Improved Prognostic Strati cation of Patients With pN3b Oral Cavity Cancer Based on Maximum Standardized Uptake Value of Metastatic Nodes, Lymph Node Ratio, and Level of Cervical Nodal Metastases


 Objective

We sought to improve the prognostic stratification of patients with pN3b oral cavity squamous cell carcinoma (OCSCC) through a combined analysis of FDG-PET parameters and clinicopathological risk factors (RFs).
Methods

Complete data on maximum standardized uptake values derived from FDG-PET of neck metastatic nodes (SUV-nodal-max) and clinicopathological RFs were available for 257 patients with pN3b disease. Participants with pN0/pN1/pN2 (n = 1318/199/213) disease were included for comparison purposes.
Results

Using the 5-year disease-free survival (DFS) rate as the outcome of interest, the optimal cutoff points for SUV-nodal-max and lymph node ratio (LNR) were 15.9 and 0.17, respectively. After stratification with SUV-nodal-max and clinicopathological RFs, the 5-year DFS rates for patients with pN3 disease were as follows: SUV-nodal-max < 15.9 (n = 226) versus SUV-nodal-max ≥ 15.9 (n = 31), 49%/21%, p = 0.000003; LNR < 0.17 (n = 230) versus LNR ≥ 0.17 (n = 27), 49%/17%, p = 0.000117; absence (n = 230) versus presence (n = 27) of neck level IV/V metastases, 49%/15%, p = 0.000004. Multivariable analyses revealed that SUV-nodal-max ≥ 15.9, LNR ≥ 0.17, and the presence of level IV/V metastases were independent prognosticators for 5-year distant metastases (DM), DFS, disease-specific survival (DSS) and overall survival (OS) rates. Based on these variables, we devised a scoring system that identified three distinct prognostic subgroups of low (score 0, n = 190), intermediate (score 1, n = 51), and high (scores 2 − 3, n = 16) risk. The 5-year rates of pN0/pN1/pN2/pN3b of our cohort and the low/intermediate/high risk subgroups of pN3b were as follows: DM, 3%/12%/20%/38% and 31%/52%/89%; DFS, 83%/72%/65%/46% and 54%/26%/0%; DSS, 92%/79%/71%/52% and 59%/36%/8%; OS, 81%/62%/54%/38% and 42%/31%/6%, respectively; all p < 0.001.
Conclusion

A scoring system based on maximum SUV-nodal, LNR, and level IV/V metastases helps improve the prognostic stratification of OCSCC patients with pN3 disease.


Introduction
The traditional therapeutic approach for oral cavity squamous cell carcinoma (OCSCC) relies on radical surgery either with or without adjuvant therapy. The main determinants of clinical outcomes in patients with OCSCC include loco-regional and distant control. The presence and extent of cervical lymph node (LNs) metastases play a crucial role in prognostic strati cation and subsequent treatment decisions. However, the clinical course of patients with neck LNs metastases remains heterogeneous -ultimately suggesting that additional prognostic re nements are necessary [1]. Numerous risk factors (RFs) have been identi ed as independent prognostic variables in patients with OCSCC and neck LNs metastases. They include the number of pathologically positive LNs (pN+) [2,3], the LN ratio (LNR, de ned as the number of pathologically positive nodes divided by the number of dissected nodes) [3][4][5][6][7][8][9][10], the log odds of positive LNs (LODDS) [10], and the level of cervical LNs metastases (with level IV/V metastases being generally associated with an unfavorable prognosis) [1,11,12]. The presence of distant failure is generally associated with dismal outcomes (mean survival in Taiwanese patients: 2.4 months) [13]. While the distant metastasis (DM) rate of patients with OCSCC and extracapsular extension (ENE) may be as high as 30%, only 1% of all cases with distant spread are correctly identi ed during the preoperative workup [14].
Compared with the previous 2010 edition [15], the 2018 AJCC staging manual introduced a novel nodal classi cation for OCSCC termed pN3b disease -which comprises pN2 disease (de ned according to the 2010 edition) and ENE [16].
While this novel classi cation was aimed to improve prognostic strati cation, our pilot clinical experience revealed that the clinical outcomes of patients with pN3b disease are characterized by signi cant interindividual differences. Starting from these premises, this study was undertaken to further re ne the prognostic strati cation of patients with pN3b OCSCC through a combined analysis of FDG-PET parameters (collected as of 2001) and clinicopathological RFs (collected as of 1996).

