1. Genetics and Risk of Breast Cancer What is the Evidence

2. Questions What is the role of mutation testing What is the risk to mutation carriers What is the evidence for intervention How does family history predict risk What lines of future research are required

3. What is the likelihood of finding a mutation ?

4. BRCA1 and BRCA2 Mutations confer an autosomal dominant susceptibility to Breast and Ovarian cancer with high penetrance In some populations there are common mutations These are not the only genes involved

5. BRCA1 and BRCA2 Population frequency (British Cases) BRCA1 0.11% BRCA2 0.12%(Peto et.al. 1999) Estimate 16% of hereditary breast cancer susceptibility is caused by these genes in UK.

6. Determining Probability of Being a GeneCarrier Empirical data Logistic Regression Analysis Bayes calculation

7. Ford et.al. 1997, Narod et.al. 1995 84% of BCLC families showed evidence of linkage to BRCA1 or BRCA2. (4 affected members). 76% of breast-ovarian families linked to BRCA1 (3 affected members, one ovarian).

8. Peto et.al. 1999 BRCA1 and BRCA2 Mutation Analysis StatusMutation detected BRCA1 BRCA2 Affected < 359/254 6/254 Affected 36-457/363 8/363 Affected <45, mother breast cancer2/54 1/54 Affected <45, 1 o with breast CA<603/52 1/52 *Affected <45, 1 o with ovarian CA3/5 0/5 *Affected 36-45, 2X 1 o with breast <601/8 3/8

9. Schattuck Eidens et.al. 1997 BRCA1 and BRCA2 Mutation Analysis StatusMutation detected BRCA1 BRCA2 Affected < 35~5% Affected 36 - 451-2% Affected <45, mother breast cancer~2% Affected <45, 1 o with breast CA<60~2% Affected <45, 1 o with ovarian CA~6% Affected 35, 1 o relative with breast + ovarian20% Affected 35, Bilateral Breast Cancer20%

10. Parmigiani et.al.1998 Bayes Risk Calculation –Uses population frequency of mutation –Uses penetrance data for gene mutation –Takes family structure into account –Assumes all non BRCA1/BRCA2 cancer is sporadic –Has been computerised

11. CASH dataFord data

12. Likelihood of identifying a mutation BRCA1 and BRCA2 Mutation Detection StatusCyrillicPeto (BRCAPro) BRCA1BRCA2BRCA1BRCA2 Affected < 352%0.3%4%2% Affected <45, 1 o with breast CA<604%2%6%2% Affected <45, 1 o with ovarian CA22%1.6%60%0% Affected <45 2X1 o relative with breast <6021%10%13%38%

13. Population Specific Mutations Ashkenazi Jewish PopulationOver 2% of population BRCA1185delAG 5382insC BRCA26174delT Icelandic population0.6% of population BRCA2999del5

14. Mutation Detection with Askenazi Jewish descent Lalloo et.al. 1998 –Breast cancer <601/4 –2 X Breast cancer <703/10 –BCLC criteria5/5 – Schattuck-Eidens et.al. 1997 –Affected age 40~12% –Affected at 40 + 1 o relative breast~20% –Affected at 40 + 1 o relative ovarian~35%

15. Male Breast Cancer Friedman et.al. 1997 (Californian Male Cases) 2/54 cases of male breast cancer had BRCA2 mutations 17% had a 1 o family history of breast cancer

16. Conclusions In a small proportion of cases, mutation testing for BRCA1 and BRCA2 would be expected to have a high pickup rate. Eg. 4 family members with breast cancer Breast cancer <age 45 with 1 o ovarian cancer

17. Conclusions Different systems exist to predict likelihood of detecting a mutation. – BRCApro –Logistic regression curves Not all of these have been validated in clinical practice.

18. Conclusions Testing for the common Ashkenazi Jewish mutations may be relevant, –In the presence of modest family history. –with isolated young onset disease.

