Gastric cancer rates as the fourth most common malignancy worldwide and is the third most common cause of death from a malignant disease, after lung and liver cancer (1). Eastern Asia accounts for more than 50% of cases registered globally. A gradual decrease in the number of cases has been observed in regions such as the United States, Western and Central Europe (2,3).

In 2012, the incidence of gastric cancer in the Czech Republic was 14.6 cases per 100,000 inhabitants. Most cases are diagnosed in the late stages of the disease and only palliative treatment remains possible. Chemotherapy is often embarked upon as part of palliative treatment and most patients receive the same medications, overall survival varies greatly from patient to patient, however. One possible cause of this variability could be the gene expression changes occurring in cancer tissue, which may alter the effect of cytostatics. Evaluating the expression of specific genes including genes for microRNA (miRNA) could help single out chemotherapy efficacy predictors in gastric cancer patients undergoing palliative treatment (4,5). By identifying patients with chemoresistant tumors, we hope to spare them the strain of inefficient chemotherapy.

The tissue samples were the same as those used by the pathologists when making the diagnoses, to measure the levels of chemotherapy response predictors. The use of efficacy predictors makes up an important part of the development of new targeted therapy drugs and their introduction into clinical practice; this is no less the case for conventional chemotherapeutics.

The ability of cancer cells to overcome the effects of chemotherapy by changing the expression of repair genes, enzymes partaking in nucleic acid metabolism and genes involved in apoptosis limits the success of chemotherapy. Eukaryotic cells react to DNA damage by activating repair mechanisms with these key functions: DNA damage detection, stopping replication of damaged DNA, repair of damaged area if possible or else the induction of apoptosis. Nucleotide excision repair (NER), base excision repair (BER), mismatch repair (MMR) and homologous recombination repair (HRR) belong to the most important repair mechanisms of eukaryotic cells (6,7). Damage to these mechanisms means that mutations are passed on to the next cell generation. Impaired function of DNA repair mechanisms is thus linked to both ageing and cancerogenesis. However, increased activity of these mechanisms can hinder chemotherapy by forestalling further effective damage to DNA and thereby preventing the activation of apoptosis in proliferating tumor cells (8).

Cisplatin acts cytotoxically by creating adducts, which participate in crosslinking DNA, and in so doing activates programmed cell death. Decrease in the scale of damage to DNA, whether as a result of fewer adducts being created (in lower drug dosage, for instance) or because of their repair, can lead to a decrease of efficacy of this chemotherapeutic substance (9).

5-Fluorouracil (5-FU) is one of the most commonly used chemotherapeutics in gastric cancer treatment regimens (10). The primary site of action for 5-FU is thymidylate synthase (TS), an essential enzyme in de novo biosynthetic pathway of deoxythymidylate (dTMP). Thymidylate synthase catalyzes the reductive methylation of dUMP (deoxyuridine-5-prime monophosphate or deoxyuridylate) to dTMP (deoxythymidine-5-prime monophosphate or deoxythymidylate) using 5,10-methylenetetrahydrofolate as a cofactor. Maintaining a dTMP pool is crucial for DNA replication and repair.

We measured the levels of mRNA of the excision repair cross-complementary group 1 gene (ERCC1), ribonucleotide reductase subunit M1 gene (RRM1), breast cancer 1 gene (BRCA1) and TS, all of which participate in repair to damaged DNA. We determined their relationship to time to progression (TTP), using the definition of progression according to RECIST (response evaluation criteria in solid tumours) criteria (11), and overall survival (OS). Table I lists the basic characteristics of the genes of interest.

In addition to genes whose products are directly involved in DNA repair and nucleic acid metabolism, we also focused on gene products that seem to affect the expression of genes involved in the above mentioned processes. MicroRNAs (also known as miRNAs or miRs), small non-coding RNA molecules 18–23 nucleotides in length, make up an immense group of regulatory molecules involved in carcinogenesis. The human genome may encode over 2,500 miRNAs, which may target ~60% of mammalian genes and are abundant in many human cell types (see miRNA database available online at www.mirbase.org). MicroRNAs participate in the post-transcriptional regulation of gene expression controlling development and maintain diverse cellular processes including proliferation, apoptosis, differentiation, motility and morphogenesis.

The effect of microRNA regulatory networks in cancer tissue can be oncogenic (by targeting tumor suppressor genes) or tumor-suppressive (by post-transcriptional repression of oncogenes) (12). However, the final effect of any particular miRNA is time and tissue dependent (13).

Many studies have described changes in expression of miRNAs and their involvement in carcinogenesis, tumor progression, invasion, metastasis and the effects of treatment in gastric cancer tissue. MicroRNAs may become valuable diagnostic markers and therapeutic targets for gastric cancer (14). Based on our research of published literature, we chose 29 miRNAs, which have a potential role in carcinogenesis or drug metabolism and therefore could be expected to influence the efficacy of treatment. A list of the main characteristics of miRNAs of interest is shown in Table II.

