Colorectal cancer (CRC) is the third most common cause of cancer-related deaths worldwide (Siegel et al., 2016). Its incidence has been increasing rapidly in some areas of the world, including Asia (Jemal et al., 2011). In Malaysia, CRC has become the second most common cancer in both men and women where men were slightly more affected than women [1.1:1] (Ferlay et al., 2012). According to Lim (2014), the cumulative lifetime risk of developing CRC for Malaysians was 1:38 in men and 1:50 in women. The Malaysian Chinese has twice higher risk compared to Indian and Malay (Lim, 2014).

Survival rate of CRC is highly dependent on the stage of disease at diagnosis (Jemal et al., 2011). The 5-year survival rate for early stage is 60–95% but this dramatically decreases to 35% for those with lymph nodes involvement, indicating that early screening and treatment are crucial to improve management of CRC patients (Sengupta et al., 2008). Dukes’ staging (Dukes, 1932) and TNM classification (Version 7) (Edge et al., 2009) are the two staging systems used in CRC worldwide. Until today, CRC staging is based solely on simple clinicopathological features such as bowel wall penetration and lymph node metastasis. Current standard therapies for CRC patients depend on the stages. For instance, radical surgery followed by adjuvant chemotherapy is the standard practice for Dukes’ C patients (Hagan et al., 2013). However, the treatment remains controversial for Dukes’ B patients because the toxicities may outweigh its benefits (Benson et al., 2004). Nevertheless, due to heterogeneity nature of cancers, clinical outcome of patients with the same histo-clinical staging could also differ.

Generally, adjuvant treatment for CRC consists of 5 fluorouracil (5-FU) and Oxaliplatin (Piedbois et al., 1992; de Gramont et al., 2000; Saltz et al., 2000; Fuchs et al., 2008). A modern choice of the treatment has emerged by using targeted personalized therapy where the molecular profiling of each individual tumor is needed (Gray et al., 2011). Due to the heterogenous nature of CRCs, identification of biomarkers for targeted therapies is now highly recommended as the genetic and epigenetic alterations have been demonstrated to correlate with prognosis and treatment responses (Silvestri et al., 2013). For instance, CRC tumors with KRAS alterations in codons 12 and 13 are irresponsive to regularly used drugs targeting EGFR, and tumors with chromosome instability frequently found in majority of CRCs are often associated with multi-drug resistance (Lee et al., 2011).

Identification of somatic alterations is one of the key approaches to understand the molecular mechanisms of CRC and for the development of novel therapeutics. Deep sequencing via NGS technology is revolutionizing our understanding of somatic changes occurring in the cancer genome. In this study, we utilized the 409-gene targeted sequencing assay, the Ampliseq Comprehensive Cancer Panel v2 (CCPv2), performed on the Ion Torrent PGM, to identify actionable somatic alterations in non-metastatic CRC.

In this study, we screened for somatic alterations in 409 genes in CRC patients with Duke’s B and Duke’s C stage using the Ion Ampliseq Comprehensive Cancer Panel on the Ion Torrent PGM. We identified several cancer driver genes and illustrate their involvement in the CRC-related pathway. Overall, 63% of CRC patient’s harbored somatic alterations in the APC gene which included 13 alterations altogether and the majority were located in exon 16. To the best of our knowledge, at the time of analysis, four from the 13 alterations have not been reported in dbSNP version 138, in CRC or other cancers.

