Diagnostic value of miR-200 family in non-small cell lung cancer: a meta-analysis

Abstract

Aim: To investigate the diagnostic potential of the miR-200 family for early detection in non-small cell lung cancer (NSCLC). Materials & methods: A systematic search was conducted of PubMed, Embase and Web of Science databases to identify studies of the miR-200 family in NSCLC. Sixteen studies meeting the inclusion criteria were included in the analysis with a total of 20 cohorts. Results: The combined sensitivity and specificity reached 73% and 85%, with an area under the curve of 0.83. Notably, miR-200b introduced heterogeneity. Subgroup analysis highlighted miR-200a and miR-141 as more sensitive, while blood-derived miRNAs showed slightly lower accuracy. Conclusion: The miR-200 family, predominantly assessed in blood, exhibits significant diagnostic potential for NSCLC, especially in distinguishing it from benign diseases.

Plain language summary

Summary points.

• Non-small cell lung cancer (NSCLC) is a major global health concern requiring effective early detection.

• The miR-200 family, known for its role in cancer processes, presents challenges and promises in NSCLC diagnosis.

• Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies guidelines were followed for a systematic search on miR-200 family expression in NSCLC.

• Databases such as Web of Science, PubMed and Embase were utilized.

• Sixteen studies, with twenty cohorts included, showed a pooled sensitivity of 73% and specificity of 85% for the miR-200 family in NSCLC diagnosis.

• Subgroup analysis revealed varied diagnostic performance among miR-200 subfamilies, with miR-200a and miR-141 showing higher sensitivity.

• Blood-derived miRNAs are slightly weaker in diagnostic accuracy compared with other sources.

• The miR-200 family exhibits good diagnostic potential for NSCLC, supported by an AUC of 0.83 and promising sensitivity of miR-200a and miR-141.

• Future research should address limitations and explore alternative sources like sputum.

• Identified limitations include potential selection bias, language restrictions and uneven study distribution among miR-200 subfamilies.

Non-small cell lung cancer (NSCLC) is a global health concern, with 2.2 million new cases and 1.8 million deaths in 2020, making it the second most diagnosed and leading cause of cancer death. Low survival rates highlight the importance of early detection strategies like low-dose CT screening, though these methods have limitations [1]. Comprising the majority of lung cancers (85%), NSCLC encompasses prevalent subtypes, namely adenocarcinoma (LUAD) and squamous cell carcinoma (SCC) [2]. The current diagnostic strategies for NSCLC include diverse methods. Imaging modalities like low-dose CT are recommended for high-risk individuals while bronchoscopy techniques, including autofluorescence bronchoscopy and narrow-band imaging, contribute to direct lesion visualization. Liquid biopsy approaches, especially miRNA analysis, are gaining prominence due to their noninvasive nature. Specific miRNAs, such as miR-429, miR-205 and miR-200b in serum samples, have shown potential for distinguishing different stages of NSCLC [3]. miRNAs are concise RNA sequences, typically 18–22 base pairs in length, that play a crucial role in post-transcriptional gene regulation. In humans, more than 2000 miRNAs that function by binding to mRNA and influencing either mRNA degradation or translation inhibition have been identified. These miRNAs can be located within cells or circulate through exosomes, making them stable and potential biomarkers. In cancer, they exhibit dual roles as oncomiRs promoting tumor growth and tumor-suppressor miRNAs inhibiting the development of tumors. The unique miRNA profiles observed in different types of tumors offer promising avenues for cancer diagnosis and therapeutic strategies [4]. The miR-200 family, comprising miR-200a, miR-200b, miR-200c, miR-141 and miR-429, is a set of highly conserved miRNAs playing a pivotal role in regulating epithelial-mesenchymal transition, a critical process in cancer metastasis. These miRNAs, encoded in two clusters, function by post-transcriptionally modulating gene expression, impacting key cellular aspects like adhesion, motility and invasion. Specifically in lung cancer, the miR-200 family emerges as a potential diagnostic tool [5]. The miR-200 family in lung cancer exhibits a dual role in influencing metastasis, angiogenesis and therapy resistance. While some members suppress tumor progression, others show tumor-promoting effects. Inconsistent expression patterns hinder their potential as diagnostic markers in tumor tissues and serum [6]. While the prognostic importance of the miR-200 family in various cancers has been assessed in several meta-analyses [7,8], its diagnostic potential remains unexplored in numerous cancer types through a meta-analytical approach. A prior meta-analysis investigated the diagnostic relevance of the miR-200 family in breast cancer. The altered expression of miR-200 in breast cancer demonstrated associations with various clinical outcomes, suggesting its potential as a diagnostic indicator. The analysis highlighted notable diagnostic accuracy, particularly emphasizing the effectiveness of miR-429 [9]. Therefore, the primary objective of this meta-analysis was to examine, for the first time, the diagnostic efficacy of the miR-200 family in NSCLC.

