Idiopathic Intracranial Hypertension Is Associated with Recurrent CSF Leak and Reoperation for Spontaneous Temporal Encephalocele

Abstract

Objective  Spontaneous temporal encephaloceles (STEs) are increasingly recognized as sequelae of idiopathic intracranial hypertension (IIH), which in turn may further complicate their management. We endeavored to review the University of Pennsylvania institutional experience on operative management of STEs, with a focus on factors which may influence surgical outcomes, particularly IIH.

Design  Retrospective chart review over 9 years from 2013 to 2022.

Setting  Single-center, two-hospital, tertiary care, academic setting.

Participants  Patients undergoing middle cranial fossa (43.9%), transmastoid (44.9%), or combined (11.2%) approaches for repair of STEs during the study period ( n  = 107).

Main Outcome Measures  Postoperative complication rates, recurrence, and diagnosis of IIH.

Results  The majority of patients were female (64.5%), with a mean body mass index (BMI) of 37 kg/m ² and mean age of 57 years. Twelve patients (9%) represented reoperations after failed primary repairs. Fourteen percent of patients undergoing primary surgical repair of STE were diagnosed with IIH, compared with 42% of patients undergoing reoperations ( p  = 0.015). In addition, there was a significant difference in the average BMI of patients undergoing primary (36.4 kg/m ² ) versus revision surgery (40.9 kg/m ² , p  = 0.04). Half of those undergoing reoperation were placed on postoperative acetazolamide compared with 11% of patients undergoing primary operations. No patient experienced recurrent leak after reoperation.

Conclusion  Based on our institutional experience, elevated BMI and the presence of IIH are significant predictors of reoperation for STE. In our experience, acetazolamide is a common adjunct management strategy in addition to reoperation for patients with recurrent cerebrospinal fluid leak in the setting of STE.

Introduction

Cranial encephaloceles are a rare disease entity with an estimated prevalence of roughly 1 in 35,000. ¹ Temporal encephaloceles are characterized by a protrusion of temporal lobe parenchyma through overlying dura mater via defects in the middle cranial fossa (MCF). Spontaneous temporal encephaloceles (STEs) represent a subgroup of these which are found independently from any preceding traumatic, neoplastic, inflammatory, or iatrogenic conditions. Depending on size, location, and associated pathological conditions, STEs may remain unidentified into adulthood. When they do manifest, they frequently present with middle ear effusions, acute otitis, conductive hearing loss, intractable epilepsy, ^{2 3} or recurrent meningitis from occult or symptomatic cerebrospinal fluid (CSF) leak.

The precise pathophysiologic mechanism underlying the development of STEs remains unknown. Persistent developmental defects along the lateral skull base have been proposed. ^{4 5} However, the occurrence of such defects alone has not been shown to be sufficient for pathological herniation, as evidenced by the significantly higher rate of such defects compared with STEs. ⁶ Perhaps most convincingly, progressive dural and osseous thinning secondary to elevated intracranial pressure (such as in the case of idiopathic intracranial hypertension, IIH) has been described as a potential etiology for STEs, although this remains controversial. ^{7 8}

Here, we report the largest single-center case series of STEs managed surgically via the MCF, transmastoid (TM), or combined approaches. We sought to compare differences in the management of patients undergoing primary repair versus revision surgery for STEs. In particular, we sought to further assess the roles of elevated BMI and IIH in the surgical treatment and risk of recurrence of STEs.

Methods

Data Collection

Data were collected in accordance with the University of Pennsylvania Institutional Review Board Policies on human subjects' research. We conducted a retrospective chart review of all consecutive patients who underwent surgical repair of STEs via MCF, TM, or combined approaches over a 9-year period from 2013 to 2022 at two hospitals within the University of Pennsylvania Health System—the Hospital of the University of Pennsylvania and Pennsylvania Hospital. Patients undergoing MCF, TM, or combined approaches for alternative diagnoses, those with secondary temporal encephaloceles—such as resulting from prior ear infections, traumas, or iatrogenic injury—and those with CSF leaks not associated with a temporal encephalocele were excluded from this study. Data reviewed included patient demographic information, surgical reports, preexisting conditions, and postoperative complications. Patients with a documented diagnosis of IIH or a new diagnosis of IIH based on elevated CSF opening pressures on outpatient lumbar punctures (LPs) were included in the IIH diagnosis group. Postoperative hearing changes were defined as improved (decrease in air conduction four-frequency pure-tone average [PTA] of ≥10 dB), unchanged (change in air conduction four-frequency PTA of ≤10 dB), or worsened (increase in air conduction four-frequency PTA ≥10 dB) comparing pre- versus postoperative audiograms. Other surgical complications (including the presence of vertigo or facial nerve injury) were extracted from postoperative clinical notes.

