Clinical significance of BRAF non-V600E mutations on the therapeutic effects of anti-EGFR monoclonal antibody treatment in patients with pretreated metastatic colorectal cancer: the Biomarker Research for anti-EGFR monoclonal Antibodies by Comprehensive Cancer genomics (BREAC) study

Patients with BRAFV600E-mutated metastatic colorectal cancer (mCRC) have a poorer prognosis as well as resistance to anti-EGFR antibodies. However, it is unclear whether BRAF mutations other than BRAFV600E (BRAFnon-V600E mutations) contribute to anti-EGFR antibody resistance. This study was composed of exploratory and inference cohorts. Candidate biomarkers identified by whole exome sequencing from super-responders and nonresponders in the exploratory cohort were validated by targeted resequencing for patients who received anti-EGFR antibody in the inference cohort. In the exploratory cohort, 31 candidate biomarkers, including KRAS/NRAS/BRAF mutations, were identified. Targeted resequencing of 150 patients in the inference cohort revealed 40 patients with RAS (26.7%), 9 patients with BRAFV600E (6.0%), and 7 patients with BRAFnon-V600E mutations (4.7%), respectively. The response rates in RAS, BRAFV600E, and BRAFnon-V600E were lower than those in RAS/BRAF wild-type (2.5%, 0%, and 0% vs 31.9%). The median PFS in BRAFnon-V600E mutations was 2.4 months, similar to that in RAS or BRAFV600E mutations (2.1 and 1.6 months) but significantly worse than that in wild-type RAS/BRAF (5.9 months). Although BRAFnon-V600E mutations identified were a rare and unestablished molecular subtype, certain BRAFnon-V600E mutations might contribute to a lesser benefit of anti-EGFR monoclonal antibody treatment.

KRAS exon 2 mutations were the first validated predictive biomarker for primary resistance to anti-epidermal growth factor receptor (EGFR) monoclonal antibodies (cetuximab and panitumumab) in patients with metastatic colorectal cancer (mCRC) (Amado et al, 2008;Van Cutsem et al, 2009). Recently, minor RAS (KRAS exons 3 and 4 or NRAS exons 2, 3, and 4) mutations observed in B15-20% of mCRC with wild-type KRAS exon 2 cases have been validated as negative predictive biomarkers for anti-EGFR antibody treatment (Douillard et al, 2013). Therefore, expanded RAS (KRAS and NRAS) testing before administration of anti-EGFR antibody treatment has become essential to maximise the therapeutic benefit in patients with mCRC Van Cutsem et al, 2015;Yoshino et al, 2015) However, the response rate (RR) to anti-EGFR monotherapy remains low, at B20% in later line, in patients with mCRC having wild-type RAS, indicating additional biomarkers beyond expanded RAS are needed (Peeters et al, 2013).
BRAF is a serine-threonine kinase, located downstream of EGFR in the Ras/Raf/mitogen-activated protein kinase (MAPK) pathway (Mercer and Pritchard, 2003;Roskoski, 2010). The hotspot of BRAF mutations in CRC is substitution from valine to glutamic acid at codon 600 (V600E), located in exon 15, leading to 130-to 700-fold increased BRAF kinase activity compared with that of wild-type BRAF; these mutations are reported in B5-12% of cases (Davies et al, 2002;Andreadi et al, 2012). In addition, BRAF V600E mutations are more frequently observed in tumours in the right-sided colon than in tumours in the left-sided colon and rectum, and are prognostic biomarkers in CRC and could be potential predictive biomarker for anti-EGFR antibody treatment in pretreated mCRC (Kawazoe et al, 2015).
Recent clinical studies have shown that the primary location of the tumour may be associated with the therapeutic effects of anti-EGFR antibody treatment. Tumours in the right-sided colon showed worse outcomes than those in the left-sided colon and rectum in patients with mCRC with wild-type RAS, suggesting that genetic alterations other than BRAF V600E could be responsible for the poor prognosis of right-sided tumours (von Einem et al, 2014;Brulé et al, 2015).
