Development and validation of a prognostic immunoinflammatory index for patients with gastric cancer

Abstract

BACKGROUND

Studies have demonstrated the influence of immunity and inflammation on the development of tumors. Although single biomarkers of immunity and inflammation have been shown to be clinically predictive, the use of biomarkers integrating both to predict prognosis in patients with gastric cancer remains to be investigated.

AIM

To investigate the prognostic and clinical significance of inflammatory biomarkers and lymphocytes in patients undergoing surgical treatment for gastric cancer.

METHODS

Univariate COX regression analysis was performed to identify potential prognostic factors for patients with gastric cancer undergoing surgical treatment. Least absolute shrinkage and selection operator-COX (LASSO-COX) regression analysis was performed to integrate these factors and formulate a new prognostic immunoinflammatory index (PII). The correlation between PII and clinical characteristics was statistically analyzed. Nomograms incorporating the PII score were devised and validated based on the time-dependent area under the curve and decision curve analysis.

RESULTS

Patients exhibiting elevated neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, and systemic immune inflammatory index displayed inferior progression-free survival (PFS) and overall survival (OS). Conversely, low levels of CD3(+), CD3(+) CD8(+), CD4(+)CD8(+), and CD3(+)CD16(+)CD56(+) T lymphocytes were associated with improved PFS and OS, while high CD19(+) T lymphocyte levels were linked to worse PFS and OS. The PII score demonstrated associations with tumor characteristics (primary tumor site and tumor size), establishing itself as an independent prognostic factor for both PFS and OS. Time-dependent area under the curve and decision curve analysis affirmed the effectiveness of the PII-based nomogram as a robust prognostic predictive model.

CONCLUSION

PII may be a reliable predictor of prognosis in patients with gastric cancer undergoing surgical treatment, and it offers insights into cancer-related immune-inflammatory responses, with potential significance in clinical practice.

Key Words: Gastric cancer; Immunity; Inflammation; Prognosis; Nomogram

Core Tip: We conducted a retrospective study in which data for lymphocyte subsets and blood inflammatory markers were collected from 291 patients who underwent surgical treatment for gastric cancer. Least absolute shrinkage and selection operator-COX regression analysis was used to construct a new immune-inflammatory biomarker. Internal validation divided into training and validation sets confirmed that patients with higher prognostic immunoinflammatory index scores had a worse prognosis.

INTRODUCTION

Gastric cancer is the fifth most prevalent malignancy globally, accounting for > 1 million new cases annually and posing a persistent public health challenge[1]. Investigations on the incidence and mortality trends of gastric cancer underscore a consistent decline worldwide[2-4]. Despite this trend, East Asia remains disproportionately burdened with the highest rates of gastric cancer incidence and mortality among geographical regions[5,6]. Advances in multimodal treatment strategies for gastric cancer have been substantiated by scientific evidence[7,8]. Surgical resection is currently the primary therapeutic option for resectable gastric cancer, while cases with advanced or inoperable disease benefit from che-motherapy, immunotherapy, and targeted therapies[9-11]. However, patients with advanced metastases face a grim prognosis, with a median survival time of approximately 1 year[12]. Traditional classification tools provide limited prognostic information for patients with gastric cancer, necessitating the exploration of convenient and effective clinical biomarkers to guide treatment and assess prognosis[13].

Cancer is recognized as a chronic inflammatory condition, in which neutrophils and monocytes infiltrate adipose tissue and other tissues, thereby contributing to nutrient depletion and disease progression[14]. Inflammation emerges as a pivotal risk factor for tumor advancement[15]. Peripheral blood biomarkers, such as neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and the systemic immune inflammatory index (SII) which is responsive to systemic inflammation, have been established as prognostic assessment tools for various types of cancer[16-19]. Beyond assessing the degree of inflammation, consideration of the body's immune status is imperative. Peripheral blood immune responses to cancer cells constitute a highly sensitive defense mechanism, outperforming conventional tumor markers in the early diagnosis of gastric cancer[20]. Furthermore, cancer cells within the tumor microenvironment activate immune cells upon entering the bloodstream. This establishes a correlation between immune cells in the tumor microenvironment and the patient’s systemic inflammatory response[21,22]. Lymphocyte subsets, serving as monitors of patient immunity, have emerged as critical biomarkers for evaluating patient prognosis[23,24].

