Friedberg, Jonathan W.

7058 Background: Follicular and marginal zone B cell lymphoma (FL; MZL) demonstrate an indolent course with heterogeneous outcomes anda potential for spontaneous remissions indicating immune system intervention. Therapeutic immunization is therefore an attractive approach but new antigens that evoke a strong immune response are needed. EO2463 expands pre-existing memory CD8+ T cells recognizing non-self protein sequences from gut bacteria which cross-react with B cell antigens and can kill HLA-A2 restricted cells (T2) loaded with target peptides. EO2463 includes 4 HLA-A2 synthetically produced epitopes which exhibit molecular mimicry with the B cell markers CD20, CD22, CD37, and CD268 (BAFF-receptor), as well as a the CD4 helper-epitope UCP2 derived from hTERT. EO2463 is being used to drive anti-tumor activity against B cell malignancies. Methods: Patients (pts) with FL and MZL, stage 1-3A, and HLA-A2, are eligible. In the safety lead-in pts with relapsed/refractory (R/R) disease were given EO2463 SC q2 weeks (w) x 4, then q4w, for a max of 12 months; at w7 lenalidomide (20 mg/day for 21/28 days up to 12 cycles) is added (EL), and if no complete remission (CR) at w19, rituximab (375 mg/m2 IV, q1w x 4, then q4w x 4) is also added (ER2). Doses evaluated: 150 μg and 300μg/peptide. Results: 3 pts received 150 μg/peptide and 6 pts 300 μg/peptide (EO2463 1 pat, EL 2 pts, ER2 6 pts): 2 MZL, 7 FL pts with median 2 lines (range 1-4) of prior systemic therapy. No related grade ≥3 adverse events were seen with EO2463 monotherapy. Most common related events were grade 1 to 2 local administration site reactions in 5/9 pts. Adverse events during combination treatment were as expected for R2. PET-CT on w6 after EO2463 monotherapy suggested clinical activity in 4/9 pts (1 PR, 1 pt size reduction 15%, 1 pt 20% reduced tracer uptake, 1 pt 5/6 target lesions reduced metabolic activity). Overall objective responses (OR) were seen in 6/9 pts (67%; 1st OR on EO2463 1 pat, EL 4 pts, ER2 1 pat), with CR in 5/9 (56%) pts; median time to response was 18w (range 8-43w), response ongoing in 5 pts, range 24-76w. Expansion of specific CD8+ T cells against mimic peptides and targeted B cell antigens were detected in all responding pts, incl. 2 pts with no measurable B cells at baseline (prior anti-CD20). Expansion of specific T cells was detected at w5 in 5/6 pts and was maintained as long as currently tested (up to w94, 43w after last EO2463 dose). No decline in expansions were seen after start rituximab. Conclusions: EO2463 (300 μg/peptide) monotherapy, EL, and ER2 are well tolerated, with encouraging clinical activity with EO2463 monotherapy and with subsequent CR in 5/9 pts on combo. Rapid and durable expansion of specific CD8+ T cells was seen in all responding pts, consistent with the preclinical hypothesis. Additional cohorts investigate EO2463 alone or in combination in newly diagnosed or R/R pts. Results will be reported at the meeting. Clinical trial information: NCT04669171 .

