Prospective controlled cohort study on the safety of a monophasic oral contraceptive containing nomegestrol acetate (2.5mg) and 17β-oestradiol (1.5mg) (PRO-E2 study): risk of venous and arterial thromboembolism

Abstract

Objective

To assess and compare the risk of venous thromboembolism (VTE) and arterial thromboembolism (ATE) in NOMAC-E2 users with levonorgestrel-containing combined oral contraceptive (COCLNG) users.

Study design

This large, prospective, observational active surveillance study used a non-inferiority design. New users of NOMAC-E2 and COCLNG were recruited in 12 countries in Australia, Europe, and Latin America. Women were followed up directly and self-reported outcomes of interest were validated via treating physicians. The main outcome of interest was VTE, specifically deep venous thrombosis of the lower extremities (DVT) and pulmonary embolism (PE). Secondary outcomes included all VTE and ATE. Data on confounders were captured and independent blinded adjudication assessed the classification of events. Incidence rates, crude (HRcrude), and adjusted (HRadj) hazard ratios were calculated.

Results

A total of 101,498 women (49,598 NOMAC-E2 users and 51,900 COCLNG users) were enrolled and followed for up to 2 years (144,901 WY of observation). NOMAC-E2 users had a higher mean age (31.0 ± 8.63 years) than COCLNG users (29.3 ± 8.53 years) but other baseline characteristics were similar between the cohorts. The main analysis comparing the risk of DVT of the lower extremities and PE in NOMAC-E2 users versus COCLNG users yielded an HRadj of 0.59 (95% CI, 0.25–1.35) (adjusted for age, BMI, family history of VTE, and current duration of use). The risk of all VTE and ATE was not higher in NOMAC-E2 users compared with COCLNG users.

Conclusion(s)

NOMAC-E2 use was not associated with a higher risk of VTE or ATE compared with COCLNG.

关于含有醋酸诺美孕酮（2.5毫克）和17β-雌二醇（PRO-E2研究）的单相口服避孕药的静脉和动脉血栓栓塞的风险 安全性前瞻队列研究 摘要

目的:评估和比较使用NOMAC-E2和含有左炔诺孕酮的口服避孕药(COCLNG)的静脉和动脉血栓栓塞的风险。

研究设计：这项大型、前瞻性、观察性的主动监视研究采用了非劣效性设计。在澳大利亚、欧洲和拉丁美洲的 12 个国家/地区进行招募服用NOMAC-E2 和 COCLNG 的新患者。妇女直接由医生随访, 患者自我报告的结果均有主治医生审核。关注的主要结果是VTE, 特别是下肢（DVT）和肺栓塞（PE）的深静脉血栓。次要结果包括所有 VTE 和 ATE。获取有关混杂因素的数据并进行独立的盲法评估。计算了发生率、粗略 (HRcrude) 和调整后 (HRadj) 风险比。

结果: 共有101498名妇女（49598名NOMAC-E2患者和51900名COCLNG患者）登记并跟踪了长达2年（观察144 901 WY）。服用NOMAC-E2的患者平均年龄（31.0 ± 8.63 岁）高于服用COCLNG 妇女（29.3 ± 8.53 岁）, 但其他基线特征在两组之间相似。主要分析比较了NOMAC-E2妇女与COCLNG妇女下肢DVT和PE的风险, 得出了0.59（95%置信区间, 0.25-1.35）的调整后风险比（根据年龄、BMI、VTE家族史和当前使用期限进行调整）。与 COCLNG 用户相比, NOMAC-E2 妇女的所有 VTE 和 ATE 的风险并没有升高。

结论: 与 COCLNG 相比, NOMAC-E2 的服用没有升高 VTE 或 ATE 的风险。

Keywords:

• NOMAC-E2
• nomegestrol acetate
• oestradiol levonorgestrel
• cardiovascular
• venous thromboembolism
• arterial thromboembolism

Introduction

Nomegestrol acetate (NOMAC) (2.5 mg) and 17β-oestradiol (E2) (1.5 mg) comprise a monophasic combined oral contraceptive (COC)¹ which is taken for 24 days followed by 4 days of placebo. NOMAC-E2 is a 19-norprogesterone-derived steroid and has a strong affinity for the progesterone receptor and strong anti-gonadotropic activity and progesterone receptor-mediated anti-oestrogenic activity, moderate anti-androgenic activity, and is devoid of oestrogenic, androgenic, glucocorticoid or mineralocorticoid activity [1]. The oestrogen contained in NOMAC-E2 is 17β-oestradiol, an oestrogen identical to the endogenous human 17β-oestradiol.

One of the serious adverse events (SAEs) associated with COC use is venous thromboembolism (VTE). Data from randomised clinical trials did not show serious health concerns for NOMAC-E2 but the statistical power to detect rare SAEs was limited [2–6]. Clinical experience suggests that serious clinical outcomes are unusual with COCs containing ethinylestradiol (EE). NOMAC-E2 has been previously reported to have a low impact on parameters of haemostasis, coagulation factors, lipids, carbohydrate metabolism, and blood pressure compared to the use of other COCs [5,7].

This large, controlled, prospective observational study was conducted to generate comprehensive data on cardiovascular and other health risks associated with the use of NOMAC-E2 in comparison with the use of levonorgestrel-containing combined oral contraceptives (COC_LNG)² during standard clinical practice. This study was designed to fulfil a post-marketing obligation to the European Medicines Agency (EMA) by the marketing authorisation holders.

