Natera has reported that its personalized Signatera molecular residual disease test identified a subgroup of patients with resected colorectal liver metastases who appeared to gain an overall survival benefit from postoperative chemotherapy. The 298-patient CIRCULATE-Japan GALAXY analysis, published in JAMA Oncology, found the association in patients who underwent upfront surgery and remained MRD-positive after resection, while MRD-negative patients showed no statistically significant survival benefit from adjuvant chemotherapy.
Why does an overall survival signal matter more than another recurrence-risk result?
The importance of the latest analysis lies less in confirming that circulating tumor DNA can identify patients at high risk of recurrence and more in connecting MRD status with the possibility of living longer after chemotherapy. Previous colorectal cancer studies had already established postsurgical MRD positivity as a powerful prognostic warning. A positive result generally indicated that microscopic disease remained after apparently curative treatment, even when conventional imaging showed no visible cancer.
The harder question has been whether that biological risk marker can also predict which patients benefit from a specific intervention. Prognostic biomarkers reveal who is more likely to experience a poor outcome, while predictive biomarkers help determine whether one treatment is more effective than another for a defined subgroup. That distinction is central to whether MRD testing becomes an observational surveillance tool or a genuine treatment-selection platform.
Among MRD-positive patients who underwent surgery without preoperative chemotherapy, adjuvant chemotherapy was associated with a 73% reduction in the adjusted risk of death compared with observation. Estimated overall survival at 48 months was 65.3% with chemotherapy and 32.9% without chemotherapy. Disease-free survival also favored treatment, strengthening the argument that the observed overall survival difference was connected to disease control rather than a statistical anomaly involving mortality from unrelated causes.
The result nevertheless requires restraint. Only 16 MRD-positive patients in the upfront-surgery cohort received adjuvant chemotherapy in the main disease-free survival comparison, while the untreated MRD-positive group was also relatively small. The overall survival confidence interval was wide, which means the true size of the treatment effect could be considerably smaller than the headline estimate.
How could postsurgical MRD testing change chemotherapy decisions after liver resection?
Surgery can offer selected patients with colorectal liver metastases a potentially curative treatment path, but recurrence remains common and the value of routinely administering chemotherapy after resection has remained uncertain. Earlier trials and pooled analyses have sometimes shown improvements in disease-free survival without producing a consistent overall survival advantage. This has left clinicians balancing the possibility of preventing relapse against neuropathy, fatigue, cytopenias, gastrointestinal toxicity and the broader burden of systemic treatment.
Signatera introduces a biological layer into that decision. Instead of relying only on the number and size of liver metastases, surgical margins, RAS status, timing of metastatic disease and other clinicopathological factors, clinicians could potentially use postsurgical circulating tumor DNA to identify evidence of residual cancer that remains invisible to scans.
The strongest immediate use case may be treatment prioritisation for MRD-positive patients. These patients had markedly worse outcomes without effective postoperative therapy, suggesting that a positive result could identify a population in which the risks of chemotherapy are more clearly justified. MRD testing may therefore help move adjuvant treatment away from broad risk estimation and toward evidence that an individual patient still carries molecular disease.
The more disruptive possibility is treatment de-escalation in patients who test MRD-negative. In the upfront-surgery cohort, MRD-negative patients had favorable overall survival regardless of whether they received adjuvant chemotherapy. At 48 months, estimated overall survival was above 90% in both treated and observed groups, and the differences were not statistically significant.
That finding does not prove that chemotherapy can safely be omitted. A negative MRD result means that the assay did not detect the selected tumor variants in the blood sample at that time. It does not establish that every residual cancer cell has been eliminated. The consequences of a false-negative result are particularly serious when the proposed action is withholding potentially beneficial treatment.
Why does the study strengthen Signatera’s case without proving a treatment standard?
The GALAXY analysis has several features that make it more informative than a small retrospective chart review. Patients came from a prospective national registry, blood collection followed defined intervals, MRD status was measured using a consistent tumor-informed platform, and the median follow-up exceeded three and a half years. The investigators also used landmark analyses to reduce immortal-time bias and adjusted treatment comparisons for clinically relevant variables including age, sex, RAS status, previous oxaliplatin exposure and metastatic burden.
However, chemotherapy was not randomly assigned. Physicians and patients decided whether postoperative treatment would be administered, creating the possibility that measured and unmeasured differences affected outcomes. Multivariable adjustment can reduce this problem, but it cannot reproduce the balance produced by randomisation.
The treatment groups were also uneven. Patients who received chemotherapy may have differed in fitness, access to care, postoperative recovery, tumour biology or clinician-assessed recurrence risk. Conversely, patients placed under observation may have had comorbidities, complications or preferences that independently influenced long-term survival.
Chemotherapy regimens and treatment duration were not completely uniform, limiting conclusions about which drugs, intensity or number of cycles produced the observed benefit. The analysis therefore supports the concept of MRD-directed treatment but does not define a standard regimen for an MRD-positive patient after colorectal liver metastasis resection.
The publication also involved several authors employed by and holding shares in Natera. The broader CIRCULATE-Japan platform received public research funding, and those public funders had no reported role in study design, analysis or publication decisions. The disclosed commercial involvement does not invalidate the findings, but it increases the importance of independent replication and randomized confirmation.
What does the neoadjuvant cohort reveal about where the survival signal may not apply?
The results cannot be applied uniformly to every patient undergoing liver metastasis surgery. In the 107-patient cohort that received chemotherapy before surgery, postoperative adjuvant chemotherapy was not associated with improved disease-free survival or overall survival, regardless of MRD status.
MRD remained strongly prognostic in this group. Patients who tested positive after surgery experienced substantially worse outcomes than those who tested negative, confirming that Signatera continued to identify residual disease risk. The missing element was evidence that additional postoperative chemotherapy changed that risk.
