Molecular Monitoring of Cell-Free Circulating Tumor DNA in Non-Hodgkin Lymphoma

August 16, 2016

In this review, we discuss the potential applications of monitoring ctDNA in patients with diffuse large B-cell lymphoma, follicular lymphoma, and mantle cell lymphoma.

The ability to precisely monitor the effectiveness of therapy for non-Hodgkin lymphoma has important clinical implications. In patients with curable lymphomas, such as diffuse large B-cell lymphoma, the eradication of all disease is necessary for cure. In patients with incurable lymphomas, such as follicular lymphoma and mantle cell lymphoma, deep and durable remissions are associated with improvements in survival. Radiographic imaging modalities such as computed tomography and positron emission tomography are the current gold standard for monitoring therapy, but they are fundamentally limited by radiation risks, costs, lack of tumor specificity, and inability to detect disease at the molecular level. Novel sequencing-based methods can detect circulating tumor DNA (ctDNA) in the peripheral blood with great sensitivity, which opens new opportunities for molecular monitoring before, during, and after therapy. Beyond monitoring, ctDNA can also be used as a “liquid biopsy” to assess for molecular changes after therapy that may identify treatment-resistant clones. ctDNA is an emerging tool that may transform our ability to offer precision therapy in non-Hodgkin lymphoma.

Introduction

Non-Hodgkin lymphomas are heterogeneous lymphoid malignancies that are managed with a range of therapeutic goals. Aggressive B-cell lymphomas such as diffuse large B-cell lymphoma (DLBCL) are typically treated with curative intent, and frontline therapy can achieve a cure in up to 70% of patients.[1] In the current rituximab era, however, patients with refractory DLBCL or those who relapse after therapy are not commonly cured with standard approaches such as autologous stem cell transplant.[2] Novel therapies are emerging for DLBCL that target mechanisms of cellular resistance, but they take years to develop and benefit only subsets of patients.[3]

Incurable lymphomas, such as follicular lymphoma (FL) and mantle cell lymphoma (MCL), exhibit a perpetually relapsing disease course, and the primary goal of therapy is induction of deep and durable remissions.[4,5] Increasingly, patients with these lymphomas are treated for extended periods with maintenance therapies. Clinical decisions regarding the duration of maintenance therapy for indolent lymphomas are usually made empirically and without direct measurement of disease burden or molecular aberrations within the tumor.

The gold standard for initial staging, response assessment, and surveillance monitoring in non-Hodgkin lymphomas is assessment with imaging modalities such as computed tomography (CT) and fluorodeoxyglucose (FDG) positron emission tomography (PET).[6] Despite widespread use, imaging scans have important drawbacks, including radiation exposure, cost, and low tumor specificity.[7-10] In addition, PET scans do not perform well as surveillance tools, and there is concern regarding the use of imaging for detection of asymptomatic relapse.[11,12] PET scans have also been extensively tested as interim biomarkers for purposes of tailoring therapy. Despite initial enthusiasm, two large studies failed to demonstrate a benefit to switching therapy in patients with DLBCL on the basis of interim PET (iPET) results.[13,14]

Precision medicine approaches to non-Hodgkin lymphomas consider the molecular profile of the tumor when targeted therapies are selected. Currently, the standard method for determining tumor genetic profiles requires tissue biopsies. Tissue biopsies, however, contain procedural risks, are subject to sampling error, and cannot account for spatial tumor heterogeneity.[15] In view of these limitations, genetic profiles obtained from peripheral blood are of great interest.

Novel sequencing-based assays can detect cell-free circulating tumor DNA (ctDNA), which encodes the immunoglobulin gene sequence unique to individual B-cell lymphomas.[16-20] Assays involving ctDNA that detect somatic mutational profiles and that can serve as a “liquid biopsy” to monitor for the development of treatment-resistant clones are also being developed for non-Hodgkin lymphomas.[21] Both patients with indolent non-Hodgkin lymphomas and those with aggressive non-Hodgkin lymphomas stand to benefit from advances in technologies that more precisely define disease relapse at earlier time points. Precision molecular monitoring of ctDNA offers lower detection limits than scans and may overcome the limitations of both imaging scans and tissue biopsies (Figure 1). In this review, we discuss the potential applications of monitoring ctDNA in patients with DLBCL, FL, and MCL.