Prior to our analysis of the relationship between gene expression and TTP and OS, we compared gene expression in cancer and adjacent non-cancer epithelial gastric tissue. We found the expression of miR-221 in cancer tissue to be significantly higher, while the expression of miR-202 and miR-509 proved to be lower compared to healthy tissue (P-value 0.013, 0.011 and 0.018, respectively). However, when drawing conclusions from this analysis, we have to take into account the low number of paired samples, caused by the lack of non-cancerous tissue in some of the FFPE samples.

The Cox regression hazard model was used to determine the relation of marker level to OS or TTP. Results for all markers are summarized in Table III. From the set of the 29 microRNAs of interest, we found high expression levels of miR-150, miR 342-3p, miR-181b, miR-221, miR-224 and low levels of miR-520h related to shorter TTP. High levels of miR-150, miR-192, miR-224, miR-375 and miR-342-3p related to shorter OS. For markers with statistically significant results, optimal cut off values were chosen and Kaplan-Meier survival distribution functions for OS and TPP were generated.

The treatment regimen of all patients included 5-FU. We noted a definite correlation between high levels of miR-150 and miR-342-p in cancerous tissue and shorter TTP as well as OS. High expression of miR-224, however, only proved to have a relation to shorter OS (Table IV and Fig. 1).

In the subgroup of patients receiving 5-FU as the only treatment, we recorded a relation between high levels of miR-181b and shorter TTP and between high levels of miR-150, miR-192 and miR-342-p and shorter OS. In the subgroup treated with both 5-FU and cisplatin, we noted that high levels of miR-221, miR-224 and low levels of miR-520 related to shorter TTP and high levels of miR-221, miR-224 and miR-375 to shorter OS (Table V and Fig. 2).

The present study focused on patients in advanced stages of gastric cancer. Therefore, these patients could not have the tumors surgically removed and underwent palliative treatment only. The main clinical concern in such cases is deciding which chemotherapeutic regimen is indicated. The question was whether we could predict the effect of chemotherapy and thereby determine, if the aggressive chemotherapeutic treatment, which inevitably decreases quality of life, would prolong survival. If a low effect of treatment is predicted, it would be appropriate to offer to those patients inclusion in new ongoing studies.

In our laboratory assessment we used tissue samples taken gastroscopically for routine diagnostic purposes and macrodissected a sample of cancerous tissue, verified by a pathologist from FFPE sections, to analyse the expression of genes influencing the effects of therapy. This approach can be easily translated into clinical practice. We conducted expression analysis from a large number of tumor cells. Cancerous tissue is heterogeneous by nature, and originates in the process of clonal evolution, and therefore examining the collective changes in expression of a variety of cancer cell types (gathered by macrodissection) provides a more complex insight regarding prognosis. We set out to ascertain the prognostic potential of chosen miRNAs, which influence apoptosis and cell proliferation and in so doing interact with the mechanism of chemotherapy indicated in the cases we examined. However, we could not leave out monitoring protein coding genes frequently investigated as possible treatment outcome predictors.

Many studies have noted the prognostic value of low ERCC1 expression in gastric cancer patients undergoing chemotherapy; a meta-analysis published in 2015 concluded ERCC1 may be a useful prognostic factor for gastric cancer and furthermore that low mRNA levels of ERCC1 appear to be associated with a significant OS benefit to patients treated with platinum-based chemotherapy (17). However, the predictive value of the ERCC1 gene for survival and response to platinum-based chemotherapy in gastric cancer remains controversial (18). We found no statistically significant relation between the levels of ERCC1 and survival. It is possible our results reflect our test group, as in all cases we examined 5-FU was part of the treatment regimen but platinum-based chemotherapy was only used in a subgroup, whose analysis was affected by the small number of patients.

The very nature of the TS enzyme makes its expression the most commonly examined in relation to 5-FU chemotherapy. We determined high levels of TS predict longer OS (Table V). Similar results were published by Wei et al (19), who conducted their analysis on a group of patients treated in the same way and also used RT-PCR to assess TS expression. On the other hand, some studies have shown high levels of TS to have the opposite effect (20).

From the set of the 29 microRNAs of interest, we found that high expression levels of miR-150, miR-181b, miR-192, miR-221, miR-224, miR 342-3p, miR-375 and low expression of miR-520h relate to unfavourable outcomes for patients (shorter TTP or OS) Table III.

MicroRNAs have the potential to become accurate, easily measurable biomarkers, with features fortuitous for diagnostic testing methods, such as stability in FFPE tissue, blood and perhaps other bodily fluids (21).

