Response rates of extra-nodal diffuse large B cell lymphoma to anti-CD19-CAR T cells: A real word retrospective multicenter study

Chimeric antigen receptor T-cells (CAR-T) are widely used for the

patients had nodal disease (ND)-only and 12 of 126 (10%) showed no disease assessed by PET-CT. There were no significant differences in CAR-T related toxicities and in the median Progression free survival (PFS) between EN patients and ND (10.76  (p = .021 and <.001, respectively), while sex, type of product administered, age and performance status did not predict PFS and OS. Of note, all the patients with involvement of gastrointestinal tract (n = 9), urinary tract (n = 9), or pharynx (n = 3) at lymphodepletion, progressed or had an early relapse. In conclusions, patients with >2 EN sites at lymphodepletion have significantly worse clinical outcomes compared to patients with <3 EN sites. Patients with specific sites of EN disease may demonstrate grim prognosis.
A retrospective study in a cohort of 32 patients treated with anti-CD19 CAR-T cells showed that soft tissue infiltration of lymphoma (n = 12) at lymphodepletion was the only factor correlated with adverse prognosis. 6 Furthermore, real life data analysis from 116 patients treated with axicabtagine ciloleucel and tisa-cel show that involvement of two or more EN sites at lymphodepletion was associated with reduced PFS and OS. 7 Several cases of DLBCL treated with CAR-T cells have described penetration of CAR-T cells to pleural cavity and cerebrospinal fluid (CSF) suggesting that CAR-T cells may enhance EN lymphoma response. 8,9 In this multicenter study, we provide a comprehensive description of the response and associated toxicity of diverse EN organs to commercially available CAR-T cells and compare to patients with ND.  10 Baseline PET-CT prior to lymphodepletion and CAR-T cell administration was used in order to assess the disease burden using the Lugano response criteria 11 and, specifically, the allocation whether patient had no EN involvement, 1-2 EN sites, or >2 EN sites. Based on these criteria we designed an organ based response criteria to ensure consistency between centers (Table 1). 11 The EN lymphoma cohort included patients either with EN involvement only or combined nodal and EN organs. The description of the EN organs involved in the PET-CT at lymphodepletion was independently extracted by two reviewers. The toxicity associated with CAR-T cell administration was recorded using the American Society of Transplantation and Cellular Therapy (ASTCT) consensus guidelines. 12

| Preparative regimen and supportive care
Once product was available for infusion, patients were admitted to the Bone Marrow Transplantation ward in designated rooms with high efficiency particulate air (HEPA) filters. Prior to infusion of CAR-T cells, patients were given cyclophosphamide (300-500 mg/m 2 ) and Fludarabine (30 mg/m 2 ) for 3 days (days À5 to À3). Cells were thawed and infused according to manufacturer recommendations.

| Characteristics of CAR-T cells related response in EN lymphoma
Among EN patients (n = 72), 69% had combined nodal and EN lym-

| Characteristics of CAR-T cells related toxicity in EN lymphoma
Incidence of overall cytokine release syndrome (CRS) was higher in patients with EN lymphoma compared to ND, however this did not reach statistically significance (76% vs. 57%, p = .06) ( commonly involved organs were BM/bone involvement (n = 5) and liver involvement (n = 5).
Incidence of overall immune effector-cell associated neurotoxicity syndrome (ICANS) was similar in the ND and the EN group. Among the four patients with severe ICANS in the EN group, two had CNS involvement and two had liver involvement. Of note, among patients with CNS involvement, high grade ICANS was noted in 2 of 11 (18%).

| DISCUSSION
The aim of this study was to consistently explore the characteristics and response rates of DLBCL involving EN sites at lymphodepletion therapy prior to CAR-T cells. In this multicenter study, we showed that for all patients included (EN and ND) The most frequent involved site in primary EN lymphoma is variable between different studies and most commonly reported organs are GI, bone, or BM. [1][2][3][4] In our cohort we assessed the frequency of EN sites at lymphodepletion and show that the most common organ was bone and BM followed by lung involvement. By analyzing treatment response segregated by organs we did not identify any organ that was significantly associated with treatment response. However, all patients with involvement of GI/urinary/kidney or pharynx at lymphodepletion experienced relapse or progression less than 3 months from CAR-T infusion. To our knowledge there is only one study (n = 32 patients) assessing EN specific involvement prior to CAR-T cells. That study reported that most patients presented with soft tissue (37.5%) and BM (25%) lymphoma involvement. 6 Interestingly, in multivariate analysis, soft tissue involvement by lymphoma was the only factor associated with PFS and OS but no other EN organs. The aformentioned study is not comparable to our trial due to major distinctions. Most notably, our trial assessed commercially available CAR-T cells and analyzed treatment response based on EN site at lymphodepletion and not at diagnosis.
Secondary CNS lymphoma involvement is an area of unknown ground in the CAR-T field with specific neurologic toxicity concerns.