Colon cancer is the second leading cause of cancer-related death worldwide, with an incidence of 9.2% [1
]. Although curative surgical resection of primary lesions and post-operative adjuvant treatment contribute to the prevention of early recurrence in patients with high-risk stage II and III colon cancer, the recurrence rate after curative surgery remains at 20%–30% [2
]. To prolong disease-free survival (DFS) after curative surgery and avoid unnecessary over-treatment of patients, more accurate prognostic tools are required. However, a few clinical prognostic factors, such as the presence of obstructive symptoms, number of retrieved regional lymph nodes, pathologic tumor and nodal stage, and features associated with micrometastatic potential including vascular, lymphatic, and perineural invasion, have been used to select patients that may benefit from adjuvant therapies [3
]. Although other molecular signatures, such as microsatellite status, BRAF (v-Raf murine sarcoma viral oncogene homolog B) and KRAS (V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog) mutations, molecular signatures, and immune infiltration, have been suggested as potential prognosticators, more sophisticated prognostic biomarkers need to be identified [1
To identify accurate and definitive molecular biomarkers for human malignancies, we previously developed a neural network-based model, Wx, which is optimized to select gene sets based on the discriminative index between combination pairs of K class [4
]. In our previous report, we applied our Wx model to a pan-cancer cohort from The Cancer Genome Atlas (TCGA) RNA-Seq dataset, consisting of 12 different types of cancer and normal samples, and obtained 14 genes distinguishing cancer tissues from normal tissues with high accuracy. Tissue elongation factor 1 alpha 1 (EEF1A1
) was one of the 14 biomarker genes extracted using the Wx model. Interestingly, it was also identified by Martinez-Ledesma et al. using the same discriminative concept to compare colon adenocarcinoma to normal tissues [5
EEF1A1 is a ubiquitous protein that functions in peptide elongation during mRNA translation and is the second-most abundant protein after actin in eukaryotic proteomes [6
]. As a translational factor, EEF1A delivers the aminoacylated-tRNA to the ribosome, as a ternary complex comprising EEF1A–GTP (Guanosine-5’-triphosphate)–aminoacylated-tRNA, to decode mRNA. There are two isoforms of EEF1A
, namely EEF1A1
, which were mapped to chromosomes 6q14 and 2q13.3, respectively [8
share more than 95% DNA and protein identity [9
]. EEF1A1 is expressed in most cells, whereas EEF1A2 is present only in the brain, heart, and skeletal muscle [10
]. Although the functional significance of their tissue-specific expression patterns is unknown, they are thought to have the same enzymatic function in protein translation. Several studies have revealed that EEF1A is not only a translation factor but also involved in many non-canonical functions including oncogenesis, protein degradation, pro-apoptotic or anti-apoptotic activity, and cytoskeleton modulation [10
]. Notably, many studies have shown that EEF1A
is a prognostic factor for several solid tumors such as ovary [11
], breast [15
], lung [18
], pancreas [19
], stomach [21
], prostate [23
], and liver cancers [24
]. Based on the previous results from the Wx neural network-based feature selection algorithm and those reporting the role of EEF1A1 in human solid cancer, we hypothesized that EEF1A1 is related to the prognosis of patients with colon adenocarcinoma. In this study, we investigated the expression of EEF1A in tissues from patients with stage II and III colon cancer and analyzed its association with patient prognosis.
2.1. Identification of Prognostic Biomarker Genes Using the Wx Algorithm with TCGA Database
Genes distinguishing cancer from normal samples were identified by applying the Wx algorithm to a pan-cancer cohort containing 6210 samples with 12 different types of cancer, using mRNA-Seq data from TCGA. In this study, we re-analyzed the mRNA-Seq data of 327 colon adenocarcinomas (287 cancer and 40 normal samples) to identify biomarker candidate genes using the Wx algorithm. The Wx algorithm ranks genes based on the discriminative index score, which reflects the classification power of distinction between groups (e.g., cancer vs. normal). The detailed method has been described previously [4
2.2. Patients and Tissue Samples
Medical records of patients with colon cancer who have undergone curative surgery at Incheon St. Mary’s hospital between 2010 and 2013 were reviewed. Their tissue microarrays (TMAs) for immunohistochemistry were obtained. If available, fresh-frozen tumor tissues and paired normal adjacent tissues from patients were used for RNA extraction. Demographic and clinicopathological data for these patients were reviewed retrospectively from the medical records. Variable factors including age, sex, sidedness of colon cancer, pathologic staging, histology, and lymphatic, venous, and perineural invasion were analyzed, and tumors were staged according to the pathological tumor/node/metastasis (pTNM) classification (8th edition) of the Union for International Cancer Control. The study was approved by the Institutional Review Board of Incheon St. Mary’s hospital, the Catholic University of Korea (OC15TISI0050). Informed consent was waived considering the retrospective study design.