Advances in Molecular Monitoring Methods for Non-Hodgkin Lymphomas

The quest to find tumor-specific molecules in the peripheral blood to monitor patients with non-Hodgkin lymphoma dates back decades. The principal barriers to widespread use have been difficulties in identifying appropriate targets with broad clinical applicability across all non-Hodgkin lymphoma subtypes. In specific subtypes, recurrent translocations can be detected in the blood by polymerase chain reaction (PCR)-based methods (Figure 2). These include the IGH-CCND1 translocation, t(11;14), in MCL and the IGH-BCL2 translocation, t(14;18), in FL. All patients with MCL and the majority of patients with FL harbor these respective translocations, which makes them attractive targets for molecular monitoring.[22,23] Allele-specific oligonucleotide PCR (ASO-PCR) techniques use patient-specific primers to detect and quantitate BCL2 and CCND1 rearrangements, with a limit of detection approaching 1 × 105 cells. Because of variations in breakpoint regions, however, PCR is not able to detect IGH-BCL2 translocations in the blood of all patients with FL. Furthermore, most non-Hodgkin lymphoma subtypes do not have recurrent translocations amenable to molecular monitoring. Given the need for patient-specific primers and the lack of universal applicability, ASO-PCR is considered laborious and not conducive to widespread use.

Tumor-specific DNA fragments can be detected noninvasively in the blood in both cellular (circulating tumor cells [CTCs]) and cell-free (ctDNA) forms (see Figure 2). In certain lymphomas, such as DLBCL, however, CTCs are rare, and recent data indicate a possible relationship between defective homing pathways and systemic dissemination.[24-26] Multiple studies have demonstrated higher levels of circulating cell-free DNA (cfDNA) in cancer patients than in healthy controls, which result from the constant shedding of DNA fragments into the peripheral blood from cancer cells undergoing apoptosis, secretion, and necrosis.[27-29] Since cfDNA can originate from both malignant and nonmalignant sources, assays for ctDNA must accurately discriminate the tumor-specific DNA fragments within the overall cfDNA concentration.[30] ctDNA is therefore defined as the tumor-specific DNA sequences found in either the plasma or the serum of the blood and may represent as little as 0.01% of total cfDNA.[31] An early study using real-time quantitative reverse transcriptase-PCR of the germline beta-globin gene revealed significantly higher median cfDNA levels in patients with DLBCL, MCL, and Hodgkin lymphoma than in healthy controls, with similar concentrations in patients with FL.[32] This study was the first to validate ctDNA as a quantifiable tumor-specific biomarker in patients with non-Hodgkin lymphoma.

Circulating Tumor DNA in Other Malignancies

ctDNA also plays an increasingly important role in the diagnosis and monitoring of patients with solid tumors. A landmark study analyzed ctDNA in patients with numerous malignancies and found that > 75% of patients with advanced pancreatic, ovarian, colorectal, bladder, gastroesophageal, breast, hepatocellular, and head and neck cancers and melanoma had detectable levels of ctDNA, including patients with localized disease.[33] Dawson et al[34] demonstrated that analysis of plasma ctDNA for tumor-specific mutations correlated better with tumor burden and was more effective at monitoring treatment response in metastatic breast cancer than conventional imaging or levels of other circulating biomarkers, such as cancer antigen 15-3. Somatic alteration panels have also been designed and have demonstrated the ability to detect ctDNA in 100% of patients with stage II–IV non–small-cell lung cancer.[35] In addition, small series have suggested that ctDNA is more sensitive than CTCs for detection in lung cancer.[36] Other solid tumors, such as colon cancer and pancreatic cancer, have been monitored with ctDNA for the purpose of deciding which patients would benefit from adjuvant therapy; the results have been promising.[37-39] These proof-of-principle studies have established ctDNA as a highly promising biomarker with numerous potential clinical applications in multiple malignancies.

Circulating Tumor DNA in Non-Hodgkin Lymphomas

All mature B-cell lymphomas, such as DLBCL, FL, and MCL, have undergone rearrangement of their immunoglobulin receptor loci (immunoglobulin heavy-chain variable [IGHV] and immunoglobulin kappa [IGK]), which makes them attractive targets with broad applicability. Malignant B cells possess a unique DNA sequence that encodes its rearranged immunoglobulin variable, diversity, and joining (VDJ) genes. Using universal primers in combination with next-generation sequencing (NGS), this unique VDJ sequence can be utilized as a quantitative biomarker of disease. After the tumor-specific DNA sequences are identified within a baseline sample, they can be detected as ctDNA in the serum or plasma as a marker for early treatment failure or after remission as an early marker for disease recurrence.[17] This technique improves on the sensitivity of PCR-based assays and can detect up to 1 clonal molecule per 1 million (1 × 106) nonclonal diploid genomes.[40] In patients with DLBCL, such NGS-based platforms have been more successful at identifying tumor-specific sequences for detection of VDJ rearrangement in ctDNA than within CTCs.[18] These NGS-based assays for ctDNA of VDJ sequences appear to be as sensitive as real-time PCR without the need for patient-specific primers.[19]

