Genomically Driven Tumors and Actionability across Histologies: BRAF-Mutant Cancers as a Paradigm

A wealth of data now suggests that molecular aberrations may be shared across multiple histologies (1). As an example, BRAF mutations can be detected in melanoma, colorectal tumors, lung and ovarian cancers, hairy cell leukemia, histiocytosis and many other related disease types (2; Fig. 1; Table 1). Indeed, a small subset of almost all types of malignancies may harbor a BRAF mutation (3, 4). Of special importance in this regard is the fact that several drugs that effectively target the BRAF-mutant protein product have been developed (Table 2). For instance, the BRAF inhibitors, vemurafenib and dabrafenib, have both been approved for BRAF-mutant melanoma based on results from the phase III BRIM-3 study (5) and the phase III BREAK-3 study (6), respectively.

A key conundrum now debated in the cancer community is whether or not targeted drugs approved for one type of histology should be administered to other histologies harboring the cognate aberration. For instance, should a BRAF inhibitor approved for BRAF-mutant melanoma be given to a patient with a BRAF-mutant tumor other than melanoma? A corollary to this question is the precise criteria needed in order to extrapolate predictive data on a biomarker for a given targeted therapy in one cancer to another cancer. These questions are of tremendous importance for the following reasons: (i) molecular aberrations, in particular amplifications, loss, and mutations, do not appear to segregate well by histology (1, 2, 4); (ii) numerous targeted drugs are becoming clinically available and they have been developed to inhibit a specific cancer signal that may be found in multiple tumor types, hence their rational application would be in tumors bearing the cognate target (3); and (iii) molecular anomalies are found in a very small percentage of diverse cancers (7), and the rarity in each histologic type presents a near-impossible challenge for classic randomized or even nonrandomized trials to determine efficacy histology by histology.

Newer study designs are beginning to accommodate these challenges. For instance, histology-agnostic trials (so-called bucket or basket trials) might include patients with a wide variety of histologies as long as they all harbor the cognate aberration. As an example, a histology-agnostic trial of the BRAF inhibitor vemurafenib can include diverse types of cancers, providing that they carry BRAF mutation (e.g., VE BASKET study; 8). However, these types of trials are still often perceived as signal finding. If a variety of histologies respond, what should be the next steps to approval and/or pay or coverage? To what extent can we be certain or do we need to be certain that each histology bearing the mutation will respond before it is acceptable to administer drugs across cancers based on their molecular, rather than histologic, classification? Does molecular classification actually represent a biology-based nosology?

Herein we review this topic, using BRAF-mutant malignancies as a paradigm. The choice of BRAF was considered apt for the following reasons: (i) BRAF mutations as well as other BRAF anomalies (amplifications, fusions) have been described in a wide variety of tumors; (ii) two BRAF inhibitors and a MEK inhibitor have already been approved for BRAF-mutant melanoma; and (iii) there is a rich literature demonstrating responses, albeit at times in small numbers of patients, with the use of BRAF inhibitors in a variety of BRAF-mutation bearing cancers (9, 10). On the other hand, BRAF-mutant colorectal cancers have proved more resistant to BRAF inhibitor monotherapy, hence striking a cautionary note. The observations in BRAF-mutant tumors may therefore inform future conceptualization of genomically driven treatment.

BRAF is mutated in about 15% of all cancers (3, 11) and BRAF mutations can be found in solid tumors, hematologic malignancies, and related disease types (Table 1). For some cancers, BRAF mutations are very frequently detected: melanoma [40%–60% of patients (12)] and hairy cell leukemia [∼100% (13)].

The predominant mutation detected in BRAF-mutated cancers is the V600E mutation, representing approximately 70% to 90% of all mutations in BRAF (12, 14–16). Substitution of glutamic acid (E) for valine (V) at codon 600 of the BRAF protein affects the activation segment of the protein by mimicking the phosphorylation of the kinase domain, causing a change in structure that favors the active conformation (14, 17). Experimental studies have confirmed that the BRAF V600E mutations are activating, resulting in increased BRAF kinase activity in in vitro studies, as well as activation of downstream effectors and oncogenic transformation in cell-based studies (12, 18, 19).

