Prognostic value of neutrophil‐to‐lymphocyte ratio in gastric cancer patients treated with immune checkpoint inhibitors: A meta‐analysis

Abstract

The neutrophil‐to‐lymphocyte ratio (NLR) has been extensively studied in patients with gastric cancer (GC) treated with immune checkpoint inhibitors (ICIs), although the results are controversial. Therefore, we performed the present meta‐analysis to systematically assess the correlation between NLR and prognosis and clinicopathological factors in GC patients undergoing treatment with ICIs. Among the electronic databases, PubMed, Web of Science, Embase, and Cochrane Library were thoroughly searched. To estimate the prognostic value of NLR for progression‐free survival (PFS) and overall survival (OS), hazard ratios (HRs) were calculated, and their 95% confidence intervals were calculated. This meta‐analysis included 10 studies with 830 patients. Based on the pooled data, a high NLR was associated with poor OS in GC patients receiving ICIs (HR = 2.13, 95% confidence interval [CI] = 1.70–2.66, p < 0.001). Elevated NLR was a significant biomarker for worse PFS in GC patients treated with ICIs (HR = 1.74, 95% CI = 1.22–2.48, p = 0.002). There was, however, no significant correlation between NLR and sex, age, and Eastern Cooperative Oncology Group Performance Status (PS) of the ECOG, differentiation, human epidermal growth factor receptor 2 (HER2) status, or PD‐L1 status in GC patients treated with ICIs. High NLR before treatment was associated with poor OS and PFS in GC patients treated with ICIs, according to our meta‐analysis. NLR is an effective biomarker for evaluating the prognosis of GC patients receiving ICIs and provide more valuable information for treatment decisions in the era of immunotherapy for GC.

Abbreviations

CI

confidence interval

ECOG

Eastern Cooperative Oncology Group

GC

gastric cancer

GEJ

gastroesophageal junction

HER2

human epidermal growth factor receptor 2

HR

hazard ratio

ICIs

immune checkpoint inhibitors

LMR

lymphocyte‐to‐monocyte ratio

NLR

neutrophil‐to‐lymphocyte ratio

NOS

Newcastle–Ottawa scale

OR

odds ratio

OS

overall survival

PD‐1

programmed cell death‐1

PD‐L1

programmed cell death ligand‐1

PFS

progression‐free survival

PLR

platelet‐to‐lymphocyte ratio

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

PS

performance status

ROC

receiver‐operating characteristic

SII

systemic immune‐inflammation index

TILs

tumor infiltrating lymphocytes

TME

tumor microenvironment

1. INTRODUCTION

Globally, gastric cancer (GC) is the sixth most prevalent cancer and the third most common cause of cancer‐related deaths. ¹ In 2020, an estimated 1,089,103 new cases of GC and 768,793 deaths due to GC occurred worldwide. ¹ GC treatment includes surgery, chemotherapy, targeted therapy, and immunotherapy. ² , ³ In recent years, immune checkpoint inhibitors (ICIs), such as monoclonal antibodies against programmed cell death‐1 (PD‐1) and programmed cell death ligand‐1 (PD‐L1), have improved survival outcomes in patients with advanced GC (AGC). ⁴ , ⁵ However, because of the lack of reliable and novel biomarkers to predict the efficacy and survival of GC patients receiving ICIs, there is still a poor prognosis for patients with AGC. The 5‐year survival of GC is ~20%–30% in most countries around the world, except in Japan and Korea. ⁶ Therefore, the identification of efficient prognostic markers for GC patients treated with ICIs is urgently needed.

Growing evidence has shown that immune responses play a pivotal role in cancer progression and development. ⁷ A series of blood‐derived parameters have been demonstrated to be effective prognostic markers for solid tumors. ⁸ These indices include the platelet‐to‐lymphocyte ratio, ⁹ systemic immune‐inflammation index, ¹⁰ lymphocyte‐to‐monocyte ratio, ¹¹ and neutrophil‐to‐lymphocyte ratio (NLR). ¹² NLR is calculated as the neutrophil count divided by lymphocyte count in peripheral blood. Previous studies have explored the prognostic effect of NLR in GC patients receiving ICIs; however, the results were controversial. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² For example, some studies identified high NLR as a significant biomarker for GC patients receiving ICIs, ¹³ , ¹⁹ whereas other researchers denied this association. ¹⁴ , ²¹ , ²² Therefore, we conducted this meta‐analysis to evaluate the prognostic significance of NLR in GC patients treated with ICIs. Furthermore, we also investigated the correlation between NLR and clinicopathological characteristics in GC patients.

