SELECT: Selenium and Vitamin E Cancer Prevention Trial

Rationale for SELECT Study Supplements chosen based on secondary results for prostate cancer from other randomized controlled trials –63% risk reduction for selenized yeast (NPC) –32% risk reduction for α-tocopherol (ATBC) –Reduction in cancer mortality for combination of selenium, vitamin E and beta carotene (Nutrition intervention trials in Linxian, China) Results supported by epidemiologic and preclinical data

SELECT Schema Calendar Year Randomized Pre-Randomization Period Follow - up Prostate cancer, other cancer, death Placebo + Selenium Placebo + Placebo Vitamin E + Placebo Vitamin E + Selenium Calendar Year 2001 – 2004 (Planned 2001 – 2006)

SELECT Study Supplements Selenium 200 μg/d from L-selenomethionine or matched placebo Vitamin E 400 IU/d all rac-α-tocopheryl acetate or matched placebo

Eligibility Criteria Age of ≥ 50 if African American, ≥55 otherwise Prostate health –No prior prostate cancer or high-grade PIN –PSA ≤ 4.0 ng/ml and DRE not suspicious for cancer within 1 year prior to randomization Concurrent supplements –Must not be taking, or planning to take, any vitamin E or selenium in addition to the Study Supplements

Eligibility Criteria Other health –No prior history of cancer (except basal and squamous cell skin cancers) unless no evidence of disease for at least 5 years –No current use of anticoagulant therapy other than <175 mg/day of aspirin or if taking Plavix < 81 mg/day –No history of hemorrhagic stroke; normal BP (SBP < 160 mm/Hg, DBP < 90 mm/Hg) –No diagnosis of retinitis pigmentosa

Statistical Elements Double-blind placebo-controlled Intent-to-treat analysis Time-to-event analysis for prostate cancers, other cancers, deaths –Cox model, Hazard rates Overall incidence rate for cardiovascular, diabetes, other adverse events –Χ 2 test, Relative risk

Follow-Up Schedule Followed twice yearly for men without prostate cancer, annually for those with prostate cancer Adherence and adverse event data collected every 6-months Annual limited physical exam including assessment of blood pressure, weight, smoking status –DRE and PSA per Study Site standard of care

Study Duration Assumed: 5 year uniform accrual Actual: 3 year accrual Follow-up: until unique calendar date Supplementation stopped in 2008

DSMC Recommendation On September 15, 2008, the independent SELECT DSMC met and made the recommendation to discontinue study supplementation –No evidence of benefit for prostate cancer incidence –Planned reduction in prostate cancer incidence extremely unlikely in all arms and would not be seen within the timeframe of the trial (p<.0001)

Eligibility Status Placebo (n=8856) Vitamin E (n=8904) Selenium (n=8910) Combination (n=8863) Excluded from Primary Analysis Participants from 2 sites removed due to poor site performance Prior prostate cancer 1512 Randomized in error 4623 Included in Primary Analysis 8696 (98%) 8737 (98%) 8752 (98%) 8703 (98%)

Eligibility Status Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Clinically ineligible 134(1%) 128(1%) 113(1%) 113(1%) Insufficient baseline data 154(2%)151(2%)166(2%)169(2%)

Follow-Up Status Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Number On or Temporarily Off Median FU (yrs)5.46 Time since last contact 0 – 7 months 6913(87%)6910(88%)6925(87%)6978(88%) 7.1 – 13 months362(5%)382(4%)369(5%)337(4%) 13.1 – 19 months136(2%)149(2%)149(2%)148(2%) 19.1 – 24 months72(1%)87(1%)75(1%)75(1%) > 24 months420(5%)385(5%)434(5%)379(5%)

Follow-Up Status Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Endpoint reached Number on or temporarily off supplements Median FU (yrs)5.46 Time since last contact 0 – 7 months 6913(87%)6910(88%)6925(87%)6978(88%) 7.1 – 13 months362(5%)382(4%)369(5%)337(4%) 13.1 – 19 months136(2%)149(2%)149(2%)148(2%) 19.1 – 24 months72(1%)87(1%)75(1%)75(1%) > 24 months420(5%)385(5%)434(5%)379(5%)

Baseline Characteristics Age Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Age (years) Median (Inter-quartile range) 62.6 (58.1 – 67.8) 62.3 (58.0 – 67.8) 62.6 (58.2 – 68.0) 62.4 (58.1 – 67.8) 50 – %4025%3374%3854% 55 – %514359%507658%505258% 65 – %264130%273331%273131% ≥ %5516%6067%5356%

