Initial pN and differentiation at recurrence as prognostic factors in patients with recurrent or metastatic tongue squamous cell carcinoma undergoing re-resection

Abstract

Objective

This study primarily aimed to identify prognostic factors for patients with recurrent/metastatic tongue squamous cell carcinoma (R/M TSCC) undergoing re-resection and to preliminarily explore the value of baseline metabolic parameters in predicting pathological response after neoadjuvant therapy (NAT).

Materials and methods

In the primary analyses, we analyzed 114 patients with R/M TSCC who underwent direct re-resection or re-resection after NAT between 2017 and 2024. Clinical, pathological, and PET/CT parameters were collected. Lasso-Cox regression was performed to identify factors influencing disease-free survival (DFS) and overall survival (OS), with internal validation via bootstrapping with 1000 resamples. Sensitivity analysis was performed to test the robustness of the main findings by restricting the analysis to the 100 patients who underwent elective neck dissection (a subgroup of the 114 patients). Patients undergoing re-resection after NAT were grouped by pathological response (pCR/MPR vs. non-MPR) for comparison.

Results

Median follow-up was 43.0 months (110 were evaluable), with 53 deaths and 46 remaining recurrence-free. Initial pathological N status and differentiation at recurrence were prognostic factors for both DFS and OS, with HRs of 2.29 (95% CI: 1.40–3.77, P = 0.001), 2.62 (95% CI: 1.48–4.62, P < 0.001), 2.75 (95% CI: 1.57–4.80, P < 0.001), and 2.87 (95% CI: 1.47–5.61, P = 0.002), respectively. The robustness of initial pathological N status for OS, differentiation at recurrence for both DFS and OS were testified by sensitivity analyses. Among the 26 patients who underwent re-resection after NAT, the pCR/MPR group showed higher SULmax (P = 0.033) and SUR (P = 0.041) and had longer DFS (P = 0.045) than the non-MPR group, while no significant difference in OS was observed between the two subgroups.

Conclusions

Initial pN positivity is associated with worse OS, and moderate to low differentiation at recurrence might be associated with worse DFS and OS in R/M TSCC patients undergoing re-resection. Baseline metabolic parameters might have the potential to identify pathological response after NAT.

Clinical relevance

Identifying risk factors for survival in R/M TSCC patients treated with re-resection is crucial for tailoring postoperative strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12903-026-08933-z.

Introduction

Tongue cancer, as the most common type of oral cavity cancer, is characterized by a high recurrence rate, with reported recurrence rates ranging from 20% to 67% [1–3]. The prognosis of recurrent tongue cancer is poor, with a reported 5-year OS of 12% to 35% [4, 5]. Salvage surgery is the optimal treatment choice for patients with recurrent tongue cancer [6]. Previous research has suggested that surgery may achieve long-term favorable survival for patients with oligometastasis [7]. For patients with recurrent/metastatic tongue squamous cell carcinoma (R/M TSCC) treated with re-resection, identifying risk factors associated with survival is crucial for adjusting treatment strategies and improving prognosis.

¹⁸F-FDG PET/CT is recommended for patients with oral cavity carcinoma after radical surgery if symptoms arise or if there is a clinical suspicion of recurrence [8]. Some studies have shown that ¹⁸F-FDG PET volumetric parameters could serve as prognostic indicators in patients with oral cavity cancer [9, 10]. However, the prognostic value of PET/CT metabolic parameters for R/M TSCC patients is unclear. We hypothesized that baseline ¹⁸F-FDG PET imaging parameters can reflect the aggressiveness and metabolic characteristics of R/M TSCC, thereby possessing the ability to predict patients’ survival.

Surgery following neoadjuvant therapy (NAT) has shown favorable 1-year and 2-year DFS and OS for patients with resectable recurrent head and neck malignancies [11]. PET/CT imaging parameters have been applied to predict pathological response in esophageal carcinoma, breast cancer, non-small cell lung cancer, and other malignancies [12–15]. However, the predictive value of baseline PET/CT imaging parameters for pathological response and the survival outcome of patients with recurrent TSCC undergoing re-resection after NAT remains unknown.

