MicroRNAs As Biomarkers For Clinical Features Of Lung Cancer

Abstract

Each year about 1.4 million people die from lung cancer worldwide. Despite efforts in prevention, diagnosis and treatment, survival rate remains poor for this disease. This unfortunate situation is largely due to the fact that a high proportion of cases are diagnosed at advanced stages, highlighting the great need for identifying new biomarkers in order to improve early diagnosis and treatment. Recent studies on microRNAs have not only shed light on their involvement in tumor development and progression, but also suggested their potential utility as biomarkers for subtype diagnostics, staging and prediction of treatment response. This review article summarizes the impact of microRNAs on lung cancer biology, and highlights their role in the detection and classification of lung cancer as well as direct targets for drug development.

Role Of MicroRNAs In Lung Cancer

Lung cancer is the leading cause of cancer-related death worldwide [1]. Tobacco smoke is unquestionably the main etiological agent, although ~25% of cases occur in never smokers, and are mostly associated with environmental carcinogens [2]. Lung cancer patients are often diagnosed at advanced stages and despite efforts made, the mortality remain high, with the five-year survival rate around 15% [3]. However, when detected at early stages, patient outcomes are substantially improved [4]. In this context, search for new biomarkers is necessary in order to improve early diagnosis.

Based on histopathologic features, lung cancer is classified as small cell lung cancer (SCLC) or non-small cell lung cancer (NSCLC), with lung squamous cell carcinomas (SqCC) and adenocarcinomas (AdC) being the major histological subtypes of NSCLC [5–9]. Both clinical and molecular characteristics of these subtypes have been associated with different patterns of genetic and epigenetic aberrations, including changes in microRNA (miRNA) expression profiles [10, 11].

MicroRNAs are small non-coding RNAs (~21–25 nucleotides in length) that interact with homologous mRNAs and regulate gene expression at the post-transcriptional level [12]. In animals, miRNAs act mainly through inhibition of the translation process, while in plants the literature indicates the major mechanism of action is enzymatic cleavage of target mRNAs [13, 14]. MiRNA genes are transcribed as long precursors called “pri-miRNA”. These precursors are cleaved in the nucleus by a complex called “microprocessor”, which contains the enzymes RNASEN (Drosha) and DiGeorge critical region 8 (DGCR8). The processed intermediate “pre-miRNA” is about 70 nucleotides in length, and adopts a hairpin-containing secondary structure through imperfect complementarities of bases between the two halves of its sequence. This pre-miRNA is transported to the cytosol and cleaved by a DICER family enzyme to release a small double stranded “miRNA”. Subsequently, this miRNA interacts with a protein of the Argonaute family (AGO1 or AGO2) to form the RNA-induced Silencing Complex (RISC) containing only the active single strand miRNA. The target mRNA is then loaded into RISC where it is silenced by either AGO2-mediated degradation or AGO1-mediated translational repression [13–15].

Alterations in miRNA expression have been shown to play major roles in cancer (reviewed in [16]). As miRNAs play important roles in development, cell proliferation and apoptosis, their deregulation has been implicated in cancer initiation and progression, indicating that miRNAs may function as tumor suppressor genes or oncogenes in different cancer types, including lung [13, 17]. Although their potential as biomarkers has been intensively explored, the contribution of miRNAs to lung cancer biology is not completely understood. Moreover, the study of microRNAs in lung cancer is an emerging field. This is evident from the increasing trend of scientific publications on this topic, including their use as biomarkers (Figure 1). In this context, the focus of this review will be the impact of miRNAs on lung cancer biology, and the pros and cons of their role as biomarkers for detection and classification of lung cancer.

Figure 1. Number of publications annotated on PubMed database from the US National Library of Medicine.

Search was performed using the following terms: “miRNA” OR “microRNA” AND “cancer” AND “lung”. Red line represents the cumulative number of entries (between 2004 and 2011). Blue line indicates number of entries per year. Green dotted line indicates the results of the same search with “biomarker” as an additional search term.

Biological Relevance Of MicroRNAs In Lung Cancer

A variety of miRNAs have been shown to play a role in lung cancer initiation and progression (Table 1 and Supplementary Table 1). Furthermore, disruption of the machinery of miRNA biosynthesis has also been documented in lung cancer. For example, reduction in DICER expression has been reported in NSCLC and has been significantly associated with poor prognosis [18]. In a mouse K-Ras-induced lung cancer model, conditional deletion of Dicer enhances lung tumor development [19]. Moreover, miRNAs might also play an important role in early events in lung tumorigenesis, since changes in miRNA expression patterns have been described in premalignant lesions of the bronchial epithelium [20]. Early involvement of miRNAs is also evident in carcinogen-induced lung cancer animal models. Treatment with tobacco carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) in rats resulted in deregulation of several miRNAs, including loss of miR-126* expression. Interestingly, cytochrome P450 2A3 (CYP2A3), which activates NNK carcinogenicity, is a target of miR-126* [21]. Changes in miR-101, miR-126* and miR-199 observed in early tumorigenesis in rats have also been reported to be underexpressed in human NSCLC [22–25]. Next we will discuss the relevance of miRNAs in deregulation of key cellular functions in lung cancer. Some of those major miRNA-regulated pathways are summarized in Figure 2.

