Response to CAR-T Therapy Can be Monitored Using Genome-Wide Sequencing of Cell-Free DNA in Patients with DLBCL

Diffuse large B-cell lymphoma (DLBCL), a subtype of non-Hodgkin lymphoma, has a five-year survival rate of up to 70%; however, approximately half of patients will ultimately relapse and/or become refractory to therapy. In this clinical setting, autologous chimeric antigen receptor T-cell (CAR-T) therapy has been approved by the United States Food and Drug Administration. While these therapies show significant promise, methods that enable monitoring of therapeutic efficacy could be clinically useful. The use of cell-free DNA (cfDNA), also known as liquid biopsy, has been utilized as a noninvasive surrogate to tumor tissue for the identification of tumor-specific mutations and for monitoring treatment response across a number of cancer types.

Here we present data that demonstrate the feasibility of using cfDNA to monitor response to CAR-T therapy in patients with refractory DLBCL. Whole blood was collected during conditioning chemotherapy prior to CAR-T administration, at the time of or shortly after administration, and throughout treatment for 12 patients. A total of 127 blood samples were collected and analyzed with a median of 10 samples (range 8-13) from each patient. For each sample, blood was separated using centrifugation to yield plasma and a buffy coat sample. Each plasma sample was subjected to cfDNA extraction using an automated, bead-based method. Extracted cfDNA was then converted to libraries for next generation sequencing and subsequently assayed with low-coverage (~0.4X) genome-wide sequencing. Copy number alteration (CNA) events were identified and characterized using analytical methods originally developed for noninvasive prenatal testing. To quantify the level of CNAs present in the plasma of cancer patients, we utilized the genomic instability number (GIN). The GIN is a metric intended to capture genome-wide autosomal deviation from empirically derived euploid dosage of the genome in circulation and is calculated as the absolute deviation of observed normalized sequencing read coverage from expected normalized read coverage summed across 50,034 autosomal segments. Cellular DNA from each buffy coat was also collected using a column-based extraction process and used as the template for quantification of the CAR-T construct utilized for the therapy.

In this initial small cohort of 12 patients, the majority (8/12) were male and the median age was 52 years (range: 38-77). At the date of data censoring, four patients had an ongoing complete response, five had a complete or partial response but have since relapsed or are now deceased, one had a mixed response, and two had progressive disease. The GIN threshold (GIN = 170) has been utilized previously to identify patients with an aberrant CNA profile consistent with the presence of a tumor. When this threshold was applied, all 12 patients had an elevated GIN at the time of CAR-T administration. To determine whether a patient was progressing despite CAR-T therapy, the GIN values for each patient were evaluated throughout treatment (median=70 days; range=23-154 days). In all four patients with an ongoing complete response, the GIN value at the time point closest to the date of data censoring was below the threshold, suggesting the GIN was consistent with clinical response. Of the five patients that ultimately progressed after an initial PR or CR, four had evidence of progression based on the GIN. In one patient that had a mixed response but ultimately progressed, the progression could be observed using the GIN 66 days before clinical relapse was noted. Finally, two patients had no evidence of a clinical response; however, one of these patients had no detectable ctDNA after treatment and represented a second discordant result in this cohort.

In addition to measuring cfDNA, the CAR construct was measured in the background of the total cellular DNA obtained from the buffy coat in those patients receiving Axi-Cel (n=11) using digital PCR. As expected, all 11 patients that received this CAR-T therapy showed no presence of the construct prior to CAR-T administration. After administration, the construct was detected in 4/11 baseline samples and all 11 patients showed the presence of the construct at some point during their treatment.

Overall, these data describe a proof-of-concept for the use of multiple liquid biopsy technologies to monitor therapeutic response in DLBCL patients receiving CAR-T therapy.