Retrospectively collected archived formalin-fixed and paraffin-embedded (FFPE) tumor tissue samples were obtained from 62 treatment-naïve patients with DGC [mean age at diagnosis ± standard deviation, 63.13±11.5 years; 36 males (58.1%) and 26 females (41.9%)] who underwent surgical resection for GC between January 2008 and October 2011 at the Department of Pathology, University Hospital Centre (Zagreb, Croatia). Histopathological diagnosis was performed according to the Lauren histological classification and the World Health Organization (WHO) criteria (3,51). With the exception of DGC tumor tissue samples, all other histopathological types of GC were not included in the study (exclusion criteria). Biopsy samples [n=62; mean age at gastroscopy ± standard deviation, 54.58±15.65 years; 35 males (56.5%) and 27 females (43.5%)] of the normal gastric mucosa obtained between October 2015 and October 2017 at the Department of Pathology, University Hospital Centre (Zagreb, Croatia) from patients without GC and H. pylori infection (exclusion criteria) who underwent gastroscopy at the Department of Gastroenterology, University Hospital Centre (Zagreb, Croatia) were used as the control. The present study protocol, including the use of archival DGC FFPE tissue samples and control FFPE samples obtained from patients without GC and H. pylori infection, was approved by the Clinical Research Ethics Committee of the University Hospital Centre Zagreb (approval no. 02/21/JG, class no. 8.1.-14/93/2; November 1, 2014) and the Research Ethics Committee of the School of Medicine, University of Zagreb (approval no. 380-59-10106-21-111/186, class no. 641/01/21-02/01). The patient demographic (i.e., age at diagnosis and sex) and clinical data (i.e., location of DGC and tumor staging parameters) were retrieved from their corresponding medical records and written informed consent was obtained from all participants involved in the present study.

Retrieved FFPE DGC and normal gastric mucosa tissue blocks prepared immediately after surgical resection/biopsy procedures (to escape the potential loss of epitopes due to a delayed time to fixation). Briefly, 2–5-mm thick tissue sections were first immersed in 4% formaldehyde (for 48 h at 4°C) followed by dehydration using increasing concentrations of ethanol [for a 1-h incubation in each concentration at room temperature (RT)] and incubation in xylene (3 changes; 1.5 h each at RT), and then embedded in paraffin wax according to a standard procedure (52). Obtained FFPE tissue blocks were first cut into 5-µm thick sections and placed on silanized glass slides (DakoCytomation; Agilent Technologies, Inc.) for immunostaining. Sections were deparaffinized (3X; 5 min) with xylene (Sigma-Aldrich; Merck KGaA) and rehydrated in an ethanol (Sigma-Aldrich; Merck KGaA) gradient and distilled water. Endogenous peroxidase was blocked using 3% hydrogen peroxide for 10 min at RT, followed by heating in a water bath for 20 min at 97°C in Dako REAL Target Retrieval Solution (S2367; Dako; Agilent Technologies, Inc.). After incubation, the slides were washed three times in 1% PBS/5% goat serum (Sigma-Aldrich; Merck KGaA). Protein expression levels and cellular localization of SFRP1, SFRP3, DVL2 and DVL3 were determined using rabbit polyclonal antibodies against human SFRP1 (dilution, 1:200; cat. no. sc-13939; Santa Cruz Biotechnology, Inc.), SFRP3 (dilution, 1:200; cat. no. sc-13941; Santa Cruz Biotechnology, Inc.), DVL2 (dilution, 1:500; cat. no. sc-13974; Santa Cruz Biotechnology, Inc.) and DVL3 (dilution, 1:100; cat. no. sc-271295; Santa Cruz Biotechnology, Inc.). Following incubation with primary antibodies at 4°C overnight, the slides were incubated with ready-to-use HRP-conjugated secondary antibodies (cat. no. K8024; EnVision Flex/HRP High pH; Dako; Agilent Technologies, Inc.) for 45 min at RT, developed in a 3,3′-diaminobenzidine chromogen solution (Dako; Agilent Technologies, Inc.) and counterstained with ready to use hematoxylin solution (Mayer's, modified; Abcam) for 35 sec at RT. Negative controls underwent the same staining procedure with the exclusion of the primary antibodies. The FFPE tissue sections of renal cell carcinomas were used as a positive control for SFRP1 and SFRP3 protein expression, while FFPE placental tissue sections were used as a positive control for DVL1 and DVL3 protein expression as previously described (53,54). The inclusion of several tissue sections obtained from one placental (female; 35 years of age) and one clear cell renal cell carcinoma (male, 55 years of age) FFPE tissue block to be used as positive controls for IHC staining was approved by the Clinical Research Ethics Committee of the Clinical Hospital ‘Sveti Duh’ Zagreb, Croatia (approval no. 01-4685; November 20, 2017), Clinical Hospital ‘Merkur’ Zagreb; Croatia (approval no. 03/1-11088, December 7, 2017) and the Research Ethics Committee of the School of Medicine University of Zagreb (approval no. 380-59-10106-19-111/181, class 641-01/19-02/01). Renal cell carcinomas and placental tissue FFPE sections were obtained from the Department of Pathology, School of Medicine, University of Zagreb (Zagreb, Croatia) and from the Department of Pathology, Clinical Hospital ‘Merkur’ (Zagreb, Croatia), respectively. Samples use as positive controls for IHC staining were collected between January and March 2018, and informed consent was obtained from both patients providing the tissue samples. Analysis of the IHC staining parameters was performed using compound light microscopy (Olympus) by two independent observers blinded to the parameters of the experiment.

