Tissue samples of primary tumors of human colorectal carcinoma and paired distant unaffected mucosa, where possible at least 20 cm from the primary tumor site, were collected from 67 patients (age range, 39–79 years old) with CRC diagnosed and treated at the Department of Surgery, Teaching Hospital in Pilsen (Pilsen, Czech Republic) during the period February 2008-August 2010 as described before (26). The following inclusion criteria were applied to the recruitment of patients into the study: i) Patients who were subject to surgery for CRC; ii) no prior chemotherapy before surgery (in order to eliminate its influence on transcript levels); iii) patients who received only first-line chemotherapy in either a palliative or adjuvant setting; and iv) patients who received adjuvant regimens based on 5-fluorouracil, leucovorin (de Gramont or FUFA), ftorafur, capecitabine and/or oxaliplatin (FOLFOX) or palliative chemotherapy based on FOLFOX regimen in combination with or without Avastin (n=3 untreated). Native tissue samples were taken during resection surgery, macrodissected, snap-frozen in liquid nitrogen and stored at −80°C until total RNA isolation. The control mucosa samples were taken from the macroscopically unaffected resection margins of colon tissues. The resection margins were microscopically evaluated and only samples free of malignant cells were further analyzed. Corresponding tumor tissue samples were verified by an independent experienced pathologist (Teaching Hospital in Pilsen, Pilsen, Czech Republic). Only histologically-verified samples were included in this study. The following data on patients were retrieved from medical records during regular hospital-based follow-up: Age, sex, date of diagnosis, tumor localization, pathological tumor node metastasis (TNM stage) according to The Union for International Cancer Control (UICC) 6th Edition (27), histological type and grade of the tumor, adjuvant (group A) or first line of palliative (group B) chemotherapy, treatment response in group B patients and disease-free interval (DFI) and overall survival time (OS) in all patients (Table I). Patients were followed by imaging techniques and assessment of circulating tumor markers (CEA and CA 19-9) every 3 months during the first two years after adjuvant chemotherapy and then every 6 months for the next 3 years. Response to the treatment was evaluated by the Response Evaluation Criteria In Solid Tumors (RECIST) criteria version 1.0 (28) based on routinely used imaging techniques (by computer tomography, with or without positron emission tomography, magnetic resonance or ultrasonography) for assessment of tumor mass. Response to the treatment was defined as a decrease of the number or volume of metastases or stabilization of the disease. In patients treated with adjuvant therapy after radical surgical resection R0 (group A), DFI served as the treatment outcome for analyses. DFI was defined as the time elapsed between radical surgical R0 resection and disease recurrence.

A cohort of 48 patients (age range, 46–80 years old) with PDAC who underwent curative intent surgery between August 2009 and January 2012 were recruited from the Institute of Clinical and Experimental Medicine, Prague and the University Hospital (Brno, Czech Republic) as described before (29). The following inclusion criteria were applied to the recruitment of patients into the study: i) Patients who were subject to surgery for PDAC; ii) had no prior chemotherapy before surgery (in order to eliminate its influence on transcript levels); and iii) pathologically confirmed PDAC diagnosis. Patients treated with preoperative chemo (radio) therapy were excluded from the study. The resection specimens from these patients were immediately transferred from the operating theater to the Pathology Department, macrodissected to differentiate tumor and paired non-neoplastic (control) tissues and then snap-frozen in liquid nitrogen. The histologically verified samples of tumors and control tissues were then stored at −80°C until RNA extraction. Histological diagnosis of PDAC was performed according to the standard classification. The clinical data including age, sex, date of diagnosis, tumor localization, pTNM stage, the histological type and grade of the tumor, resection margin status, lymphatic, vascular and perineural invasion, adjuvant chemotherapy based on gemcitabine or 5-fluorouracil (n=17 untreated) and OS were all obtained from medical records during regular hospital-based follow-up (Table II).

Study protocol was approved by the Ethical Commission of the Medical Faculty and Teaching Hospital in Pilsen (Pilsen, Czech Republic) (approval no. IGA 10230-3, 2nd September 2008) and the Institutional Review Boards of the Institute of Clinical and Experimental Medicine (Prague, Czech Republic) and the University Hospital Brno (Brno, Czech Republic) (approval received in the process of application of research project. no. GA CR P304/10/0338, 4th May 2009). Written informed consent was obtained from all individual participants included in the study.

