BRAF, a tumor-agnostic drug target with lineage-specific dependencies

Preface

The FDA accelerated approval of dabrafenib in combination with trametinib for unresectable or metastatic BRAF V600E solid tumors is the culmination of two decades of research into the landscape of BRAF mutations in human cancer, the biochemical mechanisms underlying BRAF-mediated tumorigenesis, and the clinical development of selective RAF and MEK kinase inhibitors. While RAF inhibitor-based combinations are effective in most patients with BRAF V600E mutant tumors, drug resistance is common. Here, we discuss the biochemical basis for oncogenic BRAF activation of MAP kinase signaling, pan-cancer and lineage-specific mechanisms of intrinsic, acquired, and adaptive RAF-inhibitor resistance, and novel RAF inhibitors and drug combinations designed to delay the emergence of treatment resistance and/or expand the population of BRAF-mutant patients who benefit from molecularly tailored therapies.

Introduction

Oncogenic BRAF mutations induce MAP kinase (MAPK) pathway activation and are a key mediator of tumorigenesis in multiple cancer types including melanoma, histiocytosis, and thyroid, colorectal, and lung cancers, among others.^1,2 In 2022, the combination of the ATP-competitive RAF kinase inhibitor dabrafenib and the allosteric MEK inhibitor trametinib was awarded a coveted tumor-agnostic accelerated FDA approval for the treatment of patients with unresectable or metastatic BRAF V600E mutant solid tumors who have progressed following prior treatment and have no satisfactory alternative treatment options. In contrast to prior tumor agnostic drug approvals, the FDA authorization specifically excluded patients with colorectal cancer as the BRAF/MEK combination is less clinically effective than BRAF/EGFR inhibitor combinations due to lineage specific adaptive resistance as discussed in detail below.^3,4 This tumor-agnostic approval also does not cover patients whose tumors harbor other BRAF V600 mutant alleles (V600K/R/D), which in laboratory studies are similarly sensitive to dabrafenib. Despite these limitations, the tumor agnostic approval of dabrafenib/trametinib will expand access to these targeted therapies to patients with rare cancers in which disease-specific clinical trials can be challenging to accrue. The tumor-agnostic FDA authorization of the RAF/MEK combination is also further validation of the power of the basket trial design to rapidly modify standard clinical practice across multiple cancer subtypes, and further justification for universal tumor genomic profiling of all cancer patients in need of systemic therapy.⁵

While the dabrafenib/trametinib combination has been shown to be active in multiple cancer types, the likelihood and duration of clinical response varies as a function of tumor subtype. The basis for this variation in RAF inhibitor sensitivity is incompletely understood but is likely the result of lineage-specific differences in co-mutation patterns and the expression of receptor tyrosine kinases and feedback regulators that in sum dictate the early, adaptive rebound in ERK activation that can attenuate RAF inhibitor sensitivity and promote drug resistance. This review will discuss how the underlying mechanisms through which BRAF mutations activate MAPK signaling dictate RAF kinase inhibitor sensitivity, and how insights into the adaptive and selective mechanisms responsible for RAF inhibitor resistance can inform the development more effective combination strategies.

The landscape of BRAF mutations in human cancer and the RAF inhibitor paradox

ARAF, BRAF, and CRAF comprise the RAF family of serine/threonine kinases that serve as critical nodes of the canonical mitogen activated protein kinase (MAPK) cascade.⁶ In non-stimulatory conditions, high RAS-GAP (e.g., NF1) and low RAS-GEF (e.g. SOS1) activity result in low levels of RAS activation, insufficient to induce RAF dimer formation. In this quiescent cell state, inactive wildtype RAF proteins localize to the cytosol as negatively phosphorylated, autoinhibited monomers, often bound to MEK1/2 or scaffolding proteins such as 14–3-3.^7,8 Upon extracellular growth factor binding and activation of receptor tyrosine kinases (RTKs), adaptor proteins (e.g., SHP2, GAB1, GRB2), transduce RTK signals, resulting in activation of SOS, which activates RAS (Figure 1A). Active GTP-bound RAS interacts with wildtype RAF via the RAS binding domain and cysteine-rich domain (collectively annotated as CR1),⁹ resulting in RAF dimerization and activation.

Figure 1. Mechanism of activation of wildtype and Class I/II/II BRAF mutants.

A) BRAF mutants are classified based on their dependence on upstream RTK/RAS activation and dimerization for kinase activation. B) Schematic showing BRAF V600E inhibition by αC-helix IN/DFG OUT RAF inhibitors (left) as compared to paradoxical activation of RAF dimers (right).

For BRAF, residue R509 located at the C-terminal portion of the αC-helix is fundamental for dimer interaction, with the “IN” configuration of the αC-helix favoring dimerization.^6,10 Activated RAF phosphorylates MEK1/2, which upon activation phosphorylate ERK1/2. Activated ERK1/2 in turn phosphorylate downstream effectors that regulate proliferation, survival and multiple other cellular processes.¹¹ To prevent excessive proliferation, activated ERK1/2 also induces negative feedback regulators that limit ERK pathway output. Upstream negative feedback is mediated by rapid ERK-dependent phosphorylation and inactivation of RTKs and SOS1, among other proteins and the slower transcriptional upregulation of NF1 and the SPROUTY (SPRY) family of negative feedback modulators, ultimately resulting in the termination of RAS activation. Downstream negative feedback results from upregulation of the MAPK phosphatases DUSP1/4/6 and ERK-mediated inhibitory phosphorylation of CRAF, MEK1/2, and kinase suppressor of RAS 1 (KSR1), among other mechanisms.^11,12

In 2002, mutations in the BRAF gene were identified in tumors and cell lines by Davies and colleagues at the Sanger Institute.¹ Early cellular studies confirmed that BRAF mutational hotspots such as BRAF V600E could promote transformation, and that inhibition of MEK in BRAF mutant cancer cell lines could induce cell cycle arrest and cell death and inhibit xenograft growth.^13,14 Subsequent broad scale sequencing studies have now defined the landscape of BRAF mutants across common and rare cancer subtypes.^15–18 Analysis of the first 96K tumor samples from the prospective MSK-IMPACT clinical tumor sequencing cohort indicate that somatic mutations in BRAF are present in approximately 4.5% of tumor samples. In the MSK-IMPACT cohort, BRAF mutations were most common in hairy cell leukemia, (~100%, 72/72) and thyroid cancers (43.3%, 473/1093), followed by melanoma (29.4%, 810/2755), histiocytosis (19.4%, 126/648), colorectal cancer (11.4%, 784/6882) and other mature B-cell neoplasms (2.9%, 160/5410) (Figure 2A). Pathogenic germline mutations in BRAF that exhibit lower kinase activity have also been detected in 50–75% of patients with the RASopathy cranio-facial-cutaneous-syndrome, as well as 2% of patients with Noonan/LEOPARD syndrome, a condition in which the incidence of cancer is estimated to be 3.5 fold higher than the general population.^19–21 Germline mutations in BRAF cluster in the N-terminal cysteine-rich CR1 and kinase domains, but are largely distinct from the somatic BRAF mutations most often identified in cancer patients.

Figure 2: Frequency of BRAF mutations in common cancer subtype.

A) Lollipop plot illustrating common BRAF variants in selected cancer types. The structural domains are color coded as outlined in the figure. The sizes of the circle denote the percentage of samples with that variant based on the MSK-IMPACT cohort. Variant names are color coded by BRAF mutant classes based on prior publication.²⁴ Only the most common variants were listed. B) Distribution of BRAF mutant classes in selected detailed cancer types in the MSK-IMPACT cohort.

In depth biochemical and cellular studies of over 30 BRAF mutations have led to categorization of BRAF mutants into one of three functional classes based on their biochemical properties and therapeutic implications.^22–24 Class 1 includes only BRAF codon 600 mutants. BRAF V600E is the most common RAF mutant by far, and the predominant Class I mutant, with BRAF V600K/R/D less commonly observed in cancer patients. The distribution of BRAF mutations varies widely among cancer types with BRAF codon V600 mutations representing 94% of BRAF mutations detected in thyroid cancer, 67% in melanoma, but only a minority of BRAF mutations identified in non–small cell lung cancer (25%) and prostate cancer (1%) (Figure 2B).¹⁶

BRAF, but not ARAF and CRAF, exhibits constitutive phosphorylation of the S455 residue and has a phosphomimetic D448 residue in its negatively charged N-region near the C-terminus. These differences result in higher basal kinase activity and allow for a single amino acid mutation at BRAF codon V600 in the activation segment of the kinase domain to confer constitutive and robust activation of the kinase by mimicking the phosphorylation of T599 and S602 in the activation loop. Additionally, codon 600 BRAF mutations result in kinase activation independent of upstream RAS signaling that is insusceptible to negative feedback by ERK (Figure 1A).^22,25,26 In cells with adequate levels of RAS activation, BRAF V600E mutants function as homo- and heterodimers with wildtype CRAF or ARAF. In contrast, in tumors in which wildtype RAS has been inhibited by ERK-dependent feedback, which is usually the case for tumors with BRAF V600 mutation, BRAF can function as an activated monomer.²⁷ Notably, BRAF V600 mutants are unique in this regard, as the activation of wildtype RAF and all other RAF mutants and fusions functionally characterized to date, requires their homo- or heterodimerization (see discussion of Class 2 and 3 mutants in the following section).^22,28

