Pii: s0010-7824(01)00316-x

An open label, comparative study of the effects of a dose-reduced oral contraceptive containing 20 ␮g ethinyl estradiol and 100 ␮g levonorgestrel on hemostatic, lipids, and carbohydrate metabolism J. Endrikata,*, C. Klippingb, M. Cronina, C. Gerlingera, A. Ruebiga, W. Schmidtc, aSchering AG, Mu¨llerstr, 178, D-13342 Berlin, Germany bDinox Medical Investigations, Groenewoudseweg 317, NL-6524TX Nijmegen, The Netherlands cUniversita¨tskliniken des Saarlandes, Frauenklinik und Poliklinik, D-66421 Homburg/Saar, Germany Abstract
In this open label, randomized study we compared the influence of a dose-reduced oral contraceptive containing 20 ␮g ethinyl estradiol (EE) and 100 ␮g levonorgestrel (20 EE) with a reference preparation containing 30 ␮g EE and 150 ␮g levonorgestrel (30 EE) on hemostatic,lipids, and carbohydrate metabolism variables.
Data from 48 volunteers were obtained. The direction of the change (increase or decrease) in most of the hemostatic variables were similar in both treatment groups. In particular, prothrombin fragment 1 ϩ 2 increased during treatment, reaching a median percent changeof 40% in the 20 EE group and of 17% in the 30 EE group after one year. D-Dimer fibrin split products remained virtually unchanged, withno change at Cycle 13.
The median HDL2 cholesterol levels decreased by 26% in the 20 EE group and by 39.8% in 30 EE group (p ϭ 0.0045 for group difference) after one year. The median one year change for LDL cholesterol was 3.23% in the 20 EE group, compared to 25% in the 30EE group, for VLDL 11.1% compared to 38.8%, respectively, and for total triglycerides 10.0% compared to 37.5%, respectively.
The median absolute change for the area under the curve (AUC)(0 -3h) for glucose at treatment Cycle 13 was 41.25 mmol/L ϫ min in the 20 EE group and 73.50 mmol/L ϫ min in the 30 EE group. The AUC(0 -3h) insulin at treatment Cycle 13 decreased in the 20 EE groupby 1635.0 pmolL ϫ min and increased in the 30 EE group by 11797.5 pmolL ϫ min (p ϭ 0.0491 for group difference). Both studytreatments were safe and well tolerated by the volunteers.
In conclusion, the balanced one-third dose reduction in this new oral contraceptive evoked similar effects on the hemostatic variables, but favorable results for the lipid and carbohydrate profiles. 2002 Elsevier Science Inc. All rights reserved.
Keywords: Oral contraceptives; Ethinyl estradiol; Levonorgestrel; Hemostasis; Lipids; Carbohydrate metabolism; Dose reduction 1. Introduction
containing the progestogen gestodene [1] and the otherdesogestrel [2]. Both were derived from marketed prepara- Since the introduction of oral contraceptives (OCs) in the tions containing 30 ␮g EE without reducing the progestogen early 1960s, considerable efforts have been made to reduce dose. An additional investigation, carried out with injectable the dose of the hormonal components to minimize the risk contraceptives [3], suggested that the balance between both of severe complications. These efforts attempted to reduce hormonal components could influence the pharmacodynamic the hormone content according to the principle as little as properties of the preparations. The new preparation tested in possible, but as much as necessary. In the 1990s two new 20 this trial has also been derived from an OC containing 30 ␮g ␮g ethinyl estradiol (EE) preparations were introduced, one EE combined with 150 ␮g LNG, which had been proven safeand effective for some decades [4]. However, in this newpreparation, the dose of both EE and LNG were reduced by * Corresponding author. Tel.: ϩ49-30-4681-7864; fax: ϩ49-30-4681- one-third. This dose reduction of both components maintains E-mail address: jan.endrikat@schering.de (J. Endrikat).
the established preparation's estrogen:progestogen ratio.
0010-7824/02/$ - see front matter 2002 Elsevier Science Inc. All rights reserved.
PII: S 0 0 1 0 - 7 8 2 4 ( 0 1 ) 0 0 3 1 6 - X J. Endrikat et al. / Contraception 65 (2002) 215-221 A large body of clinical data based on low-dose OCs All deviations from the scheduled tablet intakes had to be containing 20 ␮g EE has become available in the past years.
