The European Journal of Contraception and Reproductive Health Care, December 2011; 16: 444–457 Effects of a monophasic combined

Objectives To compare the effects of a combined oral contraceptive (COC) containing nomegestrol acetate and 17β-oestradiol (NOMAC/E2) on haemostasis, lipids, carbohydrate metabolism, C-reactive protein (CRP) and sex hormone-binding globulin (SHBG) with those of a COC containing levonorgestrel and ethinylestradiol (LNG/EE). Methods In a randomised, open-label study, 121 healthy women, 18-50 years of age, were randomly assigned to receive NOMAC/E2 (2.5 mg/1.5 mg) in a 24/4-day regimen (n = 60) or LNG/EE (150 μg/30 μg) in a 21/7-day regimen (n = 61) for six cycles. The primary outcome was the change from baseline to cycle 6 for all indices. Results All parameters were similar at baseline between the two groups. Over six cycles, NOMAC/E2 had less effect on most haemostatic indices than LNG/EE. Lipids were essentially unchanged with NOMAC/E2, whereas with LNG/EE high-density lipoprotein cholesterol decreased and low-density lipoprotein cholesterol and triglycerides slightly increased. NOMAC/E2 induced negligible changes in glucose and insulin parameters, in contrast to LNG/EE. A much smaller increase in CRP was observed with NOMAC/E2 than with LNG/EE. NOMAC/E2 was associated with a greater increase in SHBG. Conclusions The monophasic COC NOMAC/E2 had less influence on haemostasis, lipids and carbohydrate metabolism than the COC LNG/EE.


I N T R O D U C T I O N
Since their introduction 50 years ago, combined oral contraceptives (COCs) have become the most frequently used method of reversible contraception worldwide 1 . Over the past few decades, many advances have been achieved in COCs, including a reduction in the dose of the oestrogenic component ethinylestradiol (EE) and the introduction of more selective progestogens. This has resulted in improved tolerability 2 and a more favourable cardiovascular risk profi le 3,4 . The use of oestrogens and progestogens that are identical to or show greater resemblance with endogenous hormones represents the most recent advancement in COC development.
While the EE content in COCs is correlated with increased venous thromboembolism (VTE) risk, the infl uence of the progestogen component on VTE risk is controversial 5 . Although it has been suggested that some newer progestogens are associated with greater VTE risk 6,7 , the methodologies used in these studies have come into question 8,9 . Another study looked at confounders (weight, body mass index, smoking status, age, etc.) and found that COCs containing levonorgestrel (LNG) or norgestimate had similar VTE risk profi les compared with COCs containing gestodene or desogestrel 10 . Results from a post-marketing surveillance study (142,475 woman-years) that monitored cardiovascular outcomes among COC users (including those containing drospirenone [DRSP] or LNG) in Europe suggested that VTE risk was similar, independent of progestogen type 11 . Prescriber bias may also contribute to higher VTE rates, especially in smokers who may be more likely to receive prescriptions for COCs that contain newer progestogens 5 .
For several decades, researchers have been trying to incorporate the natural hormone 17 β -oestradiol (E2), which is identical to the endogenous oestrogen, into COCs. Although many early formulations that contained E2 were associated with poor cycle control, which limited E2 ' s clinical usefulness, initiatives to replace synthetic oestrogen with physiological hormone continued. It is still hypothesised that the use of E2 in combination with a suitable progestogen could further improve tolerability and minimise the deleterious effects of EE on hepatic metabolism 12 .
Recently, a monophasic COC containing nomegestrol acetate (NOMAC) in combination with E2 has been developed. NOMAC is a progestogen structurally similar to progesterone; it is devoid of oestrogenic, androgenic, glucocorticoid, and mineralocorticoid activity but displays an antioestrogenic activity on the endometrium and a moderate antiandrogenic activity 13 . Due to the enhanced selectivity profi le of NOMAC, this COC may provide acceptable cycle control with putative neutral effects on cardiovascular or metabolic risks 14 .
The COC containing NOMAC/E2 (2.5 mg/1.5 mg) administered in a 24/4 (active/placebo)-day regimen provides effective suppression of ovarian activity and acceptable cycle control in women of childbearing age 15,16 . NOMAC/E2 has also demonstrated suppressive effects on the ovaries, cervical mucus and endometrium that are at least as strong as those of a comparator COC containing DRSP and EE 15 .
The current study was performed to assess the effects of NOMAC/E2 on haemostasis, lipids, carbohydrate metabolism, C-reactive protein (CRP), and sex hormone-binding globulin (SHBG). The infl uence of NOMAC/E2 on surrogate markers of adrenal and thyroid function, androgens, and androgen precursors was also evaluated, but will be reported separately 17 . LNG/EE (150 μ g/30 μ g) was used as a comparator based on regulatory guidance for safety evaluation of new COCs 18 .

