Cellular Mechanisms Underlying Complete Hematological Response of Chronic Myeloid Leukemia to BRAF and MEK1/2 Inhibition in a Patient with Concomitant Metastatic Melanoma

Chronic myeloid leukemia (CML) is almost universally characterized by a reciprocal translocation between the breakpoint cluster region (BCR) gene on chromosome 22, and the Abelson murine leukemia viral oncogene homolog 1 (ABL1) gene on chromosome 9, resulting in constitutive expression of a BCR–ABL1 fusion protein (1, 2). Imatinib, a small-molecule tyrosine kinase inhibitor that binds to and sterically interferes with the kinase domain of BCR–ABL1, effectively halts its constitutive and oncogenic cell-cycle signaling activity. Kinase inhibition, as typified by imatinib (3), now forms a major part of the therapeutic options in molecularly based subtypes of advanced lung, breast and medullary thyroid cancers, melanoma, gastrointestinal stromal tumors, and hematologic malignancies such as CML.

On the basis of high rates of oncogenic activating BRAF mutations, current kinase inhibitor-based therapy in melanoma is centered upon targets within the MAP kinase (MAPK) signaling pathway. Because of the importance of MAPK signaling in virtually all tissues of the body, effects of these inhibitors on nonmalignant cells has become a field of intense research.

The MAPK pathway is also known to be an important mitogenic signaling cascade in leukocytes, and its role in myeloproliferative disease is being increasingly established; however, the precise effects of MAPK inhibitors are highly dependent on the specific clinical context in which these agents are used. MEK inhibition has been demonstrated to rapidly correct aberrant myeloproliferative activity seen in murine models of both chronic and juvenile myelomonocytic leukemia, relating to MAPK hyperactivity induced by an activating KRAS mutation (4) or inactivation of the NF1 tumor-suppressor gene (5). Other studies have revealed that RAS-driven MAPK pathway dependence may be inadvertently intensified by BCR–ABL kinase inhibitors, leading to a synergistic antileukemic effect when used in combination with a MEK inhibitor (6). Similarly, a case of accelerated RAS-mutant leukemia was reported in a patient with metastatic melanoma receiving treatment with a BRAF inhibitor (7).

We report the case of a patient with BRAF V600E metastatic malignant melanoma who was treated with a combination of the V600-mutation–specific BRAF inhibitor dabrafenib, and the MEK1/2 inhibitor trametinib. Incidentally, he had a synchronous diagnosis of t(9;22) positive CML with an initial circulating leukocyte count of >100 × 10⁹/L. Despite not receiving BCR–ABL–targeted kinase inhibitor therapy, he achieved hematologic complete remission while on dabrafenib and trametinib treatment. Correlative ex vivo and in vitro analyses confirmed a dominant effect of the MEK inhibitor rather than the BRAF inhibitor on the patient's leukocytes, mediated by a relatively selective reduction in cell number and MAPK signaling within precursor (primitive) cell populations and reduced colony-forming potential. We also demonstrate that in CML cells, in contrast to several solid tumor types, the addition of a MEK inhibitor is able to overcome the paradoxical MAPK activation induced by BRAF inhibition. These findings suggest that further investigation of the role of MAPK pathway inhibition in de novo CML in a clinical setting is warranted.

The patient was enrolled in a phase I/II clinical trial of dabrafenib and trametinib in advanced melanoma (BRF113220, GlaxoSmithKline, ClinicalTrials.gov number NCT01072175; ref. 8), approved by the Institutional Review Board of Austin Health, Australia (Austin Health Human Research Ethics Committee, project number H2011/03955). All patient-derived tissue/blood samples were provided voluntarily under the auspices of the Cancer Biobanking and Research protocol approved by the Institutional Review Board of Austin Health, Australia (H2012/04446), to which the patient had previously consented. Blood samples were collected at multiple time points prior to, during, and after the patient's period of treatment with kinase inhibitors for metastatic melanoma. Baseline leukocyte samples were also obtained from peripheral blood leukapheresis. Clinical follow-up data were as per trial protocol and standard clinical care.

Differential cell counts and blood films were obtained from diagnostic blood samples collected and analyzed through the local pathology service provider (Austin Pathology). Cytogenetic analyses were performed by the Victorian Cytogenetics Service, and BCR–ABL1 counts performed at the Royal Melbourne Hospital (Parkville, VIC, Australia).