2. MATERIALS AND METHODS

2.1. Search strategy

Meta‐analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines. ²³ A thorough search was conducted in PubMed, Web of Science, Embase, and Cochrane Library. The search items were as follows: (“neutrophil‐to‐lymphocyte ratio” OR “NLR” OR “neutrophil to lymphocyte ratio”) AND (“gastric cancer” OR “gastric carcinoma” OR “gastric neoplasms”) AND (“immune checkpoint inhibitor” OR “CTLA‐4” OR “CTLA4” OR “ICIs” OR “immunotherapy” OR “programmed death ligand‐1 inhibitor” OR “PD‐L1 inhibitor” OR “programmed death‐1 inhibitor” OR “PD‐1 inhibitor” OR “atezolizumab” OR “avelumab” OR “cemiplimab” OR “durvalumab” OR “ipilimumab” OR “nivolumab” OR “pembrolizumab”). The search timeline was from inception to July 5, 2022. The publication search was limited to English language. A manual search of the references of retrieved studies was conducted to identify possible inclusions.

2.2. Inclusion and exclusion criteria

The inclusion criteria were as follows: (1) pathological evidence confirming the diagnosis of GC; (2) GC patients receiving ICI treatment, including but not limited to PD‐1/PD‐L1, CTLA‐4, or their combined inhibitors; (3) studies reporting hazard ratios (HR) and 95% confidence intervals (CI) for NLR in terms of survival; (4) a cut‐off value dividing high and low NLR; and (5) articles published in English. The exclusion criteria were as follows: (1) reviews, case reports, meeting abstracts, and letter; (2) studies that did not provide survival information; (3) animal studies; and (4) duplicate studies.

2.3. Data extraction and quality assessment

Two investigators (LL and LP) performed independent data extraction from eligible studies. Disagreements were resolved through discussions between the two investigators. In total, we extracted the following information: name of the first author, year, country, sample size, sex, study period, age, study design, treatment, NLR cut‐off value, method of determining NLR, follow‐up, survival outcomes, study center, cancer type, survival analysis type, and HRs and 95% CIs. The Newcastle–Ottawa scale (NOS) quality assessment tool was used for assessing the methodological quality of included studies. ²⁴ The NOS consists of three aspects: selection, comparability, and outcomes. NOS has a maximum score of 9 and studies with an NOS score ≥6 is typically considered high‐quality.

2.4. Statistical analysis

To assess the prognostic value of NLR in GC patients treated with ICIs, we calculated the combined HRs and 95% CIs. Heterogeneity among studies was evaluated using the Cochrane Q test and I ² test. All meta‐analyses used a random‐effects model because statistical heterogeneity could be easily underpowered, especially for subgroup analysis, where study numbers may be small in each group. Subgroup analysis stratified by various clinicopathological factors was used to detect the source of heterogeneity. The association between NLR and clinical features was evaluated by combining the odds ratios (ORs) and 95% CIs. To verify the stability of the results, a sensitivity analysis was conducted. Tests for publication bias were performed using Begg's and Egger's tests. All analyses were performed using Stata SE 12.0 (StataCorp, College Station, Texas). A value of p < 0.05 was considered statistically significant.

2.5. Ethics statement

The Institutional Review Board approval was not necessary because all the data in the study were retrieved from public databases.

3. RESULTS

3.1. Literature search

The initial literature search enrolled 120 records, and after the removal of duplicate studies, 55 records remained. A total of 39 records were discarded after examining the titles and abstracts, and 16 studies were subsequently evaluated by reading the full text. Six studies were excluded for the following reasons: no use of ICIs (n = 2), no survival data (n = 2), not regarding GC (n = 1), and no cut‐off value for NLR (n = 1). Finally, 10 studies with 830 patients ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² were included in this meta‐analysis (Figure 1).

FIGURE 1.

The PRISMA flowchart of the literature selection process.

