Direct Method for Estimating the Fraction of T and B/Plasma Cell-Derived DNA

Background: Immunotherapies including checkpoint inhibitors and CAR-T cells have revolutionized oncology, yet significant unmet needs remain for monitoring their efficacy and safety, including immune-related adverse events. Since only a small fraction of immune cells are in circulation, current pharmacodynamic strategies to monitor such therapies to targeted tumors and other tissues have typically relied on invasive surgical sampling. We tackled this limitation by developing a novel strategy for serial, noninvasive monitoring of tissue-resident lymphocytes, by cell-free DNA immunoreceptor sequencing, which captures lymphocyte turnover.

Methods: We developed QUARTZ (QUAntification of Rearrangements in Targeted Zones), an integrated molecular strategy for profiling of B- and T-cell receptors to estimate their abundances in complex cellular mixtures. Briefly, using high-throughput targeted capture sequencing of immunoreceptor loci, QUARTZ relies on accurately measuring RAG1/2-mediated deletions in V[D]J loci across all BCR/TCR loci (IGH, IGK, IGL, TCRA/D, TCRG, TDRB), as well as AID-mediated class-switch rearrangements in IGH constant regions in B-cells (IGHAs, IGHGs, IGHE, IGHM/D). The corresponding excision rates thus allow estimation of total B- and T-cell fractions (using IGH and TCRG), as well as lymphocyte subset frequencies. For calculation of IgK vs IgL frequencies among B-cells and gd- vs ab- frequencies among T-cells, we estimated corresponding allelic exclusion rates from productive vs unproductive VDJ repertoires. To overcome stereotyped and stochastic sources of noise in high throughput sequencing data impacting performance, we applied Gaussian Process Regression and local GC content correction. We tested analytical performance using FACS-sorted purified B- and T-cells, corresponding contrived spike-in admixtures at concentrations ranging from 0.5% to 100%, and 355 cfDNA samples with known tumor fractions from 238 patients with diverse B- and T-lymphoid tumors, as estimated by CAPP-Seq. Finally, we evaluated QUARTZ’s clinical performance for monitoring CAR T-cell therapy and SARS-CoV2 mRNA immunization. Specifically, we profiled using 314 samples including from sequential cfDNA and tumor samples from B-cell lymphoma patients treated by CAR19 (n=66), as well as 7 healthy individuals before (d0, n=7) and after (d1/7/14/28; n=4) COVID-19 booster immunization.

Results: Limiting dilution studies of purified B- and T-cells revealed the QUARTZ measurements to have high accuracy, linearity, and calibration, with analytical sensitivity (LOD95) for of total B and T cells to be ~2.5% and ~0.6%, respectively. In contrast with expected circulating lymphocyte fractions in PBMCs, we found surprisingly low (<2%) contribution of cfDNA by B- or T-cells in healthy adults (n=3), consistent with longer steady state half-lives of lymphocytes relative to other cell types contributing to cfDNA. We confirmed these QUARTZ estimates using both whole genome cfDNA methylation profiling (Loyfer Nature 2023), as well as inferred expression by EPIC-Seq (Esfahani Nature Biotechnology 2022). In striking contrast, we observed surge of T cell derived cfDNA after COVID19 booster vaccination peaking between day1 and 7 (mean 4.1%).

In cfDNA from patients with diverse B- and T-cell neoplasms, we observed significant correlations between corresponding QUARTZ BCR/TCR frequencies and tumor fractions from CAPP-Seq, including in T-cell lymphoma (n=132, R=0.95) or LBCL (n=89, R=0.90). In multiple myeloma, the fraction of class-switched B-cell was correlated with tumor fraction of CAPP-Seq (n=8, R=0.89).

Among CAR19 treated LBCL patients (n=53), T-cell derived cfDNA generally peaked at d7 after CAR19 infusion, with peak cfDNA expansion significantly hampered by prior Bendamustine (p=0.01) but not prior ASCT (p=n.s.). The magnitude of expansion was also significantly associated with favorable outcomes (EFS p = 0.009, OS p = 0.01).

Conclusions: QUARTZ shows promising analytical and clinical performance as a potentially valuable noninvasive tool to accurately quantify immunodynamics of B/T cell and their subsets, including after CAR19 therapy and mRNA immunization. The significant association of QUARTZ B and T cell estimates with other measures of tumor fraction further suggests its utility for simultaneous assessment of tumor burden across diverse B/T lymphoid neoplasms.