Prognostic significance of adjuvant therapy and specific radiation dosages in Taiwanese patients with oral cavity cancer and extra-nodal extension: a nationwide cohort study

Abstract

Background

The evidence for adjuvant chemoradiotherapy (CRT) of oral cavity squamous cell carcinoma (OCSCC) with extra-nodal extension (ENE) in National Comprehensive Cancer Network (NCCN) guidelines is derived from patients with head and neck cancer. The guidelines further suggest a radiation dose ranging from 6000 to 6600 cGy. In this nationwide study, we sought to evaluate the prognostic significance of adjuvant therapy and the specific radiation dosage in Taiwanese patients with pure OCSCC and ENE.

Methods

A retrospective analysis of 1577 OCSCC patients with ENE who underwent resection and received adjuvant CRT or radiotherapy (RT) between January 2011 and December 2020 was conducted.

Results

Multivariable analysis revealed that adjuvant RT, more than four pathologically positive nodes, and radiation dosage below 6000 cGy were independent risk factors for unfavorable 5-year disease-specific survival (DSS) and overall survival (OS). Comparing patients who received CRT (n = 1453) to those treated with RT (n = 124) before and after propensity score (PS) matching, the 5-year outcomes were as follows: before PS matching, DSS (54% versus 30%, p < 0.0001), OS (42% versus 18%, p < 0.0001); after PS matching (n = 111 in each group), DSS (52% versus 30%, p = 0.0016), OS (38% versus 21%, p = 0.0019). For patients who underwent CRT, the 5-year outcomes for different radiation dose groups (6600 − 7000 cGy, n = 1155 versus 6000 − 6500 cGy, n = 199) were as follows: before PS matching, DSS (52% versus 54%, p = 0.1904), OS (43% versus 46%, p = 0.1610); after PS matching (n = 199 in each group), DSS (55% versus 54%, p = 0.8374), OS (46.5% versus 46.3%, p = 0.7578).

Conclusions

For OCSCC patients with ENE, our study shows CRT improved survivals than RT alone, underscoring the clinical significance of chemotherapy. Patients undergoing CRT with irradiation doses ranging from 6000 to 6500 cGy exhibited comparable survival outcomes to those receiving doses of 6600–7000 cGy. This observation suggests that irradiation doses exceeding the 6600 cGy may not confer the survival advantage in these patients. Further research is needed to confirm our results and explore the optimal irradiation dose for managing these patients.

Introduction

The presence of cervical lymph node metastasis (pN+) significantly impacts the survival outcomes of patients with oral cavity squamous cell carcinoma (OCSCC) [1,2,3,4]. Moreover, among patients with pN + disease, extra-nodal extension (ENE), which refers to tumor penetration through the lymph node capsule, is associated with the poorest prognosis, leading to 5-year overall survival (OS) rates of approximately 30–35% [5, 6]. Currently, the most effective method for detecting nodal metastases and ENE is nodal dissection followed by histopathologic examination [7].

The American Joint Committee on Cancer (AJCC) staging manual, Eighth Edition, has introduced a new staging category called pN3b, which comprises both pN2 (according to the AJCC staging manual, Seventh Edition) and the presence of ENE [8]. The pN3b category recognizes the considerable prognostic implications of ENE, with the exception of human papillomavirus-associated cancers.

