miR-30a-5p together with miR-210-3p as a promising biomarker for non-small cell lung cancer: A preliminary study

Abstract

BACKGROUND:

Although an immense effort has been made to develop novel diagnostic methods and treatment strategies for non-small cell lung cancer (NSCLC), the survival rate of this disease has remained virtually unchanged. Small non-coding RNAs called microRNAs (miRNAs) have appeared to be very promising biomarkers of cancer including NSCLC.

OBJECTIVE:

We investigated the expression level of six miRNAs, and subsequently we evaluated their diagnostic ability and their clinical significance.

METHODS:

We performed an analysis in 50 paired cancer and non-cancerous lung tissue samples collected from NSCLC patients. The RT-qPCR technique was used to investigate the expression profile.

RESULTS:

Obtained results indicate that miR-30a-5p, miR-126-3p and miR-486-5p are downregulated, while miR-205-5p and miR-210-3p are upregulated in NSCLC tissue. Moreover, performed stepwise discriminant analysis determined the model including miR-30a-5p and miR-210-3p which tested on the test set ( $n=$ 30) revealed an AUC of 0.969 and provided 100% sensitivity and 80% specificity in discriminating NSCLC tissue from non-cancerous lung tissue.

CONCLUSIONS:

The present preliminary study demonstrated that five tested miRNAs were deregulated in cancer tissue. Moreover, miR-30a-5p together with miR-210-3p with excellent sensitivity and acceptable specificity may distinguish cancer tissue form non-cancerous tissue and thus may become a potential diagnostic biomarker for NSCLC.

1. Introduction

Lung cancer incessantly remains not only the most common (around 13% of cancer incidences) but also the most deadly type of cancer in the world. It is responsible for more than 1.5 million deaths per year [50]. Most frequently, diagnosis of lung cancer is done at the advance stage of tumour development, when metastasis have already occurred and when surgical intervention cannot be done. Around 70% of lung cancers are unresectable, that is why small biopsy and cytology specimens are the primary method of diagnosis [51]. Late symptoms of lung cancer are the main reason of the overdue diagnosis and it is known that the advance stage of cancer is correlated with unfavourable prognosis and with very poor chance for recovery. Development of an early detection program is crucial for effective fighting with this disease and will allow to apply and perform treatment at the time when it can bring the biggest benefits. It may critically improve overall five-year survival rate which for NSCLC patients remains less than 15% [43, 8].

NSCLC is heterogeneous group comprised of subtypes: squamous cell carcinoma (SCC) and adenocarcinoma (AC) and a lesser extent large cell carcinoma (LCC). Rising number of data confirm the highly complex histological nature and intricate molecular pattern of lung cancer which can differ even within the same histological type [61].

With an advent of personalized therapies, histological subclassification is becoming essential to provide effective treatment for patients with NSCLC. The substantial differences in clinical tumor behaviour and management, depends on different histology and molecular pattern of the tumour. For example patients with SCC should not be treated with bevacizumab because of an increased risk of life-threatening haemorrhage [51]. Therefore, finding a reliable diagnostic tumour marker, not only would reduce the histologic misclassification of small biopsy specimens but also would significantly contribute to lowering lung cancer mortality.

In the diagnostic medicine, a novel and very promising class of RNA molecules called microRNA (miRNA) turn out. These single-stranded, 22–24 nucleotides in length molecules are non-coding, fundamental regulators of gene expression. They act by binding to the 3’-UTR (untranslated region) of target transcripts and initiate translational repression or mRNA degradation [14]. The increasing body of evidences have indicated that dysregulation of miRNA expression may be a key factor underlying tumorigenesis [35]. miRNAs can exert impact on signalling, differentiation, growth, transformation and other cellular processes [12, 27, 55]. Genes coding miRNA are located within introns or exons of protein-coding genes and in intergenic regions. More than 50% of genes of the known miRNAs are placed in areas of genome which are common breakpoints, highly associated with cancer like: minimal regions of loss of heterozygosity, fragile sites or close to it and regions of amplification [5]. For example miR-30a-5p is located on the chromosome 6q13, which has been reported to be a genomic fragile region, linked to the risk of lung cancer development [19], miR-205 associated with the 1q32.2 regions is frequently amplified in AC lung cancer, whereas miR-126 with the 9q34.3 region commonly deleted in lung cancer [3]. The dysregulation of miRNAs can be implicated in the initiation and progression of cancer. Moreover, accumulating evidences are suggesting that microRNAs can play an important role as biomarkers for various cancers [17, 18, 58], including NSCLC also at the early stage of the disease [9, 22, 45].