Materials & methods

Literature search

This meta-analysis was conducted following the guidelines outlined in the 2020 edition of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies statement, as detailed in the Supplementary Materials [10]. Three key electronic databases, Web of Science, PubMed and Embase, were systematically searched on October 30, 2023, tailoring the search approach to the unique features of each database. The inclusion criteria included considering only studies published in English. The search strategy included the specific terms: “lung cancer” AND “miR-200 family” AND “AUC” OR “sensitivity” AND “specificity”. Two independent reviewers applied these criteria along with pre-established inclusion and exclusion standards to evaluate the existing literature.

Study selection

The criteria for inclusion were as follows: utilization of human samples displaying miR-200 family expression in individuals diagnosed with NSCLC; confirmation of NSCLC through histopathological examination; comparison of cancer cases with normal controls or benign diseases; and availability of metrics such as true-positive, true-negative, false-positive and false-negative, either extractable or estimable from reported sensitivity and specificity, enabling the construction of a 2 × 2 contingency table. The exclusion criteria encompassed: editorials; reviews; letters, meetings or abstracts; case reports; duplicate studies; literature lacking reports of receiver operating characteristic curve and specific values for sensitivity and specificity; and in vitro and in vivo studies.

Data extraction

Two researchers independently obtained the necessary information from the eligible studies. The collected data included: first author, publication year, name of miRNA, country, subtype of NSCLC, cell lines evaluated, regulation of miRNA (oncogene miRNA or tumor-suppressor miRNA), sample type (tissue, blood or peripheral blood mononuclear cells), method of miRNA detection, number of cases and controls, reported area under the curve (AUC) and sensitivity and specificity. Any discrepancies were resolved by discussion to achieve a consensus and avoid bias.

Quality assessment

Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) is a tool designed for evaluating the quality and potential bias in diagnostic accuracy studies. It encompasses four key domains: patient selection, index test, reference standard and flow and timing. Each domain comprises specific questions and responses ‘yes’, ‘no’ or ‘unclear’, were employed to assess the risk of bias and applicability. The risk of bias was then categorized as ‘high’, ‘low’ or ‘unclear’ based on the provided answers. This assessment tool ensures a standardized and systematic approach to gauge the reliability and quality of diagnostic accuracy studies, considering factors like patient selection, test performance and reference standards [11]. Any discrepancies in assessments were resolved through discussion and consensus among reviewers. The quality of the included articles was assessed in RevMan 5.3 software.

Statistical analysis

This meta-analysis was conducted using the MIDAS module in STATA version 15.0 and OpenMeta[Analyst]. Meta-Disc software was employed to explore the threshold effect using Spearman's correlation coefficient. Pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR) and 95% CIs were calculated using a random-effects model. The diagnostic value of miRNA was assessed using the summary receiver operating characteristic (SROC) plot and AUC. Meta-regression and subgroup analyses were performed to explore sources of heterogeneity. Overall diagnostic performance was assessed using the SROC curve and AUC, with significance set at p < 0.05. Cochran's Q test and Higgins I² were used to evaluate heterogeneity, with I² >50% indicating significant heterogeneity. AUC classification criteria were 90–100% (excellent), 80–90% (good), 70–80% (fair), 60–70% (poor) and 50–60% (failure). Deeks' funnel plot was constructed to investigate publication bias. All p-values less than 0.05 were considered significant.

