2. The ‘Arimidex’, Tamoxifen Alone or in Combination (ATAC) trial: Completed Treatment Analysis Questions & Answers / Issues Management

3. Survival and distant recurrence

4. Q1: If ‘Arimidex’ is a superior drug compared with tamoxifen, why did it not show a survival advantage in this ‘definitive’ 5-year analysis?

5. ‘Arimidex’ maintains the considerable survival benefit already established for tamoxifen in this setting Primary endpoints of ATAC, DFS and tolerability, were both significantly in favour of ‘Arimidex’ –Overall survival was a secondary endpoint –Trial was not statistically powered for survival –Non-inferiority was clearly demonstrated for ‘Arimidex’ based on upper CI < 1.25

6. Q1: If ‘Arimidex’ is a superior drug compared with tamoxifen, why did it not show a survival advantage in this ‘definitive’ 5-year analysis? Absence of a benefit at this point is not unexpected ! Large randomised studies versus placebo take a long time to detect a significant advantage in overall survival –in the early 1980s it was widely but mistakenly believed that adjuvant tamoxifen did not affect survival –NSABP P14 (n=2644) took 7 years to detect a significant survival advantage for tamoxifen over placebo* * Fisher et al. J Nat Cancer Inst 2001; 93: 684-690

7. Q1: If ‘Arimidex’ is a superior drug compared with tamoxifen, why did it not show a survival advantage in this ‘definitive’ 5-year analysis? ATAC compares 2 active treatments, therefore it is likely to take even longer to detect a meaningful difference Prognostic factors are also important –many patients in ATAC had good prognoses  61% of patients were lymph node negative  64% of patients had tumours  2 cm It is therefore still too early to expect a significant survival difference between treatment arms at this time

8. *EBCTCG. Lancet 1998; 351: 1451–1467 Time to recurrence or death in the first 10 years after surgery* DeathRecurrence Patients (%) Most recurrences occur during the first 5 years after surgery, but most deaths occur during the second 5 years after surgery

9. Why did ‘Arimidex’ not show a survival advantage over tamoxifen in ATAC trial? Follow-up time (years) 123456 0 7 In the NSABP B14 1 (n=2644) it took 7 years to detect a survival advantage for tamoxifen over placebo There was no difference in survival between ‘Arimidex’ and tamoxifen in the ATAC trial at 68 months The ATAC trial population had a relatively good prognosis 61% of patients were lymph node-negative 64% of patients had tumours <2 cm in diameter It is too early to see a survival advantage! 1 Fisher B et al. J Natl Cancer Inst 1996;88:1529-1542

10. Q1: If ‘Arimidex’ is a superior drug compared with tamoxifen, why did it not show a survival advantage in this ‘definitive’ 5-year analysis? There is already a non-significant trend for reduction in breast cancer mortality with ‘Arimidex’ This together with the significant reductions in DFS and TTDR strongly suggest a benefit in breast cancer survival will be observed with time

11. Q2: If the HR+ve population is more clinically relevant than the ITT population, why isn’t the difference for TTDR significantly in favour of ‘Arimidex’ for these patients?

12. The difference in TTDR between treatment arms was not significant in the HR+ve population in the ATAC trial (p=0.06) TTDR data depends on distant recurrence events –As expected, the HR+ve population respond better to treatment and therefore there are fewer recurrences in these patients (~ 100 fewer events per arm) –The HR+ve population is also smaller than the ITT population –Therefore there are currently insufficient events to detect a significant difference HR+ve=hormone receptor-positive; ITT=intent-to-treat; TTDR=time to distant recurrence