Researchers are incessantly reporting about abnormalities concerning various miRNAs also ones abovementioned, but diagnostic value of miRNAs in NSCLC still remains not consistent and currently no single miRNA or panel of miRNAs can be applied in screening tests. The purpose of this preliminary study was to investigate the expression profile of six chosen miRNA: miR-21-5p, miR-30a-5p, miR-126-3p, miR-205-5p, miR-210-3p and miR-486-5p in NSCLC tissues and to test their diagnostic abilities.

2. Materials and methods

2.1 Patients and samples

Fifty surgically resected human paired samples of primary NSCLC and adjacent normal lung tissue were collected during surgery in accordance with protocols approved by the committee on the Ethics of Research in Human Experimentation at the Medical University of Lodz (Lodz, Poland). Informed consent was obtained in accordance with Declaration of Helsinki. All patients belonged to the Caucasian population. Forty paired samples were provided by Regional Specialized Hospital of Tuberculosis, Lung Diseases and Rehabilitation in Tuszyn (Tuszyn, Poland) and the other 10 where obtain from Nicolaus Copernicus Memorial Provincial Specialist Hospital in Lodz (Lodz, Poland). Patients underwent surgery between November 2011 and April 2015 and none of the subjects had received any therapeutic procedures prior to this study, including surgery, chemotherapy, and radiotherapy. The tissues were representative of two histological subtypes of NSCLC: 14 adenocarcinoma and 30 squamous cell carcinoma, the other 6 samples with complex composition, were assigned to subgroup called “other”. The patient samples were classified according to the seventh edition of the TNM classification of the American Joint Committee on Cancer. The patient’s gender, age, local invasion, smoking status etc. were obtained from surgical and pathological records. The clinicopathological characteristics of patients with non-small cell lung cancer are presented in Table 1. After surgery all samples were preserved and stored at $-$ 80 ${}^{\circ }$ C until needed for use. Non-cancerous tissue was used as a control.

Table 1

Clinicopathological characteristics of patients ( $n=$ 50)

Characteristic	N (%)
Sex
Male	38 (76)
Female	12 (24)
Age (years)
$⩾$ 65	26 (52)
$<$ 65	24 (48)
Histological subtype
AC	14 (28)
SCC	30 (60)
Other	6 (12)
Pathological tumor size ${}^{*}$
T1	7 (15)
T2	28 (58)
T3	10 (21)
T4	3 (6)
Pathological lymph node metastasis ${}^{*}$
N0	24 (50)
N1	13 (27)
N2	11 (23)
Pathological stage ${}^{*}$
I	8 (17)
II	27 (56)
III	13 (27)
Smoking habit
Yes	44 (88)
No	6 (12)
Pack-years
0	6 (12)
1–20	6 (12)
21–49	26 (52)
$>$ 49	12 (24)

AC, adenocarcinoma; SCC, squamous cell carcinoma; ${}^{*}$ missing data of 2 patients.

2.2 Total RNA isolation

Isolation of total RNA from frozen cancer and non-cancerous tissues was performed using TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA). 50 mg of tissue was immersed in 1 ml of TRIzol and has undergo homogenization with a TissueRuptor (Qiagen, Hilden, Germany). The procedure was followed according to the manufacturer’s instructions. Concentration, quality and purity of RNA were qualified using Picodrop Microliter UV/Vis Spectrophotometer (Picodrop Ltd., Hinxton, UK). RNA was used for reverse transcription (RT) process directly after isolation.

