Preclinical Assessment with Clinical Validation of Selinexor with Gemcitabine and Nab-Paclitaxel for the Treatment of Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) has an abysmally low survival rate (1). This necessitates the development of new treatment approaches. PDAC tumors carry aberrations in many different signaling mechanisms rendering therapy against a single pathway impractical (2). To date there is no regimen that can simultaneously activate multiple tumor suppressors (TSP) that are wild type and functional in PDAC. Therapies such as gemcitabine (GEM) plus nanoparticle albumin-bound (nab) paclitaxel partially activate these TSPs. However, the overexpression of nuclear export protein exportin 1 (XPO1) leads to nuclear expulsion and functional inactivation of these TSPs (3, 4). Coupled with low drug penetrance due to desmoplasia and immunosuppressive microenvironment, the inactivation of critical TSPs has been attributed for the lack of therapeutic response of different regimens in patients with PDAC.

Nuclear transport is mediated by a set of proteins belonging to karyopherin family (5, 6). XPO1 is member of the karyopherin family that carries out nuclear export of cargo proteins (7). XPO1 is the main exporter of a majority of TSPs and growth regulatory proteins that carry conserved leucine rich nuclear export signal (NES). The transport to the cytosol occurs through an energy-consuming process that involves Ran protein recycling (GTP-GDP; ref. 8). TSPs in particular require nuclear retention and DNA interaction to exert their tumor-suppressor regulatory function (9). Activation of TSPs (e.g., FOXO3a, p53, p27, Rb, Ikβ, and Par-4) induced by chemotherapies (e.g., platinum based, GEM, or GEM–nab-paclitaxel) is not sufficient for their tumor-suppressive function due in part to overexpression of XPO1 and mislocalization of these TSPs (10). Supporting this, we and others have shown that overexpression of XPO1 has been associated with therapy resistance and poor survival in solid tumors (4). Therefore, XPO1 inhibition and subsequent realignment of TSPs to the nucleus becomes an attractive anticancer strategy (11).

XPO1 expression has been shown to be increased in PDAC tissue and has been associated with tumor size, lymph node, and liver metastases and suggested to be independent prognostic parameter for shorter progression-free and overall survival (OS; ref. 12). Therefore, targeting XPO1 using specific small-molecule inhibitors could be a viable therapeutic strategy for PDAC (13). Using consensus induced fit docking Selective Inhibitor of Nuclear Export (SINE) compounds have been developed for oncologic applications (14–16). SINE compounds target the amino acid cysteine 528 in the XPO1 NES–binding domain, thereby blocking nuclear export function. This results in blocking TSPs in the nucleus, leading to selective apoptosis of cancer cells with damaged genomes while normal cells undergo reversible cell-cycle arrest (17). The clinical compound selinexor (KPT-330) and related analogs have broad activity against different tumor types (nanomolar IC_50s; refs. 18, 19). In earlier studies, we found that SINE can induce nuclear retention of several TSPs resulting in the suppression of PDAC growth and tumor growth retardation in mice (20). Selinexor has been shown to be well tolerated in patients with advanced malignancies (21, 22). Based on our result and numerous studies from other groups, selinexor was introduced into multiple phase I/II/III trials in different malignancies (23) and received FDA approval for penta-refractory multiple myeloma.

A large proportion of patients with PDAC receive GEM–nab-paclitaxel-based therapies (24). We have recently demonstrated that SINE compounds synergize with GEM (25). In this study, we demonstrate that selinexor synergizes with GEM–nab-paclitaxel in PDAC cellular models, stem cells, and patient-derived xenografts (PDX). We also performed a phase I study evaluating the safety and efficacy of selinexor in combination with GEM and nanoencapsulated nab-paclitaxel in patients with PDAC.

MiaPaCa-2, Panc-1, and AsPC-1 were purchased from the ATCC. L3.6pl was a generous gift from Dr. Paul J. Chiao (MD Anderson Texas Cancer Center). CRISPR/Cas9 cell pair of HEK293 wild-type and HEK XPO1^C528S mutant was developed as previously described (26). Cancer stem cells (CSC) were developed according to previous publications (27). Cell line culture conditions and authentication are detailed in Supplementary Methods.