High-throughput DNA sequencing of ctDNA encoding tumor-specific mutations also is becoming available as a liquid biopsy that can serve as a surrogate for the entire tumor genome and thus overcome the barrier of spatial and temporal tumor heterogeneity.[35] Using these novel methods, disease-specific mutations can be detected at diagnosis and followed throughout the course of the disease to quantitatively monitor tumor dynamics during treatment and to detect emergent mutations that may represent clonal evolution and/or herald resistance to targeted therapy.[41]

Clinical Applications of ctDNA

ctDNA can be used in numerous ways to improve the management and characterization of both indolent and aggressive non-Hodgkin lymphomas (Table). The application with immediate clinical implications involves replacing surveillance imaging with surveillance monitoring of ctDNA for detection of early relapse in patients with DLBCL. The lead time offered by ctDNA monitoring creates a “window of opportunity” in which earlier institution of salvage therapy can be considered (see Figure 1). Multiple other applications, however, are being studied and validated to maximize the advantages of ctDNA.

Before therapy

A significant advantage of ctDNA monitoring is the ability to measure tumor burden directly with a quantitative assay that reflects subtle changes over time. CT and PET are the current standard for defining the extent of tumor burden at diagnosis, but they are not tumor-specific and cannot be used quantitatively.

FL and MCL are both incurable with frontline therapy, and “watchful waiting” is an accepted strategy used by clinicians to delay systemic therapy in asymptomatic patients.[42-44] According to the National LymphoCare Study, 17.7% of patients with FL in the United States are treated with initial observation, and the decision to initiate therapy is related to clinical parameters such as age, Follicular Lymphoma International Prognostic Index score, and stage and grade of disease.[45] Watchful waiting strategies are not uniform, however, and it is challenging to identify patients who can safely defer initial therapy. Furthermore, the decision to transition from observation to active therapy is to a significant degree a subjective one. Evidence from analysis of the t(14;18) translocation in the blood of patients with FL suggests that levels of ctDNA rise with disease progression. In a prospective cohort study of more than 520,000 healthy participants, 168 patients were identified in whom FL later developed.[46] Compared with matched controls, the prevalence of t(14;18) positivity in the peripheral blood was higher among those in whom FL later developed. Frequencies greater than 10-4 conferred a 23× greater risk of FL development, and this remained statistically significant for up to 15 years before diagnosis. Serial monitoring of ctDNA in patients with MCL and FL who are undergoing “watchful waiting” has intuitive appeal because it is noninvasive, quantitative, and capable of directly measuring tumor-specific dynamics over time.

Pretreatment ctDNA monitoring may also capture critical biologic information. Multiple studies have shown that pretreatment ctDNA levels in patients with DLBCL correlate with markers of tumor burden such as advanced stage, lactate dehydrogenase level, and International Prognostic Index score.[17,32] It is not clear, however, whether other biologic processes affect pretreatment ctDNA levels. It is possible that indices such as tumor proliferation and/or potential for dissemination are correlated with pretreatment ctDNA levels; these questions warrant further study.

Liquid biopsies of genotypic DNA provide an average of all the genetic information within the tumor, which may complement or replace information from tissue biopsies. Investigators from the Alizadeh Laboratory at Stanford University recently reported successful genotyping of ctDNA in 45 patients with DLBCL.[47] They identified somatic mutations and/or other genetic aberrations in 87% of patients, with an average of 70 mutations per patient. Rasi et al[48] also compared targeted resequencing panels from diagnostic tissue biopsies with pretreatment plasma ctDNA in 26 patients with DLBCL. They found that the paired plasma ctDNA accurately identified 79% of the mutations from the original tissue biopsy. Importantly, plasma ctDNA was also able to detect potentially relevant mutations not identified within the original tissue biopsy. Since liquid biopsies do not carry procedural risks, they have a potential role in diagnosis even when disease involves anatomic sites that are difficult to biopsy, including the central nervous system (CNS). A recent study of patients with CNS malignancies found that 62% had detectable genomic alterations in cfDNA collected from the cerebrospinal fluid.[49] Scherer et al also recently reported three cases of isolated CNS relapse of DLBCL captured by plasma ctDNA.[47]

