Background
In recent years, cancer has become the primary cause of mortality in most countries and regions, and the incidence of human malignancies has increased substantially [
1]. Given the low survival rate of multiple cancer types, credible biomarkers for cancer prognosis are urgently required. Recently, the mammalian homologs of Lin-28, Lin28 (also called Lin28A) and Lin28B, have been considered as promising biomarkers.
Lin28A and Lin28B are highly conserved RNA-binding and microRNA-regulated proteins. In general, they selectively block the expression of let-7 microRNA family members, which act as tumor suppressors by inhibiting the expression of oncogenes and key regulators of mitogenic pathways, including RAS, MYC, and HMGA2 [
2]. Lin28A recruits TUTase to inhibit let-7 precursors to block Dicer processing in the cell cytoplasm, whereas Lin28B represses let-7 maturation through a TUTase-independent mechanism [
3]. Many studies indicated that both Lin28A and Lin28B show upregulated expression in human malignancies and that they function as oncogenes by promoting transformation and tumor progression [
4,
5].
Overexpression of Lin28A and Lin28B is associated with poor prognosis in various cancers, such as oral squamous cell carcinoma (OSCC) [
6,
7], colon cancer [
8,
9], epithelial ovarian carcinoma (EOC) [
4,
10,
11], gastric cancer [
12‐
14], hepatocellular carcinoma (HCC) [
15‐
17], breast cancer [
18,
19], esophagus cancer [
20], and other malignancies [
21‐
26]. On basis of these extensive literatures, Lin28A and Lin28B were regarded as promising prognostic factors for multiple cancers. However, the prognostic significance of Lin28A and/or Lin28B varies among different studies. To better confirm its prognostic significance, the present meta-analysis was conducted to evaluate its functional role in predicting cancerous survival of human malignancies.
Discussion
Lin28A and its homolog Lin28B belong to highly conserved RNA-binding proteins family, which are found to involve in numerous biological processes, including cell development, pluripotency, reprogramming, and oncogenesis [
31,
32]. Several recent investigations have confirmed that Lin28A and Lin28B can regulate gene expression either by directly binding to messenger RNAs (mRNAs) or by blocking microRNA biogenesis, and the underlying mechanisms include Let-7-family-dependent and Let-7-family-independent modes of action [
3].
Let-7 and lin28 (Lin28A and Lin28B) were first identified through mutagenesis screening as heterochronic genes in
Caenorhabditis elegans [
33,
34], and the expression and regulation of lin28 and let-7 are highly conserved throughout evolution [
35]. Recent studies have established the lin28/let-7 pathway as a central regulator of mammalian glucose metabolism [
36]. Although the exact roles of the let-7 family in adult mammalian tissues have not been definitely characterized, let-7 is known to serve as a tumor suppressor. Considerable evidence demonstrated that let-7 expression level is downregulated in various cancers and that let-7 overexpression restrains the growth and metastatic potential of cancer cells [
37‐
40]. As one targeting gene of let-7, lin28 expression is upregulated in various tumors, such as OSCC [
6,
7], colon cancer [
8,
9], EOC [
4,
10,
11], gastric cancer [
12‐
14], HCC [
15‐
17], and breast cancer [
18,
19]. Although many major effects of lin28 are mediated by blocking let-7 microRNA biogenesis, lin28 can also directly bind to GGAGA (G, guanosine; A, adenosine) sequences enriched with loop structures in mRNA targets, and this activity is similar to its interaction with let-7 microRNA precursors [
41]. Many of these targets function as oncogenes or tumor growth indicators, such as insulin-like growth factor 2 (IGF-2), a crucial growth and differentiation factor for muscle tissues [
42], and IGF-2 mRNA-binding protein 2, which binds several mRNAs encoding mitochondrial respiratory chain complex subunits [
43]. Therefore, Lin28A and Lin28B, along with their target genes, deserve further analysis in the future.
In this meta-analysis, we investigated the prognostic values of Lin28A and Lin28B for multiple human malignancies. Combining the outcomes of studies regarding the association of lin28 expression and tumor prognosis, we have successfully drawn many valuable results. First, increased expression level of Lin28A was considered to predict poor OS and RFS/DFS/PFS for cancer patients, with combined HR values of 1.60 (95% CI 1.38, 1.86) and 1.62 (95% CI 1.33, 1.97), respectively. Second, considering the pooled outcomes of studies on the relation between Lin28B expression and cancer prognosis, we found that elevated Lin28B level was significantly associated with poor OS and RFS/DFS/PFS of malignant diseases, with pooled HR values of 1.72 (95% CI 1.43, 2.08) and 2.35 (95% CI 1.61, 3.43), correspondingly, which also exerted statistical significance. Our findings suggested that Lin28A and Lin28B are promising biomarkers, and the detection of Lin28A and Lin28B expression in cancer patients is of potential value for monitoring patients’ survival.
However, significant heterogeneity was observed in the initial meta-analysis (Additional file
3: Figure S1 and Additional file
6: Figure S4). Therefore, influence analysis and Galbraith plot were applied for individual studies to investigate the source of heterogeneity. After excluding several studies of relatively low quality, no heterogeneity was found among the studies on cancer prognosis and lin28 expression, except for those concerning the association between Lin28B expression and RFS/DFS/PFS. Furthermore, subgroup analyses were performed to minimize the influence of heterogeneity (
I2 = 69.2%,
P = 0.006). In subgroup analyses, no heterogeneity was observed in the subtotals of Chinese Taiwan (
I2 = 0.0%,
P = 0.667) and EOC (
I2 = 0.0%,
P = 0.806). Also, no heterogeneity was shown in the subgroup analyses by different assay methods for Lin28A overexpression and OS, and RFS/DFS/PFS (Additional file
11: Figure S9A, B), and for Lin28B overexpression and OS (Additional file
11: Figure S9C). However, the subgroup of IHC still indicated of high heterogeneity for Lin28B overexpression and RFS/DFS/PFS (
I2 = 75.8%,
P = 0.016) (Additional file
11: Figure S9D). Besides, the subgroups of Asia (
I2 = 64.9%,
P = 0.036) and North America (
I2 = 70.6%,
P = 0.065) showed significant heterogeneity (Additional file
10: Figure S8). Therefore, we consider that the source of heterogeneity might result from the influence of different populations and disease types of the patients, rather than different methods.
Despite the meta-analysis was performed with rigorous statistics, our conclusion still has several limitations for the following reasons. First, the amounts of included studies in the meta-analysis was not sufficiently enough for more powerful results, as well as the study numbers for each cancer type. Second, all eligible studies were retrospective for analysis, which might impair the credibility of meta-analysis. Third, the population diversity, disease type, source of tissue sample and antibody for IHC may cause heterogeneity to a certain extent. Moreover, many researchers used a median immunohistochemical score as a division value, but the median scores slightly differed among studies. Several studies also performed a ternary method to classify the expression levels of Lin28A into high, medium, and low categories, which might result in an undervalued HR for Lin28A [
19]. All the disadvantages above might cause heterogeneity in the meta-analysis and produce deviation when evaluating the prognostic significances of Lin28A and Lin28B in human malignancies. Also, HR values of several studies were calculated using data extracted from the survival curves, which might unavoidably cause slight statistical errors.