Development and validation of nomograms integrating electrolyte, nutritional, and inflammatory indicators to predict survival in advanced esophageal cancer undergoing immunotherapy or radioimmunotherapy

Abstract

Background

The advent of immunotherapy has improved the survival of patients with advanced esophageal cancer (EC), yet there remain problems of poor overall prognosis and substantial individual heterogeneity. This study aimed to explore the correlation between pretreatment peripheral blood indicators and prognosis in patients with advanced EC undergoing immunotherapy or radioimmunotherapy, and to develop and validate prognostic nomograms for clinical risk stratification.

Methods

This retrospective study enrolled 221 patients with advanced EC who underwent immunotherapy or radioimmunotherapy. Electrolyte, nutritional and inflammatory peripheral blood indicators, as well as clinicopathological indicators, were selected as potential prognostic factors for collection. The Kaplan-Meier analysis was used to generate survival curves, the log-rank test was conducted to compare intergroup survival differences, and subgroup analysis was performed stratified by whether patients received combined radiotherapy. Univariate and multivariate Cox regression analyses were conducted to screen independent prognostic factors for overall survival (OS) and progression-free survival (PFS). Prognostic nomograms for OS and PFS were constructed based on independent prognostic factors, and internal validation was performed using the Bootstrap method for bias correction. The predictive accuracy, discriminative ability, and clinical utility of the nomograms were evaluated via the area under the curve (AUC), concordance index (C-index), calibration curves, and decision curve analysis (DCA).

Results

Multivariate Cox regression analysis showed that serum sodium [hazard ratio (HR) =0.920, 95% CI: 0.864–0.980, P=0.01], serum calcium (HR =1.026, 95% CI: 1.010–1.042, P=0.001), T stage (HR =2.883, 95% CI: 1.172–7.090, P=0.02) and systemic immune-inflammation index (SII) (HR =1.001, 95% CI: 1.001–1.001, P=0.01) were independent prognostic factors for OS. Serum sodium (HR =0.946, 95% CI: 0.901–0.993, P=0.03), prognostic nutritional index (PNI) (HR =0.953, 95% CI: 0.922–0.985, P=0.005), age (HR =0.977, 95% CI: 0.957–0.997, P=0.02), and treatment modality (whether combined with radiotherapy, HR =1.449, 95% CI: 1.048–2.004, P=0.03) were independent prognostic factors for PFS. The C-index of the OS nomogram constructed based on the above indicators was 0.667 (95% CI: 0.614–0.721), and the AUC for predicting 1-, 2-, and 3-year OS rates were 0.726, 0.731 and 0.712, respectively. The C-index of the PFS nomogram was 0.610 (95% CI: 0.560–0.660), and the AUC for predicting 3-, 6-, and 9-month PFS rates were 0.618, 0.627 and 0.643, respectively. The calibration curves and DCA results confirmed that both nomograms had ideal predictive accuracy and clinical application value.

Conclusions

Pretreatment serum sodium, serum calcium, SII and tumor (T) stage are independent prognostic factors for OS in advanced EC patients undergoing immunotherapy or radioimmunotherapy, and serum sodium, PNI, age and treatment modality (whether combined with radiotherapy) are independent prognostic factors for PFS. The OS and PFS nomogram model constructed based on the aforementioned multi-dimensional and convenient prognostic indicators has good discrimination, accuracy and clinical application value.

Highlight box.

Key findings

• This study demonstrated that tumor stage, serum sodium, serum calcium, and the systemic immune-inflammation index are independent prognostic factors for overall survival in patients with advanced esophageal cancer undergoing immunotherapy or radioimmunotherapy. For progression-free survival, the independent predictors were serum sodium, the prognostic nutritional index, age, and treatment modality. Based on these indicators, two nomograms were developed and validated for survival prediction.

What is known and what is new?

• Previous studies have established the prognostic relevance of blood electrolytes, nutritional status, and inflammatory indicators in cancer.

• Unlike previous studies that only observed a single indicator or focused on integrating inflammation and nutrition for prognosis, we innovatively incorporated electrolyte levels into a multi-dimensional prognostic prediction model that combines nutrition, inflammation, and clinical features, and constructed a model specifically for advanced esophageal cancer (EC) patients receiving immunotherapy or radioimmunotherapy, filling a gap in this field.

What is the implication, and what should change now?

• This provides a risk stratification tool for clinicians, but its clinical utility still needs to be verified by prospective, multi-center trials. In the future, the conclusions of this study can be further validated through external validation with large samples from multiple centers and prospective cohort studies. Additionally, the dynamic changes of related indicators throughout the course of immunotherapy and their association with patient prognosis can be further investigated, and the potential molecular mechanisms by which these indicators affect the prognosis of EC immunotherapy can be explored in depth by combining in vitro and in vivo basic experiments.

Introduction

Esophageal cancer (EC), a highly aggressive malignancy, is associated with poor clinical outcomes, and its incidence has risen steadily over the recent decades (1,2). The advancement and clinical implementation of immune checkpoint inhibitors (ICIs) have expanded the therapeutic options for EC and significantly improved the prognostic outcomes in patients with advanced disease (3-5). Radioimmunotherapy synergistically enhances antitumor activity through the immunomodulatory effects of radiotherapy and the systemic effects of immunotherapy (6,7). However, the prognostic response of advanced EC patients to immunotherapy or radioimmunotherapy remains highly heterogeneous, with only a small subset of patients achieving sustained clinical benefits and long-term survival (8). The treatment of EC is increasingly moving towards individualization, with a growing emphasis on precise stratification of patients (9). Valid and accessible prognostic prediction tools are therefore urgently needed to identify patients who are likely to benefit from these therapies, optimize individual treatment strategies, and avoid unnecessary treatment-related adverse events and the waste of medical resources.

In recent years, numerous prognostic models for EC have been developed and validated, aiming to stratify patient risk and predict treatment outcomes (10,11). Most existing models are constructed based on clinical pathological features [such as tumor-node-metastasis (TNM) staging, tumor location, histological type] and conventional clinical indicators, while some are derived from large databases like the Surveillance, Epidemiology, and End Results (SEER) database (12,13). Prognostic models for EC based on SEER have the advantages of large sample size and broad population representation (14). However, these models are mainly established for patients receiving traditional treatments (such as surgery, chemotherapy, or chemoradiotherapy), and their applicability to advanced EC patients receiving immunotherapy or radioimmunotherapy is limited. Furthermore, recent studies have reported several predictive models for immunotherapy of EC. Most of these models focus on programmed death ligand 1 (PD-L1) expression or inflammatory and nutritional indicators, and their coverage is still limited (11,15,16). PD-L1 expression is a recognized predictive biomarker for ICI therapy, but its detection is limited by tissue sampling, inter-laboratory differences, and high costs, making it difficult to be widely used as a routine prognostic indicator in clinical practice (17). For advanced EC patients receiving immunotherapy or radioimmunotherapy, there are few multi-dimensional and convenient prognostic models that integrate electrolytes, nutrition, inflammatory indicators, and clinical characteristics.