In this study, we performed a retrospective analysis involving 291 patients with gastric cancer who underwent surgical interventions. Comprehensive datasets comprising laboratory indexes for inflammatory biomarkers and lymphocyte subsets were obtained from these patients. By employing the least absolute shrinkage and selection operator-COX (LASSO-COX) regression analysis, we amalgamated the identified biomarkers to formulate a novel immune-inflammatory biomarker denoted as the prognostic immunoinflammatory index (PII). The primary objective of this study was to analyze the predictive capabilities of the PII specifically within the context of patients with gastric cancer who underwent surgical treatment.

MATERIALS AND METHODS

Patients

This study included patients with gastric cancer who underwent surgical treatment at Harbin Medical University Cancer Hospital between January 2016 and December 2019. The inclusion criteria were: (1) Histologically confirmed diagnosis of gastric cancer; (2) absence of extrahepatic metastasis; and (3) completion of lymphocyte subset testing. The exclusion criteria were: (1) Having received alternative treatments prior to surgery; (2) history of other malignant tumors; (3) insufficient clinical data or lack of follow-up information; and (4) presence of certain diseases or conditions affecting hematological parameters. The enrolled patients were randomly allocated based on a 7:3 ratio, into a training set (n = 204) and a validation set (n = 87). Due to the retrospective nature of this study, the requirement for informed consent was waived by the Ethics Committee of the Affiliated Cancer Hospital of Harbin Medical University.

Data collection and definition of variables

Comprehensive data were systematically obtained from all participants encompassing clinical characteristics, tumor attributes, laboratory-related examinations, and lymphocyte subset analyses. This study specifically focused on preoperative immune-inflammatory biomarkers and lymphocyte subsets, including NLR, PLR, SII, and CD3(+), CD3(+)CD4(+), CD3(+)CD8(+), CD4(+)CD8(+), CD3(+)CD16(+)CD8(+), CD3(-)CD16(+)CD56(+), and CD3(+)CD16(+)CD56(+) lymphocyte subsets. The calculation of biomarkers was performed using the following formulas: NLR = neutrophils/lymphocytes; PLR = platelets/lymphocytes; SII = neutrophils/lymphocytes × platelets; and systemic inflammation response index (SIRI) = (neutrophils × monocytes)/lymphocytes.

The primary endpoints for assessment were overall survival (OS), defined as the period from the date of surgery to that of death or last follow-up, and progression-free survival (PFS), defined as the interval between the date of surgery and that of disease recurrence or last follow-up.

PII score construction

Univariate COX regression analysis was conducted for all immune-inflammatory biomarkers and lymphocyte subsets. Variables exhibiting significance based on a threshold of P < 0.05 were retained for subsequent analyses. The prognostic significance of these retained biomarkers was evaluated using the LASSO-COX regression analysis. Ultimately, PII scores were computed based on variables that did not possess zero coefficients in the LASSO-COX regression analysis.

Statistical analysis

For categorical variables, analyses were conducted using the Pearson's chi-squared test or Fisher's test. For continuous variables, the Student’s t-test was used. Kaplan-Meier curves were employed to illustrate and compare the differences in OS and PFS utilizing the log-rank test. Nomograms representing OS and PFS were generated based on the results derived from COX multivariate regression analysis, employing the inverse stepwise selection method to minimize Akaike information criterion values. To validate the performance and clinical benefit of the nomograms in the training cohort, time-dependent area under the curve (AUC) and decision curve analysis (DCA) were utilized.

RESULTS

Baseline characteristics and PII score

A total of 291 patients participated in this study (male: 69.8%, female: 30.2%; mean age: 59.05 years). Radical surgery was performed in 274 individuals (94.1% of the study population). Demographic and clinical characteristics of the training (n = 204) and validation (n = 87) sets are shown in Table 1.

Table 1 Clinicopathological characteristics, n (%).

