MMRpro calculator: Lynch syndrome carrier probability from a family pedigree

Short version. MMRpro is a Bayesian calculator from the BayesMendel suite that estimates the probability an individual carries a pathogenic variant in one of the DNA mismatch repair (MMR) genes responsible for Lynch syndrome — principally MLH1, MSH2, and MSH6 — using family history of Lynch-associated cancers. It optionally incorporates tumour marker data (microsatellite instability and mismatch repair immunohistochemistry) to sharpen the estimate. MMRpro is a more sensitive quantitative alternative to the Amsterdam II and revised Bethesda clinical criteria, which miss a meaningful fraction of carriers. Evagene runs MMRpro directly on the pedigree you have drawn or imported, with no re-entry, and surfaces it alongside BRCAPRO and PancPRO in a single risk-analysis workflow.

Lynch syndrome in brief

Lynch syndrome, historically called hereditary non-polyposis colorectal cancer (HNPCC), is an autosomal dominant condition caused by pathogenic germline variants in the DNA mismatch repair genes — MLH1, MSH2, MSH6, and PMS2 — or by deletions of EPCAM that silence MSH2 by a read-through mechanism. It is the most common hereditary cause of colorectal cancer and is a significant cause of endometrial cancer.

Clinically, Lynch syndrome is characterised by a tumour spectrum that goes beyond colorectal and endometrial. Carriers face elevated risk of ovarian, gastric, small bowel, urothelial (renal pelvis and ureter), hepatobiliary, pancreatic, and brain cancers, and sebaceous skin tumours (Muir-Torre variant). The lifetime risks vary substantially by gene: MLH1 and MSH2 carriers generally have the highest colorectal and endometrial risks; MSH6 carriers have a somewhat lower but still substantially elevated risk, often with later onset; PMS2 risks are lower still but non-trivial.

Identifying Lynch syndrome matters because the downstream clinical changes are large: enhanced colonoscopic surveillance starting in early adulthood, consideration of risk-reducing hysterectomy and bilateral salpingo-oophorectomy once childbearing is complete, specific chemotherapy sensitivities (including response to immune checkpoint inhibitors in MSI-high tumours), and cascade testing for at-risk relatives. The tumour biology is distinctive and actionable; the clinical stakes for detection are therefore high.

Clinical criteria and their limits: Amsterdam II and Bethesda

Before quantitative risk models existed, Lynch syndrome was identified by clinical criteria. Two sets remain in common use.

Amsterdam II criteria require at least three relatives with a Lynch-associated cancer, one a first-degree relative of the other two, across at least two generations, with at least one diagnosis before age 50, and familial adenomatous polyposis excluded. Amsterdam II is specific but not sensitive: many families with Lynch syndrome do not meet it, particularly smaller families, families with substantial endometrial (rather than colorectal) burden, or families where key diagnoses occurred in older generations with limited documentation.

Revised Bethesda guidelines flag individual tumours for MSI or IHC testing on the basis of clinical features — early-onset colorectal cancer (under 50), synchronous or metachronous Lynch-associated tumours, MSI-high histology, first-degree relatives with Lynch-associated cancers, and so on. Bethesda is broader and therefore more sensitive than Amsterdam II, but still misses carriers whose families do not map onto its patterns.

Both sets of criteria remain useful for initial triage and tumour testing decisions. But as standalone screens for germline testing eligibility, their sensitivity is limited. MMRpro's advantage is that it uses the whole family structure in a principled probabilistic way rather than requiring the family to meet pre-specified patterns. A family that falls just short of Amsterdam II may still generate a high MMRpro posterior and warrant testing.

How MMRpro works

MMRpro is structurally analogous to BRCAPRO. It is a Bayesian model that starts from population allele frequencies for MLH1, MSH2, and MSH6, applies gene-specific penetrance functions for each cancer in the Lynch spectrum, and updates the prior probability of carrying a pathogenic variant based on the observed family phenotypes. The Elston-Stewart peeling algorithm propagates likelihoods through the pedigree efficiently.

The inputs MMRpro needs are straightforward:

• Sex and relationship to the counselee for every individual in the pedigree.
• Affected status for each Lynch-spectrum cancer, together with ages at diagnosis. Colorectal and endometrial cancers carry the most weight; ovarian, gastric, urothelial and others contribute as the implementation allows.
• Current age for living unaffected relatives; age and cause of death for deceased relatives.
• Synchronous or metachronous primary cancers, and polyp burden where documented (particularly for distinguishing Lynch from polyposis syndromes).
• Any prior genetic testing results for any family member.
• Optionally, tumour marker results from affected family members: microsatellite instability status (MSI-H, MSI-L, MSS) and mismatch repair protein IHC (retained vs lost expression of MLH1, MSH2, MSH6, PMS2).

The output is a posterior probability that the counselee carries a pathogenic variant in each of the modelled MMR genes, plus a cumulative risk projection for Lynch-spectrum cancers. In typical clinical practice, a combined MMR posterior probability of 5 percent or higher is often used as a threshold for germline testing, though specific thresholds vary by commissioning framework.