Two clinical trials involving patients with head and neck cancer and ENE consistently demonstrated the positive impact of adjuvant chemoradiotherapy (CRT) on 5-year OS rates, compared to radiotherapy (RT) alone [3]. Specifically, the European Organization Research and Treatment of Cancer (EORTC) #22,931 trial reported a 49% OS rate with adjuvant CRT, significantly higher than the 32% rate observed with RT alone (p = 0.0019). Similarly, the Radiation Therapy Oncology Group (RTOG) #9501 trial showed a trend toward improved OS rates, albeit with marginal significance (OS rate of 42% with adjuvant CRT, compared to 31% with RT alone, p = 0.063). These analyses took into account not only the prognostic impact of ENE but also the presence of positive margins. Based on these findings, the National Comprehensive Cancer Network (NCCN) guidelines advocate for the use of CRT as the preferred adjuvant therapy for patients with OCSCC and ENE. This recommendation is based on the potential to enhance locoregional control and improve overall treatment outcomes [3,4,5,6,7,8,9]. While the OCSCC subgroup accounted for approximately 26% of participants in the EORTC trial [10] and 27% in the RTOG trial [11], it is important to note that different tumor subsites within the head and neck area can display variations in specific characteristics and clinical outcomes.

The Taiwan Cancer Registry database (TCRD) has played a pivotal role in academic research and cancer control policies in Taiwan for over thirty years. It offers a comprehensive dataset that includes granular information on cancer stages, treatment modalities, and tumor recurrences [12]. The TCRD is currently responsible for gathering data from all major hospitals in Taiwan, ensuring coverage of approximately 99% of newly diagnosed OCSCC cases. Given the well-established prognostic implications of ENE, particularly after its inclusion in the AJCC staging manual, Eight Edition, and the lack of comprehensive studies examining the impact of adjuvant CRT in patients with OCSCC and ENE, we undertook a population-based study using the TCRD with two primary objectives. First, we sought to assess whether there is a significant difference in survival between patients with ENE who received adjuvant CRT compared to RT alone. Second, we investigated whether there is a survival disparity between patients with ENE who received radiation doses ≥ 6600 cGy and those treated with dosages between 6000 and 6500 cGy.

Materials and methods

Data sources

The study is reported in accordance with the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) checklist [13, 14]. Data were obtained from the “long-form” of the TCRD, which adheres to the guidelines specified in the Standards for Oncology Registry Entry (STORE) manual [15]. However, a limitation of TCRD is the lack of toxicity data for patients receiving adjuvant therapy and information on salvage therapy for patients experiencing disease relapse. Survival data were extracted from the Taiwanese National Health Insurance Research Dataset (NHIRD), which includes comprehensive personal information, diagnostic codes, prescription details, as well as records of surgical procedures and clinical visits for all insured individuals. The study was conducted according to the guidelines of the Declaration of Helsinki. All procedures were approved by the Chang Gung Medical Foundation Institutional Review Board (reference number: 201801398B0A3). The requirement for written patient informed consent was waived by Chang Gung Medical Foundation Institutional Review Board due to the study design. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.

Participants

From January 2011 to December 2020, a total of 47,025 patients were diagnosed with primary OCSCC based on the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) codes. Figure 1 provides a visual representation of the patient flow during the study. Exclusion criteria comprised: (1) a previous history of cancer (n = 10646), (2) initial non-surgical treatment (n = 5725), (3) unknown pathological stage (n = 967), (4) incomplete data on tumor depth, surgical margins, and ENE (n = 5286), (5) missing information on pathological lymph node metastases (n = 286), and (6) absence of data on tumor differentiation and subsites (n = 931). The study cohort ultimately included 23,184 patients of Taiwanese descent. The initial staging was based on the criteria outlined in the Seventh Edition (2010) of the AJCC Staging Manual. Subsequently, an updated classification aligned with the Eighth Edition (2018) of the AJCC Staging Manual was applied by incorporating factors such as depth of invasion (DOI) and ENE [8]. Out of the 23,184 potentially eligible patients, 26.9% (n = 6251) were diagnosed with pN + disease. Of them, 31.6% (n = 1974) exhibited ENE. Of the 1974 patients with ENE, we excluded those who received adjuvant chemotherapy alone (n = 59), surgery alone (n = 228), or a radiation dose > 7000 cGy (n = 110). As a result, the remaining 1577 patients with ENE were considered. Among them, 92.1% (n = 1,453) underwent surgery followed by adjuvant CRT, while 7.9% (n = 124) received surgery followed by adjuvant RT. In the group of patients who underwent surgery and adjuvant CRT, our focus was specifically on those who received a conventional RT dose within the range of 6000 − 7000 cGy. Out of the 1453 patients in the adjuvant CRT group, 79.5% (1155 out of 1453) received a radiation dose within the range of 6600 − 7000 cGy, whereas 13.7% (199 out of 1453) received a dose ranging from 6000 to 6500 cGy. To determine the follow-up period, we tracked each patient from the day of surgery until either the end of the study (December 2021) or the date of death, whichever occurred first.