The APC gene is located on the long arm of chromosome 5 (5q22.2), consists of 16 exons (open reading frame from exon 2 to exon 16) and encodes a large protein ∼310 kD made of 2843 amino acids (Miki et al., 1992; Pineda et al., 2010). It is an important tumor suppressor known to play important roles in both hereditary and sporadic CRCs. Germline APC alterations are the main cause of familial adenomatous polyposis (FAP) which confers 100% lifetime-risk of CRC if left untreated (Groden et al., 1991). On the other hand, somatic alterations in this gene are detected in ∼70% of sporadic CRCs (Fearnhead et al., 2001; Narayan and Roy, 2003; Jaiswal and Narayan, 2008; Fearon, 2011) which is in concordance with our findings. APC alterations almost always result in a truncated protein product with abnormal function (Fearnhead et al., 2001). The truncations of the APC protein lead to the loss of β-catenin and/or axin binding sites and prevent β-catenin degradation, resulting in abnormally high levels of cytoplasmic and nuclear β-catenin in colon tumor cells (Munemitsu et al., 1995; Faux et al., 2004). In the nucleus, β-catenin interacts with members of the T-cell factor (Tcf)/lymphoid enhancer factor (LEF) family of transcription factors to activate transcription of Wnt target genes, including cyclin D1 and myc (Korinek et al., 1997) which promote proliferation and are associated with cellular transformation. Truncating APC alterations were detected in 68–95% of CRCs (Sieber et al., 2000; Christie et al., 2013). In our cases, truncating APC alterations were detected in eight out of 12 (66.7%) CRC patients with APC mutants (APC-mt). Currently, various treatments for APC mutant patients are being explored using multiple therapeutic approaches targeting both the Wnt-dependent and Wnt-independent functions of APC including reintroduction of APC, targeting the Wnt pathway downstream of the destruction complex and inhibiting targeted pathways regulated by APC. However, identifying APC alterations is pivotal for effective treatments, as loss of APC may cause resistance to certain drugs like cisplatin through enhanced DNA repair (Lesko et al., 2014).

Our study revealed APC as a significant driver gene in our patients and since the functional roles of the majority of APC variants have not been well investigated, we further performed functional analyses on two selected variants in this gene. Both were predicted by in silico algorithms as protein truncating variants which produce protein at different sizes and contain distinct domains which were APC p.R805^∗ and APC p.Q1738^∗. Western Blot analysis further confirmed the in silico prediction that these variants indeed produced truncated APC proteins. From the in vitro functional analysis, insertion of mutant APC p.R805^∗ led to significant increase of the viability of cancer cells. Truncating APC p.R805^∗ lacks all domains involved in regulation of WNT/β-catenin signaling. When this happens, it will activate the other Tcf members like cyclin D1 and myc, therefore causing uncontrolled of cell proliferation (Aoki and Taketo, 2007), as evident in our study. Meanwhile, the variant APC p.Q1738^∗ showed a similar pattern although not significant. This might be because of the truncated site of this variant which has two out of three 15-residue repeat domains that still cover the important function for binding to β-catenin (Wang et al., 2014). In addition, based on the clonogenic assay, we also demonstrated the ability of both variants in driving in vitro cell transformation. The pathogenicity of APC variants on different protein domain might have different effect and is worth investigating for future studies.

We also identified alterations in other known CRC genes such as TP53, KRAS, FBXW7, and SMAD4, as reported by others (Cancer Genome Atlas Network, 2012; Han et al., 2013; Yu et al., 2014; Bai et al., 2015). TP53 gene, which is located at the long q arm of chromosome 17 at position 13, is altered in about 60–70% of CRCs and is important in cell cycle arrest and apoptosis processes (Fearon, 2011). Consistent with our findings, most TP53 alterations were clustered in the TP53 DNA binding domain, which encompasses exon 5 through 8 and also a critical region for the transcription factor functions of the gene (Pfeifer and Besaratinia, 2009; Cai Z.X. et al., 2014). Majority of TP53 alterations identified in our study were missense alterations and this is supported by other CRC studies (Cai Z.X. et al., 2014; Dallol et al., 2016). Most TP53 missense alterations lead to the substitution of a single amino acid in the P53 protein that can be stably expressed in cancer cells (Muller and Vousden, 2014). These alterations generally lead to a loss or diminution of the wild type activity of TP53 and mutant proteins become inactive and lose the ability to transactivate downstream target genes that regulate cell cycle and apoptosis (Muller and Vousden, 2014).