Other activating mutations in BRAF include additional mutations affecting codon 600 that result in substitutions other than glutamic acid. In BRAF-mutated melanoma, the BRAF V600K mutation is found at a frequency of approximately 7% to 19% (16, 20). Other rare mutations affecting codon 600 include BRAF V600D (0.1%), BRAF V600R (1%), and BRAF V600M (0.3%; 20). Furthermore, activating mutations in BRAF that affect codons other than 600 include L597 substitutions (0.5%), and K601E (0.7%; 20). Table 1 lists several other non-V600 mutations in BRAF and their frequencies in detected cancers (for responsiveness of non-V600E mutations to BRAF inhibitors, see section entitled “BRAF mutations other than V600E”).

In addition, inactivating or “low-activity” mutations in BRAF have been identified and characterized; they typically involve substitutions at codon 594 (19, 21), although missense mutations at other codons (including codon 466) have also been shown to result in BRAF kinase inactivation or reduced activation (18).

In addition to mutations, other types of BRAF aberrations are found in cancer, including amplification and BRAF fusions. BRAF amplification involving either the wild-type gene or mutant versions of the gene is predicted to result in increased BRAF activity in tumor cells (22). In some cases where BRAF mutations are rare, BRAF amplifications dominate. For example, while mutations in BRAF are found in only 1% of breast cancers (23), BRAF amplification has been reported in 30% of basal-like breast tumors (24). Other cancers where BRAF amplification is more frequent than BRAF mutations include ovarian serous cystadenocarcinoma (12% vs. 0.6%, respectively; 25, 26) as well as prostate adenocarcinoma (∼5% vs. 1.6%, respectively; 25, 26).

BRAF fusions such as KIAA1549-BRAF and FAM131B-BRAF are frequently found in gliomas with the KIAA1549-BRAF fusion detected in up to 70% of pilocytic astrocytomas (27, 28). The KIAA1549-BRAF is an arrangement created by a tandem duplication event, while FAM131B-BRAF is generated by a large deletion event; however, both result in constitutive activation of BRAF through duplication of the BRAF activation domain, but with deletion of the N-terminal inhibitory domain (29, 30). The KIAA1549-BRAF fusion has been reported in preclinical studies to be resistant to PLX4720, the research analog of vemurafenib, due to RAF dimerization, but remains sensitive to a second-generation BRAF inhibitor (31). In addition, one case study described a patient with a spindle cell neoplasm harboring the KIAA1549-BRAF fusion as well as a homozygous deletion of PTEN, and frameshift mutations in CDKN2A, SUFU, and MAP3K1 who had a 25% reduction in tumor volume following a combination therapy consisting of sorafenib (a weak BRAF inhibitor), temsirolimus, and bevacizumab, suggesting that the KIAA1549-BRAF fusion may be responsive to certain BRAF inhibitors in the clinic, though the precise reason for response is confounded by the other drugs in the regimen (32). The responsiveness of FAM131B-BRAF is currently not reported in the literature. While infrequent as compared with mutations, BRAF fusions have also been observed in melanoma in anywhere from 4% to 8% of “pan-negative” cases (defined as tumors negative for mutations in BRAF, NRAS, KIT, GNAQ, and GNA11). Two BRAF fusions, PAPSS1-BRAF and TRIM24-BRAF, were both shown to result in activation of the MAPK pathway, and were both reported to be sensitive to the MEK inhibitor trametinib but not the BRAF inhibitor vemurafenib as assessed by inhibition of MEK1/2 phosphorylation (33).

In summary, multiple alterations in the BRAF gene can occur. The sensitivity or lack thereof to BRAF or MEK inhibitors may vary depending on the alteration.

