaloceles. Laryngoscope 100:857–862, 1990. Schick B, Ibing R, Brors D, and Draf W. Long-term study of endonasal duraplasty and review of the literature. Ann Otol Rhinol Laryngol 110:142–147, 2001. Gassner HG, Ponikau JU, Sherris DA, and Kern EB. CSF rhinorrhea: 95 consecutive surgical cases with long term follow-up at the Mayo Clinic. Am J Rhinol 13:439–447, 1999. Hegazy HM, Carrau RL, Snyderman CH, et al. Transnasal endoscopic repair of cerebrospinal fluid rhinorrhea: A meta-analysis. Laryngoscope 110:1166–1172, 2000. Schlosser RJ, and Bolger WE. Nasal cerebrospinal fluid leaks: Critical review and surgical considerations. Laryngoscope 114:255–265, 2004. Ransom ER, Palmer JN, and Kennedy DW. Assessing risk/benefit of lumbar drain use for endoscopic skull-base surgery. Int Forum Allergy Rhinol 1:173–177, 2011. Anand VK, Murali RK, and Glasgold MJ. Surgical decisions in the management of cerebrospinal fluid rhinorrhoea. Rhinology 33:212– 218, 1995. Guido LJ, and Patterson RH Jr. Focal neurological deficits secondary to intraoperative CSF drainage: Successful resolution with an epidural blood patch. Report of two cases. J Neurosurg 45:348–351, 1976. Pepper JP, Lin EM, Sullivan SE, and Marentette LJ. Perioperative lumbar drain placement: An independent predictor of tension pneumocephalus and intracranial complications following anterior skull base surgery. Laryngoscope 121:468–473, 2011. Forsythe A, Gupta A, and Cohen SP. Retained intrathecal catheter fragment after spinal drain insertion. Reg Anesth Pain Med 34:375– 378, 2009. Caballero N, Bhalla V, Stankiewicz JA, and Welch KC. Effect of lumbar drain placement on recurrence of cerebrospinal rhinorrhea after endoscopic repair. Int Forum Allergy Rhinol 2:222–226, 2012. Wormald PJ, and McDonogh M. The bath-plug closure of anterior skull base cerebrospinal fluid leaks. Am J Rhinol 17:299–305, 2003. Hadad G, Bassagasteguy L, Carrau RL, et al. A novel reconstructive technique after endoscopic expanded endonasal approaches: Vascular pedicle nasoseptal flap. Laryngoscope 116:1882–1886, 2006.
T
O N
American Journal of Rhinology & Allergy
Y P
CONCLUSIONS
2.
3.
4. 5.
6.
7.
8. 9.
10.
11.
12.
13.
14.
15. 16.
e193
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
Seth R, Rajasekaran K III, Luong A, et al. Spontaneous CSF leaks: Factors predictive of additional interventions. Laryngoscope 120: 2141–2146, 2010. Schlosser RJ, Woodworth BA, Wilensky EM, et al. Spontaneous cerebrospinal fluid leaks: A variant of benign intracranial hypertension. Ann Otol Rhinol Laryngol 115:495–500, 2006. Reh DD, Gallia GL, Ramanathan M, et al. Perioperative continuous cerebrospinal fluid pressure monitoring in patients with spontaneous cerebrospinal fluid leaks: Presentation of a novel technique. Am J Rhinol Allergy 24:238–243, 2010. Lee TJ, Huang CC, Chuang CC, and Huang SF. Transnasal endoscopic repair of cerebrospinal fluid rhinorrhea and skull base defect: Ten-year experience. Laryngoscope 114:1475–1481, 2004. Carrau RL, Snyderman CH, and Kassam AB. The management of cerebrospinal fluid leaks in patients at risk for high-pressure hydrocephalus. Laryngoscope 115:205–212, 2005. Berker M, Aghayev K, Yu¨cel T, et al. Management of cerebrospinal fluid leak during endoscopic pituitary surgery. Auris Nasus Larynx 40:373–378, 2013. Ransom ER, and Chiu AG. Prevention and management of complications in intracranial endoscopic skull base surgery. Otolaryngol Clin North Am 43:875–895, 2010. van Aken MO, Feelders RA, de Marie S, et al. Cerebrospinal fluid leakage during transsphenoidal surgery: Postoperative external lumbar drainage reduces the risk of meningitis. Pituitary 7:89–93, 2004. Senior BA, Jafri K, and Benninger M. Safety and efficacy of endoscopic repair of CSF leaks and encephaloceles: A survey of the members of the American Rhinologic Society. Am J Rhinol 15:21–25, 2001. Casiano RR, and Jassir D. Endoscopic cerebrospinal fluid rhinorrhea repair: Is a lumbar drain necessary? Otolaryngol Head Neck Surg 121:745–750, 1999. Kirtane MV, Gautham K, and Upadhyaya SR. Endoscopic CSF rhinorrhea closure: Our experience in 267 cases. Otolaryngol Head Neck Surg 132:208–212, 2005.
O D e194
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Yeo NK, Cho GS, Kim CJ, et al. The effectiveness of lumbar drainage in the conservative and surgical treatment of traumatic cerebrospinal fluid rhinorrhea. Acta Otolaryngol 133:82–90, 2013. Lindstrom DR, Toohill RJ, Loehrl TA, and Smith TL. Management of cerebrospinal fluid rhinorrhea: The Medical College of Wisconsin experience. Laryngoscope 114:969–974, 2004. Zweig JL, Carrau RL, Celin SE, et al. Endoscopic repair of cerebrospinal fluid leaks to the sinonasal tract: Predictors of success. Otolaryngol Head Neck Surg 123:195–201, 2000. Wise SK, Harvey RJ, Neal JG, et al. Factors contributing to failure in endoscopic skull base defect repair. Am J Rhinol Allergy 23:185–191, 2009. Schlosser RJ, Wilensky EM, Grady MS, and Bolger WE. Elevated intracranial pressure in spontaneous cerebrospinal fluid leaks. Am J Rhinol 17:191–195, 2003. Schlosser RJ, Wilensky EM, Grady MS, et al. Cerebrospinal fluid pressure monitoring after repair of cerebrospinal fluid leaks. Otolaryngol Head Neck Surg 130:443–448, 2004. Schlosser RJ, and Bolger WE. Spontaneous nasal cerebrospinal fluid leaks and empty sella syndrome: A clinical association. Am J Rhinol 17:91–96, 2003. Mirza S, Thaper A, McClelland L, and Jones NS. Sinonasal cerebrospinal fluid leaks: Management of 97 patients over 10 years. Laryngoscope 115:1774–1777, 2005. Woodworth BA, Prince A, Chiu AG, et al. Spontaneous CSF leaks: A paradigm for definitive repair and management of intracranial hypertension. Otolaryngol Head Neck Surg 138:715–720, 2008. Psaltis AJ, Schlosser RJ, Banks CA, et al. A systematic review of the endoscopic repair of cerebrospinal fluid leaks. Otolaryngol Head Neck Surg 147:196–203, 2012. Banks CA, Palmer JN, Chiu AG, et al. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. Otolaryngol Head Neck Surg 140:826–833, 2009. e
T 38.
Y P
O C
O N
November–December 2013, Vol. 27, No. 6
