All trials showed statistically significant improvements in disease-free survival with relative risk reducations of about 15-40% for the use of aromatase inhibitors compared with tamoxifen either as intital therapy or after 2-3 years of tamoxifen; or around 46% for the use of an aromatase inhibitor compared with placebo after 5 years of tamoxifen. In general, similar improvements were seen in all components of disease-free survival - risk of local recurrence, distant recurrence or development of a contralateral primary tumor - although not every component of disease-free survival reached statistical significance on its own. With the short follow-up and the relatively food prognosis seen in most trials to date, absolute differences in disease-free survival are small (<4%) and absolutedifferences in distant disease-free survival are even smaller. However such differences would be expected to increase over time and to be larger in women with a poorer prognosis.
No trial has shown a statistically significant improvement in overall survival for the whole group randomised to an aromatase inhibitor. Of the trials comparing an aromatase inhibitor with tamoxifen, the trial with the most mature follow-up (ATAC) shows no difference in overall survival, while the other trials show non-significant trends favouring the aromatase inhibitor compared with tamoxifen or placebo.2 Reducing distant recurrences might be expected to translate into improved overall survival, but further follow-up is required to confirm this. Where an aromatase inhibitor was compared with placebo after 5 years of tamoxifen (MA.17), no overall survival benefit was seen in the group as a whole.7 However, a survival benefit was seen in the pre-specified subgroup with node-positive disease (p=0.04). Early reporting of the results and subsequent substantial crossover from placebo to aromatase inhibitor make it unlikely that a statistically significant survival difference will be seen in MA.17 overall.7
Aromatase inhibitors and tamoxifen were generally well tolerated in all six trials. Serious adverse events were rare and withdrawals due to adverse events were generally less frequent with aromatase inhibitors than with tamoxifen, and equally frequent with an aromatase inhibitor (letrozole) and placebo. Arthralgia, loss of bone mineral density and fractures were significantly more frequent with aromatase inhibitors compared with tamoxifen or placebo. Menopausal symptoms, venous thromboembolic events and endometrial cancer were significantly less frequent with aromatase inhibitors compared with tamoxifen. Longer follow-up is needed to determine whether differences in serum lipids and the small numbers of myocardial infarcts and strokes seen in these trials are attributable to aromatase inhibitors or tamoxifen or are merely due to chance.
Formal analyses of quality of life have been reported for two trials (ATAC and MA.17).2,7 Global assessments of quality of life were similar in women receiving anastrozole and tamoxifen in ATAC, and in women receiving letrozole and placebo in MA.17. Menopausal symptom scores were similar in women receiving anastrozole and tamoxifen in ATAC scores in both arms worsened initially and then improved over time. There were no significant differences in overall quality of life or the aggregate of endocrine symptoms in this trial. Women receiving anastrozole reported more vaginal dryness, painful intercourse, and loss of sexual interest, but fewer cold sweats and vaginal discharge than women receiving tamoxifen. In MA.17, letrozole was associated with slightly worse menopausal symptoms, sexual function, vitality, bodily pain, and physical function than placebo, but these differences affected only a small proportion of the women (6% or less).