Search

There is a clinical need to identify novel treatments for relapsed/refractory DLBCL. Cancer cells engage in novel associations with stromal and immune cells in the TME that provide crucial contributions to disease progression, immune evasion and therapeutic response. However, these hallmark capabilities have been understudied in DLBCL. Given tumor cell genetic complexity, targeting the TME has become a compelling therapeutic strategy. Immune checkpoint blockade therapy (ICB) (e.g. anti-PD-1), which can activate anti-tumor immunity, has provided a new weapon against cancer and serves as an illustrative example of therapeutically re-educating the TME. Clinical results indicate that only a fraction of DLBCL patients currently respond to ICB. Understanding ill-defined TME-driven immune suppression should help optimise ICB and identify novel therapeutic opportunities. Gene expression studies of DLBCL have identified molecular signatures present in both GCB and ABC subtypes related to the TME that correlated with outcome. The prognostically favorable stromal-1 signature reflects reprogrammed stromal cells, extracellular matrix (ECM) and an active immune response. The less favorable stromal-2 signature indicates elevated angiogenesis and blood vessel density. CAFs promote ECM remodelling and angiogenesis in solid cancers. We hypothesized that CAFs play an important role in the pathogenesis of DLBCL including the regulation of subverted host anti-tumor immunity. To assess whether DLBCL tumor cells induce a CAF phenotype in previously healthy stromal cells, we established a co-culture system with subsequent imaging of conditioned cells. Primary human lymphatic fibroblasts (HLFs) were co-cultured for 5 days in direct contact with a panel of GCB (SU-DHL4, SU-DHL6, DOHH2) and ABC (OCI-LY10, RIVA, U2932) DLBCL cell lines or healthy control B-cells. Quantitative analysis revealed a strong induction of CAF molecular marker expression including FAPα and α-SMA in all DLBCL-educated stromal cells compared to healthy B-cell exposed fibroblasts (P In conclusion, our results establish the ability of DLBCL tumor cells to reprogram HLFs into CAFs that acquire functional capabilities to modulate the TME. Notably, activated CAFs show a compensatory inhibitory response by up-regulating PD-L1 expression that may represent an important TME-driven immunosuppressive mechanism. We believe this data contributes to the understanding of the biology that underlies stromal signatures in the DLBCL TME, in particular the contribution of CAFs to immune privilege. Disclosures Ramsay: Celgene: Research Funding; MedImmune: Research Funding.

It is now generally believed that proliferation of the neoplastic clone in chronic lymphocytic leukemia (CLL) takes place in lymphoid tissues where interactions involving the B-cell receptor (BCR) and other microenvironmental elements take place. Previous studies using in-vivo labelling with deuterated water have shown that recently proliferated emigrants from lymphoid tissues express low levels of the chemokine receptor CXCR4 and high levels of CD5 (CXCR4loCD5hi). It has been proposed that, following entry into the peripheral blood (PB), these cells become quiescent, re-express CXCR4 and downregulate CD5 allowing re-entry into tissues and further rounds of proliferation. In the present study we used in-vivo labelling with deuterated glucose (6,6-2H2-glucose, D2G) to investigate the proliferation and release of CLL cells into PB. In contrast to deuterated water, this technique allows pulse labelling of a distinct cohort of cells, which can then be tracked over time in-vivo. Labelling studies were performed in 10 patients with previously untreated, non-progressive CLL. Patients underwent 10 hours of labelling with oral 2DG, after which peripheral blood and lymph node compartments were serially sampled. DNA Deuterium enrichment was measured in the entire CLL population and in flow-sorted subsets defined by CXCR4/CD5 and surface IgM (sIgM) expression. Maximum release of labelled cells into PB occurred after a median of 14 days (4-56 days). The disappearance rate was very slow, with labelled cells detectable after 56 days in half of the subjects. In one case we were able to track labelled cells in both lymph node (LN) and PB and demonstrated an increase in the fraction of labelled cells in the LN between days 7 and 28, consistent with re-entry into this compartment. Subpopulations of PB CLL cells, defined by CXCR4 and CD5 expression were studied over time to investigate the dynamic nature of these molecules in circulating cells. As previously reported, maximum rapid incorporation of deuterium was observed in the CXCR4loCD5hi fraction, equivalent to 1.15 ± 0.04 %/d fractional synthesis at 7 days. Conversely, CXCR4hiCD5lo cells remained largely unlabelled throughout the 8-week study, reaching a maximum of only 0.01 ± 0.003 %/d, suggesting that they represent a non-proliferating population, not derived from the CXCR4loCD5hi subset. In contrast, CXCR4/CD5 intermediate cells exhibited delayed and intermediate labelling, peaking at 0.1 ± 0.02 %/d at 28 days. This sequential labelling pattern suggests that they derive from the CXCR4loCD5hi subset. Since both CXCR4 and CD5 expression are modulated by BCR signalling, we went on to sort cells according to sIgM expression and found maximum labelling in the sIgM high subset with little or no deuterium incorporation in the sIgM low fraction at any time. Again, the sIgM intermediate subset showed delayed labelling, suggesting that they are derived from sIgM-high cells. These observations provide further evidence for clonal heterogeneity in CLL and suggest the existence of distinct but interdependent subpopulations. Recently-divided cells are CXCR4loCD5hi, but appear to transition to an intermediate phenotype by down-regulation of CD5 and sIgM and upregulation of CXCR4 over the ensuing weeks, but do not appear to transition to CXCR4hiCD5lo cells over the time-course of the experiment. Our findings have clinical relevance, since these functionally distinct subsets might also differ in their responsiveness to therapeutic agents, such as drugs that block BCR signalling. Figure 1. Deuterium enrichment in CLL subsets over eight week study Figure 1. Deuterium enrichment in CLL subsets over eight week study Disclosures No relevant conflicts of interest to declare.

