Toxicity of Extended Adjuvant Therapy With Aromatase Inhibitors in Early Breast Cancer: A Systematic Review and Meta-analysis

Toxicity of Extended Adjuvant Therapy With Aromatase Inhibitors in Early Breast Cancer: A... Abstract Background A number of randomized controlled trials (RCTs) have reported improvement in breast cancer outcomes from extending treatment with aromatase inhibitors (AIs) beyond the initial five years after diagnosis. However, the toxicity profile of extended AIs is uncertain. Methods We identified RCTs that compared extended AIs to placebo or no treatment using MEDLINE and a review of abstracts from key conferences between 2013 and 2016. Odds ratios (ORs), 95% confidence intervals (CIs), absolute risks, and the number needed to harm (NNH) were computed for prespecified safety and tolerability outcomes including cardiovascular events, bone fractures, second cancers (excluding new breast cancer), treatment discontinuation for adverse events, and death without recurrence. All statistical tests were two-sided. Results Seven trials comprising 16 349 patients met the inclusion criteria. Longer treatment with AIs was associated with increased odds of cardiovascular events (OR = 1.18, 95% CI = 1.00 to 1.40, P = .05, NNH = 122), bone fractures (OR = 1.34, 95% CI = 1.16 to 1.55, P < .001, NNH = 72), and treatment discontinuation for adverse events (OR = 1.45, 95% CI = 1.25 to 1.68, P < .001, NNH = 21). Longer treatment with AIs did not influence the odds of either second malignancy (OR = 0.93, 95% CI = 0.73 to 1.18, P = .56) or deaths without breast cancer recurrence (OR = 1.11, 95% CI = 0.90 to 1.36, P = .34). Conclusions Extended treatment with AIs is associated with an increased risk of cardiovascular events and bone fractures. There is no statistically significant increase in deaths without breast cancer recurrence among patients receiving longer treatment with AIs. These data should be taken into account when considering extended adjuvant AIs. The use of aromatase inhibitors (AIs) as adjuvant treatment of postmenopausal women with breast cancer is well established (1). Different schedules of treatment, including upfront AIs or sequential use of tamoxifen and AIs, have been shown to improve disease-free survival and overall survival compared with tamoxifen alone (1–6). While five years of adjuvant hormonal treatment has been the gold standard for many years (7,8), there is a growing body of evidence supporting the benefit of extended endocrine treatment both with tamoxifen (9,10) and AIs (11–13) beyond the initial five years. Treatment with AIs is associated with well-described adverse events (14–17). When used in the five years after initial diagnosis, longer exposure to AIs is associated with more cardiovascular events and bone fractures when compared with tamoxifen (15). Data regarding the toxicity profile of extended treatment with AIs when compared with no therapy are more limited. The impact of safety and tolerability is likely of greater importance for extended adjuvant therapy compared with treatment in the initial five years because the absolute risk of breast cancer recurrence falls with time (18) while with increasing age, the baseline risk of morbidity and mortality from other illnesses rises (19). As such, robust data on the toxicity of extended adjuvant therapy are desirable in order to appropriately counsel patients and provide optimal recommendations on continuation of adjuvant therapy beyond five years. Here, we report on a meta-analysis evaluating adverse events reported in phase III randomized controlled trials (RCTs) comparing extended treatment with AIs to placebo or no treatment in postmenopausal women with hormone receptor–positive early breast cancer. Methods Literature Review and Study Identification A literature search utilizing MEDLINE (host: OVID; 1996–December week 3, 2016) and databases for the Annual Meetings of the American Society of Clinical Oncology and the San Antonio Breast Cancer Symposium between 2013 and 2016 was undertaken. The search was restricted to English language articles. The terms “adjuvant,” “aromatase inhibitor,” “tamoxifen,” and “breast cancer” and similar terms were cross-searched by using the following search algorithm: ((aromatase inhibitor OR anastrozole OR letrozole OR exemestane) AND (tamoxifen) AND (adjuvant) AND (breast neoplasm MeSH OR ((breast OR mammary) AND (carcinoma OR malignan* OR neoplasm OR tumor)))) AND (randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized controlled trial [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR (“clinical trial”) [tw] OR singl* [tw] OR doubl* [tw] OR trebl* [tw] OR tripl* [tw] AND (mask* [tw] OR blind* [tw])) OR comparative study [mh] OR evaluation studies [mh] OR follow-up studies [mh] OR prospective studies [mh] OR control* [tw] OR prospective* [tw] OR volunteer* [tw] NOT (animals [mh] NOT humans [mh]). Eligible trials included all phase III RCTs comparing longer duration of AI therapy with placebo or no treatment following initial adjuvant endocrine therapy with either tamoxifen, AIs, or both in postmenopausal women. We utilized a hierarchal primary data source. Data from safety analyses were preferred, and, when not available, data from intention-to-treat analyses were utilized instead. Data Extraction Data on potentially life-threatening adverse events of any grade as reported in individual trials were extracted. Collected data included the number of patients with the following prespecified adverse events: cardiovascular events, bone fractures, and secondary malignancies (excluding second primary breast cancers), as well as the total number of patients at risk of each adverse event. These events were chosen because they are associated with substantial morbidity and mortality, are commonly reported in RCTs, and are less prone to reporting bias. Data on symptomatic adverse events unlikely to be associated with mortality, such as arthralgia and genitourinary atrophy, were not considered. Additionally, data regarding treatment discontinuation for adverse events, as well as deaths without breast cancer recurrence, were collected. Finally, we collected data reporting both disease-free and overall survival from each study. Data were extracted independently by two reviewers (HG and EA). Discrepancies were resolved by consensus. All data were extracted from primary publications, their associated online appendices, or from the conference presentation if the study was not published in full. Data Synthesis and Statistical Analysis The primary analysis compared the odds of adverse events between patients who were randomly assigned to prolonged treatment with AIs and those randomly assigned to placebo or no treatment. Data from individual studies were pooled in a meta-analysis. The odds ratios (ORs) and associated 95% confidence intervals (CIs) were computed for all adverse events and were then pooled in a meta-analysis using RevMan 5.3 (The Cochrane Collaboration, Copenhagen, Denmark). Pooled estimates of odds ratio were computed using the Mantel-Haenszel odds ratio method (21). Due to substantial clinical heterogeneity between studies in the time from diagnosis to random assignment and in exposure to prior treatments, analyses were performed using random effects modeling irrespective of the statistical heterogeneity. Statistical heterogeneity was reported using Cochran Q (which assesses whether observed differences in results are compatible with chance alone) and I2 statistics (which describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error). Statistically significant heterogeneity was defined as a Cochran Q P value of less than .10 or I2 greater than 50%. Absolute risks were calculated as the number of events per person over the follow-up period of individual trials. The difference in absolute risk between the prolonged treatment and control groups was also presented as the number needed to harm (NNH), which quantifies the number of patients who would need to be treated with longer duration of AIs to cause an adverse event in one patient who would not otherwise have experienced the adverse event. Meta-regression was used to explore the influence of median age at random assignment, duration of follow-up, and prior treatments on the odds ratio for each adverse event. Meta-regression comprised a univariable linear regression weighted by individual study sample size exploring the association between median age, duration of follow-up, and proportion of patients treated with prior tamoxifen, AIs, or cytotoxic chemotherapy on the log of the odds ratio for each adverse event. Meta-regression was performed using SPSS version 21 (IBM Corp, Armonk, NY) using the weighted least squares (mixed effect) function (44). Finally, in an exploratory analysis, we pooled the disease-free and overall survival from each trial using inverse variance and random effects modeling. Statistical tests were two-sided, and statistical significance was defined as a P value of less than .05. Results Included Studies Our search identified 1546 studies initially. After exclusions, we identified 14 individual RCTs evaluating adjuvant endocrine monotherapy. Of these, seven trials explored initial adjuvant therapy. Therefore, seven studies were eligible for analysis (see Figure 1). Eligible studies comprised 16 349 patients. Individual study characteristics are shown in Table 1 while an individual study schema is shown in Supplementary Figure 1 (available online). Four trials (MA.17R [12], MA17 [13], NSABP B33 [23], and NSABP B42 [24]) compared five years of extended AIs with placebo. One trial (ABCSG 6a [11]) compared three years of extended anastrozole with no treatment, and two trials (DATA [25] and IDEAL [26]) compared longer with shorter duration of AIs. In three of these trials (13,23,25), eligibility required no prior exposure to AIs; however, in one trial (ABCSG 6a [11]) patients were not exposed to third-generation AIs, but approximately half of the cohort had been treated with