What Is CA 19-9? What Level Is Concerning, and How Is It Related to Pancreatic Cancer?
In this article
What is CA 19-9, and what is it used for?
There is only one question that CA 19-9 can answer: "How is already-diagnosed pancreatic cancer progressing?" It answers this question well. By contrast, it can barely answer the two questions it is most often asked: "Does this healthy person have cancer?" — in healthy people, the probability that an elevated CA 19-9 truly indicates cancer is less than one percent. And "Is this mass in a patient with jaundice benign or malignant?" — in the presence of biliary obstruction, the test's ability to distinguish between them is close to a coin toss.
- The most important number in this article: In a study screening 70940 healthy people, only 4 cancers were found; the probability that an elevated CA 19-9 actually represented cancer was 0.9%. In other words, ordering it as a "check-up" test means that 99 of every 100 alarms are false.
- A normal result is not reassurance. In the population, 5-10% are genetically unable to produce this molecule at all (Lewis antigen negativity), and in operable early-stage cancers, CA 19-9 is normal in approximately 46%.
- CA 19-9 measured during jaundice cannot be interpreted. In the guideline's own words, the baseline value should be obtained "after adequate biliary drainage" . Discrimination is 0.598 before drainage, which is not significant, and rises to 0.699 afterward.
- The widely circulated "risk table" is not defensible. The claim that "above 1000 almost certainly means cancer" is wrong: benign biliary obstruction can also produce values above 1000, and complete resection (R0) was still possible in 15% of patients with values above 1000.
- CA 19-9 has no place in screening high-risk families. The methods used to monitor people with hereditary risk are MRI and endoscopic ultrasound; international protocols do not include this test.
"Cancer Antigen 19-9" (Cancer Antigen 19-9). It is a glycoprotein containing a carbohydrate structure on the cell surface called sialyl-Lewis A . It can be produced by cells in the pancreas, biliary tract, stomach, and intestine — by both cancerous and healthy tissues. It is measured with a simple blood test, and values below 37 U/mL are generally considered normal.
The term "cancer antigen" in its name is misleading: this molecule is not specific to cancer; in cancer it is produced in greater amounts . This entire article addresses the practical consequences of that distinction.
Why not test healthy people? One number
In a study screening 70940 people without symptoms, those with elevated CA 19-9 underwent further investigation. The number of cancers ultimately found was: 4. Thus, the probability that an elevated CA 19-9 truly indicated cancer — its positive predictive value — was 0.9% . The study's own conclusion was: "not useful as a screening test."
What does this mean? Of every 100 elevated CA 19-9 results in healthy people, approximately 99 are false alarms — yet each triggers the same chain of investigations: CT, MRI, sometimes endoscopic ultrasound, and weeks of anxiety. This is the hidden cost of "let's check just in case."
This has particular relevance in Turkey: CA 19-9 has become a test sold directly to consumers in Turkey, and some providers openly market it as "pancreatic cancer screening" — precisely the use rejected by all guidelines.
The Test's Actual Performance
| Question | CA 19-9 Performance |
|---|---|
| Does it detect pancreatic cancer in symptomatic patients? | Sensitivity 79-81%, specificity 82-90%. This is the setting in which the test performs best. |
| Does it detect early-stage disease? | No. In stages 1-2, sensitivity is 54% — meaning that CA 19-9 is normal in approximately 46% of operable cancers. When assessed by tumor size, sensitivity falls to around 40% for the smallest tumors (T1). |
| Does it distinguish benign from malignant disease in a patient with jaundice? | Almost not at all. Discrimination (AUC) is 0.78 against healthy controls and 0.73-0.77 against chronic pancreatitis; against benign biliary obstruction it is 0.56-0.62 — close to a coin toss when faced with the actual differential diagnosis. |
| Are there patients who do not produce it at all? | Yes. In the population, 5-10% are Lewis antigen-negative and cannot synthesize this molecule at all. In the guideline's words: "CA 19-9 will be undetectable in Lewis antigen-negative individuals (testing CEA and CA-125 may be considered)." |
A normal result does not reassure; an elevated result does not condemn.