13. Recurrence events by patient population The HR for TTDR is more in favour of ‘Arimidex’ in the HR+ve population (0.84) than the ITT population (0.86) This difference is expected to reach significance over time as more events are observed Endpoint Recurrence or death Recurrence Distant recurrence Contralateral breast cancer Events A vs T 575 vs 651 402 vs 498 324 vs 375 35 vs 59 p-value 0.01 0.0005 0.04 0.01 Events A vs T 424 vs 497 282 vs 370 226 vs 265 26 vs 54 p-value 0.005 0.0002 0.06 0.001 ITT population ‘Arimidex’ (A) n=3125; Tamoxifen (T) n=3116 HR+ve population ‘Arimidex’ (A) n=2618; Tamoxifen (T) n=2598 HR A vs T 0.87 0.79 0.86 0.58 HR A vs T 0.83 0.74 0.84 0.47 HR=hazard ratio; HR+ve=hormone receptor-positive; ITT=intent-to-treat

14. Subgroup analyses

15. Q3: There are certain subgroups of patients who appear to derive little or no benefit from ‘Arimidex’ over that seen with tamoxifen – should I limit my prescribing accordingly?

16. ‘Arimidex’ is indicated for all postmenopausal patients with hormone- sensitive disease There is no subset of women in the ATAC trial in which tamoxifen confers greater benefit The treatment effect is consistent across baseline prognostic factors, with no heterogeneity for any subgroup

17. Q3: There are certain subgroups of patients who appear to derive little or no benefit from ‘Arimidex’ over that seen with tamoxifen – should I limit my prescribing accordingly? ATAC was NOT powered to detect differences in subgroups All subgroup analyses such as these are exploratory –Findings should not be over-interpreted or influence patient selection unless they are clearly contradictory of the overall conclusion It should be the overall result of ATAC which influences clinical practice not a retrospective, exploratory subgroup analysis

18. Q4: Did patients who received adjuvant chemotherapy receive the same benefits as those who did not? Early results suggested that ‘Arimidex’ conferred no additional benefit over tamoxifen in this subgroup

19. Approximately 20% of patients in the ATAC trial received prior chemotherapy The first analysis (33 months) revealed a potential difference in outcomes based on prior chemotherapy The updated analysis (47 months) detected a detrimental HR only for patients who had received CMF (sub- optimal?), but not for those receiving anthracyclines or taxanes The ATAC completed treatment analysis (68 months) found no interaction between ‘Arimidex’ and prior chemotherapy

20. Time-to-recurrence by chemotherapy usage Intent-to-treat population Hazard ratio (A:T) and 95% CI 2001Overall 2002 2004 ‘Arimidex’ (A) betterTamoxifen (T) better 0.60.81.01.21.41.6 Chemotherapy no Chemotherapy yes Overall Chemotherapy no Chemotherapy yes Overall Chemotherapy no Chemotherapy yes

21. The efficacy of ‘Arimidex’ is superior to that of tamoxifen regardless of any baseline prognostic factor or prior adjuvant treatment Nodal status, PgR status or prior adjuvant chemotherapy is not a reason for denying patients ‘Arimidex’ !

22. Long-term safety data

23. Q5: Why were there more non-breast cancer-related deaths with ‘Arimidex’ compared with tamoxifen?

24. There were a total of 831 deaths at the time of this analysis –500 (60%) breast cancer deaths –331 (40%) non-breast cancer deaths Overall survival was similar in both treatment arms Numerically, more deaths overall were seen in the tamoxifen group. However, there were more deaths before recurrence (non-breast cancer deaths) in the ‘Arimidex’ group Q5: Why were there more non-breast cancer-related deaths with ‘Arimidex’ compared with tamoxifen?

25. The difference in the number of non-BC deaths between ‘Arimidex’ and tamoxifen is small and does NOT represent a safety concern Review of the data shows that these deaths before recurrence were spread across a variety of apparently causes and no link to ‘Arimidex’ was found –There is no increase in any one particular cause of death –The majority were not considered related to study therapy and occurred after patients had completed or discontinued treatment –There is no increase in the number of deaths with longer drug exposure Q5: Why were there more non-breast cancer-related deaths with ‘Arimidex’ compared with tamoxifen?

