Rassenti, Laura Z.

Abstract Abstract 1232 Poster Board I-254 B-Chronic Lymphocytic Leukemia (CLL) shows a strong familial risk and also co-aggregates with other indolent lymphomas. Monoclonal B-cell lymphocytosis (MBL) is an asymptomatic hematologic condition characterized by small B-cell clones, most with a surface phenotype similar to that of chronic lymphocytic leukemia (CLL). These clones are detectable at low absolute lymphocyte cell (ALC) numbers in otherwise healthy individuals using sensitive 6 or 8 color flow cytometry analysis and can be a precursor to CLL. In the general population, MBL increases with age with a prevalence of 5% in individuals over age 65. In contrast, the rate of MBL is 13-18% in first degree relatives of CLL patients in high risk families. In the largest study to date, we report the characteristics of MBL among 430 first-degree relatives of CLL patients (with no associated lymphoproliferative diseases [LPD]) in 132 high risk families. Patients and Methods Individuals studied came from “high risk” families (defined as those with two or more confirmed cases of CLL) that are participating in the Genetic Epidemiology of CLL Consortium. Multi-parameter flow cytometry analysis with a detection sensitivity of 0.02% was performed on either fresh or cryopreserved PBMC and MBL was classified according to previously published criteria (Marti et al, Br. J Haematol 130:325, 2005). Both the ALC and the absolute B cell count (B-ALC) were calculated for each individual. Survival analysis was used to compute the probability of developing MBL with age using the life table method. Results The overall rate of MBL was 17% (73/430) among first–degree relatives but the probability for developing MBL by age 90 was 51%. Males had a slightly higher (but non-significant) risk for MBL than females. MBL patients had significantly higher ALC and B-ALC than did those with normal immunophenotype. The mean ALC was 2.5×109/L among MBL patients, and 20% had an ALC greater than 3 ×109/L. The B-ALC count averaged 0.51 ×109/L. Ninety percent of the MBL cases had a CLL-like phenotype (CD20dim, CD5+, CD23+). Conclusions MBL is found at a very high rate in families selected for having 2 or more patients with CLL suggesting that MBL reflects inherited predisposition to CLL. Although most of the MBL cases in these families had low cell counts and thus have a presumed low likelihood of progressing to CLL or other LPD, a higher proportion of them had lymphocytosis compared to those with normal immunophenotype. We hypothesize that if MBL is an early step in the process of development of CLL, then germ line genes are likely to be acting early in carcinogenesis with more “hits” required before CLL develops. By looking for genes associated with MBL or CLL in families or in the population, we could substantially increase our power to identify germ line genes predisposing to CLL. Disclosures Kay: Biogenc-Idec, Celgene, Genentech, genmab: Membership on an entity's Board of Directors or advisory committees; Genentech, Celgene, Hospira, Polyphenon Pharma, Sanofi-Aventis: Research Funding.

Abstract Abstract 2443 Background: There is strong and consistent evidence that a genetic component contributes to the etiology of chronic lymphocytic leukemia (CLL). A recent genome-wide association (GWA) study of CLL identified genetic variants located on chromosomes 2q13, 2q37.1, 6p25, 11q24, 15q23, and 19q13 that increased the risk of CLL within a European population. We replicated 5 of these 6 loci in an independent sample of CLL cases and controls from the United States. We now investigate whether these loci also influences MBL, a reported precursor condition of CLL. In addition, a follow-up analysis of the initial GWA study identified four more CLL-susceptibility loci on 2q37.3, 8q24.21, 15q21.3, and 16q24.1. Herein, we also evaluate the association of these four loci with risk of CLL. Methods: Peripheral blood samples were obtained from three ongoing studies: the Genetic Epidemiology CLL (GEC) Consortium, the Mayo Clinic non-Hodgkin lymphoma (NHL)/ CLL study, and the Mayo Clinic Biobank. We implemented rigorous genotyping quality-control measures, and successfully genotyped a total of 407 CLL patients, 965 controls, and 60 MBLs from these studies. Within each locus, the previously reported single nucleotide variants (SNPs) or variants in high linkage disequilibrium (LD) with the previously reported SNPs were evaluated with risk of MBL or CLL. Tests for association was done using the Cochran-Armitage trend test, and unconditional logistic regression was used to estimate odds ratios (ORs) and corresponding 95% confidence intervals (CIs) for CLL or MBL risk Results: In our evaluation of the six initially reported CLL-susceptibility loci (2q13, 2q37.1, 6p25, 11q24, 15q23, and 19q13) with MBL risk, we found three of the six had suggestive associations (p-value < 0.20) with ORs comparable to and in the same direction as those observed from CLL risk. Our strongest finding was with rs13397985 at locus 2q37.1 (OR= 1.56; 95% CI: 1.05, 2.31; p-trend = 0.041), followed