A prognostic nomogram integrating Epstein-Barr virus DNA and specific inflammatory indices for early recurrent or metastatic nasopharyngeal carcinoma after first-line nedaplatin-based therapy

Abstract

Purpose

This study developed and validated a nomogram integrating specific inflammatory and Epstein-Barr virus (EBV) DNA quantification to predict overall survival (OS) in patients with nasopharyngeal carcinoma (NPC) who experienced recurrence or metastasis within ten years after completing first-line treatment with nedaplatin.

Methods

The nomogram was constructed utilizing data from a retrospective cohort of 120 patients treated at The People’s Hospital of Guangxi Zhuang Autonomous Region between January 2017 and December 2022. The predictive accuracy and clinical utility of the model were assessed through the concordance index (C-index), decision curve analysis (DCA), calibration plots, and time-dependent receiver operating characteristic (ROC) curves.

Results

Multivariate analysis revealed that an elevated albumin-to-globulin ratio (AGR) and lymphocyte-to-monocyte ratio (LMR) were significantly correlated with improved OS, whereas increased levels of alkaline phosphatase (ALP) and the presence of EBV DNA were associated with poorer prognostic outcomes. These four variables were integrated into a nomogram, which exhibited robust predictive accuracy, as evidenced by a C-index of 0.747 (95% CI, 0.689–0.805). This performance marked a substantial enhancement in discriminatory ability relative to the 8th edition UICC/AJCC staging system, which yielded a C-index of 0.566 (95% CI, 0.509–0.624). Additionally, the nomogram demonstrated satisfactory calibration, effectively stratified patients by risk (log-rank test, P < 0.001), and showed clinical applicability as confirmed through DCA.

Conclusions

This study successfully constructed a nomogram model for patients with NPC who experienced recurrence or metastasis within ten years following the completion of first-line treatment with nedaplatin. This model functions as an adjunct to the 8th edition UICC/AJCC staging system and provides significant support for clinical decision-making processes.

Introduction

Nasopharyngeal carcinoma (NPC) is a malignant neoplasm arising from the epithelial lining of the nasopharyngeal mucosa, it predominantly develops in the pharyngeal recess and is characterized by aggressive local invasion and a high propensity for distant metastasis [1]. Historically, NPC was referred to as the “Canton tumor” owing to its notably elevated prevalence in Guangdong Province, located in southern China, during the early twentieth century [2]. Epidemiological evidence indicates that NPC exhibits pronounced geographic clustering and ethnic variation. Data from GLOBOCAN 2022 report a global incidence of 120,434 new NPC cases and 73,485 associated deaths. Notably, China bears the greatest disease burden, accounting for 51,010 new cases and 28,408 deaths [3]. Significantly, approximately 70% of NPC cases worldwide are predominantly found in regions such as Southeast Asia, Southern China, North Africa, Greenland, and Alaska, whereas the incidence of the disease remains comparatively low in Western countries [4, 5].

The clinical management of NPC encounters three primary prognostic challenges: local tumor recurrence, distant metastatic spread, and the development of acquired drug resistance [6]. The overall rate of recurrence and metastasis within five years following initial treatment varies between 8.2% and 22.0% [7]. While this constitutes a notable advancement relative to the roughly 40% recurrence rate observed within 1 to 2 years following initial radiation therapy, the incidence of disease progression subsequent to palliative chemotherapy remains elevated, varying between 15% and 58% among patients with recurrent or metastatic disease [8, 9]. Despite cisplatin-based chemotherapy being the established standard treatment, approximately 30% of cases exhibit treatment failure attributable to the development of acquired drug resistance [10]. Significantly, Xu et al. [11] were the first to demonstrate that an immunotherapy regimen excluding cisplatin could attain a three-year failure-free survival rate of 88.5% in patients with locally advanced NPC. Recent clinical studies have been assessing a range of platinum-based compounds for use in induction chemotherapy. Notably, nedaplatin has gained attention as a promising substitute for cisplatin, attributable to its improved toxicity profile and the advantage of obviating the need for hydration therapy. Several ongoing clinical trials (e.g., NCT04472403, NCT01479504, NCT01265147, NCT04437329, and NCT03503136) are currently investigating the therapeutic efficacy of nedaplatin in NPC. Furthermore, current studies are evaluating combination treatment regimens that include nedaplatin [12–14]. Currently, there is a lack of reliable predictive tools to inform subsequent treatment decisions for patients with NPC who experience recurrence or metastasis (R/M) within ten years following the completion of first-line nedaplatin-based therapy (hereafter referred to as post-nedaplatin R/M NPC).