To estimate the effects of in vivo kinase inhibitor treatment on MAPK signaling in leukocyte subsets, peripheral blood mononuclear cells (PBMC) isolated from blood samples taken at baseline, week 8 and week 20, were subjected to multicolor flow cytometry. Two stains were applied: a “monocytic” stain, comprising CD45 (Cy5; BD Pharmingen; cat. no. 555484), CD34 (PE-Cy7; Beckman Coulter PN A51077 clone 581), HLA-DR (FITC; BD Pharmingen; cat. no. 555811), CD14 (APC; Beckman Coulter IM2580 clone RMO52), pERK (PE; Cell Signaling Technology; #5682S) or equivalent isotype controls, and live/dead stain (Pacific Blue, LIVE/DEAD fixable violet dead cell stain, Invitrogen L34955); and a “myeloid” stain, comprising CD45, CD34, HLA-DR, CD11b (APC; BD Pharmingen; cat. no. 550019), CD16 (APC-Alexa Fluor750; Beckman Coulter PN A66330 clone 3G8), pERK, or isotype controls, and live/dead stain. All cells were gated on forward and side-scatter characteristics, live cells, and CD45⁺ cells. Subpopulations of interest included blast (CD34⁺/HLA⁻DR⁺), immature/promonocyte (CD34⁻/HLA⁻DR⁻/CD14⁻), mature monocyte (CD34⁻/HLA⁻DR⁻/CD14⁺), myelocytes (CD34⁻/HLA⁻DR⁻/CD11b⁺/CD16⁻) and metamyelocytes/mature neutrophils (CD34⁻/HLA⁻DR⁻/CD11b⁺/CD16⁺). MAPK pathway activation status was measured by both the proportion of cells staining positive for phosphorylated ERK1/2 (pERK) relative to isotype control, and comparison of relative pERK staining intensities between time points in monocytic and myeloid lineage cells.

Three BCR–ABL harboring cell lines were used for in vitro studies. K562 (BCR-ABL b3-a2 fusion) was obtained from institutional stocks, MEG-01 (BCR-ABL b2-a2) was a gift from Dr. Ashley Ng (Walter and Eliza Hall Institute, Parkville, VIC, Australia), and KU-812 (BCR-ABL b3-a2 fusion) was obtained from Cell Bank Australia (Westmead, NSW, Australia); identity was confirmed by short-tandem repeat (STR) profiling. All cell lines were maintained in semiadherent culture in RPMI-1640 media (Gibco, Life Technologies, Mulgrave VIC, Australia) supplemented with 10% fetal bovine serum (Sigma), 1% glutamine (glutaMAX) and 1% penicillin/streptomycin (both from Gibco, Life Technologies) and incubated in 5% atmospheric CO₂ at 37°C.

In vitro analysis of kinase inhibitor effects on proliferation was performed using the K562 leukoerythroblastic cell line. Cells were plated out at seeding densities of 4,000 cells per well in 96-well plates and treated with dabrafenib at doses of 0, 0.01, 0.1, 1, or 5 μmol/L, trametinib at doses of 0, 0.1, 1, 10, or 100 nmol/L, or their respective combinations, for a period of 72 hours. The doses used were chosen to cover the typically achieved human plasma concentrations in vivo. Proliferation was determined colorimetrically by addition of MTS reagent (1 in 6 dilution of CellTiter 96 AQueous One Solution Cell Proliferation Assay; Promega), incubation for 1.5 hours, measurement of the absorbance at 490 nm, and comparison to the absorbance in untreated cells following background correction.

DMSO-cryopreserved aliquots of leukapheresis samples collected prior to commencement of any tyrosinase kinase inhibitor (TKI) were thawed, resuspended in IMDM/10%FCS, and then cultured, as previously described in detail (9). Cultures were set up at 5 × 10⁴ cells/mL in 0.3% Bacto Agar (Becton Dickinson; #214010), 25% fetal bovine serum (Gibco; Life Technologies) and IMDM (Invitrogen; containing glutamine, 2-mercaptoethanol, asparagines, and DEAE Dextran) in a fully humidifed 5%CO₂/air atmosphere before scoring. Day 7 colony-forming cells (CFC; mature myeloid progenitors) were scored if >30 cells were present, and 14 day BFU-E and GM-CFC (immature erythroid and myeloid progenitors, respectively) were scored if >50 cells were present. Day 7 cultures were stimulated with G-CSF (2,000 U/mL (Neupogen, Amgen), and day 14 cultures with GM-CSF (100 ng/mL; R&D Systems Inc.), IL3 (100 ng/mL; R&D Systems Inc.), SCF (100 ng/mL; R&D Systems Inc.), and erythropoietin (200 ng equivalent to 4 U/mL; Aranesp, Amgen). Cultures were in triplicate for initial titrations and duplicate for mixture experiments. EC₅₀s were calculated from log-transformed normalized data using GraphPad Prism software (GraphPad Software Inc.).