aminoglutethimide, a first-generation AI. Adverse events were defined differently in each trial (see Table 2). Table 1. Characteristics of included studies Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  * Tumor size: T1 ≤2 cm, T2 >2 cm ≤5 cm, T3 >5 cm. AI = aromatase inhibitor; NR = not reported. † Study required early accrual termination and unbinding in view of the MA.17 (13) results in October 2003. ‡ 99% received prior AIs for 4.5 to six years. § Median age not reported, but estimated from reported age ranges. Table 1. Characteristics of included studies Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  * Tumor size: T1 ≤2 cm, T2 >2 cm ≤5 cm, T3 >5 cm. AI = aromatase inhibitor; NR = not reported. † Study required early accrual termination and unbinding in view of the MA.17 (13) results in October 2003. ‡ 99% received prior AIs for 4.5 to six years. § Median age not reported, but estimated from reported age ranges. Table 2. Definition of cardiac and bone adverse events in the included trials Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Table 2. Definition of cardiac and bone adverse events in the included trials Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Figure 1. View largeDownload slide Study selection. Figure 1. View largeDownload slide Study selection. Cardiovascular Events Six studies reported data on cardiovascular events (11–13,24–26). Prolonged therapy with AIs was associated with a statistically significant increase in odds of cardiovascular events (OR = 1.18, 95% CI =  1.00 to 1.40, P = .05) (see Figure 2A). There was no evidence of statistically significant heterogeneity (Cochran Q P = .21, I2 = 30%). In absolute terms, 7.0% of patients in the prolonged AIs group had cardiovascular events, compared with 6.0% of patients in the control group. The weighted pooled absolute difference was 0.8%, with a NNH of 122 (see Table 3). Results of meta-regression are shown in Table 4. There was no association between the relative odds of cardiovascular events and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen or prior AIs. Table 3. Pooled absolute risk of adverse events* Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  * Positive values indicate excess events with prolonged aromatase inhibitors; negative values indicate excess events with placebo/no treatment. AE = adverse event; NNH = number needed to harm; NR = not reported. Table 3. Pooled absolute risk of adverse events* Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  * Positive values indicate excess events with prolonged aromatase inhibitors; negative values indicate excess events with placebo/no treatment. AE = adverse event; NNH = number needed to harm; NR = not reported. Table 4. Results of meta-regression exploring the association of adverse events with age, duration of follow-up, and proportion of patients with prior treatments Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      * All P values are two-sided. AI = aromatase inhibitor. † Cannot estimate because only two studies reported data (13,23). Table 4. Results of meta-regression exploring the association of adverse events with age, duration of follow-up, and proportion of patients with prior treatments Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      * All P values are two-sided. AI = aromatase inhibitor. † Cannot estimate because only two studies reported data (13,23). Figure 2. View largeDownload slide Forest plots for adverse events. A) Cardiovascular events. B) Bone fractures. C) Other second cancers. D) Discontinuation for adverse events. E) Death without breast cancer recurrence. Odds ratios for each trial are represented by squares, the size of the square represents the weight of the trial in the meta-analysis, and the horizontal line crossing the square represents the 95% confidence interval. The diamonds represent the estimated pooled effect. Test for overall effect based on z-test. All P values are two-sided. CI = confidence interval; OR = odds ratio. Figure 2. View largeDownload slide Forest plots for adverse events. A) Cardiovascular events. B) Bone fractures. C) Other second cancers. D) Discontinuation for adverse events. E) Death without breast cancer recurrence. Odds ratios for each trial are represented by squares, the size of the square represents the weight of the trial in the meta-analysis, and the horizontal line crossing the square represents the 95% confidence interval. The diamonds represent the estimated pooled effect. Test for overall effect based on z-test. All P values are two-sided. CI = confidence interval; OR = odds ratio. Hypertension was reported in only three studies (12,13,26). There was no association between longer duration of AIs and increased odds of hypertension (OR = 1.02, 95% CI =  0.86 to 1.21, P = .82; data not shown). There was no evidence of statistically significant heterogeneity (Cochran Q P = .42, I2 = 0%). In absolute terms, hypertension was reported in 7.3% of patients receiving longer treatment with AIs and 7.1% of patients in the control group. The weighted pooled absolute difference was 0.1%, with a NNH of 770. Bone Fractures All seven studies reported data on bone fractures (11–13,23–26). Prolonged therapy with AIs was associated with statistically significantly increased odds of bone fractures (OR = 1.34, 95% CI =  1.16 to 1.55, P < .001) (see Figure 2B). There was no evidence of statistically significant heterogeneity (Cochran Q P = .38, I2 = 6%). In absolute terms, fractures occurred in 6.3% of the prolonged AI group and 4.8% in the control group, with a weighted pooled absolute difference of 1.4% and a NNH of 72 (see Table 3). Results of the meta-regression are shown in Table 4. There was no association between the relative odds of bone fracture and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen, prior AIs, or prior chemotherapy. Second Cancers Four trials reported the occurrence of second malignancies excluding breast cancer, three studies reported on all second malignancies (23–25), and one study reported only on endometrial cancer (13). Pooled data showed that prolonged treatment with AIs was not associated with second cancers (OR = 0.93, 95% CI =  0.73 to 1.18, P = .56) (see Figure 2C). There was no evidence of statistically significant heterogeneity (Cochran Q P = .26, I2 = 25%). Exclusion of the study reporting only endometrial carcinoma did not change the estimate (OR = 0.94, 95% CI =  0.58 to 1.52, P = .81), and heterogeneity remained statistically nonsignificant (Cochran Q P = .54, I2 = 0%). The absolute risk of non–breast cancer secondary malignancies was 2.2% for prolonged treatment with AIs and 2.4% for the control group, a weighted pooled difference of –0.2% and a NNH of –625. Results of the meta-regression are shown in Table 4. There was no association between the relative odds of second cancers and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen, prior AIs, or prior chemotherapy. Discontinuation for Adverse Events Discontinuation of treatment for adverse events was reported in six studies (11–13,24–26). Pooled data showed that longer duration of treatment with AIs was associated with a 45% increased odds for treatment discontinuation for adverse events compared with placebo or no treatment (OR = 1.45, 95% CI = 1.25 to 1.68, P < .001) (see Figure 2D). There was evidence of statistically significant heterogeneity (Cochran Q P = .07, I2 = 51%). In absolute terms, discontinuation for adverse events occurred in 17.0% of patients in the prolonged treatment group and 13.4% in the control group. The weighted pooled absolute difference was 4.8%, with a NNH of 21. There was no association between the relative odds of discontinuation for adverse events and median age at random assignment, median duration of follow-up, or proportion of patients who had received prior tamoxifen or prior AIs (see Table 4). Death Without Recurrence Death without breast cancer recurrence was reported in five studies (11–13,22,23). There was no statistically significant association between death without breast cancer recurrence and longer use of AIs (OR = 1.11, 95% CI =  0.90 to 1.36, P = .34) (see Figure 2E). There was no evidence of statistically significant heterogeneity (Cochran Q P = .44, I2 = 0%). In absolute terms, 3.3% of the patients treated with prolonged AIs and 2.9% of those treated with placebo or no treatment died without breast cancer recurrence. The weighted pooled absolute difference was 0.3%, with a NNH of 371. There was no association between the relative odds of death without breast cancer recurrence and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen or prior AIs (see Table 4). In an exploratory analysis, despite improvement in disease-free survival (pooled OR = 0.78, 95% CI =  0.68 to 0.88, P< .001) (see Supplementary Figure 2, available online), no effect on overall survival was observed (OR = 1.01, 95% CI =  0.89 to 1.14, P = .92) (see Supplementary Figure 2, available online). Discussion Late relapse, which can occur decades after initial diagnosis, is a hallmark of hormone receptor–positive early breast cancer (1). Consequently, there has been interest in extended adjuvant therapy in such patients. The net benefit of any therapy is comprised of the balance between benefits and risks. In the setting of extended adjuvant endocrine therapy, an accurate assessment of this balance is essential because over time the risk of breast cancer recurrence and death falls and the contribution of concurrent illnesses to morbidity and mortality increases (19). As such, knowledge of the potential impact of extended anticancer therapy on comorbidities is important. Typically, individual trials of adjuvant endocrine therapy are designed to evaluate efficacy. Most have limited power to explore toxicities, especially those that are less common. As a result, pooling of data from multiple trials is valuable to evaluate accurately the effect of adjuvant therapy on safety and tolerability. An example of this was the observation of increased cardiovascular disease