The current prognostic evaluation of NPC is mainly based on the TNM staging system, However, this system only reflects the anatomical characteristics of the tumor (such as the extent of invasion and lymph node metastasis) and fails to incorporate key factors such as tumor biological characteristics and radiotherapy sensitivity, resulting in significant differences in the prognosis of patients with the same stage [15, 16]. Consequently, developing accurate prognostic biomarkers is crucial for advancing personalized treatment strategies. Numerous hematological markers have been investigated. Epstein-Barr virus (EBV) DNA has been established as an independent prognostic factor predicting poorer overall survival (OS) in patients with recurrent or metastatic NPC [17]. Nevertheless, its clinical utility remains substantially limited, as approximately 40% of NPC patients present with undetectable EBV DNA levels [3]. Furthermore, a single biomarker has inherent limitations in fully capturing the dynamic complexity of the tumor microenvironment. Given the well-established connection between inflammation and tumor progression, it is crucial to integrate multidimensional inflammatory and immune markers for a comprehensive assessment. The lymphocyte-to-monocyte ratio (LMR), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) have been established as predictors of treatment outcomes in head and neck cancer patients [18–20]. Furthermore, the systemic immune-inflammation index (SII, integrating platelets, neutrophils, and lymphocytes) and the systemic inflammatory response index (SIRI, combining neutrophils, monocytes, and lymphocytes) can more comprehensively assess the inflammatory status [21–23]. These hematological parameters, obtained from standard complete blood count analyses, provide a clinically accessible and reproducible approach for the dynamic assessment of systemic inflammation and its interaction with the tumor immune microenvironment. Importantly, although existing studies have mostly focused on patients who have received first-line treatment or failed cisplatin, the inflammatory characteristics and prognostic markers of post-nedaplatin R/M NPC remain underexplored.

Overall, patients with NPC exhibit significant biological heterogeneity, leading to considerable variation in prognosis among individuals at the same stage following radical treatment. Although EBV DNA and inflammatory markers have demonstrated prognostic value, their predictive efficacy remains unclear in the specific, refractory subset of patients who develop recurrence or metastasis after first-line nedaplatin-based treatment. While nomograms, as individualized prognostic tools superior to conventional TNM staging, have been widely applied in risk assessment across various cancers [24], there is currently a lack of a comprehensive prognostic model specifically designed for this clinical scenario that integrates EBV DNA with systemic inflammatory indices, including the NLR, PLR, SII, and others. Therefore, this study aims to develop and validate a prognostic nomogram specifically for this population by integrating pre-treatment EBV DNA status and key inflammatory indicators. The objective is to provide more precise risk stratification, address the existing gap in the current assessment framework, and offer a basis for guiding subsequent treatment decisions.

Materials and methods

Participants and data collection

This retrospective study was conducted in compliance with the Declaration of Helsinki and received approval from the Independent Review Board of The People’s Hospital of Guangxi Zhuang Autonomous Region (approval number: KY-KJT-2024-288). Participants were identified from a database covering the period from January 2017 to December 2022, based on the following inclusion criteria: (1) pathologically confirmed and radiologically verified recurrent or metastatic NPC, with no metastasis at initial diagnosis; (2) completion of at least two cycles of first-line nedaplatin-based standard therapy; (3) age 18 years or older and an Eastern Cooperative Oncology Group (ECOG) Performance Status of 2 or less; (4) receipt of second-line systemic treatment; (5) availability of recent peripheral blood routine tests and EBV DNA levels prior to initiation of second-line therapy; and (6) comprehensive clinical follow-up data and key prognostic indicators. Exclusion criteria encompassed: (1) presence of known autoimmune or inflammatory diseases; (2) concurrent diagnosis of other primary malignancies; and (3) pregnancy or lactation. A total of 120 patients met these criteria and were included in the analysis. Furthermore, informed consent was obtained from all participants prior to their inclusion in the study.