For in vitro phosphoprotein studies, K562, MEG-01, and KU-812 cells were resuspended in serum-free media, allowed to equilibrate for 30 minutes, and subsequently plated into 6-well plates (1.5 × 10⁶ cells/well) containing imatinib mesylate, dabrafenib, trametinib, or DMSO only (vehicle control, 0.1%–0.2%) to the final doses as indicated in results. Cells were incubated for 2 hours prior to harvest and lysis in RIPA buffer (Pierce) containing the manufacturer's recommended concentrations of PhosSTOP phosphatase inhibitor and cOmplete ULTRA EDTA-free protease inhibitor (Roche Diagnostics GmbH). Protein concentrations were estimated using the BCA method (Thermo Scientific Pierce BCA Protein Assay Kit; cat. no. 23225) and samples adjusted to uniform final concentrations, prior to the addition of NuPAGE LDS Sample Buffer (×4) and NuPAGE sample reducing agent (×10; both from Novex; Life Technologies; cat. nos. NP008 and NP009). Samples were electrophoresed on NuPAGE Novex Bis–Tris 4% to 12% precast gels with SeeBlue Plus2 prestained protein standard (Novex; Life Technologies) prior to transfer to nitrocellulose membrane by dry (iBlot Dry Blotting System, Invitrogen), or semi-wet (Invitrogen XCell II Blot Module; cat. no. EI9051) methods as necessary for optimal transfer. Erk phosphorylation was determined using mouse anti-human p44/42 MAPK (Erk1/2; 3A7) at 1:2,000, rabbit anti-human phospho-p44/42 MAPK (Erk1/2; Thr202/Tyr204; 197G2) at 1:2,000, and rabbit anti-human α/β-tubulin at 1:2,000 (Cell Signaling Technology; cat. nos. 9107, 4377, 2148) followed by IRDye 680RD goat anti-mouse IgG and IRDye 800CW goat anti-rabbit IgG secondaries at 1:40,000 and 1:20,000, respectively (LI-COR, Prod. Nos. 926-68070 and 926-32211). BCR–ABL1 activity was measured using the PathScan Bcr/Abl Activity Assay cocktail (1:500; Cell Signaling Technology; cat. no. 5300) consisting of rabbit antibodies raised against phospho-c-Abl (Tyr245), phospho-Stat5 (Tyr694), phospho-CrkL (Tyr207), and Rab11 (loading control), followed by anti-rabbit antibody as above. Wet-transfer combination treatment blots also contained mouse anti-human β-actin (1:3,000) primary followed by goat anti-mouse secondary as above as an additional loading control. After staining, membranes were dried and scanned on the LI-COR ODYSSEY Infrared Imaging System and densitometry performed in the ODYSSEY software. Densitometry results were normalized to Rab11 (ABL activity assay); ERK phosphorylation was expressed as the ratio of phosphorylated to total ERK1/2 and normalized to nontreated (DMSO) control.

For analysis of BIM isoform protein expression following kinase inhibitor treatment, K562, MEG-01, and KU-812 cells were incubated in full-serum media for 24 hours in the presence of imatinib, dabrafenib, and/or trametinib at doses indicated in the results, prior to harvest and processing as described above. Following transfer to nitrocellulose membranes usin the iBlot system, membranes were probed with mouse anti-β-actin (as above) and a rabbit monoclonal anti-BIM antibody [Y36] at 1:1,000 (ab32158; Abcam), followed by secondary staining and visualization as above. For brevity, where blots have been displayed in figures, only relevant bands of confirmed correct size according to markers are shown.

A 26-year-old man with a history of moderately severe psoriasis managed with topical steroids and prior PUVA therapy, was diagnosed with stage IIIB malignant melanoma (10) of unknown primary, involving the left inguinal lymph nodes. He underwent a left groin lymph node dissection, followed by resections of left upper thigh subcutaneous and left iliac lymph node recurrences within 6 months. He received local postoperative radiotherapy after the second recurrence. A further 10 months later he underwent resection of a solitary right lower lobe pulmonary metastasis detected on routine surveillance imaging. Surveillance FDG-PET scan performed 4 months later confirmed relapsed disease with multiple metastases involving the left adrenal, bone, lymph nodes, subcutaneous sites, and liver. At that time the patient was asymptomatic, with an Eastern Cooperative Oncology Group (ECOG) performance status (11) of 0 (unrestricted function).