Total RNA was isolated according to the procedure published elsewhere (30). Briefly, fresh frozen tumor and control tissues (~2×2×2 mm blocks) were first homogenized by mechanical disruption using a Precellys instrument (Bertin Technologies SAS; CNIM Group) at a speed of 6,500 rpm for 15 sec at room temperature. Total RNA was isolated from all samples using Trizol^® (Invitrogen; Thermo Fisher Scientific Inc.) according to the manufacturer's instructions and stored in 20 µl aliquots at −80°C. The RNA quantity was assessed in duplicates by Quant-iT RiboGreen RNA Assay kit (Invitrogen; Thermo Fisher Scientific Inc.) using the Infinite M200 multimode reader (Tecan Group, Ltd.). The quality was assessed by measurement of RNA Integrity Number (RIN) using Agilent 2100 Bioanalyzer and Agilent RNA 6000 Nano Assay Kit (Agilent Technologies, Inc.) and samples with RIN ≥3 were used for analysis. For cDNA synthesis, 0.5 µg of the isolated RNA was used with the RevertAid™ First Strand cDNA Synthesis kit (Thermo Fisher Scientific Inc.). The reaction was incubated 60 min at 42°C and terminated by heating at 70°C for 5 min. PCR amplification of ubiquitin C discriminating between product from cDNA (190 bp) and from genomic DNA (1,009 bp) was used for the cDNA quality check in terms of DNA contamination as described before (31). All cDNA samples that were free of DNA contamination (absence of 1,009 bp band in sample incubated without reverse transcriptase) were further analyzed.

RT-qPCR was performed using the LightCycler^® 96 System (Roche Diagnostics GmbH). The reaction contained 2.5 µl of cDNA samples (diluted 10-times) and 7.5 µl of kit composed of 0.5 µl of TaqMan™ Gene Expression Assay (20X), 2 µl of 5X Hot FirePol Probe qPCR Mix Plus (ROX) and 5 µl of RNAse free water (all Thermo Fisher Scientific Inc.). The thermocycling conditions used were as follows: 50°C for 2 min, denaturation at 95°C for 10 min followed by 55 cycles of denaturation at 95°C for 15 sec and annealing/extension at 60°C for 1 min. Mitochondrial ribosomal protein L19 (MRPL19), Polymerase II RNA subunit A (POLR2A) and Proteasome 26S subunit ATPase 4 (PSMC4) were used as reference genes specific for studies of human CRC and Eukaryotic translation initiation factor 2B (EIF2B1), Eukaryotic translation initiation factor 1 (ELF1), MRPL19 and POP4 homolog, ribonuclease P/MRP subunit (POP4) for PDAC based on our previously published data (26,30). Primers and probes for RT-qPCR were part of commercially provided TaqMan™ Gene Expression Assays (Thermo Fisher Scientific Inc.). The list of genes and TaqMan™ Gene Expression Assays used in the study were listed in Table SI. The non-template control contained water instead of cDNA. Negative cDNA synthesis controls (RNA transcribed without reverse transcriptase) were also employed to demonstrate possible carry-over contamination. Each sample was assayed in duplicate and the mean value used for calculations. Samples with a variation >0.5 Ct (cycle threshold) were reanalyzed. The efficiencies of all assays were between 90 and 100% and calibration curves had R²≥0.998. Transcript levels were analyzed by Roche LightCycler^® 96 System Software. Ratio of Ct of an arithmetic mean of Ct of all reference genes to a particular target gene was calculated for each sample and used for statistical evaluation (32). The qPCR study adhered to the MIQE Guidelines (Minimum Information for Publication of Quantitative Real-Time PCR Experiments) (33).

