The meta-analysis included three Chinese studies published between 2015 and 2023, comprising 764 gastric cancer cases and 554 controls. Their pooled estimates did not show statistically significant associations between CISD2 protein expression and age, sex, lymphatic invasion, TNM stage or histological grade. The notable clinicopathological signal was an inverse association with invasion depth: the pooled odds ratio was 0.57, with a 95% confidence interval of 0.41 to 0.79 and a P value of 0.001. In parallel, an independent in silico analysis of the Kaplan–Meier Plotter database associated elevated CISD2 RNA with shorter first progression and overall survival, as well as poorer survival in a HER2-negative subgroup.

CISD2 is a gene involved in regulating cellular responses to oxidative stress and maintaining iron homeostasis, and abnormal expression has been reported in multiple cancers. Its role in gastric cancer has remained uncertain because previous studies have used differing detection methods, scoring systems and thresholds for classifying expression. The researchers therefore limited their meta-analysis to studies that measured CISD2 protein by immunohistochemistry and combined staining extent with staining intensity to assess expression. To identify eligible evidence, they searched seven databases—Wan Fang, VIP, China National Knowledge Infrastructure, SinoMed, Web of Science, the Cochrane Library and PubMed—from the inception of each database through July 15, 2026.

The search returned 708 records: 202 from PubMed, 200 from Web of Science, 26 from SinoMed, 128 from Wan Fang, 20 from VIP, 82 from the Cochrane Library and 50 from China National Knowledge Infrastructure. Reference lists were also checked, but no additional studies were added. Two researchers independently screened the literature, extracted data and assessed study quality, resolving disagreements through discussion or consultation of a third researcher. Eligible studies needed full-text data and had to associate immunohistochemically measured CISD2 expression with at least one predefined feature: age, sex, invasion depth, lymphatic invasion, TNM stage or histological grade. All three included studies received a Newcastle–Ottawa Scale score of 8 out of 9. The protocol was registered in PROSPERO as CRD42023490404.

For the pooled protein-expression analysis, the researchers used odds ratios with 95% confidence intervals and selected a fixed-effect model when statistical heterogeneity was low and a random-effects model when it was higher. No significant association emerged for age, for which the odds ratio was 1.37 and the 95% confidence interval was 0.77 to 2.43, or for sex, for which the odds ratio was 1.02 and the confidence interval was 0.71 to 1.47. The estimates for lymphatic invasion, TNM stage and histological grade also failed to reach statistical significance. The invasion-depth comparison produced a pooled odds ratio of 0.57, a 95% confidence interval of 0.41 to 0.79 and a P value of 0.001; heterogeneity was comparatively low at an I-squared value of 48.5%.

The direction of the invasion-depth finding is not uniform across the three underlying studies, however. Two reported that elevated CISD2 expression was associated with more advanced invasive or metastatic features, whereas the third found no significant correlation between CISD2 expression and infiltration depth. Differences in immunohistochemical scoring thresholds may have affected classification of high and low expression, and the studies used two different antibodies that may differ in binding characteristics and detection sensitivity. The composition of T-stage subgroups also differed among the patient populations. Subgroup analyses did not identify variables that substantially explained the observed heterogeneity, while omission of any single study did not reverse the pooled estimates. The authors cautioned that this sensitivity analysis could not establish the reliability of estimates based on only three datasets.

Publication bias could not be assessed conventionally because the meta-analysis included only three studies. Funnel plots were supplied for illustration but were not interpreted statistically, and the small number of studies also limited the statistical power of subgroup and heterogeneity analyses. Heterogeneity was particularly high for the TNM-stage comparison, reaching an I-squared value of 92%, whereas the invasion-depth analysis had an I-squared value of 48.5%. The pooled estimates should therefore be interpreted with caution, particularly because the entire evidence base came from Chinese populations. Geographic and ethnic differences in gastric cancer cohorts may limit how readily the findings can be generalized elsewhere, and the imprecision of confidence intervals may make individual pooled effects appear more conclusive than the underlying data justify.

The separate prognostic analysis used the Kaplan–Meier Plotter database and measured CISD2 transcript abundance rather than protein expression. The investigators selected probe 244275_at and used the platform’s automatic optimized algorithm to divide patients into high- and low-expression groups. The overall-survival analysis included 631 gastric cancer patients, with a median follow-up of 40.7 months, and found that elevated CISD2 messenger RNA was associated with shorter overall survival, with a hazard ratio of 1.27 and a P value of 0.041. A first-progression analysis included 522 patients and produced a similar association, with a hazard ratio of 1.3 and a P value of 0.04. In a HER2-negative subgroup of 429 patients, the hazard ratio was 1.35 and the P value was 0.034.

These database findings do not directly confirm the protein-level meta-analysis because the biological measurement and clinical endpoints differ. The immunohistochemical studies assessed protein abundance in relation to tumor characteristics, whereas the Kaplan–Meier Plotter analysis examined RNA abundance in relation to survival. Taken together, the two datasets provide complementary but separate evidence: pooled immunostaining data suggested a relationship with invasion depth, and transcript-level data associated higher expression with adverse outcomes. The investigators also cited previous evidence suggesting that CISD2’s behavior may depend on context, including an association with risk in untreated gastric cancer and with favorable response to 5-fluorouracil chemotherapy. The present study did not test a treatment-selection strategy or establish a clinical role for the gene.

Several limitations constrain the conclusions. The meta-analysis depended on just three eligible studies, all conducted in China, which limited the stability and precision of pooled estimates and their generalizability to other populations. The survival analysis was also exploratory, used an automatically optimized expression cutoff and covered three separate endpoints without a formal statistical adjustment for multiple comparisons. Differences in source cohorts and the separation between RNA and protein measurements further limit direct comparison. The authors concluded that CISD2 may be a candidate marker associated with gastric cancer progression and prognosis, but large independent cohorts, prospective clinical studies and mechanistic investigations are needed to verify the associations before CISD2 can be considered a validated prognostic biomarker or applied in clinical care.

Subject of Research: Exploratory evidence on CISD2 expression, clinicopathological features and prognosis in gastric cancer

Article Title: Association between CISD2 expression and clinicopathological features and prognosis in gastric cancer: a systematic review, meta-analysis, and database validation

Article References: Liang, T., Cui, Y., & Song, H. (2026). Association between CISD2 expression and clinicopathological features and prognosis in gastric cancer: a systematic review, meta-analysis, and database validation. Frontiers in Oncology, 16, Article 1965844. https://doi.org/10.3389/fonc.2026.1965844

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1965844

Keywords: CISD2, gastric cancer, meta-analysis, systematic review, immunohistochemistry, gene expression, tumor invasion, survival prognosis, HER2-negative gastric cancer, bioinformatics, Association, between