Genetic testing performed on the previously resected lung metastasis identified a V600E mutation in the BRAF gene. The patient was considered for enrollment in a phase I/II clinical trial of combination BRAF and MEK inhibitors in advanced melanoma (BRF113220, GlaxoSmithKline, clinical trial identifier NCT01072175). Screening investigations performed as per trial protocol demonstrated a markedly elevated white cell count (WCC) of 109 × 10⁹/L, with a left-shifted white cell differential. No eosinophilia or basophilia were reported per the automated analyzer differential. Hemoglobin was 106 g/L, and the platelet count was 145 × 10⁹/L. The blood film showed a leukoerythroblastic picture with marked leukocytosis and occasional tear-drop cells, with no other specific features. Other than an elevated LDH of 579 (normal range, 98–192 IU/L), biochemistry results, including electrolytes, urea, and creatinine and liver enzymes, were within normal limits. The patient had no known medical history of blood dyscrasias and was unaware of any previous abnormal blood investigations. Of note, a full blood count (FBC) sample taken at the time of the lung metastectomy, which had been performed at a different institution, was subsequently located. This showed normal hemoglobin and platelet counts, mild leukocytosis with total WCC 25 × 10⁹/L, with a left-shifted differential, and no eosinophilia or basophilia.

The marked leukocytosis was initially considered to be a leukemoid reaction caused by paraneoplastic release of G-CSF secondary to metastatic melanoma (12), and the leukoerythroblastic film attributed to marrow involvement with melanoma. Peripheral blood was sent for molecular analysis, although results for this were anticipated to take at least 1 week. Given the rapidity with which the patient had relapsed and the high burden of metastatic disease, commencement on therapy was considered a higher clinical priority. The patient was enrolled in the BRF113220 clinical trial and randomized to receive both the BRAF inhibitor dabrafenib (GSK2118436, 150 mg twice daily), and the MEK-inhibitor trametinib (GSK1120212, 2 mg daily). Treatment commenced immediately following a cytoreductive leukapheresis.

One week after starting on trial therapy, molecular analysis of peripheral blood was completed and demonstrated the presence of BCR–ABL1 t(9;22) fusion gene, with both a major b2-a2 (p210) transcript, and a minor e1-a2 (p190) transcript detectable. A subsequent manual review of the baseline blood film and repeat differential noted mild eosinophilia and basophilia. A bone marrow biopsy confirmed a diagnosis of CML in chronic phase, and did not show evidence of infiltration by melanoma. Cytogenetic analysis demonstrated t(9;22)(q34;q11.2) without additional abnormalities or evidence of deletion of the derivative chromosome 9.

Despite this concomitant diagnosis, as the patient's overall prognosis was deemed to be more significantly affected by metastatic melanoma than CML he was permitted to remain on trial protocol therapy, providing that he did not commence imatinib for treatment of the CML concurrently, given the lack of safety data for the combination of dabrafenib/trametinib with other kinase inhibitors. As CML was in chronic phase and was to be monitored closely with weekly FBC and regular Hematology review, he continued on trial.

The patient initially tolerated BRAF- and MEK-inhibitor therapy without any adverse effects and performance status remained excellent. Despite lack of specific therapy for CML, total WCC decreased within 4 weeks of commencing on trial treatment, and continued to decline over the subsequent 6 months (Fig. 1). High resolution melt curve analysis of the patient's leukocytes confirmed the absence of a BRAF V600 mutation.

Restaging for melanoma after 8 weeks on trial demonstrated a partial response by RECIST 1.1 criteria (13), and at week 16 he had a near-complete response on CT imaging together with a complete metabolic response on FDG-PET imaging. At week 24, the patient developed a solitary site of melanoma progression in a left adrenal metastasis, but remained on trial due to continuing clinical benefit with ongoing good control of melanoma at all other sites.

During enrollment on trial, the patient required two treatment breaks. The first occurred at week 24 due to development of asymptomatic reduction in left ventricular ejection fraction. Both inhibitors were withheld for 4 weeks as per study protocol. Left ventricular dysfunction resolved without specific intervention and dabrafenib and trametinib were restarted with a protocol-mandated 25% dose reduction of trametinib.

The second treatment break occurred at week 38, at the time that the patient underwent elective laparoscopic left adrenalectomy for treatment of the solitary progressing metastasis. After a period of 48 hours following surgery, both inhibitors were recommenced at the same doses.

Weekly FBC monitoring throughout treatment confirmed a normalized WCC of 8.4 × 10⁹/L after 35 weeks on trial, with associated resolution of the left-shifted myeloid series, eosinophilia and basophilia (Supplementary Table S1), consistent with a complete hematologic response. Nevertheless, repeat PCR in peripheral blood for the BCR–ABL1 fusion gene showed persistently detectable levels (Supplementary Table S2). Following left adrenalectomy, total WCC increased, peaking at week 41 (Fig. 1 and Supplementary Table S2) with associated recurrence of a left-shifted differential. At the same time, the patient experienced a significant flare of psoriasis that required temporary intensification of topical agents (calcipotriol/betamethasone).