Chronic lymphocytic leukemia (CLL) cells rapidly undergo apoptosis in vitro, suggesting that the in vivo microenvironment provides crucial antiapoptotic signals. Overexpression of the antiapoptotic proteins Bcl-2 and Mcl-1 is a hallmark of CLL, and their expression is further enhanced in the lymphoid tissues. However, the high levels of Mcl-1 found in peripheral blood samples, coupled with its short half-life, led us to hypothesize that it must be actively maintained in the peripheral circulation. Coculture of CLL cells with human vascular endothelial cells significantly enhanced tumor cell survival, an effect that was not observed with normal B cells. This was associated with elevated levels of the antiapoptotic proteins Bcl-2, Mcl-1, and Bcl-XL and marked increased expression of CD38 and CD49d, both of which are associated with clinically aggressive disease. Because CD38, CD49d, and some Bcl-2 family genes are transcriptional targets for NF-κB, we assessed NF-κB activation following coculture with endothelial cells. DNA binding of the NF-κB subunit Rel A was significantly increased and strongly correlated with changes in transcription of CD38, CD49d, BCL2, MCL1, and BCLXL, effects that were reversed by a peptide inhibitor of Rel A. These effects were not observed following coculture with nonendothelial cell lines. Therefore, CLL cells receive specific survival signals following interaction with endothelial cells mediated through the activation of NF-κB and the induction of downstream target genes. This type of interaction in the peripheral vasculature may explain the constitutive NF-κB activation and the overexpression of Bcl-2 family proteins commonly seen in this disease. Cancer Res; 70(19); 7523–33. ©2010 AACR.

Targeted ultra-deep sequencing of chronic lymphocytic leukemia (CLL) cells has enabled the assessment of subclone development based on mutations in the IGHV-D-J signature sequence in the dominant CLL clone. We have utilized the Roche 454 FLX pyrosequencing system, which can generate long sequencing reads containing both the immunoglobulin variable region (IGHV-D-J) and part of the immunoglobulin μ constant region (IGHM) in a single sequence, to analyze the mutational characteristics of newly evolved subclones to determine if they derive from AID/APOBEC activity. APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide) is a family of cytidine deaminases that includes AID (activation-induced cytidine deaminase). AID is required for somatic hypermutation in germinal center B lymphocytes. CLL cells, like most non-germinal center B lymphocytes, generally do not express AID. However, AID expression in a small fraction of the CLL clone correlates with worse patient outcome. This observation has led to the hypothesis that abnormal AID expression promotes new off-target non-immunoglobulin mutations and DNA deletions and rearrangements leading to the development of more aggressive disease. CLL is not alone in this hypothesis, as AID is involved in the evolution of other leukemias/lymphomas and reportedly in other types of tumors such as breast and gastrointestinal cancers. Large scale cataloguing of somatic mutations by ultra-deep sequencing of a wide array of cancers has revealed an AID/APOBEC mutational signature in many cancers, including CLL (Alexandrov et al. 2013 Nature). Thus, AID/APOBEC family members may be involved in somatic mutations leading to the evolution of aggressive cancers. To test if AID/APOBEC proteins could be mutationally active in CLL, we analyzed the characteristics of new mutations found in IGHV-D-J-M in CLL cells that were activated in vivo after adoptive transfer into alymphoid NOD/Shi-scid,γcnull (NSG) mice. This CLL xenograft model activates a large fraction of CLL cells, which become AID+ and facilitates the detection of new subclones expressing mutated IGHV-D-Js by ultra-deep sequencing. Four unmutated IGHV CLL (U-CLL) and 3 mutated IGHV (>2% compared to germline) CLL (M-CLL) samples were each adoptively transferred into individual NSG mice. After expansion of CLL, the mice were sacrificed and the specific CLL clone IGHV-D-J-M was amplified from xenograft mouse spleen cDNA. Pre-transfer CLL cell cDNA was also amplified to establish a baseline comparison. Ultra-deep sequencing of these samples resulted in 2,318,800 sequence reads, which were subsequently trimmed according to the Roche 454 algorithm and further processed by custom R scripts. The sequence reads were aligned to the dominant CLL clone IGHV-D-J-M sequence to remove insertions, deletions, and poor quality ( AID mutational characteristics in new subclones were assessed using the SHMTool (http://scb.aecom.yu.edu/shmtool) algorithms to calculate mutation frequencies in IGHV-D-J relative to IGHM and at AID mutation hotspots and coldspots. Other APOBEC family members have a different mutation hotspot site, which we analyzed by custom R scripts. All CLL samples showed evidence of AID mutational activity: higher IGHV-D-J versus IGHM mutations in all cases, increased AID hotspot mutation frequency in five cases, and decreased AID coldspot mutation frequency in five cases. Increased APOBEC hotspot mutational activity was seen in four cases. This APOBEC mutational activity is consistent with increased APOBEC3 gene expression and the genomic somatic mutation pattern observed recently in CLL (Rebhandl et al. 2014 Leukemia). Thus, both U-CLL and M-CLL are capable of producing mutations characteristic of AID or APOBEC activity. These data are consistent with the hypothesis that AID/APOBEC may promote new mutations leading to the evolution of aggressive CLL. Disclosures No relevant conflicts of interest to declare.