Quantitative pretreatment levels of ctDNA may have prognostic value. In a randomized study of frontline therapy for FL, elevated pretreatment levels of IGH-BCL2 had an adverse effect on progression-free survival (PFS) compared with intermediate or low levels (high vs intermediate: hazard ratio [HR], 4.28 [95% CI, 1.70–10.77]; P = .002; high vs low: HR, 3.02 [95% CI, 1.55–5.86]; P = .001) despite having no effect on the ability to achieve remission.[50] When NGS was used for clonal VDJ sequences, 85% of patients with FL had ctDNA identified in the diagnostic tumor. A higher level of ctDNA (> 40,000 per million diploid genomes) was associated with worse PFS, and an elevated ctDNA level was the only factor significantly associated with worse outcome on multivariate analysis.[20] Further studies should validate the prognostic information from pretreatment ctDNA compared with standard clinical scores.

During therapy (interim monitoring)

The identification early in the course of treatment of patients unlikely to be cured with initial therapy (ie, interim monitoring) is of high priority. Response-adapted strategies utilize interim biomarkers with the aim of tailoring therapy based on the result. Patients with DLBCL can be offered intensification of therapy if they have a low likelihood of cure with current therapy. At the end of therapy, patients with complete remission by conventional criteria can be further subdivided on the basis of minimal residual disease (MRD) status.

Approximately 70% of patients with DLBCL can be cured with initial therapy.[1] iPET scans have been studied extensively as biomarkers that predict treatment failure when performed after 2 or 3 cycles of therapy, with conflicting results. One study tested the prognostic value of iPET scans performed after 2 cycles of therapy in 92 patients with untreated DLBCL.[51] The investigators reported a 2-year estimated event-free survival of 51% (95% CI, 34%–68%) in patients with positive PET scans after 2 cycles of therapy compared with 79% (95% CI, 68%–90%) in those with negative scans (P = .009). A prospective study of 50 patients with DLBCL, however, demonstrated that iPET scans performed after 2 or 3 cycles of therapy had a positive predictive value (PPV) of only 41% for prediction of relapse or progression.[52] Moskowitz et al[9] observed that 33 of 38 patients (87.8%) with positive iPET scans did not have active tumor tissue on biopsy.

Two recent studies that tested response-adapted strategies for DLBCL found no benefit to switching therapy on the basis of iPET results. In a large German trial of 757 patients with aggressive B-cell lymphomas, iPET results were unfavorable in 13% and were highly predictive of outcome. However, switching therapy to a more intensive regimen did not affect the incidence of treatment failure.[13] In a smaller study, patients with DLBCL underwent iPET scanning after 4 cycles of therapy and were recommended to switch to a more intensive regimen if iPET results were positive.[14] The majority of patients in whom therapy was altered on the basis of iPET results failed to achieve remission at the end of therapy.

Interim monitoring of ctDNA during therapy may also predict treatment outcomes and could improve response-adapted strategies if combined with PET. In the National Cancer Institute study of ctDNA for DLBCL, serum samples were available before therapy and at the beginning of each cycle of therapy.[17] The 5-year time to progression was 41% (95% CI, 22.2%–60.1%) in patients who had positive interim ctDNA results after 2 cycles of therapy compared with 80.2% (95% CI, 69.6%–87.3%) in those who cleared ctDNA after 2 cycles. Interim ctDNA results after 2 cycles had a PPV of 62.5% (95% CI, 40.6%–81.2%) and a negative predictive value of 79.8% (95% CI, 69.6%–87.8%) for prediction of treatment failure. When interim ctDNA results were analyzed qualitatively in patients with early disease relapse, three major patterns of clearance during therapy were observed:

• Patients who never cleared ctDNA.

• Patients with transient clearance.

• Patients with full clearance.

Those who never cleared ctDNA before clinical progression had a median time to progression of only 3.8 months.[17] Researchers at Stanford University reported a novel mathematical approach to ctDNA clearance during therapy that may better predict outcomes in DLBCL.[53] Interim ctDNA monitoring quantitatively measures tumor response dynamics in real time and identifies patients at high risk for treatment failure.

Another important application of interim monitoring could be as an early marker of resistance to targeted therapy. A recent study of panobinostat in relapsed DLBCL and transformed lymphoma demonstrated that ctDNA levels at day 15 were predictive of response to this novel targeted therapy.[54] These exciting results raise the possibility that ctDNA could serve as a surrogate endpoint in clinical trials that test novel agents in order to nominate the most active agents.

