Lumbar Drain Use during Middle Fossa Approaches for Nonneoplastic Pathology of the Skull Base

Abstract

Introduction  The middle fossa craniotomy (MFCs) is commonly utilized for spontaneous cerebrospinal fluid (CSF) leaks, encephaloceles, and superior semicircular canal dehiscence (SSCD). This study compares postoperative outcomes of MFCs with and without LD use.

Methods  A retrospective cohort study of adults over the age of 18 years presenting for the repair of nonneoplastic CSF leak, encephalocele, or SSCD via MFC from 2009 to 2021 was conducted. The main exposure of interest was the placement of an LD. The primary outcome was the presence of postoperative complications (acute/delayed neurologic deficit, meningitis, intracranial hemorrhage, and stroke). Secondary outcomes included operating room (OR) time, length of stay, recurrence, and need for reoperation.

Results  In total, 172 patients were included, 96 of whom received an LD and 76 who did not. Patients not receiving an LD were more likely to receive intraoperative mannitol ( n  = 24, 31.6% vs. n  = 16, 16.7%, p  = 0.02). On univariate logistic regression, LD placement did not influence overall postoperative complications (OR: 0.38, 95% confidence interval [CI]: 0.05–2.02, p  = 0.28), CSF leak recurrence (OR: 0.75, 95% CI: 0.25–2.29, p  = 0.61), or need for reoperation (OR: 1.47, 95% CI: 0.48–4.96, p  = 0.51). While OR time was shorter for patients not receiving LD (349 ± 71 vs. 372 ± 85 minutes), this difference was not statistically significant ( p  = 0.07).

Conclusion  No difference in postoperative outcomes was observed in patients who had an intraoperative LD placed compared to those without LD. Operative times were increased in the LD cohort, but this difference was not statistically significant. Given the similar outcomes, we conclude that LD is not necessary to facilitate safe MCF for nonneoplastic skull base pathologies.

Introduction

Middle fossa craniotomy (MFC) is a commonly used approach for surgical repair of nonneoplastic middle fossa skull base pathologies, including spontaneous cerebrospinal fluid (CSF) leak, encephalocele, and superior semicircular canal dehiscence (SSCD). ^{1 2 3} The use of preoperative lumbar drain (LD) in MFC has been a topic of much debate, as both benefits and risks of LD placement have been mentioned. ^{4 5} LD usage may facilitate brain relaxation, ⁶ which may allow optimal visualization of the skull base and minimize the risk of retraction or manipulation injury to the temporal lobe. ^{7 8} Additionally, LD use may reduce postoperative CSF volume, decreasing the risk of the development of CSF leaks. ⁶ However, LD placement is not without risk, as various complications such as headache, spinal hematoma, meningitis, retained catheter tip, operative delay for placement, and intracranial hypotension/brain herniation from excessive CSF removal have been described. ^{9 10 11 12 13}

The efficacy and need for LD placement in avoiding postoperative complications and improving outcomes have been examined in various skull base approaches, including open and endoscopic, with heterogeneous results. ^{6 14} The literature on the usage of LD in MFC remains sparse, and little consensus exists on its utility for nonneoplastic MFC approaches. ⁵ In a retrospective study with 60 patients undergoing MFC for spontaneous CSF leak or encephalocele, Nelson et al ⁵ found that perioperative LD placement significantly increased hospital length of stay (LOS) and cost. To date, no studies have sought to identify demographic and preoperative factors that may predict LD placement or directly compare perioperative and postoperative outcomes in patients undergoing LD placement and their counterparts without intraoperative LD management. Thus, the primary objectives of this study were (1) to describe demographic and preoperative factors for patients managed with LD (LD group) and those who did not (non-LD group) and (2) to compare perioperative variables and postoperative outcomes between the two groups following MFC for nonneoplastic etiology.