PDAC and HEK cells were treated with selinexor, GEM–nab-paclitaxel, or combination for 72 hours. Conditions and procedure for MTT assay have been previously described (28). Procedure for colony formation is detailed in Supplementary Methods.

PDAC, CSCs, or HEK cell pair (50,000 cells/well) growing in 6-well plates was exposed to drugs (72 hours). Cells were pelleted and processed for Annexin V FITC using standard protocol (Biovision) and analyzed using a BD SLR II flow cytometer at Karmanos Cancer Institute flow cytometry core.

Spheroids growing in sphere formation culture media were seeded in ultra-low attachment plates (Costar^R) and exposed to indicated doses of either selinexor alone, GEM–nab-paclitaxel, or selinexor–GEM–nab-paclitaxel twice a week for 2 weeks. At the end of the treatment, spheroids were counted under an inverted microscope and photographed.

Note that 3,000 cells/well were grown in chambered slides and exposed to selinexor alone (1 μmol/L) for 24 hours followed by immunofluorescence assay as described previously (28). In combination studies, cells were exposed to selinexor–GEM–nab-paclitaxel for 24 hours at indicated concentrations followed by standard immunofluorescence analysis.

Details of CRISPR/Cas9 protocol are available in our previous publication (26). For siRNA studies, cells were transfected with csiRNA, XPO1 siRNA, using standard protocols (Santa Cruz Biotechnology), and efficiency was verified by RT-PCR.

Details of RNAi, RNA isolation procedure, RT-PCR, and Western blotting are provided in Supplementary Methods section. Primer sequences used in this study are provided in Supplementary Table S1.

MiaPaCa-2 cells were treated with 1 μmol/L selinexor or 1 μmol/L selinexor plus 300 nmol/L GEM and 3 nmol/L nab-paclitaxel for 24 hours. The cell pellets were collected and processed using Label-Free Quantitative (LFQ) Proteome Profiling with nuclear and cytoplasmic enrichment using Thermo Q-Exactive HF-X Orbitrap Mass Spectrometer at Bioproximity (Manassas, VA, USA). The intensity value of total 27,528 proteins within 3,617 protein sets in each sample was read.

In vivo studies were conducted under Wayne State University's Institutional Animal Care and Use Committee (IACUC)–approved protocol in accordance with the approved guidelines. Experiments were approved by the institute's IACUC (Protocol IACUC-18-12-0887 and 16-03-059). Tumors were developed in female ICR-SCID mice (Taconic Biosciences, Rensselaer, NY, USA). Briefly, CSCs growing as spheroids were injected subcutaneously. Note that 1 × 10⁶ CD44⁺/CD133⁺/EpCAM⁺ MiaPaCa-2 cells were implanted subcutaneously (unilaterally). After 2 weeks, mice were assigned into two groups (n = 6): (i) vehicle control; (ii) selinexor at 15 mg/kg p.o. every other day ×3 weeks. On termination, the tumor tissue from 4 mice each from control or selinexor-treated group was used for protein extraction, histopathologic examination, and IHC analysis. In another set, mice bearing CSC tumors were divided into two groups (n = 5). Group 1 received vehicle and group 2 received selinexor–GEM–nab-paclitaxel as indicated.

Fresh PDAC tissue was obtained under an approved Institutional Review Board (IRB) protocol. All samples were obtained with patient consent and approval from our Institutional Ethics Committee (protocol # 1603014771). The PDAC origin was confirmed by pathologist on site and put on ice. Immediately (<20 minutes), the surgical material was subcutaneously transplanted into the flanks of 5 mice (NSG, JAX Laboratory, Sacramento, CA USA) and allowed to grow. Once tumor burden reached about 1,500 mg, mice were euthanized, and tumors were collected, cut into 50 mg pieces, and then serially propagated in additional mice. A portion of the tissue from first passage was snap frozen, and the other portion was further characterized using IHC for PDAC marker cytokeratin-18 to rule out involvement of mice tissue, and PCR for mouse versus human CHEK expression was performed. Details of IHC and PCR are provided in Supplementary Material.

