Repair of Middle Fossa CSF Using Novel Materials
Middle fossa CSF leaks are uncommon pathological entities that can be classified as acquired, congenital, or spontaneous. Acquired MFCSF leaks most commonly result from destructive pathology such as tumors, infection, trauma, and surgical procedures, the latter two being the most common. Congenital lesions such as patent Hyrtl fissure, Mondini dysplasia, and petromastoid canal fistulas occur much less frequently and typically occur in younger patients. At older ages (usually > 50 years) chronically enlarged arachnoid villi can cause pulsatile erosion of pneumatized segments of the temporal bone, leading to transmission of CSF through the defect. If there is no associated cause of a CSF leak, it is classified as spontaneous.
Cerebrospinal fluid leakage from the middle fossa can be challenging to diagnose and manage. Profuse fluid leakage from the ear canal allows for relatively straightforward diagnosis. Likewise, leaks following trauma or surgery may be anticipated, or even expected, allowing rapid establishment of the diagnosis and intervention. Conversely, if the MFCSF leak is intermittent, subtle, or spontaneous in nature, it requires a higher index of suspicion to diagnose. Persistent middle ear fullness, with or without sensorineural, conductive, or mixed hearing loss; progressive drainage of clear fluid after myringotomy; and recurrent meningitis can all be signs of an MFCSF leak. Also, fluid in the mastoid air cells can be indicative of an MFCSF leak resulting from a tegmen mastoideum defect. Persistent drainage of fluid following myringotomy can be confused with postprocedural infection. However, if drainage does not resolve with adequate antibiotic treatment, a CSF leak should be suspected. Beta-2 transferrin is a confirmatory test that is highly sensitive and specific for CSF identification, as transferrin is only converted into the beta-2 isoform within the nervous system. Historical use of glucose strips has been subject to a high false-positive rate of 45%–75% due to mucus or blood contamination.
Prompt diagnosis and treatment require a basic understanding of the structural anatomy and pathology of the middle fossa. A defect in the middle cranial fossa bone and breach of the overlying temporal dura is required to develop an MFCSF leak. Politzer first referred to the "wanting" composition of the tegmen tympani of the temporal bone in 1896. Ahrén and Thulin later studied the temporal bones in 96 cadavers and found approximately 20% to have some form of tegmen defect, a finding confirmed by Kapur and Bangash in 1986. Obviously, it is important to note that isolated tegmen defects, without a dural defect, do not lead to MFCSF leaks.
Several theories have been proposed regarding the pathogenesis of MFCSF leaks, especially those seen in the absence of trauma or congenital defects. Ommaya has proposed a theory that leaks result from erosion of the tegmen due to chronic variations in intracranial pressure. Others have implicated the enlargement of aberrant arachnoid granulations located over thin or pneumatized bone as a cause of adult-onset CSF otorrhea.
In the absence of a predicating event or an increased level of suspicion, MFCSF leaks often have a significant delay in diagnosis or are misdiagnosed altogether. Persistent MFCSF leaks can lead to considerable morbidities, including hearing loss, pneumocephalus, temporal lobe seizures, cerebral abscess, and meningitis. Secondary meningitis is well documented, with an incidence in the literature ranging from 4% to 50%. Thirty percent to 60% of postoperative meningitis and 10%–27% of post–traumatic brain injury meningitis can be directly attributed to CSF leaks. Those leaks originating from temporal bone pathological entities that persist more than 7 days have been reported to have a significantly increased risk of developing meningitis, compared with leaks repaired within 1 week.
Evidence supporting the use of prophylactic antibiotics in the prevention of meningitis has been conflicting at best. Findings from an early double-blinded study by Klastersky et al. failed to support the routine usage of prophylactic antibiotics in traumatic CSF leaks. However, the findings also failed to establish a detrimental effect. Similarly, in a recent meta-analysis Villalobos and colleagues found that prophylactic antibiotics did not decrease the risk of developing meningitis with basilar skull fractures. Conversely, Friedman et al. found that prophylactic antibiotics decreased the risk of developing meningitis by half, a finding further supported by the meta-analysis published by Brodie et al. Given the equivocal effect of prophylactic antibiotics, early surgical intervention has been strongly recommended. Repair of the temporal bone and dural defects and elimination of the CSF fistula are the ultimate goals.
