Analysis of biopsies of gastric cancer, intestinal and diffuse, and non-atrophic gastritis: an overview of loss of heterozygosity in Mexican patients

Samples

Tissue samples were obtained from 21 patients (five females and 16 males) that met the criteria for diffuse gastric cancer (DGC, n = 7) and intestinal gastric cancer (IGC, n = 7) diagnoses, and subjects with non-atrophic gastritis (NAG, n = 7) as controls. Histological assessment of the biopsies was performed by two trained pathologists independently. They assigned the phenotypic diagnosis of diffuse or intestinal tumors and non-atrophic gastritis (NAG) samples. Only samples with the same diagnosis (‘identical results’) by two independent expert pathologists were included in the analysis (Table 1).

To guide the investigation of relevant alterations, the present analysis focused on LOH events present in at least three patients (cut-off, ≥3 patients; ≥40% samples) to identify the most relevant GC alterations. The DNA extraction, DNA quality assessment, and high-density whole-genome processing microarray analysis were done according to Larios-Serrato et al. (2022).

LOH processing

The raw intensity files (.CEL) retrieved from the commercial platform Affymetrix^® CytoScan™ microarray (Affymetrix; Thermo Fisher Scientific, Inc.) were analyzed using Chromosome Analysis Suite v4.3.0.71 (ChAS, https://www.thermofisher.com/). The construction of the GRCh38 genome (December 2013) was used as a reference model and CytoScanHD_Array.na36.annot.db file for annotation. Data processing was based on the segmentation algorithm, where the Log2 ratio for each marker was calculated relative to the reference signal profile. To calculate the LOH, the data were normalized to baseline reference intensities using the ChAS reference model, including 284 HapMap samples and 96 healthy individuals. The Hidden Markov Model (HMM) was used to determine the LOH segment calls. The customized conditions were filtered to determine LOH, 3 Mb, and 50 single nucleotide polymorphisms (SNPs). The median absolute pairwise difference (MAPD) and the single nucleotide polymorphism quality control (SNPQC) score were used as the quality control parameters. Only samples with MAPD < 0.25 and SNPQC > 15 values were included in the further analysis.

Bioinformatics analysis

To generate a list of genes and frequencies for altered regions, Practical Extraction and Report Language (Perl) scripts (Wall, Christiansen & Orwant, 2000) were developed to load the LOH segment data files generated by ChAS v4.3.0.71 for each sample, including chromosomes, cytogenetic bands, Online Mendelian Inheritance in Man information, and haploinsufficiency predictions version 3 information from the DatabasE of Genomic Variation and Phenotype in Humans, using Ensembl Resources (DECIPHER v11.25). Custom scripts were developed in Perl v5.32 to obtain the frequency of LOH genes and cytobands, and the length of events.

The genes altered in three patients (cut-off, ≥ 3) with DGC, IGC, and NAG were included for analysis and visualization. The genes were compared by generating Venn diagrams with the Jvenn server (Bardou et al., 2014). Cancer Hallmarks enrichment analysis (p.adjust < 0.05) was performed with a collection of 6,763 genes (Menyhart, Kothalawala & Győrffy, 2024), available on the server http://www.cancerhallmarks.com; this database establishes the relationship of genes and Hallmarks through the collection of different publications. The Cancer Hallmarks server uses a hypergeometric test for enrichment using the greasy Python package. The results were reviewed using the Catalog of Somatic Mutations in Cancer database (COSMIC v100) (Sondka et al., 2024) and the Hallmarks of Cancer database (HOCdb database). Reactome v88 performed a metabolic pathway enrichment analysis (Fabregat et al., 2018), considering those results significant with values less than 0.05 in the false discovery rate (FDR).

Finally, an interaction network was generated based on metabolic pathways, genetics, and physical and functional associations to establish the cancer Hallmarks associated with the profile of LOH-genes IGC, NAG, and “core” IGC-DGC-NAG. Furthermore, these were determined using the STRING v12.0 prediction server (Szklarczyk et al., 2021) and Cytoscape v.3.10.0 (Shannon et al., 2003), including manual annotation of their corresponding cancer Hallmarks (adhesion, angiogenesis, inflammation, migration, metastasis, morphogenesis, proliferation, and survival).

