FERTILITY AND STERILITY Copyright 1987 The American Fertility Society Printed in U.8A. Prediction of ovulation with the use of oral and vaginal electrical measurements during treatment with clomiphene citrate* Ranjit S. Fernando, Ph.D.t Jennine Regas, M.S. t George Betz, M.D., Ph.D.t Zetek, Inc., Aurora, and University of Colorado Health Sciences Center, Denver, Colorado This study was undertaken to evaluate the usefulness of measuring salivary and vaginal electrical resistance in monitoring ovulation induced by clomiphene citrate (CC). Data from 28 cycles of 12 women treated with CC were compared with those of 18 cycles of 13 women who were ovulating spontaneously. Patterns of salivary readings in CC and spontaneous cycles were similar and showed a preovulatory peak 6.2 (mean) days before the luteinizing hormone peak. The trend in vaginal readings for CC cycles differed from that of spontaneous ovulations in that the values were depressed during and shortly after CC therapy. Occurrence of the periovulatory nadir and subsequent rise was common to both groups. The rise in vaginal readings CC cycles occurred within 24 hours of the luteinizing hormone peak in 88% of cases. Retrospective analysis showed that, based on this method, artificial insemination would have been timed appropriately in 25 of 27 cycles or in every cycle, depending on the protocol used. The results indicate that the method is equally useful for predicting and confirming ovulation in cycles in which CC was used to induce ovulation as it is in spontaneous cycles. Fertil Steril47:409, 1987 Clomiphene citrate (CC) is the drug of choice for the treatment of many anovulatory patients. 1 The standard treatment plan is to administer CC orally for 5 days beginning on the fifth cycle day. Ovulation is usually expected 5 to 10 days after the last day of CC medication. 2 Measurements of salivary and vaginal electrical resistance (SR and VR ) with a commercially Received July 7, 1986: revised and accepted November 24, 1986 *Supported in part by the National Institutes of Health United States Public Health Service grant 1 R43 HD 2022-01 to Zetek, Inc. tzetek, Inc., Aurora, Colorado. :j:reprint requests: George Betz,.D., Ph.D., Director, Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Denver, Colorado 80262. available instrument (CUE Ovulation Predictor, Zetek, Inc., Aurora, CO) have been shown to be useful for prediction and confirmation of ovulation in spontaneously ovulating women. 3 In that earlier work, a peak in SR was observed 5 to 6 days before the day of the serum luteinizing hormone (LH) peak and recordings of VR demonstrated a nadir on the day of the LH peak, with a pronounced increase the following day. It was determined that SR and VR could predict ovulation several days in advance, as well as providing near-term indications of this event. The present study was undertaken to determine whether the above changes in SR and VR reported for spontaneous cycles would be altered when ovulation was In an earlier paper3 the abbreviations SER and VER were used. They have been shortened in usage, but the two forms are interchangeable. Fernando et al. Ovulation prediction with CC 409
induced with CC treatment and to evaluate the potential usefulness of this method for monitoring ovulation induction by CC. SUBJECTS MATERIALS AND METHODS Data were obtained from 28 menstrual cycles in 12 patients. Eleven women were studied for 1 to 3 cycles each, and one was studied for 6 cycles (mean cycles per subject, 2.3). The sole inclusion criterion was that they were receiving CC for some degree of oligoovulation, amenorrhea, or luteal phase defect. Varying regimens of CC administration were used. CC (50 to 200 mg/day) was given for 5 to 7 days, beginning on cycle day 2, 3, or 5. In all but two of the cycles, the treatment duration was 5 days and the CC dosage was 50 to 100 mg/day. In 70% of these cycles, treatment began on cycle day 5. In two cycles, 200 mg/day CC was given for 7 days beginning on cycle day 6. The results were compared with control data from 18 cycles of 13 spontaneously ovulating women, who were studied for 1 to 2 cycles each (mean number of cycles per subject, 1.38). Some of these subjects were candidates for artificial insemination because of the partner's infertility and some were volunteers. All were determined to have normal ovulatory function, according to previous basal body temperature and/or monitoring of LH. These were