Survival Analysis and Prognostic Factors After Endonasal Resection of Advanced Olfactory Neuroblastomas: A Single Institution Experience

Abstract

Objectives

To explore the prognostic factors in patients with advanced olfactory neuroblastoma (ONB) underwent endoscopic surgery.

Materials and Methods

Retrospective medical records were reviewed of patients with pathologically proven ONB who underwent endoscopic surgical resection. Clinicopathological characteristics including patient demographics, treatment, complications, follow-up, and outcomes were analyzed. Kaplan–Meier overall survival (OS) and disease-free survival (DFS) curves were plotted. Univariate and multivariate Cox regression models were used to determine prognostic factors.

Results

Eighty-five patients with Kadish stage C ONB were examined. According to the various staging systems used, most patients harbored modified Kadish stage C (78.8%). Twenty-six patients (30.6%) underwent bony skull base resection, 11 (12.9%) underwent dura resection, and 24 (28.2%) underwent additional intracranial resection that included the olfactory bulb and duct. Median follow-up was 39 months. Five-year OS and DFS rates were 83.7% and 74.9%, respectively. Five-year OS was 100% in patients treated with bony skull base resection and 77.5% in those who were not (P = .052). Dura resection did not improve OS. Multivariate Cox regression analysis identified perioperative complications (P = .009), gross total resection (P = .004), orbital invasion (P = .014), postoperative radiotherapy (P = .030), and bony skull base resection (P = .019) as independent prognostic predictors.

Conclusion

For patients with advanced ONB, endoscopic surgery in conjunction with radiotherapy and chemotherapy is effective and safe. Dura resection should be performed with caution in selected patients to balance survival and complications. Postoperative radiotherapy is important to improve OS and DFS.

Graphical abstract.

Introduction

Olfactory neuroblastoma (ONB) is a rare, malignant tumor originating from the olfactory neuroepithelial mucosa that accounts for 3%-6% of sinonasal neoplasms. The diagnosis is typically indicated by positive staining for synaptophysin and other neuroendocrine markers (chromogranin A, neuron-specific enolase, and others) combined with negative staining for muscle, melanoma, and lymphoma markers.

Four main clinical staging systems have been proposed to evaluate ONB based on clinical, imaging, and intraoperative findings. Although different, the Kadish, ¹ modified Kadish, ² Dulguerov, ³ and American Joint Committee on Cancer (AJCC) ⁴ systems all classify the tumor into early and advanced stages. The Kadish system is most widely used and divides ONB into 3 stages: stage A tumors are confined to the nasal cavity, stage B tumors extend to the paranasal sinuses, and stage C tumors extend beyond the nasal cavity and paranasal sinuses and may involve the cribriform plate, skull base, intracranial cavity, and/or orbit. ¹ The modified Kadish system adds stage D, which comprises tumors with metastasis to the cervical lymph nodes and/or distant sites. ²

As indicated by the European Position Paper on Endoscopic Management of Tumors of the Nose, Paranasal Sinuses and Skull Base (EPOS) ⁵ and the National Comprehensive Cancer Network, surgical resection followed by radiotherapy is considered the treatment of choice for ONB. Open surgery has long been regarded as the gold standard; however, endoscopic surgery has become an alternative and exhibits similar oncologic control. Few studies have reported endoscopic surgery outcomes in patients with advanced ONB and little is known regarding prognostic factors.

ONB surgical principles dictate that tumor resection should include the entire bony skull base. EPOS recommended that the olfactory bulb and overlying dura should also be removed. ⁵ However, the extent of resection remains controversial, particularly whether to include the bony skull base and/or dura. Resection of the dura underlying a large anterior skull base defect increases the risk of cerebrospinal fluid (CSF) rhinorrhea and intracranial infection, which can prolong hospitalization. However, leaving the dura in place may result in residual tumor. The necessary extent of anterior skull base resection in endoscopic ONB resection remains unknown.

We performed a comprehensive review of a large cohort of patients with advanced ONB who underwent multimodality treatment including endoscopic surgery and report safety and oncological outcomes.