Why a normal result does not reassure: Lewis negativity (5-10%) + normal levels in approximately half of early-stage cancers. A patient with operable pancreatic cancer can therefore quite easily have a CA 19-9 of 20.
Why an elevated result does not condemn: the benign causes described below — especially biliary obstruction.
Biliary Obstruction: The Biggest Confounder
This is the article's most practical rule, stated verbatim in the guideline: baseline CA 19-9 should be measured "after adequate biliary drainage" . A footnote in the same text states that an elevation "may be due to biliary obstruction, infection (cholangitis), or inflammation — benign or malignant" .
The numbers support this: In a study of 121 patients, the test's discrimination before drainage was 0.598 (not statistically significant), and after drainage it rose to 0.699 (p=0.006). Even more informative was the extent of the decrease after drainage: 75.7% in patients with benign disease and 39.2% in those with malignant disease. In other words, the most informative finding is not a single value, but how much it falls after the obstruction is relieved.
Why not "adjust for bilirubin"? Because the relationship between CA 19-9 and bilirubin has been found to be weak. The solution is not a formula, but repeat measurement — after bile flow has been restored.
Studies have reported false-positive rates in obstructive jaundice ranging from 10-60% .
Why Is the Common "Risk Table" Wrong?
A table often found in Turkish sources classifies 38-100 as benign, 100-1000 as highly suspicious, and above 1000 as "almost certainly cancer." This classification is wrong in both directions.
| Common Claim | What Do We Actually Know? |
|---|---|
| "100-1000 raises strong suspicion of serious disease" | This entire range can be accounted for by benign biliary obstruction: the median value in benign cases was 267.7 , and the interquartile range extended to 639.9. |
| "Above 1000 is highly likely to be cancer" | Benign biliary obstruction, cholangitis, and acute pancreatitis can produce values above 1000. Furthermore, among pancreatic cancer patients with CA 19-9 above 1000, complete resection (R0) was still possible in 15% — so "above 1000 = inoperable" is also incorrect. |
| "If it is elevated, there must be a disease" | A study of 192 patients with CA 19-9 of 80 or above found no cause in 45 (23.4%) — none of these patients developed cancer within two years, and in 35 of the 45 patients, the value returned to normal spontaneously. Values in the same study ranged from 80 to above 1000, and none of the patients had pancreatic or biliary disease. |
The right question is not "How high was it?" but "In whom was it measured?". The same value of 300 U/mL has four different meanings in four different patients:
- In someone having a check-up without symptoms: most likely meaningless. The probability of cancer is below 1%.
- In a patient with jaundice: uninterpretable. Relieve the obstruction first, then measure it again.
- In a patient with a pancreatic mass detected on imaging: meaningful — it supports the diagnosis and provides prognostic information.
- In someone receiving treatment or who has undergone surgery: this is where the test is truly valuable — but again, the key is not one value, but the trend.
Non-Cancer Causes of Elevation
In a study of 192 patients with CA 19-9 of 80 or above, the causes were distributed as follows:
| Cause | How High Can It Rise? |
|---|---|
| Biliary obstruction, gallstones, cholangitis | Can rise above 1000. This is the benign cause that produces the most dramatic elevations. |
| Acute pancreatitis | Can produce high values. |
| Chronic pancreatitis, cirrhosis, diabetes, thyroid disorders, cystic fibrosis, endometriosis and ovarian cysts, interstitial lung diseases | Cause mild to moderate elevations. We found no verified data showing that these conditions produce values above 1000. |
Where It Is Truly Valuable: In a Patient With a Diagnosis
Before Surgery: Prognosis and the Decision to Operate
| Preoperative CA 19-9 | What Does It Indicate? |
|---|---|
| Below 37 U/mL | In a series of 1543 patients, 80% underwent surgical resection, with 5-year survival of 27% |
| 130 U/mL or above | Rate of finding otherwise unseen spread at laparoscopy that precluded surgery: 26.4% (11% below 130) |
| 1000 U/mL or above | Complete resection (R0) rate of only 15% |
| 4000 U/mL or above | Resectability: 38%, but 5-year survival of 0% |
The authors of the above series did not identify any CA 19-9 value as a criterion for exclusion from surgery. Nor do guideline texts specify a numerical threshold for "biologically borderline resectability"; the wording used is "markedly elevated CA 19-9", described as one of the high-risk features that support chemotherapy before surgery (neoadjuvant treatment).
