Exosomal samples isolated from LUAD patients and healthy controls were validated by performing Transmission electron microscopy (TEM), western blotting (WB) and nanoparticle tracking analysis (NTA). As shown, these vesicles showed typical oval-shaped extracellular vesicles with diameters of 50–150 nm by TEM (Figure 1A). WB demonstrated that the concentrations of the exosome surface protein markers, TSG101, CD63, and CD9, were enriched in serum exosomes but not in exosome-depleted supernatants (EDS) (Figure 1B). NTA showed that the diameters of most serum exosomes were mainly 91.6 nm (Figure 1C). All these results suggested that we successfully purified exosomes from the serum, which laid a foundation for further analysis.

High throughput sequencing was performed using 50 LUAD serum exosomal samples and 50 healthy controls (the 50 LUAD and 50 healthy samples were grouped and mixed into 5 repeats, respectively). The results showed that a total of 722 differentially expressed lncRNAs between the two groups were identified according to our screening disciplines, including 14 up-regulated and 708 down-regulated lncRNAs. Cluster analysis heat map, scatter plot, and volcano plot were shown to present the distinguishable lncRNAs expression profiles (Figures 1D–F).

Since up-regulated lncRNAs have greater potential to be used as early diagnostic markers or therapeutic targets, we focused on 7 most upregulated lncRNAs, including ENST00000417930, ENST00000578785, TCONS_l2_00006336, uc010jub.1, ENST00000569809, ENST00000430694, and ENST00000513626, among which, only ENST00000417930, uc010jub.1 and ENST00000513626 were successfully designed with specific primers. To confirm the High throughput sequencing results, expression levels of three serum exosomal lncRNAs were measured in 84 LUAD patients and 110 healthy controls through qRT-PCR, the results showed that only ENST00000513626 (LUCAT1) was significantly up-regulated in LUAD patients (Figure 1G), while the other two genes were not detectable due to low expression concentrations. The potential diagnostic value of serum exosomal LUCAT1 was evaluated through Receiver operating characteristic curve (ROC) analysis. As expected, the area under the ROC curve (AUC) of serum exosomal LUCAT1 for LUAD diagnosis was 0.852 (95% CI = 0.794–0.898), with a diagnostic sensitivity and specificity reaching 85.45 and 77.38%, respectively (Figure 1H). These data suggested that LUCAT1 was a promising biomarker for the diagnosis of LUAD and may play an important role in LUAD progression.

VEGF-A was the most potent pro-angiogenic factor that directly induced endothelial cell proliferation, migration and promotes tumor growth. It has been reported that miR-4316 inhibits gastric cancer proliferation and migration via directly targeting VEGF-A. However, it is unclear whether miR-4316 regulate the metastasis of LUAD through VEGF-A.

The instability of majority of the lncRNAs in serum remains a significant limitation for clinical application. Serum exosomes samples were incubated with RNase A for 0, 30, and 60 min, and then exosomal LUCAT1 expression was measured. Interestingly, the expression levels of exosomal LUCAT1 remained unchanged upon RNase A treatment (Figure 1I), which fully proved its sufficient suitability as tumor markers for LUAD diagnosis.

We also analyzed the expression data of LUCAT1 in 535 LUAD tissues and 59 normal tissues from The Cancer Genome Atlas (TCGA) database. LUCAT1 was significantly up-regulated in LUAD tumors than in non-tumor tissues (Figure 2A). Analyzing of clinical data revealed that LUCAT1 expression level was positively correlated with higher rates of Lymph Node Metastasis (LNM) (Figure 2B) and advanced Tumor Node Metastasis (TNM) stage (Figure 2C).

To further test whether expression level of LUCAT1 correlated with the outcomes of LUAD patients, the expression level of LUCAT1 was detected in 98 LUAD patients with different clinicopathological features and 82 paired normal tissues using In situ Hybridization (ISH). Consistent with TCGA results, LUCAT1 expression level was higher in LUAD tissues compared with normal tissues (Figures 2D,E). Kaplan-Meier survival analysis revealed that patients with high LUCAT1 expression had a markedly lower overall survival compared to those with low LUCAT1 expression (Figure 2F).

