ALBERT

Description: Key Points Germline activating STAT3 mutations were detected in 3 patients with autoimmunity, hypogammaglobulinemia, and mycobacterial disease. T-cell lymphoproliferation, deficiency of regulatory and helper 17 T cells, natural killer cells, dendritic cells, and eosinophils were common.

Description: Background T-cell acute lymphoblastic leukemia (T-ALL) is caused by the cooperation of multiple oncogenic lesions. Recent evidence supports that IL-7 and its receptor IL-7R contribute to T-ALL development (Zenatti et al, 2011). The two main pathways induced by IL-7R are JAK/STAT5 and PI3K/Akt/mTOR. Activating mutations to IL7R, JAK1, JAK2 or JAK3 are estimated to occur in 20-30% of all T-ALL patients (Cools 2013). STAT5 plays an important role in many hematologic malignancies but constitutive STAT5 activation often is a secondary event. Mutations in STAT5B (N642H) were recently described in LGL-leukemia in patients with an unusually aggressive and fatal form of the disease (Rajala et al, 2013). In other cancers, including ALL, patients with mutations in STAT5B have not been described. Here we report novel activating STAT5B mutations as drivers of T-ALL. Methods We performed exome sequencing of bone marrow (BM) samples from an 18-year-old female with relapsed T-ALL. Targeted next-generation amplicon sequencing and Sanger sequencing was used to analyze the region encoding the STAT5B SRC homology 2 (SH2) domain including the N642, T648 and I704 codons in a cohort of 17 adult and pediatric T-ALL patients treated at HUCH 2008-2013. For functional studies STAT5B expression vectors with the N642H, T648S or I704L mutation and an expression vector with both N642H and T648S mutations were used to transiently transfect HEK293 cells. To investigate the effect on transcriptional activity we co-transfected the mutant constructs with a STAT5 luciferase reporter plasmid and used Western blot analysis to study the phosphorylation status of the generated constructs. For drug sensitivity of STAT5B mutated cells we performed ex vivo drug testing on primary blasts from the index patient using a comprehensive set of 202 oncology drugs (approved and in clinical development). Each drug was tested over a 10,000-fold concentration range. Results Sequencing of the index patient revealed 3 different somatic missense mutations in STAT5B (T648S, N642H, I704L) and mutations in KRAS, WT1 and SUZ12. No mutations affecting the JAK genes or IL7R were detected. All STAT5B mutations were located in the SH2 domain, which mediates dimerization and activation by trans-phosphotyrosine binding. The same three STAT5B mutations were also found in the diagnostic sample and most likely represent founding events in leukemogenesis. The N642H and T648S mutations occurred on the same allele with tumor mutation frequencies of approximately 40% while the I704L mutation occurred on a different allele with a similar tumor mutation frequency. To investigate the prevalence of STAT5B mutations in T-ALL we sequenced 17 BM samples from T-ALL patients. In this cohort we could not detect any other patients carrying mutations in the STAT5B SH2 domain. Western blot analysis made with mutant constructs showed that the N642H and I704L mutations induced constitutive phosphorylation of STAT5B. Compared to wild type STAT5B the N642H and I704L mutants induced 47- and 6-fold increases in transcriptional activity, respectively, while T648S mutation had no effect in the assays. The construct with both the N642H and T648S mutations showed the highest amount of constitutive phosphorylation and induced a 56-fold increase in transcriptional activity compared to wild type STAT5B. Using ex vivo drug testing the STAT5B mutated blasts were resistant (EC50≥1 uM) to inhibitors of PI3K (e.g. idelalisib, XL147), dual inhibitors of PI3K/mTOR (PF-04691502, dactolisib) and mTOR inhibitors (temsirolimus, everolimus). Furthermore the blasts showed no response to AKT1 inhibitors (MK-2220) or JAK inhibitors (ruxolitinb, tofacitinib). In contrast, the cells were most sensitive to the BCL-2/BCL-XL inhibitor navitoclax (EC50 83 nM). Summary STAT5B mutations are uncommon in T-ALL but their occurrence underlines the significance of the IL7R-JAK-STAT5 pathway in the pathogenesis of T-ALL. While STAT5B mutant blasts were not sensitive to inhibitors targeting JAK kinases, the cells were unusually sensitive to inhibitors of target molecules of STAT5B, including anti-apoptotic BCL-2 proteins. These results suggest that BCL-2/BCL-XL inhibitors such as navitoclax are novel candidate therapies for T-ALL patients. Disclosures: Mustjoki: Novartis: Consultancy, Speakers Bureau; BMS: Consultancy, Speakers Bureau. Porkka:BMS: Consultancy, Research Funding, Speakers Bureau; Novartis: Consultancy, Research Funding, Speakers Bureau.

