To the editor

XPO1 (exportin 1) is a well-characterized nuclear export protein responsible for the nuclear-cytoplasmic transport and cellular homeostasis of up to 220 cargoes, including the tumor suppressors p53 and IκB [1, 2]. Abnormal XPO1 expression correlates with worse prognoses in human malignancies. Targeting XPO1 is a promising therapeutic approach in cancer [1, 2]. The XPO1 inhibitor selinexor has received FDA approval recently to treat refractory/relapsed (R/R) diffuse large B-cell lymphoma (DLBCL) after at least 2 lines of systemic therapy, showing an overall response rate of 28% in the SADAL trial [3]. However, it remains largely unknown whether and how XPO1 interplays with other adverse predictors in DLBCL, how to predict selinexor effectiveness, and what combination therapy is optimal in R/R DLBCL patients. Here, we evaluated the prognostic significance of XPO1 expression in 544 well-characterized DLBCL cases, and investigated the therapeutic effect of selinexor in 30 DLBCL cell lines with variable genetic background.

Patients and Methods for this study are detailed in Additional file 1. XPO1 expression was observed in 217 of 544 (40%) DLBCL patients with a mean level of 24%. High level of XPO1 expression (XPO1high; > 30%) predicted significantly poor progressive-free survival (PFS) and overall survival (OS) in DLBCL patients (Fig. 1a). DLBCL is classified into prognostic favorable germinal center B-cell-like (GCB) and unfavorable activated B-cell-like (ABC) subtypes [4]. XPO1high significantly shortened the PFS/OS in ABC-DLBCL but not GCB-DLBCL (Additional file 1: Figure S1A–B). XPO1high showed significant association with p53 overexpression (p53+) and dual p53+MYChigh expression but not clinical features (Additional file 1: Table S1), unlike a previous study using a different scoring system for XPO1 expression in 131 DLBCL patients [5].

Fig. 1
figure 1

Impact of XPO1 expression on patient survival in DLBCL. a In the entire cohort, DLBCL patients with high level of XPO1 expression (XPO1high) had significantly worse OS and PFS than those with low or negative XPO1 expression (XPO1low). b XPO1high remarkably worsened the OS/PFS of DLBCL patients with BCL2high expression. c XPO1high significantly worsened the OS/PFS of patients with dual MYChighBCL2high expression, and showed a trend of unfavorable effect on OS in patients with dual MYC/BCL2 rearrangements (MYC-R+BCL2-R+, HGBCL-DH). d In TP53-mutated (Mut) DLBCL patients without Mut-p53 overexpression, XPO1high showed a trend of unfavorable prognostic effect on OS. e In Mut-TP53 DLBCL patients with Mut-p53 overexpression, XPO1high showed favorable prognostic effect, which was not significant in overall patients but significant in the subset with low BCL2 expression. f Significantly differentially expressed genes between XPO1high and XPO1low patients with concurrent Mut-TP53 and MYChigh

Full size image

Whether XPO1high interacts with other adverse prognostic factors and whether XPO1 is a potential therapeutic target in high-risk DLBCL patients were further examined. XPO1high remarkably worsened the OS and PFS of DLBCL with BCL2high or dual MYChighBCL2high expression (Fig. 1b,c), which is known as double-expressor lymphoma with unfavorable prognosis [6]. Trends of adverse impact were also observed on PFS in MYC-rearranged (R+) patients (P = 0.097; Additional file 1: Figure S1C) and OS in patients with dual MYC-R+BCL2-R+ (Fig. 1c) with dismal prognosis, defined as high-grade B-cell lymphoma with MYC/BCL2 double-hit (HGBCL-DH) [7]. In patients with TP53 mutation (Mut-TP53) [8], XPO1high showed opposite prognostic effects in patients with and without Mut-p53 protein overexpression [9], suggesting the nuclear export may attenuate the oncogenic gain-of-function of Mut-p53. In contrast to the negative impact of XPO1high in Mut-TP53/p53-negative patients (Fig. 1d) and in TP53-wild type (Wt-TP53) patients (Additional file 1: Figure S1D), a favorable effect was associated with XPO1high in Mut-TP53/p53-positive patients, which was significant in the BCL2low subset (Fig. 1e). Gene expression profiling [4] analysis identified a distinct gene expression signature for XPO1high in patients with Mut-TP53 and MYChigh (Fig. 1f), including upregulation of SIRPA, which encodes SIRPα, a receptor for CD47 transmitting “do not eat me” signal in phagocytosis, and downregulation of several genes related to DNA repair, metabolism, splicing, or biosynthesis (Additional file 1: Table S2).

