How well does ctDNA actually detect recurrence?
ctDNA monitoring is highly specific — when it says cancer is back, it's almost always right. A major meta-analysis of 17 studies found that serial (repeated) ctDNA testing has a specificity of 98% for predicting breast cancer recurrence, meaning only about 2% of positive results are false alarms [2]. In head and neck cancer, a study reported 100% specificity: every patient with a positive ctDNA test after treatment eventually had a confirmed recurrence [5]. This high reliability means a positive ctDNA result gives doctors and patients a strong, early warning.
Sensitivity — the ability to catch a recurrence when it happens — is good but not perfect, and it depends heavily on how often you test. The same meta-analysis found that a single 'landmark' test after surgery catches only about 40% of eventual recurrences, but serial testing (every 3-6 months) raises that to 79% [2]. In a large breast cancer study of 156 patients, serial ctDNA detected 88% of relapses, with a median lead time of 10 months before scans showed anything [8]. For pancreatic cancer, longitudinal monitoring caught 94% of recurrences, with a median lead time of 3.6 months [3]. The pattern is clear: one test is not enough; repeated testing is what makes ctDNA a powerful tool.
Does ctDNA work the same for all cancer types?
No, performance varies by cancer type and subtype, and the evidence is strongest for some cancers than others. In colorectal cancer, a systematic review of 19 studies found that serial ctDNA testing detects recurrence with 83-100% sensitivity, and it picks up relapse a median of 5-11.5 months before imaging [4]. For pancreatic cancer, two studies showed ctDNA detected recurrence before scans in 67-100% of patients, with lead times of 3.6-4.2 months [3][6]. In breast cancer, the picture is more nuanced: a large study found ctDNA detected 88% of relapses overall, but it performed best in hormone-receptor-positive/HER2-negative cancers (which tend to recur late) and worst in triple-negative breast cancer, where it detected none of the recurrences in one interim analysis [1][8]. For ovarian cancer, ctDNA was detectable in 77% of patients with no visible residual disease after surgery, suggesting it can spot microscopic disease that surgeons cannot see [7]. For glioma (brain cancer), ctDNA from tumor-site fluid helped distinguish between local and distant recurrences, which has implications for treatment planning [9]. The takeaway: ctDNA works across many cancers, but its sensitivity and lead time vary, and some subtypes (like triple-negative breast cancer) may need more research.
What are the practical limitations and caveats?
Despite its promise, ctDNA monitoring is not yet a perfect crystal ball. First, sensitivity is not 100% — even with serial testing, about 10-20% of recurrences are missed [2][8]. In breast cancer, one study found that ctDNA failed to detect 4 out of 34 relapses, all of which were in hormone-receptor-positive/HER2-negative patients with local or bone-only recurrences [8]. This suggests that some low-volume or slow-growing recurrences may not shed enough DNA into the blood to be picked up. Second, the technology requires enough tumor tissue to design a personalized test; in one study, 12% of patients could not get a test because there wasn't enough tissue from their original surgery [1]. Third, a positive ctDNA result does not always mean the patient will relapse — in one breast cancer study, 5 patients had positive ctDNA but had not relapsed by the end of follow-up, though these were low-level signals [8]. Finally, while ctDNA can detect recurrence months early, it is not yet proven that acting on that early detection improves survival — that question is the focus of ongoing clinical trials [8]. So ctDNA is a powerful early-warning system, but it is best used as an adjunct to standard imaging and clinical assessment, not a replacement.
About These Sources
This answer is built on 9 peer-reviewed studies — published from 2022 to 2026, 7 from 2024 or later, 7 in Q1 journals — selected as the most relevant from 15 studies that passed quality screening, drawn from 69 papers retrieved from a database of over 500 million.
Sources used in this answer
ctDNA monitoring for breast cancer at high risk of recurrence: Interim analysis of JCOG1204A1.
In a prospective ancillary study of 211 high-risk breast cancer patients, ctDNA was detected in 11.8% overall, with 0% detection in triple-negative and 16.2% in luminal subtypes; detection increased ~2% annually in luminal cancers, suggesting correlation with late recurrence.
Accuracy of ctDNA-based minimal residual disease detection in predicting postoperative recurrence of breast cancer: a meta-analysis
A meta-analysis of 17 studies found that serial ctDNA monitoring for breast cancer recurrence has 79% sensitivity and 98% specificity, significantly outperforming a single post-surgery test (40% sensitivity, 95% specificity).
Personalized ctDNA monitoring to assess the recurrence of early stage pancreatic cancer after surgery.
In 35 resected pancreatic cancer patients, personalized ctDNA monitoring detected 94% of recurrences with a median lead time of 3.6 months; longitudinal PPV was 68% and NPV was 92%.
Abstract 1933: Improving detection of colorectal cancer recurrence using serial ctDNA measurements: A systematic literature review
A systematic review of 19 colorectal cancer studies found that serial ctDNA sampling (every 3-6 months) improved sensitivity for recurrence to 83-100%, with lead times of 5-11.5 months before imaging.
ctDNA as an Adjunct to Posttreatment PET for Head and Neck Cancer Recurrence Risk Assessment
In 29 head and neck cancer patients, post-treatment ctDNA had 100% specificity and 78% sensitivity for recurrence, with 100% PPV and 91% NPV, outperforming PET scans (78% sensitivity, 68% specificity).
Circulating tumor DNA for recurrence monitoring following resection for pancreas adenocarcinoma.
In 24 resected pancreatic cancer patients, ctDNA detected recurrence before imaging in 67% of cases (mean lead time 116 days), and preoperative ctDNA positivity was associated with significantly shorter recurrence-free survival.
Evaluating the Utility of ctDNA in Detecting Residual Cancer and Predicting Recurrence in Patients with Serous Ovarian Cancer
In 47 ovarian cancer patients, ctDNA was detectable in 77% of those with no visible residual disease after surgery, suggesting it can identify microscopic residual disease; ctDNA-positive patients had higher mortality (HR 2.32), though not statistically significant.
Serial postoperative ctDNA monitoring of breast cancer recurrence.
In 156 breast cancer patients with serial ctDNA monitoring, 88% of relapses were detected ahead of clinical/radiologic recurrence (median lead time 10 months), with a hazard ratio of 47.5 for relapse-free survival in ctDNA-positive patients.
Dynamic Monitoring of Circulating Tumor DNA to Predict the Risk of Non In Situ Recurrence of Postoperative Glioma: A Prospective Cohort Study.
In 69 glioma patients, ctDNA from tumor-site fluid helped distinguish between in-situ and non-in-situ recurrences, with EGFR mutations most common in recurrent glioblastoma and MUC/TP53 pathway mutations enriched in non-in-situ recurrences.