3.2. Characteristics of included studies

As shown in Table 1, eligible studies had the following baseline characteristics. Five studies were conducted in Japan, ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ²⁰ three studies were performed in China, ¹⁷ , ¹⁹ , ²² and two studies were performed in South Korea. ¹⁸ , ²¹ Eight studies were retrospective studies ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ²⁰ , ²² and two were prospective trials. ¹⁹ , ²¹ All the studies were published in English. In this study, the sample size ranged from 24 to 185, with a median value of 74. All the studies used anti‐PD‐L1 antibodies (Table 1). The cut‐off values of NLR ranged from 2.5 to 5, with a median value of 2.95. The methods used to determine the cut‐off value of NLR were receiver operating characteristic (ROC) analysis (n = 6), ¹⁵ , ¹⁶ , ¹⁷ , ¹⁹ , ²⁰ , ²² median value (n = 2), ¹⁴ , ²¹ and literature (n = 2). ¹³ , ¹⁸ The survival analysis types were multivariate (n = 6) ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²¹ and univariate (n = 4). ¹³ , ¹⁴ , ²⁰ , ²² Seven studies were conducted in a single center ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ²⁰ , ²² and three studies were carried out in a multicenter. ¹³ , ¹⁹ , ²¹ The association between NLR and overall survival (OS) has been reported in 10 studies ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² in GC patients receiving ICIs and eight studies ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁹ , ²¹ , ²² reported progression‐free survival (PFS). NOS scores ranged from 7 to 9, indicating that all studies were of high quality (Table 1).

TABLE 1.

Basic characteristics of included studies in this meta‐analysis.

Study	Year	Country	Sample size	Gender (M/F)	Study design	Age (years) median (range)	Treatment	Cut‐off value of NLR	Cut‐off determination	Follow‐up (month) median (range)	Survival endpoint	Survival analysis	Study center	Study period	Cancer type	NOS score
Ogata, T.	2018	Japan	24	19/7	Retrospective	64 (44–86)	Nivolumab	5	Literature	5.7	OS, PFS	Univariate	Multicenter	Jun–Dec, 2017	AGC	8
Namikawa, T.	2020	Japan	29	19/10	Retrospective	71 (49–86)	Nivolumab	2.5	Median value	32	OS, PFS	Univariate	Single center	2017–2019	AGC	7
Ota, Y.	2020	Japan	98	68/30	Retrospective	66 (33–84)	Nivolumab	3	ROC curve	4.9 (0.5–18.7)	OS, PFS	Multivariate	Single center	2014–2018	Advanced/recurrent GC	8
Yamada, T.	2020	Japan	89	42/47	Retrospective	<65 years: 27 ≥65 years: 62	Nivolumab	2.5	ROC curve	5.8 (0.6–35.6)	OS, PFS	Multivariate	Single center	2014–2019	AGC	7
Gou, M.	2021	China	137	98/39	Retrospective	59	Nivolumab/Pembrolizumab/Sintilimab/Toripalimab	3.23	ROC curve	10.9	OS, PFS	Multivariate	Single center	2016–2020	Metastatic GC	8
Kim, N.	2021	South Korea	185	120/65	Retrospective	59 (47–70)	Nivolumab/Pembrolizumab	3	Literature	7.3	OS	Multivariate	Single center	2016–2019	AGC	8
Ruan, D. Y.	2021	China	58	41/17	Prospective	60 (52–66)	Toripalimab	2.7	ROC curve	4.5	OS, PFS	Multivariate	Multicenter	2016–2021	AGC	9
Ishido, K.	2022	Japan	59	45/14	Retrospective	71 (43–86)	Nivolumab	2.5	ROC curve	5.9 (0.6–43.6)	OS	Univariate	Single center	2017–2020	AGC	7
Kim, J. H.	2022	South Korea	45	34/11	Prospective	60 (23–76)	Nivolumab	2.9	Median value	28.3	OS, PFS	Multivariate	Multicenter	2014–2016	AGC	9
Qu, Z.	2022	China	106	72/34	Retrospective	≤65 years: 77 >65 years: 29	Nivolumab/Pembrolizumab/Toripalimab/Sintilimab/Camrelizumab	3.11	ROC curve	17.5	OS, PFS	Univariate	Single center	2019–2021	AGC	8

Abbreviations: AGC, advanced gastric cancer; GC, gastric cancer; NLR, neutrophil‐to‐lymphocyte ratio; NOS, Newcastle–Ottawa scale; OS, overall survival; PFS, progression‐free survival; ROC, receiver operating characteristic.