Description: Introduction Response to treatment for multiple myeloma (MM) patients is variable and often unpredictable, which may be attributed to the heterogeneous genomic landscape of the disease. However, the effect of recurrent molecular alterations on drug response is unclear. To address this, we systematically profiled 50 samples from 43 patients to assess ex vivo sensitivity to 308 anti-cancer drugs including standard of care and investigational drugs, with results correlated to genomic alterations. Our results reveal novel insights about patient stratification, therapies for high-risk (HR) patients, signaling pathway aberrations and ex-vivo-in-vivo correlation. Methods Bone marrow (BM) aspirates (n=50) were collected from MM patients (newly diagnosed n=17; relapsed/refractory n=33) and healthy individuals (n=8). CD138+ plasma cells were enriched by Ficoll separation followed by immunomagnetic bead selection. Cells were screened against 308 oncology drugs tested in a 10,000-fold concentration range. Drug sensitivity scores were calculated based on the normalized area under the dose response curve (Yadav et al, Sci Reports, 2014). MM selective responses were determined by comparing data from MM patients with those of healthy BM cells. Clustering of drug sensitivity profiles was performed using unsupervised hierarchical ward-linkage clustering with Spearman and Manhattan distance measures of drug and sample profiles. Somatic alterations were identified by exome sequencing of DNA from CD138+ cells and skin biopsies from each patient, while cytogenetics were determined by fluorescence in situ hybridization. Results Comparison of the ex vivo chemosensitive profiles of plasma cells resulted in stratification of patients into four distinct subgroups that were highly sensitive (Group I), sensitive (Group II), resistant (Group III) or highly resistant (Group IV) to the panel of drugs tested. Many of the drug responses were specific for CD138+ cells with little effect on CD138- cells from the same patient or healthy BM controls. We generated a drug activity profile for the individual drugs correlating sensitivity to recurrent alterations including mutations to KRAS, DIS3, NRAS, TP53, FAM46C, and cytogenetic alterations del(17p), t(4;14), t(14;16), t(11;14), t(14;20), +1q and -13. Cells from HR patients with del(17p) exhibited the most resistant profiles (enriched in Groups III and IV), but were sensitive to some drugs including HDAC and BCL2 inhibitors. Samples from patients with t(4;14) were primarily in Group II and very sensitive to IMiDs, proteasome inhibitors and several targeted drugs. Along with known recurrently mutated genes in myeloma, somatic mutations were identified in genes involved in several critical signaling pathways including DNA damage response, IGF1R-PI3K-AKT, MAPK, glucocorticoid receptor signaling and NF-κB signaling pathways. The predicted impact of these mutations on the activity of the pathways often corresponded to the drug response. For example, all samples bearing NF1 (DSS=21±7.9) and 67% with NRAS (DSS=15±4.35) mutations showed higher sensitivity to MEK inhibitors compared to healthy controls (DSS=5±.21). However, sensitivity was less predictable for KRAS mutants with modest response only in 47% samples (DSS=7±2.14) . One sample bearing the activating V600E mutation to BRAF showed no sensitivity to vemurafenib, which otherwise has good activity towards V600E mutated melanoma and hairy-cell leukemia. Comparison of the chemosensitive subgroups with survival showed patients in Groups I and IV had high relapse rate and poor overall survival. The ex vivo drug sensitivity results were used to decide treatment for three HR patients with results showing good ex vivo -in vivo correlation. Summary Our initial results suggest that ex vivo drug testing and molecular profiling of MM patients aids stratification. Grouping of patients based on their ex vivo chemosensitive profile proved extremely informative to predict clinical phenotype and identify responders from non-responders. While some molecular markers could be used to predict drug response, others were less predictive. Nevertheless, ex vivo drug testing identified active drugs, particularly for HR and relapsed/refractory patients, and is a powerful method to determine treatment for this group of patients. Disclosures Silvennoinen: Genzyme: Honoraria; Sanofi: Honoraria; Janssen: Research Funding; Celgene: Research Funding; Research Committee of the Kuopio University Hospital Catchment Area for State Research Funding, project 5101424, Kuopio, Finland: Research Funding; Amgen: Consultancy, Honoraria. Porkka:Bristol-Myers Squibb: Honoraria; Celgene: Honoraria; Novartis: Honoraria; Pfizer: Honoraria. Heckman:Celgene: Honoraria, Research Funding.

