CIC Mutation as a Molecular Mechanism of Acquired Resistance to Combined BRAF-MEK Inhibition in Extramedullary Multiple Myeloma with Central Nervous System Involvement
Abstract
Combined MEK-BRAF inhibition is a well-established treat- ment strategy in BRAF-mutated cancer, most prominently in malignant melanoma with durable responses being achieved through this targeted therapy. However, a subset of patients face primary unresponsiveness despite presence of the acti- vating mutation at position V600E, and others acquire resis- tance under treatment. Underlying resistance mechanisms are largely unknown, and diagnostic tests to predict tumor response to BRAF-MEK inhibitor treatment are unavailable.
Multiple myeloma represents the second most common hematologic malignancy, and point mutations in BRAF are detectable in about 10% of patients. Targeted inhibition has been successfully applied, with mixed responses observed in a substantial subset of patients mirroring the widespread spatial heterogeneity in this genomically complex disease. Central nervous system (CNS) involvement is an extremely rare, extramedullary form of multiple myeloma that can be diag- nosed in less than 1% of patients. It is considered an ultimate high-risk feature, associated with unfavorable cytogenetics, and, even with intense treatment applied, survival is short, reaching less than 12 months in most cases. Here we not only describe the first patient with an extramedullary CNS relapse responding to targeted dabrafenib and trametinib treatment, we furthermore provide evidence that a point mutation within the capicua transcriptional repressor (CIC) gene mediated the acquired resistance in this patient.
An 81-year-old patient with κ light chain multiple myeloma (MM) was referred to our center after having a seizure and increasing M-proteins. MM had been diagnosed 2 years beforeand the patient had undergone nine cycles of bortezomib- based combination therapy (VMP) resulting in an initial good disease control. Magnetic resonance imaging of the brain andadditional experimental whole body 11C-methionine positron emission tomography (PET)-computed tomography (CT) scan, which increased MM imaging sensitivity [1], demonstrated metabolic active disease supra- and infratentorial in the clivus, as well as in the right femur as the underlying cause of the clini- cal scenario (Fig. 1A). A high load of clonal plasma cells (PCs) was detected in his cerebrospinal fluid (CSF), whereas in the bone marrow (BM), only a few CD138pos MM cells could be detected. Thus, we diagnosed extramedullary central ner- vous system (CNS) relapse, and intrathecal triple therapy (methotrexate, cytarabine, and dexamethasone) along with age-adjusted systemic chemotherapy (cytarabine and thiotepa) was initiated.Given the peculiar clinical course and the poor prognosis associated with a CNS localization, with limited effective therapeutic options available, we performed a deep molecular characterization of CSF and BM tumoral plasma cells.DNA extracted from CD138pos purified cells obtained from CSF and BM paired samples was analyzed by next- generation sequencing (NGS).
We applied the M3P (v3.0) panel [2, 3], a disease-specific in-house customized, NGS- targeted deep sequencing panel for MM (Ion torrent plat- form) that includes an 88-gene selection of most commonly mutated genes such as TP53, DIS3, FAM46C, CYLD, MAF,XBP1, MYC, MAX [4, 5], actionable drug targets (i.e., NRAS, KRAS, BRAF) [6], and genes being associated with drug resistance [7, 8] (e.g., CRBN, IKZF1/3, NR3C1, PSMB5). The average sequencing depth increased 700×. The CSF cells harbored a clonal BRAFV600E mutation (allele frequency vari- ance 52%) that was absent in the BM, highlighting spatial genomic heterogeneity [9]; no other somatic point muta- tions were detected within the M3P genes and no circulat- ing PCs were identified by peripheral blood flow-cytometry analysis.The BRAFV600E mutation in exon 15 of BRAF gene is present in between 4% and 10% of patients with MM at diagnosis [4–6, 10], rising to almost 20% at relapse [7], displaying a role in the extramedullary disease and exerting a negative impact overall survival (45 vs. 105 months, p = .04) [11, 12]. Applying the M3P gene panel, we sequenced 608 MM patients at different disease stages. Concerning BRAF, we have identified 59 (9.7%) mutated patients with a total of 25 distinctive mutations. Among our patient cohort, 21 of 59 (35%) harbored the BRAFV600E mutation; within the remaining 38 patients with 24 BRAFnon-V600E muta- tions, we found 12 alterations conferring a kinase domain activa- tion, comprising also a rare K601 mutation, and 12 leading to BRAF functional impairment (supplemental online Fig. 1) [13].