by rs17483466 at locus 2q13 (OR= 1.49; 95% CI: 0.99, 2.24; p-trend = 0.074). As expected given our previously reported findings with CLL risk, the association between rs11083846 on chromosome 19q13 and MBL risk was not significant (p-trend = 0.70). Of the four recently reported CLL-susceptibility loci SNPs located on 2q37.3, 8q24.21, 15q21.3, and 16q24.1, we found all to be associated with CLL risk but one. Specifically, the strongest association was seen for locus 8q24.21 (best tagged SNP rs1021955; OR = 1.37; 95% CI: 1.10, 1.70; p-trend = 0.005), followed by locus 16q24.1 (best tagged SNP rs305065; OR= 0.77; 95% CI: 0.61, 0.97; p-trend = 0.024). However, we found no associations for locus 15q21.3 for the previously reported SNP nor for any SNPs in LD with the previously reported SNP. Conclusions: Our MBL results provide additional robust genetic evidence that MBL is a precursor to CLL and that it shares similar underlying genetic predisposition. Also our results confirm three of the four recently reported CLL-susceptibility loci and further support the role of a genetic basis in the etiology of CLL. More research is needed to elucidate the potential manner in which these genetic loci function in CLL or MBL. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 2441 Background: CLL patients with mutated IGHVs (M-CLL) have clinical outcomes better than patients with unmutated BCRs (U-CLL). It is believed that this difference stems from the fact that somatic mutations have altered the structure of the BCR such that it no longer binds stimulatory (auto)antigens and therefore cannot deliver trophic signals to the leukemic cells. We sought to determine if this belief could be corroborated by analyzing the types (silent vs. conservative vs. non-conservative) and locations (FRs v CDRs) of mutations that occurred in IGHVs of M-CLL clones and then comparing the time to first therapy (TTFT) in patients with different IGHV features. Methods: Using IGMT software, we analyzed the IGHV sequences of 1560 cases and characterized their mutations in several respects: first, if IGHV mutations altered amino acid structure (silent vs. replacement); second, if mutations occurred in CDRs (antigen binding domains) or FRs (scaffolds of the BCR); third, if mutations were conservative or non-conservative (as determined by charge, hydropathy, size). TTFT for patients was correlated with various combinations of the above parameters. Differences in TTFT were estimated by the method of Kaplan and Meier; differences were assessed using the log rank test. Results: First we compared TTFT in patients who had only silent mutations (n=32) to those with no mutations (n=563) and found a trend toward longer TTFT in patients with silent mutations (3.0 vs. 3.6 yrs, p=0.09). Because the statistical significance of this result was likely influenced by the small numbers of clones with only silent mutations, we expanded the group to include IGHVs with replacement mutations of only the conservative type (n=83) since these would be unlikely to havemajor impact on BCR structure. Comparison of this group with those with no mutations for differences in TTFT was significant (3.0 vs. 3.9 yrs, p=0.0007). Furthermore, comparisons of patients with silent plus conservative mutations in FRs only (n=123) versus patients with no mutations showed a clear difference in TTFT (3.0 vs. 4.1, respectively; (p<0.003). In both of the above instances, non-conservative replacement mutations do not have a significantly greater impact on TTFT than the silent plus conservative combination. Conclusions: It is assumed that M-CLL patients have a clinical course superior to U-CLL patients because IGHV somatic mutations change the structure of the M-CLL BCR such that it no longer binds stimulatory antigens. For this theory to be correct, only replacement mutations and in particular non-conservative replacement mutations that would alter amino structure of the IGHV/D/J rearrangements would have relevance. Silent mutations, which by definition do not change amino acid structure, should be irrelevant in this regard, and conservative mutations, which substitute amino acids with similar chemical properties to the original amino acid, would be unlikely to result in major changes in the BCR. Therefore, the fact that the combination of only silent plus conservative mutations, which probably would not result in changes in BCR structure and therefore antigen binding, is associated with a lengthened TTFT militates against a loss of antigen binding as a reason for improved clinical course. Therefore, we suggest that another explanation for the enhanced survival of M-CLL patients should be sought. One possibility is that the cells that have the capacity to mutate their IGHVs are different in origin and/or biologic properties from those that cannot, and these account for the differences in survival between M- and U-CLL patients. (This work was supported NIH grant PO1-CA81534 of the CLL Research Consortium.) Disclosures: Byrd: Genzyme Corporation: Research Funding. Kipps: GlaxoSmithKline: Research Funding.

