Shanafelt, Tait D.

Abstract Background: B-Chronic Lymphocytic Leukemia (CLL) is predominantly characterized as a clonal B-cell disorder where a minor fraction of CLL B cells proliferate, but the majority of these cells are non-cycling. Importantly, CLL B cells also exhibit apoptosis resistance that results from a complex set of redundant mechanisms. This latter feature presents a significant problem in relation to disease progression and developing effective therapies in CLL. Previously, we and others have reported that leukemic CLL cells spontaneously secrete pro-angiogenic vascular endothelial growth factor (VEGF) that can significantly enhance CLL leukemic B cell apoptosis resistance via unknown mechanisms. In addition, the molecular mechanism of the constitutive upregulation of VEGF in CLL B cells is unknown. Methods and Results: Here, we report that CLL B cells express a high level of HIF-1α protein under normoxic conditions as compared to that of normal B cells in Western blot analysis using a specific antibody. Accumulation of HIF-1α in CLL B cells resulted in upregulation of its downstream target VEGF. To comprehend this abnormal overexpression of HIF-1α, we have examined the status of the von Hippel-Lindau gene product (pVHL) that is responsible for HIF-1α degradation by Western blot analysis and found it to be either absent or notably low level in CLL B cells as compared to that of normal B cells. Furthermore, we have shown by in vitro reporter gene assays that miR-92a, a microRNA known to be overexpressed in CLL B cells, can target the VHL 3′-untranslated region (UTR) and is likely a reason for the depressed levels of pVHL in CLL B cells. To validate our hypothesis, we transiently transfected the human embryonic kidney cell line (293T) with miR-92a and observed a dose-dependent repression of pVHL level. Post-transcriptional regulation of the VHL gene was further confirmed when we observed a subtle but definite increase of the pVHL levels monitored by immunoblot following nucleofection of the antisense oligo targeted to miR-92a into primary CLL B cells. To examine whether HIF-1α is functionally active in CLL B cells, we performed co-immunoprecipitation experiments and found that HIF-1α forms an active complex by physical association with the transcriptional co-activator p300 and phospho-STAT3 in CLL B cells. The latter molecule is known to be constitutively expressed in CLL B cells but its function has been unknown. Subsequently, we have found via chromatin immunoprecipitation analyses in CLL B cells that HIF-1α and STAT3 are bound to the VEGF promoter and also recruit RNA polymerase II, further substantiating that this complex likely functions in the activation of VEGF transcription. Summary: Overall, our data show that CLL B cells express constitutively elevated HIF-1α levels in normoxia due, at least in part, to miR-92a regulated diminished pVHL expression. This is a unique previously undescribed mechanism for enhanced HIF-1α levels in human malignancy. We believe that Increased VEGF expression is associated with the combined effects of HIF-1α, p300 and constitutively active STAT3 as a functional complex and is an explanation for the VEGF-based autocrine pathway found in CLL B cells. These findings now provide additional strategy points for interrupting the VEGF autocrine pathway in CLL B cells.

