Comprehensive Landscape of BRAF Variant Classes, Clonalities, and Co-Mutations in Metastatic Colorectal Cancer Using ctDNA Profiling

Simple Summary Patients with atypical (nonV600) BRAF mutations represent a unique category of metastatic colorectal cancer patients. In the past, these patients were considered to be similar to those with the more common BRAFV600E mutation. However, additional investigation confirmed patients with atypical BRAF mutations have a distinct molecular make up, clinical course, and treatment response to both chemotherapy and targeted therapy, compared to those of patients with BRAFV600E mutations. Here, we report the key characteristics of patients with atypical BRAF mutations identified from a large circulating tumor DNA database and a real-world clinical cohort, highlighting important differences, such as the presence of additional mutations and related survival outcomes. These findings support the need for dedicated research efforts to understand the intricacies of atypical BRAF mutations in colon cancer and promote the discovery of new therapies for these patients. Abstract Although V600E accounts for the majority of the BRAF mutations in metastatic colorectal cancer (mCRC), non-V600 BRAF variants have been shown in recent years to represent a distinct molecular subtype. This study provides a comprehensive profile of BRAF variants in mCRC using a large genomic database of circulating tumor DNA (ctDNA) and analyzing clinical outcomes in a cohort of patients with atypical (non-V600) BRAF variants (aBRAF; class II, class III, unclassified). Overall, 1733 out of 14,742 mCRC patients in the ctDNA cohort had at least one BRAF variant. Patients with atypical BRAF variants tended to be younger and male. In contrast to BRAFV600E, BRAF class II and III variants and their co-occurrence with KRAS/NRAS mutations were increased at baseline and especially with those patients predicted to have prior anti-EGFR exposure. Our clinical cohort included 38 patients with atypical BRAF mCRC treated at a large academic referral center. While there were no survival differences between atypical BRAF classes, concurrent RAS mutations or liver involvement was associated with poorer prognosis. Notably, patients younger than 50 years of age had extremely poor survival. In these patients, the high-frequency KRAS/NRAS co-mutation and its correlation with poorer prognosis underlines the urgent need for novel therapeutic strategies. This study represents one of the most comprehensive characterizations to date of atypical BRAF variants, utilizing both ctDNA and clinical cohorts.

BRAF mutations represent one of the most common aberrations in human malignancies [13], including mCRC [14].BRAF is a serine threonine kinase downstream of RAS, a part of the mitogen-activated protein kinase (MAPK) signaling pathway.The most common BRAF mutation, present in 7-10% of patients [13,14], occurs at codon 600, with a valine to glutamic acid change (c.1799T>A or p.V600E).This mutation results in the RAS-independent constitutive activation of MAPK, with the promotion of uncontrolled tumor cell proliferation and metastases formation, which lead to poorer outcomes in terms of patient survival [13,14].BRAF V600E and RAS mutations are predominantly mutually exclusive [3,15,16].
Atypical, non-V600 BRAF (aBRAF) mutations have been recognized in recent years as a unique molecular subset, partly due to the increased use of expanded molecular profiling and circulating tumor DNA (ctDNA) analysis in the management of mCRC.Preclinical data specifically characterized aBRAF mutations into class II or class III designations, based on their underlying signaling biology [17].Class II aBRAF mutants signal via constitutive dimerization and are RAS-independent [17], while Class III aBRAF mutants are characterized by low or absent kinase activity and are associated with RAS activation via receptor tyrosine kinase signaling [17].
Previous studies reported that aBRAF mutations are present in approximately 2.2% of patients with mCRC, with an improved median OS of 60.7 months compared to 11.7 months for patients with BRAF V600E mutations [18].However, considering the chronicity of aBRAF mCRC, and the fact that patients still succumb to the disease, novel treatments are still needed for personalized treatment and to preserve patients' quality of life.Furthermore, aBRAF mCRC has a distinct, antithetical clinical profile from BRAF V600E mCRC.Most patients present microsatellite stable (MSS) disease, left-sided primary tumors, lower-grade histology, and the clinical presentations of non-peritoneal spread [19,20].RAS co-mutations may also be present [19,20].
To date, there are no specific guidelines for the management of patients with aBRAF mCRC.Early-phase clinical trials are recommended after the failure of traditional systemic chemotherapy.In particular, anti-EGFR therapy is known to elicit a poor response in BRAF V600E patients [21].The potential utility of EGFR inhibitors in aBRAF patients is less clear; however, a recent study showed the response to anti-EGFR treatment to be poor in both class II and class III patients, and that these mutations were significantly more common in patients previously treated with EGFR inhibitors, suggesting that they may represent a novel resistance mechanism [22].
In this study, we aimed to provide a comprehensive landscape of BRAF mutations in mCRC, analyzing the distribution of different mutation classes, their clonalities, and the frequency of co-mutation with other genes of interest through ctDNA profiling.Moreover, we studied how these different molecular characteristics correlated with clinical outcomes in our MD Anderson cohort to confirm previous findings, suggesting that the RAS-aBRAF co-mutated phenotype may represent a more aggressive subclass of mCRC than those previously described.