Thereby, the primary aim of this study was to investigate the prognostic factors for patients with R/M TSCC. As a secondary objective, we preliminarily explored the differences in ¹⁸F-FDG PET imaging parameters and survival outcomes between the pathological response group and the non-response group in recurrent TSCC patients who underwent re-resection after NAT.

Materials and methods

Patients

We consecutively enrolled patients who were hospitalized at our institution between January 2017 and July 2024, had a history of radical surgery for oral tongue carcinoma, and had relapse identified on ¹⁸F-FDG PET/CT and verified by histopathology. Patients were excluded according to the exclusion criteria (Fig. 1). The patient enrollment flowchart was shown in Fig. 1. The median interval between ¹⁸F-FDG -PET/CT and re-treatment was 10 days, ranging from 6 to 30 days .Staging took place according to the 8th edition of the American Joint Committee on Cancer (AJCC) TNM classification [16]. R status was used to describe resection margin status [17]. The indications for adjuvant therapy after resection or re-resection was determined by surgical pathology according to National Comprehensive Cancer Network (NCCN) guidelines [18, 19].

Fig. 1.

The flowchart of patient enrollment

For patients with oligometastases, whole-body PET/CT was used to exclude widespread metastases, and R0 resection was deemed achievable. The surgical principles for performing re-resection were listed as follows: (1) In patients with local or locoregional recurrence, primary surgical resection with > 1 cm tumor-free margins (both lateral and deep) was performed. (2) In patients without clinical nodal involvement (N0), elective neck dissection (nodal levels I-III) was performed if patients were staged as rT1-2, therapeutic neck dissection (nodal levels I-IV/V) was performed if patients were staged as T3-4 (bilateral for midline tumor). (3) In patients with nodal involvement, unilateral or bilateral therapeutic neck dissection (nodal levels I-IV/V) was performed as appropriate. If elective or therapeutic neck dissection had been performed previously, the neck dissection would not cover the previously dissected nodal levels unless relapse recurred.

¹⁸F-FDG PET/CT scan

PET/CT was performed for either clinical suspicion of recurrence or restaging prior to treatment. It was performed at least 1 month after surgery or 3 months after adjuvant radiation/chemoradiation to avoid post-treatment inflammation [20, 21]. The median interval between completion of operation and PET/CT scan was 9 months (range: 1-107). ¹⁸F-FDG PET/CT scans were performed on the PET/CT scanner (Discovery MI, GE Healthcare, Milwaukee, WI, USA). All patients fasted for at least 6 h, and blood glucose levels were confirmed to be < 110 mg/dL before ¹⁸F-FDG administration. PET/CT was performed approximately 60 min after intravenous injection of ¹⁸F-FDG at a dose of 3.70 MBq/kg. First, a low-dose CT scan (110 KV, 120 mA, 512*512 matrix, pitch 1.75, reconstruction slice thickness and interval 3.75 mm) from the skull base to the aortopulmonary window or proximal thighs was performed for precise anatomical localization and attenuation correction. Subsequently, three-dimensional PET scan (thickness 3.27 mm) was performed with an acquisition time of 3 min per bed position. The acquired data were transferred to an Advantage Workstation (AW 4.7, GE Healthcare, Milwaukee, WI, USA) and reconstructed using ordered subset expectation maximization iterative reconstruction algorithm (2 iterations, 21 subsets).

FDG PET analysis

Reconstructed images were displayed as MIP image, PET, CT, and fused PET/CT in the axial, coronal, and sagittal planes on the Uws-MI workstation (United Imaging Healthcare, Shanghai). Any focal FDG uptake deviating from physiological distribution or background uptake was considered disease. We placed a semiautomatic spherical volume of interest (VOI) around the lesion and manually adjusted it to exclude areas of physiologic uptake from adjacent organs. An ROI of 20 mm in height and 10 mm in diameter was placed on the aortic arch on the fused PET/CT images to measure the SULmax of mediastinal blood pool, excluding the vessel wall. The max, mean and peak standardized uptake value corrected for lean body mass (SULmax, SULmean, and SULpeak) of each lesion were recorded. In patients with multiple metastatic lesions, the lesion with the most avid FDG uptake was used to determine the maximum, mean and peak SUL [22]. Then tumor-to-blood SULmax ratio (SUR) was calculated. The MTV of each lesion was calculated using a threshold of 40% of SULmax [23]. TLG was automatically computed by multiplying the MTV by the SULmean. We calculated MTV and TLG for all metastatic lesions. The total MTV and TLG were obtained by summing MTV and TLG of each lesion.

