Intracranial complications before and after endoscopic skull base reconstruction Richard J. Harvey, M.D.,* Jacob E. Smith, M.D.,* Sarah K. Wise, M.D.,# Sunil J. Patel, M.D.,§ Bruce M. Frankel, M.D.,§ and Rodney J. Schlosser, M.D.* ABSTRACT Background: Endoscopic skull base reconstruction (ESBR) has been widely accepted in the management of cerebrospinal fluid (CSF) leaks. However, it is not the CSF leak itself but the potential for life-threatening intracranial complications (ICCs) that is of primary clinical concern. The risk of developing complications, such as meningitis, in a skull base defect is unknown. Many ESBR are multilayered soft tissue repairs, and long-term prevention of ICCs is not well described. Methods: Retrospective chart review and telephone consultation was used to assess patients who had an ESBR from 2002 to 2008. The incidence of an ICCs (meningitis, cerebral abscess, and pneumocephalus) and associated risk factors were assessed before and after surgery. Results: One hundred six patients underwent ESBR (mean age (⫾SD), 47.7 ⫾ 18.5 years; range, 2–78 years) with 95.3% long-term follow-up (mean, 19.9 ⫾ 16.3 months). ICCs occurred in 21.7% of patients at presentation, in 2.8% of patients during the perioperative period (⬍2 weeks), and in one patient (0.9%) during the postoperative period. Risk factors for presenting with an ICC and meningitis were revision cases performed elsewhere (2 ⫽ 9.10; p ⫽ 0.007) and leaking encephaloceles (2 ⫽ 5.98; p ⫽ 0.014). Factors not associated with increased ICC were an active CSF leak at presentation (2 ⫽ 3.03; p ⫽ 0.082) and previous radiotherapy. Conclusion: ESBR offers an excellent long-term option in preventing of subsequent ICC with low perioperative complications. ESBR is Property robust with delayed (⬎2weeks) CSF leakage occurring in only 1.9% regardless of etiology. The presence of identifiable risk factors for ICC may guide the surgeon in determining the urgency of ESBR. (Am J Rhinol 22, 516 –521, 2008; doi: 10.2500/ajr.2008.22.3223) Key words: Cerebrospinal fluid, CSF leak, endoscopic sinus surgery, meningitis, pneumocephalus, reconstruction, skull base
T
endoscopic surgeon. Problems related to closure of dura may be persistent sources of morbidity in both endoscopic and open SBS.7 Small defects and CSF fistulas have an excellent rate of closure via endoscopic techniques. Greater than 90% closure rates with primary endoscopic surgery, rising to 97% with revision endoscopic surgery, have been reported.8 Larger defects in the skull base resulting from resection of neoplasms, repair of encephaloceles, and intracranial tumor resection have been described also. These case series describe extensive bone removal and subsequent intradural surgery. Successful closure is attainable, even in large defects, alPublications though most series of larger skull base defects describe CSF leak rates slightly higher than for simple defects.7,9–12 A variIP: 58.172.107.75 ety of reconstructive techniques have been described.11,13–18 The use of multilayered free grafts or pedicled mucosal repair From the *Department of Otolaryngology–Head and Neck Surgery, Medical University has been previously popular among institutions with endoof South Carolina, Charleston, South Carolina, #Department of Otolaryngology–Head and Neck Surgery, The Emory Clinic, Atlanta, Georgia, and §Division of Neurosurscopic skull base experience.13,15,19,20 gery, Department of Neurosciences, Medical University of South Carolina, Charleston, The reconstruction of larger and larger defects poses a South Carolina concern for reliable long-term separation of the intracranial Presented at the spring meeting of the American Rhinologic Society, Orlando, Florida, cavity from the paranasal sinuses. Although endoscopic skull May 1, 2008 This research was not supported by an external funding source. Dr. Schlosser receives base reconstruction (ESBR) has been widely accepted in the grant support from the Flight Attendant Medical Research Institute and the Cystic management of CSF leaks, it is not the CSF leak itself but the Fibrosis Foundation. Dr. Schlosser is a consultant for BrainLAB, Medtronic Xomed, potential for life-threatening intracranial complications (ICCs) Gyrus, and Schering Plough and serves on the speaker’s bureau for Glaxo Smith Klein. that is of primary clinical concern. The risk of developing Dr. Harvey, has served on an advisory board for Schering Plough and has received grant support from NeilMed. Dr. Wise has received grant support from XoranTech, Inc. complications, such as meningitis, after SBS is unknown. EnDr. Smith, Dr. Patel, and Dr. Frankel have no financial interests to declare doscopic surgeons often replace a rigid complex of mucopeApproval for this study was granted by the Institutional Review Board of the Medical riosteum–bone–meninges with a multilayered soft tissue cloUniversity of South Carolina sure. The long-term prevention of recurrent CSF leakage in Address correspondence and reprint requests to Richard J. Harvey, M.D., Medical University of South Carolina, Department of Otolaryngology–Head and Neck Surgery, endoscopic skull base repair has been described in several 135 Rutledge Avenue, Suite 1130, P.O. Box 250550, Charleston, SC 29425 large series.21 However, the ability to prevent subsequent E-mail address:
[email protected] long-term ICCs, especially meningitis, has not been well reCopyright © 2008, OceanSide Publications, Inc., U.S.A. ported. This study aims to describe the long-term risk of
here have been substantial advances in endoscopic skull base surgery (SBS) in the past 10 years. Endoscopic approaches to the entire ventral skull base are described. The range of pathologies now managed endoscopically is no longer limited by the umbrella of “endoscopic” but by the disease process and surgical skills available to the treating team.1–6 Reliable and safe endoscopic closure of the dura is a milestone in the evolution of both otolaryngology and neurosurgery. Previously, the ability to reconstruct a skull base defect and prevent cerebrospinal fluid (CSF) leakage had been described as the “Achilles Heel” of endoscopic SBS.7 Along with managing cerebrovascular structures, repair of OceanSide the skull base is one of the most difficult challenges facing the
516
September–October 2008, Vol. 22, No. 5
subsequent ICCs in those patients who have undergone endoscopic skull base repair and to identify patient factors that were a risk for either perioperative or long-term ICCs.
RESULTS Patient Characteristics
One hundred six patients were assessed with an age range of 2–79 years of age (mean age (⫾SD), 47.7 ⫾ 18.5 years) with 49 women (46.2%). Long-term follow-up was available for Patient Population 101/106 (95.3%) patients with a mean of 19.9 ⫾ 16.3 months. Ethnicity of our patient population consisted of 60 white A retrospective chart review and prospective telephone (57%), 40 African American (38%), 5 Hispanic (4%) patients consultation was performed on patients who underwent enand 1 Filipino (1%) patient. doscopic repair of skull base defects at the Medical University Defect size varied from 2 mm to 5.5 cm (longest axis). of South Carolina from August 2002 to February 2008. IndiIndications for endoscopic skull base repair were defects seccations for surgery included spontaneous CSF leak, CSF leaks ondary to endoscopic or neurosurgical procedures (planned secondary to trauma, planned and unplanned defects from or unplanned; 66.0%), spontaneous CSF leak (17.0%), trauma SBS, congenital skull base defects, and encephaloceles. Patient (10.4%), and congenital (5.7%). Planned surgical defects indemographics obtained consisted of age, race, and sex. Insticluded the resection of a skull base tumor; types included tutional review board approval was obtained for this work. pituitary adenoma, meningioma, chordoma, olfactory neuroblastoma, adenoid cystic carcinoma, hemangiopericytoma, Operative Technique nasopharyngeal carcinoma, squamous cell carcinoma and inESBR used a multilayered technique for the majority of verted papilloma. The location of defects were as follows: 42 cases. Current technique is placement of Duragen (Integra, sella or extended sella (39.6%), 20 sphenoid sinus (other than Plainsboro, NJ) as an underlay graft either subdurally or in sella or planum;(8.9%), 19 ethmoid (17.9%), 15 olfactory cleft/ the epidural space. Next, a free or pedicled mucosal graft is Property of cribriform (14.2%), 7 frontal sinus/recess (6.6%), and 3 middle used to cover the defect, followed by application of fibrin cranial fossa (2.8%). Five patients (4.7%) had received prior glue, Gelfoam (Pfizer, New York, NY), and placement of irradiation to the head and neck but this factor was not found Merocel (Medtronic, Mystic, CT) or balloon nasal packs. Other to be associated with higher rates of ICCs in either the perigraft materials previously used in this population had conoperative period or long term (X2 ⫽ 1.04; p ⫽ 0.309; Fisher’s, sisted of septal mucosa, septal bone, septal cartilage, tempop ⫽ 0.296). ralis fascia, fat, AlloDerm (Lifecell, Branchburg, NJ), and ca-
METHODS
daveric fascia lata. All ESBRs were performed with a multiplanar CT Image Guidance System (IGS; BrainLab, Inc., Westchester, IL). Lumbar drains are not routinely placed unless the patient has benign intracranial hypertension, prior radiotherapy, or a posterior fossa defect.