Sample characteristics

This study included tissue samples from 21 Mexican patients without treatment (naïve). Patient samples included seven DGC cases, seven for IGC, and seven more corresponding to NAG (as controls). The .CEL files and their raw intensity values obtained from the microarrays were deposited in the Center for Biotechnology Information (NCBI), with the accession key GSE117093 and BioProjet PRJNA481039.

Table 1 shows the general characteristics of the 21 patient samples, age (mean ± SD, 59.61 ± 15.94 years), sex (female 23.8% and male 76.2%), and the percentage of neoplastic cells for tumor tissues ranging between 50 and 70%. One IGC patient and three with NAG were positive for H. pylori. Data from the tumor size, number of nodes, and metastasis (TNM) classification system are presented.

Genomic detection of LOH

The LOH of the patients was estimated using the analysis described before, a meticulous process based on regions where the preponderance of SNPs does not display heterozygosity. Table 2 shows a summary of the chromosomes with the highest involvement frequency concerning the number of events (coincidences) at the LOH-regions, but not strictly perfect in the chromosomal coordinates.

Our data, which include the megabase pairs cumulative length (Mbp-cl) of our tissue samples, were also reviewed (Table S1). The LOH-gene frequency data, chromosomes, and cytobands are presented. Table S2 displays the accumulated LOH-length (Mb) values per chromosome to determine if more extended losses indicate more damage.

In DGC patients, the affected chromosomes with Mbp-cl and the specified number of LOH-events were 6, 8, 16, and X; at IGC, they were chromosomes 3, 16, and 17. Chromosomes 6 and 8 are associated with DGC, while 3 and 17 are associated with IGC (Table 2 and Table S3). Following, we found that there are 3,361 LOH-genes in DGC (Table S1); chromosomes Xq11.1/Xp22.23 in 7/7 male patients and chromosome 16p11.2 in 6/7 male patients (Table 2 and Table S3) were the most altered.

In IGC, 2,490 LOH-genes were determined (Table S1) with chromosomes Xq11.1 for 7/7 patients, Xp22.33 for 6/7 patients, 16p11.2 for 6/7 patients, 3p21.31 for 5/7 patients and 17q22 for 3/7 patients (Tables 2 and S3) as the most altered.

Finally, in NAG, 4,748 LOH genes were determined (Table S1). Chromosomes Xq11.1 for 7/7 patients, 16p11.2 for 7/7 patients, and Xp22.33 for 4/7 patients (Tables 2 and S3) were the most altered.

Interestingly, LOH lengths do not seem relevant to carcinogenesis, and 5–10 Kbp LOH lengths were more common and frequent in DGC, IGC, and NAG (Table S4).

We analyzed alterations occurring in at least three patients to identify the most relevant LOH in GC and NAG (cut-off ≥ 3). We found a similar pattern for total LOH, with events in DGC (1157), IGC (1361), and NAG (1184). In addition, DGC had the highest number of genes affected in all samples, 7/7 (1132), followed by IGC (261) and NAG (34) (Table S1).

Gastric cancer genes associated with LOH

A Venn diagram was constructed to examine the LOH-GC-relevant genes of at least three patients (cut-off ≥ 3) of the DGC, IGC, and NAG. We determined 1,153 shared LOH genes between DGC-IGC-NAG. IGC had 207 unique affected genes, while NAG only showed 28 (Fig. 1A and Table S5). From each subset (Fig. 1A), those genes with matches according to the Cancer Hallmarks Genes database, a comprehensive resource that includes 6,763 genes, are shown 241 LOH-genes were found in DGC-IGC-NAG; IGC had 55 affected unique genes and 13 genes in NAG. Figures 1B–1D represent the enrichment of the Hallmarks of Cancer. The 241 common genes DGC-IGC-NAG present more Hallmarks than NAG, which showed fewer.

Functional pathway analysis

LOH-genes-Hallmarks were used to identify metabolic pathways in each subset according to Reactome and Homo sapiens as a model organism (Table 3); those significant metabolic pathways (p-value < 0.05) were selected as the most affected.