Definitions of local recurrence, regional recurrence, and distant metastasis

Local recurrence was defined as recurrence within the same or adjacent subsites of the oral cavity. Cervical lymph node metastasis was defined as regional recurrence. Recurrence beyond local or regional sites was defined as distant metastasis.

Statistical analysis

The primary endpoints of this study were disease-free survival (DFS) and overall survival (OS). DFS was calculated from the date of re-resection to the date of earliest re- recurrence or death from any cause, whichever occurred first. OS was calculated from the date of re-resection to the date of death from any cause. Both DFS and OS were censored at the date of the last follow-up in the absence of an event.

1:1 propensity score matching (PSM) was performed to minimize confounding bias between different subgroups. The optimal cut-off values for quantitative variables in survival analysis were determined using either ROC curve analysis or a median split, whichever yielded the highest statistical significance. Lasso regression analyses were employed to identify risk factors, and fivefold cross-validation was used to select the optimal penalty parameter (lambda min).The variables corresponding to lambda min were entered into a modified Bayesian Information Criterion (BIC) backward stepwise regression. Kaplan-Meier analysis with log-rank test was performed to compare DFS and OS between subgroups. Internal validation using the bootstrap method with 1000 resamples was performed to assess and correct for potential overfitting. P value < 0.05 was considered statistically significant. Statistical analyses were performed using R software (version 4.4.2) and GraphPad Prism 10 software.

Results

Clinical and pathological data

A total of 114 patients, with Eastern Cooperative Oncology Group Performance Status (ECOG PS) score of 0 to 1, were enrolled in this study. Among the 114 patients, 95 underwent whole-body ¹⁸F-FDG PET/CT, and 19 underwent head and neck ¹⁸F-FDG PET/CT as well as concurrent contrast-enhanced CT of the chest and abdomen to rule out distant metastases. 88 patients underwent direct re-resection, while 26 patients underwent re-resection after NAT. Clinical characteristics, baseline PET/CT metabolic parameters, initial and baseline pathological features of the direct re-resection group and re-resection after NAT group are summarized in Table 1. Pathological stage at relapse of 114 patients was displayed in Supplementary Table (1). For patients who underwent direct re-resection, complete surgical specimens were used for the assessment of differentiation at recurrence, and for patients who underwent re-resection after NAT, core biopsy specimens of regional recurrence or excisional biopsy of local recurrence before NAT were used for analysis. For patients with local/locoregional recurrence, differentiation at recurrence was assessed based on the local lesion; for patients with regional or distant metastasis, differentiation at recurrence was assessed based on the metastatic site. The detailed differentiation at recurrence is displayed in Supplementary Table (2). No statistically significant differences were found between the two subgroups, except for SULmax, SULmean, MTV and TLG. The cutoff values of the metabolic parameters for DFS and OS in R/M TSCC patients are shown in Supplementary Table 3.

Table 1.

