Farooq, Umar

Abstract Glioblastoma (GBM), isocitrate dehydrogenase wild-type (IDHwt) and central nervous system diffuse large B-cell lymphoma (CNS-DLBCL) are aggressive brain tumors with overlapping MRI features, yet distinct treatment approaches. Noninvasive tools are needed to aid in differential diagnosis. Deep learning on T1 postcontrast and T2-weighted MRI sequences were used to differentiate GBM and CNS-DLBCL. A three-stage temporal study design was utilized. Model development was performed on 146 patients with CNS-DLBCL and 146 age-matched, sex-matched, and MRI year–matched patients with GBM diagnosed at Mayo Clinic between 1998 and 2019. Models were tested on independent temporal test cohorts. Initial testing included 240 independent GBM diagnosed at Mayo Clinic between 1998 and 2019. The prospective test cohort included 37 patients with CNS-DLBCL and 256 patients with GBM diagnosed at Mayo Clinic after January 1, 2020, and 36 patients with CNS-DLBCL diagnosed at an external institution. Of the patients diagnosed at Mayo Clinic, 47% had MRIs generated from non-Mayo institutions. Two different model approaches were compared: (i) ensemble approach using area under the receiver operating characteristic curve (AUC) and cross-validation for model selection and (ii) loss approach minimizing cross-entropy loss and cross-validation to evaluate prediction performance. The AUCs on the prospective test cohort were 0.84 [95% confidence interval (CI), 0.78–0.90] and 0.83 (95% CI, 0.77–0.88) for the ensemble and loss approaches, respectively. Stability of ensemble prediction improved with the increasing number of models. Stratified AUC analysis demonstrated consistent performance across sex and age. We utilized a robust temporal study design and applied 2 different analytic approaches to develop a classification model. The findings confirm the feasibility of using MRI-based deep learning models to differentiate GBM from CNS-DLBCL. Significance: GBM, IDHwt and CNS-DLBCL are aggressive brain tumors with overlapping MRI features, yet distinct treatment approaches. Noninvasive tools are needed to aid in differential diagnosis. We developed MRI-based deep learning models to differentiate GBM, IDHwt from CNS-DLBCL using a rigorous three-stage temporal design that included prospective validation. The model AUC on a prospective cohort was 0.84.

Abstract Underrepresentation of racial and ethnic subgroups in cancer clinical trials remains a persistent challenge. Restrictive clinical trial eligibility criteria have been shown to exacerbate this problem. We previously identified that up to 24% of patients treated with standard immunochemotherapy would have been excluded from recent first-line trials in diffuse large B-cell lymphoma (DLBCL) based on 5 laboratory-based criteria. These ineligible patients had worse clinical outcomes and increased deaths related to lymphoma progression, suggesting the potential exclusion of patients who could have benefited most from the novel therapies being evaluated. Using data from the prospectively enrolled Lymphoma Epidemiology Outcomes cohort study, with demographics broadly similar to the US patients diagnosed with lymphoma, we evaluated the impact of laboratory eligibility criteria from recent first-line DLBCL trials across various racial and ethnic backgrounds. There were significant differences in the baseline laboratory values by race/ethnicity with Black/African American (AA) patients having the lowest mean hemoglobin and highest creatinine clearance. Based on recent clinical trial eligibility criteria, AA and Hispanic patients had higher rates of laboratory-based ineligibility than non-Hispanic White patients. The largest gap in the clinical outcomes between eligible and noneligible patients was noted within AA patients with an overall survival hazard ratio based on POLARIX clinical trial criteria of 4.09 (95% confidence interval, 1.83-9.14). A thoughtful approach to the utility of each criterion and cutoffs for eligibility needs to be evaluated in the context of its differential impact across various racial/ethnic groups.

