Axicabtagene Ciloleucel in the Real World: Outcomes and Predictors of Response, Resistance and Toxicity

Introduction:

CD19 CAR T cells lead to durable responses in 40% of r/r aggressive B cell NHL patients. We performed a multicenter retrospective study of axicabtagene ciloleucel given in a real world setting where eligibility/management considerations may diverge from clinical trials. We evaluate efficacy and safety, and patient/disease factors associated with response and toxicity.

Methods:

Patient and treatment characteristics were summarized descriptively. Response and toxicity were reported with 95% exact binomial CIs. CyTOF was performed on frozen PBMCs using 38 metal-tagged mAbs. Multiplex IF was performed on FFPE tissue with standard, primary Abs sequentially, paired with a unique fluorochrome per published protocols. Standard IHC was done on FFPE whole tissue.

Results:

In total, 76 patients, median age 64, were included (Table 1). The majority had an ECOG PS ≤ 2. Twenty-one percent of patients had double/triple hit cytogenetics. Nearly half had an IPI ≥3 at treatment. Nearly 1/3 had a prior autologous transplant, 25% had prior ibrutinib, and 11% had prior lenalidomide. Twelve percent had bulky disease and 36% received bridging therapy following pheresis.

Seventy-three patients were evaluable for response (Table 2). At 4m median f/u, best ORR and CRR was 64% and 41% among those treated. Six patients had 6m f/u, all PRs at 1m: 3 converted to CR and 3 had PD. Eleven patients (13%) were pheresed but not treated due to PD (6), infection (2), or non-conforming cells (3). By ITT analysis, the ORR and CRR were 57% and 36%. OS at 4m among those treated was 84%; PFS will be calculated with longer f/u. In univariate analysis, PS, tumor bulk, IPI, baseline CRP and prior ibrutinib were significantly associated with lack of response (Table 3). There was no association between response and double/triple hit cytogenetics, grade 3+ CRS or NT, or the use of tocilizumab/steroids.

Among treated patients, 96% experienced CRS; in 17% this was ≥ grade 3. Two patients died from CRS (3%). Median time to onset was 1d; median duration was 6d (0-14d). NT was seen in 76% of patients; in 38% this was ≥ grade 3. One case of NT was fatal. Median time to onset was 5d; median duration was 8d (0-34d). Tocilizumab and steroids were given to 67% and 78% of patients. ICU care occurred in 30% of patients. Eleven treated patients have died: 6 from PD and 5 from toxicity. In univariate analysis, peak ferritin was associated with grade 3+ CRS and NT; peak ALC was associated with grade 3+ CRS, and peak CRP and prior autologous transplant were associated with grade 3+ NT (Table 4).

Three of 4 patients who had a biopsy with CD19 staining after relapse were positive (Fig 1). All 3 patients with PDL1⁺ tumors were refractory to CAR T cell therapy. Multiplex IF and IHC were performed on 2 primary refractory patients at progression (Fig 1). One was CD19^-/PDL1⁺; multiplex IF showed an abundance of CAR⁺ T-cells (Fig 1A,B). The second was CD19⁺/PDL1^- and multiplex IF showed no CAR⁺ T-cells (Fig 1C,D). CyTOF analysis of PBMCs at serial timepoints was performed on 6 patients (4 CRs, 1 PD, 1 PD after CR)(Fig 2, Table 5). Peak CAR T cell levels were seen at day 7 in all patients with increased expression of PD1, 41BB, and Ki67, as well as CC3 indicating apoptosis, followed by a reduction in CAR T cells by day 14. Immune subsets that associate with response will be evaluated and reported. Results from additional patients and longer f/u will be presented.

Conclusion:

Retrospective analysis of a multicenter cohort treated in the real world with axi-cel reveals important distinctions from ZUMA-1. The ORR and CR rate are lower than the 82% and 54% reported on ZUMA-1. This may reflect inclusion of sicker patients with a poorer PS, and/or with different histologies (ie transformation from non-FL). Outcomes were significantly worse in high risk lymphomas, reflected by IPI, PS, tumor bulk, and baseline CRP. Rates of CRS and NT were similar to ZUMA-1, but toxicity was not associated with tumor bulk or response. It was associated with higher peak inflammatory markers and ALC, which may reflect peak CAR T cell levels, as shown previously. Progression biopsies highlight 3 potential resistance mechanisms: loss of target antigen, an inhibitory tumor/TME, and lack of CAR T cell tumor infiltration. Immunomodulatory molecules on CAR T cells that may affect their activity and survival are upregulated early. This suggests that unique combination approaches are necessary for specific patients/tumors.

View largeDownload slide

View largeDownload slide

Close modal