Prevention Trial of Tamoxifen in Postmenopausal HRT Users
Prevention Trial of Tamoxifen in Postmenopausal HRT Users
We conducted a phase-III, double-blind, placebo-controlled trial in recently postmenopausal women using HRT for menopausal symptom relief who were randomly allocated to either placebo or tamoxifen 5 mg/day for 5 years (trial IEO S51/200, HRT opposed by low-dose tamoxifen, HOT, Clinical Trials.gov NCT01579734). Eligibility criteria were current HRT use or de novo HRT use and negative mammography within 6 months. Toxicity was evaluated using the NCI toxicity criteria v2.0. Clinical examinations were repeated every 6 months and mammography was repeated annually. Transvaginal ultrasounds were carried out at baseline and repeated in case of atypical bleeding, followed by hysteroscopy on clinical judgment. At completion of the 5-year intervention, clinical visit and mammography were repeated annually up to 10 years. The use of venflaxine was allowed. The primary end point was the incidence of breast cancer, whereas the secondary end points included endometrial cancer, coronary heart syndrome, cerebrovascular events, venous thromboembolic events (VTEs), bone fractures, all cancers.
Central pathology assessment of breast cancer included histological type, grade, peritumoral vascular invasion, ER, PgR, HER2 and Ki-67, as previously described. Molecular tumor subtypes were classified by immunohistochemistry (IHC) into four categories according to 2011 St Gallen criteria: luminal-A, ER or PgR-positive and Ki-67 < 14%; luminal-B, ER or PgR-positive and either Ki-67 ≥ 14% with negative HER2 or HER2-positive regardless of the Ki-67 level; HER2-positive and ER/PgR-negative; triple negative, all receptors negative.
The study was designed to detect a 40% reduction of breast cancer with 80% power and two-sided 5% significance level. Using the log-rank test, 126 breast cancers had to be observed on 8500 subjects recruited over 5 years with 80% compliance and a background incidence rate of 4/1000. Recalculations based on a hazard ratio (HR) of 0.5, closer to the range observed in the Italian tamoxifen trial, and practical considerations based on the dramatic drop of HRT use after the first WHI trial results led to a total of 4500 women to be recruited to attain the pre-specified assumptions. The decision to stop recruitment was recommended by the Data Safety and Monitoring Committee (DSMC) after a limited recruitment in the last 24 months.
Stratification factors were center, HRT use (current, de novo), route of estrogen administration (transdermal, oral) and prior hysterectomy (unopposed estrogen, combined estrogen-progestin therapy). An ad hoc computer program in Access was used to preserve the allocation ratio in each stratum through biased coin randomization and minimization.
The data included are based on information received as of 30 November 2011. The main analysis was carried out on an intention-to-treat (ITT) basis. Incidence rates for breast cancer were calculated by dividing the number of events by the number of event-specific person-years of follow-up. For symptoms and adverse events, person-years were calculated from the date of randomization to treatment termination, development of the specific event or dropout, whichever came first.
For other cancers and deaths, person-years were calculated from the date of randomization to the date of last follow-up or specific event. Event rates were compared by using risk ratios (RRs) and 95% confidence intervals (CIs), assuming that the events followed a Poisson distribution. Because healthy, middle-aged women were enrolled in the trial, mortality was not a major competing risk of failure. The two treatment groups were compared by the log-rank test. HRs and 95% CIs were obtained using a Cox proportional regression model after adjustment for age (in 5-year groups) and center. We tested the proportional hazards assumption by introducing a time-dependent variable and testing for its statistical significance. For breast cancer subtypes, follow-up was censored at diagnosis of any other breast cancer subtype. We carried out sensitivity analyses adjusted for nonadherence by censoring follow-up 6 months after women became nonadherent, i.e. took treatment for less than 80% of the planned intervention, or to women who completed at least 12 months of treatment. P-values were at <0.05 level for the main end points and at <0.01 level for secondary end points and subgroup analyses to account for multiple comparisons. Analyses were carried out with SAS software (version 8.2, Cary, NC).
