Preventive drug therapy and contralateral breast cancer: summary of the evidence of clinical trials and observational studies

Abstract

Background: Breast cancer patients have a lifelong 2–4-fold increased risk of developing a second primary tumor in the contralateral breast compared with the risk for a first primary breast cancer in the general female population. Prevention of contralateral breast cancer (CBC) has received increased attention during recent decades. Here, we summarize and discuss the available literature on drug preventive therapy and CBC.

Results: The endocrine-targetting drugs, tamoxifen and aromatase inhibitors are used as standard adjuvant treatment for estrogen receptor (ER)-positive breast cancer. Both are associated with relative risk reductions of CBC of up to 50%, but incur serious side effects. Several prescription drugs originally developed for other purposes, including bisphosphonates, statins, non-steroidal anti-inflammatory drugs, metformin, anti-hypertensives and retinoids, have shown anti-cancer activity in preclinical models. However, results of observational studies on CBC are sparse and inconsistent, with only statins demonstrating promise as preventive agents and a potential treatment option for ER-negative breast cancer patients.

Conclusion: Future studies are needed to assess the effect of statins in risk reduction and to identify other drugs with chemopreventive potential against CBC. Eventually, efforts must be directed towards identifying those breast cancer patients likely to benefit most from specific preventive therapies.

Background

Among women diagnosed with breast cancer, the risk of developing a second primary cancer in the contralateral breast is higher than the risk of developing a first primary breast cancer in the general female population [1,2]. The median annual cumulative incidence of contralateral breast cancer (CBC) is 0.5% (range: 0.2–0.7%) [3]. Several factors are associated with increased risk of CBC, including younger age at first breast cancer diagnosis, family history of breast cancer, and lobular histology and estrogen receptor (ER)-negative status of the first breast cancer [4].

Fear of developing CBC motivates some breast cancer patients to undergo contralateral prophylactic mastectomy [5]. This procedure has become more common recently and reduces CBC rates, but evidence of an absolute reduction in CBC risk and survival benefit associated with contralateral prophylactic mastectomy remains uncertain [6]. Other treatment options proven to reduce CBC risk are adjuvant endocrine therapy [7,8] and chemotherapy [7]. Recently, several prescribed drugs, originally developed for other medical conditions, have also shown anti-cancer potential in laboratory and observational studies [9] and may be effective in preventing CBC.

Meta-analyses and reviews have summarized the results from randomized controlled trials (RCTs) and observational studies examining preventive drug therapy for breast cancer. However, the majority of these have focused on the first breast cancer, whereas CBC has received limited attention. One review outlined treatment options to reduce CBC risk, with minor focus on chemopreventive drugs [10]. The purpose of this review is to summarize the available evidence from RCTs and observational studies of drug preventive therapy for CBC, discuss the main limitations of preventive therapy, and propose future research directions.

Adjuvant endocrine breast cancer treatment

Tamoxifen

Tamoxifen is a selective estrogen receptor modulator (SERM) that binds to and blocks the activity of the ER in the breast, while stimulating ER activity in the bone and endometrium [11]. Tamoxifen was the first targeted adjuvant breast cancer therapy and remains the standard endocrine treatment prescribed to premenopausal women with ER-positive breast cancer [12]. Among postmenopausal patients, tamoxifen is an effective alternative treatment among those intolerant to aromatase inhibitors (AIs). Tamoxifen is further approved for primary prevention among high-risk women in the United States (US) and recommended for prevention in United Kingdom (UK) [13]. In addition to reduced breast cancer recurrence and mortality after tamoxifen treatment [14], meta-analyses of RCTs summarized by the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) [7,8] and observational studies [15–26] have reported large reductions in the relative risk of CBC (10–56%) among tamoxifen users compared with non-users (Table 1 and Figure 1). Nonetheless, the absolute risk reductions were small (1.3–4% after 10–15 years) [8,25] (Table 1).

Figure 1. Forest plot of studies on preventive drug therapy and risk of contralateral breast cancer. ACE: angiotensin-converting enzyme; AI: aromatase inhibitors; CI: confidence interval; EBCTCG: Early Breast Cancer Trialists’ Collaborative Group; NSAID: non-steroidal anti-inflammatory drug; RCT: randomized controlled trial; TAM: tamoxifen; y: year. ^aMeta-analysis; ^bno confidence interval available; ^conly ER-positive and ER-unknown first breast cancer; ^dcalendar period 1999–2007; ^eincludes both new ipsilateral and contralateral breast cancer.

Table 1. Overview of clinical trials and observational studies on the association between tamoxifen and risk of contralateral breast cancer.

