Background
The increasing knowledge of cancer biology has led to the development of targeted therapies, designed to interfere with specific molecules involved in tumor growth and progression [
1,
2]. EGFR is a transmembrane receptor tyrosine kinase (TK) implicated in several cellular responses, like apoptosis, differentiation, cellular migration, and adhesion. This TK and the pathways it controls play an important role in colorectal carcinogenesis [
3‐
5], making it a good target for biological therapy of this disease [
2]. A network of various signal transduction cascades is stimulated by EGFR signaling, namely the RAS/RAF/MEK/ERK, PI3K/AKT, JAK/STAT and PLCγ pathways. Cetuximab, a human-mouse chimeric IgG1, and panitumumab, a fully human IgG2, are monoclonal antibodies (moABs) that compete with EGFR’s ligands and specifically bind to the receptor, blocking ligand-induced downstream signaling [
2]. These targeted agents have been evaluated in several clinical trials for the treatment of metastatic colorectal cancer (mCRC), either alone, in combination with fluoropyrimidine-based chemotherapy regimens, or with bevacizumab [
6‐
11], and have subsequently been approved by the European Medicines Agency (EMEA) and the U.S. Food and Drug Administration (FDA).
Several retrospective analyses of
KRAS mutational status in tumors from patients treated with cetuximab and panitumumab found an association between
KRAS codons 12 or 13 activating mutations and lack of treatment efficacy [
6‐
11]. In normal cells, the KRAS protein alternates between an inactive GDP-bound form and an active GTP-bound form. Mutations in
KRAS codons 12 and 13 originate a constitutively active protein, resulting in a continuous and self-sufficient (independent of ligand binding) KRAS signaling. These
KRAS mutations, present in about 40% of mCRC, are the only available (negative) predictors of response to anti-EGFR moABs, and this therapy is strictly indicated for patients with
KRAS wild-type mCRC [
6,
9,
12,
13]. However, absence of
KRAS exon 2 mutations does not guarantee treatment response, as only 40 to 60% of these cases respond to anti-EGFR therapy [
7,
13,
14]. Other mutations in genes encoding proteins that act downstream of EGFR, such as
KRAS,
BRAF, and
PIK3CA, may be responsible for the absence of treatment response in such cases.
In this study, 201 cases of mCRC wild-type for
KRAS codons 12 and 13 were screened for mutations in other potential biomarkers of response to anti-EGFR treatment, namely in the coding regions of KRAS switch II and G5 regions (exons 3 and 4), the P-loop and activation segment of BRAF (exons 11 and 15), and in PIK3CA’s helical and kinase domains (exons 9 and 20) [
15,
16].
Discussion
We here show that more than one-fourth of
KRAS exon 2 wild-type mCRC patients present other mutations in
KRAS,
BRAF, and/or
PIK3CA and report five novel mutations, namely four in
KRAS exon 3 and one in
BRAF exon 11. The mutational frequencies in the most commonly altered codons of
KRAS,
BRAF, and
PIK3CA genes were in accordance to those previously described in the literature. We have previously shown that HRM is a highly sensitive technique to detect mutations in mCRC, being significantly more sensitive and cheaper than standard Sanger sequencing [
17], combining high sensitivity with the ability to detect novel mutations. A
KRAS codon 61 and 146 mutation frequency of 10.4% is similar to what has been reported for
KRAS codons 12 and 13 wild-type patients (6.5% - 10.5%) [
24,
25]. The frequency of the
BRAF p.Val600Glu mutation was lower (4.0%) but comparable to the reported frequency range of 4-18% in mCRC without
KRAS codon 2 mutations [
25,
27‐
30]. Since the
BRAF p.Val600Glu mutation is associated with microsatellite instability (MSI) status and right colon tumors [
31‐
33], variations in sample characteristics between studies can account for the wide frequency range, but this is often hard to verify as many studies in mCRC do not describe the primary tumor localization nor MSI status.
