Introduction

Epidermoid cysts (EC), also known as pearly tumors or cholesteatomas, are rare benign tumors that represent 0.2 to 1.8% of all intracranial tumors [1]. They are congenital lesions caused by abnormal trapping of epidermal elements during the processes of neural groove closure and disjunction of surface ectoderm, which occur between the 3rd and 5th weeks of fetal life [2]. They display a thin epithelial-lined capsule, and their growth is mainly due to the accumulation of desquamated epithelial cells, forming shiny pearly debris, in addition to the secretion of cholesterol and keratin into the cyst. Cerebellopontine angle (CPA) cysts, being the most common location for posterior fossa EC, constitute approximately 40–50% of all intracranial EC. They represent the third most common CPA tumors after acoustic neuromas and meningiomas, comprising approximately 7% of CPA pathology [3, 4].

EC grow slowly and may penetrate to every accessible space around its origin. EC originating from the petroclival area (PCA) and the CPA share common behavior and surgical challenges. Cranial nerve (CN) disorders are the foremost symptoms in both locations. Microsurgical resection is the only effective treatment for these lesions when eligible for an active treatment [5]. Previous reports have already studied the correlation between total removal and tumor recurrence. Some authors found a higher rate of recurrence after non-total resection [3, 6,7,8], while other reports found no difference between GTR and PR [9, 10]. Moreover, little is known about the kinetic of regrowth after incomplete surgical resection.

Some other studies have focused on functional outcomes and the course of CN deficits after surgery [9, 11,12,13,14]. However, there is no existing report presenting detailed monitoring of the evolution of CN deficits over time.

The aim of this multicenter study was to assess the long-term oncological and functional surgical outcome in CPA/PCA EC patients.

Methods

All consecutive patients who underwent surgical treatment for a CPA or PCA EC between January 2001 and July 2019 in six participating French cranial base referral centers were included in this observational retrospective cohort study. The centers involved were North University Hospital, Grenoble University Hospital, Pitié Salpêtrière Hospital, Rennes University Hospital, Angers University Hospital, and La Timone Hospital.

This study has been approved by the French Neurosurgical College Institutional Review Board (reference: IRB00011687 College de neurochirurgie IRB #1: 2020/04).

Data collection

Demographics, clinical manifestations (presenting signs and symptoms, duration of symptoms, CN function), radiological investigations (location, extension, laterality, size, existence of a diffusion-weighted magnetic resonance imaging (DWI) sequence, complications), surgical procedure (surgical approach, extent of removal), postoperative imagery, postoperative complications (occurring within 30 days after surgery), neurological outcome, radiological progression/recurrence, and possible second surgery were retrospectively reviewed. Tumor size was assessed based on their largest diameter in MRI.

PCA cysts were distinguished from CPA cysts by their primary origin in the upper two-thirds of the clivus or the petrous apex, medial to the trigeminal nerve. Both could then extend forward or backward to the cavernous sinus, Meckel’s cave, the sellar region, the incisure of the tentorium, the porus, and the ventral edge of the foramen magnum.

Surgical protocol: functional sparing technique

The main goal of the surgery was to achieve optimal and safe cyst resection, without compromising neurovascular structures (functional sparing technique). The intraoperative decision to interrupt the resection was based on the evidence of critical adhesion of the tumor capsule to CN, vascular structures, or brain stem. In this situation, small fragments of the capsule were left in place to avoid the risk of neurological deficits. The surgical approach was customized for location and cyst extension and depended on the experience of each center. Neuronavigation and CN neurophysiological monitoring could be used by the surgeon according to his preoperative evaluation. The suboccipital retrosigmoid (RS) (Fig. 1) and the epidural anterior transpetrosal (TP) approach (Fig. 2) were the most frequently used. In the case of cyst extension above the tentorium, the resection could be combined with a pterional or a subtemporal approach. A median suboccipital telovelar approach could also be performed when the lesion had spread into the fourth ventricle.