Abstract Background: Chronic lymphocytic leukemia (CLL) patients exhibit variable times from diagnosis to the need to initiate therapy. Previous studies have demonstrated that the rates of CLL cell proliferation (birth) and elimination (death plus recirculation to solid tissues) could be estimated using the incorporation of deuterium (2H), taken in the form of heavy water (2H2O), into the DNA of dividing cells. In a cross-sectional study of patients with early and late stage CLL, birth rates above 0.35% per day were associated with more aggressive disease. The NIH-sponsored CLL Research Consortium (CRC) conducted a prospective study on patients with early stage disease to assess the relationship between CLL-cell birth rate and the length of time to initiation of initial therapy. Methods: Eligible patients had untreated early stage (Rai 0, 1 or 2) disease, and were no more than 3 years from diagnosis. Additional clinical parameters included age at diagnosis, gender, mutational status of the immunoglobulin heavy chain variable region (IGHV), 70-kD zeta-associated protein (ZAP-70) (binary split at 20%) and CD38 (binary split at 30%) expression, and FISH cytogenetics by the Dohner hierarchy. Heavy water was administered daily for 6 weeks, and peripheral blood was obtained for kinetic measurements at the end of weeks 3, 6, 9, 12, and 16. CLL birth rates for blood samples were measured at the same facility (KineMed, Inc, Emeryville, CA) by quantifying deuterium incorporation into the deoxyribose moiety of DNA of tumor cells, using isotope ratio mass spectrometry (IRMS). Subjects were followed for time to administration of therapy; subjects still without treatment at time of analysis were censored, as were subjects who exited the study to receive treatment outside of the established criteria. To assess the impact of birth rate and other factors on time to first therapy, Kaplan-Meier curves were compared using the logrank test for univariable analyses; stepwise Cox proportional hazards regression models were used to build multivariable models. Results: Between July 2005 and February 2009, 119 subjects were enrolled through the CRC. 107 subjects were eligible. For 10 eligible patients, exit from the study prior to completion of protocol-mandated heavy water consumption (n=5) or technical issues (n=5) made it impossible to obtain estimates of birth rate. The 97 eligible subjects for whom birth rates were available form the basis of this report. 60% of subjects were male; median age 57 years (range 40-85). Median birth rate was 0.32%/day (range 0.07-1.31%/day). 43 subjects (44%) had birth rates higher than the previously reported threshold of 0.35%/day. With a median follow up of 4 years, 33 eligible subjects (34%) required treatment. IGHV status (31 subjects unmutated), elevated ZAP-70 (31 subjects), elevated CD38 (21 subjects), and hierarchical categorization of FISH anomalies (4 del(17p), 5 del(11q), 8 tri(12), 50 del(13q)) were examined for association with differential time to first therapy in univariable analyses. All were associated with earlier need for therapy (p <0.001) except for genomic aberrations defined by FISH (p = 0.96). Birth rate > 0.35%/day was also significantly associated with shorter time to first therapy (p < 0.001). In multivariable modeling, only birth rate and IGHV status met the criteria for inclusion (Wald p-value < 0.05). Hazard ratios were 3.51 (p = 0.002) for birth rate above 0.35%/day and 5.0 for unmutated IGHV status (p < 0.001). The association between elevated birth rate and shorter time to first treatment was observed within subjects with both unmutated and mutated IGHV (see Figure 1). Conclusion: An increased birth rate of CLL cells early in the disease course is a strong predictor of an earlier need for treatment in univariable analysis, as are unmutated IGHV, elevated ZAP-70 expression and elevated CD38 expression. However, only unmutated IGHV and an increased birth rate contribute significantly to the multivariable model. This adds a direct measure of B-cell kinetics to the array of biomarkers available to subjects with CLL. Figure 1 Time to treatment by birth rate and IGHV status Figure 1. Time to treatment by birth rate and IGHV status Disclosures Murphy: KineMed, Inc: Equity Ownership. Emson:KineMed, Inc.: Employment, Equity Ownership. Li:KineMed, Inc.: Employment, Equity Ownership. McConnel:KineMed, Inc.: Equity Ownership. Turner:KineMed, Inc.: Employment, Equity Ownership. Busch:KineMed, Inc.: Equity Ownership. Hellerstein:KineMed Inc: Consultancy, Employment, Equity Ownership, Membership on an entity's Board of Directors or advisory committees. Chiorazzi:KineMed, Inc.: Stock options Other.