Abstract In this multi-institutional retrospective study, we examined the characteristics and outcomes of 160 patients with high-grade B-cell lymphoma, not otherwise specified (HGBL-NOS)—a rare category defined by high-grade morphologic features and lack of MYC rearrangements with BCL2 and/or BCL6 rearrangements ("double hit"). Our results show that HGBL-NOS tumors are heterogeneous: 83% of patients had a germinal center B-cell immunophenotype, 37% a dual-expressor immunophenotype (MYC and BCL2 expression), 28% MYC rearrangement, 13% BCL2 rearrangement, and 11% BCL6 rearrangement. Most patients presented with stage IV disease, a high serum lactate dehydrogenase, and other high-risk clinical factors. Most frequent first-line regimens included dose-adjusted cyclophosphamide, doxorubicin, vincristine, and etoposide, with rituximab and prednisone (DA-EPOCH-R; 43%); rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP; 33%); or other intensive chemotherapy programs. We found no significant differences in the rates of complete response (CR), progression-free survival (PFS), or overall survival (OS) between these chemotherapy regimens. CR was attained by 69% of patients. PFS at 2 years was 55.2% and OS was 68.1%. In a multivariable model, the main prognostic factors for PFS and OS were poor performance status, lactate dehydrogenase >3 × upper limit of normal, and a dual-expressor immunophenotype. Age >60 years or presence of MYC rearrangement were not prognostic, but patients with TP53 alterations had a dismal PFS. Presence of MYC rearrangement was not predictive of better PFS in patients treated with DA-EPOCH-R vs R-CHOP. Improvements in the diagnostic criteria and therapeutic approaches beyond dose-intense chemotherapy are needed to overcome the unfavorable prognosis of patients with HGBL-NOS.

Abstract Purpose: Older patients with relapsed/refractory (R/R) large B-cell lymphoma (LBCL) may be considered ineligible for curative-intent therapy including high-dose chemotherapy with autologous stem-cell transplantation (HDT-ASCT). Here, we report outcomes of a preplanned subgroup analysis of patients ≥65 years in ZUMA-7. Patients and Methods: Patients with LBCL refractory to or relapsed ≤12 months after first-line chemoimmunotherapy were randomized 1:1 to axicabtagene ciloleucel [axi-cel; autologous anti-CD19 chimeric antigen receptor (CAR) T-cell therapy] or standard of care (SOC; 2–3 cycles of chemoimmunotherapy followed by HDT-ASCT). The primary endpoint was event-free survival (EFS). Secondary endpoints included safety and patient-reported outcomes (PROs). Results: Fifty-one and 58 patients aged ≥65 years were randomized to axi-cel and SOC, respectively. Median EFS was greater with axi-cel versus SOC (21.5 vs. 2.5 months; median follow-up: 24.3 months; HR, 0.276; descriptive P < 0.0001). Objective response rate was higher with axi-cel versus SOC (88% vs. 52%; OR, 8.81; descriptive P < 0.0001; complete response rate: 75% vs. 33%). Grade ≥3 adverse events occurred in 94% of axi-cel and 82% of SOC patients. No grade 5 cytokine release syndrome or neurologic events occurred. In the quality-of-life analysis, the mean change in PRO scores from baseline at days 100 and 150 favored axi-cel for EORTC QLQ-C30 Global Health, Physical Functioning, and EQ-5D-5L visual analog scale (descriptive P < 0.05). CAR T-cell expansion and baseline serum inflammatory profile were comparable in patients ≥65 and <65 years. Conclusions: Axi-cel is an effective second-line curative-intent therapy with a manageable safety profile and improved PROs for patients ≥65 years with R/R LBCL.

