Single Affected Relative

Table 1 gives the estimated risks over the next years, for unaffected women of given ages with an affected first degree relative, based on the estimated relative risks from the Collaborative Group paper applied to 2001 population rates.

Table 1

Estimated risks for unaffected women with an affected first degree relative.

For women aged 40, the ten year risk is close to 3% for women with a 1^st degree relative diagnosed up to age 60. If, however, the risks are based on 1990 rates, only women a first degree relative diagnosed below age 40 would qualify. The same would apply if the criterion were raised to, say, 3.5% to reflect the increase in incidence.

Another issue that the breast cancer incidence rates in the next ten years will increase with the age of the woman, within the 35–50 age-group, and that the categorisation should in theory reflect this. Peto has argued that the risk of breast cancer is approximately constant (at about 3.5% p.a.) after the age of diagnosis of the first case. Most data (including the Collaborative Group re-analysis) seem to be broadly consistent with this. So one could argue that a more consistent definition of the moderate risk category would be: any women older than the age at diagnosis of her affected 1^st degree relative. In the guidelines, a fixed age cut-off was preferred for simplicity.

For women with only an second (or more distant) degree relative affected, excess risks will be reduced by (at least) a factor of 2 under any plausible model. On this basis, no woman would qualify as moderate risk (except possibly women with a second degree relative diagnosed below age 30, for which the data are poor).

Two affected relatives

The Collaborative Group paper quotes a relative risk of ~ 8 fold for women <50 with two first degree relatives affected over age 50. This would be sufficient to get them into the moderate risk group. Some data (e.g. from the Swedish population register) suggests this risk may be exaggerated. It is unclear on the basis of these data whether there should be an age at diagnosis cut-off. For the purposes of the guidelines, the moderate risk group is taken to include any women with two affected 1^st degree relatives at any age.

There is a discrepancy here with the Claus model. Under the Claus model, the risk drops markedly for cases aged over 60. This has to do with the shape of the incidence curve for carriers of the postulated “susceptibility allele” in this model (the probability of an affected women diagnosed with breast cancer, who has an affected relative diagnosed at the same age, carrying the susceptibility allele drops from over 30% at ages 50–59 to approximately 10% at ages 60–69. This is probably a weakness in the model, reflecting its attempt to model risk in terms of a single high risk gene. The Boadicea model does not exhibit this behaviour and gives risks exceeding the 3% threshold at all ages of diagnosis.

For women with one 1^st and one 2^nd degree relative we do not have direct estimates. On the basis of the Boadicea (polygenic) model, the relative risk for such women would be close to 3, and would therefore qualify (Antoniou et al, 2002). The Claus model also indicates that such women qualify (for ages at diagnosis below 60).

Similar arguments apply to women with two 2^nd degree relatives (e.g. grandmother and aunt). Heuristic arguments would suggest that these have half or less of the excess risk of the former group and would not generally qualify as moderate risk. Both the Claus and Boadicea models agree with this. However, women with two second degree relatives diagnosed at a young age would qualify. These include women with two second-degree relatives diagnosed below age 50. For consistency with the high-risk category, this group is taken to include cases where the average age at diagnosis is below 50.

Bilateral breast cancer

There is large volume of data indicating that the risks are greater to the relatives of bilateral cases. Based on the study of Hemminki et al, based on the Swedish registry, which is the largest, the overall RR of breast cancer in daughters of bilateral breast cancer cases is 3 fold, and is ~2.6 fold even for cases diagnosed over age 60. The most the most logical criterion would be to include all first degree relatives of bilateral cases as moderate risk. This is consistent with the argument that (from a genetic viewpoint) a bilateral case should count as two cancers. Although some criteria have included an age cut-off. the number of women presenting with an affected bilateral case where both cancers were diagnosed over 50 is likely to be small.