Baseline Characteristics Race/Ethnicity Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Race/Ethnicity White686379%689079%694279%687479% African American107812%110713%105312%107612% Hispanic (non- African American) 4926%4775%4815%4846% Hispanic (African American) 761%1031%861%951% Other1872%1602%1902%1742%

Baseline Characteristics PSA Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) PSA, ng/ml 0.1 – %420848%421848%421348% 1.1 – %265330%266130%266631% 2.1 – %122814%121114%114913% 3.1 – %6347%6527%6598% > 4.05<1%3 2 1 Unknown/missing7<1%11<1%8 15<1%

Baseline Characteristics Smoking Status Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Smoking Status Never368242%375243%378043%366642% Current6558%6598%6317%6708% Former420848%419448%421448%424249% Ever (current status unknown) 631%551%611%561% Unknown881%771%661%691%

Baseline Characteristics Education Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Education (highest level) ≤ High school grad or GED %187522%191722%189822% Some college or vocational school %238727%232727%234827% ≥ College graduate431750%439451%443051%437250% Unknown/missing951%811%781%851%

Adherence

Serum Selenium (μg/L) Data from the adherence cohort

α-tocopherol (μg/mL)* * Cholesterol-adjusted; Participants from adherence cohort

γ -tocopherol (μg/mL)* * Cholesterol-adjusted; Participants from adherence cohort

Drop-Ins

Prostate Cancer Incidence Per Study Design* and Actual * PCPT rates for years 1 – 3 and 1995 SEER rates thereafter.

Prostate Cancers Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) Total year incidence 4.43%4.93%4.56% HR (99% CI) (0.95 – 1.35) 1.04 (0.87 – 1.24) 1.05 (0.88 – 1.25) p - value---p = 0.06p = 0.62p = 0.52

Prostate Cancers 99% CI 0.95 – 1.35 p= % CI 0.87 – 1.24 p= % CI 0.88 – 1.25 p=0.52

Prostate Cancer Placebo Vitamin E Selenium Vitamin E + selenium

Prostate Cancer Method of Diagnosis Placebo (n=416) Vitamin E (n=473) Selenium (n=432) Combination (n=437) Diagnosed by biopsy40497%45897%41997%42096% Number of biopsies Reason for Biopsy (positive biopsies) Elevated PSA25964%32471%29671%26363% PSA Velocity123%102%133%164% Abnormal DRE6616%5813%4611%5613% PSA + DRE5514%4911%5613%7217%

Prostate Cancer - Gleason Score Placebo (n=416) Vitamin E (n=473) Selenium (n=432) Combination (n=437) Number graded Gleason Score %24963%21760%22060% 7 (3+4)8022%9724% % 7 (4+3)216%277%195%247% %236%206%308%

Prostate Cancer - Clinical Stage Placebo (n=416) Vitamin E (n=473) Selenium (n=432) Combination (n=437) T-Stage T1 a-c27870%34375%30173%28669% T2 a-b12230%11425%10826%12831% T3 a-b00%2 51%3 N-Stage N %127100%12599%117100% N100%0 11%00% M-Stage M %13499%12296%11998% M1a-b00%21%54%22%

Annual PSA Tests

Annual DRE Tests

Other Cancers Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Any cancer (including prostate) (0.91 – 1.17) (0.89 – 1.15) (0.90 – 1.16) Lung (0.64 – 1.55) (0.73 – 1.72) (0.76 – 1.78) Colorectal (0.69 – 1.73) (0.66 – 1.67) (0.82 – 2.00) Other primary (0.72 – 1.10) (0.77 – 1.17) (0.76 – 1.16) All p-values > 0.15

Total Cancers 99% CI 0.91 – % CI 0.89 – % CI 0.90 – 1.16

Lung CancerColorectal Cancer All Other Primary CancersDeaths Placebo Vitamin E Selenium Vitamin E + selenium

Cumulative Death Rate Per Study Design* and Actual * Estimated for placebo arm based on PCPT for the first 4 years and then adjusted upwards to 1995 US rates for all races combined.

