This retrospective study included patients with intracranial petrous apex lesions who underwent surgical treatment in the Department of Neurosurgery at the First Affiliated Hospital of Bengbu Medical University between January 2020 and August 2025. All patients were diagnosed with petrous apex lesions via preoperative magnetic resonance imaging (MRI). The cohort comprised 10 male and 17 female patients. Surgical approaches included the Kawase approach (14 patients), the retrosigmoid approach combined with neuroendoscopy (7 patients), and the pterional approach (6 patients). Postoperative pathological examination revealed meningioma in 20 patients, trigeminal schwannoma in 3 patients, cyst in 2 patients, chordoma in 1 patient, and cholesteatoma in 1 patient.

Inclusion criteria:

Exclusion criteria:

Prior to surgery, a team of neurosurgical specialists evaluated the tumor location to determine the most appropriate surgical approach. The team then comprehensively explained the specific surgical plan to the patient and their family, including the advantages, risks, and prognosis associated with the chosen approach. This ensured the patient fully understood the treatment strategy and provided informed consent.

The surgical approach was selected through standardized multidisciplinary discussions involving senior neurosurgeons, neuroradiologists, and neuropathologists. The final decision was primarily based on the following imaging and clinical criteria:

All surgical procedures in this study were performed by the same neurosurgical team at our institution, led by a senior neurosurgeon with extensive experience in skull base surgery. For each case, the surgical approach was determined through multidisciplinary discussion within this team. The primary surgeon participated in procedures utilizing all three approaches. This standardized protocol was implemented to minimize operator bias and to ensure that the observed outcomes primarily reflected the inherent characteristics of the surgical approaches themselves, rather than inter-surgeon variability.

Data analysis was performed using SPSS Statistics version 31.0. Continuous variables conforming to a normal distribution are presented as mean ± standard deviation (x̄ ± SD). Due to the small overall sample size (n < 30), the Kruskal-Wallis test (a non-parametric test) was used for continuous variables, and Fisher–Freeman–Halton’s exact test was used for categorical variables. A p-value of less than 0.05 was considered statistically significant.

Follow-up data were collected by reviewing patients’ outpatient follow-up records, supplemented by telephone interviews when necessary. The follow-up period was calculated from the date of discharge to the date of the last available clinical or imaging assessment. As this was a retrospective study, not all patients were subject to a standardized, protocol-driven follow-up schedule. Consequently, the availability and timing of follow-up assessments were inconsistent across the cohort. The primary follow-up measures collected included.

The major complications evaluated in this study were defined as follows.

This study included 27 patients. Their detailed baseline characteristics, grouped according to the three surgical approaches, are presented (Table 1). In this study, the selection of surgical approach strictly adhered to the principle of “anatomical location first.” Consequently, as shown in Table 1, there were expected and systematic differences in the core anatomical localization of the tumors among the three patient groups. All patients who underwent the Kawase approach had lesions located in the anteromedial petrous apex (14/14, 100%), with frequent transtentorial extension (42.9%). All patients who underwent the retrosigmoid approach had lesions located in the posterior petrous apex (7/7, 100%). All patients who underwent the pterional approach had lesions located in the superoanterior petrous apex (6/6, 100%), with frequent cavernous sinus involvement (66.7%). Despite these clinically determined anatomical differences, the three groups did not differ significantly in terms of age, sex, tumor laterality, maximum tumor diameter, or preoperative key neurological functional status (e.g., hearing loss, facial numbness) (all p > 0.05). An identical anatomical distribution pattern was also observed in the largest pathological subgroup of meningioma patients (Table 2). This provides a relevant baseline for subsequent observation of outcome differences among the surgical approaches.

Perioperative clinical outcomes for the three patient groups, stratified by the anatomy-driven surgical approach, are summarized in Table 3.

Given the distinct growth patterns and invasiveness of different pathological types toward surrounding structures, we further examined the surgical outcome characteristics of each pathology under the principle that “location determines approach.” This allows us to appreciate how pathological type interacts with specific “location-approach” pairings, thereby influencing therapeutic outcomes (Table 4). In parallel, we focused on analyzing the largest subgroup—meningiomas (Table 2)—to provide clearer descriptive insights.

Among the 27 patients, postoperative follow-up data were available for 21 patients (Kawase approach group: n = 11; retrosigmoid approach group: n = 5; pterional approach group: n = 5). The median follow-up time varied across groups (Kawase: 16 [3–36] months; retrosigmoid: 10 [3–18] months; pterional: 12 [6–24] months) (Table 5).

