Phase 1 study of selinexor plus carfilzomib and dexamethasone for the treatment of relapsed/refractory multiple myeloma

Summary Selinexor, an oral Selective Inhibitor of Nuclear Export, targets Exportin 1 (XPO1, also termed CRM1). Non‐clinical studies support combining selinexor with proteasome inhibitors (PIs) and corticosteroids to overcome resistance in relapsed/refractory multiple myeloma (RRMM). We conducted a phase I dose‐escalation trial of twice‐weekly selinexor in combination with carfilzomib and dexamethasone (SKd) to determine maximum tolerated dose in patients with RRMM (N = 21), with an expansion cohort to assess activity in carfilzomib‐refractory disease and identify a recommended phase II dose (RP2D). During dose escalation, there was one dose‐limiting toxicity (cardiac failure). The RP2D of twice‐weekly SKd was selinexor 60 mg, carfilzomib 20/27 mg/m2 and dexamethasone 20 mg. The most common grade 3/4 treatment‐emergent adverse events included thrombocytopenia (71%), anaemia (33%), lymphopenia (33%), neutropenia (33%) and infections (24%). Rates of ≥minimal response, ≥partial response and very good partial response were 71%, 48% and 14%, respectively; similar response outcomes were observed for dual‐class refractory (PI and immunomodulatory drug)/quad‐exposed (carfilzomib, bortezomib, lenalidomide and pomalidomide) patients (n = 17), and patients refractory to carfilzomib in last line of therapy (n = 13). Median progression‐free survival was 3·7 months, and overall survival was 22·4 months in the overall population. SKd was tolerable and re‐established disease control in RRMM patients, including carfilzomib‐refractory patients. Registered at ClinicalTrials.gov (NCT02199665)

The development of immunomodulatory drugs (IMiDs; e.g. lenalidomide, pomalidomide) and proteasome inhibitors (PIs; e.g. carfilzomib, bortezomib) as standards of care for patients with multiple myeloma (MM) has resulted in significant improvements in survival (Brenner et al, 2009;Kumar et al, 2008;Thumallapally et al, 2016). However, nearly all patients require multiple lines of therapy as they relapse or develop disease refractory to treatment. First-line and subsequent therapies usually involve IMiDs and PIs in doublet or triplet combinations with corticosteroids and other systemic therapies . As use of these combinations have become new standards of care, the treatment challenges in the relapsed/refractory (RR) setting have evolved, with increasing numbers of patients being quad-refractory to bortezomib, lenalidomide, pomalidomide and carfilzomib or penta-refractory to these four drugs and the anti-CD38 monoclonal antibody daratumumab. There is a need to develop agents with novel mechanisms of action to overcome treatment resistance (Chim et al, 2018;Sonneveld & Broijl, 2016).
Selinexor is an oral Selective Inhibitor of Nuclear Export that targets Exportin 1 (XPO1, also termed CRM1), the only known nuclear export protein for tumour suppressor proteins (TSPs) and cell-cycle regulators (e.g. p53, FOXO, IjB, p21, p27), as well as eukaryotic translation initiation factor 4E (eIF4E)-bound oncoprotein mRNAs (Conforti et al, 2015;Culjkovic-Kraljacic et al, 2012;Das et al, 2015;Gravina et al, 2014). Overexpression of XPO1 is essential for MM cell survival (Schmidt et al, 2013;Tiedemann et al, 2012). XPO1 mediates the functional inactivation of cell-cycle regulators and TSPs and promotes the export and translation of mRNA for key oncoproteins, including c-MYC, BCL-2 and Cyclin D (Culjkovic-Kraljacic et al, 2012;Gandhi et al, 2018;Nguyen et al, 2012). Inhibition of XPO1 with selinexor restores nuclear localization of TSPs and cell-cycle regulators (Nair et al, 2017;Tai et al, 2014). Selinexor elevates levels of the inhibitor of kappa B (IjBa), which forms complexes with and inhibits transcription factor nuclear factor (NF)jB, disrupting a range of signalling pathways, including inflammation, oncogenesis and cell survival. In myeloma cells, selinexor treatment has been shown to induce apoptosis, reduce levels of proto-oncoproteins and impair osteoclastogenesis (Das et al, 2015;Schmidt et al, 2013;Tai et al, 2014).