Materials and methods

The primary objective of the PRO-E2 Study was to assess and compare the risk of cardiovascular events in NOMAC-E2 users with COC_LNG users in a study population representative of the actual users of COCs. The main clinical outcome of interest was VTE, specifically deep venous thrombosis (DVT) of the lower extremities and pulmonary embolism (PE) in women without pre-defined risk factors at study entry (i.e., pregnant within 3 months of treatment initiation, a history of cancer/chemotherapy or an increased genetic risk of VTE)³.

Secondary objectives included measuring the risk of all VTE, arterial thromboembolism (ATE), depressive disorders, cholelithiasis, inflammatory bowel disease, general hepatobiliary disorders, effects on fertility, pregnancy, pregnancy outcomes, weight change, and the effect on acne. The results involving cardiovascular outcomes are described in this paper.

This was a large, multinational, controlled, prospective, active surveillance study that followed new users (starters and restarters) of NOMAC-E2 and COC_LNG. Starters were first-ever users of any COC. Restarters were restarting COC use after a minimum break of 2 months. Ethical approval was obtained as required by local law and an independent Safety Monitoring and Advisory Council (SMAC) monitored the study.

Study population

Women were recruited by health care professionals (HCPs) under real-life clinical practice conditions in Australia, Austria, Colombia, France, Germany, Hungary, Italy, Mexico, Poland, Russia, Spain, and Sweden. The planned sample size was 101,000 women (50,500 NOMAC-E2 users and 50,500 COC_LNG users). All women newly prescribed an eligible COC⁴ could participate if they had not used a COC in the past 2 months, signed a consent form, and completed a baseline questionnaire in the local language.

Baseline survey and follow-up

Women completed a baseline questionnaire to record data on age, weight, height, previous contraceptive use, medical history, family history of VTE and ATE, medication use, acne status, mood, smoking status, and educational level. They were followed up directly at 6, 12, and 24 months to capture data on contraceptive use, pregnancy, the occurrence of the outcomes of interest, other severe diseases, hospitalisations, acne status, mood, and smoking status.

Loss of information was minimised by reminding study participants via telephone, email, or post and by contacting their friends or relatives (with their consent and based on details provided by the women at enrolment). Additionally, registered letters were sent, national address registries were searched, and face-to-face interviews were conducted.

To continuously monitor safety data during the study, self-reported outcomes of interest were reviewed by the investigators. If necessary, study participants and/or the treating physicians were contacted for further information. The investigators (who were not blinded to hormonal contraceptive use) classified events as ‘confirmed’ if they were confirmed by diagnostic measures with high specificity (e.g., phlebography for DVT, spiral computed tomography for PE, or cerebral resonance imaging for cerebrovascular accidents) or by a clinical diagnosis supported by a diagnostic test with low specificity (such as D-dimer for VTE or typical electrocardiogram/blood gas tests for PE). Events were classified as ‘not confirmed’ if the diagnosis reported by the study participant was excluded by diagnostic measures, a different medical condition was diagnosed or if the study participant did not contact an HCP and no diagnostic measures were performed.

Blinded adjudication

As described in Section 2.2, the investigators classified events as ‘confirmed’ or ‘not confirmed’ on an ongoing basis throughout the study to ensure the timely reporting of events for pharmacovigilance purposes. At the end of the study, the events were reassessed by three independent adjudicators (specialised in radiology/nuclear medicine, cardiology, and internal medicine/phlebology) who, blinded to the hormonal contraceptive (HC) brand names and compositions and unaware of how the investigators previously classified the events, classified VTEs and ATEs as ‘confirmed’ or ‘not confirmed’. The adjudicators reviewed all cases independent of each other before discussing split decisions. Following this process, events were considered ‘confirmed’ if at least one adjudicator classified them as ‘confirmed’. The classification by the independent adjudicators formed the basis of the final analysis. ‘Confirmed’ events were included in the primary analysis; sensitivity analysis was performed to include ‘not confirmed’ (but potential⁵) cases.

Evaluation

A non-inferiority design was used to assess the risk of VTE in NOMAC-E2 users compared with COC_LNG users. Based upon the VTE incidence observed in other studies with a similar methodology (i.e., prospective cohort studies) [8,9], a rate of 10 VTE per 10,000 woman-years (WY) was assumed. The recruitment of 101,000 women was planned to generate 150,000 WY of observation (average follow-up of 1.5 years) to exclude a 1.5-fold risk of VTE in users of NOMAC-E2 compared with COC_LNG.

Cox regression analyses were conducted for the primary outcome of DVT of the lower extremities and PE to yield crude hazard ratios (HR_crude) and adjusted hazard ratios (HR_adj). Models included an a priori expert model (containing the pre-defined confounders age, body mass index (BMI = kg/m²), family history of VTE and current duration of HC use) with and without multiple imputations of missing data, a saturated model (which included all potential confounders as co-factors) and an automated model selection procedure (using a backward elimination process).

All statistical analyses were conducted using SAS 9.4 [10].

Results

Patient recruitment started in August 2014 and ended in September 2019. Follow-up was completed in September 2020. Of the 124,919 women who agreed to participate, a total of 23,421 were excluded (primarily because they provided no contact details, had used a COC in the last 2 months, were not prescribed a COC or the name of their baseline prescription was unknown). Therefore, a total of 101,498 women were enrolled. Overall, 91,313 women (44,559 NOMAC-E2 users and 46,754 COC_LNG users) were followed during real-world use. Women in the COC_LNG cohort were further categorised as users of monophasic COC_LNG preparations containing 20 mcg EE (COC_LNGMono20mcg), monophasic COC_LNG preparations containing 30 mcg EE (COC_LNGMono30mcg), or multiphasic COC_LNG preparations (COC_LNGMulti) (Table 1).

Table 1. Number and percentage of study participants who agreed to participate, were excluded from the study, and included in the analysis (by cohort, user status and country).

