The Role of Elective Neck Treatment in the Management of Sinonasal Carcinomas: A Systematic Review of the Literature and a Meta-Analysis

Simple Summary The impact of elective neck treatment (ENT), whether by irradiation or dissection, on the prognosis of patients with cN0 sinonasal carcinomas remains an understudied issue. A systematic review and meta-analysis of the literature were performed according to PRISMA guidelines in order to assess regional recurrence rates after elective neck treatment compared to observation in cN0 patients. Twenty-six articles for a total of 1178 clinically N0 patients were analyzed. Overall, the 5-year overall survival was 52%; 34.6% of patients underwent elective treatment of the neck and 140 regional recurrences were registered. ENT appears to confer a lower risk of regional recurrence compared to observation alone with a cumulative OR of 0.38 (95% CI 0.25–0.58). Our meta-analysis supports the efficacy of ENT in order to reduce the risk of regional recurrence but the overall impact on survival remains uncertain. Abstract The impact of elective neck treatment (ENT), whether by irradiation or dissection, on the prognosis of patients with cN0 sinonasal carcinomas (SNCs) remains an understudied issue. METHODS: A systematic review and meta-analysis of the literature were performed according to PRISMA guidelines in order to assess regional nodal relapse rate after ENT compared to observation in cN0 SNCs patients. Twenty-six articles for a total of 1178 clinically N0 patients were analyzed. Globally, the 5-year overall survival was 52%; 34.6% of patients underwent ENT and 140 regional recurrences were registered (5.9% in the ENT cohort and 15% in the observation group). ENT appears to confer a lower risk of regional recurrence compared to observation alone, with a cumulative OR of 0.38 (95% CI 0.25–0.58). Our meta-analysis supports the efficacy of ENT for reducing the risk of regional recurrence, but its overall impact on survival remains uncertain.


Introduction
Sinonasal tumors contribute to only 5% of all head and neck cancers, yet they constitute a wide spectrum of pathological entities where malignant epithelial tumors (SNCs, sinonasal carcinomas) represent more than 80% of them [1]. Their management is complex and often requires multimodal approaches that have been shown to improve survival and organ-preservation rates (e.g., orbital contents) compared to surgery alone [2][3][4]. Nonetheless, despite the enormous improvement of endoscopic surgical approaches and radiation delivery techniques, the prognosis remains disappointing, with an overall 5-year survival rate ranging from 30% to 50% [5,6].
Lymphatic drainage from the sinonasal district is mainly directed to the upper jugular, perifacial, and retropharyngeal nodes, while the sentinel lymph node approach is too far to be implemented in SNCs [7,8]. Although the incidence of regional metastasis is notably low at the time of diagnosis (about 10% in a recently published series by Peck et al. [9]), up to 33% of patients will eventually develop them during the follow-up [9,10]. In this regard, the decision whether to adopt a watchful waiting approach or to perform an elective neck treatment (ENT), whether by surgery (END, elective neck dissection) or radiotherapy (ENI, elective neck irradiation), on a clinically node-negative (cN0) SNC remains an understudied issue [11,12]. The present systematic review of the literature and meta-analysis aims to evaluate the prognostic role and clinical relevance of the elective neck treatment in the management of SNCs.

Materials and Methods
This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [13]. It was registered on the International Prospective Register of Systematic Reviews (PROSPERO, University of York, UK) with the publishing number CRD42020169882.
Languages of publication were limited to English, French, and Italian. We also searched the reference lists of all relevant systematic reviews and meta-analyses. Two reviewers (CG, LGL) independently screened articles and abstracts recovered by the search. Duplicates were removed using open-source reference management software Zotero Version 5.0.93 for Linux (Corporation for Digital Scholarship, Vienna, VA, USA). Articles deemed potentially relevant were obtained and assessed in detail by each reviewer independently, according to the above criteria. All discrepancies were resolved by consensus.

Eligibility Criteria
Inclusion criteria were as follows: prospective or retrospective studies, randomized controlled trials, and case series; studies including patients with pathologically documented SNCs (squamous cell type, intestinal-type, and non-intestinal adenocarcinoma, sinonasal undifferentiated carcinoma/SNUC, adenoid cystic carcinoma, other malignant salivary gland cancers); studies including detailed clinical data about initial neck status, neck treatment strategy, and subsequent regional recurrences with a minimum follow-up time of 6 months.
Exclusion criteria: studies including patients under 18; studies including other head and neck cancer sites and whose data were not clearly discernible; studies including non-epithelial sinonasal malignancies (mucosal melanoma, olfactory neuroblastoma, lymphomas, sarcomas, etc.); case reports or case series smaller than 5 patients.

