Rassenti, Laura Z.

Abstract The IgVH mutation status is an independent predictor of time to treatment and survival in CLL. However, the biological basis of this difference is unknown. Clinical use of the IgVH mutation status as a prognostic marker is limited due to the technical difficulties of assessing this feature. As such, we probed for differences in protein expression by CLL cells with unmutated (UM) vs mutated (M) IgVH using phage peptide display libraries (PDPL). Methods: PDPL consists of phages engineered to express random peptides (7-mer) on the coat protein. From approximately 2x1011 phages displaying all possible permutations of 7-mer peptide, we selected phages binding to UM cells by incubation with 106 UM cells and elution of bound phages. Then, from numerous phages initially binding to UM cells, we subtracted phages binding to M cells by multiple incubations with M cells (106). Thus, the phages that bound only to UM cells were selected. The procedure was repeated multiple times to increase selection specificity. The peptide sequences displayed by selected phages were evaluated. We identified phage targets on UM cells by immunoprecipitation (IP) of CLL protein with specific phage and by sequencing of precipitated protein. The expression of target protein in CLL cells was confirmed by immunobloting and flow cytometry. Results: Sequencing of phages from the 3rd round of selection (UM v. M) revealed several peptides, with the most predominant being FPSAHFL. IP identified the cellular target for FPSAHFL as vimentin, an intermediate filament protein involved in cell motility and activation. The level of vimentin expressed by CLL cells varied between patient samples (n=15). UM CLL cases were found to express higher levels of vimentin (the mean MFI was 1002; SD 381), than cases of M CLL (mean MFI 547; SD 109), p=0.0128. To evaluate the prognostic significance of vimentin, we assessed vimentin expression by flow cytometry in a separate subgroup of 40 CLL patients. Archived cell samples collected between 6/2000 – 6/2001 were used. The median follow up was 81 months (40–233). Vimentin expression (MFI) as a continuous variable was a risk factor for a shortened time to treatment (TTT) with a proportional hazard ratio of 4.152 (95% CI 1.33–15.220, p=0.0317). Using MFI closest to the median (and rounded to 2 decimal places) as a cutoff, we stratified patients in high (n=21) and low (n=19) vimentin expression groups. The median times to treatment (TTT) in patients with high vs low level expression of vimentin were 2.8 and 12.9 years respectively, p=0.0025 (Fig.). Conclusion: Using novel proteomic technology, we found that CLL cells with unmutated IgVH expressed higher levels of vimentin than CLL cells with mutated IgVH. A high-level of vimentin expression defined by flow cytometry appears to be associated with high-risk disease. Validation of these findings could allow for the vimentin level to be used as a surrogate marker for expression of unmutated IgVH in CLL. Figure Figure

Abstract Background and Significance : Chronic lymphocytic leukemia (CLL) is the most heritable hematologic malignancy; however, no common CLL predisposition genes are known. Monoclonal B lymphocytosis (MBL) is a hematologic syndrome characterized by small accumulations of B lymphocytes in the peripheral blood. MBL has a CLL-like immunophenotype, may progress to overt CLL, and is over represented in CLL families. Therefore, MBL observed in the context of familial CLL may be a marker of inherited risk for development of CLL. Detailed characterization of family-associated MBL may also provide mechanistic insights into the pathogenesis of familial CLL. Our strategy was to detail the biologic characteristics of CLL in family-associated MBL. Methods : Persons with MBL were identified by flow cytometric screening of peripheral blood from unaffected members of CLL kindreds ascertained by Genetic Epidemiology of CLL Consortium (GEC) sites. Flow cytometry was used to determine the surface immunophenotype including CD38 and intracellular ZAP-70. We defined MBL as populations of CD19+, CD5+, CD20lo, CD23+ B cells that comprised at least 2% of the CD19+ peripheral B cell compartment and did not exceed 5.0 × 109 MBL cells/L. RNA and genomic DNA from single MBL cells isolated by flow cytometric sorting were analyzed using PCR to determine immunoglobulin heavy and light chain sequences. MBL cells were sorted in bulk for FISH for loci associated with clinical CLL. Results : Twenty-two out of 190 (12%) unaffected family members were found to have MBL. We observed significant variability in the size of the MBL clone as a percentage of the CD19+ B cell compartment (mean 32%; range 2%–97%). Nonetheless, the absolute size of the MBL clone was small, 15 of 17 individuals had < 200 × 106 MBL cells/L. CD38 expression (defined as CD38 surface expression in ≥30% of MBL cells) was observed in 8 of 18 subjects tested. ZAP-70 (defined as intracellular expression in ≥ 20% of MBL cells) was expressed in 4 of 12 participants. Among 12 subjects tested, 5 MBL expressed both surface IgD and IgM, 3 expressed IgD only, 2 expressed IgM only, and 2 did not express IgD or IgM. Analysis of immunoglobulin heavy and light chains has been completed in 