Opinion|Videos|May 28, 2026

Relapsed/Refractory Setting and CAR-T Applications

Can ctDNA and interim PET guide shorter DLBCL chemo? Experts weigh de-escalation to cut toxicity, urging randomized trials.

Dr. Melody discusses ctDNA applications in CAR-T cell therapy settings, noting increased complexity compared to frontline applications. In frontline treatment, ctDNA serves as established prognostic variables similar to IPI, disease bulk, or patient comorbidities. However, second-line settings do not typically utilize prognostic variables for treatment selection as primary refractory disease itself carries prognostic significance regardless of initial risk factors.

CAR-T cell therapy presents unique challenges due to cell persistence and expansion dynamics. Unlike one-time infusions, CAR-T cells continue expanding, sometimes converting 30-day PET positivity to 90-day negativity, or conversely showing progression as cells die off without adequate expansion. TRANSFORM trial data demonstrated numerous patients with positive 30-day ctDNA continuing clearance over subsequent 60 to 90 days, complicating optimal monitoring timing and frequency determination.

Dr. Melody expresses interest in ALPHA3 trial readouts and bispecific antibody ctDNA clearance data, hypothesizing that controllable T-cell engaging therapy duration may provide clearer understanding compared to unpredictable CAR-T cell persistence. She orders ctDNA at restaging PET timepoints as additional data points, finding reassurance in patients with expanding CAR-T cells, downtrending ctDNA, and partial PET responses versus those with stable disease, non-expanding cells, and rising ctDNA.

While not changing direct treatment management, ctDNA influences monitoring intensity and relapse concern levels - what she terms "spidey senses" for clinical vigilance. Dr. Roschewski emphasizes that clinical decisions rarely depend on single data points, instead requiring multiple information sources for complex decision-making processes. This aligns with clinical reality where repeating tests and integrating various factors guides management rather than algorithmic single-test responses.


Related to this article