Risk Factors for Presentation and Perioperative ICCs
There was an active CSF leak in 77 patients (73%) at presentation. Twenty-three patients (21.7%) presented with preoperative ICCs. Of the patients who initially presented with an ICC, 11(10.4%) had meningitis, 11(10.4%) had Outcomes pneumocephalus, 3 (2.8%) had cerebral or epidural abscess, ICCs were assessed as part of the patient’s initial presenand none had subdural hematoma. The presence of an tation, perioperative period (defined as ⬍2 weeks postsuractive CSF leak at presentation was not associated with gery), and long term. ICCs were defined as those patients ICCs as part of the presentation (X2 ⫽ 3.03; p ⫽ 0.082). having meningitis, cerebral or epidural abscess, subdural heRevision surgical cases had a strong association with ICCs OceanSide matoma, or pneumocephalus. Additional information re- Publications at presentation (X2 ⫽ 9.10; p ⫽ 0.003; Fisher’s, p ⫽ 0.007) corded included success of repair, active CSF leak at presenand there was no difference in ICCs in revision cases based IP: 58.172.107.75 tation, presence of neoplasm, presence of encephalocele, on the presence of an active CSF leak (X2 ⫽ 1.91; p ⫽ 0.168; location of defect, primary surgery versus revision surgery, Fisher’s, p ⫽ 0.354). The factors influencing meningitis and and prior radiation therapy. A CSF leak was considered “acICC at presentation are listed in Table 1. Figures 1 and 2 tive” if present during any part of the patient’s activities of show the influence of etiology and location on ICCs at daily living. Open dural resections were not considered active presentation. Only one ICC occurred in a nonleaking enleaks. Follow-up consisted of review of postoperative clinic cephalocele. This 66-year-old lady, who presented with notes as well as administering a standardized questionnaire meningitis, had three synchronous encephaloceles with an by telephone to ensure that no further surgeries or complicaactive area of purulent sinusitis adjacent to an ethmoid tions had arisen beyond our last formal follow-up. The teleencephalocele. In comparison, 41% (7/17) of leaking enphone questionnaire assessed the following factors: had they cephaloceles had an ICC (Figs. 3 and 4). had an active CSF leak before surgery and/or since surgery, After undergoing ESBR, three patients (2.8%) developed meningitis before or after surgery, any revision surgery since ICCs in the perioperative period. These cases consisted of repair, and any other complications. two cases of meningitis and one case of pneumocephalus. One case of meningitis occurred 5 days after endoscopic Statistical Analysis transsphenoidal resection of a pituitary mass. The second case of meningitis occurred after a large multistage resecStatistical assessments were performed by assessment of tion of a recurrent nasopharyngeal carcinoma previously nominal data and were analyzed using chi-square and Fishers irradiated. The case of pneumocephalus occurred 6 days exact tests via SPSS software (Statistical software for social sciences; SPSS, Inc. Chicago, IL). after craniotomy and resection of clival chordoma. Initial
American Journal of Rhinology
517
Table 1 The significance of various conditions on the risk for an ICC at presentation (based on X2 analysis) Risk Factor Revision surgery Location Etiology Active leak Encephalocele present Leaking Nonleaking Prior radiotherapy ICC ⫽ intracranial complication.
Meningitis p ⫽ 0.19 p ⫽ 0.03 p ⬍ 0.001 p ⫽ 0.45 p ⫽ 0.01 p ⫽ 0.01 p ⫽ 0.22 p ⫽ 0.42
Any ICC p ⫽ 0.003 p ⫽ 0.001 p ⬍ 0.001 p ⫽ 0.08 p ⫽ 0.06 p ⫽ 0.11 p ⫽ 0.44 p ⫽ 0.31
Property of
OceanSide Publications IP: 58.172.107.75
Figure 1. (A) The pattern of defect location to presentation of meningitis. Ethmoid and frontal sinuses have the greatest percentage of meningitis presentations. (B) The pattern of defect location to any intracranial complication at presentation. Nonsella sphenoid defects were greatest in this group.