Variable	All patients	Training set	Validation set
Age (years), mean (± SD)	59.05 (10.45)	58.45 (10.38)	60.48 (10.54)
Sex,
Male	203 (69.8)	144 (70.6)	59 (67.8)
Female	88 (30.2)	60 (29.4)	28 (32.2)
Body mass index (kg/m2), mean (SD)	22.86 (3.22)	22.95 (3.04)	22.65 (3.63)
Radical resection
Yes	274 (94.2)	195 (95.6)	79 (90.8)
No	17 (5.8)	9 (4.4)	8 (9.2)
Primary tumor site
Upper 1/3	11 (3.7)	8 (3.9)	3 (3.4)
Middle 1/3	38 (13.1)	28 (13.7)	10 (11.5)
Low 1/3	208 (71.5)	142 (69.7)	66 (75.9)
Whole	34 (11.7)	26 (12.7)	8 (9.2)
Borrmann type
Ⅰ	32 (11.0)	22 (10.8)	10 (11.5)
Ⅱ	87 (29.9)	63 (30.9)	24 (27.6)
Ⅲ	153 (52.6)	109 (53.4)	44 (50.6)
Ⅳ	19 (6.5)	10 (4.9)	9 (10.3)
Lymph node positivity
Yes	138 (47.4)	99 (48.5)	39 (44.8)
No	153 (52.6)	105 (51.5)	48 (55.2)
Tumor size
< 20 mm	29 (10.0)	18 (8.8)	11 (12.6)
20–50 mm	124 (42.6)	89 (43.6)	35 (40.2)
> 50 mm	138 (47.4)	97 (47.6)	41 (47.1)
Differentiation
Poor	101 (34.7)	72 (35.3)	29 (33.3)
Moderate	149 (51.2)	106 (52.0)	43 (49.4)
Well	26 (8.9)	19 (9.3)	7 (8.0)
Unknown	15 (5.2)	7 (3.4)	8 (9.3)
Lauren type
Intestinal	143 (49.1)	105 (51.5)	38 (43.7)
Diffuse	51 (17.5)	35 (17.2)	16 (18.4)
Mixed	84 (28.9)	58 (28.4)	26 (29.9)
Unknown	13 (4.5)	6 (2.9)	7 (8.0)
TNM stage
Ⅰ	117 (40.2)	81 (39.7)	36 (41.4)
Ⅱ	71 (24.4)	53 (26.0)	18 (20.7)
Ⅲ	89 (30.6)	62 (30.4)	27 (31.0)
Ⅳ	14 (4.8)	8 (3.9)	6 (6.9)
Carcinoembryonic antigen
< 1.97 ng/mL	144 (49.5)	107 (52.5)	37 (42.5)
≥ 1.97 ng/mL	147 (50.9)	97 (47.5)	50 (57.5)
CA199
< 10.19 U/L	145 (49.8)	103 (50.5)	42 (48.3)
≥ 10.19 U/L	146 (50.2)	101 (49.5)	45 (51.7)
CA724
< 2.17 U/L	145 (49.8)	106 (52.0)	39 (44.8)
≥ 2.17 U/L	146 (50.2)	98 (48.0)	48 (55.2)
CA125II
< 10.21 U/L	145 (49.8)	102 (50.0)	43 (49.4)
≥ 10.21 U/L	146 (50.2)	102 (50.0)	44 (50.6)
NLR, mean (SD)	2.63 (2.98)	2.62 (3.03)	2.66 (2.87)
PLR, mean (SD)	148.10 (7.47)	149.64 (78.22)	144.50 (68.87)
SII, mean (SD)	668.08 (768.43)	683.94 (796.04)	630.89 (702.44)
SIRI, mean (SD)	70.69 (29.36)	70.05 (24.26)	72.18 (38.93)
CD3(+) (%), mean (SD)	70.91 (38.40)	69.00 (9.31)	75.41 (68.84)
CD3(+)CD4(+) (%), mean (SD)	40.82 (8.67)	40.79 (8.22)	40.88 (9.69)
CD3(+)CD8(+) (%), mean (SD)	23.42 (8.07)	23.49 (8.02)	23.24 (8.25)
CD4(+)/CD8(+), mean (SD)	2.04 (1.08)	1.99 (0.92)	2.17 (1.37)
CD3(+)CD4(+)CD8(+) (%), mean (SD)	0.54 (1.23)	0.61 (1.44)	0.36 (0.37)
CD19(+) (%), mean (SD)	11.32 (4.71)	11.68 (4.76)	10.48 (4.51)
CD3(-)CD16(+)CD56(+) (%), mean (SD)	16.72 (9.55)	16.25 (8.34)	17.83 (11.02)
CD3(+)CD16(+)CD56(+) (%), mean (SD)	3.13 (4.34)	3.26 (4.89)	2.83 (2.63)

CA125II: Carbohydrate antigen 125II; CA199: Carbohydrate antigen 199; CA724: Carbohydrate antigen 724; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio; SII: Systemic immune inflammatory index; SIRI: Systemic inflammation response index.