First author, year	Country	Study design	Number of breast cancer patients (CBC cases)	Data source of drug exposure	Follow-up time (years)	Definition of use	Risk estimate (95% CI, SD or SE)	Remarks
EBCTCG, 1998 [27]	Multi-center	Meta-analysis of 55 RCTs	36,689 (854)	Trial records	Mean: ∼10	≤1 year 2 years ∼5 years Any duration	RR 0.87; SD:0.13 RR 0.74; SD:0.09 RR 0.53; SD:0.09 RR 0.70; SD:0.06
EBCTCG, 2005 [7]	Multi-center	Meta-analysis of 12 RCTs	15,017 (575)	Trial records	Not reported	∼5 years vs. none	ER-positive and ER-unknown first breast cancer: RR 0.61 (0.50–0.73) ER-negative first breast cancer: RR 0.99 (0.70–1.36)
EBCTCG, 2011 [8]	Multi-center	Meta-analysis of 20 RCTs	21,457 (940)	Trial records	Median: 13	∼5 years vs. none	RR 0.71; SE: 0.06	Absolute risk reduction: 3.2% [SE: 0.8] at 15 years of follow-up for ER-positive first breast cancers
Hackshaw et al., 2011 [28]	Multi-center	RCT	3449 (118)	Trial records	Median: 10.1	5 years vs. 2 years	HR 0.70 (0.48–1.00)	Absolute risk reduction: 1.3% (95% CI: 0.00–2.3) at 10 years and 2.4% (95% CI: 0.0–4.2) at 15 years of follow-up
Davies et al., 2013 [14]	Multi-center	RCT	12,894 (886)	Trial records	Mean: 7.6	10 years vs. 5 years	RR 0.88 (0.77–1.00)
Rosell et al., 2017 [29]	Sweden	RCT	4128 (426)	Trial records	Not reported	5 years vs. 2 years	HR 0.73 (0.56–0.96)	HRs for TAM use by time since start of treatment
Cook et al., 1995 [16]	USA	Nested case-control	234 CBC cases/435 controls	Medical records	–	Ever use	OR 0.5 (0.3–0.9)	ORs by duration (≤12/≥13 months) of TAM and by time since last use (≤12/≥13 months)
Newcomb et al., 1999 [19]	USA	Cohort	85,411 (2648)	Cancer registry (SEER)	Mean: 3.2	Ever use of hormonal therapy	RR 0.79 (0.70–0.90)	Presumably tamoxifen as the study-period was 1983–1994. RRs for hormonal therapy use by time since first breast cancer diagnosis (<5/≥5 years)
Li et al., 2001 [24]	USA	Cohort	8981 (189)	Cancer registry (SEER)	Mean: 4	Ever use	HR 0.9 (0.7–1.2)	HRs for TAM use by ER status of CBC
Schaapveld et al., 2008 [20]	Netherlands	Cohort	45,229 (1477)	Medical records	Median: 5.8	Ever use of hormonal therapy	HR 0.58 (0.48–0.69)	Presumably tamoxifen
Bertelsen et al., 2008 [15]	Multi-center (WECARE)	Nested case-control	634 CBC cases/1158 controls	Medical records	–	Ever use	RR 0.66 (0.50–0.88)	RRs for TAM use by time since first breast cancer diagnosis (1–5, >5–10, >10 years)
Li et al., 2009 [23]	USA	Nested case-control	358 CBC cases/674 controls	Telephone interview and medical records	–	Ever use	OR 0.6 (0.5–0.8)	Only hormone receptor-positive first breast cancer patients ORs by duration of TAM use (<1, 1–4, /≥5 years) overall and according to ER status of CBC
Bouchardy et al., 2011 [22]	Switzerland	Cohort	4152 (63)	Cancer Registry (Geneva)	Median: 5.2	Ever use of anti-estrogens	HR 0.51 (0.26–0.99)	Presumably mainly TAM since AIs were prescribed only from 2004 and the study period was 1995–2007 HRs for TAM use by ER status of CBC
Vichapat et al., 2012 [21]	Sweden	Cohort	35,003 (894)	Regional clinical database	Median: 6.7	Ever use of endocrine therapy	HR 0.78 (0.68–0.90)	Presumably mainly TAM as the study period was 1992–2008
Mellemkjaer et al., 2014 [18]	Denmark	Cohort	37,533 (1405)	National clinical database (DBCG)	Not reported	Ever use	Premenopausal patients during 1999–2007: HR 0.44 (0.27–0.71) Postmenopausal patients by calendar year: 1977–1989: HR 1.40 (1.12–1.74) 1990–1998: HR 0.82 (0.68–0.99) 1999–2007: HR 0.61 (0.44-0.84)	HRs for current/past use by duration (</≥1.2 years) and by time since last use. HRs by ER/PR status of first breast cancer
Langballe et al., 2016 [17]^a	Multi-center (WECARE)	Nested case-control	1521 CBC cases/2212 controls	Medical records	–	Ever use	RR 0.76 (0.63–0.92)	RRs by current/past use, duration (</≥54 months) and time since last use (</≥37 months) and combinations of these. RRs by ER status of first and second breast cancer
Gierach et al., 2017 [25]	USA	Cohort	7541 (248)	Health plan administrative database and medical records	Median: 6.3	Current use	RR^b 0.76 (0.64–0.89) per year of TAM use	RRs^b for current use and time since last use by duration of TAM (<1, 1–<4, and ≥4 years). RRs^b by ER status of CBC Absolute risk reduction among 5-year ER-positive breast cancer survivors for ≥4 years of tamoxifen versus no tamoxifen: 3.3% at 10 years and 4% at 15 years of follow-up
Kramer et al., 2019 [26]	Netherlands	Cohort	83,144 (2816)	National clinical database	Median: 7.7	Ever use	HR 0.48 (0.44–0.53)

AI: aromatase inhibitor; CBC: contralateral breast cancer; CI: confidence interval; EBCTCG: Early Breast Cancer Trialists’ Collaborative Group; DBCG: Danish Breast Cancer Group; ER: estrogen receptor; HR: hazard ratio; OR: odds ratio; PR: progesterone receptor; RR: rate ratio; SD: standard deviation; SE: standard error; SEER: Surveillance Epidemiology and End Results; TAM: tamoxifen; USA: United States of America.