SFRP1 and SFRP3 protein expression was measured by performing a quantitative stereological analysis of the numerical density (Nv) and volume density (Vv) of their IHC staining pattern, while DVL2 and DVL3 expression in normal gastric mucosa and DGC tissue samples was assessed using the following semi-quantitative scoring method: 0, no staining; 1, positive staining in <10% cells; 2, positive staining in 10–50% cells; and 3, positive staining in >50% of the cells. For the statistical analysis of DVL2 and DVL3 protein expression, the DGC and control groups were subsequently subdivided into the following two groups: i) Positive staining in <10% cells combined with sections with no staining; and ii) positive staining in 10–50% of the cells. For 3 patients there are no data for DVL2 and DVL3 IHC staining due to tissue scarcity. Thus, for this analysis the total number of DGC patients was 59.

Quantitative stereological analysis of SFRP1 and SFRP3 Nv in the aforementioned IHC samples was performed using a Nikon Alphaphot binocular light microscope (Nikon Corporation) coupled with a Weibel's multipurpose test system with 42 points at a magnification of ×400, with a total length of test lines of 1.008 mm and test area (A_t) of 0.0837 mm² for each microscopic field analyzed (53). For each investigated group, an orientation/pilot stereological measurement was performed to first define the number of fields to be tested (55). The Nv of SFRP1- and SFRP3-positive cells was then determined according to the point-counting method and calculated using the following formula: N_v × ¼N=A_t × D, where N is the number of positive cells on the tested area, and D is the average ‘tangent diameter’ of the structures (55–57). Subsequently, the SFRP1 and SFRP3 Vv calculation was performed according to the following formula: Vv=Pf/Pt, where Pf represents the number of hit test points on the positive cells and Pt represents the number of all test points on the tested area (55).

