Upregulation of long non-coding RNA LOC284454 may serve as a new serum diagnostic biomarker for head and neck cancers

We used blood collection tubes containing anticoagulants, mixed gently after blood collection. The samples were centrifuged at 1000–3000 rpm for 10 min, the supernatant was collected for RNA extraction. Blood samples were transported on ice and stored in − 80 °C refrigerator. Hemolysis and hyperlipidemia samples during blood collecting and low quality RNA during RNA extraction were excluded. Unbiased both men and women patients were included, who had not received any radio-chemotherapy or surgery before diagnosis. In total, 333 serum samples were collected from Affiliated Cancer Hospital of Central South University within 2017. This study was approved by the Ethical Committee of Central South University. Written informed consent was obtained from all patients and healthy donors.

The samples were collected from 121 normal donors randomly and 212 HNC patients. Of the 212 HNC serum, 100 were NPC, 55 were oral cancer, and 57 were thyroid cancer serum samples. Sex and age distribution were summarized in Supplemental table 1.

Serum RNA was extracted using miRNeasy Serum/Plasma Kit (Qiagen, Germany). Since our commen use housekeping genes may change its expression in tumor serum, thus we introduced an external reference, pGL3 [5]. The pGL3 (1 ng, approximately 2 × 10⁸ copies) was added to serum samples according to the manufacturer’s protocol using an miRNeasy Serum/Plasma Kit (Qiagen, Germany). The extracted serum RNA was reverse transcribed using a Revert Aid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). Forward (F) and reverse (R) primers were synthesized by TSINGKE Biological Technology Company (China), as follows: LOC284454-F, 5′-ATTACAGGTGGCTCAGGTGT-3′, LOC284454-R, 5′-CTTCAGTGTGCCTCCTCAGT-3′; and pGL3-F, 5′-TCCATCTTGCTCCAACACCC-3′, pGL3-R, 5′-TCGTCTTTCCGTGCTCCAAA-3′. The probe sequences were as follows: LOC284454-P, 5′-FAM-CGTGCCTGGCTTTTCTCCACTATCTTG-BHQ1–3′ and pGL3-P, 5′-HEX-ACGCAGGTGTCGCAGGTCTTCC-BHQ1–3′. Conventional SYBR-qPCR was performed using iTaq universal SYBR Green Supermix (Bio-Rad, USA). TaqMan-qPCR was performed using iTaq Universal Probes Supermix (Bio-Rad,USA). All RT-qPCR procedures were performed using a Bio-Rad CFX96 Multicolor Real-time PCR Detection System. TaqMan-qPCR allowed the simultaneous detection of two probes in the same tube (Bio-Rad, USA).

GSE61218 is from our group, which aims to identify significantly expressed lncRNAs in NPC tissues. GSE68799 is a RNA-Seq data identified human transcriptome alterations in NPC. RNA-Seq has been proved a tool with high throughput and coverage, reliable accuracy. GSE53819 is a genome-wide expressing profiling of NPC included 18 NPC tissue samples versus 18 control samples. They are paired tumor tissues and non-cancerous controls, which we thought can reduce individual heterogeneity. After discovering that LOC284454 is highly expressed in NPC, we also wanted to know whether it is highly expressed in other head and neck cancers, so we randomly selected the GEO dataset of oral cancer (GSE30784) and thyroid cancer (GSE33630). Data were analyzed using SPSS 13.0 (SPSS Inc., USA) and GraphPad Prism 7.0 (GraphPad, USA). Student’s t-tests were used to evaluate differences between two groups of samples. Normal distribution was analysed via Graphpad Prism 7, D’Agootino-Pearson, Kolmogoov-Smirnov, or Shapiro-Wilk were used to test whether the data conforms to the normal distribution, if p > 0.1, we can use Student t-test, if no, we may use Non-parametric Wilcoxon test. P-values< 0.05 were considered statistically significant. Correlation with clinic-pathological variables were evaluated through spearman or pearson correlation test. All the results obtained were from three independent replicates. The area under the curve (AUC), sensitivity, and specificity were obtained by receiver operating characteristic (ROC) curve analysis.