3.3. Association between NLR and OS

Ten studies with a total of 830 patients reported an association between NLR and OS in GC patients receiving ICIs. The pooled results were as follows: HR = 2.13, 95% CI = 1.70–2.66, p < 0.001, suggesting that a high NLR predicted poor OS in GC patients receiving ICIs (Figure 2; Table 2). Subgroup analysis demonstrated that an elevated NLR remained a significant prognostic factor for poor OS, irrespective of country, sample size, cut‐off value, survival analysis, cancer type, study center, or ICI treatment type (all p < 0.05; Table 2).

FIGURE 2.

Forest plots of NLR expression and overall survival risk in gastric cancer patients receiving immune checkpoint inhibitors.

TABLE 2.

Subgroup analysis of the prognostic value of NLR for OS in gastric cancer patients receiving ICIs.

Subgroups	No. of studies	No. of patients	Effects model	HR (95% CI)	p	Heterogeneity
						I 2 (%)	p value
Total	10	830	Random	2.13 (1.70–2.66)	<0.001	13.2	0.322
Country
Japan	5	299	Random	1.95 (1.47–2.60)	<0.001	0	0.485
China	3	301	Random	2.85 (1.44–5.64)	0.003	52.4	0.122
South Korea	2	230	Random	1.85 (1.16–2.94)	0.009	0	0.701
Sample size
<75	5	215	Random	2.27 (1.31–3.95)	0.004	44.0	0.126
≥75	5	615	Random	2.16 (1.69–2.76)	<0.001	0	0.533
Study design
Retrospective	8	727	Random	2.14 (1.73–2.65)	<0.001	0	0.497
Prospective	2	103	Random	3.66 (0.57–23.35)	0.170	74.9	0.046
Cut‐off value
<3	5	280	Random	1.90 (1.30–2.78)	0.001	26.6	0.244
≥3	5	550	Random	2.36 (1.81–3.08)	<0.001	0	0.438
Cut‐off determination
ROC curve	6	547	Random	2.23 (1.70–2.94)	<0.001	22.6	0.264
Literature	2	209	Random	2.68 (1.11–6.46)	0.029	38.4	0.203
Median value	2	74	Random	1.47 (0.83–2.61)	0.189	0	0.539
Survival analysis
Multivariate	6	612	Random	2.20 (1.65–2.94)	<0.001	25.2	0.245
Univariate	4	218	Random	1.96 (1.29–2.9)	0.002	12.1	0.332
Cancer type
AGC	8	595	Random	1.96 (1.49–2.57)	<0.001	9.0	0.360
Metastatic/recurrent GC	2	235	Random	2.50 (1.72–3.63)	<0.001	20.1	0.263
Study center
Single center	7	703	Random	2.09 (1.69–2.60)	<0.001	0	0.590
Multicenter	3	127	Random	3.81 (1.18–12.26)	0.025	62.5	0.070
Treatment ICIs type
Nivolumab	6	344	Random	1.91 (1.47–2.49)	<0.001	0	0.605
Nivolumab or other ICIs	4	486	Random	2.51 (1.59–3.95)	<0.001	39.4	0.175

Abbreviations: AGC, advanced gastric cancer; ICIs, immune checkpoint inhibitors; NLR, neutrophil‐to‐lymphocyte ratio; OS, overall survival; ROC, receiver operating characteristic.

3.4. Association between NLR and PFS

A total of eight studies with 586 patients showed that NLR was a significant prognostic factor for PFS in GC patients receiving ICIs. The combined data showed that a high NLR was a significant biomarker for worse PFS in GC patients treated with ICIs: HR = 1.74, 95% CI = 1.22–2.48, p = 0.002 (Table 3; Figure 3). Subgroup analysis indicated that the study design did not affect the prognosis of PFS based on NLR (all p < 0.05; Table 3).

TABLE 3.

Subgroup analysis of the prognostic value of NLR for PFS in gastric cancer patients receiving ICIs.

