Journal of Neuro-Oncology (2020) 149:13–25 https://doi.org/10.1007/s11060-020-03580-y LABORATORY INVESTIGATION Repolarization of tumor infiltrating macrophages and increased survival in mouse primary CNS lymphomas after XPO1 and BTK inhibition Isabel Jiménez1,2 · Júlia Carabia1,2 · Sabela Bobillo2,3 · Carles Palacio2,3 · Pau Abrisqueta2,3 · Carlota Pagès1,2 · Juan C. Nieto1,2 · Josep Castellví4 · Francisco Martínez‑Ricarte5 · Lourdes Escoda6 · Cristóbal Perla7 · Dennis H. Céspedes Torrez7 · Joan Boix1,2 · Noelia Purroy1,2,10 · Lluís Puigdefàbregas1,2 · Joan Seoane2,8,9 · Francesc Bosch2,3 · Marta Crespo1,2 Received: 20 March 2020 / Accepted: 2 July 2020 / Published online: 20 July 2020 © The Author(s) 2020 Abstract Background Patients diagnosed with primary central nervous system lymphoma (PCNSL) often face dismal outcomes due to the limited availability of therapeutic options. PCNSL cells frequently have deregulated B-cell receptor (BCR) signaling, but clinical responses to its inhibition using ibrutinib have been brief. In this regard, blocking nuclear export by using selinexor, which covalently binds to XPO1, can also inhibit BCR signaling. Selinexor crosses the blood–brain barrier and was recently shown to have clinical activity in a patient with refractory diffuse large B-cell lymphoma in the CNS. We studied selinexor alone or in combination with ibrutinib in pre-clinical mouse models of PCNSL. Methods Orthotopic xenograft models were established by injecting lymphoma cells into the brain parenchyma of athymic mice. Tumor growth was monitored by bioluminescence. Malignant cells and macrophages were studied by immunohisto- chemistry and flow cytometry. Results Selinexor blocked tumor growth and prolonged survival in a bioluminescent mouse model, while its combination with ibrutinib further increased survival. CNS lymphoma in mice was infiltrated by tumor-promoting M2-like macrophages expressing PD-1 and SIRPα. Interestingly, treatment with selinexor and ibrutinib favored an anti-tumoral immune response by shifting polarization toward inflammatory M1-like and diminishing PD-1 and SIRPα expression in the remaining tumor- promoting M2-like macrophages. Conclusions These data highlight the pathogenic role of the innate immune microenvironment in PCNSL and provide pre- clinical evidence for the development of selinexor and ibrutinib as a new promising therapeutic option with cytotoxic and immunomodulatory potential. Keywords PCNSL · XPO1 · BTK · Innate immune system Abbreviations NHL Non-Hodgkin lymphoma PCNSL Primary central nervous system ABC-DLBCL Activated B-cell diffuse large B-cell lymphoma lymphoma BCR B-cell receptor GCB-DLBCL Germinal center B-cell diffuse large B-cell lymphoma Isabel Jiménez, Júlia Carabia and Francesc Bosch, Marta Crespo SINE Selective inhibitor of nuclear export have contributed equally to this work BBB Blood–brain barrier BLI Bioluminescence imaging Electronic supplementary material The online version of this PDX Patient derived xenograft article (https​://doi.org/10.1007/s1106​0-020-03580​-y) contains NSG NOD-SCID-γ: non-obese diabetic- severe supplementary material, which is available to authorized users. combined immune deficiency- interleukin * Marta Crespo 2 receptor gamma chain macrespo@vhio.net IHC Immunohistochemistry Extended author information available on the last page of the article SEM Standard error of the mean 13 Vol.:(0123456789) 14 Journal of Neuro-Oncology (2020) 149:13–25 COO Cell of origin suppressor proteins and cell cycle regulators together with CLL Chronic lymphocytic leukemia the activation of tumor suppressor proteins, which translates TAM Tumor-associated macrophages in cell cycle arrest and specific anti-cancer activity across M-CSF Macrophage colony-stimulating factor a wide range of hematological and solid malignancies [16]. IDO Indoleamine 2, 3 dioxygenase In July 2019, selinexor was approved by the FDA to treat MRI Magnetic resonance imaging patients with multiple myeloma while in May 2020 it was CI Combination index approved for systemic relapsed/refractory DLBCL after positive results in a phase IIb trial [17]. Also, the ability of selinexor to inhibit both the BCR and the NF-κB signaling Background pathways makes this drug interesting for studies in NHL [16, 18]. Recently, in a clinical case study, selinexor was reported Primary central nervous system lymphoma (PCNSL) is a to inhibit refractory DLBCL with CNS involvement [19]. In rare and aggressive non-Hodgkin lymphoma (NHL) local- order to provide a pre-clinical rationale for the design of new ized to the CNS in the absence of systemic involvement therapeutic strategies for patients diagnosed with PCNSL, that represents around 4% of all brain tumors and 4 to herein we evaluate the role of XPO1 and BTK inhibition in 6% of all extranodal lymphomas [1]. Approximately 95% intracerebral xenograft murine models, focusing on malig- of PCNSL are classified as activated B-cell diffuse large nant cells and the innate immune microenvironment. B-cell lymphoma (ABC-DLBCL) based on histopathol- ogy, gene expression and mutational landscape [2]. Cur- rent treatment options for PCNSL include high doses of Materials and methods chemotherapy able to cross the blood–brain barrier (BBB) combined with anti-CD20 monoclonal antibodies and the In vivo modeling of PCNSL addition of whole brain radiation in some settings; also, autologous stem cell transplantation is considered for young All animal experiments were approved by the local Ethical patients. Patients diagnosed with PCNSL respond poorly to Committee for the Use of Experimental Animals. Detailed the available treatments and often face dismal outcomes, methods including treatment schedules can be found in Sup- especially in the relapsed setting, with an estimated over- plementary information. Briefly, brains of eight-week-old all survival of 30% at 5 years [3]. This notion of the poor athymic female mice were injected with OCI-Ly10 cells sta- prognosis of PCNSL can be explained by particular biologi- bly transfected with luciferase, as previously reported [20]. cal characteristics of the tumor. First, PCNSL are charac- Tumor growth was monitored by bioluminescence imaging terized by a high frequency of concomitant MYD88 and (BLI) using IVIS® Spectrum system and Living Image soft- CD79B mutations [4] along with lesions related to B-cell ware (PerkinElmer). development and function (e.g. BLIMP1), and the NF-κB Patient derived xenograft (PDX) model was established pathway (e.g. CARD11 or TBL1XR1). The involvement of by intracerebral injection of human lymphoma cells iso- the BCR signaling in