Prognostic and clinicopathological role of RACK1 for cancer patients: a systematic review and meta-analysis

Protocol and registration

This systematic review with meta-analysis was conducted in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol (Table S1). The study protocol was drafted and registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42022351129).

Search strategy

Five databases (PubMed, Embase, Web of Science, Cochrane Library, and Scopus) were thoroughly explored from their inception to April 20, 2023, using the following keywords: “receptor of activated c kinase 1” and “neoplasms”. Potentially relevant literature was also obtained through manual searches of the reference lists of the included studies. Detailed search strategies are presented in Table S2.

Inclusion and exclusion criteria

Study selection and data extraction

Two authors (YQ Wu and SX Jiang) performed a preliminary screening of the literature based on titles and abstracts. Subsequently, two other authors (QH Wang and Y Zhang) evaluated the full-text. Any disagreements were discussed by the review group until a consensus was reached.

Following selection, two authors (QH Wang and Y Zhang) extracted the following data from the included articles: (1) first author; (2) year of publication; (3) country; (4) number of patients; (5) patients’ age; (6) detection methods; (7) treatment; (8) follow up; (9) cut-off value; (10) survival information; (11) TNM stage; (12) hazard ratios (HRs) and the 95% confidence intervals (CIs) of RACK1 for survival; (13) clinicopathological characteristics. When HR and 95% CI were not reported, the data were analyzed using survival-related Kaplan–Meier curves according to the method described by Tierney et al. (2007).

Quality assessment

Two authors (SX Jiang and M Huang) independently used the Newcastle-Ottawa Scale to assess the quality of the cohort studies, and any disagreements were resolved by a third author (Stang, 2010). Studies with a score ≥ 7 (score range, 0–9) were considered high quality.

Statistical analysis

HRs and 95% CIs were used to assess the patient prognosis. HR >1 indicated poor survival in the group with high RACK1 expression, whereas HR <1 indicated poor survival in the group with low RACK1 expression. If the results of both univariate and multivariate Cox regression analyses had been reported, multivariate models were preferred to pool the data. ORs with 95% CIs were used to determine the association between RACK1 expression and clinicopathological characteristics. The heterogeneity of the meta-analysis was tested using Cochran’s Q test and the Higgins I-squared statistic (I²). Significant heterogeneity was indicated by I² >50% or a P-value <0.1, and a random effects model was then applied; otherwise, a fixed effects model was preferred. Subgroup, meta-regression, and sensitivity analyses were performed in the presence of significant heterogeneity. Publication bias was evaluated using funnel plots, Begg’s test and Egger’s test. The level of statistical significance was set at a P-value of <0.05. Statistical analysis and visualization were performed using Revman 5.4 and Stata 12.

Process of study selection

From the five databases, 2,269 publications were identified, and seven relevant publications were manually searched. After removing 1,332 duplicate studies, 944 publications were eliminated based on their titles and abstracts. The full texts of the remaining 86 papers were reviewed further, and 22 studies were included in the meta-analysis. Of these, 13 studies evaluated the predictive value of RACK1 for cancer prognosis, and 20 studies assessed the correlation between RACK1 expression and clinicopathological features. The search and detailed selection processes are shown in the flowchart (Fig. 1).

General characteristics of the included studies

The included 22 studies were published between 2010 and 2022, with eleven studies (50%) published in the last five years (Han et al., 2018; Li et al., 2019; Liu et al., 2017; Liu et al., 2018; Qu et al., 2017; Shen et al., 2020; Wu et al., 2020; Xiao et al., 2018; Xu et al., 2022; Yu et al., 2021; Zhang et al., 2019). One study was from Japan (Nagashio et al., 2010) and the rest were from China (Cao et al., 2010; Chen et al., 2015; Han et al., 2018; Jin et al., 2014; Li et al., 2016; Li et al., 2019; Lin et al., 2014; Liu et al., 2017; Liu et al., 2018; Lv et al., 2016; Peng et al., 2016; Qu et al., 2017; Shen et al., 2020; Shi et al., 2012; Wang et al., 2015; Wu et al., 2020; Xiao et al., 2018; Xu et al., 2022; Yu et al., 2021; Zhang et al., 2019; Zhong et al., 2013). The thirteen studies used to assess the prognostic value included fifteen independent cohorts, with three studies for PC (Han et al., 2018; Li et al., 2016; Zhang et al., 2019), two studies each for GC (Chen et al., 2015; Yu et al., 2021) and lung cancer (LC) (Qu et al., 2017; Zhong et al., 2013), one study with three independent cohorts for OSCC (Liu et al., 2018), and one study each for BC (Cao et al., 2010), glioma (Lv et al., 2016), ESCC (Wang et al., 2015), cervical cancer (CC) (Wu et al., 2020), and colorectal cancer (CRC) (Xiao et al., 2018), totaling 2,620 patients. The 20 studies used to evaluate the association between clinicopathological features and RACK1 expression included four studies for LC (Nagashio et al., 2010; Qu et al., 2017; Shi et al., 2012; Zhong et al., 2013), three studies each for GC (Chen et al., 2015; Liu et al., 2017; Yu et al., 2021), PC (Han et al., 2018; Li et al., 2016; Zhang et al., 2019), and CRC (Jin et al., 2014; Li et al., 2019; Xiao et al., 2018), two studies for CC (Wu et al., 2020; Xu et al., 2022), and one study each for BC (Cao et al., 2010), ovarian cancer (OC)(Lin et al., 2014), ESCC (Wang et al., 2015), melanoma (Shen et al., 2020), and nasopharyngeal carcinoma (NPC) (Peng et al., 2016), totaling 3,043 patients. The sample sizes of the cohorts included in the meta-analysis ranged from 50 to 495 patients, with a mean of 147 patients. Two studies used reverse transcription-polymerase chain reaction (RT-PCR) to detect RACK1 expression, with immunohistochemistry (IHC) utilized in the remaining studies. The general characteristics of the included studies are shown in Table 1.