Results

Literature search

Initially, 223 articles were identified, with 47 duplicates removed, leaving 176 articles for screening. Subsequent screening based on titles and abstracts resulted in the exclusion of 133 irrelevant studies. A detailed review of the full text of the remaining 43 articles led to the exclusion of another 27. In the end, 16 articles met the inclusion criteria. The PRISMA flow chart, illustrating the inclusion and exclusion process, is presented in Figure 1.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart of included studies in quantitative synthesis.

Study characteristics

Table 1 shows the characteristics of the included studies. Sixteen studies were included in the meta-analysis [12,13,14–27]; among them, 11 studies were from China [13,14–16,18–21,24,25,27], two were from the USA [17,26], one was from Spain [12], one was from Japan [22] and one was from Iran [23]. NSCLC subtypes were LUAD in two studies [28,29], not mentioned in two studies [13,15] and a mix of LUAD and SCC/SCLC in the remaining studies [14,16,17–20,22–24,26,27]. Sources of miRNAs were blood in the majority of the studies [12–16,18,19,21,23–27], followed by tumoral tissues [20], sputum [21] and lymph nodes [22]. miR-200b was evaluated in nine studies, followed by miR-141, which was evaluated in five studies [14,20,23,24,26], then miR-200c, miR-429 and miR-200a in two [18,22], two [12,19] and one [21] studies, respectively. The control group consisted of healthy individuals in seven studies [13,16,17,19,21,23,24], and in the remaining studies consisted of patients with benign lesions [14,15,18,22,25–27] or diseases such as COPD [12].

Table 1.

Characteristics of included studies.

Author (year)	miRNA	Country	Subtype	Sample source	Control	Sensitivity	Specificity	AUC	Ref.
Halvorsen et al. (2016)	miR-429	Spain	LUAD	Blood	COPD and healthy	0.73	0.79	0.79	[12]
	miR-200b					0.83	0.73	0.83
Zou et al. (2019)	miR-200b	China	NM	Blood	Healthy	0.55	0.62	0.673	[13]
Wu et al. (2020)	miR-141-3p	China	LUAD, SCC	Blood	Healthy and benign	0.83	0.56	0.748	[14]
Xi et al. (2019)	miR-200b	China	NM	Blood	Healthy and benign	0.679	0.833	0.699	[15]
						0.74	0.61	0.724
Li et al. (2019)	miR-200b	China	LUAD, SCC	Blood	Healthy	0.833	1.0	0.87	[16]
Yu et al. (2010)	miR-200b	The USA	LUAD	Sputum	Healthy	0.629	0.785	0.823	[17]
Wang et al. (2020)	miR-200b	China	Mainly LUAD, SCC	Blood	Benign	0.346	0.929	0.637	[18]
Zhu et al. (2014)	miR-429	China	LUAD, SCC	Blood	Healthy	0.543	0.812	0.713	[19]
Zhang et al. (2015)	miR-141	China	LUAD, SCC	Tissue	Noncancer tissues	0.648	0.648	0.707	[20]
Yu et al. (2022)	miR-200a	China	LUAD, SCC	Blood	Healthy	0.833	0.633	0.721	[21]
Inage et al. (2018)	miR-200c	Japan	LUAD, SCC	Lymph nodes	Benign	0.954	1.0	0.949	[22]
Arab et al. (2017)	miR-141	Iran	LUAD, SCC	Blood	Healthy	0.827	0.98	0.918	[23]
Zhao (2018)	miR-141	China	LUAD, SCC	Blood	Healthy	0.75	0.85	0.856	[24]
Xi et al. (2018)	miR-200b	China	NM	Blood	Benign	0.643	0.867	0.708	[25]
Yang et al. (2019)	miR-141	The USA	LUAD, SCC, SCLC	Blood	Benign	0.78	0.95	0.934	[26]
	miR-200b					0.76	0.80	0.876
Wang et al. (2021)	miR-200b	China	LUAD, SCC	Blood	Benign	0.671	0.953	0.698	[27]
	miR-200c					0.697	0.944	0.769

LUAD: Lung Adenocarcinoma; NM: Not Mentioned; SCC: Squamous Cell Carcinoma; SCLC: Small Cell Carcinoma.

