What Is a Tyrer Cuzick Score & What Do the Results Mean?

A Tyrer-Cuzick score is a personalized estimate of your likelihood of developing breast cancer, expressed as both a ten-year risk and a lifetime risk percentage. The model behind it, sometimes called the IBIS (International Breast Cancer Intervention Study) model, combines your family history, genetic information, and a range of personal factors to produce a single number that clinicians use to guide screening and prevention decisions. Unlike simpler tools, the Tyrer-Cuzick model was designed from the start to account for inherited cancer genes and detailed multi-generational family history, which makes it one of the more comprehensive risk calculators available in clinical practice.

What Goes Into the Calculation

The Tyrer-Cuzick model pulls in more personal detail than most people expect. It accounts for your age, height, weight, age when your periods started, whether you have gone through menopause and at what age, whether and when you had your first child, and whether you have used hormone replacement therapy. It also asks about BRCA1 and BRCA2 gene mutation status if known. On the family history side, it goes beyond just first-degree relatives: the model incorporates second- and third-generation cancer history on both your mother’s and father’s side, including the presence of ovarian cancer in the family.¹

The original model, published in 2004, used Bayesian statistics to estimate the probability that a woman carries genes predisposing her to breast cancer, then layered on personal risk factors to refine that estimate.² Version 8 of the tool, released more recently, added breast density to the calculation, which strengthened its connection to how breast tissue itself influences risk.³

How to Read Your Results

The model produces two numbers. One is your estimated risk of developing breast cancer over the next ten years. The other is your remaining lifetime risk, calculated from your current age forward. It is the lifetime risk number that drives most clinical decisions, and there are two thresholds that matter most.

A lifetime risk at or above 20% is the widely used cutoff for recommending enhanced screening. Multiple organizations, including the American Cancer Society, the National Comprehensive Cancer Network, and the American Congress of Obstetricians and Gynecologists, recommend that women whose lifetime risk meets or exceeds this level be offered annual breast MRI in addition to annual mammography.⁴¹

A second threshold sits at around 10% lifetime risk. At this level, some healthcare systems and insurance companies recommend referral to genetic counseling services, even if the risk has not yet crossed the 20% line.⁵ For context, the average lifetime breast cancer risk for the general female population is roughly 10 to 12%, so a score in that range is close to average, not elevated.

What Happens After a High-Risk Result

Crossing the 20% threshold triggers a conversation about supplemental screening. The most common next step is adding breast MRI to your yearly mammogram. In one large screening program, about 8% of women undergoing routine mammography were classified as high-risk. Of those who were offered a consultation at a breast center, three-quarters attended, and about half of those who attended went on to complete an MRI for supplemental screening.⁶ That drop-off at each step is common and reflects the reality that being told you are “high risk” does not automatically translate into action.

A high Tyrer-Cuzick score can also open the door to discussions about chemoprevention, meaning medications like tamoxifen or raloxifene that reduce breast cancer risk. The American Society of Clinical Oncology has endorsed the Tyrer-Cuzick model as one of the validated tools for determining whether a woman is a candidate for these drugs.¹ In some imaging centers, the Tyrer-Cuzick score is also being used to flag patients who should be evaluated for genetic testing based on National Comprehensive Cancer Network criteria for hereditary cancer syndromes.⁷

How the Tyrer-Cuzick Model Compares to Other Risk Tools

The most common alternative is the Gail model, which is simpler, faster to fill out, and widely used in primary care. The Gail model asks fewer questions and does not incorporate detailed multi-generational family history or BRCA status. That difference matters a great deal for women who have strong family histories of cancer. In one study comparing the two, the Gail model underestimated risk in women with a significant family history of cancer, while the Tyrer-Cuzick model produced higher, more appropriate estimates for that group.⁸

In a head-to-head comparison using receiver operating characteristic analysis in a Chinese screening population, the Tyrer-Cuzick model had meaningfully better discriminatory accuracy, with an area under the curve of 0.786 compared with 0.665 for the Gail model.⁹ In plain terms, the Tyrer-Cuzick model was substantially better at distinguishing women who would go on to develop breast cancer from those who would not. The trade-off is that it takes longer to complete and requires more detailed family history, which can be harder to collect in a busy primary care setting.