Table 1.

Frequently deregulated miRNAs in lung cancer

miRNA	Expression				Confirmed targets	Associated with		Reference
	NSCLC		SCLC
	Tumors	Lines	Tumors	Lines		Diagnosis	Survival1
let-7	Dn	Dn	UC		HMGA2, RAS, MYC, E2Fs, CDKs	SCC	NSCLC	[24, 25, 54, 65, 77, 78, 80, 81, 105]
miR-7	Up			Up	ETS2, EGFR, Raf1			[24, 54]
miR-9	Dn/Up			Up	TGFBI, TRIM2, SIRT1, BTBD3	SCLC		[24, 25, 54, 106]
miR-15	Up	NC	Up	Dn/Up	Bcl-2, CCND1, CCNE		NSCLC	[10, 54, 77–79]
miR-16	Dn	NC		Dn/Up	Bcl-2, CCND1, CCNE		NSCLC	[10, 24, 54, 65, 79]
miR-17-3p,5p	Up		Up	Up	p21, E2F, Bim, CCND1, PTEN, Rb	NSCLC		[25, 40, 54, 77, 78, 107]
miR-17-92 cluster	Up	Up	Up	Up	p21, E2F, CCND1, PTEN, Rb, HIF-1α	SCLC	SCLC	[40, 54]
miR-20	Up			Up	p21, E2F, Bim, Cyclin D1, PTEN, Rb	NSCLC		[40, 54, 77, 78]
miR-21	Up	Up	UC	Dn	Apaf-1, PTEN, PDCD4, Faslg, Spry1-2, Btg2	NSCLC; ADC	NSCLC	[10, 23–25, 43, 44, 54, 64, 65, 77–79, 95, 107, 108]
miR-22		UC		Dn	PTEN, ESR1, TIAM1			[10, 109, 110]
miR-23a,b	Up	NC		Dn	Has2, LaminB1, Brn-3b		NSCLC	[10, 54, 79]
miR-24	Up	NC		Dn	E2F2, Myc, Net1A		NSCLC	[10, 25, 79]
miR-25	Up		Up	Up	Bim, TRAIL	NSCLC	NSCLC	[54, 60, 78, 111]
miR-27b	Dn		UC	UC	ST14, PPARα		NSCLC	[25, 54, 79]
miR-29a,b,c	Dn	NC	Dn	Dn	DNMT3 (A&B)			[10, 23, 25, 54]
miR-30	Dn			Dn	p53, B-Myb, CTGF			[22–25, 54, 65]
miR-30c, d	Dn	Up		Up	Serpine1		NSCLC	[10, 65, 79, 82, 112]
miR-31	Up	Dn		Dn	LATS2, PPP2R2A	ADC		[10, 23, 46, 64, 65]
miR-32	Dn			Dn	Bim			[25, 54]
miR-33	Dn			Dn	ABCA1, p53			[25, 54, 113, 114]
miR-34a	Dn/Up			Dn	DLL1, CDK4, Myc, CCND1		NSCLC	[23, 54, 78, 115, 116]
miR-92		UC		Up	Bim			[10, 54, 117]
miR-93		UC		Up	p21, FUS1	SCLC		[10, 54, 111, 118]
miR-95	Dn		Up	Up	SNX1	SCLC		[25, 54, 119]
miR-96	Up	Up	Up	Up	GPC3, FOXO1		NSCLC	[10, 54, 120]
miR-98		UC	Up	Up	FUS1	SCLC		[10, 54, 93, 118]
miR-99	Dn			Dn	SMARCA5, SMARCD1, mTOR			[10, 23, 65, 121]
miR-100	Dn			Dn	ATM, PLK1, MTOR, RPTOR, IGF1R, Gli1		SCC	[23, 54, 77, 78]
miR-101	Dn	Up		Up	Mcl-1			[10, 22–25, 47–49]
miR-106a	Up		Up	Up	E2F1	ADC	OS-NSCLC	[10, 54, 77, 78, 111]
miR-106b		Up		Up	p21, E2F, Bim, Cyclin D1, PTEN, Rb			[10, 54, 77, 78, 111]
miR-125	Dn		Dn	Dn	Smo, invasion and migration factors	NSCLC		[23, 25, 54, 65, 77, 122]
miR-126	Dn		Dn	Dn	Spred-1	SCC, ADC	SCC	[22–25, 64, 65, 77, 78]
miR-126*	Dn				ERBB2IP	NSCLC		[23–25, 95]
miR-128	Dn			Up	EGFR	SCLC	EGFR-TKi2	[45, 54]
miR-130a	Dn			Dn	MET		NSCLC	[23, 24, 54, 65, 79]
miR-130b	Up		Up	Up	CSF-1, TP53INP1, RUNX3	SCC		[54, 64, 65]
miR-133b	Dn			Dn	Mcl-1, BCLw			[24, 54]
miR-139	Dn				ROCK2, FOXO1, CXCR4	SCC		[24, 64, 65, 123–125]
miR-140	Dn				Sp1, BMP2	ADC		[25, 64]