PCNSL has prompted the use of the lated from a brain biopsy in eight-week-old NOD-SCID-γ BTK inhibitor ibrutinib, that, although it can cross the BBB (NSG) female mice. Next, expanded ­CD19+ tumor cells [5], achieves wide but short duration responses [6–8]. In were inoculated into the brain parenchyma of eight-week- addition, PCNSL develop in a special microenvironment old athymic female mice as specified above. Human tumor of unique immune surveillance, which could contribute to sample was obtained from a patient diagnosed with PCNSL an inefficient response of the immune system against lym- at Hospital Universitari Joan XIII, Tarragona (Spain) after phoma cells. In this regard, the few reports examining the approval from the local Clinical Research Ethics Committee tumor-infiltrating immune microenvironment show that it is according to the principles of the Declaration of Helsinki mainly composed by macrophages and by T-cells to a lesser and obtaining written informed consent from the patient. extent [9–13]. Also, an intriguing high proportion of PCNSL have genetic lesions that potentially avoid being recognized Flow cytometry and immunohistochemistry (IHC) by T-cells, namely HLA loses and PD-L1/2 amplifications analysis found in up to 80% of patients [14]. Finally, the poor prog- nosis can also be explained by the diminished capacity of Mice brains were collected in cold RPMI-1640 medium some drugs to cross the BBB. Selinexor (KPT-330), a BBB immediately after euthanasia and the two hemispheres were permeable small molecule [15], is a Selective Inhibitor of separated with a razor blade. One hemisphere was used Nuclear Export (SINE) compound that binds to the cargo for IHC and the other one was processed for flow cytom- binding pocket of XPO1 (exportin-1/CRM1) and inhibits its etry. Detailed methods can be found in Supplementary activity. This results in the nuclear accumulation of tumor information. 13 Journal of Neuro-Oncology (2020) 149:13–25 15 Statistical analysis mean radiance = 2.32·107 ph/s ± 1.86·107). Mice were dosed with 5 mg/kg of selinexor or vehicle via oral gavage three Results are expressed as the mean ± standard error of the times a week and subsequently, in order to non-invasively mean (SEM) of at least four independent experiments or monitor the tumor growth, bioluminescence was assessed subjects. The statistically significant differences between twice a week (Fig. 1e). Dose was selected based on previous groups were analyzed using the Mann–Whitney test or pre-clinical data in mouse models of different neoplasias one or two-way ANOVA, and P < 0.05 was considered sig- [25]. Treated mice showed a significantly slower increase nificant. Detailed methods can be found in Supplementary in bioluminescence signal along time (two-way ANOVA: information. p = 0.0002; Fig. 1f) indicating that the treatment with selinexor was able to notably slow down tumor growth. Spe- cific time-point analysis showed that differences were signif- Results icant as soon as 12 days after start of treatment (day 23 after injection: vehicle mean radiance 2.61·108 ph/s ± 8.64·107 DLBCL cell lines have equivalent sensitivity vs. 3.73·107 ph/s ± 1.9·107 in selinexor; p = 0.011) while to selinexor regardless of their cell of origin (COO) differences peaked at day 20 after treatment (day 31 after injection: 8.98·108 ph/s ± 3.13·108 in vehicle vs. 1.19·108 ABC-DLBCL relies heavily on NF-κB signaling and shows ph/s ± 5.58·107 in selinexor group; p = 0.0037; Fig. 1f, rep- chronic BCR activation that is needed for survival, which resentative cases can be seen in Fig. 1h). The blockage of translates into differential sensitivity to drugs targeting intracerebral lymphoma growth induced by selinexor trans- these pathways between ABC and GCB DLBCL cases [21, lated into a significantly increased survival, with a median 22]. Since increased expression of XPO1 has been related survival of 48 days in the treatment group compared to to resistance to chemotherapy and worse prognosis in dif- 34 days in the vehicle group (p < 0.0001; Fig. 1g). At final ferent neoplasias [23], we studied the potential relationship point, histopathological analysis showed multifocal and infil- between expression of XPO1 and sensitivity to selinexor in trative tumors affecting cerebral parenchyma and meninges DLBCL cell lines. Although mRNA expression of XPO1 of both cerebral hemispheres. Cells were highly prolifera- was significantly higher in ABC-DLBCL cell lines (Fig. 1a), tive (Ki-67 100%), CD20-positive and were often found in we did not find differential in vitro sensitivity to selinexor the perivascular space resembling human PCNSL histology. according to COO (Fig. 1b, c). Finally, we interrogated Remarkably, infiltration was observed in both hemispheres, the publicly available data on gene expression of primary showing no preference for the right hemisphere, where the DLBCL cases [24] and we did not observe any association original inoculation of malignant cells was performed. Also, between the COO and the expression of XPO1 (Fig. 1d). we did not observe variations in CD20 intensity among mice or within different areas of the same brain (representative Selinexor blocks tumor growth and prolongs cases shown at Fig. 1i and Supplemental Figure S1). survival in a bioluminescent orthotopic mouse model of PCNSL. The combination of selinexor and ibrutinib synergizes in vitro in DLBCL cell lines and increases We next assessed the role of XPO1 inhibition in PCNSL survival of mice with CNS lymphoma using an intracerebral orthotopic xenograft murine model established by stereotactic injection of the luciferase- The high frequency of molecular alterations in compo- expressing OCI-Ly10 cell line into the cerebral parenchyma nents of the BCR pathway can in part explain the response of nude athymic mice. OCI-Ly10 cell line was selected to BCR inhibitors in PCNSL. In this regard, ibrutinib in because it is derived from a patient diagnosed with ABC- monotherapy in patients diagnosed with relapsed or refrac- DLBCL and its genetic profile includes mutations in MYD88 tory PCNSL achieves higher response rates compared to (L265P) and CD79A (c. 4275_4316del) genes [20] (further systemic DLBCL, however, the duration of the response verified in house), frequent in PCNSL [4]. Additionally, is brief [5–7]. Alongside this, SINE compounds have also OCI-Ly10 cells have successfully been used before in a been shown to inhibit BCR signaling by downregulating the PCNSL xenograft model in athymic mice for pre-clinical protein expression of BTK via enforced IκB nuclear reten- studies [20]. Tumoral growth was monitored