7053 Background: In ECHELON-1 (NCT01712490), 6-year follow-up (FU) analyses demonstrated significant improvements in overall survival (OS) and progression-free survival (PFS) with A+AVD (brentuximab vedotin plus doxorubicin, vinblastine, and dacarbazine) versus ABVD (doxorubicin, bleomycin, vinblastine, and dacarbazine), with a comparable safety profile. Here, we report data at 7-year median FU. Methods: Analyses of OS and PFS per investigator were conducted in the intent-to-treat (ITT) population (data cut-off March 11, 2023). Patients (pts) were randomized 1:1 to receive ≤6 cycles of A+AVD (n=664) or ABVD (n=670) on days 1 and 15, every 28 days. PET scan after cycle 2 (PET2) evaluation was mandatory. Long-term safety outcomes in the safety population included resolution or improvement of peripheral neuropathy (PN), second malignancies, and pregnancies. Results: At median FU of 89.3 months (95% CI 87.0–90.2), 7-year OS rates were 93.5% (95% CI 91.1–95.2) with A+AVD and 88.8% (95% CI 85.8–91.1) with ABVD; OS favored A+AVD over ABVD (HR 0.62; 95% CI 0.42–0.90; p=0.011). Subgroup analyses showed consistent OS benefit for A+AVD, including in the age <40 years and Stage IV disease subgroups (Table). 7-year PFS rates with A+AVD vs ABVD were 82.3% (95% CI 79.1–85.0) vs 74.5% (95% CI 70.8–77.7; HR 0.68 [95% CI 0.53–0.86]; p=0.001). PN improved/resolved in most pts at last FU (A+AVD: 86%; ABVD: 87%). Median (range) time to complete resolution of PN (A+AVD vs ABVD) was 16 (0–373) vs 10 (0–343) weeks; median (range) time to improvement was 42 (2–182) vs 19 (15–142) weeks. PN was ongoing in 28% (4% grade ≥3) of A+AVD and 20% (1% grade ≥3) of ABVD pts. Second malignancies were reported in 5% of A+AVD and 6% of ABVD pts. Pts and their partners reported 84/92 livebirths/pregnancies with A+AVD and 59/73 with ABVD; no stillbirths were recorded. Conclusions: At 7-year median FU, pts with stage III/IV cHL who received A+AVD showed a sustained PFS and OS benefit vs ABVD, with PFS rates indicating potential curability. The safety profile in pts treated with A+AVD showed no new safety signals at 7 years. Clinical trial information: NCT01712490 . [Table: see text]

Abstract Although initial therapy of mantle cell lymphoma (MCL) is not standardized, bendamustine plus rituximab (BR) is commonly used in older patients. Rituximab (R) maintenance after induction is often used. Thus, the open-label, randomized phase 2 ECOG-ACRIN Cancer Research Group E1411 trial was designed to test 2 questions: (1) does addition of bortezomib to BR induction (BVR) and/or (2) addition of lenalidomide to rituximab (LR) maintenance improve progression-free survival (PFS) in patients with treatment-naïve MCL? From 2012 to 2016, 373 previously untreated patients, 87% aged ≥60 years, were enrolled in this trial. At a median follow-up of 7.5 years, there is no difference in the median PFS of BR compared with BVR (5.5 vs 6.4 years; hazard ratio [HR], 0.90; 90% confidence interval [CI], 0.70-1.16). There were no unexpected additional toxicities with BVR treatment compared with BR, with no impact on total dose/duration of treatment received. Independent of the induction treatment, addition of lenalidomide did not significantly improve PFS, with median PFS in R vs LR (5.9 vs 7.2 years; HR, 0.84; 90% CI, 0.62-1.15). Most patients completed the planned 24 cycles of LR at the scheduled dose. In summary, adding bortezomib to BR induction does not prolong PFS in treatment-naïve MCL, and LR maintenance was not associated with longer PFS compared with R alone after BR. Nonetheless, the &gt;5-year median PFS outcomes in this prospective cooperative group trial indicate the efficacy of BR followed by R maintenance as highly effective initial therapy for older patients with MCL. This trial was registered at www.clinicaltrials.gov as #NCT01415752.