The connection between BRAF-mutant, specifically BRAF V600E-mutant, cancers and response to BRAF inhibitors was first established in melanoma patients, where it was observed that anywhere from 50% to 60% of melanomas harbor the activating BRAF V600E mutation (12, 34). A phase I study reported that, in comparison with the 10% to 20% response rates for nontargeted therapies approved for the treatment of melanoma, a response rate of up to 81% was observed for BRAF V600E-mutated melanoma patients given the BRAF inhibitor vemurafenib (35). Furthermore, matched targeted therapy in heavily pretreated melanoma patients in the phase I setting (using mainly BRAF and MEK inhibitors), showed longer PFS as compared with each patient's first-line standard therapy (36). The phase II BRIM-2 study reported a best overall response rate of 53% and median duration of response of 6.8 months from treatment with dabrafenib for previously treated melanoma patients whose tumor harbored the BRAF V600E mutation (37, 38). Finally, on the basis of a phase III trial comparing vemurafenib to dacarbazine, in which it was reported that the response rate for vemurafenib was 48% as compared with the 5% response rate for decarbazine (5), vemurafenib received FDA approval for treatment of patients with melanoma whose tumors harbor the BRAF V600E mutation (39).

On the heels of vemurafenib, another BRAF inhibitor that proved to be efficacious in treating BRAF V600E-mutated melanoma patients was dabrafenib (6, 40), which received FDA approval for the treatment of patients with melanoma having the BRAF V600E mutation (41). Vemurafenib and dabrafenib are perfect examples of the superior efficacy that can be achieved by employing drugs that target a biomarker that drives oncogenesis; in patients with BRAF V600E-mutated melanoma, BRAF-directed therapy results in substantially better outcomes as compared with nontargeted therapy approaches.

Other approved drugs that act as BRAF inhibitors but are not specifically approved for BRAF-mutant cancers include regorafenib, which is approved for colorectal cancer and gastrointestinal stromal tumors (GIST); it is also currently in a phase II trial recruiting for colorectal cancer patients with any BRAF or RAS mutation (42).

Additional BRAF inhibitors that are either approved or currently in clinical development are summarized in Table 2. Some of these drugs are in trials selecting for BRAF-mutant cancers. For example, LGX818 is in a phase III trial for BRAF V600E- or BRAF V600K-positive melanoma (43) and in a phase II trial for BRAF V600-positive cancers (44).

Trametinib is currently the only approved MEK1/2 inhibitor. However, there are several other investigational MEK1/2 inhibitors being evaluated in clinical trials, including binimetinib (MEK162), cobimetinib (GDC-0973, XL518), pimasertib, refametinib, selumetinib (AZD6244), and PD-0325901.

Another drug approved for melanoma with BRAF V600E or BRAF V600K mutations is the MEK1/2 inhibitor trametinib (GSK1120212). Trametinib was approved on the basis of results from a phase III trial (NCT01245062) of 322 melanoma patients who harbored either BRAF V600E, BRAF V600K, or both mutations that were randomized to receive either a chemotherapy regimen (paclitaxel or dacarbazine) or trametinib. Patients receiving trametinib had a superior PFS as compared with patients receiving chemotherapy, with a median PFS of 4.8 months versus 1.5 months, respectively (45). Trametinib in combination with dabrafenib for melanoma with BRAF V600E or BRAF V600K mutations was subsequently approved on the basis of a trial of 162 melanoma patients who harbored either the BRAF V600E or BRAF V600K mutations, who were randomized to either trametinib 2 mg daily in combination with dabrafenib, trametinib 1 mg daily in combination with dabrafenib, or single-agent dabrafenib. The trametinib 2 mg daily in combination with dabrafenib yielded superior objective response rates and response duration (76% and 10.5 months, respectively, as compared with 54% and 5.6 months, respectively, in the single-agent dabrafenib arm; P < 0.05; 46).

In colorectal cancer, dabrafenib and trametanib combinations have also shown activity in BRAF-mutated disease (47). Of 43 patients, five (12%) achieved a partial response or better, including one (2%) complete response, with duration of response > 36 months; 24 patients (56%) achieved stable disease as best confirmed response. Ten patients (23%) remained in the study > 6 months.

There exist FDA-approved companion diagnostics for vemurafenib (COBAS 4800 BRAF V600 Mutation Test) to identify those melanoma patients harboring the BRAF V600E mutation (39, 48) and for the approved combination regimen of dabrafenib and trametinib (THxID BRAF kit) for those melanoma patients harboring the BRAF V600E or BRAF V600K mutation (41). However, these diagnostics, while well validated, are limited by their inability to detect other mutations, as well as amplifications and rearrangements in BRAF. They also cannot detect additional genomic abnormalities that coexist in most patient tumors. Other technologies such as next-generation sequencing are better suited to the more comprehensive analysis that is often needed (49).

