A cancer biomarker is a biological molecule found in blood, other body fluids, or tissues that signals something about the presence, behavior, or vulnerabilities of a cancer. Biomarkers can be proteins, gene mutations, DNA patterns, or other measurable substances that help doctors detect cancer, predict how it will behave, choose the right treatment, or track whether a treatment is working. They are the foundation of precision oncology, the approach that tailors cancer care to the specific biology of each patient’s tumor rather than treating all cancers of the same type identically.
The Three Main Roles of Cancer Biomarkers
Cancer biomarkers serve different purposes depending on when and how they’re used. The three most important categories are diagnostic, prognostic, and predictive biomarkers, and understanding the distinction matters because each one answers a fundamentally different question.
Diagnostic biomarkers help identify whether cancer is present and what type it is. These are the markers used in screening tests and in distinguishing a malignant growth from a benign one. Prognostic biomarkers tell doctors about the likely course of the disease, regardless of treatment. A tumor with certain genetic features may be inherently more aggressive, which helps determine how closely it needs to be monitored or how urgently treatment should begin. Predictive biomarkers answer a different question entirely: will this specific patient benefit from this specific therapy? A predictive biomarker identifies who is likely to respond to a particular drug and, just as importantly, who is unlikely to respond or might even be harmed by it.
The distinction between prognostic and predictive biomarkers is often confused, even in clinical settings. A prognostic marker says “this cancer tends to be more dangerous.” A predictive marker says “this cancer will respond to drug X.” Sometimes one biomarker is both, but the two concepts require different types of evidence to confirm. Identifying a predictive biomarker generally requires comparing treated and untreated patients who do and don’t carry the marker, while a prognostic biomarker can often be identified through observation alone.
How Biomarkers Guide Treatment Decisions
The clearest impact of cancer biomarkers is in matching patients with targeted therapies. Rather than relying solely on chemotherapy that attacks all rapidly dividing cells, oncologists now test tumors for specific molecular features that reveal which drugs are most likely to work.
HER2-positive breast cancer is one of the best-known examples. Tumors that overexpress the HER2 protein respond dramatically to drugs designed to block that protein. In clinical data, first-line targeted treatment in patients with HER2 gene amplification produced response rates of 34%, compared to just 7% in patients without the amplification. Before HER2-targeted therapy existed, HER2-positive breast cancer carried a particularly poor prognosis. Now it’s one of the most treatable forms.
A similar story played out in lung cancer. When drugs targeting the EGFR protein were first approved for non-small cell lung cancer, they worked in only 10 to 20% of patients. Researchers then discovered that patients whose tumors carried specific activating mutations in the EGFR gene were the ones responding. Once doctors began prescreening patients for those mutations, response rates jumped above 50%, with meaningful improvements in survival.
BRAF mutations in melanoma offer a particularly striking illustration. Tumors carrying the BRAF V600E mutation are highly sensitive to drugs that block BRAF activity. But in tumors without this mutation, those same drugs can actually accelerate tumor growth. The biomarker test doesn’t just identify who benefits; it protects patients who would be harmed.
The survival data reinforces how much this matters. In a study of late-stage cancer patients, those who received targeted treatment matched to a biomarker in their tumor had a median overall survival of about 15 months. Patients whose tumors had no druggable target survived a median of 6.5 months, and patients who had a druggable target identified but didn’t receive the matched therapy survived just 5.8 months. Biomarker-matched treatment cut the risk of death by more than half.
Biomarkers for Immunotherapy
Immunotherapy, particularly checkpoint inhibitors that help the immune system recognize and attack cancer cells, relies on its own set of biomarkers. The two most widely used are PD-L1 expression and tumor mutational burden (TMB).
PD-L1 is a protein that some cancer cells display on their surface. It essentially tells the immune system to stand down. Tumors with high PD-L1 expression tend to respond better to drugs that block this protein, because those drugs remove the “don’t attack” signal. The FDA has approved several immunotherapy drugs specifically for tumors that express PD-L1 above a certain threshold.
TMB measures how many mutations a tumor carries overall. Tumors with a high number of mutations produce more abnormal proteins, which gives the immune system more targets to recognize. Research in lung cancer patients treated with pembrolizumab found that those with high mutational burden had better response rates and more durable clinical benefit. Other markers, including a DNA repair deficiency called microsatellite instability (MSI), also predict immunotherapy response and are now routinely tested in several cancer types.