^a Update of Bertelsen et al. (2008).

^b RR: relative risk.

The preventive effects of tamoxifen depend on duration of treatment [14,17,25,27–29]. Higher reductions in CBC risk were reported in clinical trials of 5 years of tamoxifen use compared with 1 or 2 years of use [27–29] (Table 1). Furthermore, the adjuvant tamoxifen longer against shorter (ATLAS) trial showed a reduced incidence of CBC among women allocated to continue tamoxifen therapy to 10 years rather than stopping at 5 years after breast cancer diagnosis [14]. Observational studies also suggest that tamoxifen is more effective with increasing duration of treatment [17,25].

The preventive effects of tamoxifen persist after discontinuation of treatment, that is, demonstrating a ‘carry-over effect’. Reduced CBC risk estimates up to 5 years after treatment cessation have been reported in clinical trials [8], although observational studies have shown varying lengths of protection after discontinuation of treatment [16,17,25]. Whether the ‘carry-over effect’ is more pronounced with increasing duration of treatment has been evaluated by two studies showing inconsistent results [17,25]. Thus, how duration of tamoxifen affects the length and magnitude of the carry-over effect is a topic deserving further study.

No difference in CBC risk was found according to ER status of the first breast cancer in trials initiated prior to 1990 [27] and in the multi-center WECARE Study [17]. However, a meta-analysis of more recent clinical trials reported reduced CBC risk only among patients with ER-positive first breast cancer [8]. Of concern, some observational studies have reported that tamoxifen use may increase the risk of ER-negative CBC [23,24], but these findings were based on a limited number of exposed cases and not supported by three recent observational studies [17,22,25].

Overall, current evidence indicates that long-term therapy with tamoxifen is more efficient than short-term therapy, and that a ‘carry-over’ effect exists.

Raloxifen

Raloxifen is a second generation SERM, originally developed for prevention of osteoporosis in postmenopausal women [30]. Although raloxifen has demonstrated efficacy in reducing the risk of primary breast cancer, has fewer side effects than tamoxifen, and hence was approved for chemoprevention by the FDA in 2007 [31,32], its effect on CBC risk has not been investigated yet.

Aromatase inhibitors

AIs block the enzyme aromatase, thereby stopping the conversion of androgen to estrogen. In postmenopausal women, estrogen production is primarily driven by the peripheral conversion of androgens, whereas estrogen in premenopausal women is mainly produced in the ovaries [33]. Recent clinical studies have demonstrated that postmenopausal AI therapy, either as single therapy or sequentially with tamoxifen, is more effective than tamoxifen therapy alone, in terms of both prolonged disease-free survival and lower recurrence rates [34]. As a result, third-generation AIs such as anastrozole, letrozole, and exemestane, comprise the ‘state of the art’ preferred adjuvant therapeutic option for postmenopausal women with ER-positive breast cancer [35].

Several, large RCTs have assessed the incidence of CBC in women treated with AIs versus tamoxifen (Table 2 and Figure 1). A meta-analysis of randomized trials by EBCTCG showed that 5 years of AI treatment was associated with a markedly reduced risk of CBC after 10 years of follow-up compared with 5 years of tamoxifen treatment (RR: 0.62; 95% CI: 0.48–0.80) [34]. Switching to AIs after 2–3 years of tamoxifen for the remainder of the 5-year adjuvant endocrine treatment period was also associated with reduced CBC risk compared with tamoxifen alone (HR: 0.67; 95% CI: 0.51–0.87) [34]. Recently, the results of the 10-year follow-up of the Intergroup Exemestane Study also pointed towards a beneficial effect on CBC associated with switching to AIs [38].

Table 2. Overview over clinical trials and observational studies on the association between aromatase inhibitors and risk of contralateral breast cancer.