Preclinical studies have provided a rationale for combining selinexor with PIs (Kashyap et al, 2016;Nair et al, 2017;Rosebeck et al, 2016;Turner et al, 2016). The addition of selinexor to a PI has a synergistic effect on cell death of myeloma cell lines and primary plasma cells derived from patients with RRMM, and the combination demonstrated greater antimyeloma activity in a murine xenograft model than either agent alone Rosebeck et al, 2016;Turner et al, 2016). Selinexor-PI combinations were associated with inhibition of BCL2 expression, increased cleavage and inactivation of AKT, activation of caspase-10 and other caspases, and increased levels of IjBa and IjBa-NFjB complexes, leading to neutralization of NF-jB Rosebeck et al, 2016;Turner et al, 2016). NF-jB activation has been shown to be a mechanism of PI resistance (L€ u & Wang, 2013;Markovina et al, 2008).
The clinical activity of selinexor as a single agent and as part of combination regimens has been demonstrated in heavily pre-treated patients with RRMM Vogl et al, 2018). Single-agent selinexor was associated with modest activity in a phase I study with an objective response rate (ORR) of 4%, which improved to 50% when selinexor was combined with dexamethasone (Sd) at the twice-weekly recommended phase 2 dose (RP2D) . In the subsequent phase II STORM study, the Sd combination generated ORRs of 21% for patients with quad-refractory MM and 20% in penta-refractory patients .
The addition of selinexor to bortezomib and dexamethasone (SVd) in the phase I/II STOMP study generated ORRs of 43% in a cohort with PI-refractory RRMM (Bahlis et al, 2018).
Carfilzomib is approved for use in combination with dexamethasone for patients with RRMM (Berenson et al, 2014;Dimopoulos et al, 2016a;Siegel et al, 2012). Preclinical studies have demonstrated synergistic activity between selinexor and carfilzomib Rosebeck et al, 2016;Turner et al, 2016), and clinical studies further support carfilzomib as a potential therapeutic partner in RRMM. Carfilzomib was active in patients with MM previously treated with or refractory to bortezomib (Berenson et al, 2016(Berenson et al, , 2014, and the combination of carfilzomib and dexamethasone (Kd) demonstrated improved progression-free survival (PFS) and overall survival (OS) compared with the combination of bortezomib and dexamethasone (Vd) in patients with RRMM (Dimopoulos et al, 2017(Dimopoulos et al, , 2016a. Here we describe a phase 1 multicentre, open-label, investigator-initiated study to determine the maximum tolerated dose (MTD) and the RP2D of twice-weekly selinexor in combination with Kd in patients with RRMM, as well as safety, tolerability and preliminary efficacy.

Study design
This is a multicentre, open-label, phase I study (ClinicalTrials.gov, NCT02199665). The primary objectives were to determine the MTD for the combination of twice-weekly selinexor with Kd (SKd) in patients with RRMM, employing a 3 + 3 dose escalation design, followed by an expansion cohort to support a RP2D. Secondary objectives were to determine safety, tolerability and efficacy.
Patients aged ≥18 years with progressive RRMM were enrolled at five Multiple Myeloma Research Consortium sites in North America. Patients were eligible provided they had been previously treated with at least two anti-myeloma therapies, including a PI and an IMiD, had an absolute neutrophil count ≥1Á0 9 10 9 /l, a haemoglobin concentration ≥80 g/l, a platelet count ≥50 9 10 9 /l and adequate hepatic (total bilirubin ≤2 times the upper limit of normal and alanine aminotransferase ≤2Á5 times the upper limit of normal) and renal function (creatinine clearance ≥30 ml/min) within 14 days of treatment initiation. Patients were excluded if they had received prior selinexor or any other anticancer therapy within 2 weeks of treatment initiation. Concurrent anticancer therapy other than steroids was not allowed. Other exclusion criteria included unstable angina or myocardial infarction within 4 months of treatment initiation, New York Heart Association Class III/IV congestive heart failure, left ventricular ejection fraction <40%, history of severe coronary artery disease, severe uncontrolled ventricular arrhythmias or uncontrolled hypertension or uncontrolled diabetes within 14 days of treatment initiation. Patients with plasma cell leukaemia, Waldenstr€ om macroglobulinaemia, POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal protein, skin changes) syndrome or amyloidosis were also excluded.