Subgroups	No. of studies	No. of patients	Effects model	HR (95% CI)	p	Heterogeneity
						I 2 (%)	p value
Total	8	586	Random	1.74 (1.22–2.48)	0.002	63.8	0.007
Country
Japan	4	240	Random	1.25 (0.88–1.77)	0.206	18.6	0.297
China	3	301	Random	2.41 (1.56–3.72)	<0.001	48.9	0.141
South Korea	1	45	—	2.17 (1.05–4.52)	0.037	—	—
Sample size
<75	4	156	Random	1.81 (1.13–2.90)	0.014	29.6	0.235
≥75	4	430	Random	1.71 (0.99–2.93)	0.053	80.1	0.002
Study design
Retrospective	6	483	Random	1.61 (1.01–2.56)	0.045	73.3	0.002
Prospective	2	103	Random	2.15 (1.35–3.43)	0.001	0	0.974
Cut‐off value
<3	4	221	Random	1.43 (0.88–2.32)	0.148	52.4	0.098
≥3	4	365	Random	2.07 (1.27–3.38)	0.003	66.3	0.031
Cut‐off determination
ROC curve	5	288	Random	1.78 (1.15–2.77)	0.010	73.8	0.004
Literature	1	24	—	2.43 (0.92–6.42)	0.073	—	—
Median value	2	74	Random	1.31 (0.45–3.88)	0.620	68.1	0.077
Survival analysis
Multivariate	5	427	Random	1.87 (1.19–2.93)	0.006	73.7	0.004
Univariate	3	159	Random	1.46 (0.79–2.69)	0.232	36.6	0.207
Cancer type
AGC	6	351	Random	1.55 (1.10–2.19)	0.013	32.6	0.192
Metastatic/recurrent GC	2	235	Random	2.17 (0.93–5.07)	0.073	87.4	0.005
Study center
Single center	5	459	Random	1.51 (0.90–2.55)	0.118	78.1	0.001
Multicenter	3	127	Random	2.20 (1.45–3.35)	<0.001	0	0.976
Treatment ICIs type
Nivolumab	5	285	Random	1.38 (0.97–1.94)	0.070	28.2	0.233
Nivolumab or other ICIs	3	301	Random	2.41 (1.56–3.72)	<0.001	48.9	0.141

Abbreviations: AGC, advanced gastric cancer; ICIs, immune checkpoint inhibitors; NLR, neutrophil‐to‐lymphocyte ratio; PFS, progression‐free survival; ROC, receiver operating characteristic.

FIGURE 3.

Forest plots of NLR expression and progression‐free survival risk in gastric cancer patients receiving immune checkpoint inhibitors.

3.5. Correlation between NLR and clinicopathologic parameters

We further investigated the association between NLR and clinicopathological features in GC patients receiving ICIs. The pooled results demonstrated that there was no significant correlation between NLR and sex (OR = 0.99, 95% CI = 0.65–1.53, p = 0.973), age (OR = 0.77, 95% CI = 0.32–1.85, p = 0.971), Eastern Cooperative Oncology Group Performance Status (ECOG‐PS; OR = 1.41, 95% CI = 0.42–4.75, p = 0.579), differentiation (OR = 1.05, 95% CI = 0.44–2.53, p = 0.909), human epidermal growth factor receptor 2 (HER2) status (OR = 0.88, 95% CI = 0.46–1.69, p = 0.710), or PD‐L1 status (OR = 1.04, 95% CI = 0.48–2.27, p = 0.914) in GC patients treated with ICIs (Table 4).

TABLE 4.

The association between NLR and clinicopathological factors in gastric cancer patients receiving ICIs.

Variables	No. of studies	No. of patients	Effects model	OR (95% CI)	p	Heterogeneity
						I 2 (%)	p value
Sex (male vs. female)	4	390	Random	0.99 (0.65–1.53)	0.973	85.1	<0.001
Age (years) (≥65 vs. <65)	4	390	Random	0.77 (0.32–1.85)	0.971	0	0.564
ECOG PS (≥2 vs. 0–1)	4	390	Random	1.41 (0.42–4.75)	0.579	60.1	0.057
Differentiation (poor vs. moderate or well)	3	332	Random	1.05 (0.44–2.53)	0.909	72.5	0.026
HER2 status (positive vs. negative)	2	195	Random	0.88 (0.46–1.69)	0.710	0	0.862
PD‐L1 status (positive vs. negative)	2	195	Random	1.04 (0.48–2.27)	0.914	0	0.577

3.6. Sensitivity analysis

Sensitivity analysis was conducted by removing each included study one at a time to assess the impact of each study on the overall results. The OS and PFS results demonstrated that the outcomes were credible and reliable (Figure 4).