Total RNA was isolated from five consecutive 5-µm thick sections of DGC FFPE tissue blocks (n=62), and FFPE tissue blocks of the normal gastric mucosae (n=62) were used as the control. All tissue sections were deparaffinized by incubation in 1.0 ml xylene (Invitrogen; Thermo Fisher Scientific, Inc.) for 3 min at 50°C followed, by centrifugation three times for 5 min at RT at 12,000 × g each. The supernatant was then discarded, and the pellet was washed three times with 1.0 ml absolute ethanol. Subsequently, the pellet obtained from tissue sections was incubated in 350 µl protease K digestion buffer (20 mM Tris-HCl pH 8.0; 1 mM CaCl₂; 0.5% sodium dodecyl sulfate and 500 µg/ml protease K; all reagents were obtained from Sigma-Aldrich; Merck KGaA) overnight at 55°C. Total RNA was isolated using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. RNA purity and concentration were evaluated using a NanoDrop 200 spectrophotometer (Thermo Fisher Scientific, Inc.). A total of 1 µg total RNA from each sample was reverse transcribed using the high-capacity cDNA Reverse Transcription kit (Applied Biosystems; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. The mRNA expression levels of SFRP1, SFRP3, DVL2 and DVL3 were determined using a CFX-96 real-time qPCR detection system and a C100 thermal cycler (Bio-Rad Laboratories, Inc.). All qPCR reactions were performed in triplicate using TB Green™ Premix Ex Taq™ II (Tli RNaseH Plus PCR master mix; Takara Biotechnology Co., Ltd.). The following thermocycling conditions were used: Initial denaturation at 95°C for 30 sec, followed by 40 cycles of 95°C for 5 sec and 60°C for 30 sec. The CFX96 manager software version 3.1, (Bio-Rad Laboratories, Inc.) was used to generate the cycle threshold values and the data were analyzed using the 2^−ΔΔCq method (58). The relative mRNA expression levels of SFRP1, SFRP3, DVL2 and DVL3 were normalized against the mRNA expression levels of the β-actin (ACTB) gene as an endogenous control. The specificity of qPCR amplification was confirmed using melting curve analysis. The sequences of the oligonucleotide primers used were as follows: SFRP1 forward, 5′-ATCTCTGTGCCAGCGAGTTT-3′ and reverse, 5′-GGCTTCTTCTTCTTGGGGAC-3′ [accession no. NM_003012.5; product length, 94 nucleotides (nt)]; SFRP3 forward, 5′-CCTGCCCTGGAACATGACTAA-3′ and reverse, 5′-CAGACCTTCGAACTGCTCGAT-3′ (accession no. NM_001463.4; product length, 91 nt); DVL2 forward, 5′-TGAGCAACGATGACGCTGTG-3′ and reverse, 5′-GCAGGGTCAATTGGCTGGA-3′ (accession no. NM_004422.3; product length, 148 nt); DVL3 forward, 5′-ACAATGCCAAGCTACCATGCTTC-3′ and reverse, 5′-AGCTCCGATGGGTTATCAGCAC-3′ (accession no. NM_004423.4; product length, 109 nt); and ACTB forward, 5′-GGGCATGGGTCAGAAGGATT-3′ and reverse, 5′-AGTTGGTGACGATGCCGTG-3′ (accession no. NM_001101.5; product length, 101 nt).

The in silico analysis of differential gene expression (mRNA levels) of SFRP1, SRP3, DVL2 and DVL3 between the GC tissue samples and corresponding adjacent normal solid tissue or normal gastric tissue obtained from non-cancerous patients was initially performed using the University of California Santa Cruz (UCSC) Xena browser (https://xenabrowser.net; accessed, August 9, 2021) (59).

Firstly, clinical and genetic information of the GC samples (n=580) were derived from TCGA-STAD dataset (https://xenabrowser.net; accessed, August 9, 2021) (59). The obtained data were then sub-divided according to the tumor histological classification, where only data corresponding to DGC (tumor and adjacent normal tissues) with no history of neoadjuvant therapy (total number of samples n=75; DGC n=69; adjacent normal solid tissue n=6) were used. Data corresponding to IGC subtypes and other GC samples that were not histologically specified were excluded from further analysis. Secondly, using the UCSC Xena browser, the so-called TCGA TARGET Genotype-Tissue Expression (GTEx) study analysis of the GC tissues was performed, which contained both the gene expression data for GC tumor tissue (n=414) and gene expression data for normal tissue samples (n=174) obtained from individuals who do not have cancer (https://xenabrowser.net; accessed, August 9, 2021) (59). In TCGA TARGET GTEx study analysis, data for the normal adjacent tissues were omitted from the analysis so that the final number of analyzed samples was 588 (59). RNA-seq data corresponding to SFRP1, SRP3, DVL2 and DVL3 gene expression obtained from TCGA STAD study and TCGA TARGET GTEx study were normalized using the UCSC XENA browser and presented as log₂(norm count + 1) and RNA-seq data (IlluminaHiSeq TCGA Hub) were normalized and presented as log₂(norm count + 1), in which the norm_count refers to RNA-seq by Expectation Maximization values. Gene transcript expression was calculated as transcripts per million (TPM). The following criteria were applied: log₂(fold-change)|≥2, P<0.05. The expression of SFRP1, SFRP3, DVL2 and DVL3 in tumor and adjacent normal tissues was compared using Welch's t-test (unequal variances t-test), and P<0.05 was considered to indicate a statistically significant difference. Analysis was performed online using the default parameters of the UCSC Xena browser (https://xenabrowser.net; accessed, August 9, 2021) (59).