Efforts to identify inhibitors of MAPK pathway signaling that could serve as cancer therapeutics were initiated long before the discovery of BRAF mutations in human cancer, and were initially directed toward inhibition of RAS, CRAF, and MEK. One of the first ATP-competitive RAF inhibitors, ZM336372, was identified in an in vitro chemical screen as an inhibitor of RAF, yet unexpectedly activated RAF signaling in cells and was thus deemed not useful as a cancer treatment.²⁹ Clinical trials of multi-kinase inhibitors such as sorafenib demonstrated disappointing clinical activity in melanoma, likely due to their poor selectivity for BRAF and CRAF versus other kinases including VEGFR-2, VEGFR-3, platelet-derived growth factor (PDGFR)-β, FLT-3, and c-KIT.³⁰ Early clinical trials of selective MEK inhibitors also proved disappointing in melanoma,^31,32 likely in part due to a failure of most trials to restrict eligibility to patients with BRAF mutations, who were predicted to be most likely to respond based on preclinical studies.¹³ The use of MEK inhibitors as monotherapy is also limited by their narrow therapeutic index with on-target, off-tumor toxicities stemming from inhibition of MAPK signaling in normal tissues such as skin and bowel.³³

Second-generation, ATP-competitive RAF inhibitors (now called Class 1 BRAF inhibitors or BRAF monomer inhibitors) including vemurafenib (Zelboraf), dabrafenib (Tafinlar), and encorafenib (Braftovi) were much more successful clinically than prior RAF inhibitors (Figure 3).^34,35 Yet, while these second-generation RAF inhibitors demonstrate in vitro potency for all three RAF isoforms in biochemical assays, they inhibit MEK and ERK activation only in tumor cells with codon 600 BRAF mutations (Figure 1B).^22,36 In fact, in BRAF wildtype cell lines and tumors, including those with RAS mutations, vemurafenib induces a paradoxical activation of RAF signaling.^37–39 This selective inhibition of ERK signaling only in cells with BRAF V600 mutations is basis for the broader therapeutic index of vemurafenib versus MEK inhibitors and, as discussed in detail below, is the result of differential effects of the drug on BRAF monomers versus RAF dimers.^22,40

Figure 3:

Timeline of FDA-approvals of RAF inhibitors and RAF inhibitor-based combination therapies for BRAF V600 mutant tumors.

Biochemical studies of RAF dimerization and investigation of the crystal structure of the catalytic domain of BRAF bound to vemurafenib have provided molecular insights into the “RAF inhibitor paradox” (Figure 1B). Binding of vemurafenib has been proposed to stabilize one RAF protomer of the dimer in the αC-helix OUT/ DFG IN (CODI) configuration of the kinase domain. Steric hindrance prohibits the other protomer from adopting a thermodynamically unfavorable αC-helix OUT position, as this would break the dimer. This causes the second RAF protomer in the dimer to adopt the αC-helix IN position, thereby reducing the affinity of vemurafenib for the second protomer (negative cooperativity) given that vemurafenib binds in the αC-helix OUT position.^10,38,41 The active conformation of the second protomer thus promotes RAS-independent MAPK pathway output and RAF inhibitor resistance in cells with high levels of activated RAF dimers. Additionally, binding of an αC-helix OUT/ DFG-IN RAF inhibitor to RAF disrupts RAF autoinhibition and moderately promotes RAF association with RAS-GTP, subsequent phosphorylation on S338 (RAF priming), and RAS-dependent dimerization, which further contributes to pathway activation and increases the concentration at which vemurafenib inhibits ERK.^37,39,41,42

The ability of vemurafenib to activate RAF dimers is also the basis for its unique toxicity profile. Distinct from the on-target toxicities stemming from MEK and ERK inhibition, which most often include an acneiform skin rash, RAF-inhibitor induced on-target toxicities are, at least in part, attributable to paradoxical activation of ERK signaling in normal tissues and include perifollicular hyperplasia which is often accompanied by hyperkeratosis of the palms and soles (palma-plantar dysesthesia) and/or the development of papillomas (benign wart-like epidermal proliferations), and in up to one third of patients keratoacanthomas and/or cutaneous squamous cell carcinomas.^43,44 In rare cases, vemurafenib has also been shown to accelerate the growth of additional BRAF wildtype tumor types such as acute leukemias.⁴⁵

BRAF mutations are now classified based on their requirement for RAF dimerization and RAS activation. This classification schema has therapeutic implications as these biochemical properties provide insights into mutant sensitivity to FDA approved and investigation RAF inhibitors. As mentioned above, Class 1 includes only missense substitutions involving codon 600 of BRAF. Class 1 mutations are the only BRAF mutations sensitive to vemurafenib, dabrafenib and encorafenib as they are the only BRAF mutations identified to date capable of signaling as RAS-independent monomers.

Class 2 BRAF missense mutants, in frame deletions, and BRAF fusions signal as RAS-independent, constitutive RAF dimers (Figure 1A). Class 2 missense mutants are concentrated in the kinase domain activation segment near codon V600 and in residues 464–469 of the glycine-rich phosphate-binding loop (e.g. K601E/N and G464V). BRAF splice variants⁴⁶ and BRAF fusions^47,48 lack the RAS binding domain and activate RAF signaling via RAS-independent induction of BRAF dimer formation. In-frame deletions in glycine-rich phosphate-binding loop of BRAF (e.g., L485-P490del) have also been identified that function by locking the shortened αC-helix into the active “IN” conformation resulting in dimer dependent kinase activation.⁴⁹ A major focus of current drug development efforts is the identification of selective BRAF dimer inhibitors, which would be predicted to have activity in tumors driven by Class 2 BRAF mutations and fusions.

Class 3 BRAF mutants are low activity, kinase-impaired or kinase dead mutants that activate ERK by binding more tightly to activated RAS than wildtype RAF, and in doing so, recruit and hyperactivate wildtype CRAF (e.g., D594N, G466) (Figures 1 and 2). Class 3 BRAF mutants are RAS-dependent, with RAS activation often manifest through coincident RTK/RAS mutation or NF1 loss (Figure 1A).²³ The co-mutation patterns of Class 3 mutants have been shown to differ among common cancer types with co-mutation of NF1 common in melanoma, whereas RTK-based co-activation or mutation is more common in lung and colorectal cancers.²³ Tumors that harbor class 3 BRAF mutants are sensitive to inhibition of RAS, and thus comprise a new class of treatable RAF-mutant tumors with lineage-dependent sensitivity (i.e., sensitivity to inhibition of tumor type-specific mechanisms of RAS activation, often through targetable RTKs). Further, tumors with overexpression of class 3 BRAF mutants can be sensitive to vemurafenib, as the drug binds and inhibits the wildtype CRAF partner which is then unable to paradoxically activate the low activity or kinase-impaired class 3 BRAF mutant.²³ Importantly, dimerization of wildtype RAFs downstream of various upstream alterations (RTK/RAS mutation or reactivation, NF1 loss, etc.) constitutes a separate mechanism of resistance to vemurafenib (Figure 1A). In sum, the distinct mechanisms of activation of wildtype and mutant BRAF and the structural requirements of RAF inhibitor binding have important implications for ongoing efforts to develop novel RAF inhibitors that can overcome RAF dimer-mediated drug resistance and that are effective in tumors with Class 2 BRAF mutations and fusions, a population poorly served by current precision oncology-based therapeutic approaches.

The RAF/MEK inhibitor combination receives a tumor-agnostic FDA authorization

The RAF inhibitor paradox is also the molecular basis for the clinical success of the RAF/MEK inhibitor combination. Most RAF inhibitor resistance mechanisms induce ERK reactivation which can be attenuated by concurrent MEK inhibitor therapy. Early ERK rebound following vemurafenib exposure is driven by relief of ERK-mediated negative feedback which results in activation of wildtype RAF dimers (mostly CRAF dimers) that are sensitive to MEK inhibitors, but not vemurafenib.^22,40,50,51 Moreover, RAF and MEK inhibitor toxicities stemming from the overactivation (RAF inhibitor) or downregulation (MEK inhibitor) of ERK signaling in normal tissues can, if titrated correctly, offset each other making the combination more tolerable than either drug alone, albeit such abrogation of toxicity is organ and compound specific. For example, skin toxicity has been reported to be less severe with the combination, whereas hepatic and ocular toxicities have not, presumably due to difference in off-target adverse effects.^33,52,53 The unique biologic properties of RAF inhibitors that can result in induction of ERK activation in normal tissues thus explain how the RAF and MEK inhibitor combination can be both more active and less toxic than either drug along, and why the enhanced anti-tumor activity of the combination is largely if not completely restricted to patients whose tumors harbor BRAF V600 mutations.

