- Email: [email protected]
Study design for the Triphasic Randomized Clinical Trial
Grimes
provide the most confident answers to important questions. Outcomes appropriate for study by randomized clinical trials include breakthrough bleeding and metabolic effects. Randomized clinical trials should be done whenever feasible to examine important questions relating to triphasic oral contraceptives-and to all other medications. Formal, well-designed experiments operate in the interests of patients, whereas informal and poorly controlled studies jeopardize patient care. Randomized trials help to contain risks to patients posed by unpredieted adverse effects of incompletely or inaccurately evaluated treatments. Through use of randomized trials the medical community can maximize the number of persons who receive optimal care as new products evolve.IO REFERENCES 1. Daniel I: 11-6. 2. Feinstein AR. Clinical epidemiology: the architecture of
November 1989 Am J Obstet Gynecol
3. 4.
5. 6. 7. 8.
9. 10.
clinical research. Philadelphia: WB Saunders, 1985:683718. Hennekens CH, Burling]E. Epidemiology in medicine. Boston: Little, Brown and Co, 1987:178-212. Fisher B, Redmond C, Fisher ER, et al. Ten-year results of a randomized clinical trial comparing radical mastectomy and total mastectomy with or without radiation. N Engl] Med 1985;312:674-81. Francis T ]r, Korns FT, Voight RB, et al. An evaluation of the 1954 poliomyelitis vaccine trials: summary report. Am 1 Public Health 1955;45: 1-63. Wilhelmsen L, Ljungberg S, Wedel H, et al. A comparison between participants and non-participants in a primary preventive trial.] Chron Dis 1976;29:331-9. Schafer A. The ethics of the randomized clinical trial. N Engl 1 Med 1982;307:719-24. Burkman RT, Robinson lC, Kruszon-Moran D. et al. Lipid and lipoprotein changes associated with oral contraceptive use: a randomized clinical trial. Obstet Gynecol 1988;71 :33-8. Marz W, Gross W, Gahn G, et al. A randomized crossover comparison of two low-dose contraceptives. AM 1 OBSTET GYNECOL 1985;153:287-93. Chalmers I, Silverman WA. Professional and public double standards on clinical experimentation. Controlled Clin Trials 1987;8:388-91.
Study design for the Triphasic Randomized Clinical Trial Janet K. Dickerson, RN Tucson, Arizona The Triphasic Randomized Clinical Trial was a multicenter, open-label, controlled, 6-month study to determine the influence of three currently marketed triphasic oral contraceptives on lipid and carbohydrate metabolism, cyclic bleeding profiles, and other adverse drug experiences. The study protocol mandated four clinic visits: an initial visit during which study eligibility was determined, a second visit during which baselines were obtained for lipid and carbohydrate parameters, and visits at 3 and 6 months while taking the study drug. The study enrolled healthy women between the ages of 18 and 35 years who were within 20% of their ideal body weight and who had a history of normal and regular menstrual cycles. Patients were randomly assigned to one of three treatment groups: 51 patients to Tri-Levlen, 50 patients to Ortho-Novum 71717, and 56 patients to Tri-Norinyl. The control group consisted of 49 patients using non hormonal forms of birth control. Eighty-eight percent (181/206) of the patients completed all four visits and were considered evaluable for subsequent analyses. (AM J OBSTET GVNECOL 1989;161 :1392-6.)
Key words: Tri-Levlen, Ortho-Novum 7!7 17, Tri-Norinyl, randomized clinical trial The Triphasic Randomized Clinical Trial was an open-label, multicenter, controlled study to determine the effects of three marketed oral contraceptives on lipid and carbohydrate metabolism, adverse experiFrom the Department of ObstetriCs and Gynecology, Arizona Health Sciences Center. Reprint requests: Janet Dickerson, RN, Research Department of Obstetrics and G.vnecology, Arizona Health Sciences Center, 1501 N. Campbell Ave., Tucson, AZ 85724. 610116125
1392
ences, and cyclic bleeding profiles. The three treatment groups and a control group were followed up for six treatment cycles. The institutions participating in this study were the Departments of Obstetrics and Gynecology at (1) the University of Southern California, (2) the University of North Carolina, (3) Baylor College of Medicine, (4) Johns Hopkins School of Medicine, and (5) the University of Arizona. The primary investigators at each site, respectively, were Drs. Daniel Mishell and David
TRCT study design
Volume 161 Number 5
1393
Table I. Patient demographics at entry (for all patients enrolled) Group Control (n = 49)
Age (yr) Weight (pounds) Height (in) Race (n /% ) White Black Hispanic Oriental
(n
LNG = 51)
NET (A) (n = 50)
NET (B) (n = 56)
29.7 :t 4.3 136.4 :t 17.2 65. 1 :t 3.0
27.3 :t 3.8 131.5:t 18.9 65. 1 :t 2.2
26.5 :t 4.6 132.8 :t 21.6 65.2 :t 2.7
25.3 :t 4.6 129.8 :t 14.6 64.7 :t 2.4
39 / 80 4/8 6112 0/0
36/ 71 7114 8116
40/80
45/80 5/9 5/9 1/2
5/10
4/8 1/2
0 10
Data for age, weight, and height are mean :t SD. Note: Because of rounding, totals may not equal 100%.