To understand the role of LUCAT1 in LUAD cells, the expression of LUCAT1 was detected in normal lung epithelial cells (BEAS-2B) and LUAD cells (NCI-H1299 and A549) by qRT-PCR analysis. As shown in Figure 2G, LUCAT1 expression in NCI-H1299 and A549 cells was much higher than that in BEAS-2B cells. Taken together, all these results indicated that LUCAT1 was significantly overexpressed in LUAD and might be a potential biomarker for predicting prognosis in LUAD patients.

In order to better investigate the biological functions of LUCAT1 in the tumorigenesis and development of LUAD, loss- and gain-of function experiments were conducted in NCI-H1299 and A549 cells. Specific siRNAs were used to knock down LUCAT1 expression, the knockdown efficiency of LUCAT1 in two LUAD cell lines were verified by qRT-PCR (Figure 3A). As shown, LUCAT1 knockdown strikingly attenuated the invasive abilities of NCI-H1299 and A549 cells (Figure 3B). Similarly, the migration capability of NCI-H1299 and A549 cells in wound healing assays was significantly decreased after the silence of LUCAT1 (Figure 3C). Subsequently, NCI-H1299 and A549 cells were transfected with the overexpression lentiviral vector (pcDNA3.1- LUCAT1) (Figure 3D). The adverse results were obtained, that is, overexpressed LUCAT1 strikingly enhanced the invasive abilities of two LUAD cells (Figure 3E). Additionally, wound healing assay showed that LUCAT1 overexpression significantly promoted the migratory ability of NCI-H1299 and A549 cells (Figure 3F).

Also, WB analysis revealed that LUCAT1 silencing could induce a significant elevation in epithelial maker E-cadherin expression, and a decreased expression of mesenchymal makers, including N-cadherin, β-catenin, and vimentin (Figure 3G). Conversely, LUCAT1 overexpression significantly increased mesenchymal makers N-cadherin, β-catenin, and vimentin expression, and decreased the expression of epithelial protein E-cadherin (Figure 3H).

To further investigate the biological function of LUCAT1, we established in vivo xenograft nude mouse model of LUAD metastasis via a tail vin injection method. Short hairpin RNA (shRNA) was utilized to stably silence LUCAT1 expression in NCI-H1299 cell lines (Figure 4A). In line with the in vitro data, the numbers of metastatic nodules in the lungs were significantly decreased in nude mouse injected with NCI-H1299 sh-LUCAT1 cells compared those injected with NCI-H1299 sh-NC cells (Figure 4B). Metastatic nodules were dissected out for haematoxylin and eosin (HE) analysis and further confirmed the metastasis nodules (Figure 4C). Immunohistochemistry (IHC) was performed to quantity the expression of biomarkers of metastasis in mouse model. Consistent with our observations in vitro, silencing of LUCAT1 significantly enhanced E-cadherin expression and attenuated the expression of N-Cadherin in xenografts, suggesting that silencing of LUCAT1 inhibited lung metastasis (Figure 4D). And the expression levels of LUCAT1/miR-4316/VEGF-A axis in metastatic LUAD in mouse model were also investigated via qRT-PCR. As expected, metastatic nodes formed from NCI-H1299 sh-LUCAT1 cells exhibited decreased LUCAT1 expression, while the expression of miR-4316 were upregulated along with lung metastasis. Similarly, a significant difference was also observed in the expression of VEGFA between NCI-H1299 sh-LUCAT1 and sh-NC groups in metastatic nodules (Figure 4E). These data showed that LUCAT1 silencing inhibited tumor metastasis in vivo.

To explore whether LUCAT1 promotes LUAD metastasis via the competing endogenous RNAs (ceRNAs) mechanism. RNA fluorescence in situ hybridization (FISH) assay was performed using an LUCAT1 special probe, and the results revealed that LUCAT1 was mainly localized in the cytoplasm of NCI-H1299 and A549 cells (Figure 5A).

The putative candidate miRNAs binding to LUCAT1 were predicted using the bioinformatics tool miRDB (http://mirdb.org/), and miR-4316 was predicted to interact with LUCAT1. Then, we constructed a reporter vector in which the putative miR-4316-3p-binding site in the LUCAT1 sequence was mutated by base mutations, followed by co-transfection with control miRNA and miR-4316 mimics into HEK293T cells (Figure 5B). As expected, dual-luciferase reporter assay revealed that miR-4316 mimic reduced luciferase activities with the LUCAT1-WT expression vector, but not the LUCAT1-MUT, which suggested that miR-4316 could directly bind to LUCAT1 (Figure 5C). Besides, the regulation of miR-4316 by LUCAT1 in NCI-H1299 and A549 cells were analyzed, and the results showed that knockdown of LUCAT1 significantly promoted the expression of miR-4316. In contrast, the levels of miR-4316 were significantly reduced by LUCAT1 overexpression (Figure 5D). These findings indicated that LUCAT1 targeted directly miR-4316 and negatively regulated its expression.