2.3 Expression analysis of miRNAs by quantitative reverse-transcriptase-polymerase-chain-reaction (RT-qPCR)

To synthesize cDNA of selected miRNA the stem-loop RT primers for miRNA-TaqMan ${}^{\mathrm{\circledR }}$ MicroRNA assay – together with Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) were used in accordance with the guidelines provided by the manufacturer. 6 ng of total RNA was added to the reaction mix. The final volume of 15 $\mu$ l was incubated in a UNO-Thermoblock from Biometra (Gottingen, Germany) for 30 min at 16 ${}^{\circ }$ C and 30 min at 42 ${}^{\circ }$ C, followed the reverse transcriptase was inactivated for 5 min at 85 ${}^{\circ }$ C. All samples were held at $-$ 20 ${}^{\circ }$ C. qPCR was performed using TaqMan ${}^{\mathrm{\circledR }}$ MicroRNA Assay together with the TaqMan ${}^{\mathrm{\circledR }}$ Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions. Real-time PCR was performed using a Stratagene Mx3005P qPCR system (Agilent Technologies, Santa Clara, CA, USA). The final volume of each reaction was 20 $\mu$ l. All assays were carried out in triplicate and were incubated for 10 min at 95 ${}^{\circ }$ C, followed by 40 cycles of 15 s at 95 ${}^{\circ }$ C and 1 min at 60 ${}^{\circ }$ C. RNU6B (Applied Biosystems, Foster City, CA, USA) was used as an endogenous control to normalize the expression level of miRNA genes. The sequences and assay ID of tested miRNAs are presented in Table 2. Comparative Ct method (2 ${}^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$ method) was chosen to determine relative quantification (RQ) of miRNA [29].

Table 2

Sequences and assay ID of tested miRNA

miRNA	Assay ID	5’-3’ mature miRNA sequences
hsa-miR-21-5p	000397	UAGCUUAUCAGACUGAUGUUGA
hsa-miR-30a-5p	000417	UGUAAACAUCCUCGACUGGAAG
hsa-miR-126-3p	002228	UCGUACCGUGAGUAAUAAUGCG
hsa-miR-205-5p	000509	UCCUUCAUUCCACCGGAGUCUG
has-miR-210-3p	000512	CUGUGCGUGUGACAGCGGCUGA
hsa-miR-486-5p	001278	UCCUGUACUGAGCUGCCCCGAG
RNU6B	001093	CGCAAGGATGACACGCAAATTCGTGAAGCGTTCCATATTTTT

Table 3

Comparison of miRNAs expression in non-small cell lung cancer tissue and adjacent non-cancerous lung tissue. miRNA expression is given as - $\mathrm{\Delta }$ Ct calculated as a difference in Ct for reference miRNA and investigated miRNA. Benjamini and Hochberg corrected significance level is 0.0417

miRNA	Cancer tissue (- $\mathrm{\Delta }$ Ct)		Non-cancerous tissue (- $\mathrm{\Delta }$ Ct)		Fold-change (95% CI)	$P$ -value
	Mean	SEM	Mean	SEM
miR-21-5p	8.27	0.22	7.82	0.27	1.36 (0.97–1.91)	0.0721
miR-30a-5p	4.52	0.31	7.31	0.29	0.14 (0.10–0.21)	$<$ 0.0001
miR-126-3p	3.88	0.34	7.36	0.32	0.09 (0.06–0.14)	$<$ 0.0001
miR-205-5p	0.05	0.79	$-$ 3.03	0.46	8.45 (2.77–25.77)	0.0003
miR-210-3p	1.88	0.30	0.74	0.30	2.19 (1.52–3.16)	0.0001
miR-486-5p	$-$ 1.92	0.43	2.20	0.33	0.06 (0.03–0.10)	$<$ 0.0001

SEM, standard error of mean; CI, confidence intervals.