Table II. Menstrual and contraceptive history for patients at entry (for all patients enrolled) Group Control (n = 49)
Cycle duration (days) Menses duration (days) Previous oral contraceptive use (n/%) Previous pregnancy (n/%)
(n
LNG = 51)
(n = 50)
NET (A)
NET (B) (n = 56)
28.4 :t 2.3 4.6 :t 1.0 42 / 86
29 :t 1.8 4.3 :t 1.1 46/90
29:t 2.1 4.5 :t 1.0 38/76
28.8 :t 1.8 4.7 :t l.l 33/59*
27/55
24/47
24/48
15/27
Data for cycle and menses duration are mean :t SD. *Statistically significant difference from LNG, p :5 0.001.
Grimes, Drs. Watson Bowes and William Droegemueller, Dr. Ronald Young, Dr. Laura Rao Katta, and Dr. C. Donald Christian. Entry criteria
The subjects enrolled in this study were healthy women between the ages of 18 and 35 years who were within 20% of their ideal body weight and who had normal and regular menstrual cycles (Tables I and II). The definition of a "normal" cycle was one that had a duration of 25 to 32 days, and the definition of regularity was that cycle length was the same from cycle to cycle ± 2 days . Nursing mothers were excluded; pregnancy (whether it was term or aborted) in the prior 3 months also excluded potential participants. Study subjects were required to have no other contraindications to oral contraceptive use and not to have used any injectable or oral contraceptive in the previous 3 months. In practice, most subjects had not taken oral contraceptives for more than 1 year before the study. Women who had been smokers within the past year were excluded. They were also excluded if they had a history of diabetes, a diabetic primary relative, gestational diabetes, birth of an infant larger than 9 pounds, or other clinically significant diseases or were
taking chronic medication. Exclusion criteria also included clinically significant abnormal physical or laboratory findings, specifically elevated fasting serum cholesterol (>240 mg/dl), elevated serum triglyceride (> 145 mg/ml), or elevated serum glucose levels (> 120 mg/ml). Other exclusion criteria were abnormal Papanicolaou smears (class III, IV, or V) or positive culture for gonorrhea. Before being enrolled in the study, subjects were required to express a willingness to maintain their dietary and exercise habits. All subjects gave signed, informed consent. Study design
The study protocol called for four clinic visits. At visit 1, subjects were screened for study eligibility and baseline values were established. They also underwent an extensive interview to determine their medical and family histories and their habits with regard to smoking, alcohol consumption, diet, and exercise. A complete blood count, urinalysis, serum chemistry, Papanicolaou smear, and culture for gonorrhea were performed, as well as physical and pelvic examinations. At visit 2, scheduled between days 17 and 21 of the pretreatment menstrual cycle, blood samples were obtained, after a 14-hour fast, for lipid levels and glucose tolerance (including measurements of insulin). At this point subjects
1394 Dickerson
November 1989 Am J Obstet Gynecol
Table III. Lipid profiles at entry (for all patients enrolled) Group Control (n = 49)
Cholesterol (mg/dl) Triglyceride (mg/dl) High-density lipoprotein (mg/dl) Low-density lipoprotein (mgl dl) High-density lipoproteins (mg I dl) High-density lipoprotein2 (mg/dl) Apolipoprotein A-I (mg/dl) Apolipoprotein B (mg/dl)
174.3 63.4 64.3 97.3 36.9 27.5 155.0 58.2
25 .6 20.9 18.3 24.9 8.0 14.6 ± 31.4 ± 17.4 ± ± ± ± ± ±
LNG
NET (A)
(n = 51)
176.0 74.3 60.3 100.9 33.3 26.9 135.2 56.3
NET (B)
(n = 50)
± 33.7 ± 35.3 ± 13.0
173.9 69.0 64.0 96.1 35.1 28.9 140.1 54.4
± 29.8 ± 8.5
± 10.1 ± 28.6 ± 15.5
(n = 55)
± 25.4
176.4 72.6 60.0 101.8 33.3 26.9 141.6 60.6
± ± ± ±
32.5 15.9 22.6 9.6 ± 9.7 ± 33.0 ± 13.2
± 32.1 ± 24.8
± 16.0
± 29.5
± 7.3
± 14.5 ± 31.3 ± 20.0
Data are means ± SD.