miR-142-3p,5p		Up		Dn/Up	ADAM9, GRα, PKA			[10, 54]
miR-143	Dn			Dn	ERK5, KRAS, Bcl-2, DNMT3A, ELK1	ADC		[23–25, 54, 64]
miR-145	Dn			Dn	c-MYC, CDK4, eIF4E4	NSCLC; ADC		[23, 25, 54, 64, 95]
miR-146b	Up			Dn	MMP16, TRAF6, IRAK1,	ADC	SCC	[25, 54, 77, 78, 107]
miR-149	Up	Up		Up	E-cadherin	SCC		[10, 55, 65]
miR-150	Up/Dn		Dn	Dn	MUC4, P2RX7	ADC		[25, 54, 64]
miR-155	Up			Dn	TAB2, FADD, RIP-1, IKKε, ET -1	ADC	NSCLC	[25, 54, 61, 77, 78, 107]
miR-181a,b	Dn		Dn	Dn	Mcl-1, Bcl-2			[23, 54, 65]
miR-181c	Dn	NC		Up	TGFBI, TRIM2, SIRT1, BTBD3			[10, 65, 106]
miR-182	Up		Up	Up	FOXO1	NSCLC, ADC	NSCLC	[24, 54, 61, 64, 65, 77, 78, 80, 95, 120]
miR-183	Up		Up	Up	FOXO3, FOXO1, EGR1, PTEN	NSCLC	NSCLC	[24, 54, 65, 77, 95, 120]
miR-185	Dn	Dn		Dn	c-Myc, DNMT1, SMG6, CDK6, AKT1			[54, 126]
miR-191	Up			Dn	CDK6, TIMP3, NDST1	ADC	SCC	[25, 54, 77, 78, 107]
miR-192	Up				ERCC3, ERCC4, Rb1			[25, 107]
miR-195	Dn	NC		Dn/Up	CCNE, BCLw			[10, 24, 54, 65]
miR-197	Up	Up		Up	FUS1			[10, 25, 61]
miR-199	Dn	Dn	Dn	Dn	HES, ET-1, HIF-1α			[10, 21, 25, 54]
miR-200a,b	Dn/Up		Up	Up	C-ets-1, Zeb1, ECM proteins, p38α	ADC	NSCLC	[54, 64, 79, 127]
miR-200c	Up	Dn	Up	Up	C-ets-1, Zeb1, ECM proteins	SCC	NSCLC	[54, 65, 77, 128]
miR-203	Up		.		DeltaNp63, Smo	NSCLC, ADC		[25, 77, 107]
miR-205	Up	Dn		Dn	E2F1, Zeb1, Sip1	SCC; ADC		[10, 25, 65, 74, 75, 78, 93, 107]
miR-210	Up		Up	Up	HOXA1, FGFRL1, HOXA9, NDUFA4, EFNA3	NSCLC, SCC, ADC		[24, 25, 54, 57, 64, 65, 77, 95, 107, 108, 129]
miR-214	Up		Dn	Dn	EZH2, Brn-3b	ADC		[25, 54, 107]
miR-218	Dn				RICTOR	ADC		[24, 25, 64]
miR-221	Up/UC			Dn	p27, PTEN, TIMP3		NSCLC	[10, 50, 80]
miR-222	Up/UC			Dn	p27, PTEN, TIMP3			[10, 50, 54]
miR-223	Dn/Up		Dn	Dn	LMO2-L/-S, CEBP-β	NSCLC		[54, 60, 65]
miR-224	Dn/Up			Dn	DIO1			[77]
miR-324-5p	Up	Up	Up	Up/Dn	Smo, Gli1	SCC		[10, 54, 65, 130]
miR-326		Up		Up	MRP-1/ABCC1, Ets-1			[10, 131, 132]
miR-335	Dn		Up	Up	SOD2, Txnrd2, BCLw, SP1, ERα, IGF1R			[24, 54, 133–135]
miR-338	Dn	Up		Up/Dn	Runx2, Smo			[10, 24, 54, 130, 136]
miR-375	Up	Dn	Up	Up	IGF1R, SP1, LDHB, YAP1	ADC	SCLC	[54, 64, 137–140]
miR-423	Up				p21	SCC		[24, 65, 141]
miR-429	Dn			Up	ALDH1L2, PSAT1, ZEB1	SCC		[54, 65, 142]
miR-451	Dn			Dn	RAB14, apoptosis			[23, 24, 54]
miR-453	Dn				ESR1		SCC	[77, 78, 82]
miR-486	Dn				CD40	ADC	NSCLC	[64, 65, 108]
miR-494	Dn			Up	Nucleolin, Bmal1		SCC	[54, 77, 78, 143, 144]
miR-497	Dn			Dn	BCL2, CCND2			[24, 54, 145]{Xing, 2010 #85
miR-504	Up				p53, DRD1			[23, 146, 147]
miR-511	Up			Dn	TLR4, CD80		SCC	[54, 77, 78, 148]
miR-512				Dn	c-FLIP, Mcl-1			[54, 149, 150]