Fig.1
figure 1

Illustrative case of a cerebellopontine angle epidermoid cyst operated on through a retrosigmoid approach. A & B The preoperative DWI Diffusion and T2-weighted MRI showed a left cerebellopontine angle epidermoid cyst extended to the petroclival region in a 30-year-old woman. The tumor was incidentally detected in imaging studies after a traumatic brain injury. The patient was free from any symptom before surgery. C Peroperative view. The tumor was resected through the corridor between the acoustico-facial bundle and the lower CNs. The cranial nerves are exposed. D & E No CN disturbances occurred in the postoperative period. The DWI diffusion and T2-weighted MR performed 3 months after surgery did not reveal any tumor residue and remained stable 2 years after surgery

Full size image
Fig. 2
figure 2

Illustrative case of a petroclival epidermoid cyst operated on through a transpetrosal approach. A & B The preoperative T2-weighted MRI showed a right petroclival epidermoid cyst (Fig. 2A) in a 44-year-old man. He presented with a slight numbness of CN V2 & 3 and reported trigeminal neuralgia for 2 years. The superior pole of the tumor extended to the perimesencephalic cistern (B). The tumor was responsible for a mass effect on the brainstem and the ipsilateral middle cerebellar peduncle. C & D Peroperative views. The operative window after epidural anterior transpetrosal approach offered an exposure from the trigeminal nerve to the basilar and the superior cerebellar artery. E & F The patient presented postoperative transient diplopia due to CN IV disturbances and retained long term CN V3 facial hypoesthesia. The DWI diffusion and T2-weighted MR performed 3 months after surgery showed small residual fragments of tumor located at the level of perimesencephalic cistern. Those remnants experienced asymptomatic regrowth 5 years after surgery. A wait-and-rescan policy was decided

Full size image

Postoperative assessment

Clinico-radiological follow-up was planned at 3, 6, and 12 months after resection, and subsequently at 2, 3, 5, 7, and 10 years after surgery, and then once every 3 years thereafter.

The function of each CN involved in the posterior fossa was analyzed in the postoperative period and at every follow-up visit. The Barrow Neurological Institute (BNI) classification [15] was used to assess the trigeminal neuralgia, and the facial nerve function was scored according to the House and Brackmann (HB) classification [16]. Preoperative CN deficit improvement was defined as the reduction or resolution of one or more preoperative CN deficit(s) between the preoperative period and the end of follow-up.

The patient’s functional status evolution was assessed with the World Health Organization performance status score (WHO PS) [17] at each follow-up consultation.

The postoperative cystic residue was assessed by MRI scan at 3 months after surgery. The extent of resection (EOR) was considered total (gross total resection, GTR) if keratinous debris and the entire tumor capsule were removed, and was confirmed with the absence of postoperative cystic residue in DWI on the first postoperative MRI. Subtotal resection (STR) was defined as complete content removal and incomplete capsule removal, with a small residue on postoperative MRI. Partial resection (PR) was defined as incomplete resection of both cyst contents and capsule, with postoperative residue on postoperative MRI.

Progression or recurrence of the lesion was based on the apparition or increase of the EC size on the DWI sequences of the follow-up MRI scan.

Statistical analysis

All statistical analyses were conducted using IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp.

Categorical variables were presented as numbers and percentages, and continuous variables were presented as mean and standard deviation or median and range (minimum–maximum), as appropriate. Analyses were tailored to address associations among surgical data (surgical approach, EOR, tumor location, and tumor size) and the occurrence of preoperative CN deficit improvement, new postoperative CN deficit, and tumor progression/recurrence. Univariate analyses were performed using Pearson’s chi-squared test or Fisher’s exact test for categorical variables and Mann–Whitney U test for continuous variables, as appropriate. The progression-free survival after surgery and the time between progression and the second surgery were presented as Kaplan–Meier plots. A two-sided p-value of less than 0.05 was considered to indicate statistical significance. Adjusted p-value < 0.004 was considered statistically significant after Bonferroni correction for multiple comparison tests.