Survival analysis employing the Kaplan-Meier method demonstrated that, among the inflammatory biomarkers, NLR, PLR, SII, and SIRI were identified as prognostic risk factors. Elevated NLR, PLR, SII, and SIRI were associated with diminished OS. In the analysis of lymphocyte subsets, lower CD3(+), CD3(+)CD8(+), CD4(+)CD8(+), and CD3(+)CD16(+)CD56(+) lymphocyte subsets were linked to improved OS. CD19(+) lymphocytes conferred a protective effect, indicating that higher CD19(+) lymphocyte levels were associated with reduced OS (Figure 1). Kaplan-Meier analyses of the relationship between inflammatory biomarkers, lymphocyte subsets, and PFS consistently supported these findings (Figure 2). The Kaplan-Meier analysis identified all eight immune and inflammatory biomarkers as prognostic factors for OS. To isolate independent prognostic factors, all eight biomarkers were subjected to the LASSO-COX regression analysis, revealing six biomarkers with non-zero coefficients. The subsequent equation was established based on their coefficients: PII score = 0.608 × NLR + 0.277 × PLR + 0.059 × SIRI + 0.480 × CD3(+) + 0.176 × CD3(+)CD16(+)CD56(+) - 0.459 × CD4(+)CD8(+) - 0.501 × CD19(+) (Figure 3A and B). The intercorrelations between the eight immune-inflammatory biomarkers are depicted in Figure 3C. In the training set, a high PII score was associated with inferior PFS and OS (Figure 3D and E).

Figure 1 Kaplan–Meier curves for progression-free survival. A: Progression-free survival (PFS) by CD3(+) lymphocytes; B: PFS by CD3(+)CD8(+) lymphocytes; C: PFS by CD4(+)CD8(+) lymphocytes; D: PFS by CD3(+)CD16(+)CD56(+) lymphocytes; E: PFS by CD19(+) lymphocytes; F: PFS by platelet-to-lymphocyte ratio; G: PFS by neutrophil-to-lymphocyte ratio; H: PFS by systemic immune inflammatory index; I: PFS by systemic inflammation response index. PLR: Platelet-to-lymphocyte ratio; NLR: Neutrophil-to-lymphocyte ratio; SII: Systemic immune inflammatory index; SIRI: Systemic inflammation response index; PFS: Progression-free survival.

Figure 2 Kaplan–Meier curves for overall survival. A: Overall survival by CD3(+) lymphocytes; B: OS by CD3(+)CD8(+) lymphocytes; C: OS by CD4(+)CD8(+) lymphocytes; D: OS by CD3(+)CD16(+)CD56(+) lymphocytes; E: OS by CD19(+) lymphocytes; F: OS by platelet-to-lymphocyte ratio; G: OS by neutrophil-to-lymphocyte ratio; H: OS by systemic immune inflammatory index; I: OS by systemic inflammation response index. PLR: Platelet-to-lymphocyte ratio; NLR: Neutrophil-to-lymphocyte ratio; SII: Systemic immune inflammatory index; SIRI: Systemic inflammation response index; PFS: Progression-free survival; OS: Overall survival.

Figure 3 Construction of PII score using the least absolute shrinkage and selection operator-COX regression model. A: Partial likelihood deviance for least absolute shrinkage and selection operator (LASSO) coefficient profiles; B: LASSO coefficient profiles of eight immune and inflammatory related biomarkers; C: Heatmap of correlations of the immune and inflammatory related biomarkers; D: Progression-free survival by prognostic immunoinflammatory index (PII); E: Overall survival by PII. PII: Prognostic immunoinflammatory index; PFS: Progression-free survival; OS: Overall survival.

Association between PII score and clinical characteristics

The PII score construction involved determining the optimal cutoff value through receiver operating characteristic curve analysis (i.e., 70.69). Subsequently, patients were stratified into high- and low-value groups based on this cutoff value. Exploration of potential differences in clinical characteristics among distinct PII score groups ensued. Utilizing the Pearson's chi-squared test or Fisher's test, we conducted a comprehensive analysis of the clinical characteristics among different PII score groups. The outcomes revealed statistically significant variations in body mass index, primary tumor site, tumor size, carbohydrate antigen 199 (CA199), NLR, PLR, SII, SIRI, and CD3(+), CD3(+)CD8(+), and CD19(+) lymphocyte subsets (P < 0.05 for all), as depicted in Table 2.

Table 2 Association between prognostic immunoinflammatory index score and clinical characteristics, n (%).