The in silico analysis of differential SFRP1, SFRP3, DVL2 and DVL3 gene expression (mRNA levels) between DGC and corresponding adjacent solid normal tissues was also performed using the UALCAN web resource (http://ualcan.path.uab.edu/analysis.html; last accessed, August 9, 2021) as previously described (45,60). In this case, the required clinical and genetic information related to the DGC samples was also derived from the TCGA-STAD dataset. Differences in the transcriptional levels of SFRP1, SFRP3, DVL2 and DVL3 were first assessed between the adjacent normal tissue samples (n=34) and all primary GC cases (n=415) and subsequently between the adjacent normal tissue samples and GC histological subtypes [gastric adenocarcinoma non otherwise specified n=155; DGC n=69; signet-ring cell carcinoma (SRCC) GC samples n=12; IGC not otherwise specified n=73; IGC tubular n=76; IGC mucinous n=20, and IGC papillary n=7]. In all cases, comparisons between the groups were performed using the unequal variances t-test as previously described (45,60). The differential analysis of SFRP1, SFRP3, DVL2 and DVL3 transcriptional levels (mRNA) in adjacent solid normal tissues and distinct histological subgroups of GC from the UALCAN dataset was assessed using a t-test, and P<0.05 was considered to indicate a statistically significant difference (60). As a measure of targeted gene expression (SFRP1, SFRP3, DVL2 and DVL3) UALCAN web resource uses the TPM expression values to generate the visualized representations of gene expression featuring the interquartile range (IQR), including minimum, 25th percentile, median, 75th percentile and maximum values (60). Outliers were excluded from the plot. Analysis was performed online using the default parameters of the UALCAN web resource (http://ualcan.path.uab.edu/analysis.html) (60).

Statistical analysis was performed using SPSS software version 22.0 (IBM Corp.) and GraphPad Prism 5.01 (GraphPad Software, Inc.). Normal distribution was assessed using the Shapiro-Wilk test. Data are presented as counts (n) and percentages (%) for categorical variables and as the mean ± standard deviation or median and IQR for continuous variables. Statistical analyses were performed using the Fisher's exact test, unpaired independent-samples t-test, Mann-Whitney U test and Kruskal-Wallis test (with Bonferroni post hoc test for multiple comparisons when applicable), as appropriate. The correlations between SFRP1, SFRP3, DVL2 and DVL3 protein expression values and patient clinicopathological data were determined using the Spearman's rank correlation model. All statistical tests were two-sided and intergroup differences with P<0.05 were considered significant and corrected according to the Bonferroni procedure [the corrected P-value threshold for significance (P_c) was as follows: P_c = 0.05/N, where N represents the number of independent tests]. All reported P-values were uncorrected unless stated otherwise.

The basal clinical and demographic characteristics of the patients with DGC included in the present study are shown in Table I. No significant difference (DGC: 36 males (58.1%) and 26 females (41.9); control: 35 males (56.5%) and 27 females (43.5%); P>0.999) was observed in gender distribution between the DGC and control patients group (data not shown). Significant differences were observed between the DGC and control groups in terms of age at the time of surgery/biopsy (DGC, 63.13±11.50 years; control, 54.58±15.65 years, P=0.001; data not shown). However, no difference was observed when this variable was separately analyzed in male and female patient groups (DGC, 64.39±10.45 years for male patients and 61.32±12.91 years for female patients, P=0.309; control, 54.57±16.40 years for male patients and 54.59±14.94 years for female patients, P=0.996; data not shown).