Next, selinexor was assessed in 30 DLBCL cell lines, which resulted in significantly reduced cell viability with varying IC50 values (Fig. 2a). ABC-DLBCL and GCB-DLBCL cells were similarly vulnerable to selinexor (Additional file 1: Figure S1E), consistent with results in the SADAL clinical trial [3]. Biomarkers significantly associated with higher sensitivity (lower IC50) to selinexor cytotoxicity included BCL2-R and HGBCL-DH (Fig. 2b) but not MYC-R. In contrast, presence of Mut-TP53/p53+ significantly reduced the anti-lymphoma efficacy of selinexor, especially in HGBCL-DH cells (Fig. 2c; Additional file 1: Figure S1F).

Fig. 2
figure 2

Therapeutic effect of selinexor alone or in combination with a BET inhibitor INCB057643 in DLBCL cellular models. a The effect of 72-h selinexor exposure on cell viability of 30 DLBCL cell lines. Waterfall graph showed the specific IC50 value of selinexor for each cell line with either ABC or GCB subtype of DLBCL. b DLBCL cell lines with BCL2 rearrangement (BCL2-R) or HGBCL-DH were more sensitive to selinexor with a lower mean IC50 value compared with other cell lines. c The presence of mutant (Mut) p53 in DLBCL cells significantly reduced the cytotoxicity of selinexor, especially significant in HGBCL-DH cell lines. Selinexor promoted more significant apoptosis in Wt-TP53 HGBCL-DH cells than in Mut-TP53/p53-expressing HGBCL-DH cells. d INCB057643 and selinexor were cooperative in reducing cell viability and inducing apoptosis in HGBCL-DH cells with Wt-TP53 or Mut-TP53/p53+

Full size image

Limited efficacy of selinexor in HGBCL with Mut-TP53/p53+ calls for combination strategy. Previous studies showed the synergy between selinexor and venetoclax in DLBCL and double-hit lymphoma [10, 11]. However, in the SADAL trial [3], patients with MYChigh (but not BCL2high) expression had a lower overall response rate than those without. MYC expression can be inhibited by targeting the bromodomain and extra-terminal domain (BET) proteins [12]. We therefore combined selinexor with a novel BET inhibitor INCB057643. Synergistic effect was observed in DLBCL/HGBCL cells, especially in HGBCL-DH cells with Mut-TP53/p53+ (Fig. 2d), supporting INCB057643/selinexor combination as a therapeutic option for HGBCL-DH patients.

In summary, this study demonstrates that XPO1high is a valuable biomarker in DLBCL with unfavorable prognostic factors, predictive of significantly poorer outcomes in ABC-DLBCL, BCL2high DLBCL, and double-expressor lymphoma but not Mut-p53-expressing DLBCL. Targeting XPO1 with selinexor is similarly effective in GCB-DLBCL and ABC-DLBCL cells, and remarkably effective in BCL2-R+ DLBCL and HGBCL cells without Mut-TP53/p53-positivity. In DLBCL/HGBCL cells, Mut-TP53/p53-positive expression predicts resistance to selinexor. INCB057643 synergizes with selinexor in HGBCL-DH cells, overcoming resistance in Mut-TP53/p53-positive HGBCL-DH. These findings warrant future investigation on the role of XPO1, selinexor, and combined BET inhibition in R/R DLBCL and HGBCL-DH.