How the ACMG Variant Classification System Works

A variant is a spot where your DNA spells a gene differently from the usual reference, and most people carry thousands of them, nearly all harmless. When a test finds one in a gene linked to disease, the lab has to judge whether it actually causes illness, and the ACMG system is the shared rulebook US labs use to make that judgment.1Genetics in Medicine. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology The system sorts each variant into one of five labels by weighing several kinds of evidence and adding up their strength. Because the label is a weighing of evidence rather than a direct measurement, the same variant can be relabeled years after your report was written.

What the Five Labels on Your Report Mean

These labels come from guidelines published in 2015 by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.6PubMed Central. Performance of ACMG-AMP Variant-Interpretation Guidelines among Nine Laboratories in the Clinical Sequencing Exploratory Research Consortium They grew out of a 2013 effort by a group of lab directors and clinicians who set out to give all labs the same words and the same way of weighing evidence.7PubMed Central. Standards and Guidelines for the Interpretation of Sequence Variants: A Joint Consensus Recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology The rules were written for single-gene inherited conditions. A few genes only raise risk rather than cause disease outright, and some frameworks give those extra labels such as “predisposing.”8PubMed Central. Expanding ACMG variant classification guidelines into a general framework

On a real report, the label usually sits in a results box near the top, alongside the gene name and the variant code. The code (something like c.1234A>G) is the exact spelling change, and it’s what you’d give a new doctor. Be careful with reports that say “no pathogenic variants found,” because uncertain results may still be listed lower down. It’s worth asking whether your report was “negative” or “negative with VUS,” since those are different outcomes.

A common mistake is hearing “likely pathogenic” as a maybe. An uncertain result is the opposite case: it is not a weak positive and normally shouldn’t drive decisions.

A benign or likely benign label usually closes the question. The variant is regarded as a normal difference between people, so it normally doesn’t change your screening or your relatives’ testing. Many reports leave benign variants off or tuck them into an appendix, so not seeing one doesn’t mean the lab missed it.

How Labs Turn Evidence Into a Label

Each piece of evidence counts either toward harmful or toward harmless.9ACGS Best Practice Guidelines for Variant Classification in Rare Disease. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2023 It is also graded by strength: supporting, moderate, strong, or very strong.10ACGS Best Practice Guidelines for Variant Classification in Rare Disease. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2023 Combining rules then set how many pieces of each grade are needed. Two strong pieces, for example, count as much as one very strong piece.11PubMed Central. Overview of specifications to the ACMG/AMP variant interpretation guidelines

The rules have since been converted into a points system. Toward harm, very strong evidence is worth 8 points, strong 4, moderate 2, and supporting 1.12ACGS Best Practice Guidelines for Variant Classification in Rare Disease. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2023 Evidence toward harmless subtracts points: strong -4, moderate -2, supporting -1.13ACGS Best Practice Guidelines for Variant Classification in Rare Disease. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2023 A total of 10 or more means pathogenic, and 6 to 9 means likely pathogenic. At the other end, -1 to -5 is likely benign and -6 or lower is benign.14ACGS Best Practice Guidelines for Variant Classification in Rare Disease. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2023 Anything from 0 to 5 lands as uncertain.

The points have a basis in odds. Each step up in strength roughly multiplies the odds of being harmful, from about 2 to 1 for supporting to 350 to 1 for very strong.15PubMed Central. Overview of specifications to the ACMG/AMP variant interpretation guidelines Because the scale works in multiples, where doubling becomes adding, the points can simply be summed.16PubMed Central. Fitting a naturally scaled point system to the ACMG/AMP variant classification guidelines An upcoming revision of the guidelines is expected to adopt points-based scoring as standard.17Genomenon. How the ACMG Variant Classification System Works

Here is how adding works in practice. Say a variant is absent from healthy-population databases (moderate, 2 points) and a computer prediction says it’s harmful (supporting, 1 point). That totals 3, which stays uncertain. A strong lab experiment showing damage adds 4 points and brings it to 7, which is likely pathogenic. One more moderate piece would reach 9, still likely pathogenic, which shows how close calls arise.

Most single pieces of evidence, a computer prediction for instance, are graded only as supporting, so on their own they stay far below the points needed for pathogenic. A confident label usually means several independent kinds of evidence point the same way, such as rarity, lab results, and family patterns. Evidence that is really the same observation counted twice shouldn’t be added twice, so curators try to avoid stacking overlapping criteria.

The label is a judgment rather than a test result. Curators decide which criteria apply and at what strength, and two careful experts can grade the same evidence differently. This is why reports often summarize the evidence codes used, such as PVS1 or PM2.