Despite some initial doubts about this reduction in the Each participant had a thorough medical and gynecolog- estrogen dose, it has become established that 20 ␮g EE, in ical examination that included a cervical cytology exami- combination with a suitable progestogen, provides effective nation by the Papanicolau method and a pregnancy test contraceptive protection and clinically acceptable cycle before start of treatment. One central laboratory (Klinisch Chemisch Laboratorium, Canisius-Wilhelmina Ziekenhuis, However, little data on the effect of OC use on metabolic Niejmegen, The Netherlands) performed the clinical labo- variables are available, in particular data on the effects of ratory evaluations, efficacy variables as well as safety lab- dose reduction of the EE with or without simultaneous oratory examinations (liver enzymes, hematologic vari- reduction of the progestogen component [8 -10]. The prep- ables, lipids, creatinine, bilirubin, alkaline phosphatase, aration investigated here is the first with a "balanced" re- total protein, and electrolytes). This laboratory worked ac- duction, i.e., a reduction of both hormonal components by cording to the Principles of Good Laboratory Practice and regularly participated in quality assurance procedures. Onlywidely accepted test kits were applied. The hemostatic vari-ables were determined by using the following methods: 2. Materials and methods
modified Clauss assay (fibrinogen; reagents from Dade),Thromborel S (FVII Act; Dade-Behring), Coaset Factor VII The study was performed as an open label, randomized, (FVII Ag), Staclot VIIa-rTF kit (Diagnositica Stago), rabbit prospective study at one center in The Netherlands (Dinox anti-human vWF polyclonal antisera and conjugate (FVIII; B.V., Groenewoudseweg 317, NL 6525 TX) from May Dako), Antithrombin III Chromogen (Dade), ProC Global 1998 to December 1999. We compared the dose-reduced (PCAT) (Dade Behring), Enzygnost TAT micro (Behring), investigational preparation containing 20 ␮g EE and 100 ␮g Enzygnost F1ϩ2 micro (Behring), Coamatic Plasminogen levonorgestrel (20 EE) (Miranova, Schering AG, Berlin, (Chromogenix), Immulyse t-PA (t-PA Ag; Biopool), Germany) with the reference containing 30 ␮g EE and 150 Chromolize t-PA (t-PA Act; Biopool), Innotest PAI-1 ␮g levonorgestrel (30 EE) (Microgynon, Schering AG). The (PAI-1 Ag; Innogenetics), Chromolize PAI-1 (PAI-1 Act; study protocol was approved by the appropriate ethics com- Biopool), Enzygnost PAP micro (Behring), Tina-quant d- Dimer (Boehringer-Mannheim) and Fibronostika FbDP We recruited a total of 60 healthy women aged 18 to 35 years for the study. Of these, 48 volunteers were included in Blood samples for the hemostatic and lipid variables the per protocol analysis (11 prematurely discontinued the were taken during the pretreatment cycle, treatment Cycles study, and one volunteer did not follow the protocol cor- 1, 3, 6, 9, and 13 (on cycle Days 15 to 21), and the follow-up rectly). The women's wish for contraception for at least cycle (on cycle Days 21 to 28). We only report the data thirteen 28-day cycles was a prerequisite for their partici- obtained for treatment Cycles 3, 6, and 13 here, but the pation in the study. New OC users as well as women who results obtained for treatment Cycles 1 and 9 were also wanted to change their OC (switchers) were included in the study. Switchers had to have at least two OC-free cycles, The oral glucose tolerance tests (OGTTs) were carried one wash-out cycle and one pretreatment cycle, before they out in the pretreatment cycle and in treatment Cycles 6 and started to take the study medication. The exclusion criteria 13 (on cycle Days 15 to 21). Seventy-five grams of glucose were similar to the known contraindications for OC use.
were administered as dextrose solution in a fruit flavored Further exclusion criteria were the use of parenteral depot- beverage. The influence of the study treatments on carbo- contraceptives during the last 6 months before the study, hydrate metabolism was assessed by measurements of specified coexisting diseases, diagnostically unclassified plasma glucose, insulin, C-peptide, and free fatty acid lev- genital bleeding, and a history of migraine accompanying els. Blood samples were taken at a fasting state before menstruation. All volunteers gave informed consent prior to glucose administration (together with the samples required for general clinical chemical variables, screening variables, The volunteers were randomized to one of two treatment hemostatic profile, and lipid profile) and in 30-min intervals groups. The volunteers had to take the first tablet on the first day of withdrawal bleeding. The volunteers took 21 tablets Blood pressure and body weight were measured at all on the first 21 days of a cycle followed by seven pill-free study visits. Adverse events (AEs); concomitant medication days. The treatment period consisted of 13 consecutive usage; and treatment compliance, including a record of 28-day cycles, all of which started on the same week day as intake errors and cycle control patterns, were recorded by the initial cycle. The study medications were supplied in using the volunteers' diaries and through general question- calendar packs. If a woman missed the scheduled intake ing by the investigators. In the follow-up period, the vol- time, she was instructed to take the tablet until up to 12 h unteers were again asked about their general health during after the scheduled time and to record the delay in her diary.