Despite recent advances in care, central nervous system (CNS) lymphoma remains a disease with very poor long-term outcomes. We reviewed outcomes at our centre following the adoption of autologous stem cell transplant (ASCT) for patients achieving remission. Records for 78 consecutive unselected patients diagnosed with cerebral diffuse large B-cell lymphoma (DLBCL) between 2003-2014 were analyzed. 62 of these patients did not undergo ASCT. The median age for the non-ASCT group was 60 years (range 26-80, 32 males, 30 females). Of these 50 had primary CNS lymphoma (PCNSL) and 12 cases had isolated CNS relapse after systemic lymphoma. 5 patients were HIV positive, and 4 patients had post-transplant lymphoproliferative disease. 30 patients received HD-MTX alone and 3 patients received whole brain radiotherapy (WBRT) alone. 10 patients received HD-MTX with salvage WBRT on relapse and 1 patient with CNS relapse post BEAM autograft for systemic disease received HD-MTX & WBRT. 14 patients received palliative treatment and 2 unknown. The remaining 16 DLBCL patients and an additional 3 with mantle cell (2) or ALK positive anaplastic lymphoma (median age 53, range 27-67, male: female 13:6) who were in remission (VGPR/CR) by MRI with adequate performance status (PS) underwent ASCT after HD-MTX-based induction therapy, conditioned with 400mg/m2 BCNU & 10mg/kg Thiotepa. 4 received WBRT post-ASCT as adjuvant therapy. Of the 19 ASCT patients 16 presented with PCNSL, 2 late relapses confined to the CNS following prior systemic DLBCL, 1 mantle cell showed evidence of low-level disease on bone marrow biopsy and all were HIV negative. Median CD34 cell dose was 4.48 x 106. Median times to neutrophil and platelet engraftment were 12 (range 10-57) and 15.5 (range 7-57) days respectively. No deaths occurred from transplant-related complications, including sepsis. 2 year PFS was 67% and OS was 73%. 4 patients relapsed post-transplant (21%), all within 4 months of transplant; 3 subsequently died. 1 patient developed systemic disease progression at time of ASCT and 1 had evidence of a mixed response on MRI pre-ASCT. No patients in CR pre-ASCT have relapsed and 1/11 PR patients has relapsed. The longest remission is now 64 months without use of WBRT, providing evidence for durable long-term response and possible cure in an unselected group of patients. For the non-ASCT patients, median OS post diagnosis was 2.5 months overall, and 7.5 months, excluding palliative patient. Of those receiving MTX therapy 7 achieved CR, 7 PR and 23 had PD. 3 patients failed to tolerate treatment. Of those who achieved CR/PR reasons for not proceeding to ASCT included inadequate PS or lack of willingness to receive further therapy. Outcomes for those who achieve CR are good, with median PFS & OS for patients achieving CR not reached at a median of 32 months follow-up. 5/7 remain in remission; 1 patient has died. Only 1 of these patients received WBRT. The median PFS & OS for patients achieving PR were 5.5 & 8 months respectively (p= 0.041 & 0.027 vs CR). When comparing the two groups, median OS for patients receiving ASCT was not reached vs 5 months for MTX alone (p Overall we conclude MTX therapy alone induces CR in only a small number of patients and outcomes for those not achieving CR are dismal. For patients who are able to proceed to ASCT long-term remissions can be obtained. New strategies are required to enable patients to achieve CR in order to receive ASCT and thus achieve a durable long-term survival from CNS lymphoma. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

We read with interest the paper by Bonyhadi et al. ([1][1]) published February 15, 2005. We have, however, two concerns. CD3/CD28 stimulation resulted in successful expansion of T cells from B-CLL PBMC samples. The authors state that this was associated with a rapid decrease in numbers of leukemic