The guideline's wording on whether to proceed with surgery also relies on the trend: surgical exploration proceeds when the value "at least remains stable or decreases" .
During and After Treatment
- Response to chemotherapy: The most commonly used criterion is a decrease of more than 20% by week 8 — but there is no single agreed threshold, and this measure does not replace imaging.
- Normalization after surgery: This is a strong prognostic indicator. In patients whose levels normalized within the first year, median survival was 34 months , compared with 13 months in those whose levels did not normalize.
- Preoperative level: In those with levels below 37, median survival was 32-36 months, compared with 12-15 months in those with higher levels.
Postoperative Follow-Up: What Does the Guideline Say?
For postoperative follow-up, the guideline recommends assessment for 2 years at intervals of 3-6 months, followed by intervals of 6-12 months . However, CA 19-9 is included in this follow-up only at the "category 2B" level — a low-level recommendation without full consensus.
The following sentence is particularly important: "An elevated CA 19-9 without other evidence of recurrence is not a clear indication for treatment."
What this means for patients: If your CA 19-9 rises during follow-up, this alone is not a reason to start treatment. The measurement is repeated, evaluated alongside imaging, and a decision is made based on the whole picture.
Screening High-Risk Families: No CA 19-9
Defined surveillance protocols exist for people with a family history of pancreatic cancer, certain genetic alterations such as BRCA2 and PALB2, hereditary pancreatitis, and certain pancreatic cysts. The methods used in these protocols are MRI/MRCP and endoscopic ultrasound.
CA 19-9 is not included in these protocols. The relevant international consensus documents and endoscopy society guideline do not mention this test in the screening algorithm. In the words of one review, it "has not been included in the screening algorithm, although it may have an adjunctive role."
Why does this matter? Because CA 19-9 is the most readily accessible test for an anxious person with a family history of pancreatic cancer — and is especially misleading in that very person. A normal result offers no reassurance (it is normal in 46% of early-stage cases); an elevated result is most likely a false alarm.
Its Role in Other Cancers and Its Relationship to Other Markers
| Cancer | Role of CA 19-9 |
|---|---|
| Pancreatic cancer | Its main use. Prognosis, treatment response, and follow-up. |
| Biliary tract cancers (cholangiocarcinoma), ampullary and small intestinal cancers | Among its appropriate uses. |
| Appendiceal cancers | May be used. |
| Ovarian cancer and cancers of unknown primary | May be used in certain circumstances. |
| Colon and rectal cancers | ⚠ The guideline does not recommend CA 19-9 testing except for cancers originating in the appendix. The marker for these cancers is CEA. Older statements suggesting routine CA 19-9 testing in colorectal cancer are no longer current. |
| Stomach cancer | No guideline support was found. |
In a blinded study published in 2026 involving 649 people, adding two proteins (ITIH3 and CEACAM1) to CA 19-9 increased discrimination from 0.902 to 0.917. The main gain, however, lay elsewhere: at the same sensitivity, specificity rose from 14.5% to 53.3% — substantially reducing false alarms. This panel is not yet in clinical use, but it shows the direction of the field: overcoming the limitations of a single marker appears to depend more on adding other markers than on interpreting it better.