Abstract Figure Figure Background: An isolated 13q deletion in patients with chronic lymphocytic leukemia (CLL) is the single most common cytogenetic abnormality and is a favorable prognostic parameter (Dohner et al., NEJM3432:1910, 2000). Although 13q- can occur as either a heterozygous (13q-x1) or homozygous (13q-x2) deletion, it is unknown whether these two genetic types differ in their impact on clinical outcome. We previously postulated that 13q-x2 may be a more aggressive anomaly than 13q-x1 (Dewald et al., BJH121:287, 2003) and have undertaken a long-term clinical follow-up of patients with an isolated FISH detectable 13q- to test this hypothesis. Methods: We identified patients diagnosed with CLL between October 1992 to May 2008 and who had 13q deletion as their sole FISH abnormality. FISH was performed within 2 years of diagnosis and prior to treatment in all patients. Patients were sorted into three groups: 13q-x1 only, 13q-x2 only, or mosaic with both 13q-x1 and 13q-x2 cells. We assessed differences in percentage of abnormal cells, demographic characteristics, and clinical outcome among the three groups. Results: We identified 259 patients with isolated 13q deletion. Age ranged from 38 to 90 years (median=61) and included 99 women and 160 men. With respect to Rai stage, 142 patients (60%) had low risk [Rai 0], 81 (34%) intermediate risk [Rai I-II], and 13 (6%) high risk [Rai III-IV] disease. After a median follow-up of 1.9 years, 40 patients (15%) have been treated and 17 (7%) have died. Median TFS and OS for all patients were 6.9 and 9.3 years, respectively. The mean percentage of abnormal nuclei in the three groups was 44%, 53%, and 47% for 13q-x1 only, mosaic 13q-x1/13q-x2, or 13q-x2 only, respectively (p=0.22). No significant difference in age or gender was observed based on whether 13q- occurred as a heterozygous or homozygous deletion, or was mosaic. Likewise, no significant difference in survival (p=0.313, see Table) or time to treatment (p=0.53) was found among the three groups. Five year OS rates for the 13q- x1 only, 13q- x2 only, and mosaic 13q- x1/13q- x2 are 85%, 95%, and 83%, respectively Median time to treatment for 13q-x1 was not reached, and for the mosaic13q-x1/13q-x2 and 13q-x2 groups median time to treatment was 6.9 and 4.9 years, respectively. Conclusions: The presence of deletion in one versus two 13q- chromosomes in CLL patients appears to have no significant impact on treatment free or overall survival. This observation supports the grouping of CLL patients with isolated 13q- of one or both alleles into a single risk category as originally proposed in the risk categorization by Dohner et al. Taking the present findings together with those of Dohner et al and Dewald et al, we propose a hierarchical risk model of FISH abnormalities in CLL as 17p- à 11q- à 6q- à +12 à normal à 13q-, from most to least aggressive. The 13q- group would now include 13q-x1, 13q-x2, and the mosaic subgroups.

Abstract BACKGROUND: The diagnosis of CLL among asymptomatic patients without lymphadenopathy or cytopenias has historically been based on documenting a characteristic lymphocyte clone and the presence of lymphocytosis. Currently, there are no data regarding what lymphocyte type (absolute lymphocyte count [ALC] or B-cell count) or threshold should be used for this diagnosis. We analyzed the relationship of these lymphocyte counts to clinical outcome in a cohort of 459 patients to determine whether the CLL diagnosis: should be based on ALC or B-cell count what lymphocyte threshold should be used for diagnosis whether any lymphocyte counts has independent prognostic value after accounting for molecular prognostic markers. METHODS: We used the Mayo Clinic CLL database to identify all Rai stage 0 patients diagnosed with CLL between 1/1/00 and 12/31/07 who were evaluated and had flow cytometry at Mayo within 12 months of diagnosis. All patients had lymphocytosis (ALC ≥5 x 109/L) and fulfilled the 1996 criteria for CLL. Raw data from the peripheral blood flow analysis was used in conjunction with blood counts to determine ALC and absolute B-cell count at diagnosis. Estimates of survival were calculated using the Kaplan-Meier method. RESULTS: To assess whether B-cell count or ALC more strongly related to clinical outcome, we evaluated the relationship of these variables with treatment free survival (TFS) and overall survival (OS; n=459). When treated as continuous variables (i.e. measuring risk of each 1.0 x 109/L increase in cell count), both ALC and B-cell count were related to TFS (hazard ratio [HR] ALC=1.02; p<0.0001; HR B-cell count=1.02; p<0.0001) and OS (HR