Data Collection and Outcome Definitions
The following data were extracted from each study whenever available: demographics (number of patients, age, gender); clinical/surgical data: TNM stage/classification at diagnosis, anatomical subsite involved (maxillary, ethmoidal, frontal or sphenoid sinus, and nasal cavity), the reported final histopathology, treatment modalities (surgical resection irrespective of the endoscopic/open/combined approach chosen), and overall survival (OS). Information was extracted from these cohorts regarding cervical nodal status at presentation, initial treatment for nodal disease: therapeutic neck dissection for N+ cases, no treatment, or elective neck treatment (ENT) that could be both surgical (END, Elective Neck dissection) or by external RT (ENI, Elective Neck Irradiation). Finally, the regional recurrence rate was recorded and, whenever possible, it was noted whether the nodal failure was isolated or concomitant with local or distant relapse. In addition, in both ENT and watchful waiting groups, the chosen salvage neck strategy was retrieved in order to define the impact of the regional failure on prognosis.
The cumulative analysis was performed by including modalities used for treatment, nodal involvement at presentation, or subsequent recurrence. The estimated risk of nodal involvement was calculated by combining nodal involvement at presentation and subsequent recurrence in the neck, whether treated with ENT or no treatment at all.

Quality Assessment
Three reviewers (CG, AC and LGL) independently extracted data from each study, which were reviewed for consistency among the authors, and any discrepancies were resolved by consensus. The risk of bias for each study using the ROBINS tool with any discrepancies was resolved by consensus [14]. Visualization of the risk-of-bias assessments was performed by creating a traffic lights plot and a weighted bar plot using the robvis tool [15].

Statistical Methods
The software RevMan 5.3 (Review Manager, Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) was used for the meta-analysis and the creation of the forest plot. The odds ratios (ORs) of regional (neck) nodal failure and their 95% confidence intervals (CIs) were calculated for each study. Statistical heterogeneity was assessed using Cochran's Q statistic (p-value for heterogeneity) and the I 2 statistic (total percentage of variation resulting from heterogeneity). In the case of significant heterogeneity (I 2 ≥ 50) the random-effect model was used, while the fixed-effect model was used in absence of significant heterogeneity. Herein, we solely applied the fixed-effect model to obtain ORs, 95% CI, and p-values.

