Background: In chronic lymphocytic leukemia (CLL), CD49d, the alpha chain of the heterodimer CD49d/CD29 (VLA-4), is a strong negative prognosticator and key player of CLL microenvironmental interactions. The adhesive properties of VLA-4 can be rapidly inside-out activated by signals through the B-cell receptor (BCR), thus favoring the capability of the integrin to interact with its ligands. In CLL, beside the canonical antigen (Ag)-dependent mechanism, BCR signaling has been recently demonstrated to occur via an autonomous Ag-independent manner. Aims: To investigate the role of autonomous BCR signaling on constitutive VLA-4 activation state in CLL. Methods: VLA-4 activation/affinity was determined by flow cytometry (FC) using the conformation-sensitive anti-CD29 mAb HUTS21 and/or by “real-time” (FC) measuring the binding of the VLA-4 ligand LDV-FITC as reported (Tissino et al, J Exp Med, 2018) in: i) 1,984 consecutive CLL all with IGHV gene mutations/BCR features available; ii) sequential samples (0, 14, 30, 60 90 days) from CLL patients (n=26) treated in vivo with ibrutinib (IB) in real-world and from a clinical trial (NCT02827617). HUTS21 staining was also performed in the presence of: plasma depletion/replacement, soluble (s) VCAM-1, fibronectin (FN) and blocking anti-CD49d (HP1/2) mAbs. ELISA assays were used to quantify sVCAM-1 in plasma samples (n=122). BCR signaling were investigated by Ca++ influx assay in a 4-OH tamoxifen (4-OHT)-inducible murine TKO cell model (Dühren-von Minden M et al, Nature, 2012). Results: Out of 1,984 CLL, 1,070 (54%) expressed CD49d (cutoff 30%) and, among them, 250/1,070 (23%) were HUTS21+ (cutoff 20%), indicating an activated VLA-4 conformation. HUTS21 staining was: i) impaired by depletion of plasma from whole blood samples, and reconstituted by specific plasma components (sVCAM-1, FN); ii) impaired by pre-incubation with anti-CD49d HP1/2 blocking mAbs before addition of plasma, sVCAM-1 and FN. sVCAM-1 was higher in CD49d+ vs CD49d- CLL (p<0.0001); among CD49d+ cases, sVCAM-1 was lower in HUTS21+ (i.e. VLA-4 activated) cases (p=0.0096), suggesting ligand sequestration by activated surface VLA-4. CLL with mutated IGHV expressed higher levels of activated VLA-4 compared to unmutated IGHV CLL (p=0.001). Higher levels of activated VLA-4 were found in CLL using the IGHV3 and IGHV4 families, compared to cases using the IGHV1 family (p=0.043 and p=0.004). Finally, analysis of BCR stereotypy highlighted higher VLA-4 activation levels in CLL from subset#2 compared to CLL from subset#1 (p=0.02). To validate these data, murine TKO cells, expressing high VLA-4 levels, were transfected with different BCRs derived from 4 CLL with high level of constitutively activated VLA-4 (TKO-high) and 4 CLL with low level of constitutively activated VLA-4 (TKO-low). Compared to TKO-low cells, TKO-high cells showed a higher autonomous Ca++ influx (p=0.03), and consistently higher VLA-4 affinity (p=0.01). IB treatment impaired both BCR autonomous signaling and VLA-4 affinity. Notably, anti-IgM stimulation induced high Ca++ influx and high VLA-4 affinity state in both TKO-high and TKO-low, irrespective of IB treatment. According to TKO data, a decreased constitutive VLA-4 activation was observed in CLL cells collected at pre-treatment and at day 14, 30, 60 and 90 from patients on IB, confirming an IB-dependent impairment of VLA-4 activation via BCR signal. Summary/Conclusion: The presence of a constitutively activated form of VLA-4 is observed in a fraction of CD49d+ CLL, due to a continuous VLA-4 inside-out stimulation derived from autonomous BCR signaling.