The depth of response is an important predictor of outcome in all non-Hodgkin lymphoma subtypes. A complete molecular remission (CMR) is not part of the current response criteria, but two studies in FL and MCL have both demonstrated that achievement of CMR is an important endpoint after therapy.[6,55,56] Relapsed non-Hodgkin lymphoma likely originates from MRD below the current level of detection, and a recent systematic review showed that between 7% and 20% of patients with DLBCL in remission as demonstrated by PET scans will ultimately relapse.[57] Given its impact on clinical outcome and risk of subsequent relapse, the depth of remission as ascertained by means of MRD evaluation following therapy should be incorporated into current response criteria to identify high-risk patients who may benefit from further treatment intensification or maintenance therapy.

In MCL, suitable targets for MRD assessment include clonal IGH rearrangements as well as the characteristic t(11;14) translocation, which leads to the BCL1-IGH fusion gene and CCND1 protein overexpression. Molecular remission has been shown to be prognostic of outcome following immunochemotherapy with or without autologous stem cell transplant. In both the European MCL Intergroup study and the Cancer and Leukemia Group B 59909 study, MRD status following induction therapy was prognostic of outcome. In both studies, the achievement of CMR was a better predictor of outcome than clinical response, independent of other prognostic factors.[55,56] Detection of ctDNA during and following induction therapy could help tailor postinduction therapy based on the risk of subsequent relapse: high-risk patients would receive further treatment intensification while low-risk patients would enter active surveillance.

After initial therapy (surveillance)

A large retrospective trial in DLBCL found no improvement in survival with the use of routine surveillance CT scans after therapy compared with clinical evaluation alone.[10] However, no randomized studies have been performed, and clinical judgment is the main guiding principle in deciding whether to use surveillance imaging.[58] Early detection of relapsed disease in patients with aggressive lymphomas provides an opportunity to implement potentially lifesaving salvage treatment during a time when there is an overall low burden of disease.

A landmark study of 126 patients with DLBCL demonstrated that ctDNA encoding VDJ sequences outperforms CT scans at detecting disease relapse during surveillance monitoring.[17] All patients were treated with combination chemotherapy, and long-term follow-up was available, with a median of 11 years (interquartile range, 6.8–14.2 years). Patients were monitored with CT scans at prespecified intervals, along with acquisition of a paired research serum sample. Genomic DNA was extracted from the serum and assessed for the presence of ctDNA; the assessment was blinded to clinical outcomes. The lead time between detection of ctDNA and clinical detection of relapse was a median of 3.5 months (range, 0–200 months). In the 11 patients who relapsed more than 6 months after the completion of therapy, ctDNA was associated with an even longer lead time. The median lead time in these patients was 7.4 months before the clinical detection of relapse, and all but one patient (91%) had positive ctDNA assay results before relapse.

A separate study from Stanford University corroborated these results in a slightly smaller cohort of patients with DLBCL.[18] Of 75 patients with DLBCL treated with combination chemotherapy, 25 had samples available during surveillance and 5 of these patients eventually relapsed. ctDNA was detectable in the plasma of 3 (60%) of these patients, for a median of 88 days (range, 14–162 days) before clinical relapse, and 2 patients had detectable ctDNA at the time of relapse, for a specificity of 100%. Imaging with PET/CT scans in the same group of patients yielded a specificity of only 56%.[18] These studies clearly show that in patients with DLBCL, serial surveillance monitoring of ctDNA consistently outperforms imaging scans at identifying recurrent disease before clinical relapse.

FL and MCL are frequently treated for extended durations with maintenance therapies, and serial ctDNA monitoring can be used following induction therapy to help guide postinduction treatment decisions. Successful maintenance therapy could be monitored with ctDNA and continued as long as disease remains undetectable. Patients in whom ctDNA reappears during maintenance therapy might be considered for alternative therapy before overt clinical relapse. Patients who are not initially treated with maintenance therapy can be offered “delayed maintenance” at a time when disease is detectable by ctDNA monitoring but is not yet detectable by imaging scans.

Precision monitoring of ctDNA after therapy could be a foundational component of precision medicine strategies for non-Hodgkin lymphomas. Individual mutations or panels of mutations may be able to predict responsiveness to targeted therapies designed to overcome known mechanisms of treatment resistance.[41] A recent study of ibrutinib in DLBCL demonstrated that the presence of both MYD88 and CD79B mutations was highly predictive of responsiveness.[59] Although not ready for clinical use, these types of precision medicine efforts are of great research interest, and clinical trial designs are likely to explore ctDNA as an accessible source of genetic information throughout the course of disease.