In vitro rescue assays were performed to identify the biological function of LUCAT1/miR-4316 regulatory axis. NCI-H1299 and A549 cells were co-transfected with pcDNA3.1-LUCAT1 and miR-4316 mimics, the transfection efficiency of miR-4316 mimics were determined by qRT-PCR analysis (Figure 5E). The results of the transwell invasion assays showed that overexpression of miR-4316 significantly inhibited cell invasion, and rescued the increased cell invasive capability caused by LUCAT1 overexpression in NCI-H1299 and A549 cells (Figure 5F). Meanwhile, wound healing assays showed the consistent results (Figure 5G), indicating that LUCAT1 attenuating LUAD cells invasion and migration by serving as miR-4316 sponge.

MiRNAs have been reported to exert their multiple biological functions mainly through degrading their downstream mRNA. By searching with the online prediction software Target Scan (http://www.targetscan.org/vert_72/), we identified that VEGFA was a potential target of miR-4316. Also, dual-luciferase reporter assay was performed by generating wild-type or WT sequence of VEGFA (Figure 6A). The results showed a significant reduction in luciferase activities after co-transfection of miR-4316 mimic and VEGFA-WT, which provided evidence that miR-4316 could directly bind to VEGFA (Figure 6B).

VEGFA has been proved to promote the metastasis of various types of solid cancers, including LUAD. We then analyzed the TCGA data, and found that VEGFA was upregulated in LUAD tissues compared with normal tissues (Figure 6C). Besides, the expression level of LUCAT1 was positively correlated with the expression level of VEGFA in 535 LUAD samples and 59 normal tissues (Figure 6D). Similarly, a higher expression of VEGFA was also observed in our LUAD cell lines (NCI-H1299 and A549) compared with the immortalized lung epithelial cell line BEAS-2B (Figure 6E).

To determine whether VEGFA was regulated by miR-4316 in LUAD cells, mRNA of VEGFA were measured when miR-4316 was overexpressed in NCI-H1299 and A549 cells. The results indicated that VEGFA were significantly decreased by miR-4316 mimics (Figure 6F). Then, expression levels of VEGFA were also measured when LUCAT1 was overexpressed or knocked down to identify the ceRNA network between LUCAT1 and VEGFA. Results showed that transfection of si-LUCAT1 into NCI-H1299 and A549 cells decreased the mRNA levels of VEGFA via releasing miR-4316, and the opposite results were observed when LUCAT1 was upregulated (Figure 6G).

To identify the effect of LUCAT1/miR-4316/VEGF-A axis on the biological function in LUAD cells, rescue assays were performed. We examined cells co-transfected with pcDNA3.1- LUCAT1 and si-VEGFA, and found that the enhanced invasion capabilities induced by LUCAT1 overexpression were partially reversed by co-transfection with si-VEGFA in NCI-H1299 and A549 cells (Figure 6H). Wound healing assays showed that si-VEGFA also partially reversed the enhanced migration capability induced by LUCAT1 overexpression (Figure 6I). These results suggested that LUCAT1 upregulated VEGFA expression by acting as miR-4316 sponge, thereby attenuating invasion and migration of LUAD cells.

Cancer cells frequently alter their metabolic pathways to adapt to environmental challenges or facilitate tumor proliferation, metastasis (Feng et al., 2020; Yang et al., 2020). Recently, growing evidence have provided for cancer cells’ dependent on glycolysis for energy production (Vander Heiden et al., 2009). It was reported that LUCAT1 was upregulated under high-glucose conditions, and knockdown of LUCAT1 inhibited the EMT in HG-treated HK-2 cells (Zhang et al., 2020), however, whether LUCAT1 participate in glycolysis of LUAD cells needs to be confirmed.

We first set out to evaluate the potential influence of high glucose on invasion and migration in LUAD cells. Capabilities of cell’s invasion and migration were determined under normal culture conditions or with high concentration of glucose (Normal glucose: 5.5 mM glucose, High glucose: 33 mM glucose). Our data clearly showed that high glucose culture significantly promoted invasion and migration of A549 cells, in comparison with the normal culture medium (Figures 7A,B). However, when treating with 2-deoxyglucose (2-DG, 0.5 mM), an inhibitor of glucose metabolism and inhibits glycolysis by acting on hexokinase, we found that the high glucose-induced invasion and migration could be effectively reversed (Figures 7A,B).

