Expression of Demethylase Genes, FTO and ALKBH1, Is Associated with Prognosis of Gastric Cancer

Predicting Prognosis of GC by KM Plotter

KM plotter database includes 1065 GC samples with follow-up data. The median overall survival (OS) was 28.9 months, and median progression-free survival (PFS) was 18.3 months [23]. Unprocessed CEL files were MAS5 normalized in the R statistical environment (http://www.r-project.org) using the Affy Bioconductor library, and only arrays passing quality criteria were utilized. Gene expression and clinical data were integrated using PostgreSQL. The KM curve indicated the association between the candidate gene and survival, in which the samples are grouped by the median expression of each gene. KM survival plots with hazard ratio (HR), 95% confidence interval (CI), and log-rank p were obtained online.

Predicting Prognosis of GC by TCGA Database

The mRNA sequencing level 2 data (preprocessed and normalized) and clinical patient data of 415 GC cases were downloaded from TCGA database (http://download.cbioportal.org/stad_tcga_pub.tar.gz?raw=ture). The cutoffs dividing high and low expression groups were decided via receiver operating characteristic curve (ROC) analysis by SPSS Statistics 23.0 (IBM Corp., Armonk, NY, USA).

Tissue Samples

For tissue microarray (TMA) detection, 450 cases of gastric tumor and matched non-tumor tissues were collected in the department of pathology of Sun Yat-Sen University Cancer Center. To be specific, two different pieces of GC tissues and one piece of non-tumor tissue, taken at a distance of > 3 cm from the tumor margin, were collected from each individual, paraffin-embedded blocks, and arranged in recipient ones. These patients had not received any preoperative anticancer therapy. The clinicopathological features of patients are provided (Table 1). Tumor differentiation grades and clinical stages were reclassified according to the 8th American Joint Committee on Cancer (AJCC) TNM classification. The OS was calculated as the duration from the date of surgery to the date of death or last follow-up. The use of human tissue samples and clinical data was approved by the Ethics Committee of Sun Yat-Sen University Cancer Center, all patients provided signed informed consent, and the research was carried out in accordance with the Helsinki Declaration.

Immunohistochemistry (IHC) Staining and Evaluation

IHC staining was performed according to a previously described protocol [24]. Tissue sections were deparaffinized with dimethylbenzene and rehydrated via a graded alcohol series. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 15 min. Slides were boiled in tris(hydroxymethyl) aminomethane–EDTA buffer (pH 8.0) in a microwave for 30 min to retrieve antigen. Nonspecific antigens were blocked with 10% normal goat serum for 20 min. Next, the slides were incubated with monoclonal rabbit anti-FTO antibody (ab124892, 1:100 dilution, Abcam, Cambridge, UK) and anti-ALKBH1 antibody (ab126596, 1:50 dilution, Abcam) overnight at 4 °C in a moist chamber. Slides were incubated with horseradish peroxidase (DAKO ChemMate™ EnVision™ Detection Kit, Copenhagen, Denmark) for 30 min at 37 °C and subsequently incubated with 3,3′ diaminobenzidine (DAB) solution for visualization.

Two independent observers blinded to clinicopathological data conducted the immunoreactivity score (IRS) assessment for FTO and ALKBH1 expression. Scoring criteria for staining intensity were as follows: 0 (negative), 1 (weak), 2 (moderate), 3 (strong). The staining extent score was as follows: 0 (< 10%), 1 (11–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%). The final expression score, ranging from 0 to 12, was calculated as intensity score × extent score. Total scores of 6 or higher were categorized as the high expression group, and the rest as the low expression group. The staining result for each case was the average IRS determined by two pathologists. The specimens would be rescored if the difference between the two pathologists was greater than 3.

Statistics

The Chi-squared test was used to analyze the association between candidate gene expression and clinicopathological characteristics of GC patients. Survival curves were evaluated using the Kaplan–Meier method, and differences between survival curves were tested by the log-rank test. Cox proportional hazards regression model was used to examine univariate and multivariate analyses. Only significantly different variables in univariate analysis were included in the multivariate analysis. Statistical significance was based on two-tailed tests at p < 0.05. SPSS 23.0 (IBM Corp.) and GraphPad Prism 6 (San Diego, CA, USA) software were used for statistical analyses and graphical representations.