Frequently Asked Questions
- Should I have CA 19-9 tested during a check-up?
- No. In a study screening 70940 people without symptoms, the probability that an elevated CA 19-9 truly indicated cancer was 0.9% — approximately 99 of every 100 alarms were false. Conversely, a normal result offers no reassurance: 5-10% of the population cannot produce this molecule at all, and levels are normal in 46% of operable early-stage cancers. In healthy people, the test is misleading in both directions.
- My CA 19-9 is 250. Do I have cancer?
- The answer comes not from the number, but from the context . If you have jaundice, the result cannot be interpreted — the obstruction must first be relieved and the measurement repeated. A median of 267.7 has been reported in benign biliary obstruction, right in your range. If you have no symptoms, it is most likely meaningless. If imaging has detected a pancreatic mass, it is meaningful. The first step is to repeat the measurement and investigate the cause, rather than interpret the result in isolation.
- It is above 1000. Is it definitely cancer?
- No, it is not "definite." Benign biliary obstruction, cholangitis, and acute pancreatitis can also produce values above 1000. Nor is the reverse true: among pancreatic cancer patients with CA 19-9 above 1000, complete resection (R0) was still achieved in 15% — so this value does not mean "inoperable" either. A high value is taken seriously, but diagnosis depends on biopsy and imaging.
- I have jaundice and had CA 19-9 tested. Is the result reliable?
- No. In the guideline's own words, the baseline value should be measured "after adequate biliary drainage" . During jaundice, the test's ability to distinguish benign from malignant disease is 0.598, which is not statistically significant; after drainage it rises to 0.699. Mathematical adjustment for bilirubin is not possible either, because the relationship between the two is weak. There is only one solution: repeat the measurement after relieving the obstruction. Meanwhile, the most informative finding is how much the value falls after drainage.
- My level is high, but all my investigations were normal. What happens now?
- This is much more common than assumed and is a category in its own right. In a study of 192 patients with CA 19-9 of 80 or above, no cause was identified in 23.4% — none developed cancer during two years of follow-up, and in the vast majority, the value returned to normal spontaneously. The usual approach is to exclude confounding causes, repeat the measurement after some time, and monitor the patient.
- I have a family history of pancreatic cancer. Can CA 19-9 be used to monitor me?
- No. The methods used to monitor high-risk individuals are MRI/MRCP and endoscopic ultrasound; CA 19-9 is not part of international surveillance protocols. The reasons are the performance figures above: sensitivity of 54% in early-stage disease and a positive predictive value of 0.9% in healthy people. Trusting a normal CA 19-9 in someone at risk means trusting it precisely where it should not be trusted. This surveillance should be planned at an experienced center.
- I am receiving chemotherapy and my CA 19-9 is falling. Is that a good sign?
- Usually, yes. The most commonly used criterion is a decrease of more than 20% by week 8 — but there is no fully agreed threshold, and it does not replace imaging. The reverse also applies: a single increase is not enough to change treatment. The decision is made by evaluating the marker trend together with imaging.
- How can CA 19-9 be lowered?
- There is no drug or method that directly lowers CA 19-9 — nor would that make sense, because CA 19-9 is not the disease itself, but a marker of it. The way to lower it is to treat the cause of the elevation: the value falls when biliary obstruction is relieved, pancreatitis resolves, or cancer treatment produces a response. A logical error similar to trying to reduce a fever by chilling the thermometer in a freezer is often made with this test.
Glossary
- CA 19-9
- Cancer Antigen 19-9. A glycoprotein containing the sialyl-Lewis A structure. It is not specific to cancer.
- Lewis antigen negativity
- A genetic characteristic found in 5-10% of the population. These individuals cannot produce CA 19-9 at all; the test remains normal even if they have advanced cancer.
- Sensitivity / specificity
- Respectively, the proportion of people with disease detected by the test and the proportion without disease whose results remain normal.