Abstract Although it is evident that standard-dose whole-brain radiotherapy as consolidation is associated with significant neurotoxicity, the optimal consolidative strategy for primary central nervous system lymphoma (PCNSL) is not defined. We performed a randomized phase 2 clinical trial via the US Alliance cancer cooperative group to compare myeloablative consolidation supported by autologous stem cell transplantation with nonmyeloablative consolidation after induction therapy for PCNSL. To our knowledge, this is the first randomized trial to be initiated that eliminates whole-brain radiotherapy as a consolidative approach in newly diagnosed PCNSL. Patients aged 18 to 75 years were randomly assigned in a 1:1 manner to induction therapy (methotrexate, temozolomide, rituximab, and cytarabine) followed by consolidation with either thiotepa plus carmustine and autologous stem cell rescue vs induction followed by nonmyeloablative, infusional etoposide plus cytarabine. The primary end point was progression-free survival (PFS). A total of 113 patients were randomized, and 108 (54 in each arm) were evaluable. More patients in the nonmyeloablative arm experienced progressive disease or death during induction (28% vs 11%; P = .05). Thirty-six patients received autologous stem cell transplant, and 34 received nonmyeloablative consolidation. The estimated 2-year PFS was higher in the myeloablative vs nonmyeloablative arm (73% vs 51%; P = .02). However, a planned secondary analysis, landmarked at start of the consolidation, revealed that the estimated 2-year PFS in those who completed consolidation therapy was not significantly different between the arms (86% vs 71%; P = .21). Both consolidative strategies yielded encouraging efficacy and similar toxicity profiles. This trial was registered at www.clininicals.gov as #NCT01511562.

AbstractTo address the current and long‐term unmet health needs of the growing population of non‐Hodgkin lymphoma (NHL) patients, we established the Lymphoma Epidemiology of Outcomes (LEO) cohort study (NCT02736357; https://leocohort.org/). A total of 7735 newly diagnosed patients aged 18 years and older with NHL were prospectively enrolled from 7/1/2015 to 5/31/2020 at 8 academic centers in the United States. The median age at diagnosis was 62 years (range, 18–99). Participants came from 49 US states and included 538 Black/African‐Americans (AA), 822 Hispanics (regardless of race), 3386 women, 716 age <40 years, and 1513 rural residents. At study baseline, we abstracted clinical, pathology, and treatment data; banked serum/plasma (N = 5883, 76.0%) and germline DNA (N = 5465, 70.7%); constructed tissue microarrays for four major NHL subtypes (N = 1189); and collected quality of life (N = 5281, 68.3%) and epidemiologic risk factor (N = 4489, 58.0%) data. Through August 2022, there were 1492 deaths. Compared to population‐based SEER data (2015–2019), LEO participants had a similar distribution of gender, AA race, Hispanic ethnicity, and NHL subtype, while LEO was underrepresented for patients who were Asian and aged 80 years and above. Observed overall survival rates for LEO at 1 and 2 years were similar to population‐based SEER rates for indolent B‐cell (follicular and marginal zone) and T‐cell lymphomas, but were 10%–15% higher than SEER rates for aggressive B‐cell subtypes (diffuse large B‐cell and mantle cell). The LEO cohort is a robust and comprehensive national resource to address the role of clinical, tumor, host genetic, epidemiologic, and other biologic factors in NHL prognosis and survivorship.