We also detected metabolic activities and found that silencing LUCAT1 expression in LUAD cells significantly suppressed the expression of metabolic enzymes and transporters, including LDH, GLUT1, GLUT3 (Figure 7C). Conversely, LUCAT1 overexpression promoted their expressions in A549 cells (Figure 7D). To further determine whether LUCAT1 was associated with the glucose metabolic changes in A549 cells, glucose consumption and metabolites were analyzed. As shown in Figures 7E,F, knockdown of LUCAT1 led to a significant decrease in glucose consumptions and lactate productions in A549 cells. In contrast, overexpression of LUCAT1 produced the opposite effects on them in A549 cells. In addition, LUCAT1 overexpression could promote the glucose consumption and lactate productions of A549 cells, while treatment with miR-4316 mimics partially reversed the phenotypes caused by LUCAT1 overexpression (Figures 7G,I). It was consistent with our suspect that knockdown of VEGFA partially reversed the increased glucose consumption and lactate productions induced by LUCAT1 overexpression in A549 cells (Figures 7H,J). These results suggested that LUCAT1 promoted glycolysis of LUAD cells via the upregulation of VEGFA by acting as miR-4316 sponge.

A total of 160 LUAD serum samples and 134 healthy serum samples were collected from The Second Hospital of Shandong University between February 2018 and October 2020. All enrolled patients met the following inclusion criteria: primary LUAD was confirmed by histopathology and showed no evidence of other neoplastic diseases in other organs, serum samples were collected before any antitumor therapies, such as surgery, chemotherapy or radiotherapy. The serum samples of controls were prepared from age-matched healthy volunteers. Clinical features of LUAD patients and healthy controls were described in Supplementary Table S1. The tumor stage and grade complied with the 8th Union of International Control of Cancer (UICC) classification.

Fifty LUAD serum samples were randomly selected, mix them thoroughly and divide them into five parts, each of which was regarded as an independent sequencing sample, serum samples of healthy people were also processed in the same way. The remaining samples (including 110 LUAD patients and 84 healthy controls) were selected to verify the expression of LUCAT1 in serum exosomes.

All experiments involved in this study were performed in accordance with relative regulations and manners. All participants have written informed consent and this study was approved by the Clinical Research Ethics Committee of The Second Hospital, Cheeloo College of Medicine, Shandong University [Approval number: KYLL-2019(LW)042].

Serum isolated from coagulation promoting vacuum tubes within 2 h was immediately transferred to a 1.5 ml Eppendorf tube and immediately centrifuged at 1500 g for 5 min and then 13800 g for 5 min at 4°C to eliminate cell sediments. ExoQuick™ solution (EXOQ5A-1; SBI System Biosciences, United States) was mixed evenly with supernatant according to the manufacturer's instructions. Subsequently, incubate at 4°C for 30 min, followed by centrifuged twice at 4°C (1500 g, 30 min and 1500 g, 5 min), supernatants were discarded, and the exosome pellets were resuspended in 250 μL PBS and stored at −80°C for further analysis.

The morphology of isolated exosomes was imaged using the transmission electron microscopy (TEM; G2 spititi FEI; Tecnai). And the size distribution and concentration of exosomes were quantified by ZETASIZER Nano series-Nano-ZS instrument (Malvern, United Kingdom). Total exosome protein was extracted with RIPA extraction reagent (Thermo Fisher, United States) according to manufacturer’s instructions. Western blotting analysis (WB) were performed to detect the exosomal surface markers: CD9, CD63, and TSG101.

Proteins from cell lysates were extracted with RIPA buffer (Sigma-Aldrich) according to manufacturer’s instructions, separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to Immobilon-P PVDF membrane (Millipore, United States). After being blocked with 5% non-fat milk in TBS-Tween, membranes were incubated with primary specific antibodies and horseradish peroxidase-conjugated secondary antibodies sequentially. Immunoreactive bands were detected using the Clarity Western ECL kit (Bio-Rad) and quantified using Image Lab (Bio-Rad, Hercules, CA, United States). All antibodies used in this study were shown in Supplementary Table S2.