Abstract An enhanced understanding of the molecular heterogeneity of diffuse large B-cell lymphoma (DLBCL) has opened the door to clinical trials evaluating novel agents with subtype-specific activity. It is an emerging question whether core needle biopsies (CNB) can adequately meet the increasing tissue requirements of these clinical trials. This can potentially lead to selective enrollment of patients who can undergo excisional biopsy (EB). It is also important to know whether patients who can undergo extensive diagnostic work up differ in their disease characteristics and outcomes from those who cannot. In this observational study, we describe the characteristics, outcomes, and adequacy of diagnostic tissue in patients with newly diagnosed DLBCL and primary mediastinal large B-cell lymphoma who underwent EB vs CNB. Of the 1061 patients, 532 (49.8%) underwent EB and 529 (50.1%) underwent CNB. A significantly higher proportion of patients with CNB had advanced stage disease, an international prognostic index of ≥3, and inadequate tissue for molecular analyses. Patients with CNB had significantly worse 5-year event-free survival (67.6% vs 56.9%; hazard ratio [HR], 0.76; confidence interval [CI]95, 0.6-0.9, P < .001) and 5-year overall survival (76.4% vs 69.2%; HR, 0.8; CI95, 0.6-0.9, P < .001). Thus, patients who underwent CNB have poor-risk features and inferior outcomes on frontline chemoimmunotherapy, are more likely to have inadequate tissue for molecular analyses, and might not meet the tissue requirements of biomarker-driven clinical trials. Thus, the increasing tissue requirements of biomarker-driven clinical trials may result in the exclusion of patients with high-risk DLBCL who need novel agents.

Abstract Low-grade B-cell lymphomas other than follicular and small lymphocytic lymphoma (LGBCL) account for 10% of all B-cell non-Hodgkin lymphomas. Despite improvements in survival outcomes for these patients, little is known about cause of death (COD) in the rituximab era. For a better understanding, we studied 822 newly diagnosed patients with marginal zone, lymphoplasmacytic, and unclassifiable low-grade B-cell lymphoma prospectively enrolled in the University of Iowa/Mayo Clinic Specialized Program of Research Excellence Molecular Epidemiology Resource from 2002 to 2015. COD was assigned based on medical record review using a standard protocol. At a median follow-up of 107 months, 219 (27%) patients had died. The incidence of lymphoma-related deaths when pooling across subtypes was lower than non–lymphoma-related deaths (10-year incidence, 8.0%; 95% confidence interval [CI]: 6.2-10.4 vs 13.6%; 95% CI: 11.2-16.6). The incidence of lymphoma-related deaths varied by subtype, ranging from 3.7% at 10 years in extranodal marginal zone lymphoma to 19.3% in lymphoplasmacytic lymphoma/Waldenström macroglobulinemia. Patients with early progression or retreatment events, defined using event-free survival at 24 months from diagnosis, had significantly higher likelihood of lymphoma-related death compared with patients without early events (10-year estimate: 19.1% vs 5.1%, respectively; P < .001), whereas the rates for non–lymphoma-related death were comparable in patients with or without early events (10-year estimates: 11.0% vs 15.3%, respectively). In conclusion, the most common COD in LGBCLs in the first decade after diagnosis was for causes other than lymphoma. Progression or retreatment within the first 2 years of diagnosis was a strong predictor for risk of lymphoma-related death.

AbstractBecause there have been a dvances in frontline treatment for mantle cell lymphoma (MCL) over the last 2 decades, we sought to characterize the changes in frontline treatment patterns and their association with outcomes. Patients with newly diagnosed MCL from September 2002 through June 2015 were enrolled in a prospective cohort study, and clinical characteristics, treatment, and clinical outcomes were compared between patients diagnosed from 2002 to 2009 (Era 1) compared with 2010 to 2015 (Era 2). Patient age, sex, and simplified MCL International Prognostic Index (sMIPI) score were similar between the 2 groups. In patients age 65 years or younger, there was less use of rituximab plus hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone (R-Hyper-CVAD) (16.1% vs 8.8%) but more use of rituximab plus maximum-strength cyclophosphamide, doxorubicin, vincristine, and prednisone (R-maxi-CHOP) alternating with rituximab plus high-dose cytarabine (R-HiDAC), also known as the Nordic regimen, and R-CHOP alternating with rituximab plus dexamethasone, high-dose cytarabine, and cisplatin (R-DHAP) (1.1% vs 26.4%) and less use of R-CHOP or R-CHOP-like regimens (64.5% vs 35.2%) but more use of R-bendamustine (0% vs 12.1%) in Era 2 (P < .001). These changes were associated with improved event-free survival (EFS; 5-year EFS, 34.3% vs 50.0%; P = .010) and overall survival (OS; 5-year OS, 68.8% vs 81.6%; P = .017) in Era 2. In patients older than age 65 years, there was less use of R-CHOP or R-CHOP-like therapy (39.0% vs 14.3%) and nonstandard systemic therapy (36.6% vs 13.0%) but more use of R-bendamustine (0% vs 49.4%). These changes were associated with a trend for improved EFS (5-year EFS, 25.4% vs 37.5%; P = .051) in Era 2. The shift from R-CHOP or R-CHOP-like regimens to R-bendamustine was associated with improved EFS (5-year EFS, 25.0% vs 44.6%; P = .008) in Era 2. Results from this prospective cohort study provide critical real-world evidence for improved outcomes with evolving frontline patterns of care in patients with MCL.

