Skip to content

Biomarkers and liquid biopsy

TL;DR — Colorectal biomarkers have three distinct jobs: classify tumor biology, predict treatment effect and monitor residual/relapsing disease. MMR/MSI, extended RAS, BRAF V600E and HER2 have established treatment utility; CMS and many immune/stromal signatures remain prognostic or exploratory (Morris 2023, PMID 36252154; Guinney 2015, PMID 26457759). CEA is inexpensive but neither sensitive nor specific enough to diagnose recurrence alone; serial trends outperform a single threshold (Shinkins 2018, PMID 29579327). Postoperative ctDNA is one of the strongest recurrence prognosticators: DYNAMIC showed stage II chemotherapy can be reduced using a defined ctDNA strategy, but DYNAMIC-III illustrates that escalation benefit cannot be assumed from prognostic risk (Tie 2022, PMID 35657320; Tie 2025, PMID 41115959). Tumor-informed and tumor-agnostic assays differ in lead time, sensitivity, turnaround and clonal-hematopoiesis control. Blood screening and postoperative MRD are different applications and should never share performance claims (Chung 2024, PMID 38477985).

Biomarker taxonomy

Type Question Example Evidence required
Diagnostic Is cancer present? Screening cfDNA, pathology marker Sensitivity/specificity in intended population
Prognostic What happens regardless of treatment? Postoperative ctDNA Outcome association with confounder control
Predictive Does treatment effect differ? RAS for anti-EGFR Valid treatment-by-marker interaction
Pharmacodynamic Did biology change on treatment? Falling ctDNA Link between change and outcome
Surveillance Has recurrence emerged? CEA/ctDNA Lead time plus improved patient outcome

A strong prognostic biomarker may have no treatment utility. Clinical utility requires that acting on the result improves an outcome.

Established tissue biomarkers

Marker Assay Clinical use Limitation
MMR/MSI IHC, PCR or NGS Lynch triage, prognosis, checkpoint therapy Discordance and incomplete reflex pathways
KRAS/NRAS NGS/PCR Anti-EGFR resistance Acquired clones and plasma false negatives
BRAF V600E NGS/PCR/IHC context BRAF+EGFR therapy, prognosis, MLH1-loss triage V600E differs from non-V600 alterations
HER2 IHC/ISH/NGS Dual HER2 therapy Criteria/platform heterogeneity
NTRK fusion RNA/DNA NGS TRK inhibitor Extremely rare; DNA can miss fusions
DPYD germline Germline genotype Fluoropyrimidine safety Does not capture all DPD deficiency

Universal MMR IHC studies show workflow loss: pooled IHC missingness 11.81% and germline completion 76.30% among eligible patients (Eikenboom 2022, PMID 33887476). A biomarker unavailable at decision time is a system failure even if assay performance is high.

Histopathologic biomarkers

Tumor budding, lymphovascular/perineural invasion, grade, margins and treatment regression remain practical biomarkers embedded in pathology.

ITBCC standardized tumor budding as single cells/clusters ≤4 in a 0.785-mm² hotspot and judged it prognostic in pT1 and stage II disease (Lugli 2017, PMID 28548122). Its predictive utility for a specific chemotherapy is not established.

Node yield is a quality/prognostic composite, not a molecular marker. More examined nodes correlate with survival, but causality includes stage migration, surgery, pathology and host immunity (Chang 2007, PMID 17374833).

CEA

CEA is a glycoprotein shed by many colorectal cancers. Uses include pretreatment baseline, prognosis and post-treatment surveillance.

Strength Weakness
Low cost and widely available Normal in a substantial tumor subset
Serial personalized trend Smoking/inflammation can elevate
Can precede radiographic recurrence False positives accumulate with repeated tests
Complements CT Does not localize disease

FACS analysis found longitudinal CEA trends more informative than single-threshold interpretation, but lowering thresholds increased false alarms (Shinkins 2018, PMID 29579327). Meta-analysis found follow-up-test accuracy heterogeneous and no modality sufficient alone (Liemburg 2021, PMID 33704843).

Older studies established postoperative CEA as a recurrence predictor but used assays and imaging pathways different from current care (McCall 1994, PMID 8076486).

