A Practical Guide to Biomarkers for Liver Cancer

If you or someone you love is watching for liver cancer, blood tests can feel both empowering and confusing. Biomarkers are simply measurable signals in blood that reflect what's happening in the liver.¹ They are not crystal balls, but they can be smart tools when they're used the right way, alongside imaging and clinical context. Think of them like a weather forecast: trends and patterns matter much more than a single blip.

What "liver cancer" means in this guide

Most biomarker conversations center on hepatocellular carcinoma (HCC), the most common primary liver cancer.² A related but different disease is intrahepatic cholangiocarcinoma (bile duct cancer inside the liver). The two behave differently and lean on different lab signals. This guide explains both, with a focus on HCC where blood biomarkers are most developed.

Where biomarkers fit in your care pathway

Biomarkers have different jobs at different moments:

• Surveillance: aiming to flag early cancer in people at elevated risk (usually due to cirrhosis or chronic hepatitis B)³
• Diagnosis support: adding evidence when imaging shows something suspicious
• Prognosis: helping estimate tumor burden and risk of recurrence
• Treatment monitoring: tracking how a tumor is responding after ablation, embolization, systemic therapy, or transplant
• Post-treatment follow-up: watching for recurrence

One critical point: in modern liver cancer care, imaging with contrast-enhanced CT or MRI often makes the diagnosis based on hallmark features. Biomarkers complement that work, not replace it.⁴

Core blood biomarkers for HCC

Alpha-fetoprotein (AFP)

What it is: AFP is a protein made by the fetal liver. Some liver cancers switch this fetal program back on and leak AFP into the blood. That's why AFP can be a useful signal.

How it's used: AFP may be checked every 6 months in surveillance for people at higher risk and is commonly followed during treatment.⁵ A rising AFP, especially if confirmed on repeat testing, can be a clue that needs imaging attention. A falling AFP after therapy suggests tumor kill. The half-life of AFP is about a week, so meaningful changes are usually tracked over several weeks rather than days.

Limits to know: Not all HCC makes AFP. Many early cancers produce little or none. And AFP can be elevated for reasons that are not cancer, including active hepatitis, pregnancy, and some germ cell tumors.¹ That's why a "positive" AFP does not diagnose cancer and a "normal" AFP does not rule it out.

AFP-L3% (the "L3 fraction" of AFP)

What it is: AFP comes in slightly different sugar-decorated forms called glycoforms. AFP-L3 binds a lectin called Lens culinaris and is more closely linked to malignant hepatocytes.⁶ If you've ever compared collagen types in a supplement aisle, you've seen how subtle structural differences can matter. AFP-L3 is the "type" of AFP that tends to show up when cancer is driving the signal.

How it's used: Many labs report AFP-L3% alongside AFP. A higher L3 fraction may increase the likelihood that an AFP elevation is coming from tumor rather than inflamed liver.⁷ It has also been studied for risk stratification and detection of earlier-stage HCC when combined with other markers.

Limits to know: AFP-L3% is an add-on, not a stand-alone test. Its performance depends on the AFP level and on the assay platform. It is not universally recommended for routine surveillance in all guidelines. Results can vary between laboratories.

Des-gamma-carboxy prothrombin (DCP, also called PIVKA-II)

What it is: DCP is an abnormal form of prothrombin made when vitamin K–dependent processing is disrupted in tumor cells.¹ Elevated DCP may reflect biologic features of HCC that are different from the AFP story.

How it's used: DCP can complement AFP, especially for tumors that do not produce much AFP. Some regions use DCP more commonly in surveillance and diagnosis support, and it is included in several risk scores.

Limits to know: DCP can rise for non-cancer reasons, notably vitamin K deficiency and use of vitamin K antagonists such as warfarin. Always tell your care team about anticoagulants and nutritional factors when interpreting DCP.

Composite scores: bringing signals together

Biology is messy. Combining measurements often outperforms any single marker. Two examples you may hear about:

GALAD score: Uses Gender, Age, AFP, AFP-L3%, and DCP to estimate the probability of HCC.⁸ In studies across different countries, GALAD has shown strong accuracy for detecting HCC, including earlier stages. It is not yet universally adopted in guidelines but is increasingly used as a risk stratification tool alongside imaging.

GALADUS: Extends GALAD by adding ultrasound findings. This hybrid "lab plus imaging" approach mirrors real-world decision-making: blood + picture + clinical context.

Reminder: Scores help estimate probability. They do not make a diagnosis by themselves, and performance can vary by population and lab method.

Imaging biomarkers and why they matter

Liver cancer is one of the few solid tumors that can often be diagnosed non-invasively by imaging patterns. On contrast-enhanced CT or MRI, HCC typically lights up brightly in the arterial phase and then "washes out" in later phases. Radiology uses a structured system (LI-RADS) to grade how likely a lesion is to be HCC.⁹ That imaging context frames how we interpret blood tests. For example, a small nodule that is indeterminate on ultrasound but accompanied by a rising AFP may prompt earlier cross-sectional imaging. Conversely, a clearly diagnostic MRI trumps a normal AFP.

