A nuclear transport-related gene signature combined with IDH and 1p/19q better predicts the prognosis of glioma patients

The nuclear transport system has been proposed to be indispensable for cell proliferation and invasion in cancers. Prognostic biomarkers and molecular targets in nuclear transport systems have been developed. However, no systematic analysis of genes related to nuclear transport in gliomas has been performed. An integrated prognostic classication involving mutation and nuclear transport gene signatures has not yet been explored. In the present study, we analyzed gliomas from a training cohort (TCGA dataset, n = 660) and validation cohort (CGGA dataset, n = 668) to develop a prognostic nuclear transport gene signature and generate an integrated classication system. Then, we developed a nuclear transport-related risk score (NTRS) for gliomas with a training cohort.

A nuclear transport-related gene signature combined with IDH and 1p/19q better predicts the prognosis of glioma patients Background The nuclear transport system has been proposed to be indispensable for cell proliferation and invasion in cancers. Prognostic biomarkers and molecular targets in nuclear transport systems have been developed.
However, no systematic analysis of genes related to nuclear transport in gliomas has been performed. An integrated prognostic classi cation involving mutation and nuclear transport gene signatures has not yet been explored.

Methods
In the present study, we analyzed gliomas from a training cohort (TCGA dataset, n = 660) and validation cohort (CGGA dataset, n = 668) to develop a prognostic nuclear transport gene signature and generate an integrated classi cation system. Then, we developed a nuclear transport-related risk score (NTRS) for gliomas with a training cohort.

Results
Gene set enrichment analysis (GSEA) showed that glioblastoma (GBM) was mainly enriched in nuclear transport progress compared to lower-grade glioma (LGG Conclusions NTRS is associated with poor outcomes and could be an independent prognostic marker in diffuse gliomas. Prognostic classi cation combined with IDH mutation, 1p/19q codeletion and NTRS could better predict the survival of glioma patients.

Background
Eukaryotic cells are divided into the nucleus and cytoplasm by the nuclear membrane. The movement of macromolecules between the nucleus and the cytoplasm, mostly including proteins and RNAs, occurs via the nuclear transport system [25]. The nuclear transport system includes three main components: the nuclear pore complex (NPC), RanGTPase and the nuclear transport receptor (NTR) [12]. It has been reported that the nuclear transport system plays an indispensable role in cancer development and metastasis [23,24]. Targeting the nuclear transport system could be a promising therapeutic approach [1,19]. However, a single molecule cannot represent the overall activity of the system, and a systemic analysis of nuclear transport and its prognostic value in cancer involving an expression pro le is lacking.
Gliomas are the most common primary tumors of the central nervous system and are classi ed by histologic and genomic phenotype [2,17]. In fact, it is not only genomic characteristics such as IDH mutation and 1p/19q codeletion but also transcriptomic and methomic characteristics that can be used as biomarkers of molecular classi cation [4,7]. Many models of gene signatures based on RNA-seq data can predict prognosis and be employed as an independent prognostic factor [28, 30, 31]. However, integrated prognostic classi cation with classical molecular biomarkers requires further study.
In this study, using RNA-seq data from TCGA as a training cohort and data from CGGA as a validation cohort, we established a nuclear transport risk score (NTRS) and tested the correlations between NTRS and clinicopathologic characteristics. We found that NTRS was an independent biomarker of prognosis and was associated with cell cycle-related pathways. Finally, combined with IDH mutation and 1p/19q codeletion, the value of NTRS in prognostic classi cation was validated. Taken together, our results indicated that the nuclear transport-related gene signature was strongly associated with poor outcomes and could serve as a novel biomarker for prognostic classi cation in diffuse gliomas.

Data Source
The data from the TCGA training set included RNA-seq data and clinical data from patients (n=660) with LGG and GBM from cBioPortal (http://www.cbioportal.org) [4,7]. The glioma patients included in the validation set (n=668) came from CGGA (http://www.cgga.org.cn/index.jsp) [11]. Integrated diagnosis was performed according to the World Health Organization (WHO) classi cation (2016). The patient characteristics are summarized in Supplementary tables 1 and 2.