Surgical Technique

At our institution, the TM approach is favored over the MCF approach, when the pathology is amenable, as both approaches have been shown to have similar rates of hearing preservation, ^{9 10} and the TM approach avoids the morbidity of an intracranial operation, with resultant shorter lengths of hospital stay. ^{9 11} The MCF approach is preferred for large or multiple defects and anteriorly located encephaloceles ( Fig. 1 ). ¹² A combined approach was chosen based on surgeon clinical judgment when a more complete exploration of the tegmen, complete debridement of the mastoid bowl, or a particularly robust repair is thought to be indicated to reduce the likelihood of recurrence, ^{11 13} or in cases where the TM approach is pursued initially but converted to combined intraoperatively due to findings of anteriorly located or multiple defects.

Fig. 1.

Imaging examples of spontaneous temporal encephaloceles approached via MCF (A: CT, B: T2-weighted MRI) and TM (C: CT, D: T1-contrasted MRI). Note that at our institution, there were similar recurrence and complication rates regardless of approach indicating equal safety and efficacy. As such, the choice of MCF versus TM was based on the pathological anatomy, and where possible the TM was favored due to less invasiveness. CT, computed tomography; MCF, middle cranial fossa; MRI, magnetic resonance imaging; TM, transmastoid.

In patients undergoing repair via the MCF approach, the temporalis muscle is reflected to expose the root of the zygoma and the temporalis squama. Two burr holes are placed and an ∼3.5 cm diameter craniotomy is made to the superior limit of the temporal squama and the inferior limit of the zygoma. Dura is progressively elevated until the encephalocele and associated defect(s) are located. The encephalocele is elevated out of the site of dehiscence and the skull base and dural dehiscence, when present, are repaired as described later. For TM repairs, the patient undergoes mastoidectomy in standard fashion with exposure of the antrum and epitympanum. In the case of an anteriorly located defect, the facial recess may be opened and the ossicles disarticulated to facilitate access. The encephalocele(s) and associated tegmen dehiscences are located. Patients operated on using the combined approach undergo both of the aforementioned techniques to facilitate access to and repair of the STE. In all cases, repair is conducted using the following materials alone or in combination: temporoparietal fascia, collagen-based dural graft, calvarial bone, bone cement, and/or allograft materials. Choice of repair materials is based on the number and size of defects, with a preference for autograft materials when possible. For singular and small defects where robust temporoparietal fascia is harvested, this is used alone; in cases where this is deemed unsuitable for a robust repair, this is reinforced with collagen-based dural graft. For revision cases, multiple or large defects, boney reinforcement is implemented with a split-thickness calvarial bone graft being preferred, and bone cement or allograft applied when this is deemed insufficient for a robust repair.

Statistical Methods

Descriptive statistics—means and standard deviations (SDs) for normally distributed data or medians and interquartile ranges (IQRs) in cases of nonnormally distributed data—were used to summarize cohort demographics. Statistical analyses were performed to compare BMI, diagnosis of II (88.8%) H, and other variables between patients with and without revision surgery and postoperative CSF leak, and to compare variables across surgical approaches (MCF vs. TM vs. combined). Continuous variables (BMI, age at presentation, age at surgery, operative time, and CSF opening pressure) were compared between the groups by independent samples t -test. When unequal variance between groups was detected, unequal variance t -tests (Welch) were used. Categorical variables (sex, repair material, number of leak sites, postoperative hearing change, postoperative CSF leak, diagnosis of IIH, requirement for permanent CSF diversion, and use of postoperative acetazolamide) were compared between groups using chi-square tests for binary variables or analysis of variance for categorical variables. Patients with missing values of variables of interest were excluded from analyses. All statistical tests were evaluated at a significance level of α = 0.05. All statistics were performed using Stata version 15.1 (StataCorp, College Station, Texas, United States).