Introduction: The use of standardized, evidence-based classification systems is crucial for the accurate diagnosis and treatment of diseases. Moreover, standardized classification facilitates research and epidemiologic studies and promotes consistency in communication among healthcare professionals. Since 2016, the revised 4 th edition of the World Health Organization classification (WHO-HAEM4R) has been the global standard for diagnosis of lymphoid malignancies. With the emergence of new data, 2 new classification systems were developed and published in 2022: the 5 th edition of the WHO Classification (WHO-HAEM5) and the International Consensus Classification (ICC). Both WHO-HAEM5 and ICC maintain a shared fundamental concept of disease classification that integrates clinical, pathologic, and molecular data. However, they differ on nomenclature, establishment of new entities, and/or diagnostic criteria for some disease categories. To evaluate the impact of these differences on real-world diagnosis of non-Hodgkin lymphoma (NHL), we examined the diagnostic classification of NHL in the Lymphoma Epidemiology of Outcomes (LEO) Cohort Study (NCT02736357). Methods: Initiated in 2015, LEO is an ongoing prospective observational study of patients aged ≥18 years with newly diagnosed NHL. Patients are enrolled at 8 major U.S. medical centers. Clinical, epidemiologic, pathologic, and treatment data are abstracted at baseline and active follow-up is conducted for all patients. Lymphoma subtype is coded for each case based on expert hematopathology re-review of the diagnostic pathology slides, the pathology report, biomarker data, and clinical data. Of note, NHL subtype distribution is similar to SEER data. We studied all patients enrolled in LEO between 7/1/2016 (the date LEO pathology review started using WHO-HAEM4R diagnostic codes) and 5/31/2020. WHO-HAEM4R diagnoses and additional clinicopathologic data, when relevant, were used to map cases into the corresponding WHO-HAEM5 and ICC diagnoses. Differences between WHO-HAEM5 and ICC were annotated for each case as: None - same disease entity; Minor - difference in nomenclature with similar diagnostic criteria; Major - difference in disease category and/or diagnostic criteria; or Unevaluable - insufficient data to assign WHO-HAEM5 and/or ICC diagnosis. Results: LEO enrolled 6143 patients during the study period. Of these, comparison between WHO-HAEM5 and ICC was evaluable in 5730 (93.3%). Unevaluable cases included those without a specific WHO-HAEM4R diagnosis at enrollment (N=384; e.g., patients with a low-grade B-cell lymphoma that could not be definitively sub-classified) and those with a specific WHO-HAEM4R diagnosis, but with insufficient clinicopathologic data to assign a specific WHO-HAEM5 and/or ICC diagnosis (N=29). Of the 5730 evaluable cases, 5376 (93.8%) showed no difference between WHO-HAEM5 and/or ICC diagnosis, 311 (5.4%) showed minor differences (nomenclature only), and 43 (0.8%) showed major differences (Table 1). The 43 major differences all involved B-cell NHLs; 20 (46.5%) were attributable to different approaches to classifying double-hit lymphomas, 21 (48.9%) to classification of splenic/leukemic B-cell lymphomas, and 2 (4.6%) to classification of B-cell lymphomas occurring at specific anatomic sites (Table 2). Conclusion: In a large, prospective lymphoma cohort reflecting real-world clinical practice at 8 major U.S. medical centers, major diagnostic differences in the classification of NHL using WHO-HAEM5 or ICC classification criteria were seen in 0.8%, whereas the remaining 99.2% of diagnoses were either the same or showed differences in nomenclature only. The existence of 2 concurrent classification systems presents potential for discrepancies in pathologic diagnosis, clinical practice, clinical trials, and other lymphoma research. Furthermore, some of the differences between WHO-HAEM5 and ICC are clinically and potentially therapeutically significant, and their resolution requires further study. Nevertheless, our findings argue that the proportion of patients affected would be small in real-world practice settings. This appears largely related to incidence rates of specific lymphoma subtypes, with major differences between the 2 classifications predominantly affecting rare entities, while there is general concordance on the most common forms of NHL.

