In the January 15, 2012, issue of Clinical Cancer Research, Kirkwood and colleagues published a study comparing the MEK inhibitor selumetinib with temozolomide in unselected metastatic melanoma. Although selumetinib did not improve survival or response, most responders had BRAF-activating mutations, and selumetinib has since demonstrated efficacy in BRAF-mutant melanoma. This study laid the groundwork for the evaluation of BRAF/MEK inhibitors in BRAF-mutant melanoma. Clin Cancer Res; 21(24); 5412–4. ©2015 AACR.
See related article by Kirkwood et al., Clin Cancer Res 2011;18(2) January 15, 2012;555–67
Major advances in the treatment of advanced melanoma over the past 3 years are directly linked to advances in our understanding of how activating mutations in canonical signaling and tumor-mediated T-cell dysfunction moderate melanoma tumorigenesis and progression. Deep sequencing approaches established that activating mutations in MAPK pathway constituents (BRAF/NRAS) are common, mutually exclusive, and carry oncogenic potential in human cancers (1). RAS mutations (typically at G12, G13, or Q61 residues) impair GTP hydrolysis, resulting in constitutive activation of GTP-bound RAS. BRAF is a serine/threonine protein kinase, and mutations (most commonly a glutamic acid substitution for valine at codon 600, V600E) occur within the kinase domain's activation segment. Oncogenic BRAF and NRAS mutations result in constitutive activation of downstream targets, including MAPK/ERK kinases 1 and 2 (MEK1/MEK2), and unrestrained proliferation. Seminal work by Curtin and colleagues identified distinct patterns of mutations in chronically sun-damaged skin (higher mutation load, lower BRAF/NRAS–mutant incidence) and intermittent sun-exposed skin (lower mutation load, BRAF/NRAS mutant)—underscoring the concept of melanoma as a heterogeneous disease with multiple genetically distinct subtypes (2). Concurrently, Allison and Honjo identified CTLA-4 and PD-1 to be key negative regulatory immune checkpoints hijacked by tumors to subvert antitumor immunity and mediate tolerogenic tumor microenvironments (3, 4). Subsequent work has since identified multiple other positive and negative checkpoints, and their characterization in murine and human models of various cancers is ongoing.
Consequently, agents targeting the MAPK pathway and inhibitory immune checkpoints were developed and tested. In melanoma, six new agents have been approved, including two targeting BRAF (vemurafenib and dabrafenib), one targeting MEK (trametinib), and CTLA-4 (ipilimumab) as well as two targeting PD-1 (pembrolizumab and nivolumab). Several other agents in each class are under evaluation, including the BRAF inhibitor encorafenib (LGX818; Array BioPharma); MEK inhibitors cobimetinib (GDC-0973/XL518; Exelixis) and binimetinib (MEK162; Array BioPharma); PD-1 inhibitors AMP-224 and AMP-514 (Amplimmune and GlaxoSmithKline) and pidilizumab (CT-011; Cure Tech); and PD-L1 inhibitors BMS-936559 (Bristol-Myers Squibb), MEDI4736 (MedImmune and AstraZeneca), atezolizumab (MPDL3280A; Roche), and avelumab (MSB0010718C; Merck).
The use of monoclonal antibodies targeting inhibitory immune checkpoints in melanoma is marked by relatively low response rates that, however, tend to be durable. Pooled analyses from multiple phase II/III trials reveal that ipilimumab has a lower response rate (11%–15%) compared with nivolumab (30%–35%) and pembrolizumab (45%–50%), with commensurately doubled 2-/3-year survival rates for PD-1 inhibitors compared with ipilimumab (5). Combined inhibition of both CTLA-4 and PD-1 increases response rates (to ∼50%) and median progression-free survival (PFS), albeit with significant attendant toxicity (55% grade 3/4 adverse events; ref. 5). Conversely, BRAF inhibitor use is marked by higher initial response rates (45%–53%) with inevitable progression at a median of 6 to 9 months typically through the acquisition of resistance mutations that reactivate the MAPK pathway. The addition of MEK inhibitors improves response rate (64%–76%) and response duration (9–11 months), while reducing the incidence of cutaneous toxicities—although progression is similarly inevitable (5).
The initial development of MAPK pathway inhibitors proceeded concurrently with inhibitors of MEK (AZD8330/ARRY-704, PD0325901, XL518, RDEA119, AS703026, CH5126766, CL-1040, GSK1120212, and AZD6244) and RAF (GDC-0879, RAF265, ARQ 680, XL281/BMS-908662, GSK2118436, PLX4032, and sorafenib; ref. 6). Importantly, initial clinical development (CL-1040) occurred prior to the identification of BRAF mutations in human cancer, and initial trials were not selectively enriched for patients with NRAS/BRAF–mutated tumors or cancers in which these oncogenic mutations were most commonly detected, such as melanoma (6).