Ultra-deep sequencing has revolutionized our ability to acquire large amounts of genetic data. We have applied this technology towards understanding the mutational process in B-cell chronic lymphocytic leukemia (CLL), which may be a key to understanding CLL pathogenesis. Acquisition of new cytogenetic aberrations and gene mutations in the CLL clone is associated with worse patient outcome. CLL is not unique in this aspect, as new somatic mutations and DNA rearrangements are also found during the evolution of other solid and liquid tumors. In many of these, activation-induced deaminase (AID), an enzyme normally expressed in germinal center B lymphocytes to induce IGHV-D-J mutations and isotype class switch recombination, is abnormally expressed. Its mutational activity, acting outside of the Ig loci, is implicated in the evolution to more aggressive disease. In CLL, the detection of leukemic cells expressing AID ex vivo correlates with significantly shorter patient survival. To test if AID mutational activity is functional in CLL cells and therefore could contribute to CLL evolution, we analyzed mutations in IGHV-D-J, the preferred substrate for AID. Because the rate of AID-induced mutation is low and only a small percentage of CLL cells express AID ex vivo, we used ultra-deep sequencing to analyze CLL cells that were activated under conditions that simulate the CLL microenvironment. Specifically, CLL cells were activated (1) in vitro by simulating the provision of T-cell help or (2) in vivo after adoptive transfer into alymphoid recipient mice, which requires the presence of T-cells for CLL cell growth. Each of these conditions induce AID in a large fraction of CLL cells. To analyze IGHV-D-J mutations, the specific CLL clone IGHV was amplified from cDNA obtained on day 0 or from the activated CLL samples using IGHV family-specific and IGHM primers to enable subsequent comparison of IGHV-D-J with IGHM mutation frequencies. Three unmutated IGHV CLL (U-CLL) and 3 mutated IGHV (>2% compared to germline) CLL (M-CLL) samples were sequenced with the Roche 454 FLX system, resulting in a total of 1,367,522 sequence reads. After using the Roche 454 algorithm to trim sequence reads, they were prepared using custom R scripts that separated 5’ IGHV and 5’ IGHM primer sequences, aligned sequences to the CLL clone IGHV-D-J rearrangement, and removed poor quality ( To evaluate AID mutational characteristics in new subclones, SHMTool (http://scb.aecom.yu.edu/shmtool) was employed to calculate mutation frequencies in IGHV-D-J relative to the IGHM constant region, at AID mutation hotspot sites (GYW or WRC), at AID mutation coldspot sites (SYC or GRS), at C/G base pairs, and at error-prone DNA polymerase eta repair hotspot sites (WA or TW). To calculate statistical significance, we utilized a custom R script that used a bootstrap method to account for the large sample sizes provided by ultra-deep sequencing as well as to correct for differences in sequencing sample size. All samples showed an increase in IGHV-D-J versus IGHM mutations after T cell activation. Five of 6 cases showed an increase in AID hostpot mutation frequency. AID coldspot mutation frequency decreased in 3/6 CLL cases. Percent transition mutation at C/G sites was higher than random in 2/6 CLL cases, which correlated with low frequencies of DNA polymerase eta hotspot mutation. In the other 4/6 CLL cases, the lower percent transitions at C/G sites may reflect the contribution of error-prone DNA repair. In summary, we developed a method to analyze ultra-deep IGHV-D-J sequences that revealed AID mutational characteristics in both U-CLL and M-CLL cells after activation with T-cell help in vitro or in vivo. These data are consistent with the hypothesis that AID, perhaps along with error-prone DNA repair, creates new mutations leading to the evolution of aggressive CLL. Disclosures: Rai: Sanofi: Membership on an entity’s Board of Directors or advisory committees; GSK: Membership on an entity’s Board of Directors or advisory committees; Teva: Membership on an entity’s Board of Directors or advisory committees; Genentech: Membership on an entity’s Board of Directors or advisory committees; Celgene: Membership on an entity’s Board of Directors or advisory committees.