Since the approval of vemurafenib for BRAF V600E-mutated melanoma, accumulating evidence presented in published reports supports the idea that what works for BRAF V600E-mutated melanoma is often also effective for other cancers characterized by the BRAF V600E aberration (Table 3). Dabrafenib was granted the Breakthrough Therapy designation for treatment of patients with metastatic BRAF V600E mutation-positive non–small cell lung cancer (NSCLC; 50) based on a phase II study that reported a response rate of 54% for BRAF V600E mutation—positive, pretreated NSCLC patients receiving treatment with dabrafenib (51). A phase I study reported three papillary thyroid cancer patients whose disease was characterized by BRAF V600E, had either a partial response or stable disease in response to treatment with dabrafenib (52), a gastrointestinal stromal patient whose tumor harbored the BRAF V600E mutation experienced continuing tumor regression while being treated with dabrafenib (53), a child with glioblastoma multiforme harboring the BRAF V600E aberration had complete clinical regression from treatment with vemurafenib (54), glioma patients whose tumors carried the BRAF V600E mutation were reported to respond to treatment with vemurafenib (55), and several case studies have reported clinical benefit from treatment with vemurafenib for BRAF V600E-mutated hairy cell leukemia patients (9, 56–58). One study consisting of three BRAF V600E-mutated multisystemic and refractory Erdheim-Chester disease patients reported substantial and rapid clinical and biologic improvement from treatment with vemurafenib lasting 4 months (10, 59). In addition, a phase II trial of vemurafenib in BRAF V600-mutated Erdheim-Chester disease (a non-Langerhans histiocytosis) and Langerhans cell histiocytosis reported an overall response rate (defined as percentage of patients with either complete or partial response) of 36.4%, with one patient achieving a complete response (9.1%) and three patients with partial response (27.3%); none of the 11 evaluable patients had progressive disease (60).

Of interest, a basket study of vemurafenib reported clinical activity of vemurafenib in predominantly BRAF V600E-mutated nonmelanoma cancers. Complete or partial responses, tumor regression and prolonged disease stabilization were observed in several tumor types including NSCLC, Erdheim-Chester disease or Langerhans' histiocytosis, anaplastic thyroid cancer, pleomorphic xanthoastrocytoma, cholangiocarcinoma, salivary-duct cancer, ovarian cancer, clear-cell sarcoma, glioblastoma, anaplastic ependymoma, pancreatic cancer, and carcinoma of unknown primary types, and among patients with colorectal cancer who received vemurafenib and cetuximab (8). This study confirms that multiple histologic types of cancer with BRAF mutations respond to BRAF inhibitors, though precise response rates may differ (and were hard to elucidate exactly in this study since numbers of patients in each group were small). In some types of tumors (e.g., colorectal cancer), BRAF inhibitors need to be given with other drugs that impact resistance pathways.

BRAF mutations affecting codon 600 have generally been associated with sensitivity to BRAF inhibitors, with the drugs vemurafenib and dabrafenib being approved for melanoma patients with the BRAF V600E mutation (39, 61), and trametinib approved for melanoma patients with the BRAF V600E and BRAF V600K mutation (62). The BRIM7 and coBRIM studies have reported mutations at codon 600 to be sensitive to combination therapies of the MEK inhibitor cobimetinib with vemurafenib (63, 64).