¹Expression trend can be either positively or negatively correlated to overall survival

²Associated with drug response and not to overall survival

Abbreviations: Lines = cell lines, SCC = lung squamous cell carcinoma, ADC = lung adenocarcinoma, Up = increased expression, Dn = decreased expression, UC = no change

Figure 2. Main cancer-related cell functions targeted by miRNA in lung cancer.

Multiple targets of lung cancer related key miRNAs are evidenced, as well as the variety of downstream effects. Members of the Let-7 and miR17-92 families are involved in disruption of cell cycle and apoptotic processes in lung cancer. The let-7 cluster negatively regulates multiple oncogenes, including Ras and Myc. Myc is thought to negatively regulate let-7 members by binding directly to their promoters (grey dotted line). Over-expression of miR-17-92 family and miR-101 contribute to cell survival by reducing apoptosis. On the other hand, miRNAs are involved in lung cancer progression (mainly through induction of angiogenesis and activation of invasiveness). This is the case of miR-221 and miR-222 that target PTEN and TIMP3 tumor suppressors and enhance cellular migration through activation of the Akt pathway and metallopeptidases; and miR- 210, which is involved in angiogenesis by targeting EFNA3 and the HOX family of transcription factors. The miR-200 cluster restricts metastasis and EMT by targeting E-cadherin transcriptional regulators such as ZEB proteins.

MicroRNAs involved in cell cycle and apoptosis in lung cancer

The let-7 family

The tumor suppressive let-7 family comprises 13 members in the human genome that are dispersed over chromosomes 3, 9, 10, 11, 12, 19, 21, 22, and X [26]. Members are highly expressed in normal lung tissue and negatively regulate multiple oncogenes, including RAS and MYC [27, 28]. MYC is thought to negatively regulate let-7 members by binding directly to their promoters (specifically through the E-box 3 domain), representing a negative-feedback loop between these two elements [26, 29]. Expression of let-7 members also affects cell-cycle regulators, such as cyclins, CDKs and E2F transcription factors, and anti-apoptotic factors, like Bcl-xL [28, 30]. In NSCLC cell lines, inhibition of let-7 leads to increased cell division, while overexpression results on cell growth inhibition [28]. Similar effects have been observed in xenografts and transgenic mouse models [31].

Importantly, pathogenic mechanisms of let-7 disruption in lung cancer are not limited to alterations in expression levels. A single nucleotide polymorphism in let-7 complementary site 6 of the KRAS mRNA 3′-UTR is associated with an increased risk of NSCLC among moderate smokers. The frequency of this variant is ~ 20% in NSCLC patients compared to only 6% in the general population [32]. These findings illustrate that even at biologically normal levels, let-7 function can be impaired by variations in its target mRNAs.

The miR-17-92 family

The oncogenic function of the miR-17-92 cluster on chromosome 13q31.3, also known as oncomir-1, was first demonstrated in a mouse B-cell lymphoma model requiring over-expression of c-MYC [33]. Subsequent studies have further demonstrated MYC-induced expression of miR-17-92, and interestingly, induction of its paralog cluster, miR-106b-25, located on chromosome 7 [34, 35]. In addition to MYC, miR-17-92 expression can be induced by E2F proteins, which in turn are targeted by miR-17-92 members miR-17 and miR-20a, and miR-106b-25 member miR-106b. Through MYC and E2F proteins, the two clusters regulate cell cycle and apoptosis via feedback mechanisms as well as through targeting of a common set of downstream cell cycle regulators and apoptotic factors including BIM, PTEN, RB1, p21, and cyclin-dependent kinases [34–38].

In NSCLC, loss of let-7 and up-regulation of MYC are frequent events, and together can lead to an increase in miR-17-92 levels and excessive activation of E2F proteins. Excessive levels of E2F1 can induce cell cycle inhibition and apoptosis [39], yet in cancer, the net effect of this signalling network is the overall enhanced activity of E2F transcription factor activity and subsequent cell proliferation [36]. Clearly, the miR-17-92 cluster plays a key role in the cancer promoting function of E2F upregulation to promote proliferation and evade apoptosis.

Both SCLC and NSCLC exhibit over-expression of the miR-17-92 cluster, although the effects of this overexpression are perhaps more pronounced in SCLC [40, 41]. In SCLC cell lines, over-expression of miR-17-92 was shown to influence genomic instability upon RB inactivation. It was observed that RB inactivation induced γ-H2AX foci formation, a marker of DNA damage, as well as growth inhibition and generation of reactive oxygen species, effects that were offset by miR-17-92 overexpression [40]. It is possible that in an RB inactivated model of SCLC, miR-17-92 allows cell tolerance of DNA damage and increased genomic instability.