2.3. Quantitative Reverse Transcription PCR (qRT-PCR)
Total RNA was isolated from tumors and adjacent normal tissues of patients with colon cancer using the WelPrepTM Total RNA Isolation Reagent (Welgene, Daegu, Korea) and gentleMACS Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany), according to the manufacturer’s protocols. To analyze mRNA levels, qRT-PCR assays were performed using a BioFACTTM A-Star Real-time PCR Kit including SFCgreen® I (BioFACT, Daejeon, Korea) after reverse transcription with ELPIS RT Prime Kit (Elpis-Biotech, Daejeon, Korea). mRNA levels were normalized to those of ribosomal protein L32 (RPL32).
2.4. Tissue Microarray Construction and Immunohistochemistry
Formalin-fixed paraffin-embedded colon cancer samples from patients who underwent surgery were included in a TMA after careful review by two experienced pathologists. TMAs were prepared with a tissue array and were composed of 69 cores per TMA, with a core diameter of 2 mm. The cores were derived from separate sources and treated as individual cases, and not as duplicates from a single source. Cases that presented as colon cancer had sufficient material for two cores, whereas normal tissues could be used to generate one core. A set comprising 14 slides was stained by immunohistochemistry. The antibodies used for the study included a mouse monoclonal antibody against EEF1A1 (sc-21758) from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA), which was used according to the manufacturer’s instructions at a dilution of 1:100 after optimization. Immunohistochemical staining was carried out with the Ventana BenchMark ULTRA (Tucson, AZ, USA) using the ultraView Universal DAB detection kit. This included standard de-paraffinization, blocking of endogenous peroxidase and biotin, incubation with primary and secondary antibodies, and development with DAB (di-aminobenzidine). The slides were counterstained with hematoxylin and eosin (H&E).
2.5. Evaluation of Immunohistochemical Staining of EEF1A1
Immunohistochemical staining for EEF1A1 revealed its cytoplasmic expression in most tumor cells. Staining intensity was graded from 0 to 3 (0 = no expression, 1 = weak, 2 = moderate, and 3 = strong). The extent was graded from 0 to 2 (0 = 0%, 1 = 1%–50%, 2 = 51%–100% of cells) (Figure 1
). The intensity and extent scores were added to obtain a composite score. Patients with composite scores of 0–2 were classified into the low-EEF1A1 expression group and those with scores of 3–6 were included in the high-expression group. The results were analyzed by two pathologists who were blinded to the clinical data.
2.6. Statistical Analysis
Categorical values were presented as means and percentages, and non-categorical values were presented as median values and ranges. The Chi-squared test and one-way analysis of variance were performed to compare differences among means and medians, respectively. The Kaplan–Meier method was used to calculate survival rates. DFS was defined as the period from the date of curative operation of colon cancer to the date of recurrence based on medical records. Overall survival (OS) was defined as the period from the date of curative operation of colon cancer to the date of death due to any cause. Right colon cancer was defined as cancer originating from the ascending colon to the splenic flexure and left colon cancer was defined as originating from the descending colon to the sigmoid colon. A p-value < 0.05 was considered to indicate statistical significance. Differences in survival rates across each patient group were examined by the log-rank test. A Cox proportional-hazards model was used for univariate and multivariate analyses. All analyses were performed with SPSS for windows (version 25.0, SPSS, Inc., Chicago, IL, USA).
EEF1A1 was a biomarker used to distinguish between cancer and normal tissue in our previous study, using a neural network-based feature selection algorithm. In the present study, we investigated EEF1A1 expression and its prognostic significance for patients with stage II and III colon cancer who underwent curative resection. The results revealed that EEF1A1 mRNA was decreased in tumor tissue compared to that in normal tissue. Patients with low expression of EEF1A1 showed poorer survival outcomes in terms of both DFS and OS compared to those with higher expression. The expression level of EEF1A1 was found to be an independent prognostic factor for survival outcomes. Among stage II patients not administered adjuvant treatment, a poorer OS outcome was observed for those with low EEF1A1 expression than for those with high EEF1A1 expression. Among patients with stage III colon cancer, those with low EEF1A1 expression had a shorter OS than those with high EEF1A1 expression. We determined the prognostic effect of EEF1A1 on human colon cancer. To the best of our knowledge, this is the first study to investigate the prognostic value of EEF1A1 expression in human colon cancer.