Inactivation of TP53 gene is one of the most common events in CRC and plays a vital role in the tumourigenesis of colorectal epithelial cells. TP53 alterations are identified as late events in CRC development, with a loss of TP53-mediated apoptotic pathways as an important factor in the progression from an adenoma to a malignant tumor (Smith et al., 2002). The frequency and spectrum of TP53 alterations were associated with the different grades, stages and locations of the tumor (Mahdavinia et al., 2008). A study by Mahdavinia et al. (2008) found that the frequency of TP53 mutations significantly increased with tumor stages (36/94, 38.3%; 33/64, 51.5%, and 15/23, 65.2% in Stage B, C, and D, respectively) and this is also reflected in our finding. Further corroborating our finding, TP53 alteration also exhibit location-based pattern, whereby its frequency is higher in distal as compared to proximal tumors (Samowitz et al., 2002; Russo et al., 2005) (44.7 vs. 55.3% and 35% vs. 45%, respectively). Researchers are still debating the role of TP53 alterations as prognostic indicators, as some have reported that overexpression of TP53 with elevated serum carcinoembryonic antigen (CEA) or carbohydrate antigen 199 (CA199) levels is significantly associated with poor outcome [median time from progression to death (TTD) for CEA ≥ 5 U/ml and P53 positive was 23.2 months (95% CI 17.1–29.3), and that of CA199 ≥ 27 U/ml and P53 positive was 15.2 months (95% CI 1.9–28.5)] (Cui et al., 2016), whereas others have found TP53 mutations to have little or no prognostic value at all (Soong et al., 2000; Conlin et al., 2005).

SYNE1 gene is one of the biggest genes in the human genome, containing 147 exons that encode a 27,652 kb messenger RNA and 8,797 amino-acid-long protein (Gros-Louis et al., 2007). It is located at the long arm of chromosome 6 at position 25. This gene encodes a spectrin repeat containing the nuclear envelope 1 protein expressed in skeletal and smooth muscle and peripheral blood lymphocytes that localizes to the nuclear membrane (Lee et al., 2013). SYNE1 alterations are linked to cerebellar ataxia and have been associated with lung, ovarian, and CRCs (Doherty et al., 2010). With regards to epigenetics, hypermethylation of SYNE1 has been identified as a key factor in colitis-associated cancer (CAC) carcinogenesis and is a potential biomarker to identify patients at higher risk of CRC (Papadia et al., 2014). Long-standing ulcerative colitis (UC) and Crohn’s colitis are linked to an increased risk of developing CRC (Ekbom et al., 1990; Papadia et al., 2014). Melotte et al. (2015) reported methylated SYNE gene as one of the promising markers for CRC detection. SYNE1 is one of the frequently altered genes in our CRC patients (7/19) and this is concordant with a study by Yu et al. (2014) which reported 28/160 (17.5%) cases with alterations. SYNE1 also have been reported in glioblastoma (GBM) where these alterations were significantly correlated with the overexpression of several known GBM survival genes (Masica and Karchin, 2011) and the polymorphism of SYNE1 (rs2295190, G-to-T change) was associated with an increased risk of invasive ovarian cancer, with a per-T-allele odds ratio (OR) of 1.24 [95% confidence interval (CI), 1.06–1.44, P = 0.006] (Doherty et al., 2010).

The TCGA project showed that the KRAS and FBXW7 alterations rates in CRCs were 42 and 17%, respectively (Cancer Genome Atlas Network, 2012). KRAS gene is located on the short arm of chromosome 12 and encodes a 21-kD protein (Smith et al., 2002). Alterations in this gene lead to increased and unregulated cellular proliferation and malignant transformation (Smith et al., 2002). Even though our sample set had a lower frequency of KRAS alterations than expected (26%), we found that this gene had the commonest oncogenic codon 12 alterations affecting glycine 12 residue (two alterations of p.G12V; one alteration of p.G12D) as similarly reported with by other studies (Cancer Genome Atlas Network, 2012; Cai Z.X. et al., 2014; Dallol et al., 2016). KRAS p.G12V was reported as a biomarker for poor prognosis in resected non-small cell lung cancer (NSCLC) as it exhibited a worse overall survival (OS) and higher recurrence rates (Renaud et al., 2015). In CRC, KRAS p.G12V was identified in both primary and metastasis samples but the difference was not significant (Neumann et al., 2009). However, a recent clinical study found that the KRAS p.G12V alteration enhances metastases to lymph nodes, an indication of its higher aggressiveness in CRC animal model (Alamo et al., 2015). With regards to precision medicine, this alteration is reported to confer reduced sensitivity against cetuximab or panitumumab (anti-EGFR antibodies) (Di Fiore et al., 2007; Peeters et al., 2013), which is the main treatment for metastatic CRC. Hence, both agents should only be introduced into patients with RAS wild-type CRC. Alterations in other positions, such as codon 61 (p.Q61H) and 146 (p.A146T) have also been reported. However, these alterations account for a minor proportion of one to four percent of KRAS alterations, and their clinical relevance in CRC still remains unclear (Neumann et al., 2009; Tong et al., 2014).