Description: Introduction T cell acute lymphoblastic leukaemia (T-ALL) is an aggressive haematological malignancy affecting 10-15% of pediatric ALL patients. Current cure rates of pediatric patients is 80% but increasing understanding of the molecular mechanisms of T-ALL provides possibilities for more effective and targeted therapies. During the last decade targeted therapies with tyrosine kinase inhibitors (TKI) have proven to be effective in BCR-ABL1 fusion positive leukemias. Several studies have suggested TKI dasatinib to be effective also in the treatment of NUP214-ABL1 positive T-ALLs, which comprise approximately 4-10% of the T-ALL cases. Materials and Methods In silico drug screening was performed by comparing gene expression profiles of 4769 leukemic samples to a library of 13384 compounds and their known targets from the Drug signature database (DsigDB). These compounds included FDA approved therapeutic molecules and molecules under studies. Findings were validated in an ex vivo drug screen, consisting of 20 T-ALL bone marrow samples and 9 healthy bone marrow controls. Samples were treated for 72 hours with five different concentrations of dasatinib in 10-fold dilutions (0.1-1000nM), cell viability was measured and the data normalized to negative (DMSO) and positive (benzethonium chloride) controls. The effect of dasatinib was further explored in vitro by treating one NUP214-ABL1 fusion positive and six fusion negative T-ALL cell lines with dasatinib (1-1000nM). Gene expression levels of the known dasatinib targets in these cell lines were measured by Global Run On sequencing (GRO-seq) assay and qRT-PCR. Results In order to find novel targeted therapies for T-ALL, we performed an in silico drug target screen. A dasatinib-targetable gene LCK was strongly expressed in a number of T-ALL cases whereas normal T-lymphoid cells had lower expression. Chemical screen data of the target specificity of dasatinib showed high inhibition of LCK with percent of control (POC) value of 1, meaning that 0.1µM concentration of dasatinib decreases the kinase activity of LCK to 1% in comparison to control. In vitro dasatinib decreased cell viability in fusion negative Jurkat and MOLT-16 cells, and also in fusion positive cell line Peer. GRO-seq and qRT-PCR confirmed the expression of LCK and several other known dasatinib targets, including other SRC family kinases, in Jurkat and MOLT-16 cell lines. However, some LCK-expressing T-ALL cell lines were less sensitive to dasatinib. In further validation, ex vivo drug testing of patient samples revealed a marked response in 6/20 patient samples with IC50 values ranging between 1.3 - 8.2nM, while in healthy bone marrow controls IC50 values were 〉1000nM. Conclusion Our in silico drug screen identified dasatinib as a potential targeted therapy for a subgroup of T-ALL cases, and this finding was further supported by both ex vivo and in vitro studies. The exact mechanism remains to be elucidated but a number of SRC family kinases, which could potentially be targeted by dasatinib, showed expression in T-ALL samples. Disclosures Heckman: Celgene: Research Funding; Pfizer: Research Funding.