Abstract Background: Smudge cells are ruptured CLL cells seen on blood smears of CLL patients. For over a century, smudge cells were thought to represent an artifact of slide preparation. We recently showed that smudge cell formation was inversely proportional to leukemic B cell vimentin content. Vimentin is a cytoskeletal protein critical for lymphocyte rigidity and migration; high vimentin content is related to poor prognosis in CLL. Concordantly, in an initial small cohort of patients from a single institution, we found that patients with a low (<30%) percentage of smudge cells on a blood smear have a shorter time to treatment (Mayo Clin Proc.2007;82:449–53). In the current study, we evaluated the impact of smudge cell percentage on prognosis of patients with CLL seen at member institutions of the CLL Research Consortium (CRC). Methods: Archived blood smears from untreated patients with CLL were evaluated for smudge levels. All blood smears were prepared manually in a standard fashion from blood obtained by CRC Tissue Core and stained with WrightGiemsa stain. Smudge cells were defined as broken cells with no intact cytoplasm and a disrupted nuclear membrane. A total of 200 lymphocytes and smudge cells were counted on each slide and the results were expressed as the percent smudge cells. The association between the percentage of smudge cells, prognostic factors (IgVH status, CD38, ZAP-70 and FISH) and time to initial therapy (TTT) was examined. Results: We calculated smudge cell percentage on blood smears obtained prior to treatment for 337 CLL patients. The median smudge cell percentage was 32 (range 2–95%). The percentage of smudge cells was lower in CD38 positive patients (mean 33% vs. 38% in CD38 negative patients, p=0.04). No difference in smudge cell percentage was observed based on IgVH gene mutation status or Zap70 expression. Smudge cell percentage as a continuous variable was associated with prolonged TTT, exponential coefficient 0.98, p=0.04. Using our previously published cutoff of 30% to stratify patients in low and high risk categories, the estimated median time to first therapy of patients with smudge cell percentage ≤30% (n=178) was 7.8 years versus not reached in patients in patients with > 30% (n=159) of smudge cells, p=0.0036, (Figure 1). Ten years from diagnosis, 62% of patients with ≤30% of smudge cells versus 39% of patients with >30% of smudge cells required therapy. In multivariate analysis, the low percentage of smudge cells (≤30%) was an independent predictor of the shortened time to treatment (HR 1.86, 95%CI 1.09–3.16, p=0.02). Conclusion: This multicenter study confirms our initial finding that the percentage of smudge cells on a blood smear is an independent predictor of clinical outcome in patients with CLL. The estimation of smudge cell percentage on routine blood smear provides a simple and inexpensive prognostic test available to nearly all patients with a diagnosis of CLL worldwide. It also allows reanalysis and risk stratification of previously completed trials provided that blood smears have been archived. Further studies of the role of the cytoskeleton in CLL biology are warranted and ongoing in our laboratory. Figure Figure

Abstract In chronic lymphocytic leukemia (CLL), it is recognized that germinal center experience by the cell-of-origin has an influence on the malignant phenotype and impacts clinical behavior. Measurement of the mutational status of the IGHV locus is a commonly used approach that subgroups patients according to this natural history. Other biomarkers correlating with these subgroups, such as ZAP70 expression and ZAP70 DNA methylation, are also used. Recently, we and others have found vast differences in global DNA methylation profiles among CLL cases that classifies distinct subgroups related to their natural history. Here we systematically evaluate various markers of natural history of the cell-of-origin for their ability to discriminate clinical outcomes. IGHV mutational status and ZAP70 expression are currently the most widely used prognostic markers of germinal center experience. These markers are generally concordant, with IGHV-unmutated associating with ZAP70 positivity. To compare their relative prognostic