Variable	PII score		P value
	Group 1, n = 130	Group 2, n = 74
Age (years), mean (SD)	58.03 (10.52)	59.18 (10.17)	0.398
Sex			0.301
Male	95 (73.1)	49 (66.2)
Female	35 (26.9)	25 (33.8)
Body mass index (kg/m2), mean (SD)	23.38 (3.13)	22.19 (2.73)	0.004
Radical resection			0.075
Yes	127 (97.7)	68 (91.9)
No	3 (2.3)	6 (8.1)
Primary tumor site			0.016
Upper 1/3	4 (3.1)	4 (5.4)
Middle 1/3	17 (13.1)	11 (14.9)
Low 1/3	99 (76.1)	43 (58.1)
Whole	10 (7.7)	16 (21.6)
Borrmann type			0.275
Ⅰ	16 (12.3)	6 (8.1)
Ⅱ	43 (33.1)	20 (27.0)
Ⅲ	67 (51.5)	42 (56.8)
Ⅳ	4 (3.1)	6 (8.1)
Lymph node positivity
Yes	67 (47.7)	37 (50.0)
No	68 (52.3)	37 (50.0)
Tumor size			0.031
< 20 mm	14 (10.8)	4 (5.4)
20–50 mm	63 (48.5)	26 (35.1)
> 50 mm	53 (40.7)	44 (59.5)
Differentiation			0.100
Poor	43 (33.2)	29 (39.2)
Moderate	70 (53.8)	36 (48.6)
Well	15 (11.5)	4 (5.4)
Unknown	2 (1.5)	5 (6.8)
Lauren type			0.125
Intestinal	67 (51.5)	38 (51.4)
Diffuse	23 (17.7)	12 (16.2)
Mixed	39 (30.0)	19 (25.6)
Unknown	1 (0.8)	5 (6.8)
TNM stage			0.155
Ⅰ	52 (40.0)	29 (39.2)
Ⅱ	34 (26.2)	19 (25.7)
Ⅲ	42 (32.3)	20 (27.0)
Ⅳ	2 (1.5)	6 (8.1)
Carcinoembryonic antigen			0.729
< 1.97 ng/mL	67 (51.5)	40 (54.1)
≥1.97 ng/mL	63 (48.5)	34 (45.9)
CA199			0.032
< 10.19 U/L	73 (56.2)	30 (40.5)
≥ 10.19 U/L	57 (43.8)	44 (59.5)
CA724			0.056
< 2.17 U/L	61 (46.9)	45 (60.8)
≥ 2.17 U/L	69 (53.1)	29 (39.2)
CA125II			0.244
< 10.21 U/L	61 (46.9)	41 (55.4)
≥ 10.21 U/L	69 (53.1)	33 (44.6)
NLR, mean (SD)	1.69 (0.67)	4.24 (4.54)	< 0.001
PLR, mean (SD)	107.02 (31.74)	224.36 (79.72)	< 0.001
SII, mean (SD)	392.19 (180.21)	1196.46 (1134.42)	< 0.001
SIRI, mean (SD)	0.88 (0.48)	2.37 (3.27)	< 0.001
CD3(+) (%), mean (SD)	67.21 (9.87)	72.14 (7.30)	< 0.001
CD3(+)CD4(+) (%), mean (SD)	40.46 (8.14)	41.34 (8.38)	0.445
CD3(+)CD8(+) (%), mean (SD)	22.36 (7.60)	25.47 (8.40)	0.005
CD4(+)/CD8(+), mean (SD)	2.07 (0.94)	1.84 (0.87)	0.069
CD3(+)CD4(+)CD8(+) (%), mean (SD)	0.53 (0.73)	0.76 (2.20)	0.657
CD19(+) (%), mean (SD)	12.58 (4.89)	10.10 (4.09)	< 0.001
CD3(-)CD16(+)CD56(+) (%), mean (SD)	16.99 (9.71)	14.94 (6.92)	0.282
CD3(+)CD16(+)CD56(+) (%), mean (SD)	3.17 (5.48)	3.41 (3.67)	0.202

CA125II: Carbohydrate antigen 125II; CA199: Carbohydrate antigen 199; CA724: Carbohydrate antigen 724; NLR: Neutrophil-to-lymphocyte ratio; PII: Prognostic immunoinflammatory index; PLR: Platelet-to-lymphocyte ratio; SII: Systemic immune inflammatory index; SIRI: Systemic inflammation response index; TNM: Tumor-node-metastasis.