M E T H O D S
The study was a randomised, open-label, comparative, parallel-design study conducted in fi ve study centres in Finland. Advertisements were used to recruit women from the regions surrounding the study centres (Kuopio, Tampere, Turku, Espoo, and Oulu). Throughout the trial, study volunteers were compensated for their time, travel costs, and inconvenience associated with study visits.
The study complied with the ethical principles of the Declaration of Helsinki and the International Conference on Harmonisation (ICH) guideline for Good Clinical Practice. An independent Ethics Committee reviewed and approved the study protocol (NCT00511355) before the start of the study. All participants in the study provided written informed consent.

Subjects
Healthy, sexually active women aged 18 -50 years with a body mass index between 17 -29 kg/m 2 were eligible 446 The European Journal of Contraception and Reproductive Health Care

Laboratory measurements
All laboratory examinations were performed by the Bio Analytical Research Corporation (Ghent, Belgium). Procoagulatory indices . Factor II activity, factor VII coagulant activity (factor VIIc) and activated factor VII (factor VIIa) were measured using the commercial clotting assays/reagents STA-Defi cient II, STA-Deficient VII, STAclot VIIa-RTF, STA-Neoplastine CI ϩ and the automated analyser STA Compact (Diagnostica Stago). Factor VIII activity was determined by measuring factor dilution activity using the STA Analyser (Diagnostica Stago).

Haemostasis
if they had no contraindications for the use of contraceptive steroids and had not taken any other hormonal treatment (except contraceptives) during the two months preceding screening. Other exclusion criteria included an abnormal cervical smear; a clinically relevant abnormal laboratory fi nding; breastfeeding; and use of liver-enzyme-inducing drugs, investigational drugs or pharmacological agents affecting the haemostatic system (e.g., vitamin K, nonsteroidal antiinfl ammatory drugs, including aspirin).

Study design
The study consisted of one pre-treatment cycle (cycle 0), six 28-day treatment cycles (cycles 1 -6), and a post-treatment follow-up. At the screening visit, subjects were evaluated to determine their eligibility and collect general baseline data. They were instructed to discontinue the use of hormonal contraceptives, if applicable, and to use condoms whenever necessary. The pre-treatment cycle visit took place in the second half of the fi rst spontaneous menstrual cycle after the screening visit. During the pre-treatment cycle visit, blood was taken in a fasted state during the morning for baseline assessment of metabolic indices and an oral glucose tolerance test (OGTT) was performed. Eligible women were randomised in a 1:1 ratio to receive the investigational COC containing 2.5 mg NOMAC and 1.5 mg E2 or the comparator COC containing 150 μ g LNG and 30 μ g EE. Randomisation was performed using blocks with randomly permuted block sizes and an interactive voice response system. Women were stratifi ed by age class (i.e., 18 -35 years and 36 -50 years). For six consecutive 28-day cycles, the women took one tablet of the study medication orally at approximately the same time each day from day 1 to day 28. The NOMAC/E2 regimen comprised 24 days of active hormone pills followed by four days of placebo pills; the LNG/EE regimen comprised 21 days of active hormone pills followed by seven days of placebo pills. Blood collection for metabolic assessments and an OGTT were repeated between day 15 and day 21 of treatment cycles 3 and 6 when participants returned for clinic visits in a fasted state. A fi nal visit was scheduled between 8 and 14 days after the last tablet was taken in cycle 6 or after early discontinuation of treatment for the gathering of general follow-up data.

