When Are Aromatase Inhibitors Wrong for Breast Cancer Patients? Classified by F1000 as 'Changes Clinical Practice'

Elizabeth Bigger and Gretchen Kimmick


Faculty of 1000 

Amir E, Seruga B, Niraula S, Carlsson L, Ocaña A
J Natl Cancer Inst. 2011 Sep 7; 103(17):1299-309

Changes clinical practice: The risks and benefits of adjuvant aromatase inhibitors (AIs) should be weighed carefully, especially in postmenopausal women with early-stage, hormone-receptor positive breast cancer who also have coronary artery disease. Use of tamoxifen or changing to tamoxifen after 2-3 years of an AI may be preferable to 5 years of adjuvant therapy with an AI.

This meta-analysis confirms that an aromatase inhibitor (AI) is not the best therapy for all postmenopausal women with hormone-receptor positive, early-stage, breast cancer. [1] This potentially practice-changing article provides new evidence for AI toxicities, which practitioners should consider when choosing between adjuvant endocrine therapies for postmenopausal breast cancer patients. The authors conducted the study to clarify why AIs, when compared with tamoxifen, increased disease-free survival but not overall survival. AI toxicities were suspected to counteract decreased recurrence rates. The authors thus conducted a literature-based meta-analysis of randomized controlled trials to compare relative toxicities of AIs versus tamoxifen as primary adjuvant endocrine therapy in postmenopausal women with early-stage breast cancer.

Included studies were randomized phase III clinical trials comparing AIs with tamoxifen as initial adjuvant therapy and total treatment durations of 5 years. Data was included from all randomly assigned patients regardless of hormone receptor status, with an intention-to-treat analysis. The authors prespecified six adverse events: cardiovascular disease, cerebrovascular disease, venous thrombosis (DVT), bone fracture, endometrial cancer, and other secondary cancers. They also assessed hypercholesterolemia. All grades of these adverse events were recorded. Three treatment cohorts were defined for the meta-analysis: 1) 5 years of AI versus 5 years of tamoxifen (upfront AI versus tamoxifen), 2) 2-3 years of tamoxifen followed by 2-3 years of AI versus 5 years tamoxifen (switching versus tamoxifen), 3) 2-3 years of tamoxifen followed by 2-3 years of AI versus 5 years AI (switching versus AI). Odds ratios (OR) and 95% confidence intervals (CI) were calculated to compare the relative frequency of an individual adverse event. Difference in absolute risk between groups was also calculated by the number needed to harm.

Seven trials with 30,023 patients were included in the analysis. The authors found that a longer duration of AI use was associated with an increased risk of cardiovascular disease (OR 1.26, 95% CI 1.10 to 1.43, p<0.001, number needed to harm (NNH)=132) and bone fracture (OR 1.47, 95% CI 1.34 to 1.61, p<0.001, NNH=46). Longer AI exposure was also associated with decreased risk of venous thrombosis (OR 0.55, 95% CI 0.46 to 0.64, p<0.001, NNH=79) and endometrial carcinoma (OR 0.34, 95% CI 0.22 to 0.53, p<0.001, NNH=258). Five years of AI use was associated with an increased risk of death without recurrence, which was not statistically significant compared to 5 years of tamoxifen use or tamoxifen for 2-3 years followed by AI for 2-3 years (OR 1.11, 95% CI 0.98 to 1.26, p=0.9). There was no statistically significant difference in odds of cerebrovascular disease or secondary malignancies.

With regards to the cardiovascular disease risk, 4.2% of the patients in the AI group and 3.4% of patients in the tamoxifen group suffered a cardiovascular event (difference in absolute risk 0.8%, NNH=132). All treatment cohorts demonstrated increased odds of cardiovascular events in the AI group (of any AI exposure duration) versus the tamoxifen group, although the magnitude was lower for the switching cohort than for the 5 year AI cohort.

Four of the included studies formally assessed hypercholesterolemia, though they were not graded consistently. Pooled data analysis showed that a longer duration of AI use was associated with a statistically significant increase in odds of hypercholesterolemia compared with tamoxifen (OR 2.36, 95% CI 2.15 to 2.60, p<0.001). The difference was less pronounced for the switching cohort than for the upfront AI cohort.

The authors discussed the distinct toxicity profiles of AIs and tamoxifen, including relatively rare adverse events which were not detected in the analyses of individual trials, and cautioned that absolute risk (rather than relative risk) should direct clinical decision making. While the effect size was small for the increase in absolute risk of cardiovascular disease in the entire studied population, certain subpopulations may suffer higher risk. Previously the Food and Drug Administration (FDA) warned of an increased incidence of cardiovascular events for women with pre-existing heart disease for anastrozole (17%) compared with tamoxifen (10%). The data in the meta-analysis suggests that this increased cardiovascular risk may extend to all AIs. Trials of tamoxifen versus placebo also show that tamoxifen is associated with a reduction in cardiovascular events. Tamoxifen may thus have a protective effect against cardiovascular disease. Oncologists should weigh carefully the risk and benefits of AI use in patients with pre-existing heart disease, or other risk factors for heart disease. Data suggests that upfront AI use is associated with increased odds of death without breast cancer recurrence compared with the use of tamoxifen alone or switching from tamoxifen to AI. Data also suggests that switching after 2-3 years tamoxifen to an AI is associated with a reduced number of deaths compared with AI alone or tamoxifen alone. Switching may lead to a reduction in risk of cumulative toxicities.