Lenalidomide (Revlimid®), a thalidomide analogue, is an orally administered second generation immunomodulator with anti-angiogenic and anti-neoplastic properties. Initial studies treating patients with chronic lymphocytic leukemia (CLL) suggest that lenalidomide can have considerable efficacy and that its mode of action is mainly indirect, affecting non-malignant cells in the microenvironment, in particular T lymphocytes. Because a recently described xenograft model for CLL has highlighted the importance of CLL-derived, autologous T cells in promoting leukemic B-cell engraftment and growth in vivo, we have studied the influence of lenalidomide on the expansion of CLL B- and T-lymphocytes in this model. After an initial 12 day culture of FACS-isolated CLL-derived T cells with or without anti-CD3/CD28 beads plus IL-2 (30 IU/ml), T lymphocytes were transferred into alymphoid NSG mice via the retro-orbital plexus (day 0). On day 7, CLL cells were delivered retro-orbitally. These recipient animals are referred to as “T + PBMC mice”. Mice that did not receive T cells on day 0 but were given CLL PBMCs at day 7, with or without lenalidomide, served as controls (“PBMC only mice”). Recipient mice received lenalidomide (10mg/kg/day) or vehicle control daily by gavage starting at day 0. All mice were sacrificed at day 28 (28 days after T-cell and 21 days after B-cell transfer), and blood, spleen, and bone marrow were collected. On this material, four analyses were performed: [1] level of human CD45+ cell engraftment; [2] numbers and types of CLL-derived T cells; [3] numbers of CLL B cells; and [4] levels of cytokines reflective of Th1 and Th2 immune responses. There was a clear enhancement in human hematopoietic (CD45+) cell engraftment in those mice exposed to lenalidomide. This was most marked for the PBMC only mice (vehicle: 10.64%; lenalidomide: 38.53%), although it was also evident for T + PBMC mice (vehicle: 55.96%; lenalidomide: 69.65%). T-cell phenotyping was carried out, before and after cell culture and also at sacrifice. Prior to culture, CLL samples contained on average ∼96% CD5+CD19+ cells and ∼3% CD5+CD19- cells; for the latter, ∼67% were CD4+ and ∼33% CD8+. After 12-day culture, these percentages remained largely unchanged. However, the numbers and types of T cells recovered from the spleens at sacrifice were quite different after in vivo exposure to lenalidomide. For the PBMC only, the percentages of CD4+ and CD8+ cells in the spleens differed somewhat based on lenalidomide exposure (CD4: Vehicle 86% vs. Lenalidomide 61%; CD8: Vehicle 10% vs. Lenalidomide 28%). However, this change was dramatic for the T + PBMC mice (CD4: Vehicle 64.1% vs. Lenalidomide 28.9%; CD8: Vehicle 34% vs. Lenalidomide 62%). Furthermore, when the CD8+ cells from these animals were subsetted based on antigen-experience and function, it appeared that lenalidomide exposure had led to the outgrowth of a greater number of effector memory (CD45RO+ CD62L-) than central memory (CD45RO+ CD62L+) T-cells. For CLL-derived B cells, the numbers differed, based not only on lenalidomide exposure but also on prior in vitro activation. Specifically, in PBMC only mice, the addition of lenalidomide led to increased numbers of CLL B cells in the spleen (Vehicle: 7.81% vs. Lenalidomide: 14%). Conversely, in the T + PBMC mice, the numbers of B cells decreased (Vehicle: 2.36% vs. Lenalidomide: 0.34%). An analysis of Th1 and Th2-related cytokines in the plasmas of the mice at sacrifice revealed a fall in IL-4, IL-5, and IL-10 and a marked increase in IFNg, consistent with a Th2 to Th1 transition. The above data suggest that administration of lenalidomide permits greater engraftment of human hematopoietic cells in alymphoid mice. Although this enhancement involves all members of the hematopoietic lineage, T cells, in particular CD8+ effector memory T cells, emerge in excess over time. This CD8 expansion is associated with diminished levels of CLL B cells suggesting that the decrease is due to T-cell mediated cytolysis. In contrast, in the absence of prior T-cell activation, CLL T cells appear to support better CLL B-cell growth. These findings suggest that lenalidomide alters B-cell expansion in vivo depending on the activation and differentiation state of the autologous T-cell compartment. They also implicate the generation of cytolytic T cells as one mechanism whereby lenalidomide leads to clinical improvement in CLL. Disclosures: Allen: Celgene Corporation: Honoraria.

Abstract 316 Chronic lymphocytic leukemia (CLL) clones contain activated/proliferative leukemic cells in lymphoid tissues and resting cells in the periphery. Different subsets of CLL cells have distinct proliferation rates. Recently divided “proliferative” cells have a surface membrane phenotype of CXCR4DIMCD5BRIGHT (CXCR4DIM) and contain higher numbers of CD38+ and Ki-67+ cells. Circulating “resting” CLL cells express CXCR4BRIGHTCD5DIM (CXCR4BR) and genetic signatures of older, quiescent cells that need to home to lymphoid tissues or die. CXCR4DIM and CXR4BR subsets are relatively minor (1–10% of total) components of CLL clones, with the major fraction (≥90%) of CLL cells having intermediate levels of CXCR4 and CD5 (CXCR4INT). Based on these differences, we proposed a model of transitioning CXCR4DIM → CXCR4INT → CXCR4BR CLL cells in the blood. Because higher birth rates correlate with more aggressive disease, and transiting back to solid tissues permits clonal survival and re-activation, this model suggests CXCR4DIM and CXCR4BR subsets as therapeutic targets. Aiming to further understand functional differences in CLL subsets in vitro and in vivo, we found that CLL subsets differ in cell size (CXCR4DIM>CXCR4INT>CXCR4BR), in vivo apoptosis and transmigration in vitro (both CXCR4DIM< CXCR4INT< CXCR4BR). Thus, while more CXCR4BR cells undergo apoptosis, CXCR4BR cells can migrate better to tissues to receive survival signals. In vivo functional differences were then studied in a NOD/SCID/γcnull (NSG) mouse model using pre-activated CLL-derived autologous T cells. Primary CLL blood cells from 1 M-CLL and 2 U-CLL patients were sorted for CXCR4BR, CXCR4INT or CXCR4DIM fractions. Each fraction (5×106 cells) was injected into NSG mice with 5×105 CD3/28-activated autologous T cells. At weeks 2–6 post transfer, blood analyses showed more extensive expansion of CLL B and T cells in mice received CXCR4DIM than in those injected with CXCR4BR or CXCR4INT. At weeks 9–12, mice were sacrificed. Although T cells dominated in blood, spleen and bone marrow of all recipients, a larger fraction of CLL B cells existed in CXCR4BR injected mice, suggesting better long-term CLL cell engraftment capacity of this fraction. Because regulation of T cells plays key roles in CLL cell survival/growth in patients and in the NSG adoptive transfer model, we next analyzed the same fractions for their abilities to activate T cells and elicit help for engraftment and growth. Unactivated CD5+ T cells (1–1.5×105) and B-CLL fractions (3–5×106 cells) were sorted from 6 patient samples (3 U-CLL and 3 M-CLL), injected into mice and followed bi-weekly until week 6. In 5 cases, except one with few CXCR4BR and CXCR4DIM cells, CXCR4DIM injected mice had more extensive T cell growth starting from week 2. Mice injected with CXCR4BR from 2 U-CLL cases also showed T cell expansions, but at comparatively lesser levels and at later time points (from week 4–5). At week 6, CLL B cells were found in spleen and bone marrow in mice with activated T cells; the numbers of CLL B cells correlated with T cell numbers. Also, identical CXCR4 levels were found in CLL cells regardless of origination from CXCR4BR or CXCR4DIM. Notably, no human B or T cells were detected in CXCR4INT injected mice. In fact, adding CXCR4INT cells to CXCR4DIM mice suppressed CXCR4DIM induced T cell expansion and cytokine production. Specifically, mice receiving both CXCR4DIM and CXCR4INT cells had diminished T cell expansion and at least 3 fold reduced serum levels of IFNγ and IL5. Overall, our data confirm the need for activated T cells for CLL B cell growth in mice; suggest superior long term CLL B cell engraftment by CXCR4BR cells with activated T cell support, and identify a greater ability of CXCR4DIM cells to activate autologous T cells, although some U-CLL CXCR4BR cells could do so to a lesser degree. Superior activation of T cells by CXCR4DIM B cells may be due to higher numbers of CD23+, CD25+, CD27+, CD29+ and CD44+ cells in CXCR4DIM fraction that facilitate cellular interactions. Finally, unlike CXCR4BR and CXCR4DIM cells, the major fraction in patient blood, CXCR4INT, inhibited T cell activation. These results indicate previously unappreciated levels of intraclonal CLL cell heterogeneity that may have important clinical relevance, allow more precise biologic analyses, and provide a rationale for preferential therapeutic targeting of these fractions. Disclosures: No relevant conflicts of interest to declare.