Background: Classic Hodgkin lymphoma (cHL) has few pre-treatment prognostic biomarkers. Circulating tumor DNA (ctDNA) is a metrics of tumor volume and inflammation, which are both prognostic in cHL, it improves the accuracy of treatment response assessment, which is an unmet need at the interim timepoint in cHL, and it allows to accurately genotype the disease, which has technical hurdles if done of the tumor biopsy in cHL. Aims: The study aims at addressing the following questions: i) is pre-treatment ctDNA load a metrics that captures both tumor and inflammation extents in a single, measurable and radiation-free test? Can the integration of ctDNA with interim PET (iPET) improve accuracy of the sole iPET in predicting treatment outcome? Can ctDNA identify molecular groups with phenotype- and outcome-associated signatures? Methods: IOSI-EMA003 (NCT03280394) is a prospective, observational, multi-center study. The study recruited adult patients with previously untreated cHL. Blood samples were collected during staging and disease response assessment at the same time of PET scan acquisition. PET scans were centrally and blindly reviewed by a panel of nuclear medicine physicians. ctDNA was genotyped and quantified by phased variant-enhanced-LyV4.0 ctDNA CAPP-seq assay. Interim molecular response (iMR) was defined as lack tumor reporters in cfDNA collected at the interim timepoint (sensitivity 10-4). Results: A total of 215 patients were recruited. Median follow-up was 30 months. The full analysis dataset was divided into training (N=135) and validation (N=80) cohorts by using a random sample procedure. In the training cohort, ctDNA load directly associated with inflammation (B symptoms, elevated ESR), but not with disease burden: total metabolic tumor volume (TMTV), GHSG stage. The optimized threshold of pretreatment ctDNA load to stratify PFS was 850 hGE/mL of plasma (Figure 1A). In multivariable analysis, pretreatment ctDNA levels remained prognostic for PFS when controlling for either GHSG stage, TMTV, B symptoms or ESR. The validation cohort confirmed that patients with higher pretreatment ctDNA levels had inferior PFS (Figure 1B). Among intermediate and advanced stage patients, the PFS of patients with positive iPET was significantly inferior (Figure 1C). iMR was achieved in 54% patients. Patients who did not achieved iMR had a lower PFS. iMR combined with iPET had a statistically significant superior accuracy for the anticipation of progression or relapse compared to iMR or iPET alone (Figure 1D). To identify molecular subgroups within cHL, we focused on fragmentation profiles of cfDNA. cfDNA fragmentome can comprehensively represent both genomic and chromatin characteristics. Pre-treatment cfDNA fragmentation patterns were characterized by a more prominent mononucleosomal fragments abundance in 51% of patients (mono-nucleosomal cluster), whereas 33% of patients had a more prominent shift towards submono-nucleosomal fragment lengths (submono-nucleosomal cluster) (Figure 1E). Patients belonging to the submono-nucleosomal cluster had different clinical and biological nuances, including advanced stage, B-symptoms, elevated ESR, higher TMTV and total lesion glycolysis, higher mutation load, higher circulating immune suppressive cytokines and chemoattractants for monocytes and Th-cells. Accordingly, patients with submono-nucleosomal cluster had a lower PFS. Image:Summary/Conclusion: ctDNA is a validated prognostic biomarker of cHL, can improve the accuracy of interim response assessment and provide the bases for a molecular classification of cHL.

Introduction: Current prognostic models of classic Hodgkin lymphoma (cHL) incorporate pre-treatment clinical and laboratory parameters, but have low discrimination capacity and limited clinical utility. Circulating tumor DNA (ctDNA) is a sensitive cancer biomarker, but its clinical validity in cHL is unknown. Here we tested whether pre-treatment ctDNA qualification and quantification is prognostic in cHL. Methods: The IOSI-EMA-003 (NCT03280394) is a prospective, observational, multicentric, international, non-interventional trial in which liquid biopsy samples were collected from newly diagnosed cHL patients at the following time points: baseline, interim and end of treatment PET/CT assessments. The LyV4.0 ctDNA CAPP-seq assay (sensitivity: 0.1%) was used to qualify and quantify ctDNA. Clinical data quality was assured through remote medical monitoring. Baseline, interim and end of treatment PET/CT are undergoing central review. Results: A total of 135 patients were recruited. Baseline characteristics were those expected in unelected, previously untreated cHL. After a median follow-up of 33 months, 25 patients had progression events, accounting for 3-year progression free survival (PFS) of 80.0%. At baseline, ctDNA was detected in 90% of patients, with an average of 27 somatic mutation reporters per case. A summary of genes affected by nonsynonymous, synonymous, and noncoding somatic mutations is shown in Figure A. Median ctDNA load was 358.5 hGE/mL of plasma (range 0-16572.4), and correlated (p = 0.004) with clinical proxies of the tumor volume (Figure B). None of the genes mutated in ≥5% of patients, including TP53, stratified PFS (Figure C). By recursive partitioning, ctDNA load in baseline samples stratified PFS with an optimized threshold of 1500 hGE/mL of plasma (Figure D). Patients with high pre-treatment levels of ctDNA had significantly inferior rates of 3-year PFS (44.4% vs 86.9% p = 0.000006) than those with low levels (Figure E). High pre-treatment levels of ctDNA occurred: i) in 0% of early, 12.2% of intermediate and 15.9% of advanced cHL; and ii) in 6.0% of 0-2 international prognostic score (IPS) and 27.3% of >3 IPS cHL. Levels of pre-treatment ctDNA marked patients’ outcome, irrespective of whether they presented in intermediate or advanced stage. The combination of high pre-treatment ctDNA and poor IPS risk >3 allowed to sort out upfront poor risk cHL showing a 3-year PFS of only 33.3% (Figure F). In multivariable analysis, the interaction between pre-treatment ctDNA and IPS remained prognostic for PFS (HR 7.3, 95% CI 3.0-17.5, p = 0.00009) when controlling for GHSG risk. PFS prediction metrics (c-index) were 64.9% for ctDNA, 66.2% for IPS and 69.8% for IPS combined to ctDNA. The research was funded by: -Fond’Action, Lausanne, Switzerland; -Translational Research Program, No. 6594-20, The Leukemia & Lymphoma Society, New York Keywords: Diagnostic and Prognostic Biomarkers, Hodgkin lymphoma Conflicts of interests pertinent to the abstract A. Moccia Consultant or advisory role: Roche, Janssen and Takeda D. Rossi Honoraria: AbbVie, AstraZeneca, Janssen Research funding: AbbVie, AstraZeneca, Janssen.

