Confirmation of the safety of combined oral contraceptives containing oestradiol on the risk of venous thromboembolism

Hormonal contraceptives, in particular combined oral contraceptives (COCs), were designed to avoid unintended pregnancy, with the fewest possible side effects and health risks. The most important and relevant adverse effect of COC use, although very rare, is the occurrence of venous thromboembolism (VTE). The oestrogenic component of COCs, traditionally ethinylestradiol (EE), is largely responsible for the occurrence of VTE [1], and is modulated by the progestin.

Replacement of EE with 17β-oestradiol (E₂), the oestrogen naturally secreted by the ovaries, was challenging because of the difficulty in achieving satisfactory bleeding control. This was finally accomplished in 2009 when a quadriphasic combination of oestradiol valerate (E₂V) and dienogest (DNG) was introduced to the market in Europe and the USA. In 2012, a monophasic COC composed of micronised E₂ and nomegestrol acetate (NOMAC) was introduced to the European market [2]. Pharmacological data show that products containing E₂ should be associated with fewer cardiovascular events (both venous and arterial) in comparison with those containing EE [3]. To test this hypothesis in clinical practice, the occurrence of cardiovascular events is currently being compared in surveillance studies with a reference formulation, EE combined with levonorgestrel (LNG) [4].

The 2016 results of the first epidemiological studies of these E₂-based formulations were highly encouraging [5,6] (updated results were published in 2020 [7]). The International Active Surveillance Study – Safety of Contraceptives: Role of Oestrogens (INAS-SCORE) [5,7] investigated the cardiovascular risks associated with the use of the combined hormonal contraceptive E₂V/DNG. The final report described the results of 10,191 E₂V/DNG users compared with those of 41,012 women using other EE-based COCs (5796 of whom were using EE/LNG-based products). Study participants from both Europe and the USA were followed for a mean of 2.9 years for the occurrence of VTE, totalling 145,224 woman-years (Table 1). Now, just published in December 2021 (issue 6/volume 26 of The European Journal of Contraception and Reproductive Health Care [8]), Reed and colleagues, have reported similar results with a monophasic formulation containing E₂ and NOMAC. Their study results reflect those of INAS-SCORE, albeit with some general differences (as shown in Table 1). The Prospective Controlled Cohort Study on the Safety of a Monophasic Oral Contraceptive containing Nomegestrol Acetate (2.5 mg) and 17β-oestradiol (1.5 mg) (PRO-E2) carried out by Reed et al. [8] was designed to fulfil a post-marketing requirement of the European Medicines Agency and the primary outcome focussed on the risk of VTE of E₂/NOMAC compared with the regulatory gold standard LNG-containing COCs. In this study, in comparison with INAS-SCORE, a larger group of participants was followed for a shorter time resulting in a similar cumulative length of follow-up (Table 1): a group of 101,498 COCs users including 44,559 using E₂/NOMAC and 46,754 using EE/LNG were followed for a mean of 1.4 years (144,901 woman-years of observation) (Table 1). The shorter follow-up period (1.4 vs 2.9 years) was sufficient to highlight rare events such as VTE, which occur mainly in the first year of taking a COC [9]. The study was less suited, however, to uncovering effects that occur after a longer period, such as arterial cardiovascular events (e.g., stroke and myocardial infarction) [7]. These require a longer period of follow-up.

Table 1. Comparison of sample size, follow-up and results of the INAS-SCORE [5,7] and PRO-E2 [8] studies.

Variable	INAS-SCORE		PRO-E2
	E2-based COCs	EE-LNG-based COCs	E2-based COCs	EE-LNG-based COCs
Age at study inclusion, according to COC used, years	31.7	26.0	31.0	29.3
BMI, kg/m^2	23.0	24.2	23.2	23.3
First-time COC users, %	32.3		63.5
Mean follow-up, years	1.6	2.3	1.1	1.2
No. of study participants	10,191	5,796	44,559	46,754
Woman-years of follow-up	15,850	13,078	45,750	49,729
Crude no. of confirmed cases of VTE	11	13	9	15
Crude no. of confirmed and potential cases of VTE	NR	NR	20	28
No. of VTE events/10,000 users per year	6.9 (95% CI 1.9, 12.1)	9.9 (95% CI 5.5, 14.3)	2.0 (95% CI 0.9, 3.7)	3.0 (95% CI 1.7, 5.0)
Crude HR, E2 COCs vs EE/LNG COCs	0.70 (95% CI 0.30, 1.60)		0.65 (95% CI 0.28, 1.48)
Adjusted HR, E2 COCs vs EE/LNG COCs	0.40 (95% CI 0.20, 1.10)		0.59 (95% CI 0.25, 1.35)
VTE in E2 COC users	20 cases in 61,600 woman-years (3.2/10,000 users per year)
VTE in EE/LNG COC users	28 cases in 62,807 woman-years (4.5/10,000 users per year)
Pooled crude HR, E2 COCs vs EE/LNG COCs	0.73 (95% CI 0.41, 1.29)

NR: Not reported.