with AIs relative to tamoxifen as initial adjuvant therapy. No effect was seen in individual trials, but a modest increase in cardiovascular disease was observed in a meta-analysis of randomized trials (15). In contrast, population-based studies that utilized different definitions of cardiovascular events have reported conflicting data (19,28). Most of the current knowledge on the safety and tolerability of AIs is derived from data that compared AIs with tamoxifen. Consequently, the influence of extended AIs on toxicity when compared with no treatment is uncertain. There has been increasing interest in the potential interaction of AIs on cardiovascular disease. Cardiovascular disease and breast cancer share a number of risk factors (29,30), and, thanks to effective breast cancer therapy, the most common cause of death among women with early breast cancer is cardiovascular disease (31,32). Data from trials of initial adjuvant therapy have shown that longer duration of AIs is associated with increased cardiovascular events (15). A key limitation of these data is that in all trials the comparison group was tamoxifen and it was not possible to determine if the relative increase in the odds of cardiovascular events was related to a real increase in risk with AIs or some protection from tamoxifen (33). Recent data focusing on studies of AIs compared with placebo or no treatment suggested no increase in risk of cardiovascular events with AIs (34). However, this study did not include a number of the large randomized trials reported recently. In the current study, we observed a statistically significant increase in the odds of cardiovascular events with longer durations of AIs. The relative increase was modestly lower in magnitude to that observed in the initial adjuvant setting (OR = 1.18 vs 1.26, respectively); however, because of higher event rates in this older population of patients with lower competing risks of cancer death, this relative increase translated to a similar absolute risk (NNH = 122 vs 132, respectively). There are concerns that safety and tolerability data from clinical trials underestimate the effect of toxicity when treatment is administered to patients outside of clinical trials. Data show that many patients in routine practice do not meet the eligibility criteria of clinical trials (35). Eligibility criteria of randomized trials often exclude patients with preexisting cardiovascular disease (36), a group at higher risk of treatment-related toxicity. A population-based observation study focusing on cardiovascular events rather than a composite of both cardiovascular and cerebrovascular events reported that, compared with tamoxifen, AIs are associated with a doubling in the hazard of serious cardiovascular events (37), a magnitude of effect much greater than observed in clinical trials. The mechanism of effect of AIs on cardiovascular morbidity is not clear. AIs have been associated with an increase in cardiovascular risk factors such as dyslipidemia and arterial hypertension (15,38,39). In the current study, we observed a similar risk of incident arterial hypertension among those receiving longer therapy with AIs and control groups. This suggests that the link between AIs and cardiovascular events may relate to mechanisms independent of arterial hypertension such as accelerated atherosclerosis, an observation reported elsewhere (40). In postmenopausal women, AIs cause accelerated bone loss (1). Several trials of initial adjuvant therapy showed a statistically significantly higher risk of bone fractures in patients treated with AIs compared with tamoxifen (4,41–43). Again, this analysis was potentially confounded by a protective effect of tamoxifen. In the current study, the detrimental effect on bone health was confirmed, with a 34% increased odds of developing bone fractures with AIs compared with placebo or no treatment. This increase translated to a NNH of 72, modestly higher than data observed in the initial adjuvant setting (NNH = 46) (15). Potential reasons for the lower magnitude of absolute risk of bone fractures with extended AI therapy compared with initial adjuvant therapy are the protective effect of tamoxifen in the initial adjuvant setting, improved screening of patients treated with AIs with bone mineral density scans, and the treatment of those at risk with antiresorptive therapies. The impact of treatment-related bone loss may be attenuated in the future with the increasing use of adjuvant bone-targeted therapy, even in those without preexisting or treatment-related bone loss (44). Statistically significantly more patients randomly assigned to prolonged treatment with AIs discontinued therapy for adverse events. Pooled data showed that for every 21 patients assigned to longer therapy with AIs, one patient discontinued treatment due to adverse events. This finding is notable because all enrolled participants had received three to 10 years of prior endocrine therapy prior to random assignment, suggesting that serious toxicity necessitating discontinuation of therapy can occur even at a late stage of treatment. Extended adjuvant AIs are associated with a 20% to 25% relative reduction in disease recurrence, although in many cases such events are comprised of local or contralateral in-breast recurrences. In our analysis, we observed a numerical excess of deaths without breast cancer recurrence with prolonged AI therapy; however, this association was not statistically significant. This finding may explain why there is no apparent effect of extended therapy with AIs on overall survival. The increased number of noncancer deaths with longer treatment with AIs attenuates the contribution of reduced breast cancer deaths to overall survival. It is important to stress that overall survival data were immature in many studies and it is possible that the long-term effect of exposure to AIs may result in the emergence of both a survival advantage and additional adverse events with longer durations of follow-up. Of interest, we did not identify any factors that influenced the effect of prolonged treatment with AIs on safety and tolerability. Similar magnitudes of relative odds were observed for all adverse events of interest, irrespective of age, duration of follow-up, or prior treatments (including tamoxifen, AIs, and chemotherapy). These findings do not exclude an effect on absolute risk. Cardiovascular events and bone fractures increase with age. As such, despite no effect on the relative risk of these events between prolonged treatment with AIs and control, it is anticipated that older patients and those with more comorbidities with longer durations of AIs will have a higher absolute risk of adverse events than those treated with placebo or no treatment. This study has limitations. First, this is a literature-based rather than an individual patient–based meta-analysis. Consequently, data on comorbidities and concurrent medications were not available and could not be adjusted for in the analysis. Additionally, because data were extracted from reports of trials with different durations of follow-up, we were unable to report actuarial rates of toxicity. Second, the reporting and definition of toxicity was variable among the different trials and may lead to some uncertainty about estimates of odds of adverse events. Furthermore, one study (11) was open label rather than placebo controlled and therefore subject to ascertainment bias. Third, although the grade of the toxicity might clarify the severity of the adverse event on the patient, grading of adverse events was not reported uniformly, and therefore our analysis included all grades of toxicity. Fourth, inclusion criteria for each trial were different, especially with regard to prior adjuvant endocrine therapy. Some patients received only two to three years of prior tamoxifen (24), while others received up to 10 years of endocrine therapy with both tamoxifen and AIs (12), making the included patients in this meta-analysis heterogeneous. Finally, toxicity reporting is usually collected only until an efficacy event of interest occurs. As the majority of women with hormone receptor–positive breast cancer survive for several years even with metastatic disease, the impact of adverse events after recurrence remains of interest. In conclusion, extended treatment with AIs is associated with an increased risk of cardiovascular events and bone fractures. There is no statistically significant excess of deaths without breast cancer recurrence among patients receiving longer durations of AIs. These data should be taken into account when considering extended adjuvant AIs for breast cancer patients, especially patients with preexisting comorbidities or risk factors thereof. Funding None for all authors. References 1 Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Aromatase inhibitors versus tamoxifen in early breast cancer: Patient-level meta-analysis of the randomised trials. Lancet . 2015; 386( 10001): 1341– 1352. CrossRef Search ADS PubMed  2 Boccardo F, Rubagotti A, Guglielmini P, et al.   Switching to anastrozole versus continued tamoxifen treatment of early breast cancer. Updated results of the Italian tamoxifen anastrozole (ITA) trial. Ann Oncol.  2006; 17( suppl 7): vii10– vii14. Google Scholar CrossRef Search ADS PubMed  3 Coates AS, Keshaviah A, Thurlimann B, et al.   Five years of letrozole compared with tamoxifen as initial adjuvant therapy for postmenopausal women with endocrine-responsive early breast cancer: Update of study BIG 1-98. J Clin Oncol.  2007; 25( 5): 486– 492. Google Scholar CrossRef Search ADS PubMed  4 Jakesz R, Jonat W, Gnant M, et al.   Switching of postmenopausal women with endocrine-responsive early breast cancer to anastrozole after 2 years' adjuvant tamoxifen: Combined results of ABCSG trial 8 and ARNO 95 trial. Lancet.  2005; 366( 9484): 455– 462. Google Scholar CrossRef Search ADS PubMed  5 Forbes JF, Cuzick J, Buzdar A, Howell A, Tobias JS, Baum M. Effect of anastrozole and tamoxifen as adjuvant treatment for early-stage breast cancer: 100-month analysis of the ATAC trial. Lancet Oncol.  