Peripheral blood indicators have attracted increasing attention in cancer prognostic prediction due to their advantages of non-invasiveness, easy accessibility, repeatable detection, and low cost (18,19). Serum electrolytes (sodium ions, potassium ions, calcium ions and chloride ions) are critical for cellular homeostasis, immune function, and neuromuscular activity, and their imbalance significantly impacts cancer treatment efficacy and survival outcomes (20). Recent studies have shown that serum sodium may be a viable prognostic and predictive biomarker (21). For instance, it has been found that hyponatremia correlates with adverse prognoses in patients with solid tumors (22-26). One study found that a high level of serum sodium predicts immunotherapy response in patients with metastatic renal cell and urothelial carcinoma, which may be attributed to sodium’s ability to stimulate antitumor immune responses, while another study demonstrated that preoperative low serum calcium levels were predictive of a poor prognosis in patients with EC (27-29). Moreover, Bener et al. reported that calcium may serve as a promising biomarker for the detection of prostate cancer (30). Increasing laboratory evidence indicates that calcium plays a crucial role in tumor biology and participates in regulating the characteristics of tumor cells, such as uncontrolled growth, invasion, and resistance to apoptosis (31). Calcium is at the core of many biological processes, including the formation and regulation of immune responses. Ronald Rooke’s review states that from this perspective, calcium is clearly overlooked in the general cancer treatment, especially in the context of immune intervention for cancer (32). Meanwhile, inflammatory markers including elevated neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), reflect tumor invasiveness and an immunosuppressive tumor microenvironment, which can weaken ICI efficacy by promoting myeloid-derived suppressor cells (MDSC) and regulatory T cells (33,34). In the study by Shen et al., increased adiposity indices and low systemic immune-inflammation index (SII) values were associated with improved survival outcomes in patients with advanced gastric cancer receiving immunotherapy. Notably, this counterintuitive survival advantage was contingent upon SII status. In contrast, this survival benefit was not evident in patient cohorts treated with chemotherapy alone (35). Nutritional status, assessed via the body mass index (BMI) or the prognostic nutritional index (PNI), is also associated with patient outcomes, as malnutrition impairs lymphocyte proliferation and interferon gamma production and synergistically suppresses antitumor immunity in cancers such as melanoma and head and neck carcinoma (36-39). In Vitale et al.’s study, poor nutrition was found to adversely affect individuals with cancer, particularly during advanced stages of the disease (40). Overall, electrolyte disorders, chronic inflammation, and malnutrition disrupt immune homeostasis, thereby affecting the antitumor immune response. It is worth noting that electrolyte balance, nutritional status, and immune-inflammatory state are not independent of each other but are interrelated to form a complex network, jointly influencing the host’s antitumor immune response and the efficacy of immunotherapy (41,42). Although the prognostic value of electrolyte, nutritional and inflammatory indicators in cancer has been discovered, their combined application in predicting the prognosis of advanced EC patients receiving immunotherapy or radioimmunotherapy has not been reported (43-45). In the field of EC prognosis models, there is a lack of a non-invasive, easily accessible, and multi-dimensional prognostic model that can comprehensively consider electrolyte, nutritional and inflammatory states to accurately predict the prognosis of advanced EC patients receiving immunotherapy or radioimmunotherapy.

Nomograms, as a visual and quantitative prognostic prediction tool, can simplify complex statistical models into individualized risk estimates for specific clinical events, and they have been widely used in the prognostic assessment of various cancers (46,47). Compared with traditional staging systems and single-biomarker prediction, nomograms have higher predictive accuracy and better clinical applicability by integrating multiple independent prognostic factors (48,49). Our research aims to explore the predictive value of pretreatment peripheral blood electrolyte levels, nutritional status and inflammatory indicators for the prognosis of advanced EC patients receiving immunotherapy or radioimmunotherapy. We further sought to establish and validate two nomograms for predicting the 1-, 2-, and 3-year overall survival (OS) and the 3-, 6-, and 9-month progression-free survival (PFS) probabilities in these patients. We aim to provide a simple and accurate prognostic tool for clinical practice, which assists clinicians in stratifying patient risk, formulating personalized treatment strategies, and ultimately improving patient prognostic outcomes. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0361/rc).

Methods

Patient selection

Our study enrolled 221 patients with advanced EC undergoing immunotherapy or radioimmunotherapy at The First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital) from June 2016 to March 2025. Patients were staged according to the eighth edition of the TNM classification system of the American Joint Committee on Cancer (AJCC). The inclusion criteria were as follows: (I) pathologically confirmed EC; (II) having received immunotherapy or radioimmunotherapy; (III) a stage of IVA or IVB when patients received immunotherapy or radioimmunotherapy; (IV) completion of routine blood and biochemical tests within 1 week prior to these treatments; and (V) no prior history of other malignancies. Meanwhile, the exclusion criteria were (I) fewer than two cycles immunotherapy; (II) with missing follow-up data or incomplete clinicopathological and laboratory variables. The study was reviewed and approved by the Medical Ethics Committee of The First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital; approval No. 2022 S398) and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was retrospective in nature and did not involve the disclosure of patients’ private information; thus, the requirement for informed consent was waived by our Medical Ethics Committee. All patient data were anonymized and handled with strict confidentiality throughout the study.

Assessment of peripheral blood indicators and other parameters

We recorded the blood indicators related to pretreatment peripheral blood electrolytes, nutritional status, and inflammatory indicators. This included sodium, potassium, chloride, calcium, phosphorus, and magnesium levels; white blood cell count, neutrophil count, lymphocyte count, platelet count, monocyte count, albumin (ALB) level, and BMI; and calculated indicators including NLR, SII, PLR, lymphocyte-to-monocyte ratio (LMR), and PNI. PNI was calculated as follows: PNI = serum ALB level (g/L) + 5 × peripheral lymphocyte count (×10⁹/L) (50). SII was calculated as follows: SII = platelet count (×10⁹/L) × neutrophil count (×10⁹/L)/lymphocyte count (×10⁹/L) (51). In addition, we collected the baseline characteristics of the patients and tumors, including age, sex, histopathology, clinical stage, treatment modality, TNM stage, tumor location, smoking status, Karnofsky Performance Status (KPS) score, presence or absence of liver metastasis, lung metastasis, and bone metastasis. Follow-up information was obtained through reviewing the inpatient and outpatient systems of the patients and conducting telephone follow-ups. The follow-up period of this study ended in March 2025.

Study outcomes

PFS was defined as the time from treatment initiation to disease progression or death, while OS was defined as the duration of treatment initiation to death from any cause.