Note that 1 × 10⁶ MiaPaCa-2 viable cells were surgically injected in head of pancreas of 25 6- to 8-week-old female ICR-SCID mice. On day 32 after implantation, 21 palpated mice were randomly assigned to different treatment cohorts as follows: untreated (n = 5), single-agent KPT-330–treated group (n = 5 orally 10 mg/kg twice a week for 3 weeks), double combination (n = 5 GEM + PAC 30 mg/kg intravenously twice a week for 2 weeks), and triple combination group (n = 6 dosing as above).

The detailed phase Ib/II protocol is available as Supplementary Material. We obtained written informed consent from the patients, and the studies were conducted in accordance with recognized ethical guidelines and were approved by an IRB.

This phase Ib/II trial was initiated to identify the safety profile and efficiency of the regimen using GEM, nab-paclitaxel, and selinexor compared with historical controls in patients with metastatic pancreatic cancer.

The Student t test was used to compare statistically significant differences. Wherever suitable, the experiments were performed 3 times. The data were also subjected to unpaired two-tailed Student t test wherever appropriate, and P < 0.05 was considered statistically significant.

XPO1 is consistently activated in PDAC compared with normal tissue (Supplementary Fig. S1A). Earlier, we demonstrated the single-agent activity of SINE compounds (ref. 20; structures given in Supplementary Fig. S1B) as well as synergy with GEM (25). Building on these studies, we evaluated the impact of SINE compounds in combination with standard-of-care therapeutics in different cellular models of PDAC. In line with previous results, selinexor demonstrated synergy with GEM–nab-paclitaxel in various PDAC cell lines (Fig. 1A and B; Supplementary Fig. S2A and S2B for isobologram analysis for MiaPaCa-2 and L3.6pl), as well as Panc-1 and AsPc-1 (see inset table of Supplementary Fig. S2). We further evaluated synergy using a colony formation assay. The combination of selinexor–GEM–nab-paclitaxel demonstrated superior inhibition of colonies compared with GEM–nab-paclitaxel, selinexor–GEM, selinexor–nab-paclitaxel, or single-agent selinexor in two PDAC cell lines (Fig. 1C and D; Supplementary Fig. S2C and S2D). Annexin V FITC apoptosis analysis mirrored MTT and colony formation results. Compared with single-agent or dual combination treatments, selinexor–GEM–nab-paclitaxel showed enhancement in apoptosis in L3.6pl and MiaPaCa-2 cells (Fig. 1E and F; Supplementary Fig. S2E). In order to assess whether the combination shows molecular synergy, a limited set of targets were probed. The combined treatment led to superior activation of the TSP and Bim (Supplementary Fig. S2F). These findings demonstrate the improvement of selinexor–GEM–nab-paclitaxel over standard-of-care GEM–nab-paclitaxel and support the use of this combination in patients with PDAC.

We next sought to evaluate the impact of selinexor–GEM–nab-paclitaxel on GEM–nab-paclitaxel-resistant PDAC-derived CSCs. PDAC CSCs (derived from Panc-1 or MiaPaCa-2) were exposed to either selinexor or combinations of GEM–nab-paclitaxel or selinexor–GEM–nab-paclitaxel. Compared with single or dual combination, selinexor–GEM–nab-paclitaxel resulted in almost complete loss of the PDAC spheroids (Fig. 2A and B). In order to confirm these studies in relevant stem cell models, we probed for a reduction of stemness markers using RT-PCR. As anticipated, SINE compounds selinexor, KPT-8602/eltanexor, or KPT-185, but not the inactive analog KPT-301 (a Trans isomer that does not bind to cys528 of XPO1), showed marked downregulation in CSC promoters CD24, vimentin, and CD44 (Fig. 2C). Next, we examined the ability of this potent combination to induce apoptosis in the CSCs. In agreement with the spheroid data, selinexor–GEM–nab-paclitaxel treatment resulted in superior apoptosis compared with selinexor alone or GEM–nab-paclitaxel combination (Supplementary Fig. S3A). Selinexor treatment enhanced E-cadherin while simultaneously suppressed cancer stemness markers snail and vimentin (Supplementary Fig. S3B). We further investigated the impact of selinexor–GEM–nab-paclitaxel on epithelial–mesenchymal transition and apoptosis markers. Our results show that exposure of cells to either selinexor–GEM–nab-paclitaxel or eltanexor–GEM–nab-paclitaxel led to statistically significant reduction in CD44, CD133, EpCAM, and Bcl-2 with simultaneous enhancement in FOXO3a and proapoptotic Bax (Supplementary Fig. S3C). Taken together, these results demonstrate that selinexor can synergize with GEM–nab-paclitaxel and reverse stemness in therapy-resistant stem cell–like models of PDAC.