We explored the dysregulation of lncRNAs in HNCs using the GEO database. LOC284454 was significantly upregulated in several cancers, including NPC, oral cancer, and thyroid cancer. In our previous article, we performed gene expression profiling enrolled six inflammatory normal controls and 10 NPC tissues to identify differentially expressed lncRNAs (accession number GSE61218). Forty-six thousand five hundred six lncRNA probes were included. The data showed that totally 1276 lncRNAs were differentially expressed, including 405 upregulated and 871 downregulated lncRNAs in NPC tissues. We selected top 20 highly expressed lncRNAs to validate. LOC284454 was one of the most significant and had not been detected its application as a serum biomarker (heatmap showed in Fig. 1a) [11]. In NPC, we integrated three sets of gene expression profiles, including GSE53819, GSE68799, GSE61218, which further demonstrated that LOC284454 is highly expressed in NPC. GSE30784 and GSE33630 were used to analyze oral cancer and thyroid cancer, respectively. The expression levels of LOC284454 were significantly higher in NPC (Fig. 1a, P < 0.001), oral cancer (Fig. 1b, P < 0.001), and thyroid cancer (Fig. 1c, P < 0.001), compared to non-tumor tissues.

SYBR-qPCR was used to detect the expression of LOC284454 in the serum of 76 NPC patients and 51 healthy donors. LOC284454 expression level was significantly higher in the serum of patients with NPC (Fig. 2a, P < 0.001).

Next, to eliminate systematic errors and make the results more reliable, we designed a TaqMan probe for LOC284454 and tested the same serum samples using TaqMan-qPCR. This examination also revealed significantly higher expression of LOC284454 in the serum of the NPC patients (Fig. 2b, P < 0.001). This was consistent with previous conventional RT-qPCR results. Subsequent correlation analysis of the results obtained by SYBR-qPCR and TaqMan-qPCR demonstrated a good positive correlation between the two methods, which verified the reliability of this data (Fig. 2c, P < 0.001).

We also verified the expression of LOC284454 in a larger cohort of 121 normal controls and 100 NPC patients (added some new samples to the original cohort). The expression of LOC284454 in the serum of NPC patients was significantly higher than that of the normal control group (Fig. 2d, P < 0.001). Taken together, these results suggested that LOC284454 may be a potential serum marker for NPC.

The results of the GEO database results suggested that LOC284454 may be dysregulated in oral cancer and thyroid cancer. Therefore, we next detected the expression of LOC284454 in the serum of patients with these cancers using TaqMan-qPCR. LOC284454 was significantly upregulated in the serum of patients with oral cancer and thyroid cancer compared with those of the normal controls (Fig. 3a & b, P < 0.001). Notably, a significant difference was evident in the proportion of men and women with thyroid cancer (43 females and 14 males). To exclude gender effects, we analyzed the expression of LOC284454 in 43 female thyroid cancer patients and 36 normal women. The expression of LOC284454 was higher in the tumor serum than in the normal group (Fig. 3c, P = 0.024). The collective results demonstrated that the expression level of LOC284454 in the serum of patients with oral and thyroid cancers was significantly higher than normal controls.

ROC is commonly used to assess the diagnostic value of biomarkers. AUC refers to the area enclosed by the curve and the 45-degree diagonal line, which is used to quantify the diagnostic value. An AUC value < 0.5 indicates almost no diagnostic value. AUCs of 0.5 ~ 0.7, 0.7 ~ 0.9, and > 0.9 indicate low, moderate, and high diagnostic value, respectively. Values exceeding 0.9 indicate high specificity and sensitivity. When the sensitivity and specificity are the largest, we select this point as the best cut-off point. Diagnostic values of LOC284454 in these three kind of head and neck cancers are shown in Table 1. To improve the the quality of reporting diagnostic accuracy in this study, we followed the STARD statement.

The AUC values of LOC284454 in NPC (Fig. 4a), oral cancer (Fig. 4b), and thyroid cancer (Fig. 4c) were 0.931, 0.698, and 0.834, respectively, indicating that LOC284454 might be an appropriate diagnostic biomarkers for these cancers (Table 2). However, we analyzed the LOC284454 expression level as well as patients’ clinical characteristics and found that no correlation was observed between LOC284454 and pathological stages, or gender, or age distribution.