- Positive predictive value (PPV)
- The proportion of people with elevated results who actually have the disease. It becomes very low in groups where the disease is rare — which is why it is 0.9% in healthy people.
- AUC (area under the curve)
- A test's ability to distinguish between two conditions. 1.0 means perfect discrimination; 0.5 means a coin toss.
- Cholestasis / obstructive jaundice
- Obstruction of bile flow. It is the benign cause that raises CA 19-9 most dramatically.
- Biliary drainage
- Opening an obstructed bile duct through an endoscopic or interventional procedure. CA 19-9 should be measured afterward.
- R0 resection
- Complete surgical removal of the tumor without any tumor cells remaining at the surgical margins.
- Neoadjuvant treatment
- Chemotherapy or radiotherapy given before surgery.
- Category 2B
- An evidence level used in guidelines, indicating a weaker recommendation without full consensus.
DROZDOGAN Academy Commentary
CA 19-9 is genuinely valuable in the right patient. The problem is ordering it for the wrong patient — and in Turkey this is no longer merely a habit, but a service sold directly to consumers.
⚡ Why does this matter?
Pancreatic cancer is among the cancers people fear most because it is detected late, and this fear drives the search for "a blood test that catches it early." CA 19-9 is not that test. Some laboratories in Turkey openly offer it as "pancreatic cancer screening" in a bookable package — precisely the use rejected by all guidelines. This has two costs: ninety-nine out of a hundred alarms are false, and a normal result provides false reassurance.
✅ Strengths / what is improving
CA 19-9 really is useful in the right setting. Preoperative levels are clearly linked to prognosis (5-year survival of 27% below 37; 0% above 4000). Normalization within the first year after surgery is a strong indicator, with median survival rising from 13 to 34 months. Guidelines also assign it a realistic role: only category 2B in surveillance, and "not an indication for treatment on its own." Repeating the measurement after biliary drainage is a simple and effective rule. The 2026 multimarker panel study targets the right problem: rather than increasing sensitivity, reducing false alarms.
⚠️ Limitations / what should not be overstated
The test is weakest where it is used most: in patients with jaundice, its ability to distinguish benign from malignant disease is 0.56-0.62, almost a coin toss. The common "risk table" is wrong in both directions: benign biliary obstruction spans the entire 100-1000 range and can exceed 1000; meanwhile, 15% of patients above 1000 can still undergo complete resection. A "20% decrease" is widely used to assess treatment response but is not an agreed threshold. Unexplained elevations are also far from negligible: 23.4% in one series.
🩺 Our Approach in Practice
- We do not order CA 19-9 for people without symptoms — either to look for cancer or "for reassurance." Neither approach works.
- In patients with jaundice, we do not interpret values measured before drainage. We repeat the measurement after relieving the obstruction and obtain the key information from the extent of the decrease .
- We do not initiate a chain of investigations based on a single elevated value. First repeat the test, then exclude confounding causes.
- We do not base the decision to operate on CA 19-9. A high value is a risk feature supporting neoadjuvant treatment, not a criterion for exclusion from surgery.
- We do not regard rising CA 19-9 during follow-up as sufficient reason for treatment on its own — this is also the guideline's own wording.
- In high-risk families, we do not replace imaging with any blood test. A normal CA 19-9 does not replace MRI or endoscopic ultrasound.
- We do not order CA 19-9 for colorectal cancer — CEA is the marker for that cancer.
❓ Open Questions
Would routinely determining Lewis antigen status before treatment make sense, given that CA 19-9 monitoring is meaningless from the outset in a known negative patient? There is no consensus on how long unexplained elevations should be monitored. And the main question is whether multimarker panels or circulating tumor DNA can truly close the 46% blind spot in early-stage disease. At present, they cannot.
