TRIM59: A potential diagnostic and prognostic biomarker in human tumors

TRIM59 is a protein that is highly expressed in a variety of tumors and promotes tumor development. However, the use of TRIM59 as tumor diagnosis and prognosis biomarker has not been fully explored. We collected datasets from the cancer genome atlas (TCGA) and gene expression omnibus (GEO) to investigate its potential as a biomarker for diagnosis and prognosis. A total of 46 studies, including 11,558 patients were included in this study. Here, we showed that TRIM59 was significantly upregulated in 15 type of human solid tumors in comparison to their adjacent tissues. Receiver operating characteristic curve (ROC) results provided further evidence for the use of TRIM59 as a potential tumor diagnosis biomarker. Overall survival (OS) was compared between TRIM59 high expression and low expression groups. High expression of TRIM59 indicated a poor prognosis in multiple solid tumors. Taken together, these analyses showed that TRIM59 was upregulated in various types of tumors and had the potential to be used as a diagnostic and prognostic biomarker in human solid tumors.


Introduction
Cancer has always been a major global health concern and creates a heavy financial burden for patients and the health care system [1]. It is estimated that there might be 18.1 million new cancer cases and 9.6 million cancer deaths in 2018 [2]. Cancers are always characterized by a group of abnormal expressed genes, and these genes have the potential to be used as diagnosis markers or prognosis predictors [3,4].
TRIM59 is a member of TRIM protein family, which is comprised of a RING domain, bbox domain, coiled helix domain, and c-terminal non-specific domain [5]. Recent studies have shown that TRIM59 played a role in epigenetic modification [6], embryonic development [7] and autophagy [8], and was upregulated in multiple tumors [9]. It has been found that downregulation of TRIM59 significantly inhibited tumor growth in vitro experiments and animal models [10,11]. However, no large-scale clinical studies have been performed to demonstrate the relationship between TRIM59 and tumor diagnosis and prognosis. In this study, the cancer genome atlas (TCGA) and gene expression omnibus (GEO) datasets were used to analyze the expression of TRIM59 in tumors and evaluate the value of TRIM59 as a biomarker for tumor diagnosis and prognosis.

ROC, AUC and SROC analyses
R package "OptimalCutpoints" was first used to determine the optimal cutpoint in each TCGA project, at which the sample types (tumor or normal) could be best distinguished. Then the sample types were predicted by the TRIM59 expression level according to the cut point of each project using the "Youden" prediction model [12]. ROC and associated 95% confidence intervals (CIs) of each project was also calculated using the "optimal.cutpoints" function of package "OptimalCutpoints". Module "midas" of Stata 15.0 (StataCorp LLC, USA) was used for the meta-analysis of ROC curves of all the projects. Sensitivity and specificity of the metaanalysis were evaluated, and the summary of ROCs (SROC) was calculated. Deeks' funnel plot asymmetry test was used to investigate potential publication bias in SROC analysis.

Overall survival analysis
Datasets regarding TRIM59 expression and clinical characteristics of human cancers were downloaded from TCGA and GEO databases. According to the TRIM59 expression level in each data set, the samples were divided into high expression and low expression groups in comparison to the median expression level. The "survival" package was used to calculate the hazard ratio (HR) and 95% CIs for the high-expression TRIM59 group versus the low-expression TRIM59 group.

Statistical analysis
All the analyses were performed on R (Version 3.6.1) and Stata 15.0 (StataCorp LLC, USA). Normalized expression data were downloaded from TCGA or GEO datasets. Unpaired t test was used for comparison between groups. As for ROC analysis, R package "OptimalCutpoints" was first used to calculate the optimal cutpoint for tumor diagnosis, then ROC and 95% CIs were calculated for each project. Stata module "medias" was used for ROC meta-analysis. R package "survival" was used for survival analysis and package "meta" was used for survival meta-analysis.

TRIM59 was highly expressed in human tumors
Previous studies have shown that TRIM59 was highly expressed in a variety of tumors and closely related to the occurrence and development of tumors. We analyzed 15 types of tumor datasets that included sufficient numbers of tumors and adjacent normal tissues in TCGA, confirming that TRIM59 was highly expressed in tumor samples in comparison to their adjacent tissues (Fig 1).