Biomarkers for cholangiocarcinoma

Intrahepatic cholangiocarcinoma (ICC) arises from bile duct cells, not hepatocytes. Its blood signals are different:

CA 19-9

What it is: A carbohydrate antigen linked to bile duct epithelium.¹⁰

How it's used: Often elevated in cholangiocarcinoma, but also in benign conditions like bile duct obstruction or cholangitis.¹¹ Levels can swing with cholestasis and infection, so context is everything.

Limits to know: About 5 to 10 percent of people do not produce CA 19-9 at all due to their Lewis blood group. In them, the test can be misleadingly low despite active disease. Even when elevated, CA 19-9 is not specific for cancer.

CEA

Carcinoembryonic antigen can be modestly elevated in ICC but lacks specificity.¹⁰ It's usually considered supportive rather than decisive.

Risk markers that aren't "cancer markers," but still matter

Most HCC develops in the setting of chronic liver injury. Markers that reflect scarring and inflammation influence who benefits from surveillance and how easy it is to find small tumors.

Noninvasive fibrosis tools like FIB-4 (which uses age, platelets, and liver enzymes) and elastography (liver stiffness by ultrasound or MRI) help sort risk.¹² They do not detect cancer, but they help define the group in whom surveillance with ultrasound and periodic AFP is most useful. For example, people with cirrhosis from hepatitis B, hepatitis C, alcohol-related disease, or NASH are typically considered for surveillance regardless of their AFP on any given day.³

How to think about results: single numbers vs patterns

With biomarkers, movement matters. If your AFP edges up from 6 to 9 ng/mL over a year without a change in imaging, that can be very different from a jump from 50 to 300 over two months after therapy. The first might reflect background liver activity. The second suggests active tumor biology.

Two practical interpretation habits:

• Confirm unexpected changes: Repeating an outlier can rule out lab error and show whether a shift is real
• Align with imaging: Time blood draws and scans so they answer the same question in the same window

This is the same logic athletes use with workout recovery: you don't judge progress off one workout; you stack sessions and look at the curve.

Cutoffs and units: why numbers differ between labs

There isn't a single universal "positive" number for AFP, AFP-L3%, or DCP. Cutoffs vary by assay, and results can be reported in different units.¹ That means a value from Lab A may not map neatly onto Lab B's reference ranges. When possible, stick with the same lab and method for serial monitoring. If you switch labs, treat the first test after a switch as a new baseline.

Special scenarios that change the read

Pregnancy

Pregnancy raises AFP naturally because the placenta and fetal liver make it. AFP is not a reliable cancer signal during pregnancy.

Children

Pediatric liver tumors differ biologically. Hepatoblastoma, more common in very young children, often produces high AFP and is monitored differently than adult HCC. Pediatric oncology follows its own protocols and cutoffs.

Vitamin K status and anticoagulants

Because DCP reflects Vitamin K–dependent processing, dietary deficiency or vitamin K antagonist drugs can raise DCP in the absence of cancer. Your medication list and nutrition matter for interpretation.

Bile duct obstruction

CA 19-9 can surge with cholestasis from gallstones or strictures.¹¹ Levels may fall when bile flow is restored. That dynamic can help separate obstruction biology from tumor biology.

Obesity and NAFLD

Nonalcoholic fatty liver disease can lead to HCC, sometimes even without cirrhosis. Ultrasound may be less sensitive in people with higher body mass or very fatty livers, which can blunt surveillance sensitivity. In these cases, teams may lean more on blood markers and consider cross-sectional imaging when ultrasound windows are limited. Weight-loss drugs like GLP-1 agonists are changing the NAFLD landscape, though it is too early to say how they affect long-term liver cancer risk.

How biomarkers shape major decisions

Surveillance

Most guidelines recommend ultrasound at regular intervals for people at elevated risk, with or without AFP.³ Adding AFP can catch some cancers that ultrasound misses, particularly when the view is limited. However, ultrasound plus AFP still misses a fraction of early tumors. That's why consistent, repeated surveillance is the key—like checking your step count daily rather than once a month.

Diagnosis

When imaging shows a classic HCC pattern, a diagnosis can be made without biopsy in many adults with cirrhosis.⁴ If images are indeterminate, biomarkers can tip probability. For example, an enlarging nodule with a rising AFP or DCP increases suspicion and may lead to diagnostic MRI, CT, or biopsy. For cholangiocarcinoma, diagnosis relies more heavily on imaging and tissue, because CA 19-9 is too non-specific to stand alone.¹⁰

Staging and prognosis

Higher AFP levels tend to track with larger or more aggressive tumors, and persistent elevation can signal higher recurrence risk after local therapy.¹ Some transplant programs incorporate AFP into selection and prioritization models because it reflects tumor biology beyond size and number alone. AFP is not destiny, but it informs risk.