Sociodemographic variables, including age, sex, body mass index, smoking status (categorized as yes/no), and family history of NPC (yes/no), were incorporated into the analysis. Additionally, disease staging was evaluated based on the 8th edition of the Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) staging system. Routine laboratory parameters, such as the albumin-to-globulin ratio, alkaline phosphatase, and hemoglobin levels, along with inflammatory biomarkers and pre-treatment EBV DNA concentrations, were also examined [25–28]. For further analysis, the entire cohort was stratified into three subgroups based on the disease failure pattern: recurrence only, distant metastasis only, and concurrent recurrence with metastasis. All blood specimens were systematically obtained during the patient’s initial consultation, before the commencement of any diagnostic or therapeutic procedures. The inflammatory biomarkers were subsequently determined using the following calculations:

Diagnosis and treatment

The study population comprised patients diagnosed with post-nedaplatin R/M NPC. Each participant underwent a comprehensive clinical evaluation, including detailed medical history, physical examination, and baseline assessments. These assessments encompassed serum biochemical analysis, fiberoptic nasopharyngeal endoscopy, magnetic resonance imaging (MRI) of the nasopharynx and cervical region, abdominal ultrasonography, thoracic imaging, and either whole-body bone scintigraphy or positron emission tomography-computed tomography (PET-CT). The pretreatment plasma EBV DNA load was measured via real-time quantitative polymerase chain reaction (qPCR), targeting the BamHI-W fragment of the EBV genome. Results were reported as the number of EBV genomic copies per milliliter of plasma. Tumor staging for all cases was performed in accordance with the 8th edition UICC/AJCC staging system.

All patients underwent initial treatment with the intent of achieving a radical cure. Those diagnosed with advanced disease (stages III-IVB) were treated with concurrent chemoradiotherapy, supplemented in some cases by neoadjuvant or adjuvant chemotherapy. Patients presenting with early-stage disease or those deemed unable to tolerate combined modality therapy received either platinum-based chemotherapy or radiotherapy, which included two-dimensional, three-dimensional conformal, or intensity-modulated radiotherapy techniques.

Upon confirmation of recurrence or metastasis, all patients underwent individualized, comprehensive treatment plans formulated by a multidisciplinary team. Therapeutic approaches were customized according to the patient’s overall health status, disease severity, and the spatiotemporal characteristics of the recurrence or metastasis, including the sequence, anatomical location, and number of lesions. The primary treatment modality was systemic chemotherapy, supplemented as appropriate with localized interventions such as surgery or radiotherapy, immune checkpoint inhibitors, or targeted therapies. For patients receiving immunotherapy, PD-1 inhibitors (e.g., toripalimab or camrelizumab) were administered concomitantly with chemotherapy for up to six cycles during the induction phase, followed by maintenance therapy until disease progression, unacceptable toxicity, completion of a two-year treatment period, or patient withdrawal. In standard chemotherapy regimens, cisplatin could be replaced with alternative platinum-based agents (carboplatin, nedaplatin, or lobaplatin) based on clinical judgment. Furthermore, local treatments targeting isolated lesions, including radiotherapy or surgery, were permitted during systemic therapy following specialist evaluation. Additionally, some patients participated in clinical trials.