CSF closure rate was 91.5% (97/106). For the failures, endoscopic revision was performed in three cases, transcraniotomy repair was performed in two cases, ventriculoperitoneal shunt was required for two cases, and there was unknown status for one case. One further patient has active leaking and refuses further treatment. She is without an ICC at 13 months.
518
Figure 2. (A) The pattern of defect etiology to presentation of meningitis. The prevalence of congenital defects may be biased by their self-selection for clinically important defects. (B) The pattern of defect etiology to any intranasal complication at presentation.
Long-Term Outcomes after ESBR There was one patient (0.9%) that developed a delayed postoperative ICC. A 68-year-old diabetic man with a history of open skull base resection for olfactory groove meningioma presented 1 month after his craniotomy with altered mental status and significant pneumocephalus without a clinically
September–October 2008, Vol. 22, No. 5
Figure 3. A large traumatic nonleaking encephalocele requiring surgery for total nasal obstruction is an example of a long standing nonleaking patient without any intranasal complications. Despite complete absence of bone, his thick soft tissue layer had provided adequate separation from the nasal cavity.
Property of
OceanSide Publications IP: 58.172.107.75
Figure 4. An example of a high-risk patient. A previous attempt, at another hospital, to biopsy a frontal lesion had resulted in a posterior frontal table defect. Arachnoid tissue can be seen protruding through the defect. Active cerebrospinal fluid leakage was present.
apparent CSF leak. Endoscopic evaluation revealed a defect in the pericranial flap in the area of the left anterior ethmoid and frontal recess, which was repaired without complication. The patient initially did well with resolution of pneumocephalus,
American Journal of Rhinology
although 3 months later he returned with seizures and a large intracranial abscess. His frontal bone flap was found to be infected and was removed during the craniotomy for abscess drainage. At that time, there was no CSF rhinorrhea present.
519
Figure 5. ESBR has been greatly enhanced by using pedicled septal flaps. A 5 cm⫹ defect resulting from a meningioma resection is repaired with an underlay graft (A) and then bilateral septal flaps (B).
He had multiple subsequent procedures and eventually rereported that the risk persisted as long as the CSF leak was covered. active. Delayed CSF leaks occurred in 1.9% of cases. An endoThere are few studies that have assessed the prevention of scopic revision of one patient with a releaking encephalocele meningitis or any other ICC after ESBR. Bernal-Sprekelsen et was performed at 17 months and an open transcraniotomy al. reported on 39 patients with repair of actively leaking small repair was performed for a traumatic encephalocele that defects and there was 0% rate of meningitis at 65-month mean failed at 24 months. The long-term closure rate was 89.6% follow-up.24 However, delayed meningitis has been reported. Property of et al. described a 0.3% (1/111) rate of developing men(95/106) based on a single reconstruction and was 93.4% Mirza (99/106) on endoscopic revision. ingitis after successful repair.22 Isolated deficient pneumococcal IgG production was thought to be a causal factor. The management of traumatic CSF leaks has yielded the most DISCUSSION significant outcomes. A group of traumatic CSF leak patients treated successfully, either surgically or conservatively, had a Long-Term Separation of the Sinonasal and meningitis rate of 29% with a delayed presentation averaging Intracranial Cavities 499 days. However, it is unclear in this study how many of The successful endoscopic closure of skull base defects has these patients had active CSF leaks at the time of their delayed been well documented with initial closure rates of 90% and meningitis; additionally, there was a high 37% recurrent leak 8 rising to 97% with endoscopic revision from a meta-analysis. rate in this study.25 Additional outcomes from another cohort This success