Study participants	Number of women	% of women in final analysis
Agreed to participate	124,919	–
Excluded from study^a	23,421	–
ITT population	101,498	–
AT population	91,313	100
Cohorts (AT)
NOMAC-E2	44,559	48.8
COCLNG	46,754	51.2
COCMono20mcg	20,181	22.1
COCMono30mcg	17,469	19.1
COCMulti	9,104	10.0
User status (AT)
Starters	58,009	63.5
Restarters	33,304	36.5
Countries^b (AT)
Australia	535	0.6
Austria	462	0.5
Colombia	4,406	4.8
France	580	0.6
Germany	4,712	5.2
Hungary	9,407	10.3
Italy	19,683	21.6
Poland	6,263	6.9
Russia	36,092	39.5
Spain	8,656	9.5
Sweden	517	0.6

AT: “as-treated”; ITT: “intention-to-treat”; NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COC: combine oral contraceptive; COC_LNG: levonorgestrel-containing COC.

^aWomen were excluded primarily because they provided no contact details, had used a COC in the last 2 months, were not prescribed a COC or the name of their baseline prescription was unknown.

^bDue to non-compliance with study procedures, data contributed by 625 study participants in Mexico were excluded from the analyses (except a sensitivity analysis) as advised by the Safety Monitoring and Advisory Council. The local field organisation deviated from the required storage requirements and destroyed 545 of the informed consent forms. After recontacting the women, proof of consent was available for 235 women (38% of the Mexican study population). A sensitivity analysis of VTE risk with the inclusion of data from Mexico was conducted (see Table 4).

During follow-up, women could stop or switch COC use for any reason, and therefore, sub-cohorts formed: COC_Other (COCs other than NOMAC-E2 or COC_LNG), OHC (other hormonal contraceptives)⁶, and ex-users (no hormonal contraceptive use). Events were allocated to the cohort to which the woman belonged at the time of the event.

Overall, the real-world population contributed 144,901 WY: NOMAC-E2 (48,846 WY), COC_LNG (54,037 WY), COC_Other (8,300 WY), OHC (2,364 WY), ex-users (31,354 WY).

Baseline characteristics of user cohorts

Selected baseline characteristics are displayed in Table 2. NOMAC-E2 users had a slightly higher mean age (31.0 ± 8.63 years) than COC_LNG users (29.3 ± 8.53 years) but they were very similar in relation to other cardiovascular risk factors (Table 3).

Table 2. Number of women and woman-years (WY) of hormonal exposure and descriptive statistics of age, weight and BMI at study entry, by user (sub-)cohort.

		COCLNG
	NOMAC-E2	Monophasic 20 mcg EE	Monophasic 30 mcg EE	Multiphasic	All COCLNG	Total
No. of women [%]	44,559 [48.8]	20,181 [22.1]	17,469 [19.1]	9,104 [10.0]	46,754 [51.2]	91,313 [100]
WY [%]	48,846 [33.7]	24,429 [16.9]	19,413 [13.4]	10,119 [7.0]	54,037 [37.3]	144,901^a [100]
Age, mean [SD]	31.0 [8.63]	27.9 [8.69]	30.2 [8.29]	30.9 [8.15]	29.3 [8.53]	30.1 [8.62]
Age, median [range]	30.3 [11–67]	26.1 [11–62]	29.7 [12–60]	30.6 [11–65]	28.4 [11–65]	29.4 [11–67]
Weight (kg), mean [SD]	63.3 [11.67]	61.6 [12.16]	64.2 [12.37]	64.2 [11.34]	63.1 [12.15]	63.2 [11.92]
BMI, mean [SD]	23.2 [4.07]	22.9 [4.28]	23.6 [4.32]	23.5 [3.96]	23.3 [4.25]	23.3 [4.16]

^aIncludes WY contributed by women who stopped using contraceptives or switched to other contraceptives during the follow-up period.

NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COC_LNG: levonorgestrel-containing combine oral contraceptive; WY: Women-years; SD: Standard deviation; BMI: Body mass index.

Table 3. Prognostic factors for cardiovascular outcomes of interest at study entry: Total number and percentage of women by user cohort.

Factor	NOMAC-E2 (44,559 women)	COCLNG (46,754 women)	Total (91,313 women)
	n (%)	n (%)	n (%)
Current smoker	9,035 (20.3)	10,315 (22.1)	19,350 (21.2)
Heavy smoker^a	1,498 (3.4)	1,394 (3.0)	2,892 (3.2)
BMI < 20	8,763 (19.7)	9,525 (20.4)	18,288 (20.0)
BMI ≥ 20 to < 25	22,726 (51.0)	23,255 (49.7)	45,981 (50.4)
BMI ≥ 25 to < 30	9,244 (20.7)	9,746 (20.8)	18,990 (20.8)
BMI ≥ 30 to < 35	2,113 (4.7)	2,474 (5.3)	4,587 (5.0)
BMI ≥ 35	609 (1.4)	796 (1.7)	1,405 (1.5)
Deep venous thrombosis	55 (0.1)	52 (0.1)	107 (0.1)
Pulmonary embolism	1 (0.002)	7 (0.01)	8 (0.009)
Myocardial infarction	5 (0.01)	9 (0.02)	14 (0.02)
Stroke	9 (0.02)	11 (0.02)	20 (0.02)
High blood pressure (treated)	449 (1.0)	377 (0.8)	826 (0.9)
Other serious diseases	1,021 (2.3)	1,130 (2.4)	2,151 (2.4)
Any surgery	10,054 (22.6)	10,228 (21.9)	20,282 (22.2)
Family history of VTE	1,096 (2.5)	1,117 (2.4)	2,213 (2.4)
Family history of ATE	896 (2.0)	850 (1.8)	1,746 (1.9)

^a>15 cigarettes per day.

NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COC_LNG: levonorgestrel-containing combine oral contraceptive; BMI: Body mass index; VTE: Venous Thromboembolism; ATE: Arterial Thromboembolism.

Russia contributed a substantial proportion of study participants (39.5%) to the overall study population, and this warranted a comparison of their characteristics with non-Russian participants. The women from Russia had a higher mean age (32.3 ± 7.49 years compared with 28.7 ± 9.01 years for non-Russians) but other baseline risk factors were largely similar between Russians and non-Russians: mean BMI (23. 58 ± 3.828 and 23.05 ± 4.357, respectively), parity status (94.8% and 86.8%, respectively), personal history of DVT (0.08% and 0.1%, respectively) and PE (0.008% and 0.009%, respectively), family history of VTE (1.6% and 2.9%, respectively), personal history of stroke (0.02% in each cohort) and myocardial infarction (0.02% and 0.01%, respectively), treated high blood pressure (0.5% and 1.2%, respectively) and family history of ATE (1.6% and 2.1%, respectively).

Loss to follow-up

At the end of the study, a total of 16,528 women (16.3% of the study population) were lost to follow-up. This included 8,137 NOMAC-E2 users (16.4%) and 8,391 COC_LNG users (16.2%).

A total of 15,941 women (15.7% of the study population) who were not lost to follow-up dropped out of the study after indicating they no longer wished to participate. The lost-to-follow-up rate differed only marginally between the user cohorts: 7,457 NOMAC-E2 users (15.0%) and 8,484 COC_LNG users (16.3%).

VTE

DVT of the lower extremities and PE

The primary outcome was DVT of the lower extremities and PE in women without pre-defined risk factors at study entry (pregnant within 3 months of treatment initiation, a history of cancer/chemotherapy, or an increased genetic risk of VTE). The exclusion of these women from the analysis decreased the number of WY from 144,901 to 133,544. Over the course of 133,544 WY of follow-up, 34 confirmed VTEs (comprising DVT of the lower extremities and PE) were observed and included in the primary analysis. The numbers of VTE and incidences per 10,000 WY by user cohort are shown in Table 4. Characteristics of the study participants who experienced these 34 VTEs (i.e., age, BMI, smoking status, personal and family history of VTE) are shown in Supplementary Appendix 1.

Table 4. Confirmed DVT of the lower extremities and PE in women without pre-defined risk factors^a at study entry: Number of events, number of woman years and incidence per 10,000 WY (with 95% confidence intervals) by (sub-)cohort in the primary analysis and additional analyses.

DVT of the lower extremities and PE	(Sub-)cohort	Number of VTE	Number of WY	Incidence per 10,000 WY	95% CI
Primary VTE analysis
Overall	NOMAC-E2	9	45,750	2.0	0.9–3.7
	COCLNG	15	49,729^b	3.0	1.7–5.0
	COCLNGMono20	11	22,589	4.9	2.4–8.7
	COCLNGMono30	1	17,645	0.6	0.0–3.2
	COCLNGMulti	3	9,432	3.2	0.7–9.3
	COCOther	4	7,620	5.2	1.4–13.4
	OHC	1	2,073	4.8	0.1–26.8
	Ex-user	5	28,372	1.8	0.6–4.1
	Total	34	133,544	2.5	1.8–3.6
Sensitivity analyses
Excluding Russia	NOMAC-E2	8	24,372	3.3	1.4–6.5
	COCLNG	15	31,809	4.7	2.6–7.8
	COCOther	3	5,478	5.5	1.1–16.0
	OHC	1	1,844	5.4	0.1–30.2
	Ex-user	5	20,973	2.4	0.8–5.6
	Total	32	84,476	3.8	2.6–5.3
Russia only	NOMAC-E2	1	21,378	0.5	0.0–2.6
	COCLNG	0	17,920	0.0	0.0–2.1
	COCOther	1	2,142	4.7	0.1–26.0
	OHC	0	229	0.0	0.0–159.5
	Ex-user	0	7,399	0.0	0.0–5.0
	Total	2	49,068	0.4	0.05–1.5
Including Mexico^c	NOMAC-E2	9	46,169	1.9	0.9–3.7
	COCLNG	15	49,865	3.0	1.7–5.0
	COCOther	4	7,638	5.2	1.4–13.4
	OHC	1	2,094	4.8	0.1–26.6
	Ex-user	6	28,545	2.1	0.8–4.6
	Total	35	134,311	2.6	1.8–3.6

^aPregnant within 3 months of study entry, history of cancer/chemotherapy, increased genetic risk of VTE (e.g., Factor V Leiden, Protein S or C deficiency).

^bCOC_LNG cohort includes 63 WY of >40mcg EE use.

^cDue to non-compliance with study procedures, data contributed by 625 study participants in Mexico were excluded from the analyses (except a sensitivity analysis) as advised by the Safety Monitoring and Advisory Council. The local field organisation deviated from the required storage requirements and destroyed 545 of the informed consent forms. After recontacting the women, proof of consent was available for 235 women (38% of the Mexican study population). A sensitivity analysis of VTE risk with the inclusion of data from Mexico was conducted.

A Cox regression analysis comparing NOMAC-E2 with COC_LNG yielded an HR_crude of 0.65 (95% CI, 0.28–1.48) and an HR_adj of 0.59 (95% CI, 0.25–1.35) after adjusting for age, BMI, family history of VTE, and current duration of HC use (Table 5). The saturated model⁷ yielded an HR_adj of 0.61 (95% CI, 0.26–1.40). Backward elimination resulted in an HR_adj of 0.59 (95% CI, 0.26–1.35); the remaining co-factors were age, BMI, and family history of VTE.