Even with RAF/MEK inhibitor combination therapy, overall response rates vary significantly among tumor types suggesting that lineage-specific factors play a key role in conditioning therapeutic response. In BRAF V600E-mutant hairy cell leukemia and histiocytosis, clinical benefit from RAF/MEK inhibitor combination therapy approaches 100%, RECIST response rates range from 58–100% and responses are very durable (Figure 4).^54–56 The extreme counter example is BRAF V600E-mutant colorectal cancer, where less than 5% of patients respond to RAF inhibitor monotherapy or RAF/MEK combination therapy.^3,57,58 BRAF V600E-mutant melanoma lies between these two extremes with response rates to vemurafenib around 50–60% with most responders developing acquired resistance within the first year.³⁵ Despite the high degree of variability in the degree and duration of response among tumor types, in June 2022, the combination of dabrafenib plus trametinib was granted FDA-approval with a tumor-agnostic indication (excluding colorectal cancer). Results from the phase II ROAR and NCI MATCH clinical trials demonstrated response rates to RAF/MEK inhibitor combination therapy of up to 80% in 131 patients with a myriad of BRAF V600E mutant tumor types.^59–62 Similar to the basket trial of vemurafenib monotherapy,⁶³ the ‘basket’ design of these RAF/MEK inhibitor trials enabled the documentation of clinically significant responses in patients with rare cancers as well as in tumor types that have a low frequency of BRAF V600E mutations, including high- and low-grade gliomas, biliary tract cancers, and select gynecologic and gastrointestinal cancers. The recent RAF/MEK inhibitor approval extends the tumor-agnostic paradigm to six classes of alterations and mutational signatures including tumors with NTRK and RET fusions, oncogenic RET mutations, microsatellite instability high/mismatch repair deficiency (MSI-H/dMMR), high mutational burden (TMB-H), and now BRAF V600E-mutantion.⁶⁴

Figure 4: Impact of tumor lineage on response and response duration to dabrafenib plus trametinib combination therapy.

Left: Investigator-assessed overall response rate. Right: Investigator-assessed duration of response and progression-free survival. Results represent a compilation of the following clinical trials: NCT02034110 (Thyroid Cancer, Hairy Cell Leukemia, Glioma, Biliary Tract Cancer); NCT01336634 (Non-small cell lung cancer (NSCLC)); NCT01584648 (Melanoma); NCT02124772 (Langerhans Cell Histiocytosis); NCT01072175 (Metastatic Colorectal Cancer)

Overcoming RAF inhibitor resistance

Resistance to RAF inhibitors in BRAF V600 tumors has been shown to result from alterations that promote RAF dimer formation (target reactivation) and/or activate parallel or downstream signaling pathways that reduce RAF addiction (oncogenic bypass). These resistance mechanisms have been most comprehensively studied using melanomas collected pretreatment and at disease progression from patients treated with RAF inhibitors, and short-term melanoma cultures and human melanoma cell lines selected in the laboratory for RAF (+/− MEK) inhibitor resistance.

In contrast to most kinase inhibitors, gatekeeper mutations in BRAF that impair drug binding are uncommon.^65,66 Instead, resistance to RAF inhibitors in BRAF V600E-mutant cells is most often the result of the induction of RAF dimers which leads to re-activation of ERK signaling (Figure 5). In melanoma, induction of RAF dimers most often results from NRAS mutation,^67,68 increased expression/amplification of BRAF,^69–72 and expression of BRAF splice variants that lack the RAS binding domain.^46,73 Additional pre-existent or acquired alterations that likely confer drug resistance via induction of RAF dimers include mutations in KRAS,^3,74,75 overexpression of RAF1,⁷⁶ and loss of the RAS-GAP NF1.^51,77–79 Given the key role of RAF dimerization in RAF inhibitor resistance, significant effort has been made by academic investigators and industry to develop RAF inhibitors, detailed below, that can inhibit BRAF in its dimeric state (Figures 5 and 6)

Figure 5: Mechanisms of RAF inhibitor resistance.

Left (Adaptive resistance): Following RAF inhibitor treatment (depicted as vemurafenib, VEM, and representing all BRAF monomer inhibitors dabrafenib and encorafenib) of BRAF V600E mutant cells, inhibition of ERK results in a rapid decline in the expression of negative feedback regulation (see red X’s marking the loss of negative feedback signaling) which in a cell context dependent manner can result in activation of receptor tyrosine kinase (RTKs) and RAS. RAS activation induces RAF dimer formation which results in a rebound in ERK activation which can attenuate the anti-tumor activity of RAF inhibitors. The specific RTKs involved vary as a function of tumor lineage, with interpatient variable also significant among individual patients within specific tumor subtypes. Middle (Induction of RAF dimers): Various pre-existing and acquired mechanisms of drug resistance converge on induction of RAF dimers, including BRAF amplification, expression of BRAF splice variants lacking the RAS binding domain, RAS mutation and amplification, NF1 loss, among others. Right (Oncogenic Bypass): Activation of parallel or downstream signaling pathways or effectors that reduce dependency on BRAF, such as mutations in the PI3K/AKT pathway, loss of PTEN expression (represented by overlaying blue circle with a slash), overexpression of COT and MLK1, cyclin D1 amplification, and loss-of-function alterations in CDKN2A and RB1 can also induce RAF inhibitor resistance.

Figure 6: Therapeutic strategies to overcome genetic and epigenetic mediated resistance to RAF inhibitors.

A simplified signaling schematic colored to match novel pre-clinical or clinical compounds designed to inhibit the MAPK pathway, PI3K/AKT pathway, and other mechanisms of tumor survival in BRAF-mutant cells. Drugs on the left target vertical or parallel pathways. Drugs and their protein targets are colored the same. FDA-approved and investigational RAF inhibitors are depicted on the right, with the colored dots indicating selectivity for RAF monomers, dimers, or both; structural configuration of drug binding; and the extent to which each drug induces RAF priming and RAF/MEK interaction.

Paradox Breakers.

One category of novel RAF inhibitors are referred to as ‘paradox breakers”, as they were initially designed to lack the induction of ERK signaling in normal tissues induced by vemurafenib. These drugs, such as PLX7904 and PLX8394 (FORE8394), are structural analogues of vemurafenib modified with terminal sulfonamide and sulfamide substitutions.^41,80 As with vemurafenib, paradox breakers bind to and inhibit monomeric BRAF. Unlike vemurafenib, binding of a paradox breaker to one promoter within BRAF homodimers and wildtype CRAF/mutant BRAF heterodimers (where the drug is bound to CRAF) directly perturbs interactions with promoter 2 across the dimer interface thereby facilitating direct disruption of dimerization rather than binding to both protomers.⁸¹ The ability to inhibit dimers is achieved at higher concentrations than the concentration required to inhibit BRAF monomers. Despite their name, paradox breakers stabilize CRAF and ARAF homodimers and can in some cellular contexts induce paradoxical activation of RAF signaling due to negative cooperativity. Thus, in tumors with high levels of CRAF homodimers resulting from RTK activation or RAS mutation, paradox breakers would be predicted to be ineffective. Nevertheless, given their ability to spare CRAF homodimers which often govern MAPK signaling in normal tissue, paradox breakers may have a favorable safety profile. While early compounds were noted to have poor pharmacologic properties, newer formulations and compounds with greater bioavailability and longer half-life are in development. While these novel RAF inhibitors would be predicted to have a broader spectrum of activity than vemurafenib, including activity against tumors driven by Class 2 BRAF mutants, whether they will be more efficacious or have an improved therapeutic index than the RAF/MEK inhibitor combination in BRAF V600E mutant tumors is far from clear.

FORE8394 is currently being tested, in different formulations, in an ongoing, pan-cancer, expanded phase I/IIa trial of patients with advanced unresectable solid tumors with BRAF alterations (NCT02428712).⁸² A preliminary report of the first 75 patients treated, noted that one patient with melanoma and a BRAF-fusion had a complete response supporting greater potency for this compound than vemurafenib against Class 2 BRAF mutations and fusions. Clinical responses were also observed with FORE8394 in a variety of cancer types including papillary thyroid, anaplastic thyroid, small bowel, colorectal cancer (all n=1), ovarian cancer (n=2), and glioma (n=3). FORE8394 is also being tested in patients with BRAF V600E or BRAF fusion positive central nervous system metastases or high-grade gliomas (NCT05503797). A brain-penetrant paradox breaker, Compound 1a, with greater efficacy in the setting of melanoma brain metastasis than vemurafenib, is also in development.^83,84

Equipotent monomer/dimer and dimer-selective RAF inhibitors.

