This systematic review and meta-analysis was conducted in strict accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (17). The study protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD420251233582. All procedures were carried out in accordance with established methodological standards to ensure the rigor of the research process and the transparency of the reporting.

We performed a systematic and comprehensive search to identify both published and unpublished literature. Multiple databases were searched from their inception to October 20, 2025, including PubMed, MEDLINE, the Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science, and Embase. To minimize publication bias and identify ongoing studies, our search also encompassed preprint platforms, the ClinicalTrials.gov registry, ahead-of-print publications, and other grey literature (e.g., unpublished research reports and conference abstracts). Furthermore, we conducted backward reference checking of relevant studies and previous systematic reviews, and consulted experts in the field to ensure the breadth and completeness of the search. No restrictions were imposed on language, region, or country to minimize potential selection bias. Three independent reviewers (SL, LS, and XD) performed the literature screening in a blinded manner and resolved discrepancies through discussion. Any unresolved disagreements were adjudicated by a fourth reviewer (KD).Studies were considered eligible if the title, abstract, subject headings, keywords, or unique identifiers included one or more MeSH terms or keywords related to gastric cancer, lymph node ratio, or neoadjuvant treatment. These terms included: “Gastric Neoplasms,” “Stomach Neoplasms,” “gastric cancer,” “lymph node ratio,” “LNR,” “positive lymph node ratio,” “metastatic lymph nodes,” “lymph node metastasis,” “Neoadjuvant Therapy,” “neoadjuvant chemotherapy,” “preoperative chemotherapy,” and “perioperative chemotherapy.”

This systematic review and meta-analysis was guided by the PICOS framework. The population consisted of patients with histologically confirmed gastric cancer who underwent neoadjuvant chemotherapy—with or without immunotherapy—followed by curative gastrectomy and lymphadenectomy. The exposure of interest was the lymph node ratio (LNR), defined as the ratio of metastatic to examined lymph nodes, assessed either as a categorical or continuous variable. Comparisons included high versus low LNR groups or alternative LNR stratifications based on cut-off values determined in individual studies. The primary outcome was overall survival, while secondary outcome was disease-free survival. Eligible study designs were retrospective cohort studies that reported hazard ratios (HRs) and 95% confidence intervals evaluating the association between LNR and survival outcomes. Randomized clinical trials were not required due to the nature of the research question.

Three reviewers (SL, QH, and XD) independently screened the titles and abstracts retrieved from the systematic search. They then assessed the full texts of studies considered potentially eligible. Any discrepancies during the selection process were resolved through discussion to minimize bias. If consensus could not be reached, a fourth reviewer (KD) made the final determination.

Three reviewers (SL, QH, and XD) independently extracted data from each included study using a predefined standardized data extraction form. The extracted information included key study characteristics (first author, year of publication, country), patient demographics and clinical details (sample size, age, sex distribution, follow-up duration, and LNR cut-off value), as well as essential information related to neoadjuvant treatment, such as the type of neoadjuvant chemotherapy or immunochemotherapy administered.

In addition, data related to lymph node assessment were collected, including the definitions of lymph node ratio (LNR) used in each study and the corresponding LNR cut-off values. Hazard ratios (HRs) and 95% confidence intervals (CIs) for overall survival, disease-free survival, or other reported survival outcomes were extracted, with priority given to multivariable-adjusted estimates when available. When overall survival was not reported, disease-specific survival was used as an acceptable alternative outcome (17).

A fourth reviewer (KD) independently verified all extracted data to ensure completeness, consistency, and accuracy. Any discrepancies were resolved through discussion until consensus was achieved.

The methodological quality of the included cohort studies was independently assessed by three reviewers (SL, QH, and XD) using the Newcastle–Ottawa Scale (NOS) (18). This tool evaluates study quality across three major domains: (1) Selection of study participants, (2) Comparability of cohorts, and (3) Assessment of outcomes. Each study could receive up to a maximum of nine points, with higher scores indicating better methodological quality.

Studies scoring 7–9 points were considered high quality, those scoring 5–6 points were classified as moderate quality, and those scoring below 5 points were deemed low quality. Any discrepancies among reviewers were resolved through discussion, and if consensus could not be reached, a fourth reviewer (KD) adjudicated the final decision. This structured evaluation ensured consistent and rigorous assessment of methodological risk across all included studies.

