The National Institutes of Health Biomarkers Definition Working Group defined a biomarker as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention”[38].

This definition of “a characteristic that is objectively measured and evaluated” implicates that molecular biomarkers which might be useful for prediction of the success of chemotherapy in cancer patients should (1) be easily detectable and assessable in clinical patient samples; and (2) be stably expressed and refractory to degradation in these samples in order to allow proper analysis on clinical samples. Only if these characteristics are present, miRNAs can further be assessed regarding their potential as biomarkers for response prediction.

And in fact, regarding the aspect of easily accessible clinical samples, there is a large body of evidence that shows that miRNAs are indeed detectable in a number of different sample types. For example, miRNAs can be found and analyzed in fresh frozen samples such as tumor biopsies or resection specimens, or in paraffin-embedded tissues[39]. Even more interesting from a clinical point of view, miRNAs can be detected as so-called “circulating miRNAs” in a broad variety of human body fluids including urine[23], saliva[39], amniotic fluid[39] and pleura fluid[40] in healthy volunteers and cancer patients[25,40-43]. These circulating miRNAs are considered believed to originate either from immunocytes[44], as byproducts of dead/dying cells[44-46], or to be secreted actively from tumor cells via microvesicles such as exosomes[44-47].

Regarding the stability of miRNA expression in clinical samples, Lu et al[48] reported that miRNAs are quite stable in different tissue samples (paraffin-embedded sections, rapid frozen samples) and present unique expression levels in different tissue types (salvia, blood, urine). Moreover, circulating miRNAs have been shown to be highly stable in the peripheral circulation caused for example by binding to RNA-binding proteins[18,49-51] or by their excretion in microvesicles[51-54]. In addition, Fang et al[41] reported that miRNAs present tissue-specific expression patterns with a reproducible and consistent expression. Finally, studies demonstrated that circulating miRNAs are fairly resistant to different pH solutions or repeated freeze-thawing cycles, and they are stable for 24 h at room temperature[55,56].

Taken together, miRNA expression patterns can successfully be assessed and analyzed in a variety of different samples, including biopsy or resection specimen, or blood samples, what supports the hypothesis that miRNAs could be used as clinically relevant diagnostic or therapeutic molecular biomarkers.

Another aspect of using miRNAs as potential biomarkers for chemoresistance in GC includes their possible impact on MDR in general. Only if there are data available which support that miRNAs modulate drug resistance in general, these molecules might be useful for evaluation of response prediction in GC patients.

And in fact, miRNAs have been proposed to play an important role in the development of MDR in cancer progression[55-58]. Despite the fact that this field of research is still in its infancy, it has been demonstrated that miRNAs are highly related to cancer progression (including growth apoptosis, invasion and metastasis) and are responsible for cancer-related inflammation, anticancer drug resistance and regulation of cancer stem cells[59]. Most interestingly, a number of miRNAs (such as members of the let-7 family, miR-16, miR-21 or miR-451 amongst others) have been confirmed to impact on more than one anticancer drug and/or to play a role in chemotherapy resistance in more than just one tumor type[57,60-62]. In addition, several anti-cancer-drugs target the same cancer-related genes that are targeted and regulated by miRNAs, implicating a link between chemotherapy induced cancer cell death and a modulation of chemotherapy resistance via miRNAs[63-65]. However, it has to be noted in this respect that the regulation of known cancer-related genes by miRNAs is much more frequently reported simply because these genes are well studied, and not all miRNAs show similar sensitivity phenotypes in different cancer types. Finally, a number of transcripts from drug resistance-related genes may be targeted by more than one miRNA[66].

Taken together, the evidence available so far on miRNA analysis in different sample types, their stability in various tissues, and their involvement in chemotherapy resistance in general highly supports the hypothesis that miRNAs might be perfect candidates for biomarkers for chemoresistance in GC in clinical settings.

As miRNA research is so far still mostly limited to in vitro experiments, its clinical applicability at this stage is somewhat limited, and there are a number of hurdles that need to be overcome before these molecules can be used in standard clinical settings.