FIGURE 4.

Sensitivity analysis for the pooled results in this meta‐analysis. (A) OS; (B) PFS.

3.7. Publication bias assessment

Begg's and Egger's tests were used to detect publication bias in this meta‐analysis. In Begg's and Egger's tests for OS, the p‐values were 0.858 and 0.543, respectively (Figure 5). For PFS, the results were as follows: Begg's test, p = 0.902 and Egger's test, p = 0.477 (Figure 5). These results indicate that there was no significant publication bias in this meta‐analysis.

FIGURE 5.

Publication bias test. (A) Begg's test for OS, p = 0.858; (B) Egger's test for OS, p = 0.543; (C) Begg's test for PFS, p = 0.902; and (D) Egger's test for PFS, p = 0.477.

4. DISCUSSION

The prognostic value of NLR for GC patients receiving treatment with ICIs has been controversial thus far. In the current meta‐analysis, we synthesized data from 10 eligible studies that included 830 patients. The pooled results indicated that an elevated NLR was associated with poor OS and PFS in GC patients using ICIs. In contrast, there was no significant association between NLR and clinicopathological factors in GC patients receiving ICIs. Based on the publication bias test, we concluded that our results are reliable. To the best of our knowledge, this is the first meta‐analysis investigating the prognostic significance of NLR in GC patients treated with ICIs. Our meta‐analysis has implications for the clinical management of GC patients using PD‐L1 antibodies. GC patients with a high pretreatment NLR (≥3) may experience progression or recurrence after receiving ICIs and should be closely monitored and followed‐up.

In the past decade, immunotherapy, including ICIs, has emerged as a breakthrough for GC patients. ²⁵ The ATTRACTION‐2 trial showed that nivolumab improved OS compared with placebo as a third‐ or later‐line treatment for GC. ⁵ Several anti‐PD‐1 antibodies also show promising activity as a first‐line treatment for GC patients. ²⁶ ATTRACTION‐04, a recent Asian phase II–III trial, found that nivolumab in combination with chemotherapy (CAPOX or S‐1 plus oxaliplatin) significantly improved PFS compared with nivolumab alone in HER2‐negative gastroesophageal junction (GEJ) cancer or GC patients, irrespective of PD‐L1 expression, whereas OS was not significantly improved. ²⁷ Prognostic markers are important to predict the efficacy of ICIs in GC patients. In this meta‐analysis, we report that a high NLR is a significant prognostic biomarker for GC patients receiving ICIs. The NLR is a parameter reflecting inflammatory responses that can promote tumor development and metastasis. In spite of the fact that the precise mechanisms have not yet been fully elucidated, the following factors contribute to its explanation. First, neutrophils are critical components of the tumor microenvironment (TME) ²⁸ that can directly target tumor cells and thereby drive tumor development. Neutrophils can secrete a series of cytokines and chemokines, including transforming growth factor beta (TGF‐β), vascular endothelial growth factor (VEGF), interleukin (IL)‐6, IL‐8, and IL‐12, which can enhance metastasis and angiogenesis in the TME. ²⁹ Second, lymphocytes play a pivotal role in cancer immune surveillance and in preventing the development of malignancy. Tumor‐infiltrating lymphocytes (TILs) secreting interferon‐γ can induce PD‐L1 expression and promote the progression of ovarian cancer. ³⁰ Low lymphocyte counts may result in insufficient immune responses, leading to poor prognosis for solid tumor survivors. ³¹ Therefore, a high NLR, which is the result of high neutrophil and/or low lymphocyte counts, is a reasonable and reliable prognostic marker for patients with cancer.