Haass and colleagues first reported the antitumor activity of AZD6244 (ARRY-142886) in melanoma in a cell culture model where BRAF V600E–mutant (WM35, WM793, 1205Lu, and 451Lu), NRAS Q61R–mutant (SbCl2, WM2032, WM1361a, and WM852), and wild-type (C8161) lines revealed that AZD6244 inhibited melanoma growth additively with docetaxel in vitro and in vivo (7). Further studies characterized selumetinib (AZD6244/ARRY-142886) as an inhibitor of both MEK1 (IC50 14 nmol/L) and ERK1/2 phosphorylation (IC50 10 nmol/L) and led to evaluation of selumetinib in phase I trials against multiple tumor types (6).
In the phase II study we reported, 200 patients with advanced stage III/IV melanoma were assigned at random to receive either temozolomide or selumetinib (8). Most patients had cutaneous primaries (73%), but uveal (10%), mucosal (3%), and unknown primary tumors (14%) were also included. BRAF/NRAS and GNAQ mutation status were retrospectively analyzed in the cutaneous and uveal primaries, although enrollment did not stratify patients by mutational status. Of the 158 samples that had adequate archival tissue for evaluation, 73 tumors (46% of tested samples or 50% of cutaneous primaries) were BRAF mutant, 28 tumors (18% of tested samples or 19% of cutaneous primaries) were NRAS mutant, and only 4 tumors (3% of tested samples, or 20% of uveal primaries) were GNAQ mutant. The primary endpoint of this trial was PFS. Sample size of 182 (91 per arm) was calculated to provide 80% power to detect improvement in PFS with a hazard ratio (HR) of 0.74 (overall) at a one-sided 20% significance level, which would represent a HR of 0.67 among BRAF-mutant tumors. PFS comparisons were based on the assumption that BRAF- and NRAS-mutant melanoma comprised 60% and 30% of study participants, respectively, and that selumetinib would have an equal effect in NRAS-mutant and BRAF wild-type patients, a reasonable assumption given prior reports of selumetinib efficacy in the BRAF/NRAS wild-type C8161 cell line.
We reported no significant difference in response or PFS between the treatment arms among unselected patients and in patients with tumors displaying BRAF mutation. Response rates in both arms were low (6% vs. 9%)—and surprisingly lower in the selumetinib arm. Furthermore, patients assigned to selumetinib therapy had lower overall survival even after adjusting for imbalances in adverse prognostic factors between the two arms. Notably, 5 of 6 selumetinib responders and 1 of the 2 responding patients who had been crossed over to selumetinib from temozolomide following progression were identified to be BRAF mutants.
Several aspects of clinical trial design may have contributed to these findings. First, the study was likely underpowered to detect a true difference between treatment arms, given the discrepancy between the assumed (prestudy) and observed (on-study) proportions of BRAF/NRAS mutations, despite the increase in enrollment (to 200) from an original accrual target (of 182). Second, subsequent reports of atypical BRAF/KRAS (G464E and G12D, respectively) mutations in the C8161 cell line raise doubt as to selumetinib's efficacy in true BRAF/NRAS wild-type patients (9). Third, patients assigned to selumetinib had a higher incidence of known adverse prognostic features, including worse WHO performance status and higher lactate dehydrogenase levels. Although these differences were adjusted for in the final analysis of overall survival, the nonsignificant P values preclude definitive conclusions regarding the efficacy or nonefficacy of selumetinib, in either the overall or BRAF/NRAS–mutant populations. Finally, given that the study allowed crossover on investigator-assessed progression, selumetinib efficacy is difficult to ascertain as, although overall survival and PFS statistics were reported separately, unequal crossover suggests an element of investigator bias. Concurrently, two other phase II monotherapy studies in non–small cell lung cancer and colorectal cancer reported that selumetinib did not offer a response and/or survival advantage over standard chemotherapy (10, 11). Subsequent evaluation in metastatic uveal, BRAF wild-type (in combination with docetaxel), and BRAF mutant (in combination with dacarbazine) melanoma suggested that selumetinib improved PFS and response rates without significant improvements in overall survival (12–14).
