A systematical search of databases of PubMed (Medline), Embase, ISI Web of Knowledge, Springer, the Cochrane Library, Scopus, BioMed Central, ScienceDirect, Wanfang, Weipu, and China National Knowledge Internet was conducted from inception to November 1, 2022. Appropriate publications was searched in accordance with the detailed Mesh, including (“lnc SNHG” OR “long non-coding RNA SNHG” OR “Small nucleolar RNA host genes”) AND (“prognosis” OR “prognostic” OR “survival” OR “outcome”) based on PRISMA (The Preferred Reporting Items for Systematic Reviews and Meta-Analyses).

Studies were selected by two independent researchers. Inclusion criteria were as follows: (1) investigation of the association between SNHG expression and survival outcome, as well as clinical prognosis of CRC patients; (2) classification of patients into high and low expression groups in accordance with primary literature; (3) detection of SNHG expression level using validated techniques; (4) sufficient data to calculate odds ratio (OR) or hazard ratio (HR); and (5) studies written in English. Exclusion criteria were: (1) no investigation on the relationship between SNHG expression level and CRC prognosis different from a mere exploration of the involved molecular biological mechanisms; (2) reviews and meta-analyses, letters, animal studies, and conference proceedings; (3) insufficient data for prognostic analysis; and (4) duplicate studies.

For inclusion in this meta-analysis, two researches independently scored the selected studies using the Newcastle-Ottawa quality assessment scale (NOS) to assess the quality and suitability of the selected studies (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp) (16). The NOS items for selection of cohorts, comparability among included studies, and outcome were included. Study selection flow diagram is shown in Figure 1 .

For all included literatures, two researchers independently extracted the following index in detail: name of first author, article release year, country where the patient belongs, number of cases, expression level of SNHG, cutoff value, presentation of HR value, follow-up month, HR with 95% CI for overall survival (OS) and reference-free survival (RFS); OR with 95% CI for dichotomous data included TNM staging, distant lymph node metastasis (LNM), organ metastasis, pathological score, tumor size etc. If the literature provides both multivariate and univariate analyses methods, then we extract the values of multivariate analysis. In situations where an article simply presents survival curves without providing HR and its 95% CI, the HR was extracted by extracting survival curves using the Engauge Digitizer v10.4 software (17). In the event where the extracted data of the two investigators were inconsistent, a third investigator was asked to evaluate the data to decide whether to extract the data or not. Additionally, we seek out The Cancer Genome Atlas (TCGA) database of Interactive Analysis of Gene Expression Profile, and explored the GEPIA2 website to further explore SNHG expression in colorectal tumor tissues and adjacent tissues.

Review Manager V 5.4 Software and Stata SE14.0 software were utilized to implement the statistical analysis in this study. patientss included in the study were spanided into high-expression and low-expression groups on the ground of reports in the original literature. Pooling OR besides 95% CI to investigate the correlation both lncSNHGs expression and the clinical prognostic parameters of CRC patientss, such as TNM stage, LNM, DM, histological grade, tumor size, depth of invasion, etc. Pooling HR with 95% CI was implemented to explore the pertinence of lncSNHGs expression levels with OS progression-free survival (PFS) and DFS in cancer patients. The heterogeneity amid incorporated studies was appraised by I² and P value. If I² was >50% and P<0.05, we assumed there exist markedly strengthened heterogeneity in this outcome and a random-effects model was performed. At the same time, we tried to found the provenience of heterogeneity through classification analysis in view of follow-up visits, cut-off index, HR calculation procedures, and sample size. Sensitivity analyses were performed to find individual studies with large heterogeneity in OS outcomes. Clinical and statistical bias was detected using Begg's test.