Fig. 1

Flow diagram of the study patients

Data collection

The study variables were obtained from the TCRD, released in 2020, and the NHIRD, released in 2021. Data analyses were conducted in October 2023. The collected variables included oral tumor subsites, sex, age at diagnosis, DOI, nearest resection margin, tumor differentiation, pathologic T status, number of pathologically positive nodes, RT doses, and weighted Charlson comorbidity index (CCI) [16]. The data regarding OCSCC-associated morbidity and mortality were sourced from the NHIRD, enabling the calculation of disease-specific survival (DSS) and OS. However, the database lacked information on salvage treatments for patients who experienced disease relapse.

Statistical analysis

The survival durations were calculated from the date of ablative surgery. We used Kaplan-Meier plots to illustrate survival curves for OS and DSS, and their differences were evaluated with the log-rank test. A stepwise selection method was employed for multivariable Cox proportional hazards analysis, integrating all variables from the univariable analysis into the final regression model. The outcomes are presented as hazard ratios (HRs) along with their corresponding 95% confidence intervals (CIs). Analyses were conducted using SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA) and R, version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). To mitigate the differences in disease severity between patients groups, we used propensity score (PS) matching. All tests were two-tailed, and statistical significance was determined by a p value of less than 0.05.

Results

Patient characteristics

Table 1 displays the baseline characteristics of patients with OCSCC and ENE (n = 1577) who underwent surgery followed by either adjuvant CRT or RT. The two most common tumor subsites were the tongue (n = 671, 42.5%) and buccal mucosa (n = 505, 32.0%). Male patients constituted the majority of the study cohort (n = 1419, 90.0%). Before PS matching, the RT group showed a significantly higher prevalence of several risk factors compared to the CRT group, including older age, a radiation dose below 6000 cGy, and a higher weighted CCI score (all p values < 0.0001) [17].

Table 1 General characteristics of patients with extra-nodal extension who received adjuvant chemoradiotherapy and radiotherapy, before (n = 1577) and after (n = 222) propensity score matching

Specifically, the age distribution of the RT group versus the CRT group was as follows: <65 years old, 65 − 74 years old, and ≥ 75 years old: 50.0%, 26.6%, and 23.4% versus 87.6%, 11.1%, and 1.3%, respectively. Conversely, the CCI distribution of the RT group versus the CRT group was as follows: score 0, 1 − 2, and ≥ 3: 36.3%, 45.2%, and 18.5% versus 60.1%, 32.9%, and 7.0%, respectively. To mitigate baseline intergroup differences (Table 1), PS matching was applied, resulting in two PS-matched groups: one with 111 patients who received adjuvant CRT, and the other with 111 patients who had undergone RT.

Optimal cutoff values for the number of pathologically positive nodes and weighted charlson comorbidity index

The adjusted HRs for DSS and OS were calculated based on the number of pathologically positive nodes and the weighted CCI, treated as continuous variables using a penalized spline method (Supplementary Fig. 1). This approach allowed determining the non-linear relationship between the number of pathologically positive nodes and the weighted CCI with the log-transformed HRs for DSS and OS. Through recursive partitioning analysis, we identified the optimal cutoff values for the number of pathologically positive nodes (< 4 nodes versus ≥ 4 nodes) and the weighted CCI score (0 versus ≥ 1) in predicting both DSS and OS. Notably, significant increases in the HRs for DSS and OS were observed when the number of pathologically positive nodes was ≥ 4 and the weighted CCI score was ≥ 1 [18].