Methods

Study Design and Patient Population

A retrospective single-center cohort study for adults undergoing MFC for nonneoplastic etiology between 2009 and 2022 was performed. The Institutional Review Board deemed the study to be exempt (IRB #201263), and participant consent was not required. The inclusion criteria for the study were adult patients of 18 years and older undergoing a middle fossa approach for craniotomy for nonneoplastic etiology with preoperative, perioperative, and postoperative outcomes available in the electronic medical record. Pediatric patients, those undergoing MFC for tumors, and patients with unavailable data were excluded. Four neurosurgeons and six neuro-otologists were included and had generally strict practice patterns used regarding the placement of LD. Table 1 summarizes the LD use of each neurosurgeon, neurotologist, and neurosurgeon/neurotologist pair. Referral practices at this hospital do not consider patient clinical characteristics or level of complexity; cases are assigned based on surgeon availability.

Table 1. Summary of the operative characteristics of each neurosurgeon and neurotological surgeon.

	n (% Total operations, total n  = 172)	LD use (% total LD use)
Neurosurgery attendings
1	36 (21%)	1 (1%)
2	17 (10%)	15 (15.6%)
3	57 (33%)	51 (53.1%)
4	62 (36%)	29 (30.2%)
Neurotology attendings
A	24 (14%)	16 (16.7%)
B	56 (33%)	46 (47.9%)
C	26 (15%)	8 (8.3%)
D	26 (15%)	5 (5.2%)
E	7 (4%)	5 (5.2%)
F	33 (19%)	16 (16.7%)
Attending pairs
1/B	1 (1%)	0 (0%)
1/C	14 (8%)	1 (1%)
1/D	19 (11%)	0 (0%)
1/E	1 (1%)	0 (0%)
1/F	1 (1%)	0 (0%)
2/A	6 (3%)	6 (6.3%)
2/B	2 (1%)	1 (1%)
2/C	3 (2%)	3 (3.1%)
2/E	6 (3%)	5 (5.2%)
3/B	51 (30%)	45 (46.9%)
3/C	1 (1%)	1 (1%)
3/D	5 (3%)	5 (5.2%)
4/A	18 (10%)	10 (10.4%)
4/B	2 (1%)	0 (0%)
4/C	8 (5%)	3 (3.1%)
4/D	2 (1%)	0 (0%)
4/F	32 (19%)	16 (16.7%)

Abbreviation: LD, lumbar drain.

Data Collection

Data were queried using the Research Derivative from the Vanderbilt Institute for Clinical and Translational Research, and patients undergoing MFC between 2009 and 2022 were screened for inclusion. ¹⁵ A manual review and extraction of the electronic medical record for variables of interest was conducted; all extracted demographical, medical history, and outcome variables data were extracted and stored in a secure REDCap database. ^{16 17}

Exposure Variable

The primary exposure variable of interest was LD placement at the time of MFC. Included patients were dichotomized into two groups based on a review of the operative report to determine whether an LD was placed at the time of operation. Variables associated with LD placement, such as LD output, opening pressure, and duration, were recorded. LDs were placed intraoperatively prior to the surgical procedure based on neurosurgeon preference. These were most commonly removed at the end of the surgical procedure prior to patient emergence.

Outcome Variables

Demographic variables such as age, sex, race and insurance status, and relevant preoperative medical history were extracted. Preoperative symptomatology, including aural fullness, rhinorrhea, otorrhea, headache, vertigo, and tinnitus, were documented. Indication for surgery, as well as perioperative variables such as mannitol and dexamethasone usage, were recorded.

The primary outcome of interest was the presence of postoperative complications, including acute/delayed neurologic deficit, intracranial hemorrhage, stroke, meningitis, spinal hemorrhage, lumbar spine CSF leak, CSF hypotension headache, retained catheter, and epidural abscess. Secondary outcomes included operating room (OR) time, evidence of disease recurrence, and need for reoperation at last follow-up.