What ctDNA measures

ctDNA is the tumor-derived fraction of cell-free DNA. Concentration depends on tumor burden, vascularity, site, cell death and clearance.

Context Typical shedding
Bulky liver metastases High
Lung-only small metastases Lower
Peritoneal-only disease Often lower
Immediately after surgery Confounded by high non-tumor cfDNA
Microscopic residual disease Near assay detection limit

Negative ctDNA means “not detected by this assay at this time,” not “no cancer.”

Tumor-informed versus tumor-agnostic

Feature Tumor-informed Tumor-agnostic/plasma-only
Design Tracks patient-specific tumor variants Fixed panel, methylation/fragment or genomic features
Sensitivity Can aggregate multiple variants May capture evolution without tissue
Turnaround Requires tissue sequencing/design Often faster
Clonal hematopoiesis Tumor matching helps Paired white-cell filtering important
Failure Insufficient tissue/delay Lower specificity or variant attribution

Reviews stress that assay classes are not interchangeable and that lead-time, immortal-time and sampling biases can exaggerate apparent performance (Martínez-Castedo 2025, PMID 39675560; Hlauschek 2026, PMID 42508231).

Postoperative MRD

ctDNA after curative-intent surgery strongly stratifies recurrence. The critical question is whether treatment can change the poor outcome of a positive test.

DYNAMIC stage II

DYNAMIC randomized 455 stage II patients to ctDNA-guided or clinicopathologic management. Chemotherapy use fell from 28% to 15%; 2-year recurrence-free survival was 93.5% versus 92.4%, meeting noninferiority (Tie 2022, PMID 35657320).

At 5 years, recurrence-free and overall survival remained consistent with safe de-escalation within the trial algorithm (Tie 2025, PMID 40055522).

DYNAMIC proves DYNAMIC does not prove
A defined stage II strategy reduced chemotherapy Every commercial assay is equivalent
Noninferior RFS in that trial ctDNA-negative risk is zero
Positive patients received treatment Escalation beyond standard improves survival

Stage III and escalation

DYNAMIC-III randomized ctDNA-informed escalation/de-escalation. De-escalation reduced treatment, but intensifying therapy in ctDNA-positive disease did not automatically overcome risk (Tie 2025, PMID 41115959).

Post hoc IDEA-France analysis found postoperative ctDNA prognostic in stage III and related it to duration, but biomarker subgroup analysis cannot substitute for a prospective action trial (Taieb 2021, PMID 34083233).

CIRCULATE-Japan

CIRCULATE-Japan links the GALAXY observational registry with VEGA de-escalation and ALTAIR escalation trials (Taniguchi 2021, PMID 33931919). Large resectable-disease analyses confirm strong molecular-residual-disease prognostic separation (Nakamura 2024, PMID 39284954).

Timing

Time point Use Caveat
Preoperative Assay design and shedding baseline Not MRD
~2–4 weeks postoperative Early MRD Surgical cfDNA may dilute signal
Before adjuvant therapy Treatment allocation Delay versus turnaround
During therapy Molecular clearance Clearance may be temporary
End of therapy Residual risk Requires serial confirmation
Surveillance Molecular recurrence Action pathway unproven

Serial sampling improves sensitivity but increases cost and false-positive opportunity. Landmark/time-dependent analysis is required.

ctDNA after metastasis-directed therapy

After liver resection/ablation, detectable ctDNA predicts recurrence, but liver-metastasis cohorts vary by assay and treatment. A systematic review/meta-analysis supports prognostic value while emphasizing small cohorts and heterogeneity (Khan 2026, PMID 41723037).

The low shedding of lung/peritoneal recurrence can make plasma falsely negative. Organ-specific sensitivity should be reported.

Metastatic disease monitoring

Plasma NGS can identify baseline targets and acquired resistance. Under anti-EGFR therapy, serial ctDNA detects emergent RAS/BRAF/MAP2K1/EGFR alterations across lesions (Topham 2023, PMID 36007218).

CALGB/SWOG 80405 plasma analysis demonstrated that first-line cetuximab can select multiple acquired genomic alterations rather than one dominant universal mechanism (Raghav 2023, PMID 36067452).