2008; 9( 1): 45– 53. Google Scholar CrossRef Search ADS PubMed  6 Coombes RC, Kilburn LS, Snowdon CF, et al.   Survival and safety of exemestane versus tamoxifen after 2-3 years’ tamoxifen treatment (Intergroup Exemestane Study): A randomised controlled trial. Lancet . 2007;17; 369( 9561): 559– 570. Google Scholar CrossRef Search ADS PubMed  7 Early Breast Cancer Trialists Collaborative Group (EBCTCG). Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: Patient-level meta-analysis of randomised trials.  Lancet.  2011; 378( 9793): 771– 784. CrossRef Search ADS PubMed  8 Fisher B, Dignam J, Bryant J, Wolmark N. Five versus more than five years of tamoxifen for lymph node-negative breast cancer: Updated findings from the National Surgical Adjuvant Breast and Bowel Project B-14 randomized trial. J Natl Cancer Inst.   2001; 93( 9): 684– 690. Google Scholar CrossRef Search ADS PubMed  9 Davies C, Pan H, Godwin J, et al.   Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial. Lancet.  2013; 381( 9869): 805– 816. Google Scholar CrossRef Search ADS PubMed  10 Gray RG, Rea D, Handley K, et al.   aTTom: Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years in 6953 women with early breast cancer. J Clin Oncol.  2013; 31( 18 suppl): 5. Google Scholar CrossRef Search ADS PubMed  11 Jakesz R, Greil R, Gnant M, et al.   Extended adjuvant therapy with anastrozole among postmenopausal breast cancer patients: Results from the randomized Austrian Breast and Colorectal Cancer Study Group Trial 6a. J Natl Cancer Inst . 2007; 99( 24): 1845– 1853. Google Scholar CrossRef Search ADS PubMed  12 Goss PE, Ingle JN, Pritchard KI, et al.   Extending aromatase inhibitors adjuvant therapy to 10 years. N Engl J Med.  2016; 375( 3): 209– 219. Google Scholar CrossRef Search ADS PubMed  13 Goss PE, Ingle JN, Martino S, et al.   Randomized trial of letrozole following tamoxifen as extended adjuvant therapy in receptor-positive breast cancer: Updated findings from NCIC CTG MA.17. J Natl Cancer Inst.  2005; 97( 17): 1262– 1271. Google Scholar PubMed  14 Seruga B, Tannock IF. Up-front use of aromatase inhibitors as adjuvant therapy for breast cancer: The emperor has no clothes. J Clin Oncol.  2009; 27( 6): 840– 842. Google Scholar CrossRef Search ADS PubMed  15 Amir E, Seruga B, Niraula S, Carlsson L, Ocaña A. Toxicity of adjuvant endocrine therapy in postmenopausal breast cancer patients: A systematic review and meta-analysis. J Natl Cancer Inst.  2011; 103( 17): 1299– 1309. Google Scholar PubMed  16 Haque R, Shi J, Schottinger JE, et al.   cardiovascular disease after aromatase inhibitor use. JAMA Oncol.  2016; 2( 12): 1590– 1597. Google Scholar CrossRef Search ADS PubMed  17 Rydén L, Heibert Arnlind M, Vitols S, Höistad M, Ahlgren J. Aromatase inhibitors alone or sequentially combined with tamoxifen in postmenopausal early breast cancer compared with tamoxifen or placebo – meta-analyses on efficacy and adverse events based on randomized clinical trials. Breast.  2016; 26: 106– 114. Google Scholar CrossRef Search ADS PubMed  18 Early Breast Cancer Trialists' Collaborative Group. Tamoxifen for early breast cancer: An overview of the randomised trials. Lancet . 1998; 351( 9114): 1451– 1467. CrossRef Search ADS PubMed  19 Abdel-Qadir H, Austin PC, Lee DS, et al.   A population-based study of cardiovascular mortality following early-stage breast cancer. JAMA Cardiol . 2017; 2( 1): 88– 93. Google Scholar CrossRef Search ADS PubMed  20 Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med.  2004; 23( 9): 1351– 1375. Google Scholar CrossRef Search ADS PubMed  21 Higgins JPT, Deeks JJ, Altman DG. Metaanalysis of rare events. In: Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions , version 5.1.0. The Cochrane Collaboration; 2011. http://handbook.cochrane.org. Accessed January 31, 2017. 22 Stanley TD, Doucouliagos H. Neither fixed nor random: Weighted least squares meta-regression. Res Synth Methods.  2017; 8( 1): 19– 42. Google Scholar CrossRef Search ADS PubMed  23 Mamounas EP, Jeong JH, Wickerham DL, et al.   Benefit from exemestane as extended adjuvant therapy after 5 years of adjuvant tamoxifen: Intention-to-treat analysis of the National Surgical Adjuvant Breast and Bowel Project B-33 trial. J Clin Oncol.   2008; 26( 12): 1965– 1971. Google Scholar CrossRef Search ADS PubMed  24 Mamounas EP, Bandos Hanna, Lembersky BC, et al.   A randomized, double-blinded, placebo-controlled clinical trial to evaluate extended adjuvant endocrine therapy (5 years of letrozole) in postmenopausal women with hormone-receptor positive breast cancer who have completed previous adjuvant endocrine therapy: Initial results of NRG oncology/NSABP B-42. Cancer Res . 2017; 77( Suppl 4): S1– 05. 25 Tjan-Heijnen VC, Van Hellemond IE, Peer PG, et al.   First results from the multicenter phase III DATA study comparing 3 versus 6 of anastrozole after 2-3 years of tamoxifen in postmenopausal women with hormone receptor-positive early breast cancer. Cancer Res . 2017; 77( Suppl 4): S1– 03. 26 Blok EJ, van de Velde CJH, Meershoek-Klein Kranenbarg EM, et al.   Safety assessment of extended adjuvant endocrine therapy with letrozole; results of the randomized phase III IDEAL trial (BOOG 2006-05). Cancer Res . 2017; 77( Suppl 4): S1– 04. 27 Fontein DB, Nortier JW, Liefers GJ, et al.   High non-compliance in the use of letrozole after 2.5 years of extended adjuvant endocrine therapy. Results from the IDEAL randomized trial. Eur J Surg Oncol.   2012; 38( 2): 110– 117. Google Scholar CrossRef Search ADS PubMed  28 Haque R, Shi J, Schottinger JE, Chung J, et al.   cardiovascular disease after aromatase inhibitor use. JAMA Oncol.  2016; 2( 12): 1590– 1597. Google Scholar CrossRef Search ADS PubMed  29 Key TJ, Appleby PN, Reeves GK. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst.  2003; 95( 16): 1218e26. 30 Calle EE, Kaaks R. Overweight, obesity and cancer: Epidemiological evidence and proposed mechanisms. Nat Rev Cancer.  2004; 4( 8): 579e91. Google Scholar CrossRef Search ADS   31 Patnaik JL, Byers T, DiGuiseppi C, Dabelea D, Denberg TD. Cardiovascular disease competes with breast cancer as the leading cause of death for older females diagnosed with breast cancer: A retrospective cohort study. Breast Cancer Res.  2011; 13( 3): R64. Google Scholar CrossRef Search ADS PubMed  32 Yancik R, Wesley MN, Ries LA, Havlik RJ, Edwards BK, Yates JW. Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. JAMA.  2001; 285( 7): 885e92. Google Scholar CrossRef Search ADS   33 Love RR, Wiebe DA, Newcomb PA, et al.   Effects of tamoxifen on cardiovascular risk factors in postmenopausal women. Ann Intern Med.  1991; 115( 11): 860– 864. Google Scholar CrossRef Search ADS PubMed  34 Khosrow-Khavar F, Filion KB, Al-Qurashi S, et al.   Cardiotoxicity of aromatase inhibitors and tamoxifen in post-menopausal women with breast cancer: A systematic review and meta-analysis of randomized controlled trials. Ann Oncol .  2016; 28( 3): 487– 496. 35 Treweek S, Dryden R, McCowan C, et al.   Do participants in adjuvant breast cancer trials reflect the breast cancer patient population? Eur J Cancer.  2015; 51: 907– 914. Google Scholar CrossRef Search ADS PubMed  36 Srikanthan A, Vera-Badillo F, Ethier J, et al.   Evolution in the eligibility criteria of randomized controlled trials for systemic cancer therapies. Cancer Treat Rev . 2016; 43: 67– 73. Google Scholar CrossRef Search ADS PubMed  37 Abdel-Qadir H, Amir E, Fischer HD, et al.   The risk of myocardial infarction with aromatase inhibitors relative to tamoxifen in post-menopausal women with early stage breast cancer. Eur J Cancer . 2016; 68: 11– 21. Google Scholar CrossRef Search ADS PubMed  38 Mouridsen H, Keshaviah A, Coates AS, et al.   Cardiovascular adverse events during adjuvant endocrine therapy for early breast cancer using letrozole or tamoxifen: Safety analysis of BIG 1-98 trial. J Clin Oncol.  2007; 25( 36): 5715– 5722. Google Scholar CrossRef Search ADS PubMed  39 Arimidex, Tamoxifen, Alone or in Combination Trialists’ Group, Buzdar A, Howell A, et al.   Comprehensive side-effect profile of anastrozole and tamoxifen as adjuvant treatment for early-stage breast cancer: Long-term safety analysis of the ATAC trial. Lancet Oncol.  2006; 7: 633– 643. Google Scholar CrossRef Search ADS PubMed  40 Nathan L, Shi W, Dinh H, et al.   Testosterone inhibits early atherogenesis by conversion to estradiol: Critical role of aromatase. Proc Natl Acad Sci U S A.   2001; 98( 6): 3589– 3593. Google Scholar CrossRef Search ADS PubMed  41 Breast International Group (BIG) 1-98 Collaborative Group, Thürlimann B, Keshaviah A, et al.   A comparison of letrozole and tamoxifen in postmenopausal women with early breast cancer. N Engl J Med.  2005; 353( 26): 2747– 2757. Google Scholar CrossRef Search ADS PubMed  42 Howell A, Cuzick J, Baum M, et al.   Results of the ATAC (Arimidex, Tamoxifen, Alone or in Combination) trial after completion of 5 years’ adjuvant treatment for breast cancer. Lancet.  2005; 365( 9453): 60– 62. Google Scholar CrossRef Search ADS PubMed  43 van de Velde CJ, Rea D, Seynaeve C, et al.   Adjuvant tamoxifen and exemestane in early breast cancer (TEAM): A randomised phase 3 trial. Lancet.  2011; 377( 9762): 321– 331. Google Scholar CrossRef Search ADS PubMed  44 Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Adjuvant bisphosphonate treatment in early breast cancer: Meta-analyses of individual patient data from randomised trials. Lancet .  2015; 386( 10001): 1353– 1361. CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI: Journal of the National Cancer Institute Oxford University Press

Toxicity of Extended Adjuvant Therapy With Aromatase Inhibitors in Early Breast Cancer: A Systematic Review and Meta-analysis

Loading next page...
 
/lp/ou_press/toxicity-of-extended-adjuvant-therapy-with-aromatase-inhibitors-in-0t9wsCH5I6
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.
ISSN
0027-8874
eISSN
1460-2105
D.O.I.