In contrast, few reports have described mCRC with BRAF mutations other than BRAF V600E (BRAF non-V600E mutations), for which the incidence ranges from 1.6% to 5.1% (Shen et al, 2013;Ciardiello et al, 2014;Cremolini et al, 2015). BRAF non-V600E mutations can be classified on the basis of kinase activity as either high activity, intermediate activity or impaired activity (130-to 700-fold; high activity mutants, 1.3-to 64-fold; intermediate activity mutants and 30-80%, respectively) (Wan et al, 2004). Furthermore, BRAF mutation with impaired kinase activity also enhances MAPK kinase (MEK) phosphorylation by heterodimerising with wild-type CRAF (Haling et al, 2014). However, little is known regarding the clinicopathological features and anti-EGFR antibody sensitivity of BRAF non-V600E -mutated mCRC.
Here we reported the clinicopathological features of BRAF non-V600E -mutated mCRC and the clinical significance of these mutations with regard to the therapeutic effects of anti-EGFR antibody treatment in pretreated mCRC.

MATERIALS AND METHODS
Study design. The Biomarker Research for Anti-EGFR Monoclonal Antibodies by Comprehensive Cancer Genomics (BREAC) study was a multicentre, translational research study aiming to investigate novel predictive biomarkers of anti-EGFR antibody treatment in patients with mCRC harbouring wild-type or unknown KRAS exon 2 (details in Supplementary Protocol). We had the following study design; patients were divided into two independent cohorts named 'exploratory' and 'inference' cohorts according to the duration of anti-EGFR antibody treatment. The exploratory cohort included subjects who were considered as 'super-responders' or 'super-nonresponders' among the entire mCRC cohort (403 patients) who received cetuximab including treatment as salvage line between September 2008 and May 2010 at seven major institutions in Japan. We put a strong assumption that associations between relatively minor gene mutations and patient prognosis become more remarkable in the 'super-responders' plus 'nonresponders' cohort than associations observed in the entire cohort, leading to a power increase in statistical tests (Supplementary Figure S1). The possible mutations founded in the exploratory cohort were then evaluated by targeted resequencing of the patients in the inference cohort who were treated by anti-EGFR antibody during the different period from the exploratory cohort.
Study conduct. In the inference cohort, patients with mCRC were consecutively enroled between June 2010 and November 2011 from seven institutions to validate the associations of candidate biomarkers identified in the exploratory cohort with the efficacy of anti-EGFR antibody treatment in pretreated mCRC harbouring wild-type or unknown KRAS exon 2. The details of selection criteria for the inference cohort are described in the Supplementary Appendix.
This study was approved by the Institutional Review Board of each participating centre. Written informed consent was obtained from patients who were alive when initiating this study. For deceased patients and their relatives at that time, we disclosed the study design on the website of each centre and allowed the relatives to approve or deny inclusion in the study. This study was conducted in accordance with the Ethical Guidelines for the human genome and genetic analysis research of the Ministry of Education, Culture, Sports, Science and Technology, Ministry of Health, Labour and Welfare and Ministry of Economy, Trade and Industry.
Collection of clinical and pathological data. An electronic data capture system (Viedoc; PCG Solutions, Uppsala, Sweden) was used for registration of patients and collection of clinical and pathological data by the Office of Translational Research, Exploratory Oncology Research and Clinical Trial Center (EPOC), National Cancer Center, Chiba, Japan.
Patient characteristics including age, sex, site of primary lesion, histology, site of metastases, prior treatments, clinical outcome of anti-EGFR antibody treatment, subsequent treatment, and severe adverse events related to anti-EGFR antibody treatment, were collected. Sites of primary lesions were divided into right-sided colon, left-sided colon, and rectum. Right-sided tumours were defined as those arising anywhere from the caecum to the transverse colon, and left-sided tumours were defined as those arising anywhere from the splenic flexure to the rectosigmoid junction.
Primary investigators were blinded to cancer genome alterations analysed in the study; investigators evaluated the antitumour effect according to Response Evaluation Criteria in Solid Tumours (RECIST) version 1.1 (Eisenhauer et al, 2009) and confirmed the safety of the treatment based on the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 (National Cancer Institute, 2009).
Targeted capture resequencing. Archived FFPE tissue specimens collected before administration of anti-EGFR antibody were used for target resequencing. Candidate biomarkers identified from the exploratory cohort were validated by target resequencing, which covered the full length of all the candidate genes, including KRAS, NRAS, and BRAF. The details of preparation of clinical samples, DNA extraction, identification of single nucleotide variants (SNVs) and insertion-deletion mutations (INDELs) and target resequencing are described in the Supplementary Appendix. BRAF activity assay. To clarify the activity of newly identified BRAF mutations, we assessed the phosphorylation status of downstream molecules of EGFR by western blotting using HEK293 cells transfected with the BRAF mutant vector (Supplementary Appendix).