In addition, there are a few clinical reports on the responsiveness of BRAF mutations other than those affecting codon 600. A melanoma patient harboring the BRAF K601E aberration was reported to have prolonged partial response to the MEK inhibitor trametinib (65); however, two other melanoma patients harboring this aberration were reported to not derive clinical benefit from treatment with dabrafenib (40). Resistance to treatment, though, may not necessarily be due to insensitivity of the mutation, but rather to other unrelated molecular drivers that may be present in patients and supplant the role of BRAF. One patient with BRAF L597R-harboring melanoma was reported to derive clinical benefit after treatment with vemurafenib (66); however, L597R was also associated with disease progression in a melanoma patient with brain metastases who received BRAF inhibitor treatment (67). Again, it is plausible that the brain is a sanctuary in this patient, and that resistance may not be due to insensitivity to the drug. Similarly, a melanoma patient with the BRAF L597Q aberration had disease progression within 2 months of treatment with a BRAF inhibitor (67), but another melanoma patient with this aberration had a partial response to treatment with trametinib (68). Finally, a BRAF L597S–positive melanoma patient was reported to derive clinical benefit after treatment with the MEK inhibitor, TAK-733 (69).

Other activating mutations in BRAF affecting codons other than 600 that have been shown to be sensitive to either MEK inhibitors or BRAF inhibitors in preclinical studies include BRAF L597R, BRAF L597Q, BRAF L597S, and BRAF K601E (69). However, some activating mutations in BRAF have been shown to be less sensitive or even resistant to BRAF inhibitors in preclinical studies. For example, the activating BRAF L505H mutation appears to be resistant to vemurafenib (70). In addition, some mutations in BRAF, including substitutions at codon 466, are inactivating (18); BRAF inhibitors are predicted to be ineffective for such mutations.

Despite the evidence that BRAF inhibitors are efficacious in melanoma and several other different cancers characterized by the BRAF V600E mutation, at least one cancer type seems more resistant to these drugs—colorectal cancer. Fewer than 10% of BRAF V600E-mutated colorectal cancers have responded to BRAF inhibitor monotherapy in early-phase trials (40, 71). The underwhelming response of BRAF V600E-mutated colorectal cancer patients to BRAF inhibitors initially may seem to provide evidence that biomarker-driven approaches to the treatment of cancers are not sufficient unless taken within histologic context. However, a deeper analysis of the literature appears to suggest otherwise.

Preclinical studies initially discovered that, in BRAF V600E-positive colorectal cancer cells, inhibition of BRAF V600E by vemurafenib results in decreased negative feedback of the EGFR pathway (72, 73), Therefore, the attenuated clinical response to vemurafenib in BRAF V600E-mutated colorectal cancer compared with BRAF V600E-mutated melanoma could be attributable to differences in the importance of the EGFR signal present in these two cancer types (74). Specifically, these studies reported that melanoma cells express low levels of EGFR and as such are not poised, as high EGFR-expressing colorectal cancer cells are, for vemurafenib-stimulated EGFR pathway activation. Both studies demonstrated that a combination of vemurafenib with an EGFR inhibitor, such as cetuximab, erlotinib, or gefitinib, in BRAF V600E-mutated colorectal cancer cells, inhibited vemurafenib-induced feedback activation of EGFR and increased therapeutic efficacy in vitro and in tumor xenografts (72, 73).

The therapeutic efficacy of a BRAF inhibitor in combination with an EGFR inhibitor observed in preclinical models has subsequently translated into clinical efficacy seen in patients. One case study described a patient with colorectal cancer whose tumor had both the BRAF V600E mutation and EGFR amplification, with a partial remission from the combination of vemurafenib and panitumumab (an EGFR antibody; 75). Another case study reported a patient with colorectal cancer whose tumor carried the BRAF V600E mutation and proved to be refractory to treatment with several drugs including cetuximab monotherapy, but achieved symptom stabilization from the combination of cetuximab and vemurafenib (76). Another case report described a BRAF V600E-mutated colorectal cancer patient who had a 7-month long PFS and mixed response to combination therapy of sorafenib, a weak BRAF inhibitor, and cetuximab, with some areas showing dramatic improvement and other areas showing stable disease (77). Recently, a vemurafenib basket study confirmed that, in colorectal cancer, salutary effects can be attained with the combination of vemurafenib and an EGFR inhibitor in BRAF-mutated colorectal cancer (8). Studies are now ongoing with these combinations [NCT01719380 (78), NCT01791309 (79), NCT01750918 (80)], with early promising results from NCT01719380 recently being presented at the 2014 and 2015 ASCO Annual Meeting. This phase I study of the BRAF inhibitor LGX818 in combination with cetuximab, with or without the PIK3CA inhibitor BYL719, in advanced BRAF-mutated and KRAS wild-type colorectal cancer patients reported no complete responses but 7 (∼30%) partial responses from the dual combination of LGX818 with cetuximab and 6 (∼30%) partial responses from the triple combination of LGX818, cetuximab, and BYL719 (81). Finally, preliminary results from a phase Ib trial of vemurafenib in combination with irinotecan and cetuximab in BRAF-mutated advanced cancers and colorectal cancer patients reported that 6 of 17 evaluable BRAF V600E-mutated colorectal cancer patients achieved a partial response Hong and colleagues (82).