Neuroimaging is critical in the evaluation of MFCSF leaks. High-resolution, thin-cut CT scans obtained through the suspected temporal bone can be extremely valuable in localizing a bony defect and thus allowing for accurate surgical planning. Computed tomography cisternography, with or without injection of radioactive isotopes, may be an option if the bony defect is not evident on CT scans; however, meaningful results are limited to those obtained in patients with active CSF fistulas, and the modality has been shown to have a significant false-negative rate with low-volume or intermittent leaks. The decision to undertake cisternography should also be made with consideration of the increased morbidity associated with lumbar puncture and contrast reactions. Computed tomography cisternograms were acquired in 5 of the 7 patients in the present study, and in all cases the studies were deemed necessary when initial CT scans did not definitely define the full extent of the temporal bone defect. All defects were clearly delineated on the CT cisternogram. Magnetic resonance imaging has proven useful in assessing the presence of dural defects and concomitant encephaloceles. Both T1- and T2-weighted MRI sequences are useful in evaluating the presence and content of herniated cerebral tissue, but their usefulness in delineating the bony defects has been shown to be limited. In a retrospective review of 8 cases of MFCSF leakage, Pappas et al. did not use MRI to identify and correct the site of the fistula. Likewise, Lundy et al. used MRI in 2 of 19 consecutive cases and found the information acquired to be equivocal. In our patients, MRI scans were obtained in 4 of the 7 patients we treated. Scans correlated with the intraoperative presence or absence of an encephalocele in 3 of 4 patients. The fourth had an encephalocele identified intraoperatively that was not visualized on MRI. Magnetic resonance imaging did not alter our surgical approach or repair method in any case.
Conservative treatment, such as avoiding Valsalva maneuvers and straining, head of bed elevated to 30°, or placement of a lumbar drain, was not performed at our institution for MFCSF leak management. Savva et al. reviewed the cases of 92 patients with CSF leaks through the temporal bone. They found that in 82 patients the leak was caused by trauma—head injury in 29 and surgical procedure in 53. Conservative measure worked in 26 of 29 patients in whom CSF leaks were caused by a head injury. The remaining 3 patients required surgical intervention. Conversely, conservative measures worked in only 1 of the 53 patients in whom the leak resulted from surgical procedures. Of note, all patients with nontraumatic leaks in the Savva et al. series required surgical intervention.
In our series the 5 patients with acquired MFCSF leaks had previous surgical procedures. The other 2 patient had MFCSF leaks that were classified as spontaneous. Therefore, we chose not to attempt conservative treatments before surgical repair.
Following confirmation of an MFCSF leak, we advocate prompt surgical repair of the bony and dural defects to prevent associated morbidities. In the literature, the approaches and materials used to repair MFCSF leaks vary widely.
The middle fossa, transmastoid, and combined middle fossa/transmastoid approaches are the procedures most commonly described in the literature. Individual selection, however, is highly dependent not only on the advantages and disadvantages of each approach with regard to the specific location of the fistula but also the surgeon's personal experience and comfort level with the approach in question (Table 3).
The transmastoid approach has been reported to be the least technically demanding route. It also confers the additional advantages of eliminating brain retraction and providing access to fistulas originating in the posterior fossa. The disadvantages include decreased exposure of anterior tegmen defects, risk of hearing loss, higher incidence of graft failure with increased recurrence of CSF leaks, and difficulty in approaching large or multiple defects.
The middle fossa craniotomy provides a wide exposure of the entire tegmental plate, allowing identification and repair of large or multiple defects. Repair of tegmental defects can be performed without risk to the ossicular chain, thus preserving hearing. Increased exposure allows more accurate and secure placement of sealant materials, thereby decreasing the possibility of recurrence. However, many authors find the middle fossa craniotomy more technically demanding, and its disadvantages include an inability to approach defects of the posterior fossa. In addition, multiple critical structures are potentially encountered during this approach. Mobilization of the temporal lobe and associated vascular structures such as the vein of Labbé can lead to postoperative seizures or venous infarction. Extradural dissection along the floor of the middle fossa also risks damage to the geniculate ganglion and GSPN.
