CRLM indicates colorectal liver metastases; mut, mutated genotype; and wt, wild-type genotype.
For wtKRAS/wtBRAF vs mutKRAS/wtBRAF, P = .002; mutKRAS /wtBRAF vs V600E BRAF, P = .008. mut indicates mutated genotype; wt, wild-type genotype.
eTable 1. Characteristics of the Entire Cohort
eTable 2. Baseline Differences Between wtBRAF/wtKRAS vs BRAF V600E vs BRAF non-V600E Mutations
eFigure. Kaplan-Meier Estimates of Overall Survival (OS)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Margonis GA, Buettner S, Andreatos N, et al. Association of BRAF Mutations With Survival and Recurrence in Surgically Treated Patients With Metastatic Colorectal Liver Cancer. JAMA Surg. 2018;153(7):e180996. doi:10.1001/jamasurg.2018.0996
What is the prognostic association of BRAF mutations with survival and recurrence in patients with metastatic colorectal liver cancer, and how does it compare with KRAS mutations?
In this study of 853 patients with colorectal liver metastases, those with a mutant BRAF/wild-type KRAS genotype more commonly were female and 65 years or older, had right-sided primary tumors, and presented with metachronous liver metastasis. V600E but not non-V600E BRAF mutation was associated with worse overall and disease-free survival, and V600E BRAF mutations had a stronger association with overall and disease-free survival than KRAS mutations.
The presence of the BRAF V600E mutation was associated with worse prognosis and increased risk of recurrence and was the strongest prognostic determinant in the overall cohort.
BRAF mutations are reportedly associated with aggressive tumor biology. However, in contrast with primary colorectal cancer, the association of V600E and non-V600E BRAF mutations with survival and recurrence after resection of colorectal liver metastases (CRLM) has not been well studied.
To investigate the prognostic association of BRAF mutations with survival and recurrence independently and compared with other prognostic determinants, such as KRAS mutations.
Design, Setting, and Participants
In this cohort study, all patients who underwent resection for CRLM with curative intent from January 1, 2000, through December 31, 2016, at the institutions participating in the International Genetic Consortium for Colorectal Liver Metastasis and had data on BRAF and KRAS mutational status were retrospectively identified. Multivariate Cox proportional hazards regression models were used to assess long-term outcomes.
Hepatectomy in patients with CRLM.
Main Outcomes and Measures
The association of V600E and non-V600E BRAF mutations with disease-free survival (DFS) and overall survival (OS).
Of 853 patients who met inclusion criteria (510 men [59.8%] and 343 women [40.2%]; mean [SD] age, 60.2 [12.4] years), 849 were included in the study analyses. Forty-three (5.1%) had a mutated (mut) BRAF/wild-type (wt) KRAS (V600E and non-V600E) genotype; 480 (56.5%), a wtBRAF/wtKRAS genotype; and 326 (38.4%), a wtBRAF/mutKRAS genotype. Compared with the wtBRAF/wtKRAS genotype group, patients with a mutBRAF/wtKRAS genotype more frequently were female (27 [62.8%] vs 169 [35.2%]) and 65 years or older (22 [51.2%] vs 176 [36.9%]), had right-sided primary tumors (27 [62.8%] vs 83 [17.4%]), and presented with a metachronous liver metastasis (28 [64.3%] vs 229 [46.8%]). On multivariable analysis, V600E but not non-V600E BRAF mutation was associated with worse OS (hazard ratio [HR], 2.76; 95% CI, 1.74-4.37; P < .001) and DFS (HR, 2.04; 95% CI, 1.30-3.20; P = .002). The V600E BRAF mutation had a stronger association with OS and DFS than the KRAS mutations (β for OS, 10.15 vs 2.94; β for DFS, 7.14 vs 2.27).
Conclusions and Relevance
The presence of the V600E BRAF mutation was associated with worse prognosis and increased risk of recurrence. The V600E mutation was not only a stronger prognostic factor than KRAS but also was the strongest prognostic determinant in the overall cohort.