has been reported subsequently in larger series reported a 1.5% rate of delayed “CNS infections” in nonleakwith a 90, 97, and 99% closure rate for endoscopic, revision, ing patients who had undergone open surgery for penetrating 22 and re-revision cases, respectively. Many of these early recraniocerebral trauma.26 Some of these infections occurred a ports did not include long-term follow-up data. The long-term long time after the initial injury. However, many of these closure rates for both leaks and meningoencephaloceles was patients (34%) had retained bone and metal fragments and recently reviewed by Zuckerman et al.21 They reported closure may not be representative of those patients undergoing enfor their group and cumulative multicenter results as a 73% doscopic SBS. OceanSide Publications success in the 5- to 10-year follow-up group and an 84% 21 success in the 10-year-plus group. The loss to follow-up in IP: 58.172.107.75 these data was significant with only 151/383 (39.4%) patients Prevention of Subsequent ICCs followed in the 5- to 10-year group and 19/64 (29.7%) in the The perioperative risk of ICCs from ESBR is very low. 10-year-plus group. This loss to follow-up will almost cerHegazy et al. reported postoperative complications of menintainly skew these results. gitis to be 0.3%, brain abscess to be 0.7%, subdural hematoma
The Risk of Meningitis Pre- and Postsurgery Many centers have reported long-term CSF leaking patients without episodes of meningitis or ICCs. Meningitis-free CSF rhinorrhea has been described from 4 days to 15 years.8 Reported rates of meningitis in actively leaking patients at presentation has ranged from 13.923 to 36.5%.24 Traumatic causes of CSF leakage appear to be associated with a higher incidence of meningitis22,25,26 and spontaneous CSF leak with the lowest.22 Daudia et al. retrospectively assessed the “meningitis risk” for a group of patients with CSF rhinorrhea and found a 19% rate of meningitis preoperatively with an incidence of 0.3 episodes/year. Over 50% of the episodes occurred in the first 12 months of the onset of CSF rhinorrhea. The group of 111 patients had a total of 190 leaking years between them.27 They
520
to be 0.3%, smell disorders to be 0.6%, and headache to be 0.3%.8 In addition, Sethi et al. describe only one patient (1/ 284) developing meningitis after ESBR.28 The long-term risk of ICCs is largely unknown. Delayed meningitis has been reported as 0.3% (1/111)27 and 0%,24 as noted previously. The later study reported small defects only but had an excellent 65-month mean follow-up.24 The present study describes a range of skull base defects during ESBR (2 mm to 5.5 cm) and a spectrum of pathologies from spontaneous CSF leaks to large transnasal craniotomies used for endoscopic skull base tumor resection. At our institution, significant advancements in CSF leak closure rates have been made for large defects through the use of pedicled mucosal flaps in ESBR (Fig. 5). Current CSF leak rate after large skull base resections is 3.2% with pedicled flaps (Uni-
September–October 2008, Vol. 22, No. 5
versity of South Carolina, unpublished data). However, longterm prevention of ICCs is the main outcome.
Leong J-L, Citardi MJ, and Batra PS. Reconstruction of skull base defects after minimally invasive endoscopic resection of anterior skull base neoplasms. Am J Rhinol 20:476–482, 2006. CONCLUSIONS 14. Kumar A, Maartens NF, and Kaye AH. Reconstruction of the Endoscopic skull base repair has become a reliable and sellar floor using Bioglue following transsphenoidal procerobust method for closure of the CSF space and prevention of dures. J Clin Neurosci 10:92–95, 2003. subsequent ICCs. Despite removal of bone and meningeal 15. Kassam A, Carrau RL, Snyderman CH, et al. Evolution of layers, a multilayered soft tissue repair provides for good reconstructive techniques following endoscopic expanded reconstruction. Perioperative ICCs are low (2.8%), delayed endonasal approaches. Neurosurg 19:E8, 2005. CSF leakage was 1.9%, and long-term ICCs was low (0.9%). In 16. Kaptain GJ, Vincent DA, and Laws ER, Jr. Cranial base the absence of adjacent infection, small nonleaking encephareconstruction after transsphenoidal surgery with bioabsorbable implants. Neurosurgery 48:232–233, 2001. loceles may not require immediate repair. 