The Effect of Doxy: cycline on Serum Levels of Ethinyl Estradiol, Norethindrone, and Endogenous Progesterone JEFFREY L. NEELY, MD, MARIE ABATE, PharmD, MARIAN SWINKER, MD, AND RICHARD D‘ANGIO, PharmD Doxycycline and other antibiotics have been implicated in oral contraceptive (OC) failure, but information is sparse and studies of a doxycycline-OC interaction are nonexistent, Because an interaction between doxycycline and OCs, espe- cially those containing low-dose estrogen, could result in an unplanned and unwanted pregnancy, a controlled clinical trial of the effects of doxyeycline on OC hormone concentra- tions was performed. Twenty-four women aged 18-35 years ‘were recruited as volunteers from among the patients seen in a Universty-based family planning clinic. While they were fon a steady dose of the OC Orthe-Novum 1/35, serum concentrations of ethinyl estradiol, norethindrone, and endogenous progesterone were measured on days 18, 19, and 20 of the menstrual cycle (control phase). These mes- surements were repeated on days 18, 19, and 20 of the following menstrual cycle while the patient was taking doxycycline, 100 mg twice daily (treatment phase). No statistically significant diferences in serum levels of ethinyl estradiol, norethindrone, or endogenous progesterone were seen between the control and treatment phases. However, ‘there was large inter-patient and intra-patient variability in ethinyl estradiol and norethindrone levels. No elevations of endogenous progesterone occurred to suggest ovulation dut- ing antibiotic administration in either phase. Its not known what effects longer or earlier administration of doxycycline during the OC cycle would have on serum hormone concen- trations or ovulation. Pregnancies attributed to failure of (OCs because of tetracycline use could in fact be due to other ‘causes or could represent a true interaction that only mani- feat itself in a small proportion of women at risk. (Obstet Gynecol 77:416, 1991) From the Departments of Internal Medicine, Pharmacy, and Famnly Medicine, West Virginia University, Morgantown, West Virginia; and the Department of Pharmacy, University of New Mexico, Albuquerque, New Mexico. The authors would lke to acknowledge the contribution of Paul shiek, PAD, for statistical analysis ofthe dat Supported by a grant from the West Virginia University Faculty Senate and by the Johnson and Johnson Foundation. 416 0029-7844/9183.50 Unexpected pregnancies have occurred in isolated pa- tients taking oral contraceptives (OCs) concurrently with several types of antibiotics, including rifampin, ampicillin, neomycin, tetracycline, oxytetracycline, cotrimoxazole, metronidazole, cephalosporins, tri methoprim, erythromycin, sulfonamides, and griseo- fulvin.' Most of the antibiotic-OC interaction infor- ‘mation consists of case reports and limited studies in small numbers of women. The studies to date have evaluated primarily the effect of ampicillin on steroid hormone concentrations and have reported inconsis- tent results.” Research involving the interaction between tetracy- lines and OCs is scarce. The OC label warning about possible contraceptive failure while taking tetracycline initially resulted from a single case report.® Because of the inherent failure rate of approximately 1% with (OCS, it is impossible to evaluate by anecdotal reports whether an actual interaction exists. Many young women receive doxycycline, a tetracy- line derivative, for the treatment of chlamydia or gonococcal cervicitis. These sexually active women would also be likely to use OCs. Although doxycycline has been implicated in OC failure,? studies of a poten- tial doxycycline-OC interaction are nonexistent. Be- cause an interaction between doxycycline and OCs, especially those containing low-dose estrogen, could result in an unplanned, unwanted pregnancy, we performed a controlled clinical trial of the effects of doxycycline on OC hormone concentrations. Materials and Methods Twenty-four women aged 18-35 years were recruited as volunteers for a prospective study from among the patients seen in the University Health Service and outpatient clinics at the West Virginia University Obstetrics & GynecologyHealth Sciences Center. The candidates had been tak- 1g a low-dose OC containing 1 mg norethindrone and 35 pg ethinyl estradiol (Ortho-Novum 1/35; Ortho Pharmaceutical Corp., Raritan, NJ) for at least the preceding 2 months. We excluded subjects with a history of breakthrough bleeding, endocrine