Our study demonstrates that there are significant biological differences in gastroesophageal cancer between Europe and LATAM. These differences are centered on cell-cycle and molecular drivers expressed by tumor cells (HER2, p53, and Ki67), phenotypic attributes (high differentiation grade, SRC content, and MUC6 expression), and microenvironment features (FOXP3 and PD-L1).

While the frequency of diffuse/poorly cohesive adenocarcinoma was very similar between the two cohorts, we observed a slight predominance of SRC content in tumors from LATAM. This high SRC content has been previously described in LATAM series in association with young age; however, studies on the prognostic effect of this finding remain contradictory [26, 27]. Additionally, a predominance of SRC carcinoma has been noted in the LATAM population in a study conducted in the United States, suggesting a potential role of genetic factors in the development of these tumors [27]. The same authors performed a specific analysis of the Hispanic population in the SEER (National Cancer Institute’s Surveillance, Epidemiology, and End Results) database obtaining similar results.

Consensus recommendations and guidelines for GEJ cancer management from Europe and North America estimate HER2 overexpression between 10 and 20%, and 12% and 23%, respectively [28, 29]. HER2 was amplified in 17% of cases in the TCGA (The Cancer Genome Atlas) Research Network’s molecular classification of GEJ cancer [30]. Remarkably, our findings reveal a significantly lower rate of HER2 positivity in LATAM (4%), underscoring a notable regional difference that could have profound implications for targeted therapies and patient outcomes. In Costa Rica and Mexico a similar low prevalence of HER2 overexpression was described [26, 31, 32], whereas in Chile the prevalence was found to be higher [33, 34]. One possible explanation for the difference in HER2 frequency observed in both settings may be the deviation in the distribution of molecular subtypes in LATAM compared to the TCGA classification report with significant enrichment in GS cases and a relatively low frequency of CIN cases in LATAM. TP53 mutations are enriched in the chromosomal instability (CIN) molecular subtype, which concentrates most HER2-amplified cases (71% of CIN cases were TP53 mutated, according to TCGA) [30]. Additionally, the molecular classification from the Asian Cancer Research Group (ACRG) indicates that copy number variations (CNVs) in various genes, including HER2, are frequently encountered in the MSS/TP53 inactive group, where TP53 mutations are more prevalent [34]. In our series, TP53 mutations were significantly correlated with other features characteristic of the CIN subtype (intestinal histology, HER2 amplification) and the frequency of TP53 mutations in Latin America was slightly lower (40%) compared to Europe (49%). The previously discussed enrichment of SRC in Latin countries may also be a consequence of the overrepresentation of the GS molecular subtype in this population. Additionally, we observed a significant increase in SRC levels in TP53 wild-type cases and in those displaying a p53 normal immunohistochemical pattern, as previously noted in a Chilean cohort [33].

Notably, the immunohistochemical analysis of p53 revealed an aberrant pattern of expression in nearly half (46%) of the European cases, a prominent feature compared to LATAM cases (26%). This dichotomous classification of p53 immunohistochemical expression showed good specificity for predicting the TP53 mutational status (87%), as most TP53 wild-type cases were accurately classified as having a normal pattern of protein expression. However, the sensitivity of this approach was modest (65%), with TP53 mutational status being adequately predicted in only two-thirds of the cases classified as having a p53 aberrant pattern by immunohistochemistry. One limitation of this method is the intratumoral heterogeneity of the protein, which has been demonstrated in gastric and GEJ adenocarcinoma and may explain the decreased sensitivity when applied to small biopsies [36]. Despite this limitation, in our series, p53 immunohistochemical aberrant expression was associated with characteristics typical of CIN tumors (intestinal histology and HER2 overexpression), supporting the utility of p53 immunohistochemistry in identifying this more aggressive subtype [36, 37]. Furthermore, the proliferative Ki-67 index, another differential characteristic between European and LATAM tumors, was consistently associated with both the presence of TP53 mutations and p53 aberrant pattern of protein expression, aligning with our hypothesis.