using IVIS- tion in primary cells from patients with chronic lympho- Spectrum bioluminescence measurement. Eleven days after cytic leukemia (CLL) [16]. Moreover, the combination of the injection of cells, all animals had developed detectable selinexor and ibrutinib has shown in vitro synergism in CLL tumors restricted to the CNS and were randomly distributed cells [18]. Accordingly, we observed reduced BCR signaling into treatment or vehicle experimental groups (vehicle: n = 8, after treatment of OCI-Ly10 cells with selinexor and ibru- mean radiance = 1.16·107 ph/s ± 0.615·107; treatment: n = 9, tinib (Supplementary Figure S2A), as well as reduced BTK 13 16 Journal of Neuro-Oncology (2020) 149:13–25 A XPO1 expression B Apoptosis C Proliferation D Primary DLBCL ns ns 2 .5 * 25 80 3 XPO1 expression (AU) 2 .0 20 2 60 AnnV/PI ID50 (µM) XPO1 expression MTS ID50 (µM) 1 .5 15 1 40 1 .0 10 0 20 0 .5 5 -1 0 .0 0 0 -2 ABC GCB ABC GCB ABC GCB ABC GCB type III E n=73 n=114 n=52 I Brain Oral treatment with selinexor (5mg/kg) or vehicle Days 0 +4 +11 three times a week +55 CD20 Ki-67 Last animal alive Tumor monitoring (BLI) twice a week Intracerebral Tumor monitoring OCI-Ly10 cells (BLI) injection F G 1.5 × 10 9 100 Vehicle #62 Vehicle n=8 Vehicle n=8 (1.13 · 109 ph/s) BLI Intensity (ph/s) ** Selinexor n=9 Selinexor n=9 Percent survival 1.0 × 10 9 Treatment 50 ** P<0.0001 5.0 × 10 8 * 0 0 0 5 10 15 20 25 30 35 0 20 40 60 Days from injection Days from injection H Day Vehicle Selinexor Selinexor #69 (7.4 · 10 8 ph/s) +11 800 Treatment 600 +23 400 200 +31 Counts Fig. 1  In vitro and in vivo effects of selinexor in PCNSL models. a test at different time points. (*P < 0.05, **P < 0.01, ***P < 0.001. XPO1 relative expression by QRT-PCR. Cells were treated with Graphs show mean ± SEM) (g) Survival curves and (h) representa- increasing doses of selinexor or vehicle (1% DMSO) for 96 h and tive BLI images of the CNS tumors. i IHC analysis showing expres- viability and proliferation was determined by Annexin-V-PI exclu- sion of CD20 and Ki-67 in representative mice brain parenchyma and sion (b) or MTS method (c). d Relative XPO1 expression in DLBCL meninges. The bars represent 5 mm in top panels and 250 µm in bot- patients, using public data from ref [24]. e Scheme representing mice tom panels. ID50: inhibitory dose 50. ABC: activated-B cell. GCB: treatment and monitoring. f Tumor size as measured by BLI in mice germinal center B-cell. BLI: bioluminescence imaging. Ph/s: photons treated with vehicle (n = 8) or selinexor (n = 9). Data is shown until per second. (*P < 0.05, **P < 0.01, Mann–Whitney test. Graphs show day 31, last day when all animals were still alive. Two-way ANOVA mean ± SEM) analysis (P = 0.0002). Asterisks indicate the result of Mann–Whitney expression after 48 h of treatment with selinexor (Supple- between the two drugs (CI) (Supplementary Figure S2C, mentary Figure S2B). Against this background, we hypoth- right panel). esized that combining XPO1 and BTK inhibition in PCNSL We next sought to elucidate whether the synergy observed would have a synergistic therapeutic effect in our models. in vitro could be translated in vivo. Importantly, while Firstly we treated a panel of cell lines in vitro with increas- ibrutinib is mainly metabolized by cytochrome P450, the ing doses of both drugs and analyzed apoptosis after 96 h. metabolism of selinexor is independent of it, therefore it In three out of four ABC-DLBCL cell lines we observed is unlikely that their co-administration could result in any a strong synergism between the two compounds (Supple- effects on the exposure for the other drug [25, 26]. By using mentary Figure S2C); remarkably, treatment with selinexor the same animal model described above, mice were dis- sensitized GCB-SUDHL4 cells to ibrutinib, as shown by tributed into the following four groups and started therapy the combination index values indicating strong synergism 11 days after intracerebral injection of lymphoma cells: 13 Journal of Neuro-Oncology (2020) 149:13–25 17 selinexor monotherapy (5 mg/kg twice a week via oral gav- CNS lymphoma is infiltrated by tumor‑promoting age, n = 12, mean radiance = 3.95·106 ph/s), ibrutinib mono- M2‑like macrophages expressing PD‑1 and SIRPα therapy (25 mg/kg daily in drinking water, n = 9, mean radi- ance = 1.02·107 ph/s), combination therapy (n = 11, mean Analysis of the tumor-infiltrating immune microenviron- radiance = 1.02·107 ph/s) and vehicle (n = 9, mean radi- ment has shown that tumoral cells in PCNSL are accom- ance = 3.21·106 ph/s). Selinexor dose was adjusted (from panied by tumor-associated macrophages (TAMs) and three times a week to twice a week) in order to prevent T-cells to less extend, which is related to bad prognosis. potential toxicity of the drug combination, while ibrutinib [9–13] Remarkably, TAMs in mouse and human colo- dose was based on previous experience in CLL preclinical rectal cancer have been recently described to express the models [27] (Fig. 2a). Compared to vehicle, all three treat- immune checkpoint PD-1 and to recover their potential ment regimens induced an equivalent significant effect in to phagocyte tumoral cells when PD-1 is blocked [28]. tumor growth kinetics in terms of decreased growth rate To conduct an interactive study of the infiltrating innate (Fig. 2b and c). Interestingly, the combination increased the immune cells and PCNSL, we inoculated OCI-Ly10 cells survival of mice compared to vehicle, whereas there was into the brain parenchyma of nude athymic mice, an exper- no significant difference between ibrutinib and selinexor imental in vivo model that has been previously success- alone. Although the median survival increased up to 55 days, fully used to study the modulation of the innate immune the survival curve of the mice treated with the combina- response against PCNSL [20, 29]. Brains were harvested tion was not statistically different from the ones from mice after 24 days of cell injection and further processed for treated with the individual treatments (median survival of subsequent analysis. Histopathological analysis showed mice treated with vehicle: 35 days vs. survival for mice that tumors encompassing both cerebral hemispheres were treated with selinexor: 40 days, p = 0.001; vehicle vs. ibru- infiltrated by macrophages expressing the surface glyco- tinib, 43 days, p = 0.0005; vehicle vs. combination, 55 days, protein F4/80, mainly in the meninges but also in the cer- p = 0.0001; Fig. 2d). ebral parenchyma; notably, F4/80-positive macrophages Fig. 2  Treatment with selinexor and ibrutinib further increases sur- Graphs show mean ± SEM). c