COX regression analysis and nomogram construction

Variables exhibiting statistical significance (i.e., P < 0.05) in the univariate COX regression analysis were incorporated into the subsequent multivariate regression analysis. Given the inclusion of the PII score in the multivariate analysis, individual biomarkers constituting the PII score were intentionally omitted to avoid statistical redundancy. The outcomes of the multivariate COX regression analysis identified tumor-node-metastasis (TNM) stage, Borrmann type, and PII score as independent prognostic factors for PFS and OS, as shown in Table 3. Nomograms for PFS and OS were subsequently constructed, encompassing TNM stage, Borrmann type, and PII score (Figure 4). The predictive models exhibited substantial accuracy, with a concordance index of 0.798 and 0.792 for PFS and OS, respectively.

Figure 4 Nomograms for prediction of survival. A: Nomogram for progression-free survival; B: Nomogram for overall survival. TNM: Tumor-node-metastasis; PII: Prognostic immunoinflammatory index.

Table 3 COX regression analysis of overall survival and progression-free survival.

Parameter	OS				PFS
	Univariate analysis		Multivariate analysis		Univariate analysis		Multivariate analysis
	HR (95%CI)	P value	HR (95%CI)	P value	HR (95%CI)	P value	HR (95%CI)	P value
Age (years)	2.035 (1.325-3.061)	0.001	1.541 (0.982-2.417)	0.051	1.978 (1.302-3.007)	0.001	1.488 (0.948-2.335)	0.084
Sex
Male	1 (Ref.)				1 (Ref.)
Female	0.912 (0.597-1.392)	0.669			0.918 (0.601-1.401)	0.692
Body mass index (kg/m2)	0.527 (0.342-0.811)	0.004	0.853 (0.536-1.357)	0.502	0.527 (0.342-0.812)	0.004	0.838 (0.526-1.334)	0.456
CD3(+) (%)	2.051 (1.387-3.033)	< 0.001			1.995 (1.350-2.949)	0.001
CD3(+)CD4(+) (%)	1.526 (0.961-2.424)	0.073			1.528 (0.963-2.426)	0.072
CD3(+)CD8(+) (%)	1.516 (1.034-2.221)	0.029			1.513 (1.003-2.217)	0.030
CD4(+)/CD8(+)	0523 (0.341-0.802)	0.002			0.517 (0.337-0.792)	0.002
CD3(+)CD4(+)CD8(+) (%)	1.516 (0.940-2.445)	0.088			1.504 (0.933-2.425)	0.094
CD19(+) (%)	0.351 (0.171-0.721)	0.003			0.358 (0.174-0.735)	0.004
CD3(-)CD16(+)CD56(+) (%)	0.703 (0.480-1.030)	0.070			0.712 (0.486-1.043)	0.082
CD3(+)CD16(+)CD56(+) (%)	1.880 (1.207-2.931)	0.005			1.865 (1.197-2.905)	0.005
PLR
< 2.81	1 (Ref.)				1 (Ref.)
≥ 2.81	1.913 (1.241-2.949)	0.003			1.898 (1.231-2.926)	0.003
NLR
< 115.22	1 (Ref.)				1 (Ref.)
≥ 155.22	2.390 (1.588)	< 0.001			2.338 (1.554-3.518)	< 0.001
SII
< 533.41	1 (Ref.)				1 (Ref.)
≥ 533.41	1.514 (1.035-2.217)	0.031			1.515 (1.035-2.217)	0.032
SIRI
< 1.51	1 (Ref.)				1 (Ref.)
≥ 1.51	1.899 (1.263-2.857)	0.002			1.881 (1.251-2.830)	0.002
Radical resection
Yes	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
No	4.356 (2.431-7.807)	< 0.001	1.909 (0.789-4.615)	0.151	4.182 (2.335-7.492)	< 0.001	1.513 (0.595-3.846)	0.384
Borrmann type
Ⅰ	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
Ⅱ	6.025 (1.435-25.300)	0.014	2.237 (0.477-10.497)	0.307	6.081 (1.448-25.533)	0.014	2.278 (0.484-10.729)	0.298
Ⅲ	8.012 (1.958-32.780)	0.004	2.350 (0.515-10.715)	0.270	8.088 (1.977-33.091)	0.004	2.435 (0.533-11.124)	0.251
Ⅳ	27.087 (6.171-118.891)	< 0.001	5.407 (1.063-27.496)	0.042	28.997 (6.605-127.294)	< 0.001	5.301 (1.043-26.930)	0.044
Lymph node positivity
No	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
Yes	3.445 (2.264-5.242)	< 0.001	1.132 (0.545-2.351)	0.740	3.537 (2.324-5.383)	< 0.001	1.010 (0.485-2.103)	0.979
Tumor size
< 20 mm	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
20–50 mm	2.710 (0.829-8.858)	0.099	0.708 (0.192-2.607)	0.603	2.715 (0.831-8.870)	0.098	0.686 (0.185-2.544)	0.573
> 50 mm	6.917 (2.176-21.988)	0.001	0.666 (0.178-2.494)	0.546	6.883 (2.165-21.878)	0.001	0.622 (0.164-2.361)	0.485
TNM stage
Ⅰ	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
Ⅱ	3.833 (1.585-7.910)	< 0.001	3.566 (1.545-8.233)	0.003	3.875 (1.878-7.996)	< 0.001	3.433 (1.475-7.989)	0.004
Ⅲ	11.441 (5.999-21.819)	< 0.001	9.085 (3.382-24.404)	< 0.001	11.807 (6.187-22.533)	< 0.001	8.541 (3.152-23.143)	< 0.001
Ⅳ	35.899 (15.895-81.079)	< 0.001	11.978 (3.812-37.641)	< 0.001	45.844 (20.022-104.969)	< 0.001	16.297 (4.934-53.825)	< 0.001
Carcinoembryonic antigen
< 1.97 ng/mL	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
≥ 1.97 ng/mL	1.586 (1.078-2.335)	0.019	1.425 (0.926-2.195)	0.107	1.543 (1.048-2.271)	0.028	1.328 (0.867-2.034)	0.192
CA199
< 10.19 U/L	1 (Ref.)				1 (Ref.)
≥ 10.19 U/L	1.355 (0.923-1.989)	0.120			1.347 (0.918-1.977)	0.128
CA724
< 2.17 U/L	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
≥ 2.17 U/L	2.159 (1.451-3.213)	< 0.001	1.510 (0.992-2.297)	0.054	2.154 (1.448-3.205)	< 0.001	1.456 (0.957-2.214)	0.079
CA125II
< 10.21 U/L	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
≥ 10.21 U/L	1.849 (1.250-2.734)	0.002	1.198 (0.791-1.813)	0.394	1.837 (1.242-2.716)	0.002	1.171 (0.770-1.781)	0.460
PII
< 70.69	1 (Ref.)		1 (Ref.)		1 (Ref.)		1 (Ref.)
≥ 70.69	1.888 (1.289-2.766)	0.001	1.670 (1.115-2.500)	0.013	1.881 (1.284-2.755)	0.001	1.639 (1.096-2.450)	0.016

