Search

7560 Background: Tafasitamab (tafa) + lenalidomide (LEN) demonstrated efficacy in adult patients (pts) with R/R DLBCL ineligible for autologous stem-cell transplant in the pivotal Phase II study L-MIND (NCT02399085) and received accelerated approval in the United States in 2020 and conditional marketing authorization in Europe and Canada in 2021 in this setting. RE-MIND2 (NCT04697160), an observational, retrospective cohort study, compared pt outcomes from L-MIND with matched pt cohorts treated with other NCCN/ESMO recommended therapies. Methods: Methodology for RE-MIND2 has been presented previously. Hypothesis-generating analyses were conducted for pt subgroups of number of extranodal sites (ENS) (0–1 vs ≥2) and elevated lactate dehydrogenase (LDH) (yes vs no) in matched cohorts of pts receiving tafa + LEN vs systemic therapies pooled (STP), polatuzumab vedotin + bendamustine + rituximab (pola-BR), rituximab + LEN (R2), and CD19 CAR-T therapies (CAR-T). The primary endpoint was overall survival (OS). Results: Of 3,454 pts enrolled, 961, 106, 106, and 149 were treated with STP, pola-BR, R2, and CAR-T, resulting in 76, 24, 33, and 37 matched pairs for pts receiving tafa + LEN, respectively. Hazard ratios (HR) for OS show a trend toward favoring tafa + LEN in most pt subgroups (Table). Conclusion: These analyses suggest that tafa + LEN may be associated with improved OS vs selected systemic therapies for certain pts with high-risk disease and may further inform physicians’ treatment choices for pts with R/R DLBCL. These analyses are not powered for statistical comparison; small sample sizes in some subgroups result in wide confidence intervals (CI) and so results must be interpreted with caution. Data for other treatment cohorts, pt subgroups, and endpoints will be presented. Funding: MorphoSys AG. Clinical trial information: NCT04697160. [Table: see text]

7575 Background: CA-4948 is a novel oral inhibitor of interleukin-1 receptor-associated kinase 4 (IRAK4), which is essential for toll-like receptor (TLR) and interleukin-1 receptor (IL-1R) signaling in B cell proliferation. IRAK4 forms a Myddosome complex with MYD88 adaptor protein and drives overactivation of nuclear factor-kappa B (NF-κB), causing inflammation and tumor growth. CA-4948 has been reported to be well tolerated and active as monotherapy in heavily pretreated patients with relapsed/refractory (R/R) non-Hodgkin lymphoma (NHL). Preclinical studies demonstrated that tumor resistance and survival via IRAK4 activation could be delayed or reversed. CA-4948 crossed the blood-brain barrier in a murine PDX model of pCNS lymphoma, resulting in tumor response and prolonged survival. In combination with Bruton tyrosine kinase (BTK) inhibitors, CA-4948 showed in vivo synergy in B-cell NHL. Here we will present an update on the preliminary efficacy data of CA-4948+ibrutinib in R/R hematologic malignancies. Methods: This is an ongoing open-label trial (NCT03328078) of CA-4948 as monotherapy and in combination with ibrutinib. Part A1 (completed) dose escalation of CA-4948 as monotherapy; the recommended phase 2 dose (RP2D) is 300 mg BID with continuous oral dosing. Part A2 (dose escalation in combination with ibrutinib), and Part B (a basket design of 4 expansion cohorts of CA-4948 and ibrutinib: BTK-naïve MZL, DLBCL, or PCNSL and NHL with adaptive resistance to ibrutinib). The primary endpoints of Parts A1 and A2 include safety, tolerability, and RP2D. The primary endpoints of Part B include CR or ORR, with key secondary endpoints of DOR, DCR, PFS and OS following treatment of CA-4948 at dose levels of 200 (DL1) or 300 mg BID (DL2) with ibrutinib at full prescribed dose. Results: As of December 7th, 2021, 35 heavily pretreated NHL patients have received CA-4948 monotherapy (median age 66 years, range 50-87), of which six patients have been on CA-4948 for approximately 1 year or longer, suggesting CA-4948 has a long-term acceptable safety and tolerability profile at RP2D (dose level of 300 mg BID). In Part A2, 10 patients are treated with CA-4948+ibrutinib (median age 65 years, range 56-82). Median number of prior lines of anti-cancer therapies is 3 (range 1-8). No DLTs were observed at 200 or 300 mg dose levels to date. The preliminary efficacy data of seven evaluable patients with combination therapy showed 1 CR (MCL), 2 PR (MCL and MZL), 3 SD, and 1 PD, 3 of whom had failed prior ibrutinib. The preliminary data indicate the combination therapy may overcome ibrutinib resistance. Conclusions: CA-4948 as a monotherapy and in combination with ibrutinib is well tolerated with an acceptable long term safety profile and promising efficacy. Part A2 is transitioning to Part B basket cohorts of MZL, ABC-DLBCL, PCNSL and NHL with adaptive resistance to ibrutinib. Clinical trial information: 03328078.