Materials and Methods

We conducted a retrospective review of patients with pathologically proven ONB treated at Eye, Ear, Nose, and Throat (EENT) Hospital from January 2010 to September 2019. Inclusion criteria were as follows: (1) endoscopic surgery with or without open approach, including pure endoscopic endonasal approach (EEA), endoscopic skull base resection with skull base reconstruction (ESBR), and combined endoscopic cranionasal (CEC) approach; (2) pathologic diagnosis of ONB; and (3) Kadish stage C. We excluded patients who underwent palliative resection and those who had previously undergone open surgery or radiotherapy of the head and/or neck. Patients treated with open surgery alone and those with postoperative pathology other than ONB were also excluded.

The EEA was defined as pure endoscopic approach without any help from open approach. The term “ESBR” refers to the resection and reconstruction of the anterior skull base using an endoscopic approach. The CEC refers to open surgery in anterior skull base performed by neurosurgeons and assisted by otolaryngology surgeons using endoscopic approach to remove disease in the sinonasal area. The operative techniques are shown schematically in Supplemental Figure S1 and were selected according to tumor extension. The CEC approach was used in patients with extensive intracranial invasion. All patients undergoing the CEC approach have artificial dura mater implanted, with or without sutures. In the ESBR approach, reconstruction involved the use of fascia lata, with a sandwich technique applied, comprising the insertion of fat, followed by 1 layer of fascia as underlay, and another layer of fascia as overlay. Categorized by resection area, when dura or intracranial resected, fascia lata was the preference of choice. When only bony skull base or mucosa was resected, free flap or even no reconstruction material was used.

We have adopted a relatively minimally invasive resection procedure in our institution. For tumors with ethmoid or nasal invasion without skull base erosion, skeletonization and bone polishing were achieved without exposure of the dura. When skull base invasion was present, total skull base resection with exposure of the dura was performed. Intraoperative magnification of the lesion was important to detect dural adhesion. If the tumor could be detached easily, the dural surface was scraped or electrocoagulated rather than completely resected. If the tumor was adherent to the dura and not easily detached, the dura was removed. Dural margins were histopathologically examined during surgery. Dura with obvious invasion was resected with a margin of at least 3 mm. For tumors with intracranial invasion, dural and tailored brain resection was performed as necessary.

The radiotherapy techniques utilized included 3-dimensional, conformal radiotherapy and intensity-modulated radiotherapy, which were delivered using 6-MV photons. Neoadjuvant and adjuvant chemotherapy consisted of various regimens including platinum-based drugs with vincristine, epirubicin, and pirarubicin. Concurrent chemotherapy was mainly platinum based. Radiotherapy was prescribed for patients with gross tumor on clinical evaluation, positive margins, large tumor volume, or suspected lymph node metastases. Treatment was individualized according to disease extent and patient physical status. Comparison of survival outcome and proportion of dura resection among our cohort and others was performed. All patients provided informed consent for treatment. This study was approved by the institutional review board of EENT Hospital and conducted according to the principles of the Declaration of Helsinki and the Good Clinical Practice guidelines.

Clinicopathological characteristics including patient demographics, imaging findings, postoperative pathological diagnosis and stage, treatment modalities, complications, follow-up, and outcomes were recorded. Standard follow-up included enhanced computed tomography or magnetic resonance imaging and endonasal endoscopy at each visit.

Statistical analyses were performed using SPSS software version 21.0 (IBM Corp., Armonk, NY, USA) and R software version 3.5.3 (http://www.Rproject.org). The association of complications with dura resection was compared using Pearson’s χ² test. Clinicopathological factors were analyzed using univariate Cox regression. Survival curves were plotted using the Kaplan–Meier method and compared using the log-rank test. Univariate factors that achieved statistical significance in this study and those determined to be significant in previous studies were entered into a multivariate Cox regression model. Two-tailed P < .05 was considered significant.