Statistical analysis

Baseline clinical characteristics, peripheral blood electrolyte, nutritional and inflammatory indicators of the enrolled patients were described statistically as numbers with constituent ratios n (%). Continuous indicators were dichotomized into high- and low-level groups according to median values or clinically accepted cut-off values. Kaplan-Meier (KM) method was used to plot survival curves, and the log-rank test was applied to compare the differences in OS and PFS between the two groups; subgroup analyses were further performed by treatment modality (whether combined with radiotherapy). Univariate Cox regression analysis was conducted for all peripheral blood indicators and clinical characteristics. Variables with P<0.05 in univariate Cox regression analysis and several clinically important characteristics were included in multivariate Cox regression analysis to identify independent prognostic factors for OS and PFS.

Based on the identified independent prognostic factors, nomogram models were constructed to predict the 1-, 2- and 3-year OS rates and 3-, 6- and 9-month PFS rates in the overall cohort. The discriminative ability of the nomogram was evaluated using the concordance index (C-index), the time-dependent receiver operating characteristic (ROC) curve and area under the curve (AUC) value. Internal validation was performed via the Bootstrap resampling method, and calibration curves were plotted to assess the calibration of the models. Decision curve analysis (DCA) was used to evaluate the clinical net benefit and clinical applicability of the models (52,53). All statistical tests were two-tailed, and a P<0.05 was considered statistically significant. All statistical analyses were conducted with SPSS version 25.0 (IBM Corp., Armonk, NY, USA). Nomogram models were developed with R software version 4.3.1 (The R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics and survival

Continuous variables were divided into high- and low-level groups based on the median value. Peripheral blood sodium, potassium, chloride, calcium, phosphorus, magnesium, white blood cells, neutrophils, lymphocytes, platelets, monocytes, ALB, BMI, NLR, SII, PLR, LMR, and PNI were dichotomized according to their median values: 140.100 mmol/L, 4.110 mmol/L, 102.900 mmol/L, 2.270 mmol/L, 1.190 mmol/L, 0.910 mmol/L, 6.030×10⁹/L, 4.020×10⁹/L, 1.380×10⁹/L, 222.000×10⁹/L, 0.490×10⁹/L, 39.800 g/L, 21.600 kg/m², 2.791, 597.647, 156.164, 2.771, and 47.100, respectively. Age, KPS and other factors were grouped using the commonly used critical values in clinical practice (Table 1).

Table 1. Baseline characteristics of all enrolled patients.

Characteristic	Total (n=221)
Age, years
≤65	114 (51.58)
>65	107 (48.42)
Sex
Female	29 (13.12)
Male	192 (86.88)
Histopathology
Adenocarcinoma	177 (80.09)
Squamous cell carcinoma	44 (19.91)
Location
Cervical	9 (4.07)
Upper	31 (14.03)
Middle	66 (29.86)
Lower	115 (52.04)
Liver metastasis
No	199 (90.05)
Yes	22 (9.95)
Lung metastasis
No	205 (92.76)
Yes	16 (7.24)
Bone metastasis
No	211 (95.48)
Yes	10 (4.52)
T
1–2	16 (7.24)
3–4	205 (92.76)
N
0–1	32 (14.48)
2–3	189 (85.52)
M
0	68 (30.77)
1	153 (69.23)
Clinical stage
IVA	69 (31.22)
IVB	152 (68.78)
Treatment modality
Immunotherapy	116 (52.49)
Radioimmunotherapy	105 (47.51)
Smoking status
No	89 (40.27)
Yes	132 (59.73)
KPS
<70	10 (4.52)
70–80	107 (48.42)
>80	104 (47.06)

Data are presented as n (%). KPS, Karnofsky Performance Status; M, metastasis; N, node; T, tumor.

A total of 221 patients with advanced EC undergoing immunotherapy or radioimmunotherapy at The First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital) from June 2016 to March 2025 were enrolled. Among these patients, 116 received immunotherapy and 105 received radioimmunotherapy. Among them, 114 people (51.58%) were aged 65 or younger, and 107 people (48.42%) were over 65 years old; 29 people (13.12%) were female and 192 people (86.88%) were male; 44 people (19.91%) had adenocarcinoma and 177 people (80.09%) had squamous cell carcinoma; 9 people (4.07%) had cervical EC, 31 people (14.03%) had upper EC, 66 people (29.86%) had middle EC, and 115 people (52.04%) had lower EC. The median OS for the entire cohort was 19.467 months [95% confidence interval (CI): 15.393–23.540], and the median PFS was 8.867 months (95% CI: 7.569–10.164). The median follow-up time was 12.87 months (range: 1–105.27 months), with the final follow-up completed in March 2025. The baseline clinical and pathological characteristics of all 221 enrolled patients are summarized in Table 1. The comparison of characteristics among patients grouped by different pretreatment serum sodium, serum calcium, SII, and PNI is presented in Table 2. The data regarding patient and cancer characteristics associated with the levels of other indicators are provided in supplementary material 1 (available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-1.xlsx).

Table 2. Characteristics of patients and tumor by main peripheral blood electrolytes, nutritional status, and inflammatory indicators levels.