Temporal trends in the patient's WCC, differential and blood/bone marrow BCR–ABL levels are shown in Fig. 1 and Supplementary Tables S1 and S2. Rather than a proportional cytoreduction in WCC, there was a greater reduction in myelocyte, promyelocyte, eosinophil, and blast cell numbers than lymphocyte subtypes. Following the 4-week break from dabrafenib/trametinib therapy (weeks 24–28), myeloid cells at varying stages of maturation increased in number. Total WCC again peaked approximately 3 weeks following adrenalectomy, consisting largely of myelocytes, metamyelocytes, and band neutrophils, potentially suggesting a left-shift precipitated by perioperative stress.

The patient ceased kinase inhibitors late in week 48 due to clear progression of his melanoma. At that time, CML remained well controlled with a circulating leukocyte count of 14.0 × 10⁹/L. The patient subsequently received several lines of cytotoxic chemotherapy with resultant multilineage cytopenias, confounding ongoing interpretation of CML behavior.

Representative FACS plots and gates for the monocytic and granulocytic stains are shown in Supplementary Fig. S1. Proportions of cells within each defined subpopulation are shown in Supplementary Table S3; due to the relative depletion of mature myeloid cells during standard Ficoll-based PBMC isolation, the constituent leukocyte subpopulations observed by flow cytometry differ markedly from those seen in contemporaneous peripheral blood films and are likely unreliable for more mature granulocytic populations. CD34⁺/HLA⁻DR⁺ blast numbers declined by over 6-fold from baseline to week 20 (P < 0.0001, one-way ANOVA; Fig. 2A). The proportion of blasts with measurable levels of pERK declined at week 8 and remained low at week 20 (P = 0.0014 across all time points, one-way ANOVA; Fig. 2B; Supplementary Table S3). pERK staining intensity remained stable as measured by the mean fluorescence intensity (MFI) of pERK-PE (Supplementary Table S3). Mature monocytes were almost entirely (>99.0%) positive for pERK at all time points (Supplementary Table S3); however, the mean intensity of pERK staining increased significantly from baseline to week 20 (3,949 vs. 4,735). This may be due to the uneven separation of cells into pERK-high or -low subpopulations (Fig. 2C).

Cell staining to define myeloid/granulocytic maturation stages demonstrated similar results for the CD34⁺/HLA⁻DR⁺ blast population as were seen in the monocytic stain series (Supplementary Table S3). Myelocyte (CD34⁻/HLA⁻DR⁻/CD11b⁺/CD16⁻) pERK positivity reduced progressively from baseline to week 20 (P = 0.0001, one-way ANOVA; Fig. 2D; Supplementary Table S3) together with a greater than 50% reduction of mean pERK signal intensity (Fig. 2E). Metamyelocyte (CD11b⁺/CD16⁺) pERK-positive proportions fluctuated moderately, with no sustained trend in pERK staining intensity over the observed time points (Fig. 2F and Supplementary Table S3).

Next, the effect of each kinase inhibitor on hematopoietic progenitor cells from the patient was analyzed. Tested dose ranges of either drug were chosen to encompass typically achieved maximal plasma concentrations in vivo, namely 1 to 2 μmol/L for dabrafenib and 10 to 30 nmol/L for trametinib. Both drugs inhibited the formation of colony formation from leukapheresis mononuclear cells in a dose-dependent fashion (Fig. 3). Trametinib was significantly more potent than dabrafenib for each colony type analyzed. Both mature (day 7 CFC) and more primitive myeloid progenitors (day 14 GM-CFC) were inhibited by concentrations of trametinib achievable in vivo. Primitive erythroid progenitors (BFU-E) displayed similar sensitivity to trametinib; hemoglobinization and size of individual colonies were also significantly negatively affected in a dose-dependent fashion. Next, the effects of coincubation of trametinib with dabrafenib at two fixed concentrations achievable in vivo were tested. Divergent behavior was observed according to the type of progenitor cell being assayed. Dabrafenib had no significant effect on BFU-E and GM-CFC sensitivity to trametinib (Fig. 3B). However, for more mature day 7 myeloid CFC, dabrafenib 1.25 μmol/L demonstrated an additive effect with trametinib on the inhibition of day 7 CFC progenitor cell growth, with the calculated half-maximal effective concentration (EC₅₀) for trametinib reducing from 17 nmol/L (95% CI, 13–23) to 4 nmol/L (95% CI, 3–6). In contrast, in the presence of 0.31 μmol/L dabrafenib, the calculated half-maximal effective concentration (EC₅₀) of trametinib was higher, 32 nmol/L (95% CI, 25–42) than for trametinib alone, suggesting partial antagonism (Fig. 3B).

Because of limited available patient-derived sample, the effect of the BRAF and MEK1/2 inhibitors was then evaluated further in a BCR-ABL–dependent leukemic cell line, K562. Proliferation curves at constant trametinib dose (Fig. 4A) indicate the relative lack of effect of dabrafenib on proliferation unless cells are cotreated with clinically relevant doses of trametinib (≥1 nmol/L; black lines). Very high doses of trametinib (100 nmol/L) produced >50% inhibition of cellular proliferation with minimal additional contribution from dabrafenib at any dose.