Lipids
Serum levels of total cholesterol and triglycerides were determined enzymatically using cholesterol oxidase-phenol/aminophenazone (CHOD-PAP) and glycerol-3-phosphate oxidase (GPO)-PAP methods, respectively, on a Modular P Analyser (Roche Diagnostics GmbH, Mannheim, Germany). High-density lipoprotein cholesterol (HDL-C) values were measured using modifi ed enzymatic procedures on a Modular P system; low-density lipoprotein cholesterol (LDL-C) values were estimated by the Friedewald equation (i.e., total cholesterol minus HDL-C minus very low-density lipoprotein cholesterol estimated as triglyceride level divided by fi ve) 19 . Lipoprotein (a) and apolipoprotein A1 and B levels were assessed by immunoturbidimetry on a Modular P system (Roche Diagnostics).

Carbohydrate metabolism indices
For the OGTT, glucose and insulin levels were measured before (t ϭ 0) and at 30, 60, 90, 120 and 180 minutes after drinking a glucose solution (75 g/100 ml). Serum glucose levels were measured according to the hexokinase ultraviolet method on a Modular P system; serum insulin levels were estimated by chemiluminescence (CLIA test) using the automated immunochemiluminescence analyser Immulite ® (DPC Biermann, Bad Nauheim, Germany). Ion-exchange high-performance liquid chromatography was performed using the automated analyser HA-8160 (A. Menarini Diagnostics S.r.l., Firenze, Italy) for the determination of haemoglobin type A 1c (HbA 1c ).

CRP and SHBG
Serum levels of CRP were obtained with the immunochemiluminescence analyser Immulite ® (DPC Biermann, Bad Nauheim, Germany). The electrochemiluminescence immunoassay ECLIA (Roche Diagnostics) was used to measure serum levels of SHBG.

Statistical analysis
The trial was set up to confi rm the expected differences in changes from baseline between the treatment groups in several metabolic indices, especially prothrombin fragment 1 ϩ 2, D-dimer and ATIII.
In previous trials (unpublished data), standardised effect sizes (i.e., differences between the treatment groups in changes from baseline divided by the common standard deviation) were found to be in the range of approximately 0.6 -1.0, depending on the parameter and the variability range. Conservatively assuming a standardised effect size of approximately 0.6 as a difference worthwhile to detect, a sample size of 42 evaluable subjects per treatment group was needed using a two-sided statistical test with 80% power and a signifi cance level of 5%. Compensating for up to 20% premature discontinuations from treatment and accounting for a potential power loss due to the non-parametric analysis, in total 60 subjects were to be randomised per treatment group.
Analyses of all metabolic indices were performed for all randomised subjects who took at least one dose of study medication. For all parameters, median values and interquartile ranges at baseline (defi ned as the last measurement before administration of the fi rst study medication) and treatment cycles 3 and 6 are presented, as well as changes from baseline to cycle 6 (primary outcome). Interquartile ranges were preferred over standard deviations in order to limit the infl uence of outliers on the variability estimates. If a value less than the lower limit of quantifi cation (LLOQ) was observed for a metabolic variable, a value of 0.5 ϫ LLOQ for the respective laboratory test was imputed. If a value above the upper limit of quantifi cation (ULOQ) was reported, a value equal to the ULOQ of the respective laboratory test was used.
Glucose and insulin responses during the OGTT were evaluated as the area under the curve over three hours (AUC 3 ), which were calculated according to the trapezoidal rule. A correction factor was used if the last blood sample was not drawn at the exact time of 180 minutes after glucose administration. Based on the calculated AUC 3 , the incremental AUC 3 was defi ned as the difference between the AUC 3 and the concentration before OGTT (fasting concentration) multiplied by three hours.
All statistical analyses of metabolic parameters used two-sided tests performed at the 5% error level. Each parameter was analysed using a stratifi ed Wilcoxon rank sum test (Cochran-Mantel-Haenszel test) adjusted for age class, applied on the changes from baseline to cycle 6. No correction for multiplicity was applied.

Haemostasis
Prothrombin fragments 1 ϩ 2 essentially did not change in the NOMAC/E2 group, whereas a small increase in the LNG/EE group was observed ( p ϭ 0.085; Table 2; Figure 2). Although D-dimer was nearly unchanged in both groups, these results are inconclusive because more than 50% of the values were below the limit of detection.
The ETP-based APC sensitivity ratio increased by a statistically signifi cant margin from baseline to cycle 6 in both groups; however, the change was much greater with LNG/EE than with NOMAC/E2 ( p Ͻ 0.001; Table 2; Figure 2). The aPTT-based sensitivity ratio was nearly unchanged in both groups. Small changes from baseline were observed in the anticoagulatory factors ATIII, protein C, and free and total protein S in both treatment groups (Table 2), but the between-group differences for these parameters were found to be statistically signifi cant ( p Ͻ 0.01; Table 2) except for free protein S.
NOMAC/E2 and LNG/EE induced minimal changes from baseline in procoagulatory factors II, VIIa and VIII (Table 2). For factor VIIc, a minimal change from baseline was observed in the NOMAC/ E2 group, whereas a decrease was observed in the LNG/EE group ( p ϭ 0.001; Table 2).