Table IV. Carbohydrate profiles at entry (for all patients enrolled) Group
Glucose area under the curve (hr X mg/dl) Insulin area under the curve (hr X J.Lg/ dl)
Control
LNG
NET (A)
NET (B)
257.6 ± 42.3
269.4 ± 50.7
261.1 ± 52.8
263.0 ± 46.9
(n = 48) 190.2 ± 95 .0
(n = 51) 214.1 ± 163.4
(n = 50) 194.5 ± 78.4
(n = 55) 212.0 ± 89.2
(n = 49)
(n = 51)
(n = 50)
(n = 55)
Data are means ± SD.
Table V. Patient enrollment Reason for discontmuing drug (No.) Treatment group
No. patients enrolled
No. patients completed (nl %)
Adverse drug reactIOn
Control LNG NET (A) NET (B) Total
49 51 50 56 206
45/92 45/88 44/88 47/84
0 2 0 0 2
181188
I
Pregnancy*
3 1 0 1
:5
I
M tScellaneous
1 3 6 8 18
*No pregnancy occurred in patients taking oral contraceptives. The pregnancies that occurred in patients randomized to oral contraceptives occurred before the patients started taking the medication.
were randomly assigned to one of the three open-label treatment groups, or for women who did not want to use an oral contraceptive, to the control group. Subjects returned for visits 3 and 4 between days 17 and 21 of their third and sixth menstrual cycles; glucose, insulin, glucose tolerance, and lipid panels were repeated at each visit.
(P ::s 0.001) in the number of women who had previously used oral contraceptives before the study in the LNG (Tri-Levlen) group versus the NET (8) (TriNorinyl) group (Table II). Mean baseline values for carbohydrate and lipid profiles were not significantly different for the three groups (Tables III and IV).
Patient demographics The entry demographics presented in this article are based on all subjects who enrolled in the study ("intention-to-treat" basis). Average age was 27 years; the average cycle duration was 29 days, with 4 to 5 days of menstrual bleeding (Tables I and II). None of these parameters differed significantly from one group to another. There was a statistically significant difference
The control group was required to meet the same entry criteria as the test groups, with the exception that control subjects had selected non hormonal methods of contraception. Among the groups taking oral contraceptives, assignment of the particular treatment was made with a computer-generated randomization code. (The number of patients in each group is approximately, but not exactly, the same because each patient
Treatment groups
TRCT study design
Volume 161 Number 5
1395
NET (A)
c
CD 01 _0 CD
C 11>_
01 CD
o >
III
- CD ;..J
ILl
>
II> CD
.o a..
OI...J
21 Days 35~
LNG
7 Days 500J.l.9
7DlYS 750 J.l.9
7 Days 1000 J.l.9
c:
CD 01 _0 CD
c
CD_ 01 CD
o >
III
>
CD CD
- CD ii)..J
6 Days
ILl
30 J.l.9
5 Days 4°1J.9
..
OI..J
o a..
10 Days 30~
6 Days 50 J.l.9
5 Days 7SIJ.g
10 Days 1251J.9
NET (8) c
CD 01 _0 CD
cCD_
01 CD
o > - CD ii)..J
ILl
III
-
>
CD CD
OI..J
21 Days 35 J.I.9
o a..
NET (A) - Ortho-Novum@7/7/7-28
7 Days 500 J.l.9
LNG - Trl-Levlen@-28
9 Days 1000 1J.9
5 Days SOOJ.l.9
NET (B) - Trl-Norlnyl@-28
Fig_ 1. Oral triphasic dosages.
was individually assigned at random and some centers did not use all the numbers in their assigned randomization codes.) The treatment groups received one of three marketed oral triphasic oral contraceptives: (1) LNG (Tri-Levlen-28), (2) NET (A) (Ortho-Novum 71717-28), or (3) NET (B) (Tri-Norinyl-28). Each of these triphasic oral contraceptives is somewhat different in its estrogen-progestin phasing (Fig. 1). LNG alters dosages of both compounds (ethinyl estradiol and levonorgestrel) during the 21-day treatment period, whereas the other two preparations alter only the dosages of the norethindrone. Methods Each subject was required to keep a daily diary of her menstrual cycles and record compliance with pill ingestion throughout the study period. On visits 2, 3, and 4, when samples for the lipid profile and carbohydrate profile were to be obtained, patients reported to the research units after a 14-hour fast. Twenty milliliters of whole blood for the lipid estimations was collected in Venoject tubes (Terumo Medical Corp., Elkton, Md.) containing 15 mg of dry EDTA. Blood samples were placed on ice and centrifuged within I hour after collection. The plasma was aliquoted into eight separate containers: four vials containing 2.0 ml of