Results

Clinicopathological Characteristics

A total of 85 patients with Kadish stage C ONB were included in the cohort. Clinicopathological characteristics are presented in Table 1. Sixty-seven patients were men (78.8%) and 18 were women (21.2%). Median age at diagnosis was 50.9 years (range 23-83). According to the various staging systems used, most patients harbored modified Kadish stage C (78.8%), Dulguerov stage 4 (50.6%), and AJCC stage T4b (56.5%) tumors.

Table 1.

Clinicopathological Characteristics.

Characteristics	No. of patients	%
Age	Mean 50.9 (range 23-83)	SD: 14.1
Sex
Male	67	78.8
Female	18	21.2
Modified Kadish
C	67	78.8
D	18	21.2
Dulguerov
1	2	2.4
2	15	17.6
3	25	29.4
4	43	50.6
AJCC-T
2	8	9.4
3	13	15.3
4a	16	18.8
4b	48	56.5
Margin
UK	5	5.9
Negative	54	63.5
Positive	26	30.6
GTR
Yes	73	85.9
No	12	14.1
Orbital invasion
0	30	35.3
I	24	28.2
II	15	17.6
III	16	18.8
Complications
No	73	85.9
Yes	12	14.1
Surgical approach
EEA	49	57.647
ESBR	29	34.117
CEC	7	8.235
Reconstruction material
Non	45	52.9
Fascia lata	31	36.5
Artificial dura mater	6	7.1
Free flap	3	3.5
Radiotherapy
Non	13	15.3
Pre-OP	15	17.6
Post-OP	57	67.1
Chemotherapy
Non	33	38.8
Pre-OP	15	17.6
Post-OP	37	43.5
Resection area
Mucosa	24	28.2
Bony skull base	26	30.6
Dura	11	12.9
Intracranial	24	28.2

Abbreviations: AJCC, American Joint Committee on Cancer; CEC, combined endoscopic cranionasal approach; EEA, endoscopic endonasal approach; ESBR, endoscopic skull base resection with skull base reconstruction; UK, unknown; GTR, gross total resection; pre-OP, preoperative; post-OP, postoperative.

Negative resection margins were achieved in 54 patients (63.5%). Microscopic positive margins were present in 26 (30.6%) and margins were unknown in 5 (5.9%). Gross total resection (GTR) was achieved in 73 (85.9%) patients. Orbital invasion was present in 55 patients (64.6%) and classified according to Iannetti et al into 3 grades ⁶ : grade I (erosion of the medial orbital wall), 24 patients (28.2%); grade II (invasion of the periorbital fat), 15 patients (17.6%); and grade III (invasion of the medial rectus, eye globe, optic nerve, or palpebral skin), 16 patients (18.8%).

Treatment

EEA was performed in 49 patients (57.6%), and ESBR was performed in 29 (34.1%). Only 7 patients (8.2%) required the CEC approach. The skull base was reconstructed using a previously described sandwich technique ⁷ with a mucosa flap in 3 patients (3.5%), artificial dura mater in 6 (7.1%), and fascia lata in 31 (36.5%).

Eighteen patients were categorized as modified Kadish stage D; among these, 16 had lymph node involvement alone, 1 had lymph node involvement and distant metastasis, and 1 had distant metastasis without lymph node involvement (Table 2). Four of the patients with lymph node involvement did not receive radiotherapy or chemotherapy because of patient refusal; 3 of these died and 2 experienced recurrence. One of the two patients with distant metastasis received preoperative chemotherapy and experienced pathologic complete response by the time of surgery. The other patient with distant metastasis underwent surgery and postoperative radiotherapy but died from spinal metastasis 7 months after surgery. Only 1 patient with lymph node involvement underwent selective neck dissection; the remaining patients were managed with neck irradiation (median dose, 55.8 Gy).

Table 2.

Characteristics of Patients with Modified Kadish Stage D Tumors.