Conclusions

ctDNA has numerous advantages over imaging scans and tissue biopsies that may ultimately transform our ability to characterize and monitor all subtypes of non-Hodgkin lymphoma. Immunoglobulin receptor gene sequences can be detected as ctDNA in patients with DLBCL months before PET or CT scans can detect disease, which provides a “window of opportunity” for early treatment of relapsed disease. Liquid biopsies can identify somatic mutations of the tumor and promise to overcome traditional barriers to obtaining genetic information. Improvements in methodologies will advance the detection capability of ctDNA assays and better discriminate the small fraction of ctDNA within the peripheral blood.[60] Multiple other clinical applications of ctDNA are in development and promise to provide noninvasive methods for achieving the coveted goal of precision treatment of non-Hodgkin lymphoma.

Prospective studies are needed to further define the role of ctDNA in clinical practice. Paired analysis with PET/CT during interim monitoring should be performed to determine whether complementary testing improves clinical outcomes through response-adapted strategies. Noninvasive genotypic analysis should be used to better define mechanisms of resistance and to test novel therapies using a precision-directed approach. Regardless of what future studies may demonstrate, ctDNA analysis has already shown great promise in patients with non-Hodgkin lymphoma. Refinement of this technology will further enhance our ability to care for these patients.

Financial Disclosure:The authors have no significant financial interest in or other relationship with the manufacturer of any product or provider of any service mentioned in this article.

Acknowledgments:The authors acknowledge support from the intramural research program of the National Institutes of Health, especially Wyndham Wilson, Kieron Dunleavy, and the rest of the research team.

References:

1. Roschewski M, Dunleavy K, Wilson WH. Moving beyond rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone for diffuse large B-cell lymphoma. Leuk Lymphoma. 2014;55:2428-37.

2. Gisselbrecht C, Glass B, Mounier N, et al. Salvage regimens with autologous transplantation for relapsed large B-cell lymphoma in the rituximab era. J Clin Oncol. 2010;28:4184-90.

3. Roschewski M, Staudt LM, Wilson WH. Diffuse large B-cell lymphoma-treatment approaches in the molecular era. Nat Rev Clin Oncol. 2014;11:12-23.

4. Kahl BS, Yang DT. Follicular lymphoma: evolving therapeutic strategies. Blood. 2016;127:2055-63.

5. Perez-Galan P, Dreyling M, Wiestner A. Mantle cell lymphoma: biology, pathogenesis, and the molecular basis of treatment in the genomic era. Blood. 2011;117:26-38.

6. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32:3059-68.

7. Brenner DJ, Hall EJ. Computed tomography-an increasing source of radiation exposure. N Engl J Med. 2007;357:2277-84.

8. Huntington SF, Svoboda J, Doshi JA. Cost-effectiveness analysis of routine surveillance imaging of patients with diffuse large B-cell lymphoma in first remission. J Clin Oncol. 2015;33:1467-74.

9. Moskowitz CH, Schoder H, Teruya-Feldstein J, et al. Risk-adapted dose-dense immunochemotherapy determined by interim FDG-PET in advanced-stage diffuse large B-cell lymphoma. J Clin Oncol. 2010;28:1896-903.

10. Thompson CA, Ghesquieres H, Maurer MJ, et al. Utility of routine post-therapy surveillance imaging in diffuse large B-cell lymphoma. J Clin Oncol. 2014;32:3506-12.

11. Cheah CY, Hofman MS, Dickinson M, et al. Limited role for surveillance PET-CT scanning in patients with diffuse large B-cell lymphoma in complete metabolic remission following primary therapy. Br J Cancer. 2013;109:312-7.

12. Armitage JO, Vose JM. To surveil or not to surveil. J Clin Oncol. 2015;33:3983-4.

13. Duehrsen U, Hüttmann A, Müller S, et al. Positron emission tomography (PET) guided therapy of aggressive lymphomas-a randomized controlled trial comparing different treatment approaches based on interim PET results (PETAL trial). Blood (ASH Annual Meeting Abstracts). 2014;124:abstr 391.

14. Sehn LH, Hardy ELG, Gill KK, et al. Phase 2 trial of interim PET scan-tailored therapy in patients with advanced stage diffuse large B-cell lymphoma (DLBCL) in British Columbia (BC). Blood (ASH Annual Meeting Abstracts). 2014;124:abstr 392.

15. Alizadeh AA, Aranda V, Bardelli A, et al. Toward understanding and exploiting tumor heterogeneity. Nat Med. 2015;21:846-53.