data from a different set of subjects than were previously reported.3 EXPERIMENTAL MEASUREMENTS Procedures for measurement of SR and VR were as described elsewhere.3 Each day, the subject recorded the date, day of cycle, basal body temperature (BBT), SR, and YR. In nine CC cycles and three control cycles, VR measurements were not taken; in one cycle BBT was not available. However, because SR data were available, all of these cycles were included in the analysis. The ovulatory phase of the cycle was documented with serum LH assay, ultrasound, and/or urine LH assay, depending on clinical circumstances. In two cases, human chorionic gonadotropin (hcg) was administered, based on ultrasound evaluation of follicle maturity. Although ovulation is expected approximately 37 hours after the administration of hcg 4 but only 10 to 15 410 Fernando et al. Ovulation prediction with CC hours after the LH peak, 5 the day of hcg administration in these two cycles was taken as equivalent to the cycle day of the. LH peak. DATA ANALYSIS In conformity with previously reported procedures,3 we standardized the SR data relative to the preovulatory SR peak value, which was defined as a high value followed by two or more markedly lower readings. The VR data were standardized relative to the day of the VR nadir. Analysis for statistically significant differences among important features of the trends of both SR and VR was with the paired t-test. To make an objective comparison of the differences in VR trend between spontaneous and CC cycles, we adjusted the VR data within each cycle relative to the value at the periovulatory nadir of VR. We also compared the mean (± standard error of the mean [SEM)) of the adjusted VR data in spontaneous and CC cycles. The potential effectiveness of timing insemination with SR and VR was evaluated with retrospective analysis of the CC data. We tested two protocols for timing of insemination, based on the reported relationships of SR and VR to the time of ovulation.3 For the primary protocol (I), we proposed, whenever possible, only one artificial insemination (AI) in each cycle. The proposed day of insemination in protocol I was 6 days after the SR peak or on the day of the VR rise, whichever came first, with a second AI if the VR rise was more than 1 day after the first insemination. The alternative protocol (II) was two inseminations in all cycles, performed 5 and 7 days after the SR peak. The period within the menstrual cycle when conception is most likely to occur (period of maximum conception probability) has been defined as ranging from 2 days before the LH peak to 1 day after the LH peak. 6 For purposes of this study, however, we considered the timing appropriate only if the proposed day of insemination was within 24 hours of the day of the LH peak. SPONTANEOUS CYCLES RESULTS The SR and VR trends in spontaneous ovulations are shown in Figure 1. The LH peaks were distributed from 5 to 7 days after the SR peak, with 83.3% occurring on day 6. The VR data from Fertility and Sterility
175 180 165 155 Q 180 140 120 ; 145 100 '-'" 80 :! 135 iii: 60 125 40 160 180 148 180 i'136 w 140 i (II... 124 -c 120 S z 112 L 100 :5 100. 0 z is 80 -c M w 88 II' t 80. z w z '16 (3 64 40 115 20 52 20,os -6-4 -2 0 2 4 6 8 10 12 14 16 1820 DAYS FROM SR PEAK a 40' ---4-1---4-1. 0 -/I -6-4 -2-0 2 4 6 8 10 DAYS FROM VR NADIR Figure 1 Changes in SR and VR of spontaneously ovulatory cycles, showing the relationships to the timing of the LH peak. Data from all cycles are standardized to the SR and VR peaks (day 0). spontaneous cycles showed a decline to a nadir during the periovulatory period, with mean VR on the day of the nadir significantly lower (P < 0.01) than on either the preceding or the following day. In 93% of the cycles, VR nadirs were distributed between the day of the LH peak and the preceding day. CLOMIPHENE CITRATE CYCLES LH Peak Distribution The time of ovulation was determined in all but one cycle with the use of midcycle LH assay (25 cycles) or the time of hcg administration guided by ultrasound (2 cycles). In the remaining cycle, the LH peak was missed; therefore, this cycle was excluded from further analysis. The day of the LH peak ranged from cycle day 12 to 26, with 44.4% observed by day 15. More than 90% of LH peaks had occurred by day 17. The relationship between the last day of CC administration and the day of the LH peak is shown in Table 1. Trends of Salivary and Vaginal Readings Mean (± SEM) SR and VR values from CC cycles is shown in Figure 2. The mean SR value at the peak was significantly higher (P < 0.01) than the mean values on the days just before and after it. Also, the peak value was significantly higher (P < 0.01) than the mean for all other cycle days. LH peaks were observed from day 4 to day 8 (mean, 6.2) after the preovulatory peak of SR. In 81.5% of cycles, LH peaks occurred from day 5 to day 7 after this SR peak. The regression of the day of the SR peak on the day of the LH peak indicated a significant (P < 0.001) linear, and positive relationship between the two variables, with a coefficient of correlation (r) of 0.96. The mean (± SEM) VR readings for the CC-induced ovulations is shown in Figure 2, and the mean changes in VR relative to the periovulatory nadir for both CC and control cycles are shown in Figure 3. The VR values in CC cycles were depressed during the preovulatory phase, compared with the control cycles, with the mean VR value for CC cycles 5 days before the LH peak significantly (P < 0.01) lower than that for control cy- Table 1. Distribution of LH Peaks Relative to Last CC Dose Days from last CC 4 5 6 7 8 9 17 LH peaks % No. 3.7 7.4 48.1. 18.5 14.8 3;7 3.7 1 2 13 5 4 1 1 Cumulative % 3.7 11.1 59.2 77.7 92.6 96.3 100.0 Fernando et ai. Ovulation prediction with CC 411
155 100 135 150 150 90 124 135 'i' 1-45 80 113 120 [ UI iii 0140 70 102 105 -C Z -C 91 90 a..! 135 60 :5 80 7S 130 50 0 \M " z 1l1 69 60 M 125 40 E' 0: 58 45 z a:: 120 30 z :::> 0 Ci 47 30 ;! 115 20 36 15 110 10 25 O 105 0-8 --6 --4-2 0 2 4 0 8 10 12 14-8-4-2 0 2 4 6 8 10 12 14 16 1820 DAYS FROM SR PEAK DAYS FROM VR NADIR Figure 2 Changes in SR and VR of CC-induced cycles, showing the relationships to the timing of the LH peak. Data from all cycles are standardized to the SR and VR peaks (day 0)... cles at the same stage. These depressed readings were due to a pronounced decrease ofvr observed around the end of CC therapy. As in the control cycles, a peri ovulatory decline to a VR nadir was observed. Its value was significantly lower than the mean for all other days of the cycle (P < 0.01), as well as that of the preceding day (P < 0.05). On the day after the nadir, a significant increase (P < 0.01) in mean VR was seen. The periovulatory VR nadir coincided with the cycle day of the LH peak in 47.4% of cases and occurred during the preceding 2 days in another 47.4%. The VR rise was observed within ± 24 hours of the day of the LH peak in 88.2% of the cycles (Table 2), with the mean interval from LH peak to VR rise being 0.8 days. The r value between cycle days of LH peak and VR peak was 0.94. After the initial rise from the nadir, VR increased for 1 day in the CC cycles, after which a decline to lower values was observed. This is in contrast to the pattern observed in the control cycles, which showed a continuing rise during much ofthe luteal phase. Seven days after the LH peak, the VR value for CC cycles (mean, 81) was significantly lower (P < 0.01) than that for spontaneous cycles (mean, 131). Basal Body Temperature Table 3 shows the distribution of the day of the shift in BBT, from 2 days before the LH peak to 4 days after it. The time relationships of the SR peak, VR nadir and rise, and BBT shift to the cycle day of the LH peak are summarized in Figure 4. The time relationships between the various parameters and the LH peak are essentially similar for spontaneously ovulatory and CC cycles. EVALUATION OF THE PROPOSED INSEMINATION PROTOCOLS With the use of protocol I, the day(s) of proposed insemination were evaluated for appropriateness -- CC CYCLES 80r- -_-_-- =spontane=ouso=vulations 70 10.-ti.. :j-j. t,'l ' 't,,,\,ii \, '/ }-t -8-6 -4-2 0 2 4 6 8 10 12 14-7 -5-3 -1 1 3 5 7 9 11 13 DAYS FROM VR NADIR Figure 3 Relationship of the end of CC therapy to the occurrence of the CC-related VR nadir. 412 Fernando et al. Ovulation prediction with CC Fertility and Sterility