Clinical, metabolic and pathological characteristics of patients with R/M TSCC

	Direct re-resection (n = 88)	Re-resection after NAT (n = 26)	P
Sex (M/F)	82/6	25/1	0.999
Age (y)	49.61 ± 9.50	52.46 ± 11.27	0.201
Smoking (Y/N)	57/31	18/8	0.674
Drinking (Y/N)	36/52	10/16	0.823
Betel quid chewing (Y/N)	68/20	19/7	0.658
initial pT (1–2/3–4)	76/12	22/4	0.758
initial pN (neg/pos)*	65/23	15/11	0.113
initial pTNM (I -II/III - IV)	58/30	15/11	0.443
Adjuvant therapy after initial surgery (Y/ N)	12/76	6/20	0.246
The need for adjuvant therapy after initial surgery (1/2/3/4)*	38/13/22/15	9/5/9/3	0.654
Number of surgeries (1/ > 1)	77/11	20/6	0.283
Time to recurrence (mo)	9.5 (1, 107)	7.0 (1, 62)	0.663
Recurrent site (local/regional/loco-regional/ metastasis)	24/50/10/4	4/16/6/0	0.257
rT (0/1–2 /3–4)	54/18/16	16/7/3	0.665
rN (0/1–2 /3)	28/32/28	4/12/10	0.267
rM (0/1)	84/4	26/0	-
rTNM (I - III/IV)	30/58	4/22	0.088
SULmax	10.32 ± 4.69	12.32 ± 3.58	0.048
SUR	5.81 ± 2.84	6.61 ± 2.14	0.186
SULmean	5.95 ± 2.84	7.47 ± 2.34	0.014
SULpeak	8.81 ± 4.10	9.14 ± 3.64	0.713
MTV (cm3)	5.56 (0.92, 58.98)	8.72 (1.12, 43.54)	0.044
TLG	29.58 (2.94, 487.41)	50.95 (13.23, 850.23)	0.003
R status (R0/R1) *	85/3	25/1	0.999
Differentiation at recurrence (well / moderate to low)	30/58	10/16	0.682
Adjuvant therapy after re-resection (Y/N)	44/44	15/11	0.490
The need for adjuvant therapy after re-resection (1/2/3) *	5/44/39	0/15/11	0.644

* Fourteen patients without clinical nodal involvement did not undergo neck dissection during the initial surgery and were classified as initial pN negative. Classification of the need for adjuvant therapy after initial surgery or re-resection: (1) No need; (2) Completed planned adjuvant therapy; (3) Did not complete planned adjuvant therapy; (4) Unclear. R status, resection margin status, R0, no residual tumor; R1, microscopic residual tumor or a resection margin less than 1 mm

Survival outcomes

Among the 114 patients, 4 were lost to follow-up, with their last outpatient visit date recorded as the censoring date. As of November 30, 2025, 68 patients had DFS events (with 54 recurrence and 14 deaths), and 53 patients had died. The median DFS was 15 months (range: 1–85 months), and the median OS was 39 months (range: 1–104 months). The 1-year, 2-year DFS rates for R/M TSCC patients were 50.64% and 43.91%, respectively. The 1-year, 3-year, and 5-year OS rates for R/M TSCC patients were 83.82%, 52.56%, and 44.88%, respectively.

We performed PSM to reduce confounding biases between the direct re-resection group and the re-resection after NAT group. For propensity score calculation, SULmax, SULmean, MTV, TLG and rTNM were selected using binary logistic regression model. Forty-eight patients were matched via PSM using a caliper of 0.2 times the standard deviation of the logit. PSM balanced the risk factors between the two groups (Supplementary Table 4). There was no statistically significant difference in DFS or OS between the two groups in both the unmatched and the matched cohorts, as shown in Supplementary Fig. 1.

LASSO-Cox regression analysis for DFS and OS

Since no significant difference was found in DFS or OS between the direct re-resection group and the re-resection after NAT group in the propensity score-matched cohorts (Supplementary Fig. 1C-D), the two subgroups were combined for further analysis.

LASSO-Cox regression analysis was used to select factors associated with DFS. The LASSO coefficient paths for clinical, metabolic, and pathological variables are illustrated in Fig. 2A. A total of 28 variables were subjected to LASSO regression, and 11 variables were retained, including sex, SUVpeak, SUVmean, TLG, time to recurrence, initial pN status, relapse T stage (rT), relapse M stage (rM), relapse TNM stage (rTNM), differentiation at recurrence and the need for adjuvant therapy after re-resection (Fig. 2B). Modified BIC results demonstrated that initial pN status and differentiation at recurrence were prognostic factors for DFS, as displayed in Fig. 2C. The apparent C-index was 0.65, with an optimism of -0.004 after bootstrap correction, yielding a corrected C-index of 0.66 (95% CI: 0.58–0.71). The model stratified patients into high-risk and low-risk groups for recurrence (Fig. 2D). Calibration curves illustrate the concordance between model-predicted and observed event rates (Fig. 2E-F).

Fig. 2.