Patient Population
We classified all BRAF mutations in this study based on the published preclinical study as summarized above [17].The first cohort analyzed for this study included patients with mCRC in the Guardant Health (GH) database from September 2014 to May 2021.These data were used to perform a retrospective review.The Guardant360 targeted sequencing assay is a blood-based, "liquid" biopsy that identifies single nucleotide variants (SNVs), indels, copy number amplifications, and fusions within the protein-coding regions of up to 83 genes.Treatment history was not available for this cohort; therefore, a previously validated and highly specific score was used to predict the anti-EGFR exposure status [22,23].Briefly, patients were considered anti-EGFR exposed, if the ctDNA analysis revealed the specific molecular abnormalities consistent with a previous exposure [23].The second cohort (clinical cohort) included 38 patients with aBRAF mCRC (all 38 patients were confirmed via tissue next generation sequencing; 10 pts with ctDNA testing) who received treatment at MD Anderson Cancer Center between June 2018-January 2022.These patients were analyzed based on their treatment history and overall survival (OS).Next generation sequencing is a tissue-based assay performed with an in-house panel at MD Anderson Cancer Center, covering an estimated 600 genes, by utilizing patient primary tumor or metastatic site biopsies obtained as standard of care.

Data and Statistical Analysis
BRAF amplifications and synonymous variants were not included in this study.A variant was defined as clonal, if the allele frequency (VAF) was greater than 50% of the highest somatic VAF in the sample; otherwise, it was defined as subclonal.When multiple samples were available for a patient, only the earliest tested sample was included in the analysis.
Fisher's exact test was used in the analysis comparing molecular classes and variant groups.OS was defined as time from mCRC diagnosis to the date of death or last follow-up.Kaplan-Meier survival curves and long-rank tests were used to compare OS between patient groups.Statistical analyses were performed using GraphPad Prism, version 9.3.1 (GraphPad Software, La Jolla, CA, USA).A value of p < 0.05 was considered statistically significant.

Clonality and Anti-EGFR Exposure Score
Due to the lack of treatment history data in the GH cohort, a previously validate scoring system was applied to predict the exposure to anti-EGFR therapy.Patients we divided into two groups: those with predicted prior exposure (n = 3470) and those witho predicted prior exposure (n = 11,272).Among patients with BRAF class II/III/unclassifie variants, the proportion of patients with predicted prior exposure was more than two-fo greater than that in patients predicted to be non-exposed (2.1% vs. 0.8% in class II, 3.7 vs. 1.4% in class III, 6% vs. 2.6% in unclassified, Figure 3).
The BRAF-mutated patients were further grouped into those with clonal BRAF varian and those with subclonal BRAF variants.Most BRAF V600E mutations in patients with no pr dicted anti-EGFR exposure were clonal (Table 4, Figure 3).On the contrary, most aBRA variants in patients with predicted prior exposure were subclonal (Table 4, Figure 3).