ALC=1.02; p=0.04; HR B-cell count=1.02; p=0.02). Because B-cell count and ALC related to TFS and OS as a continuous variables, we next evaluated what threshold (nearest 1.0 x109/L) at diagnosis best related to an individual’s risk of requiring chemotherapeutic treatment and/or dying of CLL using the HR (log rank statistic) and c-statistic [c=1 indicates perfect discrimination between poor survivors and good survivors; c=0.5 equivalent to chance]. The B-cell threshold that best predicted OS was 11 x 109/L (HR=2.36, p=0.01; c=0.60). This threshold also predicted TFS (HR=3.02; p<0.0001; c=0.64). With respect to ALC, a threshold of 12 x 109/L was able to predict TFS (HR=2.28, p=0.003; c=0.62) but not OS (HR=1.62, p=0.12; c=0.56). The B-cell threshold used in the current diagnostic criteria (5 x 109/L) was able to predict TFS (HR=3.24, p<0.0001; c=0.63), but not OS (HR=1.64, p=0.13; c=0.55) Finally, we evaluated the ability of B-cell count to predict TFS independent of IGHV mutation status, ZAP-70 status, CD38 status, and FISH. Since not all patients had all prognostic tests performed, the predictive value of B-cell count (< or ≥11 x 109) independent of the other prognostic variable was assessed for each variable independently. B-cell count retained prognostic value independent of IGHV mutation status, ZAP-70 status, CD38 status, and FISH (all p≤0.001) [Table]. CONCLUSIONS: Although B-cell count and ALC have similar ability to predict TFS and OS as continuous variables, B-cell count may be a better predictor of TFS and OS when a defined lymphocyte threshold is used. These findings provide evidence to justify the recent proposal by Hallek et al (Blood 111:5446) to base the diagnosis of CLL on B-cell count rather than ALC. The results also provide justification for retaining the size of the B-cell count as part of the diagnostic criteria even in the era of molecular/biologic prognostic markers, but imply a threshold of 11 x 109 is the B-cell count that best predicts patient’s TFS and risk of death. Differentiating between CLL and MBL based on a patient’s likelihood of developing clinical symptoms and dying of disease could minimize unnecessary psychologic distress caused by labeling asymptomatic individuals at low risk for adverse clinical consequences as having leukemia. TABLE: TREATMENT FREE SURVIVAL HR p-value CD38 + 3.20 0.0002 B-cell ≥11 3.10 0.0001 ZAP + 8.33 < 0.0001 B-cell ≥11 6.86 < 0.0001 Unmutated 5.26 < 0.0001 B-cell ≥11 3.06 0.001 FISH (17p- or 11q-) 3.97 0.01 B-cell ≥11 3.67 < 0.0001

Abstract Background: The biology, pathology and clinical course of CD5+ chronic B cell lymphoproliferative disorders (CD5+ B-CLPD) excluding chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL) and mantle cell lymphoma (MCL) are poorly defined. These patients could either have a leukemic variant of a known B cell malignancy or a novel disease entity. Objective: To define the clinical features and pathology of CD5+ B-CLPD, compare these features to CLL, and test the hypothesis that at least some of these patients have a unique B cell malignancy. Methods: The study was conducted with Institutional Review Board permission. We studied 231 patients evaluated in the Division of Hematology at Mayo Clinic Rochester between 1996 and 2008 who had CD5+ monoclonal B cell population in the peripheral blood and/or bone marrow, and in whom CLL and MCL was not diagnosed after flow cytometry, histology and interphase fluorescence in situ hybridization (FISH) examinations. Data were abstracted from patient records and from the Mayo Clinic CLL database. Statistical analysis: To compare differences in characteristics between CD5+ and CLL patients, χ2 statistics were used for qualitative variables (gender, treatment, vital status) and t-tests were used for quantitative variables (age). We performed survival and time to treatment analyses with results displayed using Kaplan-Meier curves and p-values calculated using a log-rank test. Results: We analyzed data on 231 CD5+ B-CLPD patients diagnosed between 7/1/1976 and 5/2/2008 and 1572 CLL patients diagnosed in the same time period. Median follow-up from diagnosis was 2.5 years (range 0 to 25.1 years) for the CD5+ patients and 5.2 years (range 0 to 29.9 years) for CLL patients. CD5+ B-CLPD patients had a median age at diagnosis of 68 years (range 30–94 years) with male predominance (61.5%). One hundred and nine of 231 (52.4%) patients received treatment for their CD5+ B-CLPD and 162 (70.1%) patients were alive at the time of last follow-up. As a group, CD5+ B-CLPD patients had a similar male predominance to CLL but were older at diagnosis. CD5+ B-CLPD