Tumour markers as intermediate phenotypes

The distinctive contribution of MMRpro, relative to a pedigree-only model, is that it can incorporate tumour marker data. Two markers matter most.

Microsatellite instability (MSI) testing examines short repeat tracts in tumour DNA for evidence of replication error. A tumour with defective mismatch repair accumulates insertions and deletions in these tracts, producing a microsatellite-high (MSI-H) signature. MSI-H is strongly suggestive of mismatch repair deficiency, though it can also arise from sporadic MLH1 promoter hypermethylation (often accompanied by a BRAF V600E mutation in colorectal cancer). MSS (microsatellite-stable) tumours rarely arise on a Lynch background.

Mismatch repair immunohistochemistry (IHC) directly stains for the presence of the four MMR proteins. Loss of staining for a specific protein — say MSH2 and its heterodimer partner MSH6 — points to a specific germline candidate gene (in this example, MSH2 or, less commonly, EPCAM deletion). Loss of MLH1 requires distinguishing Lynch from sporadic hypermethylation, typically via MLH1 promoter methylation testing or BRAF mutation analysis in colorectal tumours.

MMRpro integrates these results as intermediate phenotypes. A family with a modestly elevated pedigree-only posterior who also has an MSI-H tumour showing MSH2 loss on IHC ends up with a much higher combined posterior than either source of evidence would give alone. Conversely, an MSS tumour on a suggestive pedigree brings the posterior down.

From pedigree to clinical action

A typical Lynch syndrome workflow in a modern clinical genetics service follows a sequence: pedigree construction, initial risk estimation via MMRpro and clinical criteria, tumour testing where appropriate, germline panel testing of candidates identified by the above, confirmatory variant interpretation, and cascade testing of at-risk relatives.

For carriers confirmed to have Lynch syndrome, surveillance recommendations typically include colonoscopy every one to two years from early adulthood (specific age depending on gene), consideration of risk-reducing gynaecological surgery once childbearing is complete, and monitoring for other Lynch-spectrum cancers where indicated. The specifics follow NICE, NCCN, and European Society of Medical Oncology guidelines, which update as evidence accrues.

Limitations of MMRpro

MMRpro is a strong tool within its envelope. The envelope has edges.

• Gene coverage. Most implementations model MLH1, MSH2, and MSH6 explicitly. PMS2, EPCAM deletions that silence MSH2, and rare constitutional mismatch repair deficiency (CMMRD) are not always handled to the same depth.
• Penetrance heterogeneity. MSH6 and PMS2 have lower and more variable penetrance than MLH1 and MSH2. Model estimates reflect published penetrance functions that may not generalise perfectly across all populations and age groups.
• Cancer-site accuracy. The model needs specific cancer types. "Stomach cancer" noted in a 1960s family record may or may not map cleanly onto the modern spectrum. Pedigree accuracy matters.
• Tumour marker availability. The strongest MMRpro estimates come when tumour markers are available. In families where no affected relative has had a tumour tested, MMRpro falls back to pedigree-only calculations.
• Polyposis overlap. Lynch syndrome shares phenotypic territory with attenuated familial adenomatous polyposis and MUTYH-associated polyposis at low polyp counts. MMRpro does not triage between these; it assumes the clinical question is about Lynch.

How Evagene integrates MMRpro

Evagene integrates MMRpro from the BayesMendel suite directly into the pedigree environment. The same structured pedigree that a clinician draws during a consultation — first-degree, second-degree, and third-degree relatives, their cancer diagnoses, ages, vital status, tumour marker results where available — is the input MMRpro needs. There is no separate data-entry step between drawing the pedigree and running the model.

Under the hood, Evagene invokes MMRpro via an R sidecar process that calls the validated BayesMendel package. The calculation is the same one that published research and clinical guidelines rely on, not a reimplementation. Results — posterior probabilities for each MMR gene and combined Lynch risk — appear alongside the pedigree in the web interface and can be included in AI-drafted clinical reports using bring-your-own-key large language models.

A distinctive Evagene feature is batch risk screening. For a given proband, Evagene can run MMRpro alongside BRCAPRO, PancPRO, and Mendelian inheritance analyses across the full 200-plus disease catalogue in a single operation, flagging conditions where the pedigree crosses configurable thresholds. This means Lynch-spectrum signals that might be missed if a clinician only ran BRCAPRO — because the presenting story was focused on breast cancer — are surfaced automatically. See our hereditary cancer risk assessment guide for how batch screening fits into the wider workflow.

For programmatic use, MMRpro is exposed through Evagene's REST API and MCP server, enabling EHR integrations, research pipelines, and AI agents to submit pedigrees and retrieve Lynch carrier probabilities without a human in the loop. Docs are 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
• Lynch syndrome risk calculator overview
• BRCAPRO calculator
• PancPRO calculator
• Ovarian cancer family history calculator
• Pedigree drawing software
• Clinical genetics pedigree tool
• Mendelian inheritance calculator
• GEDCOM pedigree software
• Phenotips vs Evagene comparison