Other miRNAs

Other miRNAs have also been observed to be involved in cell cycle and apoptosis regulation. This is the case for miR-15 and miR-16, which are transcriptional targets of E2F1 and limit E2F-induced excessive proliferation by targeting CYCLIN E [42]. Another example is miR-21, which is one of the most frequently over-expressed miRNAs in NSCLC. This miRNA is up-regulated by activation of EGFR and RAS pathways, and it is implicated in lung tumorigenesis in never smokers with KRAS and EGFR activating mutations and amplifications [43, 44]. Over-expression of miR-21 enhances tumorigenesis by targeting SPRY1, SPRY2, BTG2 and PDCD4, which act as negative regulators of the RAS/MEK/ERK pathway, and APAF-1, FASLG, PDCD4 and RHOB, which promote apoptosis [43]. Conversely, the anti-tumoral miR-128b, which directly targets EGFR, is frequently under-expressed in NSCLC patients [45].

MiR-31 has been shown to be up-regulated in murine lung cancers relative to adjacent normal lung tissues and in primary SqCC [23]. It may function as an oncomir by directly repressing the tumor suppressors LATS2 and PPP2R2, thereby leading to increased cell division [46]. Conversely, miR-101 is down-regulated in NSCLC [22–25], leading to enhanced expression of its MCL-1 target gene in NSCLC, thus favoring tumor progression via inhibition of apoptosis [47]. Additionally, miR-101 is affected by DNA copy number losses, which is associated with reduced expression in NSCLC. This alteration can lead to overexpression of its target gene, EZH2 (a mammalian histone methyltransferase), and consequently induce disruption of cell proliferation-related processes, among other effects [48, 49].

MicroRNAs involved in invasiveness and angiogenesis

Besides being implicated in the regulation of early steps of lung tumorigenesis, some miRNAs are involved in lung cancer progression. Examples include miR-221 and miR-222, which are over-expressed in NSCLC via C-JUN, and target PTEN and TIMP3 tumor suppressors [50]. These miRNAs induce TRAIL resistance and enhance cellular migration through activation of the AKT pathway and metallopeptidases. Similarly, miR-191, a miRNA up-regulated in NSCLC [25], targets TIMP3 and has been shown to promote epithelial-to-mesenchymal transition (EMT) in hepatocellular carcinoma [51]. This process can be hampered by over-expression of miR-130a, which targets MET, an upstream activator of C-JUN, and induces down-regulation of miR-221 and miR-222 while increasing TRAIL-sensitivity in NSCLC cell lines [52]. In fact, miR-130a down-regulation is observed in NSCLC [23].

The miR-200 cluster consists of another tumor suppressive family that restricts metastasis and EMT in lung adenocarcinoma by targeting E-CADHERIN transcriptional regulators such as ZEB proteins [53]. This family appears to undergo reciprocal regulation between lung cancer subtypes, as it has been reported to be under-expressed in NSCLC while over-expressed in SCLC [54]. The observed up-regulation in SCLC is particularly intriguing knowing the high metastatic propensity of these tumors. Additionally, miR-149 is known to directly inhibit E-CADHERIN expression [55] and has been shown to be over-expressed in both SCLC and NSCLC cell lines [10].

Some miRNAs highlighted in lung cancer may also play major roles in angiogenesis. A recent study observed that miR-21, miR-106a, miR-126, miR-155, miR-182, miR-210 and miR-424 were associated with angiogenesis in NSCLC [56]. In fact, hypoxia inducible factors enhance the expression of miR-210, which promotes angiogenesis by targeting EFNA3 [57].

Sequence variations in miRNAs and their targets

The change in expression levels is not the only mechanism by which miRNAs can disrupt biological functions. Like protein-coding genes, miRNA genes exhibit sequence variations, such as single-nucleotide polymorphisms (SNPs). Tian and colleagues have shown that the homozygous C/C rs11614913 variant of miR-196a2 was associated with a significantly elevated risk of lung cancer compared with the wild-type T/T or T/C counterparts [58]. It was suggested that the rs11614913 variant can influence miR-196a expression, which regulates metastasis by targeting HOX genes. In another example, a SNP in the 3′UTR of KRAS was associated with increased risk for NSCLC presumably through disruption of let-7 binding affinity, which in turns affects KRAS regulation [32].

Clinical Relevance Of MiRNAs In Lung Cancer

Different studies associating roles of miRNAs with deregulation of key cellular functions in lung cancer have lead to the idea that miRNAs could be used as biomarkers for this disease, especially for subtype classification, prognosis as well as directing or design of new targeted therapies. Moreover, the stability and availability of these molecules in clinical sample materials such as FFPE and serum reinforces their potential [59]. Here we present correlations established between miRNAs and clinical characteristics of lung cancer, and their potential role as biomarkers as well as therapeutic targets.