High throughput sequencing service was provided by CloudSeq Biotech (Shanghai, China) and original datasheets could be obtained through access to GEO repository (GSE191209). Briefly, rRNAs were removed from total RNA using Ribo-Zero rRNA Removal Kits (Illumina, United States) according to the manufacturer’s instructions. Subsequently, rRNA-depleted RNAs were used to constructed RNA libraries with TruSeq Stranded Total RNA Library Prep Kit (Illumina, United States). BioAnalyzer 2100 system (Agilent Technologies, United States) was performed to control the quality and quantified. Libraries mentioned above were denatured as single-stranded DNA molecules, after captured on Illumina flow cells, amplified as clusters, and finally sequenced on Illumina HiSeq Sequencer for 150 cycles. The differentially expressed RNAs were selected with log₂ (fold change) > 1 or log₂ (fold change) <−1 and with statistical significance (p value <0.05) by R package–Deseq2.

Total exosomal RNA was extracted from miRNeasyMicro Kit (Qiagen) according to the manufacturer’s instructions. Concentration and integrity of exosomal RNA were measured by NanoDrop spectrophotometer (Thermo Fisher Scientific). Purified RNA was reversely transcribed into cDNA using the PrimeScript™ RT reagent kit (Takara, Dalian, Liaoning, China) according to manufacturer’s instructions.

The expression of lncRNA/mRNA was detected by CFX96™ Real-Time System (Bio-Rad Laboratories, Hercules, CA, United States) using Power SYBR Green (Takara, Dalian, China) and GAPDH was selected as the housekeeping gene. The expression of miRNA was measured using primer of miR-4316 and U6 was used as an internal control. All the primer sequences were available in Supplementary Table S3. The relative expressions of lncRNA/mRNA/miRNA were calculated using the 2^−ΔΔct method.

The expression level of LUCAT1 in tissues was detected by ISH using a specific digoxigenin-labeled LUCAT1 probe on TMAs, which contained 98 LUAD samples and 82 paired adjacent normal samples. All detailed clinical information (including age, gender, tumor–node–metastasis stage) of these LUAD patients were obtained from Qutdo Biotech (Shanghai, China), and shown in Supplementary Table S4. Further, the quantitative scanning approach was taken to analyse the staining and expression of LUCAT1 using a Nikon microscope. The results of ISH were calculated by multiplying the value of positive staining intensity by the proportion of positively stained cells. In our study, ISH score of <1.0 was regarded as LUCAT1 low-expression group, while ≥1.0 indicated LUCAT1 high-expression. All these experiments were approved by the ethics committee of Shanghai Outdo Biotech Company.

Immortalized lung epithelial cell line BEAS-2B and LUAD cell lines (A549, NCI-H1299) were obtained from the Chinese Academy of Sciences Cell Bank (Shanghai, China). Cell lines were cultured with 1640 (Gibco) supplemented with 10% fetal bovine serum (FBS, Australia Origin, Gibco). All cell lines were cultured in a humidified chamber with 5% CO₂ and 95% air at 37°C.

Si-LUCAT1, miR-4316 mimic, si-VEGFA, si-NC, and mimics-NC were obtained from GenePharma (Shanghai, China). LUCAT1 overexpression plasmids pcDNA3.1-LUCAT1, NC plasmids pcDNA3.1-NC were synthesized by Biosune (Shanghai, China). Lipofectamine™ 2000 was employed to accomplish the cell transfections, following guidelines from the manufacturer. Supplementary Table S5 depicts the sequences of si-RNA and mimics utilized in this study.

To establish stable LUCATI knockdown cells, a lentivirus vector encoding specific sh-LUCAT1 or sh-NC sequences was constructed by Vigenebio (Jinan, China), which could express the green fluorescent protein (GFP) and puromycin resistance gene. After incubated lentivirus vector with the LUAD cells for at least 48 h, fluorescence microscope and RT-qPCR were performed to verify the knockdown efficiency.

Cell migration ability was measured by the transwell chambers (8 μm pore size; Costar), and the ability of cell invasion was evaluated by the transwell chambers with Matrigel (BD Biosciences, San Jose, CA, United States) in the upper chamber. After incubation at 37°C, 5% CO₂ for 24 h, the lower chambers were fixed in methanol, stained with 0.1% crystal violet, and imaged by Inversion Microscope (Zeiss, Germany). Cell numbers for cell invasion in three random fields were counted.