plasma each and four vials containing 0.2 ml of plasma each. The samples were frozen immediately and stored at the research site. Five milliliters of whole blood (O-hour sample) was collected in an EDTA Venoject tube for glucose and insulin assays. The volunteer was given 100 gm of glucose orally in a 100 ml solution, after which 5 ml of blood was drawn at 1, 2, and 3 hours . Samples were placed on ice and centrifuged within 1 hour of drawing. Plasma was aliquoted into two vials, each containing 0.8 ml. Remaining plasma was saved in a third vial. These samples were also frozen immediately. All samples for both lipids and glucoselinsulin were packed with dry ice in Styrofoam boxes and sent to the Atherosclerosis Laboratory, Baylor College of Medicine, Houston, Texas, for analysis. Two of the lipid aliquots containing 2 ml and the remaining plasma from the glucose tolerance test were stored at the research site. These duplicate samples were used in the event that an assay had to be repeated. A total of 157 patients were randomized to receive one of the oral contraceptives as follows : 51 to LNG, 50 to NET (A) , and 56 to NET (B). The control group consisted of 49 patients at entry. Eighty-eight percent (1811206) of patients completed the 6-month study (Table V). Twenty-five patients did not complete the
Dickerson
study because of side effects (2 patients), pregnancy (5 patients), or miscellaneous reasons (18 patients). Data were analyzed on the basis of both all patients enrolled (intent to treat) and patients who completed all three visits, with essentially the same results. Those patients who were evaluated for lipid and carbohydrate profiles at baseline (pretest) and cycles 3 and 6 were included
November 1989 Am J Obstet Gynecol
the analysis of these parameters presented in the subsequent articles of this series. Evaluation of bleeding irregularities (intermenstrual bleeding) was based on the population of patients who completed at least one cycle of treatment and are calculated on a per-cycle basis. III
The effect of triphasic oral contraceptives on plasma lipids and lipoproteins Wolfgang Patsch, MD, Spencer A. Brown, PhD, Antonio M. Gotto, Jr., MD, DPhil, and Ronald L. Young, MD Houston, Texas To determine the effect of triphasic oral contraceptives on plasma lipid transport, 150 nonsmokIng women with normolipidemia, ages 18 to 35 years, were randomly assigned to receive one of three contraceptive formulations: (1) ethinyl estradiol, 30, 40, and 30 fLg/day, each for 6, 5, and 10 days per menstrual cycle, and levonorgestrel, 50, 75, and 125 fLg/day, each for 6,5, and 10 days; (2) ethinyl estradiol, 35 fLg/day for 21 days, and phased norethindrone, 500, 750, and 1000 fLg/day each for 7 consecutive days; and (3) ethinyl estradiol, 35 fLg/day for 21 consecutive days, and norethindrone, 500, 1000, and 500 fLg/day for 7, 9, and 5 days, respectively. A control group consisting of 49 women taking a nonhormonal form of contraception was also included. After 6 months of oral contraceptive treatment, significant increases in plasma triglyceride (28% to 52%) and plasma apolipoprotein B levels (20% to 23%) were observed in each treatment group. The changes in total plasma cholesterol (3% to 10%) and low-density lipoprotein cholesterol values (0% to 11%) were less striking . Changes in total high-density lipoprotein cholesterol levels were statistically insignificant (-2% to -4%); however, high-density lipoprotein2 cholesterol levels decreased by 29% to 33% and high-density lipoprotein 3 cholesterol levels increased by 20% to 23%. Concomitantly, plasma apolipoprotein A-I values increased by 5% to 12%. No consistent significant differences among analyses were observed between any of the groups receiving different oral contraceptives for 6 months. (AM J OaSTET GYNECOL 1989;161 :1396·401.)
Key words: Oral contraceptives, lipids, cholesterol, comparative trial
A number of large epidemiologic studies have demonstrated a relationship between oral contraceptive use and myocardial infarction and other vascular disorders. I•7 Attention has been directed toward the effects of oral contraceptives on plasma lipid transport, and increased concentrations of atherogenic lipoproteins in the circulation are the best established risk factors of coronary heart disease." Several studies have demonstrated changes of fasting plasma lipid levels as a result of contraceptive drugs . Depending on the dose, potency, and estrogen-progestogen content of the drug, From the Department of MediCine, Baylor College of Mediczne, and The Methodtst Hospital. R epnnt requests: Wolfgang Patsch, MD , Department of Medicine, Baylor College of Medzcine, Mall StatIOn A-60l, 6565 Fannin, Houston, TX 77030.