Results

In the 9-year study period, 107 patients underwent repair of STEs. Ninety-five of these represented primary operations, and 12 (11.2%) were revision cases for patients who had undergone primary repair of an STE at an outside institution and suffered recurrent CSF leak—the latter will be referred to as “revision cases” henceforth. In contrast, eight patients suffered CSF leak after primary repair at our institution—referred to as “recurrent CSF leaks” henceforth. The average follow-up period for patients undergoing MCF was 16 months (range 0.5–80 months), for those undergoing TM was 18 months (0.25–66 months), and for those undergoing combined approaches was 30 months (3–96 months). There was a predominance of female patients (64.5%), and across the entire cohort, patients had a mean BMI of 37 kg/m ² and a mean age of 57 years ( Table 1 ). Among primary repairs, 41 patients (43%) underwent MCF, 42 patients (44%) underwent TM, and 12 patients (13%) underwent combined approaches for STE repair ( Table 1 ). Among revision cases, 50% (six patients) underwent MCF and 50% (six patients) underwent TM approaches for STE repair. A longer operative time was observed with combined approaches (median 174 minutes, IQR 125–350) compared with MCF (mean 133.5 minutes, SD 38.3) and TM (mean 149.9 minutes, SD 63.8) ( p  < 0.001) ( Table 2 ). There was no difference in the rates of facial nerve injury, vertigo, or other complications (wound dehiscence, contusion, gustatory dysfunction, graft failure, persistent CSF leak requiring lumbar drain placement, and nonlocalizing neurologic complaints) or the proportions of patients with postoperative hearing changes (improved, worsened, or unchanged) across surgical approaches ( p  = 0.96). All patients in our series were repaired with either allograft material alone or a combination of materials (including varying combinations of bone, fascia, collagen-based dural graft and allograft).

Table 1. Summary statistics.

Variable	Count (%)
Sex
Male	38 (35.5)
Female	69 (64.5)
Age at surgery
< 50	25 (23.4)
50–59	35 (32.7)
60–69	32 (29.9)
70 +	15 (14.0)
Body mass index range
17.5–24.9	5 (5.4)
25.0–29.9	18 (19.1)
30.0–34.9	22 (23.4)
35.0–39.9	13 (13.8)
> 40	36 (38.3)
Procedure type
MCF	47 (43.9)
TM	48 (44.9)
Combined MCF and TM	12 (11.2)

Abbreviations: MCF, middle cranial fossa; TM, transmastoid.

Table 2. Postoperative complications stratified by surgical approach.

Variable	Combined	MCF	TM
Sex
Female	8 (67)	34 (72)	27 (56)
Male	4 (33)	13 (28)	21 (44)
Mean age at presentation (y)	60.3	53.5	58.7
Revision case
No	12 (100)	41 (87)	42 (88)
Yes	0 (0)	6 (13)	6 (12)
Mean operative time (min) a	224.5	133.5	149.9
No. of intraoperative CSF leak sites	2.17	1.70	1.40
Postoperative complications
None	12 (100)	42 (88)	44 (92)
Facial nerve injury	0 (0)	1 (2)	1 (2)
Vertigo	0 (0)	1 (2)	1 (2)
Other	0 (0)	3 (6)	2 (4)
Postoperative CSF leak
No	10 (83)	44 (94)	43 (90)
Yes	2 (17)	3 (6)	5 (10)
Postoperative IIH occurrence
No	8 (67)	34 (72)	41 (85)
Definite	4 (33)	10 (21)	4 (8)
Suspected	0 (0)	3 (6)	3 (6)
Postoperative acetazolamide
No	0 (83)	38 (81)	43 (90)
Yes	2 (17)	9 (19)	5 (10)

Abbreviations: CSF, cerebrospinal fluid; IIH, idiopathic intracranial hypertension; MCF, middle cranial fossa; TM, transmastoid.

^ap  < 0.001.

Comparing primary and revision cases, the proportion of patients with diagnoses of IIH was significantly different between the groups. Five patients (41.7%) undergoing revision surgery had a definitive diagnosis of IIH, compared with 13 patients (13.7%) undergoing primary surgery ( p  = 0.015) ( Fig. 2 ). The mean BMI among patients undergoing revision versus primary cases also significantly differed (mean 40.9 vs. 36.5 kg/m ² , respectively; p  = 0.04) ( Fig. 3 ). There was no difference in the proportion of patients who underwent TM, MCF, or combined approaches in primary versus revision groups. Patients undergoing revision surgery had a greater number of sites of tegmen dehiscence identified intraoperatively than those undergoing primary repairs (mean 2.2 vs. 1.5), but the difference was not statistically significant ( p  = 0.16). Patients undergoing revision cases at our institution were more likely to undergo repair using a combination of materials (e.g., fascia, bone, allograft) compared with patients undergoing primary repairs, who were more likely to be repaired using allograft materials alone ( p  = 0.028). Among patients undergoing revision surgery, 50% were prescribed postoperative acetazolamide, compared with 11% of patients undergoing primary repair. A similar proportion of patients undergoing primary and revision surgeries underwent permanent CSF diversion via placement of a ventriculoperitoneal shunt (5 and 8% of patients undergoing primary and revision surgeries, respectively). The majority of patients who underwent permanent CSF diversion (4/6, 66%) had a diagnosis of IIH. None of the patients undergoing revision surgery (revision cases) suffered recurrent CSF leak following reoperation at our institution ( Table 3 ).