Clonality and Anti-EGFR Exposure Score
Due to the lack of treatment history data in the GH cohort, a previously validated scoring system was applied to predict the exposure to anti-EGFR therapy.Patients were divided into two groups: those with predicted prior exposure (n = 3470) and those without predicted prior exposure (n = 11,272).Among patients with BRAF class II/III/unclassified variants, the proportion of patients with predicted prior exposure was more than two-fold greater than that in patients predicted to be non-exposed (2.1% vs. 0.8% in class II, 3.7% vs. 1.4% in class III, 6% vs. 2.6% in unclassified, Figure 3).

Co-Mutations Analysis
Co-mutation analysis was conducted for BRAF, KRAS, NRAS, NF1, ERBB2, PIK3CA, and SMAD4 genes after excluding amplifications and synonymous alterations (Figure 4A,B).aBRAF variants often co-occurred with KRAS mutations, although more frequently in patients with prior anti-EGFR exposure (Figure 4A).Co-occurring KRAS G12C was only noted in one patient with a BRAF class II variant.Concomitant NRAS mutations were seen in 26.9-42.2% of patients with aBRAF variants and predicted the prior anti-EGFR exposure (Figure 4A) and were observed in only 2.7-5.8% of patients without predicted prior anti-EGFR exposure (Figure 4B).Co-mutations in BRAF and four other genes (NF1, ERBB2, PIK3CA, and SMAD4) were also more frequent in patients with predicted anti-EGFR exposure (Figure 4A); however, there was no significant difference between the BRAF V600E and aBRAF variant groups.
Blood-based tumor mutation burden (bTMB) values were available in 258 patients with BRAF variants.The median bTMB values in patients bearing BRAF V600E , class II, III, and unclassified variants were 12.44, 15.79, 12.44, and 31.58mut/MB, respectively (Table 5, Figure 5).The BRAF-mutated patients were further grouped into those with clonal BRAF variants those with subclonal BRAF variants.Most BRAF V600E mutations in patients with no predicted anti-EGFR exposure were clonal (Table 4, Figure 3).On the contrary, most aBRAF variants in patients with predicted prior exposure were subclonal (Table 4, Figure 3).

Co-Mutations Analysis
Co-mutation analysis was conducted for BRAF, KRAS, NRAS, NF1, ERBB2, PIK3CA, and SMAD4 genes after excluding amplifications and synonymous alterations (Figure 4A,B).aBRAF variants often co-occurred with KRAS mutations, although more frequently in patients with prior anti-EGFR exposure (Figure 4A).Co-occurring KRAS G12C was only noted in one patient with a BRAF class II variant.Concomitant NRAS mutations were seen in 26.9-42.2% of patients with aBRAF variants and predicted the prior anti-EGFR exposure (Figure 4A) and were observed in only 2.7-5.8% of patients without predicted prior anti-EGFR exposure (Figure 4B).Co-mutations in BRAF and four other genes (NF1, ERBB2, PIK3CA, and SMAD4) were also more frequent in patients with predicted anti-EGFR exposure (Figure 4A); however, there was no significant difference between the BRAF V600E and aBRAF variant groups.Blood-based tumor mutation burden (bTMB) values were available in 258 patients with BRAF variants.The median bTMB values in patients bearing BRAF V600E , class II, III, and unclassified variants were 12.44, 15.79, 12.44, and 31.58mut/MB, respectively (Table 5, Figure 5).