disorders, other concurrent medical illness, or allergy to tetracy- cline. The subjects gave informed written consent before study entry. During the study, the participants ‘were instructed to use an alternate method of contra- ception, which was provided free of charge. ‘The study consisted of a control and a treatment phase. The subjects were instructed throughout the phases to take their OCs in the evening at the same time each day, ie, 10-11 pm. Compliance was assessed by pill counts at office visits. The first 28-day OC cycle (21 days of contraceptive, 7 days off) represented the control phase. During the control phase, blood sam- ples were collected for the measurement of ethinyl estradiol, norethindrone, and endogenous progester- one approximately 12 hours after the OC was taken (3 hour range) on days 18, 19, and 20 of the cycle. We selected these days for hormone measurement because they were at the end of the cycle and would represent steady-state conditions for ethinyl estradiol and nor- ethindrone. They also constitute the final 3 days dur- ing which doxycycline would be given during the treatment phase. The blood was centrifuged and the plasma separated and frozen at ~20C until assayed. Because of potential intra-patient variability in the hormonal drug concentrations from day to day, three consecutive daily samples were taken to control for baseline variation. A serum pregnancy test was done before the treatment phase to ensure that no subject ‘was pregnant, During the following 28-day treatment phase, the subjects continued to receive their OC as in the control phase. The subjects were also given doxycycline (Vi- bramycin; Pfizer Laboratories, New York, NY), 100 mg by mouth twice a day for 7 days beginning on day 14 of the cycle. Day 14 was chosen arbitrarily, as it is unknown at which portion of the cycle an antibiot- ic-OC interaction would be most significant. Serum levels were drawn for the measurement of ethinyl estradiol, norethindrone, and endogenous progester- one in a manner identical to that used in the control phase. All subjects entered the control phase first in order to eliminate a possible doxycycline carry-over effect on the intestinal flora. However, the laboratories that analyzed the study samples were unaware of whether the samples were obtained during the control or treat- ment cycle. Progesterone was measured in all samples to deter- VOL. 77, NO. 3, MARCH 1991 mine whether ovulation had occurred.” The assays were performed by a radioimmunoassay technique described by Butcher,"" with modifications as outlined by Sheffel et al."? Ethinyl estradiol and norethindrone levels were measured by Hazelton Laboratories Amer- ica, Inc. (Vienna, VA), using a radioimmunoassay technique (Robard D, Hutt D. A four parameter sig- moidal curve fit with variable weighing for FIA and IRMA analysis. In: International Atomic Energy Sym- posia. Istanbul, Turkey, 1973 and Beaudry N, Listwak 8. Ethinyl estradiol radioimmunoassay validation and. methodology. Vienna, Virginia: Hazelton Biotechnolo- gies Co, project no. 265-133, 1985). The limits of detection of the ethinyl estradiol and norethindrone assays were 6.5 * 1.2 pg and 17.0 + 2.01 pg, respec- tively, The intra-assay and inter-assay variabilities were approximately 10 and 14%, respectively, for both. ethinyl estradiol and norethindrone (data available from Hazelton Laboratories). Because each subject served as her own control, the paired t test was used for data analysis. ‘All subjects were reimbursed for travel and time spent in the clinic. The study was approved by the Human Subjects Committee of West Virginia Univer- sity (Institutional Review Board). Results ‘Twenty-three women completed the study, with one (no. 17) dropping out after developing an unrelated illness. Pregnancy was not reported during the study oat the 2-month follow-up in any subject, One subject vomited on day 1 of the treatment phase, but the vomiting abated and she completed the study. Two subjects reported breakthrough bleeding during the treatment phase, compared with none during the control phase. Table 1 lists the mean (= standard deviation [SD)) concentrations of ethiny| estradiol, norethindrone, and progesterone for each subject for days 18, 19, and 20 of the control and treatment phases. The mean ethinyl estradiol concentration for all subjects in the control phase was 87.6 + 55.9 pg/mL, compared with 81.3 * 39.3 pg/mL during the treatment phase. The norethin- drone