Statistical Analysis

Descriptive statistics were generated for demographic, relevant medical history, operative and postoperative variables. Continuous variables were presented using mean and standard deviation, while categorical variables were represented with frequencies and percentages. Between-group analyses comparing LD versus non-LD groups were performed using Pearson's chi-square tests and Kruskal–Wallis rank-sum tests, as appropriate. Univariate logistic regression modeling was performed to evaluate the relationship between the presence of LD and outcome variables. Statistical significance was determined a priori to be p  < 0.05. Statistical analysis was performed using R version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Demographics and Preoperative Presentation

The final patient cohort included 172 patients, 96 (55.8%) of whom received an LD. The median follow-up was 15 months (interquartile range: 4–34), and 2.3% of patients did not attend any follow-up appointments. Demographics between the LD and non-LD groups were similar; there were no statistically significant differences in age (56.8 ± 11.3 vs. 56.4 ± 13.0 years, p  = 0.844), race ( p  = 0.862), and insurance type ( p  = 0.761). A lower proportion of patients receiving LDs were male compared to their non-LD counterparts ( n  = 25, 26.0% vs. n  = 31, 40.8%, p  = 0.040). LD and non-LD groups did not have statistically significant differences in medical comorbidities, including hypertension ( n  = 50, 52.1% vs. n  = 46, 60.1%, p  = 0.268), diabetes mellitus ( n  = 22, 22.9% vs. n  = 26, 34.2%, p  = 0.101), idiopathic intracranial hypertension ( n  = 3, 3.1% vs. n  = 6, 7.9%, p  = 0.163), and obstructive sleep apnea ( n  = 13, 13.5% vs. n  = 13, 17.1%, p  = 0.517). Finally, both groups did not have statistically significant differences in preoperative symptomology, including aural fullness ( n  = 36, 37.5% vs. n  = 30, 39.4%, p  = 0.751), rhinorrhea ( n  = 11, 11.5% vs. n  = 8, 10.5%, p  = 0.846), otorrhea ( n  = 61, 63.5% vs. n  = 54, 71.1%, p  = 0.299), or headache ( n  = 13, 13.5% vs. n  = 13, 17.1%, p  = 0.517). Most patients in both groups were operatively treated with an MFC for encephalocele repair ( n  = 81, 84.4% vs. n  = 61, 80.3%), and no statistically significant differences were noted in surgery indication ( p  = 0.149). Table 2 summarizes demographics and preoperative comorbidities of the included cohort.

Table 2. Demographics and preoperative comorbidities.

	(+) Lumbar drain ( n  = 96, 56%)	(−) Lumbar drain ( n  = 76, 44%)	p -Value
Demographics
Age (mean [SD])	56.8 (11.3)	56.4 (13.0)	0.844
Sex (male)	25 (26.0%)	31 (40.8%)	0.040
Race			0.862
White/Caucasian	87 (90.6%)	70 (92.1%)
Black/African-American	5 (5.2%)	4 (5.3%)
Other	4 (4.2%)	2 (2.6%)
Insurance type			0.761
Medicare	22 (22.9%)	23 (30.3%)
Private	44 (45.8%)	32 (42.1%)
Military	4 (4.2%)	2 (2.6%)
Uninsured/Self-pay	19 (19.8%)	12 (15.8%)
Unknown	7 (7.3%)	7 (9.2%)
Medical history
Preoperative BMI (mean [SD])	35.8 (8.3)	34.6 (9.6)	0.433
Hypertension	50 (52.1%)	46 (60.1%)	0.268
Diabetes mellitus	22 (22.9%)	26 (34.2%)	0.101
Previous stroke	1 (1.0%)	0 (0.0%)	0.372
Idiopathic intracranial hypertension	3 (3.1%)	6 (7.9%)	0.163
Obstructive sleep apnea	13 (13.5%)	13 (17.1%)	0.517
Preoperative anticoagulant/antiplatelet use	17 (17.7%)	20 (26.3%)	0.162
Presence of empty sella	2 (2.1%)	1 (1.3%)	0.709
Indication for surgery			0.149
CSF leak	9 (9.4%)	4 (5.3%)
Encephalocele	81 (84.4%)	61 (80.3%)
Semicircular canal dehiscence	7 (7.3%)	8 (10.5%)
Other	0 (0.0%)	3 (3.9%)
Preoperative symptomatology
Aural fullness	36 (37.5%)	30 (39.4%)	0.751
Rhinorrhea	11 (11.5%)	8 (10.5%)	0.846
Otorrhea	61 (63.5%)	54 (71.1%)	0.299
Headache	13 (13.5%)	13 (17.1%)	0.517
Vertigo	18 (18.8%)	7 (9.2%)	0.078
Tinnitus	24 (25.0%)	12 (15.8%)	0.140
Perioperative variables
Positioning			<0.001
Park-Bench	48 (50.0%)	5 (6.6%)
Supine	48 (50.0%)	71 (93.4%)
Tympanomastoidectomy performed	79 (82.3%)	53 (69.7%)	0.097
Mannitol use	16 (16.7%)	24 (31.6%)	0.020
Intraoperative dexamethasone use	76 (79.2%)	58 (76.3%)	0.751
Postoperative dexamethasone use	63 (65.6%)	50 (65.8%)	0.908
Intraoperative intracranial retractor use	1 (1.0%)	2 (2.6%)	0.435