Where the Model Struggles

No prediction model is perfect, and the Tyrer-Cuzick model has specific blind spots worth knowing about. One of the most clinically relevant is its performance in women who have certain high-risk breast lesions found on biopsy.

For women with lobular carcinoma in situ (LCIS), the Tyrer-Cuzick model performed no better than a coin flip at predicting who would go on to develop invasive breast cancer. The concordance statistic in one study was 0.493, meaning the model could not meaningfully distinguish between women with LCIS who would and would not develop cancer over the next ten years.¹⁰ The model generally overestimated risk in this group, though it underestimated risk for those at the lowest predicted levels.

A similar pattern appeared with atypical hyperplasia, another high-risk biopsy finding. In a study tracking women with atypia over the decade following their biopsy, the model predicted about 59 breast cancers but only 31 occurred, giving an observed-to-predicted ratio of 0.53. The concordance statistic was only slightly above chance at 0.540.¹¹ If you have been diagnosed with either of these conditions, your clinician should be aware that the Tyrer-Cuzick score is less reliable for you specifically.

Even in the general population, both the Tyrer-Cuzick and Gail models tend to overestimate risk in the highest-risk groups and slightly underestimate it in the lowest-risk groups.¹² This is a common pattern in risk prediction models and it matters because the women most affected by clinical decisions based on their score are precisely those in the highest-risk brackets, where overestimation is most pronounced.

Performance Across Racial and Ethnic Groups

The Tyrer-Cuzick model was developed primarily using data from populations of European ancestry, which raises a fair question about how well it works for everyone else. A large validation study using data from the Women’s Health Initiative found that the model was well calibrated overall, with an observed-to-expected ratio of 0.95. It performed accurately for non-Hispanic white women and African American women. However, it overestimated risk for Hispanic women, with an observed-to-expected ratio of 0.75, meaning it predicted substantially more cancers than actually occurred in that group.¹³ Results for Asian/Pacific Islander and Native American women suggested reasonable calibration, but sample sizes were too small to draw firm conclusions.¹⁴

A separate study found that Black women were significantly less likely to be classified as high-risk by version 8 of the model compared with white women. In that cohort, about 11% of Black women were classified as high-risk versus roughly 18% of white women. The gap was driven partly by differences in breast density and body mass index between the two groups: Black women had higher average BMI, which the model treats as somewhat protective (because of its association with lower breast density and hormonal factors), and lower rates of dense breast tissue, which the model treats as a risk factor.¹⁵ This is a meaningful concern because Black women in the United States face higher breast cancer mortality despite similar incidence rates, which suggests the model may be missing factors that matter for this population.

How Breast Density and BMI Affect Your Score

Since version 8 of the model incorporated mammographic breast density, these two physical measurements now have a direct effect on your result. In the study of over 15,000 women mentioned above, breast density had a moderate positive correlation with the Tyrer-Cuzick score, meaning denser breasts pushed scores higher. BMI showed a slight inverse relationship, meaning higher BMI was associated with marginally lower scores.¹⁵

This creates something of a paradox. Higher BMI is a well-established risk factor for postmenopausal breast cancer through biological mechanisms related to estrogen production in fat tissue. Yet the model’s handling of BMI in conjunction with breast density can sometimes produce counterintuitive results. A woman with high BMI but low breast density may receive a reassuringly low score even though her weight-related risk is real. Clinicians aware of this tension sometimes use the Tyrer-Cuzick score alongside their own clinical judgment rather than treating it as a standalone verdict.