10.1093/jnci/djx141
Publisher site
See Article on Publisher Site

Abstract

Abstract Background A number of randomized controlled trials (RCTs) have reported improvement in breast cancer outcomes from extending treatment with aromatase inhibitors (AIs) beyond the initial five years after diagnosis. However, the toxicity profile of extended AIs is uncertain. Methods We identified RCTs that compared extended AIs to placebo or no treatment using MEDLINE and a review of abstracts from key conferences between 2013 and 2016. Odds ratios (ORs), 95% confidence intervals (CIs), absolute risks, and the number needed to harm (NNH) were computed for prespecified safety and tolerability outcomes including cardiovascular events, bone fractures, second cancers (excluding new breast cancer), treatment discontinuation for adverse events, and death without recurrence. All statistical tests were two-sided. Results Seven trials comprising 16 349 patients met the inclusion criteria. Longer treatment with AIs was associated with increased odds of cardiovascular events (OR = 1.18, 95% CI = 1.00 to 1.40, P = .05, NNH = 122), bone fractures (OR = 1.34, 95% CI = 1.16 to 1.55, P < .001, NNH = 72), and treatment discontinuation for adverse events (OR = 1.45, 95% CI = 1.25 to 1.68, P < .001, NNH = 21). Longer treatment with AIs did not influence the odds of either second malignancy (OR = 0.93, 95% CI = 0.73 to 1.18, P = .56) or deaths without breast cancer recurrence (OR = 1.11, 95% CI = 0.90 to 1.36, P = .34). Conclusions Extended treatment with AIs is associated with an increased risk of cardiovascular events and bone fractures. There is no statistically significant increase in deaths without breast cancer recurrence among patients receiving longer treatment with AIs. These data should be taken into account when considering extended adjuvant AIs. The use of aromatase inhibitors (AIs) as adjuvant treatment of postmenopausal women with breast cancer is well established (1). Different schedules of treatment, including upfront AIs or sequential use of tamoxifen and AIs, have been shown to improve disease-free survival and overall survival compared with tamoxifen alone (1–6). While five years of adjuvant hormonal treatment has been the gold standard for many years (7,8), there is a growing body of evidence supporting the benefit of extended endocrine treatment both with tamoxifen (9,10) and AIs (11–13) beyond the initial five years. Treatment with AIs is associated with well-described adverse events (14–17). When used in the five years after initial diagnosis, longer exposure to AIs is associated with more cardiovascular events and bone fractures when compared with tamoxifen (15). Data regarding the toxicity profile of extended treatment with AIs when compared with no therapy are more limited. The impact of safety and tolerability is likely of greater importance for extended adjuvant therapy compared with treatment in the initial five years because the absolute risk of breast cancer recurrence falls with time (18) while with increasing age, the baseline risk of morbidity and mortality from other illnesses rises (19). As such, robust data on the toxicity of extended adjuvant therapy are desirable in order to appropriately counsel patients and provide optimal recommendations on continuation of adjuvant therapy beyond five years. Here, we report on a meta-analysis evaluating adverse events reported in phase III randomized controlled trials (RCTs) comparing extended treatment with AIs to placebo or no treatment in postmenopausal women with hormone receptor–positive early breast cancer. Methods Literature Review and Study Identification A literature search utilizing MEDLINE (host: OVID; 1996–December week 3, 2016) and databases for the Annual Meetings of the American Society of Clinical Oncology and the San Antonio Breast Cancer Symposium between 2013 and 2016 was undertaken. The search was restricted to English language articles. The terms “adjuvant,” “aromatase inhibitor,” “tamoxifen,” and “breast cancer” and similar terms were cross-searched by using the following search algorithm: ((aromatase inhibitor OR anastrozole OR letrozole OR exemestane) AND (tamoxifen) AND (adjuvant) AND (breast neoplasm MeSH OR ((breast OR mammary) AND (carcinoma OR malignan* OR neoplasm OR tumor)))) AND (randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized controlled trial [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR (“clinical trial”) [tw] OR singl* [tw] OR doubl* [tw] OR trebl* [tw] OR tripl* [tw] AND (mask* [tw] OR blind* [tw])) OR comparative study [mh] OR evaluation studies [mh] OR follow-up studies [mh] OR prospective studies [mh] OR control* [tw] OR prospective* [tw] OR volunteer* [tw] NOT (animals [mh] NOT humans [mh]). Eligible trials included all phase III RCTs comparing longer duration of AI therapy with placebo or no treatment following initial adjuvant endocrine therapy with either tamoxifen, AIs, or both in postmenopausal women. We utilized a hierarchal primary data source. Data from safety analyses were preferred, and, when not available, data from intention-to-treat analyses were utilized instead. Data Extraction Data on potentially life-threatening adverse events of any grade as reported in individual trials were extracted. Collected data included the number of patients with the following prespecified adverse events: cardiovascular events, bone fractures, and secondary malignancies (excluding second primary breast cancers), as well as the total number of patients at risk of each adverse event. These events were chosen because they are associated with substantial morbidity and mortality, are commonly reported in RCTs, and are less prone to reporting bias. Data on symptomatic adverse events unlikely to be associated with mortality, such as arthralgia and genitourinary atrophy, were not considered. Additionally, data regarding treatment discontinuation for adverse events, as well as deaths without breast cancer recurrence, were collected. Finally, we collected data reporting both disease-free and overall survival from each study. Data were extracted independently by two reviewers (HG and EA). Discrepancies were resolved by consensus. All data were extracted from primary publications, their associated online appendices, or from the conference presentation if the study was not published in full. Data Synthesis and Statistical Analysis The primary analysis compared the odds of adverse events between patients who were randomly assigned to prolonged treatment with AIs and those randomly assigned to placebo or no treatment. Data from individual studies were pooled in a meta-analysis. The odds ratios (ORs) and associated 95% confidence intervals (CIs) were computed for all adverse events and were then pooled in a meta-analysis using RevMan 5.3 (The Cochrane Collaboration, Copenhagen, Denmark). Pooled estimates of odds ratio were computed using the Mantel-Haenszel odds ratio method (21). Due to substantial clinical heterogeneity between studies in the time from diagnosis to random assignment and in exposure to prior treatments, analyses were performed using random effects modeling irrespective of the statistical heterogeneity. Statistical heterogeneity was reported using Cochran Q (which assesses whether observed differences in results are compatible with chance alone) and I2 statistics (which describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error). Statistically significant heterogeneity was defined as a Cochran Q P value of less than .10 or I2 greater than 50%. Absolute risks were calculated as the number of events per person over the follow-up period of individual trials. The difference in absolute risk between the prolonged treatment and control groups was also presented as the number needed to harm (NNH), which quantifies the number of patients who would need to be treated with longer duration of AIs to cause an adverse event in one patient who would not otherwise have experienced the adverse event. Meta-regression was used to explore the influence of median age at random assignment, duration of follow-up, and prior treatments on the odds ratio for each adverse event. Meta-regression comprised a univariable linear regression weighted by individual study sample size exploring the association between median age, duration of follow-up, and proportion of patients treated with prior tamoxifen, AIs, or cytotoxic chemotherapy on the log of the odds ratio for each adverse event. Meta-regression was performed using SPSS version 21 (IBM Corp, Armonk, NY) using the weighted least squares (mixed effect) function (44). Finally, in an exploratory analysis, we pooled the disease-free and overall survival from each trial using inverse variance and random effects modeling. Statistical tests were two-sided, and statistical significance was defined as a P value of less than .05. Results Included Studies Our search identified 1546 studies initially. After exclusions, we identified 14 individual RCTs evaluating adjuvant endocrine monotherapy. Of these, seven trials explored initial adjuvant therapy. Therefore, seven studies were eligible for analysis (see Figure 1). Eligible studies comprised 16 349 patients. Individual study characteristics are shown in Table 1 while an individual study schema is shown in Supplementary Figure 1 (available online). Four trials (MA.17R [12], MA17 [13], NSABP B33 [23], and NSABP B42 [24]) compared five years of extended AIs with placebo. One trial (ABCSG 6a [11]) compared three years of extended anastrozole with no treatment, and two trials (DATA [25] and IDEAL [26]) compared longer with shorter duration of AIs. In three of these trials (13,23,25), eligibility required no prior exposure to AIs; however, in one trial (ABCSG 6a [11]) patients were not exposed to third-generation AIs, but approximately half of the cohort had been treated with aminoglutethimide, a first-generation AI. Adverse events were defined differently in each trial (see Table 2). Table 1. Characteristics of included studies Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  * Tumor size: T1 ≤2 cm, T2 >2 cm ≤5 cm, T3 >5 cm. AI = aromatase inhibitor; NR = not reported. † Study required early accrual termination and unbinding in view of the MA.17 (13) results in October 2003. ‡ 99% received prior AIs for 4.5 to six years. § Median age not reported, but estimated from reported age ranges. Table 1. Characteristics of included studies Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  Trial/median follow-up  Treatment arms  Sample size, No.  