Follow up and outcomes

Patients underwent follow-up evaluations at intervals of every three months during the initial year, every six months over the subsequent four years, and annually thereafter. Comprehensive follow-up data were systematically collected for all patients through clinic visits or telephone communications, with the duration of follow-up recorded in months. The latest follow-up assessment occurred in September 2025. The primary endpoint of this investigation was OS, defined as the interval from the commencement of second-line therapy to death from any cause. Furthermore, to assess the efficacy of the initial treatment, the disease-free interval (DFI) was determined as the period from the completion of first-line curative-intent therapy to the first documented occurrence of recurrence or metastasis.

Statistical analysis

In the analysis of the combined cohort of patients with recurrent and metastatic disease, continuous variables exhibiting non-normal distributions were summarized by their medians and interquartile ranges (IQR). Categorical variables were reported as frequencies along with corresponding percentages. Group comparisons for continuous variables were conducted using the Mann-Whitney U test, whereas categorical variables were evaluated using either the chi-square test or Fisher’s exact test, depending on the appropriateness of each statistical method.

Optimal cutoff points for continuous variables were established via receiver operating characteristic (ROC) curve analysis, and these variables were subsequently dichotomized based on the identified thresholds. Plasma EBV DNA load was classified as negative or positive according to the laboratory detection limit (1.0 × 10³copies/mL). Univariate Cox proportional hazards regression analyses were performed to identify potential prognostic factors. Variables demonstrating statistical significance (P < 0.05) in univariate analyses, along with clinically relevant covariates, were incorporated into multivariate analyses. A multivariate Cox regression model was then constructed to ascertain independent prognostic factors, with results expressed as hazard ratios (HR) accompanied by 95% confidence intervals (CI).

A nomogram was developed integrating all independent prognostic factors identified through multivariate Cox regression. The nomogram’s predictive performance was evaluated in terms of discrimination, calibration, and clinical utility. Discriminative ability was assessed using Harrell’s concordance index (C-index), where values range from 0.5 (no discrimination) to 1.0 (perfect discrimination). Calibration was examined by generating calibration plots via bootstrap resampling (1,000 iterations). Overall model performance was further evaluated using time-dependent ROC curves, and clinical utility was assessed through decision curve analysis (DCA). Individual patient risk scores were calculated based on the nomogram, and an optimal cutoff value determined by survival ROC analysis was employed to stratify patients into low-risk and high-risk groups. Survival differences between these groups were analyzed using the Kaplan–Meier method, with statistical significance assessed by the log-rank test.

All statistical analyses were conducted using R software (version 4.5.0), employing key packages including rms, survival, timeROC, pROC, and ggplot2. All hypothesis tests were two-sided, with a significance threshold set at P < 0.05.

Results

Patient characteristics

This study encompassed a cohort of 120 patients diagnosed with post-nedaplatin R/M NPC, as detailed in Table 1. The median age of the participants was 48.35 years, with a predominance of males accounting for 89 individuals (74.2%). Staging based on the 8th edition UICC/AJCC staging system classification revealed that 13 patients (10.8%) were classified as stage I, 6 (5.0%) as stage II, 45 (37.5%) as stage III, and 56 (46.7%) as stage IV. Patterns of disease recurrence and metastasis were distributed as follows: locoregional recurrence in 51 patients (42.5%), distant metastasis in 62 patients (51.7%), and concurrent locoregional and distant involvement in 7 patients (5.8%). Baseline clinical characteristics indicated a median body mass index (BMI) of 22.25 (IQR: 14.76–31.25) and a median DFI of 21.53 months (IQR: 1–118). Laboratory parameters at baseline demonstrated median values of albumin-to-globulin ratio (AGR) at 1.50 (IQR: 0.82–4.56), alkaline phosphatase (ALP) at 83.21 U/L (IQR: 35.00–255.00), and hemoglobin (HGB) at 131.76 g/L (IQR: 85.00–158.00). Inflammatory markers were characterized by PLR of 312.78 (IQR: 76.78–1003.12), NLR of 5.49 (IQR: 0.84–37.67), LMR of 1.98 (IQR: 0.16–4.85), SII of 1443.82 (IQR: 88.34–7255.87), and SIRI of 3.24 (IQR: 0.10–19.35). Additionally, EBV DNA was detected in 45 patients, representing 37.5% of the study population.