We prefer the middle fossa approach for MFCSF leak repairs. Preoperative collaboration between the neurotologist and neurosurgeon is essential to ensure optimal surgical outcome. We advocate 3 steps to facilitate extradural dissection and protection of vital neurovascular structures. First, proper head positioning is critical; the head must be rotated 90° toward the opposite shoulder and extended so the temporal squama is parallel to the floor. This position allows gravity to aid in temporal lobe dissection. Second, lumbar drainage should be started only after the craniotomy is performed. Drainage of CSF prompts cerebral relaxation and allows for increased mobilization of temporal lobe dura from the middle fossa floor. Third, facial nerve monitoring is undertaken with the probe used to dissect the temporal dura from the floor of the middle fossa. This technique allows simultaneous dissection and stimulation of the GSPN and geniculate ganglion, conferring additional protection. We perform dural dissection in the traditional back-to-front manner to decrease the incidence of GSPN injury, although some authors have advocated safe dissection from front to back.
Reconstruction in all cases involved a minimal combination of HAC, collagen-based dural substitute matrix, and PEG hydrogel sealant. In 4 cases we used additional native materials, including split-thickness bone graft, temporalis muscle, and fascia. Intraoperative inspection of the defect's size and dimensions determined whether a split-thickness bone graft would be included.
Hydroxyapatite cement has been successfully used for the remodeling and repair of temporal bone defects by several authors, with Kveton et al. reporting a success rate greater than 97%. The cement is composed of tetracalcium and dicalcium phosphate salts reacting to form hydroxyapatite, the major component of skeletal bone. There are numerous physiological properties that make HAC an ideal material for temporal bone repair. Its high biocompatibility limits postoperative inflammatory response and fibrosis, while its osteoconductive and osteointegrative properties allow for formation and intercalation of new bone within implants. Finally, the product remains malleable for approximately 20 minutes, allowing for easy remodeling into bony defects, and afterward it is resistant to CSF pulsation. Complications have been reported to range from 0% to 11%, with postoperative infections occurring in up to 5% of patients when cases of nasal sinus cavities are included. Kveton and Coelho reported infections in 2.8% of their 109 consecutive temporal bone repairs. Conductive hearing loss due to ossicular chain fixation has been reported up to 33%, with the majority of deficits occurring after transmastoid repairs. Thus, the application of HAC in patients with exposed ossicles must be carefully done. None of the patients in our series had an exposed ossicular chain, but our recommendation for such cases would be to fashion a split-thickness bone graft large enough to cover the defect with or without careful placement of HAC.
In our series, temporal dura repairs involved a combination of dural substitute and PEG hydrogel sealant, with or without the addition of temporalis muscle and fascia. Whenever possible, our initial preference is primary repair with sutures; however, for defects that cannot be approximated, we prefer synthetic, autologous, or combination inlay grafts, supplemented by onlay grafts (inlay/onlay technique). In cases in which the dural defect cannot be repaired primarily because the dura is often thin and frayed and the edges cannot be cleanly reapproximated—because the location of the defect is too difficult to reach, or because the defect is too small to allow insertion of an inlay graft—a stand-alone onlay graft is placed. All repairs are subsequently augmented with PEG hydrogel sealant.
The safety and efficacy of dural reconstruction with synthetic collagen-based dural substitutes are well described. The structure of the synthetic collagen-based dural substitute presents a low-pressure surface for CSF absorption, promoting a graft–dura interface and chemical signaling for native fibroblasts providing mechanical scaffolding. Litvack et al. reported a CSF leakage rate of 6.7%, as well as a 4.2% postoperative infection rate, when using collagen-based substitutes in 425 consecutive patients.
In the literature the PEG hydrogel sealant is well described as an adjunct for the prevention of CSF leaks. Its strong adherence to tissue, biological compatibility, and absorbability make it an ideal choice in MFCSF repair. A study by Boogaarts and associates reported an intraoperative watertight seal rate of 100% and a CSF leak recurrence rate of 4.9% at 3 months in 41 patients who underwent various intracranial procedures. Later, in a large multicenter study, Cosgrove et al. reported a 1.8% postoperative CSF leak rate with PEG hydrogel augmentation of primary suture closure in 111 intracranial procedures; the incidence of infection was 7.2%.
