The first basket clinical trial testing the BRAF inhibitor vemurafenib resulted in evidence of activity in 13 unique cancer types with BRAFV600 mutations, but the response rates were variable. Therefore, different cancer histologies with the same driver oncogene display different degrees of oncogenic pathway addiction.
See related article by Subbiah et al., p. 657.
It seems logical to want to inhibit an activating mutation in a cancer-driving oncogene (e.g., BRAFV600 mutations) to treat a patient whose cancer has that genetic event, regardless of the cancer's tissue of origin. However, it took more than 10 years and the development of a new type of clinical trial, the “basket trial,” to facilitate this goal. It turns out that the degree to which a cancer is addicted to an oncogene depends on the tissue of origin. In addition to this biological complexity, the regulatory path for testing therapies targeting a specific genetic alteration across different cancer histologies was not cleared until years after initial clinical development of specific BRAF inhibitors.
In this issue of Cancer Discovery, Subbiah and colleagues report on the full cohort of patients enrolled in the first-ever basket trial designed to test a single drug, the BRAFV600 mutant inhibitor vemurafenib, in multiple indications that share the same genetic marker (1). The path to the clinical testing of BRAF inhibitors in BRAFV600 mutant–driven cancers, which should have been a slam dunk, detoured to develop a new paradigm for drug development and understand the nuanced biology underlying different clinical outcomes across early indications tested.
Before the clinical development of BRAF inhibitors, there had already been other recognized driver oncogenes expressed by cancers of different histologies. A notable example is the HER2/neu amplification present in approximately one fourth of breast cancers and also in some cases of ovarian and gastric cancers (2). The development of the anti-HER2/neu antibody trastuzumab focused on each separate cancer histology against a competing drug in the same indication, instead of focusing on the same genetic event across multiple cancer histologies. Evidence of the exquisite sensitivity of BRAFV600 mutant–metastatic melanoma to vemurafenib (3) prompted a logical desire to test this drug in patients with any cancer that harbors BRAFV600 mutations. But there was no path for such clinical development in the early 2010s.
This scenario led to the initial concept of a “basket trial,” where patients are selected based on a genetic driver event (BRAF, NTRK, RET) present across cancers of different histologies and tissues of origin, sometimes at very low frequencies, and then treated with the same targeted drug (4). During the planning of the basket trial for vemurafenib, there was no regulatory path for drug approval based on BRAFV600 mutations. There was no precedent for any drug that had received a license based on studies conducted across multiple cancer histologies, even if the drug is a highly targeted agent that would work only when a specific genetic event is present. The precedents came in 2017, with the approval of the anti–PD-1 antibody pembrolizumab in patients with cancers that harbor a very high mutational load due to microsatellite instability (MSI-high; ref. 5), and then in 2018, with the approval of larotrectinib in patients with cancers that harbor NTRK gene fusions (6).
The initial development of vemurafenib in BRAFV600-mutant cancers met hurdles which were somewhat mitigated by the time pembrolizumab and larotrectinib were being developed for MSI-high and NTRK+ cancers, respectively. For instance, because MSI-high was already recognized to have a distinct cancer biology even within the same cancer histology, MSI status was already a standard-of-care test in many centers before anti–PD-1 antibodies were found to have high activity against cancer types with high mutational burden. Moreover, prospective testing of multiple genetic alterations in cancers had been implemented in several referral cancer centers using multigene sequencing panels by the time NTRK inhibitors entered clinical testing, facilitating the identification of very rare patients whose cancers harbor NTRK fusions. In 2012, finding the rare patients across non-melanoma cancer histologies to be enrolled in the newly open vemurafenib basket trial was very slow. Compounding this problem was the lack of response to vemurafenib in the initial cohort of patients with colon cancer harboring BRAFV600 mutations. The combination of slow accrual and evidence of biological complexity behind cancer addiction to BRAFV600 mutations resulted in a long delay in acquiring the antitumor data of vemurafenib in BRAFV600-mutated cancers across wide-ranging cancer histologies.