Table 1.

Patients’ characteristics

Characteristic	n (%) or mean ± sd
Gender
Famale	31 (25.8%)
Male	89 (74.2%)
Age (years)	48.35 ± 11.26(26–72)
T
0	3 (2.5%)
I	21 (17.5%)
II	24 (20.0%)
III	37 (30.8%)
IV	35 (29.2%)
N
0	54 (45.0%)
I	14 (11.7%)
II	30 (25.0%)
III	22 (18.3%)
M
0	51 (42.5%)
1	69 (57.5%)
Stage
I	13 (10.8%)
II	6 (5.0%)
III	45 (37.5%)
IV	56 (46.7%)
Smoke
No	96 (80.0%)
Yes	24 (20.0%)
Drink
No	99 (82.5%)
Yes	21 (17.5%)
Family history
No	111 (92.5%)
Yes	9 (7.5%)
Recurrence and metastasis
Locoregional recurrence	51 (42.5%)
Distant metastasis	62 (51.7%)
Recurrence with distant metastasis	7 (5.8%)
BMI (kg/m²)	22.25 ± 3.00 (14.76–31.25)
DFI (months)	21.53 ± 19.36 (1–118)
AGR	1.50 ± 0.40 (0.82–4.56)
Albumin (g/L)	39.54 ± 3.65 (29.00–47.60)
Alkaline phosphatase (U/L)	83.21 ± 32.82 (35.00–255.00)
Hemoglobin (g/L)	131.76 ± 15.44 (85.00–158.00)
Lymphocyte (×10⁹/L)	1.05 ± 0.55 (0.28–4.22)
Monocyte (×10⁹/L)	0.57 ± 0.24 (0.20–1.78)
Neutrophil (×10⁹/L)	4.63 ± 2.66 (0.52–22.98)
Platelet (×10⁹/L)	275.09 ± 79.16 (130.00-525.00)
PLR	312.78 ± 160.14 (76.78-1003.12)
NLR	5.49 ± 4.92 (0.84–37.67)
LMR	1.98 ± 0.85 (0.16–4.85)
SII	1443.82 ± 1358.88 (88.34-7255.87)
SIRI	3.24 ± 3.53 (0.10-19.35)
EBV DNA (copies/mL)
Negative	75 (62.5%)
Positive	45 (37.5%)

BMI: body mass index; DFI: disease-free interval; AGR: albumin-to-globulin ratio; PLR: platelet-to-lymphocyte ratio; NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; SII: systemic immune–inflammation index; SIRI: systemic inflammation response index; EBV: DNA Epstein–Barr virus DNA

During a median follow-up duration of 36 months, 65 deaths were documented. The OS rates were 67.4% at 2 years and 58.4% at 3 years, respectively.

Univariate and multivariate Cox regression analyses of OS

Continuous variables, including age, DFI, AGR, ALP, HGB, PLR, NLR, LMR, SII, and SIRI, were dichotomized based on their optimal cutoff values. EBV DNA status was categorized dichotomously as either negative or positive.

Univariate analysis identified several significant risk factors for OS, including age ≥ 45.5 years (P = 0.007), DFI ≥ 17.5 months (P = 0.017), AGR ≥ 1.3 (P < 0.001), ALP ≥ 62.5 U/L (P < 0.001), HGB ≥ 133.5 g/L (P = 0.014), positive EBV DNA status (P < 0.001), PLR ≥ 286 (P = 0.003), NLR ≥ 3.2 (P = 0.020), and LMR ≥ 2.2 (P = 0.003). Subsequently, multivariable analysis was performed using a forward stepwise Cox proportional hazards regression model, incorporating univariate predictors with P-values less than 0.05. The multivariate analysis demonstrated that AGR ≥ 1.3 (P < 0.001), ALP ≥ 62.5 U/L (P = 0.045), positive EBV DNA status (P = 0.012), and LMR ≥ 2.2 (P = 0.027) were independent prognostic factors for OS (Table 2).