8 individuals. Both immunoglobulin heavy chain variable (IgVH) region mutated (n = 12) and unmutated (n = 4) sequences were observed. Four of 8 individuals had 2 or more unrelated MBL clones (range 2–5), including one individual with both unmutated and mutated clones. Among the 16 MBL clones identified in these 8 subjects, VH3 or VH4 rearrangements were observed in all MBL clones. The most commonly rearranged IgVH genes were 3–07 (3 MBL clones), 3–15 (3), and 4–34 (3). No VH1 family gene rearrangements were observed. In one individual, a VH3–07 MBL clone showed intraclonal diversification suggestive of antigen driven immunoglobulin sequence changes. Twenty productive light chain rearrangements were identified among the 16 MBL clones, with 11 Vλ and 9 Vκ genes used. We observed 6 productive rearrangements of Vλ1–51. MBL cells were bulk sorted for FISH from 9 subjects. Mono or biallelic deletion of 13q14.3 was observed in 5 subjects, the other 4 were normal. Conclusions : Our findings confirm that MBL is commonly observed among the unaffected family members from CLL kindreds. We found that some MBL clones express ZAP-70, CD38 or have unmutated IgVH genes and thus are similar to clinical CLL. The clinical outcome of these MBL clones in relation to our baseline prognostic characterization will be of interest. Small MBL clones are commonly oligoclonal. Importantly, although the immunoglobulin heavy and light chain genes rearranged in these MBL clones are all commonly used in CLL, the absence of VH1 and abundance of Vλ1–51 rearrangements do suggests differences in BCR usage between CLL and MBL. Further investigation of family associated MBL may clarify the genetics and immunobiology of familial CLL.

Abstract Approximately 10–15% of patients with CLL report a family history (FH) of CLL or a related lymphoproliferative disorder. Whether the biologic features of CLL that occurs within families differ from those that occur sporadically is unknown. A recent study addressing this question has reported that mutated IgVH is significantly more common among familial than sporadic CLL, 68% vs 47%, although the distribution of VH genes used was similar (Blood111:5691, 2008). Therefore we compared IgVH mutational status and gene usage, as well as other biologic features, between familial and sporadic CLL derived from patients enrolled on the CLL Research Consortium (CRC) Tissue Core. Of 4533 patients enrolled in the CRC database and tissue bank, 3328 had adequate family history information, and these were subdivided into 4 groups: FH of CLL (n=468), FH of other heme cancer but not CLL (n=421), FH of non-heme cancer (n=1444), and sporadic cases (no FH of any cancer) (n=995). Of those patients reporting a FH of CLL, 63% had at least one affected 1st degree relative, 20% at least one affected 2nd or greater degree relative, and 17% unknown. 68% of those reporting a FH of CLL also had other cancers in their family. Patients reporting a FH of CLL or non-CLL heme cancers were more frequently white (91%), compared to 87% of those with FH non-heme cancer and 84% of sporadic CLL patients (p=0.005). The proportion of CLL cases that were female was similar among groups, at 37.6% of patients with FH CLL, as compared to 32.5% of sporadic patients (p=0.12). The age at diagnosis of CLL also did not differ among the four groups, although a trend toward a difference was observed between the FH CLL patients, whose mean age at diagnosis was 54.9 (range 30, 93), and the sporadic patients, whose mean age at diagnosis was 56.1 (range 30, 84) (p=0.08). No significant difference was observed among groups in the frequency of IgVH mutation (defined as <98% homology to the most similar germline VH gene), which was present in 50.3% (95% CI 44.5–56.2) of patients with FH CLL, 56.3% (95% CI 49.7–62.7) of patients with FH other heme cancer, 51.9% (95% CI 48.5–55.3) of patients with FH non-heme cancer, and 51.3% (95% CI 47.0–55.5) of sporadic patients (p=0.53). The pattern of IgVH family gene usage also did not differ among the groups; the VH3 subgroup was most commonly used, at 45%, 52%, 46% and 47% of patients with FH CLL, FH other heme cancer, FH non-heme cancer and sporadic, respectively. The VH1 subgroup was next most common in all 4 groups, at 28%, 24%, 26% and 25%, respectively, followed by the VH4 subgroup, at 19%, 17%, 21% and 21%. In addition, no significant difference was observed in the distribution of ZAP-70 expression amongst the four family history subgroups (mean %ZAP-70 expression, 22.8% for FH CLL patients and 24.4% for sporadic patients, p=0.31), or in the fraction considered ZAP-70 positive (>20% expression), which was 40% for FH CLL patients and 45% for sporadic patients (p=0.16). The relative expression of CD23 and CD38 did not differ among groups, nor did levels of beta-2-microglobulin or IgG determined at time of CRC enrollment. Time from diagnosis to initial treatment for each of the subgroups is currently being analyzed. These data on patterns of biologic risk factors in this large group of familial CLLs compared to sporadic CLLs suggest that familial and sporadic CLL are biologically similar and therefore likely to have similar pathogenetic mechanisms, including perhaps an underlying genetic susceptibility.