Results
Using the selected keywords, we found 983 articles; moreover, we added 55 articles from the reference lists of all the systematic reviews and meta-analyses found in the research. After the removal of duplicates, 324 titles and abstracts were analyzed, and 206 studies were excluded. A total of 118 articles were fully read and a further 92 were excluded because they did not meet the eligibility criteria. In particular, 13 were case reports, systematic reviews, or meta-analyses; 29 included in their analyses or focused only on other non-epithelial histotypes; 5 also included carcinomas of other subsites of the head and neck; and 45 articles did not provide adequate information on neck treatment. In the end, 26 articles were eligible for the meta-analysis, and a PRISMA flow chart is represented in Figure 1. A descriptive overview of the included articles is presented in Table 1  . All the 26 articles included in the meta-analysis were retrospective case series, published from 1987 to 2018. In total, 15 studies (57.7%) were conducted in the USA (6 in Florida, 5 in Texas, 3 in California, and 1 in collaboration between West Virginia and Pennsylvania), 1 was performed in India, 1 in South Korea, 3 in the Netherlands, 2 in the UK, 2 in Germany, 1 in France, and 1 in Spain.
The entire cohort included 1320 patients, with a median of 37 cases for each series. Sex information was not available in 5 studies; in the remaining papers, there were 657 males (64.7%) and 359 females (35.3%). The median age was 58 years, with a range from 8 to 92 years. The most frequent histotype was squamous cell carcinoma (801 cases, 60.7%), followed by adenocarcinoma (190 cases, 14.4%), sinonasal undifferentiated carcinoma (171 cases, 13%), and adenoid cystic carcinoma (120 cases, 9.1%). Regarding the primary subsite, in 339 cases (31.1%) the origin was in the nasal cavity; in 639 cases (63.5%) it was in the maxillary sinus; in 102 (9.3%) it was in the ethmoidal sinus; in 9 cases (0.8%) it was the sphenoidal sinus; and in 2 it was the frontal sinus (0.2%). In 5 studies, the authors did not clearly specify these data. A descriptive overview of the included articles is presented in Table 1  . All the 26 articles included in the meta-analysis were retrospective case series, published from 1987 to 2018. In total, 15 studies (57.7%) were conducted in the USA (6 in Florida, 5 in Texas, 3 in California, and 1 in collaboration between West Virginia and Pennsylvania), 1 was performed in India, 1 in South Korea, 3 in the Netherlands, 2 in the UK, 2 in Germany, 1 in France, and 1 in Spain.
The entire cohort included 1320 patients, with a median of 37 cases for each series. Sex information was not available in 5 studies; in the remaining papers, there were 657 males (64.7%) and 359 females (35.3%). The median age was 58 years, with a range from 8 to 92 years. The most frequent histotype was squamous cell carcinoma (801 cases, 60.7%), followed by adenocarcinoma (190 cases, 14.4%), sinonasal undifferentiated carcinoma (171 cases, 13%), and adenoid cystic carcinoma (120 cases, 9.1%). Regarding the primary subsite, in 339 cases (31.1%) the origin was in the nasal cavity; in 639 cases (63.5%) it was in the maxillary sinus; in 102 (9.3%) it was in the ethmoidal sinus; in 9 cases (0.8%) it was the sphenoidal sinus; and in 2 it was the frontal sinus (0.2%). In 5 studies, the authors did not clearly specify these data. In this cohort, 305 carcinomas (29.4%) were T1/T2, 280 (24.2%) were T3, and 570 (49.4%) were T4 (110 T4a, 91 T4b, 369 not specified); 5 studies did not report the T stage. As for the lymph node involvement, only 100 patients (7.5%) had a regional metastasis at diagnosis (N+), against 1198 who were cN0. Of the N+ patients, 49.2% were N1, 50.8% were N2 (N2a in 20% of N2 cases, N2b in 46.7%, and N2c in 33.3%), and none were classified as N3. Exclusive surgical treatment of primary tumor was performed in 101 cases (8.3%), while 393 patients (32.2%) were treated with RT or with