To allow for inhibition of the full spectrum of RAF homo and heterodimers, equipotent monomer/dimer and dimer-selective RAF inhibitors are in pre-clinical and clinical development. These pan-RAF inhibitors inhibit RAF by occupying one or both RAF protomers in the αC-helix IN conformation (most often αC-helix IN/DFG OUT; alternately, αC-helix IN/DFG IN) and thus have the potential to avoid negative cooperativity. As pan-RAF inhibitors induce RAF priming and dimerization and promote RAF/MEK binding, their potency may be limited in tumors (and normal cells) in which RAS is highly activated and contributes to the induction of RAF dimers, providing the theoretical basis for a better therapeutic index than MEK and ERK inhibitors. Structurally diverse, equipotent RAF inhibitors in pre-clinical testing include GDC-0879,⁴¹ SB-590885,⁴¹ AZ628,⁴¹ BGB-659,²² BDTX-4933,⁸⁵ and LY3009120.⁸⁶ Though promising in vitro studies led to a phase I clinical trial of LY3009120, this trial was terminated due to toxicity, limited clinical activity (stable disease as best response in 8/51 patients), and a lack of major effect on pharmacodynamic markers.⁸⁶

DAY101 (formerly TAK-580/MLN2480) has demonstrated anti-tumor effects, and synergy with MEK inhibition, in models of NRAS-mutant, BRAF V600E, and BRAF splice variant expressing melanoma and in models selected for acquired resistance to monomer-selective RAF inhibitors.^47,87 DAY101 has also been shown to be brain-penetrant and effective in models of pediatric low-grade astrocytoma harboring BRAF fusions.^41,88 KIN-2787 has demonstrated efficacy in preclinical models harboring class 1, 2, and 3 BRAF mutants and in the setting of acquired resistance to RAF/MEK inhibition, with only moderate or minimal sensitivity noted in RAS mutant and RAS/RAF wildtype cells, respectively.⁸⁹ Lifirafenib (BGB-283) is a pan-RAF and EGFR inhibitor with preclinical activity in BRAF V600E-mutant colorectal models.^47,90 In a first-in-human, phase I, dose-escalation (n=35) and dose-expansion (n=96) clinical trial of lifirafenib, one patient with BRAF mutant melanoma had a complete response and eight patients had confirmed partial responses, including patients with BRAF V600E/K-mutant melanoma (n=5), BRAF V600E-mutant thyroid cancer (n=2), and BRAF V600E-mutant low grade serous ovarian cancer (n=1).⁹¹ Partial responses were also observed in two patients with KRAS-mutant NSCLC and endometrial cancer (n=1 each); however, no responses were observed in K/NRAS-mutant colon cancer patients (n=20) highlighting again the potentially influence of lineage-specific factors on BRAF inhibitor response, with this newer generation of RAF inhibitors.⁹¹ The RAF dimer inhibitor BGB-3245 is also being tested clinically with an interim analysis reporting a confirmed response rate of 18% in a phase 1 trial, including responses in a melanoma patient with a BRAF fusion and a melanoma patients with BRAF V600E who had progressed on prior BRAF/MEK inhibitor combination therapy.⁹². Finally, both lifirafenib and BGB-3245 are being tested in combination with the MEK inhibitor mirdametinib (PD0325901) in patients with RAF or RAS mutations based on anti-tumor effects in RAS-mutant pre-clinical models.⁹³

It has been recognized that some RAF inhibitors, such as the multi-kinase inhibitors regorafenib and sorafenib,⁹⁴ and RAF709,⁹⁵ selectively inhibit RAF in the dimeric form due to more stable drug binding to the dimer pocket upon dimerization. In the clinic, belvarafenib (formerly GDC-5573/HM95573/RO7223619) demonstrated evidence of efficacy in a phase I study in patients with NRAS-mutant melanoma, KRAS-mutant sarcoma, and BRAF-mutant melanoma, colorectal cancer and gastrointestinal stromal tumor (GIST).⁹⁶ A phase Ib dose-escalation study of belvarafenib in combination with cobimetinib demonstrated anti-tumor efficacy in 7/32 patients including patients with NRAS Q61-mutant melanoma (n=5), BRAFV600E-mutant melanoma (n=1), and KRASG12D-mutant colorectal cancer (n=1).⁹⁷ Similar side effects to vemurafenib were reported, though no cases of squamous carcinoma were noted. Evidence of ARAF mutations as a mechanism of resistance to belvarafenib are believed to stem from the less potent inhibition of ARAF as compared to CRAF or BRAF.⁹⁸ LXH254 has also been shown to inhibit B/CRAF dimers, but spare ARAF dimers,^98–101 which may be an important consideration in clinical settings in which ARAF mutation is the cause of acquired resistance to current RAF inhibitors or in cases of primary ARAF driver mutation such as those found in a subset of histiocytic neoplasms and lung adenocarcinomas.^102–104

The dimer-selective RAF inhibitor PHI1 is the only RAF inhibitor identified to date that promotes positive allostery (enhanced potency) for the second protomer. This has been attributed to a distinct αC-helix IN binding conformation that involves a novel allosteric site in BRAF, called back-pocket IV (BP-IV), that is essential for substrate phosphorylation.¹⁰⁵ Dimer-selective RAF inhibitors that bind the ATP pocket in the DGF-OUT inactive conformation and have shown promising activity in class 1, 2, and 3 BRAF mutants are also being developed, such as BI 882370, GNF-7 and derivative SIJ1227.^106,107 Nonetheless, triplet combination therapy of dimer-selective and monomer-selective RAF inhibitors with a MEK inhibitor, as used effectively in xenograft models of BRAF V600E colon cancer, may still be needed to more completely inhibit ERK activation and tumor growth.⁹⁴

Non-kinase inhibitors of RAF and MEK.

In lieu of competing for ATP binding, allosteric inhibitors provide an alternative strategy to inhibit mutant RAF signaling. The first-in-class allosteric MEK inhibitor avutometinib (formerly VS-6766, CH5126766, RO5126766) binds MEK and stabilizes a conformation that hinders activating phosphorylation of Ser222 and Asn221, which are required for dissociation of MEK from RAF.^50,108,109 MEK bound to VS-6766 binds tightly to RAF and cannot be released, and thus acts as a dominant negative inhibitor of RAF signaling. In a phase I trial of avutometinib in combination with the mTOR inhibitor everolimus in RAS-mutant cancers, partial responses were observed in patients with KRAS G12D low grade serous ovarian cancer (n=2), KRAS G12V and G13A NSCLC (n=1 each), and NRAS Q61K mutant thyroid cancer (n=1), with an intermittent dosing schedule chosen for phase 2 studies to improve tolerability.¹¹⁰ Avutometinib is also being studied in combination with the focal adhesion kinase (FAK) inhibitor defactinib in RAS-mutant NSCLC (RAMP-202) and in low grade serous ovarian cancer, with preliminary efficacy observed in patients (an ORR of 64% (7/11) reported for patients with KRAS mutant low-grade serous ovarian cancer).¹¹¹

Despite its unique mechanism of action, pathway inhibition by avutometinib is limited by the long off-rate of the drug.⁹⁴ To improve the durability of interactions with RAF, insights from another allosteric MEK inhibitor trametinib have been used to develop novel MEK inhibitors with unique biochemical properties. Trametinib binds directly to KSR1 at the KSR1-MEK interface and is accommodated more durably into an extended allosteric binding pocket in MEK that is generated when KSR1 is bound. Trametinib binding increases the dissociation rate of MEK from RAF, which lowers the concentration of RAF-MEK complexes and MEK phosphorylation, yet leaves RAF free to be re-activated.^109,112,113 Combining the properties of both drugs, a trametinib derivative called trametiglue was designed to bind KSR1-MEK as well as bind and trap inactive RAF with the goal of more potent and durable pathway inhibition. While this could provide greater anti-tumor activity, it could also result in greater on-target, normal tissue toxicity.¹¹²

Additional allosteric RAF inhibitors have been developed that leverage innate mechanisms of RAF dimerization and cellular degradation. Type IV kinase inhibitors, such as macrocyclic peptides and the BRAFtide 10-mer peptide, were computationally designed to indirectly inhibit kinase function via disruption of the RAF dimer interface.^114,115 BRAFtide has also been demonstrated to trigger the proteosomal degradation of class II mutants including BRAF G469A, BRAF splice variants, as well as wildtype RAF, with limited activity for BRAF V600E mutants.