Abstract Fluorescence in situ hybridization (FISH) probes and analysis methods for B-cell Chronic Lymphocytic Leukemia (CLL) vary extensively among cytogenetic laboratories. This is not unexpected, as neither national nor international standards have been established for most FISH studies. Lack of standardization is problematic when data collected at multiple institutions are used for clinical correlative studies. To circumvent such problems, the five participating laboratories in the CLL Research Consortium (CRC) designed and executed a joint CLL FISH validation study. Methods: Initially a survey was sent to assess equipment, methods and experience with FISH for CLL. In a pilot study to compare laboratory performance in scoring patient samples, slides from ten patients were prepared and sent to each participating lab to be hybridized with five probe sets (= 50 hybridizations) and analyzed according to their local protocol. In a second pilot study, slides from two patient samples and identical probe sets were sent to the participating labs where hybridization and analysis were carried out according to their local protocol. Next, technologists and directors from all participating labs attended a workshop where technologists working in pairs scored nuclei together, techniques and scoring criteria were established, and consensus reached on other concerns. In a proficiency test nine months after the workshop, slides from two patient samples (10 hybridizations) were hybridized and scored according to each lab’s protocol and results shared using a common reporting form. Results: Survey results indicated that four labs used the same commercially available CLL FISH panel, and one used a combination of probes from the same vendor plus several home-brew probes. Each lab scored between 100 and 200 nuclei per hybridization site, and each independently set normal cutoff values. The FISH panel included probes to detect 11q, 13q, and 17p deletions, trisomy 12, and IGH gene rearrangement. One lab included probes to detect 6q deletion. In the first pilot study each lab used their hybridization methods, probe sets, and scoring criteria. Differences among labs were observed due to variations in probe strategy, reporting of anomalies, and perhaps most important, scoring criteria. Probe strategy differences resulted in variable reporting of 11q- vs monosomy 11 and 12q duplication vs trisomy 12. Some participants reported 13q-x1 and 13q-x2 as subclones and some reported only 13q-. One lab reported an IGH rearrangement whereas the others scored IGH as normal. In the second pilot study each lab used the same methods and probe sets to facilitate comparison of scoring by the technologists. All labs correctly identified the abnormalities, and there were no false positive results. Minor scoring differences were attributed to variation in scoring criteria or inexperience with an unfamiliar FISH probe strategy. The proficiency test that followed the workshop demonstrated 100% concordance in identification of abnormalities. Inter-lab scoring was much improved compared to the first pilot study. The only exceptions were a 13q- range of 72–90% in one case, and a 17p- range of 38–67% in another case. Conclusion: The pilot studies identified a need to develop common scoring criteria. The subsequent workshop and proficiency test demonstrated that the collaborative effort resulted in more standardized scoring among the CRC laboratories. Our collaborative study emphasizes the need to establish rigorous standards and guidelines for FISH procedures and scoring criteria. Standardization of FISH methods among participating laboratories will enhance the confidence in FISH studies for both clinical applications and cooperative intergroup clinical research.

