Prognostic and clinicopathological significance of miR-638 in cancer patients: A meta-analysis.

INTRODUCTION: MiR-638 is believed to be involved in human cancers. However, the prognostic value of miR-638 in human carcinomas is controversial and inconclusive. Therefore, we conducted this meta-analysis to investigate the association between miR-638 expression and clinical outcomes in the patients with various cancers.
METHODS: We searched Pubmed, Embase, Wanfang, and the China National Knowledge Infrastructure (CNKI) up to September 1, 2020 to identify relevant studies. Hazard ratios (HRs) and odds ratios (ORs) with 95% confidence intervals (CIs) were used to correlate expression of miR-638 with prognosis and clinicopathological features.
RESULTS: A total of 18 studies involving 1886 patients were included in the meta-analysis. The results revealed that low miR-638 expression was significantly correlated with poor overall survival (OS) (HR = 2.09, 95% CI: 1.46-2.98, P < .001), but not with disease-free survival (DFS) (HR = 1.71, 95% CI: 0.31-9.56, P = .540). Subgroup analysis found that low miR-638 expression was associated with worse OS in patients with digestive system cancer (HR = 2.47, 95% CI: 1.85-3.30, P < .001), the reported directly from articles group (HR = 2.12, 95% CI: 1.34-3.33, P < .001), survival curves group (HR = 2.02, 95% CI: 1.07-3.80, P = .029), in studies with sample size ≥100 (HR = 2.12, 95% CI: 1.34-3.35, P = .001), and in studies with sample size <100 (HR = 2.02, 95%CI: 1.09-3.75, P = .025). Moreover, cancer patients with low miR-638 expression were prone to tumor size (OR = 1.47, 95% CI: 1.03-2.09, P = .035), earlier lymph node metastasis (present vs absent, OR = 2.26, 95% CI: 1.63-3.14, P < .001), earlier distant metastasis (present vs absent, OR = 2.60, 95% CI: 1.45-4.67, P < .001), TNM stage (III-IV vs I-II, OR = 2.01, 95% CI: 1.35-2.99, P = .001), and portal vein invasion (present vs absent, OR = 4.39, 95% CI:2.23-8.64, P < .001), but not associated with age, gender, tumor differentiation, and vascular invasion.
CONCLUSIONS: MiR-638 may serve as a promising indicator in the prediction of prognosis and clinicopathological features in patients with different kinds of cancers.

Introduction

Cancer is one of the leading causes of morbidity and mortality worldwide.[ Due to the lack of specific symptoms of early cancer, cancer is often diagnosed at an advanced stage.[ Despite continuously improving new treatment approaches including surgery, chemotherapy, radiotherapy and biological therapy, the prognosis of patients with cancer is still poor. Rising evidence demonstrated that the prognosis could be improved by using some molecular biomarkers. Therefore, identification of novel precise biomarkers is important and needed for the development of prognosis in cancer patients.

MicroRNAs (miRNAs), as a new type of biomarker, are small noncoding molecules of with a length of approximately 18 to 24 nucleotides, and can negatively regulate their target genes expression.[ Many miRNAs have been confirmed to express abnormally in human cancers and can play an important role in different biological processes, such as cell cycle control,[ proliferation,[ differentiation,[ metastasis,[ and carcinogenesis.[ Furthermore, a large number of miRNAs have been identified to function as oncogenes or tumor suppressor genes in the tumorigenesis process.[

The human miR-638 is located in the 19p13.2 region. Previous studies have indicated that miR-638 plays an important role in several tumors.[ Tang et al found that the expression of miR-638 was significantly decreased in oral squamous cell carcinoma (OSCC) tissues and cells and miR-638 could suppress the wnt/ b-catenin signaling pathway through PLD1, thus inhibiting OSCC progression.[ Low expression of miR-638 is associated with a worse survival in patients with different cancers, such as breast cancer, colorectal carcinoma, hepatocellular carcinoma, cervical cancer, and lung cancer.[ On the other hand, Zhang et al did not observe a significant association between miR-638 expression and overall survival (OS) of gastric cancer patients.[ Because of the controversy involving the association between miR-638 and survival among patients with different carcinomas, we constructed this meta-analysis to explore the correlation between miR-638 expression and prognosis in various cancers.