The report by Subbiah and colleagues is one of several articles published from the vemurafenib basket trial. The first report differentiated a higher response rate in patients with BRAFV600-mutant non–small cell lung cancer and Langerhans cell histiocytosis (both with response rates in the range of 45%), from anecdotal and lower rates of responses among patients with several other BRAFV600-mutant cancers (including colorectal cancer, pleomorphic xanthoastrocytoma, anaplastic thyroid cancer, cholangiocarcinoma, salivary duct cancer, ovarian cancer, and clear-cell sarcoma; refs. 7, 8). All the nonmelanoma response rates were lower than that for melanoma (3), but the highest response rate exceeded 95% in patients with BRAFV600 mutant–driven hairy cell leukemia, which was based on a different clinical trial that followed the report that all hairy cell leukemias harbor the BRAFV600E oncogene (9).
At the end of the day, the significance of the report by Subbiah and colleagues lies in the finding that BRAFV600 mutations are clinically actionable in 13 unique cancer types and not in the semantic portrayal of overall response rates across diverse cancer types as an indication of lineage-independent oncogene or MAPK pathway addiction. The ultimate arbiter of lineage dependence or independence of BRAF–MAPK addiction will rest with which major pathway the underlying resistance mechanisms point to. Oncogene (or pathway) addiction refers to the exquisite (clinical) dependence of cancer cell growth and survival on a singular or select few oncogenic alteration(s), in the context of a much larger number (one to two orders of magnitude) of somatic alterations. This concept justifies a limited gene panel approach for sequencing to guide the clinical choice of pathway-directed agents for cancer treatment. Thus, if the simple majority of adaptive or acquired resistance mechanisms reactivate the MAPK pathway among the 13 cancer types where responses were observed (or innate resistance mechanisms among the cancers that did not respond), then a strong case for lineage-independent BRAFV600-mutant or MAPK pathway addiction could be made. Unfortunately, tumor samples for the analysis of innate, adaptive, and acquired resistance mechanisms, through interrogation of the proteome, transcriptome, epigenome, and genome, are unavailable.
Since the initial discovery of a genetic mechanism (i.e., acquisition of RAS co-mutations) of MAPK reactivation driving acquired vemurafenib resistance in BRAFV600-mutant melanoma (10), landscape studies of clinical tumor samples support MAPK reactivation as a convergent mechanism in a simple majority of patients (11, 12). Even with the combination of a type I RAF inhibitor (vemurafenib, dabrafenib, encorafenib) and an allosteric MEK inhibitor (cobimetinib, trametinib, and binimetinib), clinical acquired resistance occurs frequently and commonly through reactivation of the MAPK pathway (13). The specific alterations that mediate reactivation vary but converge to upregulate two key protein–protein interactions: RAF–RAF and RAF–MEK (Fig. 1). Thus, alternative (combination) strategies that can better combat these upregulated protein interactions are needed.
Varying tumor immune microenvironments across a wide range of tumor types can contribute potentially to disparate response rates or durability, regardless of the debate over lineage-dependent or -independent oncogene addiction. In BRAFV600-mutant melanoma, acquired resistance to MAPK-targeted therapy selects strongly for loss of intratumoral CD8+ T cells (14), and a pretreatment inflammation signature is associated with complete responses (15). Future developments of MAPK-targeted regimens should consider cancer histology–specific tumor immune microenvironments and the dynamic impact of MAPK-targeted agents on the evolving profile of systemic and local immunity. Therefore, a BRAF slam dunk may ultimately require an assist from next-generation MAPK-targeted agents or an immunotherapy teammate.
Disclosure of Potential Conflicts of Interest
A. Ribas is a consultant at Chugai, Amgen, Merck, Novartis, and Sanofi, is a co-founder at Tango, Lutris, and PACT, is an SAB member at Arcus, Compugen, Imaginab, Merus, Rgenix, and Bioncotech, is a board member at Arcus, and reports receiving a commercial research grant from Agilent. R.S. Lo reports receiving commercial research grants from Array, OncoSec, and Merck, and has received speakers bureau honoraria from Amgen, Array, Novartis, Roche, and Merck. No other potential conflicts of interest were disclosed.
This work was supported by NIH grants R35 CA197633 (to A. Ribas), P01 CA244118 (to A. Ribas and R.S. Lo), and P30 CA016042 (to A. Ribas), R01CA176111 (to R.S. Lo), R21CA215910-01 (to R.S. Lo), The Parker Institute for Cancer Immunotherapy (to A. Ribas), the American Skin Association (to R.S. Lo), and The Ressler Family Fund (to A. Ribas and R.S. Lo).