Clinical Cohort Analysis
Our MD Anderson clinical cohort included 38 patients with aBRAF mCRC.The co hort's median age was 55, 81% patients were Caucasian, and 74% had left-sided primar tumors (45% rectal, 24% sigmoid), with 37% being exposed to at least two lines of therap (Table S1).The most common aBRAF mutation was D594G (class III).The median follow up time was 23.8 months (mo).The most common site of metastases involved was th liver.While there were no differences in OS between aBRAF classes, there was a significan difference in OS in patients with RAS co-mutations (28.3 mo, p = 0.05, Figure 6A) or live involvement (28.8 mo, p = 0.02, Figure 6B).Patients < 50 years of age had extremely poo survival with an OS of 16.3 mo and an HR of 7.51 (95% CI = 1.82-31.01,p = 0.005, Figur 6C).Treatment with anti-EGFR (Figure 6D) or the use of metastasectomy (Figure 6E) di not result in statistically significant improvement in survival.

Clinical Cohort Analysis
Our MD Anderson clinical cohort included 38 patients with aBRAF mCRC.The cohort's median age was 55, 81% patients were Caucasian, and 74% had left-sided primary tumors (45% rectal, 24% sigmoid), with 37% being exposed to at least two lines of therapy (Table S1).The most common aBRAF mutation was D594G (class III).The median follow-up time was 23.8 months (mo).The most common site of metastases involved was the liver.While there were no differences in OS between aBRAF classes, there was a significant difference in OS in patients with RAS co-mutations (28.3 mo, p = 0.05, Figure 6A) or liver involvement (28.8 mo, p = 0.02, Figure 6B).Patients < 50 years of age had extremely poor survival with an OS of 16.3 mo and an HR of 7.51 (95% CI = 1.82-31.01,p = 0.005, Figure 6C).Treatment with anti-EGFR (Figure 6D) or the use of metastasectomy (Figure 6E) did not result in statistically significant improvement in survival.

Discussion
Our study supports that aBRAF mCRC is indeed a distinct subtype of colorectal cancer with the most notable findings of ctDNA analysis, confirming that aBRAF (BRAF class II, III, unclassified variants) and co-occurrence with KRAS/NRAS mutations was increased at baseline and especially in patients predicted to have prior anti-EGFR exposure, in contrast to BRAF V600E .RAS co-mutation status was not previously consistently reported for aBRAF and highlights a unique molecular phenotype.We hypothesized this molecular