The primary mechanism of SINE compounds is nuclear retention of XPO1 cargo proteins. Therefore, we analyzed the impact of the combination treatment on nuclear retention of several TSPs. In line with the RNAi findings, selinexor–GEM–nab-paclitaxel treatment resulted in nuclear retention of several TSPs such as FOXO3a, HSP90, and PAR-4 in PDAC cell lines (Fig. 3). These results clearly demonstrate that selinexor–GEM–nab-paclitaxel combination retains the ability to localize TSPs to the nucleus.

To systematically evaluate the proteins retained in the nucleus, MiaPaCa-2 cells were exposed to 1 μmol/L selinexor alone or in combination with 300 nmol/L GEM and 3 nmol/L nab-paclitaxel for 24 hours. Then the cell pellets were collected, and the nuclear and cytoplasmic enrichments were analyzed by LFQ Proteome Profiling using Thermo Q-Exactive HF-X Orbitrap Mass Spectrometer (Bioproximity). The intensity values of 27,528 proteins within 3,617 sets in each sample were analyzed. A specific set of proteins was found to be altered in the selinexor single-agent treatment (Fig. 4A). This indicates that selinexor treatment alters a limited number of proteins. Among the 27,528 proteins, the levels of 3,535 proteins (13%) were upregulated more than 2-folds in nuclear compartment after selinexor treatment. Only 790 proteins (3%) were downregulated more than 2-fold in nuclear compartment post single-agent treatment (Fig. 4B). This highlights the propensity for nuclear enrichment of XPO1 cargo or associated proteins post selinexor treatment. Interestingly, in the cytoplasm, only 1,124 proteins (4%) were downregulated more than 2-fold after selinexor treatment, whereas 1,439 proteins (5%) were upregulated more than 2-folds (Supplementary Fig. S4A). Annotation of the molecular function of the highest percentage of proteins includes binding (Fig. 4C). Most of the nuclear-enriched proteins were composed of nucleic acid binders (Fig. 4D). These results support the hypothesis that nuclear retention of transcription factors is most affected by XPO1 inhibition. More intriguingly, comparison of selinexor alone with selinexor–GEM–nab-paclitaxel showed 29% (Fig. 4E) of the proteins to be upregulated in the nucleus compared with 13% by selinexor alone. Annotation of the nuclear target pathway indicated alterations in Wnt and cadherin signaling and p53 activation as well as other key pathways affected (Supplementary Fig. S4B). These results suggest a potential biological synergy (in terms of protein nuclear retention) by selinexor–GEM–nab-paclitaxel. In addition, XPO1 was found to be significantly increased in the nucleus and decreased in the cytosol post selinexor treatment (Supplementary Fig. S5). The levels of p53 protein were enhanced in the nucleus post combination treatment (Supplementary Fig. S6). Collectively, these results corroborate the synergy data and demonstrate that selinexor–GEM–nab-paclitaxel has the potential to increase the nuclear retention of proteins, compared with treatment with selinexor alone.