the treatment period. Furthermore, medical and gynecolog- J. Endrikat et al. / Contraception 65 (2002) 215-221 Table 1Demographic characteristics at baseline ical examinations, including cervical cytology and routine laboratory examinations, were repeated.
Any pregnancies were to have been recorded throughout The direction and magnitude of the effect on most of the hemostatic variables were similar in both treatment groups.
The median concentrations of prothrombin fragment 1 ϩ 2(baseline 20 EE ϭ 0.53 ␮g/L; 30 EE ϭ 0.61 ␮g/L) in- creased during treatment in both groups reaching 0.8 ␮g/Lat cycle 13 in both groups. The median percent change in Statistical analyses were performed for both the "full the 20 EE group was 40% and in the 30 EE group 17%.
analysis" and the "per protocol" (PP) populations. All ran- None of the measured values were above the upper normal domized volunteers who took at least one tablet of the study limit (UNL) of the reference range (2.88 ␮g/L). An appre- medication were included in the "full analysis" population.
ciable difference between the groups was not found. Also Volunteers in the "full analysis" population with major for d-dimer fibrin split products, the median concentrations protocol deviations that affected the target variables were were equivalent (200 ␮g/L) in both groups. Although, some individuals in both groups had values that were higher than The target variables included the relative changes of each Ն800 ␮g/mL (1.50 times the UNL) during the treatment individual hemostasis and lipid variable and the absolute phase, the median concentrations remained relatively con- change of the area under the curve (AUC) for glucose, stant during treatment so that a percent change at Cycle 13 insulin, and C-peptide. The AUC was determined by using the trapezoidal rule. The relative change of a variable was Significant differences between the two treatment groups defined as the difference of the value measured in treatment were found for pro- and anti-fibrinolytic variables (plasmin- Cycle 13 minus the value measured in the pretreatment ogen, tPA Ag, and PAI-Ag, PAI-Act). Whereas plasmino- cycle divided by the value measured in the pretreatment gen was significantly more increased by the 30 EE prepa- cycle times 100%. The absolute changes of the AUCs were ration (20 EE ϭ 31.1%; 30 EE ϭ 45.25), tPA Ag was calculated as the difference of the value measured in Cycle significantly less reduced by the 20 EE preparation (20 13 minus the value measured in the pretreatment cycle.
EE ϭ Ϫ31.1 ng/mL; 30 EE ϭ Ϫ47.6 ng/mL). PAI-Ag and For each target variable, the null hypothesis, which PAI-Act were significantly more suppressed by 30 EE. For stated that the mean values are equal in both treatment pro- and anti-coagulatory variables, no significant differ- groups, was tested, by using the two-sided Wilcoxon rank ences between the two groups were detected. Some pro- sum test, against its alternative, which stated that the mean coagulatory variables were increased (fibrinogen, F VII Ag, values in both treatment groups were not equal using the F VIIa) and some were reduced (F VII Act, F VIII). Bothanti-coagulatory variables, AT III and PCAT, were slightly The significance level ␣ of 5% for these tests was not adjusted for multiple testing as appropriate for exploratory The relative changes of each lipid variable from the pretreatment cycle to treatment Cycle 13 were secondary 3. Results
The baseline values of all of the lipid variables, except All 60 randomized volunteers, 30 in each group, received for lipoprotein a, were comparable in both groups. The one of the treatments. A total of 48 volunteers, 23 in the 20 median values for lipoprotein a at baseline were 20 EE ϭ EE group and 25 in the 30 EE group, were included in the PP population. The demographic characteristics of the two A decrease for HDL2 cholesterol and an increases in the groups at baseline were well matched (see Table 1).
values for total cholesterol, LDL cholesterol, VLDL cho- J. Endrikat et al. / Contraception 65 (2002) 215-221 lesterol, and total triglycerides were found in both treatment and after treatment did not suggest an influence of the study groups from baseline to the thirteenth treatment cycle.