High expression of TRIM59 showed high efficacy in the diagnosis of human tumors
To explore the value of high expression of TRIM59 in tumors, we tested if TRIM59 could be used to identify healthy and tumor samples. ROC analysis was performed based on the data obtained from TCGA. Results presented in Table 1 demonstrated that TRIM59 showed high  diagnosis efficacy in multiple cancers, especially in CHOL, with a prediction accuracy up to 100%. The pooled sensitivity and specificity were 0.84 (95% CI:0.77-0.89) and 0.90 (95% CI:0.85-0.94) respectively (Fig 2). Additionally, the area under ROC (AUC) of the SROC was 0.94 (95% CI:0.91-0.96) (Fig 3A). In order to test the potential publication bias, Deeks' funnel plot asymmetry test was performed and revealed that no significant substantial publication bias was found in SROC analysis (p = 0.75) (Fig 3B). These results suggested that the high expression of TRIM59 could be used as a diagnostic marker in different types of human cancers.

Expression of TRIM59 indicated prognosis in human tumors
To investigate the relationship between TRIM59 and tumor prognosis, OS meta-analysis was conducted. A total of 21 TCGA and 26 GEO datasets containing 11,558 patients were included (  (Fig 4).

Discussion
In this study, we used TCGA datasets to demonstrate that TRIM59 was upregulated in multiple cancers in comparison to the adjacent normal tissues. To explore the diagnosis efficacy of TRIM59, we performed ROC and SROC analyses to identify that high expression of TRIM59 showed a high diagnosis efficacy in each tumor type (AUC > 0.5), and with an AUC of summary ROC of 0.94 (0.91-0.96). We also combined GEO datasets and performed a meta-analysis to reveal that high expression of TRIM59 showed significant (p<0.05) poor prognosis in KIRP, LGG, LUAD, lung cancer and showed better prognosis in CESC and SKCM. Especially  Table 1.
https://doi.org/10.1371/journal.pone.0257445.g003 in LUAD, the expression of TRIM59 in tumors with better prognosis was still higher than that in adjacent tissues (S1 Fig). TRIM59 is closely related to cancers. A previous study used Immunohistochemistry (IHC) to determine the expression of TRIM59 in 291 cases of 37 tumor types, and found that TRIM59 expression was upregulated in tumor samples, particularly in lung, breast, liver, skin, tongue and mouth (squamous cell cancer) and endometrial cancers [9]. In subsequent studies, it was shown that upregulation of TRIM59 can promote tumor growth in tumor cell lines and animal models, while downregulation had the opposite effect. These cancers include pancreatic cancer [13], cholangiocarcinoma [14], ovarian cancer [15,16], lung cancer [17][18][19], breast cancer [10,20,21], euroblastoma [22], medulloblastoma [23], hepatocellular carcinoma [24], glioblastoma [11], colorectal cancer [25,26], bladder cancer [21], prostate cancer [27], cervical cancer [28], osteosarcoma [29], gastric cancer [30]. However, results from these studies were limited by a small number of tumor samples, or the use of in vitro experimentation or animal models could not fully address the relationship between TRIM59 and human cancers. Thanks to the gene expression data and associated prognosis information provided in public database, we identified that TRIM59 was highly expressed in most solid tumors and could indicate the prognosis in several cancers.
This study is a meta-analysis of multiple solid tumors, indicated that TRIM59 has the potential to be used as a diagnostic molecule for a variety of tumors, and special attention should be paid to the abnormal high expression of TRIM59 in specific tissues. Moreover, it plays a prognostic role in specific tumors, especially in KIRP/LGG/LUAD/Lung cancer/CESC/ SKCM. Detection of TRIM59 expression in these tumor tissues is helpful for us to evaluate the prognosis of patients.
The limitation of this study lies in the fact that it is only the data analysis of biological database, and further verification of TRIM59 expression level and follow-up information are needed in clinical samples for each specific tumor type. Basic research on TRIM59 is also needed to improve our understanding of tumors.

Conclusions
In this study, based on TCGA datasets, we revealed that TRIM59 was upregulated in 15 types of solid tumors. Additionally, TRIM59 have a high efficacy in diagnosis and prognosis prediction in various tumor types. TRIM59 have the potential to be used as diagnosis marker or prognosis predictor in tumors.