Methods

Search strategies

The present systematic review and meta-analysis was conducted and reported according to the standards of quality detailed in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.[ We searched Pubmed, Embase, Wanfang, and the China National Knowledge Infrastructure (CNKI) up to September 1, 2020 to identify relevant studies. The following keywords were adopted according to the retrieval strategy: “miR-638” OR “microRNA-638”; “cancer” OR “tumor” OR “carcinoma” OR “malignancy.” Further manual inspection was performed to improve the integrity of the eligible papers by going through the title and abstract. Moreover, references in relevant publications were also browsed. The present study was meta-analysis and did not involve the collection of samples. Therefore, ethical approval was not required.

Inclusion and exclusion criteria

Inclusion criteria were as follows:

research evaluated the association between miR-638 and cancer prognosis;

tumors were confirmed by histological or pathological examinations;

sufficient data was available for calculating the hazard ratio (HR) or odds ratio (OR) with their 95% confidence interval (CI).

The exclusion criteria were as follows:

letters, case reports, reviews, and conference abstracts without original data;

duplicate publications;

articles from which the relevant data could not be extracted.

Data extraction and quality assessment

Data extraction was conducted independently by 2 investigators from identified research in agreement with prescribed standards, during which disagreements were resolved by reaching a consensus on all contents. The extracted data elements mainly included the following information: author, publication year, country, cancer type, sample size, recruitment time, detection method, endpoints, HR obtain method, and NOS score. Additionally, clinical-pathological parameters, including age, gender, tumor size, tumor differentiation, lymph node metastasis, TNM stage, distant metastasis, portal vein invasion, and vascular invasion, were also extracted. When HRs and their 95% CIs were given in the articles, these data were extracted directly. If the prognosis was plotted as Kaplan–Meier survival curve, the data were digitized by the software Engauge Digitizer version 4.1 and calculated as described.[ The Newcastle–Ottawa Scale (NOS) was used to assess the quality of included studies.[ This method comprised 3 parameters of quality: selection (score: 0–4), comparability (score: 0–2), and outcome assessment (score: 0–3), with total scores ranging from 0 to 9. The study with total scores greater than 6 was considered high quality in the present meta-analysis.

Statistical analysis

HRs with 95% CIs were calculated the association between miR-638 expression and the OS and disease-free survival (DFS) of cancer patients. ORs with 95% CIs were used to assess the association of miR-638 expression with clinicopathological characteristics. The evaluation of statistical heterogeneity was finished by using the Cochran Q statistic and I tests.[ If the heterogeneity was significant between studies (I > 50% or P < .10), the random-effects model was used; otherwise, the fixed-effects model was used.[ Both Begg test and Egger test were used to evaluate the potential publication bias.[ The statistical analysis was performed using STATA version 12.0 software (Stata Corporation, Collage Station, Texas). All P values were two-sided and P < .05 was considered statistically significant.

Results

Literature search and study characteristics

The literature screening process is illustrated in Figure 1. Finally, 18 studies comprising 1886 patients were identified as eligible for the present quantitative analysis.[ Thirteen of the 18 articles focused on the association of miR-638 with OS, and 3 articles investigated DFS. Moreover, 15 studies reported data on the relation between miR-638 expression and clinicopathological features. The expression of miR-638 in tissue samples was measured by quantitative reverse transcription-polymerase chain reaction (qRT-PCR). All of the included studies were performed in China and were published from 2014 to 2020. NOS used for evaluating quality of included studies varied from 6 to 8. The basic characteristics of the involved studies are presented in Table 1.

Figure 1

Flow diagram of the literature selection process.

Table 1

Characteristics of the included studies.