MicroRNA as biomarkers for lung cancer early diagnosis

The implication of some miRNAs in the development of lung cancer has led to a cottage industry exploring the utility of miRNAs as biomarkers for early stage diagnosis. Moreover, some studies demonstrated that the levels of miRNAs in serum are stable, consistent among individuals, and reproducible enough to be taken into account in clinical routines. For example, miR-223 and miR-25 are over-represented in serum of lung cancer patients [60]. In fact, this study revealed a 76-miRNA lung cancer signature detected only in patient serum (relative to blood cells) when compared to healthy subjects. MiRNAs detected at elevated levels in the plasma of stage I lung cancer patients include miR-155, miR-197, and miR-182 [61]. MiRNA signatures in plasma [62] as well as in serum [63] can discriminate between screening detected early-stage (I and II) NSCLC and non-cancer controls, with a high degree of diagnostic accuracy. It has been suggested that serum signatures could even be used to detect lung cancer among asymptomatic high-risk individuals [62, 63]. MiRNAs profiles are also discernible in sputum of early-stage lung cancer patients. Indeed, studies have been able to identify specific miRNA signatures with high sensitivity and specificity in sputum of early stage AdC (81% and 92% of sensitivity and specificity, respectively) and SqCC patients (73% of sensitivity and 96% of specificity) [64, 65]. These observations have led to the consideration that analysis of miRNA in sputum could also be explored for identifying biomarkers in healthy smokers that would have the ability to distinguish individuals who will develop lung cancer in the future [66].

Despite these promising findings, it is important to note that several confounding issues may contribute to circulating miRNA variability. Disturbances in the immune system, such as inflammation, which can be independent of the tumor status, may considerably influence miRNA profiles observed in serum and plasma [67–69]. MiRNA profiles from sputum have also been observed to be modified by the inflammation status of smokers and the presence or absense of inflammatory diseases such as chronic obstructive pulmonary disease [70]. Additionally, other confounding issues are related to sample collection and processing conditions [71, 72].

Biomarkers for histopathological classification

Relevance of miRNA profiling for subclassification of solid tumors has been confirmed in several studies including lung cancer subtypes [25, 73]. Nevertheless, their accuracy compared to histopathology remains uncertain. For instance, miR-205 has been previously identified as a highly accurate marker for lung SqCC with sensitivity of 96% and specificity of 90% [74], and showed 100% concordance between the diagnoses established by histopathologic and qRT-PCR based diagnostic assay [75]. However, more recent work has indicated that the use of miR-205 alone led to the misclassification of certain cases [76]. Another subtype-specific miRNA is miR-155, which is under-expressed in SCLC [54] and over-expressed in NSCLC [25, 77]. This example illustrates the complexity of miRNA-related functions, since the same miRNA can display both oncogenic and tumor suppressive behavior depending on the cellular context.

Another strategy to distinguish lung cancer subtypes has been the application of miRNA-specific signatures. A recent study identified a set of 34 miRNAs that are differentially expressed between AdC and SqCC, while another study identified 15 miRNAs that can differentiate SqCC from normal lung [77, 78]. However, only five miRNAs of the 15 miRNA SqCC set (miR-17, miR-20a, let-7e, miR-106a, miR-106b) overlap with the subtype specific panel of 34 miRNAs. Let-7e is the only miRNA downregulated in SqCC compared to AdC [78] as well as compared to normal lung [77], while miR-17, miR-20a, miR-106a, miR-106b are upregulated in SqCC, but even more in AdC [77, 78]. Sputum is an easily accessible source of material to derive panels of subtype specific miRNAs. Early stage SqCC patients were distinguished from normal subjects by detection of over-expression of miR-205, miR-210, miR-21 and miR-708, and under-expression of miR-429, miR-139, miR-30a and miR-126 in sputum, with 73% sensitivity and 96% specificity [65]. A similar study in AdC revealed over-expression of miR-21, miR-182, miR-200b, miR-375, and under-expression of miR-486, miR-126 and miR-145 in sputum of AdC patients [64]. The use of miR-21, miR-200b, miR-375 and miR-486 alone was able to distinguish AdC patients with a sensitivity and specificity of 81% and 92%, respectively. Taken together, these studies identified several miRNAs (miR-486, miR-126, miR-139, miR-130b or miR-210) with similar expression in the two major subtypes of NSCLC.

In summary, although a new classification method based on miRNA expression should be considered in the future, current literature lacks strong evidence to support unequivocal subtype-specific miRNAs. The most recurrent subtype-specific miRNAs are listed in Table 2.