Median age, y  Tumor size (%)*  Percentage node positive, %  Prior therapy  ABCSG 6a, Jakesz et al. 2007 (11) (62.3 mo)  Anastrozole 3 y None  387 469  68.2  T1 (62.7) T2 (35.4) T3 (1.9)  32.5  Radiotherapy: NR Chemotherapy: NR Tamoxifen alone 5 y: 52.6% Tamoxifen 5 y+ aminoglutethimide first 2 y with tamoxifen: 47.4%  MA.17, Goss et al. 2005 (13) (30 mo)  Letrozole 5 y Placebo  2572 2577  62  NR  50.3  Radiotherapy: NR Chemotherapy: 45.3% Tamoxifen: 100%, median duration usage 5 y AI: 0%  NSABP B-33, Mamounas et al. 2008 (23)† (30 mo)  Exemestane 5 y Placebo  783 779  60  (38)  48  Radiotherapy: NR Chemotherapy: 55% Tamoxifen: 100%, duration 57–67 mo AIs: 0%  Dutch DATA, Tjan-Heijnen et al. 2016 (25) (49.2 mo, adapted)  Anastrozole 6 y Anastrozole 3 y  827 833  57  (54.2)  67  Radiotherapy: NR Chemotherapy: 68.4% Tamoxifen: 100%, 2–3 y AIs: 0%  IDEAL, Fontein et al. 2012 and Blok et al. 2016 (26,27) (78 mo)  Letrozole 5 y Letrozole 2.5 y  903 898  55–65§  NR  NR  Radiotherapy: 70.8% Chemotherapy: NR Total 5 y of endocrine therapy: Tamoxifen: 11.6% AIs: 24% Sequence of tamoxifen and AIs: 60% Unknown: 4.4%  MA.17R, Goss et al. 2016 (12) (75.6 mo)  Letrozole 5 y Placebo  959 959  65. 1  T1-2 (90.5) T3-4 (8.7)  51.4  Radiotherapy: NR Chemotherapy: NR Tamoxifen: 79.3% AIs‡: 100%  NSABP B-42, Mamounas et al. 2016 (24) (82.8 mo)  Letrozole 5 y Placebo  1959 1964  ≥60§  NR  42.6  Radiotherapy: NR Chemotherapy: NR Total 5 y of endocrine therapy: AIs: 60.9% Sequence of tamoxifen and AIs: 39.1%  * Tumor size: T1 ≤2 cm, T2 >2 cm ≤5 cm, T3 >5 cm. AI = aromatase inhibitor; NR = not reported. † Study required early accrual termination and unbinding in view of the MA.17 (13) results in October 2003. ‡ 99% received prior AIs for 4.5 to six years. § Median age not reported, but estimated from reported age ranges. Table 2. Definition of cardiac and bone adverse events in the included trials Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Table 2. Definition of cardiac and bone adverse events in the included trials Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Trial  Cardiac event  Bone  ABCSG 6a, Jakesz et al. 2007 (11)  Myocardial infarction  Fractures  MA.17, Goss et al. 2005 (13)  Cardiovascular disease: myocardial infarction, stroke/transient ischemic attack, new or worsening angina, angina requiring percutaneous coronary intervention, angina requiring coronary artery bypass graft, thromboembolic event, other  Fractures, including anatomic location data  NSABP B-33, Mamounas et al. 2008 (23)  Not evaluated  Fractures  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  Cardiovascular events, including arrhythmia  Fractures  IDEAL, Blok et al. 2016 (26)  Cardiac disorder: palpitations, other  Fractures  MA.17R, Goss et al. 2016 (12)  Cardiovascular events: ventricular arrhythmia, edema, cardiac left ventricular failure, ischemia/infarction, supraventricular arrhythmia, cardiac troponin I, thrombosis/ embolism, other  Fractures, including anatomic location data  NSABP B-42, Mamounas et al. 2016 (24)  arterial thrombotic event  Osteoporotic fractures  Figure 1. View largeDownload slide Study selection. Figure 1. View largeDownload slide Study selection. Cardiovascular Events Six studies reported data on cardiovascular events (11–13,24–26). Prolonged therapy with AIs was associated with a statistically significant increase in odds of cardiovascular events (OR = 1.18, 95% CI =  1.00 to 1.40, P = .05) (see Figure 2A). There was no evidence of statistically significant heterogeneity (Cochran Q P = .21, I2 = 30%). In absolute terms, 7.0% of patients in the prolonged AIs group had cardiovascular events, compared with 6.0% of patients in the control group. The weighted pooled absolute difference was 0.8%, with a NNH of 122 (see Table 3). Results of meta-regression are shown in Table 4. There was no association between the relative odds of cardiovascular events and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen or prior AIs. Table 3. Pooled absolute risk of adverse events* Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  * Positive values indicate excess events with prolonged aromatase inhibitors; negative values indicate excess events with placebo/no treatment. AE = adverse event; NNH = number needed to harm; NR = not reported. Table 3. Pooled absolute risk of adverse events* Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  Trial  Cardiovascular disease   Fractures   Treatment discontinuation for AE   Second cancers   Death without recurrence   Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  Absolute difference, %  NNH  ABCSG 6a, Jakesz et al. 2007 (11)  +0.26  +385  –0.29  –345  +9.83  +11  NR  NR  +0.84  +120  MA.17, Goss et al. 2005 (13)  +0.21  +477  +0.71  +141  +1.29  +78  –0.27  –371  –0.19  –527  NSABP B-33, Mamounas et al. 2008 (23)  NR  NR  +1.01  +100  NR  NR  –0.01  –10 000  –0.64  +157  Dutch DATA, Tjan-Heijnen et al. 2016 (25)  +0.58  +173  +2.35  +43  +8.57  +12  +0.61  164  NR  NR  IDEAL, Blok et al. 2016 (26)  +4.14  +25  +3.22  +32  +8.29  +13  NR  NR  NR  NR  MA.17R, Goss et al. 2016 (12)  +1.82  +55  +4.64  +22  +1.75  +58  NR  NR  +0.28  +358  B-42, Mamounas et al. 2016 (24)  +0.57  +176  +0.67  +150  +3.23  +31  –0.39  –257  +0.67  +150  Weighted pooled effect  +0.82  +122  +1.39  +72  +4.82  +21  –0.16  –625  +0.27  +371  * Positive values indicate excess events with prolonged aromatase inhibitors; negative values indicate excess events with placebo/no treatment. AE = adverse event; NNH = number needed to harm; NR = not reported. Table 4. Results of meta-regression exploring the association of adverse events with age, duration of follow-up, and proportion of patients with prior treatments Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      * All P values are two-sided. AI = aromatase inhibitor. † Cannot estimate because only two studies reported data (13,23). Table 4. Results of meta-regression exploring the association of adverse events with age, duration of follow-up, and proportion of patients with prior treatments Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      Variable  β  Standard error  P*  Cardiovascular events         Age  0.638  0.129  .36   Median follow-up  0.266  0.013  .61   Prior tamoxifen  0.069  1.246  .90   Prior AI  –0.038  0.825  .94   Prior chemotherapy  NE†      Fractures         Age  –0.296  0.041  .63   Median follow-up  0.315  0.006  .49   Prior tamoxifen  –0.092  0.556  .85   Prior AI  –0.172  0.337  .71   Prior chemotherapy  0.797  0.719  .41  Treatment discontinuation         Age  –0.337  0.032  .66   Median follow-up  –0.253  0.004  .63   Prior tamoxifen  0.689  0.260  .13   Prior AI  –0.501  0.206  .31   Prior chemotherapy  NE†      Second cancers         Age  –0.987  0.037  .10   Median follow-up  0.486  0.010  .51   Prior tamoxifen  –0.279  0.929  .72   Prior AI  –0.909  0.285  .09   Prior chemotherapy  0.990  0.692  .09  Death without breast cancer recurrence         Age  –0.471  0.257  .53   Median follow-up  –0.170  0.019  .79   Prior tamoxifen  0.093  1.722  .88   Prior AI  –0.615  0.923  .27   Prior chemotherapy  NE†      * All P values are two-sided. AI = aromatase inhibitor. † Cannot estimate because only two studies reported data (13,23). Figure 2. View largeDownload slide Forest plots for adverse events. A) Cardiovascular events. B) Bone fractures. C) Other second cancers. D) Discontinuation for adverse events. E) Death without breast cancer recurrence. Odds ratios for each trial are represented by squares, the size of the square represents the weight of the trial in the meta-analysis, and the horizontal line crossing the square represents the 95% confidence interval. The diamonds represent the estimated pooled effect. Test for overall effect based on z-test. All P values are two-sided. CI = confidence interval; OR = odds ratio. Figure 2. View largeDownload slide Forest plots for adverse events. A) Cardiovascular events. B) Bone fractures. C) Other second cancers. D) Discontinuation for adverse events. E) Death without breast cancer recurrence. Odds ratios for each trial are represented by squares, the size of the square represents the weight of the trial in the meta-analysis, and the horizontal line crossing the square represents the 95% confidence interval. The diamonds represent the estimated pooled effect. Test for overall effect based on z-test. All P values are two-sided. CI = confidence interval; OR = odds ratio. Hypertension was reported in only three studies (12,13,26). There was no association between longer duration of AIs and increased odds of hypertension (OR = 1.02, 95% CI =  0.86 to 1.21, P = .82; data not shown). There was no evidence of statistically significant heterogeneity (Cochran Q P = .42, I2 = 0%). In absolute terms, hypertension was reported in 7.3% of patients receiving longer treatment with AIs and 7.1% of patients in the control group. The weighted pooled absolute difference was 0.1%, with a NNH of 770. Bone Fractures All seven studies reported data on bone fractures (11–13,23–26). Prolonged therapy with AIs was associated with statistically significantly increased odds of bone fractures (OR = 1.34, 95% CI =  1.16 to 1.55, P < .001) (see Figure 2B). There was no evidence of statistically significant heterogeneity (Cochran Q P = .38, I2 = 6%). In absolute terms, fractures occurred in 6.3% of the prolonged AI group and 4.8% in the control group, with a weighted pooled absolute difference of 1.4% and a NNH of 72 (see Table 3). Results of the meta-regression are shown in Table 4. There was no association between the relative odds of bone fracture and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen, prior AIs, or prior chemotherapy. Second Cancers Four trials reported the occurrence of second malignancies excluding breast cancer, three studies reported on all second malignancies (23–25), and one study reported only on endometrial cancer (13). Pooled data showed that prolonged treatment with AIs was not associated with second cancers (OR = 0.93, 95% CI =  0.73 to 1.18, P = .56) (see Figure 2C). There was no evidence of statistically significant heterogeneity (Cochran Q P = .26, I2 = 25%). Exclusion of the study reporting only endometrial carcinoma did not change the estimate (OR = 0.94, 95% CI =  0.58 to 1.52, P = .81), and heterogeneity remained statistically nonsignificant (Cochran Q P = .54, I2 = 0%). The absolute risk of non–breast cancer secondary malignancies was 2.2% for prolonged treatment with AIs and 2.4% for the control group, a weighted pooled difference of –0.2% and a NNH of –625. Results of the meta-regression are shown in Table 4. There was no association between the relative odds of second cancers and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen, prior AIs, or prior chemotherapy. Discontinuation for Adverse Events Discontinuation of treatment for adverse events was reported in six studies (11–13,24–26). Pooled data showed that longer duration of treatment with AIs was associated with a 45% increased odds for treatment discontinuation for adverse events compared with placebo or no treatment (OR = 1.45, 95% CI = 1.25 to 1.68, P < .001) (see Figure 2D). There was evidence of statistically significant heterogeneity (Cochran Q P = .07, I2 = 51%). In absolute terms, discontinuation for adverse events occurred in 17.0% of patients in the prolonged treatment group and 13.4% in the control group. The weighted pooled absolute difference was 4.8%, with a NNH of 21. There was no association between the relative odds of discontinuation for adverse events and median age at random assignment, median duration