Correlation genes network

Cancer LOH-genes-Hallmarks associated with metabolic pathways were used to construct interaction networks (STRING, Fig. 2). The connecting lines indicate associations by metabolic pathways, expression, localization, inferred interaction, genetic interactions, data mining, and neighborhood. Likewise, each node can be related to flags (events) reporting the Hallmarks of Cancer. In this way, the network formed among IGC-DGC-NAG had 31 nodes; the genes with the highest number of Hallmarks were PAK3, IRAK1, and IKBKG, and the genes with the highest number of connections were OPHN1, WAS, TKTL and PRPS1. In the ICG network, there were 29 nodes. The genes associated with cancer Hallmarks were GNAI2, RHOA, MAPKAPK3, HYAL1, and CISH, while the most connected were RHOA, MST1R, and ATRIP. Finally, the network corresponding to NAG had five nodes, PTPRJ had the most significant number of Hallmarks, and the most connected was NUP160 (Table 4).

Design study

Our study’s choice of seven patients per group is based on rigorous selection criteria. Although our repository has approximately 600 GC samples, only those with at least 50% cancer cells (see Table 1) were included to reduce the “noise” from non-cancerous cells, which is essential for accurate results. In molecular biology and genetic studies, it is expected to face logistical and financial constraints that limit the collection of large sample sizes. Additionally, in highly controlled experiments, a smaller sample size can be sufficient to detect significant effects. Despite a limited “n”, wide confidence intervals help interpret the precision of our results. Furthermore, previous studies in the literature have used similar sample sizes and support our methodology. Finally, as an exploratory study, this research provides a foundation for larger-scale studies to generate valuable hypotheses for future experiments (Larios-Serrato et al., 2022).

Chronic gastritis is associated with inflammatory and cellular changes that can lead to GC. Notably, H. pylori infection can cause atrophic chronic gastritis, a precursor to gastric cancer (Kumar et al., 2020). Comparing gastritis with cancer samples can help explore the inflammation-to-cancer transition and the molecular changes involved in the progression of premalignant lesions to malignancies (Sui et al., 2020). Correa’s model describes gastric epithelium progression from chronic gastritis to intestinal metaplasia, dysplasia, and eventually adenocarcinoma (Correa & Piazuelo, 2012). It is relevant to use gastritis as a comparison group, since it represents an early phase within this cancer evolution model. Given the difficulty in collecting healthy gastric tissue due to ethical concerns with biopsies in asymptomatic individuals, gastritis is a clinically relevant control. Functionally, comparing inflamed (gastritis) tissues with cancerous ones may highlight genes or pathways that distinguish non-neoplastic inflammation from tumor-progressing inflammation, providing critical insights into tumorigenic transformation mechanisms.

LOH, chromosomes, and cytogenetics

GC has a high mortality rate due to its characteristics, such as late detection and silent progression; therefore, research into tumor biology is required to find early markers and carry out an opportune intervention. LOH is among the aspects that could contribute to a timely diagnosis according to prognosis (Huo et al., 2021; Battista et al., 2021). It has been proposed that LOH could function as independent prognostic markers (Koo et al., 2004), and those could even function as alternative targets for treatment (Hwang et al., 2021).

LOH is involved in different cancer types, showing its importance as a “predisposing” factor. According to other studies, specific LOH genes are more relevant than the length of affection. LOH-genes-NAG, TP53, PTEN, RUNX3 (Bellini et al., 2012; Li et al., 2005; Carvalho et al., 2005) could have a carcinogenic potential as signaling early events; of these, TP53 (Bellini et al., 2012) and PTEN (Li et al., 2005) have polymorphisms or copy number variations due to LOH-events (Battista et al., 2021; Poremba et al., 1995; Oki et al., 2005).

