An extensive literature search was conducted using the PubMed, Web of Science, Google Scholar, Cochrane Library, PMC, Wanfang, and China National Knowledge Infrastructure (CNKI) databases, with an April 1, 2023 cut-off date. The following phrases were searched in various combinations: (“STEAP1”) and (“cancer” OR “carcinoma” OR “tumor” OR “neoplasm”). In addition, the reference list of relevant literature was examined to find studies in which STEAP1 was linked to clinical events.

The inclusion criteria for this study were: (1) cancer patients, without restriction on the type of cancer; (2) an evaluation of correlations between STEAP1 expression and clinical features; (3) use of immunohistochemical techniques to evaluate STEAP1 expression; (4) classification of patients into two groups based on their STEAP1 expression level; (5) presentation of direct or indirect hazard ratios (HRs) for overall survival (OS), together with the associated 95% confidence intervals (95% CI).

The exclusion criteria for this study were: (1) studies lacking survival data; (2) not categorized by STEAP1 expression into 2 groups as low and high expression; (3) duplicate studies; (4) publications in the form of reviews, meta-analyses, case reports, fundamental studies, animal studies, letters, expert opinions, conference abstracts, and reviews.

Papers that were potentially suitable were reviewed in their entirety. Data and information from eligible studies were independently extracted by two researchers, who then separately verified the extracted data. A third researcher arbitrated any inconsistent points that arose between the two researchers.

For each eligible study, the first author’s name, detection method for STEAP1 expression, STEAP1 expression level, and clinicopathological factors including gender, age, TNM stage, distant metastasis, lymph node metastasis, tumor cell differentiation, and tumor size were recorded. When only Kaplan-Meier curves were presented in the article, Engauge Digitizer version 11.2 was used to calculate survival from the curves and to determine HRs and 95% CIs. If both univariate and multivariate analysis results were shown, the latter were used because of their superior precision.

Study quality was evaluated by the Newcastle-Ottawa Scale (NOS), which takes into account three primary categories: selection, comparability, and outcome. The spectrum of NOS scores ranges from the minimum of 0 stars to the maximum of 9 stars. A study was deemed to be high-quality if it had an NOS score of at least 6.

The Gene Expression Profiling Interactive Analysis 2 (GEPIA2.0, http://gepia2.cancer-pku.cn/#index) was used to obtain data on STEAP1 expression and prognostic value in cancer patients [13].

Cancer patients from the TCGA dataset were separated into two categories (high and low expression) using a 50% cut-off threshold, and HR was calculated according to the Cox proportional hazards model. The Kaplan-Meier plotter was utilised to analyse the survival of cancer patients, with differences between groups evaluated using the log-rank test [14].

The CancerSEA (http://biocc.hrbmu.edu.cn/CancerSEA/) dedicated single cell sequencing database was used to obtain information on the functional status of cancer cells at the single cell level. The analysis was performed based on data obtained from single cell sequencing, and associations between STEAP1 expression and biological functions of tumors was examined. STEAP1 expression patterns of individual cells within the TCGA dataset were shown using T-SNE diagrams.

The statistical analysis was performed using Stata 16.0 (StataCorp LLC, 4905 Lakeway Drive College Station, TX, USA). HRs and 95% CIs assessed the associations between STEAP1 expression and OS, lymph node metastases, TNM stage, and tumor cell differentiation. Patient groups with bad prognosis were defined by an HR $>$1. Heterogeneity was assessed using a Q test based on the Chi-square and I${}^2$ statistics. In cases where heterogeneity was minimal or nonexistent, a fixed-effects model was employed. In contrast, a random-effects model was applied when there was substantial heterogeneity. A funnel plot was applied to evaluate potential publication bias. p 0.05 was deemed statistically significant [15]. A sensitivity analysis was conducted to determine the consistency of the outcomes.

The search approach identified 386 articles in the different databases. Of these, 157 duplicate articles were eliminated, leaving 229 articles that underwent a second round of meticulous scrutiny of their title and abstract. After reading the entire text of 91 publications, 7 articles that satisfied the inclusion requirements were ultimately chosen for meta-analysis. Details regarding the selection procedure are shown as a flow chart in Fig. 1.

Fig. 1.

Flow charts of study search and selection for meta-analysis.

Based on the search strategy and selection criteria, 7 papers from Germany, Brazil and China that included 1263 patients in total were selected for further investigation (Table 1, Ref. [11, 16, 17, 18, 19, 20, 21]). The cancer types included osteosarcoma, gastric cancer, colorectal cancer, Ewing’s sarcoma, PCa, ovarian cancer, and lung adenocarcinoma.

Because of the lack of a statistically significant difference across trials, we used a fixed effects model (I${}^2$ = 0.0%), high STEAP1 expression was found to be significantly correlation with declining OS (HR = 1.87, 95% CI: 1.49–2.34, p 0.001) (Fig. 2).

Fig. 2.

Forest plot examining the relationship between six transmembrane epithelial antigen 1 (STEAP1) expression and overall survival.