preparations. The incidence of laboratory abnormalities was Although the median HDL2 cholesterol levels decreased low and not appreciably different in the pre- and the post- by 39.8% in 30 EE group, the reduction was only 26% in the treatment examinations. Individual deviations from normal 20 EE group. This difference between the treatment groups laboratory ranges were transient, generally minor, and not was significant (p ϭ 0.0045). The median percent change after 1 year of treatment for LDL cholesterol was 3.23% inthe 20 EE group compared to 25% in the 30 EE group, forVLDL 11.1% compared to 38.8%, respectively, and for 4. Discussion
total triglycerides 10.0% compared to 37.5%, respectively.
The median percent change in lipoprotein a was Ϫ7.64% in This study was conducted to compare the differences in the 20 EE group; this parameter did not change in the 30 EE the effects of two combined OC preparations, one a dose- group. However, this could also be due to differences in the reduced version of the other, on hemostatic, lipid, and car- levels for this parameter at baseline. The changes from bohydrate profiles during one year of treatment in healthy baseline to Cycle 6 were comparable to those for Cycle 13; the changes for Cycle 3 were somewhat lower (Table 3).
Both the coagulation and the fibrinolysis system ap- peared to have been upregulated to higher capacity and activity levels, for the substrates as well as for the reactionproducts. These shifts were comparable for both groups, As seen in Table 4, the median values for the fasting
thus forming a new equilibrium for coagulation and fibri- levels of all carbohydrate variables decreased slightly nolysis. A threshold for a relevant change was not defined, (Ͻ10%) in the 20 EE group, whereas either a slight increase rather the changes were described with descriptive statistics (Ͻ6%) or no change was seen in the 30 EE group. The and compared for each treatment group. The observed median absolute change for the AUC(0 -3h) for glucose at changes in all variables remained within normal ranges of treatment Cycle 13 was 41.25 mmol/L ϫ min in the 20 EE variation, which is in accordance with the current knowl- group and was 73.50 mmol/L ϫ min in the 30 EE group.
edge [11]. The magnitudes of all shifts were moderate and The median absolute change for the AUC(0 -3h) for insulin comparable with other low-dose preparations [8,12,13].
at treatment Cycle 13 decreased in the 20 EE group by Winkler stated in a review article that altered plasma 1635.0 pmol/L ϫ min and increased in the 30 EE group by concentrations of many components of the coagulation and 11797.5 pmol/L ϫ min. The latter difference was tested and fibrinolysis system have a limited predictive value for fond to be significant (p ϭ 0.0491; Table 4).
thromboembolic risk [14]. However, the Oral Contraceptiveand Hemostasis Study Group [15] specifically evaluated, in a large multicenter study, prothrombin fragments 1ϩ2 andd-dimer fibrin split products as primary outcome variables.
None of the randomized volunteers became pregnant Although these variables are considered to be involved in the development of acute deep venous thrombosis, theypermit, if at all, only a very rough estimation of theoretically increased risk of such events. Because of the lack of highlypredictive variables, we followed the strategy of the Oral A total of 414 mostly mild and transient AEs in 90% Contraceptive and Hemostasis Study Group [15] by specif- (54/60) of the volunteers were reported during the treatment ically considering these parameters. The median concentra- phase of the study. Of these, 152 AEs were reported in 87% tion of prothrombin fragments 1ϩ2, a dynamic marker (26/30) of the volunteers in the 20 EE and 262 in 93% considered to be the most representative for changes in (28/30) in the 30 EE group. The AEs that affected the most coagulation, did not change appreciably in either group volunteers were abdominal pain, nausea, flu, headache, in- during treatment, although significant increases were seen fection, and vomiting. Although overall fewer AEs were in other studies [16 -18]. Also, a measurable median change reported in the 20 EE group, major differences in the type of in d-dimer concentration, the dynamic marker for the fi- AEs reported between the treatment groups were not found.
brinolytic system, was not found in either study group after No deaths and no study drug-related serious AEs occurred.