RAF inhibitors that utilize proteolysis targeting chimera (PROTAC) technology are also being developed. PROTACs have a chemical warhead that binds to the target protein, a flexible linker, and a ligand that binds to an E3 ubiquitin ligase which marks the target for degradation by the 26S proteosome. RAF PROTACS SJF-0628¹¹⁶ (via vemurafenib) or P4B^117,118 (via the novel RAF DFG OUT inhibitor BI 882370) have the potential to block both the kinase and scaffolding or non-enzymatic effects of RAF proteins. Selective KRAS,¹¹⁹ MEK1/2,^120,121 and SOS1¹²² PROTAC degraders are also in development.¹²³ Lastly, Vem-BisAmide-2, a compound that links two vemurafenib molecules via bisamide, works by locking BRAF V600E in an inactive dimeric conformation.¹²⁴

Vertical pathway targeting and intermittent MEK inhibitor dosing

In cancers addicted to MAPK signaling, it has been postulated that ERK activation must be inhibited potently and durably to achieve maximal tumor growth arrest and cell death and to prevent or delay the emergence of drug resistance.^34,125 Given that re-activation of ERK is a common feature of RAF-inhibitor resistance, combination strategies that target multiple nodes in the MAPK pathway (vertical pathway targeting) may enhance anti-tumor activity by more potently inhibiting ERK or by preventing rebound of ERK activation. While the RAF/MEK inhibitor combination validates the vertical pathway targeting approach, the clinical utility of alternative targeting strategies, in particular those that induce additive ERK inhibition in normal tissues, may be limited by their narrow therapeutic index. Concurrent inhibition of RTKs, when they feed into RAS activation, can also be viewed as a vertical pathway targeting approach. RTK inhibition, however, can also inhibit the activation of parallel pathways such as PI3K/AKT that when activated reduce dependence or RAF and MAPK signaling for proliferation and survival (as discussed in the following section).

In KRAS-driven cancers, much attention has been focused on co-targeting the RAS-GEF SOS1 to prevent feedback-induced RAS reactivation following MEK or RAS inhibition.^126,127 The small molecule SOS1 inhibitors BI-1701963 and MRTX0902, which bind the catalytic domain of SOS1 and prevent its interaction with RAS, have advanced to early stage clinical trials in patients with KRAS mutant solid tumors, in combination with the MEK inhibitor trametinib or the allosteric-KRAS G12C selective inhibitors sotorasib and adagrasib.^128,129 The latter combinations in particular may be tolerable given the tumor selectivity of mutant-selective KRAS inhibitors and the partial redundancy of SOS2, which can compensate for SOS1 inhibition in normal cells. Additionally, inhibitors of SHP2, a nonreceptor protein tyrosine phosphatase scaffold for SOS1, are being investigated in the clinic including BBP-398, ET0038, and JAB-3068. SHP2 inhibitors have demonstrated anti-tumor activity as single agents and in combination regimens in preclinical models with loss-of-function NF1 alterations and class III BRAF, KRAS G12C mutation, and EGFR mutations.^130–132 In contrast, RAS G13D, Q61X and some BRAF V600E mutant models were resistant to single agent SHP2 inhibition or combined therapy with a MEK inhibitor. Interestingly, in an EGFR-mutant NSCLC model, combined EGFR/SHP2 inhibition was able to overcome a novel mechanism of acquired resistance to EGFR inhibitors - ARAF amplification-mediated displacement of NF1 and activation of RAS.¹³³ ARAF amplification as a mode of resistance has also been observed following treatment with RAF dimer-selective inhibitors, suggesting that SHP2 inhibitors may be effective in this context as well.⁹⁸ Combined ERK and SHP2 inhibition has also been shown to inhibit growth of cell line and xenograft models of NF1-deficient malignant peripheral nerve sheath tumors.¹³⁴ However, the potential toxicity of this combination will have to be carefully managed given its potential to potently inhibit ERK signaling in normal cells. SHP2-RAF inhibitor combination approaches for BRAF V600E mutant tumors may be better tolerated given that these agents would be predicted to have opposing effects on ERK in normal cells. Notably, the SHP2 inhibitor TNO155 is being tested as part of multiple combination regimens that include dabrafenib, trametinib, the ERK1/2 inhibitor LTT462, anti-PD-1/L1 inhibitors, and the novel RAF inhibitor LXH254 in adult colorectal patients with advanced or metastatic BRAF V600 (E, D, or K) mutations (NCT04294160).

Vertical pathway targeting through the addition of an ERK inhibitor to the RAF/MEK inhibitor combination, has also been demonstrated to potently suppress the emergence of BRAF-amplified clones in PDX models of BRAF V600E-mutant melanoma and NSCLC.¹²⁵ In this context, both inhibition of a third MAPK node as well as an intermittent dosing schedule were critical determinants of anti-tumor efficacy and tolerability. Clinical studies of ulixertinib (BVD-523), an ATP-competitive ERK1/2 kinase inhibitor, are ongoing with anti-tumor activity and a manageable safety profile reported in a first-in-human study. Specifically, 14 partial responses (among the 101 patients treated for at least two cycles of ulixertinib at or above the recommended phase II dose of 600 mg twice daily) were observed in tumors with BRAF V600 and non-V600 mutations, NRAS and MEK mutations in patients with a variety of tumor types.¹³⁵ Preclinical models also suggest that intermittent dosing of MEK inhibitors may prevent or delay the onset of drug resistance, reduce on target, off tumor toxicity, and more effectively induce cell death in tumors that have developed an addiction to the presence of RAF and MEK inhibitors.^72,136–138 In preclinical models where RAF inhibitor resistance manifests through BRAF V600E amplification, withdrawal of MEK inhibition was also shown to reverse this amplification. Further, as these cells were addicted to the presence of MEK inhibitor, withdrawal of drug led to an increase in ERK activating resulting in G1 cell cycle arrest, senescence and/or cell death.¹³⁹ In cancer patients, however, whereas intermittent RAF/MEK inhibitor regimens have demonstrated some efficacy, intermittent dosing schedules have yet to been shown to prolong progression free-survival as compared to continuous therapy.^140,141 On the contrary, drug holidays are not recommended for tumors in which KRAS G13D amplification drives resistance, as MEK inhibitor withdrawal can induce ERK phosphorylation and hyperactivation and ZEB1-mediated epithelial-to-mesenchymal transition, and further drug resistance.¹³⁹ In sum, adaptive and n-of-1 clinical trial designs will be needed to test the efficacy and toxicity differences among combinations strategies and to define their optimal dose and schedule. Advances in proteomic methodology and or functional diagnostic assays (e.g., ex-vivo short-term CRISPR screens) will also likely be needed for the development of personalization of combination regimens given the large number of potential intrinsic and acquired RAF inhibitor resistance mechanisms.

Combination strategies targeting oncogenic bypass

Tumor cells can employ multiple strategies to reduce dependence on oncogenic BRAF. As examples, mutation of MLK-1¹⁴² and RAC1¹⁴³ can result in RAF-independent activation of MEK. Mutations of MEK1 (MAP2K2) and MEK2 (MAP2K2) have also been identified in BRAF V600 and non-V600 mutant tumors. MEK1/2 mutants can be divided into three functional classes based on differences in their dependence on RAS/RAF activity and sensitivity to ERK-mediated feedback, which together confer differential sensitivity to allosteric but not ATP-competitive MEK inhibitors.^144–146 However, whether the presence of a MEK mutation is sufficient to confer treatment resistance has been debated and likely differs as a function of the specific MEK mutant allele present, with those mutants exhibiting RAF-independence most likely to confer high levels of RAF inhibitor resistance.^70,147–156 In tumors with MEK E102_I103del mutants which condition response to MEK inhibition, for instance, combination therapy with RAF and ERK inhibitors may be more advantageous than RAF/MEK therapy.¹⁴⁵ Further downstream, cyclin D1 amplification and loss-of-function alterations in CDKN2A and have been shown to confer some degree of resistance to RAF inhibition.^157–160 While combinations of MEK (and RAF) inhibitors plus CDK4/6 inhibitors have demonstrated anti-tumor activity in preclinical studies and early clinical trials, studies of the MEK plus CDK4 combination have been notable for significant and often intolerable toxicities.^159–162 Lastly, amplification of the melanocytic lineage-specific transcription factor MITF and amplification or MITF-directed upregulation of the anti-apoptotic protein BCL2A1 have also been shown to confer resistance to RAF inhibition, among others.^155,163

Resistance to RAF inhibitors can also result from mutational activation of parallel signaling pathways. Mutation of canonical PI3K pathway nodes including PIK3CA/G and AKT1/3, and the negative regulators PIK3R1/2, PTEN, and PHLPP1, are the most extensively studied mechanisms of genetic bypass in BRAF mutant-melanoma.^{70,71,156,164}. Oncogenic PIK3CA and AKT3 mutations may contribute to RAF inhibitor resistance by allowing melanoma cells to persist until they are able to reactivate MAPK signaling through other mechanisms.¹⁶⁵ Similar to MEK1 mutations, PIK3CA mutations concomitant with BRAF mutations do not necessarily preclude response to MAPK pathway inhibition.^165,166 Likewise, concurrent loss of PTEN function in patients with BRAF-mutant melanoma and colon cancer is associated with poor response to BRAF inhibitors, but a subset of melanoma patients with co-mutations in BRAF and PTEN still derive some clinical benefit.^3,167 These results highlight that co-mutations likely confer variable degrees of BRAF inhibitor treatment resistance. Building on this concept, preclinical studies suggest that loss of function mutations in RB1 and PTEN cooperate to confer higher levels of BRAF inhibitor resistance than either alone.¹⁶⁸ While dual targeting of MEK and PI3K can in some models induce apoptosis in the context of BRAF/PTEN co-mutation, there remains little clinical validation to date that combined inhibition of MAPK and PI3K pathways in patients can improve clinical outcomes in cancer patients.^168–171 This is likely due in part to the toxicities of early generation PI3K and AKT inhibitors that precluded their use in combination with MEK inhibitors. As persistent activation of TORC1, S6K, and the eIF4F eukaryotic translation initiation complex through ERK-dependent and -independent mechanisms has been suggested as the ultimate purveyor of RAF/MEK inhibitor resistance,^172,173 continued research into strategies that harness tumor-specific and tolerable RAF inhibitors combined with more selective and less toxic PI3K, mTOR, or CDK4/6 inhibitors, could prove useful, especially in patients with melanoma and colorectal cancer.