We believe that the decreased CSF pulsation and pressure along the extent of the dural repair facilitates more effective healing, despite reports to the contrary. To that end, all patients in the present series underwent lumbar drainage of 10 ml/hour for 3–5 days and remained in intensive care for the duration. Taking into consideration the intraoperative benefits of CSF diversion, we recommend the routine use of lumbar drainage in all patients undergoing repair of an MFCSF leak. Following removal of the lumbar drain, patients are monitored for an additional 1–2 days for any signs of wound leakage, otorrhea, or rhinorrhea.
Common complications seen in MFCSF leak repairs include recurrence of CSF leaks, graft failure, and wound infections. Savva et al. reported a 14.3% recurrence rate following optimal repair in 92 consecutive patients. Graft failure due to foreign body reaction and migration has been reported, whereas infection has been seen in up to 5% of MFCSF leak repairs. It is difficult to accurately assess the overall complication rate of MFCSF leak repairs due to the variability of techniques and repair materials presented in the current literature.
The repair rate in our small series was 100%, which compares favorably with larger middle fossa craniotomy series. Stenzel et al. reported a repair success rate of 91% in a series of 11 patients in whom the reconstruction was conducted exclusively via the middle fossa craniotomy. One patient in their series required reoperation for a recurrent leak. However, the exact materials used for repairs were not described. In their series of 15 patients, Gubbels et al. described the use of bone grafts, fat grafts, HAC, and fascia. All repairs were done via a middle fossa craniotomy, and the repair success rate was 93%. One patient required an additional surgical procedure for a recurrent CSF leak. Gacek et al. reported a repair rate of 93% in 16 patients with 1 patient requiring an additional surgery for a recurrent leak. The repair materials were not specifically described in their series.
Discussion
Middle fossa CSF leaks are uncommon pathological entities that can be classified as acquired, congenital, or spontaneous. Acquired MFCSF leaks most commonly result from destructive pathology such as tumors, infection, trauma, and surgical procedures, the latter two being the most common. Congenital lesions such as patent Hyrtl fissure, Mondini dysplasia, and petromastoid canal fistulas occur much less frequently and typically occur in younger patients. At older ages (usually > 50 years) chronically enlarged arachnoid villi can cause pulsatile erosion of pneumatized segments of the temporal bone, leading to transmission of CSF through the defect. If there is no associated cause of a CSF leak, it is classified as spontaneous.
Cerebrospinal fluid leakage from the middle fossa can be challenging to diagnose and manage. Profuse fluid leakage from the ear canal allows for relatively straightforward diagnosis. Likewise, leaks following trauma or surgery may be anticipated, or even expected, allowing rapid establishment of the diagnosis and intervention. Conversely, if the MFCSF leak is intermittent, subtle, or spontaneous in nature, it requires a higher index of suspicion to diagnose. Persistent middle ear fullness, with or without sensorineural, conductive, or mixed hearing loss; progressive drainage of clear fluid after myringotomy; and recurrent meningitis can all be signs of an MFCSF leak. Also, fluid in the mastoid air cells can be indicative of an MFCSF leak resulting from a tegmen mastoideum defect. Persistent drainage of fluid following myringotomy can be confused with postprocedural infection. However, if drainage does not resolve with adequate antibiotic treatment, a CSF leak should be suspected. Beta-2 transferrin is a confirmatory test that is highly sensitive and specific for CSF identification, as transferrin is only converted into the beta-2 isoform within the nervous system. Historical use of glucose strips has been subject to a high false-positive rate of 45%–75% due to mucus or blood contamination.