Results

Population study

A total of 56 consecutive patients were enrolled in this study, including 29 (52%) females and 27 (48%) males, with a median age of 38 years (20–71) at the time of diagnosis. Six patients (11%) had previously been operated on in other centers. The median duration of symptoms before diagnosis was 6 months (0–300).

Clinical and radiological features (Table 1)

Table 1 Clinical data and tumor characteristics
Full size table

Fifty-four patients (96%) were symptomatic at the time of the diagnosis. Preoperative CN disorders were present in 47 patients (84%). The foremost symptoms were the cochleo-vestibular impairment (25 patients—45%): 15 patients reported hearing loss, and 15 patients presented with vestibular deficits (balance disorder/tinnitus/vertigo). Fifteen patients (27%) suffered from trigeminal neuralgia, and 15 patients (27%) presented with facial hypoesthesia. Lower cranial nerve (LCN) disorders were observed in 9 patients (16%). Facial nerve deficits were present in 8 patients (14%) preoperatively, mostly HB grade II (11%, n = 6). Seven patients (13%) presented one or more oculomotor dysfunction in the preoperative period: there were 3 CN III (5%), 3 CN IV (5%), and 5 CN VI (9%) deficiencies.

Sixteen EC (29%) were mainly localized in the PCA, and 40 EC (71%) were localized in the CPA.

Surgical features (Table 2)

Table 2 Surgical and postoperative data
Full size table

RS approach was the most frequently performed, especially for the CPA location: 75% of CPA EC were removed through this approach, 3% were resected through a pterional approach, 10% through a TP approach, 10% through a subtemporal approach, and 3% through a suboccipital approach (n = 30, 1, 4, 4, 1/40, respectively). In the PCA location, the surgical approach was less homogeneous. A RS approach was performed in 38% of them, a pterional approach in 44% of them, a TP approach in 13% of them, and a subtemporal approach in 7% of them (n = 6, 7, 2, and 1/16, respectively).

GTR, STR, and PR were performed in respectively 14% (n = 8), 61% (n = 34), and 23% (n = 13) of patients. The EOR was assessed by the surgeon at the end of the procedure in all cases, according to the criteria defined above. The surgical perioperative evaluation was also confirmed by a DWI sequence on the 3-month postoperative MRI when available (42 patients—75%). For the remaining 14 patients, the surgery occurred before 2010 and no 3-month postoperative MRI was performed then.

The most frequent postoperative complications were aseptic meningitis (13%, n = 7), and hydrocephalus (7%, n = 4). A 28-year-old woman died in the postoperative period from a postoperative cerebellar hematoma.

Functional outcome

  1. 1.

    Preoperative CN deficit

    1. 1.1

      Course of preoperative CN deficit (Table 3)

      Among the 47 patients with CN disorders before surgery, 34 experienced a CN improvement during follow-up (72%, Fig. 3A).

      At the end of follow-up, 60% of the cochlear and vestibular preoperative dysfunction improved or were completely resolved (n = 9/15 in both groups). Six patients retained hearing loss or deafness, and 6 had a persistent vestibular deficit at the end of the follow-up.

      Respectively 67 and 53% of trigeminal preoperative neuralgia and hypoesthesia were resolved at the last follow-up control (n = 10/15 and n = 8/15). Among the remaining five persistent neuralgia (9%), three (5%) decreased according to the BNI classification (Fig. 4).

      Preoperative LCN disorders disappeared throughout the follow-up in 44% of cases (n = 4/9).

      In the same way, an improvement of preexisting facial nerve deficits was seen in 38% of cases (n = 5/8).

      Forty-three percent of patients with preoperative oculomotor nerve deficit were asymptomatic at the end of follow-up (n = 3/7), while 57% of patients (n = 4/7) retained an oculomotor deficit (CN VI: n = 2, CN IV: n = 1 and CN III: n = 1).