In a previous set of studies, mutations in cysteine 528 to serine 528 by CRISPR/Cas9 editing resulted in complete abrogation of selinexor activity (26). We used the HEK293 wild-type and mutant HEK293 XPO1^C528S pair of cell lines to evaluate the specificity of selinexor–GEM–nab-paclitaxel in PDAC cell (CRISPR/Cas9 scheme shown in Fig. 5A). In the HEK293 wild-type cell line (Fig. 5B), exposure to selinexor causes nuclear retention of the XPO1 target RanBP1. However, in HEK293 XPO1^C528S cells, we do not observe any nuclear retention of RanBP1 at similar doses of selinexor (Fig. 5B, bottom right plot). Selinexor alone or selinexor–GEM–nab-paclitaxel treatment effectively suppressed the growth of the HEK293 wild-type cells (Fig. 5C, **, P < 0.01). No statistically significant growth inhibition was observed in the HEK293 XPO1^C528S cell line by selinexor alone (Fig. 5D). In addition, no statistically significant enhancement was observed in growth inhibition in the combination treatment using selinexor–GEM–nab-paclitaxel when compared with GEM–nab-paclitaxel in the HEK293 XPO1^C528S cells (Fig. 5D). Annexin V FITC apoptosis analysis demonstrated enhancement of apoptosis in selinexor–GEM–nab-paclitaxel-treated HEK293 wild-type cells (Fig. 5E). On the contrary, there was no statistically significant apoptosis induction in the HEK293 XPO1^C528S cells treated with selinexor alone or with the triple combination (Fig. 5F). We further evaluated the impact of CRISPR/Cas9 genome editing on nuclear retention of TSPs using immunofluorescence. As shown in Fig. 5G (left plots), the HEK293 wild-type cells showed nuclear retention of HSP90 post selinexor–GEM–nab-paclitaxel treatment. However, there was no nuclear staining of HSP90 in HEK293 XPO1^C528S cells (Fig. 5G, right plot). Collectively, our results clearly show that selinexor plays a major and very specific role in enhancing the activity of GEM–nab-paclitaxel which is abrogated by mutations in NES-biding pocket of XPO1.

Selinexor either alone or with GEM has been extensively evaluated in subcutaneous and orthotopic PDAC animal tumor models (20, 25, 29). Extending these studies, we first tested the impact of using selinexor as a single agent to treat PDAC-derived CSCs. Selinexor given at 15 mg/kg every other day × 3 weeks caused statistically significant reduction in the growth of PDAC CSC xenograft (Fig. 6A; Supplementary Fig. S7A, P < 0.01). Molecular analysis of the residual tumor tissue demonstrated downregulation of XPO1 (Supplementary Fig. S7B). Having established single-agent activity, we then tested the combination regimen of selinexor–GEM–nab-paclitaxel on CSC xenograft. We recently published data showing that the CSC xenografts demonstrate resistance to both GEM and nab-paclitaxel (27). In line with our hypothesis, selinexor (at low dose of 5 mg/kg) when combined with GEM–nab-paclitaxel led to reduction in tumor growth of PDAC CSC xenograft (Fig. 6B). In view of the tumor-suppressive activity of SINE compounds in tissue culture and CSC-derived models, we evaluated their impact on PDAC PDX. We conducted IHC staining to confirm the human origin of pancreatic PDX tumor using antibodies specifically against HLA class I, human cytokeratin 19, and human Plectin-1 (Supplementary Fig. S7C). Plectin-1 is the biomarker to identify primary and metastatic PDAC. The strong staining of human HLA, cytokeratin 19, and Plectin-1 in cancer cells demonstrated that the PDX was from human PDAC. This tissue also had increased XPO1 protein expression when compared with normal tissue (Supplementary Fig. S7C). Human origin of PDX was additionally validated using RT-PCR for CHEK1 (Supplementary Fig. S7D). The PCR results showed that the PDX tumor sample and MiaPaCa-2 cells had strong human CHEK1 amplification, suggesting their human origin. The weak mouse CHEK1 band observed in PDX sample could possibly come from mouse blood cells. After confirming the human tissue origin, we evaluated the impact of selinexor or KPT-8602 on PDX tumor growth in subcutaneous xenograft. Exposure to selinexor (15 mg/kg QoDx2) or KPT-8602 (20 mg/kg QDx5) resulted in statistically significant reduction in tumor growth of this pancreatic PDX [Fig. 6C (graph; *, P = 0.04 and **, P = 0.0098) and D (tumor photographs)]. The drugs are fairly well tolerated with very minimal loss in mice body weight (Supplementary Fig. S7E). In combination studies, selinexor–GEM–nab-paclitaxel demonstrated much pronounced tumor growth inhibition compared with selinexor alone (Fig. 6E). In a MiaPaCa-2–derived orthotopic model, administration of KPT-330 selinexor (at suboptimal doses of 10 mg/kg twice a week) enhanced the activity of GEM–Abraxane (give at 30 mg/kg i.v. twice a week; Fig. 6F, tumor photographs; Fig. 6G, dot plot and tumor images in Supplementary Fig. S7F). These data suggest that selinexor can be used either alone or in combination with standard-of-care treatments for inhibition of pancreatic tumors.