2.4 Statistical analyses

The categorical variables were described as absolute and relative frequencies, whereas continuous variables as mean and standard deviation (SD), if not stated otherwise. The miRNA expression was presented as - $\mathrm{\Delta }$ Ct to maintain normal distribution of the parameter and assure positive correlation with miRNA level of expression. The comparison of miRNAs expression in cancer and non-cancerous adjacent lung tissue from NSCLC patients was performed using paired Student’s $t$ -test. In order to differentiate between cancer and non-cancerous adjacent tissue discriminant analysis was conducted. A random split into training and test sets in 7:3 ratio was done and both forward and backward stepwise discriminant analysis was performed on the training set. The obtained model was tested on the test set. Receiver operating characteristic (ROC) curves of the constructed model was prepared and area under the curve (AUC) estimated. In order to test whether the differences (or trends for ordinal variables) exist between various NSCLC subtypes in investigated miRNAs expression, two-way repeated measures analysis of variance (ANOVA) was used. The analysis was also adjusted for sex, age and number of pack-years of cigarette smoking with a mean of multi-way repeated measures analysis of covariance (ANCOVA). In these analyses, type of the tissue served as within-subject factor, whereas the group as between-subject factor and significance of an interaction between these factors was assessed. The effect size was reported with partial eta squared. In order to test the associations between miRNAs expression and overall survival (OS) Cox proportional hazards analyses were employed and hazard ratios (HR) estimated. HRs were calculated using - $\mathrm{\Delta }\mathrm{\Delta }$ Ct values in order to facilitate the interpretation of HRs. The false discovery rate (FDR) for all the analyses was controlled at the level of 0.05 with the Benjamini and Hochberg correction for testing multiple hypotheses. $P$ -values below 0.05 were considered statistically significant. The analysis was performed using STATISTICA 12.5 Software (StatSoft, Tulsa, OK, USA).

3. Results

3.1 Expression of miRNA in cancer and non-cancerous tissues

In the present preliminary study the expression of miR-21-5p, miR-30a-5p, miR-126-3p, miR-205-5p, miR-210-3p and miR-486-5p against RNU6B in 50 pairs of tissues from patients with NSCLC were assessed. The data showed that five investigated miRNAs were differently expressed in tumour tissue relative to normal samples. miR-205-5p and miR-210-3p were found to be significantly up-regulated in tumour tissue by 8-fold ( $P<$ 0.0003) and 2-fold ( $P<$ 0.0001), respectively. While miR-486-5p, miR-126-3p and miR-30a-5p were found be significantly downregulated in tumour tissue by 17-fold, 11-fold and 7-fold, respectively ( $P<$ 0.0001). Only expression level of miR-21-5p ( $P=$ 0.0721) was not statistically significant but there was noted a clear tendency for upregulation in tumor samples (Table 3).

3.2 Combinations of miRNAs which can differentiate cancer tissue from non-cancerous adjacent tissue

Discriminant analysis was conducted to determine which of the investigated miRNAs can distinguish between cancer and adjacent non-cancerous tissue. Both forward and backward stepwise discriminant analysis yielded the same highly discriminative model including miR-30a-5p and miR-210-3p (Wilk’s lambda $=$ 0.325, $p<$ 0.0001) (Fig. 1). The model tested on the test set ( $n=$ 30) provides excellent sensitivity (15/15, i.e. 100%) with acceptable specificity (12/15, i.e. 80%) for prediction of type of tissue, AUC 0.969 (95% CI 0.918–1.0), $p<$ 0.0001 (Fig. 2).

Figure 1.

Discriminant analysis of cancer and non-cancerous tissue from non-small cell lung cancer patients. The model built on training set enables satisfactory classification: Wilk’s lambda $=$ 0.325, F (2.67) $=$ 69.5, $P<$ 0.0001. The line illustrates discrimination between tumour and non-tumour tissue.

Figure 2.

The receiver operating characteristic curve of the discriminant analyses of cancer and non-cancerous adjacent tissues. For the training set the area under the curve is 0.966 (95% CI 0.913–1.0), $p<$ 0.0001, whereas for the test set the area under the curve is 0.969 (95% CI 0.918–1.0), $p<$ 0.0001. The proposed cut-off probability point for the test set provides excellent sensitivity (15/15 i.e. 100%) with acceptable specificity (12/15, i.e. 80%) for prediction of type of tissue.