Wu et al (22) found miR-150 was overexpressed in gastric cancer cell lines and tissue samples and demonstrated overexpression of miR-150 could promote proliferation and growth of cancer cells by targeting the tumor-suppressor EGR2. In undifferentiated gastric cancer, higher miR-150 levels appeared to be associated with shorter postoperative patient survival, however, miR-150 was deemed to be an insufficiently independent prognostic factor in these cases (23). In the study of Chen et al (24), miR-150 showed decreased expression in gastric cancer patients compared to healthy test subjects. In the present study, higher levels of miR-150 showed a relation to shorter TTP and OS (Table IV; Figs. 1A and B and 2A).

Compared to normal gastric tissue samples, there is an overexpression of miR-181b in gastric tumors. Lower levels of miR-181b relate to longer OS of patients on regimens based on 5-FU and platinum derivatives (25). Furthermore, overexpression of miR-181b was found to downregulate the tissue inhibitor of metalloproteinases 3 (TIMP3) (26). Our results hint at the association of higher miR-181b levels to shorter TTP of patients treated with 5-FU (Table V). The results of another study show the ambiguous nature of the effects of certain miRNA; Chen et al (27) observed miR-181b was downregulated in human gastric adenocarcinoma tissue samples compared to adjacent normal gastric tissue and also described how miR-181b could suppress tumor cell proliferation by downregulating the expression of cAMP responsive element binding protein 1 (CREB1).

Our analysis of miR-192 levels showed its high expression related to shorter OS in the group of patients treated with 5-FU (Table V and Fig. 2B). To the best of our knowledge, no other study dealing with the predictive value of miR-192 in gastric cancer patients treated with 5-FU has been published. Xu et al (28) found miR-192 to be upregulated in gastric cancer tissue samples obtained by gastrectomy. The upregulation of both miR-192 and miR-215 was related to clinical characteristic such as lymph node metastases, while the inhibition of miR-215 or miR-192 significantly decreased gastric cancer cell invasion. The results reported by Chen et al (29) demonstrate that elevated circulating miR-192 has the potential to improve the early detection of distant metastases of GC.

Recently published studies show that miR-221 is an oncogenic microRNA involved in several malignancies (30,31). We found higher miR-221 expression in tumor samples in comparison to adjacent noncancerous tissue. This is in accordance with Liu et al (32) who found miR-221 was upregulated in 88% of gastric cancer tissue samples. Moreover, we observed a relation of high miR-221 expression to shorter TTP and OS in 5-FU monotherapy treated patients (Table V). The influence of miR-221 on the effect of chemotherapy is corroborated by published experiments conducted on the human gastric cancer cell line SGC7901 showing the knockdown of miR-221 inhibited cell growth and invasion and increased the radiosensitivity of the cells (33).

We found higher levels of miR-224 indicate shorter OS (Table IV and Fig. 1D). Mao et al (34) concluded that miR-224 is overexpressed in human gastric cancer cells. Reducing the expression of miR-224 can effectively inhibit growth and promote apoptosis of gastric cancer cells. These results are also supported by the study of Liu et al (35) who investigated the expression of miR-224 in the human gastric cancer cell line SGC-7901. In examining the effects of miR-224 mimics, they observed miR-224 could negatively regulate the expression of Raf-1 kinase inhibitor protein (RKIP). RKIP contributes to the suppression of proliferation and invasion of gastric cells.

We determined higher levels of miR-342-3p correlated to shorter TTP and OS in 5-FU monotherapy treated patients (Table V and Fig. 2C); similar results were described in colorectal cancer. High levels of miR-342-3p were associated with shorter survival time (36). Kim et al (37) screened miRNAs associated with response to chemotherapy using microarrays and found miR-342-p belongs to the miRNAs, whose upregulation is associated with chemosensitivity in gastric cancer.

Our observations of the relation of high miR-375 expression to shorter OS in 5-FU monotherapy treated patients (Table V) could be explained by the findings of Liu et al (38), who showed that miR-375 downregulated p53 expression through an interaction with the 3′ UTR region of p53. In addition, they observed the expression of miR-375 desensitized cells to ionizing radiation and etoposide.

We demonstrated that higher levels of miR-520h correlated to longer TTP in 5-FU and cisplatin therapy treated patients (Table V). Shen et al (39) found that miR-520h downregulates histone deacetylase 1 and, thus, contributes to the chemotherapeutic effect of doxorubicin.

Summarizing the aforelisted results, amongst the miRNAs we examined, we found six miRNAs (miR-150, miR-181, miR-221, miR-224, miR-342-p and miR-520h) with a relation to TTP, which could serve as predictors of the effectiveness of treatment. These results merit multifactorial analysis, which we were, however, unable to perform due to the limited number of samples.

In our experience, microRNAs can generally be assessed with more precision and ease than mRNA of coding genes. This is essentially due to the fact that miRNA analysis is less demanding in terms of both quality and quantity of isolated RNA, features problematic in samples of RNA extracted from FFPE tissue. FFPE tissue samples are routinely taken and analysed during standard gastric cancer management and that is why we believe microRNAs could become clinically applicable predictors of the effectiveness of palliative treatment in gastric cancer patients.