In both studies, women using E₂-containing COCs were significantly older at inclusion compared with those using LNG-containing COCs (INAS-SCORE, 31.7 vs 26.0 years; PRO-E2, 31.0 vs 29.3 years). Participants’ body mass index (BMI), however, was similar in the different treatment groups (Table 1). It is well known that VTE occurrence is maximal during the first months of COC use in first-time COC users [10]. The proportion of new and previous COC users differs between the two studies: in INAS-SCORE only 32.3% of the study population were first-time COC users at study entry, while in PRO-E2 63.5% were first-time COC users; however, this factor was not taken into account in analyses of the two studies.

The crude occurrence of VTE found in the two studies is as follows:

• INAS-SCORE: 11 cases in 15,850 woman-years of E₂V/DNG use (6.9/10,000 users per year) vs 13 cases in 13,078 woman-years of EE/LNG use (9.9/10,000 users per year).

• PRO-E2: 9 cases in 45,750 woman-years of E₂/NOMAC use (2.0/10,000 users per year) vs 15 cases in 49,729 woman-years of EE/LNG use (3.0/10,000 users per year) (Table 1).

The difference in the incidence of VTE in the two studies is quite surprising: the occurrence of VTE in INAS-SCORE was at least three times higher, in particular for the same EE/LNG-based COCs, compared with that in the PRO-E2 study (9.9 vs 3.0/10,000 users per year). The very low incidence rate in the PRO-E2 study is very similar to that traditionally reported in combined hormonal contraceptive non-use cohorts (2–3 cases/10,000 users per year) [4,7]. The methods for VTE case evaluation in the two trials were, however, similar: after the first report of an adverse event such as a VTE, three independent adjudicators (specialised in radiology/nuclear medicine, cardiology and internal medicine/phlebology), blinded to the type of hormonal treatment used and unaware of how the investigators had previously classified events, recorded VTEs as ‘confirmed’ or ‘not confirmed’ [5,7,8]. Only in the PRO-E2 study, however, was a number of ‘potential’ VTE cases reported, when a specific case could be not adjudicated because of insufficient information (e.g., medical documentation) (Table 1). Adding potential but unconfirmed VTE cases, the figures in PRO-E2 become more similar to those previously reported, although still very reassuring: for E₂/NOMAC, 4.1 per 10,000 woman-years (95% confidence interval [CI] 2.5, 6.3); for EE/LNG, 5.2 per 10,000 woman-years (95% CI 3.4, 7.5). Reed et al. [8] discuss other potential reasons for the lower incidence found in their study (e.g., very low occurrence in the Russian dataset), but these may not completely explain the full picture.

Despite the limitations of the two studies and some methodological differences, E₂-based COCs demonstrated in both studies a similar trend of non-inferiority in comparison with EE/LNG COCs. In fact, in both trials both the crude and adjusted VTE hazard ratios (HRs) ranged between 0.40 and 0.70, with CIs clearly containing unity (Table 1).

The data show a total of 20 VTE cases in 61,600 woman-years of E₂-containing COC use (3.2/10,000 users per year) in comparison with 28 VTE cases in 62,807 woman-years of EE/LNG use (4.5/10,000 users per year), with a cumulative pooled calculated VTE crude HR of 0.73 (95% CI 0.41, 1.29) (Table 1). These are the best available safety data on VTE risk of E₂-containing COCs with different progestogenic properties (NOMAC and DNG) in comparison with EE/LNG combinations.

Unfortunately, we do not have all the necessary data to correct the cumulative rates for other confounding variables (e.g., age, BMI, smoking) in the two different studies to obtain a cumulative pooled adjusted HR: the data might even indicate a significant decrease in thrombotic risk with E₂-containing COCs. Taking into account the age-related increased risk of VTE, the results are also important because they were obtained in women using E₂-based COCs who were 2–6 years older than the group using EE-LNG COCs [4,11]. Moreover, the introduction of other new natural oestrogenic components, such as estetrol (E₄) [12], could have a similar lower VTE impact; however, we will likely need another decade to obtain results from post-marketing studies.

The current data, yet to be completed with even longer follow-up studies to determine whether the differences found reach statistical significance, should be considered by regulatory authorities regarding risk classification.