FBXW7 is a potential tumor suppressor that regulates ubiquitination and proteolysis of multiple targets (Cai Z.X. et al., 2014). The gene is located on chromosome 4 on the short arm at position 31 and encodes for a subunit of an ubiquitin protein ligase that regulates the levels of many important oncoproteins such as Cyclin E, Notch 1, c-Myc and other proteins (Wang et al., 2012; Zheng et al., 2016). FBXW7 protein domain structure consists of seven WD4 repeats, which can form eight-bladed, barrel-shaped β-propeller-like binding pockets for its substrates (Orlicky et al., 2003; Hao et al., 2007). FBXW7 alterations impair Cyclin E degradation and are associated with decreased genetic stability and impaired growth regulation, contributing to the progression of CRC (Akhoondi et al., 2007). Low FBXW7 expressions in tumor as compared to normal tissues were significantly correlated with poor prognosis of overall survival in CRCs and esophageal cancer (EC) patients (Iwatsuki et al., 2010; Kurashige et al., 2012). A similar pattern of low overall survival was also found in early stage CRC patients with FBXW7 mutants but the results was not significant when compare to wild type patients (Chang et al., 2015). About 10 to 11% of FBWX7 alterations were found in most of CRCs studies (Eliana et al., 2013; Grim, 2014; Cai Z.X. et al., 2014; Bai et al., 2015). We found this alteration in 16% (3/19) of our non-metastatic CRCs patients.

We also observed several co-occurrence of alterations among the genes including APC, TP53, KRAS, and FBXW7. The most common combination of alterations in both APC and TP53 in our patients are in concordance with a previous study which involved European CRC patients (Smith et al., 2002) but contradicted with a report in Chinese CRCs (Cai Z.X. et al., 2014). Cai Z.X. et al. (2014) identified only four out of 93 CRC patients (4.3%) harbored APC and TP53 alterations. In addition, alterations in TP53 with KRAS were rare in European CRCs (17.5%) (Smith et al., 2002), which is also in concordance with our finding. Combination of APC and KRAS was also rare which was identified in only one of our patients and this is in concordance with study by Conlin et al. (2005). Bai et al. (2015) found the combination of FBXW7 alteration with KRAS in five out of 91 CRC patients (5.5%) which were similar with our finding. However, their result on combination of FBXW7 alteration with APC and TP53 was contradicted with ours. We noticed that the co-occurrence of APC, KRAS, and TP53 alterations in the same tumor was uncommon (5%) and this result concurs with those of other studies (Smith et al., 2002; Samowitz et al., 2007; Vasovcak et al., 2011). Collectively, these results indicate the wide variability in genetic alterations found in CRCs from different populations and highlights the need to further evaluate CRCs for common patterns of alterations.

Schell et al. (2016) recently revealed that in MSS tumors, the overall survival was roughly equivalent among APC, APC/KRAS, and APC/TP53 altered groups, but lower in the APC/KRAS/TP53 group. This seems to be reflected in one of our patients. Patient CT4, who harbored the triple alterations APC/KRAS/TP53 and also a MSS tumor, had short overall survival of around 8 months. This patient was diagnosed with moderately differentiated adenocarcinoma of the colon with lymph node metastasis (Duke’s C) in June 2009, underwent sigmoid total colectomy in August 2009 and developed CRC metastasis to liver in September 2009. She died in February 2010.

Wnt signaling is the main perturbed pathway in the development of CRC carcinogenesis (Cancer Genome Atlas Network, 2012; Kandoth et al., 2013; Vogelstein et al., 2013; Ashktorab et al., 2015). This pathway was altered in 80% of CRC patients (Ashktorab et al., 2015), which is almost similar to our finding (78.9%). Among the genes altered in this pathway were APC (12/19), FBXW7 (3/19), TCF7L2 (2/19) as well as SOX11 (1/19) and these were similarly reported by Lee et al. (2014) in primary CRCs except for CTNNB1 and SOX9. We noticed that none of our samples have CTNNB1 alteration as it is less common and have been reported in only six to nine percent of CRCs patients (Forbes et al., 2015). A major proportion of CRC patients (14/19; 74%) also harbored alterations in one or more genes in P53 signaling pathway including TP53 (11/19), ATM (1/19), and ATR (1/19) (encoding DNA damage proteins); EP300 (2/19) and EP400 (2/19) (encoding a P53 coactivator), which is consistent with findings from other studies (Lee et al., 2014; Yu et al., 2014).