e19201 Background: Automated extraction of key clinical features from imaging reports, such as the site, size and FDG avidity of lesions would allow for automatic determination of disease stage, response to treatment and relapse. Since the language used in clinical notes is domain and often task-specific, generic Natural Language Processing (NLP) tools are of limited utility. We describe a strategy for a rapid construction of task-specific NLP algorithms. As a test case, we aimed to identify bone involvement with lymphoma from PET/CT reports. Methods: We use an adaptive-learning approach to facilitate the generation of task-specific lexicons by a domain expert. In the case of identifying bone lymphoma, these include words describing the primary aspect (PA: e.g., “ uptake”, “SUV”); the secondary aspect, in this case, osseous sites (SA: e.g., “skeleton”, “T5”); modifiers such as context-specific affirmation/ negation/ uncertainty (e.g., “increased”/ “likely reactive”/ “possibly disease”) and generic negation (e.g., “no evidence of”). Presented with sentences containing seed PA terms, an expert is tasked with an exhaustive identification of terms for the other lexicons. In an iterative process, unseen sentences that do not include terms previously identified are presented for lexicon expansion. The process is repeated until no further expansion is possible. An explicit rule-based algorithm is built based on the lexicons to classify sentences as supporting, negating or inconclusive (i.e., requiring manual review) for the feature of interest. We used 500 reports to develop the algorithm, evaluated performance against 516 unseen reports that were independently classified by a blinded annotator, and measured the amount of effort (number of iterations) required to complete the PA and SA lexicon expansion. Results: Of 44 reports with bone involvement identified by the algorithm, 43 had an actual involvement and 1 was a negative case (PPV: 98%). Of 380 reports flagged as negative, 379 were truly negative and 1 was positive (NPV: 99.7%). Sensitivity was 64% (43/67); specificity 84% (379/449); 18% (93) reports were flagged for manual review. It took 13 and 5 iterations to build the PA and SA lexicons, each involving the review of up to 10 sentences. Conclusions: With minimal effort, we were able to construct a high-accuracy NLP system specific for identifying bone lymphoma. This approach can be extended to other contexts (e.g., lung involvement), domains (e.g., pathology reports), or to be utilized in machine learning.