Screening Biomarkers and Their Limits
Biomarkers used for cancer screening get a lot of public attention, but they come with important caveats. The best-known example is PSA for prostate cancer. PSA-based screening in men aged 55 to 69 has been shown to prevent about 1.3 deaths from prostate cancer per 1,000 men screened over 13 years, but the trials found no reduction in overall mortality. PSA levels also rise in benign conditions like an enlarged prostate, leading to false positives, unnecessary biopsies, and overtreatment of slow-growing cancers that may never have caused harm.
CA-125, a protein sometimes elevated in ovarian cancer, faces even steeper limitations. Only about 50% of patients with stage I ovarian cancer have elevated CA-125 levels, making it unreliable for early detection, which is exactly when screening would be most valuable. CA-125 also rises in benign conditions like endometriosis, ovarian cysts, and liver disease. A large randomized trial in the UK found that screening postmenopausal women with CA-125 and ultrasound did not significantly reduce ovarian cancer deaths compared to no screening. No major American medical society currently recommends routine ovarian cancer screening in the general population. CA-125 remains useful, however, for monitoring treatment response once ovarian cancer has been diagnosed.
Serum cancer markers in general have poor positive predictive value for mass screening, meaning a positive result is often a false alarm. This is why most biomarker-based screening is reserved for people already at elevated risk, rather than the general population.
Biomarkers That Predict Drug Toxicity
Some biomarkers don’t predict whether a drug will work against the cancer. Instead, they predict whether the drug will cause severe side effects in a particular patient. These pharmacogenomic biomarkers test for inherited genetic variations that affect how the body processes certain medications.
One well-established example involves an enzyme that breaks down a chemotherapy drug used in leukemia. Patients who carry genetic variants that make this enzyme less active are at significant risk for severe toxicity, because the drug accumulates to dangerous levels instead of being cleared normally. Testing for these variants before starting treatment allows doctors to adjust the dose or choose an alternative. Similarly, a drug used to manage a complication of chemotherapy is completely contraindicated in patients with a specific enzyme deficiency more common in people of African or Mediterranean descent, because it can cause a dangerous breakdown of red blood cells.
How Biomarker Testing Works
Biomarkers are detected through two main approaches: tissue biopsy and liquid biopsy. Each has distinct strengths.
Tissue biopsy involves taking a physical sample of the tumor, usually through a needle or during surgery. This remains the gold standard for many biomarker tests because it allows direct examination of tumor cells, including their structure, protein expression, and genetic mutations. The limitations are practical. The sample comes from one spot in the tumor, and cancers are often heterogeneous, meaning different areas can have different molecular profiles. There are also small risks of pain, bleeding, and in rare cases, tumor cells spreading along the needle track.
Liquid biopsy analyzes tumor-derived material circulating in the blood, including fragments of tumor DNA, whole tumor cells, and tiny cellular packages called extracellular vesicles. The advantage is that a simple blood draw can capture genetic information from multiple tumor sites at once, providing a broader picture of the cancer’s molecular landscape. Liquid biopsies can also be repeated easily, allowing doctors to track how a tumor evolves over time and detect emerging drug resistance early. The main limitation is sensitivity: early-stage tumors shed very little DNA into the bloodstream, making them harder to detect than advanced cancers.
Multi-Cancer Early Detection Tests
A new generation of blood-based tests aims to screen for dozens of cancer types simultaneously by analyzing combinations of biomarkers. These multi-cancer early detection (MCED) tests look for DNA mutations, abnormal DNA methylation patterns, fragmented DNA, and cancer-associated proteins in a single blood sample, then use machine learning algorithms to determine whether cancer is present and where in the body it likely originated.
One FDA-approved MCED test combines genomic mutations, methylation patterns, and DNA fragmentation for early colorectal cancer detection. Another, which has received FDA breakthrough device designation, analyzes eight cancer-associated proteins and 16 gene mutations at once. Large trials have reported sensitivity ranging from 50 to 95% and specificity between 89 and 99%, depending on the cancer type and stage. A major randomized trial in the UK involving 140,000 participants is evaluating whether one of these tests can shift cancer diagnoses to earlier stages, with final results expected in 2026.
These tests combine multiple biomarker types because no single marker is reliable enough on its own. By layering DNA, protein, and methylation data together, MCED tests have reached sensitivities of 85 to 87% with specificity above 99% in some studies. Challenges remain around cost, accessibility, and the psychological impact of false positives, but the field is moving quickly toward integrating these tools into routine screening for high-risk populations.