The effect of a combined oral contraceptive (COC) on sex hormone-binding globulin (SHBG) levels may be an indicator of the venous thrombosis risk of the COC involved [1,2]. SHBG is a plasma glycoprotein, primarily produced in hepatocytes, that binds the sex steroid hormones testosterone and 17β-estradiol, but not ethinylestradiol. Users of COCs containing a third-generation progestagen have higher SHBG levels than second-generation progestagen users [1,2], reflecting the difference in venous thrombosis risk. In accordance with the hypothesis that the SHBG level is a marker of venous thrombosis risk, SHBG levels in COC users are positively associated with thrombin generation-based activated protein C (APC) resistance . APC resistance has been shown to predict venous thrombosis risk in both men and women.
If the SHBG level can be considered to be a marker for venous thrombosis, and ethinylestradiol is the main compound in COCs causing venous thrombosis, then the ethinylestradiol dose in COCs should be reflected in SHBG levels. The main aim of this study was to determine whether an increase in ethinylestradiol dose resulted in higher SHBG levels in healthy premenopausal women.
Participants were selected from a large case–control study on venous thrombosis, i.e. the Multiple Environmental and Genetic Assessment of risk factors for venous thrombosis (MEGA) study , and from a crossover study in which women were asked to switch from their current contraceptive to either a COC containing drospirenone or a COC containing levonorgestrel, i.e. the drospirenone/ethinylestradiol (DRSP) study . Female control subjects, i.e. women without venous thrombosis, were selected from the MEGA study. Data of women before they switched to a specific COC from the DRSP study were used. We excluded women with known environmental thrombotic risk factors. A total of 191 healthy premenopausal women using COCs were included (181 from the MEGA study and 101 from the DRSP study).
In the MEGA study, whether or not the women were menstruating at venipuncture was recorded; however, blood was drawn randomly during the 4-week cycle of pill use (3 weeks of pill use followed by a pill-free week). In the DRSP study, blood was drawn between days 18 and 21 of the 4-week cycle of pill use. SHBG levels (nm) were measured with an immunometric assay (Immulite; Siemens Healthcare Diagnostics, Tarrytown, NY, USA) and without knowledge of the COC used or any other of the participant’s characteristics.
The ethinylestradiol dose was categorized into 20, 30 and ≥ 35 μg per pill and triphasic preparations, which have varying ethinylestradiol and progestagen doses over 21 days. The effect of progestagen and the dose of ethinylestradiol on SHBG levels was assessed with linear regression analysis. The analysis was adjusted for study, and to ensure that the effect of ethinylestradiol dose on SHBG levels was independent of the progestagen used, we adjusted this analysis for the progestagen used in COCs. Regarding the effect of progestagen in COCs on SHBG levels, the analysis was restricted to subjects taking 30 μg of ethinylestradiol per contraceptive pill. To reduce random variation in SHBG levels, the analyses were adjusted for whether women were menstruating at the time of venipuncture, and for age and body mass index (BMI), which can influence SHBG levels. Results were expressed as mean differences with 95% confidence intervals (CIs).
Overall, women were ∼ 33 years of age (range: 18–49 years), and had a BMI of 23.3 kg m−2 (range: 15.7–37.9). The mean SHBG plasma level was 139.5 nm (95% CI 131.2–147.8, interquartile range 99.8, range 28.0–390.9). In the MEGA study, SHBG levels were compared between menstruating women and women taking a pill at venipuncture. Eleven women were menstruating at the time of venipuncture, and the mean SHBG level was 102.1 nm (95% CI 59.1–145.0), whereas the level in women who were taking a pill (N = 163) was 145.4 nm (95% CI 134.3–156.6); the mean difference was 43.4 nm (95% CI −1.0–87.7). When we restricted our analysis to women receiving 30 μg of ethinylestradiol, users of desogestrel, gestodene and drospirenone had ∼ 100 nm higher SHBG levels than users of levonorgestrel (Table 1). Adjustment for factors influencing SHBG levels did not change these results. After adjustment for progestagen, users of ≥ 35 μg of ethinylestradiol had higher SHBG levels than users of 20 μg (Table 1). Also users of triphasic contraceptives had higher SHBG levels than users of 20 μg of ethinylestradiol. The SHBG levels were only slightly higher in users of 30 μg of ethinylestradiol than in users of 20 μg. Adjustment for factors influencing SHBG levels did not change these results. The same results were obtained when the analysis of ethinylestradiol dose and SHBG levels was restricted to the most commonly used progestagens (levonorgestrel, desogestrel, and gestodene) or separately for these progestagens, although the number of women per category became very small. Furthermore, similar results were obtained when the analysis was performed per study. In addition to the progestagens levonorgestrel, gestodene, desogestrel, and drospirenone, 30 women used cyproterone acetate. The mean SHBG level in users of cyproterone acetate was high, at 215.9 nm (95% CI 199.7–232.1), much higher than in users of COCs containing levonorgestrel with 30 μg of ethinylestradiol (mean difference of 135.4 nm [95% CI 116.9–153.9], after adjustment for study and menstruation at venipuncture).