e20017 Background: FDG avidity above liver on interim PET (PET2) during frontline ABVD is considered a marker of impending treatment failure and an indication to switch to an intensified regimen. However, in clinical practice the utility of PET2 for treatment decisions is less clear. We describe outcomes of patients with positive PET2 who continued treatment with ABVD in the clinical setting. Methods: A retrospective study of all patients with newly diagnosed advanced-stage Hodgkin lymphoma treated with frontline ABVD at Memorial Sloan Kettering Cancer Center between 2008-2017. Eligibility criteria were set to correspond with the RATHL inclusion criteria (stage IIB - IV, or IIA bulky or ≥ 3 involved sites). We identified all PET2 reports indicating suspected residual uptake. All positive PET2 images were then reviewed by a single study radiologist. To increase reproducibility and avoid selection of borderline cases, we defined as PET2 positive only those cases with a lesion-to-liver (mean) SUV ratio ≥ 1.3. We also used a recently published stringent criterion of lesion-to-liver (max) ratio ≥ 1.4 (mPET2+). Progression-free and overall survival (PFS, OS) were calculated from the date of initial treatment until progression or death of any cause. Consolidative radiation was not considered a PFS event, and all progressions were verified by biopsy. Results: We identified 227 patients fitting RATHL inclusion criteria treated with ABVD. Median age was 34, with 25% (58) ≥ 45 years, 12% (26) had an IPS ≥ 4; 28% (64) stage II (5% II-X) and 38% (87) with extranodal involvement. 57 (25%) patients had a PET2 report indicating suspected residual lymphoma (PET2+), however, only 32 (14%) met the more stringent mPET2+ criterion. Most patients with PET2+ continued ABVD (84%, 48), and 9 switched to escBEACOPP (this subset of patients had substantially worse disease and are not the focus of this analysis). 21 (9%) patients received consolidative radiation. With a median follow-up of 47 months (42-54m), PET2+ patients who continued ABVD had a 3yPFS of 70% (58-85%, n = 48); mPET+ had a 3yPFS of 71% (55-92%, n = 24). Overall survival was excellent regardless of PET2 status (5yOS 97%). Conclusions: The outcome of PET2+ patients in this analysis was better than previously reported and the continuation of ABVD was appropriate for most patients. Use of a confirmatory biopsy is important for identifying true progressions. Patients with PET2+ had an excellent OS. Evaluation of the superiority of alternative regimens in PET2+ patients requires an ABVD comparator arm.

8062 Background: Immunotherapy with anti CD20 is often associated with mild easily manageable infusion reactions. In rare cases, patients experience severe drug hypersensitivity reactions (DHR) serum sickness or anaphylaxis. These in turn may lead to discontinuation of the drug. In our experience, switching to a different anti-CD20 agent is a feasible alternative to discontinuation or desensitization protocols. Methods: From our pharmacology database we identified all the patients that received rituximab and/or obinutuzumab, and/or ofatumumab, and/or all the patients who received a flat dose of less than 50 mL of the same drugs and were followed at our institution. From the medical record, we identified all the cases where the anti-CD20 antibody was changed due to allergy, serum sickness or other types of DHR, and all those who received minimal doses of anti-CD20 in the context of a desensitization protocol. DHRs were evaluated either by an allergist, or by retrospective review following the World Allergy Organization guidelines. Our primary comparison, was to assess the proportion of pts able to completed planned infusion of abs using either approach (Fisher’s exact Test). Results: Among 343 patients receiving at least two different anti-CD20 antibodies or a flat dose of < 50 mL, we identified 44 patients experiencing severe DHRs needing intervention. At the time of the reaction, 16 (36%) received the anti-CD20 as single agent, 24 (54%) in combination with chemotherapy, 4 (9%) in combination with ibrutinib or lenalidomide. In 9 (20%) patients the reaction was defined as anaphylactoid (8 rituximab; 1 obinutuzumab) and in 8 (18%) patients, all receiving rituximab, as serum sickness. Episodes of DHR were addressed with either desensitization (n = 29) or change of anti-CD20 agent (n = 25), 9 patients received both of these approaches, one patient switched anti-CD20 antibodies twice. Overall, 21 desensitizations were successful (72.4%), 8 failed; 23 changes of anti-CD20 were successful (92%) and 2 failed (p = 0.09). Conclusions: In patients with DHR use of an alternative anti-CD20 antibody is safe and is an alternative or complementary approach to anti-CD20 desensitization.