610116120
1396
increases in plasma triglyceride, cholesterol, very lowdensity lipoprotein, low-density lipoprotein, and highdensity lipoprotein values were observed.9- 16 In the great majority of these studies, monophasic preparations were tested. Data from such trials and other studies on sex hormones indicate that increases of highdensity lipoprotein cholesterol and plasma triglyceride levels are caused by the estrogen component, whereas increases of low-density lipoprotein cholesterol and a decrease of high-density lipoprotein cholesterol values reflect the androgenic potency of the progestogen in the formulation. Thus to a certain extent, effects of monophasic preparations on plasma lipid levels may be predicted from the composition of the drug. Much less is known about effects of biphasic and triphasic oral contraceptives, which are now the most frequently prescribed. We report here the results of a randomized
provide the most confident answers to important questions. Outcomes appropriate for study by randomized clinical trials include breakthrough bleeding and metabolic effects. Randomized clinical trials should be done whenever feasible to examine important questions relating to triphasic oral contraceptives-and to all other medications. Formal, well-designed experiments operate in the interests of patients, whereas informal and poorly controlled studies jeopardize patient care. Randomized trials help to contain risks to patients posed by unpredieted adverse effects of incompletely or inaccurately evaluated treatments. Through use of randomized trials the medical community can maximize the number of persons who receive optimal care as new products evolve.IO REFERENCES 1. Daniel I: 11-6. 2. Feinstein AR. Clinical epidemiology: the architecture of
November 1989 Am J Obstet Gynecol
3. 4.
5. 6. 7. 8.
9. 10.
clinical research. Philadelphia: WB Saunders, 1985:683718. Hennekens CH, Burling]E. Epidemiology in medicine. Boston: Little, Brown and Co, 1987:178-212. Fisher B, Redmond C, Fisher ER, et al. Ten-year results of a randomized clinical trial comparing radical mastectomy and total mastectomy with or without radiation. N Engl] Med 1985;312:674-81. Francis T ]r, Korns FT, Voight RB, et al. An evaluation of the 1954 poliomyelitis vaccine trials: summary report. Am 1 Public Health 1955;45: 1-63. Wilhelmsen L, Ljungberg S, Wedel H, et al. A comparison between participants and non-participants in a primary preventive trial.] Chron Dis 1976;29:331-9. Schafer A. The ethics of the randomized clinical trial. N Engl 1 Med 1982;307:719-24. Burkman RT, Robinson lC, Kruszon-Moran D. et al. Lipid and lipoprotein changes associated with oral contraceptive use: a randomized clinical trial. Obstet Gynecol 1988;71 :33-8. Marz W, Gross W, Gahn G, et al. A randomized crossover comparison of two low-dose contraceptives. AM 1 OBSTET GYNECOL 1985;153:287-93. Chalmers I, Silverman WA. Professional and public double standards on clinical experimentation. Controlled Clin Trials 1987;8:388-91.
Study design for the Triphasic Randomized Clinical Trial Janet K. Dickerson, RN Tucson, Arizona The Triphasic Randomized Clinical Trial was a multicenter, open-label, controlled, 6-month study to determine the influence of three currently marketed triphasic oral contraceptives on lipid and carbohydrate metabolism, cyclic bleeding profiles, and other adverse drug experiences. The study protocol mandated four clinic visits: an initial visit during which study eligibility was determined, a second visit during which baselines were obtained for lipid and carbohydrate parameters, and visits at 3 and 6 months while taking the study drug. The study enrolled healthy women between the ages of 18 and 35 years who were within 20% of their ideal body weight and who had a history of normal and regular menstrual cycles. Patients were randomly assigned to one of three treatment groups: 51 patients to Tri-Levlen, 50 patients to Ortho-Novum 71717, and 56 patients to Tri-Norinyl. The control group consisted of 49 patients using non hormonal forms of birth control. Eighty-eight percent (181/206) of the patients completed all four visits and were considered evaluable for subsequent analyses. (AM J OBSTET GVNECOL 1989;161 :1392-6.)
Key words: Tri-Levlen, Ortho-Novum 7!7 17, Tri-Norinyl, randomized clinical trial The Triphasic Randomized Clinical Trial was an open-label, multicenter, controlled study to determine the effects of three marketed oral contraceptives on lipid and carbohydrate metabolism, adverse experiFrom the Department of ObstetriCs and Gynecology, Arizona Health Sciences Center. Reprint requests: Janet Dickerson, RN, Research Department of Obstetrics and G.vnecology, Arizona Health Sciences Center, 1501 N. Campbell Ave., Tucson, AZ 85724. 610116125
1392
ences, and cyclic bleeding profiles. The three treatment groups and a control group were followed up for six treatment cycles. The institutions participating in this study were the Departments of Obstetrics and Gynecology at (1) the University of Southern California, (2) the University of North Carolina, (3) Baylor College of Medicine, (4) Johns Hopkins School of Medicine, and (5) the University of Arizona. The primary investigators at each site, respectively, were Drs. Daniel Mishell and David
TRCT study design
Volume 161 Number 5
1393
Table I. Patient demographics at entry (for all patients enrolled) Group Control (n = 49)
Age (yr) Weight (pounds) Height (in) Race (n /% ) White Black Hispanic Oriental
(n
LNG = 51)
NET (A) (n = 50)
NET (B) (n = 56)
29.7 :t 4.3 136.4 :t 17.2 65. 1 :t 3.0
27.3 :t 3.8 131.5:t 18.9 65. 1 :t 2.2
26.5 :t 4.6 132.8 :t 21.6 65.2 :t 2.7
25.3 :t 4.6 129.8 :t 14.6 64.7 :t 2.4
39 / 80 4/8 6112 0/0
36/ 71 7114 8116
40/80
45/80 5/9 5/9 1/2
5/10
4/8 1/2
0 10
Data for age, weight, and height are mean :t SD. Note: Because of rounding, totals may not equal 100%.