Description: Introduction Multiple myeloma (MM) is an incurable malignant plasma cell disease with the highest incidence occurring at 65-70 years of age while 10% of patients are diagnosed below 55 years of age. The International Myeloma Working Group recently proposed new risk stratification standards for MM patients: high-risk (HR), standard (SR) and low-risk (LR) groups (Leukemia 2014, 28, 269−77). Although a median overall survival of LR patients is 〉 10 years from the diagnosis, new drugs and therapeutic innovations are urgently needed for HR patients (20%) who have a median overall survival of only two years. To identify new treatment options for MM patients, we compared ex vivo drug sensitivity data from primary CD138+ cells to standard risk stratification markers. Ex vivo responses indicated a number of investigational drugs as potential novel options for HR MM patients with links to risk markers. Methods Bone marrow aspirates were collected from newly diagnosed (n=14) and relapsed/refractory (n=21) MM patients. Cytogenetics were determined by fluorescence in situ hybridization (FISH) and the patients stratified based on the presence or absence of adverse FISH markers (t(4;14) and 17p del). Plasma cells (CD138+) were enriched from freshly isolated bone marrow samples and exome sequencing performed using DNA extracted from the CD138+ cells and matched skin biopsies. Ex vivo drug sensitivity was assessed by measuring the viability of the cells after 3-day incubation with 306 different oncology drugs in a 10,000-fold concentration range. Drug sensitivity scores were calculated based on the normalized area under the dose response curve (Scientific Reports 2014, 4, 5193) and select sensitivities determined by comparing results to healthy bone marrow cells. Based on drug sensitivities, the patients were classified in four different groups (sensitive, moderately sensitive, resistant and highly resistant). Results Of the 35 patients included in this study, 11 were classified as HR (31%) and 24 as SR/LR (69%). In the HR group 6/11 (55%) had t(4;14) and 5/11 patients (45%) had 17p13 del. In the SR/LR group common abnormalities included 13 monosomy/13q del (10/24), 1q gain (10/24) and K/NRAS mutation (11/24). Within the HR group, other co-occurring abnormalities included 1q gain (9/11), 13 monosomy/13q del (6/11), K/NRAS mutation (5/11), and TP53 mutation (2/11). Based on overall ex vivo drug sensitivity profiles of all patients, the majority of HR patients were classified as moderately sensitive (8/11; 73%) while SR/LR patients had diverse responses from sensitive to highly resistant. In the HR group, the highest select sensitivities were to BH3 mimetics and PI3K/mTOR inhibitors. While the t(4;14) is predicted to lead to upregulation and increased activity of the FGFR3, which could be targeted by FGFR inhibitors, none of the t(4;14) samples showed sensitivity to these drugs. However, with the exception of one t(4;14) sample, the rest all showed good sensitivity to dual PI3K/mTOR inhibitors, but not to rapalogs, suggesting that inhibition of PI3K and the mTORC1/2 complexes is required to inhibit t(4;14) cell growth rather than mTORC1 alone. Of the 17p del patients, 3/5 were classified as moderately sensitive, 1/5 sensitive and 1/5 highly resistant based on ex vivo drug response of CD138+ cells. All showed select sensitivity to BH3 mimetics/BCL2 inhibitors (navitoclax/ABT-263 and venetoclax/ABT-199/GDC-0199), while response to other drugs varied. Therefore, blocking cell survival signaling is likely essential for this group of HR MM patients. Conclusion By assessing the ex vivo sensitivity of primary plasma cells to a large collection of oncology drugs and comparing these data to standard risk stratification markers for MM, we have been able to identify potential new treatment options for high risk MM patients including dual PI3K/mTOR and BCL2- inhibitors. Although a larger cohort of patients is required to support the correlation between specific drug sensitivities and risk markers, these preliminary data indicate that currently used risk markers may be useful to predict the use of novel treatments. Disclosures Silvennoinen: Janssen-Cilag: Research Funding; Celgene: Research Funding; Janssen-Cilag: Honoraria; Sanofi: Honoraria; Celgene: Honoraria. Porkka:BMS: Honoraria; BMS: Research Funding; Novartis: Honoraria; Novartis: Research Funding; Pfizer: Research Funding. Heckman:Celgene: Research Funding.