Multiple factors should be considered when choosing endocrine therapy for postmenopausal women with early-stage breast cancer in light of this data. If the patient has history of ischemic heart disease and lower risk of breast cancer recurrence, she should not be treated with an AI. If the patient has a higher risk of breast cancer recurrence, the practitioner should consider switching to an AI only after 2-3 years tamoxifen. Alternatively, a patient with high risk of DVT should probably avoid tamoxifen. Tobacco-using patients may have increased risk of cardiovascular disease with AIs and DVTs with tamoxifen, which complicates the decision-making process.

The authors acknowledge the limitations of the study such as the data analysis of literature rather than individual patient data and combining data from trials of different follow-up durations. There is also inevitably variable quality of data and reporting of adverse events by investigators. Adverse events are typically only collected until the endpoints are reached (primary endpoint was breast cancer recurrence in these trials), though the incidence of adverse events after breast cancer recurrence still has clinical relevance given the long lifespan of many women with both locally recurrent and metastatic breast cancer. The data unfortunately did not allow for consideration of potentially confounding host factors or use of concurrent medications. The risk of cardiovascular disease may also increase with longer follow-up, particularly given the large increase in hypercholesterolemia with AI use.

The current National Comprehensive Cancer Network (NCCN) guidelines for endocrine therapy for postmenopausal women with early-stage breast cancer recommend equally the options of AI for 5 years, tamoxifen for 2-3 years followed by AI to complete 5 years, and tamoxifen for 4.5 to 6 years followed by AI for 5 years.[2] They also offer the option of tamoxifen for 5 years if there is a contraindication for or intolerance of AIs. However, the guidelines do not mention the consideration of prior cardiovascular disease history, cardiovascular disease risk, or recurrence risk when choosing between these different therapeutic options. In their discussion of adjuvant endocrine therapy, they mention that AIs are associated with "musculoskeletal symptoms, osteoporosis, and increased risk of bone fracture, while tamoxifen is associated with an increased risk of uterine cancer and deep venous thrombosis". The current version of the breast-cancer treatment guidelines recommend that "postmenopausal women with early breast cancer receive an AI as initial adjuvant therapy, sequential with tamoxifen, or as extended therapy in those situations where endocrine therapy is to be utilized… in postmenopausal women, the use of tamoxifen alone for five years is limited to those who decline or who have a contraindication to AIs".

The meta-analysis included all grades of cardiac toxicity, but there is also concern that AIs increase high-grade adverse cardiac events compared to tamoxifen. In the Breast International Group (BIG) 1-98 trial, the overall incidence of cardiac events was similar (letrozole 4.8% versus tamoxifen 4.7%) but the incidence of grade 3 to 5 cardiac adverse events was significantly higher in the letrozole arm.[3] Recently published updated results for the BIG 1-98 treatment cohorts, with a follow-up of 12 years since randomization, now reveals a statistically significant difference in overall survival for the letrozole monotherapy arm compared with the tamoxifen monotherapy arm (hazard ratio [HR] 0.87, 95% CI 0.777 to 0.999, p=0.048) in the intention to treat analysis.[4] In their analysis by inverse probability of censoring weighted (IPCW) Cox models (to create a scenario of informative missing data), the node negative patients had a HR of 0.82 (95% CI 0.65 to 1.02), and node positive patients had a HR of 0.74 (0.63 to 0.88). There was no significant difference in disease-free survival or overall survival for switching cohorts (tamoxifen followed by letrozole, or letrozole followed by tamoxifen) compared to letrozole alone.

Another important consideration when choosing between endocrine therapies is the age of the patient. A retrospective cohort study of women age 66 and older at the time of breast-cancer diagnosis found that more patients died of cardiovascular disease than of breast cancer (cardiovascular disease deaths 15.9% [95% CI 15.6 to 16.2] followed by breast cancer deaths 15.1% [95% CI 14.8 to 15.4]). The fully adjusted relative hazard of cardiovascular disease in this group of breast cancer patients was 1.24 (95% confidence interval [95% CI] 1.13 to 1.26).[5]

In light of the meta-analysis results, the guidelines should also include cardiovascular risk factor screening and consider alternatives to AIs for women with a cardiovascular disease history, who are at high risk for developing cardiovascular disease, while also having a low risk of recurrence. For patients at higher risk of recurrence who may benefit more from AI therapy, a switching strategy combining tamoxifen and an AI would likely be the best choice. Patients at risk for coronary artery disease events should undergo risk-factor modification, particularly of hypercholesterolemia in the setting of AI use.[6]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.