Prediction of postoperative survival in patients with head and neck acinic cell carcinoma: a population-based study

Abstract

Background

Surgical intervention constitutes a pivotal therapeutic strategy for patients with head and neck acinic cell carcinoma (HNACC). However, precise long-term postoperative outcome prediction for individual patients remains suboptimal. This study aimed to develop and validate a nomogram model for “prognosis prediction and risk stratification” to evaluate postoperative outcomes in patients with HNACC.

Methods

Patients with HNACC diagnosed between 2004 and 2015 were extracted from the Surveillance, Epidemiology, and End Results (SEER) database and randomly allocated to training and validation cohorts in a 7:3 ratio. Cancer-specific survival (CSS) determinants were identified through Cox regression analysis. The nomogram’s predictive accuracy and calibration were assessed using the concordance index (C-index), calibration plots, decision curve analysis (DCA), and Kaplan-Meier survival curves. Comparative analysis with previous models was conducted via the receiver operating characteristic (ROC) curve and the C-index.

Results

A total of 931 surgically treated patients with HNACC were identified, of whom 651 comprised the training cohort for nomogram construction, while 280 constituted the validation cohort. Multivariate Cox proportional hazards regression analysis revealed tumor size, histological grade, American Joint Committee on Cancer (AJCC) stage, and M stage as independent prognostic factors for CSS. To enhance clinical decision-making, both clinically pertinent and statistically significant variables, including N stage and radiotherapy, were incorporated into the nomogram. The C-index for CSS prediction reached 0.898 in the training cohort and 0.840 in the validation cohort. The area under the curve (AUC) for 8- and 10-year survival in the training cohort was 0.905 [95% confidence interval (CI): 0.856–0.954] and 0.902 (95% CI: 0.856–0.948), respectively, while in the validation cohort, it was 0.824 (95% CI: 0.726–0.922) and 0.799 (95% CI: 0.685–0.899), respectively. Throughout the study, the nomogram demonstrated superior time-varying C-index and ROC curve performance compared to tumor-node-metastasis (TNM) staging and pathological grading.

Conclusions

The proposed nomogram model for “prognosis prediction and risk stratification” demonstrated enhanced prognostic accuracy for patients with HNACC following surgical treatment.

Highlight box.

Key findings

• This study analyzed 931 surgically treated patients with head and neck acinic cell carcinoma (HNACC) and found that tumor size, histological grade, American Joint Committee on Cancer (AJCC) stage, and M stage as independent prognostic factors for cancer-specific survival (CSS). A nomogram model for “prognosis prediction and risk stratification” was established to predict HNACC patients’ prognosis.

What is known and what is new?

• HNACC with high-grade malignancy and metastasis are accompanied with poor prognosis. The rarity and indolent nature of HNACC bring challenges in forecasting postoperative outcomes.

• Tumor grade and AJCC stage are the most influential factors for CSS in HNACC patients. The developed nomogram offers a more precise and individualized prediction of survival rates for patients with HNACC post-surgery.

What is the implication, and what should change now?

• For one, the proposed nomogram model for “prognosis prediction and risk stratification” provides an effective tool for assessing the HNACC patients’ survival. For another, further research is needed to integratnew data and research findings to optimize variable selection and maintain model relevance.

Introduction

Acinic cell carcinoma (AciCC) is a rare malignancy predominantly affecting the salivary glands of the head and neck, representing approximately 7% of parotid gland cancers (1). The incidence rate among adults is approximately 0.13 per 100,000, with a significant increase observed over the past four decades, particularly among males (2). AciCC demonstrates a higher prevalence in females and white individuals, with the typical onset occurring between the ages of 40 and 60 years (3,4).

This tumor is characterized by a slow progression and relatively favorable survival outcomes. However, cases exhibiting high-grade malignancy and metastasis to regional lymph nodes or distant organs are associated with markedly poor prognoses (5-7). Surgical resection remains the cornerstone of head and neck acinic cell carcinoma (HNACC) management, although the efficacy of adjuvant postoperative radiotherapy remains a subject of debate (7-10).