Survival outcomes and subgroup analyses after propensity score matching

The 5-year DSS and OS rates were 49% and 40%, respectively. When patients who received adjuvant RT were compared to those who underwent CRT, the 5-year DSS and OS rates were significantly different at 30% versus 54% (p < 0.0001; Fig. 2A) and 18% versus 42% (p < 0.0001; Fig. 2B), respectively. Similarly, the results of PS-matched analysis revealed that patients who underwent RT had less favorable 5-year DSS (30% versus 52%, p = 0.0016; Fig. 2C) and OS (21% versus 38%, p = 0.0019; Fig. 2D) compared to those who received CRT.

Fig. 2

Kaplan-Meier plots of 5-year disease-specific survival and overall survival for patients who were treated with adjuvant chemoradiotherapy versus radiotherapy, before (panels A and B) and after (panels C and D) propensity score matching

Univariable and multivariable Cox regression analyses

Table 2 presents the outcomes of univariable and multivariable Cox regression analyses. After accounting for potential confounding variables, multivariable analysis identified several factors that independently correlated with a less favorable DSS rate. These included treatment with adjuvant RT, an elevated DOI, a positive surgical margin, a nearest resection margin of less than 5 mm, pT4 tumors, the presence of at least 4 pathologically positive nodes, and a radiation dose of less than 6000 cGy. An unfavorable 5-year OS was independently associated with treatment with adjuvant RT, an elevated DOI, a positive surgical margin, a nearest resection margin of less than 5 mm, poor tumor differentiation, pT4 tumors, the presence of at least 4 pathologically positive nodes, a radiation dose of less than 6000 cGy, and a weighted CCI score of 1 or higher (Table 2).

Table 2 Univariable and multivariable analyses of risk factors for 5-year disease-specific survival and overall survival in patients with extra-nodal extension (n = 1577)

Survival outcomes in patients who received adjuvant chemoradiotherapy according to different radiation doses

Finally, we examined the 5-year DSS and OS rates of patients who underwent CRT with varying radiation doses, as follows: 6600 − 7000 cGy (n = 1155), 6000 − 6500 cGy (n = 199), and < 6000 cGy (n = 99). The DSS rates in three groups were 52%, 54%, and 33%, respectively (p < 0.0001), whereas the OS rates stood at 43%, 46%, and 23%, respectively (p < 0.0001). We further compared the DSS and OS rates between patients who received 6600 − 7000 cGy and 6000 − 6500 cGy. No significant differences were observed (p = 0.1904 and 0.1610, respectively; Fig. 3A-B). To account for baseline differences, we used PS matching and created two matched groups of 199 patients each (Table 3). In the PS-matched analysis, the 5-year DSS rates (55% versus 54%, p = 0.8374) and OS rates (46.5% versus 46.3%, p = 0.7578) remained comparable between the two subgroups (Fig. 3C-D). However, due to the limited sample size of patients who received less than 6000 cGy (n = 99), simultaneous PS matching across all three subgroups was not feasible.

Fig. 3

Kaplan-Meier plots of 5-year disease-specific survival and overall survival for patients who were treated with adjuvant chemoradiotherapy using different radiation doses, before (panels A and B, dose ≥ 6600 cGy versus 6000 − 6500 cGy versus < 6000 cGy) and after (panels C and D, dose ≥ 6600 cGy versus 6000 − 6500 cGy) propensity score matching

Table 3 General characteristics of patients with extra-nodal extension who received adjuvant chemoradiotherapy stratified according to the radiotherapy dosage, before (n = 1453) and after (n = 398) propensity score matching