Author (year)	Country	Cancer type	Sample size	Recruitment time	Detection method	Endpoints	HR obtain method	NOS score
Xia et al (2014)	China	NSCLC	107	2010.3–2013.6	qRT-PCR	NR	NR	6
Zhang et al (2014)	China	CRC	113	2005–2008	qRT-PCR	OS	reported directly	8
Zhang et al (2015)	China	Gastric cancer	64	2008–2012	qRT-PCR	OS	survival curves	7
Wang et al (2015)	China	NSCLC	189	2009.7–2013.3	qRT-PCR	OS	reported directly	7
Cheng et al (2016)	China	HCC	100	2008.01–2014.06	qRT-PCR	OS	survival curves	7
Zhang et al (2017)	China	HCC	113	2007.3–2009.6	qRT-PCR	OS, DFS	reported directly	8
Wei et al (2017)	China	Cervical cancer	196	2008.7–2012.1	qRT-PCR	OS, DFS	reported directly	6
Wang et al (2017)	China	Osteosarcoma	64	2008–2012	qRT-PCR	OS	survival curves	7
Yan et al (2017)	China	Colon cancer	192	2008–2014	qRT-PCR	OS, DFS	OS:reported directlyDFS: survival curves	8
Ye et al (2018)	China	HCC	78	2011.1–2012.7	qRT-PCR	OS	reported directly	7
Shi et al (2018)	China	HCC	126	2008–2011	qRT-PCR	OS	reported directly	7
Zheng et al (2018)	China	Glioma	24	2011.9–2013.12	qRT-PCR	OS	survival curves	8
Li et al (2018)	China	Breast cancer	125	2007.1–2012.12	qRT-PCR	OS	reported directly	7
Pan et al (2020)	China	RCC	118	2013-2015	qRT-PCR	OS	reported directly	7
Tang et al (2019)	China	OSCC	107	NR	qRT-PCR	NR	NR	6
Cheng et al (2017)	China	HCC	60	2008–2010	qRT-PCR	NR	NR	6
Shen et al (2017)	China	Gastric cancer	68	2010–2014	qRT-PCR	NR	NR	6
Yang et al (2020)	China	HCC	42	NR	qRT-PCR	NR	NR	6

The correlation between miR-638 expression and the prognosis of cancers

Thirteen studies with 1502 patients were included in the meta-analysis of OS. The main results of this meta-analysis are listed in Table 2. The results indicated that low expression of miR-638 was highly correlated with poor prognosis of OS (HR = 2.09, 95% CI: 1.46–2.98, P < .001) (Fig. 2). Three studies with 501 patients were included in the meta-analysis of DFS. The results indicated that no connection was identified between miR-638 expression and DFS (HR = 1.71, 95% CI: 0.31–9.56, P = .540) (Fig. 3).

Table 2

Main meta-analysis results of miR-638 expression in cancer patients.

				Heterogeneity
Analysis	Numbers of studies	HR (95%CI)	P value	χ2	I2 (%)	P value
Overall survival (OS)	13	2.09 (1.46–2.98)	<.001	33.77	64.5	.001
Cancer type
digestive system cancer	7	2.47 (1.85–3.30)	<.001	3.62	0.0	.728
other cancer	6	1.78 (0.90–3.50)	.096	25.68	80.5	<.001
HR obtain method
survival curves	4	2.02 (1.07–3.80)	.029	6.32	52.5	.097
reported directly	9	2.12 (1.34–3.33)	.001	26.89	70.2	.001
Sample size
≥100	9	2.12 (1.34–3.35)	.001	26.86	70.2	.001
<100	4	2.02 (1.09–3.75)	.025	6.46	53.6	.091
Disease-free survival (DFS)	3	1.71 (0.31–9.56)	.540	20.38	90.2	<.001
Cancer type
digestive system cancer	2	1.18 (0.05–26.74)	.916	18.10	94.5	<.001
HR obtain method
reported directly	2	0.92 (0.07–12.26)	.951	14.78	93.2	<.001

Figure 2

Forest plot of HRs for correlation between miR-638 expression and overall survival (OS).

Figure 3

Forest plot of HRs for correlation between miR-638 expression and disease-free survival (DFS).

Subgroup analysis

Next, we proceeded with subgroup analyses stratified by cancer type, HR obtain method and sample size for OS. We found that low miR-638 expression was a powerful prognostic marker for shorter OS in patients with digestive system cancer (HR = 2.47, 95% CI: 1.85–3.30, P < .001), but not with other cancers (HR = 1.78, 95% CI: 0.90–3.50, P < .001) (Fig. 4). Subgroup analysis based on the HR obtain method suggested that low expression of miR-638 predicted poor OS for both the reported directly from articles group (HR = 2.12, 95% CI: 1.34–3.33, P < .001) and, survival curves group (HR = 2.02, 95% CI: 1.07–3.80, P = .029) (Fig. 5). Furthermore, the subgroup analyses classified by sample size validated that low expression of miR-638 was a unfavorable prognostic factor in studies with sample size ≥100 (HR = 2.12, 95% CI: 1.34–3.35, P = .001), and in studies with sample size <100 (HR = 2.02, 95% CI: 1.09–3.75, P = .025) (Fig. 6).