Regarding the long-term outcomes of complications, the following were observed: For facial nerve function, among the 3 patients in the pterional group who developed this complication and were followed up, 2 showed improvement. In the Kawase and retrosigmoid groups, 1 patient each developed and was followed for this complication, and both showed improvement. For trigeminal nerve function, 1 patient in the Kawase group who developed the complication and was followed up showed improvement. In the pterional group, only 1 of the 2 patients with this injury showed improvement. Other complications, such as motor weakness and speech impairment, showed improvement in all cases. None of the 4 CSF leak cases or the 7 intracranial infection cases that occurred and were followed in the Kawase group recurred during the follow-up period.

Regarding tumor progression and recurrence, imaging progression was observed in 1 patient in the pterional group (pathology: meningioma) during follow-up. No definite recurrence was observed in the other groups within the available follow-up time.

Accurate identification of Kawase’s triangle (the posteromedial middle fossa triangle) is crucial for the Kawase approach. Its boundaries are defined medially by the trigeminal nerve (V3 segment) and the greater superficial petrosal nerve (GSPN), laterally by the arcuate eminence and the superior semicircular canal, inferiorly by the petrous segment of the internal carotid artery, and superiorly by the trigeminal ganglion (Gasserian ganglion). Kawase’s triangle represents a safe zone for bone drilling (10–12).

Removing the bone within Kawase’s triangle using a drill and craniotome creates a direct communication between the middle and posterior cranial fossae. This corridor not only provides ample room for instrument manipulation but also yields extensive exposure. It facilitates access to lesions located in the anteromedial petrous apex, the petrous segment of the internal carotid artery, the mid to upper clivus, the ventrolateral brainstem cisterns, as well as meningiomas involving both supra- and infratentorial compartments and tumors situated anteromedial to the superior cerebellar surface.

A key advantage of this approach lies in its facilitation of exposure and protection of critical cranial nerves such as the trigeminal and facial nerves. During the procedure, it allows for the full exposure of these nerves along their relevant courses. This enables clear identification and the systematic release of any tumor-induced compression, contributing to more radical resection (13–15).

The Kawase approach group exhibited a high rate of intracranial infection (64.3%, p < 0.05). This postoperative complication may be attributed to both the technical demands of the approach—including its complexity, longer operative time, and opening of pneumatized petrous air cells—and the characteristics of the lesions themselves: all patients in this group had lesions in the anteromedial petrous apex with frequent transtentorial extension, necessitating more extensive bone drilling, which likely increased the risk of infection. Consequently, this elevated infection rate should be regarded as an inherent risk associated with employing the Kawase approach for complex anteromedial petrous apex lesions. Furthermore, cerebrospinal fluid (CSF) leaks occurred exclusively in the Kawase group (28.6%), which is directly related to the need for meticulous dural closure after opening petrous air cells to prevent fistula formation (12, 16). Although the incidence of postoperative infection was high, no recurrence was observed during follow-up after appropriate management, indicating that such complications are controllable with active treatment.

The retrosigmoid approach combined with neuroendoscopy provides considerable visual access to the cerebellopontine angle and its surrounding neurovascular structures. It offers excellent exposure of the posterior cranial fossa, cerebellopontine angle, internal auditory canal, ventral brainstem, superior petrosal vein, and cranial nerves V through XII. Furthermore, neuroendoscopy facilitates clear visualization of the relationship between the tumor and adjacent vessels, effectively expanding the surgical field of view (17, 18).

For patients with posterior petrous apex lesions in this study, this technique allows for better protection of the facial, vestibulocochlear, and trigeminal nerves, helps reduce vertigo caused by vestibular apparatus injury, and minimizes overall surgical trauma. While the retrosigmoid approach combined with neuroendoscopy is highly effective for tumors located posteromedially and those extending into the cerebellopontine angle, its utility is limited for tumors situated in the anterior petrous apex, closely adherent to the clivus or the posterior portion of the cavernous sinus. In such cases, the line of sight via the retrosigmoid approach becomes obstructed, significantly increasing the technical difficulty of the procedure (19, 20). This further confirms that selecting the appropriate surgical approach based on the core anatomical location of the lesion is fundamental to achieving optimal exposure and minimizing postoperative complications.

The pterional approach is widely utilized in lateral skull base surgery. It provides extensive exposure of the sylvian fissure, sellar region, anterior clinoid process, optic nerve, internal carotid artery, and its bifurcation. This approach is particularly suitable for lesions located in the superoanterior petrous apex with extension to the posterior cavernous sinus and parasellar region (21, 22).

