HOTAIR as a diagnostic and prognostic biomarker of gastrointestinal cancers: an updated meta-analysis and bioinformatics analysis based on TCGA data

Abstract Gastrointestinal cancers are the most common type of cancer affecting humans. High expression of HOX transcript antisense intergenic RNA (HOTAIR), a long noncoding RNA (lncRNA), in various types of different tumors may be associated with poor prognosis. In the present study, we performed a meta-analysis of the relationship between HOTAIR expression and gastrointestinal cancers. Five databases were comprehensively searched for all literature until January 2023. Moreover, the target genes of HOTAIR were predicted by coexpression analysis based on The Cancer Genome Atlas (TCGA) gene expression matrix for six gastrointestinal cancer types. Finally, the mechanism through which HOTAIR affects tumors of the digestive system was systematically reviewed. Our results showed that the high HOTAIR expression group had worse outcomes with a pooled hazard ratio (HR) of 1.56 (95% confidence interval [CI] = 1.38–1.75, P<0.001). Furthermore, HOTAIR was identified as an unfavorable prognostic factor for overall survival (OS) in the esophageal carcinoma (ESCA) and gastric cancer (GC), as the HR were 1.94 and 1.58, respectively. The high correlation between the expression of homeobox C (HOXC) family genes and HOTAIR, with correlation coefficients of 0.863 (HOXC11), 0.664 (HOXC10), 0.645 (HOXC8), and 0.581 (HOXC12). The ‘cell cycle’ pathway and pathways relating to infections, namely ‘herpes simplex virus 1 infection’ and ‘complement and coagulation cascades’ were significantly enriched in Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. Also, we perform a systematic review to summarize the related oncogenic mechanism of HOTAIR. In conclusion, the HOTAIR has been identified as a potential prognostic factor in patients with gastrointestinal cancers.


Introduction
Globally, gastrointestinal cancers are a serious form of cancer and a threat to human health. For example, gastric cancer (GC) is the fifth most common cancer and the third leading cause of cancer-related deaths, pancreatic cancer is the seventh most common cause of cancer death, and liver cancer is the fourth most common cause of cancer-related deaths [1][2][3]. With the development of modern medicine, the diagnostic techniques and treatment methods for gastrointestinal tumors have improved dramatically, the 5-year survival rate remains low, and several types of gastrointestinal cancers still lack effective screening methods, such as pancreatic cancer [4]. Many studies have developed different models to find effective tumor biomarkers for early diagnosis [5,6], but the predictive ability and function of those biomarkers need further study and verification.

Inclusion and exclusion criteria
Eligible studies had to meet the following criteria: i) diagnosis of digestive system cancer must be confirmed: pathological examination and diagnosis were performed by an independent senior oncologist according to official pathological diagnostic criteria; ii) the expression level of HOTAIR was examined in control and tumor tissues; iii) the description of expression measurement of HOTAIR must be clear; iv) the cutoff value for HOTAIR must be described; v) the relationship between the level of HOTAIR and the patient survival time was analyzed; and vi) the hazard ratios (HRs) and 95% confidence intervals (CIs) for the survival rate were reported or could be calculated.
The exclusion criteria were as follows: i) studies without usable data; ii) animal studies, case reports, letters, reviews, and conference abstracts; iii) meta-analysis studies; iv) studies that included patients with benign tumors; v) studies that included patients after drug treatment; and vi) studies that did not contain or could not calculate HRs, 95% CIs, P-values, and survival data.

Data extraction and quality assessment
Two individuals of the authors extracted data from each study and independently estimated. The extracted data included the first author, year of publication, country, cancer type, sample size, tumor-node-metastasis stage (TNM), follow-up months, adjuvant therapy before surgery (yes or no), cutoff value, method of HOTAIR expression-level measurement, survival rates [OS and disease-free survival (DFS)], HRs, and 95% CIs. In cases where the HRs and 95% CIs were not directly reported, Engauge Digitizer software version 4.1 (http://digitizer.sourceforge.net/) was used to plot the Kaplan-Meier curves and extract the multiple survival rates to estimate the HRs and 95% CIs [46]. Quality assessment was performed using the Newcastle-Ottawa quality assessment scale (NOS). NOS criteria scores range from 0 (lowest) to 9 (highest), and a NOS score ≥6 is considered a high-quality study [47].