During the past 2 decades, genetic predictors of prognosis have been used with increasing frequency for patients with colorectal liver metastasis (CRLM).1 Several investigators have reported on the prognostic value of somatic mutations in the KRAS (OMIM 190070) in patients with resectable CRLM.2-6 Although only 2 studies had assessed KRAS mutational status as a prognostic factor in the 1990s, at least 14 additional studies have been published since 2000.7,8 A 2015 meta-analysis demonstrated the adverse effect of KRAS mutations on prognosis by pooling survival data from 326 patients with KRAS-mutated tumors.9
BRAF (OMIM 164757) mutations affect the same signaling pathway as KRAS mutations, the most important difference being that the genetic product of a BRAF-mutated gene exerts its influence downstream from KRAS.10 Although these similarities suggest a possible role for BRAF as a prognostic indicator of CRLM, this hypothesis has not been well studied because of the low (2%-4%) incidence of BRAF mutations in resected CRLM compared with the 30% to 40% incidence of KRAS mutations. The largest relevant study conducted to date (a multi-institutional study from Italy)11 identified only 12 BRAF mutations in a total population of 309 patients who underwent surgical resection of CRLM. According to the most recent meta-analysis,7 only 4 studies have reported overall survival according to BRAF mutational status, and disease-free survival has been examined by only a single study. Of importance, all these studies combined included only 22 patients with BRAF mutations. Although BRAF mutations were associated with adverse prognosis, the sample was too small to allow for a thorough statistical analysis.
Consequently, limited information exists regarding the association of BRAF mutations with the prognosis of resectable CRLM. To increase sample size and mitigate the limitations of previous analyses, an international, multi-institutional consortium was organized to explore the effects of BRAF in patients with CRLM (International Genetic Consortium for Colorectal Liver Metastasis [IGCLM]). Based on this collaborative effort, we aimed to investigate the clinical profile of patients with tumors with BRAF mutations and assess the prognostic association of BRAF mutations with survival and recurrence independently and compared with KRAS mutations. Last, in line with previous work on metastatic colorectal cancer (mCRC), we investigated whether different BRAF mutations (V600E vs non-V600E) may also have a distinct association with prognosis.
Owing to the multi-institutional nature of this study, all examined hypotheses and variables of interest were determined in advance. All patients who underwent curative-intent surgery for CRLM from January 1, 2000, through December 31, 2016, at 7 academic institutions participating in the IGCLM (Johns Hopkins University, Baltimore, Maryland; Stanford University School of Medicine, Stanford, California; Digestive Disease Institute, Cleveland Clinic, Cleveland, Ohio; University of Berlin–Charité, Berlin, Germany; Medical University of Vienna, Vienna, Austria; Medical University of Graz, Graz, Austria; and Haukeland University Hospital, Bergen, Norway) and had available data on BRAF and KRAS mutation status were identified. The IGCLM was officially established when the institutional review board of Johns Hopkins University approved the current study. The institutional review boards of the 7 participating centers approved the study. Owing to the retrospective nature of the study, all institutional review boards waived the requirement for obtaining informed consent.
We collected standard demographic, clinicopathologic, and genetic variables. These variables included age, sex, and characteristics of the primary colorectal tumor, including the American Joint Committee on Cancer T stage, primary tumor location (left vs right colon), and the presence or absence of lymph node metastasis. Information on the following preoperative factors was also recorded and analyzed: receipt of preoperative chemotherapy, preoperative carcinoembryonic antigen levels, and synchronous (<6 months) vs metachronous presentation of liver disease. In addition, we collected data on CRLM tumor size, number of tumors, presence of extrahepatic disease, margin status (R1 was defined as microscopically positive resection margins), data on somatic mutations (KRAS and BRAF mutation status assessed in the primary tumor or the corresponding liver lesions), and the administration of postoperative therapy. Overall survival (OS) was defined as the interval from curative-intent liver resection until death or last follow-up. Similarly, disease-free survival (DFS) was calculated from the date of resection until the first radiologic or pathologic evidence of recurrence or, in the case of no recurrence, until the date of the last follow-up.