Discussion

Recruiting a large cohort of patients with OCSCC and ensuring their long-term follow-up at a single medical institution to assess the prognostic implications of ENE over an extended time period may be challenging. The current study leveraged real-world nationwide data from Taiwan obtained through the TCRD program to address this limitation. This approach allowed us to focus specifically on OCSCC patients with ENE while minimizing selection bias. The utilization of PS matching added an extra layer of methodological rigor to our research. Remarkably, the outcomes of log-rank tests unveiled that, both before and after PS matching, patients who underwent adjuvant CRT experienced significantly better OS and DSS outcomes in comparison to those who were solely treated with RT. This finding underscores the significance of chemotherapy in the management of patients with OCSCC and ENE, as it may address the high incidence of distant metastasis within this particular patient cohort [19]. Multivariable analysis also revealed that receiving adjuvant RT and a radiation dose of less than 6000 cGy were independent risk factors for less favorable DSS and OS. Interestingly, patients who underwent CRT and received radiation doses of 6600 − 7000 cGy and 6000 − 6500 cGy had similar 5-year DSS and OS rates. These findings may provide clues for improving adjuvant therapy and determining the optimal radiation dosages for patients with OCSCC and ENE. Further randomized prospective studies are needed to validate these results and identify risk factors that may require higher doses.

Our research has significant implications for both patient management and prognostic stratification. Particularly, in cases where patients have positive lymph nodes but no evidence of ENE, caution should be exercised when considering the addition of chemotherapy to RT. Accordingly, this combination has been found to have a detrimental effect on OS compared to RT alone [20]. However, Trifiletti et al. conducted a retrospective cohort study of 10,870 patients with resected, locally advanced head and neck cancer using the National Cancer Database. Their findings suggested that adjuvant CRT provided an OS benefit compared to adjuvant RT alone, even in patients without ENE and positive margins, particularly in those with multiple metastatic neck lymph nodes [21]. Our study provides valuable insights into the potential survival advantage conferred by CRT for patients with ENE. A previous retrospective study involving 92 patients with OCSCC revealed that the distant failure rate in patients with macroscopic ENE (> 2 mm) was 44.83% [22]. Furthermore, the tumor microenvironment of primary OCSCC, characterized by high tumor budding and a pathological DOI > 10 mm, has shown a significant association with the occurrence of ENE [23]. Consequently, the presence of ENE reflects a high locoregional tumor burden and an increased likelihood of malignant cells entering the bloodstream, posing challenges for effective tumor clearance [24]. Additionally, patients diagnosed with OCSCC and ENE may frequently experience a notable decline in their quality of life and functional abilities. This can be attributed to the extensive spread of the tumor and the need for multidisciplinary treatment approaches [25]. These observations strongly support the use of systemic chemotherapy in this clinical scenario.

In both the EORTC #22,931 trial (with a RT dose of 6600 cGy) [10] and the RTOG #9501 trial (with RT doses ranging between 6000 and 6600 cGy) [11], CRT significantly enhanced the rates of local and regional control as well as disease-free survival when compared to RT alone in high-risk head and neck cancer patients, all of whom included cases with ENE. However, severe acute (grade 3 or higher) and late toxicities were prevalent in the CRT groups (with acute toxicities at 41% and 77% in the EORTC and RTOG trials respectively, and late toxicities at 40% and 21% in the EORTC and RTOG trials respectively). In the EROTC trial, only patients received less than 60 Gy were considered as protocol violation. In term of the possible acute and late toxicities, it should be more imperative to increase radiation dose to 6600 cGy. Although the EORTC #22,931 trial resulted in a significantly higher 5-year OS difference between groups compared to the RTOG #9501 trial (17% versus 11%, p = 0.0019 versus p = 0.063), our real-world data found similar survival outcomes among patients with OSCC and ENE who underwent adjuvant CRT, regardless of whether they received doses of 6600 − 7000 cGy or 6000 − 6500 cGy. These findings highlight the importance of chemotherapy and suggest that a RT dose of 6000 cGy or higher could yield similar survival outcomes. Further research is necessary to confirm this possibility and determine the optimal radiation dose for managing patients with OSCC and ENE.