The study was conducted in accordance with US Food and Drug Administration and International Conference on Harmonisation Guidelines for Good Clinical Practice, the Declaration of Helsinki, Health Canada, and any applicable local health authority, institutional review board or ethics committee requirements. All patients provided written informed consent.

Schedule and dosing
For the dose-escalation phase, three patients were assigned to each cohort, beginning at Dose Level 1 (Fig 1). Selinexor, carfilzomib and dexamethasone were all administered twice weekly, and both selinexor and carfilzomib doses were escalated. Beginning at Dose Level 1, if none of the first three patients enrolled into the cohort experienced a dose-limiting toxicity (DLT), then dose escalation proceeded to the next cohort. If any one of the three patients experienced a DLT, three more patients were added to the cohort at the same dose. If there were no additional DLTs, dose escalation proceeded to the next cohort. If two or more DLTs were observed among the initial three or expanded six patients, the dose level was considered to exceed the MTD. Because there were delays in patient accrual, the study protocol was amended to ensure that eligible patients could enrol at the time of their availability-expansion to six patients per cohort was allowed if, at a given dose level, three patients were enrolled and no DLTs were observed but all three had not completed their first cycle of treatment. Patients who did not receive all scheduled doses (unrelated to drug toxicity) during Cycle 1 were replaced for DLT evaluation per the study protocol. Given that our patients were heavily pre-treated with advanced disease, we anticipated that disease-related sequalae might prevent patients from completing the doseescalation phase without dose interruption.
Once the MTD or maximum recommend dose was determined, that dose cohort was to be expanded to a total of 12 carfilzomib-refractory patients as defined by the International Myeloma Working Group (IMWG) criteria (Rajkumar et al, 2011). If the overall DLT rate for this cohort was <30%, this dose would be declared the RP2D.

Assessments
Safety and tolerability were evaluated by means of drug-related DLTs in the dose-escalation cohorts, treatment-emergent adverse events (AEs), physical examinations and laboratory tests. DLTs were prespecified haematological and non-haematological toxicities that were considered treatment-related and occurred during Cycle 1. Haematological DLTs were: febrile neutropenia; grade 4 neutropenia lasting >7 days; grade 4 thrombocytopenia lasting >7 days despite dose delay; and grade 3/4 thrombocytopenia associated with bleeding. Non-haematological DLTs were: grade ≥2 neuropathy and any grade ≥3 toxicity (lasting for ≥3 days for gastrointestinal events and >7 days for fatigue or anorexia) despite maximal supportive care except for electrolyte abnormalities, hair loss and elevation of alanine aminotransferase or aspartate aminotransferase. Dose modifications were allowed, but any treatment toxicities that required a dose reduction during Cycle 1, or any toxicities that delayed initiation of Cycle 2 by >7 days were also considered DLTs.
Treatment-emergent AEs were graded according to the National Cancer Institute-Common Terminology Criteria for Adverse Events v4.0 (National Cancer Institute 2010). Patients that discontinued treatment underwent a final assessment at 28 days after the last dose of a study drug. Patients were followed for survival for up to 2 years after the end of treatment.
Efficacy measures included response according to the IMWG criteria (Durie et al, 2006;Rajkumar et al, 2011)progressive disease (PD), stable disease (SD), minimal response (MR), partial response (PR), very good partial response (VGPR), complete response (CR) and stringent complete response (sCR). Response assessments were completed at Cycle 2 Day 1 and Day 1 of subsequent cycles. All response criteria required confirmation with two consecutive assessments.