patients required their first treatment significantly sooner than CLL patients (median 2.0 years vs. 6.8 years, (p< 0.001) and a higher percentage of CD5+ B-CLPD patients required treatment (52.4% vs. 37.5%, p<0.0001). In addition, CD5+ B-CLPD patients had a shorter survival from diagnosis than CLL patients (median 8.4 years vs. 12.5 years, p<0.001). In 60 patients the CD5+ B-CLPD disease was the leukemic phase of a known lymphoma (lymphoplasmacytic lymphoma, n=34; nodal marginal zone lymphoma, n=6; splenic marginal zone lymphoma, n=16; MALT lymphoma, n=4), 46 patients had incomplete data, and 125 patients did not have pathological and flow cytometry features of a known lymphoid malignancy. Next we compared clinical outcome among the 125 patients with unclassifiable CD5+ B-CLPD to the CLL patients. The unclassifiable CD5+B-CLPD patients were older at diagnosis (median 69 vs. 65 years, p=0.0003) but there were no significant differences in time to treatment (median 9.0 years vs. 6.8 years, p=0.93) or the proportion of patients receiving treatment (30% vs. 37.5%, p=0.11). The survival from diagnosis was shorter for unclassifiable CD5+ B-CLPD patients (median 6.3 vs. 12.5 years, p<0.001). Conclusions: In this large series of CD5+ B-CLPD patients we found that a large subset did not meet diagnostic criteria for known B-cell malignancies and could have a novel disease. This will need to be investigated further. We reviewed the pathology of all available archived tissue and intend to examine additional clinical and prognostic information including molecular prognostic markers. Our data provide a clinical rationale for the careful differentiation between patients with CLL and CD5+ B-CLPD.

Abstract Next‐generation sequencing identified about 60 genes recurrently mutated in chronic lymphocytic leukemia (CLL). We examined the additive prognostic value of the total number of recurrently mutated CLL genes (i.e., tumor mutational load [TML]) or the individually mutated genes beyond the CLL international prognostic index (CLL‐IPI) in newly diagnosed CLL and high‐count monoclonal B‐cell lymphocytosis (HC MBL). We sequenced 59 genes among 557 individuals (112 HC MBL/445 CLL) in a multi‐stage design, to estimate hazard ratios (HR) and 95% confidence intervals (CI) for time‐to‐first treatment (TTT), adjusted for CLL‐IPI and sex. TML was associated with shorter TTT in the discovery and validation cohorts, with a combined estimate of continuous HR = 1.27 (CI:1.17‐1.39, P = 2.6 × 10 −8 ; c ‐statistic = 0.76). When stratified by CLL‐IPI, the association of TML with TTT was stronger and validated within low/intermediate risk (combined HR = 1.54, CI:1.37‐1.72, P = 7.0 × 10 −14 ). Overall, 80% of low/intermediate CLL‐IPI cases with two or more mutated genes progressed to require therapy within 5 years, compared to 24% among those without mutations. TML was also associated with shorter TTT in the HC MBL cohort (HR = 1.53, CI:1.12‐2.07, P = .007; c ‐statistic = 0.71). TML is a strong prognostic factor for TTT independent of CLL‐IPI, especially among low/intermediate CLL‐IPI risk, and a better predictor than any single gene. Mutational screening at early stages may improve risk stratification and better predict TTT.

Abstract To study the impact of dose modification and temporary interruption of ibrutinib in routine clinical practice, we conducted a retrospective study of consecutive CLL patients treated with ibrutinib outside the context of a clinical trial at Mayo Clinic, (Rochester, MN) from 11/2013 to 12/2017. Of 209 patients, 131 (74%) had unmutated IGHV , 38 (20%) had TP53 disruption, and 47 (22%) were previously untreated. A total of 87/209 (42%) patients started reduced dose ibrutinib (<420 mg daily; n = 43, physician preference; n = 33, concomitant medications; and n = 11, other). During 281 person‐years of treatment, 91/209 patients had temporary dose interruption (54%, nonhematologic toxicity; 29%, surgical procedures; 10%, hematologic toxicity; and 7%, other). After a median follow‐up of 24 months, the estimated median event‐free survival (EFS) was 36 months, and median overall survival (OS) was not reached. On multivariable analyses, temporary ibrutinib interruption (hazard ratio [HR]: 2.37, P = .006) and TP53 disruption at ibrutinib initiation (HR: 1.81, P = .048) were associated with shorter EFS, whereas only TP53 disruption (HR: 2.38, P = .015) was associated with shorter OS. Initial ibrutinib dose and dose modification during therapy did not appear to impact EFS or OS. These findings illustrate the challenges associated with continuous oral therapy with ibrutinib in patients with CLL.