Quality assessment

The quality of the included studies was evaluated using the QUADAS-2 tool (Figure 2). In the patient selection domain, some risks of bias was observed for three studies, including Inage et al. [22], Yang et al. [26] and Zhang et al. [20], due to receiving treatment, including some patients with SCLC, using normal tissues of patients with cancer (instead of healthy patients), respectively. In addition, some unclear risks of bias in the patient selection domain existed as the inclusion and exclusion criteria for patient selection were not clearly defined [16,18,23]. However, the overall risks and applicability concerns for the patient selection domain were less than 40%, as demonstrated in Figure 2B. No risk of bias or applicability concern was identified for the index test section. In the reference standard domain, the risk of bias existed only in one study due to not providing enough information on how patients were diagnosed with NSCLC [16]. Lastly, some unclear risks of bias were deemed for the flow and timing section due to not mentioning if there was an appropriate interval between the index test and reference standard (Figure 2A).

Figure 2.

Quality Assessment of Diagnostic Accuracy Studies 2.

(A) Results of quality assessment per study. (B) Results of quality assessment per domain. Figure A depicts risk of bias assessment and applicability concerns across various domains of the QUADAS questions, encompassing patient selection, index test and reference standard. Notably, the flow and timing domain is solely evaluated for risk of bias and not for applicability concerns. On the other hand, Figure B provides an overview of overall study quality, considering both risk of bias and applicability concerns within each domain. Red bars indicate a high risk of bias, with the noteworthy observation that the patient selection domain exhibited an overall high-risk rate of less than 20%. Conversely, other domains received acceptable risk of bias scores and demonstrated low applicability concerns.

Diagnostic meta-analysis

Twenty cohorts were included in the meta-analysis. The pooled sensitivity, specificity, PLR, NLR and DOR with 95% CI were 0.73 (95% CI: 0.67–0.78), 0.85 (95% CI: 0.78–0.91), 5.0 (95% CI: 3.2–7.9), 0.32 (95% CI: 0.26–0.40) and 16 (95% CI: 8–29), respectively. The coupled forest plot showing sensitivity and specificity with heterogeneity indicators (Higgins' I² and Cochran's Q) is shown in Figure 3. The SROC plot is shown in Figure 4, yielding an AUC of 0.83 (95% CI: 0.80–0.86), indicating good diagnostic performance.

Figure 3.

Forest plots showing pooled sensitivity and specificity.

Figure 4.

Summary receiver operating characteristic curve with confidence and prediction regions around each study.

The 95% confidence region provides a band around the SROC curve, expressing confidence that the true curve lies within this range. It accounts for variability in estimated parameters. On the other hand, the 95% prediction region is wider and incorporates both parameter uncertainty and future variability in study outcomes. This region allows prediction with 95% confidence where the SROC curve for a new study is likely to fall. The overall measured area under the curve is 0.83, which indicates a good diagnostic ability of the test.

SROC: Summary receiver operating characteristic curve.

Heterogeneity

The Cochran's Q and Higgins I² tests revealed significant heterogeneity in pooled sensitivity (p = 0.00; I² = 75.82%) and specificity (p = 0.00; I² = 88.24%) values. The presence of a threshold effect as a potential cause of heterogeneity was ruled out, as indicated by a Spearman's correlation coefficient of -0.101 with a p-value of 0.672. Meta-regression was employed to explore potential sources of heterogeneity (see Table 2). Among the examined covariates, including sample size (with a cutoff of 150), country, sources of miRNAs (blood or tissues), subfamily of miR-200 family and the use of patients with benign diseases in control groups, only the miR-200b subfamily contributed to the heterogeneity of the results (p = 0.05).

Table 2.

Results of subgroup analysis using meta-regression.