To assess clinical relevance, fixed-effects subgroup analyses were conducted based on sample size, number of follow-up months, and data extraction methods (Fig. 3). High STEAP1 expression was associated with falling OS in cancer patients from China (HR = 1.88; 95% CI: 1.48–2.39; p 0.001) (Fig. 3a). Sample size did not affect the observed predictive value of STEAP1 expression (Fig. 3b). Moreover, the prognostic significance of STEAP1 expression for OS remained consistent in each of the cancer types studied, regardless of the length of follow-up or the method of data extraction (Fig. 3c,d).

Fig. 3.

Stratified analyses to determine the relationship of STEAP1 expression and overall survival (OS). (a) Subgroup study of hazard ratios (HRs) of OS based on country. (b) Subgroup study of HRs of OS based on sample size. (c) Subgroup study of HRs of OS based on follow-up time. (d) Subgroup study of HRs of OS based on the method of data extraction.

A meta-analysis of the correlations between STEAP1 expression and various clinicopathological features was conducted using data from 7 studies involving 765 tumor specimens. As shown in Fig. 4, no significant correlations were found between high STEAP1 expression and TNM stage (OR = 1.67, 95% CI: 0.67–4.16, p 0.001, Fig. 4a), tumor cell differentiation (OR = 3.09, 95% CI: 0.85–11.24, p 0.001, Fig. 4b), or gender (OR = 1.37, 95% CI: 0.77–2.43, p 0.001, Fig. 4c). However, high STEAP1 expression in tumor tissue was significantly associated with the presence of lymph node metastasis (OR = 3.19, 95% CI: 1.26–8.09, p 0.001, Fig. 4d).

Fig. 4.

Forest plot for the relationship between STEAP1 expression with clinical characteristics. Forest plot for the relationship between STEAP1 expression with (a) TNM stage, (b) tumor cell differentiation, (c) gender, and (d) lymph node metastasis.

We completed a sensitivity analysis to assess the stability of the meta-analysis for the expression of STEAP1 and OS. The pooled HR was not substantially altered by excluding individual studies, thus demonstrating the consistency of the findings (Fig. 5a).

Fig. 5.

Sensitivity analysis and publication bias for STEAP1 expression and OS. (a) Sensitivity analysis, and (b) funnel plot of publication bias for OS.

The funnel plot and Begg’s bias test were applied to assess the potential for publication bias. A nearly symmetrical form was observed (Fig. 5b). Moreover, the Begg’ s test (p = 0.806) and the Egger test (p = 0.37) of OS in the included studies indicated there was no appreciable publication bias.

To confirm and validate our findings, GEPIA2.0 (Peking University, Beijing, China) was employed to examine STEAP1 expression levels in different cancer types. Kaplan-Meier analysis indicated a substantial correlation between elevated STEAP1 expression and decreased OS in all cancer types (Fig. 6a). High STEAP1 expression was strongly associated with bad prognosis in lung adenocarcinoma (LUAD), adrenocortical carcinoma (ACC), pancreatic adenocarcinoma (PAAD), brain low grade glioma (LGG), and kidney renal clear cell carcinoma (KIRC) (Fig. 6b–f). Overall, these results support the conclusion that increased STEAP1 expression is associated with worse OS in several different tumor types. Hence, STEAP1 expression could potentially serve as a prognostic indicator in several types of cancer.

Fig. 6.

Prognostic value of STEAP1 expression in various cancer types. GEPIA2.0 was utilized to examine the relationship between STEAP1 expression and OS in all cancer types (a), Lung adenocarcinoma (LUAD) (b), pancreatic adenocarcinoma (PAAD) (c), adrenocortical carcinoma (ACC) (d), brain low grade glioma (LGG) (e), kidney renal clear cell carcinoma (KIRC) (f).

Sequencing the transcriptome of a single cell is a valuable method for analysing the potential roles of molecules at level of the single cell [22, 23]. The CancerSEA website was applied here to investigate the relationship between the change of STEAP1 expression and other biological functions at the single cell level (Fig. 7a). STEAP1 expression was favorable correlation with angiogenesis in lung adenocarcinoma (LUAD). Moreover, a negative correlation was reported between STEAP1 expression and apoptosis in acute myelocytic leukemia (AML) patients (Fig. 7a), whereas a positive correlation was observed with the cell cycle, DNA repair, invasion, proliferation, and stemness, suggesting a possible mechanism of action in this cancer type. Based on the analysis of Fig. 7a, stemness was positively correlated with STEAP1 expression in AML, uveal melanoma (UM), and head and neck squamous cell carcinoma (HNSCC), while a negative correlation was observed between quiescence and STEAP1 expression in AML and UM, both of which are statistically significant. STEAP1 expression profiles at the single cell level in UM are shown by t-SNE plots in Fig. 7b. DNA repair, DNA damage, apoptosis, and metastasis were inversely correlated with STEAP1 expression in UM (Fig. 7c).

Fig. 7.

STEAP1 expression correlates with tumor biological behavior at the single cell level. (a) The CancerSEA tool was applied to explore the relationship between STEAP1 expression and different biological functions in tumors. * p 0.05; ** p 0.01; *** p 0.001. (b) STEAP1 expression in single cells of uveal melanoma (UM) as shown by t-SNE plots. (c) Correlation of STEAP1 expression with tumor biological behavior in UM.