At the time of surgery, 50% of the tumors were localized in the antrum (30.6%) or minor curvature (21.0%; data not shown). In addition, the majority of the DGC cases were either in the T3-T4 (88.7%), N1-N3 (77.4%) and/or M0 (72.6%) clinical stages, indicating advanced stages of disease (Table I).

For 4 patients, clinicopathological data were incomplete.

IHC analysis revealed higher protein expression levels of SFRP1 and SFRP3 in the normal gastric mucosa compared with the tumor tissue samples. In the normal gastric mucosa samples, SFRP1 protein was mainly localized in the cytoplasm and nuclei of gland cells, while SFRP3 staining was also detected in the plasma membrane (Figs. 1 and 2). Cytoplasmic and nuclear localization of the SFRP1 protein was also detected in the DGC tissue sections, while SFRP3 localization in the DGC tissues was predominantly cytoplasmic, although nuclear and membrane positivity could also be observed (Figs. 1 and 2).

Higher SFRP1 and SFRP3 protein expression in control tissues was also demonstrated using quantitative stereological analysis. Specifically, the Nv values for SFRP1 [control, 372,732.07 (325,775.89–437,725.94); DGC, 67,516.88 (49,371.91–105,720.87); P<0.001] and SFRP3 [control, 344,669.50 (300,769.64–390,305.91); DGC, 72,657.05 (40,774.08–124,892.33); P<0.001] protein expression were significantly higher in normal gastric mucosa samples compared with DGC tissue samples (Fig. 3). Additionally, a similar trend was identified in Vv analysis (Fig. 3) in terms of both SFRP1 [control, 0.20 (0.16–0.32); DGC, 0.07 (0.04–0.09); P<0.001] and SFRP3 [control, 0.19 (0.16–0.27); DGC, 0.06 (0.04–0.08), P<0.001]. These results of Nv and Vv analyses remained significant even after Bonferroni's correction was applied (P<P_c; P_c=0.05/2=0.025).

When SFRP1 and SFRP3 protein expression values were compared separately among the members of the DGC or control tissue group, no significant difference was observed for either Nv or Vv after Bonferroni's correction was applied (data not shown). However, in both DGC and control tissue samples, slightly higher values for SFRP1 expression were observed compared with those of SFRP3 expression (data not shown). In addition, no statistically significant association was found between the Nv or Vv values of SFRP1 and any of the demographic or clinicopathological characteristics analyzed (Table II). By contrast, Vv values of SFRP3 protein expression were significantly associated with T stage (P=0.004; Mann-Whitney U test) and DGC clinical stage (P=0.004; Mann-Whitney U test). In particular, the values of SFRP3 Vv were higher in the T3-T4 group [0.06 (0.05–0.08)] compared with the T1-T2 stage group [0.04 (0.0328–0.06)], while also being higher in the III–IV group [0.07 (0.05–0.08)] compared with the I–II clinical stage group [0.04 (0.03–0.06); Table III]. Therefore, this suggested that higher SFRP3 protein expression levels may be associated with more advanced DGC stages.

Positive IHC staining for DVL2 and DVL3 in the normal gastric mucosa and DGC tissues was found mainly in the cytoplasm of gland cells (Figs. 4 and 5). In control tissues, both DVL2 and DVL3 proteins exhibited mostly weak (1+; DVL2, 56.9%; DVL3, 63.8%) to moderate (2+; DVL2, 27.6%; DVL3, 34.5%) staining, while no staining (IHC=0) was detected in 15.5% (DVL2) and 1.7% (DVL3) of the cases (data not shown). Furthermore, a predominantly weak (1+; DVL2, 44.1%; DVL3, 11.9%) to moderate (2+; DVL2, 55.9%; DVL3, 88.1%) staining pattern was also observed in DGC. However, negative DVL2 and DVL3 staining patterns were not observed in the analyzed DGC tissue sections (data not shown). No staining pattern with >50% stained cells (IHC=3+) could be observed in any of the analyzed DGC or control tissue sections (data not shown). Therefore, for the statistical analysis of DVL2 and DVL3 protein expression, the DGC and control groups were subsequently subdivided into the following two groups: i) Positive staining in <10% cells combined with sections with no staining; and ii) positive staining in 10–50% of the cells.