Another perspective for this issue derives from studies showing that, in colorectal cancer patients, aberrations in the KRAS/BRAF axis often coexist with aberrations in the PI3K/AKT/mTOR axis (83–85), with each being a resistance pathway for the other (86, 87). These co-mutations occur in other cancers as well, but not as frequently.

Finally, CRAF activation may occur as a resistance mechanism. Recently, BRAF-mutated colorectal cancer has been shown to respond to combinations of a BRAF inhibitor dabrafenib together with the MEK inhibitor trametinib. Of 43 patients, five (12%) achieved a partial response or better, including one (2%) complete response, with duration of response > 36 months; 24 patients (56%) achieved stable disease as best confirmed response. Ten patients (23%) remained in the study > 6 months (47).

Targeting multiple pathways of resistance in BRAF-mutated colorectal cancer may be the most efficacious route as suggested by recent preliminary results from a phase II study of the triple combination of vemurafenib, anti-EGFR antibody panitumumab, and trametinib versus dual combination therapy of vemurafenib with trametinib. The triple combination had acceptable tolerability and improved response rate as compared with doublet therapy in BRAF V600E-mutated colorectal cancer patients, with 9 (26%) of patients treated with the triple combination achieving either a partial or complete response as compared with 2 (10%) treated with the doublet combination (88). Table 4 summarizes clinical studies using BRAF and/or MEK inhibitors in the treatment of BRAF-mutated colorectal cancer.

Taken together, it appears likely that BRAF mutations may indeed be actionable in colorectal cancers, but that a more complete picture of the molecular portfolio of the patient needs to be taken into account, and customized combinations created, a situation that is almost certainly relevant to other tumor types in which resistance pathways in addition to BRAF are also activated. These observations may also be relevant to BRAF-mutated melanoma, in that, patients who relapse or are resistant upfront to BRAF inhibitors may still need such a drug in their treatment regimen, albeit in combination with other drugs that target molecular aberrations that coexist and mediate resistance. Indeed, even melanoma patients whose tumors are sensitive to BRAF inhibitors probably need customized combinations, since most do not achieve complete remissions, and relapse after a few months is routinely observed. Several preclinical reports have identified mechanisms of resistance to BRAF inhibitors in BRAF-mutated melanoma that would suggest the use of specific BRAF inhibitor-containing combinations. These include but are not limited to generation of drug-tolerant microenvironments characterized by high integrin β1/FAK/Src signaling via paradoxical BRAF inhibitor-mediated activation of melanoma-associated fibroblasts (89; suggestive of use of BRAF inhibitor in combination with FAK or Src inhibitor), upregulation of many ERBB pathway genes in response to BRAF inhibition (90; suggestive of use of BRAF inhibitor in combination with ERBB inhibitors), and BRAF inhibitor-induced activation of cryoprotective autophagy (91; suggestive of use of BRAF inhibitor in combination with autophagy inhibitor).

