Current approaches to assess response to therapy and disease recurrence in diffuse large B-cell lymphoma (DLBCL) have many limitations. The detection of minimal residual disease by positron emission tomography (PET) and/or computed tomography (CT) requires a sufficient volume of disease, and false positives are common. Frequent imaging is very costly and is associated with exposure to ionizing radiation and the risk of secondary malignancies. Identifying minimal persistent lymphoma or early recurrence using assays for disease-specific circulating tumor DNA may offer a safe and effective means of disease assessment.
In their recent study published in Lancet Oncology, Dr. Mark Roschewski and colleagues retrospectively evaluated the utility of circulating DNA to identify persistent and recurrent disease in patients with DLBCL. Previously untreated patients with DLBCL, treated on three clinical trials using EPOCH (etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin), with or without rituximab, underwent CT imaging with concurrent serum collection at baseline, during therapy, and at the end of chemotherapy. Surveillance was performed every three, four, and six months during years one, two, and three, respectively, then annually during years four and five. Available pretreatment formalin-fixed paraffin-embedded biopsy specimens were analyzed for tumor-specific immunoglobulin gene rearrangements using next-generation sequencing of VDJ gene sequences, which were then quantified. For patients without available tissue, circulating tumor DNA was isolated from serum.
One hundred twenty-six patients had identifiable tumor specific clonotypes, with 94 isolated from 109 patients with available diagnostic tissue, and 32 isolated from 86 patients with banked serum specimens. The concentration of baseline circulating tumor DNA correlated with tumor burden as reflected by the International Prognostic Index, stage, and lactate dehydrogenase. After two cycles of therapy, 108 patients underwent interim assessment of circulating DNA. The five-year progression-free survival of patients with detectable versus undetectable DNA was 42 percent versus 80 percent, yielding a positive predictive value of 63 percent and a negative predictive value of 80 percent.
Thirty-six patients experienced disease progression. Twenty-five patients relapsed early, at less than six months from treatment, and 11 patients relapsed late. Of the patients with early progression, those who recurred earliest did not clear circulating tumor specific DNA. At the completion of frontline chemotherapy, none of those with late relapse had measurable tumor DNA. Ten of 11 patients subsequently developed detectable levels—-eight had no concurrent clinical evidence of disease. Seven patients who had clinical progression did not have measurable circulating tumor DNA—four had localized mediastinal lymphoma with a small area of residual disease by PET, and one had an isolated central nervous system disease. Of the 98 patients who remained in remission, only two had low-level detectable circulating tumor DNA measurable on one assessment, which was not reproduced. Overall, the positive predictive values and sensitivity of circulating tumor DNA were both 88 percent, and the negative predictive value and specificity were 98 percent. Circulating tumor DNA was detectable at a median of 3.5 months prior to the appearance of clinical disease and was earlier for patients with primary refractory disease or early recurrence.
In Brief
Monitoring circulating tumor DNA in patients with DLBCL has a number of important advantages over imaging-based strategies, including accuracy, cost, and safety. Although no clear evidence to date demonstrates an improvement in survival for patients treated for asymptomatic relapse detected on imaging, many patients and providers are uncomfortable with clinical follow-up exclusively. Additionally, following serum disease–specific DNA has great potential for application in the design of novel clinical trials. PET-driven risk-adapted studies in DLBCL have been difficult to interpret given the low positive predictive value of interim scans. Using slow clearance of circulating tumor DNA to select those patients for whom to intensify therapy, while reducing treatment in early complete responders, may improve outcomes and limit toxicity. Lastly, using serum tumor DNA as a predictive biomarker may lead to more rapid assessment of the efficacy of novel agents and combinations.
Competing Interests
Dr. LaCasce indicated no relevant conflicts of interest.