With regards to sample acquisition and miRNA extraction for examples from blood samples, it has been demonstrated that several aspects and factors impact on results of miRNA analyses. These include the appropriate selection of samples (e.g., plasma or serum), collection tubes (EDTA, citrate, heparin), extraction methods (phenol/chloroform, silica: distinct differences in required fluid volume, yield, procedural contaminants etc.), quality and quantity control, fasting or blood draw timing, all potentially. In addition, there are a number of different miRNA profiling methods including qRT-PCR, microarrays, sequence specific hybridization in solution followed by miRNA molecule counting based on reporter probes, and direct sequencing available, all of which with more or less relevant advantages and disadvantages regarding sensitivity and specificity, absolute quantification/accuracy and flexibility and throughput[67]. In general, it can be postulated, that there is a lack of standardized of experimental techniques at this stage, and this impacts clearly on the possibility to compare data of different studies.

With regards to miRNAs as therapeutic tools, in vivo data on animal experiments are limited, and there are to our best knowledge so far only two clinical studies on a potential use of miRNAs as therapeutic tools in human beings available or on its way. Both studies do not refer to GC patients. In a phase 2a clinical trial, Janssen et al[68] evaluated the safety and efficacy of Miravirsen^® (a locked nucleic acid-modified DNA phosphorothioate antisense oligonucleotide that sequesters mature miR-122 in a highly stable heteroduplex, thereby inhibiting its function) in 36 patients with chronic HCV genotype 1 infection. The authors could demonstrate prolonged dose-dependent reductions in HCV RNA levels without evidence of viral resistance and without dose-limiting adverse events and no escape mutations in the miR-122 binding sites of the HCV genome[68]. It is worth to mention that in two phase I safety studies in humans with Miravirsen^® showed that Miravirsen^® was well tolerated with no dose-limiting toxicities[69,70].

The second clinical trial (phase I trial), which is currently recruiting healthy patients as a control group, uses MRX34 (a liposome-formulated mimic of the tumor suppressor miR-34) in patients with inoperable primary liver cancer or metastatic cancer with liver involvement[71]. In this trail, the safety, pharmacokinetics and pharmacodynamics MRX34 is being evaluated in healthy patients[72].

Taken together, there are many hurdles that need to be overcome before miRNAs can be safely applied as clinical biomarkers or therapeutic tools. Standardization of technical approaches, confirmation of in vitro results in animal experiments and finally safety assessment and treatment trials in humans have to show whether these molecules find their way into the daily clinic. However, first results are promising indeed.

The potential to predict response to chemotherapy treatment is a promising clinical application of miRNAs. Successful prediction of treatment response would allow a more tailored, individual therapy planning, as patients who do not respond to this treatment modality would not have to undergo futile treatment with potential severe side-effects. However, response prediction of chemoresistance implies that miRNAs exhibit significant and reproducible differences in expression between patients that respond well to chemotherapy compared to patients that show no or minimal response to this treatment option. With other words, miRNA expression pattern should differ between chemotherapy resistant GC tumors and sensitive tumors.

And in fact, some first authors reported results from in vitro experiments that showed differing expression pattern of several miRNAs in chemotherapy resistant GC cell lines when compared to sensitive controls (Table 1), with some miRNAs being upregulated in resistant cancer cells, and others being downregulated. For example, Wang et al[77] demonstrated an upregulation of miR-19a/b in a vincristine resistant GC cell line (SCG7901/VCR) and a doxorubicin resistant GC cell line (SGC7901/ADR). In addition, Yang et al[78] reported that miR-21 was upregulated in cisplatin resistant GC cells (SCG7901/DDP) vs controls. Furthermore, miR-106a was shown to be upregulated in a cisplatin resistant GC cell line (SCG7901/DDP)[79]. Finally, miR-195 and miR-378 were found to be upregulated in 5-azazytidine resistant GC cell lines (MGC803, SGC7901 and AGS)[80].

Xia et al[62] on the other hand showed that miR-15b and miR-16 both were downregulated in a vincristine resistant GC cell line (SGC7901), and Shang et al[81] showed that miR-508-5p was downregulated in two GC cell lines resistant towards doxorubicin (SGC7901/ADR) and vincristine (SGC7901/VCR). Furthermore, Zhu et al[82] demonstrated in vincristine resistant GC cell lines miR-181b, miR-200b/c and miR-497 to be downregulated compared to controls[83].