The protein expression levels of DVL2 (P<0.001) and DVL3 (P<0.001) were significantly higher in DGC tissues compared with in control tissue sections (Fig. 6), which remained significant even after Bonferroni's correction was applied (P_c=0.05/3=0.017).

When the protein expression values of DVL2 and DVL3 were compared separately among the members of the DGC or control tissue group, no statistically significant difference was observed. However, in both DGC and control tissue samples, slightly higher DVL3 protein expression values were observed compared with those of DVL2 (data not shown). No statistically significant association was observed between either DVL2 or DVL3 protein expression and any of the clinicopathological parameters of the patients with DGC analyzed in the present study (Table IV).

To measure the mRNA expression levels of SFRP1, SFRP3, DVL2 and DVL3, RT-qPCR was performed using RNA isolated from the DGC and normal gastric mucosa tissue samples. SFRP1, SFRP3, DVL2 and DVL3 were found to be transcriptionally active in both DGC and control tissue samples. However, contrary to observations made from IHC staining, RT-qPCR analysis only revealed significantly higher (2.81-fold; P<0.05) expression levels of SFRP1 mRNA in the control tissues and significantly higher (1.99-fold; P<0.05) expression levels of DVL3 mRNA in the DGC tissue samples (Fig. 7). Although SFRP3 and DVL2 mRNA expression was slightly higher in the control and DGC tissue samples, respectively, no statistical significance was observed. In addition, no statistically significant difference between the DGC and control tissue samples was observed in terms of the mRNA expression levels of the ACTB gene used as the reference control (data not shown).

Spearman's correlation analysis among all cases combined (DGC + control) revealed the following negative and statistically significant correlations: i) SFRP1 Nv and DVL2 (r=−0.310; P=0.001) or DVL3 (r=−0.470; P<0.001) protein expression; ii) SFRP1 Vv and DVL2 (r=−0.209; P=0.024) or DVL3 (r=−0.390; P<0.001) protein expression; iii) SFRP3 Nv and DVL2 (r=−0.339; P<0.001) or DVL3 (r=−0.480; P<0.001) protein expression; and iv) SFRP3 Vv and DVL2 (r=−0.252; P=0.006) or DVL3 (r=−0.431; P<0.001) protein expression (data not shown). Furthermore, a positive but not significant correlation was found between DVL2 and DVL3 expression (r=0.157; P=0.090).

In terms of the clinicopathological data, statistically significant correlations were found only between SFRP3 Vv and tumor tissue T stage (r=0.304; P=0.017), in addition to between SFRP3 Vv and clinical stage (r=0.336; P=0.008) in DGC tissues, but this was not confirmed for SFRP3 Nv values (T stage: r=0.227, P=0.078; clinical stage: r=0.174, P=0.179). Statistically significant correlations between the SFRP3 Vv values and T stage (r=0.357; P=0.005), in addition to between SFRP3 Vv and clinical stage (r=0.368; P=0.003), were also found in a separate analysis of the DGC group (data not shown). In the DGC subgroup, there was also a negative but not significant correlation between DVL2 and DVL3 protein expression (r=−0.220; P=0.094; data not shown).

To verify the experimental findings from the IHC and RT-qPCR analyses, in silico analysis of SFRP1, SRP3, DVL2 and DVL3 gene expression (mRNA levels) in GC tissues based on publicly available RNA-seq data derived from TCGA-STAD project was performed. Using the web-based UCSC Xena browser analysis of target gene expression was first performed on the selected paired samples of tumor and adjacent normal tissues of DGC cases without a history of neoadjuvant therapy (Fig. 8A-E), in line with the patients with DGC recruited for the present study. Similar to the results of the present study, the expression levels of SFRP1 (P<0.001) were significantly lower in the primary tumor tissues compared with in solid adjacent normal tissues (Fig. 8F). By contrast, no statistically significant difference was observed for SFRP3 (P=0.171) and DVL3 (P=0.125), while, opposite to the results of the present study, DVL2 (P=0.005) was expressed at significantly lower levels in the primary tumor tissues compared with the solid adjacent normal tissues (Fig. 8F).