Abstract The SCHOLAR-1 international retrospective study highlighted poor clinical outcomes and survival among patients with refractory large B-cell lymphoma (LBCL) treated with conventional chemotherapy. Axicabtagene ciloleucel (axi-cel), an autologous anti-CD19 chimeric antigen receptor (CAR) T-cell therapy, demonstrated durable responses in patients with refractory LBCL in the pivotal phase 1/2 ZUMA-1 study (NCT02348216). Here, we compared SCHOLAR-1 with the 2-year outcomes of ZUMA-1. Prior to comparison of clinical outcomes, propensity scoring (based on a broad set of prognostic covariates) was used to create balance between ZUMA-1 and SCHOLAR-1 patients. In the pivotal phase 2 portion of ZUMA-1, 101 patients received axi-cel and were evaluable for response and survival. In SCHOLAR-1, 434 and 424 patients were evaluable for response and survival, respectively. ZUMA-1 patients were more heavily pretreated than were SCHOLAR-1 patients. The median follow-up was 27.1 months in ZUMA-1. The objective response rate (ORR) and complete response rate were 83% and 54% in ZUMA-1 vs 34% and 12% in SCHOLAR-1, respectively. The 2-year survival rate was 54% in ZUMA-1 and 20% in SCHOLAR-1, and a 73% reduction in the risk of death was observed in ZUMA-1 vs SCHOLAR-1. These results were consistent with those of an additional standardization analysis in which strata were limited to 2 prognostic factors (refractory categorization and presence/absence of stem cell transplant after refractoriness to chemotherapy) to conserve sample size. Despite the limitations of a nonrandomized analysis, these results indicate that axi-cel produces durable responses and a substantial survival benefit vs non–CAR T-cell salvage regimens for patients with refractory LBCL.

SummaryZUMA‐1 (NCT02348216) examined the safety and efficacy of axicabtagene ciloleucel (axi‐cel), an autologous CD19‐directed chimaeric antigen receptor (CAR)‐T cell therapy, in refractory large B‐cell lymphoma. To reduce treatment‐related toxicity, several exploratory safety management cohorts were added to ZUMA‐1. Specifically, cohort 6 investigated management of cytokine release syndrome (CRS) and neurologic events (NEs) with prophylactic corticosteroids and earlier corticosteroid and tocilizumab intervention. CRS and NE incidence and severity were primary end‐points. Following leukapheresis, patients could receive optional bridging therapy per investigator discretion. All patients received conditioning chemotherapy (days −5 through −3), 2 × 106 CAR‐T cells/kg (day 0) and once‐daily oral dexamethasone [10 mg, day 0 (before axi‐cel) through day 2]. Forty patients received axi‐cel. CRS occurred in 80% of patients (all grade ≤2). Any grade and grade 3 or higher NEs occurred in 58% and 13% of patients respectively. Sixty‐eight per cent of patients did not experience CRS or NEs within 72 h of axi‐cel. With a median follow‐up of 8·9 months, objective and complete response rates were 95% and 80% respectively. Overall, prophylactic corticosteroids and earlier corticosteroid and/or tocilizumab intervention resulted in no grade 3 or higher CRS, a low rate of grade 3 or higher NEs and high response rates in this study population.