Introduction: PET/CT is the standard-of-care for staging and therapy response assessment in lymphoma. Fluorodeoxyglucose (FDG)-avidity in extranodal marginal zone lymphoma (EMZL) remains controversial with some variability by extranodal location. The current Lugano classification recommends the use of CT scans only in staging and follow-up for EMZL. However, emerging data from small studies have challenged this suggestion. In this study we aimed to determine the ability of PET/CT to detect FDG-avid EMZL during staging work-up. Methods: First, we retrospectively analyzed the University of Miami (UM) MZL database (period 2/2016-1/2023) searching for patients who underwent staging PET/CT. Images were reviewed by expert radiologists to ascertain avidity of tumor and not surrounding tissue, with specific attention to ocular and GI locations; if scans were not available (n=15), we retrieved data from reports (results presented in part at ICML2023). Next, we aimed to validate the initial UM results additionally analyzing data from imaging reports for patients prospectively enrolled in the Lymphoma Epidemiology of Outcomes (LEO) MZL cohort (period 7/2015-5/2020), excluding LEO patients from Miami to avoid overlap. Patients with high-grade transformation at diagnosis were excluded. For patients with &gt;1 extranodal site, each location was counted independently. We considered FDG-avid disease if SUVmax was ≥2 as this value is commonly observed between mediastinal blood pool (BP) and liver background. To normalize data, we calculated ratios of lymphoma SUVmax versus both BP and liver background, as available. Patients without measurable disease were excluded from analysis. Results: Among 210 patients with EMZL in the UM cohort, 152 had staging PET/CT and were analyzed. Twenty-two patients demonstrated &gt;1 extranodal site providing a total of 187 locations. The LEO cohort included 411 patients from 7 centers, with 235 having available imaging reports. Fifty-two patients in this cohort demonstrated &gt;1 extranodal site providing a total of 306 locations. Disease distribution is depicted in Table 1. Among 187 EMZL locations in the UM cohort, the most common were gastric (n=33, 17.6%), ocular (n=31, 16.6%), lung (n=30, 16%), skin and soft tissue (n=16, 8.6%, each). Similarly, lung (n=63; 20.6%), soft tissue (n=46; 15%), gastric and ocular (n=34 each; 11%) were common locations in the LEO cohort. EMZL demonstrated FDG-avidity (SUVmax ≥2) across all locations with identical values in both cohorts (UM= 78.1% & LEO= 76.5%). FDG-avidity by location is shown in Table 1A & B with salivary gland (100% & 100%), lung (93.3% & 95.2%), soft tissue (93.8% & 100%), and airways (90.9% & 100%) demonstrating high avidity in both cohorts. Contrarily, skin was commonly non-FDG avid (93.8% & 81%). The median size of FDG-avid lesions was 2.4cm (range, 0.7-17.4) in UM and 2.9cm (range, 0.1-12) in LEO cohorts. In patients with &gt;1 extranodal site FDG-avidity was nearly universal across sites (UM=93% & LEO=100%). We observed a significant correlation between lymphoma size and FDG-avidity for UM (r=0.33; P=0.001), and for LEO (r=0.55; P&lt;0.0001). In patients with background FDG-avidity data in the UM cohort (n=150), a BP index and liver index ≥1 detected lymphoma in 79.3% & 71.5% cases, respectively. We were unable to calculate indexes in LEO cohort since only limited data on BP (n=26) and liver (n=19) background was available in imaging reports. Conclusions: This study represents the largest analysis evaluating accuracy of PET/CT to determine initial tumor burden in 387 patients with EMZL. We demonstrate that EMZL is largely an FDG-avid disease across two cohorts, and PET/CT should be included in the staging of these patients while acknowledging possible limited sensitivity in GI locations due to physiologic uptake. Furthermore, we observed that patients with multiple extranodal locations demonstrate uniform FDG-avidity across all sites except for skin involvement. Based on our results we recommend the incorporation of PET/CT into standard staging of EMZL in planned revised Lugano classification guidelines. In those demonstrating initial FDG-avidity, a lower SUVmax threshold of SUVmax ≥2, comparison to mediastinal BP rather than to liver background activity and evaluating for resolution of focal FDG-avidity should be considered to assess treatment response in routine care and clinical trials.