Table 5. Cox regression analyses of the risk of DVT of the lower extremities and PE in women without pre-defined risk factors^a at study entry, by comparison group: Primary model and alternative analyses.

Comparison groups and models	Crude hazard ratio		Adjusted hazard ratio
	Point estimate	95% CI	Point estimate	95% CI
NOMAC-E2 vs COCLNG
A priori expert model (primary model)	0.65	0.28–1.48	0.59^b	0.25–1.35
A priori expert model (multiple imputation)^c	0.65	0.28–1.48	0.59^b	0.25–1.35
Saturated model	0.65	0.28–1.48	0.61	0.26–1.40
Backward elimination model^d	0.65	0.28–1.48	0.59	0.26–1.35
NOMAC-E2 vs COCLNGMono20mcg
A priori expert model (primary model)	0.41	0.17–0.99	0.31^b	0.13–0.75
A priori expert model (multiple imputation)^c	0.41	0.17–0.99	0.31^b	0.13–0.75
Saturated model	0.41	0.17–0.99	0.33	0.13–0.85
Backward elimination model^d	0.41	0.17–0.99	0.31	0.13–0.75

NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COC_LNG: levonorgestrel-containing combine oral contraceptive; CI: Confidence interval.

^aPregnant within 3 months of treatment initiation, history of cancer/chemotherapy or an increased genetic risk of VTE (e.g., Factor V Leiden, Protein S or C deficiency).

^bAdjusted for age, BMI, family history of VTE and current duration of hormonal contraceptive use.

^cValues of BMI were imputed.

^dRemaining co-factors included age, BMI, family history of VTE.

Only 2 of the 34 confirmed VTEs in the primary analysis occurred in Russia (1 NOMAC-E2 user and 1 COC_Other user) over 49,068 WY. To determine what impact the low Russian incidence (0.4 per 10,000 WY; 95% CI, 0.05–1.5) had on the overall VTE incidence, a sensitivity analysis was performed by excluding Russian data. The VTE incidence in non-Russian countries was 3.8 per 10,000 WY (95% CI, 2.6–5.3). A Cox regression analysis to compare NOMAC-E2 with COC_LNG in non-Russian countries yielded an HR_crude of 0.70 (95% CI, 0.30–1.65) and an HR_adj of 0.59 (95% CI, 0.25–1.41) (adjusted for age, BMI, family history of VTE and current duration of HC use).

The incidence of VTE was lower in the COC_{LNGMono/30 mcg} cohort (1 VTE results in 0.6 per 10,000 WY; 95% CI, 0.01–3.2) than the COC_{LNGMono/20 mcg} cohort (11 VTEs results in 4.9 per 10,000 WY; 95% CI, 2.4–8.7). A Cox regression analysis comparing NOMAC-E2 with COC_LNGMono20mcg yielded an HR_crude of 0.41 (95% CI, 0.17–0.99) and an HR_adj of 0.31 (95% CI, 0.13–0.75) (adjusted for age, BMI, family history of VTE, and current duration of HC use) which would indicate statistical significance given that the confidence intervals did not contain unity. Cox regression analyses to compare NOMAC-E2 with COC_LNGMono30mcg or COC_LNGMulti were not performed because the number of VTEs in each comparison group was lower than the minimum of 5 events as defined in the statistical analysis plan (SAP).

The Russian study population (who had a lower risk of VTE) contributed more substantially to the COC_{LNGMono/30 mcg} cohort compared to the COC_{LNGMono/20 mcg} cohort. Therefore, to determine whether the contribution of data by the Russian study population impacted upon the differences between the COC_{LNGMono/30 mcg} and COC_{LNGMono/20 mcg} cohorts, an analysis was conducted which excluded the Russian data. This resulted in an incidence of 1.4 per 10,000 WY (95% CI, 0.04–7.9) in the COC_{LNGMono/30 mcg} cohort and 5.1 per 10,000 WY (95% CI, 2.6–9.2) in the COC_{LNGMono/20 mcg}.

All VTE

Overall, 228 VTEs were reported of which 46 confirmed VTEs (including, but not restricted to, DVTs of the lower extremities and PE) were observed, 155 VTEs were excluded after adjudication and 27 VTEs were not confirmed. Of 46 confirmed VTEs: NOMAC-E2 users (12 VTEs, 2.5 per 10,000 WY; 95% CI, 1.3–4.3), COC_LNG (20 VTEs, 3.7 per 10,000 WY; 95% CI, 2.3–5.7), COC_Other (5 VTEs, 6.0 per 10,000 WY; 95% CI, 2.0–14.1), OHC (1 VTE, 4.2 per 10,000 WY; 95% CI, 0.1–23.5) and ex-users (8 VTEs, 2.6 per 10,000 WY; 95% CI, 1.1–5.0) (Table 6). A Cox regression analysis for the secondary outcome of all VTE was not planned or conducted.

Table 6. All Confirmed VTE: Number of events, number of woman years and incidence per 10,000 WY (with 95% confidence intervals) by (sub-)cohort in the overall analysis and additional analyses.