LASSO-Cox analysis of DFS in R/M TSCC patients who underwent re-resection. A Tuning parameter (λ) selection in the LASSO regression using 5-fold cross-validation. B Profiles of the LASSO coefficients of the 28 candidate features, with 11 ultimately selected by LASSO-Cox regression. C Forest plot showing the independent risk factors for DFS identified by backward stepwise regression on the modified BIC. D Kaplan-Meier analysis comparing DFS between the high-risk and low-risk groups for re-recurrence, as stratified by LASSO-Cox model. (E-F) Calibration plots assessing the agreement between predicted and observed outcomes for 1-year and 2-year DFS, respectively

Of the 28 variables subjected to LASSO regression, 10 variables were retained at the optimal lambda value selected via five-fold cross-validation, including TLG, initial pN status, number of surgeries, adjuvant therapy after initial surgery, the need for adjuvant therapy after initial surgery, recurrent site, rT, rTNM, differentiation at recurrence, and the need for adjuvant therapy after re-resection (Fig. 3A-B). Through modified BIC regression, initial pN status and differentiation at recurrence were prognostic factors for OS, as shown in Fig. 3C. The apparent C-index was 0.68, with an optimism of 0.005 after bootstrap correction, yielding a corrected C-index of 0.67 (95%CI: 0.61–0.74). The model stratified patients into high-risk and low-risk groups for mortality (Fig. 3D). Calibration curves illustrate the concordance between model-predicted and observed event rates (Fig. 3E-F).

Fig. 3.

LASSO-Cox analysis of OS in R/M TSCC patients who underwent re-resection. A Tuning parameter (λ) selection in the LASSO regression using 5-fold cross-validation. B Profiles of the LASSO coefficients of all 28 candidate features, with 10 features ultimately selected by LASSO-Cox regression. C Forest plot showing the independent risk factors for OS identified by backward stepwise regression on the modified BIC. D Kaplan-Meier analysis comparing OS between the high-risk and low-risk groups, as stratified by the LASSO-Cox model for OS. (E-F) Calibration plots assessing the agreement between predicted and observed outcomes for 1-year and 3-year OS, respectively

Sensitivity analysis

To evaluate potential confounding related to initial pN status, we performed sensitivity analyses restricted to patients who underwent elective or therapeutic neck dissection (n = 100). The same analytical pipeline as the primary analysis was applied, including LASSO-Cox regression with 5-fold cross-validation and modified BIC backward stepwise regression. The results showed that initial pN status (HR = 2.73, 95% CI: 1.62–4.89, P < 0.001) and differentiation at recurrence (HR = 2.87, 95% CI: 1.38–5.99, P < 0.001) remained independent risk factors for OS. Differentiation at recurrence (HR = 2.41, 95% CI: 1.33–4.39, P = 0.004) remained an independent risk factor for DFS; however, initial pN status was discarded from the final model for DFS. The optimism-corrected C-index for the OS was 0.67 (95% CI: 0.60–0.75), and for the DFS was 0.60 (95% CI: 0.55–0.66).

Survival analysis for patients undergoing re-resection after NAT

Among the 26 patients who underwent re-resection after NAT, the neoadjuvant regimens included immunotherapy combined with targeted therapy, chemoimmunotherapy, chemotherapy combined with targeted therapy, and chemotherapy alone. NAT cycles ranged from 1 to 6. Imaging assessment was conducted according to RECIST 1.1 [24]. Pathological response was evaluated and categorized as pathological complete response (pCR), major pathological response (MPR), partial pathological response (pPR), or non-pathological response (NPR) based on established criteria [25, 26]. Detailed treatment and survival information for each patient is shown in Fig. 4. Radiological response was assessed and categorized as CR in 2 patients, PR in 12, SD in 6, PD in 5, and NE in 1. The distribution of pathological response was as follows: pCR in 4 patients, MPR in 3 patients, pPR in 3 patients, and NPR in 16 patients, as shown in Fig. 4. Patients achieving pCR/MPR were classified as the pathological response group, while those with pPR/NPR were classified as the non-response group. Further analysis suggested that the pathological response group had significantly higher baseline SULmax (P = 0.033) and SUR (P = 0.041) compared with the non-response group, as detailed in Table 2. The pathological response group exhibited longer DFS and OS than the non-response group. However, only the difference in DFS reached statistical significance in this limited sample (P = 0.045), while the OS did not (P = 0.102), as shown in Fig. 5.