Abstract We performed immunophenotypic and IgVH mutation analyses on the chronic lymphocytic leukemia (CLL) cells of 307 patients for expression of CD38, ZAP-70, and IgVH mutation status to determine the relative strength of each in predicting the need for early treatment. For this we assessed the time from diagnosis to initial therapy for progressive and/or symptomatic disease. Patients confirmed to have remained untreated were censored for this endpoint. Expression of CD38 was defined as the proportion of CD38-stained CLL cells that had fluorescence intensity above a defined threshold. Similarly, cases were defined as ZAP-70+ when 20% or more of the ZAP-70-stained CLL cells had fluorescence intensity above a defined threshold, as described (NEJM351:893, 2004). CLL cells were classified as expressing mutated IgVH genes if they expressed IgVH with <98% homology with a known germline IgVH gene. The median percentage of cells expressing CD38 in our cohort was 11.5% (range 0.1% to 99.5%). In patients with multiple samples, the CD38 expression level appeared constant over time, with a median S.D. of 0.8%. We defined patient samples that had CD38 expression levels at or below the median as having low expression. There was a significant association between low expression of CD38 and lack of ZAP-70 or expression of a mutated IgVH (p < 0.0001 for each assessment by the Fisher exact test). Median level of CD38 among cases expressing mutated IgVH was 4.0%, compared to 27.4% among cases expressing unmutated IgVH (p < 0.0001 by the Wilcoxon rank sum test). Median level of CD38 among ZAP-70 negative cases was 5.3%, compared to 27.2% among ZAP-70+ cases (p < 0.0001 by the Wilcoxon rank sum test). Patients with low-level expression of CD38 had a median time to initiation of therapy of 8.2 years; whereas, patients with high-level expression of CD38 had a median time to initiation of therapy of 4.0 years (p<0.0001). Although high-level expression of CD38 was associated with a shorter time from diagnosis to initiation of therapy (p < 0.0001 in a Cox proportional hazards regression), the inclusion of IgVH mutation status (p < 0.0001) or expression of ZAP-70 (p < 0.0001) in these models reduced the contribution of CD38 status to a p-value of 0.051 (ZAP-70 model) or 0.052 (mutation status model). Although ZAP-70 expression and mutation status each contributed significantly to the hazard of initiating therapy in a Cox proportional hazards model, high-level expression of CD38 did not add significantly to the model that already addresses ZAP-70 and mutational status, p=0.425. As such, although knowledge of the CD38 expression level can be used as a predictor of the time from diagnosis to the initiation of therapy it provides only marginal value if either the IgVH mutational status or expression of ZAP-70 is known and no value if knowledge of both ZAP-70 expression and IgVH mutational status is available.

Abstract Abstract 288 BACKGROUND. Chronic lymphocytic leukemia (CLL) patients with mutated IGHV genes (M-CLL) have better outcomes than patients with unmutated IGHV genes (U-CLL). It has been proposed that this difference reflects the fact that IGHV mutations alter the structure of the B-cell antigen receptor (BCR) such that it no longer binds stimulatory (auto) antigens and therefore cannot deliver trophic signals to the leukemic cells. For this theory to be correct, only replacement (R) mutations and in particular non-conservative R mutations that would more likely alter amino structure of the IGHV/D/J rearrangements would have relevance. Silent (S) mutations by definition do not change amino acid structure and could not alter antigen binding. We sought to investigate this hypothesis by analyzing the types (S, conservative R, non-conservative R) and distribution of mutations that occur in IGHVs of M-CLL clones and then comparing the time to first therapy (TTFT) in patients with different IGHV features. This analysis expanded an initial study of 1569 CLL cases in the US to include 1858 patients from Europe for a total of 3427 cases. METHODS. Using IGMT software and tools, we analyzed the rearranged IGHV sequences of 3427 cases and characterized their mutations in several respects: first, if IGHV mutations altered amino acid structure (S vs. R); second, if mutations occurred in CDRs (antigen binding domains) or FRs (scaffolds of the BCR); third, if R mutations were conservative or non-conservative as determined by charge, hydropathy, and size. TTFT for patients was examined with various combinations of the above parameters. Differences in TTFT were estimated by the method of Kaplan and Meier and assessed using the log rank test. RESULTS. First, TTFT was compared for 4 groups of patients with the following mutation profiles: no mutations; only S mutations (median 1 per sample); only R mutations (median 1 per sample); and mixed S and R mutations (median 16 per sample). These 4 categories were significantly different (P<0.0001), with the median TTFT being 2.0 yrs for no mutations; 2.6 yrs for R only; 2.8 yrs for S only; and 7.3 yrs for mixed. All comparisons used patients with no mutations (n=1234) as the reference