R E S U L T S
A total of 121 women were randomised to receive either NOMAC/E2 or LNG/EE ( Figure 1). All 60 women in the NOMAC/E2 group received treatment; three of the 61 women in the LNG/EE group did not receive treatment because of a pre-treatment adverse event (AE) ( ' acne ' ), withdrawn consent or other personal reasons (found a new job). Seven women (11.7%) in the NOMAC/E2 group and six women (10.3%) in the LNG/EE group discontinued the treatment prematurely. Of these 13 discontinuers, eight women (four in each group) discontinued due to an AE, one due to pregnancy wish, one moved to another city, and three women were lost to follow-up. Overall tablet intake compliance was high in both treatment groups: 93.1% and 87.7% of women in the NOMAC/E2 and LNG/EE groups, respectively, took the daily tablet on at least 95% of treatment days.
The demographic and clinical characteristics were similar between the two treatment groups at baseline (Table 1), with the exception of smoking prevalence, which was higher among women in the NOMAC/E2 group. Baseline values for all metabolic indices in the two treatment groups were also similar (Tables 2 -4).

Lipids
No clinically relevant changes were seen in total cholesterol, HDL-C, LDL-C, or total triglycerides among women receiving NOMAC/E2 during six treatment cycles (Table 3; Figure 2). LNG/EE treatment did not change total cholesterol either, but it decreased HDL-C, increased LDL-C, and increased total triglycerides. The decrease from baseline in HDL-C and the increase from baseline in LDL-C and total triglycerides observed in the LNG/EE group were statistically signifi cantly different from changes observed in the NOMAC/E2 group ( p Ͻ 0.05; Table 3). No changes in lipoprotein (a) were observed in the NOMAC/E2 group, and small changes were observed in the LNG/EE group, the difference being statistically signifi cant ( p Ͻ 0.001).
Women in the NOMAC/E2 group had a significantly greater increase from baseline in apolipoprotein A1 ( p ϭ 0.006) and a signifi cantly smaller increase in apolipoprotein B ( p Ͻ 0.001) than those in the LNG/EE group.

Carbohydrate metabolism
Over six cycles of treatment, women receiving NOMAC/E2 had negligible changes from baseline in the AUC 3 and incremental AUC 3 for both glucose and insulin (Table 4). In the LNG/EE group, increases from baseline were observed for all four parameters. The differences between treatment groups were statistically signifi cant for all of these indices ( p Յ 0.002; Table 4). No changes in HbA 1c were observed in either group.   FEU, Fibrinogen equivalent units; ETP, endogenous thrombin potential; aPTT, activated partial thromboplastin time; APC, activated protein C; Factor VIIa, activated Factor VII; factor VIIc, coagulated activated Factor VII. * The Cochran-Mantel-Haenszel test adjusted for age class using standardised midranks, applied on the absolute changes from baseline to cycle 6. A p -value Յ 0.05 indicates that the difference from baseline to cycle 6 (NOMAC/ E2 vs LNG/EE) was statistically signifi cant. † The p -value is not presented because Ͼ 50% of subjects had D-dimer values below the lower limit of quantifi cation (LLOQ) of 0.22 mg/l FEU.

Contraceptive effi cacy and tolerability
No pregnancies occurred during the trial in either treatment group. NOMAC/E2 was generally well tolerated, with a similar AE profi le as LNG/EE. AEs reported with an incidence Ն 5% (in any treatment group) in women who received either NOMAC/E2 or LNG/EE, respectively, were upper respiratory tract infection (6 and 5 subjects), headache (3 and 7 subjects), acne (2 and 4 subjects), infl uenza (1 and 4 subjects), metrorrhagia (3 and 1 subject), and vaginal candidiasis (3 and 1 subject). One serious adverse event (SAE), worsening of a congenital mitral valve leak, was reported in the NOMAC/E2 group. The subject was withdrawn from the study. No SAEs were reported in the LNG/EE group. Eight women (four women in each group) discontinued treatment during the study because of an AE. In the NOMAC/E2 group, reasons for discontinuation due to an AE included depression and nausea; one subject experienced a combination of tachycardia, pain in the calf, and weakness in the limb. In the LNG/EE group, AEs that led to discontinuation included decreased sexual desire, nausea, and headache.