Abstract 2545 B-cell chronic lymphocytic leukemia (CLL) clones often acquire new mutations and cytogenetic aberrations over time. In other human cancers, including lymphomas and solid tumors, activation-induced deaminase (AID), which normally causes immunoglobulin (Ig) somatic hypermutation (SHM) and isotype class switch recombination (CSR) in germinal center B cells, is expressed and functions abnormally to cause mutations promoting aggressiveness. In CLL, AID mRNA expression in the leukemic cells correlates with increased adverse cytogenetic aberrations and worse clinical outcomes. Furthermore, CLL cells activated by culture with CD32-transfected murine L cells, anti-CD40 and interleukin-4, produce AID protein with associated functions: DNA breaks, Ig CSR, and Ig heavy chain (IGH) variable (IGHV) gene SHM. To evaluate AID-mediated SHM in CLL more accurately, ultra-deep sequencing was performed on CLL clone's IGH cDNA prior to and after in vitro activation in one unmutated CLL (U-CLL) case (CLL1278, 0.0% mutated IGHV3–30) and one mutated CLL (M-CLL) case (CLL1299, 4.9% mutated IGHV3–23). Additionally, to examine activation of CLL IGH mutation in vivo, ultra-deep sequencing was performed on cells from one U-CLL case (CLL1083, 0.0% mutated IGHV4-b) prior to and after adoptive transfer into the NOD/SCID/γcnull (NSG) mouse, a xenograft model of CLL, where upregulation of AID protein occurs in CD5+CD19+ human CLL cells. Libraries were created for ultra-deep sequencing using the 454 FLX system (Roche) by PCR amplification with IGHV family-specific framework1 (Lprimer) and IGH constant region Cμ (IGHM) (Rprimer) primers on cDNA obtained from CLL cells prior to (day 0) or after in vitro culture for 7 (CLL1278) or 14 days (CLL1278; CLL1299) or from NSG spleen CLL cells collected 35 days after transfer (CLL1083). The resulting 461,153 sequence reads were processed to generate separate datasets with fixed sequence block lengths for each primer. The Lprimer sequence blocks included only 5'IGHV sequence, while the Rprimer sequence blocks encompassed 3'IGHV, IGH diversity, and IGH joining genes (IGHVDJ) as well as 5'IGHM sequence. Individual subclone sequences that occurred at least twice were extracted from each of the datasets and the unique de novo subclones not shared between day 0 and activation were analyzed for mutations. All three CLL cases showed increases in 5'IGHV and 3'IGHVDJ subclones with activation. After in vitro activation, for CLL1278, 123,518 total sequence reads produced 68 unique subclones as compared to 33 at day 0; and for CLL1299, 163,358 total sequence reads produced 78 unique subclones as compared to 61 at day 0. Likewise, after in vivo activation in the NSG mouse, for CLL1083, 174,472 total sequence reads produced 91 unique subclones as compared to 56 at day 0. In contrast, all three CLL cases showed decreases in 5'IGHM subclones after activation. After in vitro activation, CLL1278 and CLL1299 decreased from 22 and 20 unique day 0 subclones to 13 and 16 unique subclones. Similarly, CLL1083 showed a decrease from 20 unique day 0 subclones to 11 unique subclones after transfer into the NSG mouse. After normalization for read number and block sequence length, all three CLL cases showed an increase in 5'IGHV mutation with CLL cell activation (fold change relative to 5'IGHM = 3.4, 2.2, and 4.6 for CLL1278, CLL1299, and CLL1083, respectively). This increase in IGHV mutation relative to IGHM following activation is consistent with AID activity. Furthermore, examination of mutation sites in these subclones revealed an increase in mutations in AID hotspot motifs (GYW or WRC) in the 5'IGHV of all three CLL cases with activation (fold change = 2.0, 1.9, and 2.5 for CLL1278, CLL1299, and CLL1083, respectively), which was not observed further downstream in 3'IGHVDJ and 5'IGHM. Thus, by analyzing a very large number of IGH sequences in CLL cells after activation in vitro or in vivo, a pattern of de novo mutations consistent with AID activity is found. Furthermore, since both U-CLL and M-CLL clones exhibited AID activity, these findings indicate that AID-mediated SHM is not limited by CLL IGHV mutation status. Finally, these data support a model of AID-promoted mistargeted mutations, which may lead to adverse cytogenetic aberrations and unfavorable outcomes in CLL. Disclosures: No relevant conflicts of interest to declare.