The Kinds of Evidence Labs Look At

One of the most useful clues is how common the variant is in healthy people. A variant found in 5% or more of people is treated as harmless on that fact alone, unless it’s already known to cause disease.18PubMed Central. Variant interpretation using population databases: Lessons from gnomAD Being more common in healthy people than expected for how rare the disease is counts as strong evidence of being harmless.19PubMed Central. Variant interpretation using population databases: Lessons from gnomAD There are exceptions: some recessive diseases have common harmful variants above 1%, such as the main cystic fibrosis variant, so labs check known lists.20PubMed Central. Variant interpretation using population databases: Lessons from gnomAD At the other extreme, being absent from large population databases counts as moderate evidence toward harm in the standard rules.21PubMed Central. VarCards2: an integrated genetic and clinical database for ACMG-AMP variant-interpretation guidelines in the human whole genome

Absence only tells you a variant is rare, and nearly everyone carries some rare harmless variants unique to their family. That’s why rarity counts as moderate rather than strong, and why it has to be combined with other evidence to reach a pathogenic label. How much it’s worth also depends on the size of the database, since absence from a small one tells you much less.

Another kind of evidence is whether the change breaks the gene. A change predicted to stop the gene from working, called loss-of-function, can count as very strong evidence.22PubMed Central. Recommendations for interpreting the loss of function PVS1 ACMG/AMP variant criterion Classic examples cut a gene’s instructions short, such as a premature stop signal or a shifted reading frame. This evidence is strongest when losing the gene’s function is the known way it causes disease, so it isn’t applied equally to every gene.

Not every such change earns full strength. In one large analysis, at least 27.6% did not reach very strong.23bioRxiv. AutoPVS1: An automatic classification tool for PVS1 interpretation of null variants About 41% of changes affecting how the gene is spliced together got reduced strength, more than other types.24bioRxiv. AutoPVS1: An automatic classification tool for PVS1 interpretation of null variants Some changes near the very end of a gene may still leave a working protein, which is one reason they can be graded lower.

Lab experiments testing what a variant does can count strongly for or against harm when the test is well validated.25PubMed Central. Consultation informs strategies for improving the use of functional evidence in variant classification A test checked against at least 11 known harmful and harmless variants can count as moderate evidence.26PubMed Central. Recommendations for application of the functional evidence PS3/BS3 criterion using the ACMG/AMP sequence variant interpretation framework Computer prediction tools can add evidence too, but only at score cut-offs stricter than many tool makers suggest.27PubMed Central. Calibration of additional computational tools expands ClinGen recommendation options for variant classification with PP3/BP4 criteria

Family patterns matter as well. If the variant appears in relatives who have the condition and not in those who don’t, that supports harm. For this reason a lab may ask to test affected relatives. One family rarely settles the question, but several families can, and agreeing to relative testing can help your own label move out of uncertain.

Finally, expert panels adjust the general rules for particular genes. One panel, for example, tailored rules for the hemoglobin genes behind recessive blood disorders.28PubMed Central. Adapting the ACMG/AMP variant classification framework: A perspective from the ClinGen Hemoglobinopathy Variant Curation Expert Panel So the same evidence can weigh differently in different genes, and asking whether your gene has an expert-panel rule set is reasonable.

What a Variant of Uncertain Significance Means for You

An uncertain result sits between harmful and harmless, and its meaning for your health is unclear.29PubMed Central. How Variants of Uncertain Significance Impact Clinical Decisions: A Systematic Review Yet people given one have had higher rates of preventive mastectomy and pregnancy termination than similar people with benign results.30PubMed Central. How Variants of Uncertain Significance Impact Clinical Decisions: A Systematic Review An uncertain result can complicate decisions and lead to unnecessary care.31PubMed Central. The challenge of genetic variants of uncertain clinical significance: A narrative review Screening and surgery choices are better based on your personal and family history and your doctor’s advice than on the variant itself.

Many of these get resolved, and usually not toward harm. In one hereditary breast, ovarian, and colon cancer study, about 11% of reclassified VUS became actionable.32PubMed Central. A multicenter study of clinical impact of variant of uncertain significance reclassification in breast, ovarian and colorectal cancer susceptibility genes

A few questions are worth bringing to your doctor or counselor. Would your care plan differ if the variant didn’t exist? Should relatives be tested to help clarify it? How will you hear if it’s reclassified? And can a genetic counselor review the report with you?

Extra monitoring may be reasonable when your history supports it, whereas irreversible steps based only on an uncertain result are much harder to justify. A strong family history may justify screening regardless of the variant.

If you or your relatives have had the cancer or heart condition that prompted testing, an uncertain result doesn’t cancel it. Screening guidelines based on your history apply as if the test had found no variant of concern, which can mean earlier or more frequent screening than for the general public. Nor does the result mean the test failed to explain your history, because the cause may be another gene or something testing cannot yet detect.