In patients who have undergone surgery for OCSCC with ENE, nodal characteristics – including the number of ENE-positive nodes, lymph node density, and ENE grading – may contribute to risk stratification and the optimization of adjuvant therapy [26]. Recent studies have shed light on the prognostic significance of nodal metastases, suggesting that it may vary based on the number of pathologically positive nodes and the extent of ENE. For instance, Arun et al. [27] found that patients with multiple nodes exhibiting ENE had a poorer OS, although disease-free survival was not significantly different, compared to those with ENE in a single node. In addition, Joshi et al. [22] reported a stark difference in the 2-year survival rates of patients with OCSCC based on the size of ENE. Patients with microscopic ENE (≤ 2 mm) had a survival rate of 72.6%, whereas those with macroscopic ENE (> 2 mm) had a survival rate of 0%. While adjuvant CRT has been shown to improve survival in patients with ENE > 2 mm, its benefit for those with microscopic ENE remains unclear [28]. Our study further contributes to this body of knowledge by demonstrating that harboring at least four pathologically positive nodes is an independent risk factor for both OS and DSS in patients with OCSCC and ENE. Moreover, among patients with pN3b disease, we have previously shown that having at least eight pathologically positive nodes or at least five nodes with ENE independently predicted poorer survival outcomes and a higher risk of distant metastases [4]. These findings suggest that future research may need to focus on optimizing treatment strategies based on the number and pathological extent of ENE.

The current study has several significant strengths. Unlike previous investigations such as the EORTC and RTOC trials, which included a variety of tumor subsites and treatment methods, our research specifically focused on surgically treated OCSCC. Additionally, we conducted a thorough examination of a significant number of patients with histopathologically confirmed ENE, enabling a comprehensive analysis of a uniformly treated group [29]. Finally, by utilizing the TCRD as the data source and implementing PS matching, we effectively minimized the risk of spurious findings commonly associated with limited sample sizes. However, we must acknowledge several limitations. The retrospective design may introduce selection bias, coding errors, and missing data. The rationale behind the decision not to administer adjuvant CRT to a small subset of patients with ENE (8% [124/1577]) remains unclear. Patients in the RT group were older, with a mean age of 63.23 years compared to 52.72 years in the CRT group, and had a higher CCI, with 63.7% having a CCI of ≥ 1 compared to 39.9% in the CRT group (Table 1). Potential explanations for this treatment decision may include advanced age, reduced compliance due to performance status and/or comorbidities, patient refusal, and insufficient family support. While we have performed PS matching on both cohorts, we were unable to perform 2:1 or 3:1 matching due to a significant decrease in the number of patients in both cohorts. In addition, the impact of ENE on survival outcomes in OCSCC is not solely determined by its presence, but also by its extent [4]. Regrettably, our dataset did not provide the specific count of nodes exhibiting evidence of ENE. In addition, our sample was drawn from an area where areca quid chewing is endemic, which could introduce a potential geographic bias and restrict the generalizability of our findings. Finally, while cisplatin-based protocols are well-established in the systemic treatment of head and neck malignancies [4, 5, 9, 19], our study datasets lacked information on specific chemotherapy regimens, including administration dosages and frequencies. This limitation prevented us from assessing the uniformity of treatments across the study cohort and conducting a comprehensive analysis of the prognostic impact of various chemotherapy protocols.

Conclusions

Adjuvant CRT, as opposed to adjuvant RT alone, provides better 5-year DSS and OS for patients with surgically treated OCSCC and ENE. Interestingly, survival outcomes were similar between patients who were treated with a radiation dose of 6600 − 7000 cGy and those who received 6000 − 6500 cGy as part of adjuvant CRT. Nonetheless, to confirm the reliability of these results, it is essential to conduct additional large-scale prospective studies.