Table II. Menstrual and contraceptive history for patients at entry (for all patients enrolled) Group Control (n = 49)
Cycle duration (days) Menses duration (days) Previous oral contraceptive use (n/%) Previous pregnancy (n/%)
(n
LNG = 51)
(n = 50)
NET (A)
NET (B) (n = 56)
28.4 :t 2.3 4.6 :t 1.0 42 / 86
29 :t 1.8 4.3 :t 1.1 46/90
29:t 2.1 4.5 :t 1.0 38/76
28.8 :t 1.8 4.7 :t l.l 33/59*
27/55
24/47
24/48
15/27
Data for cycle and menses duration are mean :t SD. *Statistically significant difference from LNG, p :5 0.001.
Grimes, Drs. Watson Bowes and William Droegemueller, Dr. Ronald Young, Dr. Laura Rao Katta, and Dr. C. Donald Christian. Entry criteria
The subjects enrolled in this study were healthy women between the ages of 18 and 35 years who were within 20% of their ideal body weight and who had normal and regular menstrual cycles (Tables I and II). The definition of a "normal" cycle was one that had a duration of 25 to 32 days, and the definition of regularity was that cycle length was the same from cycle to cycle ± 2 days . Nursing mothers were excluded; pregnancy (whether it was term or aborted) in the prior 3 months also excluded potential participants. Study subjects were required to have no other contraindications to oral contraceptive use and not to have used any injectable or oral contraceptive in the previous 3 months. In practice, most subjects had not taken oral contraceptives for more than 1 year before the study. Women who had been smokers within the past year were excluded. They were also excluded if they had a history of diabetes, a diabetic primary relative, gestational diabetes, birth of an infant larger than 9 pounds, or other clinically significant diseases or were
taking chronic medication. Exclusion criteria also included clinically significant abnormal physical or laboratory findings, specifically elevated fasting serum cholesterol (>240 mg/dl), elevated serum triglyceride (> 145 mg/ml), or elevated serum glucose levels (> 120 mg/ml). Other exclusion criteria were abnormal Papanicolaou smears (class III, IV, or V) or positive culture for gonorrhea. Before being enrolled in the study, subjects were required to express a willingness to maintain their dietary and exercise habits. All subjects gave signed, informed consent. Study design
The study protocol called for four clinic visits. At visit 1, subjects were screened for study eligibility and baseline values were established. They also underwent an extensive interview to determine their medical and family histories and their habits with regard to smoking, alcohol consumption, diet, and exercise. A complete blood count, urinalysis, serum chemistry, Papanicolaou smear, and culture for gonorrhea were performed, as well as physical and pelvic examinations. At visit 2, scheduled between days 17 and 21 of the pretreatment menstrual cycle, blood samples were obtained, after a 14-hour fast, for lipid levels and glucose tolerance (including measurements of insulin). At this point subjects
1394 Dickerson
November 1989 Am J Obstet Gynecol
Table III. Lipid profiles at entry (for all patients enrolled) Group Control (n = 49)
Cholesterol (mg/dl) Triglyceride (mg/dl) High-density lipoprotein (mg/dl) Low-density lipoprotein (mgl dl) High-density lipoproteins (mg I dl) High-density lipoprotein2 (mg/dl) Apolipoprotein A-I (mg/dl) Apolipoprotein B (mg/dl)
174.3 63.4 64.3 97.3 36.9 27.5 155.0 58.2
25 .6 20.9 18.3 24.9 8.0 14.6 ± 31.4 ± 17.4 ± ± ± ± ± ±
LNG
NET (A)
(n = 51)
176.0 74.3 60.3 100.9 33.3 26.9 135.2 56.3
NET (B)
(n = 50)
± 33.7 ± 35.3 ± 13.0
173.9 69.0 64.0 96.1 35.1 28.9 140.1 54.4
± 29.8 ± 8.5
± 10.1 ± 28.6 ± 15.5
(n = 55)
± 25.4
176.4 72.6 60.0 101.8 33.3 26.9 141.6 60.6
± ± ± ±
32.5 15.9 22.6 9.6 ± 9.7 ± 33.0 ± 13.2
± 32.1 ± 24.8
± 16.0
± 29.5
± 7.3
± 14.5 ± 31.3 ± 20.0
Data are means ± SD.
