Hematology · Medical Literature
CAR T-cell Kinetics, Persistence, and Clinical Outcomes in Adult Patients with Relapsed/Refractory B-cell ALL Treated with Obecabtagene Autoleucel in the FELIX Study.
Abstract
Assessments of chimeric antigen receptor (CAR) T-cell pharmacokinetics by flow cytometry (FC) and a droplet digital PCR (ddPCR) assay were compared in adult patients with relapsed/refractory (R/R) B-cell acute lymphoblastic leukemia (B-ALL) following treatment with obecabtagene autoleucel (obe-cel) in the phase Ib/II FELIX study (NCT04404660). CAR T-cell persistence and B-cell aplasia (BCA) were then correlated with event-free survival (EFS). Peripheral blood (PB) samples collected from 127 obe-cel-infused patients were tested by FC and ddPCR. The Spearman correlation coefficient was used to measure the correlation between the number of CAR T-positive cells by FC and ddPCR. The impact of CAR T-cell persistence and BCA (B cells <20 cells/μL by FC in PB) on EFS was assessed using Cox proportional hazards regression. ddPCR was observed to statistically correlate with both the surface (0.60, P < 0.0001) and intracellular FC assays (0.74, P < 0.0001). A higher sensitivity for detecting CAR T-cell positive samples was observed with ddPCR, with 58.8% and 41.3% of samples negative by surface and intracellular FC, respectively, being positive by ddPCR. Loss of CAR T-cell persistence (HR, 2.7; 95% CI, 1.4-5.4) and, to a lesser degree, B-cell recovery (HR, 1.7; 95% CI, 0.7-3.8) as time-dependent variables and at month 3 were associated with poorer EFS. ddPCR demonstrated enhanced sensitivity over FC methods for detection of obe-cel persistence. Additionally, ongoing persistence and BCA were associated with longer EFS and may be taken into consideration, together with clinical parameters, in informing decision making. In patients with R/R B-ALL treated with obe-cel, ddPCR assessment of CAR transgene levels produced results consistent with FC while providing superior sensitivity. ddPCR-detected CAR T-cell persistence at month 3 and at any time after infusion correlated with longer EFS versus loss of persistence, therefore potentially informing clinical decision making.
Clinical relevance
In adult patients with relapsed/refractory B-cell ALL treated with obecabtagene autoleucel (obe-cel) in the FELIX trial, does droplet digital PCR (ddPCR) provide superior sensitivity for monitoring CAR T-cell pharmacokinetics compared with flow cytometry (FC), and do CAR T-cell persistence and B-cell aplasia (BCA) independently associate with event-free survival (EFS)? __SNAPSHOT__ {"studyDesign":"Post hoc pharmacokinetic and biomarker analysis embedded within a phase Ib/II open-label, single-arm, multicenter study (FELIX; NCT04404660)","setting":"34 sites across the United States, United Kingdom, and Spain","sampleSize":"n=127 (all obe-cel–infused patients from FELIX); 1,715 peripheral blood samples tested by ddPCR, 734 by surface FC, and 703 by intracellular FC","population":"Adults (≥18 years) with CD19-positive R/R B-cell ALL; median age 47 years (range 20–81), 52% male, 29.9% Hispanic or Latino; included morphologic disease (≥5% bone marrow blasts), MRD-positive remission, or isolated extramedullary disease cohorts","intervention":"Obe-cel: autologous 4-1BB-ζ CD19-directed CAR T-cell therapy with a fast off-rate (low-affinity) CAT19 binder; administered as tumor burden–guided split-dose infusions on Days 1 and 10 (target total dose 410×10⁶ CAR T-cells) following lymphodepletion (fludarabine 30 mg/m² ×4 days + cyclophosphamide 500 mg/m² ×2 days); CAR T-cell kinetics monitored serially by both surface FC, intracellular FC, and ddPCR","comparator":"Not applicable (single-arm study); intra-patient methodologic comparison between ddPCR and two FC modalities (surface and intracellular) for CAR T-cell detection","primaryOutcome":"Correlation between CAR T-cell enumeration by FC and ddPCR (Spearman coefficient); secondary: association of CAR T-cell persistence and BCA with EFS by Cox proportional hazards regression (post hoc, descriptive secondary analyses—p-values not generated for these endpoints)","followUp":"Median follow-up 21.5 months (range 8.6–41.4 months) for the pharmacokinetic analysis cohort","fundingSource":"Industry (Autolus Therapeutics; multiple authors report personal fees, grants, or employment from Autolus)"}
Key findings
ASSAY CORRELATION: ddPCR correlated statistically with both surface FC (Spearman ρ=0.60, P<0.0001) and intracellular FC (Spearman ρ=0.74, P<0.0001), with intracellular FC showing modestly stronger agreement. However, ddPCR demonstrated substantially superior analytical sensitivity: 58.8% of samples deemed negative by surface FC and 41.3% of samples deemed negative by intracellular FC were positive by ddPCR, indicating that flow cytometric methods meaningfully underestimate CAR T-cell detection, particularly at low transgene levels. PERSISTENCE AND EFS: Loss of CAR T-cell persistence (modeled as a time-dependent covariate) was associated with significantly worse EFS (HR 2.7; 95% CI 1.4–5.4). B-cell recovery (B cells ≥20 cells/µL) was also associated with worse EFS, though with a wider confidence interval spanning unity (HR 1.7; 95% CI 0.7–3.8), indicating less certainty for BCA as an independent prognostic marker. LANDMARK EFS ESTIMATES: Among patients in ongoing remission at month 3, 9-month EFS was 85.4% (95% CI 72.9–92.4) with ongoing persistence versus 64.2% (95% CI 33.9–83.4) without; 12-month EFS was 73.5% (95% CI 59.3–83.4) versus 56.2% (95% CI 26.9–77.6). At month 6, median EFS was not yet reached in patients with ongoing BCA versus 22.1 months (95% CI 8.1–NE) in those with B-cell recovery; 12-month EFS probabilities were 81.7% versus 69.6%, respectively. TEMPORAL RELATIONSHIP: Loss of CAR T-cell persistence typically preceded B-cell recovery; at month 3, 63.2% of patients who had lost persistence still maintained BCA, declining to 22.2% by month 6. Conversely, virtually all patients with ongoing persistence at months 3 and 6 had concurrent BCA (96.7% and 100%, respectively). NOTE: The survival analyses were explicitly designated as descriptive and post hoc; p-values were not generated for Cox regression outputs, and all HR estimates should be interpreted as hypothesis-generating rather than confirmatory.