Resistant clones decay off anti-EGFR with estimated half-life around 4.4 months, supporting molecular rechallenge selection (Parseghian 2019, PMID 30462160). Clinical utility depends on whether plasma-guided sequence improves survival, not just response (Patelli 2023, PMID 37436743).

Screening cfDNA is a separate use case

ECLIPSE reported 83.1% CRC sensitivity, 89.6% specificity for advanced neoplasia and 13.2% advanced-precursor sensitivity (Chung 2024, PMID 38477985). Screening performance in asymptomatic average-risk people cannot be inferred from high-shedding metastatic disease.

Cell-free DNA diagnostic reviews show large variation by methylation, fragmentation, mutation panel and case-control design (Petit 2019, PMID 30694754). Methylated-region panels can detect cancer but require prospective screening denominators (Young 2021, PMID 33478584).

Meta-analysis and bias

A network meta-analysis found ctDNA MRD strongly associated with recurrence and compared assay/timing strategies, but indirect comparisons inherit study and treatment differences (Hoang 2025, PMID 40293388).

Bias How it inflates performance
Lead time Earlier detection appears longer survival without changing death
Immortal time Patient must survive to later sample
Treatment confounding Positive patients receive more therapy
Case-control spectrum Advanced cases versus healthy controls overstates screening accuracy
Selective sampling Missing samples from sicker patients
Publication bias Negative assays less visible

Analytical standards

  • Predefine limit of detection and report plasma volume.
  • Report tumor fraction and number of tracked variants.
  • Filter clonal hematopoiesis, preferably with paired leukocytes.
  • State timing from surgery and treatment.
  • Report failure/no-result rates.
  • Blind outcome assessment to exploratory biomarker where feasible.
  • Lock threshold before validation.
  • Validate externally and prospectively.

Clinical-utility scoreboard

Application Status
Metastatic RAS/BRAF target testing Established
Anti-EGFR resistance mapping Clinically useful in selected rechallenge; evolving
Stage II de-escalation Randomized DYNAMIC evidence
Stage III escalation Not established merely by prognostic positivity
Routine surveillance replacing CT/CEA Not established
Blood screening replacing established strategies Outcome evidence immature
Rectal complete-response prediction Investigational

Open questions

  • Which assay/timing combination maximizes MRD sensitivity without unacceptable false positives? (Hoang 2025, PMID 40293388)
  • What treatment clears ctDNA and improves survival in positive stage III disease? (Tie 2025, PMID 41115959)
  • Can serial ctDNA safely reduce CT imaging in any surveillance group? (Liemburg 2021, PMID 33704843)
  • How should low-shedding lung and peritoneal recurrence be handled? (Khan 2026, PMID 41723037)
  • Will blood screening’s participation gain compensate for 13.2% advanced-precursor sensitivity? (Chung 2024, PMID 38477985)

