PancPRO calculator: familial pancreatic cancer risk from a family pedigree

Short version. PancPRO is a Bayesian calculator from the BayesMendel suite that estimates the probability an individual carries a hypothetical autosomal dominant pancreatic cancer susceptibility gene and projects their future pancreatic cancer risk, based on family history. It complements — rather than replaces — gene-specific panel testing for the genes now known to contribute to inherited pancreatic cancer: BRCA2, PALB2, ATM, BRCA1, CDKN2A, STK11, the Lynch mismatch repair genes, and hereditary pancreatitis genes. Individuals identified as high-risk may be candidates for pancreatic surveillance under CAPS consortium recommendations. Evagene runs PancPRO directly on the pedigree canvas, with no re-entry, alongside BRCAPRO and MMRpro.

Why pancreatic cancer risk matters

Pancreatic ductal adenocarcinoma carries one of the worst prognoses of any solid malignancy. Late-stage presentation is the norm, curative surgery is feasible in a minority of cases, and five-year survival has historically sat in single digits, though recent cohort data put it slightly higher in selected populations. Early detection is therefore of disproportionate clinical value, and the families and individuals at highest risk are the ones most likely to benefit from it.

Most pancreatic cancers are sporadic. However, a clinically meaningful subset — commonly estimated at around 5 to 10 percent of cases — occurs in families with a recognisable hereditary pattern, either through a defined hereditary cancer syndrome or through the phenotype known as familial pancreatic cancer (FPC) that does not yet have an identified driver gene in many kindreds.

Familial pancreatic cancer: the FPC definition

The working research definition of familial pancreatic cancer is a kindred with at least two first-degree relatives affected with pancreatic ductal adenocarcinoma, in the absence of another recognised hereditary cancer syndrome that could explain the clustering. Kindreds meeting this definition have substantially elevated risk relative to the general population, and the risk scales with the number of affected first-degree relatives.

The qualitative risk pattern generally described in the literature is roughly as follows: one affected first-degree relative confers a modest multiplicative increase in lifetime risk over the population baseline; two first-degree relatives produce a larger increase; three or more first-degree relatives represent a markedly elevated risk, particularly when at least one diagnosis occurred at young age. These figures are kindred-dependent and should not be taken as precise quantitative estimates outside of published cohort data; PancPRO is the tool designed to produce a principled estimate for a specific pedigree.

FPC is a heterogeneous phenotype. Some FPC kindreds are driven by recognised cancer predisposition genes. Others have no identified pathogenic variant on current panels, indicating that part of the genetic architecture remains to be characterised. The lack of a single responsible gene is one reason PancPRO was built around a hypothetical autosomal dominant susceptibility locus rather than a specific gene.

Genes associated with inherited pancreatic cancer

Several recognised hereditary cancer syndromes include pancreatic cancer in their tumour spectrum. The common contributors are as follows.

• BRCA2 — The most frequently identified gene in FPC kindreds. BRCA2 pathogenic variants elevate pancreatic cancer risk moderately above population baseline; the lifetime risk is lower than the associated breast or ovarian risk but non-trivial. See our BRCAPRO calculator guide for how BRCA1/2 carrier probability is estimated from family history.
• PALB2 — An important contributor to FPC. PALB2 functions in the same homologous recombination pathway as BRCA2.
• ATM — Pathogenic variants are associated with elevated pancreatic and breast cancer risk.
• BRCA1 — Less strongly associated with pancreatic cancer than BRCA2 but present on most contemporary hereditary pancreatic cancer gene panels.
• CDKN2A — The familial atypical multiple mole melanoma (FAMMM) syndrome. CDKN2A carriers face elevated melanoma and pancreatic cancer risk.
• STK11 — Peutz-Jeghers syndrome. STK11 carriers have elevated risk of multiple cancers including pancreatic, breast, gastrointestinal, and gynaecological.
• MLH1, MSH2, MSH6, PMS2 — The Lynch syndrome mismatch repair genes include pancreatic cancer in their tumour spectrum. See our MMRpro calculator guide and Lynch syndrome risk calculator.
• PRSS1, SPINK1, CFTR — Hereditary pancreatitis genes. Chronic pancreatic inflammation elevates pancreatic cancer risk over time; PRSS1 pathogenic variants cause the most clearly defined risk.
• TP53 (Li-Fraumeni) — Rare in FPC but occasionally identified.

A contemporary hereditary pancreatic cancer panel typically includes most of the above. Even with a comprehensive panel, a substantial fraction of FPC kindreds have no identified pathogenic variant, leaving a role for family-history-based models such as PancPRO to estimate risk in the absence of a specific molecular answer.

How PancPRO works

PancPRO is structurally analogous to BRCAPRO and MMRpro. It uses the Bayesian peeling algorithm at the heart of the BayesMendel framework to update a prior probability of carrying a hypothetical autosomal dominant pancreatic cancer susceptibility gene, using the pedigree's pancreatic cancer phenotype observations as likelihood contributions.