How Often Labels Change

At one large testing lab, most VUS reclassifications happened in 2021 to 2023.33PubMed Central. Clinical Variant Reclassification in Hereditary Disease Genetic Testing The pace rose as labs added machine-learning predictions and updated population databases.34PubMed Central. Clinical Variant Reclassification in Hereditary Disease Genetic Testing Across studies, the average time to reclassification was about 590 days, ranging from 13 days to 20 years.35PubMed Central. Variant reclassification and recontact research: A scoping review Updated rules can also move labels in both directions: in one heart gene, newer expert rules changed the actionable label for 18% of variants, with as many upgrades as downgrades.36PubMed. Clinical and laboratory reporting impact of ACMG-AMP and modified ClinGen variant classification frameworks in MYH7-related cardiomyopathy

Labels change because new evidence arrives, such as more people tested, published lab experiments, and newly reported families carrying the variant. Updates to population databases can show a variant is more common in healthy people than first thought, which can push it toward benign. Revised expert rules can change a label even when no new data about the variant exists, because the same evidence gets weighed differently.

When cancer-gene VUS were upgraded, some patients began more intensive screening such as breast MRI or colonoscopy, or had risk-reducing surgery.37PubMed Central. A multicenter study of clinical impact of variant of uncertain significance reclassification in breast, ovarian and colorectal cancer susceptibility genes Still, only about 4.6% of reclassified VUS in that study changed a patient’s medical care.38PubMed Central. A multicenter study of clinical impact of variant of uncertain significance reclassification in breast, ovarian and colorectal cancer susceptibility genes

A reclassification usually arrives as an amended report naming the same variant with a new label. An upgrade may prompt referral for screening or testing of relatives. A downgrade may end extra monitoring that was started for that variant. Keep the original report so you can compare the two.

Who Tells You When a Label Changes

No legal duty requires labs or doctors to tell you when a variant is reclassified, even if it matters for your care.39PubMed Central. GENETIC DUTIES Canadian geneticists describe recontact as a shared responsibility among the patient, the ordering doctor, and the lab.40PubMed Central. Canadian College of Medical Geneticists: clinical practice advisory document – responsibility to recontact for reinterpretation of clinical genetic testing Labs often lack your contact details and clinical information, so the ordering doctor is usually better placed to explain changes.41PubMed Central. Is There a Duty to Reinterpret Genetic Data? The Ethical Dimensions Specialists who saw you only briefly may not stay in touch, which can leave updates unnoticed.42PubMed Central. Is There a Duty to Reinterpret Genetic Data? The Ethical Dimensions

In practice, much of keeping your result current falls to you. Hold on to a copy of your report with the lab name and variant code, and update your contact details with both the lab and the ordering clinic. Ask your doctor every few years to check for reclassification, and mention the result to any new doctor so it isn’t lost.

How Much Labs Agree on a Label

When three labs classified the same variants, they fully agreed on 54% at first.43PubMed Central. Variant Classification Concordance using the ACMG-AMP Variant Interpretation Guidelines across Nine Genomic Implementation Research Studies About 11% had disagreements that could change medical advice.44PubMed Central. Variant Classification Concordance using the ACMG-AMP Variant Interpretation Guidelines across Nine Genomic Implementation Research Studies After discussing the evidence, agreement rose to 84%.45PubMed Central. Variant Classification Concordance using the ACMG-AMP Variant Interpretation Guidelines across Nine Genomic Implementation Research Studies Classifications from clinical testing agree much more than research-based ones, and newer ones more than older ones.46Genetics in Medicine. Sources of discordance among germ-line variant classifications in ClinVar Automated tools match experts best on gene-breaking changes and least on harder types like single-letter changes, so expert review still matters.47PubMed. Comparative evaluation of manual and automated ACMG/AMP variant classification: implications for clinical genetic practice

Labs often have different evidence, because each sees its own patients and may hold private data, such as family information, that others don’t. They also apply the same criteria with different judgment, such as how strongly to weigh a prediction tool or how to treat a borderline experiment. These differences matter most near a category boundary, where a single point of evidence can move a variant from one label to the next.

That’s when a second opinion helps. If two labs gave you different labels, a genetic counselor can compare their evidence. Shared databases like ClinVar show other labs’ labels for the same variant. The most common close call is between likely pathogenic and uncertain, and older reports are worth rechecking against newer classifications.

Why Ancestry Can Affect Your Result

Non-European groups are underrepresented in genetic databases, which leads to many more VUS for them.48PubMed Central. Is there a way to reduce the inequity in variant interpretation on the basis of ancestry? Whether a variant is rare is judged against those reference databases, so a harmless variant common in your group can look rare if your ancestry is poorly represented. Rarity then counts as evidence toward harm, which can produce an uncertain label or, in some cases, a mistaken harmful one.

The effect shows up in the data. In heart-muscle disease genes, over a third of downgraded variants were common in non-Europeans but rare in Europeans.49PubMed Central. Underrepresentation of Diverse Ancestries Drives Uncertainty in Genetic Variants Found in Cardiomyopathy-Associated Genes As more people of a given ancestry are added, those same variants appear commonly and get downgraded. Newer database releases that separate ancestries more finely may move some VUS to benign or likely benign.50PubMed Central. Improved allele frequencies in gnomAD through local ancestry inference

If your ancestry is underrepresented, ask whether your variant is common in any ancestry group, and request reanalysis after major database updates. Be especially cautious about acting on a harmful label that rests mostly on rarity. Joining research studies can also help improve databases for others.