One of the lowest survival rates of any cancer, because it is usually found late. The first genome-wide scan of it returned the blood group gene, confirming an observation from decades of hospital records, and the later ones returned the genes that build the pancreas and make its digestive enzymes.
What this condition connects to
Solid lines are connections this site curates. Dashed lines mean the two ends share a research paper — worth knowing, and not a claim that one explains the other.
Prevalence
The first genome-wide association study of pancreatic cancer identified a single locus, rs505922 in the first intron of ABO, with a per-allele odds ratio of 1.20, consistent with earlier epidemiological evidence that blood group O carries a lower risk than groups A or B (Amundadottir et al., Nature Genetics 2009, PMID 19648918). A later study of 7,683 cases and 14,397 controls identified loci at PDX1 and BCAR1/CTRB1/CTRB2 and an independent signal in TERT (Wolpin et al., Nature Genetics 2014, PMID 25086665), and a 2015 study of 9,925 cases and 11,569 controls added SUGCT and TP63 (Childs et al., Nature Genetics 2015, PMID 26098869).
Inheritance
Common variants, each shifting an already small baseline risk slightly. Distinct from the rare high-penetrance mutations in BRCA2, PALB2, ATM, BRCA1, STK11, CDKN2A and the mismatch-repair genes, which run in families in a visible pattern and are the basis for surveillance. None of those genes is on this page.
The pancreas sits behind the stomach and does two jobs: it secretes the enzymes that digest food, and it makes insulin. Cancer of it is uncommon relative to breast, bowel or lung cancer, but it is among the leading causes of cancer death, because it usually causes no symptoms until it has spread and there is no screening test for the general population.
That is the fact this page has to be read against. The variants here each shift risk a little. None of them detects the disease, and nothing on this page is a substitute for the two things that do change outcomes, which are described further down.
The first answer was a blood group
The first genome-wide association study of pancreatic cancer, published in 2009, genotyped 1,896 cases and 1,939 controls from twelve prospective cohorts and then combined them with a further 2,457 cases and 2,654 controls. It found one locus, and the locus was ABO — the gene that determines blood group.
The paper says why this is not a surprise: it is consistent with earlier epidemiological evidence that people with blood group O have a lower risk of pancreatic cancer than people with groups A or B. That observation came from decades of counting blood groups among patients, long before anyone could read a genome, and a method with no way of being steered towards it reproduced it. The variant, rs505922, sits in the first intron of the gene and carries a per-allele odds ratio of about 1.20; this site holds it under this trait, and the ABO blood group page explains what the gene does the rest of the time.
Then the genes that build the pancreas
Five years later a study of 7,683 cases and 14,397 controls of European descent found four new loci, and two of them are the pancreas's own genes:
PDX1 — the transcription factor that directs the formation of the pancreas in the embryo and keeps insulin-producing cells working afterwards. The site's catalogue labels this variant by its nearest gene, URAD; the paper names PDX1.
CTRB1 and CTRB2 — the chymotrypsinogen genes, whose product the pancreas secretes into the gut to digest protein. This is the locus with the largest effect on the page, an odds ratio of about 1.46. The catalogue's label here is ZFP1, the nearest gene; the paper names BCAR1, CTRB1 and CTRB2.
TERT — telomerase, the enzyme that maintains the ends of chromosomes, and a region that turns up in the genetics of many cancers. The same study found an independent signal in exon 2 of the gene, and a 2016 follow-up found a further one nearby at CLPTM1L.
A scan that could have returned anything returned the developmental regulator of the organ and its digestive enzymes. The remaining variants on the page come from the same consortium's later work — SUGCT and TP63 from a 2015 study of 9,925 cases and 11,569 controls, and MYC from the 2016 imputation — and from a 2011 study in Chinese populations, which is where BACH1 comes from and which the European studies later confirmed at several loci.
Clinical detail
Two kinds of inherited risk, and this page is the smaller one
Pancreatic cancer has a rare inherited form. The 2015 study lists the genes: BRCA2, ATM, PALB2, BRCA1, STK11, CDKN2A and the mismatch-repair genes. A change in one of those is a high-penetrance mutation, it runs in families in a visible pattern, and it is the kind of finding that gets someone into a surveillance programme.
None of those genes is on this page. The variants here are the other kind — common, each moving an already small baseline risk by a fraction — and they do not qualify anyone for anything. If pancreatic cancer runs in your family, the useful question is about those rare genes and a specialist can ask it. The variants below cannot answer it.
What changes an outcome
Because there is no population screening test, what changes outcomes is being seen at the right moment. Two things get pancreatic cancer looked for early enough to matter.
A family history. Two or more close relatives with pancreatic cancer, or a known mutation in one of the genes above, is the usual basis for being offered surveillance by imaging. That decision is made by a specialist on the family history, not on a genotype from a consumer test.
The symptoms. Yellowing of the skin or eyes without pain, unexplained weight loss, pain in the upper abdomen that goes through to the back, or diabetes that appears for the first time in later life without an obvious reason. None of these is specific and all of them are worth a prompt appointment. Nothing on this page detects the disease and nothing on this page should delay that appointment.
What this page cannot do
It cannot screen for pancreatic cancer. Nothing can, for the general population, which is the central problem of this disease.
It cannot tell you your family's risk. That comes from the rare high-penetrance genes and the pattern of relatives affected, and this page holds neither.
It cannot tell you your blood group. The ABO variant here tags the group, and the effect on risk is small; if you want to know your blood group, a blood test says so directly.
Related here
The ABO blood group page covers the gene the first finding sits in. TERT and CLPTM1L also appear on the lung cancer page, because the same telomere-maintenance region shifts risk for several cancers at once.
Related variants MyGeneLog checks for
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Pancreatic Cancer comes down to these specific, well-studied positions — not a diagnosis.
Can a DNA test tell me if I will get pancreatic cancer?
No. The variants here each shift risk slightly and none of them detects the disease. There is no screening test for pancreatic cancer in the general population, which is the central problem of this cancer, and a consumer genotype does not change that.
Pancreatic cancer runs in my family. What should I do?
Ask about the rare inherited genes rather than about this page. Two or more close relatives, or a known mutation in a gene such as BRCA2, PALB2 or CDKN2A, is the usual basis for being offered surveillance, and that decision is made by a specialist on the family history.
Does my blood group really affect my risk?
A little. People with blood group O have a somewhat lower risk than those with A or B, an observation made from hospital records long before genetics could test it, and the first genome-wide study of this cancer reproduced it by finding the ABO gene. The effect is small and does not change what anyone should do.
Why do the studies find pancreas genes?
Because they are looking at a pancreatic cancer with a method that has no idea what it is looking at. The loci include PDX1, which builds the pancreas in the embryo, and the chymotrypsinogen genes CTRB1 and CTRB2, whose product the pancreas secretes to digest protein. Getting the organ's own genes from an unbiased scan is what makes the scan believable.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.