Characteristic	Total (n=221)	Serum sodium, mmol/L				Serum calcium, mmol/L				SII				PNI
		≤140.1 (n=111)	>140.1 (n=110)	P		≤2.27 (n=116)	>2.27 (n=105)	P		≤597.65 (n=110)	>597.65 (n=111)	P		≤47.1(n=111)	>47.1 (n=110)	P
Age, years	64.55±8.08	65.41±8.05	63.67±8.05	0.11		65.47±7.98	63.52±8.10	0.07		64.41±8.07	64.68±8.11	0.80		65.95±8.12	63.14±7.82	0.009
KPS	80.00 (80.00, 90.00)	90.00 (80.00, 90.00)	80.00 (80.00, 90.00)	0.44		85.00 (80.00, 90.00)	80.00 (80.00, 90.00)	0.69		85.00 (80.00, 90.00)	80.00 (80.00, 90.00)	0.14		80.00 (80.00, 90.00)	85.00 (80.00, 90.00)	0.16
Histopathology				0.19				0.97				0.48				0.19
Adenocarcinoma	44 (19.91)	26 (23.42)	18 (16.36)			23 (19.83)	21 (20.00)			24 (21.82)	20 (18.02)			26 (23.42)	18 (16.36)
Squamous cell carcinoma	177 (80.09)	85 (76.58)	92 (83.64)			93 (80.17)	84 (80.00)			86 (78.18)	91 (81.98)			85 (76.58)	92 (83.64)
Location				0.82				0.27				0.08				0.77
Cervical	9 (4.07)	6 (5.41)	3 (2.73)			7 (6.03)	2 (1.90)			4 (3.64)	5 (4.50)			5 (4.50)	4 (3.64)
Upper	31 (14.03)	16 (14.41)	15 (13.64)			19 (16.38)	12 (11.43)			15 (13.64)	16 (14.41)			13 (11.71)	18 (16.36)
Middle	66 (29.86)	32 (28.83)	34 (30.91)			34 (29.31)	32 (30.48)			25 (22.73)	41 (36.94)			35 (31.53)	31 (28.18)
Lower	115 (52.04)	57 (51.35)	58 (52.73)			56 (48.28)	59 (56.19)			66 (60.00)	49 (44.14)			58 (52.25)	57 (51.82)
Liver metastasis				0.19				0.045				0.19				0.03
No	199 (90.05)	97 (87.39)	102 (92.73)			100 (86.21)	99 (94.29)			102 (92.73)	97 (87.39)			95 (85.59)	104 (94.55)
Yes	22 (9.95)	14 (12.61)	8 (7.27)			16 (13.79)	6 (5.71)			8 (7.27)	14 (12.61)			16 (14.41)	6 (5.45)
Lung metastasis				0.62				0.18				0.29				0.62
No	205 (92.76)	102 (91.89)	103 (93.64)			105 (90.52)	100 (95.24)			100 (90.91)	105 (94.59)			102 (91.89)	103 (93.64)
Yes	16 (7.24)	9 (8.11)	7 (6.36)			11 (9.48)	5 (4.76)			10 (9.09)	6 (5.41)			9 (8.11)	7 (6.36)
Bone metastasis				0.74				0.87				0.34				0.76
No	211 (95.48)	107 (96.40)	104 (94.55)			110 (94.83)	101 (96.19)			107 (97.27)	104 (93.69)			105 (94.59)	106 (96.36)
Yes	10 (4.52)	4 (3.60)	6 (5.45)			6 (5.17)	4 (3.81)			3 (2.73)	7 (6.31)			6 (5.41)	4 (3.64)
T				0.99				0.41				0.99				0.04
1–2	16 (7.24)	8 (7.21)	8 (7.27)			10 (8.62)	6 (5.71)			8 (7.27)	8 (7.21)			4 (3.60)	12 (10.91)
3–4	205 (92.76)	103 (92.79)	102 (92.73)			106 (91.38)	99 (94.29)			102 (92.73)	103 (92.79)			107 (96.40)	98 (89.09)
N				0.43				0.94				0.007				0.24
0–1	32 (14.48)	14 (12.61)	18 (16.36)			17 (14.66)	15 (14.29)			23 (20.91)	9 (8.11)			13 (11.71)	19 (17.27)
2–3	189 (85.52)	97 (87.39)	92 (83.64)			99 (85.34)	90 (85.71)			87 (79.09)	102 (91.89)			98 (88.29)	91 (82.73)
M				0.81				0.70				0.23				0.23
0	68 (30.77)	35 (31.53)	33 (30.00)			37 (31.90)	31 (29.52)			38 (34.55)	30 (27.03)			30 (27.03)	38 (34.55)
1	153 (69.23)	76 (68.47)	77 (70.00)			79 (68.10)	74 (70.48)			72 (65.45)	81 (72.97)			81 (72.97)	72 (65.45)
Clinical stage				0.92				0.60				0.18				0.29
IVA	69 (31.22)	35 (31.53)	34 (30.91)			38 (32.76)	31 (29.52)			39 (35.45)	30 (27.03)			31 (27.93)	38 (34.55)
IVB	152 (68.78)	76 (68.47)	76 (69.09)			78 (67.24)	74 (70.48)			71 (64.55)	81 (72.97)			80 (72.07)	72 (65.45)
Treatment modality				0.04				0.003				0.31				0.31
Immunotherapy	116 (52.49)	66 (59.46)	50 (45.45)			72 (62.07)	44 (41.90)			54 (49.09)	62 (55.86)			62 (55.86)	54 (49.09)
Radioimmunotherapy	105 (47.51)	45 (40.54)	60 (54.55)			44 (37.93)	61 (58.10)			56 (50.91)	49 (44.14)			49 (44.14)	56 (50.91)
Sex				0.86				0.48				0.57				0.17
Female	29 (13.12)	15 (13.51)	14 (12.73)			17 (14.66)	12 (11.43)			13 (11.82)	16 (14.41)			18 (16.22)	11 (10.00)
Male	192 (86.88)	96 (86.49)	96 (87.27)			99 (85.34)	93 (88.57)			97 (88.18)	95 (85.59)			93 (83.78)	99 (90.00)
Smoking status				0.37				0.72				0.85				0.53
No	89 (40.27)	48 (43.24)	41 (37.27)			48 (41.38)	41 (39.05)			45 (40.91)	44 (39.64)			47 (42.34)	42 (38.18)
Yes	132 (59.73)	63 (56.76)	69 (62.73)			68 (58.62)	64 (60.95)			65 (59.09)	67 (60.36)			64 (57.66)	68 (61.82)

Continuous data following a normal distribution are expressed as the mean ± SD. The comparison between the two groups was conducted with the t-test for two independent samples. Continuous data with a skewed distribution are represented as median (Q₁, Q₃) and were compared with the Mann-Whitney test. Categorical data are expressed as n (%). The Chi-squared test or Fisher exact probability method was used for comparisons between groups. KPS, Karnofsky Performance Status; M, metastasis; N, node; PNI, prognostic nutritional index; Q₁, 1st quartile; Q₃, 3st quartile; SD, standard deviation; SII, systemic immune-inflammation index; T, tumor.

Serum sodium and calcium levels predicted the survival benefit of immunotherapy

Patients were stratified into high- and low-level groups according to the median serum sodium level. KM analysis indicated that high serum sodium levels were significantly associated with longer OS and PFS as compared to low serum sodium levels (OS: P=0.01; PFS: P=0.006) (Figure 1A,1B). Subsequently, we conducted a subgroup analysis based on treatment modality. The results showed that in the subgroup of patients receiving immunotherapy, higher serum sodium levels were associated with a significantly improved PFS, but there was no difference in OS (PFS: P=0.006; OS: P=0.07) (Figure 1C,1D). However, in the subgroup treated with radioimmunotherapy, no significant differences in OS or PFS were observed between patients with different serum sodium levels (supplementary material 2 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-2.pdf).

Figure 1.

Survival curves for pretreatment serum sodium levels. (A,B) Survival curves for pretreatment serum sodium levels in the overall study cohort. (C,D) Survival curves for pretreatment serum sodium levels in patients with advanced EC receiving immunotherapy. CI, confidence interval; EC, esophageal carcinoma; HR, hazard ratio.

