BRAF V600E-Mutant Colorectal Cancers with Liver-Only Metastases: A Retrospective Study.


 BACKGROUND BRAF V600E-mutant colorectal cancers (CRCs) are associated with shorter survival than BRAF wild-type tumors. Therapeutic decision-making for colorectal liver metastases (CRLM) harboring this mutation remains difficult due to the scarce literature. Our aim was to study a large cohort of BRAF V600E-mutant CRLM patients in order to identify prognostic factors associated with overall survival. METHODS We retrospectively identified BRAF V600E-mutant CRCs diagnosed with liver-only metastases, resected or not, between April 2008 and December 2017, in 25 French centers. Clinical, molecular, pathological characteristics and treatment features were collected. Overall survival (OS) was defined as the time from CRLM diagnosis to death from any cause. Cox proportional hazard models were used for statistical analysis. RESULTSAmong the 105 patients included, 79 (75%) received chemotherapy, 18 (17%) underwent upfront CRLM surgery, and 8 (8%) received exclusive best supportive care. CRLM surgery was performed in 49 (46.7%) patients. CRLM were mainly synchronous (90%) with bilobar presentation (61%). The median OS was 34 months (range, 28.9–67.3 months) for resected patients and 10.6 (6.7–12.5) months for unresected patients (P < 0.0001). Most patients received doublet chemotherapy (72%), and 34% received a combination with bevacizumab as a first-line treatment. In multivariate analysis, primary tumor surgery (hazard ratio (HR) = 0.349; 95% confidence interval (CI) 0.164–0.744, P = 0.0064) and CRLM resection (HR = 0.169; 95% CI 0.082–0.348, P < 0.0001) were associated with significantly better OS. CONCLUSIONSIn the era of systemic cytotoxic chemotherapies, liver surgery seems to extend OS. Therefore, CRLM resectability should be assessed, regardless of mutational status.


Background
Approximately 50% of patients with colorectal cancer develop colorectal liver metastases (CRLM), and their outcomes are intimately related to CRLM resectability: the 5-year overall survival (OS) rate ranges from 30-50% after CRLM surgery, whereas it is lower than 10% for unresectable CRLM [1,2] . However, 50-85% of patients experience relapse after CRLM resection, and the curative intent of metastasectomy is accomplished in approximately 20% of cases [3][4][5] . In the era of precision medicine, efforts are aimed at a better selection of patients who might bene t from metastasectomy. Several clinical scoring systems based on clinicopathological parameters have been proposed; however, but their clinical value is still questioned [6,7] .
Colorectal cancers (CRCs) harboring BRAF V600E mutations are aggressive cancers with rapid metastatic spread that more frequently involves peritoneal and nodal invasion than liver metastases. Until recently, their management was based on limited data, mainly from subgroup analysis of randomized clinical trials. This subgroup of patients is less responsive to standard chemotherapies. In the CALGB/SWOG 80405 trial assessing the addition of the targeted agent cetuximab or bevacizumab or both to doublet chemotherapy FOLFOX or FOLFIRI, the median OS for BRAF-mutant patients remained poor compared to that of BRAF wild-type patients: 13.5 months versus 30.6 months, respectively (P < 0.001) [8] . In addition, a recent meta-analysis of ve randomized clinical trials demonstrated that intensive upfront chemotherapy with triplet FOLFOXIRI plus bevacizumab did not improve survival among BRAF V600E-mutant patients [9] .
This gap in survival rates has been also observed after CRLM resection in several retrospective subgroup analyses. In the latest study, the 3-year OS rates for BRAF-mutant and wild-type patients were 54% and 82.9%, respectively [10] . However, these numbers must be interpreted with caution as BRAF-mutant CRCs with liver-only metastases represent a limited population, and only 5% of patients undergoing CRLM resection harbor these mutations [11][12][13] .
Therefore, our knowledge about BRAF-mutant patients with CRLM is currently limited to those patients undergoing resection or with extra-hepatic metastases receiving chemotherapies. In this study, we aimed to report outcomes of a large cohort of BRAF V600E-mutant patients with exclusive CRLM, resected or not, to identify potential prognostic factors. Oncologie DIGEstive). BRAF V600E mutated-patients were identi ed from molecular biology platforms and each case was screened in order to identify and include patients with liver-only disease. Patients with synchronous extra-hepatic resectable disease metastases were excluded. The study was conducted according to the ethical standards in line with the French regulation. French Data Protection Authority (CNIL agreement n° DEC18-409 (2018_01)) provided a waiver of informed consent for this retrospective study and permitted the publication of anonymized data.