Male Breast Cancer

The evidence for an increased risk associated with a first degree relative with male breast cancer has been found in many studies, with relative risks typically >2 fold. There are few data on the effect of age, though the largest study (Rosenblatt et al. 1991) study did estimate a 3 fold relative risk for cases <60 and a lower relative risk for older cases. A cut-off of <60 is defensible, but no age cut-off would be simpler and more consistent with the specialist referral.

Ovarian Cancer

Most data suggest that the increased risk associated with a family history of both breast and ovarian cancer is mostly due to BRCA1 and BRCA2 mutations.

A complication here is that the effect of age at breast cancer diagnosis on BRCA1 prevalence is much stronger than for BRCA2 or for the familial risks. Thus the prevalence in cases drops from ~5% in the 30–39 age-group to less than 1% in the 50–59 age-group. Nevertheless, the prevalence among cases with a family history of ovarian cancer will be substantial even for older cases (although this will also depend on the age of diagnosis of the ovarian cancer, another complication). For example, even for cases aged 60–69 at diagnosis with an affected relative with ovarian cancer aged 40–49, the probability of a BRCA1 mutation is of the order of 10% (based on various calculations, including the estimates of Antoniou et al, 2003 and Boadicea predictions). On this basis, women in this category should be included as moderate risk regardless of the age at diagnosis of the breast cancer.

The logical criterion here would be all women with two 10 (or one 10, one 20) affected relatives with breast cancer and/or ovarian cancer.

2 case families

2 1^st degree relatives. It is clear that the two cases at any age does not qualify for the high risk category, on the basis of an 8% risk over the next ten years (the Collaborative Group paper estimates a risk of ~5%) or on the basis of mutation carrier probability. Two cases at any age might qualify on the basis of a cumulative risk of 25% by age 80.

According to the collaborative group paper, the ten-year risk would exceed 8% if one case is diagnosed below age 40. Thus, women with two relatives affected at a sufficiently young age would qualify as high risk. It is uncertain where the correct age cut-off should be. For the purposes of the guidelines, an average age of 50 has been chosen.

One 1^st and one 2^nd degree relative. Again there is uncertainty in this situation. A single gene model such as Claus would predict a similar risk to that for 2 1^st degree relatives and so would include, but a more complex model such as Boadicea predicts substantially lower risks. For the purpose of the guidelines the same average of 50 criterion has been adopted for consistency, but this requires further research.

Ovarian cancer

Here it is clear that, on the basis of known BRCA1/2 risks, at least for ovarian cancers diagnosed in the 40–49 age-group, the carrier probability will exceed 20% for families with two ovarian cancers. For families with or one breast cancer and one ovarian cancer diagnosed in the 40–49 age-group, the carrier probability may be approximately 20% (~17% on the basis of recent BRCA1/2 risks; Antoniou et al, 2003). By similar arguments, the carrier probabilities will be too low if the breast cancer case is diagnosed above age 50.

The carrier probabilities are lower if ovarian cancer is diagnosed below age 30 (rare) or above age 50 (for BRCA1) or 60 (for BRCA2).

Male breast cancer

Similar arguments apply for male breast cancer. 5–10% of cases harbour a BRCA2 mutation, so a single case would not be sufficient for high risk, but a male breast cancer in conjunction with a female case (possibly at any age) would be sufficient.

Families with 3+ cases

Calculations based on a variety of models indicate that families with 3 cases (for example, mother, sister and aunt) do not necessarily qualify as high risk on any basis, whilst families with three cases diagnosed at a young age (for example, below age 50) do qualify. For the purposes of the guideline we have chosen an arbitrary cut-off of an average of less than 60 years.

Families of Ashkenazi Jewish descent

Families of Ashkenazi Jewish descent are treated differently because the BRCA1/2 carrier probabilities are higher. Systematic studies indicate carrier probabilities in excess of 20% for breast cancer cases diagnosed below age 40 (but not at older ages) and for ovarian cancer cases. It should be noted that Ashkenazi Jewish women do not generally have a higher risk of breast cancer than other women in the U.K. The reason for referral in this case related solely to the greater BRCA1 carrier probabilities.