Figure 4

Subgroup analysis of the HR of overall survival (OS) by cancer type.

Figure 5

Subgroup analysis of the HR of overall survival (OS) by HR obtain method.

Figure 6

Subgroup analysis of the HR of overall survival (OS) by sample size.

Nevertheless, there was no significant association between miR-638 expression and DFS in patients with digestive system cancer (HR = 1.18, 95% CI: 0.05–26.74, P = .916), or the reported directly from articles group (HR = 0.92, 95% CI: 0.07–12.26, P = .951).

Association between miR-638 expression and clinicopathological characteristics

Meta-analysis of the relationship between miR-638 expression and clinicopathological characteristics failed to show a significant association of low miR-638 expression with age (OR = 0.85,95% CI: 0.63–1.15, P = .293), gender (OR = 1.06, 95% CI: 0.83–1.34, P = .656), tumor differentiation (poor vs well+moderate, OR = 1.42, 95% CI: 0.97–2.06, P = .068), or vascular invasion (present vs absent, OR = 2.35, 95% CI: 0.81–6.83, P = .117) (Table 3).

Table 3

Results of the association of miR-638 expression with clinicopathological features.

Clinicopathological parameter	N	OR (95% CI)	P value	Heterogeneity test (Q, I2, P-value)
Age (<60 vs ≥60 years )	7	0.85 (0.63-1.15)	.293	2.48, 0.0%, .871
Gender (male vs female)	13	1.06 (0.83-1.34)	.656	13.76, 12.8%, .316
Tumor size (≥5 vs <5)	7	1.47 (1.03-2.09)	.035	10.31, 41.8%, .112
Tumor differentiation (poor vs well+moderate)	4	1.42 (0.97-2.06)	.068	2.02, 0.0%, .732
Lymph node metastasis (Present vs Absent)	6	2.26 (1.63-3.14)	<.001	1.18, 0.0%, .947
TNM stage (III-IV vs I-II)	13	2.01 (1.35-2.99)	.001	31.57, 62.0%, .002
Distant metastasis (Present vs Absent)	4	2.60 (1.45-4.67)	<.001	4.78, 37.3%, .188
Portal vein invasion (Present vs Absent)	3	4.39 (2.23-8.64)	<.001	0.05, 0.0%, .975
Vascular invasion (Present vs Absent)	3	2.35 (0.81-6.83)	.117	7.29, 72.6%, .026

In contrast, low miR-638 expression was significantly related to tumor size (OR = 1.47, 95% CI: 1.03–2.09, P = .035), earlier lymph node metastasis (present vs absent, OR = 2.26, 95% CI: 1.63–3.14, P < .001), earlier distant metastasis (present vs absent, OR = 2.60, 95% CI: 1.45–4.67, P < .001), TNM stage (III-IV vs I-II, OR = 2.01, 95% CI:1.35–2.99, P = .001), and portal vein invasion (present vs absent, OR = 4.39, 95% CI: 2.23–8.64, P < .001) (Table 3).

Sensitivity analysis

For the purpose of assessing the reliability and stability of our results, sensitivity analysis was contacted by sequentially omitting any individual cohort analysis. Fortunately, the pooled HR for OS was not influenced, which meant increased credibility (Fig. 7).

Figure 7

Sensitivity analysis of miR-638 expression and overall survival (OS).

Publication bias

Begg test was performed to evaluate the publication bias of the meta-analysis. The results indicated that there was no significant publication bias in this meta-analysis for OS (P = .381) and DFS (P = .058) (Fig. 8).

Figure 8

Begg funnel plots for the studies included in meta-analysis. OS (A) and DFS (B).