19. What is the Risk to Mutation Carriers ?

20. Risk to Mutation Carriers Derived from Linkage –Easton et.al. 1993 –Ford et.al. 1994,1998 Empirical data from common mutations –Struewing et.al. 1997 –Steinumn et.al. 1998

21. Risk To Mutation Carriers (%) Easton et.al. Ford et.al. Struewing Steinum ISD BRCA1 BRCA2 BRCA1+2 BRCA2 (Population) Br Ov Br Ov Br Ov Br Ov Br O v By age 40 19 0.6 12 0.0 - - - - - - By age 5050 22 28 0.4 33 7 15 - - - By age 6054 30 48 7.4 - - - - - - By age 65 - - - - - - - - 5.5 0.9 By age 7085 63 84 27 56 16 35 - - - By age 75 - - - - - - - - 7.9 1.5

22. Conclusions BRCA1 and BRCA2 mutations confer a high risk of breast and ovarian cancer. All studies have potential sources of bias, the true risk will depend on the population and mutation type.

23. Modifying Risk to Gene Carriers

24. Modifying Risk ScreeningBreast examination Mammography Ovarian Ultrasound Hormonal ManipulationTamoxifen Surgical InterventionMastectomy Oophorectomy

25. Screening / Mammography Proven benefit when age > 50 in individuals at population risk –Meta-analysis, Kerlikowske JAMA 1995 Conflicting Evidence for population screening ages 40 to 49 –Some studies for, some against High Risk Screening –Uncontrolled longitudinal follow up of high risk cohort

26. Screening / Mammography Chart et.al. 1997 - Canadian High Risk Programme –1044 women categorised as high, moderate or low risk –6 year follow up, mammography and breast examination –in high risk group 7/381 had tumours at presentation –5/381 high risk developed tumours on follow up Lalloo et.al. 1998 - Manchester high risk breast clinic –1259 women with a lifetime risk of breast cancer > 1 in 6 –7 tumours prevalent (4 were in situ), 9 tumours incident –2 tumours were detected by self examination between screens –6 of incident tumours were palpable

27. Tamoxifen prevention Studies The Breast Cancer Prevention Trial (P1) –Women at “increased Risk”, 13388 cases, 5 year follow up –Tamoxifen reduced breast cancer risk (RR 0.5) –Increased endometrial cancer and pulmonary embolus + cataract –Overall mortality not significantly lower Royal Marsden Chemoprevention Trial –8 year follow up of 2471 women, power 90% for 50% effect –No detectable effect on breast cancer

28. Hormone Replacement Therapy No study has looked at HRT in BRCA mutation carriers Generally, HRT confers a small increased risk of breast cancer. HRT decreases cardiovascular and osteoporotic events. In one large meta-analysis (anonymous 1997) positive family history did not show a significantly increased risk of breast cancer in HRT users as opposed to non-users. Numbers analysed were small. Breast cancer in BRCA1 carriers is often oestrogen receptor negative.

29. Prophylactic Surgery Mastectomy –Various modelling approaches looking at cost/benefit –Hartmann et.al. 1999 Retrospective study Estimated 90% reduction in breast cancer incidence Did not take other post-operative morbidity into account Oophorectomy –Rebbeck et.al. 1998 (ASHG abstract) reduction in breast cancer in BRCA1 mutation carriers –Papillary serous carcinoma of peritoneum may arise in BRCA1 carriers after oophorectomy. ( Schorge et.al.1998, Piver et.al. 1993.)

30. Conclusions The benefits of prophylactic tamoxifen remain unproven. Family history of breast cancer is not necessarily a contraindication for HRT. More studies are needed

31. Conclusions Prophylactic mastectomy can have a role in patients at high risk of breast cancer. Prophylactic oophorectomy may reduce risk of ovarian cancer and breast cancer. Peritoneal tumours may still arise.

32. Conclusions Screening of high risk patients can be effective in detecting breast cancer. Overall benefit of screening remains to be demonstrated.

33. Presenting Risk

34. Lifetime Absolute Risk Your lifetime risk of dying is 100%

35. Relative Risk Your Risk of dying is 1X that of the population

36. Absolute Risk over Time Your risk of dying over the next 10 years is 2%

37. Presenting Risk Absolute Risk over a given time is –Easy to understand –Easy to base decisions upon Relative Risk can be converted to absolute risk –Assuming relative risk is constant over time –Assuming individual belongs to the population studied (Dupont and Plumber 1996)

