Determining Prognosis in Aggressive Lymphomas: Integrating Liquid Biopsy Into Imaging Assessment

Get Permission

The incorporation of blood-based measurements—ie, “liquid biopsies”—into imaging assessment may refine the accuracy of prognosis in aggressive lymphomas, as described by David Kurtz, MD, PhD, Assistant Professor in the Division of Oncology, Stanford University Medical Center, in a talk at the 2022 Pan Pacific Lymphoma Conference in Kauai, Hawaii.1

In particular, circulating tumor DNA (ctDNA)—a method to detect tumor DNA molecules from the blood plasma—has emerged as a useful biomarker across many cancer types, including lymphoma. These liquid biopsies can measure tumor burden, assess treatment response, and track the earliest traces of cancer leading to relapse, including genetic features that may predict eventual relapse, he said.

“Liquid biopsies are not, however, going to immediately replace our PET/CT [positron-emission tomography/computed tomography] scans…. PET/CT scans and ctDNA can actually augment each other,” he said.

Accuracy of ctDNA

Levels of ctDNA in previously untreated patients correlate highly with key clinical indices, including the International Prognostic Index (IPI) and metabolic tumor volume from PET scans. Perhaps more importantly, ctDNA has been shown to give a more objective measure of disease burden than a measure that has been gaining attention, the “diagnosis-to-treatment interval.” Paradoxically, in multiple recent studies, patients who have a longer time between diagnosis and initial treatment of their lymphoma have better outcomes: this indicates that patients deemed by their providers to need prompt treatment have worse outcomes.2

“Liquid biopsies are not … going to immediately replace our PET/CT scans…. PET/CT scans and ctDNA can actually augment each other.”
— David Kurtz, MD, PhD

Tweet this quote

“This is a major problem in clinical trial development, because many trials do not capture patients with shorter [diagnosis-to-treatment interval]—meaning trials might not capture a real-world population of people with lymphoma,” said Dr. Kurtz. “However, [diagnosis-to-treatment interval] is impacted by many factors, including patient and provider preferences as well as available resources, so ctDNA is a more objective measurement. This gives us a way to truly compare if patients in clinical studies are similar to patients we see on a day-to-day basis.”

Dr. Kurtz concluded that ctDNA could have immediate clinical utility in quantifying disease burden, improving staging and pretreatment risk stratification, and guarding against selection biases.

CAPP-Seq and ctDNA for Response Assessment

Multiple prior studies have shown the prognostic significance of an interim ctDNA during treatment for diffuse large B-cell and other lymphomas: change in ctDNA at interim time points is highly prognostic for eventual outcome. One means of assessing ctDNA is via Cancer Personalized Profiling by Deep Sequencing (CAPP-Seq), which finds and tracks cancer-related mutations in the blood, thereby providing sensitive detection of ctDNA.

Dr. Kurtz and the Stanford group have shown that dynamic changes in ctDNA on CAPP-Seq can stratify patients who achieve a complete response vs those who do not.3 Studying a cohort of over 200 patients, the researchers observed that during the first few cycles of immunochemotherapy, ctDNA levels changed rapidly, with a 2-log decrease after one cycle (labeled an “early molecular response”) and a 2.5-log decrease after two cycles (“major molecular response”) predicting better response to therapy. This could stratify patient outcomes, including best response to treatment and event-free survival after front-line therapy. The data from this and subsequent studies have suggested, he said, “that a rapid decline in ctDNA is associated with favorable outcomes.”

The potential for ctDNA in assessing response is also being shown with chimeric antigen receptor (CAR) T-cell therapy. According to a 2021 study, 71% of patients with disease progression after axicabtagene ciloleucel infusion had detectable ctDNA at day 28 as measured by immunoglobulin sequencing.4

Newer Test: PhasED-Seq

Although these initial results have substantial promise, there are still significant blind spots for current ctDNA technologies. In fact, Dr. Kurtz showed, about 50% of patients destined for treatment failure actually have undetectable ctDNA after two cycles. “We are not capturing all patients who go on to have relapses,” he acknowledged.

To address these problems, more sensitive assays are in development. One such test is Phased Variant Enrichment and Detection by Sequencing, or PhasED-Seq.5 By detecting a special class of alterations called “phased variants”—or multiple mutations on a single DNA strand—PhasED-Seq is 10 to 100 times more sensitive than current ctDNA assays. “This increased analytic sensitivity directly leads to increased clinical sensitivity,” he said.