Case No.	Baseline LN	Baseline distant metastasis	Radiotherapy	Chemotherapy	Reccurence	Death	Distant metastasis in follow up
1	+	−	Preoperative	Preoperative	No	No	No
2	+	−	Postoperative	Preoperative	No	No	No
3	+	−	Preoperative	Preoperative	No	No	Yes
4	+	−	None	None	Yes	No	Yes
5	−	+	Postoperative	None	No	Yes	Yes
6	+	−	Postoperative	Postoperative	No	No	No
7	+	−	Postoperative	Postoperative	Yes	Yes	Yes
8	+	−	None	None	No	Yes	No
9	+	−	None	None	No	Yes	No
10	+	−	None	None	Yes	No	No
11	+	−	Postoperative	Postoperative	Yes	No	Yes
12	+	−	Postoperative	Preoperative	No	No	No
13	+	−	Postoperative	Postoperative	No	No	No
14	+	−	Postoperative	Postoperative	No	No	No
15	+	+	None	Preoperative	No	No	No
16	+	−	None	None	Yes	Yes	No
17	+	−	Postoperative	Postoperative	Yes	Yes	Yes
18	+	−	Preoperative	Preoperative	No	No	No

Abbreviation: LN, lymph node.

Adjuvant therapy was used in 58 (68.2%) patients. Neoadjuvant radiotherapy and chemotherapy were used in 15 (17.6%); 12 of these patients (14.1% overall, 80% of those treated with neoadjuvant therapy) received both before surgery. Overall, 72 patients (84.7%) received radiotherapy; among these, 15 (17.6% overall, 20.8% of those who received radiotherapy) received it before surgery. The radiation dose to the primary site ranged from 58 to 72 Gy for the preoperative group and 58 to 69 Gy for the postoperative group. Chemotherapy was administered to 52 patients (61.1%): 15 (17.6% overall, 28.8% of those who received chemotherapy) before surgery and 37 (43.5% overall, 71.1% of those who received chemotherapy) after surgery. Preoperative chemotherapy without radiotherapy was administered in 3 patients. One achieved complete remission with pathologic complete response after chemotherapy. The other 2 patients had stable disease after chemotherapy and underwent postoperative radiotherapy. All 3 patients were alive without disease at last follow-up.

With respect to extent of resection, 24 patients (28.2%) underwent extended resection of the mucosa alone, 26 (30.6%) underwent bony skull base resection, 11 (12.9%) underwent bony resection plus dura resection, and 24 (28.2%) underwent additional intracranial resection that included the olfactory bulb and duct.

Complications

Twelve patients experienced perioperative complications: meningitis in 3, CSF rhinorrhea in 1, vision loss in 1, hypotension in 3, heart failure in 2, delirium in 1, and brain edema in 1. The patient with CSF rhinorrhea experienced resolution with conservative treatment. The patient with brain edema became comatose and died quickly despite intensive treatment. Incidence rate of patients with dura resection was significantly higher than that of patients without (25.7% vs 6%, P = .01).

Mucosa flap was used in 3 patients (3.5%), artificial dura mater in 6 (7.1%), and fascia lata in 31 (36.5%). When we divided the reconstruction methods into artificial dura mater/fascia lata group and mucosa/nonreconstruction material, no evidence showed that the reconstruction method may affect incidence of postoperative complications (Pearson χ² = 1.246, P = .264).

Oncological Outcome

Median follow-up in the entire cohort was 39 months. Two patients were lost to follow-up. One was a 58-year-old woman with a modified Kadish stage C tumor who received neoadjuvant radiotherapy and chemotherapy and failed to follow-up after surgery. The other was an 83-year-old man with a modified Kadish stage C tumor who did not receive neoadjuvant therapy; his adjuvant therapy status was unclear.

In the entire cohort, 5-year overall survival (OS) and disease-free survival (DFS) rates were 83.7% and 74.9%, respectively; 10-year OS and DFS rates were 72.1% and 41.7%, respectively (Figure 1A and B). Patients with modified Kadish stage D tumors had significantly worse OS than those with stage C tumors (34% vs 91.4%; P = .003; Figure 1C). DFS was also significantly worse in patients with stage D tumors (27.5% vs 83.7%; P = .002; Figure 1D).