D I S C U S S I O N
This randomised study compared the effects of NOMAC/E2 administered in a 24-day regimen and LNG/EE in a 21-day regimen on metabolic indices. contraceptive effi cacy and safety since their introduction in 1960. However, epidemiological evidence linking COC use with an increased risk of cardiovascular morbidity, particularly of VTE, remains a cause for concern. Several factors have been associated with VTE risk in women taking EE-containing COCs, including the EE dose 20,21 and type of progestogen used 21 -25 . Whether observed differences in VTE risk among COC formulations can be explained by confounding factors and biases in study design and After six cycles of treatment, NOMAC/E2 had minimal infl uence on parameters related to haemostasis, lipids, and carbohydrate metabolism. Overall, NOMAC/E2 had less effect on these indices than the comparator LNG/EE. Both NOMAC/E2 and LNG/ EE were associated with moderate increases in SHBG and increases in CRP. Changes in CRP were particularly marked with LNG/EE. An estimated 100 million women worldwide use COCs 1 , which have demonstrated an excellent overall HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; CRP, C-reactive protein; SHBG, sex hormone-binding globulin. * The Cochran-Mantel-Haenszel test adjusted for age class using standardised midranks, applied on the absolute changes from baseline to cycle 6. A p -value Յ 0.05 indicates that the difference from baseline to cycle 6 (NOMAC/ E2 vs LNG/EE) was statistically signifi cant.
methodology remains the subject of debate 5 . Reductions in the dose of EE ( Ͻ 50 μ g) have resulted in a decreased incidence of VTE 21 , ischaemic stroke 25,26 , and myocardial infarction 25 . Further reduction of the EE dose to 20 μ g may further decrease the VTE incidence (suggested in a case-control study 21 ), but the evidence for this is limited. Even very low doses of EE (i.e., 10 μ g) have been shown to adversely affect haemostatic parameters 27 . COC formulations containing oestrogen that are structurally similar or identical to endogenous E2 have been sought for several decades because they are expected to offer an improved safety profi le. A number of clinical studies have demonstrated that E2 and its derivative oestradiol valerate have less effect on hepatic protein synthesis than EE and have more favourable lipid and haemostatic profi les 27 -29 . This is confi rmed by data from the present study, in that NOMAC/E2 had less infl uence on indices of haemostasis, lipid and carbohydrate metabolism than LNG/EE (150 μ g/30 μ g), as well as by the results of a previous study, which showed that NOMAC/E2 induced smaller changes on coagulation and fi brinolysis markers relative to the lower-dosed COC LNG/EE 100 μ g/20 μ g 30 . The apparent improved safety profi le of E2-containing COCs needs to be confi rmed in clinical studies with VTE as endpoint.
COCs containing EE have been associated with pronounced APC resistance 31,32 as well as elevations in procoagulatory factors and reductions in anticoagulatory factors 33,34 . Reduced sensitivity for ETP-based APC has been claimed to be associated with VTE risk in men and women, even in the absence of the Factor V Leiden mutation 35,36 . In the current study, NOMAC/ E2 resulted in changes in ETP-based APC sensitivity ratios that were substantially and signifi cantly smaller than ratios observed with LNG/EE.
Evidence suggests that the individual composition of different COCs, in terms of hormone dose and progestogen type, also infl uences their respective effects on lipids and lipoproteins 37,38 . Although the changes in lipoprotein metabolism induced by EE are complex, the net effect is generally considered favourable, primarily due to increased HDL concentrations. Progestogens with androgenic properties may reverse the lipid benefi ts of oestrogen, whereas natural progesterone, progesterone-like derivatives and DRSP may be more likely to preserve these benefi ts 39 . In the present study, NOMAC/E2 did not induce any meaningful changes in the lipid profi le, whereas women receiving LNG/ EE demonstrated a reduction in HDL-C and an elevation in LDL-C and total triglycerides. These results are consistent with those observed in previous studies of LNG/EE 40 -42 . Hormonal contraceptives (including COCs) have been associated with subclinical disturbances in carbohydrate metabolism, including impaired glucose tolerance and insulin resistance 43 that may increase the risk of type 2 diabetes and vascular disease. Studies evaluating the effects of oestrogen on carbohydrate metabolism have reported contradictory fi ndings 44 -50 . The androgenic properties of the proge stogens used in COCs may infl uence responses to glucose tolerance tests; non-androgenic progestogens appear to exert a neutral effect on carbohydrate metabolism 51 . In the present study, negligible changes in glucose and insulin parameters were observed in women taking NOMAC/E2, whereas signifi cant increases in insulin and glucose AUC 3 were found in women taking LNG/EE. The increases associated with LNG/EE are in line with those observed in previous studies 40,42 , but it is doubtful whether they are clinically meaningful in terms of predicting greater risk.