Generation of NTRS
The nuclear transport gene set (n=338) was collected from the Molecular Signature Database v7.0 (http://software.broadinstitute.org/gsea/msigdb). Univariate Cox regression analysis was carried out to pre lter genes associated with nuclear transport and 251 genes correlated with survival (P ≤0.01). Seven genes and their regression coe cients were calculated according to least absolute shrinkage and selection operator (LASSO) regression [28]. The risk score was calculated according to the formula presented in Figure 1B.

Statistical Analysis
The optimal cut-off value for NTRS was determined via ROC curve analysis. Brie y, in the ROC curves, the x-axis was plotted as "1-speci city" (false positivity), and the y-axis was plotted as the "sensitivity" (true positivity). The optimal cut-off value was determined on the basis of the Youden index (Y), which was the point with maximum sensitivity and speci city (Y = sensitivity+ speci city − 1) [9]. Student's t test was performed to compare the NTRS values of two different groups. Tukey's multiple comparisons test was performed to compare the NTRS values of more than two groups. Differences in clinicopathological characteristics between groups were tested with chi-squared tests. Patient survival was analyzed via the Kaplan-Meier method. Univariate and multivariate Cox regression analyses were performed to evaluate independent prognostic factors by using SPSS software. ROC curve analysis was performed to predict overall survival (OS). P<0.05 was considered statistically signi cant. (*p<0.05, **p<0.01, ***p<0.001).

Results
Identi cation of a 7-gene nuclear transport-related signature for the prognosis of glioma.
First, we analyzed the expression of the nuclear transport gene set with the TCGA dataset. GBM showed distinct nuclear transport phenotypes from LGG (Supplementary Fig. 1). Gene set enrichment analysis (GSEA) based on the TCGA and CGGA datasets also con rmed that the GBM group was enriched for transcriptional programs related to nuclear transport (Fig. 1A). To develop a gene signature based on the nuclear transport pathway, we rst screened the glioma samples and nuclear transport-related genes in the training cohort. From the matrix of 660 gliomas and 336 genes, we selected 251 genes associated with OS (P ≤ 0.01) by univariate Cox regression analysis (Fig. 1B). Seven genes were selected via LASSO regression analysis, and the nuclear transport risk score (NTRS) in the training cohort was obtained ( Fig. 1C, D). To analyze the relationships between NTRS and clinical characteristics, 660 patients from the training cohort and 668 patients from the validation cohort with clinical information were selected. The distribution of clinical characteristics, genetic characteristics and the expression of 7 genes in the patients are shown ( Fig. 2A). As we expected, NTRS increased according to glioma grade (Supplementary Fig. 2A) and was higher in patients who were over 50 years old without IDH mutation or 1p/19q codeletion ( Supplementary Fig. 2B-D). Furthermore, in the subtype classi ed according to histology or molecular markers, NTRS was elevated in subgroups with shorter survival times, such as patients with the glioblastoma subtype or the subtype without IDH mutation and 1p/19q codeletion ( Supplementary  Fig. 2E, F). These ndings were validated in the CGGA dataset (Fig. 2B). In brief, NTRS was signi cantly associated with clinical and genetic characteristics that have been reported as prognostic markers in gliomas.