Study Population and Design
BRAF and RAS mutational statuses were determined from either primary CRC samples or CRLM tissuesas several studies have demonstrated a high molecular concordance between primary CRC and liver metastases [14,15] -using PCR or next-generation sequencing.
The following clinical, molecular, and pathological characteristics were collected at baseline: age at CRLM diagnosis, sex, KRAS and NRAS mutations, mismatch repair (MMR) status, primary tumor site, surgery of primary tumor, tumor and nodal stages according to the American Joint Committee on Cancer, synchronous (< 6 months) versus metachronous CRLM diagnosis, CRLM distribution and number, and initial resectability status. In addition, treatment features (CRLM surgery and systemic therapies) and survival were assembled.

Treatment Features and De nitions
The treatment decision for each patient was made during multidisciplinary meetings in each institution.
According to the CRLM resectability status and performance status, patients received preoperative chemotherapy, upfront liver surgery, palliative chemotherapy, or best supportive care. Patients were then followed-up every 2-3 months through physical examination, biological tests, and computed tomography scan. OS was de ned as the time from CRLM diagnosis to the time of death or the date of last follow-up. Post-operative OS was de ned as the time from CRLM resection.

Statistical Analysis
For descriptive analysis, quantitative parameters are presented as median and quartiles and qualitative parameters as percentages. CRLM resected and unresected groups were compared using the χ2 or Fisher's exact test, as appropriate. Survival rates were estimated by the Kaplan-Meier method and were compared using the log-rank test.
Among patients treated exclusively with chemotherapy (n = 48), 63% received a second line (n = 30) and 35% received a third line (n = 17). From the second line onward, targeted therapies were more frequently used. In total, 24 patients (50%) received concomitantly or successively the following cytotoxic drugs: uoropyrimidine, oxaliplatin, and irinotecan. Of note, 54% of patients with unresected CRLM underwent primary tumor surgery (Table 1).
Finally, 10 out of 105 patients (10%) participated in clinical trials, four of which involved immune checkpoint inhibitors (ICIs) or targeted therapies.