Table 2. Distribution of VR Rise Relative to LH Peak Days from VR rises LH peak % No. Cumulative % -1 17.6 3 17.6 0 35.3 6 52.9 1 35.3 6 88.2 2 11.8 2 100.0 LEGEND CC CYCLES o SPONTANEOUS OVULATIONS BBT SHIFT"""""'-; VR RISE t-+-i of timing. In 25 of the 27 cycles, insemination was judged to be correctly timed at ± 24 hours of the cycle day of the LH peak. In 63% of the cycles, only one insemination would have been necessary. With protocol II, at least one of the two inseminations would have been performed within the appropriate time in all cycles. DISCUSSION Although the expected day of ovulation ranges from 5 to 10 days after the last CC dose, CC administration is considered useful for attempting to regulate the timing of ovulation in women undergoing donor AI2. and has also been used in timing oocyte recovery.7 However, data in our study showed that LH peaks were distributed from 4 to 17 days after the last CC dose, indicating that scheduling of AI based on the last day of CC would not always be reliable. SR trends from CC and spontaneous cycles are similar (Figs. 1 and 2). The SR peak is as readily identified in CC cycles as in spontaneous cycles. In the spontaneous cycles represented here, the distribution of LH peaks relative to the SR peak was narrower than for CC cycles, with a 6-day interval between the two events in 83.3% of cycles. However, because a previous study of spontaneously ovulating women 3 showed a distribution similar to that ofthe CC cycles of the present study, we believe that the difference in the distribution between the two groups reported here is not significant. Table 3. Distribution of BBT Shift Relative to LH Peak BBT shifts Dsfrom peak % No. Cumulative % -2 3.8 1 3.8-1 3.8 1 7.6 0 11.5 3 19.1 1 26.9 7 46.0 2 15.4 4 61.5 3 30.8 8 92.3 4 7.7 2 100.0 SR PEAK VR NADIR"""-"; -8-7 -6-5 -4-3 -2-1 0 2 DAYS FROM LH PEAK Figure 4 Time intervals (mean and 95% confidence interval) from the SR peak, VR nadir, VR rise, and BBT shift to the LH peak for CC-induced and spontaneous ovulatory cycles. The first increase in estrogen and the appearance of mucus are recognized to be closely associated with the development of the dominant follicle. 8 The mean interval from the SR peak to the LH peak, approximately 6.2 days for both CC and spontaneous cycles, is similar to the reported interval from the first estradiol rise 9 and that from first observed mucus to ovulation.lo However, the interval from first observed mucus to ovulation ranged from 4 to 9 days in spontaneous cycles,io whereas that from estradiol rise to follicle maturity ranged from 5 to 7 days in cycles in which ovulation was induced. ll Thus the observed SR peak appears to be a more sensitive indication than mucus evaluation for the onset of follicular activity. In contrast to the VR trend in spontaneous cycles, which showed a progressive decline to a nadir during the peri ovulatory period, the pattern of the CC cycles showed initially depressed readings for several days (Figs. 2 and 3). Low VR indicates increased volume and/or increased electrolyte concentrations of cervicovaginal fluid and is usually seen in response to estrogens. 3 These low readings were unexpected, in view of the wellknown anti estrogenic effect of CC on cervical mucus parameters. 1, 7 This decrease in readings is most pronounced near the end of CC therapy, when concentrations of CC are probably at their highest. This observation leads to the speculation Fernando et al. Ovulation prediction with CC 413
that the decreased values are the result of the weak estrogenic effects inherent in CC. 12 The VR nadir coincided with the cycle day of the LH peak or preceded it by 1 or 2 days in 94.8% of CC cycles; this is a broader distribution than in the control cycles. CC therapy has been found to affect the mucus in 15% to 50% of cycles studied. 1 Thus it is possible that the greater variability of the VR nadir relative to the LH peak in CC cycles than in control cycles may be due to the variable effect of CC on the characteristics of cervical mucus. The VR rise following this preovulatory nadir appeared to be closely associated with the time of ovulation, as suggested by the significant correlation between the days of the VR rise and the LH peak. In contrast to the VR trend of the luteal phase of control cycles, a gradually declining pattern of VR was observed during the luteal phase of CC cycles (Fig. 3). In the normal menstrual cycle, declining VR values are usually seen in response to the mucus changes associated with high estrogenic activity, whereas increased VR values are observed during the luteal phase, when progesterone (P) is dominant even though estrogen is present at moderate levels. 3 The unusually low luteal phase readings in CC-induced ovulations may suggest a higher-than-usual estrogen-to-p ratio. It has been shown that viable embryos failed to implant in women who had abnormally high estrogen/p ratios during the luteal phase. 