Figure 1.

Kaplan–Meier curves for (A) overall survival or (B) disease-free survival in the entire cohort and Kaplan–Meier curves for (C) overall survival or (D) disease-free survival according to modified Kadish stage.

Five-year OS was 100% in patients treated with bony skull base resection and 77.5% in those who were not (P = .052, Figure 2A). DFS did not significantly differ according to bony skull base resection (P = .739; Figure 2B). OS and DFS rates did not significantly differ according to dura mater resection (P = .646 and .931, respectively; Figure 2C and D).

Figure 2.

Kaplan–Meier curves according to bony skull base resection for (A) overall survival and (B) disease-free survival. Kaplan–Meier curves according to dura mater resection for (C) overall survival and (D) disease-free survival.

Five-year OS was significantly higher in patients treated with surgery plus radiotherapy than those treated with surgery alone (87.2% vs 65.6%; P = .0059; Figure 3A). OS was significantly higher in patients treated with preoperative radiotherapy than those treated postoperatively (Figure 3B). The difference in 5-year DFS was greater: DFS was 80.1% in patients treated with surgery plus radiotherapy and only 44.4% for those treated with surgery alone (P = .00055; Supplemental Figure S2A).

Figure 3.

Kaplan–Meier overall survival curves according to (A) radiotherapy (with or without) and (B) timing of radiotherapy (preoperative, postoperative, or none). (C) Univariate (left) and multivariate (right) analysis for overall survival.

Abbreviations: CEC, combined endoscopic cranionasal approach; CI, confidence interval; EEA, endoscopic endonasal approach; ESBR, endoscopic skull base resection with skull base reconstruction; GTR, gross total resection; HR, hazard ratio; RT, radiotherapy.

Nine patients developed distant metastasis within 30 months of diagnosis and one developed metastasis at 54 months. The 1-, 3-, and 5-year distant metastasis-free survival (DMFS) rates were 90.4%, 81.3%, and 78.7%, respectively. Eleven patients underwent surgery alone: 4 died and 4 experienced disease relapse.

Univariate and Multivariate Cox Regression Analysis of OS and DFS

In the univariate analyses, surgical approach, Dulguerov stage, chemotherapy, and surgical margin were not associated with OS. Modified Kadish stage, perioperative complications, and GTR were strong predictors of lower OS (P < .05). Orbital invasion, radiotherapy, and bony skull base resection showed a predictive trend. A multivariate Cox regression model constructed using modified Kadish stage, perioperative complications, GTR, orbital invasion, radiotherapy, and bony skull base resection as variables showed that perioperative complications, GTR, orbital invasion, postoperative radiotherapy, and bony skull base resection were significant predictors of OS (Figure 3C).

Univariate DFS analyses identified modified Kadish stage (P = .004), perioperative complications (P = .021), GTR (P = .011), and radiotherapy (P = .008) as significant predictors (data not shown). Patients without orbital invasion showed better DFS but the difference was not significant (P = .132). Bony skull base resection and dura resection were not significantly associated with DFS (P = .740 and .931, respectively). Multivariate analysis revealed that GTR, orbital invasion, and radiotherapy remained independent predictive factors for DFS.

Although chemotherapy was not significantly associated with distant metastasis (P = .387, data not shown), a trend toward better DMFS was observed in patients who received preoperative chemotherapy (P = .153; Supplemental Figure S3). As with DFS, DMFS was not significantly associated with bony skull base resection (P = .511) or dura resection (P = .635).

Comparison of Survival Among Cohorts

In our cohort, only 41.1% of cases underwent dura resection, while the proportions were 69.2% and 73.8% at the Beijing ⁸ and Changsha ⁹ centers, respectively, which were significantly higher than in our cohort (Supplemental Table S1). Interestingly, the 5-year OS and DFS rates were comparable among these 3 cohorts, showing that no additional survival benefit was obtained from dura resection in the era of endoscopic surgery. Comparing with the CFR approach, ¹⁰ which requires dura resection in all patients, we achieved an extremely close 5-year DFS (in the literature, ONB has a 5-year DFS of 81%, in this study, the data were 79%). Minimal conservative endoscopic surgery possibly had comparable survival outcomes to the CFR approach.