The model's inputs are pedigree structure, pancreatic cancer affected status and age of diagnosis for each affected relative, current age for unaffected relatives, and age and cause of death for deceased relatives. Ethnicity is used to select appropriate population allele frequencies. Because pancreatic cancer has a relatively low lifetime risk in the general population (approximately 1 to 2 percent), even a modest family cluster — two first-degree relatives with pancreatic cancer, for example — produces a noticeably elevated posterior carrier probability.

The output is the posterior probability of carrying the susceptibility locus and a projection of cumulative pancreatic cancer risk by specified future ages. These outputs support decisions about whether to offer gene-panel testing, referral to a high-risk pancreatic surveillance programme, or risk-reducing behavioural counselling (smoking cessation, which is particularly important given smoking's multiplicative interaction with inherited risk).

Surveillance: the CAPS recommendations

The International Cancer of the Pancreas Screening (CAPS) consortium publishes consensus recommendations for surveillance of high-risk individuals. The recommendations identify groups for whom surveillance may be considered: FPC kindred members with a specified pattern of affected relatives, carriers of specific hereditary cancer syndrome genes with or without a family history of pancreatic cancer, and individuals with hereditary pancreatitis.

Surveillance is typically delivered via annual imaging with endoscopic ultrasound or MRI/MRCP, with the starting age depending on genetic background and family history. Surveillance is resource-intensive, requires specific radiological and endoscopic expertise, and carries its own risks — false positives can lead to invasive investigation of lesions that never become clinically relevant. CAPS and equivalent guidelines emphasise that surveillance should be delivered within a multidisciplinary high-risk pancreatic clinic and framed as part of a research-linked or registry-linked programme where feasible.

PancPRO does not by itself determine surveillance eligibility — that decision follows CAPS criteria and local commissioning. But PancPRO provides the quantitative basis for the risk discussion, particularly in FPC kindreds without an identified gene, where the family history is the primary evidence.

Limitations of PancPRO

PancPRO has a specific clinical niche and specific limitations.

• Hypothetical locus. PancPRO models a single autosomal dominant susceptibility gene rather than the specific multi-gene architecture now known to underlie hereditary pancreatic cancer. The carrier probability it returns should be read as a family-history-weighted risk score, not as a probability of carrying a specific variant in a specific named gene.
• Complementarity with gene panels. In a service where hereditary pancreatic cancer gene panels are available, those panels give a direct molecular answer where one exists. PancPRO is most useful where a gene panel has been negative, or before panel testing to triage who should be tested.
• Pedigree completeness. Pancreatic cancer in older generations was often poorly documented historically. Cases recorded as "stomach cancer" or "abdominal cancer" may or may not have been pancreatic. Model accuracy degrades with imprecise cancer-site coding.
• No environmental factors. Smoking, heavy alcohol use, long-standing diabetes, chronic pancreatitis, and obesity all contribute to pancreatic cancer risk. PancPRO does not model these; they should be considered alongside the family-history estimate.

How Evagene integrates PancPRO

Evagene integrates PancPRO from the BayesMendel suite directly into the pedigree environment. The same structured pedigree used to capture cancer family history — relationships, sex, affected status, cancer types, ages, vital status — is passed directly to PancPRO. There is no separate calculator to open, no data to re-enter, and no format to wrangle. The model runs in an R sidecar process that invokes the validated BayesMendel package, so the calculation is the same one research and guideline development rely on.

A distinguishing feature of Evagene for hereditary pancreatic cancer workflows is that PancPRO is not run in isolation. Evagene's batch risk screening executes PancPRO alongside BRCAPRO (because BRCA2 is the most common hereditary contributor to FPC), MMRpro (because Lynch syndrome is in the differential for families with pancreatic plus colorectal or endometrial burden), and Mendelian inheritance analyses, across the full 200-plus disease catalogue. For a pedigree with pancreatic, breast, and melanoma diagnoses, this means the model that would identify a CDKN2A/FAMMM pattern is run automatically rather than depending on the clinician remembering to suspect it.

Results appear alongside the pedigree and can feed AI-drafted clinical reports using bring-your-own-key large language models (Anthropic Claude or OpenAI GPT). Evagene also exposes PancPRO through its REST API and MCP server, so downstream systems — EHR integrations, research pipelines, AI agents — can submit pedigrees and retrieve risk estimates programmatically. Documentation is at evagene.net/help. Evagene is browser-based, zero install, and free during Alpha via the waiting list.

Frequently asked questions

Related reading

• Hereditary cancer risk assessment
• Pancreatic cancer family history overview
• BRCAPRO calculator
• MMRpro calculator
• Lynch syndrome risk calculator
• Pedigree drawing software
• Clinical genetics pedigree tool
• Mendelian inheritance calculator
• GEDCOM pedigree software
• Phenotips vs Evagene comparison