Encouraged by our in vitro, in vivo, and ex vivo studies, we obtained IRB approval to proceed with phase Ib/II clinical trial (NCT02178436) to test the safety and efficiency of Selinexor 60 mg in a combination with GEM 1,000 mg/m² and nab-paclitaxel 125 mg/m² (days 1, 8, and 15 of a 28-day cycle) as a first line for metastatic PDAC. The patients were eligible for this trial if they are >18 years old, have metastatic PDAC, did not receive prior systemic oncologic treatment or radiotherapy with adequate renal, hepatic, and hematopoietic function. After obtaining the informed consents from the patients and in accordance with Declaration of Helsinki, we treated 9 patients with this combination. Two (40%) out of 5 evaluable patients had partial response, and 2 patients (40%) had stable disease with only 1 patient had a progressive disease, per RECIST 1.1 (Supplementary Table S2). One of the patients with partial response demonstrated exceptionally durable response with a progression-free survival (PFS) of 16 months and an OS of 22 months using this regimen. This patient (66 years old, male) presented with abdominal, back pain, weight loss of 25 pounds, and grade 2 fatigue. Diabetes mellitus was diagnosed a year prior to his presentation. His overall performance status was 1 on Eastern Cooperative Oncology Group (ECOG) scale. Initial evaluation revealed a pancreatic tail mass, peritoneal nodules, and multiple liver lesions. CT-guided biopsy of one of liver metastases confirmed an adenocarcinoma compatible with a pancreatic primary. Serum CA19-9 was elevated at 1,444,000 units/mL (normal range is <36 units/mL). C-reactive protein was 15.2 mg/L (normal range is <9.10 mg/L). Treatment was started with GEM 1,000 mg/m², nab-paclitaxel 125 mg/m², and selinexor 60 mg/m² given on days 1, 8, and 15 of a 28-day cycle. After three cycles, nab-paclitaxel was discontinued due to grade 3 neuropathy and peripheral edema. The patient continued treatment at a reduced dose of GEM (750 mg/m²) combined with 60 mg/m² dose of selinexor. CT scans were performed every 8 weeks with regular evaluations of serum CA19-9. The patient achieved a partial response based on RECIST 1.1 after 2 months of starting therapy. Continued shrinkage was seen in each of the subsequent restaging CT scans for at least 8 months and was followed by disease stability (Fig. 6H). The patient also experienced resolution of pain, improvement of performance status to 0, and weight gain. Serum CA19-9 level dropped precipitously to below 100 (Fig. 6I). Although the patient remained only on selinexor with a reduced dose of GEM for most of the treatment period, and had a PFS of 16 months and OS of 22 months. The patient maintained an excellent ECOG PS of 1 throughout the treatment. It is worth noting that molecular profiling of a pretreatment liver metastasis biopsy based on the CARIS 592 gene panel revealed mutations in three tumor-suppressor genes (NF1 E18 c.2002-1G>T, SMAD4 E9 p.R361H, and TP53 E8 p.R282W) in addition to the expected KRAS-G12D oncogene mutation (Supplementary Tables S3–S6).