Since the normal solid tissues adjacent to the tumor tissues may, even if they appear histologically normal under the microscope, present transcriptionally and translationally unique intermediate states (morphologically normal but molecularly altered cells) between non-tumor-bearing healthy tissues and nearby tumor tissues (61), so-called ‘TCGA TARGET GTEx study analysis’ of the GC tissue was also performed (Fig. 9), which included expression data for both solid adjacent normal tissues and normal tissues obtained from individuals with no malignant disease (59,62,63). The data for the solid adjacent normal tissue were omitted from this analysis (Fig. 9C). In this case, the grouping of GC cases based on their histological subtypes was not possible and the obtained data, therefore, corresponded to overall GC cases and not specifically to the DGC subtype itself (Fig. 9A-F).

In line with the RT-qPCR data of the present study and aforementioned in silico UCSC Xena browser analysis performed on the paired samples of tumor and adjacent normal tissues, the expression levels of SFRP1 (P<0.001) were significantly higher in normal gastric tissue samples compared with in tumor GC tissues, while no difference was observed for SFRP3 (P=0.476) expression (Fig. 9G). By contrast, the expression levels of DVL2 (P=0.02) and DVL3 (P<0.001) were significantly higher in the GC tissue samples than in normal gastric tissue samples (Fig. 9G).

Data from TCGA-STAD project were also used for the analysis of SFRP1, SFRP3, DVL2 and DVL3 gene expression between the DGC tumor tissue and adjacent solid normal tissue samples using the UALCAN web browser (Figs. 10 and 11).

The expression levels of SFRP1 were significantly lower in both primary GC tumors (P<0.001) and DGC tumor tissues (P<0.05) compared with in the adjacent solid normal tissue samples (Fig. 10A and B). Additionally, the expression levels of SFRP3 were significantly lower in overall cohort of primary GC tumors (P<0.05) compared with in the adjacent solid normal tissue samples (Fig. 10C). However, no significant difference in SFRP3 expression was detected between the tissues of DGC tumor subtype and the adjacent solid normal tissue samples (Fig. 10D).

Significantly higher expression levels of SFRP1 were also detected in SRCC of the stomach (P>0.05; group IV in Fig. 10B) compared with those in the adjacent solid normal tissue samples, while no significantly different expression of either SFRP1 or SFRP3 gene expression was detected between the DGC and SCRR histological subtypes (Fig. 10B and D).

Previously, this histologic subtype of gastric adenocarcinoma was, by Lauren's classification, considered as a part of the DGC type or designated as infiltrative, undifferentiated or high-grade GC subtype as defined by Nakamura, Ming and International Union Against Cancer classification systems, respectively (4,64–66). At present, in the WHO classification of gastrointestinal tumors, the SRCC type is defined as a poorly cohesive carcinoma (51).

The expression levels of DVL3 were significantly higher in both primary GC tumors (P<0.001) and DGC tumor tissues (P<0.001) compared with levels in the adjacent solid normal tissue samples (Fig. 11C and D). By contrast, DVL2 expression was significantly higher in primary GC samples (P<0.05) compared with in the adjacent solid normal tissue samples, while no significant difference in DVL2 expression was detected between the DGC tumor tissues and the adjacent solid normal tissue samples (Fig. 11A and B). Significantly higher expression levels of DVL3 were also detected in the SRCC histological subtype samples (P<0.01; group IV in Fig. 11D) compared with the adjacent solid normal tissue samples. No significantly different expression levels of either DVL2 or DVL3 gene expression were detected between the DGC and SCRR histological subtypes (Fig. 11B and D).