Background: Overall, ten independent retrospective studies addressed the question of the tolerability of ibrutinib in the real-world setting. Discontinuation rates ranged from 5% to 29%. Such heterogeneity may reflect differences in the case mix, follow up and accessibility to next treatments. Dose reduction/transient interruption rates ranged from 22% to 32%. The impact of intolerance on ibrutinib effectiveness according to baseline CLL biology is largely unknown. We assessed whether outcomes of patients with high-risk CLL is affected by ibrutinib discontinuation, interruption, or dose reduction due to intolerance. Methods: The IOSI-EMA-001 and the IOSI-EMA-003 observational prospective studies (NCT02827617; NCT03280394) enrolled patients with CLL treated with ibrutinib according to prescribing indications in Switzerland and Italy. Pre-treatment demographics, disease data, mutation analysis by LyV4.0 CAPP-seq assay, information on treatment discontinuation, interruption and dose reduction, and efficacy outcomes were collected. Results: In stotal, 90 patients were included in the per protocol population. Baseline features include age >65 years in 71% of cases, male gender in 61%, previous treatment in 47%, beta-2-microglobulin >5 mg/L in 40%, lactate dehydrogenase >ULN in 55%, TP53 disruption in 83%, and unmutated IGHV in 78%. Baseline mutations of SF3B1 (30%), NOTCH1 (28%), ATM (15%), EGR2 (13%), MGA (11%), POT1 (11%), and BIRC3 (10%) were reported in ≥10% of cases. The median follow-up of patients was 2.8 years. Median time on ibrutinib was 29 months. Discontinuation of ibrutinib due to any reason except progressive disease (PD) was reported in 21% of patients and their median time on ibrutinib was 17 months. Median time to discontinuation due to any reason except PD was 16 months. At least 1 interruption was reported in 31% of patients, and 6% interrupted treatment >1 time. The median consecutive days of interruption was 28 and the median total days of interruption was 38. Dose reduction was required by 25% of patients, including 5% who reduced to 140 mg/d. Median dose intensity (proportion of administered vs planned doses of ibrutinib 420 mg/d) was 93.4%. Three-years PFS of the per protocol cohort was 80.3% (CI 69.9-92.2). Early discontinuation due to any reason except PD, treatment interruption, regardless of duration (≥1, ≥8, ≥14 and ≥21 consecutive days), and dose reduction had no impact on PFS. By recursive partitioning, PFS stratification based on the best cut-off for dose intensity did not show a negative outcome, indicating no compounded effect of both dose reduction and interruption. Conclusions: Ibrutinib treatment modifications seem not to have major impact on PFS in patients with high-risk CLL. The impact of ibrutinib dose intensity during the first months of therapy on PFS, and the impact of ibrutinib tolerability on time to next treatment will be explored in this dataset. Keywords: Tumor Biology and Heterogeneity, Diagnostic and Prognostic Biomarkers, Ongoing Trials Conflicts of interests pertinent to the abstract E. Zucca Honoraria: AbbVie; AstraZeneca; Janssen Research funding: AbbVie; AstraZeneca; Janssen D. Rossi Honoraria: AstraZeneca; AbbVie; Janssen Research funding: AstraZeneca; AbbVie; Janssen

(range 0 ‐ 320), no SARS ‐ CoV ‐ 2 reinfections were reported and, among IgG + cases, none of the patients became seronegative. Conclusions: CLL remains a high ‐ risk disease for COVID ‐ 19 regard-less of best understanding of SARS ‐ CoV ‐ 2 management and improved health ‐ care conditions during the 2W. Of note, patients in W&W have better OS compared to those previously treated or in active treatment at COVID diagnosis, suggesting that CLL treatment is worsening COVID ‐ 19 outcomes. Finally, PCR clears earlier in W&W patients than in treated cases. 8 patients in order to identify potential additional differences between the lymph node biopsy and the circulating PB CD19 + cell compartment. This algorithm showed 100% concordance with FISH karyotype and allowed the detection of at least one CNVs difference in 3/8 patients (37.5%). Conclusions: These results suggest that the multiregional sequencing of the different anatomical compartments of SLL is essential to gain a comprehensive view of the disease mutational landscape. This obser-vationmayhaveclinicalrelevancewhentreatmenttailoringisbasedon specific gene mutations used as molecular predictors that might be present in only one specific anatomical compartment of the disease. EA – previously submitted to EHA 2021. The research was funded by: Molecular bases of disease dissemination in lymphoid malignancies to optimize curative thera-peutic strategies, (5 x 1000 No. 21198), Associazione Italiana per la Ricerca sul Cancro Foundation Milan, Italy No conflicts of interest pertinent to the abstract.