Polygenic Risk Scores

A polygenic risk score (PRS) estimates how common genetic variants distributed across the genome combine to influence a person's likelihood of developing a condition. Unlike hereditary disease results, which identify rare high-impact variants in a single gene (such as BRCA1/BRCA2), a PRS reflects the cumulative effect of thousands of common variants, each with a small individual effect. Most people carry some of these variants; the question is how they add up relative to the population.

Because PRS captures a different dimension of genetic risk, a patient can have a high PRS without any hereditary variant, or carry a hereditary variant without an elevated PRS. The two types of results are complementary, not redundant. However, for some conditions a PRS result may not be provided when a hereditary variant is already detected. For example, patients with a BRCA variant detected via genotyping will not receive a breast cancer PRS. The model was developed in a cohort that excluded individuals with those variants and cannot reliably estimate risk for this group.

Yes, an increased likelihood result is clinically meaningful regardless of family history. Studies have shown that PRS provides individual risk stratification beyond family history. PRS and family history should be thought of as separate but complementary measures when assessing risk.

An increased likelihood PRS, even in the absence of family history, should prompt a more thorough risk assessment and a conversation about preventive strategies appropriate for that condition.

No. A typical likelihood result means the patient's genetic score falls within the range seen in most of the population, which does not mean they have no risk for the condition. Conditions like heart disease, type 2 diabetes, and breast cancer are influenced by many factors beyond genetics, including lifestyle, environment, age, and family history. A patient with a typical PRS can still develop the condition.

Think of it as one layer of information: a typical result is reassuring but not a green light, and should not change standard preventive care recommendations.

A PRS result is best used as a risk stratification tool to inform the conversation and prioritize follow-up, not as a standalone clinical decision point. For patients with an increased likelihood result, it can prompt a more comprehensive risk assessment, reinforce preventive behaviors, or inform shared decision-making around screening. For patients with a typical likelihood result, it provides context but should not override other clinical indicators.

PRS results are most useful when integrated with family history, biomarker data, and established risk calculators for the relevant condition. They are not diagnostic and should not be used to start, stop, or change treatment on their own.

No. A PRS complements but does not replace existing risk assessment approaches. Family history captures both genetic and shared environmental risk; biomarkers reflect current physiologic status; validated risk calculators integrate multiple factors to estimate absolute risk. A PRS adds a genetic dimension that these tools may not fully capture, particularly for patients with no family history or unremarkable biomarkers who may still carry elevated genetic risk.

The most complete risk picture comes from combining PRS with these other inputs, not substituting one for the other.

Yes, and it is an important consideration. Current PRS models show differences in predictive accuracy across genetic ancestry groups, reflecting the demographics of research cohorts on which most models were trained. 23andMe develops separate models for distinct ancestry groups and continues to expand its diverse reference datasets to improve performance across populations.

That said, the qualitative meaning of a result (typical likelihood or increased likelihood) is the same regardless of a patient's ancestry.

The most common reason is an ancestry limitation. PRS models require a sufficiently large, well-characterized dataset for a given genetic ancestry group to generate a reliable score. When the available data is insufficient to meet our performance standards for a population, the report is not returned rather than returning a result with reduced reliability.

If your patient does not have a result for a specific condition, it is because their genetic ancestry profile did not match a population for which a validated model currently exists. This is not a test failure. It reflects the current state of available research data, and 23andMe continues to expand its datasets, so coverage may improve over time.

PRS results can change over time as models are updated. Unlike a genotype, which is fixed, a PRS is a calculated score based on a statistical model that incorporates newly available research, expanded datasets, and methodological improvements. When a model is updated, the recalculated score may place a patient in a different likelihood category than before.

A result change reflects improving science, not a change in your patient's DNA. When a result changes, updated educational content is provided to help patients understand what the change means. This can be a useful opportunity to revisit their overall risk picture and preventive care plan.