17. Esposito F, Dusick JR, Fatemi N, and Kelly DF. Graded repair of cranial base defects and cerebrospinal fluid leaks in REFERENCES transsphenoidal surgery. Neurosurgery 60(suppl 2):295–303, 1. Stammberger H, Anderhuber W, Walch C, and Papaefthy2007. miou G. Possibilities and limitations of endoscopic manage18. Arita K, Kurisu K, Tominaga A, et al. Size-adjustable titament of nasal and paranasal sinus malignancies. Acta Otonium plate for reconstruction of the sella turcica. Technical rhinolaryngol Belg 53:199–205, 1999. note. J Neurosurg 91:1055–1057, 1999. 2. Stamm AC. Transnasal endoscopy-assisted skull base sur19. Frank G, Pasquini E, Doglietto F, et al. The endoscopic gery. Ann Otol Rhinol Laryngol Suppl 196:45–53, 2006. extended transsphenoidal approach for craniopharyngio3. Snyderman CH, Kassam AB, Snyderman CH, and Kassam mas. Neurosurgery 59(suppl 1):ONS75–ONS83, 2006. AB. Endoscopic techniques for pathology of the anterior 20. de Divitiis E, Cavallo LM, Cappabianca P, et al. Extended cranial fossa and ventral skull base. J Am Coll Surg 202:563, 2006. Property ofendoscopic endonasal transsphenoidal approach for the removal of suprasellar tumors: Part 2. Neurosurgery 60:46–58, 4. Snyderman C, Kassam A, Carrau R, et al. Acquisition of 2007. surgical skills for endonasal skull base surgery: A training 21. Zuckerman J, Stankiewicz JA, and Chow JM. Long-term program. Laryngoscope 117:699–705, 2007. outcomes of endoscopic repair of cerebrospinal fluid leaks 5. Mehta RP, Cueva RA, Brown JD, et al. What’s new in skull and meningoencephaloceles. Am J Rhinol 19:582–587, 2005. base medicine and surgery? Skull Base Committee Report. 22. Mirza S, Thaper A, McClelland L, and Jones NS. Sinonasal Otolaryngol Head Neck Surg 135:620–630, 2006. cerebrospinal fluid leaks: Management of 97 patients over 10 6. Maroon JC. Skull base surgery: Past, present, and future years. Laryngoscope 115:1774–1777, 2005. trends. Neurosurgery 19:E1, 2005. 23. Lanza DC, O’Brien DA, and Kennedy DW. Endoscopic re7. Laws ER, Kanter AS, Jane JA, Jr, and Dumont AS. Extended pair of cerebrospinal fluid fistulae and encephaloceles. Latranssphenoidal approach. J Neurosurg 102:825–827, 2005. ryngoscope 106:1119–1125, 1996. 8. Hegazy HM, Carrau RL, Snyderman CH, et al. Transnasal 24. Bernal-Sprekelsen M, Alobid I, Mullol J, et al. Closure of endoscopic repair of cerebrospinal fluid rhinorrhea: A metacerebrospinal fluid leaks prevents ascending bacterial menanalysis. Laryngoscope 110:1166–1172, 2000. ingitis. Rhinology 43:277–281, 2005. 9. De Divitiis E, Cavallo LM, Cappabianca P, and Esposito F. 25. Bernal-Sprekelsen M, Bleda-Vazquez C, and Carrau RL. AsExtended endoscopic endonasal transsphenoidal approach cending meningitis secondary to traumatic cerebrospinal for the removal of suprasellar tumors: Part 2. Neurosurgery fluid leaks. Am J Rhinol 14:257–259, 2000. 60:46–58, 2007. 26. Gonul E, Baysefer A, Kahraman S, et al. Causes of infections 10. Frank G, Pasquini E, Doglietto F, et al. The endoscopic and management results in penetrating craniocerebral injuextended transsphenoidal approach for craniopharyngioOceanSide Publications ries. Neurosurg Rev 20:177–181, 1997. mas. Neurosurgery 59(suppl 1):ONS-75–ONS-82, 2006. 27. Daudia A, Biswas D, and Jones NS. Risk of meningitis with 11. Dusick JR, Esposito F, Kelly DF, et al. The extended direct IP: 58.172.107.75 cerebrospinal fluid rhinorrhea. Ann Otol Rhinol Laryngol endonasal transsphenoidal approach for nonadenomatous 116:902–905, 2007. suprasellar tumors. J Neurosurg 102:832–841, 2005. 28. Sethi DS, Chan C, and Pillay PK. Endoscopic management of 12. Maira G, Anile C, Albanese A, et al. The role of transsphecerebrospinal fluid fistulae and traumatic cephalocoele. Ann noidal surgery in the treatment of craniopharyngiomas. J Neurosurg 100:445–451, 2004. Acad Med Singap 25:724–727, 1996. e
American Journal of Rhinology
13.
521