16. Armand P, Oki Y, Neuberg DS, et al. Detection of circulating tumour DNA in patients with aggressive B-cell non-Hodgkin lymphoma. Br J Haematol. 2013;163:123-6.

17. Roschewski M, Dunleavy K, Pittaluga S, et al. Circulating tumour DNA and CT monitoring in patients with untreated diffuse large B-cell lymphoma: a correlative biomarker study. Lancet Oncol. 2015;16:541-9.

18. Kurtz DM, Green MR, Bratman SV, et al. Noninvasive monitoring of diffuse large B-cell lymphoma by immunoglobulin high-throughput sequencing. Blood. 2015;125:3679-87.

19. Ladetto M, Bruggemann M, Monitillo L, et al. Next-generation sequencing and real-time quantitative PCR for minimal residual disease detection in B-cell disorders. Leukemia. 2014;28:1299-307.

20. Sarkozy C, Huet S, Carlton V, et al. Quantitative assessment of circulating clonal IG-VDJ sequences in plasma of follicular lymphoma at diagnosis is highly predictive of progression free survival (PFS). Blood (ASH Annual Meeting Abstracts). 2015;126:abstr 2675.

21. Diaz LA Jr, Bardelli A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol. 2014;32:579-86.

22. Raffeld M, Jaffe ES. Bcl-1, t(11;14), and mantle cell-derived lymphomas. Blood. 1991;78:259-63.

23. Weinberg OK, Ai WZ, Mariappan MR, et al. “Minor” BCL2 breakpoints in follicular lymphoma: frequency and correlation with grade and disease presentation in 236 cases. J Mol Diagn. 2007;9:530-7.

24. Muppidi JR, Schmitz R, Green JA, et al. Loss of signalling via Gα13 in germinal centre B-cell-derived lymphoma. Nature. 2014;516:254-8.

25. Muringampurath-John D, Jaye DL, Flowers CR, et al. Characteristics and outcomes of diffuse large B-cell lymphoma presenting in leukaemic phase. Br J Haematol. 2012;158:608-14.

26. Mancuso P, Calleri A, Antoniotti P, et al. If it is in the marrow, is it also in the blood? An analysis of 1,000 paired samples from patients with B-cell non-Hodgkin lymphoma. BMC Cancer. 2010;10:644.

27. Jahr S, Hentze H, Englisch S, et al. DNA fragments in the blood plasma of cancer patients: quantitations and evidence for their origin from apoptotic and necrotic cells. Cancer Res. 2001;61:1659-65.

28. Leon SA, Shapiro B, Sklaroff DM, Yaros MJ. Free DNA in the serum of cancer patients and the effect of therapy. Cancer Res. 1977;37:646-50.

29. Fleischhacker M, Schmidt B. Circulating nucleic acids (CNAs) and cancer-a survey. Biochim Biophys Acta. 2007;1775:181-232.

30. Mandel P, Metais P. Les acides nucléiques du plasma sanguin chez l’homme. C R Seances Soc Biol Fil. 1948;142:241-3.

31. Schwarzenbach H, Hoon DS, Pantel K. Cell-free nucleic acids as biomarkers in cancer patients. Nat Rev Cancer. 2011;11:426-37.

32. Hohaus S, Giachelia M, Massini G, et al. Cell-free circulating DNA in Hodgkin’s and non-Hodgkin’s lymphomas. Ann Oncol. 2009;20:1408-13.

33. Bettegowda C, Sausen M, Leary RJ, et al. Detection of circulating tumor DNA in early- and late-stage human malignancies. Sci Transl Med. 2014;6:224ra24.

34. Dawson SJ, Tsui DW, Murtaza M, et al. Analysis of circulating tumor DNA to monitor metastatic breast cancer. N Engl J Med. 2013;368:1199-209.

35. Newman AM, Bratman SV, To J, et al. An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage. Nat Med. 2014;20:548-54.

36. Guibert N, Pradines A, Casanova A, et al. Detection and monitoring of the BRAF mutation in circulating tumor cells and circulating tumor DNA in BRAF-mutated lung adenocarcinoma. J Thorac Oncol. 2016 May 7. [Epub ahead of print]

37. Diehl F, Schmidt K, Choti MA, et al. Circulating mutant DNA to assess tumor dynamics. Nat Med. 2008;14:985-90.

38. Sausen M, Phallen J, Adleff V, et al. Clinical implications of genomic alterations in the tumour and circulation of pancreatic cancer patients. Nat Commun. 2015;6:7686.

39. Tie J, Wang Y, Tomasetti C, et al. Circulating tumor DNA analysis detects minimal residual disease and predicts recurrence in patients with stage II colon cancer. Sci Transl Med. 2016;8:346ra92.

