Introduction

Familial clustering of breast cancer is a common indication for referral to clinical genetics services. Whilst shared environmental factors contribute, they do not fully explain the risk, and genetic predisposition is thought to be a major factor. This can be due to rare, highly penetrant mutations, or multiple low penetrance variants [1, 2]. Risk assessment includes variant analysis for known cancer risk genes where appropriate, or assessment by family history (FH). The UK National Institute of Health and Care Excellence (NICE) provides guidance for classification and management of people with a FH of breast cancer (CG164) [3]. Patients are stratified according to FH into near population risk (PR), moderate risk (MR) and high risk (HR) based on percentage lifetime risk and 10-year risk from age 40. Risk stratification uses empirical criteria provided (shown in Table 1), or other models such as the Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA), a computer program that is used to calculate the risks of breast and ovarian cancer in women based on their FH [4]. NICE recommends additional screening for women at MR and HR, as seen in Table 1, in the form of mammograms or MRI. This is of relevance for younger women who are not yet enroled in the UK National Breast Screening Programme (NBSP), which offers 3-yearly mammograms to all women aged 50–70. To our knowledge, there has been no attempt to validate the empirical NICE criteria in women attending clinical genetics services regarding their breast cancer risk.

Table 1 NICE risk criteria and interventions
Full size table

Subjects and methods

Female patients referred to clinical genetics services for breast cancer risk from 2000 to 2010 were included in the study. Patients were aged under 50 at initial consultation, with no personal history of breast and/or ovarian cancer. FH information was collected from clinical genetics services records. BRCA (referring to both BRCA1 and BRCA2) mutation carriers were identified through the national BRCA testing service. Women who went on to develop breast cancer were identified by linkage to pathology records.

All women were risk categorised into PR, MR and HR as outlined in the NICE guidelines [3]. NICE guidelines do not state that affected relatives must be from the same side of the family. However, it is acknowledged that many clinicians interpret the guidelines this way. Therefore, all analyses were performed based on a risk categorisation which (1) did not assume and (2) assumed same-side FH as necessary to meet risk criteria. The result of BRCA testing was also considered for appropriate risk categorisation. This was time intensive with each case taking between 5–15 min for risk assignment. As this was done retrospectively using clinical notes, time taken for clinical consultation and confirmation of diagnoses of affected family members is not included.

Percentage 10-year risk was calculated for each risk category and for BRCA mutation carriers, for ages 40–49 and ages 50–59 years inclusive. Incidence of breast cancer per patient year of follow-up within each group was calculated, and extrapolated to give the 10-year absolute breast cancer risk. Kaplan–Meier Survival Analysis (KMSA) was used to assess the rate of breast cancer development across different risk categories and age ranges. Patients were censored at completed time of follow-up or at breast cancer diagnosis. The HR group was analysed both including and excluding BRCA carriers.

Results

In total, 1409 patients were eligible for inclusion with a total of 15,414 patient years of follow-up. Using both sides of the FH to calculate risk, 505 women were PR (35.8%), 522 MR (37%) and 382 HR (27.1%), including 12 BRCA1 and 10 BRCA2 carriers. Using only a same-side FH, there were 554 (39.3%) PR, 490 (34.8%) MR and 365 (25.9%) HR women.

Thirty women developed an invasive cancer prior to May 2016. The frequency and percentage 10-year absolute risk are shown in Table 2. Not assuming a same-side FH, the highest absolute risk between the ages of 40 and 49 was in the HR group, both including (3.56% (3.34–3.80%) and excluding BRCA carriers (2.49% (2.28–2.70%). From ages 50 to 59, the MR group had the highest percentage absolute risk, at 7.05% (6.78–7.31%).

Table 2 Frequency and absolute risk of breast cancer by NICE risk category
Full size table

Between ages 40 and 49, none of the groups met the 10-year risk suggested by NICE guidelines. Assuming a same-side FH, a similar pattern of absolute risk is seen, with no group reaching the screening threshold suggested by NICE.

Table 3 shows the results of KMSA. Not assuming same-side FH, there is no significant difference in the rate of breast cancer development between the PR and MR groups from 40 to 49 (p = 0.431). A risk difference between these two groups emerges after the age of 50 (p = 0.037). When same-side FH is assumed, there is no significant difference in breast cancer rates between the PR and MR groups overall (p = 0.134) or across any age range (<39 years p = 0.283, 40–49 years p = 0.791, 50–59 years p = 0.11).

Table 3 Kaplan–Meier analysis of the rate of breast cancer diagnosis comparing NICE risk categories by age range
Full size table

Both not assuming and assuming same-side FH, there is a difference in breast cancer rates between the PR and HR women from 40 to 49 (p = 0.036 and p = 0.042, respectively). However, this significance is lost on exclusion of BRCA carriers (p = 0.136 and p = 0.171, respectively). There is no significant difference in the rate of breast cancer between these groups from the ages of 50 to 59 not assuming or assuming same-side FH (p = 0.149 and p = 0.063).

The MR and HR groups combined were compared with the PR group to try and detect a significantly increased rate of breast cancer in women deemed at any increased risk. Not assuming same-side FH, the MR/HR group (excluding BRCA carriers) had a significantly increased rate of breast cancer from 50 to 59 years (p = 0.049). There was no detectable difference in breast cancer rates between MR and HR women at any time.

Discussion

Before the age of 50, neither the MR nor HR groups have a risk that reached the suggested NICE 10-year threshold. KMSA showed the rate of breast cancer development under the age of 50 to be significantly greater for those with a BRCA mutation but, crucially, not for other MR or HR women in the cohort compared to the PR group.

Our study has used a real clinical cohort, based on routine clinical practice for patients referred over a 10-year period. In this context, empirical NICE risk criteria do not appear to achieve effective risk stratification of those without a highly penetrant mutation before the age of 50. In the MR group, there was a detectable increase in cancer risk after the age of 50; however, additional screening is not mandated for this group. When interpreted as requiring a same-sided FH, empirical criteria fail to detect this difference.

It is recognised that the moderately increased risk of breast cancer observed in some families may be due to a multifactorial, polygenic risk model. The greater ability of the guidance to identify at-risk women when both sides of a FH are used in risk estimation may reflect this model of inheritance, with risk alleles being transmitted from both sides of the family. Future routine clinical practice is likely to require the analysis of genetic variants contributing to polygenic risk to achieve better performing risk estimation models. This is currently under investigation [5, 6].

NICE guidelines do suggest the use of other methods of risk stratification, specifically BOADICEA [3]. There is evidence that other methods such as BOADICEA may be effective in risk stratification [7], although there is no direct published comparison with NICE empirical criteria.

This study has used a simple methodology to assess current clinical practice in UK cancer genetics. Of 1409 patients being screened over a 16-year period, 30 developed invasive breast cancer. In this cohort, the ability of the current guidance to identify at-risk women, once highly penetrant mutations are excluded, is poor. Though we have a moderate cohort size, we feel that these results are important and should encourage further investigation of the effectiveness of these national guidelines. It would appear beneficial to refine risk stratification methods to focus resources on women who will benefit most from early screening.