Bioinformatics analysis
Based on the gene expression matrix of six gastrointestinal tumors [liver hepatocellular carcinoma (LIHC), stomach adenocarcinoma (STAD), esophageal carcinoma (ESCA), pancreatic adenocarcinoma (PAAD), COAD, and cholangiocarcinoma (CHOL)] download from The Cancer Genome Atlas (TCGA, as mentioned in our previous study [48,49]. The mRNA expression profiles and the corresponding clinical data of 312 GI cancer patients of Asian ancestry were obtained. Raw mRNA expression data were normalized by [log2 (data+1)] for further statistical analysis and the 'edgeR' package was used to identify differentially expressed lncRNAs and protein-coding genes (PCGs) [50]. Then, the lncRNA-mRNA coexpression network was constructed to predict the potential biological functions of lncRNA [51]. We examined the correlation between the expression level of lncRNA HOTAIR and each PCG using two-sided Pearson correlation coefficients and the z-test. The PCGs positively or negatively correlated with the lncRNA HO-TAIR were considered lncRNA HOTAIR-related PCGs (|Pearson correlation coefficient| > 0.5 and P<0.001). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was used for functional enrichment analysis, which was conducted by the 'stringi' package, 'biocmanager' package, 'pathview' package, and 'clusterprofiler' package. All those pathways were mapped based on the KEGG database and searched for significantly enriched KEGG pathways at P<0.05 level. R software (R version 4.2.1) was used for statistical analysis and to plot the data.

Statistical analysis
HRs and 95% CIs were extracted to investigate the association between the expression level of HOTAIR and the survival rates of patients with malignant digestive system tumors. An HR value not equal to 1 was considered statistically significant, while an HR value greater than 1 indicated a worse survival outcome in the high HOTAIR expression group. The random and fixed effects models were compared and used to estimate the pooled HR. The Q-test was performed, and the I 2 statistic was determined to quantify the heterogeneity across the studies. An I 2 statistic of more than 50% and a Q-test result with P<0.05 indicated strong heterogeneity, and studies were grouped to reduce heterogeneity further. In addition, a sensitivity analysis was performed. Then, we investigated publication bias using funnel plots, Egger's test, and Begg's test [52]. All meta-analyses were performed using Stata 15.0 software (Stata Corp. College Station, TX, U.S.A.). P<0.05 was considered to indicate that the results were statistically significant.
In studies with DFS data, the high HOTAIR group was related to worse outcomes with HR = 2.62 (95% CI = 1.27-3.96, P<0.001) and no significant heterogeneity was identified (I 2 = 0%, P=0.825) ( Figure 5). Subsequently, a sensitivity and publication bias analysis was performed, and the results showed that the analysis was robust and reliable. Subgroup analysis was not performed owing to the limited number of studies.

Independent prognostic value of HOTAIR for OS in different cancer types
To determine the relationship between the HOTAIR expression level and OS in patients with gastrointestinal system cancer, we classified studies according to the cancer type and analyzed the HR and 95% CI by using the fixed-effects model. The results showed that HOTAIR was a significant prognostic indicator of the OS of patients with ESCA (HR

Construction of lncRNA-mRNA coexpression network to predict potential biological functions of HOTAIR
Using TCGA data, we determined the correlation between the level of expression of lncRNA HOTAIR and each PCG (Figure 7, Supplementary Table S3). Coexpression networks can predict lncRNA function since genes involved in the same biological process are often coexpressed. The results showed that the six PCGs most positively correlated with HOTAIR expression were HOXC11, HOXC10, HOXC8, C2orf70, HELB, and HOXC12, with the correlation coefficient are 0.863, 0.664, 0.645, 0.59, 0.585, and 0.581 (P<0.001) (Figure 7). Among them, HOXC11, HOXC10, HOXC8, and HOXC12 interact with HOTAIR in gastrointestinal tumors to jointly promote tumor formation and development [44,83,84]. Next, we investigated the pathways through which HOTAIR significantly accumulates in tumors by mapping genes to KEGG pathways (Figure 8, Supplementary Table S4). HOTAIR is most enriched in the 'cell cycle' pathway and pathways relating to infections, namely 'herpes simplex virus 1 infection' and 'complement and coagulation cascades', which are associated with key functions of HOTAIR in cancer. The results showed that  HOTAIR targets the JAK/STAT signaling pathway, a mitochondria-dependent signaling pathway involving Bcl-2 and Bax, to interfere with tumor growth and apoptosis. Meanwhile, HOTAIR affects coagulation and complement factors, thereby promoting cell migration and proliferation. Additionally, HOTAIR functions in the TGF-β/TAK1/MAPK and