Fig. 2.

Proportion of patients with IIH in primary versus revision cases. Although 13 out of 95 patients undergoing primary surgery had a definitive diagnosis of IIH (13.7%), more than one-third of patients undergoing revision surgery (5/12 = 41.7%) had a definitive diagnosis of IIH. This difference in proportions was statistically significant between primary and revision surgery groups ( p  = 0.015). IIH, idiopathic intracranial hypertension.

Fig. 3.

BMI differences between primary versus revision cases. Mean BMI among patients undergoing revision surgery (40.9 kg/m ² ) was roughly four points higher than for patients undergoing primary surgery (36.5 kg/m ² , p  = 0.04).

Table 3. Differences in outcomes between primary and revision cases.

Variable	Primary ( N  = 95)	Revision ( N  = 12)
Sex
Female	61 (64)	8 (67)
Male	34 (36)	4 (33)
Mean age at presentation (y)	56.8	55
Mean BMI (kg/m 2 ) a	36.4	40.9
Mean operative time (min)	147	176.8
No. of intraoperative CSF leak sites	1.55	2.18
Postoperative complications
None	87 (92)	11 (83)
Facial nerve injury	2 (2)	0 (0)
Vertigo	1 (1)	1 (17)
Other	5 (5)	0 (0)
Recurrent CSF leak
No	87 (92.5)	12 (100)
Yes	8 (7.5)	0 (0)
Postoperative IIH occurrence a
No	82 (86)	7 (58)
Definite	13 (14)	5 (42)
Postoperative acetazolamide b
No	85 (89)	6 (50)
Yes	10 (11)	6 (50)

Abbreviations: BMI, body mass index; CSF, cerebrospinal fluid; IIH, idiopathic intracranial hypertension.

^ap  < 0.05.

^bp  < 0.001.

^cp  < 0.01.

Eight patients (7.5%) undergoing primary repair at our institution suffered recurrent CSF leak postoperatively. Management of these eight patients was as follows: six underwent reoperation, one was recommended for reoperation and sought second opinion, and one resolved with shunting. The mean BMI was higher among patients who suffered recurrent leak after primary repair at our institution (mean 40.8 kg/m ² , SD 7.6) compared with those who did not (mean 36.6 kg/m ² , SD 8.9), but the difference was not statistically significant ( p  = 0.179). There was no difference in the proportion of IIH diagnoses among patients who suffered CSF leak recurrence after primary repair ( n  = 1 patient, 12.5%) compared with those who did not ( n  = 17 patients, 17.2%) ( p  = 0.734). There were similar proportions of patients who underwent repairs via the MCF, TM, and combined approaches among groups who did and did not suffer recurrent CSF leak after primary repair at our institution ( p  = 0.930), and the proportions of patients who underwent repair via a combination of materials versus allograft alone were also similar across groups ( p  = 0.665). The number of leak sites identified intraoperatively did not differ significantly comparing patients who suffered recurrent leak after primary repair at our institution (mean 1.4, SD 0.7) and those who did not (mean 1.6, SD 1.4) ( p  = 0.619).

Discussion

Here, we describe our institutional experience with repair of STEs using the MCF, TM, and combined approaches in 107 consecutive cases. To our knowledge, this represents the largest single-center case series of STE repairs in the published literature to date. We summarize our institutional experience and have leveraged the increased statistical power afforded by our large sample size to investigate a long-suspected yet unverified role of elevated intracranial pressure in the pathophysiology of STEs.