Study setting
We retrospectively reviewed the clinical records of all patients diagnosed with rst primary OCSCC (n = 2256) who were consecutively referred to the Chang Gung Memorial Hospital (CGMH) from January 1996 to February 2019. Between August 2001 and February 2019, a total of 1310 patients underwent preoperative FDG-PET imaging. Owing to the prospective collection of data on tumor depth of invasion (DOI) and ENE, patients were staged according to the AJCC staging manual, eighth edition [16]. Patients who did not undergo neck dissection (ND, n = 133), with unavailable data on ENE (n = 3), and who had pN3b disease but did not undergo FDG-PET imaging (n = 133) were excluded -resulting in a nal cohort comprising 1987 cases. While the focus of this study was on the prognostic strati cation of pN3 disease, patients with pN0 − 2 disease were included for comparison purposes. The distribution of pathological nodal stage was as follows: pN0, n = 1318; pN1, n = 199; pN2, n = 213; and pN3b, n = 257, respectively. We were unable to identify any case with pN3a disease. Figure 1 depicts the ow of participants through the study. The presurgical evaluation and staging workup have been previously described in detail [1]. Clinicopathological RFs were collected in a prospective fashion by investigators blinded to clinical outcomes. In accordance with the recommendations set forth by the CAP Cancer Reporting Protocol [17], a dedicated checklist was used by two experienced headand-neck pathologists to independently review histopathological data. The study protocol was granted ethical approval by the local Institutional Review Board (references: CGMH 101-4457B and 202100048B0). The requirement for written consent was waived due to the study design. PET and CT data were consecutively acquired 50 min following 18 F-FDG injection. PET image reconstruction was carried out using an ordered-subset expectation maximization algorithm. A nuclear medicine team not blinded to patient clinical data reviewed all images in a prospective fashion. 18 F-FDG-avid foci were thoroughly identi ed and recorded. All images showing at least one area of increased 18 F-FDG uptake (i.e., with an intensity higher than that of the surrounding tissue) were considered as positive -provided that the foci did not correspond to areas of physiological tracer distribution on hybrid images. Regions of interest (ROIs) were positioned over visible lesions, as well as on simultaneously displayed axial, coronal, and sagittal images. Maximum standardized uptake value (SUV-max) was de ned as the highest activity concentration per injected dose per body weight (kg) after correction for radioactive decay. SUV-max values of the primary tumor and neck lymph nodes (SUV-nodal-max) were calculated in accordance with a previously described methodology [18].

Surgery and adjuvant therapy
Primary tumors were removed with ≥ 1 cm margins (both peripheral and deep margins). Patients with cN + disease underwent level I-IV or I-V NDs, whereas cN-patients received level I-III NDs. Patients who harbored pathological RFs were generally treated with post-operative radiotherapy (RT, 60 Gy) or concurrent chemoradiotherapy (CCRT, 66 Gy) [19][20][21]. RFs were assessed using the National Comprehensive Cancer Network (NCCN) guidelines until 2008 [22]; thereafter, the CGMH guidelines were adopted [23]. The radiation eld consisted of the entire tumor bed area (with 1-to 2-cm margins) and regional lymphatics. We used the following chemotherapy regimens: intravenous cisplatin 50 mg/m 2 biweekly plus daily oral tegafur 800 mg and leucovorin 60 mg, cisplatin 40 mg/m 2 weekly, or cisplatin 100 mg/ m 2 every three weeks [21). Patients who refused the proposed approaches or had unexpected evidence of disease stage modi cations in the postoperative period were treated with surgery alone.

Follow-up schedule and data collection
Postoperative follow-up visits were performed at prede ned time intervals according to the date of surgery, according to the following scheme: every 1-3 months throughout the rst postoperative year; every 2-4 months throughout the second year; and every 4-6 months between the third and the fth years. Patients who survived for more than 5 years were followed every 6-12 months. At each visit, data on the following clinical events were collected: local control (LC), neck control (NC), DM, disease-free survival (DFS), disease-speci c survival (DSS), and overall survival (OS).

Statistical analysis
The study variables are summarized using descriptive statistics -including frequencies, percentages, means, medians, ranges, and standard deviations (SD). All participants were followed up until February 2021 (minimum follow-up duration after surgery: 24 months).

Results
Optimal cutoff values for SUV-nodal-max and lymph node ratio in patients with pN3b disease A total of 16 variables -including SUV-nodal-max, LNR, and level IV/V metastases -were analyzed in relation to clinical outcomes of patients with pN3b disease. Using the 5-year DFS rate as the endpoint of interest, the optimal cutoff values for the SUV-nodal-max and LNR were 15. General characteristics of patients with OCSCC and pN3 disease Table 1 depicts the general characteristics of patients with pN3 disease.
The study endpoints included the 5-year rates of LC, NC, DM, DFS, DSS, and OS. All outcomes were calculated as the time from the date of surgery to the date of the event of interest. We calculated Kaplan-Meier probabilities to estimate time to each event, and a log-rank test was performed to assess differences between the groups. To evaluate the association between RFs and clinical outcomes, we performed univariable and multivariable Cox proportional hazard regression analyses. All variables examined at the univariable level were entered as covariates into the multivariable model. We censored for patient death and study completion in outcome analyses. Results are presented as hazard ratios (HRs) with their 95% con dence intervals (CIs). For all analyses, statistical signi cance was de ned as a two-tailed p value < 0.05. Table 1 General characteristics of patients with OCSCC and pN3b disease strati ed according to SUV-nodalmax < 15.9 versus ≥ 15.9 and lymph node ratio < 0.   Abbreviations: SUV, standardized uptake value; LNR, lymph node ratio; HR, hazard ratio; CI, con dence interval; ns, not signi cant.