In this study, the pterional approach group exhibited a significantly higher incidence of trigeminal nerve injury, facial nerve injury, and speech impairment (p < 0.05). However, all lesions in this group were located in the superoanterior petrous apex and frequently involved the cavernous sinus region. These anatomical areas are intrinsically associated with relevant cranial nerves; for example, cavernous sinus pathology often affects branches of the trigeminal nerve. Therefore, the observed high rate of postoperative neurological deficits may be attributable to the inherent nature of the lesions themselves, the challenging anatomy of the region, and/or the combined effects of the exposure provided by the pterional approach and intraoperative manipulation, including retraction of the temporal lobe and cranial nerves. This suggests that for lesions in this specific location, the risk of cranial nerve injury may be inherently elevated regardless of the surgical approach chosen. Under such high-risk anatomical conditions, the pterional approach did not demonstrate a clearly superior protective effect (23, 24). Furthermore, follow-up data from patients in this group showed that some who experienced nerve injuries exhibited functional improvement over time, suggesting that these deficits may be partially related to reversible factors such as surgical retraction.

This study encompasses common petrous apex pathologies, including meningiomas, schwannomas, cholesteatomas, cysts, and chordomas. It must be re-emphasized that surgical approach selection in this study consistently adhered to the principle of “anatomical location first.” Even within the largest meningioma subgroup, the lesion locations differed across the corresponding approach groups. Therefore, the pathological subgroup analysis conducted herein aimed to examine how distinct pathological characteristics interact with the technical specifics of a given approach—after the lesion location had predetermined the fundamental surgical corridor—thereby influencing the ultimate extent of resection and the profile of complications.

In this study, the three “location–approach” pairings exhibited distinct patterns of hospital stay duration: 40 days in the Kawase group (anteromedial petrous apex), 27 days in the retrosigmoid group (posterior petrous apex), and 29 days in the pterional group (superoanterior petrous apex). The variation in length of stay—particularly the prolonged hospitalization in the Kawase group—serves as an important descriptive indicator reflecting the overall treatment course. This difference is primarily attributable to the following factors.

First, it is closely related to the management and care of complications. The Kawase group had the highest rate of intracranial infection (64.3%) and accounted for all cerebrospinal fluid (CSF) leak cases (28.6%). Antibiotic therapy and continuous lumbar drainage are required to treat these complications significantly prolonged hospitalization. Second, the anatomical complexity of the lesions dictates postoperative monitoring needs. Anteromedial petrous apex lesions are deeply situated, adjacent to the brainstem and critical neurovascular structures, typically necessitating closer neurological monitoring and a longer recovery process postoperatively. In contrast, although the pterional group had a higher incidence of new cranial nerve deficits, such functional impairments often recovered during outpatient follow-up and had less impact on acute-phase hospital stay. The endoscope-assisted retrosigmoid group, with its more balanced complication profile and minimal tissue trauma, had the shortest hospital stay. Therefore, hospitalization duration also reflects an important clinical characteristic of the inherent risks, complication spectrum, and recovery process associated with surgery in specific anatomical regions, providing valuable reference for preoperative assessment.

Based on our findings, the retrosigmoid approach combined with neuroendoscopy provides significantly enhanced visualization, offering a substantially larger field of view and greater depth of field compared to the microscope alone (29, 30). With endoscopic assistance, surgical maneuvers become more precise, which significantly reduces the risk of injury to vessels, nerves, and the internal auditory canal. This technique also enables more accurate and complete tumor resection, thereby lowering the likelihood of recurrence. Neuroendoscopy is minimally invasive, requiring only a small incision and utilizing natural anatomical corridors, which also minimizes the need for brain retraction (31–33).

In the future, integrating neuroendoscopy with the Kawase approach could address the anatomical limitations inaccessible via the endoscopic retrosigmoid route. After drilling Kawase’s triangle, the use of a neuroendoscope can expand the exposure of the petrous apex, petroclival region, and the ventrolateral brainstem. Further extensive drilling of the petrous apex will correspondingly increase the exposed area. This hybrid strategy not only maximizes the surgical field and reduces technical difficulty but also maintains a minimal trauma profile for the patient. The combination of the Kawase approach and neuroendoscopy holds the potential to optimize both the gross total resection rate and the control of postoperative complications.

Our study provides a clear delineation of the respective advantages and limitations of the three surgical approaches. To maximize the rate of gross total resection and minimize patient complications, a hybrid strategy, partially combining elements of these approaches, can be employed. This is not a simple superposition of techniques, but rather a sequential and staged process tailored to the individual patient’s anatomy and the specific phase of the procedure, allowing for controlled exposure and resection of the tumor (34).

Our work is founded upon a hierarchical decision-making framework based on the principle of “anatomical location first.” A hybrid approach can be strategically planned as follows: For tumors involving the anteromedial petrous apex and both the middle and posterior cranial fossae, the Kawase approach combined with neuroendoscopy is recommended. For tumors located in the posterior petrous apex, posterior fossa, and cerebellopontine angle in patients with pre-existing hearing loss, the retrosigmoid approach combined with neuroendoscopy is an effective option. For large tumors extensively involving the petrous apex with simultaneous extension into the parasellar region, middle fossa, and posterior fossa, a combination of the pterional and Kawase approaches may be utilized.