Author, year	Country	Study design	Number of breast cancer patients (CBC cases)	Data source of drug exposure	Follow-up time (years)	Definition of use	Risk estimate (95% CI)	Remarks
EBCTCG, 2015 [34]	Multi-center	Meta-analysis of 9 RCTs	1) 9,885 (236) 2) 11,798 (226) 3) 12,799 (136)	Trial records	Mean: ∼5–10	1) 5 years AI vs. 5 years TAM 2) 2–3 years TAM + 2–3 years AI vs. 5 years TAM 3) 5 years AI vs. 2–3 years TAM + 2–3 years AI	1) RR 0.62 (0.48–0.80) 2) RR 0.67 (0.51–0.87) 3) RR 1.12 (0.80–1.57)	Only hormone receptor-positive first breast cancer patients
Goss et al., (2013) [36]	Multi-center	RCT	7576 (79)	Trial records	Median: 4.1	5 years of exemestane vs. 5 years of anastrozole	–	Only hormone receptor-positive first breast cancer patients Exemestane: 46 CBC cases Anastrozole: 33 CBC cases
Goss et al., 2016 [37]	USA/Canada	RCT	1918 (44)	Trial records	Median: 6.3	Additional 5 years of letrozole vs. placebo	HR 0.42 (0.22–0.81)	Only hormone receptor-positive and hormone receptor-unknown first breast cancer patients
Morden et al., 2017 [38]	Multi-center	RCT	4724 (131)	Trial records	Median: 10	2–3 years TAM + 2–3 years AI vs. 5 years TAM	HR 0.73 (0.52–1.03)	Only hormone receptor-positive and hormone receptor-unknown first breast cancer patients
Tjan-Heijnen et al., 2017 [39]	Netherlands	RCT	1912 (19)	Trial records	Median: 4.2	2–3 years TAM + 3 years anastrozole vs. 2–3 years TAM + 6 years anastrozole	–	Only hormone receptor-positive first breast cancer patients 3-year of AI: 15 CBC cases 6-year of AI: 4 CBC cases
Blok et al., 2018 [40]	Netherlands	RCT	1821 (37)	Trial records	Median: 6.6	Additional 5 years vs. 2.5 years of letrozole	HR 0.39 (0.19–0.81)	Only hormone receptor-positive first breast cancer patients Includes both new ipsilateral and CBC
Colleoni et al., 2018 [41]	Multi-center	RCT	4851 (44)	Trial records	Median: 5	Additional 5 years of continuous letrozole use vs. intermittent letrozole use	–	Only hormone receptor-positive first breast cancer patients Continuous AI use: 17 CBC cases Intermittent AI use: 27 CBC cases
Francis et al., 2018 [42]	Multi-center	RCT	4690 (36)	Trial records	Median: 8	Exemestane + OFS vs. TAM + OFS	–	Only premenopausal hormone receptor-positive first breast cancer patients Exemestane + OFS: 22 CBC cases TAM + OFS: 33 CBC cases
Gierach et al., 2017 [25]	USA	Cohort	7541 (248)	Health plan administrative database and medical records	Median: 6.3	1) Ever use of AI without TAM 2) Ever use of AI with TAM	1) RR^a 0.48 (0.22–0.97) 2) RR^a 0.86 (0.46–1.51)	RRs^a by ER status of CBC
Kramer et al., 2019 [26]	Netherlands	Cohort	83,144 (2,816)	National clinical database	Median: 7.7	1) Ever use of AI without TAM 2) Ever use of AI with TAM	1) HR 0.32 (0.23–0.44) 2) HR 0.45 (0.36–0.56)

AI: aromatase inhibitors; CBC: contralateral breast cancer; CI: confidence interval; EBCTCG: Early Breast Cancer Trialists’ Collaborative Group; ER: estrogen receptor; HR: hazard ratio; OFS: ovarian function suppression; RCT: randomized controlled trial; RR: rate ratio; TAM: tamoxifen; USA: United States of America.

^a RR: relative risk.

There is evidence supporting the benefit of extended AI use beyond the initial 5 years of endocrine therapy from the NCIC Clinical Trial Group (CTG) – MA.17R trial, that extended first-line endocrine therapy with 5 additional years of letrozole versus placebo, and reported a reduced incidence of CBC among women receiving letrozole [37]. In addition, the investigation on the duration of extended adjuvant letrozole treatment (IDEAL) trial found stronger risk reductions for second primary breast cancers for 5 additional years compared with 2.5 additional years of letrozole [40]. In a recent systematic review by Burstein et al., it was stated that ‘a substantial portion of the benefit from extended AI therapy was derived from prevention of second breast cancers’ [43].

Although not recommended in current guidelines, premenopausal women may also benefit from AI therapy when given in combination with ovarian suppression [44]. The suppression of ovarian function trial and tamoxifen and exemestane trial (TEXT) were designed to investigate the efficacy of 5 years of adjuvant exemestane plus ovarian suppression versus adjuvant tamoxifen plus ovarian suppression in premenopausal women with hormone receptor-positive breast cancer. In this study, fewer CBC cases occurred in the exemestane-ovarian suppression group as compared with the tamoxifen-ovarian suppression group [42].

To date, it remains unclear whether one AI is superior over another in CBC prevention. Although the NCIC CTG – MA.27 trial, comparing steroidal (exemestane) versus nonsteroidal (anastrozole) AIs, showed a slightly higher number of CBC cases among women treated with exemestane, future studies are needed before definite conclusions can be drawn [36].

A recent cohort study by Kramer et al. reported more pronounced reductions in CBC risk among patients who received AIs (HR: 0.32, 95% CI: 0.23–0.44) than among those who received tamoxifen (HR: 0.48, 95% CI: 0.44–0.53) when both were compared with no endocrine therapy (Table 2 and Figure 1) [26]. Previously, the first observational study on AIs by Gierach et al. reported a reduced CBC risk associated with use of AIs without tamoxifen (RR: 0.48, 95% CI: 0.22–0.97); however, not when administered in combination with tamoxifen (RR: 0.86, 95% CI: 0.46–1.51) (Table 2 and Figure 1) [25]. Moreover, the authors only observed a protective effect of AI against ER-positive CBC [25]. To date, no clinical trials have reported independent risk estimates by hormone receptor status of the contralateral tumor; however, this will be part of the ongoing Long-term Anastrozole versus Tamoxifen Treatment Effects study (NCT01745289). The finding in the study by Gierach et al. [25] is consistent with two trials, investigating first breast cancer risk among unaffected high-risk postmenopausal women, which reported a beneficial effect of AIs only for ER-positive first breast cancer [45,46].