Statistical analyses were performed using Review Manager (version 5.4.1) and Stata (version 17.0). For overall survival and disease-free survival, pooled hazard ratios (HRs) with 95% CIs were calculated using the generic inverse-variance method (19). Heterogeneity was evaluated using Cochran’s Q test (P < 0.10 indicating significance) and quantified with I² statistics (20). A random-effects model was applied if significant heterogeneity was observed (P < 0.10 for Q test OR I² ≥ 50%); otherwise, a fixed-effect model was used (21).

To better understand the potential sources of heterogeneity in the prognostic effect of lymph node ratio after neoadjuvant therapy, subgroup analyses of the primary outcome were conducted based on the following criteria: LNR cutoff (≥0.30 vs <0.30) and type of neoadjuvant therapy (chemotherapy vs chemo–immunotherapy).The threshold of 0.30 was selected primarily because it was the most frequently employed cutoff across the included studies (10, 22, 23), allowing for the most robust cross-study comparisons, and because it aligns with a value often identified in prior gastric cancer prognosis literature as a critical inflection point for survival outcomes (22).

robustness of the pooled OS effect was assessed using leave-one-out sensitivity analyses, repeating the meta-analysis after sequential exclusion of each study (Table 1).

Because only seven studies were included, formal assessment of publication bias was not performed. A two-sided p-value < 0.05 was considered statistically significant. However, we acknowledge that the absence of unpublished negative or null-result studies could potentially lead to an overestimation of the pooled effect size. To mitigate this risk, our search strategy explicitly included preprint databases and trial registries.

In the initial search, a total of 1,081 records were identified across all databases. After removing 267 duplicates, 640 unique studies underwent title and abstract screening, of which 573 were excluded. Sixty-seven full-text articles were assessed for eligibility, and 39 were retrieved in full. Following exclusion of studies without neoadjuvant therapy, those lacking lymph node ratio data, unavailable full texts, protocols, or duplicate datasets, seven retrospective cohort studies met the inclusion criteria and were included in the final meta-analysis (Figure 1). The seven studies (10, 12, 22–26), published between 2019 and 2024, comprised 2,437 patients with histologically confirmed gastric cancer who underwent neoadjuvant chemotherapy followed by curative gastrectomy. Sample sizes ranged from 95 to 1,551 patients, and the mean or median age ranged from 57 to 65 years. All studies focused on locally advanced gastric adenocarcinoma, with follow-up durations spanning 19 to 42 months. LNR cut-off values varied across studies, including thresholds of 0.10, 0.15, 0.25, 0.26, 0.30, 0.33, 0.40, and 0.50, with 0.30 being the most frequently used. The detailed characteristics of included studies are presented in Table 2. Quality assessment using the Newcastle–Ottawa Scale rated seven studies as high quality (scores 7–9), as shown in Table 3.

Six studies (10, 12, 23–26) reported overall survival (OS), while one study (22) provided disease-specific survival (DSS) as an alternative outcome. The pooled analysis demonstrated that patients with a high lymph node ratio experienced significantly poorer overall survival compared with those with a low lymph node ratio, with a combined hazard ratio of 3.48 (95% CI 2.27–5.33) (Figure 2). Three studies (12, 23, 26) reported disease-free survival, and high LNR was consistently associated with inferior DFS, yielding a pooled hazard ratio of 1.94 (95% CI 1.24–3.06) (Figure 3).

Subgroup analysis based on the commonly used LNR threshold of 0.30 revealed a statistically significant interaction between studies using cut-offs ≥0.30 and those using cut-offs <0.30 (p for interaction = 0.03) (Figure 4). In contrast, subgroup analysis stratified by neoadjuvant treatment modality demonstrated no statistically significant interaction between chemotherapy-only regimens and chemo–immunotherapy regimens (p for interaction = 0.49) (Figure 5).

Sensitivity analyses confirmed the robustness of these findings, with leave-one-out pooled hazard ratios ranging from 2.91 to 4.05, all remaining statistically significant, and with consistent results across fixed- and random-effects models (Table 1).