    2. 1.2

      Prediction of preoperative CN deficit improvement at last follow-up consultation

      Eighty-eight percent (n = 7/8) of patients who received GTR experienced an improvement in their initial CN deficit, compared to 53% (n = 18/34) for STR and 69% (n = 9/13) for PR (p = 0.196). Patients who experienced postoperative improvement of their CN deficit harbored smaller tumor volumes than those who remained stable over time (median 40 mm (range 17–72) vs median 45 mm (range 24–82); p = 0.127). There was no significant association between the surgical approach or the EC location and the improvement of preoperative deficit (respectively 67% of improvement for TP vs 64% for RS approach (p = 1.00) and 65% of improvement for CPA vs 50% for PCA location (p = 0.369)).

  2. 2.

    New postoperative deficit

    1. 2.1

      Course of  new postoperative CN deficit (Table 4)

      New neurological CN deficits occurred during the immediate postoperative course in 27 (48%) patients (Fig. 3B). Among the patients with oculomotor CN, vestibular CN, and LCN postoperative disorders, only 2 still experienced an oculomotor deficit at the end of follow-up, and none retained LCN or vestibular CN impairment anymore.

      In contrast, only 14% of patients with postoperative cochlear deficit improved at the end of follow-up (n = 1/7). To resume, 44 patients were free of cochlear deficits at the last follow-up examination, including 9 preoperative cochlear deficits and the new postoperative deficit that improved.

    2. 2.2

      Prediction of early postoperative CN deficit

Table 3 Preoperative CN deficits evolution after the surgery
Full size table
Fig. 3
figure 3

Evolution of cranial nerves deficit. A Evolution of cranial nerves preoperative deficit. Line chart representing the proportion of patients with CN deficit during follow-up among those with preoperative CN disorders. Only the most frequent CN deficits are illustrated: oculomotor (●), trigeminal hypoesthesia ( +), trigeminal neuralgia (✕), and cochlear nerve (▼). B Evolution of new postoperative cranial nerves deficit. Line chart representing the proportion of patients with new CN deficit during follow-up. Only the most frequent CN deficits are illustrated: oculomotor (●), cochlear (✕), and lower cranial nerves ( +)

Full size image
Fig. 4
figure 4

Evolution of preoperative trigeminal neuralgia. Bar chart illustrating the evolution of preoperative trigeminal neuralgia during follow-up, according to the Barrow Neurological Institute (BNI) Classification (from grade 1 to grade 5). Grade 1 defines asymptomatic patients, who are not shown is this figure. No grade 5 patients were identified in our cohort

Full size image
Table 4 New CN postoperative deficits and their evolution
Full size table

In the univariate analysis, patients who underwent a TP approach had an increased risk of new postoperative deficit compared to the ones treated with a RS approach (100% vs 42%, p = 0.020). All of them were oculomotor deficits, including 2 CN III deficits (1 transitory deficit and 1 persistent at discharge), 5 CN IV deficits (4 transitory deficits, and only one persistent at discharge), and 2 CN VI transitory deficits. When a RS approach was performed, the new postoperative oculomotor rate was only 22% versus 100% after a TP approach (p = 0.001). After Bonferroni correction, significant differences (adjusted p value < 0.004) only existed between the TP approach and oculomotor rate. Tumor size, location, and extent of resection were not statistically associated with de novo early postoperative CN deficits (Table 5).

Table 5 Predictive factors
Full size table

Functional status

At the last follow-up examination, 49 patients (88%) were fully independent (WHO PS 0 and 1) compared to 45 patients (80%) in the postoperative period. The number of ambulatory patients unable to work (WHO PS at 2) decreased over time, with 3 patients (5%) at the end of follow-up compared to 8 patients (14%) in the early postoperative period. Only 2 patients displayed poor general condition (WHO PS 3 and 4) at the end of follow-up (4%). Finally, two patients died during follow-up: one in the postoperative period as described previously and the other after 45 days of intensive care for postoperative cerebral ischemia. To note, the second comatose patient survived and was ambulatory at the end of follow-up (WHO PS at 2).