The rarity and indolent nature of HNACC present considerable challenges in forecasting postoperative outcomes. As an established prognostic tool for cancer individuals, the nomogram synthesizes multiple significant variables to predict patient outcomes (11,12). Although various nomograms have demonstrated robust prognostic accuracy (13,14), a substantial gap exists in models specifically tailored to HNACC prognosis, as existing models lack comprehensive clinical data relevant to this context. Moreover, no validated postoperative prognostic model for HNACC has been reported. Consequently, developing a reliable, precise, and comprehensive prognostic model that accurately reflects the multifaceted nature of HNACC is essential. This study, therefore, aimed to construct and validate a “prognosis prediction and risk stratification” nomogram model to estimate postoperative survival in patients with HNACC, utilizing a robust clinical dataset (Figure 1). We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1251/rc).

Figure 1.

Roadmap providing a comprehensive overview of the research process utilized in this population-based study. C-index, concordance index; CSS, cancer-specific survival; HNACC, head and neck acinic cell carcinoma; ROC, receiver operating characteristic; SEER, Surveillance, Epidemiology, and End Results; TNM, tumor-node-metastasis.

Methods

Patient selection and data collection

Data for this study were retrieved from the Surveillance, Epidemiology, and End Results (SEER) public database, which comprises 18 population-based cancer registries. The analysis included patients initially diagnosed with primary HNACC between 2004 and 2015, utilizing the SEER*8.4.1.0 dataset (covering the period from 2000 to 2019). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. As cancer was publicly reported in each state, patient informed consent was not required for SEER data usage. HNACC cases were identified using the AciCC code (8550/3) from the updated third edition of ICD-O, Ninth Revision (ICD-O-3) classification, combined with head and neck topography codes within the SEER database. Eligible patients were those who underwent surgical intervention. Cases identified exclusively via autopsy or death certificate, patients under 20 years of age, tumors exceeding 10 cm, and records lacking complete data on race, tumor-node-metastasis (TNM) staging, surgical procedures, radiotherapy, chemotherapy, or cause of death would be excluded. The patient selection flow is depicted in Figure 2. Following the inclusion and exclusion criteria, data from 931 patients with HNACC diagnosed between 2004 and 2015 were extracted. The cohort was randomly partitioned into a training set (n=651) for nomogram development and internal validation, and a validation set (n=280) for external validation.

Figure 2.

Flowchart depicting the patient selection process. SEER, Surveillance, Epidemiology, and End Results; TNM, tumor-node-metastasis.

The analysis of variables

The variables analyzed included age, sex, race, tumor size, tumor site, TNM staging based on the American Joint Committee on Cancer (AJCC) sixth edition, surgical intervention, radiotherapy, and chemotherapy. Age was categorized into three groups: 20–59, 60–74, and ≥75 years. Tumor size was stratified as <2, 2–3.9 cm, and ≥4 cm. Tumor sites were classified as the parotid gland or other head and neck regions. Lymph node involvement was categorized as N0, N1, and N2/N3.

Statistical analysis

Prognostic factors influencing HNACC outcomes were identified using univariate and multivariate analyses, employing Cox proportional hazards regression to calculate hazard ratio (HR) with 95% confidence intervals (CIs). The primary endpoint, cancer-specific survival (CSS), was assessed using the Cox regression model to identify key prognostic determinants. A nomogram was subsequently developed to predict 8- and 10-year CSS rates based on the aforementioned independent prognostic variables. Internal and external validation of the nomogram was performed using bootstrapping methods (15). The model’s predictive accuracy was quantified using the concordance index (C-index). Discriminative ability was assessed through the receiver operating characteristic (ROC) curve, calculating the area under the curve (AUC) as a measure of model performance. Calibration curves were constructed to evaluate the agreement between predicted and observed outcomes. Additionally, decision curve analysis (DCA) was employed to quantify the net clinical benefit of the nomogram at various threshold probabilities, thereby determining its practical applicability.

Results

Patient characteristics

Table 1 presents a comprehensive comparison of patient demographics between the training cohort (n=651) and the validation cohort (n=280). The majority of patients in both cohorts were under 60 years of age, accounting for 70.5% (n=459) in the training cohort and 72.5% (n=203) in the validation cohort. White individuals constituted the predominant racial group, with 511 cases (78.5%) in the training cohort and 228 cases (81.4%) in the validation cohort, significantly outnumbering Black and other racial categories. Gender distribution revealed 402 females (61.8%) and 249 males (38.2%) in the training cohort, while the validation cohort consisted of 171 females (61.2%) and 109 males (38.9%). The parotid gland was the most frequent primary site, comprising 93% (n=866) of the cases, with similar distributions in both cohorts: 92.9% (n=605) in the training cohort and 93.2% (n=261) in the validation cohort. Tumor sizes were predominantly below 4 cm in both groups. A notable proportion of patients in the training cohort (n=420) underwent neck dissection, while 638 patients had no record of chemotherapy or an indeterminate chemotherapy status. Statistical analysis revealed no significant differences between the training and validation cohorts regarding age, race, sex, primary site, tumor size, grade, AJCC stage, TNM grade, neck dissection, chemotherapy, or radiotherapy (all P>0.05).