Based on the median serum calcium level, patients were categorized into high- and low-level groups. KM analysis indicated no statistically significant differences in OS or PFS between the two groups. However, subsequent subgroup analysis revealed that, among patients receiving immunotherapy, lower serum calcium levels were associated with a significantly improved PFS, although no significant difference was observed for OS (PFS: P=0.03; OS: P=0.12). Similarly, in the subgroup receiving radioimmunotherapy, no significant differences in OS or PFS were detected across different serum calcium levels (supplementary material 2 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-2.pdf).

In addition, it was found that in the subgroup receiving radioimmunotherapy, higher serum phosphorus levels were significantly associated with improved PFS and OS (PFS: P=0.03; OS: P=0.01). No correlation was found between the levels of other electrolytes and the prognosis of patients (supplementary material 2 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-2.pdf).

Prognostic value of nutritional status and inflammatory indicators

Similarly, we divided patients into high- and low-level groups according to the median levels of nutritional status and inflammatory indicators. KM analysis indicated that the factors significantly associated with improved OS included high levels of ALB, LMR, KPS and PNI; low levels of neutrophils, NLR, SII and PLR, as well as T stage (ALB: P<0.001; LMR: P=0.01; KPS: P=0.047; PNI: P=0.03; neutrophils: P=0.03; NLR: P=0.017; SII: P=0.02; PLR: P=0.03; T stage: P=0.02). Moreover, subgroup analysis revealed that a lower T stage and the absence of liver metastasis were significantly correlated with better OS in the immunotherapy subgroup (T stage: P=0.02; liver metastasis: P=0.04). In the radioimmunotherapy subgroup, lower SII and NLR levels, as well as higher ALB levels, were significantly correlated with improved OS (SII: P=0.03; NLR: P=0.02; ALB: P<0.001) (supplementary material 2 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-2.pdf).

For PFS, patients with higher LMR and PNI levels and lower NLR levels exhibited longer PFS (LMR: P=0.007; PNI: P=0.03; NLR: P=0.005). In the subgroup analysis, high levels of ALB, LMR, and PNI, as well as low levels of neutrophil, were significantly associated with improved PFS in patients receiving radioimmunotherapy (ALB: P=0.009; LMR: P=0.02; PNI: P=0.005; neutrophils: P=0.02). In the immunotherapy group, lower NLR levels were significantly correlated with improved PFS (NLR: P=0.02) (supplementary material 2 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-2.pdf).

Serum sodium and calcium levels as independent prognostic factors

In the univariate Cox regression analysis, OS was significantly associated with indicators related to serum sodium [hazard ratio (HR) 0.918, 95% CI: 0.861–0.978; P=0.009], serum calcium (HR =1.031, 95% CI: 1.015–1.047; P<0.001), ALB (HR =0.944, 95% CI: 0.900–0.990; P=0.02), SII (HR =1.001, 95% CI: 1.001–1.001; P=0.003), PLR (HR =1.003, 95% CI: 1.001–1.005; P=0.02), PNI (HR =0.951, 95% CI: 0.916–0.988; P=0.009), T stage (HR =2.830, 95% CI: 1.151–6.956; P=0.02), and KPS (HR =0.976, 95% CI: 0.959–0.992; P=0.004) (Table 3). Several indicators were significantly correlated with PFS in the univariate Cox regression analysis, including serum sodium (HR =0.947, 95% CI: 0.901–0.994; P=0.03), serum calcium (HR =1.019, 95% CI: 1.004–1.034; P=0.01), SII (HR =1.001, 95% CI: 1.001–1.001; P=0.02), and PNI (HR =0.967, 95% CI: 0.937–0.998; P=0.04) (Table 4). No statistically significant correlations were found between other patient characteristics or tumor-related features and OS or PFS (P>0.05) (supplementary material 3 available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-3.pdf).

Table 3. Univariate and multivariate Cox regression analyses of OS in the overall cohort.

Variable	Univariate						Multivariate
	β	SE	Z	P	HR (95% CI)		β	SE	Z	P	HR (95% CI)
Histopathology
Adenocarcinoma					1
Squamous cell carcinoma	−0.139	0.237	−0.589	0.56	0.870 (0.547–1.384)
Location
Cervical					1
Upper	0.393	0.754	0.522	0.60	1.482 (0.338–6.491)
Middle	0.391	0.730	0.536	0.59	1.479 (0.354–6.188)
Lower	0.463	0.719	0.644	0.52	1.589 (0.388–6.507)
Liver metastasis
No					1
Yes	0.355	0.279	1.274	0.20	1.426 (0.826–2.463)
T
1–2					1						1
3–4	1.040	0.459	2.267	0.02	2.830 (1.151–6.956)		1.059	0.459	2.306	0.02	2.883 (1.172–7.090)
N
0–1					1
2–3	0.340	0.278	1.222	0.22	1.405 (0.814–2.422)
M
0					1
1	0.053	0.215	0.248	0.80	1.055 (0.693–1.606)
Treatment modality
Immunotherapy					1
Radioimmunotherapy	−0.076	0.195	−0.391	0.70	0.927 (0.633–1.357)
Sex
Female					1
Male	0.058	0.306	0.190	0.85	1.060 (0.581–1.933)
Age	0.002	0.012	0.172	0.86	1.002 (0.979–1.026)
Smoking status
No					1
Yes	−0.000	0.198	−0.001	>0.99	1.000 (0.679–1.473)
Serum sodium	−0.086	0.033	−2.626	0.009	0.918 (0.861–0.978)		−0.083	0.032	−2.584	0.01	0.920 (0.864–0.980)
Serum calcium	0.031	0.008	3.918	<0.001	1.031 (1.015–1.047)		0.026	0.008	3.236	0.001	1.026 (1.010–1.042)
ALB	−0.058	0.024	−2.391	0.02	0.944 (0.900–0.990)
SII	0.001	0.000	2.932	0.003	1.001 (1.001–1.001)		0.001	0.000	2.458	0.01	1.001 (1.001–1.001)
PLR	0.003	0.001	2.334	0.02	1.003 (1.001–1.005)
PNI	−0.050	0.019	−2.598	0.009	0.951 (0.916–0.988)
KPS	−0.025	0.009	−2.847	0.004	0.976 (0.959–0.992)

ALB, albumin; CI, confidence interval; HR, hazard ratio; KPS, Karnofsky Performance Status; M, metastasis; N, node; OS, overall survival; PLR, platelet-to-lymphocyte ratio; PNI, prognostic nutritional index; SE, standard error; SII, systemic immune-inflammation index; T, tumor.

Table 4. Univariate and multivariate Cox regression analyses of PFS in the overall cohort.