Description: Background The t(5;11)(q35;p15.5) translocation resulting in fusion of the nucleoporin NUP98 and methyltransferase NSD1 (NUP98-NSD1) genes is a recurrent aberration observed in pediatric and adult AML. The NUP98-NSD1 fusion often co-occurs with the FLT3-ITD mutation and characterizes a group of cytogenetically normal AML patients with very poor prognosis. Despite advances in the understanding of the biology of NUP98-NSD1-positive AML, its therapeutic success rate has remained low. We aimed to identify novel candidate drugs for NUP98-NSD1-positive AML by testing primary patient cells and in vitro cell models with a high-throughput drug sensitivity platform. Methods Leukemic blasts were Ficoll separated from bone marrow (BM) aspirates of an AML patient positive for t(5;11)(q35;p15.5) and FLT3-ITD. RNA extracted from primary cells was used for RNA sequencing and gene expression analysis. NUP98-NSD1 cDNA was amplified from primary cell RNA and expressed from a lentiviral vector (LeGO-iCer2) also encoding the cerulean fluorescent marker. The NUP98-NSD1/LeGo-iCer2 and empty LeGo-iCer2 viruses were used to establish stably expressing Ba/F3 cell lines. Primary murine (BALB/c) BM cells were transduced with NUP98-NSD1 and FLT3-ITD retroviruses alone or in combination (NNF) in vitro (“preleukemic”) or passaged in vivo (“leukemic”) as previously described (Thanasopoulou et al, 2014). For screening, 309 small molecule inhibitors including FDA/EMA-approved and investigational oncology drugs were plated on 384-well plates in a 10,000-fold concentration range. Cells were dispensed on the pre-drugged plates and incubated at 37°C for 72h, and then cell viability measured using the CellTiter-Glo® luminescent assay. Drug response curves were generated and a drug sensitivity score determined (Yadav et al, 2014). Select drug sensitivity was calculated for each drug by comparing results between primary leukemic and healthy donor BM cells or between the cell constructs and empty vector transduced controls cells. Results Primary patient cells and murine BM cells expressing FLT3-ITD alone or in combination with NUP98-NSD1 were selectively sensitive to specific FLT3 inhibitors (e.g. quizartinib, sorafenib and lestaurtinib), and broad-spectrum receptor tyrosine kinase inhibitors targeting FLT3-ITD (e.g. cabozantinib, crenolanib, foretinib, midostaurin, MGCD-265 and ponatinib). Furthermore, these cells were highly sensitive to checkpoint kinase 1/2- inhibitor AZD7762. The primary murine cells expressing both NUP98-NSD1 and FLT3-ITD showed higher sensitivity to all of the above-mentioned drugs compared to cells expressing either of the events alone indicating functional synergy. A very distinct drug response pattern was observed in the leukemic NNF cells cultured in vivo compared to the same cells cultured in vitro suggesting that microenvironment may also affect the observed drug responses. Interestingly, the preleukemic murine cells expressing NUP98-NSD1 with or without FLT3-ITD as well as the primary patient cells showed extreme vulnerability to BCL2/BCL-xL inhibitor navitoclax. Furthermore, primary murine cells expressing NUP98-NSD1 alone showed high select sensitivity to JAK-inhibitors ruxolitinib, BMS-911543, AZD1480 and tofacitinib indicating the fusion may stimulate JAK/STAT-signaling. Similar sensitivity was also observed in the Ba/F3-cells expressing NUP98-NSD1. In support of these findings, gene expression analyses showed high expression of anti-apoptotic factors BCL2, BCL-xL and MCL1 in the patient cells. MCL1 is regulated by STAT3 while BCL-xL is regulated by STAT5, which were also highly expressed. Conclusions In summary, we have observed an enhanced response to specific and non-specific FLT3 inhibitors in cells expressing NUP98-NSD1 and FLT3-ITD together compared to cells expressing either of the two alone. This coincides with previous findings that functional co-operation between NUP98-NSD1 and FLT3-ITD is important in AML (Thanasopoulou et al, 2014). We have seen high in-vitro-in-vivo correlation between primary patient cells and murine cells expressing NUP98-NSD1 and FLT3-ITD. Moreover, we have identified potential candidate compounds targeting oncogenic signaling activated by these two events. These data form a basis for clinical evaluation of candidate compounds for NUP98-NSD1-positive AML. Disclosures Porkka: Bristol-Myers Squibb: Honoraria, Research Funding; Novartis: Honoraria, Research Funding. Heckman:Celgene: Research Funding.