Systematic literature review of the molecular mechanism
The lncRNA HOTAIR plays an important role in the formation and progression of gastrointestinal tumors. Most of the reported effects of HOTAIR on tumors in vitro and in vivo are related to cell apoptosis, EMT, cell cycle, drug resistance, metastasis, cell proliferation, invasion, and migration ( Figure 9). First, studies on gastrointestinal tumors suggest that HOTAIR is vital for cancer cell survival and deficiency of HOTAIR leads to cancer cell apoptosis [69,85]. Meanwhile, HOTAIR can inhibit tumor cell apoptosis via the Bax/Bcl-2 pathway by promoting the expression of related protein Bcl-2 and inhibiting the expression of proto-oncogene Bax [86]. These results are similar to the results of the bioinformatics analysis based on TCGA data in the present study. Second, HOTAIR down-regulates the expression of E-cadherin and miR-217, then activates the DACH1/JAK3/STAT3 and Wnt/β-catenin pathways, which are involved in the tumor's EMT ability, which is important for primary tumor formation and metastasis [45,63,87]. Third, HOTAIR knockdown can prevent GC cell proliferation, influence cell cycle distribution, and increase P21 and P53 protein levels [70]. HOTAIR also affects the cell cycle by increasing the expression of the P53 tumor-suppressor gene and decreasing the expression of the PTEN tumor-suppressor gene, thereby activating the AKT/P21 signaling pathway and hyperactivating cell division [42]. The role of HOTAIR in influencing the cell cycle was also examined by our bioinformatics analysis. HOTAIR can up-regulate P53, CYCA, and CYCD by increasing the expression of P300. Fourth, many studies have reported a relationship between tumor resistance and HOTAIR. HOTAIR down-regulates Mir-130a-3p and reduces its binding to its target gene ATG2B, thereby affecting tumor chemosensitivity [88]. Studies have also shown that the methylation level of the methylenetetrahydrofolate reductase (MTHFR) gene is regulated by HOTAIR, which reduces the chemosensitivity of cells to 5-fluorouracil [89]. Fifth, HOTAIR also affects tumor metastasis by affecting the Wnt/β-catenin, TGF-β, VAGE, and PI3K/AKT/MAPK pathways; the role and importance of these signaling pathways in the metastatic potential of various cancers have been demonstrated [90,91]. Sixth, HOTAIR promotes cell proliferation by decreasing the expression of miR-454-3p and activating STAT3 and its downstream target gene CYCD [92]. Seventh, HOTAIR participates in regulating the expression of many invasion-related genes, including MMP1 and MMP3 [69]. Eighth, HOTAIR affects intracellular formation and transport of tumor-related exosomes via up-regulation of ras-related protein Rab-35 (RAB35), promotes the phosphorylation of synaptosome-associated protein 23 (SNAP23), and induces the translocation of vesicle-associated membrane protein 3 (VAMP3) and SNAP23 to the cytomembrane, and the formation of transmembrane protein soluble N-ethylmaleimide-sensitive fusion factor attachment protein receptor (SNARE), which plays an important role in enhancing the transport of multivesicular bodies (MVBs) related to the formation of exosomes [93]. Overall, HOTAIR participates in gastrointestinal tumor progression through the above eight aspects and may be a potential prognostic factor for digestive system malignancies.