Quality assessment

The overall quality of the included studies was good, with none of them scoring <6 (Table S3). Of the 22 studies evaluated, 11, 10, and 1 were individually scored 8, 7, and 6, respectively. The primary sources of bias include insufficient follow-up time, unclear cohort determination processes, and inconsistent interventions applied to the same cohort.

Association between RACK1 expression and overall survival (OS)

Thirteen studies and 2,620 patients were included in the meta-analysis to assess the association between RACK1 expression and the prognosis of patients with cancer. The findings indicated that high RACK1 expression was significantly associated with poor OS despite the detection of significant heterogeneity (Fig. 2; HR = 1.62; 95% CI [1.13–2.33]; P = 0.009; I² = 89%). Subgroup analysis, meta-regression, and sensitivity analysis were performed to address the observed heterogeneity.

Association between RACK1 expression and disease-free survival/recurrence-free survival (DFS/RFS)

Two studies, including 192 patients, reported DFS or RFS data. Furthermore, the results demonstrated a significant correlation between high RACK1 expression and poor DFS/RFS, and no significant heterogeneity was indicated (Fig. 3; HR = 1.87; 95% CI [1.22–2.88]; P = 0.004; I² = 0%).

Association between RACK1 expression and clinicopathological characteristics

To evaluate the association between RACK1 expression and clinicopathological features, 20 cohorts containing 3,043 patients were included in this meta-analysis. The pooled results indicated that high RACK1 expression in patients with cancer was associated with lymphatic invasion/N+ stage (OR = 1.74; 95% CI [1.04–2.90]; P = 0.04; I² = 79%) of tumors. No relationship was observed between RACK1 expression and age, gender, T stage, M stage, TNM stage, tumor differentiation, tumor size, or nerve invasion, as presented in Table 2 and Fig. S1. Significant heterogeneity was observed in each group when the effect sizes were combined, except when the relationship between gender and RACK1 expression was explored.

Subgroup analysis and meta-regression

Subgroup analysis for RACK1 expression and OS was based on cancer type (digestive or non-digestive system cancers) and sample size (sample size ≥ 100 or <100), as well as the source of HRs(reported and estimated) and the Cox analysis method (multivariate and univariate). Except for the subgroups regarding the digestive system, estimated HR, and univariate Cox analysis, all other results of the subgroup analysis suggested that high RACK1 expression could predict poor OS (Table 3 and Fig. S2).

Considering that different cancer types may have different propensities for lymphatic invasion, we analyzed the association between RACK1 expression and lymphatic invasion in specific cancer types. The findings indicated high RACK1 expression to be significantly associated with lymphatic invasion in LC (OR = 2.36; 95% CI [1.56–3.56]; P < 0.001; I² = 44%), but not statistically associated with PC and CRC. For other cancer types, definite conclusions could not be drawn because only one study for each could be included in the analysis (Table 4 and Fig. S3).

To further explain the heterogeneity among the studies focusing on RACK1 expression and OS, meta-regression was performed, and a P value <0.05 was utilized as the assessment standard for judging whether a certain factor serves as the source of heterogeneity. The findings implied that the source of HR (P = 0.037) might be a source of heterogeneity, contributing 29.49% of the heterogeneity. The detailed results of the subgroup analysis and meta-regression are presented in Table 3.

Sensitivity analysis

Sensitivity was evaluated to verify the stability of our conclusions when significant heterogeneity occurred in assessing the correlation of RACK1 expression with OS and lymphatic invasion. The sequential exclusion of each study did not affect the conclusion regarding the OS, indicating that the results were relatively stable and reliable (Fig. 4A). However, inconsistent combined results were observed for lymphatic invasion after removing specific literature; some pooled ORs indicated that RACK1 expression was not associated with lymphatic invasion (Fig. 4B). As a result, the conclusions regarding lymphatic invasion might be unstable.

Publication bias

Visually inspecting the funnel plot, we found asymmetry in the OS analysis, and no asymmetry was found for DFS/RFS or lymphatic invasion (Fig. 5). Begg’s and Egger’s tests were performed to assess the funnel plot asymmetry. No publication bias was observed in the association between RACK1 expression and lymphatic invasion (Begg’s test: P = 0.692; Egger’s test: P = 0.258) of cancer. However, publication bias was found in the studies on the correlation between RACK1 expression and OS using Egger’s test (Begg’s test: P = 0.092; Egger’s test: P = 0.004). As a result, caution was required when interpreting OS outcomes. No quantitative assessment of publication bias for DFS/RFS was conducted owing to insufficient literature.