PIK3CA mutations were present in 10.9% of the tumors, which is similar to previous reports [
24,
34‐
36]. Interestingly, both
BRAF (
P=0.000) and
PIK3CA (
P=0.011) mutations were significantly more frequent in colon than in sigmoid or rectal carcinomas. On the other hand, an association was found between
KRAS mutations and older age at diagnosis (
P=0.034), which was not observed for
BRAF or
PIK3CA. These findings should be confirmed in larger series in order to evaluate its significance.
KRAS codon 61 oncogenic mutations occur at an essential position for GTP hydrolysis and decrease RAS-mediated GTP hydrolysis [
37], resulting in transformation efficiencies that vary up to 1000-fold [
38]. It has been demonstrated
in vivo that codon 61Leu, Lys, and Arg induce a strong oncogenic phenotype, whereas 61 His is a moderately transforming mutant [
38]. Aminoacid Ala146 is involved with guanine base interaction and mutations in this codon do not affect GTPase activity, but are associated with an increased GDP to GTP exchange. Expression of p.Ala146Thr mutations
in vivo results in elevated RAS-GTP and phosphorylated ERK compared to wild-type
KRAS, albeit lower than that caused by codon 12 mutations [
39]. However, there is no data available to determine the influence in RAS protein structure of the novel deletion (p.Ala59del) and the two novel large in frame duplications (p.Cys51_Ser65dup and p.Thr58_Met72dup) we found in exon 3, but the fact that they are located in the switch II region is an indicator that they may activate RAS by impairing GTP hydrolysis. Of notice, few
KRAS duplications and deletions have been reported: only three in exon 2 and two in exon 3. No functional studies exist regarding the role of
KRAS p.Gly60Val or p.Glu49Lys mutations, but it is known that the Gly60 residue interacts with γ-phosphate of GTP and is a conserved amino acid in the superfamily of GTPases [
40], facts that argue in favor of Gly60Val pathogenicity.
Both
BRAF p.Val600Glu and p.Lys601Glu mutations occur in the activation site and originate proteins with high kinase activity.
In vitro,
BRAF p.Val600Glu and p.Lys601Glu proteins have higher TK activity than the wild-type protein (500-fold and 120-fold higher, respectively) [
41]. Mutants p.Val600Glu also show a six-fold higher ERK signaling
in vivo, when compared to the wild-type protein [
41]. These high TK activity mutants are thought to simulate the conformational changes caused by activation segment phosphorylation, resulting in protein ligand-independent constitutive activation. On the other hand, the Gly466 is the second glycine of the P-loop GXGXXG motif (G=glycine; X=variable), conserved in more than 99% of all kinases [
15]. It is an important catalytic residue and its substitution to glutamic acid (p.Gly466Glu) originates a protein with higher ERK signaling than wild-type BRAF but a diminished, although constitutively active, TK activity [
41]. It has been proposed that increased ERK signaling occurs via an association between BRAF and CRAF and their ability to stimulate ERK is dependent on CRAF activation [
41]. It has been demonstrated that p.Gly466Glu cells depend on CRAF for ERK signaling: they induce strong CRAF activation and CRAF depletion significantly suppresses ERK signaling [
41].