impact, we selected CLL cases from a large sample cohort collected as a prospective study maintained by the CLL Research Consortium that were discordant (IGHV-mutated and ZAP70(+) or IGHV-unmutated and ZAP70(-); n=192). We found a significant difference in time-to-first-treatment (TTFT; P=0.002) and overall survival (OS; P=0.032) between discordant groups, with curves separating according to IGHV as opposed to ZAP70 (Fig.1A). DNA methylation-based classification of CLL according to their natural history has not been compared to IGHV status. This classification strategy subgroups the majority of cases into low-programmed CLL (LP-CLL) or high-programmed (HP-CLL) subgroups that generally correlate with IGHV-unmutated/ZAP70(+) and IGHV-mutated/ZAP70(-), respectively, along with a minority of cases in an intermediate (IP-CLL) subgroup. We used our previously-described DNA methylation assay, based on MassARRAY interrogating 7 genomic loci including ZAP70, to group the 192 samples with discordant IGHV/ZAP70 results with 327 samples analyzed previously from the same cohort. DNA methylation (LP-CLL vs. HP-CLL) outperformed IGHV classification (unmutated vs. mutated) in terms of separating TTFT curves (n=432; LP-CLL vs. HP-CLL median difference=4.8 years, unmutated vs. mutated IGHV median difference=2.9 years) and OS curves (n=491; LP-CLL vs. HP-CLL median difference=12.9 years, unmutated vs. mutated IGHV median difference=6.8 years). To evaluate whether DNA methylation subgroups provide additional prognostic information to IGHV, we compared TTFT and OS within IGHV subtypes. We found that the ability of DNA methylation subgrouping to separate OS curves was maintained in the IGHV-unmutated subtype (P<0.001; Fig.1B) and TTFT in both IGHV-unmutated and mutated subtypes (P<0.001 and P=0.016, respectively). Conversely, IGHV subtypes did not significantly separate outcomes within DNA methylation subgroups (P>0.5 for all). Further investigation of the features of DNA methylation subgroups as defined revealed a strong bias in the frequency of several recurrent adverse prognostic somatic mutations. In particular 17p deletions and EGR2, SF3B1 and XPO1 mutations occurred frequently within the LP-CLL subgroup. We next explored whether the presence of each mutation correlated with clinical outcome within its corresponding DNA methylation subgroup, and found heterogenous results. EGR2 mutations separated TTFT in the LP-CLL subgroup (P=0.0009), but 17p, XPO1 and SF3B1 did not (P>0.3). This suggests that the cell-of-origin should be taken into account to accurately assess the clinical impact of somatic mutations. In summary, we demonstrate that among discordant cases for IGHV/ZAP70, IGHV appears to be a stronger prognostic factor for both TTFT and OS. When then evaluating the relative importance of DNA methylation subgrouping and IGHV status on TTFT and OS, we find that DNA methylation subgrouping is the more powerful prognostic factor. Among the markers of the cell-of-origin, DNA methylation is a strong tool for risk-stratifying patients and should be considered in development of risk prediction models. Figure 1. A) Kaplan-Meier analysis of TTFT of IGHV-mutated and ZAP70(+) versus IGHV-unmutated and ZAP70(-) (P=0.002, median difference=4.4 years). B) Kaplan-Meier analysis of OS in patients with unmutated IGHV separated by DNA methylation subgroup (P<0.001). Figure 1. A) Kaplan-Meier analysis of TTFT of IGHV-mutated and ZAP70(+) versus IGHV-unmutated and ZAP70(-) (P=0.002, median difference=4.4 years). B) Kaplan-Meier analysis of OS in patients with unmutated IGHV separated by DNA methylation subgroup (P<0.001). Disclosures Wierda: Acerta: Research Funding; Novartis: Research Funding; Gilead: Research Funding; Genentech: Research Funding; Abbvie: Research Funding. Brown:Infinity: Consultancy; Janssen: Consultancy; Gilead Sciences: Consultancy; Roche/Genentech: Consultancy; Celgene: Consultancy; Sun BioPharma: Consultancy; Pfizer: Consultancy; Abbvie: Consultancy. Kipps:Roche: Consultancy, Honoraria; Celgene: Consultancy, Honoraria, Research Funding; Gilead: Consultancy, Honoraria, Speakers Bureau; Pharmacyclics, LLC, an AbbVie Company: Consultancy, Honoraria; AbbVie: Consultancy, Honoraria, Research Funding.