Description: Introduction New drugs have improved survival for multiple myeloma (MM) patients, however, patient outcome remains highly variable, unpredictable and often very poor. To identify novel treatments and potential biomarkers, we applied high throughput ex vivo drug sensitivity testing combined with exome and transcriptome sequencing to samples collected from newly diagnosed and relapsed MM patients. Integration of results from the different platforms indicated several oncogenic signaling pathways driving drug response and highlighted the importance of a multi-targeted approach for treatment. Methods Bone marrow (BM) aspirates (n=48) were collected from MM patients (newly diagnosed n=14; relapsed/refractory n=26) and healthy individuals (n=8). CD138+ plasma cells were enriched by Ficoll separation followed by immunomagnetic bead selection. Cells were screened against 306 oncology drugs with the drugs tested in a 10,000-fold concentration range. Drug sensitivity scores were calculated based on the normalized area under the dose response curve (Yadav et al, Sci Reports, 2014). Importantly, MM selective responses were determined by comparing data from MM patients with those of healthy BM cells. Clustering of drug sensitivity profiles was performed using unsupervised hierarchical ward-linkage clustering with Spearman and Manhattan distance measures of drug and sample profiles. Somatic mutations were identified by exome sequencing of DNA from CD138+ cells and skin biopies from each patient, while gene expression profiles were derived from RNA sequencing of CD138+ cells. Results Cluster analysis of drug response profiles segregated the samples into four MM specific groups (Figure). Group I patients (n=12) were highly sensitive to many drugs, including several signal transduction inhibitors such as those targeting PI3K-AKT, MAPK and IGF pathways, as well as HSP90 and BCL2 inhibitors plus epigenetic/chromatin modifiers such as BET and HDAC inhibitors. Group II (n=15) showed a more modest response profile and were moderately sensitive to signal transduction inhibitors and epigenetic modifiers. Group III (n=9) were largely insensitive to most drugs in the panel except for BCL2 and proteasome inhibitors, while group IV (n=3) were resistant to all drugs except BCL2 inhibitors. Many samples were selectively sensitive to navitoclax (55%), dual PI3K/mTOR inhibitors (45%) and aminopeptidase inhibitors (20%), which had little effect on healthy control or MM CD138- cells. Only 33% of the samples responded to glucocorticoids. The majority of samples including healthy BM controls were sensitive to proteasome and CDK inhibitors, suggesting low selective cytotoxicity. However, drug sensitivity profiles of healthy control and CD138- cell populations were distinct from MM CD138+ samples indicating that observed CD138+ drug responses were specific for malignant plasma cells. In addition, we observed that drugs with overlapping target profiles tended to cluster together, indicating sample responses were similar to related drugs. Diagnostic and relapse samples were spread across the different response groups. Samples with mutations to genes involved in PI3K and NF-κB signaling tended to cluster in group I, while most samples with t(4;14) fell in Group II. Samples with RAS mutations were present in all response groups and no correlation with MEK inhibitor sensitivity was observed. 17p deletion samples were also found in all response groups, however, those with additional TP53 mutation tended to have increased drug sensitivity. Summary Our results indicate that PI3K/mTOR, MAPK, IGF1R, NF-κB and cell survival (e.g. BCL2, BCLXL) signaling are important pathways mediating MM ex vivo drug response. This matched with genomic and transcriptomic data, which identified alterations of genes involved in these pathways. Although additional work is needed to correlate ex vivo drug sensitivity with in vivo treatment response, our initial results suggest the possibility that MM patients could be subjected to stratified treatment based on combined ex vivo drug testing and molecular profiling. In addition, these results highlight the multiple signaling pathways active in MM and emphasize the need for improved combination strategies for treatment. Figure: Subgrouping of MM patient samples (I-IV) based on selective drug response profiles. H/D/R denotes healthy, diagnostic and relapse, respectively. Figure:. Subgrouping of MM patient samples (I-IV) based on selective drug response profiles. H/D/R denotes healthy, diagnostic and relapse, respectively. Disclosures Silvennoinen: Research Funding of Finland Government, Research Funding from Janssen-cilag, research funding from Celgene: Research Funding; Janssen-Cilag, Sanofi, Celgene: Honoraria. Wennerberg:Pfizer: Research Funding. Kallioniemi:Medisapiens: Consultancy, Membership on an entity's Board of Directors or advisory committees. Porkka:Bristol-Myers Squibb: Honoraria, Research Funding; Novartis: Honoraria, Research Funding. Heckman:Celgene: Research Funding.