RT-CHT. The majority of cases (726 patients, 59.5%) were treated with a combined therapy composed of surgery and adjuvant or neoadjuvant RT/CHT-RT. In one study, the kind of treatment of primary tumor was not reported. Complications after neck irradiation were rarely described: Lee et al. [33] and Hinerman et al. [29], respectively, mentioned a case of left brachial plexopathy with severe neck fibrosis and a carotid blow-out; moreover, Castelnau-Marchand et al. [21] reported 51 cases of dysphagia, 61 of radio-mucositis, and 61 patients who suffered from radiodermatitis related to the neck RT. The mean 2-year and 5-year overall survival rates were 66% and 52%, respectively. The median follow-up time was 54 months, with a range from 10 to 132 months.
Meta-analysis is based on the evaluation of 1178 clinically N0 patients. Of these, 407 (34.6%) underwent ENT while, in 771 cases (65.4%), the neck was not treated, and an "observation" strategy was chosen. In this cohort, 140 nodal failures were observed, with a regional recurrence rate of 5.9% (24/407) in the ENT group and 15% (116/771) in the observation group (Table 2). ORs for regional recurrences after elective neck treatment ranged from a minimum of 0.03 to a maximum of 1.39. The cumulative OR was 0.38 (95% CI 0.25-0.58; p < 0.0001), indicating a 62% lower risk of regional recurrence in patients undergoing neck treatment compared to patients who were directed to observation only ( Figure 2). The funnel plot (Figure 3) did not show asymmetry, indicating the absence of publication biases and heterogeneity between the studies. Finally, the graphical representation of the risk of bias between the included studies is represented in Figures 4 and 5. In this cohort, 305 carcinomas (29.4%) were T1/T2, 280 (24.2%) were T3, and 570 (49.4%) were T4 (110 T4a, 91 T4b, 369 not specified); 5 studies did not report the T stage. As for the lymph node involvement, only 100 patients (7.5%) had a regional metastasis at diagnosis (N+), against 1198 who were cN0. Of the N+ patients, 49.2% were N1, 50.8% were N2 (N2a in 20% of N2 cases, N2b in 46.7%, and N2c in 33.3%), and none were classified as N3. Exclusive surgical treatment of primary tumor was performed in 101 cases (8.3%), while 393 patients (32.2%) were treated with RT or with RT-CHT. The majority of cases (726 patients, 59.5%) were treated with a combined therapy composed of surgery and adjuvant or neoadjuvant RT/CHT-RT. In one study, the kind of treatment of primary tumor was not reported. Complications after neck irradiation were rarely described: Lee et al. [33] and Hinerman et al. [29], respectively, mentioned a case of left brachial plexopathy with severe neck fibrosis and a carotid blow-out; moreover, Castelnau-Marchand et al. [21] reported 51 cases of dysphagia, 61 of radio-mucositis, and 61 patients who suffered from radiodermatitis related to the neck RT. The mean 2-year and 5-year overall survival rates were 66% and 52%, respectively. The median follow-up time was 54 months, with a range from 10 to 132 months.
Meta-analysis is based on the evaluation of 1178 clinically N0 patients. Of these, 407 (34.6%) underwent ENT while, in 771 cases (65.4%), the neck was not treated, and an "observation" strategy was chosen. In this cohort, 140 nodal failures were observed, with a regional recurrence rate of 5.9% (24/407) in the ENT group and 15% (116/771) in the observation group (Table 2). ORs for regional recurrences after elective neck treatment ranged from a minimum of 0.03 to a maximum of 1.39. The cumulative OR was 0.38 (95% CI 0.25-0.58; p < 0.0001), indicating a 62% lower risk of regional recurrence in patients undergoing neck treatment compared to patients who were directed to observation only ( Figure 2). The funnel plot (Figure 3) did not show asymmetry, indicating the absence of publication biases and heterogeneity between the studies. Finally, the graphical representation of the risk of bias between the included studies is represented in Figures 4 and 5