26. The imbalance is not unexpected! Due to the significantly reduced number of withdrawals due to adverse events and disease recurrence compared with tamoxifen, more patients stay on ‘Arimidex’ for longer These patients therefore become at higher risk of death from other causes! Q5: Why were there more non-breast cancer-related deaths with ‘Arimidex’ compared with tamoxifen?

27. Q6: What are the risks of experiencing bone fractures on ‘Arimidex’ compared with tamoxifen?

28. Fracture risk increases with postmenopausal status and increasing age –in postmenopausal women, fractures due to osteoporosis typically occur in the spine, hip and wrist –fractures of the hip have a high morbidity and mortality Tamoxifen is known to have ‘bone-protecting’ effects, due to its partial oestrogen-agonist mode of action Important to establish the risk:benefit ratio for ‘Arimidex’ vs tamoxifen as adjuvant therapy in postmenopausal women

29. The fracture risk with ‘Arimidex’ is predictable and manageable The RR of fracture remains steady, stabilising after 2 years, and shows no evidence of worsening over time –the ATAC Completed Treatment Analysis shows that the OR of fracture was 1.49; the RR was 1.60 in previous analyses –importantly, there is no significant difference between ‘Arimidex’ and tamoxifen for fractures of the wrist, or even hip –spinal fractures were more frequent with ‘Arimidex’ but these are vertebral compression fractures OR=odds ratio; RR=relative risk Q6: What are the risks of experiencing bone fractures on ‘Arimidex’ compared with tamoxifen?

30. 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 6122436424830 Months 18 Wrist, spine and hip fracture rates measured at 6-monthly intervals 6-monthly fracture rates/ 100 patients Locker, Eastell. Proc ASCO 2003; Abs 98 Stability of fracture rates over time Tamoxifen 20 mg od ‘Arimidex’ 1 mg od Fracture rates stabilise after 2 years of treatment and relative risk remains constant

31. HRT (total) = 14.75 Placebo (total) = 19.10 Healthy women Age = 63 (45% 50-69) WHI (n=16,608) Tamoxifen = 18.05 Placebo = 18.40 Breast cancer prevention Age = 35-49 (39%), >50 (61%) PI ( n=13,175) ‘Arimidex’ = 22.6 Tamoxifen = 15.6 Early breast cancer (adjuvant) Age = 64 (mean) ATAC (n=6186) Fracture rate/1000 patient years Setting Average age (years) Clinical study ATAC Trialists’ Group, 2004; Fisher et al. J Natl Cancer Inst 1998; 90: 1371–1388; Women's Health Initiative Writing Group. JAMA 2002; 288: 321–333 Indirect fracture rate comparison Fracture rates in the ATAC trial fall within the broad range of fracture rates in other large trials or surveys, indicating that any increase associated with ‘Arimidex’ is modest

32. 2002 clinical practice guidelines for the diagnosis and management of osteoporosis in Canada, CMAJ 2002; 167 Standard care for all postmenopausal women: osteoporosis guidelines Recommend healthy lifestyle changes –smoking cessation –moderating caffeine and alcohol intake –regular weight-bearing exercise –calcium and vitamin D supplements Screen for osteoporosis risk factors DXA scan on all patients with risk factors Bone health is manageable

33. ASCO Tech Assessment 2004: “Aromatase inhibitor–associated bone loss may represent a preventable and treatable condition” Clinical trial evidence indicates that bisphosphonates are effective in maintaining bone density in breast cancer patients on hormonal therapy In otherwise healthy women, a strong body of evidence supports early detection and therapy of osteoporosis The ASCO bisphosphonate guideline identifies postmenopausal breast cancer patients on AIs to be at high risk for osteoporosis and recommends baseline BMD evaluation. It is hoped that implementation of these guidelines will reduce fracture rates associated with adjuvant AI use Based on recent results and ongoing studies, adjuvant bisphosphonates may become a more standard component of the treatment of women with early-stage breast cancer

34. Issue Management Document