Abstract 2362 Poster Board II-339 There is growing evidence that interactions in the tumour microenvironment promote the survival, proliferation and drug resistance of primary chronic lymphocytic leukaemia (CLL) cells. Furthermore, progressive CLL is frequently associated with lymphadenopathy, pointing to a crucial role for signals emanating from the tissue microenvironment in the accumulation of malignant B-cells. Proliferation of CLL cells appears to be confined to specialized structures called pseudofollicles, which contain a number of cell types including CLL cells, T-cells, vascular endothelial cells and stromal cells. Using two separate model systems designed to mimic this microenvironment, we characterised the phenotypic changes induced in primary CLL cells and assessed both viability and proliferation when compared with corresponding control cultures. In the first model system we co-cultured CLL cells with mouse embryonic fibroblasts transfected with human CD40L supplemented with soluble IL-4. In the second model system we utilised the microvascular endothelial cell line HMEC-1 with and without the addition of stimulated autologous T-cells. Microarray analysis of this cell line confirmed that it expresses high levels of the CD38 ligand CD31 as well as the cytokines IL-2, IL-4, IL-6 and the chemokines CXCL1 and CXCL2. Both model systems resulted in enhanced CLL cell survival when compared to control cultures (P Disclosures: No relevant conflicts of interest to declare.

Chronic lymphocytic leukemia (CLL) is a highly variable disorder whose outcome can be predicted by a number of clinical and biological markers. The level of expression of CD38 by peripheral blood leukemic cells is one such prognostic biomarker however despite widespread use in clinical practice, its role in the pathogenesis of CLL remains unclear and an area of active research. A relationship between CD38 expression and proliferation of CLL cells both in the blood and tissues has been demonstrated by several groups and we recently reported that expression is higher in tissues that contain proliferation centers such as splenic white pulp and bone marrow compared to red pulp and peripheral blood which do not. Using an in-vitro model designed to mimic some of the cellular interactions that take place in proliferation centers, we showed that contact with activated autologous CD4+ T cells causes upregulation of CD38 and proliferation of the tumor. In addition, staining of CLL lymph node sections demonstrated that the highest CD38 levels are found in perivascular areas and that overall there are more microvessels in lymph nodes from CD38 positive patients. CD38 expression in CLL is thus linked to tumor proliferation and regulated by interactions in the microenvironment that involve activated T cells and the microvasculature. In order to investigate these interactions, we further developed the in-vitro proliferation center model by incorporating the human microvascular endothelial cell line HMEC-1. CLL cells were incubated for 24–48 hours in the absence or presence of HMEC-1 and/or activated autologous T cells prior to staining for CD19 and CD38 and analysis by flow cytometry. In the presence of HMEC-1, activated T cells or both, CD38 expression by CLL cells increased from an unstimulated mean fluorescence intensity of 3693 to 4747 (n=10, p = 0.001), 18,691 (n= 10, p = 0.03) and 20,729 (n=10, p = 0.01) respectively confirming our previous results and demonstrating an additional effect of endothelial cells. During the course of these experiments it was also noted that HMEC-1 cells markedly improved the viability of the tumor cells. After 7 days, CLL cells purified by negative selection and co-cultured with HMEC-1 had a viability of 86.5%, assessed by flow cytometric measurement of CD19, annexin-V and 7AAD staining, compared to 25.5% in control medium (n=15, p It is well recognized that anti-apoptotic proteins such as Bcl-2 are up-regulated in CLL however the mechanism is unclear. Analysis of CLL cells purified by negative selection following co-culture with HMEC-1 revealed a marked increase in the expression of the anti-apoptotic proteins Bcl-2 (in 6/6 patients) and Bcl-XL (5/6) as well as inhibition of PARP cleavage (6/6) compared to CLL cells cultured in control medium. These results demonstrate that factors derived from the microvasculature enhance the viability of CLL cells by up-regulating Bcl-2 family members. Interfering with these interactions, for example using drugs that target angiogenesis, might thus be a useful adjunct to the therapy of CLL.