Methods
Subjects and Treatment
We conducted a phase-III, double-blind, placebo-controlled trial in recently postmenopausal women using HRT for menopausal symptom relief who were randomly allocated to either placebo or tamoxifen 5 mg/day for 5 years (trial IEO S51/200, HRT opposed by low-dose tamoxifen, HOT, Clinical Trials.gov NCT01579734). Eligibility criteria were current HRT use or de novo HRT use and negative mammography within 6 months. Toxicity was evaluated using the NCI toxicity criteria v2.0. Clinical examinations were repeated every 6 months and mammography was repeated annually. Transvaginal ultrasounds were carried out at baseline and repeated in case of atypical bleeding, followed by hysteroscopy on clinical judgment. At completion of the 5-year intervention, clinical visit and mammography were repeated annually up to 10 years. The use of venflaxine was allowed. The primary end point was the incidence of breast cancer, whereas the secondary end points included endometrial cancer, coronary heart syndrome, cerebrovascular events, venous thromboembolic events (VTEs), bone fractures, all cancers.
Pathology and Immunohistochemistry
Central pathology assessment of breast cancer included histological type, grade, peritumoral vascular invasion, ER, PgR, HER2 and Ki-67, as previously described. Molecular tumor subtypes were classified by immunohistochemistry (IHC) into four categories according to 2011 St Gallen criteria: luminal-A, ER or PgR-positive and Ki-67 < 14%; luminal-B, ER or PgR-positive and either Ki-67 ≥ 14% with negative HER2 or HER2-positive regardless of the Ki-67 level; HER2-positive and ER/PgR-negative; triple negative, all receptors negative.
Sample Size and Statistical Analysis
The study was designed to detect a 40% reduction of breast cancer with 80% power and two-sided 5% significance level. Using the log-rank test, 126 breast cancers had to be observed on 8500 subjects recruited over 5 years with 80% compliance and a background incidence rate of 4/1000. Recalculations based on a hazard ratio (HR) of 0.5, closer to the range observed in the Italian tamoxifen trial, and practical considerations based on the dramatic drop of HRT use after the first WHI trial results led to a total of 4500 women to be recruited to attain the pre-specified assumptions. The decision to stop recruitment was recommended by the Data Safety and Monitoring Committee (DSMC) after a limited recruitment in the last 24 months.
Stratification factors were center, HRT use (current, de novo), route of estrogen administration (transdermal, oral) and prior hysterectomy (unopposed estrogen, combined estrogen-progestin therapy). An ad hoc computer program in Access was used to preserve the allocation ratio in each stratum through biased coin randomization and minimization.
The data included are based on information received as of 30 November 2011. The main analysis was carried out on an intention-to-treat (ITT) basis. Incidence rates for breast cancer were calculated by dividing the number of events by the number of event-specific person-years of follow-up. For symptoms and adverse events, person-years were calculated from the date of randomization to treatment termination, development of the specific event or dropout, whichever came first.
For other cancers and deaths, person-years were calculated from the date of randomization to the date of last follow-up or specific event. Event rates were compared by using risk ratios (RRs) and 95% confidence intervals (CIs), assuming that the events followed a Poisson distribution. Because healthy, middle-aged women were enrolled in the trial, mortality was not a major competing risk of failure. The two treatment groups were compared by the log-rank test. HRs and 95% CIs were obtained using a Cox proportional regression model after adjustment for age (in 5-year groups) and center. We tested the proportional hazards assumption by introducing a time-dependent variable and testing for its statistical significance. For breast cancer subtypes, follow-up was censored at diagnosis of any other breast cancer subtype. We carried out sensitivity analyses adjusted for nonadherence by censoring follow-up 6 months after women became nonadherent, i.e. took treatment for less than 80% of the planned intervention, or to women who completed at least 12 months of treatment. P-values were at <0.05 level for the main end points and at <0.01 level for secondary end points and subgroup analyses to account for multiple comparisons. Analyses were carried out with SAS software (version 8.2, Cary, NC).
Source...