1 year of treatment. The median concentrations remained Physical and gynecological examinations showed only unchanged in both groups during treatment, and a percent few abnormal findings, all of which were not considered change at Cycle 13 was not seen in either study group treatment related. The results of the pre- and post-treatment physical examination was basically the same. An apprecia- The changes found in the lipid profile after one year of ble treatment influence on body weight and blood pressure treatment were characteristic for low-EE dose combined OCs [19,20]: moderate, but measurable, increases in the The overall levels of the safety laboratory values before values for total cholesterol, LDL cholesterol, VLDL cho- Table 2Hemostatic variables (median and interquartile ranges) Table 3Lipid profile (median and interquartile ranges) J. Endrikat et al. / Contraception 65 (2002) 215-221 Table 4Carbohydrate profile, fasting and AUCs (0 -3h) (median and interquartile ranges) * It should be noted that this column presents the median of the individual changes, which is in general not equal to the change of the medians for each time point presented in the columns on the left hand side.
lesterol, and total triglycerides and decreases in HDL cho- bined OC-induced changes are considered clinically irrele- lesterol were found in both treatment groups from baseline vant. Nonetheless, any new low-dose OC should have the to the thirteenth treatment cycle. Despite large variations in least possible impact on carbohydrate metabolism [19].
the change of all values, slight differences in the magnitude Evidence exists that EE increases insulin resistance of the median changes in the lipid profile between the whereas progestogens have selective effects on insulin se- treatment groups were found. These differences were only cretion and elimination [22]. This dose-reduced 20 EE prep- significant for HDL2. For all lipid variables, with the ex- aration appears to have a more favorable effect on the ception of lipoprotein a, less change was seen in the 20 EE carbohydrate profile compared to the higher-dosed OC.
This study was conducted to assess the metabolic effects It is interesting that the one-third dose reduction from 30 of a dose-reduced OC containing 20 ␮g EE ϩ 100 ␮g ␮g EE ϩ 150 ␮g levonorgestrel to 20 ␮g EE ϩ 100 ␮g levonorgestrel. The new preparation was designed as a bal- levonorgestrel 20 EE yielded an improved lipid pattern with anced, one-third dose reduction of a preparation containing respect to cardiovascular risk: all adverse lipids (e.g., LDL 30 ␮g EE ϩ 150 ␮g levonorgestrel. Similar effects on the cholesterol, total triglycerides, lipoprotein a) were less in- hemostatic profile were found for both preparations, al- creased and the positive lipid HDL cholesterol was less though a striking advantage for the lower dose preparation decreased by the 20 EE preparation compared to the 30 EE.
was not found. However, favorable results for the lower The fasting levels for glucose, insulin, c-peptide, and free dose preparation were found for the lipid and carbohydrate fatty acids were decreased in the 20 EE group and either minimally elevated or unchanged in the 30 EE group.
The median absolute change for the AUC(0 -3h) for glucose at treatment Cycle 13 was increased in both treat- References
ment groups. However, none of the volunteers had glucosevalues during the OGTT that were suggestive of subclinical [1] Du¨sterberg B, Ellman H, Mu¨ller U, et al. A three-year clinical hyperglycemia (2-h value Ͼ11.1 mmol/L).
investigation into efficacy, cycle control and tolerability of a new The median absolute change for the AUC(0 -3h) for low-dose monophasic oral contraceptive containing gestodene. Gy- insulin at treatment Cycle 13 decreased in the 20 EE group (Ϫ1635.0 pmol/L ϫ min), but increased in the 30 EE group [2] Nevinny-Stickel J. German trial of an oral contraceptive containing (11797.5 pmol/L ϫ min). This difference was significant 0.150 mg desogestrel plus 0.020 mg ethinylestadiol. Acta ObstetGynecol Scand 1988;44(Suppl.):19 -21.
(p ϭ 0.0491). Although large variations for both AUCs [3] Newton JR, d'Arcangues C, Hall PE. A review of 'once-a-month' were found, measurable differences in the changes in the combined injectable contraceptives. Obstet Gynecol 1994;14:1-34.
carbohydrate profile between the treatment groups were [4] Teichmann A, Corbin A. Clinical experience with a monophasic oral detected. For both carbohydrate variables, but in particular contraceptive containing 150 ␮g levonorgestrel and 30 ␮g ethi- for insulin, less change was seen in the 20 EE group com- Levonorgestrel, Thieme 1998:81-91.
[5] Bannemerschult R, Hanker JP, Wu¨nsch C, et al. A multicenter, Former studies with higher-dosed OCs revealed a re- uncontrolled clinical investigation of the contraceptive efficacy, cycle duced glucose tolerance during their use [9,21]. Although a control, and safety of a new low dose oral contraceptive containing 20 chronically reduced insulin sensitivity may be accompanied ␮g ethinyl estradiol and 100 ␮g levonorgestrel over six treatment by adverse metabolic and cardiovascular changes, the com- cycles. Contraception 1997;56:285-90.