Finally, Wu et al[84] analyzed miRNA expression patterns in hydroxycamptothecin (HCPT)-resistant and HCPT-sensitive GC cell lines. miR-224 and miR-338-3p were only expressed in HCPT-resistant cells, and miR-141, miR-200a, miR-200b, miR-372, and miR-373 were only expressed in HCPT-sensitive cells[84].

Even more interesting from a clinical point of view, there is some first evidence that alterations in miRNA expression pattern might be able to discriminate between responders and non-responders in clinical patient samples (in vivo). Kim et al[85] for example presented data based on biopsy samples from 90 GC patients which were collected prior to chemotherapy. The authors identified a signature of several miRNAs (namely miR-363, miR-518f, miR-519e, miR-520a, and miR-520d) that was correlated to resistance to cisplatin and 5-fluorouracil therapy[85]. With regards to blood based circulating miRNAs, the current literature draws a promising picture: Zhu et al[86] showed in a recent review including 22 studies with a sample size ranging from 37 to 164, that a total of 35 circulating miRNAs were differentially expressed between GC patients and healthy controls. Most interestingly, 6 of these miRNAs (miR-21[78], -27a[87], -106a[79], -195[80], -200c[83] and -378[80]) were shown to be deregulated in chemotherapy resistant GC cell lines as demonstrated in the in vitro experiments in the paragraph above, and to be potentially involved in MDR.

The application of miRNAs as modifiers of chemotherapy sensitivity is an even more interesting approach for a potential clinical use of these molecules, especially as additive treatment with conventional chemotherapeutics. This implicates that artificial manipulation of miRNA levels lead to increased sensitivity or reduced resistance towards various chemotherapeutic drugs. And again, there is growing evidence that modulation of miRNA expression in fact affects resistance of various tumors to chemotherapeutic treatment. However, as this field of research is still very young, results on GC are limited and refer so far only to in vitro experiments. In vivo studies on human patients with GC are not available yet.

Similarly to the data presented in the section about miRNAs with diagnostic potential, some miRNAs impact on sensitivity towards chemotherapy if their levels were artificially upregulated, others if their levels were downregulated. For example, upregulation of miR-21 or miR-106a was demonstrated to increase cisplatin resistance of GC cells[78], and Deng et al[80] showed that upregulation of miR-195 or miR-378 led to an enhanced 5-azacytidine resistance in normal gastric cells. Upregulation of miR-449 or miR-508-5p was demonstrated to positively impact on sensitivity towards cisplatin (miR-449) respectively vincristine or doxorubicin (miR-508-5p)[81,88]. Interestingly, in accordance with these findings about the modulation of sensitivity towards chemotherapeutic drugs via miRNAs, Bandres et al[89] reported that upregulation of miR-451 led to an increased sensitivity of cancer cells towards radiotherapy by down-regulating the macrophage migration inhibitory factor (MIF).

Only one research group reported on the effect of miRNA downregulation on chemotherapy resistance: Zhao et al[87] found that increased doxorubicin sensitivity in GC cells is connected with downregulation of miR-27a.

The authors of the aforementioned studies performed additional experiments in order to elucidate the underlying mechanisms by which the respective miRNAs finally impacted on sensitivity and resistance towards the different chemotherapeutic drugs. Table 2 and Figure 1 present an overview about the results of these investigations, and highlight the downstream targets and pathways that mediate the effects of manipulated miRNA expression in GC cells and their response to anticancer treatment.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 MicroRNAs and their influence on multidrug resistance in gastric cancer. The figure presents an overview about the targets and pathways that mediate the effects of miRNA manipulation on multidrug resistance in gastric cancer (Data modified from DIANA miRPathv2.0[90]). The arrows indicate the up- or down-regulation of miRNA patterns in gastric cancer cells in vitro. The dashed boxes indicate nucleus related targets. Interrogation marks represent to date unknown intermediate steps. STAT: Signal transducers and activators of transcription; mTOR: Mammalian target of rapamycin; PKB: Protein kinase B.