Practice implications
This analysis establishes that ddPCR is analytically superior to both surface and intracellular FC for detecting low-level obe-cel transgene in peripheral blood, with clinically meaningful false-negative rates for FC—particularly surface FC. This finding has direct implications for post-treatment monitoring: FC-based persistence assessments may prematurely classify patients as having lost CAR T-cell persistence, potentially influencing clinical decisions (e.g., timing of stem cell transplantation or salvage therapy). The association between ongoing CAR T-cell persistence and improved EFS is consistent with the biological rationale for obe-cel's engineered low-affinity binder and adds important prognostic context. However, because these are post hoc, descriptive, secondary analyses of an already single-arm study with no p-values generated for the survival comparisons, the persistence–EFS association is hypothesis-generating and should not be used as a standalone clinical decision rule. The BCA–EFS association is less certain given the HR confidence interval crosses 1.0. Key populations not well-represented include pediatric patients, patients with isolated extramedullary disease as a primary cohort, and those with certain high-risk cytogenetics (e.g., Ph-like ALL). Importantly, the decision to proceed to consolidative stem cell transplantation in patients who lose persistence or BCA is not directly addressed by this analysis. Only the abstract and supplementary data available through open sources were reviewed; the full published dataset was not accessible for independent verification of all reported statistics. Industry sponsorship by Autolus should be noted when interpreting conclusions favorable to obe-cel monitoring paradigms. __LIMITATIONS__ Post hoc, descriptive, secondary analyses: all survival analyses are explicitly non-inferential and p-values were not generated, limiting statistical rigor and precluding determination of whether observed differences meet conventional significance thresholds.,Single-arm study design: absence of a randomized comparator means the association between persistence and EFS cannot exclude confounding by favorable tumor biology or other patient-level factors that independently drive both persistence and outcomes.,Analytical comparison between ddPCR and FC is observational; the clinical impact of adopting ddPCR-based monitoring over FC on patient outcomes has not been prospectively tested.,The confidence interval for the BCA–EFS hazard ratio (0.7–3.8) crosses 1.0, indicating the association between B-cell recovery and worse EFS is not statistically established in this analysis.,Samples were not uniformly collected at all time points for all assays, potentially introducing informative censoring bias in the pharmacokinetic data.,Generalizability is limited to adults treated at specialized academic centers; pediatric patients, those with isolated extramedullary disease, and heavily pre-treated or Ph-like ALL populations are incompletely represented.,Industry sponsorship (Autolus Therapeutics) and multiple author conflicts of interest with the sponsor are present.,EFS—while clinically relevant—is a composite surrogate endpoint; overall survival data and the relationship of monitoring strategy changes to OS have not been reported in this analysis.,The ddPCR lower limit of detection cutoff and its calibration to clinical relapse risk have not been independently validated in external cohorts. __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ In adult patients with R/R B-ALL treated with obe-cel in the FELIX study, ddPCR demonstrated superior sensitivity over both surface and intracellular flow cytometry for detecting CAR T-cell persistence, with FC yielding false-negative rates of 41–59% relative to ddPCR—a finding with direct implications for post-infusion monitoring. Loss of CAR T-cell persistence was associated with a 2.7-fold higher hazard of relapse or death (95% CI 1.4–5.4), and ongoing B-cell aplasia trended toward better EFS, though these are post hoc, non-inferential, descriptive analyses without formal p-values. Clinicians monitoring obe-cel-treated patients should be aware that FC-based assays may underestimate residual CAR T-cell activity, and CAR T-cell persistence status at 3 months may serve as a clinically meaningful prognostic signal, pending prospective validation before incorporation into treatment algorithms.