An interesting finding in our study is the differential expression profile of MUC6 between Europe and LATAM. MUC6 is a gel-forming mucin constitutively expressed in the stomach, pancreas, duodenum, and female reproductive tract [38]. In gastric mucosa, MUC6 has been identified as an oncogenic driver, mutated in around 20% of all cases [35, 39], and its downregulation associated with tumor progression [40]. Notably, we found a significant positive relationship between PD-L1 and MUC6 expression, irrespective of the CPS cut-off point. It has been suggested that mucins may play a role in immune regulation in solid tumors through various mechanisms, including direct interaction with dendritic cells, macrophages, and natural killer cells; interaction with Toll-like receptors; the generation of neoantigens; or their transcriptional regulation by pro-inflammatory cytokines [41,42,43]. We identified a significant association between MUC6 expression and MMRd, a relationship previously reported in colorectal carcinoma [44]. In our series, the strong correlation between PD-L1 and MUC6 was primarily driven by the immune-active MSI group. However, within the non-immune GS/CIN subtypes, a subtle correlation between these markers persisted, highlighting a potentially nuanced interplay that warrants further investigation. This is especially important in the era of immunotherapy for gastric cancer, where significant scientific effort is focused on understanding the mechanisms regulating the immune response and refining biomarkers for patient selection. Additionally, PD-L1 expression was significantly higher in Europe, and a low prevalence of PD-L1 expression was previously observed in an exploratory analysis conducted in Chile [34].

Our analysis identified significant distinctions between immune-related tumors (EBV/MSI) and non-immune tumors (GS/CIN) across multiple dimensions, including morphology, PD-L1 expression, proliferative activity, SRC component, and the presence of CD8-positive immune cells. These differences may hold functional relevance. Nevertheless, despite these findings, the distribution of immune and non-immune subtypes was comparable between LATAM and European cohorts. This similarity suggests that other intrinsic factors may be responsible for the observed regional differences.

Finally, our differential analysis of H. pylori prevalence and abundance revealed no differences between LATAM and Europe. Therefore, at least in our study, none of the observed regional differences can be attributed to H. pylori infection.

We conducted a detailed morphological and protein expression analysis using IHC and ISH to explore intrinsic differences between Latin American and European gastric cancer (GC) cohorts. To achieve this, we utilized a molecular surrogate classification based on morphology and IHC, a method that has demonstrated both clinical value and feasibility [20]. However, it is important to recognize the limitations of this approach. The surrogate classification may not fully align with the more intricate TCGA molecular classification, which integrates multi-omics data for a comprehensive analysis. This highlights the need for further studies to refine and validate surrogate methods in diverse populations.

Another limitation of our study is that all analyses were conducted on endoscopically obtained biopsies, which reflect only the superficial portion of the tumor. This limitation is particularly relevant in GEJ cancer, where tumor heterogeneity is a prominent characteristic. Additionally, our study lacks data on Claudin-18.2 and FGFR2b expression—key targets in emerging antibody–drug therapies—further constraining its scope. Finally, it is important to consider that for PD-L1 analysis we have utilized the 22C3 antibody–based LDT on the Ventana BenchMark platform. While this may be considered a limitation, published harmonization studies have shown comparable results when compared with the regulatory-approved, PD-L1 IHC 22C3 pharmDx test, supporting the validity of our approach [45].

In summary, the LEGACy study underscored significant biological distinctions between European and LATAM patients with gastric and GEJ cancer:

1.

Distribution of molecular subtypes: our results suggest a deviation from TCGA results, notably showing a decreased proportion of CIN and HER2-positive cases, and an overrepresentation of GS tumors, in LATAM. The increased SRC content observed in LATAM cases may align with this trend.

2.

TP53 mutations: While the immunohistochemical approach for predicting TP53 mutations has limitations, we did find a significant correlation between p53 aberrant expression and other CIN features: intestinal histology and HER2 overexpression.

3.

Correlation between MUC6 and PD-L1: Both MUC6 and PD-L1 overexpression was seen more frequently in European cases. Across both cohorts, MMRd cases showed high expression levels of MUC6 and PD-L1. MMRp cases displayed a weaker but still significant association between these two markers. This lends support the hypothesis of MUC6 acting as an immune regulator, which suggests avenues for further investigation.

4.

H. pylori infection relevance: differences in H. pylori distribution or abundance were not observed in the comparison between European and LATAM cases; hence the observed differential features cannot be attributed to this infection.

Altogether, the findings of this study underscore a need for region-specific approaches in gastric and GEJ cancer treatment. In LATAM, where tumors exhibit lower HER2 and PD-L1 expression, current treatment strategies may be less effective. Further research tailored to the unique biological profiles in different regions is crucial for addressing these disparities effectively.

Centers

ANAX: Anaxomics SL (Spain).

ECPC: European Cancer Patient Coalition (Belgium).

GENPAT: Genpat SL (Paraguay).

IAF: Institute Alexandre Fleming (Argentina).

INCAN: National Cancer Institute (Mexico).

INCLIVA: Biomedical Research Center INCLIVA (Spain).

IPATIMUP: Institute of Molecular Pathology and Immunology of the University of Porto (Portugal).

PUC: Pontificia Universidad Católica de Chile (Chile).

ULEI: University of Leipzig (Germany).

VHIO: Valld’HebrónInstitut of Oncology (Spain).

VUMC: Amsterdam UMC location VU University Medical Center (Netherlands).