Influence of the levonorgestrel-releasing intrauterine system on the risk of breast cancer: a systematic review

With 2.3 million new cases per year worldwide, BC is the most common cancer among women [1]. Fortunately, the mortality rate decreased due to implementation of screening and improved therapies during the last decades. Multiple risk factors are known including age, genetic factors, breast density, number of births, alcohol consumption, as well as endogenous and exogenous hormone exposure e.g. in the context of hormone replacement therapy (HRT) or contraception [2]. Regarding HRT, there is already a considerable amount of information available: long-term WHI data showed a significantly decreased BC risk with oestrogen-only HRT. Meanwhile, women who took combined oestrogen and progestogen HRT had an increased risk of being diagnosed with BC [3]. While the risk of HRT and also combined oral contraceptives is relatively well understood, the effect of progesterone alone on the female breast remains controversial.

LNG-IUS is a contraceptive method based on progesterone but is also used for treatment of menorrhagia or endometrial protection in postmenopausal women undergoing oestrogen therapy. It has been shown that progestins combined with oestrogens substantially increase BC risk [4]. Even though the amount of levonorgestrel released by LNG-IUS is low and seems to achieve the lowest systemic progestin levels of all progestin-only methods and most common contraceptives [5, 6], side effects after insertion of LNG-IUS include amongst others breast tenderness [6]. A possible influence on the breast and therefore on the BC risk cannot yet be ruled out with certainty.

Since there are several studies with inconsistent results addressing the possible influence of LNG-IUS on BC, the objective of this systematic review was to collect and analyze the existing literature concerning this matter.

To evaluate the association between the use of LNG-IUS and BC risk, a systematic review was performed. The screening and exclusion process was based on the PRISMA 2020 guidelines. The literature was peer-reviewed using the exclusion criteria.

All studies associated with BC risk and the use of LNG-IUS were included. Only articles in English, German or French were revised.

Duplicates as well as papers and reports without original data were considered ineligible and were excluded from the search. Data from patients known to be at high risk for developing BC were also not considered.

To identify all potentially relevant documents on the topic, systematic literature searches were designed and executed for the following information sources: besides the standard medical bibliographic databases Medline, Embase and the Cochrane Library, CINAHL and one interdisciplinary database, Web of Science, were searched. In addition, ClinicalTrials.gov, a database of clinical trials, was checked for unpublished trials on the topic. All searches were run on February 24th, 2021.

Search terms were identified by looking at subject headings, titles, abstracts and author keywords from a list of core references. An initial search strategy in Medline was drafted by a medical information specialist and tested against these core references to see if they were included in the search results. After refinement and consultation, search strategies were set up for each information source based on database-specific controlled vocabulary (thesaurus terms/subject headings) and textwords. No limits have been applied in any database considering study types, languages, publication years or any other formal criteria. Animal studies have been excluded.

The search concepts included were 1. “breast cancer” (in pre-, peri- and post-menopausal women) and 2. “levonorgestrel-releasing intrauterine systems”. Synonyms, acronyms and similar terms were used for all concepts in the textword search, as well as the thesaurus terms.

The final detailed search strategies are presented in the “Appendix”.

All identified citations were imported into EndNote and duplicates were removed. Subsequently, the screening of titles and abstracts as well as the eligibility assessment were performed. In the first screening process, all 326 studies were screened by title and abstract. 198 studies could thus already be excluded. In a second round, the remaining studies were tested against the inclusion criteria and assessed for eligibility. In the end, 10 studies were according to the four-eyes principle, considered eligible and therefore used for this systematic review (Fig. 1).

Overall, 326 studies were identified. Out of these, 10 studies [7‐16] met the inclusion criteria and were thus selected (Table 1).

The other studies were excluded due to e.g. focusing on contraception for women with a history of BC or discussing HRT for BC patients.

Six of the selected studies were cohort studies [8‐11, 13, 16], three were case–control studies [12, 14, 15] and one a systematic review/meta-analysis [7]. Only one study was retrieved from Embase/Ovid [12], the rest was found in the database Medline [7‐11, 13‐16].

Most of the studies were performed in Finland [11‐13, 15, 16]. One additional case–control study contained data from Finland as well as from Germany [14]. One cohort study stemmed from Norway [9], one from Denmark [10] and one from Israel [8]. The systematic review/meta-analysis was conducted in Brazil [7].

Primary endpoints of all studies were the occurrence of BC [7, 8, 10‐12, 14‐16], except for two studies where they further focused on other cancer risks such as ovarian and endometrial [9] as well as other types of cancer [13].

The sample size ranged from 17.360 [16] to 1.797.932 [10] women, mean time of follow up was 6.5 years [8] up to 12.5 years [9]. The studies assessed both pre- and postmenopausal women, as the youngest participants were only 15 years old at the time of study entry [10], the oldest 76 years [9]. Six out of the ten selected studies focused more on pre- or perimenopausal women [8, 10, 11, 13, 14, 16], two more on postmenopausal women [9, 15] and two included both age categories [7, 12].