CA125II: Carbohydrate antigen 125II; CA199: Carbohydrate antigen 199; CA724: Carbohydrate antigen 724; HR: Hazard ratio; NLR: Neutrophil-to-lymphocyte ratio; OS: Overall survival; PFS: Progression-free survival; PII: Prognostic immunoinflammatory index; PLR: Platelet-to-lymphocyte ratio; SII: Systemic immune inflammatory index; SIRI: Systemic inflammation response index; TNM: Tumor-node-metastasis.

The assessment of predictive performance involved the generation of time-dependent AUC and DCA. The 1-, 3-, and 5-year AUC values of the nomogram for PFS were 0.834, 0.841, and 0.863, respectively (Figure 5A). Similarly, the 1-, 3-, and 5-year AUC values of the nomogram for OS were 0.830, 0.821, and 0.850, respectively (Figure 5B). DCA demonstrated that the PII-nomograms of PFS and OS were slightly better than the TNM staging system in terms of predictive power (Figure 6). DCA further affirmed the clinical utility and patient benefit of the nomogram prediction model constructed using the PII, as illustrated in Figure 7.

Figure 5 Time-dependent receiver operating characteristic curve analysis of the nomogram fo prediction of survival. A: 1-, 3-, and 5-year area under the curves for progression-free survival; B: 1-, 3-, and 5-year area under the curves for overall survival. AUC: Area under the receiver operating characteristic curve.

Figure 6 Decision curve analysis of the nomogram for prediction of survival. A: Decision curve analysis for progression-free survival; B: Decision curve analysis for overall survival.

Figure 7 Decision curve analysis of the nomogram for prediction of 12-, 36-, and 60-mo survival. A: 12-, 36-, and 60-mo decision curve analysis for progression-free survival; B: 12-, 36-, and 60-mo decision curve analysis for overall survival.