References

  1. Morris VK, et al. Treatment of Metastatic Colorectal Cancer: ASCO Guideline. J Clin Oncol. 2023;41(3):678-700. PMID 36252154
  2. Guinney J, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21(11):1350-6. PMID 26457759
  3. Shinkins B, et al. Serum carcinoembryonic antigen trends for diagnosing colorectal cancer recurrence in the FACS randomized clinical trial. Br J Surg. 2018;105(6):658-662. PMID 29579327
  4. Tie J, et al. Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon Cancer. N Engl J Med. 2022;386(24):2261-2272. PMID 35657320
  5. Tie J, et al. Circulating tumor DNA-guided adjuvant therapy in locally advanced colon cancer: the randomized phase 2/3 DYNAMIC-III trial. Nat Med. 2025;31(12):4291-4300. PMID 41115959
  6. Chung DC, et al. A Cell-free DNA Blood-Based Test for Colorectal Cancer Screening. N Engl J Med. 2024;390(11):973-983. PMID 38477985
  7. Eikenboom EL, et al. Universal Immunohistochemistry for Lynch Syndrome: A Systematic Review and Meta-analysis of 58,580 Colorectal Carcinomas. Clin Gastroenterol Hepatol. 2022;20(3):e496-e507. PMID 33887476
  8. Lugli A, et al. Recommendations for reporting tumor budding in colorectal cancer based on the International Tumor Budding Consensus Conference (ITBCC) 2016. Mod Pathol. 2017;30(9):1299-1311. PMID 28548122
  9. Chang GJ, et al. Lymph node evaluation and survival after curative resection of colon cancer: systematic review. J Natl Cancer Inst. 2007;99(6):433-41. PMID 17374833
  10. Liemburg GB, et al. Diagnostic accuracy of follow-up tests for detecting colorectal cancer recurrences in primary care: A systematic review and meta-analysis. Eur J Cancer Care (Engl). 2021;30(5):e13432. PMID 33704843
  11. McCall JL, et al. The value of serum carcinoembryonic antigen in predicting recurrent disease following curative resection of colorectal cancer. Dis Colon Rectum. 1994;37(9):875-81. PMID 8076486
  12. Martínez-Castedo B, et al. Minimal residual disease in colorectal cancer. Tumor-informed versus tumor-agnostic approaches: unraveling the optimal strategy. Ann Oncol. 2025;36(3):263-276. PMID 39675560
  13. Hlauschek D, et al. Common analysis pitfalls in longitudinal ctDNA studies: lead time, sensitivity and immortal time bias. EBioMedicine. 2026;130:106407. PMID 42508231
  14. Tie J, et al. Circulating tumor DNA analysis guiding adjuvant therapy in stage II colon cancer: 5-year outcomes of the randomized DYNAMIC trial. Nat Med. 2025;31(5):1509-1518. PMID 40055522
  15. Taieb J, et al. Prognostic Value and Relation with Adjuvant Treatment Duration of ctDNA in Stage III Colon Cancer: a Post Hoc Analysis of the PRODIGE-GERCOR IDEA-France Trial. Clin Cancer Res. 2021;27(20):5638-5646. PMID 34083233
  16. Taniguchi H, et al. CIRCULATE-Japan: Circulating tumor DNA-guided adaptive platform trials to refine adjuvant therapy for colorectal cancer. Cancer Sci. 2021;112(7):2915-2920. PMID 33931919
  17. Nakamura Y, et al. ctDNA-based molecular residual disease and survival in resectable colorectal cancer. Nat Med. 2024;30(11):3272-3283. PMID 39284954
  18. Khan AU, et al. Circulating Tumor DNA as a Biomarker for Recurrence After Curative Resection of Colorectal Cancer Liver Metastases: A Systematic Review and Meta-Analysis. Clin Colorectal Cancer. 2026;25(2):181-190.e1. PMID 41723037
  19. Topham JT, et al. Circulating Tumor DNA Identifies Diverse Landscape of Acquired Resistance to Anti-Epidermal Growth Factor Receptor Therapy in Metastatic Colorectal Cancer. J Clin Oncol. 2023;41(3):485-496. PMID 36007218
  20. Raghav K, et al. Acquired Genomic Alterations on First-Line Chemotherapy With Cetuximab in Advanced Colorectal Cancer: Circulating Tumor DNA Analysis of the CALGB/SWOG-80405 Trial (Alliance). J Clin Oncol. 2023;41(3):472-478. PMID 36067452
  21. Parseghian CM, et al. Anti-EGFR-resistant clones decay exponentially after progression: implications for anti-EGFR re-challenge. Ann Oncol. 2019;30(2):243-249. PMID 30462160
  22. Patelli G, et al. Circulating Tumor DNA to Drive Treatment in Metastatic Colorectal Cancer. Clin Cancer Res. 2023;29(22):4530-4539. PMID 37436743
  23. Petit J, et al. Cell-Free DNA as a Diagnostic Blood-Based Biomarker for Colorectal Cancer: A Systematic Review. J Surg Res. 2019;236:184-197. PMID 30694754
  24. Young GP, et al. Evaluation of a panel of tumor-specific differentially-methylated DNA regions in IRF4, IKZF1 and BCAT1 for blood-based detection of colorectal cancer. Clin Epigenetics. 2021;13(1):14. PMID 33478584
  25. Hoang T, et al. Utility of circulating tumor DNA to detect minimal residual disease in colorectal cancer: A systematic review and network meta-analysis. Int J Cancer. 2025;157(4):722-740. PMID 40293388