Treatment monitoring

After ablation or embolization, AFP that falls in line with its half-life suggests a good local response. A plateau or new rise prompts imaging to look for residual disease. During systemic therapy, directional changes in AFP or DCP can add context to scans, which sometimes lag behind biology.

Real-world examples of interpretation

Scenario 1: A person with cirrhosis has stable ultrasound exams for two years. AFP sits around 7 to 10 ng/mL, with small bumps during a flare of hepatitis. This pattern often reflects background liver activity rather than cancer. Continued routine surveillance is appropriate, and imaging drives decisions.

Scenario 2: A new 1.8 cm liver nodule appears indeterminate on ultrasound. AFP rises from 12 to 48 ng/mL over three months and AFP-L3% is elevated.⁷ This combination raises the probability of early HCC and typically prompts diagnostic MRI or CT to look for arterial enhancement and washout.

Scenario 3: After ablation of a 2.5 cm HCC, AFP drops from 180 to 30 ng/mL over six weeks and then to 8 ng/mL at three months, with a clean MRI. This trend is reassuring. AFP becomes a personalized tumor marker to track over time.

Scenario 4: A cholestatic picture with jaundice, very high CA 19-9, and a bile duct stricture. After stenting, CA 19-9 falls by half. That fall supports the idea that obstruction was driving the marker. Further imaging and tissue assess for cholangiocarcinoma if suspicion remains.

Common pitfalls and how clinicians avoid them

• Chasing a single AFP blip: repeat testing and aligning with imaging reduces false alarms
• Ignoring medications: warfarin can raise DCP, making it look falsely "tumor-like"
• Relying on CA 19-9 alone: obstruction and infection can inflate levels, especially in biliary disease¹¹
• Mixing labs and methods: switching assay platforms mid-course can make trends look misleading¹
• Forgetting life stage: pregnancy and pediatrics have different baselines and playbooks

Modern immunoassays are robust, but no test is immune to real-world quirks. Hemolysis, lipemia, and very high bilirubin can interfere with some measurements. Timing relative to therapy matters, especially with AFP's half-life. The reference interval on your report reflects your lab's method and validation. When you read a number, read the units and the lab's reference ranges too.

Sex, age, and biology

HCC is more common in men than women.⁸ Age raises baseline risk in chronic liver disease. These variables matter enough that the GALAD score formally includes sex and age. They do not change the interpretation of a single lab value but do influence the pre-test probability that a given pattern represents cancer.

When tissue is still the issue

Biomarkers and imaging can lean strongly in one direction, yet sometimes the picture stays uncertain. In those cases, a liver biopsy can provide the definitive answer and enable molecular testing. For cholangiocarcinoma, tissue diagnosis is often essential because blood markers lack specificity. For HCC, tissue is used when imaging is inconclusive or when targeted therapy decisions require genomic information.

What to discuss with your care team

• Which surveillance plan fits my specific risk and imaging visibility
• Which biomarkers we will follow and how often
• What counts as a meaningful change for me, given my baseline
• How medications, supplements, or nutrition might skew results
• Which lab will run my tests so serial results are comparable

The research horizon

Liver cancer biology is evolving fast. Liquid biopsy approaches, including circulating tumor DNA and methylation signatures, are being studied to detect cancer earlier and to map tumor mutations from a blood draw.¹³ Early results are promising in some cohorts, but clinical use is still limited and assays are not yet standardized across centers. Multi-omic models that blend proteins, DNA fragments, and imaging features may become the next-generation "weather forecast," offering better early detection for people with fatty liver or hard-to-image livers, though more research is needed.

A note on guidelines and global practice

Society guidelines (such as from AASLD and EASL) align on the big picture: use regular imaging for surveillance in high-risk groups, consider AFP as an adjunct, and rely on diagnostic CT or MRI features for many HCC diagnoses.¹⁴ ¹ ⁵ The use of AFP-L3%, DCP, and composite scores is more common in some regions than others. Your team will tailor the toolkit based on local practice standards, lab availability, and how your liver looks on imaging.

Key takeaways you can hold onto

• Biomarkers are supporting actors: valuable for trends, not definitive alone
• AFP, AFP-L3%, and DCP tell different biological stories that complement imaging¹
• CA 19-9 helps more with cholangiocarcinoma but is easily skewed by bile duct obstruction
• Life stage, meds, and lab methods can nudge numbers up or down
• Consistency beats intensity: same lab, regular intervals, aligned with scans

Closing thought

Biomarkers for liver cancer work best when they're treated like instruments in a band. Each has its voice. Together, with imaging and clinical judgment, they can play a clear tune. Use them to follow the pattern over time, not to panic over a single note. And remember, interpretation is the job of your care team, grounded in your liver's story, not just the lab's numbers.