Deaths Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N HR N HR (99% CI) N HR (99% CI) N HR (99% CI) All deaths (0.77 – 1.13) (0.82 – 1.19) (0.77 – 1.13) Cancer deaths (0.60 – 1.18) (0.74 – 1.41) (0.67 – 1.30) Cardiovascular deaths (0.61 – 1.15) (0.66 – 1.24) (0.60 – 1.13) Other deaths (0.83 – 1.60) (0.75 – 1.47) (0.78 – 1.51)

Cardiovascular Events Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) All Cardiovascular events* (0.88 – 1.09) (0.92 – 1.13) (0.89 – 1.10) * Grades 3 – 5 (includes deaths)

Cardiovascular Events Non-fatal Strokes Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Hemorrhagic (0.18 – 2.20) (0.33 – 2.98) (0.37 – 3.19) Ischemic (0.51 – 1.44) (0.55 – 1.49) (0.75 – 1.90) NOS (0.24 – 1.32) (0.17 – 1.11) (0.37 – 1.73)

Cardiovascular Events Other Non-fatal Events Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Grade (0.89 – 1.17) (0.95 – 1.25) (0.87 – 1.15) Grade (0.82 – 1.38) (0.78 – 1.31) (0.82 – 1.37)

Diabetes Diagnoses Placebo (n=7156) Vitamin E (n=7215) Selenium (n=7210) Combination (n=7248) Number (%)669 (9.3%)700 (9.7%)724 (10.0%)660 (9.1%) Relative Risk (99% CI) (0.91 – 1.18) 1.07 (0.94 – 1.22) 0.97 (0.85 – 1.11) p-value--p = 0.47p = 0.16p = 0.61 Based on self-report of diabetes or reported use of diabetes medications of the glitazone class. Excludes prevalent cases at randomization (n=3625) and participants not evaluated for diabetes because they either died or were lost to follow-up prior to the assessment (n=2434).

Events Known to Be Associated with Study Supplements Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Alopecia (0.83 – 1.36) (1.01 – 1.62) (0.91 – 1.47) Dermatitis Grades 1 – (0.98 – 1.32) (1.00 – 1.35) (0.92– 1.25) Grades 3 – (0.46 – 4.83) (0.56 – 5.44) (0.66 – 6.09) = p < 0.01

Events Known to Be Associated with Study Supplements Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Halitosis (0.97 – 1.36) (0.99 – 1.38) (1.06 – 1.46) Nail changes (0.90 – 1.11) (0.94 – 1.26) (0.93 – 1.15) = p < 0.01

Events Known to Be Associated with Study Supplements Placebo (n=8696) Vitamin E (n=8737) Selenium (n=8752) Combination (n=8703) N RR N RR (99% CI) N RR (99% CI) N RR (99% CI) Fatigue Grades 1 – (0.89 – 1.19) (0.95 – 1.26) (0.90– 1.20) Grades 3 – (0.59 – 2.45) (0.40 – 1.88) (0.38 – 1.81) Nausea Grades 1 – (0.72 – 1.21) (0.94 – 1.52) (0.77– 1.28) Grade (0.06 – 1.85) (0.30 – 3.34) (0.25 – 3.10)

Study Strengths Robust statistical design –Ample power for 5 comparisons between arms –Very large study population, including African Americans –Baseline characteristics well balanced –Design assumptions largely met Rigorous implementation –SELECT Workbench –Web-based forms with detailed edit checks –Semi-annual staff training; targeted mentoring

Study Limitations Not designed to test different supplement formulations and doses Unable to assess effects in reducing advanced or fatal prostate cancer –Active annual screening reduced incidence of advanced disease across all study arms Unable to assess effects in a vitamin E and/or selenium deficient population –Study population well-nourished at baseline

Conclusion SELECT has definitively demonstrated that, in these doses and formulations, selenium, vitamin E, and selenium + vitamin E do not prevent prostate cancer in the generally healthy, heterogeneous population of men in SELECT.

Other Findings No difference in rates of lung, colon or other primary cancers No differences in deaths

Other Findings Vitamin E: Statistically nonsignificant increase in prostate cancer (HR 1.13, 99% CI 0.95 – 1.35, p=.06) Selenium: Statistically nonsignificant increase in type 2 diabetes mellitus (RR 1.07, 99% CI 0.94 – 1.22, p=.16) Neither difference seen in vitamin E + selenium group

Vitamin E: Why Not a Positive Finding? Dose –The higher dose (400 IU/day) may be less effective than the lower dose (50 IU/day) –Effect on suppression of γ-tocopherol may be greater with the higher dose Population – Vitamin E may be most effective in smokers; SELECT population is only 7.5% smokers compared to ATBC which was 100% Inherent problems with secondary/exploratory findings of other trials; it’s why we do carefully controlled trials to answer the question

Selenium: Why Not a Positive Finding? Formulation –l-selenomethionine was chosen over high-selenium yeast –Selenomethionine is the major component of the high- selenium yeast –Pilots showed large batch-to-batch variation in the yeast Population – Prior study had men chosen because they were deficient in selenium; selenium was most protective in those with the lowest levels – Our population were replete in baseline selenium Inherent problems with secondary/exploratory findings of other trials; it’s why we do carefully controlled trials to answer the question