3.3 Association between expression levels of miRNAs in different clinicopathological characteristics and survival analysis

There were no significant differences between AC and SCC subtypes of NSCLC, size of tumour, stage of involvement of nearby lymph nodes and in the expression of all investigated miRNAs. The differences remained insignificant and of negligible effect size even after adjustment for sex, age and number of pack-years of cigarette smoking. Only in expression level of miR-21-5p was noted tendency for downregulation in cancer tissue with increased tumour size comparing to non-cancerous tissue (no adjusted and adjusted for sex, age and pack years analysis, $p=$ 0.0253 and $p=$ 0.0192 respectively) but this result did not survive correction for testing multiple hypothesis (Table 4). No significant association was found between miRNAs expression and OS of patients suffering from NSCLC in univariate Cox proportional hazard models as well as the models adjusted for sex, age and number of pack-years of cigarette smoking (Table 5).

4. Discussion

Six highly implicated in lung cancer genesis and development miRNAs were selected for testing in this preliminary study and RT-qPCR was employed to evaluate their expression in NSCLC tissues. Performed discriminant analysis revealed that miR-30a-5p together with miR-210-3p can distinguish cancer from normal adjacent tissue. Moreover, ROC analysis revealed the high diagnostic value of combination of these two miRNAs. Several research groups have reported about the altered expression of miR-30a-5p [16, 25, 64] and miR-210-3p in NSCLC [38], but never before, according to our knowledge, the combination of these two miRNAs was mentioned as an independent biomarker. Interestingly, careful insight into the discriminant model determined in this study revealed that miR-30a-5p and miR-210-3p are almost perfectly correlated in non-cancerous lung tissue (r $=$ 0.95), but not so much in cancerous tissue (r $=$ 0.61). This may indicate that potential common pathway of regulation of expression of these two miRNAs is involved, which warrants further research.

Kumarswamy et al. [25] found significant differences in miR-30a-5p expression between cancer and their adjacent non-cancer lung tissues ( $n=$ 64). They revealed that expression of miR-30a-5p is inversely correlated to invasive capabilities. They indicted that miR-30a-5p may act as a tumour suppressor and has ability to modulate an epithelial mesenchymal transduction and to inhibit invasion and metastasis through targeting Snail1 gene (a known transcriptional repressor of E-cadherin). In studies conducted by Jiang et al. [23] miR-30a-5p was downregulated 3.9 fold times in NSCLC tissue compared to adjacent normal tissue ( $n=$ 22). Moreover, authors presented a mechanism of  the  inhibition  of  BCL11A  protein  expression  by

Table 4

Differences in expression of investigated miRNAs between various non-small cell lung cancer subtypes and pathological characteristics. The analyses were performed with two-way repeated-measures analysis of variance (no adjustment) or with multi-way repeated measures analysis of covariance (to control for sex, age and number of pack-years of cigarette smoking). Benjamini and Hochberg corrected significance level for cancer is 0.0083 (false discovery rate 0.05)