Overall, AI therapy appears to be a good option for the prevention of CBC in postmenopausal breast cancer survivors with studies consistently showing more pronounced CBC risk reductions with AIs than with tamoxifen. Due to the relatively recent introduction of AI therapy in breast cancer patients, issues concerning preventive effects for CBC during treatment according to duration of use and persistence of these effects after discontinuation of AI therapy have not been investigated thoroughly. Future long-term follow-up of RCTs and well-designed observational studies with adequate sample size are needed to answer these questions.

Potentially preventive non-cancer agents

Some drugs initially developed for the prevention and treatment of non-cancer diseases received increased attention due to their potential anti-cancer properties observed in preclinical studies.

Bisphosphonates

Bisphosphonates are the first-line treatment for osteoporosis [47]. They are also effective in the management and prevention of metastatic bone disease and bone loss induced by breast cancer-specific treatments [48–51]. There is a consensus that bisphosphonates should be recommended as part of the adjuvant breast cancer treatment among high-risk postmenopausal women, or among premenopausal women receiving ovarian suppression therapy [52,53]. More recently, several preclinical in vitro and in vivo studies have demonstrated that bisphosphonates may exhibit anti-tumor activities, involving pro-apoptotic and anti-proliferative pathways [54,55]. Moreover, bisphosphonates can indirectly affect the bone marrow microenvironment, the levels of cytokines and growth factors, and tumor angiogenesis [54,55]. The most recent meta-analysis by Ou et al., based on more than 65,000 breast cancer patients from four case-control and four cohort studies, reported a 16% reduction in risk of first primary breast cancer associated with bisphosphonate use (95% CI: 0.77–0.90) [56].

The existing evidence on the association between bisphosphonate use and CBC risk stems from several clinical trials and two observational studies (Table 3 and Figure 1). In the EBCTCG meta-analysis involving 18,766 breast cancer patients from 26 individual bisphosphonate trials [57], adjuvant bisphosphonate treatment was not associated with a reduced CBC risk (RR: 0.96, 95% CI: 0.74–1.25) [57]. The majority of patients included in these trials had ER-positive first breast cancer and were accordingly treated with tamoxifen or AI. Therefore, a potential association with use of bisphosphonates may have been masked by the established CBC preventive effects of endocrine therapy.

Table 3. Overview of clinical trials and observational studies on the association between non-cancer drugs and risk of contralateral breast cancer.

Drug	First author, year	Country	Study design	Number of breast cancer patients (CBC cases)	Data source of drug exposure	Definition of use	Follow-up time (years)	Risk estimate (95% CI)	Remarks
Bisphosphonates	EBCTCG, 2015 [57]	Multi-center	Meta-analysis of 26 RCTs	18,766 (254)	Trial records	Ever use	Median: 5.6	RR 0.96 (0.74–1.25)
	Monsees et al., 2011 [58]	USA	Case-control	351 CBC cases/652 controls	Medical records	Ever use	–	OR 0.53 (0.30–0.93)	Only hormone receptor-positive first breast cancer patients ORs by current/former use and duration of use (≥6, ≥12, and ≥24 months)
	Kwan et al., 2016 [59]	USA	Cohort	16,781 (494)	Health plan administrative database and medical records	≥2 prescriptions	Mean: 6.4	HR 0.96 (0.67–1.38)	Only hormone receptor-positive first breast cancer patients treated with tamoxifen HRs by duration of use (<1, 1–<2, and ≥2 years)
Statins	Ahern et al., 2011 [60]	Denmark	Cohort	18,769 (520)	Prescription registry	≥1 prescription of simvastatin	Median 6.8	HR 0.54 (0.33–0.90)
	Boudreau et al., 2014 [61]	USA	Cohort	4216 (143)	Health plan administrative database	≥1 prescription of statins	Median 6.3	HR 0.92 (0.56–1.52)	HR by type of statin and duration of use (<1, 1–2.9, +3 years)
	Langballe et al., 2018 [62]	Denmark	Cohort	52,723 (1382)	Prescription registry	≥2 prescription of statins	Median: 4.8	HR 0.88 (0.73–1.05)	HRs by current/past use, duration (</≥5 year), intensity (</≥1 DDD), consistent/inconsistent use, and type of statin. Ever use by ER status of first breast cancer and CBC
Metformin	Oppong et al., 2014 [63]	USA	Cohort	141 (7)	Medical records	Ever use within 6 months prior to baseline	Median: 7.3	–	CBC rate at 5 years Metformin: 3.4 (95% CI: 0–7.9) No metformin: 7.9 (95% CI: 0.1–15.1)
	Calip et al., 2015 [64]	USA	Cohort	4216 (153)	Health plan administrative database	≥1 prescription	Median: 6.3	HR 1.52 (0.58–3.98)	Combined HR for ipsilateral and CBC
NSAIDs	Bens et al., 2018 [65]	Denmark	Cohort	52,723 (1444)	Prescription registry	≥2 low-dose aspirin prescriptions	Median: 4.8	HR 0.91 (0.76–1.09)	HRs by current/past use, duration, consistent/inconsistent use, and tablet strength. Ever use by ER status of first breast cancer and CBC
	Bens et al., 2018 [66]	Denmark	Cohort	52,723 (1,444)	Prescription registry	≥2 non-aspirin NSAID prescriptions	Median: 4.8	HR 0.98 (0.87–1.11)	HRs by duration, intensity, and COX-2 selectivity. Ever use by ER status of first breast cancer and CBC
Anti-hypertensive drugs	Boudreau et al., 2014 [61]	USA	Cohort	4216 (143)	Health plan administrative database	≥1 prescription	Median: 6.3	ACE inhibitors: HR 1.66 (1.06–2.58) Beta blockers: HR 0.77 (0.50–1.19)
	Chen et al., 2015 [67]	USA	Case control	359 CBC cases/691 controls	Medical records	≥6 months	–	ACE inhibitors: OR 1.2 (0.9–1.8) Beta-blockers: OR 1.0 (0.7–1.4)	Only hormone receptor-positive first breast cancer patients
Fenretinide	Veronesi et al., 2006 [68]	Italy	RCT	1739 (148)	Trial records	∼5 years vs. none	Median: 14.6	HR 0.90 (0.65–1.26)	HRs by menopausal status