Here, 11 LOH-genes were determined, besides the small sample which were selected according to their apparition frequency in the analyzed samples, their participation in metabolic pathways (p-value < 0.05), their established interactions (networks), and their enrichment in Cancer Hallmark’s genes database (p.adjust < 0.05). Thus, PTPRJ and NUP160 were determined into NAG samples, RHOA, GNAI2, and MAPKAPK3 for ICG, and no unique or relevant genes were identified for DCG. NAG LOH-genes that are relevant to carcinogenesis participate in proliferation and growth, while those for IGC are on genomic instability, tissue invasion, metastasis, and the arrest of cell death; and DGC genes are for energy metabolism, destruction of immune evasion, and replicative immortality. Other genes were shared between IGC and NAG-IGC-DGC, whose p-values are close and could be considered similar LOH events, since they are involved in sustained angiogenesis. On the other hand, IGC genes also promote inflammation, and although the p-values are not significant, there was a difference in the NAG-IGC-DGC group. Then, those molecular, cellular, and metabolic LOH alterations should be monitored in GC patients. These findings must be validated to develop tests with molecular profiles for diagnosis, prognosis, and response to treatment, as well as, most importantly, screening tests.

When analyzing the metabolic pathways associated with LOH genes, the common ones were signal transduction, immune system, cell cycle, and extracellular matrix organization, which would be involved in early GC stages. Developmental biology is added to IGC samples because of the semaphorins interactions (Table 3, genes PLXNB1 and RHOA), which have been found to function as tumor suppressors and inhibit tumor progression by various mechanisms; however, they also can function as inducers and promoters of tumor progression (Neufeld et al., 2016). Another metabolic pathway found here (IGC) was protein and carbohydrate metabolism, where processes such as post-translational modifications (phosphorylations) and hyaluronan metabolism (degradation proteins Hyal 1–3) are affected. Hyaluron plays a fundamental role in tissue architecture and the regulation of cellular function, which could be related to proliferation and migration, since hyaluronan accumulation at the extracellular matrix and its derived fragments due to the altered expression of hyaluronidases enhance cancer progression, remodeling the tumor microenvironment (Kobayashi, Chanmee & Itano, 2020), just like in GC (Vizoso et al., 2004).

Marker genes

The results described above determined five shared genes among the three sample groups analyzed (IGC-DGC-NAG): IRAK1, IKBKG, PAK3, TKTL1, and PRPS1. These genes are associated with various typical functions of cancer development and progression. IRAK1 is involved in the interleukin 1 receptor, then regulates inflammation genes in immune cells. It is one of the most emerging Hallmarks of Cancer (because of avoiding immune destruction or tumor inflammation), affecting the disease development due to its possible oncogenic and immunological functions (Liu et al., 2022). IKBKG is linked to the NF-kB pathway, which is crucial for cancer since its involvement in cell survival and proliferation (Yin, Wang & Lu, 2020; Gong et al., 2020) is another of the principal Cancer of Hallmarks. PAK3 can regulate cell growth and migration, potentially contributing to metastasis; it can also regulate Circ 0000190, a circular RNA that inhibits CG through the caspase-3, p27, and cyclin D axis (Wang et al., 2020).

On the other hand, TKTL1 is associated with metabolism reprogramming, since it is involved in glucose metabolism in GC cells (Kämmerer et al., 2015). Moreover, it regulates other events such as proliferation, metastasis, epithelial-mesenchymal transition, resistance to chemoradiotherapy, and survival (Ahopelto et al., 2020). Finally, the relapse-specific mutations in phosphoribosyl pyrophosphate synthetase 1 gene (PRPS1), a rate-limiting purine biosynthesis enzyme that confers significant drug resistance to combination chemotherapy in acute lymphoblastic leukemia (Wang et al., 2018), suggest that these genes shared could indicate a core of early genes, given that they are also found in NAG, a non-cancerous lesion.

The NDUFS3 gene, which encodes a subunit of complex I of OXPHOS, was also identified among the shared genes. This gene also plays a significant role in metabolic reprogramming, an early event in carcinogenesis that occurs in the first phases of disease development. Metabolic reprogramming is a crucial aspect of the Warburg effect. The common feature of this altered metabolism is increased glucose uptake and fermentation of glucose to lactate, which is a common feature in cancer. Also, the H. pylori infection, like some of the NAG samples in this study, shows increased complex I, further highlighting the role of NDUFS3 (Feichtinger et al., 2017) and metabolic reprogramming as an early carcinogenesis event. Furthermore, in the MKN28 and MKN45 cell lines derived from moderately differentiated tubular adenocarcinoma and undifferentiated adenocarcinoma of medullary type, a decrease in this complex I of the mitochondrial respiratory chain is also observed (Puurand et al., 2012).