group. We identified statistically longer TTFT (2.8 vs 2.0 yrs; P=0.04) in patients with only S mutations (n=62). Patients with only R mutations (n=218) also had superior outcomes (2.6 yrs vs. 2.0 yrs; P=0.008). There was no statistical significance in TTFT between patients with all S vs. all R mutations (2.8 yrs vs. 2.6 yrs; P=0.71). Because exploring the relevance of BCR structural change on TTFT required us to focus on a small subset of patients with only silent IGHV mutations, our sample size was small. Therefore, we expanded the group to include IGHVs with R mutations limited to the FR region (S + R in FR only; n=241) since these mutations would be unlikely to have a major impact on BCR structure, particularly sparing the binding site. When compared to patients with no mutations, this group demonstrated a significantly longer TTFT (2.8 yrs vs. 2.0 yrs, P=0.0002). This finding was upheld when not permitting any non-conservative R mutations in this group, leaving only patients with S plus conservative R mutations in FRs only (n=124); TTFT in this case was 2.8 yrs vs. 2.0 yrs, respectively (P=0.0006). CONCLUSIONS. Our findings show that patients with only S mutations have better outcomes than patients with no mutations, suggesting that a change in BCR structure that could lead to a loss of antigen binding is an unlikely reason for improved clinical course. This is further supported by the finding that the combination of only S plus conservative R mutations located solely in FRs, which probably would not result in major BCR structural changes and therefore antigen binding, is associated with a lengthened TTFT. Therefore, we suggest that the currently accepted paradigm to explain the enhanced survival of M-CLL patients needs to be re-evaluated. Disclosures: Brown: Calistoga: Consultancy, Research Funding; Celgene: Honoraria, Research Funding; Genzyme: Research Funding; GSK: Research Funding. Kipps:Igenica: Equity Ownership, Membership on an entity's Board of Directors or advisory committees; Celgene: Consultancy, Research Funding; Abbott Industries: Research Funding; Pharmacyclics: Membership on an entity's Board of Directors or advisory committees; Genentech: Research Funding; GSK: Research Funding; Gilead Sciences: Consultancy, Research Funding; Amgen: Research Funding.

Abstract Previous work has suggested a relationship between disease activity and in vivo birth rates of chronic lynphocytic leukemia (CLL) cells, as measured by incorporation of deuterium into DNA during heavy water (2H2O) consumption. By extension, CLL-cell kinetics may be a useful prognostic marker in early stage disease, with an increased birth rate predicting shorter time to disease progression and poor prognosis. We are conducting a multi-center study to evaluate the relevance of proliferation rates to both known prognostic markers and to disease progression during 2.5 years of clinical follow up in previously untreated patients with early stage disease. Patients are given a daily dose of 2H2O for 6 weeks and CLL-cell birth rates are measured. We examined the leukemia cells of 36 patients for known prognostic markers (ZAP 70 expression, CD38 expression, IgVH mutation status) and CLL birth rates. Subjects were Rai Stage 0 (n=15), 1 (n=17), or 2 (n=4); mean age 57 +/− 9 (range 41–78); mean time from diagnosis to enrollment 2.1 +/− 2.2 y (range 0.1–10 y). ZAP-70 expression (20% cutoff) was positive in 44%, CD38 expression (30% cutoff) was positive in 35%, and use of unmutated IgVH genes was found in 45% of the subjects. CLL-cell birth rates ranged from 0.09 to 0.97% per day. Utilizing the previously published cut-off value of 0.35% newly divided cells per day, 33% of subjects had a high proliferation rate. Data on all biomarkers were available for 33 subjects. Nine subjects had 4/4 favorable prognostic indicators and 5 subjects had 4/4 unfavorable prognostic indicators. Agreement between prognostic markers, as assessed by kappa coefficient, was strongest for the ZAP-70 v. IgVH mutational status comparison. Comparisons were also significant for ZAP-70 v. CD38 and IgVH mutational status v. CD38. Birth rate was not significantly correlated with any of the other prognostic markers, suggesting it may be a useful independent predictor of disease progression. Isolation of CLL-cells from the bone marrow at the end of the 6 wk labeling period in 5 subjects revealed matching proliferation rates in the marrow and blood in 3 subjects. In 2 subjects, marrow enrichment was lower, suggesting a more slowly proliferating pool in the marrow, the pathophysiologic significance of which is unclear. In 5 subjects, kinetic analysis revealed a very delayed entry of newly divided cells into the blood compartment, presumably from tissue sites of proliferation. This parameter may reflect pool size of CLL-cells in tissues and correlation with lymph node burden is being evaluated. In summary, analysis of CLL-cell birth rates in vivo in these early stage patients provides unique information about the biology of CLL. Correlations with disease progression during the follow-up period will help determine the utility of this test as a prognostic biomarker.