Fig. 4.

Presentation of treatment regimens, neoadjuvant treatment cycles, treatment response, pathological remission, and survival outcomes in 26 patients who underwent re-resection after NAT

Table 2.

The comparison of baseline metabolic parameters in patients who underwent re-resection after NAT

	Pathological response group (n = 7)	Non-response group (n = 19)	P
SULmax	14.75 ± 4.51	11.42 ± 2.81	0.033
SULpeak	11.97 ± 4.97	8.09 ± 2.43	0.088
SULmean	8.93 ± 2.67	6.93 ± 2.02	0.051
SUR	8.00 ± 2.50	6.09 ± 1.80	0.041
MTV (cm3)	11.94 (5.04, 43.54)	7.97 (1.11, 36.75)	0.169
TLG	72.78 (25.57, 850.22)	43.43(13.63, 219.22)	0.188

Fig. 5.

Comparison of DFS (A) and OS (B) between the pCR/MPR group and the non-MPR group

Discussion

This study investigated the prognostic factors in patients with R/M TSCC who underwent re-resection. We found that initial pN status was a risk factor for OS, and differentiation at recurrence was a risk factor for both DFS and OS. For patients underwent re-resection after NAT, the pathological response group showed higher SULmax and SUR compared to the non-response group.

For R/M TSCC patients, the 1-year and 2-year DFS rate were 50.64% and 43.91%, respectively. The 5-year OS rate was 44.88%, which is consistent with the reported rate ranging from 37% to 43% in the literature [27, 28]. In our study, the differences in DFS and OS between patients receiving re-resection after NAT and those receiving direct re-resection were not statistically significant. Research by Hanna GJ et al. indicated favorable 1-year and 2-year DFS and OS for patients with resectable recurrent head and neck malignancies treated with neoadjuvant immunotherapy followed by surgery [11]. This suggests that re-resection after NAT may be a feasible option for recurrent TSCC patients, although further evaluation of this strategy is warranted.

We found that differentiation at recurrence is a risk factor for both DFS and OS, and initial pN status is a risk factor for OS both in the primary analysis and sensitivity analysis, which is in line with the literature [29, 30]. Notably, the assessment of differentiation at recurrence based on varying sites reduced data comparability and robustness of the results. Furthermore, for patients who underwent re-resection after NAT, biopsy specimens may not fully reflect the differentiation grade. In addition, for patients with locoregional recurrence, only the differentiation status of the local recurrent lesion was analyzed. These represent methodological limitations. Initial pN status is a risk factor for DFS in the primary analysis but is discarded in the sensitivity analysis, suggesting it is a less robust predictor for DFS. Patients with initial nodal metastasis are often associated with a higher recurrence risk and poorer prognosis than patients without. Furthermore, although adjuvant therapy was not identified as an independent prognostic factor for DFS and OS in the LASSO-Cox model, the proportion of patients with initial positive pN status (19/31, 61.3%) who did not complete planned adjuvant therapy after resection was high, which could indirectly contribute to poorer survival after re-resection. This study also found that differentiation at recurrence is a risk factor for survival, which is consistent with the finding of Weckx A et al. [31]. Although we did not find time to recurrence was associated with the survival in R/M TSCC patients, it has been reported that a late-onset recurrence was a protective indicator for survival in recurrent oral squamous cell cancer patients [6, 32].

In our study, TLG showed statistical significance for OS in the LASSO selection. However, TLG was discarded after modified BIC regression. It has been reported that MTV and TLG were associated with OS in recurrent oral cavity patients treated with salvage surgery [9, 33]. Although we did not observe a significant role of volumetric parameters (MTV and TLG) in identifying patients at a higher risk of mortality after re-resection, large-scale studies to further explore the association between volumetric parameters and survival in R/M TSCC treated with re-resection are required.

To the best of our knowledge, little is known about the association between pathological response and PET metabolic parameter in recurrent TSCC patients who underwent re-resection after NAT. We observed that the pathological response group showed longer DFS, higher SULmax and SUR compared to the non-response group. These preliminary observations suggest a potential link between baseline tumor metabolism and pathological response. However, given the small sample size and the exploratory design of this subgroup analysis, these observations require validation in larger, prospective cohorts. Despite the fact that we did not observe statistical significance in MTV and TLG between the pathological response group and the non-response group, a recent study has shown that patients achieving MPR have higher TLG [34]. Another study has shown that ΔSUV correlates with pathological response [35]. We were unable to investigate the association between ΔSUV and pathological response due to the lack of post-NAT PET/CT examinations. Future studies should address this question.