Table 2.

Univariate and multivariate Cox regression analyses of overall survival

Cox-regression	Univariate		Multivariate
	HR (95% CI)	P value	HR (95% CI)	P value
Gender
Famale	Reference
Male	0.87 (0.51–1.51)	0.628
Age (years)
< 45.5	Reference
≥ 45.5	2.07 (1.22–3.52)	0.007
Smoke
No	Reference
Yes	0.89 (0.50–1.59)	0.700
Drink
No	Reference
Yes	0.88 (0.48–1.63)	0.693
Family history
No	Reference
Yes	1.13 (0.46–2.83)	0.787
Stage
III	Reference
IV	0.96 (0.57–1.61)	0.880
Recurrence and metastasis
Locoregional recurrence	Reference
Distant metastasis	1.45 (0.86–2.42)	0.164
Recurrence with distant metastasis	2.39 (0.91–6.30)	0.078
BMI (kg/m²)
Normal	Reference
Abnormal	0.97 (0.59–1.62)	0.922
DFI (months)
< 17.5	Reference
≥ 17.5	0.53 (0.32–0.89)	0.017
AGR
< 1.3	Reference		Reference
≥ 1.3	0.24 (0.14–0.41)	< 0.001	0.39 (0.23–0.69)	< 0.001
ALP (U/L)
< 62.5	Reference		Reference
≥ 62.5	3.84 (1.75–8.43)	< 0.001	2.32 (1.02–5.30)	0.045
HGB (g/L)
< 133.5	Reference		Reference
≥ 133.5	0.52 (0.31–0.88)	0.014	0.62 (0.36–1.06)	0.079
EBV DNA (copies/mL)
Negative	Reference		Reference
Positive	2.68 (1.64–4.39)	< 0.001	1.92 (1.15–3.20)	0.012
PLR
< 286	Reference
≥ 286	2.13 (1.30–3.49)	0.003
NLR
< 3.2	Reference
≥ 3.2	1.96 (1.11–3.45)	0.020
LMR
< 2.2	Reference		Reference
≥ 2.2	0.41 (0.23–0.74)	0.003	0.52 (0.29–0.93)	0.027
SII
< 1094	Reference
≥ 1094	1.46 (0.89–2.38)	0.131
SIRI
< 2.9	Reference
≥ 2.9	1.65 (0.99–2.73)	0.054

CI: confidence intervals; HR hazard ratio; BMI: body mass index; DFI: disease-free interval; AGR: albumin-to-globulin ratio; ALP: alkaline phosphatase; HGB: hemoglobin; EBV: DNA Epstein–Barr virus DNA; PLR: platelet-to-lymphocyte ratio; NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; SII: systemic immune–inflammation index; SIRI: systemic inflammation response index

Nomogram and model performance

Fig. 1 presents a nomogram developed to predict OS in post-nedaplatin R/M NPC. This predictive model was established utilizing the prognostic variables of AGR, ALP, LMR, and EBV DNA status. Each factor was assigned a weighted score reflective of its relative impact on patient outcomes. The aggregate score, calculated by summing the individual points corresponding to each variable, correlates inversely with prognosis, whereby a higher total score signifies a less favorable survival outlook.

Fig. 1.

Nomogram for Predicting OS in post-nedaplatin R/M NPC. To utilize the nomogram, first identify each patient’s measurement on the respective variable axis. From this point, extend a vertical line upward to the ‘Points’ axis to determine the score associated with each variable. Subsequently, aggregate all individual scores and locate the cumulative value on the ‘Total Points’ axis. Finally, draw a vertical line downward from this total to the survival probability scales at the bottom of the nomogram to estimate the predicted overall survival probabilities at 12, 18, and 24 months. OS: overall survival; AGR: albumin-to-globulin ratio; ALP: alkaline phosphatase; LMR: lymphocyte-to-monocyte ratio; EBV: epstein–barr virus