Tumor control

No patient was lost to follow-up in the first 1-year postoperative period, except for the deceased patient. The median clinical and radiological follow-up were 46 months (0–409). Twenty-six patients (46%) showed evidence of tumor progression or recurrence during the follow-up period, after a median duration of 63 months (7–210). Fifty percent (n = 13) of the patients presenting with EC progression underwent a second surgery. Re-operated patients were symptomatic in 85% of cases before second surgery (n = 11/13). The remaining two patients were operated on again because of the rapidity of the cyst expansion. The median delay between the recurrence and the new surgery was 5 months (0–41) (Fig. 5).

Fig. 5
figure 5

Tumor control. A Kaplan–Meier plot illustrating progression-free survival. The shaded areas indicate 95% CI. B Kaplan–Meier plot illustrating the delay between progression/recurrence and a second resection surgery. The shaded areas indicate 95% CI

Full size image

RS approach and smaller tumor size tended to be associated with less recurrences (p = 0.672 and 0.245 respectively, see Table 5). Postoperative residual volume was not associated with tumor progression or recurrence: 50% of GTR patients experienced tumor regrowth, as well as 50% of STR patients and 31% of PR patients (p = 0.394). Similarly, the CPA or PCA location of the EC did not influence the recurrence rate (p = 1).

Discussion

The kinetics of postoperative recovery of CN disorders is an original aspect of this work, which demonstrates the favorable effect of surgery on the functional outcome of CN function, with a low rate of long-term morbidity since postoperative CN deficits were mostly transient.

This study is one of the largest surgical series of CPA-PCA EC published so far, with a total of 56 patients. The literature dealing with the same topic is scarce and only a few retrospective cohort studies include more than 30 patients [1, 3, 7, 9, 18,19,20].

CN preoperative deficit

We report here the very good evolution of preoperative CN disorders after surgery, with an improvement of 72% of preexisting deficits, regardless of the approach or the location (PCA or CPA) of the cyst.

Both the inflammatory effect induced by the cystic content and the direct mechanical compression over the cisternal segment of the CN by the tumor may explain preoperative CN disorders favorable outcome after surgery, even with small volume cysts. Contrary to schwannoma or meningioma cases wherein the CN are shift by the boundaries of the tumor capsule, EC invade the cisternal space by adapting their shape to the local morphology, encompassing nerves and vessels. Thus, surgery might improve symptoms by reducing cystic content and inflammatory process over the nerve in addition to reducing the mass effect. In this way, GTR has a greater impact than STR and PR on the improvement of preoperative CN deficits in our series (88% of improvement VS 53% and 69%, p = 0.196): maximal resection of cystic content and capsule fragments results in a higher reduction of local irritation. The effects of local irritation by the cholesterol seeping through the cyst wall have already been reported in the previous study, in cases of hyperactive dysfunction such as trigeminal neuralgia or hemifacial spasm [1, 13]. Vascular compression of the nerve, either by a displaced artery or by nerve displacement toward the artery by the tumor, has also been relieved [1, 9].

The beneficial effects of surgery on these preoperative CN deficits had also been demonstrated in several series. Of the 17 CPA EC reported by Czernicky et al., 11 patients experienced improvement or resolution of their preoperative deficits, in particular with trigeminal neuralgia, LCN deficits, and facial nerve deficits [12]. In their cohort of 37 EC patients, Gopalakrishnan et al. demonstrated a significant improvement in trigeminal and lower cranial nerve dysfunction after surgery and half of the CN VIII and oculomotor deficits [3]. Of the 21 patients with preoperative CN dysfunction reviewed by Schiefer and Link, 33% were resolved (n = 7) after the surgery and 43% were improved (n = 9) [10]. In Vernon et al. series of 139 patients, 74% of them improved compared with their preoperative clinical status. Prior to our series, none of these studies attempted to analyze the association between these CN improvements and surgical approach or EOR. Of note, none of them included PCA location of the cyst, and the TP approach was not used.