38. How Can Risk be Estimated from Family History ?

39. Risk Analysis - Situation A Mother and Sister Affected with Breast Cancer 60

40. Risk Analysis - Situation B Three Relatives Affected with Breast Cancer

41. Risk Analysis - Situation C Mother Affected with Ovarian Cancer and Sister with Breast Cancer

42. Risk Analysis - Situation D Mother affected with bilateral breast cancer 40/55 40

43. Risk Analysis - Situation E Two second degree relatives with breast cancer < age 60

44. Risk Analysis - Situation F Mother affected with breast cancer age 45 and ovarian age 65

45. Estimation of Risk Empirical Data Studies –OPCS data set3295 cases of breast cancer –CASH data set4730 cases of breast cancer –Meta-analysis - Pharoah et.al.74 published studies Modelling of Data –CASH data –Gail Model2,852 cases of breast cancer Linkage/Computer Analysis –Cyrillic (Uses CASH data)

46. Empirical Estimation of Risk Advantages –No model is assumed –Information is directly applicable Disadvantages –Data is population specific –Data only covers a small range of situations –Large studies are needed for meaningful data

47. CASH data (Claus et.al. 1990) Kaplan-Meier estimates of cumulative risk –By age of onset of breast cancer in 1 o relative Hazard ratios for other family histories –Sister and mother affected RR 5.9 (3.9-8.9) –Two sisters affected RR 3.6 (2.1-6.1) –One mother, 2 sisters RR 17 (9.4-31)

48. Meta-Analysis Pharoah et.al.1997 Applicable to limited situations –Affected 1 o relativeRR 2.1 (2.0-2.2) –Mother and sister RR 3.6 (2.5-5.0) –Sister affected <50RR 2.7 (2.4-3.2) –Mother affected <50RR 2.0 (1.7-2.4)

49. Estimation of Risk Using Models Advantages –Data can be widened to a greater range of situations Disadvantages –Can generalise to situations where data is not applicable –Risks are often based on a small number of data points –Risks calculated are population specific

50. Modelling of Breast Cancer Risk “Gail” Model (BCDDP data) –Incorporates age at menarche, parity, and affected 1 o relatives –Curves given to estimate 10, 20 and 30 year risk –No mode of inheritance assumed CASH data model –Using age of onset and first degree relative data only –Segregation analysis suggests dominant major locus –Give cumulative risk curves based on relatives and age of onset

51. Linkage/Computer Analysis Likelihood of developing breast cancer = Likelihood of dominant mutation in family + Likelihood of carrying mutation + Likelihood of developing cancer if mutation carrier (Mendel to determine LOD score, CASH data penetrance figures/ current age)

52. Linkage/Computer Risk Analysis Advantages –Generates a risk for a complex situation Uses age of onset Uses unaffected individuals –Easy to apply –Removes subjective element Disadvantages –Assumes single dominantly inherited gene –Assumes one set of penetrance values for a single gene –Prone to “rubbish in, rubbish out” phenomena –Essentially unvalidated

53. So Which is Best ?

54. 20 Year Risk Comparison

55. Conclusions Different systems for risk estimation can give different results. Empirical risk calculation systems can only apply to well defined situations. Computerised risk assessment is based on assumptions that are not necessarily valid.

56. Where do we start screening ?

57. Current Guidelines SIGN Guideline - 3 times population risk –1 o relative with bilateral breast cancer* –1 o relative with breast cancer <40 –1 o male relative with breast cancer –1 o relative with breast and ovarian –2 first or second degree relatives with breast cancer < 60 –3 first or second degree relatives with breast cancer

58. Bilateral breast cancer in 1 o relative 2-5% of breast cancer is bilateral CASH data (USA families) –risk same as if unilaterally affected relative Tulinius et.al. 1992 (Icelandic families) –RR 4.4* (3.39-6.49), RR 9 if first onset <45 Houlston (British families, OPCS data) –RR 4.78 (0.12 to 26.62) postmenopausal onset –RR 7.78 (0.94 to 28.08) premenopausal onset

59. 10 Year Risk Estimates

60. Conclusions Setting the criteria for screening should depend on –Estimation of absolute risk (age dependant) –Effectiveness of screening (may be age specific) –Resources available –A sensible risk estimation system

61. Future Research Validation of risk estimation –Detailed comparison of methods of risk analysis –Application to pedigrees with known outcome (A retrospective-prospective study !)

62. Future Research Validation of screening protocols for high risk individuals.

63. Future Research ( Long term) Audit of effectiveness of screening protocols and accuracy of risks calculated. This will be greatly facilitated by an effective computerised database.