Representative BLI images in mice vival of mice with CNS lymphoma. a Scheme representing mice from every treatment arm. d Survival curves of mice in the four treat- treatment and monitoring. b Tumor size as measured by BLI inten- ment groups. Survival curves were generated using the Kaplan and sity. Data is shown until day 29, last day when all animals were still Meier method, and statistically compared by the log-rank test. HR alive. (*P < 0.05, **P < 0.01, ***P < 0.001, Mann–Whitney test. hazard ratio, CI confidence interval, BLI bioluminescence imaging 13 18 Journal of Neuro-Oncology (2020) 149:13–25 were completely absent in the areas of the brain that were the co-expression of PD-1 and SIRPα was also higher in not invaded by tumoral cells (Figs. 3a and Supplemen- the M2 subset (Fig. 3f), pointing out towards a severe inhi- tal figure S1) as well as in healthy brains from control bition of macrophage activity in CNSL. mice (Fig. 3a). Iba-1 staining further identified microglial The response of the innate immune system to PCNSL cells and TAMs, which showed an amoeboid morphol- cells derived from a patient was further analyzed. For that, ogy when interacting with tumoral cells, consistent with we developed an orthotopic PDX model using NSG mice an active state (Fig. 3a) [30]. TAMs can be polarized to initially expand the freshly obtained primary malig- towards a pro-inflammatory (M1) or a tumor-promoting nant cells, as previously described by Rubenstein et al. and (M2) state, depending on microenvironment and external following the detailed protocol described in Supplemen- stimuli [31]. By flow cytometry, we analyzed the propor- tary methods [33] Next, we inoculated 2·10 5 lymphoma tion of M1 and M2 TAMs and their expression of immune cells into the brain parenchyma of nude athymic mice checkpoints in brains from mice with PCNSL. First, we [34]. Since the median survival of this mouse model was observed that TAMs were evenly distributed between M1 22 days, infiltration by immune cells was analyzed after and M2 (Fig. 3c). Of note, TAMs expressed PD-1, mainly 18 days of tumor injection allowing infiltration by innate the tumor promoting M2 subset (Fig. 3d). This suggests immune cells. In this model, TAMs were also found only that the direct interaction of M2 macrophages with the amongst tumoral cells (Figs. 4a and Supplemental Figure tumor triggers the upregulation of PD-1 and thus impairs S3) as assessed by IHC. TAMs from the PDX model dis- their phagocytic capacity, as has been recently discovered played an immunophenotypic profile resembling the one in an analogous role to tumor-infiltrating T-cells using found in TAMs from the cell line xenograft model. Along both immunocompetent syngeneic and athymic xenograft this line, a similar proportion of M1 and M2 (Fig. 4b) mouse models [28]. SIRPα is a well described regulatory and a more frequent expression of PD-1 and SIRPα in checkpoint on macrophages, its interaction with CD47 M2 tumor-promoting macrophages was observed (Fig. 4c, on malignant cells hampering the phagocytosis by mac- d, e). In contrast to the OCI-Ly10 model, patient-derived rophages [32]. Herein we observed that SIRPα was also PCNSL cells did express the SIRPα ligand CD47 (97.61% preferentially expressed by M2 TAMs (Fig. 3e) and that of CD20 cells ± 0.62). Fig. 3  OCI-Ly10 CNS lymphomas are infiltrated by innate immune ing strategy for the analysis of TAMs. Percentage of macrophages cells. a Representative IHC images from brains obtained from three (M1/M2) (c) expressing PD-1 (d), SIRPα (e) and co-expressing both mice inoculated with OCI-Ly10 cells (24 days after injection). The (f) bar represents 500 µm, except for fourth and last rows (50 µm). b Gat- 13 Journal of Neuro-Oncology (2020) 149:13–25 19 Fig. 4  PDX CNS lymphomas are infiltrated by innate immune cells. centage of macrophages (M1/M2) (b) expressing PD-1 (c), SIRPα a Representative IHC images from brains obtained from two mice (d) and co-expressing both (e). (*P < 0.05, **P < 0.01, ***P < 0.001, inoculated with patient-derived PCNSL cells (18 days after injection). Mann–Whitney test. Graphs show mean ± SEM) The bar represents 100 µm except for the four last rows (50 µm). Per- Treatment with selinexor and ibrutinib favors macrophages, the drug combination significantly reduced TAM polarization toward pro‑inflammatory the frequency of PD-1-positive, SIRPα-positive (Fig. 5c, d, M1‑like and diminishes PD‑1 and SIRPα expression e) and double-positive M2 macrophages (Fig. 5f). In agree- in M2‑like TAMs ment the (CI) that the reduction of the expression of PD-1, SIRPα and their co-expression was synergistic (CI < 1). This BTK protein has been shown to be crucial for tumor-promot- was accompanied by a reduction in PD-L1-expressing malig- ing function of macrophages in different neoplasias, espe- nant cells (Fig. 5g, h) that was attributable to ibrutinib action cially in CLL, where modulation of TAMs has been shown since it was also observed under ibrutinib monotherapy. to be also a relevant mode of action of ibrutinib [35, 36]. In the PDX model, the study of immunomodulation was Therefore, after showing that the combination of selinexor performed 18 days after cell injection preceded by 12 days and ibrutinib restrains tumor growth and prolongs mice of oral gavage treatment as described earlier (Fig. 6a). survival, and since both drugs are able to inhibit BTK, we Both treatments alone or in combination were able to hypothesized that these drugs could also cooperate to mod- change the M1/M2 balance towards a more anti-tumoral ify the innate immune response in PCNSL. In this regard, or inflammatory response (Fig. 6b). Moreover, treatment pre-clinical PCNSL models have previously demonstrated with ibrutinib only or with the drug combination was how immunomodulating drugs are able to shift macrophages able to diminish the frequency of PD-1-positive M2 mac- polarization as well as have direct antitumoral effect. [20, rophages (Fig. 6c). The frequency of SIRPα-positive M2 29] To test that, we treated mice bearing OCI-Ly10-CNS macrophages was also diminished by both individual treat- lymphomas with selinexor 5 mg/kg twice a week, ibrutinib ments, as well as the double positive M2 cells (Fig. 6d and 25 mg/kg daily or the combination of the two drugs for two e). In this mouse model we did not observe any effect in weeks by oral gavage (Fig. 5a). We observed that selinexor the expression of PD-L1 by the malignant cells, while the and the combination shifted the M1/M2 ratio towards pre- percentage of malignant cells was also not affected by the dominance of anti-tumoral M1 (Fig. 5b). Interestingly, short term treatment (Fig. 6f and g). Expression of CD47 while none of the individual treatments induced significant by patient-derived PCNSL cells was significantly down- changes in the frequency of PD-1 or SIRPα-positive M2 regulated after treatment with the combination (Fig. 6h). 