Discussion

This retrospective single-center study represents one of the largest reported cohorts of patients with advanced ONB. In contrast to previously reported ONB incidence peaks in the second and sixth decades, disease incidence in our cohort was even across all ages. Although a sex difference in disease incidence has been reported in previous studies, the number of men was 3 times greater than women in our study, which may indicate a selection bias was present.

A meta-analysis conducted by Bonnecaze that analyzed and compared OS in patients with advanced ONBs reported that 5-year survival was highest in those treated with combined surgery and radiotherapy (72.9%; range, 60.0%-82.2%). ¹¹ Other institutions have reported variable 3- and 5-year OS rates ranging from 70.7% to 89.7%.^12-14 In our cohort, the 5-year OS was 83.7%, which is comparable. For patients with Kadish stage C and D disease, 5-year OS in our cohort was 91.4% and 34%, respectively, which is similar to 5-year OS in patients treated with radiotherapy with or without surgery (78.3% and 41.9%, respectively). ¹⁵ The slightly higher OS in stage C disease patients in our cohort may be explained by the fact that unresectable tumors tend to be treated with neoadjuvant therapy before surgery.

A previously developed nomogram from our institution based on a 154-patient cohort that included those with early and advanced ONB indicated that orbital invasion, radiotherapy, and modified Kadish stage were important factors affecting OS. ¹⁶ These same factors were also found to have a significant impact on OS among patients with advanced ONB in the present cohort.

Preoperative radiotherapy is preferable for highly advanced tumors in clinical practice, partially because it may decrease tumor burden, which is beneficial to preserve adjacent critical structures such as the skull base and orbit. ¹⁵ However, from a surgeon’s perspective, no convincing data have been previously reported. Our results suggest that postoperative radiotherapy is an independent predictive factor for both OS and DFS.

For advanced ONB, particularly those with extension to the dura or orbital muscles, achieving clear surgical margins is difficult. In our cohort, no significant difference in OS or DFS was observed between patients with negative and positive margins, which are counterintuitive; however, postoperative radiotherapy may be responsible to some extent. By analyzing the timing of radiotherapy, we found that patients who received postoperative radiotherapy were more likely to have a better OS regardless of margin status.

The role of chemotherapy remains uncertain. Chemotherapy is more likely offered to patients with advanced-stage tumors or lymph node metastasis. Song et al reported that the locoregional recurrence rate decreases after chemotherapy. ¹⁷ Although EPOS has proposed preoperative chemotherapy followed by surgery for Kadish stage C/D tumors, ⁵ this remains controversial because of a lack of evidence. Miller et al reported that patients who received platinum-based adjuvant chemotherapy do not exhibit improved survival compared to surgery plus adjuvant radiotherapy alone. ¹⁸ In a Surveillance, Epidemiology, and End Results database review, the use of chemotherapy in ONB management was associated with inferior disease-specific survival and OS. ¹⁹ In contrast, our study observed no differences in OS or DFS between patients treated with and without chemotherapy. However, chemotherapy may be beneficial in selected patients to reduce tumor volume and achieve negative margins. ²⁰

A significant association between surgical treatment and survival in patients with stage C and D disease has been previously reported. Song et al reported that the risk of death or disease progression is elevated in patients who do not undergo surgery. ¹⁵ However, this finding should be interpreted with caution, as selection bias may have been present. Patients who do not undergo surgery may have unresectable disease, while surgically treated patients are more likely to have resectable disease that is less invasive and extensive, which ensures a better outcome.