BRAF exon 15 mutations confer sensitivity to target therapies such as vemurafenib, dabrafenib, and trametinib [14, 15]. Heuck et al. reported BRAF-MEK targeted therapy approach in 58 patients with MM with either BRAF, NRAS, and KRAS mutations or high-risk gene expression profiling; out of58 patients, 11 displayed an extramedullary localization. Ten patients received a combination therapy with trametinib, and two of them were additionally treated with dabrafenib or vemurafenib. Interestingly, within the patients with a measurable disease (40 patients), 16 patients achieved a reduction of at least 50% of the MM protein, and 9 achieved a complete remission evaluated by PET-CT. Raab et al. described a case of a patient with MM with a disseminated disease harbor a BRAFV600E muta- tion. The patient was successfully treated with vemurafenib upfront and with a subsequent combination therapy with borte- zomib at disease relapse owing to a clonal selection of NRAS mutants’ resistant subclones. [16, 17]. These published real-life based evidences and ongoing clinical trials (i.e., BIRMA trial) com- bining BRAF and MEK inhibitors, highlight the clinical relevance of circumventing the paradoxical RAS pathway activation upon BRAF inhibition already described in melanoma [18, 19]. Lohr et al. tested in vitro the combination of trametinib and dabrafenib in several MM cell lines harboring distinct BRAF or RAS mutations; the U266 BRAFK601N proved most sensitive and displayed a similar paradoxical feedback loop of RAS-activation [5].Recently a combination of dabrafenib and trametinib effectively eliminated in BRAFV600E mutant melanoma brain metastases, demonstrating that the drug can cross the blood brain barrier [20, 21].
Although neutropenic, because of the cytarabine-based che- motherapy, the patient developed a Gram-positive septice- mia. Taking into account the risk profile, the therapy-related infectious episode, and the sequencing results, and according to German law and ethical approval (Einzelheilversuch), the patient started a combinational targeted therapy with contin- uous BRAF-MEK inhibitor (dabrafenib 150 mg twice daily and trametinib 2 mg daily). Neurological examination revealed a significant clinical improvement on the basis of the absence of pathological signs and symptoms, which was confirmed by11C-methionine PET subtotal tumor shrinkage (Fig. 1B).Regrettably, only 3 months after the treatment initiation, 11C-methionine PET revealed local MM recurrence and dis- seminated bone while on continuous therapy (Fig. 1C). To con- firm the disease relapse, we repeated the CSF assessment, revealing, as expected, a high mononucleated tumoral plasma cells load. The patient underwent palliation with hyper- fractionated radiotherapy of the cerebrum (cumulative irradi- ation dose: 30 Gy); because of compromised performance status of the patient, no further systemic therapy could be applied, and best supportive care was adopted until patient exitus occurred at the end of October 2017.Genotyping Results and Interpretation of the Molecular Results To investigate the underlying mechanisms of resistance develop- ment upon targeted MEK-BRAF inhibitor therapy, we performed a whole exome DNA sequencing (Illumina platform) on the pretherapy sample and on CD138pos purified MM cells obtained from the CSF after confirmed disease relapse. Sequencing depth of 115× was applied.
A total number of 97 nonsilent coding variants (missense, nonsense, indels, splice) with an allele frequency higher than 5% were identified, of which 74 were shared between the timepoints. Sequencing revealed 19 additional point mutations acquired at relapse. According to published guidelines, we performed an extensive literature revision, and we systematically selected four potential clinically relevant non- synonymous point mutations (Table 1) [22]. Dispatched RND transporter family member 2 (DISP2; p.P1271L) is a key regula- tor of the hedgehog signaling pathway [23, 24] and has been associated with the development of bortezomib resistance in MM [25]. It further impacts fibroblast growth factor receptor 3 signaling to RAS pathway, thus potentially mediating the paradoxical activation of the downstream pathway in a BRAF- independent manner [26]. CREB binding protein (CREBBP; p.V429F) represents an epigenetic modulator able to control the TP53 apoptosis machinery activation [27] and the downstream regulation of the RAS-RAF pathway [28]. The pyrimidinergicreceptor P2Y4 (P2RY4, p.R314Q) is an upstream regulator of PLCβ/PI3K pathway able to cross-talk with the EGFR-RAS path- way [29], a well-known mechanism of resistance described in BRAFV600E mutated melanoma [30].
We also identify a missense mutation in capicua transcriptional repressor (CIC; p.A984P) mapped on chromosome 19 with an allelic variance of 17% (Table 1). CIC represents a transcriptional repressor gene directly involved in the downstream regulation of the RAS-RAF pathwayable to drive the development of BRAF-MEK inhibitor resis- tance [31]. Based on this correlation, we hypothesized that the acquisition of CIC mutation may mechanistically underlie the BRAF-MEK resistance in our patient.Functional and Clinical Significance of CIC in Cancer Next, we aimed to functionally validate the molecular signifi- cance of CIC alteration in mediating resistance to BRAF and MEK inhibitors. Wang et al. demonstrated in lung, colon, pancreatic [31, 32], and melanoma [31, 33] human cancer models the piv- otal role of low CIC expression in inducing resistance to vemurafenib and trametinib; however, evidences of its role of resistance induction in MM or other hematological malignanciesare lacking. Le Blanc et al. reported that CIC missense mutations result in a gene expression downregulation [34]. Interrogating the Multiple Myeloma Research Foundation CoMMpass study we found a similar correlation: the 10 patients that harbored aCIC mutation (Fig. 2A) had a significant gene downregulation compared with the unmutated ones (p = .03). Mutations in the proline-rich (Fig. 2B) region are reported to impair the protein expression [35].