Background. FL is associated with a heterogeneous clinical course, and currently available prognostic indicators are unable to reliably separate patients who exhibit early therapy resistance (POD24, defined as progression within 24 months from diagnosis) from those who achieve “functional cure” following initial therapy. NGS methods developed in recent years enabled quantification of MRD. The ClonoSeq assay (Adaptive Biotechnology, Inc.) has been FDA-cleared to track MRD in several hematologic malignancies. Here we demonstrate that MRD analysis by ClonoSeq assay predicts outcomes in a prospective cohort of patients with advanced stage FL enrolled on SWOG S0016 clinical trial. Methods. SWOG S0016 was a phase III randomized study that compared the safety and efficacy of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone) with CHOP-RIT (CHOP + I 133-tositumomab) in patients with FL. Between 2001 and 2008, 531 patients with previously untreated advanced-stage FL (bulky stage II, III-IV) of any pathologic grade were randomly assigned to receive 6 cycles of R-CHOP or CHOP-RIT. We used high-throughput sequencing (HTS) of the IGH and IGK/L loci (ClonoSeq assay) from tumor biopsy specimenstaken prior to treatment to define CDR3 rearrangements that tag the tumor. Subsequently, 1-year bone marrow specimens were screened for residual disease. The baseline dominant clone was defined as ≥10 estimated number of genomes (ENG) and comprised ≥5% of the library. MRD-positive state (MRDp) was defined as a minimum Hamming distance (HD) ≤6 in a patient's post-library (in any clone also detected in a pre-library) and conserved pre-existing SHM (if HD&gt;0). Undetectable MRD (MRDu, at &lt;10 -4) was defined as CRD3 HD&gt;6 in a post-library with a depth &gt;10,000 ENG. Five-year progression-free survival was the primary endpoint (PFS; defined as the time from registration to the first observation of progression or death due to any cause). Landmark analysis was defined at 1 year after registration and used for imputing the 5-year PFS and OS. Overall survival (OS) was defined as the time from registration to date of death due to any cause or last follow-up. PFS and OS estimates and 95% confidence interval were calculated using the Kaplan-Meier method and compared for MRDp and MRDu disease using two-sided log-rank test. HR and 95% CI were calculated using Cox regression model. Results. Tumor biopsy tissue was available for 189 patients. Of these, 24 patients were not evaluable (16 did not consent for banking, 6 were ineligible and 4 had an early event). In 36 patients, pre-libraries did not yield a trackable clone. In total, MRD status could not be ascertained in 47 patients due to absence of a clear trackable clone in either pre- or post-library, contributing to a 28.8% failure rate. Of 116 patients which yielded a trackable clone in both libraries, 83 were MRDu and 33 were MRDp. Baseline characteristics (age, sex, race, β2M, B symptoms, bulky disease, grade 3, stage, and FLIPI risk) were similar between MRDp and MRDu patients; however, MRDp patients were more likely to exhibit morphologic baseline bone marrow involvement by FL (82% vs. 55%, p=0.01). The estimated 5-year PFS was 72% in MRDu patients and 30% in MRD+ patients (Figure); 10-year PFS was 56% and 17%, correspondingly. Meanwhile, 5-year OS was not different (96% for MRDu, 81% for MRD+, p=0.91). Patients with MRDu had a significantly higher best overall response rate than MRDp patients (96% vs 79%, p=0.005), but complete response (CR) rates were similar (41% vs. 30.3%). Patients who received CHOP-RIT had better PFS regardless of MRD status (HR=0.48, p=0.01 for MRDu; HR=0.38, p=0.02 for MRDp), while OS was not statistically different. Five-year PFS was superior in patients who achieved MRDu (75% for CR and 70% for PR) compared with MRDp patients, regardless of response (34%). Patients who had MRDp status had a significantly increased risk of experiencing POD24 relative to MRDu patients (RR=6.5, 95% CI 3.0-14.0; p&lt;0.0001). Conclusion. Here we for the first time demonstrate that MRDu status, as assessed by Clonoseq, predicts improved 5- and 10-year PFS in patients with FL initially treated with chemoimmunotherapy. MRD+ status in the bone marrow at a 1-year timepoint was associated with POD24. Despite a relatively high failure rate to detect trackable sequences using current technology (~29%), MRD assessment by NGS is a promising prognostic tool in FL. *co-senior authors