Dr. Kurtz showed how PhasED-Seq could help in the management of patients by describing the clinical case of a patient with stage IV diffuse large B-cell lymphoma who was in remission after treatment but relapsed 450 days later. CAPP-Seq could detect residual disease about 150 days before relapse but failed to detect disease at the end of standard treatment, when additional therapy could be considered. In contrast, the newer PhasED-Seq technology never lost a handle on the cancer, detecting disease at the key end-of-treatment landmark: over 300 days prior to relapse. “This is the point when we may want to make a treatment decision, possibly adding consolidated treatment at the end of therapy,” he said.

Combining ctDNA, Imaging Responses, and Continuous Individualized Risk Index

Rather than supplanting PET/CT imaging, ctDNA and PET scans can actually augment each other, Dr. Kurtz continued. In his 2018 study in the Journal of Clinical Oncology,3 he showed that these two modalities could be independently prognostic. Perhaps most significantly, the group of patients with residual disease on both PET scan and ctDNA had the worst outcomes—their complete response rate was 0%, and 24-month event-free survival rate was only 11%.

“When combined, PET imaging and molecular response identifies a group of patients [in whom front-line therapy is destined to fail],” he said. “The modalities compensate for each other’s ‘blind spots’ and augment each other.”

One of the largest unanswered questions, however, is how to integrate these risk factors into a single, clinically useful tool. Prior risk-assessment methods use only one factor and provide a snapshot in time. “Serial testing is a major advantage of ctDNA compared to other biomarkers, but it is still unclear how to best utilize it,” Dr. Kurtz noted.

The optimal use of ctDNA may be as a component of what Dr. Kurtz and his team have labeled the Continuous Individualized Risk Index. This is a more personalized and presumably more accurate approach to risk assessment that integrates pretreatment risk factors (IPI, subtype, and ctDNA level) and dynamic risk factors (early and major molecular responses and interim PET scan) measured over time.

As the authors described,6 the Continuous Individualized Risk Index is “a method to dynamically determine outcome probabilities for individual patients utilizing risk predictors acquired over time…. Similar to ‘win probability’ models in other fields, [the Continuous Individualized Risk Index] provides a real-time probability by integrating risk assessments throughout a patient’s course.”

Applying the Continuous Individualized Risk Index to patients with diffuse large B-cell lymphoma, Dr. Kurtz and colleagues demonstrated improved outcome prediction compared to conventional risk models.6 He described how the Continuous Individualized Risk Index could be used to generate personalized prognostic models in two patients with diffuse large B-cell lymphoma sharing the exact same set of risk factors prior to treatment, making robust and accurate predictions with only a few additional pieces of interim data, such as interim PET and molecular response assessment. The Continuous Individualized Risk Index prediction model can be obtained online (

“Ultimately, we envision that such dynamic risk assessment will facilitate personalized medicine and enable innovative therapeutic paradigms,” Dr. Kurtz said. 

DISCLOSURE: Dr. Kurtz has consulted for Roche, Genentech, Adaptive Biotechnologies, and Foresight Diagnostics; owns stock in Foresight; and is a patent holder for a serologic assay.


1. Kurtz DM: Incorporating serologic response measures into imaging modalities. 2022 Pan Pacific Lymphoma Conference. Presented July 18, 2022.

2. Alig S, Macaulay CW, Kurtz DM, et al: Short diagnosis-to-treatment interval is associated with higher circulating tumor DNA levels in diffuse large B-cell lymphoma. J Clin Oncol 39:2605-2616, 2021.

3. Kurtz DM, Scherer F, Jin MC, et al: Circulating tumor DNA measurements as early outcome predictors in diffuse large B-cell lymphoma. J Clin Oncol 36: 2845-2853, 2018.

4. Frank MJ, Hossain NM, Bukhari A, et al: Monitoring of circulating tumor DNA improves early relapse detection after axicabtagene ciloleucel infusion in large B-cell lymphoma: Results of a prospective multi-institutional trial. J Clin Oncol 39:3034-3043, 2021.

5. Kurtz DM, Soo J, Co Ting Keh L, et al: Enhanced detection of minimal residual disease by targeted sequencing of phased variants in circulating tumor DNA. Nat Biotechnol 39:1537-1547, 2021.

6. Kurtz DM, Esfahani MS, Scherer F, et al: Dynamic risk profiling using serial tumor biomarkers for personalized outcome prediction. Cell 178:669-713, 2019.