Statistical analysis. The efficacy endpoints were progression-free survival (PFS), defined as the duration from the initiation of anti-EGFR antibody treatment to disease progression or death from any cause; overall survival (OS), defined as the duration from the initiation of anti-EGFR antibody treatment to death from any cause; RR, defined as the proportion of patients who had a complete or partial response with anti-EGFR antibody treatment; and disease control rate (DCR), defined as the proportion of patients who had a complete or partial response or stable disease. For PFS and OS, survival curves according to each mutational status were estimated by the Kaplan-Meier method and were compared using log-rank test.
Univariate and multivariate Cox regression analyses were performed to evaluate the prognostic impact of any RAS/ BRAF V600E /BRAF non-V600E mutant (herein referred to as RAS/ BRAF mutant) vs wild-type. Covariates in the regression analyses included RAS/BRAF (mutant vs wild-type), age, gender, ECOG PS, histology, primary site, primary tumour resection, adjuvant chemotherapy, metastasis (synchronous vs metachronous), combined use of irinotecan, and prior oxaliplatin treatment. Considering the limited number of death events, backward elimination procedure, setting the removal criteria as a P-value of o0.20, was performed; four covariates (gender, ECOG PS, primary site, and combined use of irinotecan) were forcibly retained as potential confounding factors.
All statistical analyses were performed using SAS Release 9.3 (SAS Institute, Inc., Cary, NC, USA). All P values were obtained from two-sided statistical tests with a significance level of 0.05.

RESULTS
Summary of the exploratory cohort. In the exploratory cohort, 92 patients with mCRC, comprising 57 super-responders and 35 nonresponders to anti-EGFR antibody treatment (90 KRAS exon 2 wild type and 2 unknown), were selected (Supplementary Figure  S2). FFPE clinical samples of both cancerous and noncancerous areas were subjected to whole exon sequencing. Briefly, the exomes were captured using the SureSelect Human All Exon V4 þ UTRs Kit (Agilent Technologies) and sequenced using a HiSeq 2000 system (Illumina) to generate 100 bppaired-end data. The average base coverage of the targeted regions in the tumour and normal samples was 162.5-fold (range: 10.2-389.7) and 166.2-fold (range: 2.7-377.4), respectively. We identified 182.7 ± 97.1 (range: 37.0-509.0) (5.7±3.0 per Mb, range: 1.2-15.9) somatic SNVs and 8.1 ± 4.2 (range: 1.0-23.0) (0.3 ± 0.1 per Mb, range: 0.0-0.7) somatic INDELs in the tumour tissues. Thirty-one candidate biomarker genes, including KRAS/NRAS/BRAF, in which mutations significantly deviated from either super-responders or nonresponders, were selected for further analysis with the inference cohort. Detailed data of the exploratory cohort is described elsewhere. Here we focused on the association of expanded RAS and BRAF mutation with efficacy endpoints in the inference cohort.
Genomic alternations and patient characteristics according to RAS/BRAF status in the inference cohort. A total of 184 patients were selected in the inference cohort. Target resequencing of the candidate biomarker genes, including KRAS/NRAS and BRAF, was successful in 156 patients, while 28 clinical samples were not analysed due to insufficient FFPE samples (n ¼ 6) and sequencing failure (n ¼ 22). The average base coverage of the targeted regions in the tumour and normal samples was 671.9-fold (range: 66.8-1735.0) and 731.6-fold (range: 70.4-1699.5), respectively. We identified 1.4 ± 1.3 (range: 0.0-8.0) variants in the 31 candidate genes from tumour tissues. Additionally, six patients were excluded due to ineligibility (n ¼ 5) and acquisition of the specimen after anti-EGFR antibody treatment (n ¼ 1). Accordingly, 150 patients were included in the biomarker analysis population (Supplementary Figure S3). Baseline patient characteristics and clinical outcomes were similar between the whole population (N ¼ 184) and biomarker analysis population (N ¼ 150; data not shown).
Baseline patient characteristics, based on the RAS and BRAF mutational status, are shown in Table 2. Both BRAF V600E and BRAF non-V600E mutant tumours were more commonly associated with the right-sided colon (44.4% and 57.1%, respectively) than the RAS/BRAF wild-type and RAS mutant tumours (13.8% and 27.5%, respectively). BRAF non-V600E mutant tumours tended to have more lymph node metastases (71.4%) than with other mutational subtypes, RAS and BRAF V600E mutations (27.5% and 11.1%, respectively).