All VTE	(Sub-)cohort	Number of VTE	Number of WY	Incidence per 10,000 WY	95% CI
Overall	NOMAC-E2	12	48,846	2.5	1.3–4.3
	COCLNG	20	54,037	3.7	2.3–5.7
	COCLNGMono20	14	24,429	5.7	3.1–9.6
	COCLNGMono30	3	19,413	1.5	0.3–4.5
	COCLNGMulti	3	10,119	3.0	0.6–8.7
	COCOther	5	8,300	6.0	2.0–14.1
	OHC	1	2,364	4.2	0.1–23.5
	Ex-user	8	31,354	2.6	1.1–5.0
	Total	46	144,901	3.2	2.3–4.2
Russia only	NOMAC-E2	1	22,283	0.45	0.0–2.5
	COCLNG	0	19,301	0.0	0.0–1.9
	COCOther	1	2,342	4.3	0.1–23.8
	OHC	0	241	0.0	0.0–151.7
	Ex-user	1	8,130	1.2	0.0–6.9
	Total	3	52,298	0.6	0.1–1.7
Excluding Russia	NOMAC-E2	11	26,563	4.1	2.1–7.4
	COCLNG	20	34,736	5.8	3.5–8.9
	COCOther	4	5,958	6.7	1.8–17.2
	OHC	1	2,122	4.7	0.1–26.2
	Ex-user	7	23,224	3.0	1.2–6.2
	Total	43	92,603	4.6	3.4–6.3

NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COC_LNG: levonorgestrel-containing combine oral contraceptive; COC_Other: Combined oral contraceptive other than NOMAC-E2 or COC_LNG; OHC: Other hormonal contraceptive (i.e., progestin-only method, injections, implants, levonorgestrel-releasing intrauterine devices, contraceptive patches); WY: Women-years.

Only 3 of the 46 confirmed VTEs were in Russia (0.6 per 10,000 WY; 95% CI, 0.1–1.7). The Russian data were excluded in a sensitivity analysis (which was then based on the remaining 43 VTEs) and resulted in an overall incidence of 4.6 per 10,000 WY (95% CI, 3.4–6.3) in non-Russian countries. The numbers and incidences per cohort were: NOMAC-E2 (11 VTEs, 4.1 per 10,000 WY; 95% CI, 2.1–7.4), COC_LNG (20 VTEs, 5.8 per 10,000 WY; 95% CI, 3.5–8.9), COC_Other (4 VTEs, 6.7 per 10,000 WY; 95% CI, 1.8–17.2), OHC (1 VTE, 4.7 per 10,000 WY; 95% CI, 0.1–26.2) and ex-users (7 VTEs, 3.0 per 10,000 WY; 95% CI, 1.2–6.2).

In addition to the 46 confirmed VTEs, there were 27 potential VTEs that could not be adjudicated because sufficient information (e.g., medical documentation) could not be obtained. In a sensitivity analysis, these 27 VTEs were combined with the ‘confirmed’ VTEs to form a dataset of 73 VTEs: NOMAC-E2 (20 VTEs, 4.1 per 10,000 WY; 95% CI, 2.5–6.3), COC_LNG (28 VTEs, 5.2 per 10,000 WY; 95% CI, 3.4–7.5), COC_Other (7 VTEs, 8.4 per 10,000 WY; 95% CI, 3.4–17.4), OHC (1 VTE, 4.2 per 10,000 WY; 95% CI, 0.1–23.5) and ex-users (17 VTEs, 5.4 per 10,000 WY; 95% CI, 3.2–8.7).

Overall, 35 of the 46 confirmed VTEs were considered idiopathic VTEs. The numbers and incidence rates for each (sub-)cohort were as follows: 10 VTEs in NOMAC-E2 users (2.0 per 10,000 WY; 95% CI, 1.0–3.8), 15 VTEs in COC_LNG users (2.8 per 10,000 WY; 95% CI, 1.6–4.6), 5 VTEs in COC_Other users (6.0 per 10,000 WY; 95% CI, 2.0–14.1), 1 VTE in an OHC user (4.2 per 10,000 WY. 95% CI, 0.1–23.5) and 4 VTEs in ex-users (1.3 per 10,000 WY; 95% CI, 0.35–3.3).

ATE

Overall, there were 16 ATEs: NOMAC-E2 (4 ATEs, 0.8 per 10,000 WY; 95% CI, 0.2–2.1), COC_LNG (7 ATEs, 1.3 per 10,000 WY; 95% CI, 0.5–2.7), COC_Other (1 ATE, 1.2 per 10,000 WY; 95% CI, 0.0–6.7) and ex-users (4 ATEs, 1.3 per 10,000 WY; 95% CI, 0.35–3.3) (Table 7). A Cox regression analysis was not performed (the number of events was lower than defined in the SAP).

Table 7. Arterial thromboembolism: Numbers, incidence rates and 95% confidence intervals per (sub-) cohort.

	NOMAC-E2 (48,846 WY)		COCLNG (54,037 WY)		COCOther (8,300 WY)		OHC (2,364 WY)		Ex-user (31,354 WY)		Total (144,901 WY)
Category	n	Incidence^a (95% CI)	n	Incidence^a (95% CI)	n	Incidence^a (95% CI)	n	Incidence^a (95% CI)	n	Incidence^a (95% CI)	n	Incidence^a (95% CI)
ATE	4	0.8 (0.2–2.1)	7	1.3 (0.5–2.7)	1	1.2 (0.0–6.7)	0	0.0 (0.0–15.6)	4	1.3 (0.35–3.3)	16	1.1 (0.6–1.8)
Thereof:
Myocardial infarction	1	0.2 (0.0–1.1)	3	0.6 (0.1–1.6)	1	1.2 (0.0–6.7)	0	0.0 (0.0–15.6)	1	0.3 (0.0–1.8)	6	0.4 (0.15–0.9)
Ischaemic stroke	2	0.4 (0.05–1.5)	4	0.7 (0.2–1.9)	0	0.0 (0.0–4.4)	0	0.0 (0.0–15.6)	3	0.96 (0.2–2.8)	9	0.6 (0.3–1.2)
Transient ischaemic attack	1	0.2 (0.0–1.1)	0	0.0 (0.0–0.7)	0	0.0 (0.0–4.4)	0	0.0 (0.0–15.6)	0	0.0 (0.0–1.2)	1	0.1 (0.0–0.4)

^aIncidence rate per 10,000 WY.