Abstract Abstract 1785 Background: Chronic lymphocytic leukemia (CLL) is a malignant lymphoproliferative disorder characterized by the progressive clonal expansion of CD5+ B cells. Approximately 8–10% of CLL patients have a first degree relative with CLL, indicating that genetic factors can contribute to disease susceptibility. Genome wide association (GWA) studies have identified 16 “common variants,” single-nucleotide polymorphisms (SNPs) with a population prevalence of >5%, that confer risk for CLL. Among patients with CLL, the clinical course can be highly heterogeneous. Studies have defined disease-characteristics that vary among patients that can be used as markers to help predict clinical outcome. African American (AA) CLL patients have a poorer overall survival than Caucasians. However, there is little information on whether there are distinctive, ethnic-associated disease-characteristics that contribute to differences in clinical outcome. Moreover, studies of the genetics of CLL have involved primarily Caucasian patients. For these reasons, we examined for ethnic-related differences in the distribution and prevalence of CLL-associated SNPs and adverse disease-associated characteristics among AA with CLL relative to those of Caucasian patients with this disease. Methods: The clinical and biologic characteristics of 43 AA CLL patients ascertained at Duke University and the Durham VA Medical Center were tabulated and compared to Caucasian CLL patients at the same centers. Fifteen SNPs were genotyped in this “discovery” cohort of 43 AA CLL patients using TaqMan SNP genotyping assays. As a “validation” cohort, 70 AA CLL patients ascertained by the CLL Research Consortium (CRC) were assayed using the same platform. Three control groups were considered in analysis: (1) previously published Caucasian CLL patients, (2) control AA genotypes reported in the HapMap database and (3) control AA with no history of malignancy from other GWA studies. The 3' untranslated region of CPEB1 was sequenced in AA CLL patients using conventional Sanger sequencing. Results: Among the Duke/Durham VA CLL patients, no significant differences were observed in age, WBC count, platelet count, or hemoglobin at diagnosis; lymphocyte doubling time; Rai stage; ZAP70 expression; or CD38 expression between AA and Caucasian patients. AA had a larger proportion of trisomy 12 and fewer 13q.14 deletions than Caucasians: 40.0% of AA had trisomy 12 and 15.0% had del 13q, compared to 14.6% and 40.7% of Caucasians, respectively. IGHV mutation status also varied between races, with 62.2% of AA having unmutated IGHV, compared to 38.4% of Caucasians. Disparities in clinical outcomes were similar to that reported in the SEER database: overall survival was 8.8 years in AA patients and 17.8 years in Caucasians, and time to first treatment was 2.8 years among AAs and 4.5 years for Caucasians. Risk allele frequencies among 15 genotyped SNPs showed that 14 SNPs had risk allele frequencies that statistically differed between AA patients and Caucasian CLL patients. In 9 of 14 SNPs, the risk allele was less common in AA than in Caucasians but similar to that AA controls. In the other 5 SNPs, the risk allele was more common in AA CLL than Caucasian CLL, and similar to that of AA controls in 4 of the SNPs. A single SNP, rs783540 (CPEB1) was found to have an increased frequency in AA patients with CLL when compared to both Caucasian CLL and control AA. As an exploratory analysis to identify possible pathogenic variants, the 3' UTR of CPEB1 was sequenced as there are no common non-synonymous SNPs in CPEB1. Preliminary review of the sequencing data revealed that a G insertion SNP (rs34816185) that alters a predicted miRNA binding site is in linkage disequilibrium with rs783540 (D'= 0.978). Conclusions: Most SNPs associated with CLL risk are less frequent in AA than in Caucasian CLL patients. However, rs783540 (intron 2, CPEB1) risk allele was more frequent in AA CLL than that previously reported in Caucasian CLL or AA controls. Initial data suggests that rs783540 is linked to a SNP in the 3' UTR of CPEB1, which may have functional significance. Studies to better delineate the functional and clinical implications of these germline variants are underway. Our data provide the first evidence of differential genetic risk as a determinant of CLL biology among different ethnic groups. Disclosures: No relevant conflicts of interest to declare.