Similarly, the effect of up-titrating trametinib in combination with each given dose of dabrafenib (Fig. 4B) clearly demonstrates the dominance of trametinib in driving a dose-dependent inhibition of proliferation.

Notably, at low dabrafenib doses (<1 μmol/L), the inhibitory effect of trametinib on proliferation was reduced relative to dabrafenib-untreated cells (Fig. 4A), while at higher dabrafenib doses of 1 and 5 μmol/L there was additive antiproliferative effect. This biphasic effect is consistent with the in vitro analyses of the patient's mature myeloid progenitor cells (i.e., day 7 CFC).

Overall, there was a dominant antiproliferative effect of kinase inhibitor therapy mostly attributable to the effect of trametinib at clinically relevant doses. This is consistent with a MEK-mediated effect in this cell line that lacks an activating BRAF V600 mutation.

As expected, imatinib treatment of K562 cells markedly reduced phosphorylation of BCR–ABL1 with consequent downstream reduction of STAT5 phosphorylation and ERK1/2 phosphorylation, confirming that the MAPK pathway responds to upstream modulation of BCR–ABL1 activity (Fig. 5A). Trametinib monotherapy for 2 hours in K562 cells inhibited ERK1/2 phosphorylation relative to basal (untreated) conditions in a dose-dependent manner. Consistent with potential paradoxical activation of MEK/ERK signaling in RAS-activated cells that lack activating BRAF V600 mutations, dabrafenib monotherapy markedly increased ERK1/2 phosphorylation with a maximal response between 0.5 and 1 μmol/L (Fig. 5A and B). In combination with dabrafenib, trametinib doses attenuated (1 nmol/L) or abrogated (10–100 nmol/L) this paradoxical activation, and reduced ERK phosphorylation to below basal levels (Fig. 5C and D). These findings were further confirmed in studies of the MEG-01 and KU-812 cell lines, both of which displayed even greater sensitivity to imatinib with an identical pattern of paradoxical ERK activation in response to dabrafenib, which could be overcome by trametinib cotreatment (Fig. 5E and F).

To assess for potential direct off-target effects of dabrafenib and/or trametinib on BCR–ABL1 activity, we analyzed phosphorylation of BCR–ABL1, as well as the BCR–ABL1 interacting adaptor protein CRKL and the direct downstream target STAT5 in the same cells used to study ERK phosphorylation. No statistically significant effect on activation of BCR–ABL1 was observed in response to dabrafenib or trametinib treatment, either alone (Fig. 5A and B) or in combination (Fig. 5C and D). Dabrafenib induced a dose-dependent attenuation of STAT5 phosphorylation that was statistically significant for 5 μmol/L dabrafenib (P < 0.05 vs. control) although was no longer significant with the addition of trametinib (Supplementary Fig. S2). Neither kinase inhibitor had significant effect on CRKL phosphorylation. Analogous studies in MEG-01 and KU-812 cells using the principal clinically relevant doses demonstrated no significant off-target effects of dabrafenib or trametinib on downstream targets of BCR-ABL.

In all three cell lines tested (K562, MEG-01, and KU-812), BIM-EL was markedly more abundantly expressed than the shorter isoforms (Supplementary Fig. S3). Following 24 hours of exposure to imatinib, increased expression of BIM-EL became evident in K562 (marginally), of all three isoforms in MEG-01, with minimal change in expression of any BIM isoform seen in KU-812 cells (Supplementary Fig. S3). No significant effects on BIM expression were seen with dabrafenib and/or trametinib in K562 or MEG-01 cells; however, dabrafenib in combination with the higher dose of trametinib (10 nmol/L) increased BIM isoform expression in KU-812 cells only.

Imatinib is an extremely effective therapy in chronic phase of CML, with over 65% of patients achieving complete cytogenetic response (CCR) after 12 months of therapy and 5-year overall survival rates of approximately 90% (14, 15). Data from the International Randomized Study of Interferon versus STI571 (IRIS) trial comparing IFNα and imatinib (STI571), demonstrated impressive event-free, disease-specific, and overall survival advantages for patients receiving first-line imatinib (16), together with both deep and durable molecular responses, including major metabolic responses (MMR) in 85% to 90% at 5 to 6 years (17). Subsequent generations of inhibitors, such as dasatinib and nilotinib, and more recently ponatinib, are also in clinical use (18). These agents are, however, susceptible to such setbacks as inadequate target specificity, unfavorable side-effect profiles, reduced efficacy due to intra- and intertumoral molecular and genetic heterogeneity, and development of resistance through a variety of mechanisms.