concentration in the control phase averaged 4825 * 3358 pg/mL, compared with 4919 = 2809 pg/mL. during the treatment phase. Progesterone concentra- tions ranged from an overall mean of 0.5 + 0.2 ng/mL. during the treatment phase to 0.6 + 0.2 ng/mL during the control phase. These concentrations imply lack of ovulation since luteal phase progesterone levels are above 2.0 ng/mL." None of the individual subjects’ progesterone concentrations exceeded 0.8 ng/mL dur- ing the treatment phase. Four serum hormone concen- trations were not used during calculation of the mean Neely et al OCs and Doxycycline 417Table 1, Mean Ethinyl Estradiol, Norethindrone, and Endogenous Progesterone Levels thinylestadiol (pg/ml) ‘Norethindrone (pg/mL) Progesterone (ng/mL) ‘Subject no Control ‘Treatment Control ‘Treatment Control ‘Treatment 1 12732117, 119.3 = 127 5370 = 533 5120 = 199 05201 06201 2 6611 = 15.0 86.9 237.7 4310 = 524 4830 = 972 06 +02 0401 3 22118 644273 5020 = 191 3920 = 485 0401 04 04 4 1653 = 402 15732219 6320 = 3039 7490 = 1924 06 +02 05 £01 5 137 #38 109.2 = 396 4567 = 101 4521 = 816 os +01 07 +01 e 709.2 = 104.2 409225 17,375 2 13,9 5793 = $033 o7=01 o7 04 7 5.9 = 367 143 = 687 2238 = 1532 ‘4890 + 2785, 09 +01 a7 +01 3 val = 177 1343 = 30 5620 + 2239 4593 + 817, 05 #0. 05 +01 1398 = 709 888 +68 4830 + 2630 4623 = 520 05 =0.1 05 +01 10 71.9 = 333 53116 3977 = 260 3900 252 4201 a4 =00 Mw 71.0 +137 4682135 207 = 811 3973 = 380 03 £0. 04 = 00 ne 5822 134 73.02 45 2503 = 124, 4260 = 1428 04 = 00 04 = 0.0 B 61231 51.6 = 26. 453 = 452 3677 = 594, 04s 01 03=01 4 45.0 = 96 ne =23 2140 = 306 567 = 561 o4 = 04 5 474s 184 622163 6769 = 1576 6563 = 3138 os = 04 6 "45275 3203 4787 = 1448, 4523 + 26 0602 rd 1B 749 = 489 1253 + 662 2017 « 1330 12,182 = 9392 07 = 0a » 1504 + 139.1 27 = 125 6750 = 4462 4293 + 350 0801 » 85.7 = 257 80.6 = 203 $537 = 2550 4523 = 1506 04 = 01 2 mi =a1 614 = 66 3240 = 241 3277 = 262 06201 2 580239 70227 4173 = 631 4230 = 609 05 04 2 407 = 121 282 = 04 2163 = 837 1930 = 234 06 = 0.0 4 634277 5216 5857 = 660 4897 = 569 04200 Total mean, 87.6 * 55.9 813 +393 4925 = 3358, 4919 + 2509 os =02 aa are presented es mean = SD, * Subject vomited on day 1 of treaiment phase; one value for day 2 control norethindrone wat unavailable, 1 One value fr day 2 contol norethindrone was discarded. + One value for day 3 teatment norethindrone was discarded $ Subject dropped from study because of unrelated illness. * One value for day 2teatment ethinyl estrada was discarded values because of difficulties with specimen collection or grossly aberrant reported values (Table 1). Because the concentration values seen during days 18, 19, and 20 in some subjects did not appear to follow @ normal distribution, we used a nonparametric test, the Wilcoxon sign rank, in addition to the t test for statistical analysis. No significant differences were de- tected between the control and treatment phases for ethinyl estradiol (P = .49, paired test; P = .96, Wilcoxon sign rank test), norethindrone (P = .356, paired f test; P = .69, Wilcoxon sign rank test), progesterone (P = .319, paired f test; P = .33, Wilcoxon, sign rank test). Discussion Antibiotics such as tetracycline or penicillins are thought to interact indirectly with contraceptive ste- roids by suppressing the intestinal bacterial flora. This interferes with enterohepatic circulation of the steroids by diminishing the availability of hydrolytic enzymes to regenerate the parent steroid molecule. Conse- quently, plasma concentrations of estrogens are pos- 418 Neely et al OCs and Doxyeycline tulated to be abnormally low, allowing ovulation and/or breakthrough bleeding.*"* Study of the antibiotic-OC interaction is complicated by the variability reported in the contraceptive steroid pharmacokinetic studies to date. In addition, the amount of decrease in steroid hormone levels that would constitute clinical significance is not known. Although one review concluded that there is a second- ary peak in the plasma concentration-time profile of ethinyl estradiol, due possibly to enterohepatic circu- lation," this effect has not been noted by others.1®7 Extensive inter-subject variability has also been re- ported in the plasma concentrations obtained follow- ing doses of norethindrone and ethinyl estradiol “2° with norethindrone levels 10-12 hours after a I-mg dose ranging from 1600-15,200 pg/mL.