Genomic DNA was isolated from primary colorectal cancer (CRC) or CRLM tissue specimens and was used as a template for sequencing the BRAF gene locus (V600E and non-V600E mutations) and KRAS codons 12, 13, and 61 using standard techniques previously described.3 Patients from Haukeland University Hospital only underwent sequencing of KRAS exon 2 (harboring codons 12 and 13) and BRAF. Of note, the data on BRAF mutations from Haukeland University Hospital have been previously reported.12 We elected to use primary CRC or CRLM tissue to determine KRAS and BRAF mutational status; this decision was based on a number of studies that have consistently demonstrated a high concordance rate for KRAS and BRAF mutational status (>90%) between primary and metastatic lesions.13-16
We estimated differences between categorical values using the χ2 test or Fisher exact tests, whereas differences between continuous values were assessed using the Mann-Whitney or Kruskal-Wallis test, as appropriate. Survival estimates for the study population were generated using the Kaplan-Meier method. A Cox proportional hazards regression model was used to assess the association of several variables with OS and DFS. Variables that were statistically significant in univariable analysis (P < .2) were retained in the multivariable model. For the multivariable analysis, multiple imputations were performed using the mice package for R software (version 3.3.1; https://cran.r-project.org/). The prognostic power of independent factors was assessed by calculating the β coefficient as previously described.17 All analyses were performed using SPSS (version 22.0; IBM Corp) and R (version 3.3.1). All tests were 2-sided, and P < .05 defined statistical significance.
Of 853 patients who met inclusion criteria (510 men [59.8%] and 343 women [40.2%]; mean [SD] age, 60.2 [12.4] years), 4 patients with double BRAF-KRAS mutations were excluded, resulting in a final study population of 849 adult patients in the study cohort (Figure 1). A total of 326 patients demonstrated a wild-type (wt) BRAF/mutated (mut) KRAS genotype; 480, wtBRAF/wtKRAS genotype; and 43, mutBRAF/wtKRAS genotype. Among the patients with the mutBRAF/wtKRAS genotype, 33 had a V600E substitution mutation, 6 had a non-V600E mutation, and 4 had a nonspecified variant of BRAF mutation. Clinicopathologic, genetic, and treatment-related characteristics of the entire cohort are summarized in eTable 1 in the Supplement.
The baseline characteristics of patients with wtBRAF/wtKRAS and mutBRAF/wtKRAS genotypes are presented in Table 1. Among those with available data, patients with a BRAF mutation were significantly more likely to be 65 years or older (22 [51.2%] vs 176 [36.9%]) and female (27 [62.8%] vs 169 [35.2%]). Primary CRC tumors in patients with the mutBRAF/wtKRAS genotype were also significantly more likely to be right sided (27 [62.8%] vs 83 [17.4%]) and of a more advanced T stage (41 [95.3%] vs 381 [81.9%]); metachronous liver metastases were significantly more common in this patient group (28 [64.3%] vs 229 [46.8%]).
Subsequently, we compared the baseline characteristics of the wtBRAF/wtKRAS group with each of the 2 major subgroups of the mutBRAF/wtKRAS group (eTable 2 in the Supplement). Patients with non-V600E mutations were more similar to patients with wtBRAF genotypes compared with patients who had the V600E BRAF mutation. Specifically, all 6 patients with a non-V600E mutation were younger than 65 years, and 3 (50.0%) were female, with a similar pattern being observed in patients with wtBRAF genotypes. The location of primary CRC was evenly distributed among patients with non-V600E mutations (2 of 6 rectal, 2 of 6 right sided, and 2 of 6 left sided). Synchronous liver metastases were significantly more common in patients with non-V600E mutations (4 of 6 [66.7%]) compared with patients with the V600E BRAF mutation (10 of 33 [30.3%]).