Therefore, we established a CIC knockdownin vitro model, using a small interfering RNA as specific gene silencing technique. We employed the U266 MM cell line har- boring an activating BRAFK601N mutation, usually sensitive to BRAF-MEK inhibition [5], as a commercially available MM model harboring a BRAF activating mutation. Of note, upon CIC gene silencing, we observed drug resistance induction to BRAF-MEK inhibition (Fig. 3A); in detail, we cultured the silenced and not- silenced MM cells with trametinib and dabrafenib, either as sin- gle agents or in combination, and we observed resistance induc- tion to the combination of the two drugs (row factor, 91.16%; p < .0001, two-way ANOVA test). Next, we investigated whether this drug-resistance phenotype also coincided with a more inva- sive behavior. Thus, we performed a motility and migration assay. CIC knockdown in U266 BRAFK601N cells significantly enhanced MM migration in a scratch wound healing assay (Fig. 3B). These findings prompted us to investigate potential mechanism able to explain the CIC-impairment-related biologi- cal effects. In particular, CIC is the direct master regulator of sev- eral transcription factors such as ETV4 and ETV5 two oncogenes able to modulate the RAS downstream pathway [31–33]. More- over, indirectly CIC induces an invasiveness related protein namely MMP24 [32, 33]. Consequently, we confirmed by West- ern blotting an upregulation of ETV4, ETV5, and MMP24 protein expression in the CIC-knockdown U266, as mediators of drug resistance and MM invasiveness (Fig. 3C). The upregulation of these transcription factors can activate the MAPK signaling, in an independent p-MEK manner [31], providing an escape mech- anism from BRAF-MEK inhibition (Fig. 3D) [36].[12, 38–42]. Furthermore, given the scanty evidences available about the disease biology behind the EMD [43], CIC onco- suppressive functions might be also expressed as ancillary mechanism that sustain the EMD phenotype in MM [39].Prospective clinical trials including the BRAF and MEK inhibi- tion are ongoing in multiple myeloma in Europe (NCT02834364) and in the U.S. (NCT03091257) as well as in different solid cancers such as melanoma and colon cancer (NCT02974803, NCT03668431); these ongoing studies represent the ideal oppor- tunity to determine and validate the role of CIC mutations as potential disease biomarker in a large clinical and controlled- prospective setting.GAllelic frequency: percentage of reads referred to the mutated alleleAverage sequencing depth: mean number of unique reads for each single nucleotide aligned to a reference sequenceSpatial genomic heterogeneity: presence of distinctive genomic alterations in different anatomical sitesBRAF: B-Raf Proto-Oncogene, Serine/Threonine KinaseMEK: Mitogen-Activated Protein Kinase Kinase 1CIC: Capicua Transcriptional RepressorDISP2: Dispatched RND Transporter Family Member 2CREBBP: CREB Binding Protein P2RY4: Pyrimidinergic receptor P2Y4 ETV4: ETS Variant 4ETV5: ETS Variant 5 MMP24: Matrix Metallopeptidase 24CIC has recently been identified as a candidate gene related to MEK-BRAF resistance development [31, 33]. Our clinical observations, the subsequent in vitro MM model, and the public datasets interrogations support that the acquisition of CIC mutation and its subsequent downregulation confers MEK-BRAF inhibitors resistance for the first time in MM.As large BRAF-RAS treated cohorts in MM are not avail- able, we screened for published datasets to answer the ques- tion of whether mutation acquisition in CIC under BRAF-RAS targeted therapy can be observed in a significant number of patients being resistant to BRAF inhibitors. Remarkably, Van Allen et al. recently published a comprehensive genomic char- acterization of 45 patients with metastatic melanoma resis- tant to BRAF inhibitors; 5 of them (11%) harbored a somatic CIC mutation (4 missense and 1 frame shift). Intriguingly, two of the single nucleotide variations out of these five were acquired at time of relapse (Fig. 2B). One out of these five patients harboring a pretherapy CIC mutation experienced a very early disease relapse under dabrafenib therapy [37].Given that almost 30% of patients with MM harbor mutations affecting the BRAF-RAS pathway, this may represent a poten- tial biomarker to predict therapy response. Based on prior published findings [31, 33, 34], public available datasets [37] and previous [31–34] and original in vitro validations BGB-3245 pinpoint CIC mutation as one of the mechanisms of drug resistance of BRAF-MEK inhibition therapy. Extramedullary (EMD) dissemi- nation in MM typically correlates with very poor prognosis, especially when the clinical onset manifests at disease relapse1417 of 3.03.2016 to A.G.S.; the Apulian Regional project: medicina di precisione to A.G.S.; ZKF Würzburg (Z-3R/2) to L.R. and M.D.V.; and a BTHA grant to K.M.K. The German SKELMET/ μBone consortium supported by the German Research Council (DFG FOR 1586, SPP 2084) through an Investigator grant toA.B. Bayerische Forschungsstiftung consortium FortiTher (WP2TP3), the Deutsche Forschungsgemeinschaft mBone con- sortium (2084/1, 401253051 to A.B.).