Table 2.

miRNAs differentially expressed in histological subtypes of lung cancer

miRNA	Expression in subtype			Confirmed targets	References
	SqCC	AdC	SCLC
let-7	Dn			HMGA2, RAS, MYC, E2Fs, CDKs	[24, 25, 54, 65, 77, 78, 80, 81, 105]
miR-9			Up	TGFBI, TRIM2, SIRT1, BTBD3	[24, 25, 54, 106]
miR-17-92 cluster			Up	p21, E2F, CCND1, PTEN, Rb, HIF-1α	[40, 54]
miR-31		Up		LATS2, PPP2R2A	[10, 23, 46, 64, 65]
miR-93			Up	p21, FUS1	[10, 54, 111, 118]
miR-95			Up	SNX1	[25, 54, 119]
miR-98			Up	FUS1	[10, 54, 93, 118]
miR-128			Up	EGFR	[45, 54]
miR-130b	Up			CSF-1, TP53INP1, RUNX3	[54, 64, 65]
miR-139	Dn			ROCK2, FOXO1, CXCR4	[24, 64, 65, 123–125]
miR-140		Dn		Sp1, BMP2	[25, 64]
miR-143		Dn		ERK5, KRAS, Bcl-2, DNMT3A, ELK1	[23–25, 54, 64]
miR-146b		Up		MMP16, TRAF6, IRAK1,	[25, 54, 77, 78, 107]
miR-149	Up			E-cadherin	[10, 55, 65]
miR-150		Dn		MUC4, P2RX7	[25, 54, 64]
miR-155		Up		TAB2, FADD, RIP-1, IKKε, ET-1	[25, 54, 61, 77, 78, 107]
miR-182		Up		FOXO1	[24, 54, 61, 64, 65, 77, 78, 80, 95, 120]
miR-191		Up		CDK6, TIMP3, NDST1	[25, 54, 77, 78, 107]
miR-192		Up		ERCC3, ERCC4, Rb1	[25, 107]
miR-200a,b		Up		C-ets-1, Zeb1, ECM proteins, p38α	[54, 64, 79, 127]
miR-200c	Up			C-ets-1, Zeb1, ECM proteins	[54, 65, 77, 128]
miR-203		Up		DeltaNp63, Smo	[25, 77, 107]
miR-205	Up			E2F1, Zeb1, Sip1	[10, 25, 65, 74, 75, 78, 93, 107]
miR-218		Dn		RICTOR	[24, 25, 64]
miR-324-5p	Up			Smo, Gli1	[10, 54, 65, 130]
miR-375			Up	IGF1R, SP1, LDHB, YAP1	[54, 64, 137–140]
miR-423	Up			p21	[24, 65, 141]
miR-429	Dn			ALDH1L2, PSAT1, ZEB1	[54, 65, 142]
miR-486		Dn		CD40	[64, 65, 108]

Note: Targets cited are confirmed by experiments

Biomarkers for prognostics and treatment prediction

MiRNAs have been associated with patient prognosis and response to chemo- and radiotherapy in lung cancer. Over- and under-expression of miRNAs has been associated with recurrence of stage I disease and disease-free survival in NSCLC patients [79, 80]. In terms of survival, higher expression levels of some miRNAs (such hsa-miR-221 and hsa-let-7a) seem to have protective functions, while others (hsa-miR-137, hsa-miR-372, and hsa-miR-182*) are associated with decreased survival [80]. Additionally, poor survival of NSCLC patients has been linked to low expression of the let-7 family, and miRNA processing enzymes DICER and DROSHA [18, 81]. Some studies have tested the utility of miRNA-signatures as non-invasive serum biomarkers for overall survival in NSCLC patients who received surgical treatment and adjuvant chemotherapy. A 4-miRNA signature (miR-486, miR-30d, miR-1 and miR-499) was identified as a predictor of overall survival [82]. Moreover, expression of let-7 is related to survival time after curative resection [81].

Alterations in miRNA levels are also correlated with prognosis in a subtype-specific manner. Poor survival in AdC has been correlated with low expression of let-7a-2 and high expression of miR-155, while changes in miR-21 and other miRNA levels have been recently associated with shortened survival time in patients with SqCC [25, 77, 78]. Additionally, elevated expression of miR-31 inversely correlates with the survival of Chinese SqCC patients [83].

MiRNAs have been shown to modulate chemo- and radiosensitivity of lung tumor cells [84, 85]. The let-7 family of miRNAs modulates response to cytotoxic anticancer therapy and also affects response to radiation of lung cancer cells through a RAS dependent mechanism [85]. Loss of heterozygosity of miR-128b, which directly targets EGFR, correlates significantly with positive NSCLC patient response and survival with EGFR-TKI treatment [45]. Another study found that low plasma levels of miR-21 correlated with sensitivity to platinum-based chemotherapy [86]. In lung cancer cell models, miR-181a and miR-630 conferred sensitization to cisplatin [87], while downregulation of miR-200 was associated with chemoresistance to cisplatin and cetuximab [88].