13 20 Journal of Neuro-Oncology (2020) 149:13–25 Fig. 5  Treatment with selinexor and ibrutinib favors M1-like (e) or co-express both markers (f). g Percentage of C­ D20+ cells in response in tumor-associated macrophages in OCI-Ly10-derived CNS ­ D20+ the brains from mice treated for two weeks. h Percentage of C lymphomas. a Scheme representing mice treatment and monitoring. malignant cells expressing PD-L1 in the different treatment groups. b Percentage of M1 and M2 TAMs by flow cytometry. c Histograms (*P < 0.05, **P < 0.01, ***P < 0.001, Mann–Whitney test. Graphs of ­PD1+ M2 and SIRPα+ M2 of one representative mouse from each show mean ± SEM). CI combination index, BLI bioluminescence group. Frequency of M2 macrophages that express PD-1 (d), SIRPα imaging Accordingly, CI calculations show that the combination methods). M2 macrophages derived from 8 healthy donors did not improve upon individual treatments for any of the had a mean expression of PD-1 of 81.15% + /−8.8 and parameters except for the expression of CD47 on malig- mean expression of SIRPα of 45.53% + /−9.3. Firstly, we nant cells. In order to identify direct immunomodulatory made sure that the drugs did not affect survival of mac- effects of selinexor and ibrutinib on human macrophages, rophages at the concentrations used (data not shown). we treated peripheral blood-derived macrophages in vitro Next, in agreement with what we observed in vivo, we with increasing doses of selinexor, ibrutinib or the com- observed downregulation of the expression of both PD-1 bination for 30 min before inducing differentiation to M2 and SIRPα caused by individual drugs or the combination. using macrophage colony-stimulating factor (M-CSF) (Supplementary Figures S4A, S4B and S4C). However, and IL-10 (see Supplementary information for detailed 13 Journal of Neuro-Oncology (2020) 149:13–25 21 Fig. 6  Treatment with selinexor and ibrutinib favors M1-like ­ D20+ expressing PD-L1 in the differ- Percentage of malignant cells C response in tumor-associated macrophages in CNS lymphoma PDXs. ent treatment groups. Percentage of malignant cells expressing CD47 a Scheme representing mice treatment and monitoring. b Percent- (h) and co-expressing PD-L1 and CD47 (i). (*P < 0.05, **P < 0.01, age of M1 and M2 TAMs by flow cytometry. Frequency of M2 ***P < 0.001, Mann–Whitney test. Graphs show mean ± SEM). CI macrophages that express PD-1 (c), SIRPα (d) or co-express both combination index ­ D20+ cells in the brains from mice. g markers (e). f Percentage of C this did not translate into increased phagocytic activity Discussion (Supplementary Figure S4D). Also, using the same experimental setting we analyzed Blockage of XPO1-mediated nuclear transport using the effect of selinexor, ibrutinib or the combination in SINEs like selinexor has been shown to be an effective interfering with M2 polarization by analyzing additional anti-neoplastic approach in a variety of malignancies. M1 and M2-like markers and IL-10 production. We found [17, 37, 38] XPO1 inhibition forces nuclear localization an increase in the expression of the activation and M1-like of tumor suppressors and also interferes with additional marker CD86 and a decrease in the M2-like marker CD163 signaling pathways, including NF-κB and BCR, which are as well as lower levels of PD-L1 and the anti-inflammatory crucial for survival of malignant B cells in general and for cytokine IL-10 after treatment with selinexor and ibrutinib PCNSL cells in particular. The clinical use of selinexor (Supplementary Figures S4E-H). However, we did not see in lymphoma has been studied in a phase I trial studying any significant effect in the expression of CD206 or HLA- patients diagnosed with relapsed/refractory NHL and a DR (Supplementary Figures S4I–J). In vitro modulation phase IIb study in patients with DLBCL [17], which has of additional surface markers and cytokines is consistent led to a recent approval by the FDA in such an adverse with the loss of pro-tumoral M2 properties after treatment setting. Additionally, based on our pre-clinical experience, with selinexor and ibrutinib. we recently used selinexor in a compassionate way for a Altogether these results indicate that the combination patient diagnosed with DLBCL who developed an isolated of selinexor and ibrutinib is able to block tumoral growth, CNS relapse after several lines of treatment. After a month to significantly increase the median survival of mice with of treatment a partial response was already observed while PCNSL and to modulate the innate immune microenviron- after 5 months of selinexor the patient remained asympto- ment towards a more anti-tumoral stage, likely reinvigor- matic and the MRI (magnetic resonance imaging) showed ating the anti-tumoral phagocytic function of the tumor a complete resolution of the brain tumors [19]. Ibrutinib infiltrating macrophage population in vivo. 13 22 Journal of Neuro-Oncology (2020) 149:13–25 is also able to cross the BBB and is active against CNS immunotherapies that aim at potentiating the autologous lymphoma cells. In this setting, ibrutinib has been assayed anti-tumoral immune response. In this regard, it has been alone [6, 7] or in combination with chemotherapy [5], previously shown how immunomodulation by pomalidome showing high response rates but relatively short remis- in mouse models of PCNSL results in reprogramming of sions, while other BTK inhibitors have showed similar M2 macrophages into M1 [20]. In the clinical setting, both efficacy [39]. Based on all these data, herein we proposed pomalidomide and lenalidomide are showing preliminary to combine selinexor with ibrutinib in models of PCNSL. therapeutic activity in a phase I study in patients diagnosed Exploitation of the immune response to a neoplastic pro- with PCNSL (combined with dexamethasone) [29]. Also, cess is currently a widespread strategy to treat cancer. To lenalidomide in combination with rituximab showed signifi- achieve this, different approaches are being pursued, spe- cant clinical activity in relapsed/refractory PCNSL patients cially focused on harnessing the anti-tumoral capacity of [49, 50]. Combination therapies that not only directly attack T lymphocytes via checkpoint inhibition [40]. Intriguingly, the survival of malignant cells but also alter the immune evading a T-cell mediated