Statistical analysis
Continuous and categorical variables were summarized with descriptive statistics. ORR (≥PR) and a clinical benefit rate (CBR; ≥MR) were estimated, and 90% confidence intervals (CIs) were generated for the RP2D cohort. If the data deviated strongly from normality as judged by boxplots and normal probability plots, non-parametric, Wilcoxon signed-rank tests were performed in place of t-tests. Given the small sample size, multiplicity adjustments were not made to the alpha levels; these analyses were considered exploratory and hypothesis-generating only. Time-to-event endpoints were assessed by the Kaplan-Meier method using GraphPad Prism 7.03 software (La Jolla, CA, USA).

Results
Twenty-one patients with RRMM were enrolled between July 2014 and September 2016. The data cut-off for this analysis was 1 September 2017. Patients were aged between 55 and 74 years, 43% were ≥65 years of age (Table I). The median time since the initial MM diagnosis was 4Á5 years. Twelve (57%) patients were identified as high risk per IMWG criteria, including 5 (24%) with del (17p). Patients received a median of 4 (range, 2-10) prior lines of therapy. Previous treatments included carfilzomib (95%) and pomalidomide (81%) ( Table II). All patients (100%) were refractory to last line of therapy, including 13 (62%) who were refractory to carfilzomib (4 to carfilzomib/dexamethasone, 9 to carfilzomib/pomalidomide/dexamethasone). Seventeen (81%) patients were dual-class refractory (refractory to a PI and an IMiD) and quad-exposed (bortezomib, carfilzomib, lenalidomide and pomalidomide). Baseline characteristics and prior therapies by dose level are presented in Tables SI and SII. Of the 21 patients enrolled, 18 were evaluable for DLTs (received one full course of treatment or stopped treatment due to a DLT). All 21 patients were included in overall toxicity, survival and response assessments. One patient with a history of congestive heart failure (CHF) was retrospectively determined to be ineligible due to pre-existing amyloidosis which was unknown at study entry. This patient enrolled in the dose-escalation phase of the study and was included in safety and efficacy assessments.

Determination of the MTD
During the course of Cycle 1, there were no DLTs at Dose Level 1 (30 mg/m 2 selinexor; 20/27 mg/m 2 carfilzomib; 20 mg dexamethasone). Two patients in the Dose Level 1 cohort did not receive all scheduled treatment doses (i.e. dose modifications unrelated to toxicity) and were replaced per the study protocol, resulting in a total of five patients in this cohort. There were no DLTs in three patients enrolled at Dose Level 2a (40 mg selinexor; 20/36 mg/m 2 carfilzomib;  20 mg dexamethasone). A DLT was experienced by one of the first three patients enrolled at Dose Level 2b (60 mg selinexor; 20/27 mg/m 2 carfilzomib; 20 mg dexamethasone); the patient with a history of CHF who retrospectively was found to have cardiac amyloidosis experienced cardiac failure during Cycle 1. Therefore, three additional patients were assigned to the cohort. One patient did not receive all scheduled doses (unrelated to toxicity) and was replaced, resulting in a total of seven patients for Dose Level 2b, with no additional DLTs during the dose escalation stage. Further dose escalation was not pursued based on AE rates, tolerability and anti-myeloma activity. Assessment by dose level showed rapid disease control at all dose levels but no notable trend in the rate or depth of response to support further escalation, while rates of some AEs increased after Cycle 1, as did dose reduction. In the safety population, the rate of grade 3/4 anaemia was 5% during Cycle 1 but increased to 33% after Cycle 1. In the dose escalation cohorts, dose reductions were required by 80% (4/5) of patients enrolled at Dose Level 1, 100% (3/3) at Dose Level 2A, and 29% (2/7) at Dose Level 2b (Table SIII). Based on these findings and previous experience with selinexor (Bahlis et al, 2018;Chen et al, 2018;Vogl et al, 2018), Dose Level 2b was selected as the maximum recommended dose and selected for expansion. Six additional patients were enrolled into the Dose Level 2b cohort for a total of 13 patients, of whom 12 were carfilzomib-refractory. Among these 6 patients, there were 2 DLTs during Cycle 1 (grade 3 diarrhoea and grade 3 decrease in platelet count), yielding a DLT rate of 25% for the 12 patients who completed 1 cycle of Dose Level 2b (below the predefined limit of 30%). Dose Level 2b was selected as the RP2D.