Table 1. Characteristics of patients with head and neck acinic cell carcinoma in the training cohort and validation cohort.

Characteristic	Total (n=931)	Training cohort (n=651)	Validation cohort (n=280)	P
Age, n (%)				0.69
<60 years	662 (71.1)	459 (70.5)	203 (72.5)
60 to <75 years	216 (23.2)	156 (24.0)	60 (21.4)
≥75 years	53 (5.7)	36 (5.5)	17 (6.1)
Race, n (%)				0.40
White	739 (79.4)	511 (78.5)	228 (81.4)
Black	88 (9.5)	67 (10.3)	21 (7.5)
Other	104 (11.2)	73 (11.2)	31 (11.1)
Gender, n (%)				0.84
Female	573 (61.5)	402 (61.8)	171 (61.2)
Male	358 (38.5)	249 (38.2)	109 (38.9)
Site, n (%)				0.88
Parotid gland	866 (93.0)	605 (92.9)	261 (93.2)
Other	65 (7.0)	46 (7.1)	19 (6.8)
Tumor size, n (%)				0.66
<2 cm	406 (43.6)	288 (44.2)	118 (42.1)
2 to <4 cm	421 (45.2)	294 (45.2)	127 (45.4)
≥4 cm	104 (11.2)	69 (10.6)	35 (12.5)
Tumor grade, n (%)				0.14
I	212 (22.8)	160 (24.6)	52 (18.6)
II	134 (14.4)	97 (14.9)	37 (13.2)
III–IV	47 (5.0)	33 (5.1)	14 (5.0)
Unknown	538 (57.8)	361 (55.5)	177 (63.2)
AJCC stage, n (%)				0.58
I	451 (48.4)	320 (49.2)	131 (46.8)
II	304 (32.7)	214 (32.9)	90 (32.1)
III	106 (11.4)	68 (10.4)	38 (13.6)
IV	70 (7.5)	49 (7.5)	21 (7.5)
TNM grade
T, n (%)				0.70
T1	468 (50.3)	332 (51.0)	136 (48.6)
T2	323 (34.7)	227 (34.9)	96 (34.3)
T3	97 (10.4)	64 (9.8)	33 (11.8)
T4	43 (4.6)	28 (4.3)	15 (5.3)
N, n (%)				0.98
N0	869 (93.3)	607 (93.2)	262 (93.6)
N1	35 (3.8)	25 (3.8)	10 (3.6)
N2–3	27 (2.9)	19 (2.9)	8 (2.8)
M, n (%)				0.62
M0	926 (99.5)	647 (99.4)	279 (99.6)
M1	5 (0.5)	4 (0.6)	1 (0.4)
Neck dissection, n (%)				0.16
No/unknown	317 (34.0)	231 (35.5)	86 (30.7)
Yes	614 (66.0)	420 (64.5)	194 (69.3)
Radiotherapy, n (%)				0.35
No/unknown	560 (60.1)	398 (61.1)	162 (57.9)
Yes	371 (39.9)	253 (38.9)	118 (42.1)
Chemotherapy, n (%)				0.63
No/unknown	911 (97.9)	638 (98.0)	273 (97.5)
Yes	20 (2.1)	13 (2.0)	7 (2.5)

AJCC, American Joint Committee on Cancer; TNM, tumor-node-metastasis.

Screening for independent prognostic factors

The prognostic factors of 651 patients with HNACC were evaluated through univariate and multivariate Cox regression analyses. Variables exhibiting P values <0.1 in the univariate analysis were subsequently included in the multivariate model. As detailed in Table 2, tumor size, tumor grade, AJCC stage, and M stage emerged as independent prognostic factors for postoperative CSS in patients with HNACC. To enhance clinical decision-making, the prognostic model incorporated not only statistically significant variables but also clinically relevant factors lacking statistical significance. Therefore, N stage and radiotherapy were added, resulting in a total of six variables used to construct the nomogram for predicting CSS at 8 and 10 years post-surgery (Figure 3). The analysis indicated that tumor grade and AJCC stage were the most influential factors in the nomogram model.