immune response seems to be a function are therefore an interesting approach when aiming common feature of PCNSL since a high percentage of cases at achieving long lasting responses. In this regard, inhibit- are affected by both MHC-I loss and/or PD-L1/2 amplifica- ing BTK can have this double effect in B-cell malignan- tion [14], and the infiltration by T lymphocytes is scarce cies, since BTK protein is not only involved in malignant while present [9–12]. However, some immunotherapies have B-cell survival but is also required for the tumor-promoting already shown to be effective in PCNSL, such as anti-CD20 effect of macrophages [35, 36]. Taking this into account, and, more recently, anti-PD-1 therapy, with both preclinical we hypothesized that combining ibrutinib with selinexor [41] and clinical evidences, although with only informa- would also be effective in harnessing the innate immune tion for four patients, where responses lasted a median of response mediated by TAMs in PCNSL. In fact, selinexor 15 months [42]. In agreement, anti-PD-1 is highly effective and ibrutinib combination treatment was able to not only in Hodgkin’s lymphoma [43] even though the expression of increase mouse survival but to shift the innate immune PD-1 on T-cells is heterogeneous and PD-L1/2 amplification response towards a more inflammatory phenotype, specifi- and lack of MHC-I expression on tumoral cells are com- cally defined by downregulation of PD-1 and SIRPα in M2 mon, characteristics that should hamper a T-cell mediated macrophages and increased proportion of M1 macrophages response [44]. In this regard, a role for the innate immune as well as modulation of additional M1 and M2-like proper- system in the development of PCNSL is further supported ties consistent with loss of pro-tumoral M2 characteristics. by recent discovery of PD-1 expression in TAMs [28] and Confirmation of these results and additional studies in the the fact that these immune cells have also been found to be interaction of malignant cells and the immune system in suppressed by the MHC-I system in cancer cells, render- PCNSL using different in vivo models, including syngeneic ing malignant cells that downregulate MHC-I to avoid T mice, is needed to further confirm the potential clinical value cell surveillance exposed to macrophage phagocytosis [45]. of the combination of selinexor and ibrutinib in patients Therefore, paralleling the few PCNSL patients treated with diagnosed with PCNSL. anti-PD-1 achieving a complete response, this effect may be related to a macrophage-mediated anti-tumoral effect after PD-1 pharmacological blockage. Supporting that, herein we describe the presence of brain PD-1-positive M2 Conclusions macrophages in two orthotopic mouse models of PCNSL, including PDXs. The recognition of human malignant cells Our results show that selinexor blocks tumor growth and by mice macrophages has been previously demonstrated in prolongs survival in a bioluminescent mouse model, while mice models of PCNSL [20, 46] and other tumoral mod- its combination with ibrutinib further increases survival. els such as colon cancer [28], pancreatic adenocarcinoma Alongside this, treatment with this combination not only had [47] and T-cell lymphoma [48]. TAMs in CNLS have been a direct cytotoxic effect in malignant cells but also favored found to be supportive of the tumoral growth and related to an anti-tumoral innate immune response by shifting polari- prognosis of patients [9, 13]. Also, indoleamine 2,3 dioxyge- zation of tumor-infiltrating macrophages toward inflamma- nase (IDO) and IL-10, which may be markers of macrophage tory M1 and diminishing PD-1 and SIRPα expression in the infiltration, are related to prognosis or response to immu- remaining tumor-promoting M2 macrophages, highlighting nomodulatory therapy [9, 13, 49]. The observed expression the pathogenic role of the innate immune microenvironment of PD-1 and SIRPα by innate immune cells responding to in PCNSL. Herein we provide pre-clinical evidence for the and interacting with CNS lymphoma cells in vivo indicates development of selinexor and ibrutinib as a new therapeu- that their anti-tumoral effect is partially impaired but also tic option with cytotoxic and immunomodulatory potential opens the opportunity to potentially target these cells by for patients diagnosed with PCNSL, aiming at a durable 13 Journal of Neuro-Oncology (2020) 149:13–25 23 response to improve the fatal prognosis of patients diagnosed References with this disease. 1. Maher EA, Fine HA (1999) Primary CNS lymphoma. Semin Acknowledgements Authors want to thank Dr. Francesc Graus and Oncol 26(3):346–356 Josep González from IDIBAPS, University of Barcelona, Spain and Dr. 2. Camilleri-Broët S, Crinière E, Broët P et al (2006) A uniform Andreu Gabarrós, from IDIBELL, University of Barcelona, Spain, for activated B-cell-like immunophenotype might explain the poor their commitment and support in this project. Authors thank the Cellex prognosis of primary central nervous system lymphomas: analysis Foundation for providing research facilities and equipment. of 83 cases. Blood 107(1):190–196 3. Kerbauy MN, Moraes FY, Lok BH et al (2017) Challenges and Author contributions Designed research and supervised the work: MC opportunities in primary CNS lymphoma: a systematic review. and FB. Performed experiments: IJ, JC, SB, C.Pagès. JCN, JB, NP, Radiother Oncol J Eur Soc Ther Radiol Oncol 122(3):352–361 LP. Analyzed and interpreted data: MC, FB, JS, IJ, JC. Contributed 4. Nakamura T, Tateishi K, Niwa T et al (2016) Recurrent muta- primary samples and analysis: SB, PA, JC, LE, CP, DHC, FM-R. Wrote tions of CD79B and MYD88 are the hallmark of primary cen- the manuscript: IJ, JC, MC, FB. Revised the manuscript: all authors. tral nervous system lymphomas. Neuropathol Appl Neurobiol 42(3):279–290 5. Lionakis MS, Dunleavy K, Roschewski M et al (2017) Inhibi- Funding This work was supported by research funding from the tion of B cell receptor signaling by ibrutinib in primary CNS Instituto de Salud Carlos III, Fondo de Investigaciones Sanitarias lymphoma. Cancer Cell 31(6):833–843.e5 (PI17/00950, M.C., PI17/00943, F.B, PI18/01392, P.A., PI16/01278, 6. Grommes C, Pastore A, Palaskas N et al (2017) Ibrutinib J.S) cofinanced by the European Regional Development Fund (ERDF); unmasks critical role of bruton tyrosine kinase in primary CNS Fundación Asociación Española Contra el Cáncer (M.C. and P.A.) and lymphoma. Cancer Discov 7(9):1018–1029 Gilead Fellowships (GLD16/00144, GLD18/00047, F.B). M.C. holds 7. Soussain C, Choquet S, Blonski M et al (1990) Ibrutinib mono- a contract from