Background: Prognostic modeling allows personalized risk prediction for individual patients (pt). The A-HIPI model in advanced stage classical Hodgkin lymphoma (AS-HL), developed and validated by the HoLISTIC Consortium (www.hodgkinconsortium.org), generates the individualized probability of a progression-free survival (PFS) event or death (OS) within the first 5 years (y) from diagnosis in pts based on continuous variables (www.qxmd.com/calculate/calculator_869/a-hipi). Clinical prognostic tools in lymphoma (eg, IPS, IPI, FLIPI, etc) typically use groupings of categorical values to define risk. Grouping a continuous value often results in loss of information, and most tools are not predictive for individual pts. However, discrete groupings have clinical utility & practicality for a) defining pt populations for clinical trials & real world studies, b) stratification within clinical trials, and c) crafting treatment guidelines. We studied potential approaches for utilizing the A-HIPI model to generate risk groups with input on strengths & limitations from the HoLISTIC modeling team & clinical experts. Methods: The A-HIPI model was developed via TRIPOD guidelines on 4,022 pts treated on 8 international clinical trials for AS-HL (Rodday. JCO 2023). External validation was performed on a dataset of 1,431 pts from 4 prospective registries. The 5y PFS (PFS5) in the A-HIPI development dataset was 77% (95% CI: 76-78); the 5y OS (OS5) was 92% (95% CI: 91-93). This represented the average outcome for a pt with AS-HL pt naïve to other clinical data at presentation. The distribution of PFS5 & OS5 predictions were heavily skewed (ie, asymmetric distribution) in both the A-HIPI discovery and validation dataset. While not unexpected due to the excellent PFS & OS in this disease setting, this presents challenges in the delineation of risk groups as depicted below . Three approaches were examined for the generation of A-HIPI risk groups. Proposed cutoffs were defined using the distribution of A-HIPI risk scores and data from the model-building cohort (ie, clinical trials). Validation was done using the A-HIPI validation cohort (ie, HL registries). Results: Approach 1: Risk groups based on clinical thresholds. Clinicians were queried what estimates of PFS5 would constitute high vs low risk. The positive, right-skewed distribution of A-HIPI risk scores limited this approach ( Figure), as cutoffs of PFS5 &lt;70 and PFS5&gt; 90 would only identify 15% and &lt;1% of the population, respectively. Approach 2: Risk groups based on deviation from “average” pt. The 5-y PFS was 77% (95% CI: 76-78). We explored defining “standard risk” based on this confidence interval as well as clinical boundaries, with pts above or below this classified as decreased or increased risk, respectively. Results of a +/-5% clinical boundary are presented in the Table. Approach 3: Risk groups based on “ranking” of pts. Here we ranked the A-HIPI risk scores of the 4022 AS-HL pts in the model building cohort and used the distribution of the risk scores as a benchmark. The risk profile for a future pt is then compared to this distribution (eg, how do you rank compared to your peers). Continuous results are often presented this way (eg, tertiles or quartiles). It also allows flexibility for the user to define the size of the risk groups and/or clinical threshold of interest. Application of this approach showed good alignment between the predicted model percentiles and the observed distribution of scores in the validation cohort ( Table). This is also reflected in calibration curves presented in the primary manuscript. An online application (eg, R-Shiny) will be provided at the meeting to allow users to define their own cutoffs to aid in pt prognostication as well as identify populations for clinical trial development. Conclusions: There are challenges with defining risk groups from individual risk prediction modeling in AS-HL. Different applications and purposes, the skewed distribution of events/risk estimates, as well as varying clinical definitions of high risk, make it challenging to define consensus expert-based risk groupings for AS-HL. A flexible “rank-based” approach appeared to provide the most clinical utility & data granularity, which may be leveraged for clinical trial design and pt stratification. Further analysis and discussion of how to optimally define high-risk and low-risk populations in AS-HL will be needed as new therapeutic options emerge.