Table 2. Univariate and multivariate Cox regression analysis of cancer-specific survival in head and neck acinic cell carcinoma.

Variables	Univariate analysis				Multivariate analysis
	HR	95% CI	P		HR	95% CI	P
Age, years
<60
60 to <75	1.052	0.561–1.976	0.87
≥75	1.425	0.509–3.995	0.50
Race
White
Black	0.728	0.290–1.826	0.50
Other	0.000	0.000–inf	>0.99
Gender
Female
Male	1.653	0.974–2.805	0.06
Site
Parotid gland
Other	0.471	0.115–1.935	0.30
Tumor size, cm
<2
2 to <4	5.104	2.262–11.517	<0.001		4.448	1.707–11.593	0.002
≥4	10.665	4.295–26.487	<0.001		2.104	0.718–6.169	0.17
Tumor grade
I
II	3.294	0.824–13.174	0.09		4.147	1.015–16.939	0.048
III–IV	65.016	19.461–217.205	<0.001		26.59	6.941–101.87	<0.001
Unknown	3.300	0.991–10.995	0.052		3.283	0.973–11.079	0.055
AJCC stage
I
II	2.372	0.902–6.237	0.08		0.834	0.282–2.465	0.74
III	11.171	4.501–27.726	<0.001		6.58	2.092–20.701	0.001
IV	37.686	16.103–88.200	<0.001		4.552	1.458–14.21	0.009
TNM grade
T
T1
T2	2.035	0.952–4.352	0.07
T3	10.854	5.292–22.258	<0.001
T4	10.672	4.345–26.212	<0.001
N
N0
N1	6.783	3.121–14.739	<0.001		0.78	0.305–1.991	0.60
N2–3	33.362	17.791–62.560	<0.001		2.368	0.849–6.609	0.10
M
M0
M1	24.618	7.528–80.506	<0.001		6.199	1.614–23.809	0.008
Neck dissection
No/unknown
Yes	3.422	1.617–7.243	0.001
Radiotherapy
No/unknown
Yes	3.927	2.195–7.029	<0.001		1.678	0.89–3.164	0.11
Chemotherapy
No/unknown
Yes	20.036	9.894–40.57	<0.001

AJCC, American Joint Committee on Cancer; CI, confidence interval; HR, hazard ratio; TNM, tumor-node-metastasis.

Figure 3.

Nomogram predicting the 8- and 10-year survival probability in patients with head and neck acinic cell carcinoma post-surgery. The predicted 8- or 10-year survival probability was calculated by summing the scores associated with each variable and mapping the aggregate score to the probability scale at the bottom. AJCC, American Joint Committee on Cancer; M, metastasis; N, node.

Validation of the prognostic nomogram

The HNACC nomogram demonstrated high prognostic accuracy, with a C-index of 0.898 (95% CI: 0.858–0.939) in the training cohort and 0.840 (95% CI: 0.767–0.914) in the validation cohort. The AUC values for predicting 8- and 10-year outcomes were 0.905 and 0.902, respectively, in the training cohort, and 0.824 and 0.799, respectively, in the validation cohort. The nomogram exhibited superior predictive performance compared to TNM staging and pathological grade, as evidenced by the time-dependent ROC curve for HNACC (Figure 4). Calibration curves indicated a high concordance between predicted and observed 8- and 10-year CSS in both cohorts (Figure 5). These findings confirm the nomogram’s robust discriminative ability and calibration accuracy. The DCA curve (Figure 6) demonstrated a consistent net benefit across threshold probabilities in both cohorts. Furthermore, the nomogram maintained superior discriminative capability over various survival intervals, as indicated by the time-dependent C-index for CSS, surpassing the performance of TNM staging and pathological grade in both cohorts (Figure 7A,7B).

Figure 4.

AUC values from ROC analyses comparing the nomogram, TNM staging, and pathological grade. (A,B) 8-year CSS in the training and validation cohorts; (C,D) 10-year CSS in the training and validation cohorts. AUC, area under the curve; CI, confidence interval; CSS, cancer-specific survival; ROC, receiver operating characteristic; TNM, tumor-node-metastasis.

Figure 5.