Basic characteristics of enrolled documents according to our search using the keywords, 263 articles were screened in the initial trial, excluding 74 articles that were repeatedly published. One-hundred and thirty-seven papers did not assess the interrelation between lncRNA SNHGs expression and CRC prognosis, 14 papers were based on animal experiments, 6 papers did not classify lncRNA SNHGs into high and low expression patients, and 7 papers had insufficient data. This study finally included 25 suitable papers, including 2,342 CRC cases published between 2016 and 2022, with the number of patients ranging between 30–338 (14, 18–31). All the patients were from China, and were diagnosed as CRC by pathology or histology. High and low lncRNA SNHG expression was verified using the real-time quantitative reverse transcription reaction. Twenty-two papers included survival data of patients, and three papers only provided clinical pathological characteristics of patients ( Table 1 ). The NOS score among covered investigations ranged between 6–9 based on overall evaluation of literature quality in detail ( Table 2 ).

Twenty-two researches comprising 2,134 patientss were obtained to revealed the relativity between SNHG expression and the survival outcome of CRC patients. Twelve papers directly reported HR values and 95% CI; 10 papers simply provided survival curves and were extracted via the Engauge Software (indirect extraction). Pooled HRs demonstrated that high SNHG expression predicts poor cancer prognosis (HR = 1.635; 95% CI, 1.405–1.864; P<0.0001) ( Figure 2 ). We took into consideration different cut-off index (mean or median), follow-up month, and HR estimation method (Directly and Indirectly). The classification analysis was carried out to reveal the marked heterogeneity between different groups based on HR estimation method (Directly and Indirectly), follow-up month (<60 months and not less than 60 months), number of patients (>100 and <100), cut-off value (mean and median), NOS score (9 and <9), and analysis method (univariate and multivariate analyses). A positive association was expose between increasing SNHG expression and short OS in the multivariate analysis (HR = 1.527; 95% CI, 1.276–1.779), univariate analysis (HR = 2.186; 95% CI, 1.616–2.756), HRs withdraw straight from studies (HR = 1.527; 95% CI, 1.276–1.777), or explicitly abstracted from the survival curve (HR = 2.224; 95% CI, 1.639–2.808), cancer with >100 (HR = 2.300; 95% CI, 1.729–2.871) and <100 (HR = 1.505; 95% CI, 1.254–1.757), not less than 60 months of follow-up (HR = 1.907; 95% CI, 1.546–2.267) and <60 months (HR = 1.448; 95% CI, 1.149–1.746) ( Table 3 ). Three studies comprising 301 patients, 2 studies comprising 173 patientss, and 1 study with 141 patientss were obtained to evaluate the connection between SNHG expression and cancer survival outcome, including PFS, DFS, and recurrence free survival (RFS). Pooled HRs showed that increasing SNHG expression reminds worse PFS (HR = 2.35; 95% CI, 1.46–3.78) ( Figure 3A ), DFS (HR = 1.85; 95% CI, 1.27–2.71) ( Figure 3B ), and RFS (HR = 2.70; 95% CI, 1.17–6.23) ( Figure 3C ).

Twenty publications with 1,873 patientss were obtained to inquire the dependency between lncRNA SNHGs expression and clinical stage of CRC patientss. The publications showed that increasing SNHG expression predicts an poor TNM stage (OR = 1.750; 95% CI, 1.481–2.067; P=0.0001) ( Figure 4 ). A classification analysis redouble revealed that high SNHG expression means a terminal TNM stage in a group that used mean value as the cut-off value (OR = 1.656; 95% CI, 1.315–2.085; P=0.0001) and median value as the cut-off value (OR = 1.859; 95% CI, 1.460–2.368; P=0.0001). At the same time, enhanced SNHG expression indicates advanced TNM stage in both high NOS score (OR = 1.841; 95% CI, 1.490–2.274; P=0.0001) and low (OR = 1.609; 95% CI, 1.226–2.110; P=0.0001) NOS score groups. A fixed effects model was exerted to derive from a small heterogeneity level (I² = 19.2, P=0.216) ( Table 4 ).