CBC: contralateral breast cancer; CI: confidence interval; DDD: daily defined dose; EBCTCG: Early Breast Cancer Trialists’ Collaborative Group; ER: estrogen receptor; HR: hazard ratio; OR: odds ratio; RCT: randomized controlled trial; RR: rate ratio; USA: United States of America.

A nested case-control study including 351 CBC cases and 662 controls, showed that ever use of any bisphosphonate was associated with a substantial reduction in CBC (OR: 0.53, 95% CI: 0.30–0.93) [58]. Furthermore, the risk decreased with increasing duration of bisphosphonate use among current users (per year: OR: 0.71, 95% CI: 0.52–0.96) [58]. In contrast, a more recent US cohort study did not support an inverse association between bisphosphonate use and CBC risk (HR: 0.96, 95% CI: 0.67–1.38) [59]. Both observational studies were conducted solely among hormone receptor-positive breast cancer patients, and the question remains whether bisphosphonates could affect CBC development in hormone receptor-negative breast cancer patients.

Statins

Statins (HMG-CoA reductase inhibitors) are cholesterol-lowering drugs that are effective in preventing cardiovascular disease [69]. Preclinical studies of breast cancer cell lines and animal models have suggested that statins also impede tumor initiation, growth, and metastasis [70–72]. In particular, the lipophilic statins have shown stronger anti-neoplastic effects in ER-negative cell lines [71,73]. However, meta-analyses of RCTs and observational studies investigating overall statin use and incidence of first breast cancer have consistently reported neutral associations [74–77]. The most recent meta-analysis from 2017 [77] included sub-group analyses by statin solubility and duration of use, and suggested a reduced breast cancer risk associated with long-term use of any statin (RR: 0.74; 95% CI: 0.52–1.04) and the highly lipophilic statin, simvastatin (RR: 0.90; 95% CI: 0.80–1.01). The few breast cancer incidence studies, specifically investigating lipophilic statins, overall [78] or by duration [79] according to ER status reported associations close to unity.

To date, only three cohort studies have examined the association between statin use and risk of CBC [60–62] (Table 3 and Figure 1). The first study including 18,769 women diagnosed with primary breast cancer from the Danish Breast Cancer Group (DBCG) database reported a 46% reduced risk of CBC (95% CI: 0.33–0.90) associated with use of simvastatin [60]. The second study included 4216 women from the US with primary breast cancer and had limited statistical precision [61]. The most recent study, including 52,723 breast cancer patients from the DBCG database within an extended period, presented detailed information on timing, type of statins, and patterns of use as well as ER status [62]. In this study, ever use of statins was associated with a slight reduction in CBC risk (HR: 0.88; 95% CI: 0.73–1.05) together with a more marked risk reduction for long-term statin use (HR: 0.64; 95% CI: 0.43–0.96). However, this risk reduction was not seen among long-term high intensity users or consistent users. A substantial reduction in CBC risk was observed among patients diagnosed with ER-negative first breast cancer (HR: 0.67; 95% CI: 0.45–1.00).

Thus, current evidence on statins and CBC risk is sparse, but there is a suggestion of a reduced CBC risk associated with statin use [60,62], particularly among patients diagnosed with ER-negative breast cancer [62]. Future observational studies are warranted to establish whether statins could be useful for preventing a new CBC, especially among women with ER-negative disease.

Metformin

Metformin, a widely used oral antidiabetic drug, has received substantial interest as a potential anti-cancer agent [80]. Preclinical studies suggest that metformin inhibits the proliferation of breast cancer cells and delays the development of mammary tumors in mice [81]. This contrasts to the anti-diabetic drug insulin, which has been found to increase breast cancer cell proliferation [82]. Recent meta-analyses of observational studies by Col et al. [83] and Zhang et al. [84] reported breast cancer risk reductions of 17% (95% CI: 0.71–0.97) and 6% (95% CI: 0.91–0.98), respectively, associated with metformin use among women without prior breast cancer.