Concurrent blockade of BRAF and adaptive reprogramming of cellular networks

Rapid adaptive changes in gene expression play an important role in attenuating response to RAF inhibitors. In BRAF-mutant cells, the foremost process of adaptive resistance to RAF inhibition lies in the relief of ERK-dependent transcription of negative feedback regulators.⁴⁰ The ensuing permissive signaling environment results in rapid reactivation of RTKs and RAS, the formation of RAF dimers, and reactivation of ERK. This rapid adaptive response to RAF inhibitors is the also the mechanistic basis for the success of vertical pathway co-targeting, both the co-targeting of upstream RTKs, RAS, and the downstream co-targeting of MEK, ERK or CDK4/6.¹⁷⁴ Addition of SHP2 inhibitors to RAF/MEK inhibitor therapy has also been demonstrated to block RTK-mediated adaptive ERK reactivation in BRAF-mutant melanoma models suggesting that the early use of vertical pathway targeting may be critical to avoid clonal selection of cancer cells with high levels of non-genomic RAF inhibitor resistance.¹⁷⁵ The induction of RAF dimers following rapid RTK reactivation is also a rationale for the development of novel RAF inhibitors that inhibit RAS-dependent RAF dimers as detailed above.^22,76 .

As with genomic mechanisms of RAF inhibitor resistance, much of the epigenetic remodeling and adaptive changes in signaling networks that follow RAF inhibition result in reactivation of ERK signaling. As an example, upregulation of the MAP3Ks MLK-1 and COT have been shown to be sufficient to confer BRAF-independent, but MEK-dependent, rescue of MAPK pathway output following RAF inhibition.^142,176 Transcriptional upregulation and secretion of FGF1 ligand upon RAF/MEK inhibitor treatment has also been shown to promote resistance through autocrine activation of FGFR and stimulation of local fibroblasts to secrete HGF, the ligand for MET.^177,178 Elevated expression of other RTKs such as PDGFRβ, IGF1R, and INSR can all restore MAPK signaling in melanoma, and co-inhibition of each RTK with RAF has been shown to overcome resistance in select preclinical models.^{67,71,179–181} Which of these RTKs, if any, play a dominant role in individual patients with RAF inhibitor resistance has not been fully elucidated and is likely to be lineage dependent and potentially heterogeneous within cancer types and possibly among different sites of disease in individual patients. Single cell sequencing studies also suggest that only a small subset of melanoma cells exhibits effective RTK and ERK reactivation following treatment with RAF/MEK inhibitor combination therapy and it is this rare subset of persister cell that form major resistant clones.¹⁷⁵ In sum, the above complexity together with the inability of current clinical diagnostic tools to robustly assess the activation status of individual RTKs has made the personalization of RTK and RAF inhibitor combinations challenging.

Many of the lineage-specific modes of adaptive resistance identified to date involve the HER kinase family of RTKs. In contrast to BRAF V600E-mutant melanoma and colon neuroendocrine carcinomas (co-NEC), lineages in which EGFR is epigenetically suppressed via promoter methylation,^182,183 higher basal levels of phosphorylated EGFR in BRAF V600–mutant colorectal cancer underlie the rapid reactivation of EGFR, RAS and CRAF upon RAF inhibition.^184,185 To overcome this adaptive resistance, combination therapy with RAF and EGFR inhibitors was demonstrated to synergistically reduce tumor proliferation in preclinical cell line and xenograft models.^184,185This lead to the pilot study of the vemurafenib/panitumumab (anti-EGFR antibody inhibitor) combination in patients with BRAF V600E-mutant colorectal cancer and in parallel the modification of the vemurafenib basket trail to include a vemurafenib/cetuximab (anti-EGFR antibody inhibitor) arm in the colorectal cancer cohort.^63,186 Further testing of triplet therapy of encorafenib, cetuximab, and binimetinib (MEKi) or doublet encorafenib/cetuximab therapy in the BEACON study demonstrated longer overall survival rates and higher response rates than standard chemo-based therapy in patients with BRAF V600E-mutated metastatic colorectal cancer who progressed on prior treatments, warranting approval of the combination encorafenib/cetuximab therapy as a new standard of care.¹⁸⁷ Mechanisms of resistance to combination RAF/EGFR inhibitor therapy in the BEACON trial were often KRAS mut, NRAS mut, and MET amplification.¹⁸⁸ As these resistance mechanisms all lead to the induction of RAF dimerization, preclinical studies suggest that combined EGFR and RAF-dimer specific inhibitors could prove to be more effective than current RAF inhibitors in this clinical context.⁴ Analogous to EGFR in colon cancer, adaptive upregulation and activation of HER3 has been observed in BRAF-V600E mutant thyroid cancers treated with RAF inhibitors, prompting preclinical testing of HER kinase inhibitors and anti-HER3 monoclonal antibodies in combination with RAF inhibitors in this cancer type.^189–191

Transcriptional activation of parallel pathways has also been shown to contribute to RAF inhibitor resistance. As an example, upregulation of HIPPO/YAP signaling was demonstrated to drive resistance following RAF/MEK inhibition in BRAF V600E-mutant models, with YAP inhibition able to abrogate adaptive resistance in the setting of concurrent RAF/MEK inhibitor treatment.^192,193 ERK1/2 plays a critical role in driving the transcription, phosphorylation and activation of many anti-apoptotic BCL2 family members, while suppressing pro-apoptotic BH3-only proteins, and these processes are further co-opted upon mutational activation of the MAPK pathway.¹⁹⁴ In melanoma, prior studies have demonstrated that transcriptional upregulation of the anti-apoptotic BCL2 family member MCL1 and suppression of pro-apoptotic BIM, driven by BRAF V600E-STAT3 activation and PTEN loss, respectively, contributed to RAF inhibitor resistance.^164,195 RAF inhibitor-induced relief of feedback has also been shown to reinforce expression of MCL1, through RTK-mediated, yet MEK-independent processes.¹⁹⁶ As such, BCL2 family proteins have become attractive targets for development of potent and selective inhibitors. Preclinical studies have demonstrated that combined use of the BCL-2 inhibitor navitoclax (ABT-263) and copper chelation restored sensitivity to RAF/MEK inhibition in RAF/MEK-resistant models of BRAF V600E-mutant melanoma.¹⁹⁷ Building on earlier work demonstrating synthetic lethality of combined BCL-XL and MEK inhibitors in KRAS-mutant models,¹⁹⁸ combined RAF or MEK inhibitors with the MCL1 inhibitor AZD5991 was shown to be synthetic lethal via BAK/BAX/BIM/BMF-induced apoptosis in BRAF V600E- or NRAS-mutant melanoma models, and AZD5991 combined with ERK inhibition restored sensitivity in RAF/MEK inhibitor-resistant melanoma models.¹⁹⁹ The apparent success of this combined regimen is based on the low levels of the pro-survival protein BCL-XL observed in melanomas, thus biasing these tumors to rely on MCL1 for survival signaling. Conversely, due to the higher ratio of BCL-XL:MCL-1 observed in colorectal cancer, MEK inhibition in combination with BCL2/BCL-w/BCL-XL inhibition was more effective in those models.¹⁹⁹ As an additional mechanism of parallel pathway adaptive resistance, crosstalk with the WNT/CTNNB1 (β-catenin) pathway has emerged as an important regulator of cell growth and proliferation in cancer cells. In BRAF V600E-mutant colorectal cancer, RAF inhibition was shown to induce EGFR/ERK-independent, FAK-dependent upregulation and activation of WNT/CTNNB1 (β-catenin), a prominent mediator of colorectal cancer pathogenesis.²⁰⁰ Other work, however, has suggested that loss-of-function mutations in the E3 ubiquitin ligase RNF43, which would serve to elevate WNT signaling, were a biomarker of favorable response to EGFR/RAF inhibition selectively in microsatellite-stable, BRAF V600E-mutant, metastatic colorectal cancer.²⁰¹ The latter suggests a non-canonical WNT mechanism that may parallel the WNT-mediated suppression of MAPK signaling seen in intestinal stem cells,²⁰² though more studies will be needed to understand the complex contribution of WNT signaling to RAF inhibitor response and resistance.