Each sample was assayed in duplicate and the mean values were used for statistics. Ct ratio which was an arithmetic mean of Ct for all reference genes to a particular target gene was calculated for each sample. Gene expression levels are presented as mean ± SD (standard deviation) of this ratio (Fig. 1). Statistical analysis was performed using SPSS v.16.0 (SPSS, Inc.) as previously described (21,22). Briefly, distribution of gene expression data was evaluated by the Kolmogorov-Smirnov test. Gene expression data did not follow a normal distribution, and hence, non-parametric tests (the Wilcoxon signed-rank test, the Mann-Whitney test and the Kruskal-Wallis test) were used for evaluation of differences between groups of patients divided by the clinical data. Differences of gene expression levels in patients divided by sex, tumor localization and size, lymph node and distant metastasis, pathological stage and grade, resection margins and response to therapy were evaluated by Mann-Whitney and Kruskal-Wallis tests. The Spearman's rank correlation test was used for assessment of correlations between continuous variables as patient age and gene expression levels. Differences in gene expression levels between tumor and control samples were evaluated by the Wilcoxon signed-rank test. Receiver operator curve (ROC) analysis was performed for evaluation of the power of expression biomarkers to discriminate the patients with different survival functions. To divide the patients into low and high expressing groups, gene expression levels were divided by the median and cutoff values calculated using the ROC analysis performed using ROC Plotter (34). Median cutoff values were the following: 0.990, 1.160 and 1.020 for PRC1, KIF14 and CIT DFI analysis in CRC tumors, respectively; 0.990, 1.157 and 1.015 for PRC1, KIF14 and CIT OS analysis in CRC tumors, respectively and 1.083, 1.221 and 1.052 for PRC1, KIF14 and CIT OS analysis in PDAC tumors, respectively. Cutoff values based on the ROC analysis were the following: 1.000, 1.170 and 1.020 for PRC1, KIF14 and CIT OS analysis in CRC tumors, respectively and 1.084, 1.204 and 1.044 for PRC1, KIF14 and CIT OS analysis in PDAC tumors, respectively. Survival function was plotted by the Kaplan-Meier method and the log rank test used for survival comparisons between groups of patients. To provide better estimates of survival probabilities and cumulative hazard, multivariate analysis was performed using Cox regression adjusted to stage. The correction for false discovery rate (FDR) was applied according to Benjamini and Hochberg (35). All P-values are from two-sided tests. P<0.05 was considered to indicate a statistically significant difference. Basic data about protein expression of PRC1, KIF14 and CIT in CRC and PDAC were extracted from The Human Protein Atlas (https://www.proteinatlas.org).

First, transcript expression levels of cytokinesis genes were assessed in tumor and control tissues from patients with CRC by RT-qPCR. PRC1, KIF14 and CIT were significantly upregulated in tumors compared with paired control tissue samples (PRC1, P=2.4×10⁻¹⁰; KIF14, P=3.6×10⁻¹⁰; CIT, P=2.4×10⁻¹⁰; Wilcoxon signed-rank test; Fig. 1A). KIF14 and CIT expression was highly significantly correlated in tumor tissues (P=3.0×10⁻⁰⁹, correlation coefficient ρ=0.708; Spearman correlation test) while correlations between PRC1 and KIF14 and PRC1 and CIT were weaker (P=4.2×E⁻⁰⁵, ρ=0.532 and P=3.2×E⁻⁰⁵, ρ=0.538, respectively; Spearman correlation test). In control tissues, correlation between PRC1 and KIF14 was stronger (P=3.7×10⁻⁰⁷, ρ=0.633; Spearman correlation test) compared with the rest of the correlations (P=0.005, ρ=0.378 for PRC1 with CIT and P=2.1×E⁻⁰⁴; ρ=0.487 for KIF14 with CIT; Spearman correlation test). All associations passed the FDR test for multiple testing (q=0.016). Data about protein expression from The Human Protein Atlas demonstrated that in general, PRC1 and KIF14 are overexpressed in colon or rectal carcinomas compared to normal tissues. On the other hand, CIT was downregulated in tumors (Table SII). The above results suggested that all cytokinesis genes were significantly upregulated in colorectal carcinomas compared to control tissues and that their levels mutually correlated.