Prognostic scoring system
On multivariable analysis, we identi ed SUV-nodal-max ≥ 15.9, LNR ≥ 0.17, and the presence of level IV/V metastases as independent RFs

Discussion
The 2018 AJCC staging manual has introduced a novel pN3b classi cation for patients with OCSCC. While the presence of pN3b disease is generally believed to portend poor outcomes, we found that a scoring system based on SUV-nodal-max ≥ 15. cutoff values for prognostic strati cation in patients with pN3b disease. Importantly, on analyzing the association of LNR with clinical outcomes, we found that this factor was not only associated with survival endpoints, but it also predicted the 5-year DM rates.
Accordingly, for patients with pN3 disease and LNR ≥ 0.17, we observed a 5-year DM rate of 66% − with all events occurring within the rst two postoperative years.
The prognostic role of the cervical nodal metastases level in patients with OCSCC has been evaluated in a limited number of studies [1,11,12]. These investigations found that, compared with patients with level I − III metastases, cases with level IV/V metastases (i.e., lower neck metastasis) have less favorable outcomes in the subgroups of patients with pN1 − 3 disease (DSS) and pN2 disease/ENE (NC, DM, DFS, DSS, and OS). While distant metastases are rarely identi ed (1% of all cases) during the preoperative workup of patients with level IV/V metastases, the risk of distant spread at 5 years is as high as 60% [12]. The 5-year DM rate observed in our study for cases with level IV/V metastases (69%) is in line with previous observations -albeit being slightly higher as a result of our focus on pN3b disease.
Several investigations involving patients with OCSCC have examined the clinical value of SUV-max calculated from FDG-PET images for preoperative staging [24], identifying second primary tumors [25], allocating patients to adjuvant therapy following tumor removal [26], and assessing response to treatment [27]. Our study is the rst to analyze the prognostic role of FDG-PET in patients with OCSCC and pN3 disease. Here, we found that a SUV-nodal-max ≥ 15.9 was independently associated with less favorable 5-year NC, DM, DFS, DSS, and OS rates. Notably, the 5-year DM rate of patients with SUV-nodal-max ≥ 15.9 was as high as 63%, and all events occurred within the rst 18 postoperative months. Figure 5 shows While being relatively rare (16/257, i.e., 6% of all patients with pN3b disease in our study), patients in the high-risk group had a markedly high 5-year DM rate (89%). Therefore, they should be considered as candidates for novel treatment approaches in clinical trials or receive palliative systemic therapy -rather than traditional CCRT. However, it can be argued that reliable data concerning LRN values cannot be obtained in the preoperative phase. In this scenario, we propose that patients with cN3 disease and concomitant evidence of SUV-nodalmax ≥ 15.9 and level IV/V metastases on imaging studies (FDG-PET and MRI/CT) should undergo non-surgical treatment to minimize morbidity related to radical tumor excision.
Several caveats of our study need to be considered. First, its single-center design may have limited the external validity of the results; in addition, the retrospective nature of our investigation could be associated with information bias. Second, the study sample was ethnically homogeneous and all patients were living in a betel quid chewing endemic area; for that reason, more studies are necessary to con rm our ndings in Western countries. Third, two different treatment guidelines were used throughout the study (NCCN guidelines before 2008 and CGMH guidelines thereafter). Future analyses strati ed by treatment guidelines should work to address this limitation. Finally, on analyzing the variables associated with clinical outcomes, data on clinicopathological RFs were prospectively collected as of 1996; however, FDG-PET imaging was systematically introduced only ve years thereafter (2001). Despite these limitations, these data represent a promising step in understanding the prognostic value of SUV-nodal-max, LNR, and the level of cervical nodal metastases in patients with OCSCC and pN3b disease.
On the one hand, our scoring system enables an objective assessment that is suitable for clinical prognostication. On the other hand, the use of our tool has the potential to tailor treatment at the individual level and can nd application in clinical trial design.

Declarations
Funding support This study did not receive speci c funding.

Con ict of interest statement
The authors declare no con icts of interest.
Availability of data and material The study protocol was granted ethical approval by the local Institutional Review Board (references: CGMH 101-4457B and 202100048B0). The requirement for written consent was waived due to the study design.

Consent to participate and Consent for publication
This manuscript has not been published or presented elsewhere in part or in entirety and is not under consideration by another journal. All authors have contribution to (1) data analysis and interpretation; (2) manuscript drafting or critical revision for important intellectual content; and (3) nal approval of the manuscript. We have read and understood your journal's policies, and we believe that neither the manuscript nor the study violates any of these. Figure 1 Flow of patients through the study.