Variable	Univariate						Multivariate
	β	SE	Z	P	HR (95% CI)		β	SE	Z	P	HR (95% CI)
Histopathology
Adenocarcinoma					1
Squamous cell carcinoma	0.105	0.196	0.537	0.59	1.111 (0.757–1.631)
Location
Cervical					1
Upper	−0.306	0.503	−0.609	0.54	0.736 (0.275–1.973)
Middle	−0.000	0.471	−0.001	>0.99	1.000 (0.397–2.518)
Lower	0.097	0.461	0.211	0.83	1.102 (0.446–2.720)
Liver metastasis
No					1
Yes	0.357	0.240	1.485	0.14	1.428 (0.892–2.287)
T
1–2					1
3–4	0.236	0.301	0.786	0.43	1.267 (0.702–2.284)
N
0–1					1
2–3	0.357	0.226	1.582	0.11	1.429 (0.918–2.225)
M
0					1
1	0.022	0.174	0.130	0.90	1.023 (0.728–1.437)
Treatment modality
Immunotherapy					1						1
Radioimmunotherapy	0.305	0.159	1.923	0.055	1.357 (0.994–1.852)		0.371	0.165	2.244	0.03	1.449 (1.048–2.004)
Sex
Female					1
Male	0.304	0.250	1.216	0.22	1.355 (0.830–2.213)
Age	−0.017	0.010	−1.726	0.08	0.984 (0.965–1.002)		−0.024	0.010	−2.283	0.02	0.977 (0.957–0.997)
Smoking status
No					1
Yes	0.119	0.162	0.733	0.46	1.126 (0.819–1.548)
Serum sodium	−0.055	0.025	−2.184	0.03	0.947 (0.901–0.994)		−0.055	0.025	−2.229	0.03	0.946 (0.901–0.993)
Serum calcium	0.019	0.007	2.557	0.01	1.019 (1.004–1.034)
SII	0.001	0.000	2.264	0.02	1.001 (1.001–1.001)
PNI	−0.034	0.016	−2.111	0.04	0.967 (0.937–0.998)		−0.048	0.017	−2.834	0.005	0.953 (0.922–0.985)
KPS	−0.013	0.007	−1.782	0.08	0.987 (0.974–1.001)

CI, confidence interval; HR, hazard ratio; KPS, Karnofsky Performance Status; M, metastasis; N, node; PFS, progression-free survival; PNI, prognostic nutritional index; SE, standard error; SII, systemic immune-inflammation index; T, tumor.

At the end of the follow-up period, a total of 113 positive events (deaths) were recorded for the OS endpoint among the 221 patients finally enrolled in this study. In accordance with the events per variable (EPV) criterion for clinical prediction model construction—which stipulates that at least 10 endpoint events should correspond to each candidate predictive variable in multivariate Cox regression modeling to inherently control the risk of overfitting—a total of 11 variables were included in the multivariate Cox regression analysis. These variables consisted of those with a P<0.05 in the univariate Cox regression analysis, as well as important clinical characteristics associated with the prognosis of EC [node (N) stage, presence of liver metastasis, and age]. EPV =113/11 (≈10.273), which met the requirement of EPV ≥10. This ensured effective control of the model overfitting risk and guaranteed the stability of the prognostic model. Multivariate Cox regression analysis showed that the independent predictors of OS were serum sodium (HR =0.920, 95% CI: 0.864–0.980; P=0.01), serum calcium (HR =1.026, 95% CI: 1.010–1.042; P=0.001), SII (HR =1.001, 95% CI: 1.001–1.001; P=0.01), and tumor (T) stage (HR =2.883, 95% CI: 1.172–7.090; P=0.02) (Table 3).

During the follow-up period, among the 221 patients finally included in this study, there were a total of 163 positive events at the PFS endpoint (disease progression or death). Based on the EPV rule, we included 15 variables in the multivariate Cox regression analysis, which comprised the variables with a P<0.05 in the univariate Cox regression analysis and important clinical characteristics related to the prognosis of EC (treatment modality, TNM stage, KPS, presence or absence of liver metastasis, age, sex, histopathology, smoking history, and tumor location). EPV=163/15 (≈10.867), which met the requirement of EPV ≥10. This effectively controlled the risk of model overfitting and ensured the stability of the model. Multivariate Cox regression analysis showed that the independent predictors of PFS were serum sodium (HR =0.946, 95% CI: 0.901–0.993; P=0.03), PNI (HR =0.953, 95% CI: 0.922–0.985; P=0.005), age (HR =0.977, 95% CI: 0.957–0.997; P=0.02), and treatment modality (HR =1.449, 95% CI: 1.048–2.004; P=0.03) (Table 4) (the complete results of univariate and multivariate Cox regression analyses are provided in supplementary material 3, available at https://cdn.amegroups.cn/static/public/10.21037jtd-2026-1-0361-3.pdf).

Development and validation of nomogram models

The independent prognostic factors identified from multivariate Cox regression analysis, including serum sodium, serum calcium, T stage, and SII, were used to establish a prognostic nomogram model for predicting 1-, 2-, and 3-year OS probabilities (Figure 2). Similarly, we used serum sodium, PNI, treatment modality, and age to construct a prognostic nomogram model for predicting 3-, 6-, and 9-month PFS probabilities (Figure 3).

Figure 2.

Nomogram for predicting the OS probability for the overall cohort. The specific point for each variable of the patient lies on each variable axis. A vertical line is drawn upward to determine the point at which each variable accepts; the sum of these points is located on the total points axis, and a vertical line can be drawn down to the survival axis to determine the 1-, 2-, and 3-year OS probabilities. OS, overall survival; SII, systemic immune-inflammation index; T, tumor.

Figure 3.

Nomogram for predicting the PFS probability of the overall cohort. The 3-, 6-, and 9-month PFS probabilities can be determined according to the method described in Figure 2. PFS, progression-free survival; PNI, prognostic nutritional index.

The C-index of the OS prognostic nomogram was 0.667 (95% CI: 0.614–0.721), which was close to the clinically practical threshold of 0.70, indicating a moderate overall discriminative ability. The AUC values for predicting 1-, 2-, and 3-year OS rates were 0.726 (95% CI: 0.645–0.806), 0.731 (95% CI: 0.635–0.827), and 0.712 (95% CI: 0.598–0.827), respectively, all exceeding 0.70, which demonstrated satisfactory discriminative performance at these follow-up time points. Calibration curves for 1-, 2-, and 3-year OS prediction generated via the Bootstrap resampling method showed a good agreement between the nomogram-predicted probabilities and actual survival probabilities. This indicated that the nomogram model had no significant overfitting bias after internal validation and possessed good predictive accuracy. The results of DCA revealed that the OS nomogram exhibited favorable clinical utility within clinically relevant threshold probabilities (Figure 4).

Figure 4.

Validation of the nomogram model for OS probability. (A-C) ROC curves of the ability of nomogram to predict the 1-, 2-, and 3-year OS probability. (D-F) Calibration curves of nomogram to predict the 1-, 2-, and 3-year OS probability. (G-I) DCA of the nomogram’s ability to predict the 1-, 2-, and 3-year OS probability. AUC, area under the curve; CI, confidence interval; DCA, decision curve analysis; OS, overall survival; ROC, receiver operating characteristic.