Table IV. Carbohydrate profiles at entry (for all patients enrolled) Group
Glucose area under the curve (hr X mg/dl) Insulin area under the curve (hr X J.Lg/ dl)
Control
LNG
NET (A)
NET (B)
257.6 ± 42.3
269.4 ± 50.7
261.1 ± 52.8
263.0 ± 46.9
(n = 48) 190.2 ± 95 .0
(n = 51) 214.1 ± 163.4
(n = 50) 194.5 ± 78.4
(n = 55) 212.0 ± 89.2
(n = 49)
(n = 51)
(n = 50)
(n = 55)
Data are means ± SD.
Table V. Patient enrollment Reason for discontmuing drug (No.) Treatment group
No. patients enrolled
No. patients completed (nl %)
Adverse drug reactIOn
Control LNG NET (A) NET (B) Total
49 51 50 56 206
45/92 45/88 44/88 47/84
0 2 0 0 2
181188
I
Pregnancy*
3 1 0 1
:5
I
M tScellaneous
1 3 6 8 18
*No pregnancy occurred in patients taking oral contraceptives. The pregnancies that occurred in patients randomized to oral contraceptives occurred before the patients started taking the medication.
were randomly assigned to one of the three open-label treatment groups, or for women who did not want to use an oral contraceptive, to the control group. Subjects returned for visits 3 and 4 between days 17 and 21 of their third and sixth menstrual cycles; glucose, insulin, glucose tolerance, and lipid panels were repeated at each visit.
(P ::s 0.001) in the number of women who had previously used oral contraceptives before the study in the LNG (Tri-Levlen) group versus the NET (8) (TriNorinyl) group (Table II). Mean baseline values for carbohydrate and lipid profiles were not significantly different for the three groups (Tables III and IV).
Patient demographics The entry demographics presented in this article are based on all subjects who enrolled in the study ("intention-to-treat" basis). Average age was 27 years; the average cycle duration was 29 days, with 4 to 5 days of menstrual bleeding (Tables I and II). None of these parameters differed significantly from one group to another. There was a statistically significant difference
The control group was required to meet the same entry criteria as the test groups, with the exception that control subjects had selected non hormonal methods of contraception. Among the groups taking oral contraceptives, assignment of the particular treatment was made with a computer-generated randomization code. (The number of patients in each group is approximately, but not exactly, the same because each patient
Treatment groups
TRCT study design
Volume 161 Number 5
1395
NET (A)
c
CD 01 _0 CD
C 11>_
01 CD
o >
III
- CD ;..J
ILl
>
II> CD
.o a..
OI...J
21 Days 35~
LNG
7 Days 500J.l.9
7DlYS 750 J.l.9
7 Days 1000 J.l.9
c:
CD 01 _0 CD
c
CD_ 01 CD
o >
III
>
CD CD
- CD ii)..J
6 Days
ILl
30 J.l.9
5 Days 4°1J.9
..
OI..J
o a..
10 Days 30~
6 Days 50 J.l.9
5 Days 7SIJ.g
10 Days 1251J.9
NET (8) c
CD 01 _0 CD
cCD_
01 CD
o > - CD ii)..J
ILl
III
-
>
CD CD
OI..J
21 Days 35 J.I.9
o a..
NET (A) - Ortho-Novum@7/7/7-28
7 Days 500 J.l.9
LNG - Trl-Levlen@-28
9 Days 1000 1J.9
5 Days SOOJ.l.9
NET (B) - Trl-Norlnyl@-28
Fig_ 1. Oral triphasic dosages.
was individually assigned at random and some centers did not use all the numbers in their assigned randomization codes.) The treatment groups received one of three marketed oral triphasic oral contraceptives: (1) LNG (Tri-Levlen-28), (2) NET (A) (Ortho-Novum 71717-28), or (3) NET (B) (Tri-Norinyl-28). Each of these triphasic oral contraceptives is somewhat different in its estrogen-progestin phasing (Fig. 1). LNG alters dosages of both compounds (ethinyl estradiol and levonorgestrel) during the 21-day treatment period, whereas the other two preparations alter only the dosages of the norethindrone. Methods Each subject was required to keep a daily diary of her menstrual cycles and record compliance with pill ingestion throughout the study period. On visits 2, 3, and 4, when samples for the lipid profile and carbohydrate profile were to be obtained, patients reported to the research units after a 14-hour fast. Twenty milliliters of whole blood for the lipid estimations was collected in Venoject tubes (Terumo Medical Corp., Elkton, Md.) containing 15 mg of dry EDTA. Blood samples were placed on ice and centrifuged within I hour after collection. The plasma was aliquoted into eight separate containers: four vials containing 2.0 ml of
plasma each and four vials containing 0.2 ml of plasma each. The samples were frozen immediately and stored at the research site. Five milliliters of whole blood (O-hour sample) was collected in an EDTA Venoject tube for glucose and insulin assays. The volunteer was given 100 gm of glucose orally in a 100 ml solution, after which 5 ml of blood was drawn at 1, 2, and 3 hours . Samples were placed on ice and centrifuged within 1 hour of drawing. Plasma was aliquoted into two vials, each containing 0.8 ml. Remaining plasma was saved in a third vial. These samples were also frozen immediately. All samples for both lipids and glucoselinsulin were packed with dry ice in Styrofoam boxes and sent to the Atherosclerosis Laboratory, Baylor College of Medicine, Houston, Texas, for analysis. Two of the lipid aliquots containing 2 ml and the remaining plasma from the glucose tolerance test were stored at the research site. These duplicate samples were used in the event that an assay had to be repeated. A total of 157 patients were randomized to receive one of the oral contraceptives as follows : 51 to LNG, 50 to NET (A) , and 56 to NET (B). The control group consisted of 49 patients at entry. Eighty-eight percent (1811206) of patients completed the 6-month study (Table V). Twenty-five patients did not complete the