Abstract ROR1 is a type-1 tyrosine kinase-like orphan-receptor that ordinarily is expressed during embryogenesis, but that also is found on leukemia cells of patients (pts) with chronic lymphocytic leukemia (CLL). In prior studies we found ROR1 served as a receptor for Wnt5a, which could promote survival/growth of CLL cells. We found Wnt5a in the plasma is significantly higher in pts with CLL (3.2±1.6 ng/ml (mean ±S.D.), N = 36) than in healthy adults (0.14±0.16 ng/ml, N=14, p<0.01) and also may be elaborated by accessory cells in the CLL microenvironment. On the other hand, reducing expression of ROR1 on CLL cells via siRNA adversely affected leukemia-cell survival. We hypothesized that expression of ROR1 in CLL could play a role in the pathogenesis and/or progression of this disease. Early studies failed to define differences in the expression level of ROR1 on CLL cells of different pts in a small cohort. However, upon further analyses, we observed heterogeneity in the expression levels of ROR1 among samples of different pts in larger cohort. We investigated whether the level of ROR1 expression was associated with disease progression in a cohort of 1,568 CLL pts followed by CLL Research Corium (CRC). CLL samples were uniformly examined for the expression of ROR1 using the same preparation of an Alexa-647-conjugated anti-ROR1 mAb (4A5) and a standardized procedure by the CRC Tissue Core. We used 10 CLL cases, and tested ROR1 expression at 3 different time-points for each case. We found ROR1 expression was stable over time in all CLL cases. Then, we randomly segregated pt samples into either of two cohorts, one a training set of 797, and the other a validation set of 773 cases. For the 797 cases in the training dataset, we used the profile-likelihood method in a Cox regression model of treatment-free survival (TFS) from diagnosis to determine the optimal threshold level of ROR1 expressionthat could segregate pts into 2 subgroups. In the training cohort, 294 pts were stratified into a "ROR1-Low" subgroup and 503 pts were assigned to a "ROR1-High" subgroup, defining a threshold absolute mean fluorescence intensity of CLL cells stained for ROR1 of 29. The subgroup of pts in the ROR1- High subgroup had a median TFS of 6.0 years, whereas those pts in the ROR1-Low subgroup had a median TFS of 11.0 years. We applied this threshold to segregate the 773 samples of the validation cohort into ROR1-High and ROR1-Low subgroups. We found that high-level expression of ROR1retained its strong predictive adverse effect on TFS in this validation set. The subgroup of pts in the ROR1- High subgroup had a median TFS of 5.7 years, whereas those pts in the ROR1-Low subgroup had a median TFS of 11.8 years. The immunoglobulin heavy-chain variable region (IGHV) mutation status was known for 593 of the 797 samples in the training set (74%) and 586 of the 773 samples in the validation set (76%). A significantly lower percentage of the ROR1-Low cases (34%, or 69 of 203) used unmutated IGHV than did the ROR1-High cases (54%, or 209 of 390) (p<0.001). Nonetheless, using a multivariate Cox regression model with either a non-delayed entry or a delayed entry, we observed that high-level expression of ROR1 had predictive value for short TFS (non-delayed entry HR 1.4, p<0.05; delayed entry HR 1.9, p<0.001) even when the IGHV mutation status was taken into consideration (non-delayed entry HR 2.8, p<0.001; delayed entry HR 2.7, p<0.001). Taken together, this study demonstrates that there is heterogeneity in the expression of ROR1 on CLL B cells. Furthermore, we find that a high level expression of ROR1 is associated with early disease progression in pts with CLL. Table 1. Number of cases with high versus low ROR1 and mutated versus unmutated IGHV in the training and validation cohort Training Dataset Validation Dataset ROR1 MFI distribution (Median) 0~475 (35.5) 0~424(34.6) ROR1Low with Mutated IGHV 139 132 ROR1Low with Unmutated IGHV 64 84 ROR1High with Mutated IGHV 181 156 ROR1High with Unmutated IGHV 209 214 Total With Known IGHV Mutation Status 593 586 Disclosures Keating: Celgene Corp.: Consultancy; Glaxo-Smith-Kline Inc.: Other: Advisory board. Kay:Celgene Corporation: Membership on an entity's Board of Directors or advisory committees, Research Funding; Pharmacyclics: Research Funding; Tolero Pharma: Research Funding; Genentech: Research Funding; Hospira: Research Funding; Gilead: Membership on an entity's Board of Directors or advisory committees, Research Funding. Byrd:Acerta Pharma BV: Research Funding. Gribben:Celgene: Consultancy, Honoraria; Janssen: Honoraria; Roche/Genentech: Honoraria; Pharmacyclics: Honoraria; Gilead: Honoraria. Kipps:Pharmacyclics Abbvie Celgene Genentech Astra Zeneca Gilead Sciences: Other: Advisor.