In the Ras signaling pathway, aside from the dominant KRAS alterations (5/19), we had a patient with NRAS alteration (1/19) and this gene was previously reported to be altered in about five to eight percent of CRC patients (Forbes et al., 2015). Ras proteins are important in the signaling pathways that coordinate cell proliferation, differentiation, regulation of cell cycle as well as angiogenesis (Adjei, 2001; Zheng et al., 2016) and alterations in this gene are likely to result in constitutive activation and impaired regulatory functions (Downward, 2003).

SMAD4, SMAD2, MYC, and ARID1A were among the gene alterations found related to the TGF-β signaling pathway in our study. SMAD4 and SMAD2 alterations have been observed in 10–15% and 5% of CRCs, respectively (Fearon, 2011). SMAD4 codes for a protein that is involved as a downstream regulator of the TGF-β signaling pathway (Bai et al., 2015), and acts as a trimer and forms complexes with the receptor-phosphorylated SMAD2 and SMAD3; where these heteromeric complexes enter the nucleus to regulate apoptosis and cell cycle (Woodford-Richens et al., 2001). SMAD4 alterations caused the dysfunctional protein to interfere with proper signaling and gene transcription of target genes critical in cell cycle regulation (Bai et al., 2015). ARID1A alterations were reported in about nine percent in CRCs patient (Cancer Genome Atlas Network, 2012). In another study, inactivating mutations in ARID1A were shown to be frequent in microsatellite unstable CRC (Cajuso et al., 2014).

In the PI3K signaling pathway, we identified PIK3R1 and PIK3CB alterations. Class I PI3Ks are the best-characterized enzymes and include a catalytic subunit (p110α, p110β, and p110δ) and a regulatory subunit (p85α, p55α, p50α, p85β, and p55γ). The regulatory subunit p85 α is encoded by PIK3R1, while the catalytic subunit p110β is encoded by PIK3CB (Yuan and Cantley, 2008). The PI3K pathway play crucial roles in regulating cellular processes including protein synthesis, cell growth, proliferation, angiogenesis, cell cycle and survival (Baselga, 2011). PIK3R1 alterations and p85α loss were associated with PI3K pathway activation and increased oncogenic potential (Cizkova et al., 2013), while PIK3CB gene is implicated in PTEN-deficient tumourigenesis (Wee et al., 2008).

Cancer is a heterogenous disease and for the application of precision medicine it is vital to have the exact genetic information of the individual tumor. A number of gene markers are already guiding treatment decisions in daily practice. For instance, the identification of KRAS mutation at codons 12 and 13 in metastatic CRC predicts the lack of benefit from EGFR-targeted antibodies (Van Cutsem et al., 2011), and for the non-small cell lung cancer patients with EGFR del746_A750 or L858R, they are very responsive to EGFR tyrosine kinase inhibitors (Mok et al., 2009). Most recently, the novel PARP inhibitor olaparib is indicated for treating advanced ovarian cancer patients who have mutations in BRCA1 or BRCA2 (Munroe and Kolesar, 2016) and also the identification of KDR alteration as a novel predictive biomarker of exceptional response to low dose regorafenib reported by Loaiza-Bonilla et al. (2016) in patients with advanced CRC.

We have successfully profiled the gene alterations of a number of our local CRC patients and almost all patients harbored actionable mutations. Individualized cancer gene sequencing may be the next critical step in improving treatment options and increasing patient survival by selecting the appropriate therapeutic regime. By identifying gene alterations in individual cancers, specific treatments targeted against the detected gene alterations may prove to have greater benefits for cancer patients.

S-NA performed the experiments and data analysis. N-SM was involved in data interpretation, drafting the manuscript and overseeing the experiments. KS was heavily involved in data analysis. SS and MI were involved in optimization of library preparation and sequencing. NA was involved in functional analyses. NM and RJ were involved in critical review of the manuscript. IS and LM are colorectal surgeons involved in specimen collection and IR is a pathologist. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.