For wound healing assays, cells were seeded into 6-well plates, and a 200 μL pipette tip was used to create an artificial scratch when incubated to 100% confluence. Subsequently, the cells were cultured in serum-free medium, wound closure images were captured in the same field at 0 and 24 h. The fraction of cell coverage across the line were used to calculate the cell healing rates.

For the in vivo tumor metastasis experiments, BALB/c nude mice were divided into two groups (n = 5) to establish the lung metastasis model. 1 × 10⁶ NCI-H1299 cells stably transfected with sh-LUCAT1 or shRNA-NC in 0.1 ml PBS were subcutaneously injected into the tail vein of the mice above. Six weeks after injection, the mice were euthanatized and dissected. Lungs of the mice in two groups were removed and photographed, metastatic nodules in lungs were counted by naked eye and investigators were blinded to the group allocation. The criteria used to identify metastatic nodules were as follows: whitish and uniformly colored, round, more than 1 mm in diameter, and the appearance quality was different from the surrounding normal tissue. After tissue were fixed in paraformaldehyde, embedded in paraffin, hematoxylin and eosin (HE) staining were performed to confirm the metastatic nodes. The animal studies were carried out in accordance with NIH Guidelines for the Care and Use of Laboratory Animals and approved by the Animal Care Committee of The Second Hospital, Cheeloo College of Medicine, Shandong University. The ethics approval number of Animal experiments was KYLL-2019(LW)043.

Cy3-labeled LUCAT1 and U6 sense probe were obtained from GenePharma (Shanghai, China). Fluorescence processed RNA FISH assay in Situ Hybridization Kit (GenePharma, Shanghai, China) were performed according to the manufacturer's instructions. Briefly, NCI-H1299 and A549 cells were washed, fixed and treated by 0.1% Triton X-100. After incubated with labelled FISH probe pre-mixed solution overnight in the dark, 4,6-diamidino-2-phenylindole (DAPI) (Solarbio, Beijing, China) was used to stain the cell nucleus for 10 min. Finally, fluorescence images were produced by a confocal microscope (Carl Zeiss Microscopy, LLC, United States). All the probe sequences used in this study were available in Supplementary Table S6.

For luciferase reporter assays, we constructed wild-type (WT) LUCAT1/VEGFA reporter plasmid in which the putative miR-4316-binding site in the LUCAT1sequence, and mutated-type (MUT) LUCAT1/VEGFA reporter plasmid in which the putative miR-4316-binding site in the LUCAT1sequence was mutated by base mutations with pmirGLO promoter vector. Luciferase reporter vectors were stably co-transfected with miRNA-4316 mimic or mimic-NC by Lipofectamine™ 2000 Transfection Reagent (Invitrogen, United States) according to the manufacturer’s guidelines. Luciferase intensity was measured using a dual-luciferase reagent (Promega) according to the manufacturer’s instructions. Renilla luciferase intensity was used as the control. Sequence of luciferase reporter associated with this article were listed in Supplementary Table S7.

A549 cells were seeded into six-well plates at a density of 1 × 10⁶ cells/well and cultured overnight. Glucose assay kits (Solarbio) and lactate assay kits (Solarbio) were used to detect the glucose consumption and lactate production in A549 cells, according to the manufacturer’s instructions, respectively. All data were normalized by the cell numbers.

A549 cells were cultured in RPMI-1640 medium (Gibco, Shanghai, China) supplemented with 10% fatal bovine serum (FBS) (Sagecreation, Beijing, China). After reaching 70–80% confluence, A549 cells were washed and then growth arrested in serum-free RPMI-1640 medium for 24 h to synchronize the cell growth. Then, d-glucose (Solarbio, Beijing, China) at a final concentration of 33mM, or 2-Deoxy-D-glucose (2-DG) (Solarbio) at a concentration of 0.5 mmol/L were added to the culture medium for an additional 48 h until further use.

The experimental data were statistically analysed using Graph Pad Prism 6.0 and SPSS 13.0 software. The mean values of two groups were assessed by Student’s t test or the Mann-Whitney U test as appropriate. Kaplan–Meier plots and log-rank tests were used for the survival analysis. The diagnostic performance of LUCAT1 was performed on MEDCALC 15.2.2 (Med-Calc, Belgium). Statistical significance was achieved when p values were less than 0.5. All results were expressed as the means ± standard deviation (SD) of three independent experiments.