® Time since pill ingestion accounted for only 25% of the variability in ethinyl estradiol levels reported in one study. Unfortunately, multiple-dose pharmacokinetic stud- ies of intra-subject variability (ie, variability within the same subject on repeated occasions) have not been published, although their desirability has been indicat- ed.” A single-dose study comparing three bioequiva- Obstetrics & Gynecologylent ethinyl estradiol-norethindrone preparations re- ported substantial inter- and intra-subject variability in the hormone concentrations achieved.” Although we found no statistically significant effect of doxycycline on ethinyl estradiol or norethindrone serum concen- trations, substantial patient variability existed. Intra- individual differences in the three study-day mean values between the control and treatment phases ranged from a low of 1 pg/mL to a high of 168 pg/mL for ethinyl estradiol and a low of 37 pg/mL toa high of 11,582 pg/mL for norethindrone. The SDs of the 3-day means for individual subjects were also generally large, ranging from less than 1.0 pg/mL to 139.1 pg/mL for ethinyl estradiol and 101.7-13,329 pg/mL for nore- thindrone. ‘The day-to-day variations in norethindrone and ethi- nyl estradiol concentrations could have resulted from differences in the times the OC was taken, although the subjects were told to take the tablets at approxi- mately the same time each evening. Because mealtime was not controlled in the subjects, it is also possible that on some days a rebound in blood hormone con- centrations occurred, due to bile “dumping” after a ‘meal.”" Differences in metabolism or intestinal absorp- tion could also explain the extensive inter-patient vari- ability and, to some extent, the intra-patient varia~ bility.” Given the variation in hormone concentrations that can exist in individuals, it has been speculated that patients with unusually low steroid hormone levels may be the ones at risk for an antibiotic-OC interac- ton 151623 Studies of other antibiotics have not found a signif- icant interaction with OCs. Controlled clinical trials with ampicillin have failed to show a consistent, sig- nificant effect on contraceptive steroid serum levels.” In six patients taking tetracycline, 250 mg orally twice daily, tetracycline decreased ethinyl estradiol plasma levels, but to a nonsignificant degree."* The mean plasma concentrations of ethinyl estradiol were 54.5 + 11.9 pg/mL in the control cycle and 49.3 + 7.5 pg/mL during tetracycline therapy. Among four women tak- ing an OC and oral tetracycline 250 mg four times daily, the tetracycline resulted in an increase in fecal excretion and a decrease in urinary excretion of ethinyl estradiol in two women (Swenson L, Goldin B, Gor- bach SL. Effect of antibiotics on fecal/urinary excretion of ethinyl estradiol, an oral contraceptive [abstract]. Gastroenterology 1980;78:1332). This phenomenon suggests decreased enterohepatic circulation and pos- sibly decreased contraceptive effectiveness. This lack of significance could be a result of a small number of patients (je, type II statistical error), particularly given the large SDs in hormone concentrations in our study. In the present study, ovulation did not appear to VOL. 77, NO. 3, MARCH 1991 have occurred during week 3 of the cycle as a result of doxycycline administration. Questions remain, how- ever, such as the possible effect on progesterone co! centrations if the antibiotic had been taken at a di ferent time during the cycle or for a longer period. It has been suggested that pregnancy is more likely if an. OC pill is skipped early in the cycle; itis possible that the same would be true for an antibiotic-OC interac- tion, Itis also not known whether a delayed effect on ‘ovulation would have developed with ovulation and elevated progesterone concentrations occurring during, the subsequent 7-day break. Pregnancies that have been attributed to OC failure secondary to a tetracycline-OC interaction could reflect the background failure rate of the OCs or represent an. ‘uncommon idiosyncratic response. Most of the case reports of OC failure have not mentioned the concur- rent use of other medications; therefore, other drug interactions in these failures are a possibility. Alterna- tively, a true interaction between antibiotics and OCs might manifest itself only in women with unusually low steroid hormone concentrations during OC ad- ministration. Subjects 23 and 14 (Table 1) might repre- sent such examples. Ideally, further controlled studies should be performed in subjects preliminarily identi- fied as having low steroid concentrations. Antibiotic administration early in the cycle could also be tried. Because of the reported variability in drug concentra- tions, large numbers of women would be needed, along with more frequent sampling for hormone levels in a controlled setting. References 1, Skolnick JL, Stoler BS, Katz DB, Anderson WH. Rifampin, oral ‘contraceptives, and pregnancy. JAMA 1976;236:1382. 