In this article, we show that the combination of nuclear export inhibitor selinexor with GEM and nab-paclitaxel can suppress PDAC proliferation and stemness potentially through nuclear retention of multiple TSPs. The activity was quite notable in PDAC stem cell and PDX. In addition, a patient with stage IV PDAC treated with combination achieved a PR in our ongoing phase Ib/II (NCT02178436) clinical study.

With a 5-year survival rate of approximately 7%, PDAC is considered to be among the most lethal malignancies (30). This dismal picture is due to both the lack of early diagnosis and the absence of effective treatments for patients with advanced disease. Therapy resistance has been linked to low drug delivery, low tumor vascularity, and complex immune-suppressive tumor microenvironment (TME; ref. 31). In addition, PDAC tumors have been shown to harbor a resistant population of cells termed CSCs (32). Combination of all of these factors causes PDAC to have high intrinsic resistance to therapeutic treatments (33). Despite this knowledge, attempts to overcome drug resistance by reprogramming the stromal microenvironment to revascularize tumors and enhance drug penetrance have resulted in only marginal benefits in the clinic. This could be due to the fact that most of the studies so far have focused on targeting individual or few set of molecules in the pathways supporting CSCs and DR. Multiple meaningful perturbations are required to impact the TME. Holistic strategies need to be designed that can simultaneously target the pathways that sustain CSCs and the DR. Based on the studies presented here and supporting evidence from other laboratories, we postulate that XPO1 is one potential target of multiple pathways that maintain PDAC. We provide evidence that targeted inhibition of XPO1 via SINE compounds can induce meaningful perturbations in multiple survival and therapy-resistant signaling pathways in PDAC.

SINE blocks nuclear export by covalently binding to cysteine 528 in the XPO1 NES cargo-binding domain in a slowly reversible manner. This results in blocking TSPs in the nucleus which leads to selective apoptosis of cancer cells (17). The clinical SINE compound selinexor and related analogs have a broad activity against different tumor types (18, 19). These compounds have been independently studied for their impact on tumor stromal component, angiogenesis, and the immune system as well as synergy with several chemo and targeted therapies. Selinexor has been shown to impair cellular cross-talk in the TME by blocking osteoclastogenesis in multiple myeloma (34). The drug can suppress proangiogenic and pro-osteolytic cytokines and reduced osteoclastogenesis in prostate cancer models (35). In studies more relevant to PDAC, selinexor has been shown to exert enhanced anticancer activity when combined with T-cell checkpoint inhibitors in solid tumor models (36). Furthermore, SINE compounds can effectively overcome therapy resistance in several solid tumor cell lines (10). Our studies on CSCs are supported by additional findings in acute myeloid leukemia (AML) tumor models showing activity of selinexor against blasts and leukemia-initiating cells both in vitro and in vivo (37).

We used a systematic proteomic analyses approach to better understand selinexor synergy. Compared with single-agent activity, we observed an increase in nuclear retention of TSPs in the selinexor–GEM–nab-paclitaxel combination treatment group. This data suggests that the combination may be selective as opposed to being more cytotoxic. Rather, it shows that selinexor–GEM–nab-paclitaxel combination may exert a more selective synergy through enhanced nuclear retention and activation of tumor suppressors. Annotation of these nuclear targets suggests increased activation of DNA-binding proteins. These results perhaps indicate that selinexor and the combination with GEM–nab-paclitaxel work through nuclear reactivation of TSP signaling. Collectively, the evidence we present here indicates that XPO1 inhibition by selinexor can be a viable therapeutic strategy. In addition, XPO1 inhibition can induce meaningful perturbations on the components critical for supporting the PDAC TME to overcome therapy resistance.