At a median follow-up of 28.3 months (interquartile range [IQR], 13.5-50.7 months), 377 of 853 patients (44.2%) had died. The 1-year OS rate was 87.5%; 3-year OS rate, 61.6%; and 5-year OS rate, 43.2%. On univariable analysis, the presence of a BRAF mutation (all subtypes included) was associated with significantly worse OS (eFigure, A, in the Supplement) (P = .003). However, although the presence of a V600E mutation was associated with significantly worse OS (hazard ratio [HR], 2.39; 95% CI, 1.53-3.72; P < .001), survival among patients with non-V600E mutations did not differ from that among patients with wtBRAF genotypes (eFigure, B, in the Supplement) (HR, 1.34; 95% CI, 0.43-4.19; P = .61).
The univariable and multivariable analyses of OS are summarized in Table 2. The following factors were independently associated with OS on multivariable analysis: 65 years or older, primary tumor lymph node metastasis, prehepatectomy chemotherapy, carcinoembryonic antigen level of more than 8.5 ng/mL (to convert to micrograms per liter, multiply by 1.0), resection of extrahepatic disease, synchronous liver metastasis, tumor size, presence of multiple CRLMs, positive surgical margin, presence of KRAS mutation, postoperative chemotherapy, and presence of BRAF mutation.
The presence of non-V600E BRAF mutations was not associated with OS in the univariable (HR, 1.34; 95% CI, 0.43-4.19; P = .61) or multivariable (HR, 1.75; 95% CI, 0.54-5.60; P = .35) analysis. However, the presence of the V600E BRAF mutation (HR, 2.76; 95% CI, 1.74-4.37; P < .001) was the strongest prognostic factor for OS in the multivariable analysis.
The prognostic association of the V600E mutation, as modeled by the β coefficient, was almost twice as large as that of the second strongest prognostic factor and more than 3 times greater than that of the presence of a KRAS mutation (β for V600E vs KRAS, 10.15 × 10−1 vs 2.94 × 10−1). Kaplan-Meier OS curves according to KRAS and BRAF mutation status (wtBRAF/wtKRAS vs wtBRAF/mutKRAS vs V600E BRAF mutation) are displayed in Figure 2.
During the study period, 473 of 849 patients (55.7%) developed recurrence or metastasis. The univariable and multivariable analyses of DFS are summarized in Table 3. The following variables were independently associated with DFS on multivariable analysis (Table 3): primary tumor lymph node metastasis, primary rectal cancer, prehepatectomy chemotherapy, resection of extrahepatic disease, presence of multiple CRLMs, positive surgical margin, presence of KRAS mutation, and postoperative chemotherapy.
Similar to the OS analysis, BRAF mutation (all subtypes included) was associated with an increased risk of recurrence in the multivariable analysis (HR, 1.62; 95% CI, 1.07-2.47; P = .02). When the molecular subtypes of BRAF mutation were examined separately, the presence of the V600E BRAF mutation (HR, 2.04; 95% CI, 1.30-3.20; P = .002) was the strongest prognostic factor of DFS identified in the multivariable analysis. This was not the case for non-V600E mutations (HR, 0.67; 95% CI, 0.22-2.06; P = .48). Similar to OS, the prognostic association of the V600E mutation, as modeled by the β coefficient, was almost twice as large as that of the second strongest prognostic factor and more than 3 times greater than that of a KRAS mutation (β for V600E vs KRAS, 7.14 × 10−1 vs 2.27 × 10−1).
In this international collaborative effort, we assembled the largest cohort, to our knowledge, of surgically treated patients with CRLM and BRAF mutations to be reported in the literature to date. In fact, the number of patients with BRAF mutations in the present study (n = 47) exceeded the cumulative population of the most recent meta-analysis (n = 22) by a factor greater than 2.7 The presence of a BRAF mutation was a negative independent prognostic factor for survival and recurrence, thus confirming earlier, preliminary reports18 as well as the largest study to date conducted by Schirripa et al.11 Although the number of patients with non-V600E BRAF mutations can be considered to be too small to draw definitive conclusions, our results indicate that different BRAF mutations may have a distinct association with survival. Specifically, the V600E mutation was associated with worse prognosis, whereas the presence of non-V600E mutations was not associated with significantly different outcomes compared with tumors with wtBRAF genotypes (for OS and DFS). Nonetheless, the latter finding should be interpreted with caution because of the small number of patients in the non-V600E subgroup. Of importance, our analysis indicated that the BRAF V600E mutation may be not only a stronger prognostic factor than KRAS, thus confirming the distinct biological features of the 2 mutations, but also the strongest determinant of prognosis in patients with resectable CRLM. To our knowledge, this is the first time that such results are reported in an exclusively surgical cohort.
