Circulating Tumor DNA as a Precision Medicine Tool for Lymphoma

Circulating Tumor DNA as a Precision Medicine Tool for Lymphoma

Dr. Mark Roschewski

Dr. Wyndham H. Wilson

Mark Roschewski, MD, and Wyndham H. Wilson, MD, PhD

Article Highlights

  • Selection of targeted therapy for lymphoma remains largely empirical, highlighting the need for precision methods to directly monitor tumor dynamics at the molecular level.
  • Recent studies have demonstrated the utility of monitoring circulating tumor DNA (ctDNA) for immunoglobulin receptor gene sequences and genotypic DNA in diffuse large B-cell lymphoma.
  • Clinical validation studies are ongoing to prove clinical utility and to further explore the various roles of ctDNA. Challenges include the optimization of collection, processing, and turn-around times, and research and development are needed to enhance sensitivity, specificity, and reproducibility.

On December 13, 2016, former President Barack Obama signed the widely acclaimed 21st Century Cures Act into law. Among other things, this legislation pledged $4.8 billion toward former President Obama’s Precision Medicine Initiative and former Vice President Joe Biden’s Cancer Moonshot Initiative. These ambitious projects aim to jump-start technology-enabled personalized medicine and advance the most promising biomedical research initiatives, including innovative cancer research that is high risk and high reward.

“Precision medicine” can be defined as an approach based on a patient’s unique germline and somatic genetics and environment. In lymphoma, precision medicine is typically defined as the selection of targeted therapy based on genetic and/or functional characteristics of the tumor, although one cannot lose sight of the importance of the host on drug action and metabolism. The biology of all tumors is dynamic, however, and current barriers to precision medicine include tumor heterogeneity, clonal evolution, and imprecise monitoring tools. Currently, the selection of targeted therapy for lymphoma remains largely empirical, highlighting the need for precision methods to dynamically and directly monitor disease at a molecular level. Such tools must have robust sensitivity, specificity, and reproducibility.

A powerful emerging technology involves the ability to quantify small fragments of cell-free tumor-derived DNA in the peripheral blood, known as circulating tumor DNA (ctDNA). Next-generation sequencing (NGS) of ctDNA can detect disease with much greater sensitivity than conventional imaging scans and has the potential to precisely monitor clonal evolution of targetable mutations throughout the disease course (Fig. 1).1 Employing ctDNA as a liquid biopsy may also overcome inherent sampling errors of tissue biopsies and identify mechanisms of drug resistance in real time. Indeed, ctDNA holds great promise to fundamentally change how we approach patient treatment and clinical management.2

ctDNA Assays for Immunoglobulin Receptors

Modern NGS-based platforms can identify and quantify tumor-specific rearrangements of the immunoglobulin receptor (VDJ) gene sequences in B-cell lymphomas.3-5 Two important studies have demonstrated the utility of ctDNA for VDJ in diffuse large B-cell lymphoma (DLBCL).4,5 The first study, from the National Cancer Institute (NCI), included 126 patients with untreated DLBCL and a clonotypic VDJ sequence identified prior to therapy.4 Patients were treated with chemotherapy for curative intent, and serum samples were prospectively collected prior to therapy, at the beginning of each cycle of therapy, at the end of induction therapy, and at each follow-up visit. Importantly, the serum samples taken during follow-up were paired with CT scans until 5 years from the end of therapy. Various patterns of ctDNA kinetics during therapy were observed, and the kinetics of ctDNA clearance was predictive of clinical outcome. Almost half of the patients cleared ctDNA after only one cycle, and 78% were negative after two treatment cycles. Patients who cleared ctDNA after two cycles were significantly more likely to be progression free at 5 years compared with patients who were ctDNA positive (80.2% vs. 41.7%, p < 0.0001).

Because ctDNA is a noninvasive test that can be measured serially, it allows for direct measurement of tumor dynamics during therapy (Fig. 1).1 Early identification of patients at the highest risk of treatment failure while undergoing therapy is the basis for risk-adapted treatment. Risk-adapted strategies were recently reported in two prospective DLBCL studies using interim PET scans but did not improve clinical outcomes.6,7 Because ctDNA is tumor specific and PET scans are not, a combination of these different modalities may provide even greater precision in identifying patients at risk for treatment failure.