Discussion
Our study supports that aBRAF mCRC is indeed a distinct subtype of colorectal cancer with the most notable findings of ctDNA analysis, confirming that aBRAF (BRAF class II, III, unclassified variants) and co-occurrence with KRAS/NRAS mutations was increased at baseline and especially in patients predicted to have prior anti-EGFR exposure, in contrast to BRAF V600E .RAS co-mutation status was not previously consistently reported for aBRAF and highlights a unique molecular phenotype.We hypothesized this molecular phenotype would translate into a more aggressive clinical course, which was supported by the results of our MD Anderson clinical cohort.We identified aBRAF mCRC to be more common in younger patients with a median age of 55, more likely to have left-sided (sigmoid/rectal) primaries and liver metastases with worse outcomes for co-mutation with RAS and young onset.Anti-EGFR exposure and metastasectomy did not statistically improve survival outcomes in this subset.These findings reiterate the current void in treatment options for aBRAF mCRC and supports the need for novel therapeutic development.
As of 2023, there are no specific guidelines for the management of patients with aBRAF mCRC, with early-phase clinical trials generally recommended after the failure of traditional systemic 5-FU-based chemotherapy.Additionally, the efficacy of EGFR receptor antibodies remains inconclusive with data, suggesting that response may in fact be driven by the underlying aBRAF class [17,[22][23][24][25]. Previous reports noted that class III aBRAF could benefit from anti-EGFR exposure; however, a subsequent study highlighted a lack of durable response and suggested class II aBRAF as a negative predictive biomarker for anti-EGFR receptor therapy [22][23][24][25].In the present study, we showed significant differences between BRAF V600E and class II/III/unclassified variants in terms of prevalence and clonality, following the previous anti-EGFR treatment.The most common aBRAF variants that are usually encountered in clinical practice are class II G469A and class III D594G [22], which are consistent with our ctDNA findings.It should be noted that the present study showed a notable prevalence for unclassified variants, whose biology is unclear.Therefore, further investigation into the biology of these alterations is warranted.Our results show that both class II and III aBRAF mutations are associated with concomitant RAS aberrations at a frequency higher than that reported previously [18,19,25].Prior anti-EGFR exposure significantly increased sub-clonal class II, III, and unclassified aBRAF mutation frequencies.This finding supports the hypothesis that aBRAF mutations may represent a resistance mechanism following EGFR inhibition in CRC [22].
In the current series, our MD Anderson cohort suggested EGFR inhibition in combination with chemotherapy may have limited efficacy in aBRAF mCRC.Although ongoing studies will further investigate this prospectively, such as the EPOC1703 study where both aBRAF class III patients who are EGFR receptor antibody naïve or refractory will be included and treated with the BEACON regimen [26], our data reveal that a significant proportion of patients with aBRAF have concomitant RAS mutations, thus limiting their eligibility and subsequent response to such a targeted approach.The data from our clinical cohort show that patients with double mutations in aBRAF and RAS, noted in both class II and class III aBRAF, have in fact inferior OS when compared to RAS wild-type (wt) aBRAF patients.Our results suggest that this double mutation phenotype represents in fact a more aggressive subset of mCRC, with a current void in the available clinical trials.Considering the high frequency of co-mutations and the unfavorable prognosis shown by double-mutated patients, novel approaches are therefore urgently needed.Interestingly, the analysis of bTMB data revealed that patients affected by all considered BRAF variants presented a median TMB value higher than 10 mut/MB, which may have implications for exposure to immune checkpoint inhibitors (ICI) therapy.However, these findings need to be interpreted with caution as blood-based TMB is typically higher than tissue-based TMB, with 16 mut/MB in blood roughly correlating to 10 mut/MB in tissue.Following the results of the KEYNOTE 158 trial [27], in 2020, the FDA approved anti-PD-1 inhibitors for any type of solid tumors with TMB ≥ 10 with tissue NGS.Our findings may be particularly useful for patients with unclassified aBRAF variants, since there are currently no specific therapeutic guidelines for this subtype, and they present the highest median bTMB value (31.58).A recent study suggested 28 mut/MB as a potential cutoff value to determine when ICI therapy may be more likely to be beneficial in CRC via a blood-based TMB measurement, but this needs to be corroborated by further research [28].
Several clinical trials targeting aBRAF mutations are currently ongoing.One therapeutic strategy attempted in these trials is to inhibit the MAPK pathway with novel MEK or ERK inhibitors, alone or combined with RAF inhibitors [29][30][31].The examples of studies pursuing this strategy are NCT02465060, NCT02607813, NCT04249843, NCT02428712, and NCT03839342 [31].Another strategy currently under investigation in early-phase clinical trials utilizes a novel Src homology phosphatase 2 (SHP2) inhibitor in cancers with class III mutations (NCT04045496; NCT03518554).SHP2 is a major scaffold protein downstream of numerous receptor tyrosine kinases, promoting RAS/MAPK signaling in cancers with class III BRAF mutations with concomitant RAS mutations.We have shown that the survival for these patients is much worse than for those presenting aBRAF RAS wt in our MD Anderson cohort.Unfortunately, none of these patients in this cohort had a concomitant KRAS G12C mutation.Therefore, a combination approach with a novel RAF dimer and G12C inhibitor would not be a feasible immediate path forward, while pan RAS inhibitors or checkpoint inhibition may be considered as viable options in the future.
This study presents some limitations.First, this is a retrospective, non-randomized cohort analysis.Second, because of the rarity of atypical aBRAF mutations, only 38 patients were identified for inclusion in our MD Anderson clinical cohort.Third, we were not able to verify if the patients from the ctDNA cohort underwent anti-EGFR therapy as treatment information is not available.Therefore, we applied a previously validated method that allowed to predict if patients were previously treated with anti-EGFR drugs.