The Push to Add Polygenic Risk Scores

One of the most active areas of research around the Tyrer-Cuzick model involves adding polygenic risk scores, which capture the combined effect of many common genetic variants, each contributing a tiny amount to overall breast cancer risk. The idea is that the model already handles high-penetrance genes like BRCA1 and BRCA2 well, but misses the cumulative influence of hundreds of smaller genetic signals.

A large validation study found that when a multiple-ancestry polygenic risk score was combined with the Tyrer-Cuzick model, the resulting combined score had roughly twice the discriminatory accuracy of either tool alone. Among women the Tyrer-Cuzick model had classified as high-risk, about a third turned out to be low-risk when the polygenic data was factored in. And among those the model had classified as low-risk, about 4% were actually high-risk by the combined score. When the two tools disagreed, the combined score was more accurate at predicting who actually developed breast cancer.¹⁶

An independent analysis in the UK Biobank confirmed that adding a polygenic risk score boosted the Tyrer-Cuzick model’s discriminatory accuracy from a C statistic of 0.57 to 0.67.¹² That finding comes with a caveat, though: a population-based comparison found that the Tyrer-Cuzick model with polygenic risk scores tended to overestimate risk at the highest risk levels, particularly in older women.¹⁷ The technology is promising but still being refined. Most clinics do not yet routinely incorporate polygenic scores into Tyrer-Cuzick assessments, though commercial tests that offer this combination are beginning to reach the market.

For women who carry moderate-penetrance gene variants like CHEK2 or ATM, combining the Tyrer-Cuzick model with a polygenic score and variant status paints a more nuanced picture than any single tool. In one analysis, about a quarter of CHEK2 carriers had a remaining lifetime risk below 20% when all three inputs were combined, which could spare them from screening and prevention measures they do not actually need.¹⁸

Making Sense of a Risk Number

Receiving a breast cancer risk percentage can be harder to process than it sounds. Research on women in the UK NHS screening population found that only about a third of women, whether they were classified as high-risk or low-risk, could correctly identify the general population’s lifetime breast cancer risk when asked. Women who had previously received formal risk counseling did noticeably better, with about half giving an accurate answer.¹⁹ The encouraging part of that study was that receiving risk information did not cause low-risk women to opt out of future mammograms, which had been a concern.

For women on the other end of the spectrum, the psychological weight can be considerable. A study of women with a sister diagnosed with breast cancer found that those whose lifetime risk exceeded 20% were more than twice as likely to report moderate or severe cancer-related distress compared with women below that threshold. Perceived risk, rather than any objective clinical measure, was the strongest predictor of that distress.²⁰ A 22% lifetime risk and a 19% lifetime risk may lead to very different clinical pathways because of where the 20% cutoff falls, but the biological difference between those two numbers is negligible. Clinicians who work with these scores regularly emphasize that the number is a guide to action, not a prophecy, and that the sharp thresholds are practical conventions rather than biological boundaries.

The Quality of What You Put In

Because the Tyrer-Cuzick model relies heavily on self-reported family history, the quality of the output is only as good as the information you provide. Many people have incomplete knowledge of their relatives’ cancer histories, particularly on the paternal side or in families where cancer was not openly discussed. Reporting an aunt’s cancer diagnosis as breast cancer when it was actually a different type, or forgetting that a grandmother had cancer at all, can shift the result meaningfully. Height, weight, and reproductive history are more straightforward, but even small inaccuracies in age at menopause or age at first birth feed into the calculation.

The model also cannot account for factors it does not ask about. It does not include alcohol intake, physical activity, or radiation exposure history, all of which are recognized breast cancer risk factors. It does not capture breast density changes over time, only a single measurement. And while it asks about hormone replacement therapy use, the type, duration, and timing of HRT all carry different risk profiles that the model handles somewhat bluntly. If you are filling out a Tyrer-Cuzick questionnaire, being as thorough and accurate as possible with your family and personal history is the single most useful thing you can do to get a meaningful result. Gathering information from relatives before your appointment, rather than guessing on the spot, makes a real difference.