Subgroups		N	Sensitivity	PSEN	Specificity	PSPEC	Bivariate analysis
							Log (likelihood RT)	p-value
Sample size	>150	8	0.74 (0.67–0.81)	0.01	0.87 (0.78–0.95)	0.16	0.26	0.88
	<150	12	0.72 (0.64–0.79)		0.84 (0.75–0.94)
Region	China	15	0.70 (0.64–0.76)	0.00	0.84 (0.76–0.92)	0.05	3.71	0.16
	Others	5	0.80 (0.72–0.88)		0.90 (0.80–0.99)
Source	Blood	17	0.72 (0.67–0.78)	0.09	0.85 (0.78–0.92)	0.37	0.13	0.94
	Others	3	0.74 (0.60–0.89)		0.87 (0.73–1.00)
Control	Healthy	8	0.71 (0.62–0.79)	0.00	0.82 (0.70–0.93)	0.02	0.96	0.62
	Benign	12	0.74 (0.67–0.80)		0.88 (0.81–0.95)
miR-200 family	miR-200a	2	0.89 (0.80–0.99)	0.73	0.91 (0.76–1.00)	0.71	5.87	0.05
	Others	18	0.71 (0.66–0.76)		0.85 (0.78–0.92)
	miR-200b	10	0.69 (0.61–0.76)	0.00	0.84 (0.74–0.94)	0.05	2.13	0.35
	Others	10	0.76 (0.70–0.83)		0.87 (0.79–0.95)
	miR-200c	1	0.70 (0.46–0.93)	0.40	0.95 (0.84–1.00)	0.18	1.34	0.51
	Others	19	0.73 (0.67–0.78)		0.84 (0.78–0.91)
	miR-141	5	0.77 (0.69–0.85)	0.06	0.85 (0.72–0.97)	0.16	1.34	0.51
	Others	15	0.71 (0.65–0.77)		0.86 (0.78–0.93)
	miR-429	2	0.64 (0.47–0.82)	0.05	0.81 (0.57–1.00)	0.44	1.16	0.56
	Others	18	0.74 (0.68–0.79)		0.86 (0.79–0.92)

Analysis performed using STATA software.

Subgroup analysis

Different factors were considered for subgroup analysis and are described in Table 2.

Study sample size

Studies with sample sizes larger than 150 were shown to have a higher pooled sensitivity (0.74 vs 0.72) and specificity (0.84 vs 0.87), and this difference was more significant for pooled sensitivity (p = 0.01).

Region

Studies conducted in China had a significantly lower pooled sensitivity (0.70 vs 0.80) and specificity (0.84 vs 0.90; p < 0.05) compared with other countries.

Type of control group

miR-200 family miRNAs differentiated patients with NSCLC from patients with benign diseases compared with healthy individuals with a significantly higher (p < 0.05) sensitivity (0.74 vs 0.71) and specificity (0.82 vs 0.88).

Sources of miRNAs

Blood (plasma)-derived miRNAs were shown to have slightly lower sensitivity (0.72 vs 0.74) and specificity (0.85 vs 0.87), but the differences were not statistically significant (p > 0.05).

miR-200 subfamilies' diagnostic performance

To illustrate the diagnostic performance of each miR-200 subfamily individually, a subgroup analysis was performed in OpenMeta[Analyst] software, as shown in Figure 5. In terms of sensitivity, miR-200a had the highest pooled value (sensitivity = 0.86), followed by miR-141 (sensitivity = 0.76), miR-200c (sensitivity = 0.69), miR-200b (sensitivity = 0.68) and miR-429 (sensitivity = 0.64). However, in terms of specificity, miR-200c had the highest value (specificity = 0.94), followed by miR-200a (specificity = 0.91), miR-141 (specificity = 0.84), miR-200b (specificity = 0.81) and miR-429 (specificity = 0.80). Therefore, the miR-200 family was shown to have a greater specificity in diagnosing NSCLC. It should be noted that the results in this section were slightly different from those conducted in STATA software due to different modeling in meta-analyses.