Co-targeting cellular plasticity and the immune microenvironment

Epigenetic, transcriptomic, and metabolic changes often underlie the adaptive reprogramming induced by BRAF mutation and RAF and MEK inhibition.²⁰³ As an example, BRAF V600E-mutant colorectal cancer cells treated with EGFR and RAF inhibitors have been shown to display ‘adaptive mutability’ resulting in increased genetic diversity, which would be predicted to favor drug tolerance and the emergence of tumor cells with co-mutations that confer high levels of drug resistance.²⁰⁴ This transient cell state employs error-prone DNA repair in lieu of highly regulated mismatch repair (MMR) or homologous recombination to generate microsatellite instability and DNA damage.²⁰⁴ In BRAF-mutant melanoma, RAF and MEK inhibition has been shown to drive adaptive switching from a hyper-differentiated, proliferative state (often MITF^high/AXL^low) to a drug dedifferentiated, invasive state (often MITF^low/AXL^high) that enables drug tolerance and resistance.^205–208 In breast cancer models, MEK inhibition was shown to promote transcriptional rewiring through the induction of genome-wide enhancer formation involving the seeding of BRD4, MED1, H3K27 acetylation, and p300, providing rationale for cotargeting of these P-TEFb transcriptional elongation regulatory complex members.²⁰⁹ Further, metabolic reprogramming and induction of oxidative phosphorylation via the MITF-PGC1α axis has been shown to drive adaptive resistance to RAF and MEK signaling.^210–214 Overall, tumor cells often exhibit cellular plasticity which enables fluid rewiring of genome stability and transcriptional states, and exposes metabolic dependencies.

As BRAF mutants and RAF or MEK inhibitors can also have immunomodulatory effects, additional adaptive responses to RAF/MEK inhibition center on crosstalk with the tumor microenvironment in ways that would be permissive for tumor growth.^215,216 As an example, cancer-associated fibroblasts have been implicated in stimulating MAPK signaling through release of HGF and FGF7.^217–219 Aged fibroblasts can also promote reactive oxygen species-mediated DNA-damage in tumor cells through the secretion of the WNT antagonist sFRP2.²²⁰ Paradoxical activation of stromal cells by RAF inhibitors can induce matrix production and remodeling within the tumor microenvironment leading to elevated integrin β1/FAK/Src signaling in melanoma cells and drug tolerance.²²¹ In melanoma cells, RAF inhibition induces YAP/MRTF-dependent autocrine remodeling of their secreted extracellular matrix associated with tumor stiffening, mechanosensing, and therapy resistance.²²² In sum, adaptative mechanisms of RAF inhibitor resistance may be dependent on interactions between tumor and stromal cells or ligand secretion by stromal cells within the tumor microenvironment which could theoretically be co-targeted together with RAF and MEK to prevent or delay the emergence of drug resistant tumor cells. Critically, greater understanding of the contribution of the immune microenvironment to the success or failure of RAF inhibitor therapy or RAF/immunotherapy combination regimens would also benefit from the development of better preclinical models, such as humanized mouse models that better reflect the genomic and biologic complexity of human cancer.

Given the success of immune checkpoint inhibitors in melanoma and other cancers, increasing efforts to study the relationship among CTLA-4, PD-1, and PD-L1 and RAF inhibition in BRAF-mutant tumors is underway.²²³ In patients with BRAF-mutant melanoma, RAF/MEK inhibitor responses are more rapid than those observed with immune checkpoint blockade, but significantly less durable. Efforts to combine both modalities have been largely disappointing to date mainly due to overlapping toxicities which have required treatment discontinuation.^224–227 However, randomized trials of novel combinations of RAF/MEK and PD1/PD-L1 inhibitors are ongoing (see Table 1) and have begun to show promise, such as the triplet therapy of atezolizumab (anti-PD-L1), vemurafenib, and cobimetinib, which is now FDA approved as first-line treatment for unresectable advanced BRAFV600 mutation-positive melanoma.²²⁸ Investigation into the molecular basis for the enhanced anti-tumor effects of immunotherapy plus RAF targeting combinations have pointed to molecular changes induced by RAF and MEK inhibitors within immune networks and the tumor microenvironment. As an example, a single-arm phase 2 trial of combined spartalizumab (PDR001, anti-PD-1), dabrafenib and trametinib in 37 patients with BRAFV600E colorectal cancer demonstrated a confirmed overall response rate of 24.3%, with single-cell RNA sequencing of 23 paired pretreatment and on-treatment mCRC tumor biopsies revealing global upregulation of IFN-stimulated transcriptional programs and antigen processing and presentation pathways and more complete MAPK inhibition in those patients with better clinical outcome.²²⁹

Table 1.

Single agent RAF inhibitors and combination therapies approach in clinical trials

Single Agent
Drug	Indications	Phase	NCT/Reference
Vemurafenib	Adult HCL, relapsed or refractory	II	NCT01711632 Tiacci, E. et al.43
	Adult advanced solid tumors, V600	II	NCT02304809
	Pediatric advanced solid tumors, V600	II	NCT03220035
	Pediatric glioma, V600E	I	NCT01748149
DAY101	Pediatric LGG, any RAF alteration	III	NCT05566795
	12 yrs.+, recurrent or progressive solid tumors, alterations in the key proteins of the MAPK pathway	II	NCT04985604
	6 mos. - 25 yrs., recurrent or progressive LGG, activating RAF alteration	II	NCT04775485
	6 mos. - 25 yrs., advanced solid tumors, activating RAF fusion	II	NCT04775485
Belvarafenib	Solid tumors, BRAF Class II/III	II	NCT04589845 Kim, T. W. et al.96
FORE8394	10 yrs.+, advanced unresectable solid tumors	I/IIa	NCT02428712 Janku, F. et al.82
Lifirafenib	Adult advanced solid tumors	I	Desai, J. et al.91
BGB-3245	Adult advanced solid tumors	Ia/Ib	NCT04249843
KIN-2787	Adult solid tumors, known BRAF alterations	I/Ib	NCT04913285 Miller, N. L. G. et al.89
	Adult melanoma, NRAS mutations	I/Ib	NCT04913285
XP-102	Adult NSCLC/ Thyroid Cancer / CRC / Melanoma, V600	Ia	NCT05275374
Compound 1a		Preclinical	Wichmann, J. et al.70
BDTX-4933		Preclinical	Ng, P. Y. et al.85
GNF-7		Preclinical	Kim, N. et al.89
SIJ1227		Preclinical	Kim, N. et al.89 Choi, S. H. et al.90
BRAFtide		Preclinical	Beneker, C. M. et al.97
SJF-0628		Preclinical	Alabi, S. et al.116
P4B		Preclinical	Posternak, G. et al.101
Combinations
Drugs	Indications	Phase	NCT/Reference
Encorafenib + Binimetinib	Adult metastatic NSCLC, V600E	II	NCT03915951
	Adult metastatic pancreatic cancer, V600E	II	NCT04390243
	Adult relapsed/ refractory HCL, V600	II	NCT04324112
Encorafenib + Binimetinib + Pembrolizumab	Adult advanced melanoma, V600E/K	III	NCT04657991
Encorafenib + Binimetinib + Palbociclib	Adult advanced melanoma, V600	Ib	NCT04720768
Encorafenib + Cetuximab +/− Binimetinib	Adult metastatic CRC, V600E	III	NCT02928224
Encorafenib + Cetuximab + Binimetinib	Adult localized CRC, V600E	II	NCT05510895
Encorafenib + Cetuximab +/− Nivolumab	Adult metastatic/unresectable CRC, V600E	II	NCT05308446
	Adult metastatic/unresectable CRC, V600E, Microsatellite Stable	I/II	NCT04017650 Morris, V.K. et al.237
Dabrafenib + Trametinib	Adult relapsed/ refractory HCL, V600E	II	NCT02034110 Kreitman, R. J. et al.238
	Adult Erdheim Chester Disease, V600E	II	NCT02281760
Dabrafenib and Trametinib + Ipilimumab and Nivolumab	Adult stage III-IV melanoma, V600	III	NCT02224781 Atkins, M. B. et al.231
Dabrafenib + Trametinib + PDR001	Adult advanced Melanoma, V600	III	NCT02967692
	Adult metastatic CRC, V600E	II	NCT03668431 Tian, J. et al. 229
XP102 + Trametinib	Adult NSCLC/ Thyroid cancer / CRC / Melanoma, V600	I/IIa	NCT05275374
Lifirafenib + Mirdametinib	Adult advanced/unresectable solid tumors	Ib	NCT03905148
BGB-3245 + Mirdametinib	Adult advanced solid tumors	I/IIa	NCT05580770
Belvarafenib + Cobimetinib/Cetuximab	Adult advanced/ metastatic solid tumors	Ib	NCT03284502 Shin, S. J. et al.97
Avutometinib + Defactinib	Adult recurrent NSCLC, BRAF mutant	II	NCT04620330
	Adult endometrial/ cervical/ ovarian cancer, RAS/ BRAF mutant	II	NCT05512208
Vemurafenib + HL-085	Adult advanced melanoma, V600E/K	IIa/IIb	NCT05263453
	Adult metastatic CRC, V600E	II	NCT05233332
	Adult advanced solid tumors, V600	I	NCT03781219
Vemurafenib + Cobimetinib	Adult advanced solid tumors, V600	IIa	NCT02091141
	Adult advanced solid tumors, V600	II/III	NCT05768178