PIK3CA helical domain mutants p.Glu542Lys, p.Glu545Lys, and p.Gln546Lys and kinase domain mutants p.Met1043Ile, p.His1047Arg, p.His1047Leu, and p.His1047Tyr all display enhanced lipid kinase activity compared to the wild-type p110α, and p.Glu542Lys, p.Glu545Lys, and p.His1047Arg induce AKT phosphorylation at higher levels than the normal protein [
42‐
48]. Furthermore, p.Glu545Lys and p.His1047Arg promote cell growth and invasion in CRC cell lines, and mutations p.His1047Leu and p.His1047Tyr induce oncogenic transformation in primary cell cultures of chicken embryo fibroblasts [
44,
46]. In CRC, Met1043 is less frequently altered than His1047 (0.8%
vs 7.1%) [
49]. Amino acids 1043 and 1047 are located on the same protein helix and probably affect protein function by altering activation loop conformation, leading to elevated kinase activity [
42,
50]. The above referred helical domain mutations occur at residues involved in the interaction with the adaptor protein and are thought to abrogate its inhibitory effect by increasing the positive charge on the surface of the helical domain. It has also been demonstrated that p85 does not inhibit these mutants
in vitro[
45,
50]. Finally, the Glu to Asp substitution in codon 545 has not been functionally studied, but both amino acids involved are polar and negatively charged, thus making it unlikely that this substitution will produce the same effect on p110α as those described above. In this study we also observed that
PIK3CA codon 545 substitutions account for 9.8% of
PIK3CA mutations in CRC [
49] and, since the carcinoma carrying the
PIK3CA p.Glu545Asp mutation did not present mutations in either
KRAS or
BRAF, it is conceivable that this mutation confers some selective advantage.
In six cases, we found two different mutations in the various exons studied, most commonly coexistence of a
PIK3CA mutation with either a
KRAS or a
BRAF mutation. Coexisting mutations of
KRAS/
BRAF and
PIK3CA have been reported in several studies [
24,
30,
31,
36], with
PIK3CA exon 20 mutations more frequently co-occurring with mutations of unknown significance or with
KRAS codon 146 mutations [
24]. Additionally, we found one instance of coexistence of the
PIK3CA p.His1047Arg mutation with the novel mutation
BRAF p.Val471Ala, a conserved residue of RAF proteins, having no functional studies available to allow inferences on its oncogenic potential. Finally, one case harbored two
KRAS mutations, namely the novel p.Glu49Lys and the p.Ala146Thr mutations. The coexistence of two mutations in the same gene or in different genes may be explained by a synergistic contribution of both mutations to pathway activation or the occurrence of each mutation in different subclones as a result of tumor clonal divergence.
According to a recently published large multicentric study involving retrospective mutation analysis on
KRAS,
BRAF,
NRAS, and
PIK3CA in mCRC and the impact of mutations in these genes on cetuximab treatment efficacy [
24], tumors with
KRAS codon 61 mutations have lower response rates and
PIK3CA exon 20 mutations are associated with a worse outcome after cetuximab treatment, with
NRAS mutations (codons 12, 13 and 61) being predictive of nonresponse to this targeted therapy. On the other hand, this retrospective study indicates that
KRAS codon 146 and
PIK3CA codon 9 mutations do not affect cetuximab efficacy. This study also confirmed the inefficacy of cetuximab in patients with
BRAF p.Val600Glu mutations, with response rates of 8%
vs 38% for
BRAF mutated and wild-type, respectively [
24,
25,
27,
28,
51]. No associations with treatment response have been published for
BRAF exon 11 mutations or any other
KRAS exon 3 mutations besides those in codon 61, essentially because they are rare. We could not make an evaluation of the predictive value of these mutations in our patients at the time of writing due to the low number of mutated cases that have been treated with cetuximab, but in face of the findings of De Roock
et al.[
25] our mutation data indicates that at least 10.9% of our mCRC patients wild-type for
KRAS codon 12 and 13 would not benefit from anti-EGFR targeted therapy. Further prospective or functional studies will be necessary to evaluate the predictive value of the remaining mutations, including the novel
KRAS and
BRAF mutations we here report.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JGG carried out most molecular genetic studies and drafted the manuscript. IV, PR, PP, CP, MP, and AP helped to set up, carry out, and interpret the molecular genetic studies. PL and RH provided histopathological data. MF, AR, PF, MM, NS, AA, JM, FA, CC and LLS provided patient clinical data. LLS and JGG performed the statistical analysis. MRT coordinated the study and helped to draft the manuscript. All authors read and approved the final manuscript.