Association of ipsilateral radiation therapy with contralateral lymph node failure in patients with squamous cell carcinoma of the oral cavity: A systematic review and meta‐analysis

Ipsilateral neck radiotherapy (INRT) is controversial in some patients with oral cavity cancer due to concern for contralateral neck failure (CNF).


| INTRODUCTION
Oral cavity squamous cell carcinomas (OCSCC), like other cancers of the head and neck, require careful selection of neck treatment volumes to maximize oncologic control and minimize treatment related toxicity that can impact quality of life. There are randomized, prospective clinical trials to guide surgical management of the neck. [1][2][3] However, the selection of neck radiotherapy targets is guided by retrospective institutional series on patterns of failure since prospective data are lacking. [4][5][6][7][8][9][10][11][12] Historically, radiation targets included the bilateral neck, 13 even when the contralateral neck was uninvolved, due to the potential risk for contralateral lymph node drainage 14 and poor outcomes associated with salvage therapy in OCSCC. [15][16][17] This work was presented at the ACRO 2023 Annual Meeting on 16 March 2023. However, bilateral neck irradiation is associated with increased toxicity and can impact quality of life. [18][19][20] In the last 20 years, irradiation of only the involved ipsilateral neck has emerged as an alternative management strategy. [21][22][23] While this approach is accepted for lateralized tonsil squamous cell carcinoma with 1 or fewer involved ipsilateral lymph nodes, 24,25 its use in OCSCC is not well defined: American and European consensus guidelines provide different recommendations as to when ipsilateral neck irradiation should be employed. 26,27 In this systematic review and meta-analysis, we aim to better understand the efficacy of ipsilateral neck irradiation for patients with OCSCC. We focused on identifying the rate of contralateral neck failure (CNF) associated with this treatment approach and potential characteristics and treatment factors that may influence the risk of CNF.

| METHODS
A systematic literature review was performed following PRISMA guidelines. 28 Review and analyses were designed prospectively and registered with PROSPERO (CRD42022349267). In total, three databases (Pubmed, Embase, and Web of Science) were queried using standardized search terms to identify publications that contained oral cavity cancer (including subsites oral tongue, floor of mouth, retromolar trigone, gingival mucosa, and buccal mucosa), ipsilateral radiation, and contralateral neck failure. Only peer-reviewed articles published between 1 January 1980 and 7 January 2022 were included for screening. After removing duplicate publications, titles and abstracts were screened by at least two reviewers. In the case of discordance, a third reviewer was added, and inclusion for full text review was determined by consensus opinion. At both the abstract and full-text review stages, articles were included for analysis if they contained at least 10 patients treated with ipsilateral neck radiation for oral cavity cancer. Case reports/ series, systematic reviews, cancer database studies, studies in foreign languages, and those that did not have full text available were excluded. Furthermore, studies that only used bilateral neck radiation, included patients treated with re-irradiation, or those that did not provide details of RT neck target (i.e., number receiving unilateral RT, number receiving bilateral RT), were excluded. Patients with bilateral neck disease (stage N2c) were also excluded from analysis. To reduce confounding, if two or more studies had overlapping patient populations, only the more recently published article was included for statistical analysis. A full list of search terms and exclusion criteria is enclosed in Data S1 (eMethods).
Following full text review, rates of CNF failure were recorded. Patient and treatment characteristics-such as radiation treatment modality, AJCC staging edition, pathologic tumor (T) stage, pathologic nodal (N) stage, use of neck dissection, use of chemotherapy, pathologic risk factors (grade, margin status, lymphovascular invasion [LVSI]), perineural invasion (PNI), extranodal extension (ENE), and depth of invasion (DOI)-were also extracted. Quality of studies included for statistical analysis was quantified using the MINORS criteria. 29 The primary outcome was the pooled rate of CNF following ipsilateral RT, defined as nodal failure within the contralateral, unirradiated neck, as determined by clinical, pathologic, or radiographic assessments. Secondary outcomes included pooled rates of CNF by T-stage, N-stage, and use of bilateral RT. Pooled rates of CNF were estimated using random effects models. Publication bias was evaluated using Egger's regression test for funnel plot symmetry. Heterogeneity among the studies was assessed using Cochran's Q tests for heterogeneity and Higgins I 2 statistic. Sensitivity analyses to assess the effect of tumor lateralization, diagnostic imaging, American Joint Committee on Cancer (AJCC) staging edition, and oral cavity subsite on CNF were performed. Differences between subgroups were examined using the omnibus test of moderators (Q M ). Exploratory linear regression analyses were performed to examine the relationship between CNF and use of neck dissection, use of chemotherapy, tumor grade, and presence of ENE, LVSI, or PNI. Differences in rates of CNF in patients receiving ipsilateral or bilateral RT was assessed using a mixed effect meta-analysis comparing log odds ratios across studies. For this analysis, only studies that reported patients treated with ipsilateral RT or bilateral RT were included. Detailed explanations of each analysis are presented within Data S1 (eMethods). Statistical significance was defined as p < 0.05. All tests were performed in R using "metafor" package (version 3.0-2). 30 3 | RESULTS