Ministerio de Ciencia, Innovación y Universidades therapy for relapse or refractory primary CNS lymphoma and (RYC-2012-12018). S.B. is the recipient of a postdoctoral fellowship primary vitreoretinal lymphoma: final analysis of the phase II from Fundación Alfonso Martin Escudero. “proof-of-concept” iLOC study by the Lymphoma study asso- ciation (LYSA) and the French oculo-cerebral lymphoma (LOC) Data availability The datasets used and/or analyzed and materials from network. Eur J Cancer Oxf Engl 2019(117):121–130 the current study are available from the corresponding author on rea- 8. Grommes C, Tang SS, Wolfe J et al (2019) Phase 1b trial of sonable request. an ibrutinib-based combination therapy in recurrent/refractory CNS lymphoma. Blood 133(5):436–445 Compliance with ethical standards 9. Nam SJ, Kim S, Kwon D et al (2018) Prognostic implications of tumor-infiltrating macrophages, M2 macrophages, regula- tory T-cells, and indoleamine 2,3-dioxygenase-positive cells in Conflicts of interest M. C. has received research funding from Kary- primary diffuse large B-cell lymphoma of the central nervous opharm, Pharmacyclics, Roche, Arqule and AstraZeneca. F.B. has re- system. Oncoimmunology 7(7):e1442164 ceived research funding and honoraria from Roche, Celgene, Takeda, 10. He M, Zuo C, Wang J et al (2013) Prognostic significance of AstraZeneca, Novartis, Abbie and Janssen. J.S. is a co-founder of Mo- the aggregative perivascular growth pattern of tumor cells in saic Biomedicals and Northern Biologics. J.S. received grant/research primary central nervous system diffuse large B-cell lymphoma. support from Mosaic Biomedicals, Northern Biologics and Roche/ Neuro-Oncol 15(6):727–734 Glycart. PA has received honorarium for advisory and speaker fac- 11. Chang C, Lin C-H, Cheng A-L, Medeiros LJ, Chang K-C (2015) ulty from Janssen, Roche, Celgene, and Abbvie. All remaining authors Primary central nervous system diffuse large B-cell lymphoma have declared no conflicts of interest. has poorer immune cell infiltration and prognosis than its peripheral counterpart. Histopathology 67(5):625–635 Ethics approval and consent to participate All animal experiments 12. Venetz D, Ponzoni M, Schiraldi M et al (2010) Perivascular were approved by the local Ethical Committee for the Use of Experi- expression of CXCL9 and CXCL12 in primary central nervous mental Animals. The use of primary cells from patients was approved system lymphoma: T-cell infiltration and positioning of malig- by the local Clinical Research Ethics Committee according to the nant B cells. Int J Cancer 127(10):2300–2312 principles of the Declaration of Helsinki and after obtaining written 13. Sasayama T, Tanaka K, Mizowaki T et al (2016) Tumor-asso- informed consent from the patient. ciated macrophages associate with cerebrospinal fluid interleu- kin-10 and survival in primary central nervous system lym- Informed consent Not applicable. phoma (PCNSL). Brain Pathol Zurich Switz 26(4):479–487 14. Chapuy B, Roemer MGM, Stewart C et al (2016) Targetable Open Access This article is licensed under a Creative Commons Attri- genetic features of primary testicular and primary central nerv- bution 4.0 International License, which permits use, sharing, adapta- ous system lymphomas. Blood 127(7):869–881 tion, distribution and reproduction in any medium or format, as long 15. Etchin J, Berezovskaya A, Conway AS et al (2017) KPT-8602, a as you give appropriate credit to the original author(s) and the source, second-generation inhibitor of XPO1-mediated nuclear export, provide a link to the Creative Commons licence, and indicate if changes is well tolerated and highly active against AML blasts and leu- were made. The images or other third party material in this article are kemia-initiating cells. Leukemia 31(1):143–150 included in the article’s Creative Commons licence, unless indicated 16. Zhong Y, El-Gamal D, Dubovsky JA et al (2014) Selinexor sup- otherwise in a credit line to the material. If material is not included in presses downstream effectors of B-cell activation, proliferation the article’s Creative Commons licence and your intended use is not and migration in chronic lymphocytic leukemia cells. Leukemia permitted by statutory regulation or exceeds the permitted use, you will 28(5):1158–1163 need to obtain permission directly from the copyright holder. To view a 17. Kuruvilla J, Savona M, Baz R et al (2017) Selective inhibition copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/. of nuclear export with selinexor in patients with non-Hodgkin lymphoma. Blood 129(24):3175–3183 13 24 Journal of Neuro-Oncology (2020) 149:13–25 18. Hing ZA, Mantel R, Beckwith KA et al (2015) Selinexor 36. Ping L, Ding N, Shi Y et al (2017) The Bruton’s tyrosine kinase is effective in acquired resistance to ibrutinib and syner- inhibitor ibrutinib exerts immunomodulatory effects through gizes with ibrutinib in chronic lymphocytic leukemia. Blood regulation of tumor-infiltrating macrophages. Oncotarget 125(20):3128–3132 8(24):39218–39229 19. Bobillo S, Abrisqueta P, Carpio C, et al. 2017. Promising activity 37. Das A, Wei G, Parikh K, Liu D (2015) Selective inhibitors of of selinexor in the treatment of a patient with refractory diffuse nuclear export (SINE) in hematological malignancies. Exp Hema- large B-cell lymphoma and central nervous system involvement. tol Oncol 4:7 Haematologica. 38. Gravina GL, Mancini A, Colapietro A et al (2017) Pharmacologi- 20. Li Z, Qiu Y, Personett D et al (2013) Pomalidomide shows sig- cal treatment with inhibitors of nuclear export enhances the anti- nificant therapeutic activity against CNS lymphoma with a major tumor activity of docetaxel in human prostate cancer. Oncotarget impact on the tumor microenvironment in murine models. PLoS 8(67):111225–111245 ONE 8(8):e71754 39. Nagane M, Narita Y, Mishima K, et al. (2019). Phase 1/2 study 21. Yang Y, Shaffer AL 3rd, Emre NCT et al (2012) Exploiting syn- of tirabrutinib (ONO/GS-4059), a Next-Generation Bruton’s thetic lethality for the therapy of ABC diffuse large B cell lym- Tyrosine Kinase (BTK) inhibitor, monotherapy in patients with phoma. Cancer Cell 21(6):723–737 relapsed/refractory primary central nervous system Lymphoma 22. Wilson WH, Gerecitano JF, Goy A et al (2012) The Bruton’s (PCNSL). Blood, 134(Supplement_1): 1586–1586. Tyrosine Kinase (BTK) Inhibitor, Ibrutinib (PCI-32765), has 40. Robert C, Ribas A, Wolchok JD et al (2014) Anti-programmed- preferential activity in the ABC subtype of relapsed/refractory de death-receptor-1 treatment with pembrolizumab in ipilimumab- novo diffuse large B-Cell lymphoma (DLBCL): interim results