Among patients who underwent re-resection, four underwent oligometastasectomy: three had lung oligometastases and one had thyroid oligometastasis. Of the three patients with lung metastases, two survived without recurrence (survival of 20 and 60 months), and one died 18 months after re-resection. The patient with thyroid metastasis remains recurrence-free at 59 months of follow-up. It has been reported that the risk factors of distant metastasis are lymph node metastasis, locoregional recurrence, lymphovascular invasion, extranodal extension [36–38]. The common sites of distant metastasis are lung, skin, liver, bone and brain. Distant metastasis can be treated with local therapies (surgery and/or radiation), or systemic therapy (cytotoxic, immunotherapy, targeted therapy or combined therapy). Previous research suggests that surgery may achieve long-term favorable survival for patients with oligometastasis [7]. From our experience, whole-body ¹⁸F-FDG PET/CT contributes to accurately evaluating whether a patient has oligometastasis or polymetastases before treatment. Patients with oligometastasis treated with re-resection may achieve favorable outcomes.

This study has several limitations: First, the sample size was limited, thus, we chose LASSO-Cox regression to correct for overfitting. However, the model needs to be validated externally in prospective studies with larger sample size. Second, we did not include extranodal extension status during the initial surgery as a potential prognostic factor for survival analysis due to the extremely imbalance distribution of data (1 vs. 113), which is a major limitation of this study. Third, the recurrent sites of patients varied, precluding further analysis of pathological factors after re-resection (such as depth of invasion, perineural invasion, lymphovascular invasion and extra nodal extension). Fourth, some initial pathological factors were lacking because primary surgery of some patients was not conducted in our institution. Fifth, only a small subset of patients (n = 26) underwent re-resection after NAT, with heterogeneous treatment regimens, NAT cycles, and absence of post-NAT PET/CT scans. Thus, our observations warrant cautious interpretation and further studies. Sixth, lesions smaller than 1 cm on PET/CT were excluded because accurate semi-automatic delineation of the VOI was difficult. This may introduce selection bias and limit the generalizability of our findings to very small lesions. Finally, a small proportion of patients only underwent head and neck ¹⁸F-FDG PET/CT. However, these patients also underwent concurrent contrast-enhanced CT scans of the chest and abdomen to exclude the possibility of distant metastasis.

Conclusions

Initial positive pN positivity is associated with worse OS in R/M TSCC patients undergoing re-resection. Additionally, moderate to low differentiation at recurrence might be associated with worse DFS and OS. PET/CT metabolic parameters might have the potential in predicting pathological response to neoadjuvant therapy.

Abbreviations

DFS

Disease-free survival

NAT

Neoadjuvant therapy

NPR

Non-pathological response

MPR

Major pathological response

MTV

Metabolic tumor volume

OS

Overall survival

pCR

Pathological complete response

PET/CT

Positron Emission Tomography / Computed Tomography

pPR

Partial pathological response

R/M TSCC

Recurrent/metastatic tongue squamous cell carcinoma

SUL

Standardized uptake value normalized by lean body mass

TLG

Total Lesion Glycolysis

Authors' contributions

Xiaoping Yu and Xiaohuang Yang were involved in study concept and design. Hui Cao was responsible for collecting survival data. Kai Zheng and Wu Li were responsible for collecting clinical, PET-CT metabolic parameters and pathological data. Kai Zheng and Xiaohuang Yang were involved in analysis and interpretation of data. Xiaohuang Yang was responsible for drafting and revising the manuscript. Hui Ye and Xiaoping Yu were responsible for manuscript revision and study supervision. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by Hunan Provincial Natural Science Foundation of China (grant number: 2026JJ82072).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Hunan Cancer Hospital (approval number: KY20250721). The entire collection process adhered strictly to the ethical principles outlined in the Declaration of Helsinki. Written informed consent was waived due to its retrospective nature.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.