Discussion
Owing to their potential roles in cancer biology, lncRNAs represent a promising new class for the diagnosis and prognosis of various cancers and have the potential for use as predictors, therapeutic targets, and cancer biomarkers [94]. The different sensitivities of human tumors may be related to different lncRNA expression levels or lncRNA polymorphisms [95]. For example, PCA3 (prostate cancer antigen 3, lncRNA) is considered a biomarker in prostate cancer [96], and the high expression of lncRNAs CCAT2, MALAT1, and NEAT1 is related to worse OS in breast cancer [97]. Moreover, HOTAIR affects the occurrence and development of cancer in many types of malignancies, including digestive system malignant tumors [91]. Though many studies have reported that HOTAIR plays an important role in digestive system tumors [34], the literature lacks an updated systematic analysis and summary. In the present article, we collected studies on HOTAIR in almost all kinds of digestive system malignant tumors and used meta-analysis to clarify the prognostic value of HOTAIR. Furthermore, we used two-sided Pearson correlation coefficients to identify lncRNA HOTAIR-related PCGs. Finally, we predicted and summarized the main related molecular mechanism of the function of HOTAIR in cancer biology by bioinformatics analysis and systematic review.
A meta-analysis of the prognostic value of HOTAIR was reported previously, but the present study only addressed esophageal, gastric, colorectal, and hepatocellular carcinomas [34]. Our review covers the recently published literature on pancreatic cancer and CHOL. The results showed that high HOTAIR expression levels were associated with poorer OS with an HR of 1.56 (95% CI = 1.38-1.75, P<0.001). The heterogeneity Q statistic (P>0.05) and I 2 value (0.00%) indicate that there is no statistical evidence for heterogeneity among the 32 studies, and the subgroup analysis revealed the relationship between HOTAIR expression and OS, while country, sample size, and analysis methods did not show significant differences. We also performed sensitivity and publication bias analyses. The results of Begg's and Egger's tests indicated publication bias between these studies, so the trim and filling method was used to determine the stability of the analysis results. Finally, we classified tumor types and found that high HOTAIR expression was associated with worse OS in STAD and ESCA. HOTAIR may play an important role in these two gastrointestinal tumors. Indeed, a previous meta-analysis demonstrated that patients with up-regulated levels of HOTAIR showed a Figure 9. Association between HOTAIR and the development of digestive system malignancies via its regulation ability of cell apoptosis, EMT, cell cycle, drug resistance, metastasis, cell proliferation, invasion, and migration Abbreviations: AKT, serine/threonine kinase 1; ATG2B, autophagy-related 2B; ATG7, autophagy-related 7; Bcl-2, B-cell chronic lymphocytic leukemia/lymphoma-2; Bax, Bcl-2-associated X protein; Cyclin D, D-type cyclins; DACH1, dachshund family transcription factor 1; GPC5, glypican 5; HOXD, homeobox D cluster; JAK3, janus kinase 3; MMP1, matrix metalloproteinase-1; MMP3, matrix metalloproteinase-3; MTHFR, methylenetetrahydrofolate reductase; MVBs, multivesicular bodies; PTEN, phosphatase and tensin homolog; RAB35, member RAS oncogene family; SNAP23, synaptosome associated protein 23; SNARE, soluble N-ethylmaleimide-sensitive fusion factor attachment protein receptor; STAT3, signal transducer and activator of transcription 3; TGF-β, transforming growth factor-β; VAMP3, vesicle-associated membrane protein 3; WNT, wingless-type MMTV integration site family.
poor OS rate as compared with patients with lower levels of HOTAIR, suggesting the predictive value of HOTAIR for CRC prognosis [36]. In the present study, the result showed weak significance in CRC as HR is 1.46 (95% CI = 0.99-1.93). Meanwhile, another meta-analysis of HOTAIR in LIHC provides a strict and classic protocol for analysis, although it did not provide specific and detailed analysis data [35]. Due to few studies about HOTAIR in LIHC, our result is also inadequate. Therefore, through our meta-analysis results, we can see that HOTAIR is an important risk factor in well-studied GC, CRC, and ESCA, and can be used as a potential prognostic biomarker. However, there is not enough evidence in tumors such as LIHC, CHOL, and PAAD, and further research is needed.
However, we also noticed that HOTAIR plays an important role in other tumors, not just gastrointestinal tumors [98]. High expression levels of HOTAIR in many cancers, such as lung cancer [25], cervical cancer [26], breast cancer [27], gliomas [28], prostate cancer [99], ovarian cancer [100], and oral cancer [101], have been reported in many studies. In lung cancer, HOTAIR is overexpressed and correlated with tumor metastasis and poor prognosis, which promotes proliferation, survival, invasion, metastasis, and drug resistance in lung cancer cells [25]. In cervical cancer, HOTAIR plays an oncogenic role by promoting cell proliferation, migration, invasion, and autophagy, inhibiting cell apoptosis, stimulating angiogenesis, accelerating cell cycle progression, and inducing EMT [26]. In breast cancer, HOTAIR expression is augmented in primary breast tumors and metastases, and the HOTAIR expression level in primary tumors is a powerful