miRNA	Subtype AC vs. SCC ( $n=$ 44)		Tumour size ${}^{*}$ ( $n=$ 48)		Nodes ${}^{*}$ ( $n=$ 48)		Pathological stage ${}^{*}$ ( $n=$ 48)
	No adjustment	Adjusted for sex, age and pack-years	No adjustment	Adjusted for sex, age and pack-years	No adjustment	Adjusted for sex, age and pack-years	No adjustment	Adjusted for sex, age and pack-years
miR- 21-5p	$F$ (1,42) $=$ 0.62 $p=$ 0.4344 part. ${\eta }^2=$ 0.01	$F$ (1,39) $=$ 1.11 $p=$ 0.2982 part. ${\eta }^2=$ 0.03	$F$ (1,44) $=$ 5.36 $p=$ 0.0253 part. ${\eta }^2=$ 0.11	$F$ (1,41) $=$ 5.95 $p=$ 0.0192 part. ${\eta }^2=$ 0.13	$F$ (1,45) $=$ 2.42 $p=$ 0.1269 part. ${\eta }^2=$ 0.05	$F$ (1,42) $=$ 2.02 $p=$ 0.1626 part. ${\eta }^2=$ 0.05	$F$ (1,45) $=$ 0.60 $p=$ 0.4439 part. ${\eta }^2=$ 0.01	$F$ (1,42) $=$ 0.77 $p=$ 0.3852 part. ${\eta }^2=$ 0.02
miR- 30a-5p	$F$ (1,42) $=$ 0.54 $p=$ 0.4652 part. ${\eta }^2=$ 0.01	$F$ (1,39) $=$ 0.02 $p=$ 0.8836 part. ${\eta }^2<$ 0.01	$F$ (1,44) $=$ 1.18 $p=$ 0.2833 part. ${\eta }^2=$ 0.03	$F$ (1,41) $=$ 1.93 $p=$ 0.1725 part. ${\eta }^2=$ 0.04	$F$ (1,45) $=$ 0.28 $p=$ 0.5961 part. ${\eta }^2=$ 0.01	$F$ (1,42) $=$ 0.09 $p=$ 0.7614 part. ${\eta }^2<$ 0.01	$F$ (1,45) $=$ 0.01 $p=$ 0.9399 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.03 $p=$ 0.8583 part. ${\eta }^2<$ 0.01
miR- 126-3p	$F$ (1,42) $=$ 0.62 $p=$ 0.4365 part. ${\eta }^2=$ 0.01	$F$ (1,39) $=$ 0.07 $p=$ 0.7990 part. ${\eta }^2<$ 0.01	$F$ (1,44) $=$ 0.47 $p=$ 0.4957 part. ${\eta }^2=$ 0.01	$F$ (1,41) $=$ 0.72 $p=$ 0.4008 part. ${\eta }^2=$ 0.02	$F$ (1,45) $=$ 0.08 $p=$ 0.7822 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.24 $p=$ 0.6293 part. ${\eta }^2=$ 0.01	$F$ (1,45) $<$ 0.01 $p=$ 0.9892 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.03 $p=$ 0.8659 part. ${\eta }^2<$ 0.01
miR- 205-5p	$F$ (1,42) $=$ 1.14 $p=$ 0.2921 part. ${\eta }^2=$ 0.03	$F$ (1,39) $=$ 0.01 $p=$ 0.9241 part. ${\eta }^2<$ 0.01	$F$ (1,44) $=$ 0.92 $p=$ 0.3436 part. ${\eta }^2=$ 0.02	$F$ (1,41) $=$ 2.20 $p=$ 0.1458 part. ${\eta }^2=$ 0.05	$F$ (1,45) $<$ 0.01 $p=$ 0.9527 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.26 $p=$ 0.6126 part. ${\eta }^2=$ 0.01	$F$ (1,45) $=$ 0.69 $p=$ 0.4110 part. ${\eta }^2=$ 0.02	$F$ (1,42) $=$ 1.88 $p=$ 0.1774 part. ${\eta }^2=$ 0.04
miR- 210-3p	$F$ (1,42) $=$ 0.94 $p=$ 0.3387 part. ${\eta }^2=$ 0.02	$F$ (1,39) $=$ 0.12 $p=$ 0.7310 part. ${\eta }^2<$ 0.01	$F$ (1,44) $=$ 3.09 $p=$ 0.0857 part. ${\eta }^2=$ 0.07	$F$ (1,41) $=$ 3.88 $p=$ 0.0556 part. ${\eta }^2=$ 0.09	$F$ (1,45) $=$ 0.13 $p=$ 0.7196 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.03 $p=$ 0.8651 part. ${\eta }^2<$ 0.01	$F$ (1,45) $=$ 0.18 $p=$ 0.6701 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.37 $p=$ 0.5475 part. ${\eta }^2=$ 0.01
miR- 486-5p	$F$ (1,42) $=$ 0.93 $p=$ 0.3392 part. ${\eta }^2=$ 0.02	$F$ (1,39) $=$ 0.02 $p=$ 0.8973 part. ${\eta }^2<$ 0.01	$F$ (1,44) $=$ 0.08 $p=$ 0.7731 part. ${\eta }^2<$ 0.01	$F$ (1,41) $=$ 0.34 $p=$ 0.5649 part. ${\eta }^2=$ 0.01	$F$ (1,45) $<$ 0.01 $p=$ 0.9432 part. ${\eta }^2<$ 0.01	$F$ (1,42) $=$ 0.05 $p=$ 0.8232 part. ${\eta }^2<$ 0.01	$F$ (1,45) $=$ 0.38 $p=$ 0.5406 part. ${\eta }^2=$ 0.01	$F$ (1,42) $=$ 0.12 $p=$ 0.7314 part. ${\eta }^2<$ 0.01

AC, adenocarcinoma; SCC, squamous cell carcinoma; ${}^{*}$ in case of ordinal variables the significance of linear trends was tested.

