The present study included biopsy samples from 40 patients divided into four groups on the basis of the biopsy diagnosis: CRA, malignant polyp, CRC without lymph node metastases (CRC N0), or CRC with lymph node metastases (CRC N+). The gender and age of the patients, and the location of the lesions for each group, are listed in Table I. Only cases with clear biopsy diagnoses were included. In the CRA group, there were 4 tubular adenomas with high grade dysplasia, 4 tubulovillous adenomas with high grade dysplasia and 2 tubulovillous adenomas with low grade dysplasia. The malignant polyps group consisted of 5 tubulovillous adenomas, 3 tubular adenomas, 1 villous adenoma and 1 tubular adenoma, all with high grade dysplasia and with malignant transformation, evidenced by invasion of the dysplastic glands in the submucosa (pT1 stage). The TNM stages of the CRC N0 group were T2N0M0, T3N0M0 and T4aN0M0 in 2, 6 and 2 patients, respectively. The TNM stages of the CRC N+ group were T3N1Mx, T4aN1Mx and T4aN2Mx in 5, 2 and 3 patients, respectively.

As a control group, healthy colon mucosa from surgical margins from 20 patients with CRC N0 and CRC N+ was included.

All tissue samples were fixed for 24 h in 10% buffered formalin prior to the paraffin embedding. After fixation and embedding, tissues were cut into 3–4 µm sections and stained with hematoxylin and eosin for routine histopathological examination. For the purposes of the present retrospective study, representative paraffin blocks were retrieved from the archives of the Institute of Pathology, Faculty of Medicine, University of Ljubljana (tissues originally collected from December 2013 to October 2018).

Four 10 µm-thick FFPE sections were cut from the archival paraffin blocks for the isolation of total RNA. Total RNA extraction was performed using the AllPrep DNA/RNA FFPE kit (Qiagen GmbH), according to the manufacturer's instructions. Apart from the deparaffinization solution (hexadecane; Sigma-Aldrich; Merck KGaA) and the ethanol (Merck KGaA), all the reagents were from Qiagen GmbH. The RNA concentration and quality was evaluated using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Inc.). The A260/A230 nm intensity ratio was >1.0 and the A260/A280 ratio was >1.8. Reverse transcription and amplification using SYBR Green technology of the housekeeping gene GAPDH (95 bp) was performed as quality control. Reverse transcription was performed as described below. Each qPCR reaction contained 1 µl 10X GAPDH primers (QuantiTect primer assay Hs_GAPDH_1_S; cat. no. QT00079247; Qiagen GmbH), 5 µl iTaq Universal SYBR Green (Bio-Rad Laboratories, Inc.) and 4 µl of cDNA. Thermocycling conditions were as follows: 2 min at 50°C, 2 min at 95°C, and 45 cycles of 15 sec at 94°C, 30 sec at 55°C and 1 min at 72°C. This approach is similar to the approach that we use for FusionPlex Sarcoma, a next-generation sequencing panel, which includes the PreSeq RNA QC assay (39) after reverse transcription. These quality control assays basically use primers that amplify a universally expressed transcript (in the present study, GAPDH) to quantify the amount of amplifiable RNA of sufficient length in a sample. The presence of lower Cqs in the results is strongly indicative of successful cDNA preparation and predicts successful sequencing. All of the samples that were included in the present study passed amplification of GAPDH using SYBR Green (initial quality control) and those that did not amplify were not included in the study (the number of samples initially isolated was >2-fold the number of samples that passed quality control and were subsequently included in the final manuscript). Finally, for the selected genes, TaqMan probes (Thermo Fisher Scientific, Inc.) were designed to amplify and detect PCR products <84 bp long (Table SI).

Tissue cores from the tumor center and the invasive front were punched from FFPE tissue blocks using a 600 µm needle. Three punches were used for each isolation. Total RNA extraction was performed using the MagMax FFPE DNA/RNA Ultra kit (Applied Biosystems; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions, with one modification: Protease digestion was performed overnight with mixing for 15 sec at 300 rpm every 4 min instead of 1 h. Apart from the deparaffinization solution (xylene; Sigma-Aldrich; Merck KGaA) and the ethanol (Merck KGaA), all the reagents were from Applied Biosystems (Thermo Fisher Scientific, Inc). RNA quantity was assessed on a NanoDrop 1000 (Thermo Fisher Scientific, Inc.) by measuring the absorbance at A260.