Response and treatment outcomes
Most patients achieved disease control after 1 cycle (CBR of 67% and ORR of 38%). For best response during the course of treatment, the CBR was 71%, ORR was 48% and the VGPR rate was 14% (Table IV). There were no CRs. The patient with pre-existing amyloidosis who experienced a DLT discontinued SKd prior to response evaluation; this was considered a non-response. For patients receiving the RP2D (n = 13), CBR was 62% (90% CI: 0Á36-0Á83), ORR was 38% (90% CI: 0Á17-0Á65) and the rate of VGPR was 15% (90% CI: 0Á03-0Á41); two additional patients achieved SD. For dual-refractory/quad-exposed patients (n = 17), the CBR was 76%, ORR was 53% and the VGPR rate was 18%, and for patients who were refractory to carfilzomib in last line of therapy (n = 13), the corresponding values were 77%, 62% and 15%. The one patient who was tri-refractory/penta-exposed achieved VGPR.
Durability and depth of responses are presented in Fig. 2. The median duration of response for patients who achieved ≥MR and ≥PR were 2Á9 and 3Á4 months, respectively, for all response-evaluable patients, 3Á1 and 3Á0 months for the RP2D cohort, 2Á8 and 3Á3 months for the carfilzomib-refractory cohort and 3Á1 and 3Á0 months for the high-risk cohort.
Median PFS and OS were 3Á7 and 22Á4 months, respectively, for all enrolled patients (Fig. 3), 3Á7 and 22Á4 months
-Refractory to PI and IMiD/exposed to bortezomib, lenalidomide, carfilzomib and pomalidomide. In the phase I/II STOMP study (N = 42 RRMM), selinexor was administered twice-weekly at 60 or 80 mg or once-weekly at 80 or 100 mg in combination with Vd during the dose-escalation stage; this was followed by a dose expansion with the RP2D (Bahlis et al, 2018). The investigators selected 100 mg once-weekly dose of selinexor as the RP2D, and nearly all patients (39/42) received bortezomib 1Á3 mg/ m 2 weekly at treatment initiation rather than at the standard twice-weekly dose schedule. The results suggested that the rates for some treatment-related haematological and GI AEs improved with once-weekly selinexor. The rate of grade 3/4 thrombocytopenia decreased from 69% during the dose-escalation stage (60 or 80 mg selinexor twice-weekly/onceweekly) to 31% with the RP2D, anaemia decreased from 25% to 4%, and grade 3 diarrhoea decreased from 13% to 4% (Bahlis et al, 2018).
The clinical activity of SKd is promising and compares favourably with activity in studies of selinexor alone and selinexor with dexamethasone (Sd) Vogl et al, 2018). However, we also recognize the limitations of crossstudy comparisons and differences between study populations. The current study enrolled heavily pre-treated patients, One patient with dose-limiting toxicity that resulted in treatment discontinuation prior to response evaluation was not included. (B) Duration of response. Three patients had PD at first response assessment and were not included; two patients remained in SD and were not included; one patient with dose-limiting toxicity that resulted in treatment discontinuation prior to response evaluation was not included. DL, dose level; MR, minimal response; PD, progressive disease; PR, partial response; SD, stable disease; VGPR, very good partial response.

Change in M-protein relative to baseline
all of which were refractory to their last line of therapy: 95% were refractory to carfilzomib, 62% were refractory to carfilzomib in the last line of therapy and 57% had high-risk cytogenetics. In a phase I study of patients with RRMM (n = 81) or Waldenstr€ om macroglobulinaemia (n = 3), selinexor (3-60 mg/m 2 or fixed dose of 40 or 60 mg) showed modest activity as a single agent with an ORR of 4% (n = 57) and a CBR of 21%, which improved to 22% and 33%, respectively, when twice-weekly selinexor (45 or 65 mg/m 2 ) was combined with 20 mg of dexamethasone (n = 27). At the 45 mg/m 2 selinexor dose, the ORR for the combination was 50% . In the STORM study, twice-weekly Sd (80/ 20 mg) resulted in an ORR of 21% in patients with quad-or penta-refractory RRMM . Response rates with SKd also appear consistent with those of SVd from the STOMP study. ORR with SVd was 63% overall and 43% for PI-refractory patients (n = 21) compared with 84% for PInonrefractory patients (n = 19) (Bahlis et al, 2018).