of follow-up, or proportion of patients who had received prior tamoxifen or prior AIs (see Table 4). Death Without Recurrence Death without breast cancer recurrence was reported in five studies (11–13,22,23). There was no statistically significant association between death without breast cancer recurrence and longer use of AIs (OR = 1.11, 95% CI =  0.90 to 1.36, P = .34) (see Figure 2E). There was no evidence of statistically significant heterogeneity (Cochran Q P = .44, I2 = 0%). In absolute terms, 3.3% of the patients treated with prolonged AIs and 2.9% of those treated with placebo or no treatment died without breast cancer recurrence. The weighted pooled absolute difference was 0.3%, with a NNH of 371. There was no association between the relative odds of death without breast cancer recurrence and median age at random assignment, median duration of follow-up, or the proportion of patients who had received prior tamoxifen or prior AIs (see Table 4). In an exploratory analysis, despite improvement in disease-free survival (pooled OR = 0.78, 95% CI =  0.68 to 0.88, P< .001) (see Supplementary Figure 2, available online), no effect on overall survival was observed (OR = 1.01, 95% CI =  0.89 to 1.14, P = .92) (see Supplementary Figure 2, available online). Discussion Late relapse, which can occur decades after initial diagnosis, is a hallmark of hormone receptor–positive early breast cancer (1). Consequently, there has been interest in extended adjuvant therapy in such patients. The net benefit of any therapy is comprised of the balance between benefits and risks. In the setting of extended adjuvant endocrine therapy, an accurate assessment of this balance is essential because over time the risk of breast cancer recurrence and death falls and the contribution of concurrent illnesses to morbidity and mortality increases (19). As such, knowledge of the potential impact of extended anticancer therapy on comorbidities is important. Typically, individual trials of adjuvant endocrine therapy are designed to evaluate efficacy. Most have limited power to explore toxicities, especially those that are less common. As a result, pooling of data from multiple trials is valuable to evaluate accurately the effect of adjuvant therapy on safety and tolerability. An example of this was the observation of increased cardiovascular disease with AIs relative to tamoxifen as initial adjuvant therapy. No effect was seen in individual trials, but a modest increase in cardiovascular disease was observed in a meta-analysis of randomized trials (15). In contrast, population-based studies that utilized different definitions of cardiovascular events have reported conflicting data (19,28). Most of the current knowledge on the safety and tolerability of AIs is derived from data that compared AIs with tamoxifen. Consequently, the influence of extended AIs on toxicity when compared with no treatment is uncertain. There has been increasing interest in the potential interaction of AIs on cardiovascular disease. Cardiovascular disease and breast cancer share a number of risk factors (29,30), and, thanks to effective breast cancer therapy, the most common cause of death among women with early breast cancer is cardiovascular disease (31,32). Data from trials of initial adjuvant therapy have shown that longer duration of AIs is associated with increased cardiovascular events (15). A key limitation of these data is that in all trials the comparison group was tamoxifen and it was not possible to determine if the relative increase in the odds of cardiovascular events was related to a real increase in risk with AIs or some protection from tamoxifen (33). Recent data focusing on studies of AIs compared with placebo or no treatment suggested no increase in risk of cardiovascular events with AIs (34). However, this study did not include a number of the large randomized trials reported recently. In the current study, we observed a statistically significant increase in the odds of cardiovascular events with longer durations of AIs. The relative increase was modestly lower in magnitude to that observed in the initial adjuvant setting (OR = 1.18 vs 1.26, respectively); however, because of higher event rates in this older population of patients with lower competing risks of cancer death, this relative increase translated to a similar absolute risk (NNH = 122 vs 132, respectively). There are concerns that safety and tolerability data from clinical trials underestimate the effect of toxicity when treatment is administered to patients outside of clinical trials. Data show that many patients in routine practice do not meet the eligibility criteria of clinical trials (35). Eligibility criteria of randomized trials often exclude patients with preexisting cardiovascular disease (36), a group at higher risk of treatment-related toxicity. A population-based observation study focusing on cardiovascular events rather than a composite of both cardiovascular and cerebrovascular events reported that, compared with tamoxifen, AIs are associated with a doubling in the hazard of serious cardiovascular events (37), a magnitude of effect much greater than observed in clinical trials. The mechanism of effect of AIs on cardiovascular morbidity is not clear. AIs have been associated with an increase in cardiovascular risk factors such as dyslipidemia and arterial hypertension (15,38,39). In the current study, we observed a similar risk of incident arterial hypertension among those receiving longer therapy with AIs and control groups. This suggests that the link between AIs and cardiovascular events may relate to mechanisms independent of arterial hypertension such as accelerated atherosclerosis, an observation reported elsewhere (40). In postmenopausal women, AIs cause accelerated bone loss (1). Several trials of initial adjuvant therapy showed a statistically significantly higher risk of bone fractures in patients treated with AIs compared with tamoxifen (4,41–43). Again, this analysis was potentially confounded by a protective effect of tamoxifen. In the current study, the detrimental effect on bone health was confirmed, with a 34% increased odds of developing bone fractures with AIs compared with placebo or no treatment. This increase translated to a NNH of 72, modestly higher than data observed in the initial adjuvant setting (NNH = 46) (15). Potential reasons for the lower magnitude of absolute risk of bone fractures with extended AI therapy compared with initial adjuvant therapy are the protective effect of tamoxifen in the initial adjuvant setting, improved screening of patients treated with AIs with bone mineral density scans, and the treatment of those at risk with antiresorptive therapies. The impact of treatment-related bone loss may be attenuated in the future with the increasing use of adjuvant bone-targeted therapy, even in those without preexisting or treatment-related bone loss (44). Statistically significantly more patients randomly assigned to prolonged treatment with AIs discontinued therapy for adverse events. Pooled data showed that for every 21 patients assigned to longer therapy with AIs, one patient discontinued treatment due to adverse events. This finding is notable because all enrolled participants had received three to 10 years of prior endocrine therapy prior to random assignment, suggesting that serious toxicity necessitating discontinuation of therapy can occur even at a late stage of treatment. Extended adjuvant AIs are associated with a 20% to 25% relative reduction in disease recurrence, although in many cases such events are comprised of local or contralateral in-breast recurrences. In our analysis, we observed a numerical excess of deaths without breast cancer recurrence with prolonged AI therapy; however, this association was not statistically significant. This finding may explain why there is no apparent effect of extended therapy with AIs on overall survival. The increased number of noncancer deaths with longer treatment with AIs attenuates the contribution of reduced breast cancer deaths to overall survival. It is important to stress that overall survival data were immature in many studies and it is possible that the long-term effect of exposure to AIs may result in the emergence of both a survival advantage and additional adverse events with longer durations of follow-up. Of interest, we did not identify any factors that influenced the effect of prolonged treatment with AIs on safety and tolerability. Similar magnitudes of relative odds were observed for all adverse events of interest, irrespective of age, duration of follow-up, or prior treatments (including tamoxifen, AIs, and chemotherapy). These findings do not exclude an effect on absolute risk. Cardiovascular events and bone fractures increase with age. As such, despite no effect on the relative risk of these events between prolonged treatment with AIs and control, it is anticipated that older patients and those with more comorbidities with longer durations of AIs will have a higher absolute risk of adverse events than those treated with placebo or no treatment. This study has limitations. First, this is a literature-based rather than an individual patient–based meta-analysis. Consequently, data on comorbidities and concurrent medications were not available and could not be adjusted for in the analysis. Additionally, because data were extracted from reports of trials with different durations of follow-up, we were unable to report actuarial rates of toxicity. Second, the reporting and definition of toxicity was variable among the different trials and may lead to some uncertainty about estimates of odds of adverse events. Furthermore, one study (11) was open label rather than placebo controlled and therefore subject to ascertainment bias. Third, although the grade of the toxicity might clarify the severity of the adverse event on the patient, grading of adverse events was not reported uniformly, and therefore our analysis included all grades of toxicity. Fourth, inclusion criteria for each trial were different, especially with regard to prior adjuvant endocrine therapy. Some patients received only two to three years of prior tamoxifen (24), while others received up to 10 years of endocrine therapy with both tamoxifen and AIs (12), making the included patients in this meta-analysis heterogeneous. Finally, toxicity reporting is usually collected only until an efficacy event of interest occurs. As the majority of women with hormone receptor–positive breast cancer survive for several years even with metastatic disease, the impact of adverse events after recurrence remains of interest. In conclusion, extended treatment with AIs is associated with an increased risk of cardiovascular events and bone fractures. There is no statistically significant excess of deaths without breast cancer recurrence among patients receiving longer durations of AIs. These data should be taken into account when considering extended adjuvant AIs for breast cancer patients, especially patients with preexisting comorbidities or risk factors thereof. Funding None for all authors. References 1 Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Aromatase inhibitors versus tamoxifen in early breast cancer: Patient-level meta-analysis of the randomised trials. Lancet . 