TPS8072 Background: Diffuse large B-cell lymphoma (DLBCL) escapes host immune responses via inhibition of the B2M and CD58 pathways and upregulation of PD-1 ligands. However, treatment results with PD-1 blockade have been disappointing. One potential mechanism is the recruitment of intra-tumoral regulatory T-cells (Tregs) which suppress anti-tumor immunity and inhibit NK cell cytotoxicity. Targeting regulatory T cells with the CCR4 antibody mogamulizumab (KW-0761) represents a molecularly informed strategy to overcome this resistance. Mogamulizumab has been safely administered in combination with pembrolizumab (MK-3475) in patients with solid malignancies and may promote CD8 T-cell dependent effector and NK cell-dependent cytotoxicity in lymphomas. Methods: This is a multi-center NIH-ETCTN phase Ib/randomized phase II study. The phase I will evaluate the safety and tolerability of mogamulizumab in combination with pembrolizumab in patients with R/R DLBCL and determine the recommended phase II dose (RP2D). A traditional 3+3 design with a starting dose of pembrolizumab 200mg IV on day 1 of a 21-day cycle and mogamulizumab 1mg/kg IV on days 1, 8, 15 and then 1.5 mg/kg IV every 21 days. The phase II will evaluate the efficacy of the combination by PFS (primary endpoint) ORR and CR (secondary endpoints). This will be a randomized 1:1 study with allowed crossover, comparing the combination to single agent pembrolizumab Correlative studies will evaluate the association of tumor infiltrating CD8 and NK cells with response to treatment, somatic mutations in B2M and CD58 and with MHC-I expression on DLBCL cells. Functional characterization of circulating immune cells in the peripheral blood and measurement of pro and anti-inflammatory cytokines will be used to assess the levels, activation status and effector function of Tregs and circulating T cells. Inclusion criteria include measurable disease, ≥2 prior lines of therapy including or ineligible for autologous stem cell transplant, ECOG ≤ 2 and normal organ function. Prior or planned allogeneic stem cell transplant, as well as prior treatment with an anti PD-1/PD-L1/CTLA4 antibody, preexisting autoimmune disease or CNS involvement by lymphoma are exclusion criteria. The study aims to enroll up to 12 patients on the phase I and up to 58 patients on the phase II and can be opened at any ETCTN participating site. To date it has been opened at two sites and is accruing the first patients for the phase I portion. Clinical trial information: NCT03309878 .

8006 Background: Venetoclax (Ven)-Obinutuzumab (O) is approved for chronic lymphocytic leukemia (CLL) achieving frequent undetectable minimum residual disease (uMRD; Fischer NEJM 2019). Ven-Ibrutinib is synergistic with frequent uMRD but with grade >3 neutropenia in 33-48% patients (pts; Tam ASH 2019; Jain NEJM 2019). Zanubrutinib (B) is a highly specific BTK inhibitor that demonstrated 100% occupancy in lymphoid tissues, so may be preferred to combine with OVen. We hypothesize that treatment (tx) with BOVen using an MRD driven discontinuation strategy will achieve frequent uMRD and durable responses. Methods: In this multicenter, investigator initiated phase 2 trial (NCT03824483), eligible pts had previously untreated CLL requiring tx per iwCLL, ECOG PS 1, PLT >75 (ANC >0, PLT >20 if due to CLL). BOVen was administered in 28D cycles: B 160 mg PO BID starting D1; O 1000 mg IV D1 or split D1-2, 8, 15 of C1, D1 of C2-8; Ven ramp up initiated C3D1 (target 400 mg QD). Tx duration was determined by a prespecified uMRD endpoint (min 8 cycles). MRD was assessed in peripheral blood (PB; flow cytometry, sensitivity >10−4) starting C7D1 then every 2 cycles. Once PB uMRD was determined and confirmed in bone marrow (BM), tx continued 2 additional cycles. Adverse events (AE) were assessed per CTCAE v5. Median (med) time to uMRD (primary endpoint) was estimated using the Kaplan-Meier method. Results: The study accrued 39 pts (3-10/19): med age 59 years (23-73), 3:1 male, CLL IPI >4 26/39 (67%), unmutated IGHV 28/39 (72%), 17p del/ TP53 mutated 4/39 (10%), all pts were evaluable for toxicity with 37 evaluable for efficacy. At a med follow up of 8 months (mo; 3-10), 25/37 (68%) pts achieved PB uMRD. Med time to PB uMRD is 6 mo (4-8+). Another 8/37 (22%) had PB MRD < 0.1%. Of 25 with PB uMRD, 19 had BM uMRD with 10/19 completing 2 additional cycles and discontinued; 3 had BM MRD (all