Dickerson
study because of side effects (2 patients), pregnancy (5 patients), or miscellaneous reasons (18 patients). Data were analyzed on the basis of both all patients enrolled (intent to treat) and patients who completed all three visits, with essentially the same results. Those patients who were evaluated for lipid and carbohydrate profiles at baseline (pretest) and cycles 3 and 6 were included
November 1989 Am J Obstet Gynecol
the analysis of these parameters presented in the subsequent articles of this series. Evaluation of bleeding irregularities (intermenstrual bleeding) was based on the population of patients who completed at least one cycle of treatment and are calculated on a per-cycle basis. III
The effect of triphasic oral contraceptives on plasma lipids and lipoproteins Wolfgang Patsch, MD, Spencer A. Brown, PhD, Antonio M. Gotto, Jr., MD, DPhil, and Ronald L. Young, MD Houston, Texas To determine the effect of triphasic oral contraceptives on plasma lipid transport, 150 nonsmokIng women with normolipidemia, ages 18 to 35 years, were randomly assigned to receive one of three contraceptive formulations: (1) ethinyl estradiol, 30, 40, and 30 fLg/day, each for 6, 5, and 10 days per menstrual cycle, and levonorgestrel, 50, 75, and 125 fLg/day, each for 6,5, and 10 days; (2) ethinyl estradiol, 35 fLg/day for 21 days, and phased norethindrone, 500, 750, and 1000 fLg/day each for 7 consecutive days; and (3) ethinyl estradiol, 35 fLg/day for 21 consecutive days, and norethindrone, 500, 1000, and 500 fLg/day for 7, 9, and 5 days, respectively. A control group consisting of 49 women taking a nonhormonal form of contraception was also included. After 6 months of oral contraceptive treatment, significant increases in plasma triglyceride (28% to 52%) and plasma apolipoprotein B levels (20% to 23%) were observed in each treatment group. The changes in total plasma cholesterol (3% to 10%) and low-density lipoprotein cholesterol values (0% to 11%) were less striking . Changes in total high-density lipoprotein cholesterol levels were statistically insignificant (-2% to -4%); however, high-density lipoprotein2 cholesterol levels decreased by 29% to 33% and high-density lipoprotein 3 cholesterol levels increased by 20% to 23%. Concomitantly, plasma apolipoprotein A-I values increased by 5% to 12%. No consistent significant differences among analyses were observed between any of the groups receiving different oral contraceptives for 6 months. (AM J OaSTET GYNECOL 1989;161 :1396·401.)
Key words: Oral contraceptives, lipids, cholesterol, comparative trial
A number of large epidemiologic studies have demonstrated a relationship between oral contraceptive use and myocardial infarction and other vascular disorders. I•7 Attention has been directed toward the effects of oral contraceptives on plasma lipid transport, and increased concentrations of atherogenic lipoproteins in the circulation are the best established risk factors of coronary heart disease." Several studies have demonstrated changes of fasting plasma lipid levels as a result of contraceptive drugs . Depending on the dose, potency, and estrogen-progestogen content of the drug, From the Department of MediCine, Baylor College of Mediczne, and The Methodtst Hospital. R epnnt requests: Wolfgang Patsch, MD , Department of Medicine, Baylor College of Medzcine, Mall StatIOn A-60l, 6565 Fannin, Houston, TX 77030.
610116120
1396
increases in plasma triglyceride, cholesterol, very lowdensity lipoprotein, low-density lipoprotein, and highdensity lipoprotein values were observed.9- 16 In the great majority of these studies, monophasic preparations were tested. Data from such trials and other studies on sex hormones indicate that increases of highdensity lipoprotein cholesterol and plasma triglyceride levels are caused by the estrogen component, whereas increases of low-density lipoprotein cholesterol and a decrease of high-density lipoprotein cholesterol values reflect the androgenic potency of the progestogen in the formulation. Thus to a certain extent, effects of monophasic preparations on plasma lipid levels may be predicted from the composition of the drug. Much less is known about effects of biphasic and triphasic oral contraceptives, which are now the most frequently prescribed. We report here the results of a randomized