Figure 5.

Subgroup analysis based on miR-200 family for sensitivity and specificity.

Publication bias

No significant publication bias was found in the included studies based on Deeks' asymmetry test (p = 0.74; Figure 6).

Figure 6.

Deeks' funnel plot showed no significant publication bias (p = 0.74).

Deeks' funnel plot is a visual tool used to assess the symmetry of study effect sizes in a meta-analysis, and a nonsignificant p-value indicates no statistically significant evidence of publication bias. In this context, the lack of asymmetry in the funnel plot suggests that the included studies were not selectively reported based on their results. This finding enhances the credibility of the meta-analysis results, suggesting that any observed effects are likely due to genuine study differences rather than biased reporting.

Discussion

Lung cancer, a major global health concern [30], poses diagnostic challenges [31], but miRNAs show promise as diagnostic tools due to their stability in bodily fluids [32]. The diagnosis of lung cancer faces challenges primarily due to late detection, invasive procedures and dependence on the expertise of bronchoscopists [33]. Current diagnostic tools include bronchoscopy [34], invasive biopsies [35], CT scans [36] and PCR techniques for detecting genetic mutations [37]. Liquid biopsies analyzing genetic, transcriptomic and epigenetic markers in body fluids offer a noninvasive approach [38]. Despite progress, challenges persist, emphasizing the need for more reliable and cost-effective early diagnostic tests to improve outcomes for patients with lung cancer [31]. Dysregulation of oncogenes and tumor suppressors is frequently seen in different cancers, particularly in NSCLC [39–41]. A significant portion of the human genome is transcribed into ncRNAs, and these ncRNAs play crucial roles in various cellular processes and regulatory mechanisms in cancers by acting as tumor suppressors or oncogenes [42,43]. miRNAs are small RNA molecules that regulate gene expression at transcriptional and post-transcriptional levels, influencing various biological processes [44]. Diverse testing methods, including real-time PCR, microarray analysis and advanced software, are employed to identify abnormal gene and miRNA expression [45]. Diagnostic tools such as the miR-test show notable accuracy in discerning different subtypes of lung cancer. miRNAs, known for their resistance to degradation, provide a noninvasive and efficient method for early detection, subtype classification and monitoring of lung cancer. This underscores their pivotal role in the field of diagnostics [46,47]. The miR-200 family, which includes miR-200a, miR-200b, miR-200c, miR-141 and miR-429, plays a crucial role in cancer by overseeing essential disease characteristics. Changes in miR-200 family expression, often linked to both transcriptional and post-transcriptional processes, impact the development and progression of cancer [48,49]. Examination of expression patterns in cancer tissues, body fluids and exosomes indicates the potential for diagnostic and prognostic applications [5]. While increased miR-200 family expression often associates with adverse survival outcomes, its nuanced role in distinct cancer types and its links to evolving cancer hallmarks underscore the need for additional investigation [8,50]. Generally acknowledged as tumor suppressors, these miRNAs exert regulatory control over fundamental biological processes such as cell proliferation, differentiation and apoptosis. Notably, the miR-200 family plays a significant role in impeding epithelial-mesenchymal transition and suppressing tumor metastasis while also influencing cancer stem cell behavior and chemoresistance reversal. However, the miR-200 family's functional dynamics are nuanced, contingent on factors such as cellular context, tumor stage and the subcellular localization of their targets. Despite the prevailing recognition of their tumor-suppressive role, the miR-200 family's influence on cancer prognosis appears multifaceted and context-dependent, with contradictory results reported in the literature [7]. Consequently, the miR-200 family emerges as a promising avenue for comprehending, diagnosing and potentially treating a diverse array of cancers [48]. A previous meta-analysis established a positive correlation between the miR-200 family and the survival of patients with cancer with a hazard ratio of 1.2 [7]. In patients with ovarian cancer, miR-200 family expression is detectable in samples obtained from tumoral tissue, plasma and urine, indicating a helpful liquid biopsy tool for cancer diagnosis [51]. In addition, the miR-200 family was shown to diagnose breast cancer with a pooled sensitivity and specificity of 0.70 and 0.72, respectively, based on a meta-analysis of five studies [9]. In the current research, these diagnostic values, particularly specificity, were significantly higher, indicating that the miR-200 family can be an excellent diagnostic tool for NSCLC.