Discussion
In the literature, the question of whether to perform an elective treatment on a clinical node-negative patient with SNCs remains open, and the present work may be considered as a natural prosecution of two other published meta-analyses that included only SCC [11] or SNUC [12]. Actually, squamous cell, adenocarcinomas, SNUC, and adenoid cystic carcinomas (ACC) are the most frequent histotypes that can yield nodal metastases in 28%, 25%, 12%. and 10% of cases, respectively [42]. A recent evaluation of the US National Cancer Database has estimated an overall risk of nodal disease for sinonasal cancer (all histotypes) of 13.2% at the moment of diagnosis, with sinus localization (vs. nasal cavity), black race, and uninsured patients as the principal risk factors [43]. By including over 1320 patients, we found a significantly higher nodal recurrence rate in patients who did not receive an upfront ENT (15.0%) versus patients who were treated with surgical dissection or irradiation (5.9%); while these findings might apparently favor ENT as the standard of treatment in cN0 SNCs, it remains unclear whether the added morbidity of ENT does actually outweigh the risk of having a regional failure [42,44].
An inherent issue of SNCs remains their biological heterogeneity that clinically translates into a different prognosis, radiosensitivity, and tendency to regional spread.
For example, a prospective propensity score matching study found no benefits in terms of event-free survival for END in cN0 ACCs; only a quarter of the cases were in the sinonasal cavities, but it should be noted that the vast majority of patients had postoperative RT, with fields probably involving at least the I levels [45]. Another study from the US found improved OS only for END performed in T3-T4 major salivary glands ACC; in this large dataset of over 2800 patients, the sinonasal cavity represented the second anatomical subsite in terms of frequency, but it was also the least likely to receive END and to harbor occult metastases [46]. Finally, in a recently published series of exclusively sinonasal ACC treated by endoscopic endonasal approaches, nodal recurrence occurred in only 1 out of 30 cases, and it was concomitant to distant metastases; again, adjuvant RT was delivered in 83% of patients but the fields were not specified [47].
For SNUC lesions, in the series from the Mayo Clinic, in those patients who received ENI, no regional recurrences were documented with a median follow-up time of 6.9 years; only one patient died of regional failure, but their initial neck status was not specified [48]. In a meta-analysis of SNUC patients, nodal metastases were present in 16% at diagnosis and, thus, the authors always advocated for the inclusion of the neck in the planning of irradiation fields [49].
Instead, for adenocarcinomas, a recent SEER analysis revealed no significant differences in terms of OS or DSS when comparing active surgical management to observation in the N0 neck (OS p = 0.83, DSS p = 0.82) [50]. Most notably, the presence of nodal metastases at diagnosis represented a negative prognostic factor at multivariate analysis.
Currently, the most recent NCCN guidelines recommend ENT only in patients with T3 to T4a tumors of the maxillary sinus [51]. The rationale for prophylactic neck treatment is based on the classical model proposed by Weiss et al. [52] more than twenty years ago; however, the cut-off of a preoperative risk of occult nodal disease of 20% was based only on squamous-cell carcinomas and whatever the anatomical subsite. Given the rarity of these malignant tumors, the majority of the included studies have a large time span, with the earliest cases dating back to the 1960s [32]. This feature obviously introduces an unavoidable bias in terms of correct pretreatment staging and, as already discussed in the work by Mirghani et al. [44], the highest rates of occult nodal disease were found in older publications. Whether this was related only to a late diagnosis with associated high T SNCs is not clear, given that, in recent years, delayed presentation of these tumors is still frequent, and T3-T4 cancers still represent the majority of cases [53].
The association between higher T stages and the risk of nodal occult disease should be always kept in mind; for instance, in a series of 299 patients with sinonasal malignancies, the higher cumulative incidence of regional metastases was significantly linked to dural, orbit, or infratemporal fossa involvement, i.e., T4 tumors by definition [9]. An unresolved question is whether there has been a reduction of late presentation at diagnosis for these particular tumors. Unfortunately, because of missing data, changes in the TNM staging system, and of the long timespan of the published series, this was basically never assessed, even in studies specifically addressing the temporal trends of SNCs' epidemiology [6,54]. In the near future, it remains to be verified whether improvements in both radiological and endoscopic techniques would allow for a parallel change in this setting [55,56].
It must be remembered that the NCCN recommendations are not consistently followed at all institutions and that, for instance, due to the possible involvement of retropharyngeal nodes, END is not performed, not even for maxillary sinus cancer, where some authors may deem it necessary in order to isolate blood vessels for microvascular anastomosis after maxillectomy [57,58]. Interestingly, retropharyngeal lymph nodes were almost never mentioned (see Table 2) as sites of nodal failure, and this is despite their known prognostic impact [7]. Regional recurrences portend a significantly worse prognosis for SNCs [59], but a major obstacle to the standard implementation of ENI is represented by the fact that isolated regional failures are very rare and, therefore, there is a lack of perceived benefit [42]. Actually, patients who experienced nodal relapse frequently died because of subsequent distant metastasis or combined local disease progression [33,42,58,59].
Our work suffers from some of the well-known issues related to sinonasal malignant tumors: because of their rarity and of the heterogeneity of the many described histotypes, only limited and small studies are usually reported. Furthermore, there is always the risk that some advanced-stage squamous cell carcinomas of the maxillary sinus and of the nasal cavity could actually represent tumors of the oral cavity or the skin, respectively. Another inherent bias comes from the imaging techniques that were chosen to stage the pretreatment nodal status; as was reported in Table 1, some groups only relied on physical examination or simple x-rays, which have notoriously low diagnostic accuracy. In addition, it was not possible to analyze the specific cause of death because they were almost never reported, nor was it always clear whether successful salvage neck dissection could be more easily accomplished in the ENT or observation group. These factors make it very difficult to ascertain how regional relapses actually contribute to mortality, and we strongly hope that a future published series will not forget to report this information.

Conclusions
This is the first meta-analysis investigating the role of ENT in the management of clinically node-negative sinonasal carcinomas. Our results show that the "observation" approach on the neck portends a significantly higher regional recurrence rate, and they suggest that, despite continuous improvement in imaging techniques, nodal micrometastases can be present at the moment of diagnosis. In this regard, an elective dissection or irradiation should be taken into consideration in order to minimize the risk of recurrence in the neck and the need for salvage surgery. Specific studies are still needed, specifically in order to clarify whether a reduction in the nodal recurrence rate does actually imply a reduction in the disease-specific mortality rate.
Author Contributions: C.G.: data curation, formal analysis, investigation, methodology, validation, visualization, and writing-original draft. L.G.L.: data curation, formal analysis, investigation, methodology, validation, visualization, and writing-original draft. C.B.: data curation, formal analysis, investigation, methodology, validation, visualization, and writing-original draft. A.C.: data curation, methodology, validation, visualization. G.M.: project administration, resources, supervision, and review and editing. O.G.: conceptualization, data curation, project administration, resources, supervision, and review and editing. All authors have read and agreed to the published version of the manuscript.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data can be obtained upon reasonable request to the Corresponding Author.

Conflicts of Interest:
All authors declare they have nothing to disclose.