RRs to anti-EGFR antibody treatment according to RAS/BRAF mutation status in the inference cohort. The RR was 20.7% in all patients. The RRs in patients with RAS, BRAF V600E , and BRAF non-V600E mutations were lower in comparison with patients harbouring wild-type RAS/BRAF (2.5%, 0%, and 0% vs 31.9%, respectively). In addition, the proportion of SD more than 6 months in patients with BRAF non-V600E mutant was 14.3%, which was similar to that in patients with RAS or BRAF V600E mutations rather than wildtype RAS/BRAF (Table 3).
Survival and safety analysis according to RAS and BRAF status in the inference cohort. The median follow-up time was 12.1 months as of the cutoff date of December 24, 2014. The median PFS and OS of all patients were 4.0 months (95% confidence Right-sided tumours were defined as those arising anywhere from the caecum to the transverse colon, and left-sided tumours were defined as those arising anywhere from the splenic flexure to the rectosigmoid junction. interval (CI), 3.4-4.8 months) and 12.4 months (95% CI, 9.8-14.0), respectively.
BRAF activity assays for the newly identified mutations Q524L and L525R. To evaluate the kinase activity of newly identified BRAF mutants, mutant-and wild-type BRAF-expressing vectors were transiently transfected into EGFR-expressing HEK293 cells. The transfection efficiency was more than 70%, as assessed by EGFR-expressing control plasmid vector transfection (data not shown). Western blot analysis showed that extracellular signalregulated kinase (ERK) phosphorylation in cells with BRAF V600E overexpression was significantly increased compared with that in cells with wild-type BRAF overexpression (Figure 2). BRAF L525R induced increased ERK phosphorylation to a level similar to that induced by BRAF V600E . However, BRAF Q524L activity was similar to that of wild-type BRAF.
In the control vector-transfected cells, cetuximab reduced ERK phosphorylation in a concentration-dependent manner. Additionally, cetuximab reduced ERK phosphorylation level, which was enhanced by wild-type BRAF or BRAF Q524L expression. On the other hand, ERK phosphorylation enhanced by BRAF V600E and BRAF L525R was not affected by cetuximab, suggesting that cells with BRAF V600E or BRAF L525R mutants were resistant to cetuximab-induced inhibition of EGFR.

DISCUSSION
To the best of our knowledge, this is the first report for the clinical significance of BRAF non-V600E mutations focusing on the therapeutic effects of anti-EGFR monoclonal antibodies in patients with pretreated mCRC.
Few studies have reported the clinicopathological features of BRAF non-V600E mutations because BRAF V600E -mutant tumours are most frequently observed in CRC (Shen et al, 2013;Ciardiello et al, 2014). According to studies in Western countries, the incidence of BRAF non-V600E in mCRC was reported to range from 1.6% to 2.7% (Ciardiello et al, 2014;Cremolini et al, 2015). In contrast, the incidence of BRAF non-V600E in 676 Chinese patients with mCRC was reported to be 5.1%, consistent with that in our cohort (4.7%) (Shen et al, 2013). On the other hand, the racial differences in terms of the incidence of BRAF V600E mutations were reported from  the analysis of large-scale adjuvant trial in US, suggesting the incidence appeared to be lower in Asians than in blacks or whites (Yoon et al, 2015). Instead, the incidence of BRAF non-V600E mutations might be higher in Asian than that in Caucasian patients. Two meta-analyses suggested that primary tumours in the rightsided colon showed worse prognoses than those in the left-sided colon and rectum in patients treated with anti-EGFR antibodies (Arnold et al, 2017;Holch et al, 2017). In addition, integrated analysis of two randomised panitumumab studies showed that they had consistent results, even when the BRAF V600E mutations were excluded (Boeckx et al, 2017). In our series, a similar tendency was observed clinical outcomes; specifically, patients with wild-type RAS mCRC with primary tumours in the right-sided colon had poorer prognoses compared with those having primary tumours in the left-sided colon and rectum, although the difference was not significant. However, if limited to wild-type RAS/BRAF tumours, there were no clear differences in OS among sites of primary lesions (Supplementary Figure 4). Thus, the unresponsiveness of primary tumours in the right-sided colon to anti-EGFR antibodies in later line might be partially explained by underlying BRAF non-V600E -mutated tumours.