Discussion

Accumulated evidences exerted that miR-638 acted critical roles in physiological and pathological processes via their regulation of a wide variety of genes. For example, Cheng et al[ showed that miR-638 was downregulated in hepatocellular carcinoma and could repress tumor growth and inhibit angiogenesis by downregulating vascular endothelial growth factor (VEGF). Moreover, downregulated miR-638 has been shown to induce cell invasion and proliferation by regulating SRY (sex determining region Y)-box 2 (SOX2), which is related to epithelial-to-mesenchymal transition (EMT) in the development of non-small cell lung cancer.[ Similarly, another study found that loss of miR-638 repressed cell proliferation and colony formation in patients with osteosarcoma by targeting suppress proviral integration site for Moloney murine leukemia virus 1 expression.[ Those results suggested that miR-638 may serve as a tumor suppressor. However, miR-638 has also been reported to promote melanoma progression and metastasis by suppressing p53 pro-apoptotic signaling as an oncogene.[ Furthermore, Ren et al also identified miR-638, as a potential oncogene, promoting tumorigenic properties, including cell proliferation, migration, and invasion in esophageal squamous cell carcinoma (ESCC) and breast cancer.[ Thus, miR-638 may serve as an oncogene according to the kind of tumors.

The current study presented the first meta-analysis to comprehensively evaluate the relationship between miR-638 expression and prognosis and clinicopathological characteristics of tumors. In the present study, a total of 18 eligible studies containing 1886 patients were enrolled in this meta-analysis. The results provided strong evidence that low expression of miR-638 was significantly correlated with shorter OS. However, similar result was not seen in DFS. The negative outcome of correlation between miR-638 and DFS might result from the fewer number of the included study and smaller sample size. Subgroup analysis found that low miR-638 expression was associated with worse OS in patients with digestive system cancer (HR = 2.47, 95% CI: 1.85–3.30, P < .001), the reported directly from articles group (HR = 2.12, 95% CI: 1.34–3.33, P < .001), survival curves group (HR = 2.02, 95% CI: 1.07–3.80, P = .029), in studies with sample size ≥100 (HR = 2.12, 95% CI: 1.34–3.35, P = .001), and in studies with sample size <100 (HR = 2.02, 95% CI: 1.09–3.75, P = .025).

In this meta-analysis, the association between expression levels of miR-638 and clinicopathological characteristics was evaluated. Cancer patients with low miR-638 expression were prone to tumor size (OR = 1.47, 95% CI: 1.03–2.09, P = .035), earlier lymph node metastasis (present vs absent, OR = 2.26, 95% CI: 1.63–3.14, P < .001), earlier distant metastasis (present vs absent, OR = 2.60, 95% CI: 1.45–4.67, P < .001), TNM stage (III-IV vs I-II, OR = 2.01, 95% CI: 1.35–2.99, P = .001), and portal vein invasion (present vs absent, OR = 4.39, 95% CI: 2.23–8.64, P < .001), but not associated with age, gender, tumor differentiation, and vascular invasion.

This meta-analysis also has some limitations, and the results should be interpreted with caution. First, data presented in the current meta-analysis were not applicable to all countries worldwide, because all the included studies were derived from China. Second, part of the HR value was calculated using a survival curve, which may lead to some error. Third, although all studies used qRT-PCR to evaluate the expression of miR-638, the cut-off value differed among studies, which might cause bias in the meta-analysis. Fourth, because of the relatively small sample size, we were unable to aggregate results based on a single type of tumor. Therefore, larger-scale, multicenter and high-quality studies are desperately necessary to confirm our findings.

Conclusions

In conclusion, our meta-analysis demonstrated that low expression of miR-638 was significantly correlated with poor OS and may serve as an effective predictive biomarker for tumor prognosis. Future larger scale prospective and standard investigations should be conducted to confirm our results.

Author contributions

Conceptualization: Fanliang Kong.

Data curation: Lixia Hu, Mengqin Huang.

Formal analysis: Lixia Hu, Mengqin Huang, Qianqian Yuan.

Methodology: Lixia Hu, Qianqian Yuan.

Project administration: Lixia Hu, Mengqin Huang, Fanliang Kong.

Software: Qianqian Yuan, Lixia Hu.

Writing – original draft: Lixia Hu, Mengqin Huang, Qianqian Yuan.

Writing – review & editing: Fanliang Kong.