Prompt diagnosis and treatment require a basic understanding of the structural anatomy and pathology of the middle fossa. A defect in the middle cranial fossa bone and breach of the overlying temporal dura is required to develop an MFCSF leak. Politzer first referred to the "wanting" composition of the tegmen tympani of the temporal bone in 1896. Ahrén and Thulin later studied the temporal bones in 96 cadavers and found approximately 20% to have some form of tegmen defect, a finding confirmed by Kapur and Bangash in 1986. Obviously, it is important to note that isolated tegmen defects, without a dural defect, do not lead to MFCSF leaks.
Several theories have been proposed regarding the pathogenesis of MFCSF leaks, especially those seen in the absence of trauma or congenital defects. Ommaya has proposed a theory that leaks result from erosion of the tegmen due to chronic variations in intracranial pressure. Others have implicated the enlargement of aberrant arachnoid granulations located over thin or pneumatized bone as a cause of adult-onset CSF otorrhea.
In the absence of a predicating event or an increased level of suspicion, MFCSF leaks often have a significant delay in diagnosis or are misdiagnosed altogether. Persistent MFCSF leaks can lead to considerable morbidities, including hearing loss, pneumocephalus, temporal lobe seizures, cerebral abscess, and meningitis. Secondary meningitis is well documented, with an incidence in the literature ranging from 4% to 50%. Thirty percent to 60% of postoperative meningitis and 10%–27% of post–traumatic brain injury meningitis can be directly attributed to CSF leaks. Those leaks originating from temporal bone pathological entities that persist more than 7 days have been reported to have a significantly increased risk of developing meningitis, compared with leaks repaired within 1 week.
Evidence supporting the use of prophylactic antibiotics in the prevention of meningitis has been conflicting at best. Findings from an early double-blinded study by Klastersky et al. failed to support the routine usage of prophylactic antibiotics in traumatic CSF leaks. However, the findings also failed to establish a detrimental effect. Similarly, in a recent meta-analysis Villalobos and colleagues found that prophylactic antibiotics did not decrease the risk of developing meningitis with basilar skull fractures. Conversely, Friedman et al. found that prophylactic antibiotics decreased the risk of developing meningitis by half, a finding further supported by the meta-analysis published by Brodie et al. Given the equivocal effect of prophylactic antibiotics, early surgical intervention has been strongly recommended. Repair of the temporal bone and dural defects and elimination of the CSF fistula are the ultimate goals.
Neuroimaging is critical in the evaluation of MFCSF leaks. High-resolution, thin-cut CT scans obtained through the suspected temporal bone can be extremely valuable in localizing a bony defect and thus allowing for accurate surgical planning. Computed tomography cisternography, with or without injection of radioactive isotopes, may be an option if the bony defect is not evident on CT scans; however, meaningful results are limited to those obtained in patients with active CSF fistulas, and the modality has been shown to have a significant false-negative rate with low-volume or intermittent leaks. The decision to undertake cisternography should also be made with consideration of the increased morbidity associated with lumbar puncture and contrast reactions. Computed tomography cisternograms were acquired in 5 of the 7 patients in the present study, and in all cases the studies were deemed necessary when initial CT scans did not definitely define the full extent of the temporal bone defect. All defects were clearly delineated on the CT cisternogram. Magnetic resonance imaging has proven useful in assessing the presence of dural defects and concomitant encephaloceles. Both T1- and T2-weighted MRI sequences are useful in evaluating the presence and content of herniated cerebral tissue, but their usefulness in delineating the bony defects has been shown to be limited. In a retrospective review of 8 cases of MFCSF leakage, Pappas et al. did not use MRI to identify and correct the site of the fistula. Likewise, Lundy et al. used MRI in 2 of 19 consecutive cases and found the information acquired to be equivocal. In our patients, MRI scans were obtained in 4 of the 7 patients we treated. Scans correlated with the intraoperative presence or absence of an encephalocele in 3 of 4 patients. The fourth had an encephalocele identified intraoperatively that was not visualized on MRI. Magnetic resonance imaging did not alter our surgical approach or repair method in any case.