A key question for the treatment of cancer in the emerging genomics era is whether or not we can extrapolate predictive data on a biomarker for a given targeted therapy in one cancer to another cancer. In this scenario, clinical research-informed treatment options from one histology with that biomarker are applied to another cancer type with that same biomarker where such research is not available or limited in scope. The corollary to this question is whether or not the biomarker defines the tumor and represents a nosology in and of itself. Herein, we have used BRAF-mutant cancers as one example in support of such a restructuring of cancer classification, but others have proposed a similar genetic-centric nosology such as ALKomas for tumors encompassing multiple organs but harboring aberrations in ALK that have either been shown to or are likely to respond to ALK inhibitors (92). While not clear-cut, increasingly the data show that the presence of certain genomic drivers, such as BRAF aberrations, predicts response to cognate inhibitors across multiple (though not all) tumor types. Even in tumors such as colorectal cancer that are considered an important exception, emerging data suggest that BRAF is a relevant target, but must be prosecuted in the context of combination therapy (e.g., a BRAF inhibitor together with a EGFR or MEK inhibitor) that impacts pertinent coactivated pathways. Indeed, an important lesson from the “resistance” of BRAF-mutant colon cancer may be extrapolated to “sensitive” tumors such as melanoma, in that most patients with BRAF-mutant melanoma do not achieve complete remissions on BRAF inhibitors and they often relapse within months; these “sensitive” tumors, like “resistant” BRAF-mutant colorectal cancer, may require combination regimens to overcome the therapeutic plateau.

It seems that the current research points to a fork in the road. On the one hand, the approach might be to continue to explore and approve genomic marker-driven treatments histology by histology. The advantage to this strategy is more definitive data in each defined histology. The disadvantages, however, are substantial. Most importantly, it will be nearly impossible from the point of view of resources to perform studies for each biomarker within each histology or each organ of origin for a tumor. Therefore, one of the key points that should be established is the overall response rates in bucket, histology-agnostic biomarker driven trials, and whether these response and benefit rates suggest that patients are better off being treated on the basis of the biomarker, regardless of histology, or by classic treatments. A related alternative is to approve the targeted therapy across tumor types, but require post-approval outcome collection, perhaps within a phase IV setting such that indications can be refined if needed. The basic principle of clinical research should be maintained, that is, showing that, overall, patients have improved outcomes with a certain approach. Importantly, nonresponding patients regardless of histology may have additional genetic drivers, be it co-mutations in a given biomarker, or alterations in other biomarkers, that may provide a biologic rationale for their lack of response and suggest new drug development opportunities or combinations therapies that may be more effective for these patients, with BRAF-mutated colorectal cancer being an emerging example of the latter.

In conclusion, it is increasingly clear that many oncogenic drivers do not segregate by organ of tumor origin. In these cases, a new method of classifying cancers on the basis of the genomic aberration itself may be useful. New trial designs, such as histology-agnostic, genomically defined bucket trials are being used more frequently. Using the example of BRAF aberrations, the literature shows that this powerful driver is present in multiple tumor types and that the use of cognate inhibitors can be effective across many, even if not all, tumor types. Touted exceptions such as colorectal cancer, however, may still be responsive to BRAF inhibitors, albeit at lower rates. Furthermore, when important coexisting pathways such as EGFR or CRAF that drive innate resistance are taken into consideration, and combination therapy given (e.g., combine BRAF and EGFR inhibitor regimens or BRAF and MEK inhibitors in colorectal cancer), impressive responses can be seen. By the same token, even in histologic types with high response rates, there are nonresponders as well as a large numbers of patients that relapse after BRAF inhibitor monotherapy. In order to abrogate resistance, these tumors may require combination strategies that include a BRAF inhibitor and a drug that impacts other concomitantly activated driver pathways. Indeed, in melanoma, the combination of a BRAF and MEK inhibitor is associated with outcomes superior to BRAF inhibitor therapy alone (46). Taken together, these observations suggest that consideration for genomically based approval strategies are rational and merit consideration.

F. Janku has commercial research grants from Biocartis, Trovagene, Foundation Medicine, Illumina, Novartis, Biomed Valley Discoveries, Agios, Astellas, Plexxikon, and Deciphera; and is a consultant/advisory board member for Trovagene, Deciphera, Novartis, and Foundation Medicine. J. Rodon is a consultant/advisory board member for Novartis and Lilly. R. Kurzrock is the co-founder of RScueRX; has other commercial research support from Genentech, Merck Serono, Foundation Medicine, Pfizer, Guardant, and Sequenom; has ownership interest (including patents) in RScueRX; and is a consultant/advisory board member for Sequenom. No potential conflicts of interest were disclosed by the other authors.