In the present case, a complete hematologic response was seen after a 28-week period of treatment with BRAF and MEK1/2 kinase inhibitors. Concurrently, metastatic melanoma was well-controlled, with near-complete radiologic response by week 16. In addition, the patient ceased kinase inhibitors late in week 48 due to clear progression of his melanoma, despite well-controlled leukocyte counts at that time. Although a humorally mediated effect secondary to the melanoma may have altered the patient's CML disease course, the kinetics of the CML response were delayed and discordant from the melanoma disease course, suggesting a direct antileukemic effect of the BRAF inhibitor, the MEK inhibitor, or both. In particular, the gradual decline in circulating leukocytes and restoration of a normal mature differential count suggests an effect on a primitive or precursor cell population. In support of this, in vitro analysis of the patient's leukocytes revealed dose-dependent inhibition of colony formation across multiple precursor cell populations of varying maturity, with a dominant and potent effect of trametinib relative to dabrafenib.

Evaluation of phospho-ERK1/2 levels in patient-derived leukocytes as a surrogate for MAPK pathway activity revealed only modest differences over time in most subpopulations tested, with the exception of myelocytes and mature monocytes. Flow cytometric analysis of patient-derived leukocyte samples is subject to several potential confounding effects on subset enumeration and quantification of residual phosphoprotein levels introduced by sample collection, handling, purification, and freeze-thaw, which largely depleted the dominant circulating cell type (neutrophils) in this patient. Nonetheless, these data suggest that differences in relative sensitivity to MAPK inhibition exist between leukocyte subclasses and maturation stages. Although mature subpopulations were relatively depleted in this analysis, the slow kinetics of response and breadth of cell types affected is most consistent with a more primitive target cell population being ultimately responsible for the observed hematologic response in this patient.

The majority of the antiproliferative effect observed in treatment of the K562 CML cell line was secondary to MEK inhibition, consistent with other studies using MEK inhibitors in this cell line (6, 19–22). Although MEK inhibition appears to have a biphasic dose-response effect on erythroid differentiation in K562 cells (23), it demonstrated a dose-dependent inhibitory effect on erythroid precursors in our patient-derived cells, suggesting a difference in underlying MAPK and erythropoietin receptor (EPO-R) signaling in these differing contexts. In addition, regardless of MEK inhibitor dose, intermediate doses of dabrafenib tended to antagonize the antiproliferative effect of trametinib on K562 cells, resulting in a net proliferation either close to, or higher than control. Western blot analysis demonstrated the paradoxical activation of ERK phosphorylation in K562 cells treated with dabrafenib alone; this effect was dose-dependently reversed by trametinib cotreatment.

Several other studies have demonstrated that pharmacologic inhibition of MAPK signaling can increase cell death and decrease cell proliferation in CML cells in vitro (6, 19–22, 24). In a competing fates model of BCR–ABL1 inhibition by inhibitors, such as imatinib and dasatinib, cells typically commit to apoptosis more rapidly than they can restore responses to growth factor–induced survival signals (25). Accordingly, addition of the MEK inhibitor PD184352 to imatinib increased apoptosis in BCR–ABL1–positive human myeloid leukemia cells (22) and in another study, enhanced mitochondrial damage and caspase activation following dasatinib treatment, including in cells overexpressing BCR–ABL1 and in some forms of BCR–ABL1 point mutations (20). Another study reported a novel CRAF dysregulation mechanism that would, theoretically, be amenable to MEK/ERK inhibition (26). The proapoptotic molecule BIM is known to be an important mediator of the response to imatinib in CML; we did not find a consistent change in BIM expression in three model CML cell lines in response to dabrafenib/trametinib, suggesting that while differences exist between the cell lines, BIM is unlikely to represent a critical role in the beneficial hematologic effects of MAPK targeting in this setting (27, 28).

These findings are interesting considering what is known about the paradoxical MAPK-activating effects of BRAF inhibitors in cells that are wild-type for BRAF. As we and others have previously reported, MAPK-activated cells, commonly those harboring oncogenic RAS mutations, may in fact be stimulated to proliferate when exposed to V600 mutation-selective BRAF inhibitors (7, 29). Insofar as RAS appears to be a downstream target of BCR–ABL1, CML cells may thus be considered RAS-activated and therefore susceptible to BRAF inhibitor-driven proliferation (Fig. 6). Furthermore, CML cells with acquired resistance to BCR–ABL1–targeted kinase inhibitors may also display paradoxical MAPK activation. This has been observed in CML cells harboring the common BCR-ABL^T315I secondary-resistance mutation during continued exposure to kinase inhibitors such as imatinib, due to weak off-target binding of inhibitor to BRAF. This phenomenon could be exploited by the addition of a MEK inhibitor, which appeared to induce synthetic lethality in this context (6).