Conservative Treatment
Conservative treatment, such as avoiding Valsalva maneuvers and straining, head of bed elevated to 30°, or placement of a lumbar drain, was not performed at our institution for MFCSF leak management. Savva et al. reviewed the cases of 92 patients with CSF leaks through the temporal bone. They found that in 82 patients the leak was caused by trauma—head injury in 29 and surgical procedure in 53. Conservative measure worked in 26 of 29 patients in whom CSF leaks were caused by a head injury. The remaining 3 patients required surgical intervention. Conversely, conservative measures worked in only 1 of the 53 patients in whom the leak resulted from surgical procedures. Of note, all patients with nontraumatic leaks in the Savva et al. series required surgical intervention.
In our series the 5 patients with acquired MFCSF leaks had previous surgical procedures. The other 2 patient had MFCSF leaks that were classified as spontaneous. Therefore, we chose not to attempt conservative treatments before surgical repair.
Choice of Approach
Following confirmation of an MFCSF leak, we advocate prompt surgical repair of the bony and dural defects to prevent associated morbidities. In the literature, the approaches and materials used to repair MFCSF leaks vary widely.
The middle fossa, transmastoid, and combined middle fossa/transmastoid approaches are the procedures most commonly described in the literature. Individual selection, however, is highly dependent not only on the advantages and disadvantages of each approach with regard to the specific location of the fistula but also the surgeon's personal experience and comfort level with the approach in question (Table 3).
The transmastoid approach has been reported to be the least technically demanding route. It also confers the additional advantages of eliminating brain retraction and providing access to fistulas originating in the posterior fossa. The disadvantages include decreased exposure of anterior tegmen defects, risk of hearing loss, higher incidence of graft failure with increased recurrence of CSF leaks, and difficulty in approaching large or multiple defects.
The middle fossa craniotomy provides a wide exposure of the entire tegmental plate, allowing identification and repair of large or multiple defects. Repair of tegmental defects can be performed without risk to the ossicular chain, thus preserving hearing. Increased exposure allows more accurate and secure placement of sealant materials, thereby decreasing the possibility of recurrence. However, many authors find the middle fossa craniotomy more technically demanding, and its disadvantages include an inability to approach defects of the posterior fossa. In addition, multiple critical structures are potentially encountered during this approach. Mobilization of the temporal lobe and associated vascular structures such as the vein of Labbé can lead to postoperative seizures or venous infarction. Extradural dissection along the floor of the middle fossa also risks damage to the geniculate ganglion and GSPN.
We prefer the middle fossa approach for MFCSF leak repairs. Preoperative collaboration between the neurotologist and neurosurgeon is essential to ensure optimal surgical outcome. We advocate 3 steps to facilitate extradural dissection and protection of vital neurovascular structures. First, proper head positioning is critical; the head must be rotated 90° toward the opposite shoulder and extended so the temporal squama is parallel to the floor. This position allows gravity to aid in temporal lobe dissection. Second, lumbar drainage should be started only after the craniotomy is performed. Drainage of CSF prompts cerebral relaxation and allows for increased mobilization of temporal lobe dura from the middle fossa floor. Third, facial nerve monitoring is undertaken with the probe used to dissect the temporal dura from the floor of the middle fossa. This technique allows simultaneous dissection and stimulation of the GSPN and geniculate ganglion, conferring additional protection. We perform dural dissection in the traditional back-to-front manner to decrease the incidence of GSPN injury, although some authors have advocated safe dissection from front to back.
Repair Materials
Reconstruction in all cases involved a minimal combination of HAC, collagen-based dural substitute matrix, and PEG hydrogel sealant. In 4 cases we used additional native materials, including split-thickness bone graft, temporalis muscle, and fascia. Intraoperative inspection of the defect's size and dimensions determined whether a split-thickness bone graft would be included.