Reverse transcription, preamplification and qPCR were performed to measure the expression levels of genes. The extracted total RNA was reverse transcribed using the OneTaq-PCR kit (New England Biolabs, Inc.) according to manufacturer's instructions. Each reverse transcription reaction contained 60 ng of extracted total RNA, 6 µM random primers mix, 5 µl 2X reaction mix and 1 µl 10X enzyme mix. First, RNA and random primers were mixed and incubated for 5 min at 70°C, then the reaction and enzyme mix was added and incubated for 5 min at 25°C, 60 min at 42°C and 4 min at 80°C.

Following cDNA synthesis, a preamplification procedure was performed using TaqMan PreAmp mastermix (Applied Biosystems; Thermo Fisher Scientific, Inc.). Each reaction contained 5 µl cDNA, 5 µl of 0.2X TaqMan Gene Expression Assays (Table SI) and 10 µl 2X TaqMan PreAmp Mastermix. TaqMan Gene Expression Assays were pooled, followed by dilution to 0.2X using Tris-EDTA buffer solution, pH 8.0 (Sigma-Aldrich; Merck KGaA). The thermocycling conditions were as follows: 10 min at 95°C and 10 cycles of 15 sec at 95°C and 4 min at 60°C.

Prior to qPCR, to calculate the efficiency of qPCR reactions, pools of each group were created by mixing RNA samples. RNA pools were reverse transcribed and preamplified, as aforementioned. Then, the preamplified cDNA was diluted in four steps, ranging from 5-fold dilution to 5,000-fold or 625-fold and qPCR was performed as described below. All qPCR reactions were conducted on a Rotor-Gene Q system (Qiagen GmbH), and each sample was run in triplicate. The thermocycling conditions were as follows: 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 sec at 95°C and 1 min at 60°C.

qPCR was performed using TaqMan technology with FastStart Essential DNA Probe Master (Roche Diagnostics). Each preamplification reaction (20 µl) was diluted with ddH₂O to a final volume 100 µl (5X). Each qPCR reaction contained 4.5 µl of diluted preamplification mixture, 5 µl of 2X FastStart Essential DNA Probe Master (Roche Diagnostics) and 0.5 µl of TaqMan Gene Expression Assay. The TaqMan Gene Expression assays used in the present study are listed in Table SI. Importin 8 (IPO8) and β2 microglobulin (B2M) were used as reference genes. All qPCR reactions were conducted on a Rotor-Gene Q system (Qiagen GmbH), and each sample was run in duplicate. The thermocycling conditions were as follows: 2 min at 50°C, 10 min at 95°C and 40 cycles of 15 sec at 95°C and 1 min at 60°C.

For immunohistochemistry analysis, commercially available antibodies against SPP1, SPARC, SPON2 and DCN were used. Detailed information about the antibodies is presented in Table II. Sections of 4 µm thickness were cut from paraffin blocks, deparaffinized, and processed for automated antigen retrieval and staining with the antibodies in an automatic immunostainer (Ventana BenchMark; Roche Diagnostics). Sections were treated with biotinylated secondary antibody (8 min at 37°C), followed by incubation with peroxidase-conjugated streptavidin (both part of the UltraVIEW DAB Detection kit; Roche Diagnostics). Visualization of the immunoreaction was established with 3,3′-diaminobenzidine. The sections were then counterstained with haematoxylin.

To calculate relative gene expression, Cq values were corrected according to Latham (40). To obtain ∆Cq, the geometric mean of Cq values of the reference genes (IPO8 and B2M) was deducted from the gene of interest (DCN, EPH4, FN1, SPARC, SPON2 and SPP1). Differences in expression in the paired tissue samples of tumour and adjacent healthy mucosa from the CRC N0 group were analyzed for significance with the Wilcoxon signed rank test. For all other comparisons of ΔCqs between different groups of samples (healthy mucosa, CRA, malignant polyp, CRC N0 and CRC N+), ANOVA was used with two different post-hoc tests, Bonferroni and Hochberg's GT2 (41). Additionally, the Spearman's Rank correlation coefficient (ρ) was used to calculate the statistical significance of association between the expression of analyzed genes (ΔCq values) and various lesions (CRA, malignant polyp CRC N0 and CRC N+). All the statistical analyses were performed using SPSS analytical software version 24 (IBM Corp.), with a cut-off of P≤0.05.

The majority of genes tested in the present study were downregulated in CRA compared with healthy mucosa: DCN (16.1-fold, P<0.001), FN1 (8.1-fold, P=0.022), SPARC (2.6-fold, not significant), SPON2 (2.7-fold, not significant) and SPP1 (21.1-fold, P=0.006), whereas EPHA4 was slightly upregulated but not statistically significant. Results are summarized in Fig. 1 and Table III.