| Characteristics of included studies
After removing duplicates, 2345 abstracts were screened, and 114 articles underwent full text review ( Figure 1). In total, 15 studies (n = 1825) were included for statistical analysis. [31][32][33][34][35][36][37][38][39][40][41][42][43][44][45] Baseline characteristics, treatments, and outcomes from the included studies are summarized in Table S1-S6, Supporting Information. Included studies were published from 2003 to 2021, most studies were retrospective (13 of 15 studies) and had median follow-up ranging from 20 to 66 months. Median study quality was 11 (Table S7) and no significant evidence of publication bias was identified (p = 0.81; Figure S1). Radiation treatment was designed using 2D RT (three studies), 3D-conformal RT (3D-CRT, five studies), or intensitymodulated RT (IMRT, six studies) and covered the appropriate lymph node basins. Treatment was delivered in the definitive (1 study) or postoperative (14 studies) setting. Diagnostic imaging (CT, MRI, or PET/CT) was reported in seven studies. The edition of the AJCC staging manual that was used was reported in 11 studies: the 7th edition was most commonly utilized (seven studies), followed by the 5th (two studies), 6th (one study), and 8th editions (one study). Definition of tumor (T) and nodal (N) stage was similar between AJCC 5th and 7th editions, while the 8th edition incorporated DOI and ENE (Table S8). Neck dissection (ipsilateral ± contralateral neck dissection) and use of chemotherapy were reported in 9 and 11 studies, respectively. The most commonly reported reason for using ipsilateral neck RT was for tumors that were lateralized (Table S9).
Pathologic T-and N-stage was available for 41 of the 56 patients (73%) with CNF following ipsilateral RT. Patients with more advanced disease constituted the highest percentage of CNF: T4, 56%; N2a-N2b/N3, 66% (Figure 4) Table S9. Among these studies that primarily included patients with lateralized and/or small primary tumors, 397 patients received ipsilateral neck RT and 369 patients received bilateral neck RT. Patients with bilateral or contralateral neck involvement (AJCC 7th ed. N2c disease) were excluded from this analysis. The rate of CNF following ipsilateral neck RT was 7.2% (95% CI 3.2-11.2), while the rate of CNF following bilateral neck RT was 4.0% (95% CI 0.40-7.6; Figure S9). Compared to bilateral treatment, ipsilateral neck radiation was not associated with a significantly greater risk of CNF (log odds ratio 0.216 [95% CI À0.917 to 1.348], p = 0.81; Figure S10). Subgroup analysis by tumor/nodal stage could not be performed due to lack of information specifically in the studies reporting outcomes after both ipsilateral and bilateral neck RT.

