Q&A: Beyond imaging: Leveraging molecular tools to detect MRD for personalized cancer care

Dr. Tan: MRD can fit into multiple points across care. In the neoadjuvant setting, serial ctDNA testing can help assess whether a patient is responding to therapy before surgery. If ctDNA is not cleared by the time of surgery, that can indicate worse prognosis and suggest residual systemic disease. In the adjuvant setting, MRD can help identify who may still harbor disease after tumor removal and who may benefit from additional therapy. In surveillance, MRD can complement imaging to detect recurrence earlier. And in the metastatic setting, ctDNA can help answer one of the most common patient questions: “Is my treatment working?” rather than waiting months for the next scan.

Dr. Raez: Breast cancer provides a strong example of MRD’s potential clinical value, particularly in triple-negative disease. In the neoadjuvant setting, pathologic complete response has traditionally been one of our best prognostic markers, but it’s imperfect. MRD adds another layer of information. Some patients who do not achieve pathologic complete response still do well if they are ctDNA-negative, suggesting that MRD may help distinguish which patients truly need more therapy from those who may be spared unnecessary treatment.

Dr. Raez: MRD technology today is not the same as it was several years ago. The field has moved toward ultrasensitive assays capable of detecting disease at much lower levels. That matters because improved sensitivity allows us to identify molecular disease that older methods would miss. This is especially important in solid tumors, where ctDNA levels may be very low.

Dr. Raez: Lung cancer studies have shown that both pre-surgical and post-surgical ctDNA positivity are strongly associated with worse relapse-free and overall survival. That’s important because it reinforces that MRD is not only relevant after surgery. Baseline positivity before surgery can itself be prognostic, and longitudinal monitoring improves accuracy even further. Repeated testing over time is generally more informative than a single landmark draw.

Dr. Tan: It’s also notable because lung adenocarcinoma has historically been viewed as a poor shedder of ctDNA. So seeing strong performance with ultrasensitive approaches is encouraging and speaks to how far the technology has advanced.

Dr. Tan: In colorectal cancer, Tempus’ tumor-naive xM assay has shown strong prognostic value, particularly in patients with oligometastatic liver disease treated with systemic therapy followed by surgery. In that setting, persistent ctDNA positivity after therapy and surgery was associated with much worse relapse-free survival. Serial testing was especially informative—patients who remained negative or converted from positive to negative did well, while those who stayed positive or converted to positive had poorer outcomes.

Dr. Tan: In kidney cancer, particularly clear cell RCC, one challenge is that these tumors are often very low shedders. Even so, ultrasensitive tumor-informed testing has shown promising negative predictive value in post-surgical patients. We’ve also seen molecular recurrence precede radiographic recurrence by many months in some patients. In bladder cancer and other GU settings, this has major implications for treatment de-escalation and for identifying which patients may truly need more therapy.

Dr. Raez: In many cancers, we still rely heavily on imaging, but imaging has limitations. MRD can often detect recurrence earlier than scans because it identifies molecular disease before lesions are large enough to visualize. Ideally, in the future, MRD could become the primary surveillance tool, with imaging used as a complement rather than the other way around.

Dr. Tan: There’s also the opposite problem in some cancers. In breast cancer, for example, we may be underscanning during surveillance because routine imaging is not typically performed unless symptoms arise. A blood-based MRD assay could potentially identify recurrence much earlier, before patients become symptomatic.

Dr. Tan: ctDNA can be especially useful in immunotherapy, where imaging can sometimes be misleading because of pseudoprogression. If ctDNA decreases early after treatment begins, that can indicate a true biologic response even if scans appear worse initially. This can help distinguish true progression from inflammation-related imaging changes.

Dr. Tan: In advanced refractory patients receiving immunotherapy in early-phase trials, early molecular response—such as a significant decrease in ctDNA within a few weeks—has been associated with much better progression-free and overall survival. Patients without that molecular response did much worse. That suggests ctDNA may become increasingly important not only in practice but also in clinical trial design.

Dr. Raez: In general, tumor-informed testing is more sensitive and is usually preferred when tissue is available. The challenge is that it requires adequate tumor tissue and time to build the personalized assay. Tumor-naive testing is appealing because it does not require tissue and can provide faster answers. So, in practice, we tend to favor tumor-informed testing when feasible, and use tumor-naive testing when tissue is limited or unavailable.

Dr. Raez: That can be a difficult conversation, because imaging is still widely viewed as the standard. But biologically, it makes sense—molecular disease can be present long before a lesion is large enough to be seen. Patients need to understand that a positive MRD result may indicate recurrence earlier than imaging can. In the future, I believe MRD will increasingly guide when imaging is performed, rather than imaging guiding how we interpret MRD.

Dr. Tan: More and more trials are now studying treatment decisions based on molecular recurrence rather than waiting for radiographic progression. That’s a major shift, and it may open the door to earlier intervention.

Dr. Tan: I like to keep it practical. In metastatic disease, I want a baseline sample and then repeat testing roughly every 6 weeks, often aligned with treatment cycles. That baseline is critical because if you later get a negative result, it’s much more meaningful if you first established that the patient was positive. In the adjuvant setting, if a patient is expected to be negative, I usually test every 3 months. If they become positive, I increase frequency to every 6 weeks.

Dr. Tan: I think it absolutely has the potential to be. In cancers like bladder cancer, melanoma, lung cancer, and others, ctDNA may ultimately prove as good as—or even better than—traditional markers like pathologic complete response for predicting outcomes. It may also help us make de-escalation decisions, avoiding unnecessary therapy and its associated toxicity.

Dr. Raez: I agree. In hematologic malignancies, molecular remission has already been standard for decades. Solid tumor oncology is catching up. We are increasingly seeing how MRD can refine adjuvant decisions, improve surveillance, and help monitor treatment effect in ways that are more sensitive and potentially more actionable than imaging alone.