A second important finding of the NCI study was the performance of serial ctDNA monitoring after therapy compared with CT scans for the early detection of disease relapse. Recent studies have suggested that routine surveillance imaging with CT scans is not very effective or cost-efficient for detecting early relapse in DLBCL.8,9 However, effective surveillance monitoring strategies remain compelling, as some aggressive lymphomas can be cured with salvage therapy and are greater in patients with lower tumor burdens.10 Because disease recurrence occurs from a persistent tumor below limits of clinical detection, it was hypothesized that ctDNA for VDJ would identify disease relapse prior to CT scans and create a window of opportunity for earlier institution of salvage therapy (Fig. 2).11

In this regard, the NCI study showed that surveillance monitoring of ctDNA for VDJ detected recurrence a median of 3.5 months (range 0-200) prior to CT scans.4 In patients who experienced relapse later than 6 months from the end of therapy, ctDNA was identified in 91% (10 of 11), and 80% (8 of 10) were detected prior to clinical disease on CT scans. These findings were validated by investigators from Stanford University who studied plasma-derived ctDNA for VDJ in patients with DLBCL and compared the results with PET scans.5 In another study of B-cell lymphomas after allogeneic transplantation, ctDNA for VDJ was detected months before clinical relapse.12

The clinical and research role of molecular remission is an important area of investigation in aggressive B-cell lymphomas. Current response criteria rely exclusively on imaging scans, and yet up to 20% of patients with DLBCL have negative PET and CT scans at the end of treatment relapse, indicating the persistence of disease.13 This raises the obvious response category of “complete molecular remission” in DLBCL and other aggressive B-cell lymphomas. Unlike indolent lymphomas, where molecular remissions generally do not indicate cure, minimal residual disease eradication is required for cure of aggressive B-cell lymphomas.14,15 In contrast, achieving molecular remission may inform decisions on maintenance therapy for indolent lymphomas.

Quantitative levels of ctDNA at diagnosis may be a useful prognostic indicator. Multiple studies have now shown that ctDNA concentrations correlate with clinical markers of tumor burden such as lactate dehydrogenase (LDH), International Prognostic Index (IPI), and disease stage.4,5,16 One study in follicular lymphoma reported that quantitative levels of ctDNA was predictive of progression-free survival and independent of other prognostic indices.17 Future studies are clearly needed to further address the biologic relevance of quantitative levels of ctDNA. It is possible that lower levels of ctDNA are surrogates for active antitumor immunity and may identify patients more likely to benefit from immunotherapy as recently reported in mantle cell lymphoma.18

ctDNA Assays for Genotypic DNA: Liquid Biopsies

DLBCL is molecularly heterogeneous, and its molecular profile may predict responsiveness to targeted therapy.19-21 Tissue biopsies are the gold standard for determining molecular features of a tumor, but they are invasive, are prone to sampling error, and can be challenging to obtain. To overcome these problems, recent studies in DLBCL have characterized tumor heterogeneity using genomic profiling of ctDNA in the plasma.16,22,23 The question to be addressed is whether the ctDNA liquid biopsy can more accurately identify the tumor heterogeneity in all disease sites and accurately identify molecular changes over time. A recent study from Stanford University investigated the role of genotypic profiling of ctDNA in 76 patients with DLBCL using a novel sequencing-based method known as Cancer Personalized Profiling by Deep Sequencing (CAPP-Seq).16 In addition to immunoglobulin heavy-chain regions, CAPP-Seq detects lymphoma-relevant single nucleotide variants (SNV), insertions/deletions, and breakpoints in DLBCL-relevant genes such as BCL2, BCL6, and MYC.24 In cases where the tumor genotype was known, ctDNA profiling with CAPP-Seq detected somatic alterations in 100% of patients with 99.8% specificity. Furthermore, the majority (91%) of SNVs in driver genes could be detected in the pretreatment plasma, with a better yield in patients with higher ctDNA concentrations and advanced-stage disease. Genotypic profiling of ctDNA with CAPP-Seq identified a relevant tumor-derived biomarker in 87% of patients and was particularly successful when the mutant allele fraction was over 20%. The prognostic importance of pretreatment ctDNA levels was supported by its association with metabolic tumor volume, LDH, Ann Arbor stage, and IPI and was independently associated with progression-free survival.16