Abstract BACKGROUND. The detection of minimal residual disease (MRD) at the level of 0.01%/10-4 or above is a strong independent predictor of reduced progression-free (PFS) and overall survival (OS) in patients with CLL treated with chemoimmunotherapy. Although newer agents such as B-cell receptor pathway inhibitors can result in prolonged survival without achieving complete response, there remains a important role for MRD analysis in assessing therapeutic strategies aimed at disease eradication and cure. This is particularly important in front-line trials for fit patients which now require at least five years of follow-up if PFS is used as an endpoint. The feasibility of using MRD as a surrogate or intermediate endpoint for accelerated approval of new treatments is under review by regulatory agencies but further prospective validation is required. At the same time technology is rapidly evolving and high-throughput sequencing (HTS) technologies now detect MRD at the 0.0001%/10-6 level. It is therefore important to determine the most effective approaches for quantifying MRD that are compatible with previous studies but sufficiently sensitive for current treatments. AIMS. This collaborative project had two objectives. First, to identify the simplest and most flexible flow cytometry panel capable of detecting MRD at the 0.01%/10-4 or lower, that is compatible with published data and independent of instrument/reagent manufacturer. Second, to compare the flow cytometry approach with HTS analysis using the ClonoSEQ assay (Adaptive Biotechnologies, Seattle, WA). METHODS AND RESULTS. A core panel of antibodies for MRD detection was identified by testing an 8-marker combination in 52 samples (27 post-treatment and 25 dilution study) and re-analysing data with serial exclusion of single markers to determine redundancy. A 1-tube core panel of CD19, CD20, CD5, CD43, CD79b, and CD81 was identified and validated against the previously published 2-tube 6-marker and 4-tube 4-marker ERIC panels in 76 samples (19 post-treatment and 57 dilution study). The results showed good concordance (for log-transformed data above the LoQ, linearity=0.977, Pearson correlation co-efficient=0.983, average difference=0.026 log, 95% limit of agreement 0.20log) and a limit of detection of 0.001%/10-5 for the 1-tube core panel. Inter-operator variation was similar to CML MRD monitoring with both experienced operators, or inexperienced cytometrists after ~1 hour of specific education, achieving a 95% limit of agreement less than 0.3log in samples with MRD levels ranging from 0.0001 – 100%. The flow cytometry approach was compared with the ClonoSEQ HTS assay in 109 samples (21 dilution study and 88 post-treatment samples, complete data currently available on 13/88). The assay was applicable to the vast majority CLL patients, often with two clonal markers. There was 94% concordance at the 0.01% (10-4) threshold (15 samples with ≥0.01% CLL by both methods, 14 samples with <0.01% by both methods, 1 sample with 0.03% CLL by HTS and <0.003% CLL by flow cytometry, and 1 sample with 0.005% CLL by HTS and 0.012% by flow cytometry. HTS detected CLL IGH sequences in 22% (7/31) samples with no detectable CLL cells by flow cytometry (i.e. CLL level 0.0001-0.001%, 3/13 patient samples and 4/18 dilution samples). HTS demonstrated a relatively high variability in quantification, as seen in previous studies, but with a clear superiority in the limit of detection and good linearity (linearity=0.905, Pearson correlation co-efficient=0.870, average difference=0.078 log, 95% limit of agreement 1.5 log). CONCLUSIONS. The 1-tube 6-marker flow cytometry core panel is compatible with published studies, manufacturer-independent and flexible, providing directly quantitative results to 0.001%/10-5 without requiring a pre-treatment sample. HTS requires further work to standardise the quantitative analysis and prospective validation but the ClonoSEQ assay is applicable to >95% of CLL patients, does not require viable cells and is extremely sensitive, detecting residual disease in a significant proportion of cases with <0.01% CLL. The results indicate that flow cytometry and HTS are complementary technologies with a combined approach offering the most reliable way of quantifying CLL at the 0.01%/10-4 threshold while allowing higher sensitivity in clinical trials aimed at disease eradication. Disclosures Rawstron: Roche: Honoraria; Biogen Idec: Consultancy; Gilead: Consultancy, Honoraria; Abbvie: Honoraria; BD Biosciences: Intrasure reagent Patents & Royalties; Celgene: Honoraria; GSK: Honoraria. Williamson:Adaptive Biotechnologies: Employment, Equity Ownership. Sanders:Adaptive Biotechnologies: Employment, Equity Ownership. Robins:Adaptive Biotechnologies: Consultancy, Equity Ownership, Patents & Royalties. Hallek:Celgene: Honoraria, Research Funding; Mundipharma: Honoraria, Research Funding; Roche: Honoraria, Research Funding; Janssen: Honoraria, Research Funding; GSK: Honoraria; Gilead: Honoraria. Hillmen:Roche: Honoraria, Research Funding; GSK: Honoraria, Research Funding; Janssen: Honoraria, Research Funding; Pharmacyclics: Honoraria, Research Funding; Gilead: Honoraria, Research Funding.