- Corrections: In the previous text, the risk table was removed and replaced with a context-based framework. The table was wrong in both directions: benign biliary obstruction spans the 100-1000 range and can exceed 1000; meanwhile, the claim that "above 1000 is highly likely to be cancer" is inconsistent with the fact that complete resection was achieved in 15% of patients above 1000. The statement that CA 19-9 may be tested in colon and rectal cancers was corrected: the guideline does not recommend this except for cancers originating in the appendix. General statements about liver, lung, and ovarian cancers were restricted to uses supported by guidelines.
- Additions: The positive predictive value in healthy people (0.9%; 70940 individuals). Stage-specific sensitivity (54% in early-stage disease) and the resulting finding of "normal in 46%." The inability to interpret the test during biliary obstruction and the rule to repeat it after drainage. The prognostic and surgical decision-making value of preoperative levels. The association between postoperative normalization and survival. Follow-up intervals and the category 2B status of CA 19-9 in follow-up. The absence of CA 19-9 from surveillance protocols for high-risk families. Unexplained elevations and their course. The multimarker panel study.
- Source correction: The screening positive predictive value comes from a Korean study; the widespread attribution of this figure to Japan could not be verified.
- Figures not provided because they could not be verified: Pooled sensitivity and specificity estimates from meta-analyses (primary texts were inaccessible; the ranges provided come from reviews). The ethnic distribution of Lewis antigen negativity. Precise study-specific thresholds for chemotherapy response. The ASCO text on surveillance intervals (inaccessible, so only the available guideline was cited). No data were found showing CA 19-9 above 1000 in cirrhosis, diabetes, thyroid disease, endometriosis, or interstitial lung disease.
- Turkey: No verifiable Turkey-specific clinical data or guideline was found. The observation that the test is marketed directly to consumers in Turkey as "screening" is based on commercial promotional pages and is not a scientific source; no prevalence rate is therefore given. No information is provided about reimbursement by Turkey's Social Security Institution (SGK).
- Internal links are estimated addresses and should be checked before publication.
This article is for information only and does not replace individualized medical advice. Your physician should interpret test results together with your clinical condition and imaging findings.
References
- Kim JE, Lee KT, Lee JK, et al. Clinical usefulness of carbohydrate antigen 19-9 as a screening test for pancreatic cancer in an asymptomatic population. J Gastroenterol Hepatol. 2004;19(2):182-186.
- Ballehaninna UK, Chamberlain RS. The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: an evidence based appraisal. J Gastrointest Oncol. 2012;3(2):105-119.
- Early Detection Research Network reference set analysis. PLOS ONE. 2015.
- National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Pancreatic Adenocarcinoma, v2.2025.
- Serum CA 19-9 before and after biliary drainage in the differential diagnosis of pancreatic head masses. BMC Cancer. 2021.
- Hartwig W, Strobel O, Hinz U, et al. CA19-9 in potentially resectable pancreatic cancer: perspective to adjust surgical and perioperative therapy. Ann Surg Oncol. 2013;20(7):2188-2196.
- Preoperative CA 19-9 and occult metastatic disease at staging laparoscopy. Ann Surg Oncol. 2008.
- Elevated CA 19-9 in patients without malignancy: causes and outcomes. Sci Rep. 2020.
- Goggins M, Overbeek KA, Brand R, et al. Management of patients with increased risk for familial pancreatic cancer: updated recommendations from the International Cancer of the Pancreas Screening (CAPS) Consortium. Gut. 2020;69(1):7-17.
- American Society for Gastrointestinal Endoscopy. Guideline on pancreatic cancer screening in high-risk individuals. 2022.
- National Comprehensive Cancer Network. NCCN Guidelines: Colon Cancer, v4.2025 (tumor markers section).
- CA 19-9, ITIH3 and CEACAM1 biomarker panel for pancreatic cancer detection. Cancers. 2026.
Was this article helpful?
Your feedback helps us improve our content.
Reader comments
Share your comment
Your name and comment are published only after administrator approval. Your email address is never shown to visitors.
Be the first to comment.