Yet, only two studies have investigated the influence of metformin use on second breast cancer events including CBC (Table 3 and Figure 1). A US cohort study by Oppong et al. including breast cancer patients with type II diabetes reported lower CBC incidence among metformin users compared with non-users. However, the study was too small to provide an appropriate risk estimate with only seven CBC cases occurring during follow-up [63]. A larger US cohort study by Calip et al. did not observe an association between metformin use and risk of a second primary breast cancer including both contralateral and ipsilateral cases [64]. Any cancer preventive effects of metformin might be masked if used in combination with insulin. In the available studies, Calip et al. adjusted for insulin in their multivariable model [64], while in the study by Oppong et al. the proportion of combined users was small [63]. The ongoing phase III study (NCIC CTG – MA.32 trial) investigating the effect of metformin versus placebo in addition to standard adjuvant therapy, may provide further knowledge on the association between metformin use and CBC risk.

Non-steroidal anti-inflammatory drugs (NSAIDs)

Aspirin and non-aspirin NSAIDs, including traditional nonselective NSAIDs and cyclooxygenase (COX)-2 inhibitors, are widely prescribed drugs with cancer preventive properties. These drugs inhibit the activity of the COX enzymes, which are important in platelet function and in the production of inflammatory prostaglandins [85]. Various in vitro and in vivo studies have shown that aspirin and non-aspirin NSAIDs inhibit the proliferation of breast cancer cells and suppress tumor growth [86,87]. The most recent meta-analyses reported slightly decreased breast cancer risk associated with aspirin [88,89] and in one of these with COX-2 inhibitors [88], but not with overall non-aspirin NSAID use [88].

The effect of aspirin and other NSAID use on CBC risk has been studied in two Danish cohort studies (Table 3 and Figure 1). Among 52,723 breast cancer patients, HRs for CBC of 0.91 (95% CI: 0.76–1.09) and 0.98 (95% CI: 0.87–1.11) were observed for low-dose aspirin (75–150 mg) and non-aspirin NSAID use, respectively [65,66]. The ongoing Add-Aspirin trial (NCT02804815), assessing the efficacy of adjuvant aspirin (100 mg or 300 mg versus placebo) in breast cancer patients, may provide further understanding of the potential of aspirin in prevention of CBC among breast cancer survivors.

Antihypertensive medications

Beta-blockers (β-blockers), angiotensin converting enzyme (ACE) inhibitors, and angiotensin receptor blockers (ARBs) are widely used drugs in the treatment of hypertension and heart disease. Laboratory studies of human cancer cell lines and animal models have demonstrated anti-proliferative and anti-angiogenic properties of propranolol, a nonselective β-adrenergic receptor antagonist [90]. ACE-inhibitors and ARBs may exert anti-neoplastic effects by inhibiting the renin-angiotensin system and its downstream signaling proteins such as vascular endothelial growth factor and the transcription factor, NFκB [91,92]. Despite the promising laboratory results, observational studies on risk of first breast cancer with anti-hypertensive use have provided mixed results [93].

Two US-based observational studies investigated the association between anti-hypertensive use and CBC risk (Table 3 and Figure 1). Both studies were small including 143 and 352 CBC cases, respectively [61,67]. The first study based on a cohort of 4,216 breast cancer patients suggested a decreased risk of CBC associated with β-blocker use (HR: 0.77; 95% CI: 0.50–1.19), whereas ACE-inhibitors use was associated with an increased risk of CBC (HR: 1.66, 95% CI: 1.06–2.58) [61]. The second study [67] reported no reduced risk with use of β-blockers (OR: 1.0, 95% CI: 0.7–1.4), while the association for ACE inhibitors (OR: 1.2, 95% CI: 0.9–1.8) was consistent with the first study.

Retinoids

Fenretinide, a synthetic derivative of retinoic acid, inhibits breast cancer development in animal models [94]. In an RCT, Veronesi et al. assessed the efficacy of 5-year treatment with fenretinide in preventing second breast malignancies in women with early breast cancer (Table 3) (Figure 1). Overall, no reduction in CBC risk was observed in a post-intervention study with a median follow-up of 14.6 years (HR: 0.90, 95% CI: 0.65–1.26) [68]. A beneficial effect of fenretinide on CBC risk was, however, suggested in premenopausal women (HR: 0.63, 95% CI: 0.38–1.04) irrespective of the hormone receptor status of the first breast cancer [68].

Discussion

Short summary of the results

Research in the field of preventive therapy for breast cancer has made important discoveries during the past decades; however, CBC has received limited focus. In this review, we summarized the available literature regarding the association between various drugs and CBC risk among breast cancer patients. To date, the most favorable candidates for prevention of CBC are the endocrine therapies, tamoxifen, and AIs. Widespread use of these drugs among patients with ER-positive first breast cancer has possibly led to declining CBC rates [95], whereas limited therapeutic options, and thus CBC prevention, are available for patients with ER-negative first breast cancer.

Various drugs, originally developed for other purposes, have shown preventive effects against breast cancer in laboratory and/or observational studies. Among these, statins show perhaps the most promising results supporting a reduction in CBC risk [60,62]. Moreover, mounting evidence supports reduced mortality [96] and risk of recurrence [96,97] associated with statin use following a breast cancer diagnosis. A number of pilot trials (planned or ongoing) are currently listed at www.clinicaltrials.gov with the primary aim of investigating the effect of statins on several breast cancer biomarkers [72]. However, before statins can be approved for use in breast cancer patients, clinical trials in the adjuvant setting for breast cancer are needed [70]. There is some evidence supporting a reduced CBC risk associated with use of the other drugs summarized in this review; however, more studies are needed to explore these further.