MicroRNAs as therapeutic agents and targets in lung cancer

The multiple molecular pathways targeted by miRNAs make them candidates for developing novel molecular therapeutics agents [89]. Antagonizing a single miRNA could affect a number of vital signalling pathways in the tumor cell simultaneously; however, the same characteristic could have undesirable side effects on other pathways. The potential broad biological impact of microRNA perturbation needs to be considered when considering miRNAs as therapeutic targets.

Currently, direct strategies to target miRNAs in cancer involve the use of oligonucleotides or virus-based constructs to either suppress the expression of oncogenic miRNAs or to replace the loss of expression of tumor suppressor miRNAs. Indirect strategies involve the use of drugs to regulate miRNA expression by targeting their transcription and processing [90]. In this context, let-7 has been intensively studied. In mouse models, delivery of let-7 miRNA via adenoviral infection induces regression of lung tumors [31], and intranasal let-7 administration reduces lung tumor formation in KRAS mutant models [91]. Additionally, ectopic let-7g miRNA expression in xenografts and in a mouse lung tumor model was found to reduce tumor burden [92]. Even if under-expression of the let-7 family is stronger in SqCC [93], let-7 is associated with survival in both subtypes and could still be a useful target in both subtypes [25].

The therapeutic potential of other miRNAs has been explored in lung cancer. In vivo knockdown of oncogenic miR-31 has been shown to repress lung cancer in FVB syngeneic mice [46]. Likewise, administration of miR-34a (either locally or systemically) inhibited NSCLC tumor growth in mouse models by blocking CDK4, c-Met and Bcl-2 expression [94]. Promising results on cell growth inhibition have also been obtained by administering miR-145 to EGFR mutant human AdC cell lines [95] and efficacy of miR-133b delivery in lung tissue has been proved in mice [96]. The potential use of the miR-17-92 cluster as a therapeutic target was also recently demonstrated in lung cancer cell models. Antagonization of miR-17 and miR-20a was found to induce apoptosis in vitro [97].

To date, most of the investigations into the therapeutic use of miRNAs in lung cancer have been performed in cell lines or animal models. Results suggest a reasonable degree of toxicity of these molecules; however, a final decision for human use must be based on results derived from clinical trials. Clinical trials have already been initiated to investigate the safety and efficacy of miRNA treatment in human patients, such as the NCT00979927 protocol (http://clinicaltrials.gov). This study aims to assess the potential of the SPC3649 oligonucleotide as an antagonist of miR-122, a liver specific miRNA, in HCV therapy. The potential success of this clinical trial represents a proof of concept for the field of miRNA therapeutics, although lung cancer clinical trials have yet to begin.

Conclusions and Perspectives

There is a great need for the identification of non-invasive biomarkers for early detection, subtype classification and drug response prediction in lung cancer. The presence and stability of miRNAs in body fluids, such as in peripheral blood or sputum, and in FFPE samples makes them attractive candidates for this purpose. Furthermore, new techniques such as nanopore-based detection of miRNA in plasma samples of lung cancer patients reinforces their potential [98].

Despite all the advantages, several concerns need to be addressed before these molecules can be reliably translated as cancer biomarkers into clinical settings. First, research in this field is still in its infancy, with pilot studies showing promising results; however, most of them still involve a small number of individuals which could lead to the omission of confounding issues (e.g. effects derived from other diseases and lifestyle) that can alter miRNA expression. Another concern is related to the variability of miRNA profiles in different body fluids, as it is the case between profiles from serum and plasma, as well as for profiles between body fluid and tumor samples [99]. It is noteworthy to consider that miRNA profiles may change due to reasons not directly related to the tumor itself. For example, miRNA signatures can be influenced by miRNA expression from immune system and inflammation status [67–70]. On the other hand, miRNA expression levels from the peripheral immune system are dependent on the tumor status and can revert after tumor resection [100].

Subtype specificity of miRNAs should also be carefully considered, as some miRNAs (such as miR-486, miR-126, miR-139, miR-130b, and miR-210) have been identified as markers for AdC as well as for SqCC. Nevertheless, a few miRNAs such as the let-7 cluster in SqCC, miR-155/miR-191 in AdC, and miR-375/miR-98 in SCLC, show consistent results between studies and have a great potential to be used as reliable biomarkers for subtype identification.

A major concern related to the development of miRNA-based therapies is the complexity and widespread effects exerted by miRNAs. Since a single miRNA can regulate various genes, the potential off-target effects of miRNA therapeutics can lead to unpredictable side effects. Moreover, miRNAs can function in a cell type-specific manner, as some miRNAs can act as oncogenes in some subtypes of lung cancer and as tumor suppressors in others. It is also important to note that little information is available concerning epigenetic modulation of miRNAs [101–104] as well as SNPs or nucleotide variations in miRNAs and their target binding sites [32, 58].

Recent research has only begun to unveil the utility of some miRNAs as biomarkers and therapeutic targets. Future studies should focus on a better understanding of miRNA functions, regulation, and on maximizing the advantages related to target diversity. Certainly, the next decade will reveal new important features of these players in lung tumorigenesis, which will help with their translation from the bench to the clinical practice.