| Treatment and clinical factors influencing CNF
To elucidate risk factors associated with CNF, exploratory linear regression analyses were performed. Margin status (free, close, positive), tumor grade (grade 1, 2, 3), ENE, LVSI, PNI, use of contralateral neck dissection, chemotherapy use, and median follow-up were reported in 10, 12, 6, 11, 10, 8, 14 and 13 studies, respectively. Relationship between each risk factor and CNF, with corresponding coefficient of determination (R 2 ), are shown in Figures S11 and S12. For each study, the proportion of each risk factor is taken from the entire study population regardless of treatment received as data from patients who received only ipsilateral RT was not available for analysis. From these exploratory analyses, CNF appeared to increase with increasing proportion of ENE (R 2 = 0.35; Figure S11A) and use of contralateral dissection (R 2 = 0.20; Figure S11B).
Conversely, grade 1 histology was correlated with a decreased rate of CNF (R 2 = 0.10; Figure S11C). Margin status, LVSI, PNI, grade, chemotherapy use, and median follow-up all demonstrated a low degree of correlation with CNF ( Figure S12).

| DISCUSSION
In this systematic review and meta-analysis, we evaluated the rate of CNF after ipsilateral RT for OCSCC. This is an important endpoint given that salvage treatment for OCSCC is associated with poor outcomes [15][16][17] and that the risk of CNF is frequently cited to justify the use of ipsilateral RT in other head and neck cancers. European guidelines 26 for the management of OCSCC utilize nodal staging as the basis for treatment recommendations and suggest ipsilateral RT be considered for patients with 1 or more ipsilateral neck nodes ≤6 cm (AJCC 7th ed. N2a-N2b disease) so long as the primary tumor is well lateralized. However, American Society of Clinical Oncology (ASCO) guidelines 27 consider both tumor and nodal staging: ipsilateral RT is recommend for patients with T1-T2 tumors, primary tumors that do not approach midline, or those with 0-1 ipsilateral neck nodes ≤3 cm (AJCC 7th ed. N0-N1 disease).
Among all included studies, we found that the rate of CNF following ipsilateral RT is low (5.7%), and did not appear to differ significantly by oral cavity subsite, the staging edition utilized, or whether studies reported primary tumor lateralization. While the relationship between diagnostic imaging use and CNF rate demonstrated a trend toward significance, with studies reporting diagnostic imaging use having a higher CNF rate, one possible explanation is that these studies were older and utilized less sophisticated radiation planning techniques (e.g., 2D-RT or 3D-CRT). Ultimately, rather than be applied generally, the use of ipsilateral RT needs to be reserved for the patients at the lowest risk of CNF. Based on our data, CNF varies based on nodal stage and advanced T stage, and the use of ipsilateral RT in these situation must be carefully assessed.
Based on nodal stage, CNF following ipsilateral RT was lowest (1%-4%) in patients with 0-1 ipsilateral neck nodes (pathologic N0-N1 disease), while patients with multiple ipsilateral neck nodes involved or nodes greater than 3 to 6 cm (pathologic N2a-N2b/N3 disease) had a significantly higher rate of contralateral failure (17%). The relationship between number of involved lymph nodes and CNF could not be addressed due as this information was not reported by the included studies. This finding is most consistent with the ASCO guidelines 27 and suggests that patients with more than 1 ipsilateral lymph node should receive bilateral neck irradiation. Analysis by tumor stage was less straightforward. Pooled rate of CNF was lower for pathologic T1-T2 disease (3%) than pathologic T3-T4 disease (9%) but these rates were not significantly different statistically. This is likely due to the fact that only four studies were able to be used for this subgroup analysis: the majority of these studies limited ipsilateral RT only to lateralized primary tumors 31,40,44 or to primary tumors located in the buccal mucosa, gingiva, or retromolar trigone. 31,41,44 Examination of the failures following ipsilateral RT also suggests that the risk of CNF is high for patients with more advanced tumor stage. Among the CNF cases with recorded combined staging, T4 disease constituted over half (56%) of all failures and had the highest proportion of failures at a given nodal stage.