Comparison of Resected and Non-Resected CRLM groups
Clinical and pathological characteristics according to CRLM treatment status are summarized in Table 1.
Patients with resected CRLM (n = 49) were signi cantly more likely to present less than 10 metastases (P < 0.0001) with unilobar distribution (P = 0.0036) and initially resectable (P < 0.0001  [17] . In addition, a recent case-matched study demonstrated that BRAF mutations did not increase the risk of relapse after CRLM surgery compared to BRAF wild-type disease (HR 1.16, 95% CI 0.72-1.85; P = 0.547) [10] . The high proportion of patients undergoing resection in our cohort should re ect the fact that the assessment of mutational status was probably not performed in patients with poor prognosis.  [17][18][19][20] . OS results remain lower than the previous publication from Bachet et al. with a median OS of 52.7 months (n = 66). The exclusion of non-V600E BRAF mutated-patients in our study may explain this difference [10] . Relapse free survival and progression free survival were not included in our study as the de nitions would differ for resected and unresected groups.
The positive results of primary tumor resection in the unresected CRLM group were surprising and should be interpreted with caution due to the small number of patients involved (30 versus 26). Indeed, a recent study showed that primary tumor resection followed by chemotherapy was not superior to chemotherapy alone (HR 1.10 [0.76-1.59], one-sided P = 0.69). The trial was terminated early for futility reason [21] .
In our cohort, a small proportion of patients received an upfront triplet regimen with or without bevacizumab (12/79, 15%), with a median OS of 16.6 months (6.7-not reached). Of note, the majority of the cohort presented a good performance status (84% in the group treated exclusively with chemotherapy). An intensive regimen has been assumed bene cial in BRAF-mutant patients with unresectable liver-limited disease to date, based on a pooled analysis of a small number of patients (n = 20) [22] . However, a recent meta-analysis of ve randomized trials comparing FOLFOXIRI plus bevacizumab with doublet plus bevacizumab in 105 BRAF V600E-mutant patients showed no increased bene t in the intensive therapeutic arm [9] .
Among the patients treated with chemotherapy only, 63% received second-line treatment; this rate is superior to those reported in the COIN trial (33%) (23) and in a matched case-control study (51%) [24] .
Beyond the second line, candidates for treatment decreased dramatically to 35%, and it is important to pinpoint that only 50% of patients received the three major cytotoxic drugs: 5-FU/leucovorin, oxaliplatin, and irinotecan. Our study showed that even if BRAF-mutant metastatic disease is con ned to one organ, the prognosis remains poor when the patient is treated with chemotherapy only.
Our study population mostly received standard chemotherapies, not new practice-changing therapies.
Recently, mitogen-activated protein kinase pathway-targeted therapies have demonstrated better e cacy.
In the BEACON trial, the combination of encorafenib, a BRAF inhibitor, and cetuximab (anti-EGFR) with or without binimetinib, a MEK inhibitor, was associated with a signi cantly longer OS than standard chemotherapy after at least one prior line in a large cohort of patients with BRAF V600E mutations [25] .
The same regimen as a rst-line is currently under investigation in a phase II trial (ANCHOR-CRC) [26] .
ICIs represent another therapeutic option, especially for MSI-H mCRC. The phase III KEYNOTE-177 study demonstrated that rst-line pembrolizumab was associated with signi cant progression-free survival improvement over chemotherapy in MSI-H mCRC (median progression-free survival of 16.5 versus 8.2 months, HR = 0.60; 95% CI 0.45-0.80, P = 0.0002). The bene t of pembrolizumab was consistent in the BRAF V600E-mutant subgroup [27] . In our cohort, very few patients received ICI after chemotherapy failure (n = 3), explained by the period of inclusion. MSI-H status was not reliable for any conclusions due to insu cient data. In a recent study, MSI-H status was associated with signi cantly longer OS in a BRAFmutant mCRC population (n = 194) treated with standard chemotherapies [17] .
The major weakness of our study is related to the differences between the resected and unresected CRLM groups with the bias of less aggressive disease in the resected group. However, this should be counterbalanced by increased liver surgery ability and moreover, the initial resectability or unresectability status might have been subject to variability between the centers. The reasons for non-resectability were not speci ed, and patients could be considered unresectable solely based on the presence of BRAF mutations.
The missing data in our cohort represent an important limitation, and some known prognostic factors,

Declarations
Ethics approval and consent to participate: The study was conducted according to the ethical standards in line with the French regulation. French Data Protection Authority (CNIL agreement n° DEC18-409 (2018_01)) provided a waiver of informed consent for this retrospective study and permitted the publication of anonymized data.
Non-deceased patients at the time of the study were informed and declared non-opposition to medical data collection and the data were anonymously analyzed.

Consent for publication:
French Data Protection Authority (CNIL agreement n° DEC18-409 (2018_01)) provided a waiver of informed consent for this retrospective study and permitted the publication of anonymized data.
Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests :
The authors declare that they have no competing interests  Overall survival according to primary tumor resection status among unresected CRLM group (n = 56)