Table 5

Cox proportional hazards models for associations between miRNAs expression and overall survival. The analyses were performed as univariate and were adjusted for sex, age and number of pack-years of cigarette smoking

miRNA	Overall survival OS ( $n=$ 33)
	Univariate models		Adjusted for sex,
			age and pack-years
	HR (95% CI)	$P$ -value	HR (95% CI)	$P$ -value
miR-21-5p	0.74 (0.51–1.08)	0.1233	0.79 (0.61–1.03)	0.0769
miR-30a-5p	0.97 (0.81–1.17)	0.7794	0.87 (0.69–1.10)	0.2354
miR-126-3p	0.99 (0.83–1.16)	0.8596	0.93 (0.76–1.13)	0.4456
miR-205-5p	1.00 (0.94–1.06)	0.9871	0.99 (0.93–1.06)	0.8499
miR-210-3p	0.90 (0.72–1.12)	0.3317	0.85 (0.69–1.04)	0.1255
miR-486-5p	1.00 (0.90–1.12)	0.9337	0.97 (0.85–1.12)	0.7138

HR, hazard ratio; CI, confidence intervals.

miR-30a in in vitro studies. BCL11A protein is specifically upregulated in NSCLC tissues what correlates with disease-free survival and OS in early stage SCC patients. Xu et al. [56] found that miR-30a-5p expression is negatively correlated with expression of BCL-2 protein, a regulator of apoptosis, involved in tumorigenesis and chemoresistance in NSCLC tissues. Zhu et al. [64] found that CD73/NT5E is a direct target of miR-30a-5p, they reveal that expression level of this miRNA influence the cancer cells ability to proliferate. Moreover, according to Meng et al. [33] miR-30a-5p may regulates cell apoptosis and cell invasion with migration properties through controlling the PI3K/AKT signalling pathways. Results of research conducted by Tang et al. [49] on a bigger set of samples ( $n=$ 125) are in line with ours and indicate downregulation of miR-30a-5p in tumour tissue. Moreover, expression of miR-30a in their studies turned out to be negatively correlated with tumor size, lymphatic metastasis, clinical TNM stage and OS.

Researchers reported several times about increased miR-210-3p expression is different cancers [6, 32, 39]. Hypoxia, which is a hallmark of many solid tumors, is inducing the gene expression of miR-210 via hypoxia-inducible factors HIF-1a and HIF-2a [6, 24, 32]. Moreover, this miRNA targets many genes which are involved in carcinogenesis and are implicated in processes regulating apoptosis, angiogenesis and tumor growth [20]. miR-210 inhibits expression of transcription factors MNT, an antagonist of MYC and cell-cycle regulator E2F3 what results in the induction of cell proliferation [10, 60]. miR-210-3p has been reported to be increased in lung tumor tissues and sputum from NSCLC patients and to be correlated with patient’s survival [13, 44]. Recent research conducted by Daugaard et al. revealed that miR-210-3p is upregulated in AC tissues of patients showing distant metastasis and indicates the high potential of this miRNA to become a biomarker for formation of distant metastasis ( $n=$ 52) [11].

This preliminary study also confirmed the different expression of miR-126-3p, miR-486-5p and miR-205-5p in NSCLC tissue compared to non-cancerous adjacent tissue. Shao et al. [42] reported about downregulated level of miR-486-5p in NSCLC tissues ( $n=$ 38). This miRNA is known from acting as a tumor suppressor by targeting CDK4, an oncogene regulating cell cycle G1/S phase progression, IGF1R responsible for controlling of cell survival and proliferation [37] and ARGAP5, a protumorigenic gene [54]. miR-126-3p is reported to act as a tumor suppressor through targeting VEGF-A and EGFL7 involved in angiogenesis and metastasis development [28, 47]. Recently Song et al. [46] have also highlighted that miR-126 is exerting an impact on cells ability to proliferate, migrate and invasion. In studies on cell line A549 they show that miR-126 through targeting PIK3R2 can influence the PTEN/PI3K/AKT signalling pathway of multiple biological processes related to metabolism, cell growth and proliferation.