7522 Background: Ibrutinib therapy is associated with an increased risk (~15%) of AF. An accurate risk assessment model for the development of AF in pts starting ibrutinib is not established, and outcomes after AF are not well described. Methods: An IRB approved retrospective review of CLL pts treated with ibrutinib at Mayo Clinic between Oct 2012 and Nov 2018 was conducted. Results: 299 pts were identified with a total of 565 years of ibrutinib exposure (Table). After a median follow-up of 24 months (range 0–70 months), 51 pts developed treatment-emergent AF (13 [25%] CTCAE 3 or higher). Our study assessed 3 clinical prediction models, the Framingham (Schanbel:Lancet 2009), Italian (Visentin:Blood 2018;132:3118), and Shanafelt (Shanafelt:Leuk Lymp; 2017) risk scores. Based on a lower Akaike information criteria, the Italian score was best able to predict risk of treatment emergent AF (2-year risk of AF with score 0 6%; 1-2 8%; 3-4 26%; 5+ 47%). Thirty (61%) pts were treated with medical therapy for AF (27 rate control; 2 rhythm control; 1 both). 16 (31%) pts underwent interventional therapy (3 AV node ablation; 11 cardioversion; and 2 pacemaker). Twelve (23%) pts temporarily held ibrutinib and resumed their original dose, 22 (43%) pts continued reduced dose ibrutinib and 11 (22%) continued their initial ibrutinib dose. Six (12%) pts permanently discontinued ibrutinib. Of 51 pts with treatment-emergent AF, 41 (80%) had a CHA2DS2-VASc score of ≥2 (41% received anticoagulation alone; 12% received antiplatelet therapy alone; 10% received both). No pt developed a thrombotic stroke after treatment-emergent AF. Two major bleeds, (1 GI and 1 intracranial) occurred, one in a pt on concomitant antiplatelet and anticoagulation therapy, and one in a pt receiving neither. The development of AF was associated with shorter event-free survival (HR 2.5, 95%CI 1.5-4.2, p<.001) and shorter overall survival (HR 3.5, 95%CI 2.0-6.3, p<.001). Conclusions: The Italian score was the best predictor of AF development in this cohort of ibrutinib treated pts. Although the vast majority of pts who develop AF after ibrutinib are able to continue therapy, the occurrence of AF appears to be associated with shorter EFS and OS. [Table: see text]

7015 Background: Besides clinical staging, a number of biomarkers predicting OS in CLL have been identified. The multiplicity of markers, limited information on their independent value, and a lack of understanding of how to interpret discordant markers are major barriers to use in routine clinical practice. We developed an integrated prognostic index using the database of the German CLL Study Group (GCLLSG), which was subsequently validated in a cohort of untreated CLL patients (pts) from the Mayo Clinic. Methods: The analysis was based on a dataset collected between 1997 and 2006 in 3 GCLLSG phase III trials. The external validation was performed on a series of newly diagnosed CLL pts managed at Mayo Clinic. Results: The GCLLSG dataset (1,948 physically fit pts at early and advanced stage; median age: 60 yr (range 30-81); median observation time 63.4 mo) was used as a training dataset. 7 parameters were identified as independent predictors for OS: sex, age, ECOG status, del 17p, del 11q, IGHV mutation status, thymidine kinase and β2-microglobulin. By using a weighted grading a prognostic index was derived separating four different pts groups: low risk (score 0 - 2), intermediate risk (score 3-5), high risk (score 6-10) and very high risk (score 11-14) with significant different OS rates (95.2%, 86.9%, 67.7% and 18.7% OS after 5 yr for the low, intermediate, high and very high risk group respectively (p<0.001). This prognostic index was validated in a cohort of 676 newly diagnosed, untreated pts from the Mayo Clinic (median age 61.5 yr (range 32 - 89); median observation time 47.0 mo). The 4 risk groups were reproduced with 98.3%, 95.4%, 75.4% and 10.8% OS after 5 yr. The prognostic index predicts OS independent of Rai/Binet stage and provides accurate estimations regarding time to first treatment (TTF). C-statistic is 0.75. Conclusions: Using a multi-step process including external validation, we developed a comprehensive prognostic index combining clinical, serum, and molecular information into a single risk score for pts with untreated CLL. The prognostic index provides more accurate prediction of both TTF and OS. To our knowledge it is the first prognostic model in CLL to reach the C-statistic threshold (c > 0.70) necessary to have utility at the level of the individual.