NOMAC-E2: Nomegestrol acetate and 17β-oestradiol; COCLNG: levonorgestrel-containing combine oral contraceptive;

COCOther: Combined oral contraceptive other than NOMAC-E2 or COCLNG; OHC: Other hormonal contraceptive (i.e., progestin-only method, injections, implants, levonorgestrel-releasing intrauterine devices, contraceptive patches); WY: Women-years.

Thromboembolic events (TE)

All confirmed TEs (i.e., VTEs and ATEs) were analysed to facilitate comparisons with other studies. There were 62 TEs (4.3 per 10,000 WY; 95% CI, 3.3–5.5) as follows: NOMAC-E2 (16 TEs, 3.3 per 10,000 WY; 95% CI, 1.9–5.3), COC_LNG (27 TEs, 5.0 per 10,000 WY; 95% CI, 3.3–7.3), COC_Other (6 TEs, 7.2 per 10,000 WY; 95% CI, 2.7–15.7), OHC (1 TE, 4.2 per 10,000 WY; 95% CI, 0.1–23.5) and ex-users (12 TEs, 3.8 per 10,000 WY; 95% CI, 2.0–6.7). A Cox regression analysis comparing the risk of TE in users of NOMAC-E2 versus COC_LNG yielded an HR_crude of 0.68 (95% CI, 0.36–1.27) and an HR_adj of 0.64 (95% CI, 0.34–1.20) after adjusting for age, BMI, current duration of HC use, family history of VTE and family history of ATE.

Discussion

The PRO-E2 study was a large, prospective, observational multinational study designed to assess the risk of cardiovascular events in NOMAC-E2 users compared with COC_LNG users. Baseline characteristics and cardiovascular risk factors at study entry were similar between the user cohorts, although NOMAC-E2 users had a slightly higher mean age (31.0 ± 8.63 years) compared to COC_LNG users (29.3 ± 8.53 years) and reported a higher level of education. A 1.5-fold risk of VTE (DVT of the lower extremities and PE) in NOMAC-E2 users compared with COC_LNG users was excluded (yielded an HR_adj of 0.59 (95% CI, 0.25–1.35) (adjusted for age, BMI, family history of VTE and current duration of use). This result was supported by several sensitivity analyses. There were no substantial differences between the NOMAC-E2 cohort and the COC_LNG cohort regarding their risk of ATE and all thromboembolic events.

The incidence of VTE was lower than in similar previous studies [8,11] and is primarily driven by the low incidence in the Russian participants. When the Russian data were excluded, the incidence of confirmed VTE was 5.8 per 10,000 WY in the COC_LNG cohort. This corresponds with what is currently reported as the VTE incidence for COC_LNG (5–7 per 10,000 WY) [12,13]. The investigators conducted qualitative interviews with VTE experts in Russia to explore whether differences in behaviour or health care systems could explain the low incidence in that country. The experts advised that Russian women usually start COC use only after they have had children; women predisposed to VTE are identified during pregnancy and subsequently refrain from COC use. In contrast, women in Western European countries usually start COC use while they are nulliparous (and not yet identified as having a predisposition to VTE). Russian women do not need a prescription to buy COCs. Perhaps women who visit their HCP for advice rather than obtaining a COC over the counter are more health-conscious (only women who visited their HCP were enrolled in the study). Also, women are routinely screened for VTE risk prior to receiving their COC prescription. Taken together, these factors could explain why VTEs are rarely diagnosed in COC users. In Russia, diagnostics are readily available and free of charge to women with clinical signs of VTE. Unfortunately, no official source of VTE incidence exists in Russia and it is not possible to compare the incidence observed in the PRO-E2 Russian study population with that of the general population. Nevertheless, the investigators are confident that VTEs in Russia were captured; there was a higher incidence of SAEs in Russia than in countries excluding Russia (data not shown). Therefore, it is highly unlikely that there was an under-reporting of adverse events in Russia.

The risk of DVT of the lower extremities and PE was lower in COC_LNGMono30mcg users than in COC_LNGMono20mcg users, with an incidence rate of 0.6 and 4.9, respectively. A total of 57% of all COC_{LNGMono/30 mcg} users were in Russia compared with only 6.2% of all COC_{LNGMono/20 mcg} users. It was therefore considered that the overall low incidence rate of VTE observed in the Russian study population may have influenced the low incidence rate in the COC_{LNGMono/30 mcg} users. However, after excluding data from Russia from the analysis, the incidence of VTE remained lower in the COC_LNGMono30mcg cohort (1 VTE, 1.4 per 10,000 WY; 95% CI, 0.04–7.9) compared with the COC_{LNGMono/20 mcg} cohort (11 VTEs, 5.1 per 10,000 WY; 95% CI, 2.6–9.2).

The low risk of VTE observed in NOMAC-E2 users is not completely unexpected as previous studies demonstrated that NOMAC-E2 use had fewer adverse effects on coagulation and fibrinolysis parameters than COC_LNG use [3,5]. This also aligns with the results of another observational study conducted by the same investigators to assess oestradiol use and VTE risk [14]. The International Active Surveillance Study – Safety of Contraceptives: Role of Oestrogens (INAS-SCORE) followed users of dienogest (DNG)/oestradiol valerate (EV) and other COCs (including COC_LNG). The risk of VTE was lower in the DNG/EV cohort than the COC_Other cohort, with an HR_crude of 0.9 (95% CI, 0.4–1.7) and an HR_adj of 0.4 (95% CI, 0.2–0.9) after adjusting for age, BMI, current duration of HC use and family history of VTE. The comparison of DNG/EV to COC_LNG resulted in an HR_crude of 0.8 (95% CI, 0.3–2.0) and an HR_adj of 0.4 (95% CI, 0.2–1.1) after adjusting for the same potential confounding variables.