Validity of NTRS as an independent prognostic marker in glioma
To investigate the prognostic value of NTRS, we rst calculated the cut-off value by maximizing the Youden index through ROC analysis. The patients were divided into NTRS High and NTRS Low groups ( Fig. 3A). Subsequently, we validated the correlation between the NTRS group and clinicopathological factors in the TCGA dataset and CGGA dataset (  (Fig. 3E). These data indicated that NTRS is a promising prognostic marker for gliomas.
NTRS High gliomas exhibit elevated cell cycle and immune responses.
To analyze the association between NTRS and a poor prognosis of glioma patients, we performed gene set enrichment analysis (GSEA) coupled with enrichment map analysis to visualize the enriched GO biological processes. The NTRS High group was enriched in transcriptional programs related to the cell cycle, DNA replication and immune responses (Fig. 4A, B). Based on the identi ed differentially expressed genes (P < 0.05), GO analysis veri ed that the cell cycle and immune responses were signi cantly enriched in NTRS High patients (Fig. 4C). These transcriptomic data indicated that NTRS High gliomas exhibit increased proliferative activity, which might result in a worse outcome.
NTRS is a potential marker for prognostic classi cation, combined with IDH mutation and 1p/19q  . 5A). Subsequently, we performed survival analysis in different subgroups. The NTRS High group exhibited shorter survival among patients with WHO grade III gliomas classi ed by IDH mutation and 1p/19q codeletion (Fig. 5B). These results indicated that NTRS could be more effective as a marker when combined with other prognostic markers for gliomas. To test this hypothesis, we analyzed the prognostic value in subgroups strati ed by IDH mutation and 1p/19q codeletion. In both the subgroup with IDH mutation without 1p/19q codeletion and the subgroup without IDH mutation, overall survival (OS) was decreased in patients with a high NTRS (Fig. 6A). These results were further con rmed in the validation cohort (Fig. 6B). In conclusion, by combining data on IDH mutation and 1p/19q codeletion with NTRS, we established a prognostic classi cation model for survival prediction in glioma patients (Fig. 6C).

Discussion
The nuclear transport system has been proven to be critical for tumorigenesis and the development of cancer [1]. Nuclear transport could serve as a therapeutic target in several cancer types [8,13,23]. Many genes involved in nuclear transport have been reported to be associated with the prognosis of cancer patients [3]. These results indicate that nuclear transport may serve as a marker of prognosis in cancer. In this study, we used RNA-seq data from the TCGA and CGGA databases to generate a seven-gene nuclear transport risk score (NTRS) to predict the prognosis of glioma patients. We further con rmed that NTRS was an independent prognostic marker and better predicted overall survival compared to traditional factors. Our work establishes a novel nuclear transport-based gene signature for the prediction of glioma patient survival.
One shortcoming of this work was the lack of clinical validation and functional research on NTRS. With the development of RT-PCR, Nanostring and next-generation sequencing (NGS), gene signatures have been broadly applied in the clinic for the prediction of recurrence and the response to therapy [6,10,22]. Gene signature panels based on NTRS should be developed, and real-world research (RWR) involving multiple centers should be performed in the future. Although all seven genes were signi cantly associated with survival in multiple datasets (TCGA, CGGA and Rembrandt) and several of the genes have been reported to be functional in gliomas [14][15][16], further experiments are needed to study the function and mechanism of these seven genes, which will be performed in the future.
Since the publication of the 2016 WHO classi cation of tumors of the central nervous system, integrated classi cation has been generally applied to glioma. With the availability of public databases, the integration of data on histology and mutation, methylation and mutation or mRNA expression and mutation can divide patients into different subgroups [4,5,7,27]. Furthermore, with the development of arti cial intelligence and machine learning, digital images obtained via magnetic resonance imaging and histopathological analysis can be used to predict not only overall survival but also IDH mutation and 1p/19q codeletion [18,21]. In the near future, the diagnosis of gliomas will involve the combination of multidimensional data. At the molecular level, glioma panels including mutation, methylation and gene expression data will be rapidly developed. In this study, we made a preliminary attempt to combine NTRS with IDH mutation and 1p/19q codeletion data for prognosis. The patients in the ve subgroups exhibited signi cantly different outcomes (Fig. 6C). Our research demonstrated that the nuclear transport-related gene signature could serve as a novel marker for prognostic classi cation in combination with IDH mutation and 1p/19q codeletion.

Conclusions
Risk score based on nuclear transport system is signi cantly associated with poor clinicopathologic characteristics and is an independent prognostic marker in diffuse gliomas. Combined with IDH mutation, 1p/19q codeletion and the nuclear transport risk score could better predict the overall survival of glioma patients.

Funding
This research was supported by grants from the National Natural Science Foundation of China (81602196 and 81702456) and the Chongqing Basic Research Project (cstc2016jcyjA2194). These funds were used for the design of the study, analysis of the data and writing of the manuscript.