Abbreviation: mos, months; HCL, hairy cell leukemia; LGG, low-grade glioma; NSCLC, non-small cell lung cancer; CRC, colorectal cancer

Two treatment regimens, pulsatile (but not continuous) MEK inhibition or anti-PD-1/PD-L1 dosing prior to MEK inhibition, have been found to be most supportive of CD8+ cytotoxic T cell activation.^136,230 In validation of such findings, data from the phase III DREAMseq trial in patients with advanced BRAF-mutant melanoma demonstrated a 20.3% improvement in overall survival when treatment with nivolumab and ipilimumab preceded dabrafenib and trametinib therapy. These data are also likely to provide clarity as to the optimal temporal sequencing of targeted and immune checkpoint inhibitors for BRAF V600E melanomas.²³¹ Interestingly, combinations of novel equipotent or dimer-selective RAF inhibitors with MEK inhibitors have also demonstrated durable MAPK pathway inhibition along with immunomodulatory effects like expansion of CD8+ T cells, supporting the future clinical testing of combinations of novel RAF inhibitors with PD-1/L1-targeted therapies.²³² Nonetheless, adaptive reprogramming of both proliferative and immune signaling upon RAF/MEK inhibition, including concurrent dysregulation of the MET/YAP1/LEF1 axis as well as CD8+ T cell exhaustion and loss of antigen presentation, is one mechanistic rationale for potential cross-resistance between targeted and immunotherapies.²³³ Single cell sequencing and novel, real-time, dynamic technologies will thus be required to capture the states of tumor and T cells most sensitive to therapeutic intervention.^234,235 In sum, the crosstalk between tumor and immune molecules may serve as a reference to guide the development of diagnostic biomarkers for risk stratification and facilitate the identification of novel therapeutic targets for testing as part of future combination strategies.

Conclusion

The clinical success of vemurafenib, dabrafenib and encorafenib lie in their selective inhibition of ERK signaling in BRAF V600 expressing tumor cells which results from their selectivity for activated BRAF V600E/K/R/D monomers.^22,36,38 The latter biochemical property is also the basis for their vulnerability to adaptive and selective resistance mechanisms that result in reactivation of ERK. This ERK reactivation can arise as a result of rapid relief of upstream and downstream negative feedback (adaptive resistance), and/or the clonal selection of tumor cells with RAS mutation, BRAF amplification, NF1 loss, and RAF splice variants, among other less common resistance alterations.^22,40,46 As analyses of tumor cells and preclinical studies in cell lines and mice indicated that RAF inhibitor resistance is typically the result of ERK pathway reactivation, it was hypothesized that vertical pathway co-targeting of RAF and MEK would have superior antitumor effects than RAF inhibitor monotherapy in a BRAF V600E mutant context.⁴⁰ The RAF/MEK combination is, however, notable and distinct from most other vertical pathway targeting strategies in that RAF and MEK inhibitors often have opposing effects in normal tissues (RAF inhibitors induce, whereas MEK inhibitor downregulates ERK signaling in normal cells).³⁸ The dabrafenib and trametinib combination is therefore something rare in oncology, a combination which is both less toxic and more effective than either agent alone.

The biochemistry and unique vulnerabilities of current RAF inhibitors will therefore need to be carefully considered as the field seeks to improve on the dabrafenib/trametinib combination through the development of novel RAF inhibitors with activity against RAF dimers, and triple drug combinations designed to prevent the emergence of drug resistance through more potent ERK inhibition (vertical pathway targeting) or through co-targeting of oncogenic bypass mechanisms. While newer RAF dimer inhibitors would be predicted to have broader clinical activity than current RAF inhibitors, one cannot assume a priori that they will have similar clinical utility. This is because inhibition of RAF dimers in normal tissues may result in greater toxicity than that observed with vemurafenib, dabrafenib or encorafenib. While a concern, early preclinical data suggest that it may be possible to develop RAF inhibitors that inhibit some but not all RAF dimers. If so, optimally designed RAF dimer inhibitors could have tumor selective properties that endow them with a broader therapeutic index than MEK and ERK inhibitors, which lack tumor selectivity for ERK inhibition and thus have a narrow therapeutic index. Furthermore, while the therapeutic index of RAF dimer inhibitors may be less than the RAF/MEK inhibitor combination in a BRAF V600E mutant context, the development of such drugs will be critical to expanding the benefits of precision oncology to tumors with Class 2 BRAF mutants, such as BRAF fusions. Similarly, RAF dimer inhibitors with selectivity for CRAF (RAF1) homo- and heterodimers could be more active than MEK inhibitors in tumors with Class 3 BRAF mutations, in particular when combined with upstream inhibitors of RTKs and RAS.

In sum, the tumor agnostic FDA approval of the dabrafenib/trametinib combination is an important milestone in the development of effective precision oncology-based therapeutic strategies for patients with BRAF V600E mutant tumors. Unfortunately, even with the combination, treatment responses are typically partial and acquired resistance common. Several general approaches are now being pursued to identify three and four-drug combinations that prevent or delay the emergence of drug resistance. These can be divided into 1) combinations designed to more maximally and durably inhibit ERK signaling (vertical pathway co-targeting), 2) combinations that co-target parallel signaling pathways that reduce dependence on mutant RAF for proliferation, survival, or other oncogenic phenotypes, and 3) combinations that modulate the host immune response with the goal of achieving not only tumor regression but potentially disease cure.

As the co-mutation pattern and adaptive response of BRAF mutant tumors to RAF inhibitors varies as a function of cancer lineage, we predict that future RAF inhibitor combinations will need to be developed with lineage specific dependences in mind to achieve optimal clinical results. Intratumoral and lesion-to-lesion heterogeneity among patients with BRAF mutations, even those with the same cancer type, will also likely prove to be a significant hurdle to the design of clinical trials testing novel RAF inhibitor-based combination strategies, as we currently lack diagnostic assays capable of robustly identifying which RTK or parallel signaling pathways is activated in an individual patient. Furthermore, as RTK activation may not be apparent until after the start of drug treatment (following relief of upstream feedback), it may not be possible to predict pre-treatment using archival tumor tissue which RTKs or parallel signaling pathways would be most appropriate to co-target in individual patients.

Future progress will therefore require the development of more robust and reproducible transcriptomic, proteomic or functional profiling methods, as well as the routine use of on-treatment tumor biopsies to guide treatment optimization, an approach that would be both invasive and costly. Alternately, analysis of serial ctDNA samples could provide a less invasive strategy to identify occult or emerging genomic alterations that may serve as biomarkers of clinical outcome or mechanisms of resistance such as NRAS mutation. As an example, ctDNA analysis of colorectal cancer patients treated with encorafenib/cetuximab as part of the BEACON trial revealed decreased overall survival in patients with higher baseline BRAF V600E variant allele frequency.²³⁶ Current ctDNA methods cannot however detect many of the alterations that confer RAF inhibitor resistance such as BRAF splice variants or changes in the tumor microenvironment. Therefore, the greatest near-term success may be achievable by cotargeting pathways for which a genomic biomarker such as PIK3CA or PTEN mutation can be detected by currently available next generation sequencing methods. Finally, dosing schedule will also likely be critical to the future success of RAF inhibitor combinations, as shown preclinically for the combination of RAF and MEK inhibitors with immunotherapy.²³⁰ In sum, while the above hurdles are significant, the tumor agnostic approval of the dabrafenib in combination with trametinib for BRAF V600 tumors represents the successful culmination of over two decades of integrated bed-to-bench and back collaboration, and provides hope that ongoing translational investigation and rational drug development will soon lead to clinically meaningful improvements in clinical outcomes for patients with a broader diversity of RAF-driven tumors.

Figure 7: Status of RAF inhibitor clinical trials.

RAF inhibitors and combinations of FDA-approved and investigational agents in clinical trials are organized per cancer type. Colored circles indicate FDA-approval or phase of clinical trial. Colored lettering provides additional information about each drug, as indicated in the legend. Abbreviations: Vemurafenib, VEM; Dabrafenib, DAB; Belvarafenib, BELVA; Encorafenib, ENCOR; Trametinib, TRAM; Cobimetinib, COBI; Binimetinib, BINI; Cetuximab, CETUX; Palbociclib; PALBO; Pembrolizumab, Pembro.