Conclusions
In conclusion, we have highlighted clear differences in clonality between patients with atypical, non-V600 mutations and with traditional BRAF V600E mutations.We also summarized key aspects of aBRAF mutations as potential resistance mechanisms in RAS/RAF wt patients treated with anti-EGFR therapy from our ctDNA cohort.These results may help inform future anti-EGFR re-challenge strategies in future clinical trial design.Additionally, for the first time, we report on the co-mutation status as being a characteristic in both class II and class III atypical mutations based on our ctDNA cohort and highlight this feature representing an aggressive subtype of double-mutated mCRC (RAS & aBRAF) as a particularly difficult-to-treat patient population validated in our MD Anderson clinical cohort.Furthermore, it is important to highlight the need for additional research efforts regarding the preclinical characterization of unclassified variants.The unclassified cohort represents a group of patients whose underlying signaling biology is unclear to date, and such information would be informative for treatment decisions and early phase clinical trial triage regarding novel target/agent selection.Anti-EGFR exposure and metastasectomy do not appear to statistically impact survival outcomes in this subset, albeit we had a small cohort available to perform this analysis.However, all of these insights coupled together highlight the critical need for innovative clinical trials that consider these molecular and clinical intricacies.Additional innovative targeted approaches for aBRAF mCRC that address the co-mutation status may provide a viable path forward in this aggressive subset of colorectal cancer.These data represent the foundational framework for understanding the intricacies of aBRAF mCRC and highlight the need for continued dedicated therapeutic development for these unique patients.Informed Consent Statement: This study used an internal cohort created retrospectively with institutional review board approval at the University of Texas MD Anderson Cancer Center (MDACC), and a waiver of written consent was obtained.All databases used were deidentified, and the study adhered to MDACC institutional review board guidelines for deidentified databases.

Figure 1 .
Figure 1.Number of patients expressing class II (A) and class III (B) variants.Clonal variants are presented in blue and subclonal variants in red.

Figure 1 .
Figure 1.Number of patients expressing class II (A) and class III (B) variants.Clonal variants are presented in blue and subclonal variants in red.

Figure 4 .Figure 4 .
Figure 4. Frequency of co-mutations in patients (pts) predicted to have (A) or not have (B) prio exposure to anti-EGFR therapy.

Figure 6 .
Figure 6.Kaplan-Meier survival curves comparing aBRAF patients in the clinical cohort based on RAS co-mutations (A), liver involvement (B), and the age at diagnosis, higher or lower than 50 (C), a history of anti-EGFR treatment (D), and a history of metastasectomy (E).

Figure 6 .
Figure 6.Kaplan-Meier survival curves comparing aBRAF patients in the clinical cohort based on RAS co-mutations (A), liver involvement (B), and the age at diagnosis, higher or lower than 50 (C), a history of anti-EGFR treatment (D), and a history of metastasectomy (E).

Funding:
This research received no external funding.Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of University of Texas MD Anderson Cancer Center (MDACC) (project code: LAB09-0373; approval date: 22 June 2009).

Table 1 .
Summary of patient characteristics from the ctDNA database/GH cohort.

Table 2 .
Clonality of BRAF variants in the ctDNA cohort.

Table 2 .
Clonality of BRAF variants in the ctDNA cohort.

Table 3 .
VAF values for clonal or subclonal variants of the different BRAF classes.

Table 4 .
Clonality of BRAF variants in presence or absence of predicted anti-EGFR exposure.

Table 3 .
VAF values for clonal or subclonal variants of the different BRAF classes.

Table 5 .
Sample size and bTMB median values for the different BRAF classes.

Table 4 .
Clonality of BRAF variants in presence or absence of predicted anti-EGFR exposure.

Table 5 .
Sample size and bTMB median values for the different BRAF classes.