Summary: Dual MAPK and CDK4/6 targeting is an emerging strategy in melanoma, but toxicity and acquired resistance are limitations. In this issue, two groups (Teh and colleagues and Romano and colleagues) show that therapeutic resistance mechanisms converge on the PI3K pathway, and inhibition of this pathway's mediators can overcome this resistance. Cancer Discov; 8(5); 532–3. ©2018 AACR.
See related article by Romano et al., p. 556.
See related article by Teh et al., p. 568.
Therapies that target mediators of the MAPK pathway have been shown to be effective in the treatment of patients with metastatic melanoma (1). The most efficacy has been found with dual BRAF/MEK inhibitor therapy in patients with BRAFV600 mutations; however, activity has also been seen with MEK and ERK inhibitors in patients with NRAS and non-V600 BRAF mutations (1, 2). Unfortunately, the great majority of patients with melanoma who receive MAPK inhibition will progress, and more-effective targeted therapies are needed. One possible strategy is dually targeting MAPK and the cyclin- dependent kinases 4 and 6 (CDK4/6). Genetic aberrations of cell-cycle regulators such as CCND1, CDKN2A, and CDK4 are common in melanoma, are potentially susceptible to CDK4/6 inhibition, and are associated with poorer outcomes in the setting of MAPK pathway inhibition (3, 4). Additionally, preclinical data show improved outcomes, in vitro and in vivo, in NRAS-mutant melanoma cell lines (5).
In patients, combined CDK4/6 and MAPK inhibition has been challenging. First, it is difficult to give full doses of each drug continuously, which leads to reduced drug exposure that may limit efficacy. As an example, the best-tolerated dose of the combination of encorafenib (BRAF inhibitor), binimetinib (MEK inhibitor), and ribociclib (CDK 4/6 inhibitor) requires a dose reduction of encorafenib (from 450 to 200 mg), which likely limits the activity of the combination (6). Specifically, in patients with BRAF-mutant melanoma, the response rate of all three drugs (∼50%) is lower than the response rate of the encorafenib and binimetinib doublet (∼65%). Dose finding has also been challenging with the doublet of binimetinib and ribociclib in patients with NRAS-mutant melanoma, and the recommended phase II dose utilizes a ribociclib dose (200 mg daily) that is one third the dose approved for women with breast cancer (7). The preliminary efficacy has been modest at this dose level, suggesting that optimization of dual MEK and CDK4/6 inhibition is needed.
In this issue of Cancer Discovery, two groups present data that address the major challenges of dual CDK4/6 and MAPK inhibition in melanoma, namely, optimal treatment scheduling and therapeutic resistance (8, 9). Teh and colleagues (8), understanding the challenges of toxicity with continuous dosing, explored the utility, in an NRAS-mutant melanoma model, of three dosing schemas incorporating intermittent dosing: group 1, continuous MEK inhibitor with intermittent CDK4/6 inhibitor; group 2, continuous CDK4/6 inhibitor with intermittent MEK inhibitor; group 3, intermittent therapy with both CDK4/6 and MEK inhibitors. Interestingly, all three regimens were similarly well tolerated (using the fair but crude readout of animal weight), but the animals in group 1 experienced the most complete responses and control of tumor growth, compared with group 2 which had improved outcomes compared with group 3. Upon identifying the optimal dosing schedule, of the three tested, resistant tumors from each group were compared, and those from group 1 had striking reactivation of phosphorylated ERK (pERK) and phosphorylated S6 (pS6), suggesting reactivation of not only the MAPK pathway but also the PI3K pathway. Further analysis of group 1 resistant tumors showed marked copy-number gain of NRAS that seemed to drive pS6 activity and thus resistance. A focused drug screen, evaluating inhibitors of the PI3K pathway and downstream MAPK pathway mediators (ERK and RSK) against resistant tumors, identified that only the mTORC1/2 and beta-sparing PI3K inhibitor effectively inhibited pS6. Finally, use of a dual mTORC1/2 inhibitor demonstrated single-agent activity against resistant cell lines in vivo and more dramatic activity in combination with a MEK and a CDK4/6 inhibitor.