We also compared ipsilateral RT to bilateral RT in terms of CNF rate and we did not find a significant difference in CNF based on RT neck target (ipsilateral versus bilateral). One possible explanation for this is the heterogeneous nature of the studies included for this analysis. For example, four of the seven studies treated only lateralized tumors, which would be less likely to drain to the contralateral neck and carry an inherently lower risk of CNF. This selection bias would be expected to artificially decrease the effect size and observed "benefit" of bilateral neck RT with regard to the reduction of CNF risk. Additionally, studies that treated buccal, gingival, or retromolar trigone tumors (i.e., more lateralized tumor sites) appeared to favor ipsilateral RT, 38,43,44 while studies treating predominantly oral tongue tumors (i.e., midline tumor sites) appeared to favor bilateral RT. 33,39,42 Overall, our study had several strengths. First, this is the largest analysis of outcomes following ipsilateral RT for OCSCC. The patient population is heterogeneous, spans over multiple decades, and included multiple oral cavity subsites. Additionally, this analysis was designed prospectively, used appropriate statistics, and can be used to power future randomized studies.
This study does, however, have limitations. Because it is a study level meta-analysis, and not a patient level analysis, detailed comparisons between subgroups were not possible for every outcome of interest. For example, studies did not report CNF rates by specific tumor location. Given that tumor location is a surrogate for proximity to midline, and thus a higher likelihood of contralateral regional lymphatic drainage, we attempted to account for this by calculating pooled rates of CNF after separating studies with more lateralized tumors (e.g., retromolar trigone, gingiva) from those with more central tumors (tongue, floor of mouth) ( Figure S2). This, however, does not provide a precise estimate of CNF by each subsite within the oral cavity. Additionally, most of the included studies were retrospective, which constitutes its own selection bias within each study. Moreover, for the regression analyses, meta-analytical techniques could not be used due to the structure of available data. Thus, these correlations are at most hypotheses generating and require additional validation. Finally, the treatment of OCSCC required careful consideration of multiple risk factors such as tumor proximity to midline, margin status, PNI, LVSI, and DOI. While the presence of these features is associated with higher risk of primary tumor recurrence, and thus represent indications for adjuvant RT, they do not dictate whether to treat the ipsilateral or bilateral neck. While determining the association of these risk factors with CNF after ipsilateral RT would be useful clinically, we were unable to address or account for them due to available data. For example, DOI, was only reported in four studies, each of which used a different cutoff point (Table S4). Additionally, primary tumor distance from midline, which is also considered when deciding irradiation volumes, was not reported by the included studies, and data were not available to compare CNF after ipsilateral versus bilateral RT on the basis of tumor/ nodal stage. This dearth of information highlights need for consistent reporting of risk factors and outcomes in studies of patients with head and neck cancer.
Ultimately, these data further inform the risk-benefit discussion between physicians and patients with OCSCC. By limiting the radiation target, ipsilateral RT has the potential to improve treatment toxicity and patient reported outcomes. Given the concern for treatment failure, the decision to reduce the size of the irradiation field must be made by weighing several factors including tumor size, nodal involvement, proximity to midline, DOI, and lateralization. Our study demonstrates that the risk of CNF is lower than what is commonly assumed for patients with limited neck disease. However, due to the underlying heterogeneity of the included studies, we are unable to assess the impact of other risk factors on CNF. This precludes us from further delineating the subset of OCSCC patients who would benefit from ipsilateral RT. In light of these limitations, patients with more extensive neck involvement should receive bilateral neck RT. A prospective, randomized trial would be most appropriate to assess differences in CNF rates between bilateral and ipsilateral RT (NCT03622164). In the absence of prospectively collected data, our findings allow for more comprehensive decision making for patients with OCSCC.

| CONCLUSION
For patients with OCSCC, ipsilateral neck RT is associated with a low rate of CNF in those with 1 (<3 cm) or fewer neck nodes involved. The risk of CNF appears to increase in patients with T4 disease and those N2-N3 disease. This study provides the strongest level of evidence to date to support the use of ipsilateral neck RT in patients with well-lateralized OCSCC with limited neck involvement. Bilateral RT should be considered for patients with advanced tumor or nodal disease.