predictor of metastases and death [24]. Therefore, HOTAIR may be a potential therapeutic target in breast cancer [102]. In gliomas, HOTAIR mainly promotes cell proliferation and migration, and inhibits apoptosis. In vitro and in vivo studies have shown that HOTAIR regulates cell cycle-related genes and related signaling pathways, such as the Wnt/β-catenin axis. Additionally, it can promote angiogenesis and affect the permeability of the blood-brain barrier, thereby modulating the effectiveness of chemotherapeutic drugs [28]. In prostate cancer, a study demonstrated that HOTAIR promotes the invasion and metastasis of PCa by decreasing the inhibitory effect of hepaCAM on MAPK signaling [99]. In ovarian cancer, inhibiting HOTAIR expression in ovarian cancer cells prevents tumorigenesis and metastasis [103], and HOTAIR up-regulates c-Myc in breast and ovarian cancers, thereby promoting cancer cell proliferation [8]. Additionally, HOTAIR is associated with drug susceptibilities, such as platinum resistance [104] and carboplatin resistance [105]. In oral cancer, HOTAIR mediates the suppression of cell proliferation and promotion of apoptosis [106] and promotes the invasion and metastasis of oral squamous cell carcinoma through metastasis-associated gene 2 (MTA2) [107]. Meanwhile, HOTAIR has been extensively studied as an oncogene, and functional SNPs in HOTAIR are associated with cancer risk, including lung, gastric, esophageal, cervical, breast, and prostate cancers [108][109][110][111]. Many studies reported that HOTAIR polymorphisms could affect cancer biology. The HOTAIR rs920778 polymorphism is associated with ovarian cancer susceptibility and poor prognosis in the Chinese population [112]. The HOTAIR rs7958904 polymorphism is associated with CRC morbidity and mortality and is a potential CRC biomarker [113]. In conclusion, there is increasing evidence that HOTAIR acts as an oncogene in various cancers and that its up-regulation can lead to malignant transformation and tumorigenesis [67].
Through systematic review and bioinformatics analysis, we found that HOTAIR was involved in the regulation of cancer cell apoptosis, cell proliferation, and the induction of cell cycle arrest, EMT, migration, invasion, metastasis, and resistance in gastrointestinal malignancies. Down-regulation of HOTAIR can induce apoptosis of GC cells and significantly inhibit the proliferation, invasion, and metastasis of GC cells [70], the same findings have been found in other gastrointestinal tumors, although the signaling pathways that may be involved are inconsistent. Meanwhile, we integrated the gene matrix of six major types of Asian patients with gastrointestinal tumors in the TCGA database and found HOTAIR was closely related to the following pathways through the lncRNA-mRNA coexpression network. The 'cell cycle' pathway and pathways relating to infections, namely 'herpes simplex virus-1 infection' and 'complement and coagulation cascades' were significantly enriched in KEGG analysis. Several studies have shown that HOTAIR inhibition leads to G0/G1 cell cycle arrest, thereby inhibiting tumor cell proliferation. Since most GCs are associated with multiple pathogenic infections, several oncogenic viruses play important roles in the malignant progression of GC, and oncolytic viruses appear to be a new therapeutic agent class that inducing antitumor immune responses by selectively killing tumor cells and inducing systemic antitumor immunity [114]. Herpes simplex virus-1 (HSV-1) is a double-stranded DNA virus belonging to the alpha-herpesviruses subfamily, several investigators have used the attenuated HSV vector betaH1 in transplanted human lung, breast, gastric, and colon tumors. Tumor regression and similar apoptosis were observed in all tumors [115]. Finally, coagulation-related genes play an important role in gastrointestinal tumors, and coagulation-related gene models provide new insights and targets for the diagnosis, prognosis prediction, and therapeutic management of patients with GC [116], similar results were also found in CRC [117] and ESCA [118]. In conclusion, our study suggests that HOTAIR has an important association with these pathways and needs to be further elucidated.

Limitations
Several limitations are unavoidable when performing a meta-analysis. First, publication bias exists due to the more frequent publication of positive results. Second, the results of some small studies are unreliable. Third, different analysis methods were used in the selected studies, and some did not use multivariate analysis. Fourth, the treatment after surgery varied across the selected studies, which might affect the survival time of patients. The relationship between HOTAIR and the survival rate of patients still needs confirmation. Fifth, many studies were performed in Asian populations, especially in China populations, while fewer studies were performed in Europe and America.

Conclusion
In conclusion, we demonstrated the predictive power of HOTAIR lncRNAs in gastrointestinal malignancies and showed that high HOTAIR expression is associated with worse OS. Then, the molecular mechanisms were systematically evaluated. HOTAIR plays an important role in cancer biology and deserves further study.

Data Availability
The datasets analyzed in the present study are available from the published papers that have been cited in the present manuscript.