A remarkably similar story was told by Romano and colleagues (9), although from a different starting point. In this paper, the investigators begin with a case report of a patient with NRAS-mutant melanoma treated on the phase I/II trial of ribociclib and binimetinib (R/B) who initially responded and then quickly developed progressive disease. Analysis of whole-exome sequencing from tumors collected at key time points (pre-R/B, on-R/B, and then three postprogression samples) was performed with a goal to identify genetic alterations not present in the pretreatment sample but identified at an allelic frequency of more than 25% in progression samples. Only one gene mutation was found that fit this selection criterion and was a known oncogene, PIK3CAE545K. Knowing this, the group retrospectively analyzed circulating tumor DNA isolated at various time points and were able to identify the presence of this mutation within 16 days of commencing R/B. This rapid rise suggests the emergence of a subclone under the evolutionary pressure of treatment, and so a deeper analysis of the pretreatment tumor looking for this mutation was performed, including subdivision of the tumor into seven sections and incorporating blocker displacement amplification. Sure enough, PIK3CAE545K was identified in three adjacent regions of the seven total regions and persisted in every postprogression sample tested. Next, preclinical studies were performed demonstrating that PIK3CAE545K increases S6K1 and pS6 in NRAS-mutant melanoma lines and that this is associated with resistance to dual MEK and CDK4/6 inhibition. Furthermore, the addition of an mTOR inhibitor or, better yet, an S6 kinase inhibitor overcomes this resistance.
Summarizing the work of these two groups is quite simple: combined MEK and CDK4/6 inhibition drives therapeutic resistance in NRAS-mutant melanoma by multiple mechanisms that activate the PI3K pathway generally and S6 kinase specifically. What to do about this is also straightforward: Targeting PI3K pathway mediators with small-molecule inhibitors overcomes this resistance, but rolling out this approach as a treatment strategy will not be easy. In patients previously treated on MAPK and CDK4/6 inhibition, repeat mutational analysis should be performed to look for abnormalities that might predict PI3K pathway activation (e.g., NRAS amplification and PIK3CAE545K mutation), and agents targeting that pathway should be considered for those patients. Yet, although sequential therapy is doable, it possibly is not optimal, because the best data in the preclinical modeling suggest that up-front combination therapy is a more efficacious approach.
The big challenge with up-front combination therapy is that based on the doublet toxicity data, it is safe to say that continuous triplet therapy will not be tolerable. So, clinical trials testing a third agent in combination with MEK and CDK4/6 inhibition, for example, will need to adopt dose-escalation strategies that will maximize drug exposure but limit dose-limiting toxicities. These studies will need to have built-in flexibility, allowing for “schedule escalation” of various continuous/intermittent schedules of agents, ideally at therapeutic doses, as opposed to the traditional dose-escalation schemes that involve titrating upward on one drug, then the other, which rarely end up leading to adequate therapeutic exposure of both (or all three) agents. Data from preclinical studies will be critical to help guide the development of these strategies, such as the work by Teh and colleagues, suggesting that any approach of MEK plus CDK4/6 inhibition in patients should incorporate continuous MEK inhibition, because there are too many variables with too many agents to sufficiently test all possibilities in the clinic. Finally, the challenges of developing multidrug regimens that provide efficacious drug exposure yet are tolerable may require truly outside-the-box thinking. For example, better detection of very low-frequency subclonal populations or treatment-emergent (through true clonal evolution) populations might allow for the development of nonintuitive treatment strategies that target the emerging and more dangerous clones than the dominant clones predicted to respond to a given single-agent or doublet therapy (10). What is clear is that high-level translational and preclinical work, as exemplified by these two papers, continues in an iterative process to better inform clinicians who launch clinical trials that help better inform the basic and translational researchers.
Disclosure of Potential Conflicts of Interest
R.J. Sullivan is a consultant/advisory board member for Merck, Amgen, Novartis, and Array BioPharma.