DISCUSSION

Studies have revealed the intricate connection between immunity, inflammation, and cancer[25,26]. Inflammation assumes a pivotal role across tumorigenesis, growth stages, and developmental phases, exerting a lasting impact on subsequent treatment modalities and patient survival[27]. The carcinogenic potential of Helicobacter pylori, a recognized risk factor for gastric cancer, is closely associated with its induction of chronic inflammation, thereby promoting tumor development[28]. This symbiotic relationship between gastric cancer and inflammation underscores chronic inflammation as a predisposing factor for gastric cancer, with tumor cell heterogeneity reciprocally enhancing the expression of inflammatory cytokines and chemokines[29]. Notably, the role of inflammation extends to tumor metastasis, as evidenced by studies indicating a significant impact on cancer-related mortality, particularly in gastrointestinal cancers and distant metastases; prolonged aspirin use contributes to reduced overall mortality[30,31]. In light of these considerations, the identification of convenient and effective biomarkers for assessing the degree of inflammation in patients remains crucial.

Numerous clinical studies have investigated the prognostic implications of inflammatory biomarkers and lymphoid subgroups in patients with gastric cancer. For instance, Miyamoto et al[32] conducted a study involving 154 consecutive patients with gastric cancer, demonstrating that a high NLR correlated with poorer clinical outcomes. Zurlo et al[33] corroborated these findings in their assessment of NLR as a prognostic factor in patients with locally advanced gastric cancer. PLR has also emerged as an independent predictor of mortality and morbidity in gastric cancer[34]. Comparative studies on NLR and PLR as prognostic factors for gastric cancer, such as the retrospective study performed by Kim et al[35] involving 2000 patients who underwent radical surgery, concluded that NLR exhibited greater prognostic power than PLR. The SII, encompassing lymphocytes, platelet count, and neutrophils, provides a comprehensive estimate of the body's immune response and inflammatory status[36,37]. Zhang et al[38] and He et al[39] proposed SII as a good biomarker for diagnosing gastric cancer compared with NLR and PLR, highlighting its significant correlation with OS in this setting. It has been shown that SIRI predicts patient prognosis in a variety of cancer types[40,41]. In our hospital, the study of lymphocyte subpopulations further affirmed their role as predictive biomarkers for gastric cancer, surpassing the predictive capability of inflammatory biomarkers[42]. Additionally, lymphocyte subsets, in conjunction with the nutritional indicator prognostic nutritional index, emerged as prognostic biomarkers, reflecting the immune and nutritional status of patients[43].

While serum immune indicators are easily detectable in clinical settings, individual immune and inflammatory biomarkers exhibit limitations in prognostic risk analysis. Therefore, the integration of indicators is imperative to establish a more comprehensive biomarker for assessing the immune and inflammatory status of patients. In our study, the initial dichotomization of collected immune and inflammatory biomarkers, followed by univariate COX regression analysis, revealed potential prognostic factors associated with the prognosis of gastric cancer. These factors were subjected to the LASSO-COX regression analysis, culminating in the derivation of the PII score, with non-zero coefficients serving as the basis for the formula. The PII, encompassing a broad spectrum of patient immune-inflammatory levels, demonstrated a significant correlation with OS in the training set, with high PII scores indicting poorer OS. Notably, PII scores exhibited correlations with tumor characteristics, including primary tumor site and tumor size. The 1-, 3-, and 5-year AUC values for the nomograms of PFS and OS, constructed by incorporating TNM stage, Borrmann type, and PII, exhibited remarkable performance. DCA further confirmed the beneficial impact of the PII score on patients with gastric cancer.

The immune response protects the body from the development of tumors; however, other immune responses (e.g., chronic inflammation) also play a role in this process[44]. The present study focused on inflammatory biomarkers and lymphoid subpopulations because altering the tumor microenvironment (including cytokines and chemokines) may alter the immune cells in patients receiving conventional treatment for gastric cancer[45]. Since inflammatory markers take into account absolute lymphocyte and neutrophil counts, the number of these immune cells may be affected by the cytokine environment. Detection of blood lymphocyte subsets also reflects the immune cell status. Therefore, the combination of inflammatory biomarkers and lymphocyte subsets utilized in this study to predict the prognosis of patients with gastric cancer is also of interest.

Despite the robust predictive accuracy of our model, certain limitations warrant consideration. The wide distribution of patients with gastric cancer, particularly across diverse ethnicities[46], highlights the need for multicenter studies in the future to validate the present findings. Additionally, the impact of Helicobacter pylori on serum immune-inflammatory factors was not investigated in this study. Furthermore, the retrospective nature of this study may have led to data omissions, potentially reducing the number of patients included in the analysis.

CONCLUSION

The PII may provide a comprehensive assessment of the immune-inflammatory status of the body, offering valuable insights for individualized prognostic evaluations in patients with gastric cancer. Elevated PII scores were associated with poorer clinical outcomes.