Interacting partners and downstream modulators of BCR–ABL1 signaling are numerous and include RAC and RHO family members, STAT5, CRKL, and the PI3K/AKT pathway in addition to MAPK members. BCR–ABL1 activates MAPK signaling in a GRB2 and SOS-1–dependent manner (24, 30–36), but leads to marked MEK-dependent negative feedback of receptor-level growth factor signaling, as well as direct negative feedback on RAS itself (Fig. 6; ref. 25). Although we did not find a significant direct effect of dabrafenib or trametinib on BCR–ABL1 activation as measured by phosphorylation of ABL-Tyr245 and CRKL, possibly due to the kinetics of our assay, we found clear evidence that dabrafenib dose-dependently reduced STAT5 phosphorylation. This effect was likely due to paradoxical activation of MEK in the context of BRAF inhibitor monotherapy, with consequent intensification of MEK-dependent negative feedback on STAT5 phosphorylation, as has been described in other studies (25). Although trametinib cotreatment had a very small counter-effect on STAT5 phosphorylation, trametinib monotherapy had little effect on STAT5 phosphorylation relative to untreated cells (Supplementary Fig. S2). This suggests that the set-point of this feedback loop is such that negligible negative feedback on STAT5 occurs until MEK activity is raised above basal levels, in this case by dabrafenib. Whether some of the functional differences observed between BRAF monotherapy and combined BRAF/MEK inhibition may relate to this divergent effect on STAT5, either in preference or in addition to divergent effects on MAPK pathway signaling, is unclear and warrants further investigation.

In our case, despite complete hematologic response, there was no significant decline in BCR–ABL1 gene expression either in blood or by repeat bone marrow cytogenetics, indicating that MEK inhibition was not effective at inducing molecular or cytogenetic response. Given the clear relationship between depth of molecular response and long-term outcome in CML (16), MEK inhibition as a single-agent therapy would thus seem unsuitable, but could be considered in combination with other therapies, or potentially in cases of therapeutic resistance such as the BCR–ABL1 T315I mutation, for which current first- and second-generation BCR–ABL1 kinase inhibitors are ineffective (37–39). Although unintended MAPK activation may be undesirable in other cancer settings, it may paradoxically be advantageous in therapy-resistant CML by driving cells into the downstream proapoptotic block afforded by concurrent MEK inhibition. Exactly how dependent the efficacy of this approach is on the degree of upstream MAPK activation is unclear and may influence its therapeutic potential.

This case documents a unique observation of complete hematologic response to MEK inhibition in a patient with chronic-phase CML and concomitant metastatic melanoma. It confirms differential effects of MEK inhibition on defined leukocyte subsets, highlighting the inhibitory effects on CML cell proliferation and colony formation. Although previous preclinical work suggests a role for MEK inhibition in acquired BCR–ABL1 inhibitor resistance, the response seen in our patient occurred without prior exposure to, or selection against, imatinib or related kinase inhibitors. These findings may support a more prominent therapeutic role for MAPK targeting in CML that is not limited to exploitation of paradoxical MAPK activation seen in some drug-resistant CML populations. MEK inhibition appears to affect a broad spectrum of hematopoietic lineages in CML, but our correlative analyses indicate that its primary clinically relevant effect may be on a primitive subpopulation. Further clinical evaluation of MAPK targeting as add-on therapy to conventional BCR–ABL1 kinase inhibitors, and in the setting of resistance to these agents is warranted.

M.C. Andrews is a consultant/advisory board member for GlaxoSmithKline. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.C. Andrews, N. Turner, A.W. Roberts, A. Behren, J. Cebon

Development of methodology: M.C. Andrews, A. Behren

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.C. Andrews, N. Turner, J. Boyd, A.W. Roberts, A.P. Grigg, J. Cebon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.C. Andrews, N. Turner, J. Boyd, A.W. Roberts, A. Behren

Writing, review, and/or revision of the manuscript: M.C. Andrews, N. Turner, J. Boyd, A.W. Roberts, A.P. Grigg, A. Behren

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Boyd

Study supervision: A. Behren, J. Cebon

Other (involved in decisions care of the patients and extensively reviewed and revised the manuscript on more than one occasion): A.P. Grigg

The authors thank the patient and his family for their support of this work; Tina Cavicchiolo and Noel Micallef for expert clinical trial management; Nektaria Dimopoulos and James Lynam for their assistance and insights with flow cytometry; Laura J. Vella for guidance with Western blotting, and Hongdo Do for HRM analysis.

M.C. Andrews is supported by a National Health and Medical Research Council of Australia Postgraduate Medical Scholarship (#1055456) and A.W. Roberts by a Practitioner Fellowship (#637309) and Program Grant (#1016647). This work was funded in part by Ludwig Cancer Research, and Operational Infrastructure Support Program funding of the Victorian State Government.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.