40. Faham M, Zheng J, Moorhead M, et al. Deep-sequencing approach for minimal residual disease detection in acute lymphoblastic leukemia. Blood. 2012;120:5173-80.

41. Roschewski M, Staudt LM, Wilson WH. Dynamic monitoring of circulating tumor DNA in non-Hodgkin lymphoma. Blood. 2016;127:3127-32.

42. Martin P, Chadburn A, Christos P, et al. Outcome of deferred initial therapy in mantle-cell lymphoma. J Clin Oncol. 2009;27:1209-13.

43. Armitage JO, Longo DL. Is watch and wait still acceptable for patients with low-grade follicular lymphoma? Blood. 2016;127:2804-8.

44. Portlock CS, Rosenberg SA. No initial therapy for stage III and IV non-Hodgkin’s lymphomas of favorable histologic types. Ann Intern Med. 1979;90:10-3.

45. Friedberg JW, Taylor MD, Cerhan JR, et al. Follicular lymphoma in the United States: first report of the national LymphoCare study. J Clin Oncol. 2009;27:1202-8.

46. Roulland S, Kelly RS, Morgado E, et al. t(14;18) translocation: a predictive blood biomarker for follicular lymphoma. J Clin Oncol. 2014;32:1347-55.

47. Scherer F, Kurtz DM, Newman AM, et al. Noninvasive genotyping and assessment of treatment response in diffuse large B cell lymphoma. Blood (ASH Annual Meeting Abstracts). 2015;126:abstr 114.

48. Rasi S, Monti S, Zanni M, et al. Liquid biopsy as a tool for monitoring the genotype of diffuse large B-cell lymphoma. Blood (ASH Annual Meeting Abstracts). 2015;126:abstr 127.

49. Pentsova EI, Shah RH, Tang J, et al. Evaluating cancer of the central nervous system through next-generation sequencing of cerebrospinal fluid. J Clin Oncol. 2016 May 9. [Epub ahead of print]

50. Zohren F, Bruns I, Pechtel S, et al. Prognostic value of circulating Bcl-2/IgH levels in patients with follicular lymphoma receiving first-line immunochemotherapy. Blood. 2015;126:1407-14.

51. Lin C, Itti E, Haioun C, et al. Early 18F-FDG PET for prediction of prognosis in patients with diffuse large B-cell lymphoma: SUV-based assessment versus visual analysis. J Nucl Med. 2007;48:1626-32.

52. Cashen AF, Dehdashti F, Luo J, et al. 18F-FDG PET/CT for early response assessment in diffuse large B-cell lymphoma: poor predictive value of international harmonization project interpretation. J Nucl Med. 2011;52:386-92.

53. Kurtz DM, Scherer F, Newman AM, et al. Dynamic noninvasive genomic monitoring for outcome prediction in diffuse large B-cell lymphoma. Blood (ASH Annual Meeting Abstracts). 2015;126:abstr 130.

54. Assouline SE, Nielsen TH, Yu S, et al. Phase 2 study of panobinostat +/- rituximab in relapsed diffuse large B cell lymphoma and biomarkers predictive of response. Blood. 2016 May 10. [Epub ahead of print]

55. Liu H, Johnson JL, Koval G, et al. Detection of minimal residual disease following induction immunochemotherapy predicts progression free survival in mantle cell lymphoma: final results of CALGB 59909. Haematologica. 2012;97:579-85.

56. Pott C, Hoster E, Delfau-Larue MH, et al. Molecular remission is an independent predictor of clinical outcome in patients with mantle cell lymphoma after combined immunochemotherapy: a European MCL Intergroup study. Blood. 2010;115:3215-23.

57. Adams HJ, Nievelstein RA, Kwee TC. Prognostic value of complete remission status at end-of-treatment FDG-PET in R-CHOP-treated diffuse large B-cell lymphoma: systematic review and meta-analysis. Br J Haematol. 2015;170:185-91.

58. Nabhan C, Smith SM, Cifu AS. Surveillance imaging in patients in remission from Hodgkin and diffuse large B-cell lymphoma. JAMA. 2016;315:2115-6.

59. Wilson WH, Young RM, Schmitz R, et al. Targeting B cell receptor signaling with ibrutinib in diffuse large B cell lymphoma. Nat Med. 2015;21:922-6.

60. Newman AM, Lovejoy AF, Klass DM, et al. Integrated digital error suppression for improved detection of circulating tumor DNA. Nat Biotechnol. 2016;34:547-55.