Then, PRC1, KIF14 and CIT levels in tumor samples were compared in groups of patients stratified by clinical data and survival functions. No significant associations between intra-tumoral expression of PRC1, KIF14 and CIT and age, sex, primary tumor size, location and invasive depth (pT), regional lymph node involvement (pN), distant metastasis (cM), grade or localization of tumor (colon, sigma, or rectum) were observed (Table III). KIF14 expression significantly differed between patients stratified by the disease stage (P=0.020; Kruskal-Wallis test; Table III). However, the lack of trend between all the compared stages suggested that this association was not clinically relevant (Table III). In addition, it did not pass the FDR correction for multiple testing (q=0.002). No association between gene expression and response of patients to palliative therapy (progressive disease vs partial response or stable disease, group B, n=30) was found (Table III). Similarly, transcript expression levels of PRC1, KIF14 and CIT did not associate with DFI of subgroup of patients treated with adjuvant chemotherapy (group A, n=34) or with OS of all patients (both groups together) as evaluated by the gene expression median (Fig. 2) or cutoff using the ROC analysis (Fig. S1, for OS only). Multivariate analysis by Cox regression adjusted to disease stage also failed to show significant associations between intra-tumoral expression levels of PRC1, KIF14 and CIT and survival of patients with CRC (Table IV). Hence, gene expression levels were not associated with survival time in both univariate and multivariate analyses. Taken together, these analyses suggested a lack of significant associations between intra-tumoral expression of cytokinesis genes and clinical data of patients with CRC. Transcript expression of these genes was not associated with prognosis of patients assessed by DFI and OS.

Expression levels of cytokinesis genes were assessed in tumor and control tissues from patients with PDAC by RT-qPCR. PRC1, KIF14 and CIT were significantly upregulated in tumors compared with paired control tissue samples (PRC1, P=3.1×10⁻⁰⁵; KIF14, P=3.4×10⁻⁰⁶; CIT, P=5.2×10⁻⁰⁵; Wilcoxon signed-rank test; Fig. 1B). PRC1 and KIF14 expression was highly significantly correlated in tumor tissues (P=1.1×10⁻¹¹, ρ=0.852; Spearman correlation test) while correlations between KIF14 and CIT and PRC1 and CIT were weaker (P=0.004, ρ=0.456 and P=0.001, ρ=0.513, respectively; Spearman correlation test). All genes correlated together in the same way in control tissues. However, correlations were of lower significance (P=8.5×10⁻⁰⁴, ρ=0.519 for PRC1 with KIF14; P=0.004, ρ=0.461, for PRC1 with CIT and P=0.046, ρ=0.326, for KIF14 with CIT; Spearman correlation test) compared with tumors. All associations except correlation of KIF14 with CIT in control tissues passed the FDR test for multiple testing (q=0.016). Data about protein expression from The Human Protein Atlas demonstrated that in general, PRC1, KIF14 and CIT were overexpressed in pancreatic carcinomas compared to normal tissues (Table SII). The above analyses suggested that all cytokinesis genes were significantly upregulated in pancreatic carcinomas compared with control tissues and that their levels mutually correlated.

PRC1, KIF14 and CIT levels in tumor samples were compared in groups of patients stratified by clinical data and survival functions. No significant associations between intra-tumoral expression of PRC1, KIF14 and CIT and individual clinical data of patients age, sex, tumor localization (head versus body or tail), pT, pN, stage, grade, angioinvasion and perineural invasion or resection margins (R0 versus R1) were observed (Table V). Expression of PRC1 or KIF14 did not associate with OS of all patients as evaluated by the gene expression median (Fig. S2A) or cutoff using the ROC analysis (Fig. S3). Notably, patients with lower expression of CIT had significantly worse OS compared with patients with higher expression level (P<0.05 for median and P<0.05 for the ROC analysis cut offs; log rank test; Fig. 3), but this result did not pass the FDR test (q=0.016). The OS of the subgroup of patients with PDAC treated with adjuvant chemotherapy (n=31) was not significantly associated with expression of PRC1, KIF14 and CIT (Fig. S2B). No significant association of expression levels with OS was observed in multivariate analyses using the Cox regression adjusted to stage (Table VI). Hence, gene expression levels were not associated with survival time in both univariate and multivariate analyses. Taken together, these analyses suggested a lack of significant associations between intra-tumoral expression of cytokinesis genes and clinical data of patients with PDAC. Transcript expression of these genes were not associated with prognosis of patients assessed by OS.