Prognostic Values of m⁶A-Related Genes in KM Plotter Cohort

First, nine candidate genes were analyzed using the KM plotter database. The clinicopathological information of GC patients in this cohort was summarized previously (Table S1). The survival curves revealed that high mRNA expression of METTL3, METTL14, FTO, ALKBH1, and ALKBH5 and low mRNA expression of WTAP, YTHDF2, and YTHDF3 were associated with worse OS (Fig. 2) and worse PFS (Fig. 3), respectively. The mRNA expression of YTHDF1 seemed to have little effect on the prognosis of GC patients.

Prognostic Values of m⁶A-Related Genes in TCGA Cohort

Next, we tested the associations between mRNA expression of the above genes and prognosis of GC patients in TCGA database to select even more reliable targets. The characteristics of GC patients in TCGA cohort are described in Table S2. The results suggested that higher mRNA levels of FTO and ALKBH1 lead to worse OS (Fig. 4a, b), while higher mRNA level of FTO, but not ALKBH1, was correlated with worse PFS (Fig. 4c, d), indicating that FTO and ALKBH1 were more likely to exert a significant influence on progression of GC compared to the other m⁶A-related genes. This also indicated that T stage, N stage, M stage, TNM stage, differentiation grade, FTO, and ALKBH1 expression had significant impact on OS in univariate analysis, while only FTO and ALKBH1 expression levels remained observably significant in multivariate analysis (Table S3). Patients diagnosed with stage III or IV had observably shorter OS than those with stage I or II (22.0 vs. 59.5 months, Fig. 4e). In addition, patients showing high expression levels of both FTO and ALKBH1 (OS: 19.9 vs. 72.2 months) and patients with high expression level of one gene (OS: 26.5 vs. 72.2 months) had observably shorter OS than patients with low expression levels of both genes (Fig. 4f).

Independent Validation of Prognostic Value of FTO and ALKBH1

Considering the findings of the in silico analysis, we validated the prognostic value of protein expression of demethylase genes FTO and ALKBH1 using TMA-based IHC to detect 450 paired GC and corresponding adjacent non-tumor tissues. Different staining intensities of FTO and ALKBH1 are displayed (Fig. 5a, b). Interestingly, as opposed to their mRNA levels, both FTO and ALKBH1 protein levels were significantly downregulated in signet ring cells compared to the adjacent normal gland cells (Fig. 5c, d). The proportion of strong and moderate staining was significantly lower in GC tissues than in adjacent non-tumor tissues for both FTO (p < 0.0001) and ALKBH1 (p < 0.0001) (Fig. 6a, b). Consistently, the IRSs of FTO and ALKBH1 were markedly lower in GC tissues than in adjacent non-tumor tissues (Fig. 6c, d), and lower protein expression of FTO, but not of ALKBH1, was closely associated with a shorter OS of GC patients (HR = 0.74, 95% CI: 0.56–0.98, and p = 0.0364) (Fig. 6e, f). In addition, the lower protein level of ALKBH1 was closely correlated with larger tumor sizes (≥ 5 cm, p < 0.001) and more advanced TNM stages (III/IV, p = 0.033) (Table 1). In the TMA-IHC cohort, T stage, N stage, M stage, TNM stage and the protein expression of FTO, but not ALKBH1, were correlated with OS in univariate analysis, while only the M stage, TNM stage, and FTO expression were significant in multivariate analysis, indicating that FTO protein expression was as valid as M stage and TNM stage in prognosis prediction in GC patients (Table 2).

Below is the link to the electronic supplementary material.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

The use of human tissue samples and clinical data was approved by the Ethics Committee of Sun Yat-Sen University Cancer Center (Guangzhou, China), all patients provided signed informed consent, and the research was carried out in accordance with the Helsinki Declaration. This article does not contain any studies with animals performed by any of the authors.