Leptomycin B (LMB), first isolated by researchers as a novel antibiotic from Streptomyces, was found to nonreversibly target Cyc528 in XPO1 (38). LMB (or elactocin) was used in a single phase I study but was discontinued due to severe toxicity that included anorexia and malaise with a narrow therapeutic window (39). Similarly, selinexor alone or combinations with GEM–nab-paclitaxel may have undesirable side effects resulting from nuclear retention of unwanted oncogenes in cancer cell nucleus or toxicity to normal cells. However, data from several early and later stage trials show that selinexor is well tolerated in patients with cancer (22, 40). Selinexor is currently being evaluated in >60 phase I–III studies. More than 2,700 patients have been exposed to selinexor alone or in combination. Objective responses were observed in patients with solid or hematologic malignancies. In these studies, the reported drug-related adverse events (AE) are mostly anorexia, nausea, vomiting, and fatigue (grades 1 and 2) and are well managed with dosing holidays and supportive care. Although the most frequently reported grade 3 or 4 AEs were anemia, thrombocytopenia, fatigue, leukopenia, and lymphopenia, major clinically significant organ or cumulative toxicities were not observed. The phase I data have also validated changes in XPO1 mRNA expression in patient blood samples as well as FOXO nuclear retention as two pharmacodynamic response markers in paired tissue biopsies. In the first study involving 189 patients, evaluable XPO1 expression changes were observed using RT-PCR on leukocytes along with FOXO3a retention in paired patient tissue biopsies using IHC (41). In another study, selinexor plus fludarabine and cytarabine was used for the treatment of refractory or relapsed leukemia or myelodysplastic syndrome. Fifteen of 16 patients were confirmed to have at least 2-fold induction of XPO1 mRNA expression in the blood and FOXO3a nuclear retention in the tumors (42). Our preclinical work in PDAC also confirmed FOXO3a nuclear retention in PDAC cellular models as well as XPO1 protein downregulation in mice xenografts.

In the phase Ib study presented in this report, the interim analysis shows that dose-limiting toxicity was not achieved when selinexor was administered at a dose up to 60 mg/m² with GEM–nab-paclitaxel. Among the responding evaluable patients, we had 80% disease control rate. One can argue that the exceptional response that was observed in one of the patients on this trial who was maintained only on selinexor with reduced dose GEM after only 3 cycles of initial nab-paclitaxel is also observed historically with GEM–nab-paclitaxel combination. However, we hypothesize that this effect could be primarily due to selinexor and its synergy with GEM in this patient as nab-paclitaxel was omitted earlier in the course of the treatment due to severe peripheral sensory neuropathy. The complete phase I trial is part of our forthcoming publication and is beyond the scope of this article. Collectively, these in vitro, ex vivo and early phase clinical data support the expansion phase of SINE compounds for the treatment of PDAC.

P.A. Philip reports receiving other remuneration from Karyopharm Therapeutics (provided funds for the phase I trial on selinexor). No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.S. Azmi, I. Muqbil, A. Mahipal, M. Kamgar, V. Shidham, R.M. Mohammad, P.A. Philip

Development of methodology: A.S. Azmi, J.E. Neggers, D. Daelemans, M. Kamgar, R.M. Mohammad

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.S. Azmi, H.Y. Khan, A. Aboukameel, I. Muqbil, M. Kamgar, M.N. Al-Hallak, S. Kim, P.A. Philip

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.S. Azmi, H.Y. Khan, G. Dyson, R.M. Mohammad

Writing, review, and/or revision of the manuscript: A.S. Azmi, H.Y. Khan, A. Aboukameel, I. Muqbil, A. Aboukameel, D. Daelemans, A. Mahipal, G. Dyson, M. Kamgar, M.N. Al-Hallak, A. Tesfaye, S. Kim, V. Shidham, R.M. Mohammad, P.A. Philip

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.S. Azmi, J.E. Neggers, V. Shidham

Study supervision: A.S. Azmi, A. Tesfaye, R.M. Mohammad, P.A. Philip

Work in the lab of A.S. Azmi is supported by NIH R37 grant 1R37CA215427. The authors thank the SKY Foundation and Perri Family Foundation for supporting part of this study.

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.