Role of miR-205 is carcinogenesis is not clear, it depends on cancer type and can be downregulated [21] or upregulated [7]. In NSCLC it is observed to be overexpress in tumor tissue [2, 4, 31]. Lebanony et al. [26] ( $n=$ 122) and Patnaik et al. [36] ( $n=$ 77) are indicating that miR-205 may become a specific biomarker for distinguishing SCC from AD. In vitro studies conducted by Bai et al. [2] presented that miR-205 may act as a tumor suppressor through inhibiting proliferation of cells by targeting PTEN. However Cai et al. [4] report that miR-205 through targeting PHLPP2 and PTEN leads to activation of both AKT/FOXO3a and AKT/mTOR pathways, consequently resulting in accelerated proliferation and enhanced angiogenesis in NSCLC.

Some studies indicate panels of miRNAs which can serve as biomarkers of NSCLC. Tan et al. [48] reported about combination of 5 miRNAs (miR-210, miR-182, miR-486-5p, miR-30a, and miR-140-3p) which contains three of tested by us miRNAs, that can distinguish cancerous SCC lesions from adjacent normal tissues ( $n=$ 148). Zhu et al. [65] noted the downregulated levels of miR-486-5p and miR-30a ( $n=$ 80). Moreover, meta-analysis conducted by Vosa et al. [52], covers highly deregulated miRNAs in NSCLC across 20 different studies and is highlighting diagnostic validity of investigated in this study miRNAs.

Nevertheless, the present preliminary study has several limitations, mostly resulting from the limited number of samples. This survey did not cover patients with metastasis, thus the correlation of miRNA-30a-5p or miR-210-3p (which are reported to be involved in process of metastasis formation [11, 25, 49]) with this variable could not be determined. Additionally the analysis of OS was performed on modest sample size ( $n=$ 33), what might be the main reason why none of the selected miRNAs were associated with OS. An attempt to determine the role of selected miRNAs in subclassification of NSCLC did not succeed, most probably because of the poor sample representation of each subtype.

miR-21 is reported to be overexpressed in NSCLC patients tissue, moreover, enhanced expression of this miRNA is reported to affect the bad prognosis and shorter survival of the patients [16, 31, 40, 41, 53, 57, 59]. miR-21 targets many genes, it is regulating EGFR-Akt pathway activity, the expression of tumour suppressors genes PDCD4, TPM1 and apoptosis-related proteins, and thus has impact on proliferation and metastasis of cancer cells [1, 15, 30, 62, 63]. Interestingly, in our research miR-21-5p showed only tendency to increase in expression in tumor tissue (1.36 fold) but, this value did not reach a statistical significance ( $P=$ 0.072). Additionally this miRNA showed tendency to be downregulated in cancer tissue with increased tumour size comparing to non-cancerous tissue ( $P=$ 0.025), but this result after correction for testing multiple hypothesis also did not reach statistical significance.

Carcinogenesis is a complex process in which miRNAs undoubtedly plays an important role. miRNAs are proved to be more stable and reproducible than mRNAs [34], what gives them opportunity to become a perfect biomarkers. Assessment of the expression of indicated in this preliminary study miRNAs in blood samples form NSCLC patients may provide additional evidence supporting the utility of these miRNAs as reliable diagnostic biomarkers. Additionally, further investigation of the role of these miRNAs in cancer and their mechanisms of action is substantiated and would increase knowledge concerning cancer biology.

In conclusion, we have consolidated the claim that miRNAs selected to this preliminary study have potential to become diagnostic biomarkers for NSCLC, especially miR-30a-5p together with miR-210-3p due to a high sensitivity in distinguishing cancer from non-cancerous adjacent tissue. Nevertheless because of the small sample size these preliminary results require confirmation by a further study.