13 These abnormal ratios were the result of elevated estrogen levels, rather than low P. Dissimilarity of the luteal phase VR trends between the two groups in our study might indicate an estrogen dominance during the luteal phase in the CC subjects and should be investigated further. EFFECTIVENESS FOR TIMING OF INSEMINATION We performed a retrospective analysis to determine the efficacy of VR and SR measurements for appropriate timing of insemination, assuming that an insemination is timed correctly if it is within 24 hours of the serum LH peak. Although self-testing for urine LH has resulted in a reduction in both cost and inconvenience, compared with serum analysis of LH, a disadvantage is that the detection of the urine LH surge is variably delayed relative to the surge in plasma. 14 This delay can result in narrowing the time window during which AI would have the highest probability of success. It may also explain the observation 414 Fernando et a1. Ovulation prediction with CC of ultrasonic evidence of follicle collapse in 13 of 23 cycles before the expected time of ovulation based on the urinary LH surge. 15 With the use of protocol I, in CC-treated subjects, AI would have been timed within 24 hours of the LH peak in 25 of the 27 cycles (92.6%). In the remaining two cycles, AI would have been performed in one subject on day - 2 relative to LH peak and on day + 2 in the other. With protocol II, although two inseminations per cycle would have been performed, at least one of the procedures would have been correctly timed in each cycle. In summary, our data suggest that SR and VR can be useful in monitoring CC-induced ovulation. Differences in VR trends between CC and control cycles are definable and may be caused by the effects of the drug on cervical mucus. The trends of the oral readings between the control and CC-treated subjects were similar. Retrospective analysis of the data indicates that the method has potential for high accuracy in timing AI. REFERENCES 1. Hammond MG: Monitoring techniques for improved pregnancy rates during clomiphene ovulation induction. Fertil Steril 42:499, 1984 2. Speroff L, Glass RH, Kase NG: Induction of ovulation. In Clinical Gynecologic Endocrinology and Infertility, 3rd ed. Baltimore, Williams & Wilkins, 1983, p 528 3. Albrecht BH, Fernando RS, Regas J, Betz G: A new method for predicting and confirming ovulation. Fertil Steril 44:200, 1985 4. Testart J, Frydman R: Minimum time lapse between luteinizing hormone surge or human chorionic gonadotropin administrations and follicular rupture. Fertil Steril 37:50,1982 5. Garcia JE, Jones GS, Wright GL: Prediction of the time of ovulation. Fertil Steril 36:308, 1981 6. Schiphorst LEM, Collins WP, Royston JP: An estrogen test to determine the times of potential fertility in women. Fertil Steril 44:328, 1985 7. Maxson WS, Pittaway DE, Wentz AC: Antiestrogenic effect of clomiphene citrate: correlation with serum estradiol concentrations. Fertil Steril 42:356, 1984 8. Moghissi KS: Prediction and detection of ovulation. Fertil Steril 34:89, 1980 9. Brown JB, Harrison P, Smith MA, Burger HG: Correlations between the mucus symptoms and the hormonal markers of fertility throughout reproductive life. Melbourne, Advocate Press, 1981 Fertility and Sterility
10. Billings EL, Billings JJ, Brown JB, Burger H: Symptoms and hormonal changes accompanying ovulation. Lancet 1:282,1972 11. Levran D, Lopata A, Nayudu PL, Martin MJ, McBain JC, Bayly CM, Speirs AL, Johnston WIH: Analysis of the outcome of in vitro fertilization in relation to the timing of human chorionic gonadotropin administration by the duration of estradiol rise in stimulated cycles. Fertil Steril 44:335, 1985 12. Clark JH, Markaverich BM: The agonistic-antagonistic properties of clomiphene: a review. Pharmacol Ther 15:467, 1982 13. Gidley-Baird AA, O'Neill C, Sinosich MJ, Porter RN, Pike IL, Saunders DM: Failure of implantation in human in vitro fertilization and embryo transfer patients: the effects of altered progesterone/estrogen ratios in humans and mice. Fertil Steril 45:69, 1986 14. Frydman R, Testart J, Feinstein M: Comparison between plasmatic and urinary detection of LH surge in human oocyte recovery. Presented at the Third World Congress on Human Reproduction, Berlin, 1981, p 195 15. Queenan JT, O'Brien GD, Bains LM, Simpson J, Collins WP, Campbell S: Ultrasound scanning of ovaries to detect ovulation in women. Fertil Steril 34:99, 1980 Fernando et al. Ovulation prediction with CC 415