The data on LNG-IUS use and BC incidence was obtained through self-administered questionnaires [9, 12, 14, 16], medical records/nationwide registries [8‐15] or study databases [7].

Six studies found a positive association between the use of LNG-IUS and BC [7, 10‐13, 15]. One study only stated a slightly increased risk for invasive BC in the subgroup of women aged 40–45 years [8]. Three studies observed no correlation between LNG-IUS and BC [9, 14, 16].

This systematic review showed that (1) the use of LNG-IUS seemed to increase BC risk, (2) with ORs up to 1.52; 95% CI 1.34–1.72 [7], this elevated risk was more evident in postmenopausal women and with longer durations of use and (3) confounding factors could have influenced the results.

Six of the ten studies included in this review stated a positive association between LNG-IUS and BC, although the risk increase was small. However, the included studies were very heterogenous. As each looked at various study cohorts with women at a different age and using LNG-IUS for different reasons (e.g. contraception, treatment for menorrhagia, HRT during menopause), it was difficult to compare them directly and draw clear conclusions. It would be necessary to verify the results in further studies, taking possible confounders into account. An extensive comparison to other hormonal contraceptive methods and their respective BC risk would also be interesting.

The BC risk increase becomes more evident in postmenopausal women and ranged up to OR = 1.52; 95% CI 1.34–1.72, I² = 0%, p = 0.84 [7]. In combination with oestradiol within the framework of postmenopausal HRT it was even increased to OR = 2.15; 95% CI 1.72–2.68 [15]. The LNG-IUS could have influenced this finding, but it could also be distorted by other risk factors which become more numerous with higher age and thus increase the risk for BC in postmenopausal women.

Longer durations of use also seem to be a risk factor. After two or more purchases of an LNG-IUS, the SIR was up to 1.73; 95% CI 1.37–2.15, p < 0.001 [11].

Nonetheless, the selected studies were probably susceptible to confounding factors. Since LNG-IUS is used as part of HRT, a confounder might have been the fact that HRT increases the risk for BC [17, 18]. Among current users of HRT, adjusted RR = 1.66; 95% CI 1.58–1.75, p < 0.0001 [18]. However, past users of HRT were no longer exposed to this risk [17, 18]. As some of the study cohorts used LNG-IUS in addition to HRT, it could very likely have influenced the results. Moreover, LNG-IUS is therapeutically used against abnormal bleeding and menorrhagia, which is more common in obese women [19] who are also at higher risk for BC [20]. Another confounder could have been a selection bias because LNG-IUS is more often prescribed to women with a family history of BC [12].

Apparently, the prevalence of BC risk factors in LNG-IUS users is higher than in non-users. For example, they are more likely to have used previous hormonal contraception [9] and have a higher socioeconomic status [12, 14].

Not only does the risk of BC seem to be increased, but other types of cancer also appear to be influenced by LNG-IUS: with 188 more observed than expected cases, the overall cancer risk was elevated [13]. On a positive note, LNG-IUS seemed to significantly decrease other hormonal cancer risks. Two studies found a reduced risk for ovarian and endometrial cancer [9, 13]. This finding was supported by previous studies, where the LNG-IUS seemed to have an antiproliferative and protective effect on the endometrium [5, 21]. On the other hand, there was insufficient evidence that LNG-IUS decreases the occurrence of ovarian cancer [22].

However, the individual results pointed towards an increased BC risk in LNG-IUS users, especially in postmenopausal women and with longer duration of use. This should particularly be considered in patients with a family history of BC or other risk factors present like obesity and higher age. Nonetheless, positive effects of the LNG-IUS such as a reduced risk for endometrial (and possibly ovarian) cancer and the high effectiveness as a contraceptive method should also be put into the balance and weighed against the probable increase in BC risk.

Moreover, other hormonal contraceptive methods, e.g. combined oral contraceptives (COC) seemed to have an increased BC risk too: in women who were currently using COC, RR of a BC diagnosis = 1.24 (95% CI 1.15–1.33, p < 0.00001) [23]. Consequently, the effects of LNG-IUS should be put into perspective and compared with other forms of contraception.

The results of our review show an increased BC risk in LNG-IUS users. The risk increase was especially marked in postmenopausal women and with longer duration of use. However, the findings must be considered with caution as included studies were very heterogenous and possibly influenced by many confounding factors. Nevertheless, the individual risk profile of patients should be taken in account for counselling. According to our results hormone-free alternatives may be discussed in women at high risk for BC.

Furthermore, the positive effects as risk reduction for other hormonal cancers, therapeutical and contraceptive benefits should also be included in the shared decision-making process.

Further well-designed studies with a focus on certain demographic and epidemiologic group are needed to explore the association between LNG-IUS and BC risk.