Traditionally, resection of the cribriform plate, dura, and olfactory bulb has been the standard surgical approach for ONB, regardless of tumor stage, ²¹ mainly because craniofacial resection has been considered the gold standard for sinonasal tumors since it was first introduced by Ketcham et al in 1963. ²² With the development of the combined craniofacial approach, dural preservation has attracted interest. ² This approach is performed by both a neurosurgeon and an otolaryngologist. In tumors without dural invasion, the dural incisions are closed with sutures. In contrast, the dura in tumors with dural invasion is resected and replaced with a temporalis fascia graft. In the Ketcham cohort, up to 22.7% of patients may experience infection. Although with advances in reconstruction technology, the incidence of meningitis has significantly reduced, from a modern point of view, the meningitis rate was about 4.5% in craniofacial resection, still higher than that for endonasal resection where the meningitis rate was about 0%. ²³ Intact dura appears to be important for preventing infection. With endoscopic surgery, the “bottom-up” surgical technique allows cribriform plate resection without unnecessary resection of dura or intracranial tissue. ²¹

Another reason that the cribriform plate and dura are typically resected when treating ONB lies in that olfactory neuroepithelium is located in the superior nasal vault and cribriform plate. However, the olfactory neuroepithelium is also dispersed throughout the superior and middle turbinates. ²⁴ Therefore, resection of the cribriform plate or higher structures may not be necessary in all cases.

With respect to survival and recurrence, the findings from our cohort suggest that bony skull base resection should be performed in the advanced ONB patients. Those who underwent skull base resection had a better outcome than those who did not. In our relatively minimally invasive resection procedure, dura resection was not performed if dural invasion was not present, and survival in these patients was similar to that reported in previous studies utilizing aggressive resection. By itself, this study has limitation in subgroup analysis because of the relatively small size and shorter follow-up time in dura resection group. When examined among cohorts, the survival outcome did not vary according to the different proportions of dura resection observed in the era of endoscopic surgery. Other limitations included selection bias and information bias because of the retrospective nature, so additional prospective studies are needed to make this conclusion more evident.

Moreover, complications may be another reason to use dural resection with caution. In the modern endoscopic era, incidence of meningitis and CSF leak range from 0% to 18.2% in ONB surgeries.^8-10,23,25 In our cohort, 41.2% (35/85) of patients underwent dural resection. The total incidence of intracranial complication accounts for 4.7%, with 1/50 (2%) meningitis and 0% CSF rhinorrhea in nondura resection group, while 2/35 (5.7%) meningitis and 1/35 (2.9%) CSF rhinorrhea in dural resection group. Dura resection may bring about higher incidence of meningitis or CSF rhinorrhea. However, when compared with the findings of other 2 studies,^8,9 no significant difference in incidence was noted regarding different proportions of dura resection. Caution is warranted, and further evidence is needed before drawing the conclusion that dural resection is associated with a higher risk of meningitis or CSF rhinorrhea.

Standard treatment of ONB remains controversial. With improvements in endoscopic surgical techniques and equipment, endoscopic surgery can achieve favorable outcomes with a low complication rate. Moreover, proton beam therapy has gained popularity in treating head neck cancers and can achieve satisfactory results. Nakamura et al reported 28 patients with Kadish stage C disease who received proton beam therapy with curative intent and achieved a 76% 5-year OS, ²⁶ suggesting proton beam therapy can be an alternative treatment for advanced ONB. Few cellular, molecular, or genetic ONB studies have been performed because of the rarity of the disease. However, Classe et al recently investigated tumor-infiltrating lymphocytes and PDL1 expression in ONBs, which has provided rationale for the use of immune checkpoint inhibitors in selected ONBs.^27,28 A greater understanding of ONB molecular pathogenesis to reveal potential therapeutic options is needed.

Conclusion

Bony skull base resection can be considered in patients with advanced ONB. Additional dura resection should be performed with caution in selected patients to balance survival and complications. Postoperative radiotherapy is important to improve OS and DFS. Future studies are needed to confirm the benefit of chemotherapy. Preliminary results revealed that dural resection does not provide a survival benefit in patients with Kadish C ONB with skull base involvement but without apparent gross dural involvement; however, further prospective case-control studies or randomized controlled trials are needed to provide additional evidence.