RCTs versus observational studies

The evidence for potential preventive drug effects against cancer comes from two main types of studies, RCTs and observational studies. Randomized trials are often considered the optimal design; however, each study type has their own strengths and weaknesses. Randomization in trials ensures similarities between groups with respect to both measured and unmeasured confounding factors. Theoretically, study groups only differ in the type of treatment they receive, allowing causal conclusions to be drawn. On the other hand, RCTs often apply strict in- and exclusion criteria. Thus, a trial population is selected and may have limited generalizability. Data from clinical trials on CBC risk are only available for tamoxifen, AIs, and bisphosphonates. Although the definition of recurrences often includes new contralateral tumors, some of the trials investigating breast cancer recurrence did not report separate risk estimates for CBC, or had limited information on tumor characteristics of the contralateral tumors. Moreover, many trials had limited study size and length of follow-up, and were thus underpowered to provide meaningful CBC risk estimates. Observational studies can be valuable in examining the effects of preventive drugs on long-term outcomes in large human populations. However, these studies are prone to confounding, particularly confounding by indication as well as other drug exposures, lifestyle or socio-economic factors that are often challenging to account for. As for RCTs, a limited number of observational studies on recurrences focused specifically on the association with CBC. Results from RCTs and observational studies can be difficult to compare as duration of use and definitions of current and past use often vary. However, when large datasets are available, observational studies can also examine various levels of duration and time since last use. Thus, although RCTs are the gold standard for establishing causal conclusions, observational studies provide much needed information on the preventive effects in real-life settings.

Limitations of drug preventive therapy

Preventive drug therapy, while promising, is also not without limitations. First, the efficacy of tamoxifen in reducing recurrence, mortality, and CBC risk is highly dependent on patient adherence. The most recent systematic review reported discontinuation rates of up to 73% at the end of 5 years of endocrine treatment [98]. A small UK trial observed high compliance rates for bisphosphonate treatment with more than 90% of the breast cancer patients taking their monthly doses [99]. Although not investigated specifically in breast cancer patients, studies on statin adherence have provided inconsistent results [100–102]. The data sources used to define adherence differ in the literature, from self-report by patients, drug concentration assessment in serum, to medical records review. Adherence is influenced by multiple factors such as patient, health and physician-related factors, and side effects. In contrast to observational studies, breast cancer patients enrolled in clinical trials are often highly motivated to participate and are closely monitored, which positively influences adherence. Therefore, the drug effects seen in RCTs may not be replicated in real-life observational studies simply due to lower adherence.

Breast cancer patients treated with tamoxifen or AIs suffer from side effects ranging from relatively mild menopausal symptoms to more severe adverse events. Side effects are also seen for non-cancer drugs. Detailed information on side effects of the drugs discussed in this review is presented in the Supplementary Material.

The use of preventive drugs should be considered individually by balancing their benefits and toxicity. For most breast cancer patients, the absolute risk of CBC is low. Consequently, the large reductions in relative risks of CBC observed for adjuvant endocrine therapy can be misleading, since the absolute risk reductions are small. In the clinical guidelines for breast cancer adjuvant endocrine therapy, the primary consideration is risk of recurrence. For patients at low-risk of recurrence, currently not treated with adjuvant therapy [103], use of endocrine therapy may not be justified on the benefits of CBC prevention alone. Thus, it is important to identify subgroups of women with breast cancer at high-risk of CBC, who are likely to benefit most from preventive therapy.

Personalized preventive therapy

Future studies should focus on developing risk prediction models to identify high-risk patients [104]. In that context, we need to identify biomarkers that could be included in these models to more accurately predict the risk of CBC. Biomarkers may ultimately be used to predict whether a woman will benefit from a specific treatment or not thereby personalizing preventive therapy for breast cancer survivors. Women receiving such personalized treatment may also have a greater willingness to accept possible side effects related to preventive drugs.

In this review, we primarily focused on preventive drugs and CBC development among breast cancer patients, but CBC has been proposed to serve as a surrogate marker for development of first primary breast cancer. For example, the first evidence for tamoxifen as a preventive agent in healthy women at high-risk of breast cancer originates from trials of breast cancer patients reporting reduced incidence of contralateral breast tumors [105]. Therefore, a favorable effect on CBC risk would provide a strong rationale for the effectiveness of a preventive drug in unaffected women at high-risk of first primary breast cancer.

Conclusion

Major progress has been achieved in the prevention of CBC mainly by incorporating adjuvant endocrine therapy with tamoxifen and AIs into routine clinical practice. Nevertheless, these drugs are associated with serious side effects and are not indicated for patients with ER-negative first breast cancer. At present, the effectiveness of other non-cancer agents in reducing CBC risk is limited. Yet, statins show promise, have fewer, and less severe adverse effects than endocrine therapies, and may be a successful treatment option for ER-negative breast cancer patients. For all preventive drugs, important questions regarding the optimal duration and dosing schedule require further investigation. Furthermore, better understanding of the anti-cancer mechanisms demonstrated in experimental studies and the identification of biomarkers will help us move forward towards more personalized medicine.

Disclosure statement

Lene Mellemkjaer has an immediate family member employed at Novo Nordisk, and has an immediate family member who owns stocks in Novo Nordisk. All other authors have declared no conflicts of interest.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by 'Danmarks Frie Forskningsfond' under grant number [11-108395].