Postoperative new CN deficit

New CN deficits occurred during the immediate postoperative course in almost 50% of our patients. Most of them, apart from cochlear impairment, tended to resolve during the follow-up period. The unique cisternal cytoarchitecture of the VIII CN (i.e., centrally myelinated) could explain its higher surgical vulnerability in comparison to the other CN. Moreover, postoperative CN VIII impairment is related to direct nerve dissection or vasculature damage during surgery because of the adhesion between the lesion and the nerve and is therefore less likely to recover than preoperative CN VIII deficit [21].

TP approach tended to increase postoperative impairment, especially oculomotor deficits (p = 0.001). However, most of them were transient and the TP approach remains advantageous for some PCA locations, particularly when the cyst crosses the midline or straddles the basilar artery. Thus, this approach must remain systematically considered to ensure the best surgical exposure.

The high rate of postoperative CN injury, resulting from the adhesion of the tumor capsule to the nerve, is a well-known complication of this surgery [18]. Very few studies have detailed the evolution of the CN deficit over time [1, 3, 9, 10, 12, 18, 19]. Vernon et al. reported 41% of new postoperative deficits, which resolved completely on long-term follow-up in 21% patients, improved significantly in 10% patients, and remained at an unchanged level in 9% of patients. Czernicki et al. provided comparable results in terms of frequency and evolution of CN deficit, with 58% of new postoperative CN injury (n = 10) and persistent cochlear deficits during follow-up. Similar to our series, the EOR had no effect on the occurrence of new postoperative deficits [12]. In the Gopalakrishnan series of CPA EC, only 13% (n = 5) of the 38 new postoperative deficits persisted at long-term follow-up [3]. They reported a higher incidence of new neurological deficits in patients undergoing total removal compared to subtotal removal, but at the same time, the former group experienced a better improvement in preoperative neurological deficits compared to the latter one [3]. Two series have reported a lower rate of postoperative deficit [1, 18]. The cohort of 30 patients undergoing retrosigmoid surgery associated with whole course neuroendoscopy of Hu et al. experienced only 7% of new deficits [18]. Finally, two papers have reported a higher rate of new postoperative deficit, but with a very good improvement over time [9, 10]. Despite these differences in terms of postoperative CN deficit, our cohort covered a larger group of accurately and sequentially monitored patients than previously reported, which was the main objective of this work, and strengthen our findings.

General condition

The long-term general condition was good to excellent for most patients. Only 2 patients presented a WHO PS score at 3 or 4 (capable of limited self-care or completely disabled) at the end of follow-up. In the literature, two other series have recorded long-term general condition, and their results were similar [10, 12]. Eighty-eight percent of patients were able to carry out all usual activities (modified Rankin score of 0 or 1) in Czernicki’s cohort and 84% for Schiefer’s cohort. These results are linked to a cautious surgical strategy, avoiding maximalist resection of the fragments adhering to the brain stem, vessels, or nerves. The fact that EC are mostly managed in young and healthy patients favorably outweigh the outcome.

Tumor control and onco-functional balance

Twenty-six patients (46%) showed evidence of tumor progression during the follow-up period, after a median duration of 63 months. The factors that might predict a stable behavior instead of a keep growing one could not be unveiled by our study. Based on our experience, patients with postoperative residual lesions had no increased risk of progression or recurrence during follow-up (p = 0.394). This result could be due to many biases inherent to the design of the study and to the short follow-up of only 46 months. In our work, GTR rate (14%) is lower and recurrence rate (46%) higher than those previously reported for posterior fossa EC: Farhoud et al. (32 patients, 59% GTR rate, 0 recurrence), Samii et al. (40 patients, 75% GTR rate, 8% recurrence rate), Kobata et al. (30 patients, 57% GTR rate, 7% recurrence rate), Vernon et al. (139 patients, 73% GTR rate, 8% recurrence rate), Yawn et al. (47 patients, 46% GTR rate, 8% recurrence rate), or Schiefer et al. (24 patients, 54% GTR rate, 25% recurrence rate) for instance [1, 7, 9, 10, 19, 20]. The recurrence rate in the Gopalakrishnan et al. series was higher after long-term follow-up: 45% of patients showed evidence of tumor recurrence after a mean duration of 9.3 years [3]. Indeed, our definitions of GTR and recurrence were strict, and non-total resections and recurrence rates could have been overestimated.