The discriminative performance of the nomogram for OS was evaluated using the (C-index, which yielded a value of 0.747 (95% CI: 0.689–0.805). This value was significantly superior to the C-index of 0.566 (95% CI: 0.509–0.624) observed for the 8th edition UICC/AJCC staging system. Time-dependent ROC analysis demonstrated that the nomogram consistently outperformed the 8th edition UICC/AJCC staging system in terms of the area under the curve (AUC) across all evaluated time points, indicating more robust and stable predictive accuracy (Fig. 2A). Calibration curves generated through 1,000 bootstrap resampling iterations revealed a high concordance between the nomogram-predicted survival probabilities at 12, 18, and 24 months and the actual observed outcomes, suggesting excellent model calibration without evidence of significant overfitting (Fig. 2B). Furthermore, DCA indicated that the nomogram provided a broader range of threshold probabilities and greater clinical net benefit relative to the 8th edition UICC/AJCC staging system in predicting OS (Fig. 2C).

Fig. 2.

Validation of the prognostic nomogram for post-nedaplatin R/M NPC. Time-dependent ROC curves comparing the nomogram and the 8th edition UICC/AJCC staging system for OS prediction, with AUC values illustrated over time (A). Calibration curves for the nomogram predicting 12-, 18-, and 24-month overall survival. The 45-degree diagonal line represents ideal prediction (B). Decision curve analysis comparing the nomogram and the 8th edition UICC/AJCC staging system for OS prediction. The vertical axis represents net benefit, and the horizontal axis indicates threshold probability (C). OS: overall survival; AUC: area under the curve

Prognostic nomogram for risk stratification

To transform the nomogram into clinically applicable risk categories, patients were divided into low-risk and high-risk groups based on the optimal cutoff value derived from their scores. Kaplan-Meier analysis demonstrated significantly shorter OS in the high-risk group compared to the low-risk group (P < 0.001). The median survival was 26 months (95% CI: 20–39) in the high-risk group, while it was not reached (95% CI: 70-NA) in the low-risk group, indicating robust risk stratification (Fig. 3).

Fig. 3.

Overall survival in subgroups based on binary classification of the total nomogram score

Discussion

Tumor heterogeneity is an important challenge for personalized treatment, and accurate prognosis assessment is the prerequisite for implementing stratified treatment. This study shows that pretreatment AGR < 1.3, LMR < 2.2, elevated ALP ≥ 62.5 U/L, and positive plasma EBV DNA were independently associated with poorer OS.

The prognostic significance of these biomarkers is based on their fundamental biological mechanisms. The AGR encompasses both nutritional status and systemic inflammation. Pro-inflammatory cytokines, such as interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), inhibit albumin production while increasing globulin levels, resulting in a decreased AGR that indicates the presence of cachexia as well as compromised immune function [29–33]. The LMR indicates a dysregulation within the tumor immune microenvironment; specifically, lymphopenia reflects diminished anti-tumor immune responses, whereas monocytosis is associated with tumor-related inflammation and increased tumor burden [34]. This immune dysregulation is intricately associated with the systemic inflammatory response. Research has demonstrated that a decreased AGR correlates with elevated white blood cell and neutrophil counts. Moreover, neutrophils may contribute to tumor progression by releasing pro-angiogenic factors, including vascular endothelial growth factor, in response to inflammatory stimuli [35, 36]. Increased levels of ALP frequently indicate the presence of bone metastases, which represent the most prevalent site of distant dissemination in NPC, and reflect an imbalance in bone remodeling homeostasis [37–39]. Quantitative measurement of plasma EBV DNA serves as a highly specific biomarker for tumor burden. Elevated levels are strongly correlated with treatment failure and function as an independent prognostic indicator, especially in cases of metastatic disease [17, 40].