The role of infl ammation in the pathogenesis of cardiovascular disease (CVD) has been increasingly recognised in recent years, and infl ammation markers such as the non-specifi c acute-phase protein CRP have been identifi ed as useful predictors of CVD risk 52 . In clinical studies, elevations in CRP levels have been observed with the use of various COC formulations 53 -55 . Both NOMAC/E2 and LNG/EE increased CRP levels in this randomised metabolic study, with a signifi cantly greater increase seen with LNG/EE. Whereas the median CRP level observed in the NOMAC/E2 group at cycle 6 (0.70 mg/l) remained well below the threshold indicating increased cardiovascular risk ( Ն 2 mg/l) 56 , the median CRP level in the LNG/EE group at cycle 6 (1.70 mg/l) approached this threshold.
SHBG is a carrier protein synthesised in hepatocytes; its synthesis has been shown to be highly sensitive to oestrogens and androgens 57 . EE administered orally markedly increases SHBG levels in a dosedependent fashion, whereas progestogens, due to their androgenic properties, decrease these levels to a varying extent depending on the steroid dose and type. The net effect of a COC formulation on SHBG thus appears to refl ect the balance between oestrogenic and androgenic activity 57 . In this study, NOMAC/E2 induced an approximate 44% increase from baseline in SHBG, a change that was signifi cantly greater than the 22% rise induced by LNG/EE. However, compared with other COCs, where increases in SHBG vary from 150% for COCs containing norgestimate to 400% for COCs containing cyproterone acetate, the SHBG increases observed in this study were small 57 . The rise in SHBG induced by LNG/EE is most likely the result of the increase due to EE and the decrease due to LNG, a progestogen with residual androgenicity. The rise in SHBG induced by NOMAC/E2 is most likely only due to the oestrogen component of this combination because NOMAC is a non-androgenic progestogen.
This is one of the fi rst studies reporting the effect of a COC that contains E2 instead of EE on several metabolic indices. The VTE risk and metabolic effects of the comparator COC LNG/EE (150 μ g/30 μ g) have been well established, and it is therefore an appropriate comparator. The infl uence of NOMAC/E2 on haemostatic indices has also been studied in comparison to the lower-dosed COC containing LNG/EE (100 μ g/20 μ g), with essentially similar results 30 . Although direct comparisons between NOMAC/E2 and COCs containing progestogens other than LNG are not yet available, it is likely that these COCs are associated with greater changes in these indices than NOMAC/E2, in view of their generally more pronounced effects in comparison to LNG/EE 58,59 .
The main limitation of this study is the use of surrogate endpoints for metabolic indices. In addition, none of the haemostatic indices measured in this study have been established as defi nitive markers of thrombosis, and the clinical relevance of the differences found in this study can only be determined by performing large, clinical studies with VTE as endpoint. The same holds true for other indices, including lipids and carbohydrate metabolism.
In summary, NOMAC/E2 had minimal infl uence on haemostatic, lipid and carbohydrate metabolism indices and caused less change in these parameters than LNG/EE. Large clinical studies will be needed to Declaration of interest: U. M. Å gren, W. F. Sommer, and E. Mommers are employees of MSD. At the time this study was performed, U. M. Å gren did not work for MSD, but she was an employee of FSHS Kuopio, Finland. M. Anttila, K. M ä enp ä ä -Liukko, M. L. Rantala, and H. Rautiainen declare that they have no confl icts of interest. They also declare that they received no compensation or honoraria in association with this article. determine whether these differences in surrogate endpoints are also clinically relevant.