Breast cancer is the most common cancer among women and the incidence in Denmark is among the highest in the world. Improvements of treatment and the implementation of breast cancer screening have contributed to the reduced breast cancer mortality during the last decades. However, breast cancer screening has been criticised for producing false positives – findings on the X-ray picture, which proves to be benign when biopted, and overdiagnosis – finding breast cancers which the woman would not have noticed in her lifetime, had she not been screened. Intuitively, it seems logical that if you have high risk of breast cancer you are more likely to benefit from screening than if you have low risk, and if you have low risk, the potential harms from screening might outweigh the benefits. So risk in itself seems to be an important player if we want to maximize benefits from screening.
Today, women are invited to participate in screening programs, which typically consist of an X-ray examination, a mammography, with regular intervals. Despite the development of different risk models, screening according to individual risk has never been implemented in the general population.
Common genetic variants and lifestyle
The most comprehensive risk model combines a woman’s age, genetic variants, family history of breast cancer and other cancers, lifestyle, reproductive history, use of oral contraception/hormones and mammographic density into a single absolute risk of getting breast cancer within the next 5 and 10 years.
During many genetic studies of breast cancer including 100,000s of women, scientists have found many common genetic variants with importance for breast cancer risk. We all carry those from conception. Some many, some less. If you carry few of these, the risk of developing breast cancer is lower compared to a woman who carries more of these variants. Individually, each variant contribute little, but if combined into a polygenic score, they provide important risk information.
Other factors are also important for risk of breast cancer: use of alcohol, body mass index, age at menarche, age at birth of first child, age at menopause, family history and the appearance of the breast tissue on the mammogram. Use of oral contraceptives and hormone therapy at menopause are also important for risk of breast cancer.
Mathematically, it has been a challenge to combine information of all these factors into a single reliable risk estimate. Eg. you inherit the common genetic variants from your parents, and if your family history is also incorporated in the model, how do we ensure that the part of the family history accounted for by the common genetic variants is not counted twice? These and other issues were resolved during the development of the model.
A risk model for breast cancer might be very useful in the general population, but of course only if the model produces the right predictions. Therefore, we analyzed existing data from 50,000 Danish women, who had already been examined many years ago and of whom some had developed breast cancer since then. By applying the risk model, we could therefore test the model’s predictions. We estimated all the 50,000 women’s risk of getting breast cancer the next 5 and 10 years after they were enrolled in the study and compared this estimate with the breast cancers, which were actual found in these women in the period.
If the model worked, we would have a novel medical instrument at our disposal for the general population. The possibility to measure each woman’s risk of getting breast cancer in the future might seem unpleasant. On the other hand, used proactively, risk information could perhaps also be very useful to maximize benefit and minimize harms from screening by adjusting it to risk.
The modified BOADICEA model provided valid risks among the Danish women
BOADICEA (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) is a multifactorial breast cancer risk model, that combines a woman’s age, information on genetics, breast and other cancer family history, lifestyle/hormonal risk factors and mammographic density into a single absolute risk estimate.
We set up different models with increasing information (an age-alone model, a model with only polygenic score, a model with only lifestyle risk factors and a combined model with age and polygenic score and lifestyle risk factors). The combined model provided the best prediction of 5-year risk.
When considering the 50% of the 50,000 women who had the lowest risk from the model, we found that these women only developed 5.2% of the breast cancers. If future screening programs are adjusted to risk, it is important not to overlook the rare breast cancers in the low-risk group. We therefore looked closely into these low-risk women, who did develop breast cancer. They did not have any disease history, obvious risk factors or clear family history of breast cancer. They were generally younger and too young to participate in the population screening program. Our assumption is that they might carry genetic predisposition beyond the common genetic risk variants but this needs to be examined more closely in further studies.
The next steps and into the clinics
BOADICEA has already been validated in the Dutch, Swedish, UK and now also in Danish women. Most studies point to the genetical component as being more predictive than only lifestyle risk factors and the combined models to be the most predictive.
So now we know that the risk works, but there are still several aspects to address before implementation of risk stratified screening in everyday clinic. For example, which risk-cut offs and screening intervals produce the most optimal outcome in terms of reduction of advanced breast cancer and breast cancer mortality? Another aspect is how to communicate the risk estimate as a part of the screening without inflicting psychological harms? These are questions that need to be addressed further randomized prospective studies.