Abstract Abstract 3865 ZAP-70 expression is a prognostic marker differentiating chronic lymphocytic leukemia (CLL) patients with indolent disease from those rapidly progressing after diagnosis and requiring treatment. We previously demonstrated in four independent clinical cohorts that DNA methylation at a single CpG dinucleotide in the ZAP-70 promoter region impacts on ZAP-70 transcriptional regulation and is prognostic in CLL (Claus et al. J Clin Oncol. 2012;30(20):2483–91). Here, we confirm and extend our previous findings by investigating the significance of ZAP-70 methylation relative to established prognostic biomarkers in CLL: IGHV mutation status or expression of ZAP-70 protein or CD38. DNA methylation analysis of the ZAP-70 core promoter region was conducted in mononuclear cells of 295 untreated CLL patients, previously reported by the CRC (Rassenti et al. Blood 2008;112:1923–30) using MALDI-TOF mass spectrometry (MassARRAY, Sequenom). The 295 patients had a median age of 55 years (range 26–82), 69% were male, and 46% were Rai stage 0, 47% stage I/II, and 7% stage III/IV at diagnosis. ZAP-70 protein and CD38 expression were measured by four-color flow cytometry and IGHV mutational status was assessed, as previously described. By hierarchical clustering, the previously identified single CpG dinucleotide (“CpG unit 1”, position +223 relative to the transcription start) separated patients with high methylation levels across the entire region from those with lower methylation levels at this specific site. With a median follow-up of 3.9 years in 50 patients who had not started treatment and 5.6 years among those still alive (n=196), patients with lower methylation levels at CpG unit 1 had shortened time to first treatment (TTT) and overall survival (OS) (both p<0.0001), confirming the prognostic relevance of this site. In our previous study, an optimal cut-off for CpG unit 1 methylation across four data sets was between 10% and 20%. In this cohort, the optimal cut-off across the two endpoints was between 15% and 25%. A natural gap in CpG unit 1 methylation levels occurred between 20% and 25%, adding potential biological significance, as lower values may represent patient cell samples in which both alleles are unmethylated. Applying a 20% cut-off, TTT and OS were significantly different for those with lower and higher methylation levels (both p<0.0001). The 5-year TTT estimates were 22% (95 % CI: 0.16–0.28) and 64% (95% CI: 0.52–0.74), respectively, for patients with low versus high methylation levels; 5-year OS estimates were 80% (95% CI: 0.73–0.85) and 90% (95% CI: 0.81–0.95). Low ZAP-70 methylation was significantly associated with CLL-cell expression of ZAP-70, CD38, or unmutated IGHV status (each p<0.0001). In a model of TTT, expression of ZAP-70 modified the strong effect of ZAP-70 methylation. In patients with CLL cells lacking ZAP-70 expression, high methylation was protective (hazard ratio [HR]=0.37, 95% CI: 0.23–0.58). In contrast, methylation had no significant effect in ZAP-70+ CLL (HR=1.47, 95% CI: 0.81–2.68); TTT was short regardless of methylation levels (Figure 1). IGHV mutation status and expression of CD38 did not provide significant additional prognostic information. The effect modification described for TTT was not observed for OS (p=0.57). Here, ZAP-70 methylation was the strongest risk factor (hazard ratio=0.32, 95% CI: 0.19–0.54; p<0.0001), and expression of ZAP-70, CD38 or unmutated IGHV did not significantly improve the ability to explain differences in OS. Hence, the prognostic value of ZAP-70 CpG unit 1 methylation was confirmed in a large patient cohort with extensive follow-up. We defined a clinically applicable threshold for ZAP-70 methylation assessment and determined its relative value to current prognostic markers in CLL. ZAP-70 methylation predicted longer TTT among patients with CLL cells lacking ZAP-70 expression, but not among those with ZAP-70+ CLL cells, although this could reflect sampling bias due to the small number of patients (n=16) in our cohort with both high methylation and expression of ZAP-70. Lastly, methylation was the strongest risk factor for OS, and neither IGHV mutation status nor expression of ZAP-70 or CD38 significantly improved the ability to predict prognosis. In conclusion, ZAP-70 DNA methylation represents a clinically valuable biomarker for predicting TTT and OS in previously untreated patients with CLL. Disclosures: Gribben: Gilead: Honoraria; Mundipharma: Honoraria; GSK: Honoraria; Pharmacyclics: Honoraria; Roche: Honoraria; Celgene: Honoraria. Kay:Celgene: Research Funding.