Several explanations remain plausible. Patients selected for preoperative chemotherapy often have more extensive, synchronous or biologically aggressive disease. They may also have already been exposed to oxaliplatin or other agents, reducing the effectiveness of repeating a similar regimen after surgery. Persistent MRD after preoperative systemic treatment and resection could represent treatment-resistant disease rather than disease that remains sensitive to further conventional chemotherapy.
The divergence between cohorts is clinically valuable because it limits overinterpretation. Signatera may function as a strong prognostic test across both groups while serving as a predictive treatment-selection test only in a more specific setting. A positive MRD result after upfront surgery may identify chemotherapy-sensitive residual disease, while positivity after neoadjuvant treatment could signal a need for a different strategy, intensified therapy, a clinical trial or closer surveillance.
How does Signatera’s tumor-informed design compare with broader ctDNA approaches?
Signatera uses a personalized, tumor-informed process. Tumor tissue and matched normal blood are initially sequenced to identify patient-specific somatic variants. A customized assay then tracks up to 16 selected variants in subsequent plasma samples, with MRD positivity requiring detection of at least two tumor-specific variants above a predefined threshold.
This design aims to improve specificity by distinguishing genuine tumour-derived alterations from germline variants and age-related clonal changes in blood cells. It also allows circulating tumour DNA to be quantified over time, potentially showing whether molecular disease is clearing, persisting or increasing during treatment.
The operational trade-off is complexity. Tumor-informed testing requires suitable tissue, matched blood, bespoke assay construction and centralized laboratory processing before serial monitoring can begin. Turnaround time, sample logistics and access to archived tumour material can become important in real-world settings where adjuvant decisions must be made within a limited postoperative window.
Tumor-naive or tissue-free MRD platforms avoid bespoke assay development by searching directly for broader genomic or epigenomic signals in plasma. They may offer simpler onboarding, but their sensitivity and specificity can differ depending on tumour type, disease burden, assay methodology and sampling schedule. The GALAXY analysis was not a head-to-head comparison, so it cannot establish that Signatera is superior to competing MRD technologies.
An additional biological limitation is variable ctDNA shedding. The study observed different recurrence patterns among MRD-positive and MRD-negative patients, with lung recurrences appearing more frequently among patients who initially tested negative. Small lung lesions may release less detectable circulating tumour DNA than liver disease, creating a risk that a blood-based assay underestimates residual cancer in certain anatomical settings.
What could accelerate adoption, reimbursement and regulatory acceptance in colorectal cancer?
The data arrive as Natera is expanding Signatera from a laboratory testing platform toward a more formally regulated oncology diagnostic. Signatera received regulatory approval in Japan for colorectal cancer in the adjuvant setting in June 2026, with commercial launch dependent on final pricing and reimbursement decisions. The GALAXY programme provides a locally generated evidence base that may support adoption among Japanese oncology centres already familiar with the platform.
The regulatory position is different in the United States. Signatera CDx received approval as a companion diagnostic for identifying molecular residual disease-positive patients with muscle-invasive bladder cancer who may benefit from adjuvant atezolizumab. That approval demonstrates that the underlying tumor-informed platform can meet companion diagnostic requirements, but it does not automatically extend the approved indication to colorectal cancer.
The colorectal liver metastasis study may strengthen future regulatory submissions by linking MRD status not only with recurrence but also with differential overall survival after chemotherapy. Regulators and payers generally place greater weight on evidence that a diagnostic changes treatment outcomes rather than merely predicting risk.
Reimbursement will still depend on the exact population, intended use and treatment consequence. Existing coverage arrangements for colorectal MRD testing do not necessarily resolve access in every resected stage IV or oligometastatic setting. Payers may ask whether testing reduces ineffective chemotherapy, improves survival, changes imaging use or generates sufficient savings to offset the cost of personalized assay development and serial testing.
Commercial adoption will also require reliable integration into pathology, surgery and oncology workflows. Tumor tissue must be transferred promptly, blood must be drawn within the validated postoperative window, results must return before adjuvant decisions are finalized, and clinicians must understand how to interpret both positive and negative findings. A clinically useful assay can still fail to change practice when logistics are slow or treatment algorithms remain uncertain.
Which unanswered questions will determine whether MRD-guided care becomes routine?
Randomized evidence remains the central requirement. Future trials need to assign MRD-positive patients to defined treatment strategies and determine whether escalation, continuation or alteration of therapy improves overall survival. Separate de-escalation studies must establish whether MRD-negative patients can avoid chemotherapy without increasing late recurrence or mortality.
Serial testing may prove more useful than a single postoperative result. An initially positive result that becomes negative during chemotherapy could function as an early measure of treatment effectiveness. Persistent positivity could identify resistance before radiographic progression, creating an opportunity to change therapy or enrol the patient in a clinical trial.
The field also needs clearer answers about optimal sampling time, testing frequency and the management of discordant results. A negative blood test with suspicious imaging, rising carcinoembryonic antigen or high-risk pathology should not be interpreted in isolation. Likewise, a positive MRD result without visible disease raises difficult questions about when treatment should begin and which regimen should be selected.
The most defensible interpretation is that Signatera has crossed an important evidentiary threshold without completing the journey to routine treatment selection. The new data provide a credible overall survival signal in MRD-positive patients undergoing upfront surgery for colorectal liver metastases and reinforce the possibility that MRD-negative patients are being exposed to chemotherapy without measurable benefit.
The study does not yet justify automatic chemotherapy decisions based solely on a blood test. Its value lies in defining a biologically selected population for prospective validation and showing that the long-running debate over postoperative chemotherapy may be better resolved through residual-disease biology than through clinicopathological risk factors alone.