Lessons learnt from these studies have fundamentally altered studies the next-generation MEK inhibitors. The MEK 1/2 inhibitor tremetinib (GSK1120212) was evaluated in a tiered approach that is characteristic of the current paradigm of targeted therapy development: (i) highly specific and potent candidate (trametinib MEK 1/2 IC50 0.92 nmol/L/1.8 nmol/L vs. selumetinib MEK1 IC50 14 nmol/L) was defined through high-throughput screening; (ii) maximum tolerated dose (MTD) and recommended phase II dose (RP2D) was defined in a nonselected cohort; (iii) RP2D was evaluated in selected tumor types enriched for the target of interest; and (iv) biologically active dose ranges were further characterized (15). Melanoma patients were included in the latter three phases, although enrollment was stratified by BRAF status, reflecting our growing understanding of the importance of pretreatment stratification by mutational status. Given reports of disease stabilization and tumor regression among BRAF-mutated (V600E/V600K) melanoma patients, subsequent phase II/III evaluation was restricted to BRAF V600E- or V600K-mutated melanoma (16).
Evaluation of BRAF inhibitors in melanoma has paralleled that of MEK inhibitors. Initial forays utilized sorafenib (BAY43-9006) in unselected patients—with only minimal (single-agent) or modest (in combination with chemotherapy) clinical activity that did not correlate with BRAF mutational status (17). Subsequent analysis suggested that while initially considered a selective RAF inhibitor, sorafenib targeted VEGFR-2, VEGFR-3, PDGFR, Flt-3, c-KIT, and FGFR-1 (18). Subsequent evaluation centered on more potent agents with greater BRAF selectivity. Vemurafenib (PLX4032/RG7204) was developed as a more potent agent with greater selectivity for mutant BRAF compared with wild-type BRAF (where it actually activates ERK). Significant activity in the phase I trial (69% in dose-escalation and 81% in dose-expansion phases) limited to BRAF V600E–mutated patients prompted rapid further evaluation in phase II/III trials and led to regulatory approval by 2011. The development of dabrafenib (GSK2118436) has been similarly rapid.
Underscoring the “One Companion Diagnostic–One Drug” paradigm, recent development of targeted therapies has been marked by concurrent development of in vitro companion diagnostic tests in concordance with guidelines from U.S. and European regulatory agencies. Companion diagnostic tests must be rigorously validated prior to adoption as they are prone to many of the same vagaries of testing as the drugs they accompany. To wit, the Cobas BRAF V600 assay was used in the pivotal BRIM studies to evaluate BRAF V600E mutation in formalin-fixed, paraffin-embedded tissues and subsequently approved for this purpose concurrently with vemurafenib. Subsequently, vemurafenib was noted to have similar efficacy in patients with V600K BRAF mutations—missed in the prior report as the Cobas assay was poor at detecting these less common mutations. Dabrafenib's approval in 2013 was accompanied by a companion diagnostic (THxID BRAF test) that detected both V600E and V600K mutations.
Since selumetinib was first evaluated for melanoma in 2006, the field has advanced rapidly, with the approval of multiple more highly selective BRAF and MEK inhibitors between 2011 and the present. Multiple modes of resistance to BRAF inhibition have been identified, including MAPK pathway–dependent (alternative splicing, increased gene amplification, and copy number gains, development of oncogenic mutations in NRAS or MEK), and MAPK pathway–independent [COT (MAP3K8) overexpression] mechanisms (19). Combining MEK and BRAF inhibitors increases response rate and prolongs duration of response by subverting multiple escape mechanisms. BRAF/MEK inhibitors are being combined in BRAF-mutant malignancies (primarily melanoma) with a myriad of other agents, including small-molecule inhibitors of Bcl-2 (navitoclax, NCT01989585), Akt (GSK2141795, NCT01902173); autophagy inducers hydroxychloroquine (NCT02257424) and metformin (NCT02143050); VEGF antagonist bevacizumab (melanoma, NCT01495988); and EGFR antagonist panitumumab in colorectal cancer (NCT01750918). Furthermore, BRAF/MEK inhibitors are being combined with CTLA-4–blocking agents, including ipilimumab (NCT01940809) and tremelimumab (NCT02027961); PD-1/PD-L1–blocking agents, including nivolumab (NCT02357732), pembrolizumab (KEYNOTE-022, NCT02130466), and MPDL3280A (NCT01656642); and in a BRAF/MEK inhibitor versus CTLA-4/PD-1 inhibitor sequence study (NCT02224781). It is hoped that these combinatorial strategies will improve upon the outcomes achieved with MEK/BRAF inhibition alone.
Disclosure of Potential Conflicts of Interest
J.M. Kirkwood is a consultant/advisory board member for Bristol-Myers Squibb, Merck, and Roche/Genentech. No potential conflicts of interest were disclosed by the other author.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NCI or the NIH.
Conception and design: D. Davar, J.M. Kirkwood
Writing, review, and/or revision of the manuscript: D. Davar, J.M. Kirkwood
This work was supported by the NCI of the NIH under award number P50CA121973 (to J.M. Kirkwood).