The PRO-E2 study has shown that the VTE and ATE risk associated with NOMAC-E2 use is no higher than with COC_LNG use. This conflicts with the current class labelling which cites a higher VTE risk with newer generation COCs compared with those containing LNG. Available evidence suggests that greater emphasis should be placed on differences between the oestrogen components (E2 vs EE) when comparing the risk of VTE among COC users. This is confirmed in this study where a relatively lower risk of DVT of lower extremities was observed among NOMAC-E2 use in comparison to COC_{LNGMono/20 mcg} users (HR_adj of 0.31 [95% CI, 0.13–0.75]) These results are similar to those reported in other studies highlighting the lower risk of VTE among NOMAC-E2 users compared to COC_{LNGMono/20 mcg} users due to the haemostatic impact and effects on coagulation factors, lipid and carbohydrate metabolism, and blood pressure [5,7,15].

There were limitations to the study. The introduction of the General Data Protection Regulation (GDPR) in May 2018 adversely impacted the lost-to-follow-up rate. Previously, if contact with study participants was lost, the contact details of a friend or relative (provided at study entry) were used to try to re-establish contact with the study participant. The more stringent data protection regulations restricted the ability to use these additional contacts. The lost-to-follow-up rate was further impacted by the restrictions caused by the emergence of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. Prior to the pandemic, face-to-face interviews were conducted if the women did not respond to other methods of contact (e.g., phone calls and emails). During the pandemic, face-to-face interviews (i.e., interviewing women at their homes if they could not be reached by other methods) were halted due to travel restrictions and health concerns. Additionally, the collection of registered letters was hindered by the closure of post offices and travel restrictions. Although the lost-to-follow-up rate was higher than expected (based on experience in similar previous studies), the rates were non-differential between the user cohorts.

The study had several strengths. A broad spectrum of COC-prescribing HCPs (gynaecologists, general practitioners, midwives) recruited women for the study, thereby enhancing the generalisability of the findings. Only new users were enrolled; selection bias introduced by the inclusion of prevalent users was eliminated. Following up directly with the study participants optimised the capture of VTEs and ATEs. Many COC-prescribing HCPs would not be aware that a woman suffered a VTE or ATE after study entry because the woman was either treated by another physician or she did not return to the HCP who recruited her. Relying upon patient reports ensured that almost all outcomes of interest were captured. The investigators carefully validated self-reported events with medical documentation to eliminate the classification of non-events as ‘confirmed’ simply because the women misunderstood the questionnaire or their diagnosis. All events (‘confirmed’ and ‘not confirmed’) were subsequently subjected to blinded adjudication, thereby reducing misclassification bias. The study design also enabled the capture of important potential confounders, including family history of VTE and ATE. Furthermore, precise information on HC use (specific HCs, stopping and starting patterns) was captured and women contributed WY to several (sub-)cohorts depending upon their real-life use.

Acknowledgements

The authors would like to acknowledge the valuable scientific contribution of the Safety Monitoring and Advisory Council. Their critical review of the data and challenging discussions substantially enhanced the overall quality of the study and the interpretation of the results. SMAC members included David Grimes, Michael Lewis, Robert Reid, Samuel Shapiro⁸, Carolyn Westhoff, Ulrich Winkler, and Stephanie Teal. Advisors to the SMAC included Jochen Albrecht and Fritz von Weizsäcker⁹. The authors would also like to thank the colleagues in the local field organisations who conducted the day-to-day field work. Their remarkable dedication and flexibility despite the challenging circumstances caused by the pandemic ensured the success of the study. The authors would also like to thank the ZEG Berlin staff members (statisticians, data managers, event validation staff, medical advisers) whose contribution to the study was invaluable.

Disclosure statement

Carol Koro and Julia DiBello are employees of Merck Sharp &hx0026; Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA, and hold stock/stock options in Merck & Co., Inc., Kenilworth, NJ, USA. Merck Sharp & Dohme Corp. was exclusively licensed by Theramex to sell Zoely in certain territories (other than the U.S. and Canada) until January 2020, and held Marketing Authorisations for the product in the non-EU markets of such territory. In January 2020, these rights were transferred back to Theramex. The rights to sell the product in the U.S. and Canada pursuant to a license from Teva remained with Merck until spin-off of Organon on June 2, 2021.

Additional information

Funding

Merck Sharp and Dohme, a subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA, and Theramex Ireland Limited, Dublin, Ireland provided funding for this research.

Notes

1 Marketed as ‘Zoely’.

2 COC_LNG as the comparator was a requirement of the European Medicines Agency as it was considered to be associated with the lowest risk of VTE.

3 This definition of the primary outcome and the exclusion of women with pre-defined risk factors conformed with requirements of the Pharmacovigilance Risk Assessment Committee of the EMA for this study.

4 1) NOMAC-E2, 2) COC_LNG monophasic preparation containing 20–30mcg of ethinylestradiol, 3) COC_LNG multiphasic preparation containing up to 40mcg of ethinylestradiol.

5 Potential cases = sufficient information (i.e., medical documentation) could not be obtained to confirm these events.

6 Progestin-only OC, vaginal ring, injection, intrauterine device, contraceptive patch or implant.

7 Potential sources of confounding included user status (starter or restarter), lifestyle factors (e.g. smoking), reason(s) for COC prescription, personal and family history of cardiovascular outcomes, medical history (including gynecological history) and concomitant medication use.

8 Died 19 April 2016.

9 Died 19 November 2019.