Hydroxyapatite cement has been successfully used for the remodeling and repair of temporal bone defects by several authors, with Kveton et al. reporting a success rate greater than 97%. The cement is composed of tetracalcium and dicalcium phosphate salts reacting to form hydroxyapatite, the major component of skeletal bone. There are numerous physiological properties that make HAC an ideal material for temporal bone repair. Its high biocompatibility limits postoperative inflammatory response and fibrosis, while its osteoconductive and osteointegrative properties allow for formation and intercalation of new bone within implants. Finally, the product remains malleable for approximately 20 minutes, allowing for easy remodeling into bony defects, and afterward it is resistant to CSF pulsation. Complications have been reported to range from 0% to 11%, with postoperative infections occurring in up to 5% of patients when cases of nasal sinus cavities are included. Kveton and Coelho reported infections in 2.8% of their 109 consecutive temporal bone repairs. Conductive hearing loss due to ossicular chain fixation has been reported up to 33%, with the majority of deficits occurring after transmastoid repairs. Thus, the application of HAC in patients with exposed ossicles must be carefully done. None of the patients in our series had an exposed ossicular chain, but our recommendation for such cases would be to fashion a split-thickness bone graft large enough to cover the defect with or without careful placement of HAC.
In our series, temporal dura repairs involved a combination of dural substitute and PEG hydrogel sealant, with or without the addition of temporalis muscle and fascia. Whenever possible, our initial preference is primary repair with sutures; however, for defects that cannot be approximated, we prefer synthetic, autologous, or combination inlay grafts, supplemented by onlay grafts (inlay/onlay technique). In cases in which the dural defect cannot be repaired primarily because the dura is often thin and frayed and the edges cannot be cleanly reapproximated—because the location of the defect is too difficult to reach, or because the defect is too small to allow insertion of an inlay graft—a stand-alone onlay graft is placed. All repairs are subsequently augmented with PEG hydrogel sealant.
The safety and efficacy of dural reconstruction with synthetic collagen-based dural substitutes are well described. The structure of the synthetic collagen-based dural substitute presents a low-pressure surface for CSF absorption, promoting a graft–dura interface and chemical signaling for native fibroblasts providing mechanical scaffolding. Litvack et al. reported a CSF leakage rate of 6.7%, as well as a 4.2% postoperative infection rate, when using collagen-based substitutes in 425 consecutive patients.
In the literature the PEG hydrogel sealant is well described as an adjunct for the prevention of CSF leaks. Its strong adherence to tissue, biological compatibility, and absorbability make it an ideal choice in MFCSF repair. A study by Boogaarts and associates reported an intraoperative watertight seal rate of 100% and a CSF leak recurrence rate of 4.9% at 3 months in 41 patients who underwent various intracranial procedures. Later, in a large multicenter study, Cosgrove et al. reported a 1.8% postoperative CSF leak rate with PEG hydrogel augmentation of primary suture closure in 111 intracranial procedures; the incidence of infection was 7.2%.
Postoperative Care
We believe that the decreased CSF pulsation and pressure along the extent of the dural repair facilitates more effective healing, despite reports to the contrary. To that end, all patients in the present series underwent lumbar drainage of 10 ml/hour for 3–5 days and remained in intensive care for the duration. Taking into consideration the intraoperative benefits of CSF diversion, we recommend the routine use of lumbar drainage in all patients undergoing repair of an MFCSF leak. Following removal of the lumbar drain, patients are monitored for an additional 1–2 days for any signs of wound leakage, otorrhea, or rhinorrhea.
Complications
Common complications seen in MFCSF leak repairs include recurrence of CSF leaks, graft failure, and wound infections. Savva et al. reported a 14.3% recurrence rate following optimal repair in 92 consecutive patients. Graft failure due to foreign body reaction and migration has been reported, whereas infection has been seen in up to 5% of MFCSF leak repairs. It is difficult to accurately assess the overall complication rate of MFCSF leak repairs due to the variability of techniques and repair materials presented in the current literature.
Comparison
The repair rate in our small series was 100%, which compares favorably with larger middle fossa craniotomy series. Stenzel et al. reported a repair success rate of 91% in a series of 11 patients in whom the reconstruction was conducted exclusively via the middle fossa craniotomy. One patient in their series required reoperation for a recurrent leak. However, the exact materials used for repairs were not described. In their series of 15 patients, Gubbels et al. described the use of bone grafts, fat grafts, HAC, and fascia. All repairs were done via a middle fossa craniotomy, and the repair success rate was 93%. One patient required an additional surgical procedure for a recurrent CSF leak. Gacek et al. reported a repair rate of 93% in 16 patients with 1 patient requiring an additional surgery for a recurrent leak. The repair materials were not specifically described in their series.