Notably, in this meta‐analysis, we found that a high NLR was associated with poor survival in GC patients undergoing treatment with ICIs. However, our results did not recommend immunotherapy, and ICIs are promising treatment methods for AGC patients. Our results should be interpreted with caution in the following respects: First, all GC patients were diagnosed with AGC, recurrent GC, or metastatic GC (Table 1). The prognosis of AGC patients is extremely poor. In the AGC patient group, median PFS and OS were reported as 4.9–6.0 months and 10.5–14.1 months, respectively. ³² The survival duration is relatively short, and there is limited room for improvement of survival using ICIs. Second, immunotherapy remains an effective therapeutic method for AGC patients. ICI treatment has demonstrated a highly variable objective response rate in GC patients, ranging from 10% to 26%. ³³ Third, in HER2 positive AGC patients, a double‐blind, placebo‐controlled Phase III KEYNOTE‐811 study is still ongoing to evaluate the efficacy and safety of pembrolizumab or placebo in combination with trastuzumab and chemotherapy as first‐line treatment. ³⁴ Moreover, in the included studies, ICIs were used as third‐line (or later) treatments for patients with unresectable, recurrent, or metastatic GC.

We investigated the prognostic role of NLR for OS and PFS according to cancer type, as shown in the subgroup analysis in Tables 2 and 3. Cancer type did not affect the prognostic role of NLR for OS (Table 2). A high NLR was associated with poor PFS in AGC patients (Table 3). However, the direct association between NLR and cancer type could not be analyzed based on the data of the included studies. Each study enrolled patients with one cancer type (AGC or metastatic/recurrent GC). Future studies are needed to explore the correlation between NLR and GC type.

Through meta‐analysis, many studies have indicated that NLR has prognostic value in several cancer patients receiving ICIs. ³⁵ , ³⁶ A meta‐analysis involving 27 studies showed that in patients treated with immunotherapy for advanced‐stage cancer, a high pretreatment NLR appears to be a promising prognostic biomarker. ³⁵ In a meta‐analysis of 1700 patients, researchers reported that elevated blood NLR was associated with shorter PFS and OS in non‐small‐cell lung cancer treated with PD‐1/PD‐L1 inhibitors. ³⁷ Chen et al. showed that an elevated NLR was associated with inferior OS and PFS in patients with renal cell carcinoma treated with ICIs, based on a meta‐analysis including 12 studies. ³⁸ In a recent meta‐analysis with 14 studies, Takenaka et al. demonstrated that a higher NLR was associated with poor OS, PFS, response, and disease control in patients with head and neck squamous cell carcinoma treated with ICIs. ³⁹ In a meta‐analysis encompassing 1164 patients, Zhang et al. reported that higher modified Glasgow Prognostic Score (mGPS)/GPS was associated with poor prognosis in patients with advanced cancer undergoing therapy with ICIs. ⁴⁰ In this meta‐analysis, we pooled data from 830 patients and found a significant prognostic value of NLR for GC patients undergoing treatment with ICIs. Our results are in line with those of previous studies on NLR and prognosis in other cancer types. Furthermore, we are the first to report these findings in patients with GC.

Our meta‐analysis has several limitations. First, all eligible studies were from Asia, and specifically from the following three countries: China, Japan, and South Korea. However, we did not restrict the nationality or region of the eligible studies, and only considered English publications. Therefore, our results may be applicable to Asian patients, and the prognostic role of NLR in non‐Asian GC patients treated with ICIs needs to be verified. Second, the results were pooled based on 10 eligible studies but not for each individual patient. Therefore, selection may exist. Third, the studies included in the analysis varied in their cut‐off values of NLR, ranging from 2.5 to 5. A standard cut‐off value for NLR is needed in future studies to avoid the inherent heterogeneity caused by inconsistent data in this meta‐analysis.

5. CONCLUSION

In summary, our meta‐analysis demonstrated that a high pretreatment NLR was significantly associated with poor OS and PFS in GC patients receiving ICIs. NLR is an effective biomarker for assessing the prognosis of GC patients receiving ICIs and provides valuable information for making treatment decisions in the era of immunotherapy for GC. Due to study limitations, prospective clinical trials including patients of diverse ethnicities are needed to validate our findings.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Supporting information

Data S1. The PRISMA checklist for this meta‐analysis.

Li L‐L, Pan L‐S. Prognostic value of neutrophil‐to‐lymphocyte ratio in gastric cancer patients treated with immune checkpoint inhibitors: A meta‐analysis. Kaohsiung J Med Sci. 2023;39(8):842–852. 10.1002/kjm2.12694