Abstract Abstract 1614 Poster Board I-640 Background Cytogenetic abnormalities in B-cell CLL are important prognostic indicators for predicting disease progression and treatment response. Until recently, only interphase cytogenetics has been readily applicable to clinical use in CLL due to the inability of traditional mitogens to promote effective metaphase cytogenetic identification. Recently, cytokine or CpG oligonucleotide stimulation has been utilized to promote efficient metaphase analysis. One concern with this approach is the actual clonality and reproducibility of such clones when examined by different cytogenetic laboratories. The CLL Research Consortium (CRC) conducted a prospective trial to test (1) whether CpG would enhance detection of abnormal clones and (2) whether the CRC cytogenetic laboratories would achieve similar results when testing the same patient samples. Methods Initially, one laboratory compared stimulation of CLL cells using CpG versus pokeweed mitogen (PWM) + 12-O-tetradecanoyl-phorbol-13-acetate (TPA) in the same samples. To test the reproducibility of CpG stimulation, fresh blood samples from 1 normal control and 12 CLL patients were collected at one site and distributed (blinded to phenotype and prior cytogenetic results) to 5 CRC cytogenetic laboratories where they were cultured for 3 days using CpG stimulation, harvested, banded and analyzed utilizing standard laboratory procedures. When possible, each laboratory analyzed 20 metaphases per case. Interphase FISH analysis for CLL (including probes for chromosomes 6q, 11q, 12, 13q, 17p, and IGH/CCND1) was also done on each sample. Results In the PWM+TPA versus CpG comparison, more cases were found to have clonal abnormalities with CpG stimulation (147 of 229, 64%) than with PWM+TPA (110 of 229, 48%; P=0.0005). All clonal abnormalities detected in the PWM+TPA cultures were also observed in the CpG cultures. In the 5-laboratory CpG comparison, the results were compiled and compared to the concurrent FISH results. The CpG karyotypes were concordant with the FISH results. The control case had a normal CpG-stimulated karyotype in all 5 laboratories. Three cases exhibited a complex karyotype; one of these also exhibited 11q-, one had 17p-, and one had neither. The cytogenetic descriptions varied modestly among labs, but all identified the complex clones. Every case including the control exhibited some nonclonal abnormal cells, but no pattern was detected among the 5 labs. When an abnormal clone was found by only 1 or 2 labs, it was present in only a very small number of cells. A small 13q- clone was observed in 2 cases in 1 and 2 labs (3 of 20 and 5 of 50 cells, respectively), and was confirmed by interphase FISH. A single lab found 2 of 30 cells with a t(12;15) in one case. Conclusions Abnormal clones in CLL are more readily detected with CpG stimulation than with traditional B-cell mitogens. The clonal abnormalities revealed by CpG stimulation are consistent with the “gold standard” interphase FISH results, and are reproducible among different cytogenetic laboratories. We have no evidence that CpG oligonucleotides create new clonal B-cell populations in vitro during 3 days of cell culture. CpG stimulation in vitro reveals clonal cytogenetic abnormalities and complexity in CLL that should be evaluated as potential independent prognostic parameters. Disclosures Kay: Biogenc-Idec, Celgene, Genentech, genmab: Membership on an entity's Board of Directors or advisory committees; Genentech, Celgene, Hospira, Polyphenon Pharma, Sanofi-Aventis: Research Funding.

AbstractAlterations in global DNA methylation patterns are a major hallmark of cancer and represent attractive biomarkers for personalized risk stratification. Chronic lymphocytic leukemia (CLL) risk stratification studies typically focus on time to first treatment (TTFT), time to progression (TTP) after treatment, and overall survival (OS). Whereas TTFT risk stratification remains similar over time, TTP and OS have changed dramatically with the introduction of targeted therapies, such as the Bruton tyrosine kinase inhibitor ibrutinib. We have shown that genome-wide DNA methylation patterns in CLL are strongly associated with phenotypic differentiation and patient outcomes. Here, we developed a novel assay, termed methylation-iPLEX (Me-iPLEX), for high-throughput quantification of targeted panels of single cytosine guanine dinucleotides from multiple independent loci. Me-iPLEX was used to classify CLL samples into 1 of 3 known epigenetic subtypes (epitypes). We examined the impact of epitype in 1286 CLL patients from 4 independent cohorts representing a comprehensive view of CLL disease course and therapies. We found that epitype significantly predicted TTFT and OS among newly diagnosed CLL patients. Additionally, epitype predicted TTP and OS with 2 common CLL therapies: chemoimmunotherapy and ibrutinib. Epitype retained significance after stratifying by biologically related biomarkers, immunoglobulin heavy chain mutational status, and ZAP70 expression, as well as other common prognostic markers. Furthermore, among several biological traits enriched between epitypes, we found highly biased immunogenetic features, including IGLV3-21 usage in the poorly characterized intermediate-programmed CLL epitype. In summary, Me-iPLEX is an elegant method to assess epigenetic signatures, including robust classification of CLL epitypes that independently stratify patient risk at diagnosis and time of treatment.