The clinical activity of SKd also can be related to the results from historical studies in RRMM (Berdeja et al, 2015;Berenson et al, 2016Berenson et al, , 2014Dimopoulos et al, 2016b;Papadopoulos et al, 2015;Richardson et al, 2013Richardson et al, , 2014San Miguel et al, 2013;Shah et al, 2015;Siegel et al, 2012). Many of these studies enrolled patients who had been previously treated with bortezomib or lenalidomide with dexamethasone to assess the efficacy and safety of carfilzomib and pomalidomide regimens. Response rates have ranged from 32% to 77% for pomalidomide or carfilzomib in combination with dexamethasone (Berenson et al, 2016;Dimopoulos et al, 2016b;Papadopoulos et al, 2015;San Miguel et al, 2013) and 50% for the triplet of carfilzomib, pomalidomide and dexamethasone (Shah et al, 2015).
While response rates and time to response in our study are quite promising given the refractory status of the patient population, duration of response and PFS were shorter than anticipated in view of other selinexor studies. However, response rates, duration of response and PFS in these other selinexor studies were generally less robust in patients who were refractory to one of the drugs used in combination (Bahlis et al, 2018;Chen et al, 2017a;Gasparetto et al, 2017;White et al, 2017). Median PFS was 9Á0 months for all evaluable patients in the STOMP study but 6Á1 months for PI-refractory patients, compared with 17Á8 months for PI-nonrefractory patients (Bahlis et al, 2018). Further, OS in our study appeared prolonged at a median of 22Á4 months when considering PFS results. It is possible that SKd selected for less aggressive MM clones at the time of progression, allowing for a durable response with subsequent therapy (e.g., daratumumab), or clones that were sensitive to subsequent treatment.
Observations from this study, particularly consistent and rapid disease control (67% CBR in the first cycle), indicate that SKd could be a 'bridge' to subsequent therapy, allowing patients to at least transiently overcome resistance, restore disease control and prepare for subsequent therapy. Further    studies are needed to better understand these observations and to determine the mechanism for loss of disease control and to evaluate whether a different dosing or schedule with SKd can improve durability of response. Based on observations from the STOMP trial (Bahlis et al, 2018), we are currently evaluating once-weekly SKd in carfilzomib-refractory, carfilzomib/PI-na€ ıve and non-refractory patient populations. Other studies in RRMM are assessing selinexor in combination with liposomal doxorubicin (NCT02186834) and ixazomib and dexamethasone (NCT02831686), and additional cohorts from the STOMP trial have demonstrated the tolerability and activity of twice-or once-weekly selinexor with dexamethasone in combination with lenalidomide (SRd), pomalidomide (SPd) or daratumumab (SDd) (Chen et al, 2017a;Gasparetto et al, 2017;White et al, 2017). The ongoing phase III BOSTON study (NCT02831686) is evaluating SVd in patients with RRMM who are PI-relapsed/-na€ ıve.
With more patients receiving first-and second-line combinations of PIs, IMiDs and monoclonal antibodies, it is becoming more challenging to effectively treat patients in the RRMM setting (Nooka et al, 2015). The addition of selinexor to Kd demonstrated manageable safety and tolerability and promising efficacy in a heavily pre-treated population of patients with RRMM. Significant clinical activity was observed in dual-refractory/quad-exposed patients, and in patients who were carfilzomib-resistant in their last line of therapy. Further studies are needed to understand the high rate of response but relatively short response duration, and to identify patients for whom SKd might serve as an effective 'bridge' regimen to subsequent therapies or as a line of therapy that provides durable disease control.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.