The incidence of BRAF mutation in our study population was 5.5%, in line with previous reports on resected CRLM.11,18,19 Of interest, BRAF mutations are reported to occur more frequently in mixed mCRC cohorts (5%-11%) that include patients with resectable and unresectable metastatic disease.20-22 This disparity may reflect the inherent biological aggressiveness of BRAF mutations, which leads to a decreased incidence of liver-limited disease and a higher likelihood of multiorgan involvement, thus precluding curative resection for many patients.23 In our cohort, patients with a BRAF mutation more frequently were female and 65 years or older, had right-sided primary tumors, and were more likely to present with metachronous liver metastases. The latter finding is interesting because the reverse association has been observed in cohorts that include patients who are not candidates for resection.22
Patients with mutBRAF/wtKRAS genotypes had a median overall survival of only 26 months compared with 60 months for patients with wtBRAF/wtKRAS tumors. Of importance, the study cohort had sufficient size to allow us to control for pertinent prognostic factors with the aid of multivariable analysis. These findings contribute significantly to the literature because the previous 4 studies on the implications of BRAF mutations in CRLM11,18,19,24 generated conflicting results. Our findings are in line with the study by Schirripa et al11 and with reports from even larger patient populations with unresectable mCRC.20-22,25,26 Furthermore, to our knowledge, this study is only the second to date to investigate the association of BRAF mutations with recurrence after CRLM resection.11 The median time to recurrence was only 9.9 months, consistent with mCRC studies that reported a similar, poor median DFS of 5.7 months.
Although the presence of the V600E BRAF mutation was strongly associated with inferior prognosis, non-V600E mutations demonstrated no significant association with prognosis. Although these results have not been reported previously in a surgically treated CRLM cohort, they are consistent with those of 2 recent studies in unresectable mCRC.27,28 Cremolini et al27 demonstrated that patients with mCRC harboring BRAF codon 594– and 596–mutated tumors (n = 10) had better OS than did patients with the V600E BRAF mutation (n = 77). A subsequent collaborative study from the Mayo Clinic and MD Anderson Cancer Center28 confirmed these results in the largest cohort of patients with mCRC and non-V600E BRAF mutations reported to date. Reports from basic and translational science29-34 provide a possible explanation for these findings by suggesting that non-V600E BRAF mutations may confer only intermediate (at best) kinase activity compared with an increase in kinase activity as high as 700-fold in the presence of V600E mutations.
After characterizing the relative prognostic association of different mutations in the BRAF gene, we used a β coefficient analysis to compare the relative prognostic weight of the most potent BRAF mutation, namely, V600E, with that of KRAS mutations. V600E had more than 3 times the prognostic weight of a KRAS mutation. This finding contradicts previous reports that mutations in KRAS and BRAF may have similar phenotypic implications.35 Our results suggest that BRAF and KRAS mutations should not be used interchangeably as markers of aggressive tumor biology. Furthermore, our study suggests that the presence of the V600E BRAF mutation is associated with poor prognosis and may serve as a more useful tool for preoperative patient selection than KRAS mutation status. Although this clinical study is the first, to our knowledge, to compare KRAS with V600E BRAF mutations directly, the results are consistent with those of preclinical studies and can be readily interpreted.36 Both of these somatic mutations constitutively activate the extracellular signal-regulated kinase pathway that in turn phosphorylates and regulates the functions of numerous cellular components. However, V600E BRAF–mutated cells show a 138-fold increase in oncogenic activity compared with wtBRAF, thus suggesting that the oncogenic activity of the V600E BRAF mutation is greater than that of the KRAS G12V point mutation.37,38 Because the G12V mutation is the most prognostic among all point mutations in CRLM, the V600E mutation would likely have a stronger association with survival than other KRAS mutations.2