of refractory advanced melanoma: a randomised dose-comparison a multicenter, open-label, phase 2 study. ASH Annu Meet Abstr cohort of a phase 1 trial. Lancet Lond Engl 384(9948):1109–1117 120(21):686 41. Qiu Y, Li Z, Pouzoulet F et al (2018) Immune checkpoint inhi- 23. Turner JG, Dawson J, Sullivan DM (2012) Nuclear export of bition by anti-PDCD1 (anti-PD1) monoclonal antibody has sig- proteins and drug resistance in cancer. Biochem Pharmacol nificant therapeutic activity against central nervous system lym- 83(8):1021–1032 phoma in an immunocompetent preclinical model. Br J Haematol 24. Rosenwald A, Wright G, Chan WC et al (2002) The use of molec- 183(4):674–678 ular profiling to predict survival after chemotherapy for diffuse 42. Nayak L, Iwamoto FM, LaCasce A, et al. 2017. PD-1 blockade large-B-cell lymphoma. N Engl J Med 346(25):1937–1947 with nivolumab in relapsed/refractory primary central nervous 25. Abdul Razak AR, Mau-Soerensen M, Gabrail NY et al (2016) system and testicular lymphoma. Blood. First-in-class, first-in-human phase i study of selinexor, a selective 43. Ansell SM, Lesokhin AM, Borrello I et al (2015) PD-1 blockade inhibitor of nuclear export, in patients with advanced solid tumors. with nivolumab in relapsed or refractory Hodgkin’s lymphoma. J Clin Oncol Off J Am Soc Clin Oncol 34(34):4142–4150 N Engl J Med 372(4):311–319 26. de Vries R, Smit JW, Hellemans P et al (2016) Stable isotope- 44. Reichel J, Chadburn A, Rubinstein PG et al (2015) Flow sorting labelled intravenous microdose for absolute bioavailability and and exome sequencing reveal the oncogenome of primary Hodg- effect of grapefruit juice on ibrutinib in healthy adults. Br J Clin kin and reed-sternberg cells. Blood 125(7):1061–1072 Pharmacol 81(2):235–245 45. Barkal AA, Weiskopf K, Kao KS et al (2018) Engagement of MHC 27. Ponader S, Chen S-S, Buggy JJ et al (2012) The Bruton tyrosine class I by the inhibitory receptor LILRB1 suppresses macrophages kinase inhibitor PCI-32765 thwarts chronic lymphocytic leuke- and is a target of cancer immunotherapy. Nat Immunol 19(1):76–84 mia cell survival and tissue homing in vitro and in vivo. Blood 46. Yushi Q, Li Z, Von Roemeling CA et al (2016) Osteopontin is a 119(5):1182–1189 multi-faceted pro-tumorigenic driver for central nervous system 28. Gordon SR, Maute RL, Dulken BW et al (2017) PD-1 expres- lymphoma. Oncotarget 7(22):32156–32171 sion by tumour-associated macrophages inhibits phagocytosis and 47. Massó-Vallés D, Jauset T, Serrano E et al (2015) Ibrutinib exerts tumour immunity. Nature 545(7655):495–499 potent antifibrotic and antitumor activities in mouse models of 29. Tun HW, Johnston PB, DeAngelis LM et al (2018) Phase 1 study pancreatic adenocarcinoma. Cancer Res 75(8):1675–1681 of pomalidomide and dexamethasone for relapsed/refractory pri- 48. Horwitz SM, Koch R, Porcu P et al (2018) Activity of the PI3K-δ, mary CNS or vitreoretinal lymphoma. Blood 132(21):2240–2248 γ inhibitor duvelisib in a phase 1 trial and preclinical models of 30. Colonna M, Butovsky O (2017) Microglia function in the central T-cell lymphoma. Blood 131(8):888–898 nervous system during health and neurodegeneration. Annu Rev 49. Rubenstein JL, Geng H, Fraser EJ et al (2018) Phase 1 inves- Immunol 35:441–468 tigation of lenalidomide/rituximab plus outcomes of lena- 31. Sica A, Schioppa T, Mantovani A, Allavena P. (2006). Tumour- lidomide maintenance in relapsed CNS lymphoma. Blood Adv associated macrophages are a distinct M2 polarised population 2(13):1595–1607 promoting tumour progression: potential targets of anti-cancer 50. Ghesquieres H, Chevrier M, Laadhari M et al (2019) Lenalid- therapy. Eur J Cancer Oxf Engl, 1990, 42(6): 717–727 omide in combination with intravenous rituximab (REVRI) in 32. Matlung HL, Szilagyi K, Barclay NA, van den Berg TK (2017) relapsed/refractory primary CNS lymphoma or primary intraocu- The CD47-SIRPα signaling axis as an innate immune checkpoint lar lymphoma: a multicenter prospective “proof of concept” phase in cancer. Immunol Rev 276(1):145–164 II study of the French Oculo-Cerebral lymphoma (LOC) Network 33. Rubenstein JL, Wong VS, Kadoch C et al (2013) CXCL13 plus and the Lymphoma Study Association (LYSA)†. Ann Oncol Off J interleukin 10 is highly specific for the diagnosis of CNS lym- Eur Soc Med Oncol 30(4):621–628 phoma. Blood 121(23):4740–4748 34. Pouzoulet F, Alentorn A, Royer-Perron L et al (2019) Primary Publisher’s Note Springer Nature remains neutral with regard to CNS lymphoma patient-derived orthotopic xenograft model cap- jurisdictional claims in published maps and institutional affiliations. ture the biological and molecular characteristics of the disease. Blood Cells Mol Dis 75:1–10 35. Nguyen P-H, Fedorchenko O, Rosen N et al (2016) LYN kinase in the tumor microenvironment is essential for the progression of chronic lymphocytic leukemia. Cancer Cell 30(4):610–622 13 Journal of Neuro-Oncology (2020) 149:13–25 25 Affiliations Isabel Jiménez1,2 · Júlia Carabia1,2 · Sabela Bobillo2,3 · Carles Palacio2,3 · Pau Abrisqueta2,3 · Carlota Pagès1,2 · Juan C. Nieto1,2 · Josep Castellví4 · Francisco Martínez‑Ricarte5 · Lourdes Escoda6 · Cristóbal Perla7 · Dennis H. Céspedes Torrez7 · Joan Boix1,2 · Noelia Purroy1,2,10 · Lluís Puigdefàbregas1,2 · Joan Seoane2,8,9 · Francesc Bosch2,3 · Marta Crespo1,2 1 6 Experimental Hematology, Vall d’Hebron Institute Department of Hematology, Hospital Universitari Joan of Oncology (VHIO), Vall d’Hebron Barcelona Hospital XXIII, Tarragona, Spain Campus, C/Natzaret, 115‑117, Barcelona 08035, Spain 7 Department of Neurosurgery, Hospital Universitari Joan 2 Department of Medicine, Universitat Autònoma de XXIII, Tarragona, Spain Barcelona, Bellaterra 08193, Spain 8 Translational Research Program, Vall d’Hebron Institute 3 Department of Hematology, Experimental Hematology, of Oncology (VHIO), Vall d’Hebron Barcelona Hospital Vall d’Hebron Hospital Universitari, Vall d’Hebron Institute Campus, C/Natzaret, 115‑117, 08035 Barcelona, Spain of Oncology (VHIO), Vall d’Hebron Barcelona Hospital 9 Institució Catalana de Recerca I Estudis Avançats (ICREA), Campus, Passeig Vall d’Hebron 119‑129, Barcelona 08035, CIBERONC, Barcelona, Spain Spain 10 4 Department of Medical Oncology, Dana-Farber Cancer Department of Pathology, Vall d’Hebron Hospital Institute, Boston, Massachusetts and Broad Institute of MIT Universitari, Vall d’Hebron Barcelona Hospital Campus, and Harvard, Cambridge, MA, UK Passeig Vall d’Hebron 119‑129, Barcelona 08035, Spain 5 Department of Neurosurgery, Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Passeig Vall d’Hebron 119‑129, Barcelona 08035, Spain 13