Seventeen publications with 1,389 patientss were incorporated to inspect the pertinence between the lncRNA SNHGs expression and LNM of CRC patientss. These studies showed that high SNHG expression was noteworthy relativity to lymph node invasion (OR = 1.645; 95% CI, 1.338–2.021; P=0.0001) ( Figure 5 ). In a subgroup analysis, we detected that elevated lncRNA SNHGs expression demonstrated significant association with lymph node metastasis in a group that used mean value as the cut-off index (OR = 1.507; 95% CI, 1.184–1.919; P=0.0001) and median value as the cut-off index (OR = 2.073; 95% CI, 1.393–3.086; P=0.0001). A remarkable relevance was observed in both high (OR = 1.793; 95% CI, 1.347–2.388; P=0.0001) and low (OR = 1.496; 95% CI, 1.111–2.014; P=0.0001) NOS score groups. A fixed effects model was adopted for small heterogeneity level (I² = 8.3, P=0.357) ( Table 4 ).

A total of 18 studies with 1,694 patients described the association between the correlation of the expression level of lncRNA SNHGs and distant metastasis in patients with CRC. A significantly positive correlation was revealed between increased lncRNA SNHGs expression and earlier distant metastases (OR = 1.601; 95% CI, 1.299–1.973; P=0.0001) ( Figure 6 ). A subgroup analysis further found a positive association between elevated SNHG expression and earlier distant metastases in the group that used mean value as the cut-off value (OR = 1.815; 95% CI, 1.363–2.418; P=0.0001) and median value as the cut-off value (OR = 1.381; 95% CI, 1.016–1.876; P=0.0001). A similar positive relationship was observed in groups with high (OR = 1.604; 95% CI, 1.227–2.096; P=0.0001) and low (OR = 1.596; 95% CI, 1.143–2.229; P=0.0001) NOS scores. A fixed effects model was adopted due to small heterogeneity level (I² = 46.6, P=0.016) ( Table 4 ).

Twelve publications comprising 1,158 patients were included to assess the relationship between the expression level of lncRNA SNHGs and tumor size in CRC patients. A positive association was observed between high lncRNA SNHGs expression and bigger tumor size (OR = 1.297; 95% CI, 1.061–1.584; P=0.011) ( Figure 7 ). A subgroup analysis was based on a cut-off value and NOS score, and significantly positive correlations were also confirmed in groups that used median value as the cut-off value (OR = 1.372; 95% CI, 1.046–1.800; P=0.0001) and those with high NOS score (OR = 1.450; 95% CI, 1.150–1.829; P=0.0001). Meanwhile, no significance was observed in groups that used mean value as the cut-off value (OR = 1.212; 95% CI, 0.901–1.630; P=0.0001) and those with low NOS score (OR = 0.927; 95% CI, 0621–1.383; P=0.0001). A fixed effects model was adopted due to small heterogeneity level (I² = 46.6, P=0.016) ( Table 4 ).

The relationship between lncRNA SNHGs expression and other clinicopathological indicators, including histological grade, age, and gender were also assessed. A significantly positive correlation was observed between high SNHG expression and depth of invasion (HR = 1.603; 95% CI, 1.218–2.110; P=0.001) and bad histological grade (HR = 1.416; 95% CI, 1.111–1.805; P=0.005). Meanwhile, no significance was observed between lncRNA SNHGs expression and age (OR = 1.055; 95% CI, 0.916–1.216; P=0.456) and gender (OR = 0.951; 95% CI, 0.817–1.107; P=0.61) ( Table 4 ).

Sensitivity analyses were performed to explore methodological heterogeneity in these studies, and to explore whether single study data significantly affected the overall outcome. We did not find a study data that affected the overall outcome significantly ( Figure 8 ). A potential publication bias was explored using the Begg’s test, and unobvious publication bias was found in the included studies ( Figure 9 ).

Based on TCGA-COAD and TCGA-READ datasets of TCGA database (https://portal.gdc.cancer.gov/repository?facetTab=cases) and the GEPIA website (http://gepia.cancer-pku.cn/), we analyzed the expression of SNHGs in tumor tissues and adjacent tissues. The results showed that the expression level of SNHGs in tumor tissues was higher than that in adjacent tissues. This result is consistent with the results of this meta-analysis. ( Figure 10 ).