DCN was significantly downregulated in the malignant polyp group compared with healthy mucosa (4.1-fold, P=0.031; Fig. 1). FN1 and SPP1 also appeared downregulated in malignant polyps compared with healthy mucosa, however these changes were not significant. The other three genes exhibited a trend towards upregulation in malignant polyps, albeit not significant. Results are summarized in Fig. 1 and Table III.

Expression of genes in CRC N0 compared with healthy colon mucosa exhibited a significant upregulation of the levels of SPARC (12.17-fold, P<0.001) and SPON2 (5.65-fold, P<0.001). DCN, EPHA4 and SPP1 expression levels were slightly upregulated and FN1 was downregulated, however these changes were not significant. Results are summarized in Fig. 1 and Table III.

Expression of DCN, EPHA4, SPARC, SPON2 and SPP1 showed similar patterns in the CRC N+ and CRC N0 groups, both compared to their corresponding healthy mucosa. Statistical significance was reached for SPARC (17.3-fold, P<0.001), SPON2 (12.17-fold, P<0.0017) and SPP1 (20.65-fold, P<0.001) expression levels. In contrast to CRC N0, FN1 was upregulated (5.5-fold) in CRC N+ when compared with healthy mucosa, although this change was not statistically significant. Results are summarized in Fig. 1 and Table III.

Statistically significant difference between the CRA and malignant polyp groups was observed for the SPON2 and SPARC genes. In addition, statistically significant difference between the CRA and the CRC N0 and CRC N+ groups was observed for all genes tested, except for EPHA4. Results are summarized in Table III.

Expression of most of the genes gradually increased with the level of malignancy, from CRA to malignant polyps and to CRC. Accordingly, the results demonstrated a statistically significant association between the various lesions (CRA, malignant polyp, CRC N0, CRC N+) and the expression levels of DCN, EPHA4, FN1, SPARC, SPON2 and SPP1 (ρ=0.660, p<0.001; ρ=−0.259, p=0.039; ρ=0.495, p<0.001; ρ=0.531, p<0.001; ρ=0.505, p<0.001; and ρ=0.677, p<0.001, respectively).

Expression of DCN was slightly upregulated in CRC N0 compared with CRC N+. The difference in DCN expression between CRC N0 and CRC N+ was statistically significant (P=0.025). There was no significant change in the expression levels of EPHA4, FN1, SPARC, SPON2 and SPP1 between the CRC N0 and CRC N+ groups. The expression levels of EPHA4 were upregulated compared with healthy mucosa, but the same in CRC N0 and CRC N+. FN1 expression was slightly downregulated in CRC N0 and upregulated in CRC N+, compared with healthy mucosa. The expression of SPP1, SPON2, and SPARC was upregulated both in CRC N0 and CRC N+ compared with healthy mucosa, and expression of these genes was slightly but not significantly higher in CRC N+ than in CRC N0. The P-value of the difference of the SPP1 levels between the CRC N0 and CRC N+ groups was just above the cut-off (P=0.083).

Expression of DCN, FN1 and SPARC was upregulated at the invasive front of the tumour compared with the central parts in the CRC N+ group, whereas in the CRC N0 group, expression at the invasive front was similar to the central parts (Fig. 2). The expression levels of SPON2 and SPP1 were downregulated at the invasive front compared with the central parts and similar in both CRC N0 and CRC N+ (Fig. 2). The highest difference in expression levels was observed in the case of the EPHA4 gene; however, there was no statistically significant difference, even though its expression was downregulated in CRC N0 and upregulated in CRC N+ at the invasive front compared with the central parts of the tumour (Fig. 2).

Expression of most of the genes tested in the present study appeared increased with the level of malignancy, from CRA to malignant polyps and to CRC. Therefore, immunohistochemistry was performed for the DCN, SPARC, SPON2 and SPP1 proteins, since these four genes exhibited statistically significant differences in expression using RT-qPCR between CRC N0 and its corresponding healthy mucosa (DCN, 4.22-fold, P=0.028; SPARC, 12.17-fold, P=0.007; SPON2, 5.65-fold, P=0.017; and SPP1, 5.24-fold, P=0.036; Fig. 1).

Immunohistochemistry analysis for the investigated proteins revealed two patterns. One pattern was observed for the SPON2 and SPP1 protein expression: Positive staining was evident in the epithelial cells of the healthy colon mucosa and in dysplastic glands, both in CRA and CRC tissues, and the intensity of staining progressively increased with the severity of the lesions, being the most intense in CRC (Fig. 3).

The other pattern was observed for the DCN and SPARC protein expression: Positive staining was evident in the lamina propria in healthy mucosa and in the stroma of CRA and CRC (Fig. 3). In this pattern, too, the intensity of staining increased with the severity of the lesions, being most intense in CRC (Fig. 3).