The nomogram developed in this study demonstrates clear clinical applicability, specifically tailored for a patient population characterized by poor prognosis and a previous absence of effective predictive instruments—namely, individuals with post-nedaplatin R/M NPC. This model effectively addresses a critical gap in the current clinical landscape. By incorporating routinely accessible laboratory biomarkers, particularly pre-treatment AGR, ALP, LMR, and EBV DNA status, the nomogram offers substantial clinical feasibility and potential for widespread adoption. In terms of predictive accuracy, internal validation revealed a C-index of 0.747, markedly outperforming the 8th edition UICC/AJCC staging system, which yielded a C-index of 0.566. Additionally, the model enhances risk stratification capabilities, facilitating more precise identification of high-risk patients who may benefit from intensified therapeutic interventions or more rigorous follow-up regimens. Moreover, the predictive performance of this nomogram is comparable to that of recently published prognostic models for NPC. For example, it aligns closely with the model proposed by Zeng et al. [41] for patients with synchronous metastatic disease (C-index: 0.743), the progression-free survival prediction model by He et al. [42] for patients undergoing first-line immunochemotherapy (C-index: 0.670), and the model developed by Xiao et al. [43] for patients with lung metastases (external validation C-index: 0.762). This comparability further corroborates the robustness and precision of our model in risk assessment within this distinct refractory patient cohort.

This study presents several methodological strengths. Firstly, it included patients with post-nedaplatin R/M NPC, with the DFI serving as a principal variable. A shorter DFI is recognized as indicative of more aggressive tumor biology and a tendency toward rapid emergence of chemotherapy-resistant clones, factors that are significantly correlated with poor prognosis [44]. Given that the DFI was statistically comparable among the study cohorts, this design effectively mitigated confounding bias potentially arising from variations in prior treatment protocols. Secondly, the developed nomogram incorporates multiple prognostic factors—namely ALP, AGR, LMR, and EBV DNA status—and exhibits enhanced predictive performance relative to traditional staging systems.

Several limitations inherent to this dataset warrant consideration. The modest sample size potentially restricts the statistical power of the analyses and elevates the likelihood of overfitting in any predictive models constructed. Additionally, the retrospective nature and single-center origin of the data may introduce selection bias and limit the external validity of the results. While internal validation methods, such as the C-index and time-dependent ROC curves, can be employed, the lack of an independent external validation cohort precludes definitive assessment of the models’ robustness.

Conclusion

This study demonstrates the substantial prognostic significance of pre-treatment AGR, ALP, LMR, and EBV DNA status in patients with NPC who experienced recurrence or metastasis within ten years after completing first-line treatment with nedaplatin. The nomogram incorporating these variables exhibits superior predictive performance compared to the 8th edition UICC/AJCC staging system, thereby offering an enhanced framework for risk stratification and serving as a foundation for individualized, risk-adapted therapeutic approaches.

Author contributions

Meiyan Tang: Conceptualization; data curation; writing-original draft. Yuanjun Zhan: Data curation; formal analysis; methodology. Suosu Wei: Data curation; software; validation. Zhuan Jin: Data curation; methodology. Jiao Lan: Resources; investigation. Ying Qin: Resources; investigation. Jiazhang Wei: Resources; investigation. Jingjin Weng: Conceptualization; supervision. Fei Liu: Project administration; Writing-review and editing; funding acquisition.

Funding

The authors declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by Guangxi Natural Science Foundation (2024GXNSFAA010066, 2025GXNSFDA069010), National Natural Science Foundation of China (81960493).

Data availability

The datasets produced and/or examined in the present study are not accessible to the public owing to privacy constraints specified in the ethics approval. However, they can be obtained from the corresponding author upon reasonable request, contingent upon obtaining the necessary approval from the relevant ethics committee.

Declarations

Ethics approval and consent to participate

This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Independent Review Board of The People’s Hospital of Guangxi Zhuang Autonomous Region (number: KY-KJT-2024-288). Informed consent was obtained from all individual participants included in the study.

Consent for publication

All authors have read and approved the final version of the manuscript and agree to its submission to Discover Oncology for publication.

Competing interests

The authors declare no competing interests.