Abbreviations: BMI, body mass index; CSF, cerebrospinal fluid; SD, standard deviation. Bold p -Values denote the the significant findings in this table.

Intraoperative Lumbar Drain Usage

Perioperatively, nearly all patients without an LD were positioned supine ( n  = 71, 93.4%), a position in which only half of the LD cohort was placed ( n  = 48, 50.0%; p  < 0.001). The remainder of patients in both cohorts were placed in Park–Bench positioning. Furthermore, LD patients were less likely to receive intraoperative mannitol than their non-LD counterparts ( n  = 16, 16.7% vs. n  = 24, 31.6%, p  = 0.020). No statistically significant differences in intraoperative ( p  = 0.751) or postoperative ( p  = 0.908) dexamethasone usage were observed. Furthermore, the use of intraoperative intracranial retractors was rare in both the LD and non-LD groups ( n  = 1, 1.0% vs. n  = 2, 2.6%, p  = 0.435; Table 2 ). The difference in operative times between LD and non-LD groups (371.8 ± 84.9 vs. 349.2 ± 71.2 minutes) did not reach statistical significance ( p  = 0.070; Table 3 ).

Table 3. OR time, hospital stay, and postoperative complications.

	LD	No LD	p- Value
OR time (minutes)	371.8 ± 84.9	349.2 ± 71.2	0.069
Hospital stay (days)	2.8 ± 2.1	2.4 ± 1.8	0.068
Postoperative complications
Acute neurologic deficit	0	1 (1.3%)	0.442
Delayed neurologic deficit	2 (2.1%)	0	0.504
Intracranial hemorrhage	2 (2.1%)	3 (3.9%)	0.666
Postoperative stroke	0	1 (1.3%)	0.442
Postoperative meningitis	0	0	–
LD-specific complications
Spinal hemorrhage	0	–	–
Lumbar spine CSF leak	0	–	–
CSF hypotension headache	0	–	–
Retained catheter	0	–	–
Epidural abscess	0	–	–

Abbreviations: CSF, cerebrospinal fluid; OR, operating room; LD, lumbar drain.