2015; 386( 10001): 1341– 1352. CrossRef Search ADS PubMed  2 Boccardo F, Rubagotti A, Guglielmini P, et al.   Switching to anastrozole versus continued tamoxifen treatment of early breast cancer. Updated results of the Italian tamoxifen anastrozole (ITA) trial. Ann Oncol.  2006; 17( suppl 7): vii10– vii14. Google Scholar CrossRef Search ADS PubMed  3 Coates AS, Keshaviah A, Thurlimann B, et al.   Five years of letrozole compared with tamoxifen as initial adjuvant therapy for postmenopausal women with endocrine-responsive early breast cancer: Update of study BIG 1-98. J Clin Oncol.  2007; 25( 5): 486– 492. Google Scholar CrossRef Search ADS PubMed  4 Jakesz R, Jonat W, Gnant M, et al.   Switching of postmenopausal women with endocrine-responsive early breast cancer to anastrozole after 2 years' adjuvant tamoxifen: Combined results of ABCSG trial 8 and ARNO 95 trial. Lancet.  2005; 366( 9484): 455– 462. Google Scholar CrossRef Search ADS PubMed  5 Forbes JF, Cuzick J, Buzdar A, Howell A, Tobias JS, Baum M. Effect of anastrozole and tamoxifen as adjuvant treatment for early-stage breast cancer: 100-month analysis of the ATAC trial. Lancet Oncol.  2008; 9( 1): 45– 53. Google Scholar CrossRef Search ADS PubMed  6 Coombes RC, Kilburn LS, Snowdon CF, et al.   Survival and safety of exemestane versus tamoxifen after 2-3 years’ tamoxifen treatment (Intergroup Exemestane Study): A randomised controlled trial. Lancet . 2007;17; 369( 9561): 559– 570. Google Scholar CrossRef Search ADS PubMed  7 Early Breast Cancer Trialists Collaborative Group (EBCTCG). Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: Patient-level meta-analysis of randomised trials.  Lancet.  2011; 378( 9793): 771– 784. CrossRef Search ADS PubMed  8 Fisher B, Dignam J, Bryant J, Wolmark N. Five versus more than five years of tamoxifen for lymph node-negative breast cancer: Updated findings from the National Surgical Adjuvant Breast and Bowel Project B-14 randomized trial. J Natl Cancer Inst.   2001; 93( 9): 684– 690. Google Scholar CrossRef Search ADS PubMed  9 Davies C, Pan H, Godwin J, et al.   Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial. Lancet.  2013; 381( 9869): 805– 816. Google Scholar CrossRef Search ADS PubMed  10 Gray RG, Rea D, Handley K, et al.   aTTom: Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years in 6953 women with early breast cancer. J Clin Oncol.  2013; 31( 18 suppl): 5. Google Scholar CrossRef Search ADS PubMed  11 Jakesz R, Greil R, Gnant M, et al.   Extended adjuvant therapy with anastrozole among postmenopausal breast cancer patients: Results from the randomized Austrian Breast and Colorectal Cancer Study Group Trial 6a. J Natl Cancer Inst . 2007; 99( 24): 1845– 1853. Google Scholar CrossRef Search ADS PubMed  12 Goss PE, Ingle JN, Pritchard KI, et al.   Extending aromatase inhibitors adjuvant therapy to 10 years. N Engl J Med.  2016; 375( 3): 209– 219. Google Scholar CrossRef Search ADS PubMed  13 Goss PE, Ingle JN, Martino S, et al.   Randomized trial of letrozole following tamoxifen as extended adjuvant therapy in receptor-positive breast cancer: Updated findings from NCIC CTG MA.17. J Natl Cancer Inst.  2005; 97( 17): 1262– 1271. Google Scholar PubMed  14 Seruga B, Tannock IF. Up-front use of aromatase inhibitors as adjuvant therapy for breast cancer: The emperor has no clothes. J Clin Oncol.  2009; 27( 6): 840– 842. Google Scholar CrossRef Search ADS PubMed  15 Amir E, Seruga B, Niraula S, Carlsson L, Ocaña A. Toxicity of adjuvant endocrine therapy in postmenopausal breast cancer patients: A systematic review and meta-analysis. J Natl Cancer Inst.  2011; 103( 17): 1299– 1309. Google Scholar PubMed  16 Haque R, Shi J, Schottinger JE, et al.   cardiovascular disease after aromatase inhibitor use. JAMA Oncol.  2016; 2( 12): 1590– 1597. Google Scholar CrossRef Search ADS PubMed  17 Rydén L, Heibert Arnlind M, Vitols S, Höistad M, Ahlgren J. Aromatase inhibitors alone or sequentially combined with tamoxifen in postmenopausal early breast cancer compared with tamoxifen or placebo – meta-analyses on efficacy and adverse events based on randomized clinical trials. Breast.  2016; 26: 106– 114. Google Scholar CrossRef Search ADS PubMed  18 Early Breast Cancer Trialists' Collaborative Group. Tamoxifen for early breast cancer: An overview of the randomised trials. Lancet . 1998; 351( 9114): 1451– 1467. CrossRef Search ADS PubMed  19 Abdel-Qadir H, Austin PC, Lee DS, et al.   A population-based study of cardiovascular mortality following early-stage breast cancer. JAMA Cardiol . 2017; 2( 1): 88– 93. Google Scholar CrossRef Search ADS PubMed  20 Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med.  2004; 23( 9): 1351– 1375. Google Scholar CrossRef Search ADS PubMed  21 Higgins JPT, Deeks JJ, Altman DG. Metaanalysis of rare events. In: Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions , version 5.1.0. The Cochrane Collaboration; 2011. http://handbook.cochrane.org. Accessed January 31, 2017. 22 Stanley TD, Doucouliagos H. Neither fixed nor random: Weighted least squares meta-regression. Res Synth Methods.  2017; 8( 1): 19– 42. Google Scholar CrossRef Search ADS PubMed  23 Mamounas EP, Jeong JH, Wickerham DL, et al.   Benefit from exemestane as extended adjuvant therapy after 5 years of adjuvant tamoxifen: Intention-to-treat analysis of the National Surgical Adjuvant Breast and Bowel Project B-33 trial. J Clin Oncol.   2008; 26( 12): 1965– 1971. Google Scholar CrossRef Search ADS PubMed  24 Mamounas EP, Bandos Hanna, Lembersky BC, et al.   A randomized, double-blinded, placebo-controlled clinical trial to evaluate extended adjuvant endocrine therapy (5 years of letrozole) in postmenopausal women with hormone-receptor positive breast cancer who have completed previous adjuvant endocrine therapy: Initial results of NRG oncology/NSABP B-42. Cancer Res . 2017; 77( Suppl 4): S1– 05. 25 Tjan-Heijnen VC, Van Hellemond IE, Peer PG, et al.   First results from the multicenter phase III DATA study comparing 3 versus 6 of anastrozole after 2-3 years of tamoxifen in postmenopausal women with hormone receptor-positive early breast cancer. Cancer Res . 2017; 77( Suppl 4): S1– 03. 26 Blok EJ, van de Velde CJH, Meershoek-Klein Kranenbarg EM, et al.   Safety assessment of extended adjuvant endocrine therapy with letrozole; results of the randomized phase III IDEAL trial (BOOG 2006-05). Cancer Res . 2017; 77( Suppl 4): S1– 04. 27 Fontein DB, Nortier JW, Liefers GJ, et al.   High non-compliance in the use of letrozole after 2.5 years of extended adjuvant endocrine therapy. Results from the IDEAL randomized trial. Eur J Surg Oncol.   2012; 38( 2): 110– 117. Google Scholar CrossRef Search ADS PubMed  28 Haque R, Shi J, Schottinger JE, Chung J, et al.   cardiovascular disease after aromatase inhibitor use. JAMA Oncol.  2016; 2( 12): 1590– 1597. Google Scholar CrossRef Search ADS PubMed  29 Key TJ, Appleby PN, Reeves GK. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst.  2003; 95( 16): 1218e26. 30 Calle EE, Kaaks R. Overweight, obesity and cancer: Epidemiological evidence and proposed mechanisms. Nat Rev Cancer.  2004; 4( 8): 579e91. Google Scholar CrossRef Search ADS   31 Patnaik JL, Byers T, DiGuiseppi C, Dabelea D, Denberg TD. Cardiovascular disease competes with breast cancer as the leading cause of death for older females diagnosed with breast cancer: A retrospective cohort study. Breast Cancer Res.  2011; 13( 3): R64. Google Scholar CrossRef Search ADS PubMed  32 Yancik R, Wesley MN, Ries LA, Havlik RJ, Edwards BK, Yates JW. Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. JAMA.  2001; 285( 7): 885e92. Google Scholar CrossRef Search ADS   33 Love RR, Wiebe DA, Newcomb PA, et al.   Effects of tamoxifen on cardiovascular risk factors in postmenopausal women. Ann Intern Med.  1991; 115( 11): 860– 864. Google Scholar CrossRef Search ADS PubMed  34 Khosrow-Khavar F, Filion KB, Al-Qurashi S, et al.   Cardiotoxicity of aromatase inhibitors and tamoxifen in post-menopausal women with breast cancer: A systematic review and meta-analysis of randomized controlled trials. Ann Oncol .  2016; 28( 3): 487– 496. 35 Treweek S, Dryden R, McCowan C, et al.   Do participants in adjuvant breast cancer trials reflect the breast cancer patient population? Eur J Cancer.  2015; 51: 907– 914. Google Scholar CrossRef Search ADS PubMed  36 Srikanthan A, Vera-Badillo F, Ethier J, et al.   Evolution in the eligibility criteria of randomized controlled trials for systemic cancer therapies. Cancer Treat Rev . 2016; 43: 67– 73. Google Scholar CrossRef Search ADS PubMed  37 Abdel-Qadir H, Amir E, Fischer HD, et al.   The risk of myocardial infarction with aromatase inhibitors relative to tamoxifen in post-menopausal women with early stage breast cancer. Eur J Cancer . 2016; 68: 11– 21. Google Scholar CrossRef Search ADS PubMed  38 Mouridsen H, Keshaviah A, Coates AS, et al.   Cardiovascular adverse events during adjuvant endocrine therapy for early breast cancer using letrozole or tamoxifen: Safety analysis of BIG 1-98 trial. J Clin Oncol.  2007; 25( 36): 5715– 5722. Google Scholar CrossRef Search ADS PubMed  39 Arimidex, Tamoxifen, Alone or in Combination Trialists’ Group, Buzdar A, Howell A, et al.   Comprehensive side-effect profile of anastrozole and tamoxifen as adjuvant treatment for early-stage breast cancer: Long-term safety analysis of the ATAC trial. Lancet Oncol.  2006; 7: 633– 643. Google Scholar CrossRef Search ADS PubMed  40 Nathan L, Shi W, Dinh H, et al.   Testosterone inhibits early atherogenesis by conversion to estradiol: Critical role of aromatase. Proc Natl Acad Sci U S A.   2001; 98( 6): 3589– 3593. Google Scholar CrossRef Search ADS PubMed  41 Breast International Group (BIG) 1-98 Collaborative Group, Thürlimann B, Keshaviah A, et al.   A comparison of letrozole and tamoxifen in postmenopausal women with early breast cancer. N Engl J Med.  2005; 353( 26): 2747– 2757. Google Scholar CrossRef Search ADS PubMed  42 Howell A, Cuzick J, Baum M, et al.   Results of the ATAC (Arimidex, Tamoxifen, Alone or in Combination) trial after completion of 5 years’ adjuvant treatment for breast cancer. Lancet.  2005; 365( 9453): 60– 62. Google Scholar CrossRef Search ADS PubMed  43 van de Velde CJ, Rea D, Seynaeve C, et al.   Adjuvant tamoxifen and exemestane in early breast cancer (TEAM): A randomised phase 3 trial. Lancet.  2011; 377( 9762): 321– 331. Google Scholar CrossRef Search ADS PubMed  44 Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Adjuvant bisphosphonate treatment in early breast cancer: Meta-analyses of individual patient data from randomised trials. Lancet .  2015; 386( 10001): 1353– 1361. CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.

Journal

JNCI: Journal of the National Cancer InstituteOxford University Press

Published: Aug 4, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off