Calibration curves illustrating the accuracy of the nomogram. (A,B) 8-year CSS in the training and validation cohorts; (C,D) 10-year CSS in the training and validation cohorts. CSS, cancer-specific survival.

Figure 6.

DCA demonstrating the clinical benefit of the nomogram. (A) Predicting 8-year CSS; (B) predicting 10-year CSS. CSS, cancer-specific survival; DCA, decision curve analysis.

Figure 7.

Time-dependent C-index comparing nomogram, TNM staging, and pathological grade for predicting CSS. (A) Training cohort; (B) validation cohort. C-index, concordance index; CSS, cancer-specific survival; TNM, tumor-node-metastasis.

Risk stratification

Optimizing patient management through risk stratification, patients were classified into high-risk and low-risk groups based on median risk scores. Kaplan-Meier survival analysis revealed significantly improved CSS in the low-risk group compared to the high-risk group (Figure 8A). This survival advantage was also observed in the validation cohort, where the low-risk group consistently demonstrated superior CSS (Figure 8B).

Figure 8.

Kaplan-Meier survival curves stratified by risk group according to total scores. (A) Survival curves for the training cohort; (B) survival curves for the validation cohort.

Discussion

Recurrence rates following surgical resection of HNACC remain high despite its status as the primary treatment modality. The limited incidence of HNACC constrains the availability of prognostic data post-surgery. This large cohort study represents the newest postsurgical patients with HNACC analysis to evaluate the prognostic factors associated with this malignancy. Through Cox regression analysis, four key variables and various clinicopathological factors were identified to construct and validate a nomogram model for prognosis prediction and risk stratification. This model, integrating tumor and demographic characteristics, is designed to facilitate clinical decision-making and prognostic assessment.

The distribution of HNACC anatomical sites in this cohort was consistent with previous studies, with the parotid gland being the most commonly involved site, followed by other salivary glands and diverse head and neck regions (2,8,16-18). The surgical strategy for parotid AciCC—total, partial, or superficial parotidectomy—depends primarily on the tumor’s anatomical relationship with the facial nerve, a factor that critically influences patient outcomes (19,20). Accordingly, patients were stratified into parotid and non-parotid groups based on site distribution. Cox regression analysis revealed no significant difference in survival outcomes between these groups. However, the SEER database’s lack of specific information regarding surgical techniques for parotid AciCC limits further investigation into surgical approach efficacy.

Surgical resection remains the standard therapeutic option for HNACC, although the effectiveness of selective cervical lymph node dissection remains contentious. Given the increased risk of lymph node metastasis in advanced stages (21), selective neck dissection may offer clinical benefit. Scherl et al. (5) recommended neck dissection in parotid AciCC cases, citing the frequent occurrence of occult cervical lymph node metastasis. Additionally, van Herpen et al. (22) demonstrated that high-grade tumors and larger tumors had a high rate of occult neck metastases, and routine elective treatment should be considered for these tumors. In our data, 93.3% of the patients analyzed were N0, but 64.5% of them underwent neck dissection. It meant that in most patients with N0, no positive lymph nodes were found by dissection, which belonged to “overtreatment”. Else, this study found no clear survival advantage for patients with HNACC undergoing selective neck dissection, consistent with prior SEER-based analyses (18). This outcome may reflect the typically indolent nature of AciCC (8,23). Therefore, the selective intervention should be carried out according to individual risk factors (like tumor grade), rather than “one size fits all” cleaning.

The role of radiotherapy in HNACC management remains contentious. Some studies have reported favorable outcomes, including a 5-year disease-free survival rate of 91% for patients with parotid AciCC receiving postoperative radiotherapy (24). Moreover, a matched cohort study suggested postoperative radiotherapy for all major salivary gland malignancies (25). However, SEER database analyses have indicated that adjuvant radiotherapy does not significantly enhance survival outcomes (7,26), a finding corroborated by independent cohort studies (23,27,28). In this cohort, 293 patients (31%) who underwent radiotherapy exhibited poorer outcomes compared to those treated with surgery alone. This result warrants cautious interpretation, as radiotherapy is often administered in cases involving adverse prognostic factors such as positive surgical margins and high pathological grade (29).