Effects of Lumbar Drain Usage on Postoperative Outcomes

There was no statistical difference in LOS between the LD versus the non-LD groups (2.8 ± 2.1 vs. 2.4 ± 1.8 days, p  = 0.070). The majority of the LD group had their LD removed immediately after surgery ( n  = 70, 72.9%), and for those who did not have their LD immediately removed following surgery, the average time to removal was 2.7 ± 1.7 days. Postoperative complications were similar between LD and non-LD groups, with no statistically significant differences in acute neurological deficit (0% vs. 1.3%, p  = 0.442), delayed neurological deficit (2.1% vs. 0%, p  = 0.504), intracranial hemorrhage (2.1% vs. 3.9%, p  = 0.656), stroke (0% vs. 1.3%, p  = 0.442), and meningitis (0% vs. 0%). LD-specific complications, including spinal hemorrhage, lumbar spine CSF leak, CSF hypotension headache, retained catheter, and epidural abscess, did not occur for any patient in the LD group ( Table 3 ). On univariate logistic regression analysis predicting postoperative complications and outcomes, the use of LD was not significantly associated with overall immediate postoperative complication rates (OR: 0.38, 95% CI: 0.05–2.02, p  = 0.280). Furthermore, LD use was not significantly associated with long-term CSF leak recurrence (OR: 0.75, 95% CI: 0.25–2.29, p  = 0.610), reoperation (OR: 1.47, 95% CI: 0.48–4.96, p  = 0.510), or permanent CSF diversion (OR: 0.15, 95% CI: 0.02–1.32, p  = 0.090) at last follow-up ( Table 4 ).

Table 4. Univariate logistic regression on the effect of LD use on postoperative outcomes.

	OR (95% CI)	p -Value
Overall postoperative complications	0.38 (0.05–2.02)	0.28
CSF leak recurrence	0.75 (0.25–2.29)	0.61
Reoperation	1.47 (0.48–4.96)	0.51
Permanent CSF shunt	0.15 (0.01–1.32)	0.09

Abbreviations: CI, confidence interval; CSF, cerebrospinal fluid; OR, operating room; LD, lumbar drain.

Discussion

Key Findings

The present study sought to compare perioperative variables and postoperative outcomes in LD usage for patients undergoing MCF for nonneoplastic skull base pathology. Perioperatively, patients who had an LD placed were significantly less likely to receive intraoperative mannitol, while no differences in dexamethasone usage were observed. Although there was a trend toward longer OR time and hospital LOS, LD placement did not statistically increase OR time or hospital LOS. Following operative intervention, no differences in postoperative complications, CSF leakage rates, permanent CSF diversions, and need for reoperation were observed between the LD and non-LD groups. Taken together, the results of this study suggest that LD placement has no effect on immediate postoperative outcomes.

Postoperative Complications and Need for Reoperation

Within our cohort, we found no statistically significant differences between the LD and non-LD groups with regard to overall postoperative complications and the need for reoperation. While our study is the first to determine the utility of LD placement in patients undergoing MFC for nonneoplastic etiologies, other studies have sought to determine the safety and efficacy of LD placement in related skull-base operations. In a retrospective cohort study of 150 patients undergoing open posterior fossa craniotomy, Bien et al found that LD usage significantly lowered rates of postoperative CSF leakage. ⁶ Furthermore, a prospective randomized controlled trial comparing LD to non-LD usage in 230 patients undergoing endoscopic endonasal approaches demonstrated significantly lower rates of postoperative CSF leaks in patients receiving an LD. ¹⁸ However, the majority of published studies have demonstrated no appreciable differences in postoperative CSF leakage rate between the groups, ^{14 19 20 21} while some have even reported higher complication rates in patients undergoing upfront LD placement. ^{14 22 23} In a prospective randomized controlled trial of 38 patients comparing the upfront placement of LD for anterior endoscopic approaches, Huo et al reported no differences in postoperative CSF leakage rates between patients in which an LD was placed at the beginning of the case versus those who did not have an LD placed but observed significantly higher rates of pneumocephalus, meningitis, hypertensive headaches, subdural hygroma, and LD disconnection/blockage/replacement. ¹⁴ These previously mentioned studies do evaluate LD usage in a variety of neurosurgical and skull base etiologies, but none of these studies look specifically at MCF for nonneoplastic lesions and few represent as large a patient sample.