Several indirect indicators of IIH have been associated with STE, including obesity, ¹⁴ empty sella syndrome, ¹⁵ and increased incidence in women. ¹⁶ A 2012 study by Brainard et al was the first to directly investigate the association between IIH and STE ¹⁷ that had been proposed speculatively in prior literature. ^{18 19} Although this study demonstrated a likely association, with sample size of n  = 26, it was underpowered for statistical analysis. We found statistically significant relationships between both diagnosis of IIH and BMI with primary versus revision surgery for STE and CSF leak. There was a predominance of obese patients in this series, with an average BMI of 37 kg/m ² and more than one-third (38%) of patients with BMI more than 40 kg/m ² . Still, the mean BMI was greater in patients undergoing revision compared with primary STE repair, and there was a statistically significantly higher proportion of patients with a confirmed diagnosis of IIH in the revision surgery group compared with the primary repair group. Among the 18 patients with confirmed definitive diagnosis of IIH, 9 (50%) underwent preoperative LPs at our institution—among these, 7 had opening pressures > 20 cm H ₂ O, while the 2 had normal opening pressures and the remaining 9 had remote diagnoses of IIH. Four out of the 89 patients without a diagnosis of IIH had opening pressures assessed via preoperative LP, all of which were lower than 20 cm H ₂ O. Patients with demonstrated elevated intracranial pressure were more likely to be placed on Diamox.

Patients who underwent revision surgery at our institution were more likely to be repaired using a combination of materials compared with allograft alone, and were also more likely to be placed on postoperative acetazolamide. Notably, no patient who underwent revision surgery at our institution suffered a second recurrence postoperatively. The lack of any cases of recurrence in the revision surgery group precludes further analysis as to the utility of acetazolamide or multilayered repair in preventing further recurrence.

Among patients undergoing primary STE repair at our institution who suffered recurrent CSF leak postoperatively, the mean BMI was higher compared with patients undergoing primary repairs at our institution who did not suffer recurrent leaks, consistent with the notion that BMI influences the likelihood of recurrent leak after surgery—although in this case, the difference in BMI between the groups did not reach statistical significance. In addition, despite the relatively higher BMI, there was no difference in the proportion of IIH diagnoses comparing those who did and did not suffer leak recurrence after primary repair at our institution. However, the lack of significance of these associations may be due to the overall small number of patients in our series who suffered recurrent CSF leak after primary repair at our institution ( n  = 8 patients, 7.5%), reducing our power to detect significant relationships and increasing the likelihood of type II error.

We were unable to identify any other factors predictive of recurrent CSF leak, including surgical approach (TM, MCF, or combined), number of leak sites, or repair material, which similarly may be due to the low number of recurrences in this series and thus reduced statistical power to detect significant associations.

Our results are congruent with the longstanding notion that elevated BMI and IIH are associated with STE. Yancey et al previously reported on 94 patients who underwent repair for spontaneous (44%) and secondary temporal encephaloceles. In this series, all leaks were initially repaired via a TM approach, and if multiple or large defects were encountered intraoperatively, a combined approach was undertaken. This study found that mean BMI was higher in patients with spontaneous CSF leaks and class III obesity was more common in spontaneous CSF leak compared with secondary etiologies. Similar findings have been reported by other authors; we note that these were relatively smaller studies which were underpowered to identify statistically significant associations. ^{21 22}

Rates of postoperative facial nerve palsy in this series were similar to previous reports in the literature. Hoang et al. identified 3/25 patients (12%) with facial palsy post-MCF repair, compared with 1/47 (2%) of MCF cases in our series. ¹² Jeevan et al reported a series of 26 patients undergoing combined repair, of whom 1 (3.8%) developed postoperative facial weakness compared with the 0% incidence among patients undergoing combined repair in our series. ^{12 13} We could not identify any other published reports of facial nerve injury following TM approaches for STE, but the incidence of iatrogenic facial nerve injury associated with otologic procedures is estimated to be between 0.6 and 3.6%. ²³ This is congruent with our study findings in which 1/48 (2%) developed postoperative facial palsy following the TM approach to STE repair.

Conclusion

The present study provides a comparatively large institutional review of a rare disease state, STE, and provides new insights into the relationships between BMI, IIH, and recurrent CSF leak following STE repair. In this series, patients undergoing revision compared with primary surgery were significantly more likely to have a diagnosis of IIH and elevated BMI. There was also a higher proportion of IIH among patients who suffered recurrent CSF leak following primary repair at our institution compared with those who did not, but the relationship did not reach statistical significance, possibly due to small numbers of cases. Repair is possible using a combination of materials, and the choice of surgical approach and materials does not appear to increase the risk of postoperative CSF leak.