|Variable||N (%)||Mean SHBG levels (95% CI)*||Adjusted difference (95% CI)*||Adjusted difference (95% CI)†|
|Levonorgestrel||99 (60)||80.3 (72.3–88.2)||Reference||Reference|
|Desogestrel||36 (22)||193.0 (179.9–206.2)||112.8 (97.3–128.2)||116.9 (101.1–132.7)|
|Gestodene||13 (8)||160.9 (138.8–183.0)||80.6 (57.3–104.0)||81.5 (56.3–106.6)|
|Drospirenone||17 (10)||191.3 (171.8–210.9)||111.1 (89.8–132.3)||114.3 (93.1–135.5)|
|20 μg||31 (11)||101.6 (80.4–122.8)||Reference||Reference|
|30 μg||165 (60)||115.3 (103.9–126.8)||13.8 (−7.1–34.6)||13.9 (−8.3–36.2)|
|≥35 μg||45 (16)||247.0 (200.6–293.4)||145.4 (87.1–203.7)||136.4 (64.5–208.3)|
|Triphasic||32 (12)||152.5 (132.7–172.4)||51.0 (22.8–79.1)||50.9 (20.7–81.1)|
One other study evaluated the effects of different COCs on SHBG levels, and provided information on ethinylestradiol dose per progestagen . However, no direct comparisons between ethinylestradiol dose and SHBG levels were made; therefore, no conclusions could be drawn on whether the ethinylestradiol dose in different COCs was reflected in SHBG levels. The positive association between ethinylestradiol dose and SHBG levels is in line with previous findings regarding venous thrombosis risk. Lidegaard et al.  reported that the risk of venous thrombosis was higher in users of 50 μg of ethinylestradiol (odds ratio [OR] 1.6, 95% CI 0.9–2.8) and lower in users of 20 μg (OR 0.6, 95% CI 0.4–0.9) than in users of oral contraceptive preparations containing 30–40 μg of ethinylestradiol. In the MEGA study, we also demonstrated that, among users of oral contraceptives containing levonorgestrel, the risk of venous thrombosis was higher in users of 50 μg of ethinylestradiol than in users of 30 μg (OR adjusted for age of 2.2, 95% CI 1.3–3.7) . The risk of venous thrombosis was lower in users of 20 μg than in users of 30 μg, for both users of gestodene (OR 0.3, 95% CI 0.2–0.7) and users of desogestrel (OR 0.7, 95% CI 0.4–1.2).
Unfortunately, ethinylestradiol levels could not be measured directly, because the blood was drawn at random during the 4-week cycle of pill use in the MEGA study and without considering the hours after a pill was taken, which both have a significant influence on ethinylestradiol levels . The hours after a pill was taken do not influence the SHBG levels, because of the half-life of SHBG of ∼ 7 days. However, data were available on factors that were previously shown to influence SHBG levels and on whether women were menstruating at the time of venipuncture. Furthermore, regarding the analysis of the relationship between ethinylestradiol dose and SHBG levels, we would have preferred to restrict our analysis to one progestagen; however, the number of women per category became very small, leading to unreliable estimates. We combined two studies that differed in their design, which may have affected our results. However, the same results were observed in an analysis performed per study. The strengths of our study were that we included a relatively large number of COC users who were using many different types of prescription. Furthermore, both mean SHBG levels and the difference in SHBG levels between different progestagens in COC users were in the same range as observed in other studies.
In conclusion, SHBG levels reflect the ethinylestradiol dose used in COCs, independently of the progestagen used. As ethinylestradiol is important in the pathogenesis of venous thrombosis in COC users, these findings strengthen the idea that SHBG levels in COC users may be seen as a marker for venous thrombosis risk.