The physiological role of EEF1A1 in eukaryotic protein synthesis involves recruiting amino-acylated tRNA to the ribosome during the elongation phase of translation. It also plays various non-canonical roles including cytoskeleton modulation, protein degradation, apoptosis, and oncogenesis [12
]. Particularly, EEF1A1 has been reported to be a pro-apoptotic protein in previous studies. Ruest et al. reported that myotube death is accelerated by the induction of the homologous gene [26
Borradaile et al. demonstrated that induction of EEF1A1
is involved in lipotoxic cell death secondary to oxidative stress [27
]. Bosutti et al. compared the mRNA levels of EEF1A1
, and other pro-apoptotic genes (p66
) in skeletal muscles between severely traumatized patients and healthy volunteers [28
]; the results showed upregulation of muscle EEF1A1
in the traumatized patient group, which was significantly related to the proteolysis rate. Thus, pro-apoptotic function is a possible mechanism underlying the protective effect of EEF1A1 against tumor development or progression. However, the tumor suppressive role of EEF1A1 in colon adenocarcinoma remains unclear. EEF1A1 expression has been determined to be independent of KRAS
mutations in our study, suggesting that the expression of EEF1A1 is unlikely to involve KRAS
-related colon adenocarcinoma. However, one study suggested that posttranslational changes in EEF1A1 might affect the KRAS pathway. Liu et al. suggested an association between EEF1A1 and KRAS
-related tumorigenicity [29
], demonstrating that neoplastic cell growth induced by oncogenic KRAS signaling may be associated with methylation of EEF1A1
at the mRNA or protein level. Additionally, the results of the current study revealed that EEF1A1 expression was associated with tumor location, and right-sided colon cancers tended to show lower expression of EEF1A1, although the correlation with tumor sidedness was not significant. The biological role of EEF1A1 in colon adenocarcinoma and its correlation with tumor sidedness will be evaluated in our next study.
Based on our results, high expression of EEF1A1 indicates a favorable survival prognosis for patients with colon cancer, and at the mRNA level, EEF1A1
was found to be downregulated in tumor tissues compared to that in normal tissues. Lin et al. revealed that EEF1A1 protein is overexpressed in estrogen/progesterone receptor (ER/PR)-positive and lymph node-negative breast cancers, which have relatively favorable outcomes compared to ER/PR-negative and lymph node-positive cases [30
]. However, at the mRNA transcript level, low EEF1A1
expression was an independent poor-prognostic factor with ER/PR-positive tumors; breast cancer-specific post-transcriptional regulatory events may be responsible for this discordance between the expression of EEF1A1
mRNA and protein [30
]. Further, Hassan et al. investigated the mRNA transcript levels of EEF1A
isoforms across different cancers such as lung, breast, gastric, brain, prostate, liver, and colon cancers using the Oncomine and TCGA databases [31
]. In colon cancer, EEF1A1
mRNA expression levels were significantly low in patients with a high risk of recurrence compared to those in patients with a low risk, which was also consistent with our findings. Interestingly, based on our results, patients who did not undergo treatment for stage II colon cancer showed differences in OS according to EEF1A1 expression. In patients with stage II disease, EEF1A1 could serve as a potential biomarker for selecting candidates who may benefit from adjuvant treatment.
In contrast to our results, studies have reported that EEF1A
downregulation leads to decreased expression of pAkt1, consequently inhibiting proliferation and invasion and promoting apoptosis in HCC (Hepato cellular carcinoma) cells [16
]. In stomach cancer, EEF1A1 has been reported to promote gastric cancer cell migration and invasion [21
]. Such contradictory reports regarding the role of EEF1A1 in tumorigenesis may be due to the fact that it plays fundamental roles in various cellular functions [12
]. Even in the same cancer, EEF1A1 may act in an opposing manner with respect to different subtypes or stages. For example, Hassan et al. suggested that higher EEF1A1
], which is otherwise a favorable biomarker for breast cancer, can be an indicator of poor OS, particularly in the basal subtype.
Our study has some limitations, such as its retrospective design. Further, the KRAS status was not verified in all populations. However, we analyzed a relatively homogeneous population, which compensates for these limitations. To the best of our knowledge, this is the first study to analyze the prognostic role of EEF1A1 in human colon cancer.