J. Endrikat et al. / Contraception 65 (2002) 215-221 [6] Hite RC, Bannemerschult R, Fox-Kuchenbecker, Turck R, Brill K.
[14] Winkler UH. Blood coagulation and oral contraceptives. Contracep- Large observational trial of a new low-dose oral contraceptive con- taining 20 ␮g ethinylestradiol and 100 ␮g levonorgestrel (Miranova) [15] The Oral Contraceptive and Hemostasis Study Group. An open label, in Germany. Europ J Contracept Reproduct Health Care 1999;4:7-13.
randomized study to evaluate the effects of seven monophasic oral [7] Endrikat J, Du¨sterberg B, Ruebig A, Gerlinger C, Strowitzki T.
contraceptive regimens on hemostatic variables. Contraception 1999; Comparison of efficacy, cycle control, and tolerability of two low- dose oral contraceptives in a multicenter clinical study. Contraception [16] Meijers J, Middeldorp S, Tekelenburg W, et al. Increased fibrinolytic activity during use of oral contraceptives is counteracted by an en- [8] Winkler UH, Schindler AE, Endrikat J, Du¨sterberg B. A comparative hanced factor XI-independent down-regulation of fibrinolysis.
study of the effects on the hemostatic system of two monophasic gestodene oral contraceptives containing 20 ␮g and 30 ␮g ethi- [17] Prasad R, Koh S, Viegas O, Ratnam S. Effects on hemostasis after nylestradiol. Contraception 1999;53:75- 84.
two-year use of low dose combined oral contraceptives with gesto- [9] Crook D, Godsland I. Safety evaluation of mordern oral contracep- dene or levonorgestrel. Clin Appl Thromb Hemostat 1999;5:60 -70.
tives. Contraception 1998;57:189 -201.
[18] Middeldrop S, Meijers JC, van den Ende, et al. Effects on coagulation [10] Kluft C, De Maat MP, Heineman LA, Spannagl M, Schramm W.
of levorgestrel- and desogestrel-containing low dose oral contracep- Importance of levonorgestrel dose in oral contraceptives for effects tives: a cross-over study. Thromb Haemost 2000;84:4 - 8.
on coagulation. Lancet 1999;354:832-3.
[19] Gaspard IJ, Demeyer F, Jaminet CB, Scheen AJ, Lefebbvre PJ. In- [11] Consensus Development Meeting.: Combined oral contraceptives, fluence of two low-dose oral contraceptives containing ethinylestra- and cardiovascular disease. The Consensus statement issued after the diol (20 ␮g) and desogestrel or gestodene on carbohydrate metabo- Second European Conference on Sex Steroids and Metabolism, Am-sterdam, November 1995. Gynecol Endocrinol 1995;10:1-5.
lism during 1 year of use. Gynecol Endocrinol 1996;10(Suppl 2): [12] Winkler UH, Ho¨lscher T, Schulte H, Zierleyn JP, Collet W. Schindler AE. Ethinylestradiol 20 versus 30 ␮g combined with 150 ␮g deso- [20] Robinson GE, Registrar S. Low-dose combined oral contraceptives.
gestrel: a large comparative study of the effects of two low-dose oral Br J Obstet Gynaecol 1994;101:1036 - 41.
contraceptives on the hemostatic system. Gynecol Endorcinol 1996; [21] Watanabe RM, Azen CG, Roy S, Perlamn JA, Bergman RN. Defects in carbohydrate metabolism in oral contraceptive users without ap- [13] Winkler UH, Daume E, Sudik R, et al. A comparative study of the parent metabolic risk factors. J Clin Endrocrinol Metab 1994;78: hemostatic effects of two monophasic oral contraceptives containing 30 ␮g ethinylestradiol and either 2 mg chlormadinone acetate or 150 [22] Godsland IF, Walton C, Felton C, Proudler A, Patel A, Wynn V.
␮g desogestrel. Europ J Contracept Reprod Health Care 1999;4:145- Insulin resistance, secretion, and metabolism in users of oral contra- ceptives. J Clin Endrocrinol Metab 1992;74:64 -70.

Source: http://www.dinoxgroup.com/fileadmin/dinox/datei/publications/Contrac_EE_Levo_on_metabolism.pdf