Operating Room Time and Length of Stay

While not reaching statistical significance, there was a trend toward increased OR time and longer LOS in the LD group compared to the non-LD group. While prior studies have compared OR time between open and minimally invasive approaches to tegmen defects, few studies to date have examined OR time with respect to LD use in MFC alone. ²⁴ Within our cohort, OR time was approximately 20 minutes longer in the LD group; however, this difference did not reach statistical significance. While the cost implications associated with this difference in OR time are beyond the scope of this study, it is an appropriate future direction for follow-up studies examining the utility of LD in MFC. In addition, our study found no significant difference in LOS between the LD and non-LD groups; existing literature remains heterogenous on the relationship between LD usage and LOS. While an increased LOS was observed in a retrospective cohort study by Nelson et al comparing LD versus no LD in MCF for spontaneous CSF leak repair, this study was limited by a small sample size of 60 patients, with only 15 patients undergoing LD placement. ⁵ Huo et al also demonstrated an association between LD placement and longer LOS in a cohort of 38 patients ¹⁴ ; however, these patients underwent anterior endoscopic skull-base surgeries versus the MFC approach described in our cohort. Future follow-up studies validating the results of this study are indicated to determine whether LD placement prolongs LOS in nonneoplastic patients undergoing MFC.

Limitations and Future Directions

The present study is not without limitations. First, the retrospective nature of this study subjects it to standard biases associated with this design. Many patients travel a significant distance to our institution and prefer to conduct routine follow-ups with an otologist closer to their home. While these patients do return to our center for major complications such as recurrent CSF leaks or those requiring hospitalization, this represents an inherent limitation to our study design. Given that the patients in our study were treated by six neuro-otologists and four neurosurgeons at our institution, varying surgeon preferences may have contributed to selection bias in the study. Although our referral practices assign patients based on surgeon availability, this does not completely account for confounders that may be present based on individual surgeon techniques and preferences. Additionally, we found that women had an LD placed significantly more than men which is unexplained and may be a result of the small sample size of our cohort. Furthermore, given the retrospective nature of the study, important variables were unable to be accounted for in our analysis. We compared outcomes based on three pathologies—SSCD, encephalocele, and CSF leak—but these pathologies may not have similar outcome profiles. Differences between pathologies may make our findings less robust, although these pathologies have been studied together in other literature. ²⁵ Although the sample size is robust compared to present studies in the literature on the subject, a prospective multi-institutional study with a greater number of patients is indicated to validate the findings of this study. Given the low number of complications observed in this cohort of patients, a very large group of patients is required to have appropriate power for statistical significance. Furthermore, no statistically significant differences were observed in OR time between the two groups; this may reflect the study being performed at an academic institution with variable trainee experience serving to skew operative times. Also, data regarding nerve root injury or postprocedural back pain were not reported as this did not appear to be well documented in the electronic medical record. Finally, CSF hypotension headache is a known and common complication of LD use, and our study was limited by the retrospective nature of data collection, possibly having missed cases of these complications that were not documented. However, we ensure that charts were reviewed thoroughly for any mention of headaches or when an intervention was recommended or provided for such. Despite these limitations, our study presents the first direct comparison of demographics and preoperative symptomatology, as well as perioperative variables and postoperative outcomes, between LD and non-LD patients undergoing MFC for nonneoplastic etiologies.

Conclusion

In patients undergoing MFC for nonneoplastic etiologies, LD placement was not associated with statistically significant differences in postoperative complications, CSF leakage rates, necessity for permanent CSF diversion, or need for reoperation. While there was a trend toward longer OR time and hospital LOS in the LD group, there was no statistical difference in OR time or hospital LOS between LD and non-LD groups. Taken together, our data suggest that intraoperative LD placement is safe and not associated with increased postoperative complications, CSF leak rates, permanent CSF diversion, and need for reoperation. Further inquiry with a prospective, multi-institutional cohort is indicated to validate these results.