Our analysis is limited by the retrospective nature of the study and its exclusive focus on patients with surgically resectable disease. As such, a degree of selection bias was largely unavoidable. However, because of the scarcity of BRAF mutations, assembling a prospective cohort of this size would be difficult. Second, although our study was comparable in terms of cohort size to a 2015 study,27 in general, the number of patients with non-V600E BRAF mutations is relatively small. Although a type II error is a strong possibility given the small size of the non-V600E subgroup, results similar to ours have been reported by 2 adequately powered studies in BRAF-mutated, unresectable mCRC.27,28 With respect to follow-up, serum carcinoembryonic antigen measurements and radiologic imaging were performed in all participating centers every 3 to 6 months during the first 2 to 3 years after surgery and every 6 months or annually thereafter. However, because the timing and duration of surveillance may also be influenced by patient risk profiles, some heterogeneity in posthepatectomy surveillance was likely present in our multi-institutional cohort. Another limitation of the study is the lack of data on the mismatch repair system. As shown in studies on primary CRC, the mismatch repair system is important in the interpretation of BRAF mutations.39 In addition, detailed information on systemic therapy, especially with respect to chemotherapy regimens and treatment cycles, were not available. Of interest, we found that prehepatectomy chemotherapy was independently associated with worse OS. Although previous studies from Memorial Sloan-Kettering Cancer Center40 and MD Anderson Cancer Center41 have reported similar findings, the retrospective design of the study, the heterogeneity of the chemotherapeutic protocols used, and the lack of randomization preclude any reliable interpretation.
The size of this study population with BRAF-mutated tumors exceeded that of previous studies, and the study was adequately powered to indicate that patients with BRAF mutations may be at an increased risk of recurrence and death. A novel (in surgical CRLM cohorts) finding is that the V600E mutation alone (rather than V600E and non-V600E mutations together) may confer a distinctly aggressive clinical phenotype, thus driving the adverse outcomes associated with BRAF mutation. However, this finding needs to be interpreted with great caution because of the small number of patients with non-V600E mutations in the cohort. Although 2 previous studies of unresectable mCRC27,28 reported similar results, additional confirmation from larger cohorts is needed. No recommendations can be made regarding the selection of surgical or medical treatment for patients with BRAF-mutated CRLM based on our findings. Future cohort studies free of selection bias that will incorporate a complete denominator of patients with CRLM (namely, those with resected, unresected, and unresectable disease) and appropriately designed clinical trials are warranted to address this issue.
Accepted for Publication: February 17, 2018.
Corresponding Author: Matthew J. Weiss, MD, Johns Hopkins University, 600 N Wolfe St, Halsted 608, Baltimore, MD 21287 (email@example.com).
Published Online: May 16, 2018. doi:10.1001/jamasurg.2018.0996
Author Contributions: Drs Margonis and Buettner contributed equally to this study. Dr Weiss had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Margonis, Sasaki, Pawlik, Cameron, Aucejo, Wolfgang, Weiss.
Acquisition, analysis, or interpretation of data: Margonis, Buettner, Andreatos, Kim, Wagner, Sasaki, Beer, Schwarz, Løes, Smolle, Kamphues, He, Pawlik, Kaczirek, Poultsides, Lønning, Burkhart, Gerger, Kreis, Weiss.
Drafting of the manuscript: Margonis, Buettner, Andreatos, Pawlik.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Margonis, Buettner, Kim, Sasaki.
Administrative, technical, or material support: Wagner, Beer, Schwarz, Kamphues, Pawlik, Kaczirek, Poultsides, Lønning, Burkhart, Gerger, Wolfgang.
Study supervision: Pawlik, Cameron, Aucejo, Kreis, Wolfgang, Weiss.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Bodossaki Foundation (Dr Margonis) and the Drs Keith and Valda Kaye Research Fund and the Carolyn Pastorini Research Fund (Dr Weiss).
Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: Dr Pawlik is a deputy editor of JAMA Surgery. He was not involved in the editorial evaluation or decision to accept this article for publication.