Subtypes of BRAF mutations in the kinase domain can be classified into high, intermediate, and impaired activity subtypes based on their kinase activity (Wan et al, 2004). The BRAF V600E mutation belongs to the high activity subtype, whereas the BRAF G469A , BRAF L485F , and BRAF V600R mutations observed in this study belong to the intermediate subtype and the BRAF D594G mutation belongs to the impaired subtype. Moreover, the BRAF L525R mutant observed in this study had enhanced kinase activity, and the enhanced downstream signal of BRAF L525R may contribute to primary resistance to cetuximab, consistent with the lack of response to anti-EGFR antibody treatment in our series. In contrast, the newly identified BRAF Q524L mutant identified in this study had intermediate kinase activity and did not induce resistance to cetuximab in an in vitro cell model. However, such in vitro experiments in non-CRC epithelial cell lines may not fully predict the clinical outcome.
One possible explanation for the similar behaviours of BRAF non-V600E and BRAF V600E -mutated tumours in terms of unresponsiveness to anti-EGFR antibody treatment may be the incomplete blockade of the MEK pathway by modestly upregulated kinase activity of BRAF and/or by additional signalling through  . BRAF activity assay. Phosphorylation status of ERK was assessed by western blotting. HEK293 cells were transiently transfected with wild-type, V600E, Q524L, or L525R BRAF and were then treated with 0.5 or 5 mg/ml cetuximab.
wild-type CRAF (Wan et al, 2004). Therefore, it is necessary to establish patient-derived xenograft models harbouring BRAF non-V600E mutations to clarify the mechanisms of primary resistance to anti-EGFR antibody treatment in BRAF non-V600E -mutated mCRC.
Regarding BRAF D594G , classified as an impaired subtype in the study, Cremolini et al reported 10 cases with BRAF mutations in codons 594 or 596, the number of which was similarly small showing a favourable prognosis from first line (Cremolini et al, 2015). More recently, it was reported that OS was significantly longer for 101 patients with BRAF non-V600E mutations than for the control group of 99 patients with BRAF V600E mutations (60.7 vs 11.4 months) (Jones et al, 2017); however, our analysis focused on pretreated population. In addition, according to the European Consortium, De Roock et al reported that two patients harbouring BRAF D594G -mutated mCRC achieved a partial response to cetuximab monotherapy (De Roock et al, 2010). In contrast, the two patients with BRAF D594G -mutated tumours in our study did not achieve objective response to anti-EGFR antibody treatment. Considering small patient's number of each reports as well as heterogeneous population, it is difficult to conclude predictive impact of BRAF non-V600E mutation. The overall data in this study supported that BRAF non-V600E mutations were prognostic, as a similar magnitude to the presence of BRAF V600E and RAS mutations in later line, and the outcome appeared similar to patients with RAS mutations as well, who do not benefit from anti-EGFR therapy. The present study has some limitations. It was a retrospective study with a small number of subgroups of BRAF non-V600E mutations, using archived FFPE samples. In addition, it is difficult to conduct further analyses by subdividing the group into 'non-V600E kinase activity' and 'non-V600E non-kinase activity' subgroups, due to the small number of BRAF non-V600E mutations. The emergence of secondary RAS and BRAF gene mutations in ctDNA was recently reported after treatment with anti-EGFR antibody (Bettegowda et al, 2014). However, in the case of patients receiving systemic chemotherapy without anti-EGFR antibody, secondary gene alterations are rare (Kawamoto et al, 2012); therefore, we believed that adding the archived FFPE samples before anti-EGFR antibody administration would be reasonable. Further investigation in large-scale data set from such as randomised control trials is necessary to clarify the significance of anti-EGFR antibody treatment for each BRAF non-V600E mutational variants as the role of the predictive value.
In conclusion, although identified BRAF non-V600E mutations were rare and unestablished molecular subtype in mCRC, overall clinical outcomes of BRAF non-V600E mutations in the kinase domain, similar to those of RAS-and BRAF V600E -mutant tumours, appeared to be significantly worse than those in wild-type RAS/ BRAF tumours. Certain BRAF non-V600E mutations might contribute to a lesser benefit of anti-EGFR monoclonal antibody treatment. This novel knowledge provides an intriguing background to investigate new target approaches in patients with BRAF non-V600E mutations and represents substantial progression toward more precision medicine.
This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ r The Author(s) named above 2017