For the PFS prediction nomogram, the C-index was 0.610 (95% CI: 0.560–0.660), indicating moderate overall discriminative ability. The AUC values for predicting 3-, 6-, and 9-month PFS rates were 0.618 (95% CI: 0.513–0.722), 0.627 (95% CI: 0.546–0.708), and 0.643 (95% CI: 0.565–0.721), respectively. All AUC values fell within the range of 0.6–0.7, which indicated that the nomogram had moderate discriminative performance for PFS prediction at these follow-up time points. Calibration curves for 3-, 6-, and 9-month PFS prediction corrected by the Bootstrap resampling method showed no significant overfitting bias of the model after internal validation, and thus exhibited acceptable predictive accuracy. Results of DCA further confirmed that the PFS nomogram had favorable clinical utility within clinically relevant threshold probabilities (Figure 5).

Figure 5.

Validation of the nomogram model in predicting PFS probability. (A-C) ROC curves of the nomogram to predict the 3-, 6-, and 9-month PFS probability. (D-F) Calibration curves of the nomogram to predict the 3-, 6-, and 9-month PFS probability. (G-I) DCA of nomogram to predict the 3-, 6-, and 9-month PFS probability. AUC, area under the curve; CI, confidence interval; DCA, decision curve analysis; PFS, progression-free survival; ROC, receiver operating characteristic.

Discussion

In this study, KM analysis revealed that several pretreatment peripheral blood electrolytes, nutritional status and inflammatory indicators such as serum sodium, PNI, and LMR were significantly correlated with the PFS and OS in patients with advanced EC undergoing immunotherapy or radioimmunotherapy (P<0.05). Furthermore, multivariable Cox regression analysis demonstrated that serum sodium, serum calcium, SII and T stage were independent predictors for OS (P<0.05). Multivariable Cox regression analysis also indicated that serum sodium, PNI, age, and treatment modality were independent predictors for PFS (P<0.05). According to the independent predictors derived from multivariate Cox regression analysis, we developed and verified two nomograms to predict OS and PFS probabilities in this patient cohort.

Our multivariate analysis confirmed that a low serum sodium level (<140.100 mmol/L) was an independent adverse prognostic factor for both PFS and OS in patients with advanced EC undergoing immunotherapy or radioimmunotherapy (P<0.05), a result corroborated by the KM analysis (P<0.05). This finding is consistent with the retrospective studies conducted by Wang et al. and Lu et al., which reported that a decrease in serum sodium levels before treatment was associated with an unfavorable prognosis in patients with EC (54,55). A low serum sodium level (hyponatremia) has been consistently associated with adverse prognosis across a range of conditions, including cancers (22). A recent review synthesized the current evidence supporting NaCl’s immunomodulatory effects and found that sodium chloride exerts immunomodulatory effects, underscoring its potential as both a biomarker and an adjunctive therapeutic tool in cancer immunotherapy (56). Although the data for EC are limited, there are similar research results in the field of renal cell carcinoma. More recently, serum sodium has been identified as a prognostic factor in patients with renal cell carcinoma treated with tyrosine kinase inhibitors and nivolumab (57). Sahin et al. conducted a retrospective analysis across a diversity of tumor types and found that patients with higher baseline sodium levels experienced significantly prolonged OS and PFS, as well as superior antitumor responses, during immunotherapy treatment (58). These findings highlight the potential of serum sodium levels to serve as an indicator of immunotherapy response. These studies suggest that higher serum sodium levels may enhance the efficacy of tumor immunotherapy and improve the prognosis of patients. Scirgolea et al. reported that sodium chloride promotes the effector differentiation of CD8⁺ T cells and the production of interferon gamma, known to be crucial for antitumor immune response (59). A high salt diet (HSD) has been shown to influence the activity and differentiation of T helper 17 cells and can also cause macrophages to enter a specific activated state (60,61). Research further indicates that an HSD can inhibit the growth of tumors in mice, with the related mechanism primarily involving the enhancement of the antitumor immune response and the regulation of MDSC differentiation (62,63). The above research results suggest that an increase in serum sodium levels may enhance the antitumor immune response, thereby improving the efficacy of cancer immunotherapy and consequently enhancing the prognosis of patients.

Although the KM analysis in our study only attested to a significant correlation between serum calcium and the PFS of patients receiving immunotherapy (P=0.03), the multivariate Cox regression analysis showed that it had a significant predictive value for the OS of all the study patients (P<0.05). Lower serum calcium levels were significantly associated with longer OS. In Wang et al.’s study, P2RX4 promoted the progression of liver cell carcinoma by enhancing calcium influx (64). Other research has found that calcium serves as a key secondary messenger that regulates the signaling pathways associated with tumor metastasis. The transient accumulation of calcium was found to promote tumor metastasis, while calcium-dependent proteins and calcium-related channels also play significant roles in this malignant process (65). However, Shiratori et al.’s research on EC revealed that low preoperative serum calcium concentrations are associated with poorer prognostic outcomes in these patients (29). This discrepancy might be attributed to the different treatment modality and pathological stages of the patients in these studies.

Regarding indicators for nutritional status, our KM analysis demonstrated that high PNI was associated with better PFS and OS in our patient cohort (P<0.05). In the multivariate Cox regression analysis, we found that PNI was an independent prognostic factor for PFS (P<0.05). This finding is consistent with the results reported by Chen et al. (66). Interestingly, the PNI has been demonstrated to correlate with immunotherapy treatment efficacy across a variety of other tumor types (67,68). Lei et al. found that PNI can serve as a prospective indicator for the success of immunotherapy in patients with non-small cell lung cancer, and constructed an effective nomogram model to predict the prognosis of these patients (69). A lower PNI might suggest that the body is in a state of immunosuppression or immunodeficiency, which is associated with an unfavorable prognosis. Our subgroup analysis further indicated that a higher serum ALB level was associated with the prolonged PFS and OS of patients with advanced EC receiving radioimmunotherapy (P<0.05). This may be attributed to certain physiological alterations induced by radiotherapy. The findings reported by Chen et al. support serum ALB as an independent prognostic factor for OS in patients with advanced EC receiving immunotherapy (66). Hypoalbuminemia may impair the activation and functional activity of antitumor immune cells by reducing amino acid levels within the immune microenvironment (70). In terms of mechanism, the patients’ nutrition has the potential to modulate the tumor microenvironment, which in turn may influence their response to immunotherapy (71). However, in the multivariable Cox regression analysis, we were unable to confirm serum ALB as an independent predictor of PFS or OS. This might be due to the fact that the effect of serum ALB is also influenced by other factors.