According to the NAG interaction network, PTPRJ and NUP160 genes could regulate cellular and molecular processes contributing to inflammation, immune response, energy and metabolism, and proliferation. PTPRJ encodes a protein from the tyrosine phosphatase family and dephosphorylates CTNND1, related to cell proliferation, adhesion, and migration (Du & Grandis, 2015; Holsinger et al., 2002), as with inflammation and regulation of the immune response. NUP160 encodes for a nuclear porin complex protein that regulates the transport of macromolecules between the nucleus and the cytoplasm (Xu & Powers, 2009); its deregulation can modify gene expression and cell signaling.

IGC network (Fig. 2) shows four multiple interaction genes in cancer signaling and progression process; RHOA encodes for a protein of the Rho-GTPase family and is associated with signaling (Nam, Kim & Lee, 2019), Ras homologous A (RHOA) as a significant signaling hub in GC regulates several cellular processes, including the cytoskeletal structure and cell adhesion. Recently, it was shown that RHOA can be targeted by small molecule inhibitors in cancer, implicating it as a potential druggable target (Kaibuchi, Kuroda & Amano, 1999). Therefore, it is linked to cancer progression. GNAI2 encodes the alpha subunit of the Guanine nucleotide-binding proteins (G proteins), and in stomach adenocarcinoma it is considered a prognostic marker (Li et al., 2024). Also, it is essential for the transduction of extracellular signals and is related to cellular processes such as growth, migration, regulation of proliferation, differentiation, and the response to external stimuli. MAPKAPK3 has been proposed as a therapeutic target (Cazes et al., 2022) that regulates cell stress response, proliferation, and survival; it has also been reported as an element of diagnosis, prognosis, and prediction (Niloofa, De Zoysa & Seneviratne, 2021).

Moreover, the MST1R (or RON) gene encodes macrophage-stimulating receptor-1 or Ron family tyrosine kinase receptors, which can act in the HIF-1α and β-catenin pathways. This receptor is involved in oncogenesis by regulating cell migration, adhesion, and survival; it plays an essential role in the inflammatory response and metastasis. In this way, RHOA and GNAI2 are related to the signaling and regulation of critical cellular processes in cancer. In contrast, the MST1R/RON gene is involved in the tumor progression of various types of cancer, and GC mutations have been reported (Purwar et al., 2023).

LOH and clinic

In certain tumor types, the LOH impacts over 20% of the genome, diminishing allelic diversity across numerous genes (Zhang & Sjöblom, 2021). Clinically, mutations associated with LOH on specific chromosomes can serve as biomarkers for assessing cancer risk and prognosis, may guide the development of new therapies, and offer insights into its evolution (Cavenee et al., 1983). LOH analysis is particularly relevant in oncology and is a biomarker in breast (Kaur et al., 2018), ovarian (Ryland et al., 2015), and leukemia cancers (Fitzgibbon et al., 2005). It is also linked to hereditary cancer syndromes, making it valuable in risk assessment programs for individuals with a family history of cancer (Lozac’hmeur et al., 2024).

Challenges in LOH

The study of LOH in cancer presents significant methodological and biological challenges. Variability in detection platforms, such as SNP arrays and genomic sequencing, leads to inconsistent sensitivity levels, increasing the risk of false positives or negatives. Additionally, distinguishing LOH from complex structural variants, like duplications or point mutations, complicates precise identification. Cellular context further challenges LOH analysis; it does not permanently fully inactivate tumor suppressor genes, mainly if the remaining gene copy retains partial function. Genes with haploinsufficiency add complexity, as LOH may not yield significant functional consequences. Contamination from non-cancerous cells in solid tumor samples also affects LOH detection accuracy. In contrast, genetic diversity in polyclonal tumors can skew LOH frequency estimates, complicating the interpretation of its relevance to tumor progression. Given the scarcity of LOH-focused studies, further research is essential to identify LOH-affected genes and understand their roles in cancer, potentially aiding diagnostic and therapeutic advances (Zhang & Sjöblom, 2021).