6571 Background: Chronic lymphocytic leukemia (CLL) is associated with an increased risk of developing second cancers. However, it is unknown whether CLL alters the natural history of these cancers once they occur. Methods: All patients with breast (n=583,838), colon (n=412,932), prostate (n=632,922), lung (n=489,290), kidney (n=95,902), pancreas (n=82,121) and ovarian (n=61,958) cancer reported to the Surveillance, Epidemiology, and End Results (SEER) Program from 1990 to 2007 were identified. Overall survival (OS; death due to any cause) and cancer-specific survival (death due to cancer site of interest) were examined, comparing patients with or without pre-existing CLL. Age- and sex-adjusted hazard ratios (HRs) were calculated. Results: OS for patients with pre-existing CLL was inferior for patients with breast (HR=1.61; p<0.001), colon (HR=1.65; p<0.001), kidney (HR=1.41; p<0.001), prostate (HR=1.75; p<0.001), and lung (HR=1.22; p<0.001) cancer after adjusting for age and sex. After excluding CLL-related deaths, OS remained shorter among patients with breast (p<0.001), colon (p<0.001), kidney (p=0.03), prostate (p<0.001), and lung (p<0.001) cancer. Cancer-specific survival was inferior for patients with breast (HR=1.29; p=0.03), colon (HR=1.75; p<0.001), and lung cancer (HR=1.17; p<0.001) who had pre-existing CLL after adjusting for age and sex. Conclusions: Several common cancers, including breast, colon, and lung, have inferior overall and cancer-specific survival when there is coexistent CLL. [Table: see text]

6533 Background: Although it can be rewarding, caring for patients with cancer is also demanding and stressful. Oncologists work long hours, supervise administration of highly toxic therapy, and continually observe death and suffering. While isolated studies have explored burnout in national samples of U.S. oncologists (last in 2003), little is known about the personal and professional characteristics related to burnout and career satisfaction. Methods: Between October 2012 - January 2013, ASCO conducted a national surveyof ~3000 U.S. oncologists evaluating burnout and career satisfaction. The survey sample was constructed from the member registry to include an equal number of men/women and an equal proportion of individuals from all career stages (<10 yrs, 10-19 yrs, >20 yrs). Validated tools were used to assess burnout and well-being. Results: As of January 2012, 1041 (35%) oncologists had returned surveys (median age 52 years; 50% women). Among responders, 355 (34%) were in academic practice (AP) and 447 (43%) in private practice (PP) with the remainder in other settings (e.g. military, veterans, industry). Oncologists worked an average of 51 hours/week and saw a mean of 51 outpatients/week. Although average hours worked for oncologists in PP and academic practice were similar (AP=54.6/wk; PP=54.2/wk), profound differences in other practice characteristics were observed with respect to sub-specialization (AP=81%; PP=17%), time devoted to clinical care, patients seen per week (AP=37; PP=74), time devoted to education, and method of compensation. Relative to 2011, 23% of oncologists reported a >10% decline in take home pay in 2012 (>10% decline AP=8%; PP=34%). Overall, 454 (45%) of oncologists had at least one symptom of burnout (e.g. emotional exhaustion, depersonalization) on the Maslach Burnout Inventory (AP=47%; PP=51%; p=0.19). While a majority were satisfied with their career (83%) and specialty (80%) choice, both measures of career satisfaction were lower for PP relative to AP (all p<0.005). Conclusions: Burnout is common among oncologists. Detailed multivariate analyses to identify personal and professional characteristics associated with burnout and career satisfaction are underway and will be presented.