Furthermore, we observed statistically significant associations between inflammatory indicators and survival outcomes. Our multivariable Cox regression analysis identified SII as an independent prognostic factor for OS in patients with advanced EC receiving immunotherapy or radioimmunotherapy (P<0.05), which was confirmed in the KM analysis (P<0.05). In line with the study by Dong et al., we found that SII was associated with the survival benefit in patients with advanced EC who received immunotherapy (P<0.05) (72). Recent studies have found that SII is an independent predictor of pathological complete response in patients with locally advanced EC who have received neoadjuvant chemotherapy and immunotherapy (73). In terms of mechanism, neutrophils, lymphocytes, and platelets reflect immune-inflammatory status, which is linked to tumorigenesis, progression, and metastasis (74,75). Furthermore, we observed that a higher LMR was significantly associated with a longer OS and PFS in the overall cohort (P<0.05). Similarly, previous studies have confirmed that lymphocytes play a critical role in promoting antitumor immunity, with a higher LMR generally being associated with better survival and response to immunotherapy (76). In our analysis, a lower PLR was significantly associated with a longer OS in the overall cohort (P<0.05). Furthermore, a lower NLR was correlated with improved OS in patients with advanced EC who had received radioimmunotherapy (P<0.05); meanwhile, lower neutrophil levels were associated with improved PFS among patients with advanced EC who had received radioimmunotherapy (P<0.05). This is line with the findings reported by Deng et al., who found that NLR and PLR levels were correlated with poorer PFS and OS in immunotherapy-treated patients with EC (77). An elevated PLR reflects either an increased platelet count or decreased lymphocyte count, which may potentially indicate tumor recurrence and metastatic progression. Platelets generate a procoagulant microenvironment that facilitates the amplification of cancer-associated coagulation cascades. Additionally, they can be recruited to aggregate around tumor cells, facilitating immune evasion and thereby promoting tumor progression and metastasis (78). Extensive clinical and preclinical evidence has established the critical role of lymphocytes in antitumor immune response, and lymphopenia has been associated with poor prognostic outcomes in patients with recurrent and metastatic EC receiving immunotherapy (79). Numerous studies have also demonstrated that neutrophils actively contribute to the promotion of angiogenesis and the induction of immunosuppressive effects (80,81).

It is worth noting that our analysis revealed younger age as an adverse prognostic factor for PFS in patients with advanced EC receiving immunotherapy (P<0.05), which is contrary to existing reports (82,83). The chronic inflammation associated with aging activates tumor-associated macrophages, which further suppress antitumor immunity and diminish the therapeutic efficacy of ICIs (84). This discrepancy may be attributed to the limited sample size and potential selection bias inherent in this study.

We additionally found that combined radiotherapy was an independent risk factor for patients with advanced EC receiving immunotherapy (P<0.05). Despite extensive literature on the synergistic effects and underlying mechanisms of combining radiotherapy with immunotherapy, the field of radioimmunotherapy remains in its early stages (85,86). One study found that among patients with a history of immune-related adverse events or prior ICI-induced pneumonitis, 61% and 83%, respectively, developed grade 2 or higher radiation pneumonitis after radiotherapy (87). Several large-sample retrospective studies have confirmed the positive effect of aggressive primary tumor radiotherapy for patients with advanced EC, although optimal patient selection criteria remain to be further refined (88,89). Treatment outcomes in patients with advanced EC are highly variable due to significant disease heterogeneity, the influence of multiple factors including radiotherapy parameters (dose, fractionation, site, and technique), treatment timing and sequencing, and the choice of immunotherapy agents.

Unlike previous studies that only observed a single indicator or focused on integrating inflammation and nutrition for prognosis, we innovatively incorporated electrolyte levels into a multi-dimensional prognostic prediction model that combines nutrition, inflammation, and clinical features, and constructed a model specifically for advanced EC patients receiving immunotherapy or radioimmunotherapy, filling a gap in this field (11,15,16). The novelty of our research also lies in conducting stratified analyses based on whether radiotherapy was combined, clarifying the differences in the role of prognostic indicators under different treatment strategies, which is distinct from the homogenized analysis of the entire population in most studies. The two multi-dimensional nomogram models that we developed have been verified by various methods to confirm its good predictive performance, and the included indicators are all routine clinical detection items, providing a practical assessment tool for the prognosis of advanced EC patients undergoing immunotherapy or radioimmunotherapy.

There are some limitations in this study: first, it is a single-center retrospective study with a relatively limited sample size. Future multi-center, prospective studies are needed for external validation to further test the broader applicability of the model; second, this study only included baseline indicators before treatment. Continuous monitoring during the treatment process might more accurately reflect the patient’s response to immunotherapy; third, this study did not deeply explore the specific molecular mechanisms, which require further clarification through subsequent basic experiments; fourth, the differences in the sensitivity of hematological indicator detection and the detection methods among different medical institutions may affect the external applicability of this model.

Despite these limitations, our study still holds significant clinical implications. The results of this study confirm the prognostic predictive value of pretreatment peripheral blood electrolyte, nutritional, inflammatory indicators, and clinical characteristic indicators, which are characterized by low detection cost, simple operation, and rapid result generation for patients with advanced EC undergoing immunotherapy or radioimmunotherapy. Moreover, the two multi-dimensional nomogram prediction models constructed in this study can effectively predict the 1-, 2-, and 3-year OS rates and 3-, 6-, and 9-month PFS rates of this population. This provides a risk stratification tool for clinicians, but its clinical utility still needs to be verified by prospective, multi-center trials. In the future, the conclusions of this study can be further validated through external validation with large samples from multiple centers and prospective cohort studies. Additionally, the dynamic changes of related indicators throughout the course of immunotherapy and their association with patient prognosis can be further investigated, and the potential molecular mechanisms by which these indicators affect the prognosis of EC immunotherapy can be explored in depth by combining in vitro and in vivo basic experiments.

Conclusions

T stage, serum sodium, serum calcium, and SII were independent prognostic factors for OS in patients with advanced EC receiving immunotherapy or radioimmunotherapy. The independent prognostic factors for PFS in these patients included serum sodium, PNI, age, and treatment modality. Based on these results, we developed and validated two nomograms that can accurately predict the 1-, 2-, and 3-year OS and the 3-, 6-, and 9-month PFS probabilities in these patients. These nomograms may be helpful in conducting risk stratification for patients in clinical practice, but they need to be validated in larger-scale trials.

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 reviewed and approved by the Medical Ethics Committee of The First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital; approval No. 2022 S398) and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was retrospective in nature and did not involve the disclosure of patients’ private information; thus, the requirement for informed consent was waived by our Medical Ethics Committee. All patient data were anonymized and handled with strict confidentiality throughout the study.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0361/dss

jtd-18-04-405-dss.pdf ^{(97.1KB, pdf)}

DOI: 10.21037/jtd-2026-1-0361