Abstract Abstract 1241 Poster Board I-263 Background and Significance Monoclonal B lymphocytosis (MBL) is a hematologic syndrome characterized by accumulations of clonal B lymphocytes in the peripheral blood. Most MBL have a CLL-like immunophenotype, though other, less common phenotypes are observed. Although some MBL, particularly those with MBL cell counts >1.9 × 109 / L, progress to CLL, “low count” MBL (<0.2 × 109 MBL / μL) have little potential to progress to MBL and may have different biologic characteristics from MBL with lymphocytosis or Rai Stage 0 CLL. Our hypothesis was that detailed characterization of family-associated MBL may also provide biologic insights into the pathogenesis of CLL. Methods Individuals with MBL were identified by flow cytometry screening of fresh and cyropreserved peripheral blood from unaffected members of CLL kindreds ascertained by Genetic Epidemiology of CLL Consortium (GEC) sites. We defined MBL as populations of CD5+, CD19+, CD20lo, CD23+ B cells that comprised at least 0.02% of the PBMC and did not exceed 5.0 × 109 MBL cells / L. Flow cytometry was used to determine the surface immunophenotype including prognostic parameters of CD38, intracellular ZAP-70, immunoglobulin isotype, CD49d, and the ratio of CD69:71. mRNA and genomic DNA from single MBL cells isolated by flow cytometric sorting were analyzed using PCR to determine immunoglobulin heavy and light chain sequences (IGVH status). MBL cells were sorted in bulk for FISH determination of genetic loci associated with clinical CLL. Results Fifty-four unaffected family members were found to have MBL, of these 48 (89%) showed a typical CLL immunophenotype (CLL-like MBL). We observed significant variability in the size of the MBL clone as a percentage of the CD19+ B cell compartment (mean 30%, median 16%; range 2% - 97%). CD38 positive (defined as CD38 surface expression in ≥ 30% of MBL cells) was observed in 6 of 38 (16%) subjects tested. ZAP-70 positive (defined as intracellular expression in ≥ 20% of MBL cells) was expressed in 5 of 28 (18%) participants. CD49d positive (defined as surface expression in ≥ 45% of MBL cells) was observed in 4 of 17 (24%) subjects. The ratio of surface expression of CD69:CD71, a measure of cellular activation that also correlates with an unmutated IGVH, was ≥2.0 in 3 of 34 (9%) subjects tested. Among 36 subjects tested, 9 (25%) MBL expressed both surface IgD and IgM (defined as surface expression ≥ 40% for both IgM and IgD), 12 (33%) expressed IgD only, 2 (6%) expressed IgM only, and 13 did not express IgD or IgM (36%). Analysis of IGVH status has been completed in 10 individuals. Both immunoglobulin heavy chain variable (IGVH) region mutated (n = 14) and unmutated (n = 5) sequences were observed. Six of 10 individuals had 2 or more unrelated MBL clones (range 2 - 4), including two individuals with both unmutated and mutated clones. Among the 19 MBL clones identified in these 8 subjects, VH3 or VH4 rearrangements were observed in all MBL clones. The most commonly rearranged IGVH genes were 3-07 (3 MBL clones), 3-15 (3), and 4-34 (3). No VH1 family gene rearrangements were observed. MBL cells were bulk sorted for FISH from 14 subjects. Mono or biallelic deletion of 13q14.3 was observed in 9 subjects, 4 were normal, and one showed trisomy 12. Conclusions Our data affirms that CLL-like MBL are commonly observed among the unaffected family members from CLL kindreds. We found that family associated MBL clones (most of which were small clones) express ZAP-70, CD38, and CD49d at an apparently lower frequency than observed in CLL. Unlike in CLL, the surface immunoglobulin isotype showed co-expression of IgM and IgD in only one third of cases. Moreover, small MBL clones are commonly oligoclonal and predominantly express mutated IGVH genes with an IGVH usage that also appears different from CLL. Taken together, these findings suggest that small MBL clones, though phenotypically similar to CLL, have important biologic differences from CLL that may explain the limited potential of these clones to progress to CLL. The clinical outcome of these MBL clones in relation to our baseline prognostic characterization will be of interest. In addition the further investigation of family associated MBL is being conducted and may clarify the genetics and immunobiology of familial CLL. Disclosures No relevant conflicts of interest to declare.