An important application of ctDNA liquid biopsies is the ability to identify mutations that predict response or resistance to targeted therapy.19,20,25 Recently, it was shown that the presence of CD79B and MYD88 mutations in DLBCL predict responsiveness to ibrutinib, whereas CARD11 mutations predict resistance.19 Additionally, the emergence of BTK mutations while on ibrutinib monotherapy is a common mechanism of resistance to ibrutinib in both chronic lymphocytic leukemia and mantle cell lymphoma.26,27 In a study by Scherer et al., ctDNA genotyping detected mutations in the BTK binding site prior to relapse in two of three patients with DLBCL who were receiving ibrutinib monotherapy.16 Indeed, such studies indicate that serial analysis of ctDNA for mutation allele frequency can detect clonal evolution and shifts in the dominant subclone (Fig. 3).2,28-30

Another recent study prospectively investigated the utility of genotyping ctDNA in the plasma of 50 patients with DLBCL uniformly treated with R-CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone).22 All patients had plasma samples drawn at baseline, on day 1 of each cycle of therapy, at treatment completion, and at disease progression. Tumor DNA extracted from diagnostic tissue biopsies was available for comparison in 36 patients. Using CAPP-Seq, the authors validated the findings of the Stanford group and showed that plasma genotyping had almost universal specificity and very high sensitivity when the circulating mutant allele fraction was over 20%. Interestingly, the authors also identified mutations in the ctDNA that were not identified in the tumor biopsy. During therapy with R-CHOP, patients whose disease responded to therapy had rapid clearance of ctDNA, while patients whose disease was resistant to therapy had detectable persistence of DLBCL mutations in the plasma. These findings further confirm the potential of ctDNA genotyping as a robust method for studying clonal evolution and characterizing treatment-resistant clones.

Challenges With ctDNA

Although ctDNA has not yet proven clinical utility, clinical validation studies are ongoing to define its role. Technical considerations such as collection, processing, and turnaround time have yet to be optimized, and research and development is needed to enhance sensitivity and specificity.31 Determining the clonotypic VDJ sequence is optimal when adequate pretreatment tissue is available, and it is dependent on quality DNA extraction, which is affected by biopsy size, tumor content, and/or tumor necrosis. Standardized collection and handling procedures of blood are needed to minimize normal DNA contamination and DNA fragmentation. Specialized collection tubes, for example, can reduce DNA degradation by nucleases and contamination from white blood cell DNA.32 Processing delay and variations in temperature of blood samples before centrifugation also affect DNA concentration.33 Finally, because low allele frequencies of ctDNA are more challenging to detect, discordant results between tissue and liquid biopsies will need to be adjudicated.34-36


ctDNA is a promising precision monitoring tool for clinical research studies and clinical management. Monitoring ctDNA for VDJ has demonstrated the ability to directly assess tumor kinetics during therapy, detect occult disease prior to imaging, and define response depth. Molecular profiling of ctDNA as a liquid biopsy will enable the study of tumor heterogeneity and clonal evolution as the backbone of precision treatment approaches. As the technology continues to evolve, it will be imperative to standardize collection and processing procedures for ctDNA to fully incorporate its clinical and research utility. Because ctDNA is noninvasive, quantifiable, and highly tumor specific, it holds great potential for transforming clinical care paradigms and trial designs in the era of precision medicine.  

About the Authors: Dr. Roschewski is a clinical investigator in the Lymphoid Malignancies Branch of the Center for Cancer Research, National Cancer Institute, National Institutes of Health. Dr. Wilson is the deputy chief of the Lymphoid Malignancies Branch of the Center for Cancer Research, National Cancer Institute, National Institutes of Health.