2, Barnett ML. Inhibition of orsl contraceptive effectiveness by ‘concurrent antibiotic administration. J Periodontol 1985;56:18-20. 3, Silber TJ. Apparent oral contraceptive failure associated with antibiotic administration. J Adolesc Health Care 1983:4:287-9. 4, Back Dj, Grimmer SFM, Orme MLE, Proudlove C, Mann RD, Breckentidge AM. Evaluation of Committee on Safety of Medi- cines yellow card reports on oral contraceptive-drug interactions with anticonvulsants and antibiotics. Br J Clin Pharmacol 1986; 25507-32 5. Back Dj, Breckenridge AM, Maciver M, et al. The effects of ampicillin on oral contraceptive steroids in women. Br J Clin Pharmacol 1982;14:43-8, 6, Friedman Cl, Huneke AL, Kim MH, Powell J. The effec of ampicilin on oral contraceptive effectiveness. Obstet Gynecol 1980:5.33-7. 7, Joshi JV, Joshi UM, Sankholl GM, etal. A study of interaction of low-dose combination oral contraceptive with ampiciin and ‘metronidazole. Contraception 1980;22:643-52. ‘8 Hudson CP, Callen JP. The tetracycine-oral contraceptive con- troversy. J Am Acad Dermatol 19827:269-70. 9, Fleischer AB, Resnick SD. The effect of antibiotics on the efficscy of oral contraceptives. Arch Dermatol 198;7:269-70. Neely et al OCs and Doxycycline 41910. uy, 2. B. 14 1. 16. v. 18. 18. 420 Neely et al Israel R, Mishell DR, Stone SC, Thomeycrott IH, Moyer DL. Single hiteal phase serum progesterone assay as an indicator of ‘ovulation. Am J Obstet Gynecol 1972/112:1083-6 Butcher RL. Changes in gonadotropins and steroids associated with unilateral ovariectomy of the rat. Endocrinology 1977/10: 830-40. Shefel CE, Pratt BR, Ferrell WL, Inskeep EK. Induced corpora lutea in the postpartum beef cow IL. Efects of treatment with progestogen and gonadotropins. } Anim Sci 1982;54830-6, Sciarra J, ed. Gynecology and obstetrics. Vol 5, Endocrinology, Infertility and genetics 3rd ed. Philadelphia: Harper and Row, 198315. (Orme ML, Back DJ. Drug interactions between oral contraceptive steroids and antibiotics. Br Dent J 1986;160:169-70. Back Dj, Breckenridge AM, Crawford FE, Maciver M, Orme MLE, Rowe PH. Interindividual variation and drug interactions with hormonal steroid contraceptives. Drugs 1981/21:46-61 Kanarkowski R, Tornatore KM, D’Ambrosio R, Gardner MJ, Jusko WJ. Pharmacokinetics of single and multiple doses of ethinyl estradiol and levonorgestrel in relation to smoking. Clin Pharmacol Ther 1988/3;23-31 Stancayk FZ, Bennet PF, Mishell DR, Ortiz A, Gentaschein EKE, Goebelsmann U. A radioimmunoassay for norethindrone (net): Measurement of serum net concentrations following insertion of net-containing oral contraceptive steroids. Contraception 1978:18: 5-33. Beck DJ, Breckenridge AM, Crawford FE, et al. Kinetics of norethindrone in women. Clin Pharmacol Ther 1978;24439-47. Stadel BV, Sternthal PM, Schlesselman JJ, etal. Variation of OCS and Doxycycline ‘ethinyl estradiol blood levels among healthy women using oral contraceptives. Fertil Steril 1980;33257-60, 20. Brody SA, Turkes A, Goldzieher JW. Pharmacokinetics of three bioequivalent norethindrone/mestranol—S0 micrograms and three norethindronelethiny estradio!—35 micrograms OC formu lations: Are “low-dose” pills relly lower? Contraception 1989;40: 269-84. 21, Newburger J, Goldzicher JW. Pharmacokinetics of ethinyl estra- dliok A current view. Contraception 1985;3233-44 22 Stanczyk FZ, Mroszezak EJ, Ling T, et al. Plasma levels and pharmacokinetics of norethindrone and ethinyl estradiol admin- istered in solution and as tablets to women. Contraception 1983;28241-51, 23, Fotherby K. Variability of pharmacokinetic parameters for con- tnaceptive steroids J Steroid Biochem 1983;19:817-20. 24. Fraser IS, Jansen RP. Why do inadvertent pregnancies occur in ‘oral contraceptive users? Effectiveness of oral contraceptive rege ‘mens and interfering factors. Contraception 1983,27531-51. Reprints are not available. Received June 28, 1990, Received in revised form November 5, 1990. Accepted November 14, 1990. Copyright © 1991 by The American College of Obstetricians and Gynecologists Obstetrics & Gynecology