Background: We previously reported thatearly events in follicular lymphoma (FL), defined as progression of disease within 24 months (POD24) or early transformation to a more aggressive histology, are significantly associated with increased risk of lymphoma related death. To refine clinical management, the FLIPI24 prognostic model was developed to predict the risk of early events after starting immunochemotherapy (IC) (Maurer et al, 2022). The FLIPI24 model was found to be significantly superior to standard clinical models for prediction of both event-free survival at 24 months (EFS24) and overall survival (OS) in internal and external validation sets of patients receiving IC. We sought to evaluate the prognostic ability of the FLIPI24 model in patients with FL not receiving IC. Methods: The cohort consists of prospectively enrolled patients in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER) and the multicenter Lymphoma Epidemiology of Outcomes (LEO) Cohorts. Data were collected from medical records by standard protocol with all events verified by medical record review. Eligible patients were those diagnosed with grades 1-3A FL and initiated observation, rituximab monotherapy, radiation, or other non-IC therapy. The FLIPI24 model utilizes 5 continuous variables: age (linear 60-90 years with inflection at age 75), HGB (linear 8-17 g/dL), WBC (linear 4-11x10 9/L), LDH/ULN (linear 0.5-5), and B2M (linear 1-10 mg/L). FLIPI24 risk is grouped as follows: low risk (≤10%], low-average risk (10-15%], average risk (15-20%], high risk (20-40%], very high risk (&gt;40%) for an early event. EFS was defined as time from diagnosis to progression, relapse, (re)treatment, histologic transformation, or death due to any cause. OS was defined as the time from diagnosis until death from any cause. Cox models and Kaplan Meier curves were used to evaluate association between FLIPI24 and EFS or OS. Results: 1542 patients initiating non-IC management approaches at diagnosis from 2002-2020 were identified. 51% were &gt;60 years (IQR 52-70), and 49% were male. Most patients (86%) had FL grade 1-2. 61% of patients met any criteria for high tumor burden per GELF, BLNI, or GITMO based on available variables. Treatment approaches were 833 observation (54%), 315 rituximab monotherapy (20%), 186 radiation (12%), and 208 other non-IC approaches (14%). At median follow-up of 58 months, EFS24 was 73.8% (95% CI: 71.6%, 76.1%), 5-year OS was 90.9% (95% CI: 89.3%, 92.5%) and 253 (16%) patients died. The FLIPI24 score identified poor outcomes among patients with high/very high-risk score (FLIPI24 &gt;20%), with median EFS of 1.8 years (95% CI, 1.3, 3.28) and 5 year OS of 65.1% (95% CI: 55.9-75.8) (Figure 1). When evaluating patient subsets, for patients with high tumor burden (defined based on available variables) and high/very high-risk score, median EFS was 1.6 years and 5 year OS was 64.6% (95% CI: 54.9%-76.1%). Among patients with low tumor burden FL and high/very high-risk score, median EFS was 2.0 years (95% CI: 0.84-NA) and 5 year OS was 66.2% (95% CI: 42.4%-100%). Patients undergoing observation with high/very high-risk score similarly had shortened median EFS of 1.53 years (95% CI: 1.25-NA) and 5 year OS of 68.9% (95% CI: 56.4%-84.1%). The FLIPI24 had superior concordance for OS (c=0.726) compared to the FLIPI (c=0.646) when evaluated in patients not treated with IC. Results were comparable in the subsets of patients observed (c=0.706 vs 0.596) and receiving rituximab monotherapy (c=0.764 vs 0.721). Conclusion: The FLIPI24 model has been previously validated to provide an individual risk score at diagnosis for the likelihood of experiencing an event within 24 months from starting IC. Here we demonstrate that a high FLIPI24 score predicts for poorer outcomes following non-IC strategies including observation, treatment with radiation, and rituximab monotherapy. The FLIPI24 has better prognostic performance for OS than FLIPI in this non-IC population. High FLIPI24 in patients not treated with IC identifies a patient population who should be considered for IC and/or novel frontline therapies when feasible.