Langerhans cells (LCs) are antigen presenting cells found in the epidermis. They are thought to originate from bone marrow precursors although studies in mice suggest they have the ability to self renew in situ. Murine models have also shown that recipient LCs are required for the development of acute graft versus host disease (GVHD) following allogeneic hematopoietic stem cell transplantation (allo-HSCT). Engraftment of LCs following human allo-HSCT has previously been assessed using migration from isolated epidermal sheets, which may be inaccurate because of the potential for differential migration of donor and recipient cells. We therefore studied LC engraftment using fluorescence immunophenotyping and simultaneous in-situ hybridisation for X/Y chromosomes in sex mismatched transplants. Skin biopsies were performed at days 28, 56 and 100, 6 months and 1 year post transplant on 8 patients receiving alemtuzumab based RIC allo-HSCT regimens. Ten micron cryosections were taken, LCs labelled with anti-CD1a and X/Y chromosomes detected with the CEP X/Y probe kit. Slides were examined on a Zeiss LSM510 Meta confocal microscope and Z-stack images were collected to cover a volume of 100μm3. Chimerism of purified CD15+ and CD3+ peripheral blood cells was determined in parallel by polymerase chain reaction amplification of short tandem repeat sequences. Results are tabulated below and show that compared to CD15+ myeloid blood cells, donor LC engraftment is markedly delayed (day 28, 56 and 100 paired t-tests give P Post transplant chimerism analysis Cell Type Day 28 Day 56 Day 100 6 Months 1 Year Median % donor chimerism at specific time points following allo-HSCT. LC 18.5 (n=8) 73 (n=7) 90 (n=7) 93.5 (n=6) 97 (n=3) CD3+ 99 (n=8) 80.5 (n=6) 66.5 (n=6) 79.5 (n=4) 100 (n=3) CD15+ 100 (n=8) 100 (n=6) 100 (n=6) 100 (n=4) 100 (n=2)

Immune dysfunction is a hallmark of B-cell chronic lymphocytic leukemia (B-CLL) which occurs through loss of normal cell function as the malignant clone expands, as a result of therapy or because of immunoregulatory properties of the tumor itself. It has previously been shown that B-CLL cells are poor stimulators of the allogeneic mixed lymphocyte reaction (MLR) and we first determined whether this is due to lack of stimulatory activity or active immunosuppression by examining the effect of B-CLL contact and tumor supernatant (TSN) on a 3rd party MLR. Incorporation of B-CLL cells in a 5 day MLR inhibited 3H proliferation by responders in 2/10 cases, whereas TSN inhibited in all 10 cases. Studies in which normal T cells were stimulated by CD3/28 beads for 72 hours in the absence or presence of TSN showed a reduction in cell cycle entry measured by PI and FITC staining with 26+/−4.6% and 13.7+/−4.3% of cells in S +G2M in the absence and presence of TSN respectively (p

High levels of CD38 expression in B-cell chronic lymphocytic leukaemia (B-CLL) confer a poor prognosis. Although its role in B-CLL is unknown, signalling through CD38 has been implicated in cell survival, trafficking and proliferation. Since proliferation in B-CLL is thought to take place within both bone marrow (BM) and secondary lymphoid tissue, we investigated whether CD38 expression might vary in response to stimuli that occur in these tissue compartments. Firstly, we compared the percentage CD38 expression of CD5/19 cells on 35 paired PB and BM aspirate B-CLL samples. The mean CD38% was significantly higher in BM than PB in all samples (27% vs 19%, p=0.009) including samples with a PB CD38 of 7% or more (33% vs 42%, p=0.047), indicating that factors present in the BM up regulate CD38 expression. Next, CD38 expression and cell division of B-CLL peripheral blood mononuclear cells (PBMCs) were examined in an in vitro system aimed at mimicking the proliferation centre microenvironment where leukaemic cells are situated in close proximity to activated T lymphocytes. Positively selected T cells from 15 B-CLL patients were activated overnight with CD3/28 beads and subsequently cultured with autologous B-CLL PBMCs. Both the percentage of CD19+ CD38+ cells (29.9% vs 59.9%, p=0.003) and CD38 mean fluorescence intensity (75.1 vs 830.8, p=0.005) increased over the 6 day culture period. B-CLL cell division was assessed using the dye carboxyfluorescein diacetate succinimidyl ester (CFSE) in the same co-culture system. This showed that co-culture with autologous activated T-cells can result in B-CLL cell division, and is preceded by CD38 up regulation. In addition, significantly more B-CLL cells underwent at least one division from patients with an initial CD38 level of 7% or more, as compared to under 7% (24.6% vs 10.9%, p=0.031). To further investigate the relationship between B-CLL cell proliferation, CD38 expression and the role of T-cells we examined tissue sections known to contain paraimmunoblasts and other proliferating B-CLL cells. Four colour confocal microscopy using CD3, Ki67, CD38 and CD23 to label frozen B-CLL lymph nodes was employed. Large Ki67+ CD23+ cells were present in close proximity to CD3+ T-cells and these large B-CLL cells had higher CD38 expression than the surrounding small B-CLL lymphocytes. These results support the proposal that CD38 expression in B-CLL is dynamic and may reflect exposure to T-cell derived stimuli which contribute to proliferation in the BM or LN microenvironment. A possible explanation for the poorer prognosis of patients with higher CD38 expression may be that their disease has more proliferative potential.