Description: Introduction: Constitutive hyperactivation of the STAT3 and STAT5B transcription factors is often observed in cancer. Lately, activating STAT3 mutations have been identified in hematological malignancies including large granular lymphocytic (LGL) leukemia (prevalence 40%), aplastic anemia (7%) and CD30+ T-cell lymphoma (17%). Furthermore, recent studies highlight the importance of STAT5B mutations in the pathogenesis and prognosis of T-cell malignancies such as T-cell prolymphocytic leukemia (36%), T-cell acute lymphoblastic leukemia (8%) and hepatosplenic T-cell lymphoma (33%). While STAT3 and STAT5B mutations lead to constitutive STAT3/STAT5B signaling, several other known gene mutations and mechanisms may also cause JAK/STAT-pathway activation. These findings indicate that inhibiting the JAK/STAT pathway with targeted drugs could be used as a treatment option. Here, we aimed to identify drugs that inhibit STAT3 or STAT5B function and determine if mutant STAT3/5B and wild type STAT3/5B cells respond differently to the tested drugs. In addition, we wished to ascertain if STAT3 inhibition is sufficient to induce apoptosis in patient derived LGL cells with constitutively active STAT3. Methods: High-throughput drug sensitivity testing was performed with a compound collection containing over 300 approved and investigational oncology drugs including many kinase inhibitors (such as those targeting JAK, SRC, VEGFR, mTOR, MEK, and CHK) and small molecule STAT3 inhibitors (Stattic, LLL12, Sta-21). All drugs were tested in 5-8 different concentrations over a 10,000-fold concentration range. Mutant STAT3 (Y640F) and mutant STAT5B (N642H) transformed Ba/F3 cells as well as HEK293 cells containing a STAT5 (pGL4.52[luc2P/STAT5 RE/Hygro]) or STAT3 specific luciferase reporter gene element (HEK-SIE) were used in the screens. In addition, drug sensitivities of five LGL leukemia patient samples were also assessed. Primary patient cells and the Ba/F3 cells were incubated in 384-well plates for three days with the drugs after which cell viability was measured with CellTiter-Glo. STAT3/5B induced luciferase activity in the HEK cells was analyzed after the cells were incubated for 6 or 24 hours with the drugs using the ONE-Glo luciferase assay system. Results: A significant decrease in luciferase activity was detected in STAT3 mutant Y640F, STAT5B mutant N642H and wild type STAT3 transfected HEK-SIE cells in the presence of PI3K/mTOR inhibitors such as PF-04691502 and INK128. In addition, PI3K/mTOR inhibitors significantly decreased the viability of mutant STAT3 and STAT5B transformed Ba/F3 cells compared to wild type cells. Interestingly, JAK inhibitors (e.g. ruxolitinib, gandotinib) did not inhibit mutant STAT3 activity in the HEK-SIE cells, whereas IL6-induced wild type STAT3 activity was completely blocked. A BET family inhibitor (JQ1+) and glucocorticoids (e.g. dexamethasone, methylprednisolone) showed specific and strong cytotoxicity to mutant STAT3 and STAT5B transformed Ba/F3 cells. Although JQ1+ inhibited luciferase activity of STAT5B N642H cells, no effect on the luciferase activity of STAT3 Y640F transfected HEK cells was detected, suggesting that JQ1+ may have a direct effect on mutant STAT5B function while the effect on mutant STAT3 transformed cells may be indirect. Cells from LGL leukemia patients showed high sensitivity against glucocorticoids, the histone deacetylase inhibitor quisinostat, JQ1+ and PF-04691502 when compared to healthy CD8+ T-cells. However, no increase in apoptosis was observed with JAK or other mTOR inhibitors. Conclusions: Our results suggest that JAK inhibitors lack efficacy in STAT3 mutated diseases. However, our ex vivo and in vitro drug screens highlight some other promising agents including PF-04691502 (PI3K/mTOR inhibitor) and JQ1+ (BET family inhibitor) that inhibited mutant STAT5B and STAT3 activity in the cell line models and were effective against primary LGL patient cells. Additional experiments are ongoing to determine how these drugs function to block STAT signaling and induce cell death. As the STAT3 and STAT5 pathways are activated in many other cancer types as well, the results may be applicable to a variety of different malignancies. Figure 1 Figure 1. Table 1. Drug sensitivity scores of the different cell line models and patient samples. DSS value range 0-50. (0 = no drug response with any conc., 50 = maximal drug response with every conc.) Disclosures Mustjoki: Bristol-Myers Squibb: Honoraria, Research Funding; Novartis: Honoraria, Research Funding.

Description: Key Points Somatic mutations were discovered for the first time in the SH2 domain of the STAT5b gene in LGL leukemia. The mutations are activating and lead to increased phosphorylation and transcriptional activity of STAT5b.