Postoperative tumor control results are heterogeneous in the literature. Like our study, some reports found no difference in tumor progression after complete or incomplete resection, even after longer follow-up periods (respectively 11.5 years and 4.3 years of follow-up) [9, 10]. In contrast, some authors found a higher rate of recurrence after non-total resection [3, 7, 8]. A recent large meta-analysis including 691 patients with intracranial epidermoid tumors found that STR was associated with a 7 times higher rate of regrowth than tumors that underwent GTR [8]. However, the analyses were not stratified according to the intracranial location of the cysts, such as infratentorial sites. Moreover, capsule adherence to the neurovascular structures in PCA and CPA locations makes total removal extremely challenging. Additionally, the propensity of EC for regional spreading toward neighbor cisterns hampers the ability to expose the full volume of the cyst “behind the corner” using regular approaches. The use of angled endoscopes (endoscopic assisted microsurgery) could be of help to check for fragments that could be overlooked under a microscope, as suggested above [18, 22].

These findings confirm that altogether, operative findings and high field MR do not reach the level of sensibility to insure the cure of the disease. EC grow linearly, not exponentially [23], and the overlooked micro-fragments of cyst walls have the propensity to regrowth which is empirically known [1, 10, 12,13,14, 18, 24].

Our results (see Kaplan–Meier plot, Fig. 5A) underline that recurrence or regrowth might be expected in the ultra-late period, regardless of the EOR of the cyst. Patients should be aware and a clinic-radiological sequential follow-up must be planned in long term. Only half of the growing residual tumors justified additional surgery (see Kaplan–Meier plot, Fig. 5B). The expectation for clinical symptomatology in 85% of re-operated patients can explain the half-year delay before considering surgery.

Weaknesses

This is a retrospective study, and so there may be missing data in our work. Also, the center effect has not been tested. Multivariate analysis and cox-regression analysis were not performed, because it is not appropriate to this limited patient sample. We deliberately merged the findings of PCA and CPA EC in our work; these locations are not supposed to carry the same risks when surgically approached. However, testing this variable did not impact the functional nor the oncological results.

Finally, it would have been interesting to obtain precise volumes and growth rates of the preoperative and residual EC. DWI sequence is essential for the diagnosis of EC and differentiation from other lesions, as the content of the epidermoid cysts shows prominent diffusion restriction due to the layered microstructure of the debris [12]. However, this sequence is mediocre in terms of anatomical resolution and did not allow precise calculation of cyst size. The interest of new sequences coupled to the DWI in order to increase the three-dimensional resolution while keeping an important sensitivity and specificity for the diagnosis of EC is an axis to develop in the future.

Last, the length of follow-up did not reach the long term which weaken the analysis of tumor growth potential over time.

Conclusion

This multicenter retrospective cohort study of large posterior fossa EC demonstrates the favorable impact of a functional sparing surgery policy over preexisting CN deficits. In addition, the high rate of early new postoperative CN deficits observed in this study underlines the complexity of such surgery. Most of these disturbances recovered, except for the cochlear nerve. The counterpart of this surgical strategy was the evidence of tumor remnant or recurrence in half of the study group, which was demonstrated by a regular sequential MR follow-up protocol. The linear progression of EC and the need of additional surgery in half of the recurrent patients support the need for a lifetime DWI MR surveillance, which need to be early mentioned to the patient.