TNM staging remains a fundamental approach for evaluating tumor treatment strategies and prognosis. AciCC typically exhibits low rates of lymph node and distant metastasis, rendering tumor size a pivotal predictor of recurrence. Scherl et al. (5) identified tumors exceeding 3 cm as an independent marker of poor prognosis. Moreover, numerous studies have demonstrated that pathological grade provides superior survival prediction compared to TNM staging (7). Previous investigations (27,30-35) have consistently highlighted pathological grade as a key prognostic factor in AciCC, yet its application as a clinical biomarker remains limited. In the current study, multivariate Cox regression analyses identified tumor grade and AJCC stage as the most significant post-surgical survival determinants in HNACC. The “prognosis prediction and risk stratification” nomogram model developed herein demonstrates clear advantages over both TNM staging and pathological grade.

The proposed nomogram model serves as a critical tool for patient counseling and clinical decision-making, offering enhanced predictive accuracy for post-surgical outcomes in HNACC. This study represents a pioneering effort in evaluating post-surgical prognostic factors for HNACC and constructing a corresponding nomogram. Predictive performance was assessed through C-indexes and calibration curves, while DCA confirmed its clinical applicability. ROC curve analysis and time-dependent C-index assessments demonstrated that the nomogram model significantly outperformed traditional TNM staging and pathological grade in accuracy. Kong et al. previously reported satisfactory calibration and discrimination with a C-index of 0.715 (95% CI: 0.679–0.751) for adrenocortical carcinoma, validated internally with a C-index of 0.672 (95% CI: 0.637–0.707) through incorporating four predictors (age at diagnosis, T stage, N stage, and M stage) for overall survival (36). In contrast, the present study reported superior results, with a C-index of 0.898 (95% CI: 0.858–0.939) in the training cohort and 0.840 (95% CI: 0.767–0.914) in the validation cohort. Similarly, Pan et al. developed a nomogram incorporating independent prognostic factors such as age, tumor site, operation, N stage, M stage, and TNM stage, demonstrating robust predictive capabilities with AUC values for the primary cohort of 0.80, 0.80, and 0.82 for 5-, 7-, and 9-year survival rates, respectively, and validation cohort AUC values of 0.73, 0.84, and 0.83 (37). In contrast, the present study reported higher AUC values for the nomogram model, with 0.905 and 0.902 for 8 and 10 years in the training cohort and 0.824 and 0.799 in the validation cohort, respectively.

In summary, the developed nomogram offers a more precise and individualized prediction of survival rates for patients with HNACC post-surgery, facilitating more effective risk stratification and aiding clinicians in optimizing patient management.

Despite its strengths, there are several limitations in this study. The SEER database does not provide detailed information regarding surgical type, surgical margins, and neck dissection, leading to their exclusion from the final model. However, the established nomogram model remains a valuable tool because the included variables in the nomogram are readily available and easy to generalize. Additionally, as a retrospective analysis, the validation of the “prognosis prediction and risk stratification” nomogram model through C-index, calibration curves, and ROC curves requires further prospective studies for external validation to enhance robustness. Moreover, in Table 1, the data labeled as ‘no’ and ‘unknown’ for variables such as ‘neck dissection’, ‘radiotherapy’, and ‘chemotherapy’ were not distinctly separated. Given that combined radiotherapy and chemotherapy have been shown to significantly influence survival in HNACC, the high proportion of “unknown” data may compromise the stability and predictive capacity of statistical models, potentially leading to an underestimation or overestimation of treatment effects. Addressing this issue through sensitivity analysis or data imputation methods is crucial to minimize bias in future research. Furthermore, some variables significant in univariate analysis were excluded from multivariate analysis due to high collinearity, which could induce multicollinearity and compromise model stability and interpretability, as observed with variables such as neck dissection and TNM grade. Chemotherapy was also excluded from the model owing to a substantial proportion of “unknown/no” data, as its inclusion could negatively impact statistical accuracy. To mitigate these limitations, future studies could benefit from expanding the research scope or utilizing multi-center data to increase sample size, thereby supporting multifactorial analysis of a broader set of variables. Continuous model verification and updates are recommended, integrating new data and research findings to optimize variable selection and maintain model relevance.

Conclusions

In conclusion, the “prognosis prediction and risk stratification” nomogram model demonstrates superior accuracy, clinical utility, and prognostic precision compared to conventional risk classification methods, establishing itself as an indispensable tool for predicting CSS following HNACC surgery. This model aids clinicians in assessing postoperative outcomes and formulating more tailored and effective treatment strategies for patients with HNACC.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.