In this meta-analysis, we combined data from 16 studies with 20 cohorts to explore the diagnostic accuracy of the miR-200 family for the first time in NSCLC. The findings indicate that the miR-200 family exhibits good diagnostic capability (AUC = 0.83) in distinguishing patients with NSCLC from those without cancer. The pooled specificity for miR-200 in detecting NSCLC is promising, reaching nearly 85%. Moreover, the pooled sensitivity is 73%, suggesting that miR-200 performs notably better in NSCLC detection compared with breast cancer.

Subgroup analysis highlights that miR-200a and miR-141 demonstrate superior sensitivity compared with other family members. All miR-200 members exhibit a pooled sensitivity higher than 80%, with miR-200c showing the highest sensitivity at 94%. Subgroup analysis also suggests that the miR-200 family can effectively differentiate patients with NSCLC from those with benign diseases, making these miRNAs potentially valuable for suspected patients, offering enhanced diagnostic performance. Additionally, blood-derived miRNAs had slightly weaker diagnostic accuracy compared with other sources such as tissue, sputum and lymph nodes. Notably, most studies focused on evaluating these miRNAs in the blood of patients. Consequently, further research with alternative sources, especially sputum, is essential to provide a more comprehensive understanding of miR-200 family diagnostic accuracy in NSCLC.

This meta-analysis is subject to several limitations. First, potential selection bias may exist due to the constraints of the electronic database search, limiting the inclusivity of relevant studies. Additionally, the restriction to English-language publications may introduce bias, excluding valuable data from non-English sources. The predominantly Chinese origins of the included studies raises concerns about the generalizability of the findings. Furthermore, an uneven distribution of studies among miR-200 subfamilies, with some subtypes having fewer studies than others, may impact the overall assessment of diagnostic performance. The predominant use of blood as the source for miR-200 family evaluation, with limited exploration of alternative sources like sputum, also limits the comprehensiveness of this analysis. Lastly, while heterogeneity was identified in the miR-200b subtype, sources of heterogeneity in other subtypes were not delineated, urging caution in extrapolating these results across the entire miR-200 family. Addressing these limitations in future research will enhance the robustness and applicability of insights into the diagnostic role of the miR-200 family in NSCLC.

Conclusion

In summary, these meta-analytic findings strongly indicate that the miR-200 family holds significant promise as a reliable and effective diagnostic tool for NSCLC. The analysis, involving 16 studies and 20 cohorts, revealed good diagnostic ability with an AUC of 0.83, showcasing the potential of miR-200 family members, including miR-200a, miR-200b, miR-200c, miR-141 and miR-429, in distinguishing patients with NSCLC from those without cancer. Furthermore, the findings support that utilizing miR-200 family members for suspected patients, especially those with benign diseases, could yield higher diagnostic performance. Despite certain limitations in study inclusivity and source diversity, these results highlight the promising role of the miR-200 family in enhancing the diagnostic landscape for NSCLC. Continued research and exploration, particularly into alternative sources like sputum, will further contribute to validating and refining the diagnostic potential of the miR-200 family in lung cancer detection.

Supplementary data

To view the supplementary data that accompany this paper please visit the journal website at: www.futuremedicine.com/doi/suppl/10.2217/bmm-2024-0087

Author contributions

L Yang: Conceptualization, methodology, writing original draft, review and editing. Z Ling: Methodology, writing original draft, review, editing and formal analysis.

Financial disclosure

This study was funded by Postgraduate Research & Practice Innovation Program of Jiangsu Province (no: SJCX23_09690).

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, stock ownership or options and expert testimony.

Writing disclosure

During the preparation of this work, the authors used ChatGPT by OpenAI to improve paper readability. After using this tool/service, the authors reviewed and edited the content as needed and took full responsibility for the publication's content.