Detecting Residual Disease in Lymphoma: Current and Emerging Tools

Overview: In this discussion, Dr David Russler-Germain explains how circulating tumor DNA (ctDNA)–based minimal residual disease (MRD) testing is emerging as a more sensitive and accurate tool than PET/CT imaging for predicting relapse risk and assessing remission in diffuse large B-cell lymphoma (DLBCL).

David Russler-Germain, MD, PhD: Using diffuse large B-cell lymphoma, or DLBCL, as the predominant and most common aggressive lymphoma that we treat with curative intent, we use tools such as PET scans to assess patients' end-of-treatment response in order to actually estimate whether they're likely to be cured or not. The goal of frontline treatment is to eradicate the disease, have patients achieve a complete metabolic response by PET scan, and move on to a surveillance period where they're off therapy. And hopefully in years we can call that remission a cure. Unfortunately, roughly 40% of patients with DLBCL relapse after frontline therapy. And albeit PET scans are better than CAT scans, we are still unable to optimally identify which patients are at the highest risk of recurrence. 

PET/CT imaging is a dual modality imaging assessment, which both provides anatomic measurements of disease burden using the CT parameters of measuring how large by one centimeter, two centimeters, et cetera, a residual lesion is, but also has a functional component. So the 18FDG is a radiotracer sugar equivalent that is taken up by highly metabolically active tissues, namely cancers, for example, that allow us to assess not only how much a disease has shrunk anatomically by the CT component of the imaging, but whether it's "cold" or dead even based on the PET imaging. 

We use what's called the Deauville five-point scale, which intuitively ranges from one to five to semi-quantitatively adjudicate whether patients are in remission or not, with scores of one through three being complete metabolic response and scores of four or five being residual disease. Unfortunately, it's a little bit in the eye of the beholder as PET scan technologies evolve and are getting more sensitive. But unfortunately not reaching quite the level of sensitivity where complete metabolic responses are definitively able to prognosticate a cure. 

As great as our modern PET/CT imaging technologies are, they still only detect disease down to the two- to three-cubic millimeter resolution. And while certainly that is extremely helpful in adjudicating whether patients have reasonably responded to disease or not, several studies have shown that patients even with definitive complete metabolic responses have a high recurrence rate upwards of 10 to 20%. And conversely, patients who have residual disease seen on PET scan still may have even a 50% or greater chance of being cured because of the risk of false positive signals by PET scan. And so when you factor in that there are patients with underlying low-grade lymphomas, non-nodal tissues such as the nervous system, GU system of especially testicle, as well as gastric lymphomas are very tricky to discern on PET/CTs, improving upon our PET response assessments is a really crucial aspect in the field. 

Perhaps the most robust data we have regarding the prognostic capacity of end-of-treatment PET/CT scans comes from the randomized phase three GOYA study comparing rituximab plus CHOP versus obinutuzumab plus CHOP in frontline DLBCL. And in this cohort of over a thousand patients, several notable observations could be made. First was the roughly 20% false negative end-of-treatment PET rate. In essence, patients called in a complete response still had a one in five chance of subsequent disease recurrence within three years. And secondarily, patients with end-of-treatment positive PETs had roughly a 50% chance of actually being cured, in essence, never having their disease come back despite not being in complete response at the end of treatment. 

And so, multiple attempts have been made to improve upon PET scans as far as our end-of-treatment adjudication goes. Some studies have looked at whether adding bone marrow biopsies are necessary or not to look for actual residual cells. And that has not panned out to be a modern standard of care. Patient people have also thought about flow cytometry-based assays, both in the aforementioned bone marrow compartment as well as in peripheral blood. But as everyone knows, DLBCL is predominantly not a peripheral blood circulating disease. And that's why in the era of next-generation sequencing and quantitative PCR, multiple non-invasive blood-based biomarkers looking for genomic alterations inherent to the disease cells are proving to be dramatically, perhaps more useful than end-of-treatment PET scans. 

In the last 10 to 15 years, there have been immense strides made in using molecular biology techniques to track residual disease in DLBCL, albeit predominantly on the research side and only recently starting to enter the clinic. Early on, as we know from multiple other diseases, the malignant B-cells in diffuse large B-cell lymphoma almost always have a canonical immunoglobulin rearrangement that could be trackable using either PCR or sequencing-based technologies. And so, several assays have looked both in the circulating cell component as well as in more recently cell-free DNA or circulating tumor DNA compartment in the plasma for these prototypical or sort of barcoded immunoglobulin heavy or light chain rearrangements for tracking residual malignant B-cells, often to the level of one in a hundred thousand circulating cells. 

More recently, with the advent of next-generation sequencing on a broad scale, such as whole-exome and whole-genome sequencing, we now have a better understanding of the molecular profile and commonly recurrent mutations in diffuse large B-cell lymphoma, as well as others. Allowing the development of targeted sequencing panels to identify sort of trackable variants from a patient's baseline plasma sample from ctDNA or from their actual tumor tissue. And then those same mutations can then be looked for in a subsequent analyte such as the plasma at mid-treatment or end of treatment for an MRD or measurable residual disease assessment. 

Each of the seminal ctDNA sequencing studies described in 2025 regarding the performance of ctDNA MRD testing at end of treatment in frontline DLBCL management showed a very convincing signal all in the same direction, that MRD-based prognostication at end of treatment outperforms PET scans. And that's not to say they aren't complementary, but what we've observed is that in multiple cohorts, the end of treatment positive PET patients still had roughly a 50% chance of being free from recurrence at two to three years after the end of treatment. Whereas patients with detectable MRD at end of treatment had only a 20 to 25% chance of remaining free from progression at two years. Conversely, the patients who had complete metabolic responses on their end-of-treatment PET scans still had a 10 to 20% chance of recurrence based on PET alone. Whereas that risk was less than 10% approaching even five or only 4% in some studies for patients with undetectable MRD by ctDNA sequencing.